Cancer risk evaluation method and cancer risk evaluation system

Abstract

[Problem] To provide a cancer risk evaluation method that makes it possible to objectively and highly accurately estimate a cancer incidence risk of a subject by using a urine specimen that is a biological sample that can be collected much more easily than blood. [Solution] Concentration data of an evaluation element group in a urine specimen 2 collected from a subject is acquired (step S1), the concentration data is applied to a discrimination function for discriminating whether the subject belongs to a control group or a case group to calculate a correlation between concentrations of the evaluation element group (step S2), an index indicating whether the subject suffers from cancer is generated on the basis of the correlation (step S3), and an evaluation result of a cancer incidence risk of the subject is created on the basis of the index (step S4). As the evaluation element group, a combination of 20 elements of Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl is used.

Description

Cancer risk assessment method and cancer risk assessment system

[0001] The present invention relates to a method and system for assessing cancer risk, and more specifically, to a method and system for assessing the risk of developing cancer in a person using an index obtained by utilizing the concentration balance of a group of elements contained in human urine (the correlation between the concentrations of the group of elements).

[0002] Diagnostic methods for cancer include methods that involve direct visualization or palpation (palpation, endoscopy, etc.), methods that use images of the inside of the body (X-ray, CT scan, MRI scan, PET scan, etc.), and methods that examine blood or cells (blood tests, cytology, biopsy, etc.).

[0003] However, methods involving direct visual inspection or palpation have the drawback of being limited to specific areas (affected areas) such as the breasts, rectum, stomach, and large intestine. Furthermore, while imaging diagnostics are simple, they have the disadvantages of low detection sensitivity and radiation exposure to the patient. In this respect, blood tests are preferable because they are less burdensome for the patient and have high detection sensitivity. In particular, if a diagnosis can be made by analyzing blood collected from the patient, it is even more preferable because it reduces the burden on the patient and can be implemented in mass screenings.

[0004] Accordingly, the applicant of the present application has developed a "cancer risk evaluation method" as a pre-evaluation method for cancer risk, in which concentration data of a group of evaluation elements in blood (serum) collected from a subject and age data of the subject are applied to a discriminant function for discriminating whether the subject belongs to a control group or a case group for calculation, and an indicator of whether the subject suffers from any cancer is obtained based on the correlation of the evaluation element group obtained thereby, and filed a patent application for the method. In this method, a combination of 17 elements including Na, Mg, P, S, K, Ca, Fe, Cu, Zn, Se, Rb, Sr, As, Mo, Cs, Co, and Ag is used as the evaluation element group. This method has advantages such as: an indicator for cancer risk evaluation can be easily obtained through a blood test; cancer risk can be easily evaluated using the indicator; cancer risk of general examinees can be estimated with high accuracy; and it can be easily applied to mass screening. This method has already been registered as Japanese Patent No. 6946133 (see Patent Document 1).

[0005] On the other hand, based on the experience of developing the "cancer risk evaluation method" using human blood described in Patent Document 1 and the findings obtained therefrom, the applicant of the present application came up with the idea that human urine could be used for pre-evaluation of cancer risk, and examined the feasibility of the idea. This is because if the idea is realized, it would be more convenient because applicants for evaluation do not need to visit facilities such as medical institutions or blood centers to have blood collected, and further cost reduction can be achieved.

[0006] Based on the above examination results regarding the feasibility, the applicant of the present application has continued intensive research on the possibility of developing a new cancer screening method using human urine, and has consequently completed the present invention.

[0007] As research studies for clarifying the relationship between the concentration of element groups present in human urine and cancer risk, there are reports disclosed in Non-Patent Documents 1 and 2.

[0008] Specifically, Non-Patent Document 1 reports the results of an investigation into the correlation between heavy metals detected in the urine of breast cancer patients and the urinary metabolome (all metabolites contained in the body), and their relationship to the development of breast cancer. The report states that urinary metabolites were determined using nuclear magnetic resonance spectroscopy, and heavy metals were detected in urine samples using an inductively coupled plasma mass spectrometer. It also states that Cd in the urine of breast cancer patients increased by approximately twofold compared to the control group, and that Cr and As also increased in the urine of the same patients. Furthermore, it states that numerous small molecule metabolites changed in the urine of the same patients compared to the control group. Based on these findings, it is suggested that environmental exposure to Cd, As, or Cr may affect the urinary concentrations of metabolites that may be involved in the development of breast cancer.

[0009] Furthermore, Non-Patent Literature 2 reports the results of measuring the concentrations of Fe, Cu, Zn, Ni, Sr, and Cs in the urine of subjects (randomly selected from patients with precancerous lesions, patients with gastric cancer, and a normal control group) using inductively coupled plasma mass spectrometry (ICP-MS) to infer the association between human urinary exposure to a group of elements and precancerous lesions and gastric cancer. It states that weighted quantile regression showed a positive correlation between urinary exposure to the element mixture and both precancerous lesions and gastric cancer, and that the OR (odds ratio) for exposure to the mixture was 1.34 for precancerous lesions and 1.38 for gastric cancer. It also states that the Qgcomp model and the BKMR model showed a statistically significant positive correlation between the mixture and both precancerous lesions and gastric cancer.

[0010] Patent No. 6946133

[0011] Yuhao Men et al., Evaluation of heavy metals and metabolites in the urine of patients with breast cancer, ONCOLOGY LETTERS, 19: 1331-1337, 2020, DOI: 10.3892 / ol.2019.11206Shiqing Qian et al., Association analyzes between urinary concentrations of multiple trace elements and gastric precancerous lesions and gastric cancer in Anhui province, eastern China, 1-12, August 15, 2024, DOI 10.3389 / fpubh.2024.1423286

[0012] Non-patent document 1, mentioned above, investigated the correlation between heavy metals present in the urine of breast cancer patients and urinary metabolome, and the relationship between these and the development of breast cancer. It showed that environments that induce exposure to Cd, As, or Cr affect the urinary concentrations of metabolites that may be involved in the development of breast cancer. However, the report stops there and does not teach or suggest that the correlation between the concentrations of element groups present in the urine of breast cancer patients can be used to assess the risk of developing breast cancer.

[0013] Non-patent document 2, mentioned above, shows that several trace elements present in human urine, specifically Fe, Cu, Zn, Ni, Sr, and Cs, are positively correlated with both precancerous lesions and gastric cancer. However, non-patent document 2 goes no further than that, and does not teach or suggest that the correlation between the concentrations of these elements present in the urine of precancerous lesions and gastric cancer patients can be used to assess the risk of developing gastric cancer.

[0014] In view of the above-mentioned conventional circumstances, the inventors diligently conducted research based on the knowledge gained during the research and development of the "cancer risk assessment method" disclosed in Patent Document 1, the knowledge gained from subsequent further research on the same method, and the knowledge gained from reports described in Non-Patent Documents 1 and 2. As a result, they discovered the possibility of a method for evaluating the risk of developing cancer in humans using an index obtained by utilizing the correlation between the concentrations of elements present in human urine, in other words, a new cancer screening method using human urine, and thus arrived at the present invention.

[0015] Therefore, the object of the present invention is to provide a cancer risk assessment method and a cancer risk assessment system that can objectively and accurately estimate the cancer risk of a subject using urine, a biological sample that can be collected far more easily than blood.

[0016] Another object of the present invention is to provide a cancer risk assessment method and a cancer risk assessment system that can be used for auxiliary diagnosis of cancer in patients visiting medical institutions.

[0017] Another object of the present invention is to provide a cancer risk assessment method and a cancer risk assessment system that can predict not only the cancer risk of a subject but also the location (type) of cancer.

[0018] Another object of the present invention is to provide a cancer risk assessment method and a cancer risk assessment system that can be easily applied to mass screenings.

[0019] Other objects of the present invention not explicitly stated herein will become apparent from the following description and accompanying drawings.

[0020] (1) The cancer risk assessment method of the present invention comprises the steps of: obtaining concentration data of a group of evaluation elements contained in a urine sample taken from a subject; applying the concentration data of the group of evaluation elements to a discriminant function for determining whether the subject belongs to a control group or a case group, and calculating the correlation between the concentrations of the group of evaluation elements; generating an index for identifying whether the subject has cancer based on the correlation; and evaluating the subject's risk of developing cancer based on the index and creating an assessment result, wherein the group of evaluation elements is a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl.

[0021] In the cancer risk assessment method of the present invention, the concentration data of the evaluation element group contained in the urine sample collected from the subject is applied to the discriminant function to calculate the correlation between the concentrations of the evaluation element group. Here, the evaluation element group is a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl.

[0022] Then, based on the correlation, an index is generated to identify whether or not the subject has any kind of cancer, and the cancer risk of the subject is evaluated based on the index to generate the evaluation result. As the index, for example, a discriminant score calculated by applying the concentration data to the discriminant function (discriminant expression) generated based on the correlation is used.

[0023] Thus, the cancer risk assessment method of the present invention allows for the objective and highly accurate assessment of the subject's risk of developing cancer by directly understanding the subject's physical condition through the measurement and analysis of the subject's urine sample, which is a biological sample of the subject.

[0024] Furthermore, since the evaluation results are obtained through the measurement and analysis of the urine sample, which is a biological sample of the subject, an objective evaluation is provided. For this reason, medical institutions can refer to the evaluation results and use them as an auxiliary diagnostic tool for cancer in patients who visit their medical institutions.

[0025] Furthermore, by using the concentration data of the evaluation elements contained in the urine sample collected from the subject, and performing automatic calculations on a computer, it is possible to determine whether the subject belongs to the control group or the case group, in other words, whether or not the subject has any kind of cancer. Therefore, even if there are many subjects, it is possible to make this determination easily and quickly. Thus, the cancer risk assessment method of the present invention can be easily applied to mass screenings.

[0026] Furthermore, based on the aforementioned correlation, it is possible to determine which element selected from the group of evaluation elements consisting of the 20 elements is significant for discrimination. Moreover, since the element significant for discrimination changes depending on the location (type) of the cancer, it is also possible to infer which location (type) of cancer the subject has if it is suspected that they have some kind of cancer.

[0027] (2) In a preferred example of the cancer risk assessment method of the present invention, a discrimination score calculated by applying the concentration data to a discriminant formula generated based on the correlation is used as the index, the discrimination score is compared with a predetermined reference value to determine whether the subject belongs to the control group or the case group, and the assessment result is made to include an inference as to whether the subject has cancer or not based on the result of the discrimination.

[0028] (3) In another preferred example of the cancer risk assessment method of the present invention, the probability of a subject developing cancer is estimated by comparing the subject's discrimination score with the relationship between the discrimination score and the probability of developing cancer, and the estimated probability of a subject developing cancer is included in the assessment result.

[0029] (4) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether the male subject has any cancer without specifying the site of cancer, and the assessment result includes a determination of whether the male subject has any cancer.

[0030] (5) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether the female subject has any cancer without specifying the site of cancer, and the assessment result includes a determination of whether the female subject has any cancer.

[0031] (6) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has gastric cancer without specifying the sex, and the assessment result includes the determination result of whether or not the subject has gastric cancer.

[0032] (7) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has colorectal cancer without specifying the sex, and the assessment result includes the determination result of whether or not the subject has colorectal cancer.

[0033] (8) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has pancreatic cancer without specifying the sex, and the assessment result includes the determination result of whether or not the subject has pancreatic cancer.

[0034] (9) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has esophageal cancer without specifying the sex, and the assessment result includes the determination result of whether or not the subject has esophageal cancer.

[0035] (10) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has malignant lymphoma without specifying the sex, and the assessment result includes a determination of whether or not the subject has malignant lymphoma.

[0036] (11) In yet another preferred example of the cancer risk assessment method of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has lung cancer without specifying the sex, and the assessment result includes a determination of whether or not the subject has lung cancer.

[0037] (12) In yet another preferred example of the cancer risk assessment method of the present invention, one or more discriminant formulas are used as discriminant formulas, along with a first discriminant formula generated to assess whether the male subject has any type of cancer without specifying the site of cancer, a second discriminant formula generated to assess whether the subject has stomach cancer, a third discriminant formula generated to assess whether the subject has colorectal cancer, a fourth discriminant formula generated to assess whether the subject has pancreatic cancer, a fifth discriminant formula generated to assess whether the subject has esophageal cancer, a sixth discriminant formula generated to assess whether the subject has malignant lymphoma, and a seventh discriminant formula generated to assess whether the subject has lung cancer. If the first discriminant formula determines that the male subject has some kind of cancer, the evaluation result will include the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on one or more discriminant formulas selected from the group.

[0038] (13) In yet another preferred example of the cancer risk assessment method of the present invention, one or more discriminant formulas selected from the group consisting of a first discriminant formula generated to assess whether the subject has any type of cancer without specifying the site of cancer, a second discriminant formula generated to assess whether the subject has stomach cancer, a third discriminant formula generated to assess whether the subject has colorectal cancer, a fourth discriminant formula generated to assess whether the subject has pancreatic cancer, a fifth discriminant formula generated to assess whether the subject has esophageal cancer, a sixth discriminant formula generated to assess whether the subject has malignant lymphoma, and a seventh discriminant formula generated to assess whether the subject has lung cancer, are used as the discriminant formulas. If the first discriminant formula determines that the female subject has some kind of cancer, the evaluation result will include the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on one or more discriminant formulas selected from the group.

[0039] (14) In yet another preferred example of the cancer risk assessment method of the present invention, if, based on the correlation, four elements P, K, Ca, and Cu selected from the 20 elements used as the group of elements for assessment are significant in the determination, and the subject is male, then the assessment result includes the inference that the male subject has some kind of cancer.

[0040] (15) In yet another preferred example of the cancer risk assessment method of the present invention, based on the correlation, (a) without correction for urinary creatinine in the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Co, Cu, Mo, Cd, Cs, and Tl—are significant in the discrimination, or (b) with correction for urinary creatinine in the urine sample, 7 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Mo, Cd, and Cs—are significant in the discrimination, and if the subject is female, the evaluation result includes the inference that the female subject has some kind of cancer.

[0041] (16) In yet another preferred example of the cancer risk assessment method of the present invention, based on the correlation, if (a) without urinary creatinine correction for the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Cu, Zn, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, or (b) with urinary creatinine correction for the urine sample, 8 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, then the assessment result includes the inference that the subject has gastric cancer.

[0042] (17) In yet another preferred example of the cancer risk assessment method of the present invention, if seven elements Na, P, K, Cu, Mo, Ba, and Tl selected from the 20 elements used as the group of elements for assessment are significant in the determination based on the correlation, the assessment result includes the inference that the subject has colorectal cancer.

[0043] (18) In yet another preferred example of the cancer risk assessment method of the present invention, if six elements Na, P, K, Ca, Cu, and Mo selected from the 20 elements used as the group of elements for assessment are significant in the determination based on the correlation, then the assessment result includes the inference that the subject has pancreatic cancer.

[0044] (19) In still another preferred example of the cancer risk evaluation method of the present invention, based on said correlation, when (a) 7 elements selected from the 20 elements used as said element group for evaluation, which are Li, B, P, Cu, As, Mo, and Cd, are significant for discrimination without urinary creatinine correction for said urine specimen, or (b) 6 elements selected from the 20 elements used as said element group for evaluation, which are Li, B, P, Cu, Mo, and Cd, are significant for discrimination with urinary creatinine correction for said urine specimen, a presumption that said subject suffers from esophageal cancer is included in said evaluation result.

[0045] (20) In still another preferred example of the cancer risk evaluation method of the present invention, based on said correlation, when 4 elements selected from the 20 elements used as said element group for evaluation, which are B, P, Cu, and Mo, are significant for discrimination, a presumption that said subject suffers from malignant lymphoma is included in said evaluation result.

[0046] (21) In still another preferred example of the cancer risk evaluation method of the present invention, based on said correlation, when 8 elements selected from the 20 elements used as said element group for evaluation, which are Na, Mg, P, K, Cu, Rb, Ba, and Tl, are significant for discrimination, a presumption that said subject suffers from lung cancer is included in said evaluation result.

[0047] (22) The cancer risk evaluation system of the present invention comprises: a data storage unit that stores concentration data of an element group for evaluation contained in a urine specimen collected from a subject; a calculation unit that applies the concentration data of said subject stored in said data storage unit to a discriminant function for discriminating whether said subject belongs to a control group or a case group, and calculates a correlation between the concentrations of said element group for evaluation; and an evaluation result generation unit that generates an index for identifying whether said subject suffers from cancer based on said correlation calculated by said calculation unit, and evaluates the cancer risk of said subject based on said index to generate an evaluation result, wherein said element group for evaluation is a combination of 20 elements consisting of Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl.

[0048] In the cancer risk assessment system of the present invention, in the arithmetic unit, the concentration data of the evaluation element of the subject stored in the data storage unit is applied to the discriminant function to calculate the correlation between the concentrations of the evaluation element group contained in the urine sample. Here, as the evaluation elements, a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl is used.

[0049] Then, in the evaluation result generating unit, based on the correlation calculated by the arithmetic unit, after generating the index for identifying whether the subject suffers from any type of cancer, the cancer risk of the subject is evaluated based on the index to generate an evaluation result. As the index, for example, a discriminant score calculated by applying the concentration data to a discriminant generated based on the correlation is used.

[0050] As described above, in the cancer risk assessment method of the present invention, by directly grasping the physical condition of the subject through measurement and analysis of the urine sample, which is a biological sample of the subject, the cancer risk of the subject can be evaluated objectively and with high accuracy.

[0051] In addition, since the evaluation result is obtained through measurement and analysis of the urine sample, which is a biological sample of the subject, the evaluation result includes objective assessment. For this reason, when a medical institution refers to the evaluation result, it can be effectively used for auxiliary diagnosis of cancer for examinees who visit the medical institution.

[0052] Furthermore, by using the concentration data of the evaluation element group in the urine sample collected from the subject and performing automatic calculation by the arithmetic unit, it is possible to determine which of the control group and the case group the subject belongs to, in other words, whether the subject suffers from any type of cancer. Therefore, even if there are a large number of subjects, the determination can be performed easily and quickly. Accordingly, the cancer risk assessment system of the present invention can be easily applied to mass screening.

[0053] Furthermore, based on the correlation obtained by the calculation unit, it is determined which element selected from the evaluation element group consisting of 20 combinations is significant for discrimination. Moreover, since the element significant for discrimination changes depending on the location (type) of the cancer, it is also possible to infer which location (type) of cancer the subject has if it is suspected that the subject has some kind of cancer.

[0054] (23) In a preferred example of the cancer risk assessment system of the present invention, a discrimination score calculated by applying the concentration data to a discriminant formula generated based on the correlation is used as the index, the discrimination score is compared with a predetermined reference value to determine whether the subject belongs to the control group or the case group, and the assessment result includes an inference of whether or not the subject has cancer based on the result of the discrimination.

[0055] (24) In another preferred example of the cancer risk assessment system of the present invention, the probability of a subject developing cancer is estimated by comparing the subject's discrimination score with the relationship between the discrimination score and the probability of developing cancer, and the estimated probability of a subject developing cancer is included in the assessment result.

[0056] (25) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant is a discriminant generated to assess whether the male subject has any cancer without specifying the site of cancer, and the assessment result includes a determination of whether the male subject has any cancer.

[0057] (26) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether the female subject has any cancer without specifying the site of cancer, and the assessment result includes a determination of whether the female subject has any cancer.

[0058] (27) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has gastric cancer without specifying the sex, and the assessment result includes a determination of whether or not the subject has gastric cancer.

[0059] (28) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has colorectal cancer without specifying the sex, and the assessment result includes a determination of whether or not the subject has colorectal cancer.

[0060] (29) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has pancreatic cancer without specifying the sex, and the assessment result includes a determination of whether or not the subject has pancreatic cancer.

[0061] (30) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has esophageal cancer without specifying the sex, and the assessment result includes a determination of whether or not the subject has esophageal cancer.

[0062] (31) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has malignant lymphoma without specifying the sex, and the assessment result includes a determination of whether or not the subject has malignant lymphoma.

[0063] (32) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is a discriminant formula generated to assess whether or not the subject has lung cancer without specifying the gender, and the assessment result includes a determination of whether or not the subject has lung cancer.

[0064] (33) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is at least one selected from the group consisting of a first discriminant formula generated to assess whether the male subject has any type of cancer without specifying the site of cancer, a second discriminant formula generated to assess whether the subject has stomach cancer, a third discriminant formula generated to assess whether the subject has colorectal cancer, a fourth discriminant formula generated to assess whether the subject has pancreatic cancer, a fifth discriminant formula generated to assess whether the subject has esophageal cancer, a sixth discriminant formula generated to assess whether the subject has malignant lymphoma, and a seventh discriminant formula generated to assess whether the subject has lung cancer. If the first discriminant formula determines that the male subject has some kind of cancer, the evaluation result will include the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on one or more discriminant formulas selected from the group.

[0065] (34) In yet another preferred example of the cancer risk assessment system of the present invention, the discriminant formula used is at least one selected from the group consisting of a first discriminant formula generated to assess whether the subject has any type of cancer, without specifying the site of cancer, a second discriminant formula generated to assess whether the subject has gastric cancer, a third discriminant formula generated to assess whether the subject has colorectal cancer, a fourth discriminant formula generated to assess whether the subject has pancreatic cancer, a fifth discriminant formula generated to assess whether the subject has esophageal cancer, a sixth discriminant formula generated to assess whether the subject has malignant lymphoma, and a seventh discriminant formula generated to assess whether the subject has lung cancer. If the first discriminant formula determines that the female subject has some kind of cancer, the evaluation result will include the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on one or more discriminant formulas selected from the group.

[0066] (35) In yet another preferred example of the cancer risk assessment system of the present invention, if four elements P, K, Ca, and Cu selected from the 20 elements used as the group of evaluation elements are significant for discrimination based on the correlation, and the subject is male, the assessment result includes the inference that the subject has some kind of cancer.

[0067] (36) In yet another preferred example of the cancer risk assessment system of the present invention, based on the correlation, (a) without urinary creatinine correction for the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Co, Cu, Mo, Cd, Cs, Tl—are significant in the discrimination, or (b) with urinary creatinine correction for the urine sample, 7 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Mo, Cd, Cs—are significant in the discrimination, and if the subject is female, the evaluation result includes the inference that the subject has some kind of cancer.

[0068] (37) In yet another preferred example of the cancer risk assessment system of the present invention, based on the correlation, if (a) without urinary creatinine correction for the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Cu, Zn, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, or (b) with urinary creatinine correction for the urine sample, 8 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, then the assessment result includes the inference that the subject has gastric cancer.

[0069] (38) In yet another preferred example of the cancer risk assessment system of the present invention, if seven elements selected from the 20 elements used as the group of evaluation elements, namely Na, P, K, Cu, Mo, Ba, and Tl, are significant in the determination based on the correlation, then the assessment result includes the inference that the subject has colorectal cancer.

[0070] (39) In yet another preferred example of the cancer risk assessment system of the present invention, if six elements Na, P, K, Ca, Cu, and Mo selected from the 20 elements used as the group of evaluation elements are significant in the determination based on the correlation, the assessment result includes the inference that the subject has pancreatic cancer.

[0071] (40) In yet another preferred example of the cancer risk assessment system of the present invention, based on the correlation, if (a) seven elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, As, Mo, and Cd—are significant in the discrimination without correction for urinary creatinine in the urine sample, or (b) six elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, Mo, and Cd—are significant in the discrimination with correction for urinary creatinine in the urine sample, then the assessment result includes the inference that the subject has esophageal cancer.

[0072] (41) In yet another preferred example of the cancer risk assessment system of the present invention, if four elements B, P, Cu, and Mo selected from the 20 elements used as the group of evaluation elements based on the correlation are significant in the determination, then the assessment result includes the inference that the subject has malignant lymphoma.

[0073] (42) In yet another preferred example of the cancer risk assessment system of the present invention, if eight elements selected from the 20 elements used as the group of evaluation elements, namely Na, Mg, P, K, Cu, Rb, Ba, and Tl, are significant in the determination based on the correlation, then the assessment result includes the inference that the subject has lung cancer.

[0074] (43) As described above, the cancer risk assessment method and cancer risk assessment system of the present invention can assess the cancer risk of the subject in eight patterns using the concentration data of the 20 elements of the evaluation group contained in the urine sample of the subject: - "Whether or not the male subject has any cancer" - "Whether or not the female subject has any cancer" - "Whether or not the subject has gastric cancer regardless of gender" - "Whether or not the subject has colorectal cancer regardless of gender" - "Whether or not the subject has pancreatic cancer regardless of gender" - "Whether or not the subject has esophageal cancer regardless of gender" - "Whether or not the subject has malignant lymphoma regardless of gender" - "Whether or not the subject has lung cancer regardless of gender".

[0075] Furthermore, for each of the eight patterns mentioned above, there are cases where urine creatinine correction is applied to the concentration data and cases where it is not. Therefore, the cancer risk assessment of the subject can be performed in a total of 16 patterns.

[0076] Furthermore, since the 16 patterns can be appropriately combined and implemented according to the needs, the cancer risk assessment method and cancer risk assessment system of the present invention can perform cancer risk assessment in a very large number of patterns as needed, and therefore it is clear that they are extremely useful.

[0077] The cancer risk assessment method and cancer risk assessment system of the present invention have the following advantages: (a) it is possible to objectively and accurately estimate the risk of cancer in a subject using urine, a biological sample that can be collected much more easily than blood; (b) it can be used as an auxiliary diagnostic tool for cancer in patients who visit medical institutions; (c) it is possible to estimate not only the risk of cancer in a subject but also the location of the cancer; and (d) it can be easily applied to mass screenings.

[0078] This is a flowchart of the cancer risk assessment method of the present invention. This is a functional block diagram showing the configuration of the cancer risk assessment system of the present invention. This is a flowchart of one embodiment of the cancer risk assessment method of the present invention. This is a table showing the breakdown of all subjects (healthy individuals and cancer patients) and disease names (site of cancer) that provided urine samples used in Examples 1 to 16 of the cancer risk assessment method of the present invention. This is a table showing the age distribution of each group (healthy individuals group and site-specific cancer patient group) of all subjects that provided urine samples used in Examples 1 to 16 of the cancer risk assessment method of the present invention. This is a table showing the mean, standard deviation, minimum, and maximum age of each group (healthy individuals group and site-specific cancer patient group) of all subjects that provided urine samples used in Examples 1 to 16 of the cancer risk assessment method of the present invention. This is a table (without urine creatinine correction) showing the mean and standard deviation of the concentration data (natural logarithm transformed) of 20 elements (assessment element group) contained in the urine sample for all subjects that provided urine samples used in Examples 1 and 3 of the cancer risk assessment method of the present invention, by gender and group (healthy individuals group and all cancer patient group). This table shows the mean and standard deviation of the concentration data (natural logarithm transformed) of 20 elements (evaluation element group) contained in the urine samples (natural logarithm transformed) for all subjects who provided urine samples used in Examples 2 and 4 of the cancer risk assessment method of the present invention, broken down by gender and group (healthy control group and all cancer patient group) (with urinary creatinine correction). This table shows the mean and standard deviation of the concentration data (natural logarithm transformed) of 20 elements (evaluation element group) contained in the urine samples (natural logarithm transformed) for all subjects who provided urine samples used in Examples 5, 7, 9, 11, 13, and 15 of the cancer risk assessment method of the present invention, broken down by group (healthy control group and each cancer patient group) (without urinary creatinine correction). This table shows the mean and standard deviation of the concentration data (natural logarithm transformed) of 20 elements (evaluation element group) contained in the urine samples for all subjects who provided urine samples used in Examples 6, 8, 10, 12, 14, and 16 of the cancer risk assessment method of the present invention, separated by group (healthy control group and each cancer patient group) (with urinary creatinine correction).The diagram shows the discriminant formulas obtained in Examples 1, 3, 5, and 7 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. (a) shows the discriminant formula obtained from healthy men and all male cancer patients (Example 1), (b) shows the discriminant formula obtained from healthy women and all female cancer patients (Example 3), (c) shows the discriminant formula obtained from healthy men and gastric cancer patients (Example 5), and (d) shows the discriminant formula obtained from healthy men and colorectal cancer patients (Example 7). The diagram also shows the discriminant formulas obtained in Examples 9, 11, 13, and 15 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. (a) shows the discriminant formula obtained from healthy men and pancreatic cancer patients (Example 9), (b) shows the discriminant formula obtained from healthy men and esophageal cancer patients (Example 11), (c) shows the discriminant formula obtained from healthy men and malignant lymphoma patients (Example 13), and (d) shows the discriminant formula obtained from healthy men and lung cancer patients (Example 15). The diagram shows the discriminant formulas obtained in Examples 2, 4, 6, and 8 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. (a) shows the discriminant formula obtained from healthy men and all male cancer patients (Example 2), (b) shows the discriminant formula obtained from healthy women and all female cancer patients (Example 4), (c) shows the discriminant formula obtained from healthy men and gastric cancer patients (Example 6), and (d) shows the discriminant formula obtained from healthy men and colorectal cancer patients (Example 8). The diagram also shows the discriminant formulas obtained in Examples 10, 12, 14, and 16 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. (a) shows the discriminant formula obtained from healthy men and pancreatic cancer patients (Example 10), (b) shows the discriminant formula obtained from healthy men and esophageal cancer patients (Example 12), (c) shows the discriminant formula obtained from healthy men and malignant lymphoma patients (Example 14), and (d) shows the discriminant formula obtained from healthy men and lung cancer patients (Example 16). This table shows the discriminant coefficients obtained from the discriminant analysis between the healthy male group and the male all-cancer patient group, and the elements that were significant in the discriminant analysis, obtained in Example 1 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from the discriminant analysis between the healthy male group and the male all-cancer patient group, and the elements that were significant in the discriminant analysis, obtained in Example 2 (with urinary creatinine correction) of the cancer risk assessment method of the present invention.This table shows the discriminant coefficients obtained from discriminant analysis between a group of healthy women and a group of women with all types of cancer, and the elements that are significant in the discriminant analysis, obtained in Example 3 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a group of healthy women and a group of women with all types of cancer, and the elements that are significant in the discriminant analysis, obtained in Example 4 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a group of healthy women and a group of women with all types of cancer, and the elements that are significant in the discriminant analysis, obtained in Example 5 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a group of healthy women and a group of patients with gastric cancer, and the elements that are significant in the discriminant analysis, obtained in Example 6 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a group of healthy women and a group of patients with gastric cancer, and the elements that are significant in the discriminant analysis, obtained in Example 7 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a healthy control group and a colorectal cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 8 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 9 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 10 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 11 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 12 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from the discriminant analysis between the healthy control group and the malignant lymphoma patient group, and the elements significant in the discriminant analysis, obtained in Example 13 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This table shows the discriminant coefficients obtained from the discriminant analysis between the healthy control group and the malignant lymphoma patient group, and the elements significant in the discriminant analysis, obtained in Example 14 (with urinary creatinine correction) of the cancer risk assessment method of the present invention.This is a table showing the discriminant coefficients obtained from the discriminant analysis between the healthy control group and the lung cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 15 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. This is a table showing the discriminant coefficients obtained from the discriminant analysis between the healthy control group and the lung cancer patient group, and the elements significant in the discriminant analysis, obtained in Example 16 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This is a figure showing the results obtained from the discriminant analysis between the healthy male control group and the male all-cancer patient group in Example 1 (without urinary creatinine correction) of the cancer risk assessment method of the present invention, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. This is a graph showing the relationship between the discriminant score obtained from the discriminant analysis between the healthy male control group and the male all-cancer patient group and the probability of cancer in Example 1 (without urinary creatinine correction) of the cancer risk assessment method of the present invention. The first example of the cancer risk assessment method of the present invention (without urinary creatinine correction) shows the results obtained from discriminant analysis between a group of healthy men and a group of all cancer patients in men, where (a) is a graph showing the ROC (Receiver Operating Characteristic) curve of the group of all cancer patients in men, and (b) is a table showing the AUC (Area Under the Curve), standard deviation, lower limit, and upper limit of the group of all cancer patients in men. The second example of the cancer risk assessment method of the present invention (with urinary creatinine correction) shows the results obtained from discriminant analysis between a group of healthy men and a group of all cancer patients in men, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. The third example of the cancer risk assessment method of the present invention (with urinary creatinine correction) shows a graph showing the relationship between the discriminant score obtained from discriminant analysis between a group of healthy men and a group of all cancer patients in men and the probability of cancer. In Example 2 of the cancer risk assessment method of the present invention (with urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a group of healthy men and a group of men with all types of cancer, where (a) is a graph showing the ROC curve of the group of men with all types of cancer, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the group of men with all types of cancer.In Example 3 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a group of healthy women and a group of women with all cancers, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. In Example 3 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the graph shows the relationship between the discriminant scores obtained from discriminant analysis between a group of healthy women and a group of women with all cancers and the probability of cancer. In Example 3 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a group of healthy women and a group of women with all cancers, where (a) is a graph showing the ROC curve of the group of women with all cancers, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the group of women with all cancers. In Example 4 of the cancer risk assessment method of the present invention (with urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a group of healthy women and a group of women with all cancers, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. In Example 4 of the cancer risk assessment method of the present invention (with urinary creatinine correction), the graph shows the relationship between the discriminant scores obtained from discriminant analysis between a group of healthy women and a group of women with all cancers and the probability of cancer. In Example 4 of the cancer risk assessment method of the present invention (with urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a group of healthy women and a group of women with all cancers, where (a) is a graph showing the ROC curve of the group of women with all cancers, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the group of women with all cancers. The first figure shows the results obtained from the discriminant analysis between a healthy control group and a gastric cancer patient group in Example 5 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and predictive value of the two groups. The second figure shows the relationship between the discriminant scores obtained from the discriminant analysis between a healthy control group and a gastric cancer patient group in Example 5 of the cancer risk assessment method of the present invention (without urinary creatinine correction) and the probability of cancer.The first figure shows the results obtained from the discriminant analysis between a healthy control group and a gastric cancer patient group in Example 5 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a graph showing the ROC curve of the gastric cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the gastric cancer patient group. The second figure shows the results obtained from the discriminant analysis between a healthy control group and a gastric cancer patient group in Example 6 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. The third figure shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and a gastric cancer patient group and the probability of cancer in Example 6 of the cancer risk assessment method of the present invention (with urinary creatinine correction). The first figure shows the results obtained from discriminant analysis between a healthy control group and a gastric cancer patient group in Example 6 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a graph showing the ROC curve for the gastric cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit for the gastric cancer patient group. The second figure shows the results obtained from discriminant analysis between a healthy control group and a colorectal cancer patient group in Example 7 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores for the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy for the two groups. The third figure shows the relationship between the discriminant score obtained from discriminant analysis between a healthy control group and a colorectal cancer patient group and the probability of cancer in Example 7 of the cancer risk assessment method of the present invention (without urinary creatinine correction). The first figure shows the results obtained from the discriminant analysis between a healthy control group and a colorectal cancer patient group in Example 7 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a graph showing the ROC curve for the colorectal cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit for the colorectal cancer patient group. The second figure shows the results obtained from the discriminant analysis between a healthy control group and a colorectal cancer patient group in Example 8 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores for the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy for the two groups. The third figure shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and a colorectal cancer patient group and the probability of cancer in Example 8 of the cancer risk assessment method of the present invention (with urinary creatinine correction).The first example of the cancer risk assessment method of the present invention (with urinary creatinine correction) shows the results obtained from discriminant analysis between a healthy control group and a colorectal cancer patient group, where (a) is a graph showing the ROC curve for the gastric cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit for the gastric cancer patient group. The second example of the cancer risk assessment method of the present invention (without urinary creatinine correction) shows the results obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, where (a) is a histogram showing the frequency distribution of the discriminant scores for the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy for the two groups. The third example of the cancer risk assessment method of the present invention (without urinary creatinine correction) shows a graph showing the relationship between the discriminant score obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group and the probability of cancer. The first figure shows the results obtained from the discriminant analysis between a healthy control group and a pancreatic cancer patient group in Example 9 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a graph showing the ROC curve of the pancreatic cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the pancreatic cancer patient group. The second figure shows the results obtained from the discriminant analysis between a healthy control group and a pancreatic cancer patient group in Example 10 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. The third figure shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and a pancreatic cancer patient group and the probability of cancer in Example 10 of the cancer risk assessment method of the present invention (without urinary creatinine correction). In Example 10 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a healthy control group and a pancreatic cancer patient group, where (a) is a graph showing the ROC curve of the pancreatic cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the pancreatic cancer patient group. In Example 11 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a healthy control group and an esophageal cancer patient group, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and hit rate of the two groups.This graph shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and an esophageal cancer patient group and the probability of cancer in Example 11 of the cancer risk assessment method of the present invention (without urinary creatinine correction). This figure shows the results obtained from the discriminant analysis between a healthy control group and an esophageal cancer patient group in Example 11 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a graph showing the ROC curve of the esophageal cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the female all cancer patient group. This figure shows the results obtained from the discriminant analysis between a healthy control group and an esophageal cancer patient group in Example 12 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and hit rate of the two groups. This graph shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and an esophageal cancer patient group and the probability of cancer in Example 12 of the cancer risk assessment method of the present invention (without urinary creatinine correction). This figure shows the results obtained from the discriminant analysis between a healthy control group and an esophageal cancer patient group in Example 12 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a graph showing the ROC curve of the esophageal cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the esophageal cancer patient group. This figure shows the results obtained from the discriminant analysis between a healthy control group and an esophageal lymphoma patient group in Example 13 of the cancer risk assessment method of the present invention (without urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and hit rate of the two groups. This graph shows the relationship between the discriminant score obtained from the discriminant analysis between a healthy control group and a malignant lymphoma patient group and the probability of cancer in Example 13 (with urinary creatinine correction) of the cancer risk assessment method of the present invention. This figure shows the results obtained from the discriminant analysis between a healthy control group and a malignant lymphoma patient group in Example 13 (with urinary creatinine correction) of the cancer risk assessment method of the present invention, where (a) is a graph showing the ROC curve of the malignant lymphoma patient group, and (b) is a table showing the AUC, standard deviation, lower limit and upper limit of the malignant lymphoma patient group.The first figure shows the results obtained from discriminant analysis between a healthy control group and a malignant lymphoma patient group in Example 14 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores for the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy for the two groups. The second figure shows the relationship between the discriminant scores obtained from discriminant analysis between a healthy control group and a malignant lymphoma patient group in Example 14 of the cancer risk assessment method of the present invention (with urinary creatinine correction), and the probability of cancer. The third figure shows the results obtained from discriminant analysis between a healthy control group and a malignant lymphoma patient group in Example 14 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a graph showing the ROC curve for the malignant lymphoma patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit for the malignant lymphoma patient group. In Example 15 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a healthy control group and a lung cancer patient group, where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and accuracy of the two groups. In Example 15 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the graph shows the relationship between the discriminant scores obtained from discriminant analysis between a healthy control group and a lung cancer patient group and the probability of cancer. In Example 15 of the cancer risk assessment method of the present invention (without urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a healthy control group and a lung cancer patient group, where (a) is a graph showing the ROC curve of the lung cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the lung cancer patient group. The first figure shows the results obtained from the discriminant analysis between a healthy control group and a lung cancer patient group in Example 16 of the cancer risk assessment method of the present invention (with urinary creatinine correction), where (a) is a histogram showing the frequency distribution of the discriminant scores of the two groups, and (b) is a table showing the sensitivity, specificity, and predictive value of the two groups. The second figure shows the relationship between the discriminant scores obtained from the discriminant analysis between a healthy control group and a lung cancer patient group in Example 16 of the cancer risk assessment method of the present invention (with urinary creatinine correction) and the probability of cancer.In Example 16 of the present invention's cancer risk assessment method (with urinary creatinine correction), the figure shows the results obtained from discriminant analysis between a healthy control group and a lung cancer patient group, where (a) is a graph showing the ROC curve of the lung cancer patient group, and (b) is a table showing the AUC, standard deviation, lower limit, and upper limit of the lung cancer patient group.

[0079] The details of the present invention and preferred embodiments will be described below with reference to the accompanying drawings.

[0080] (Outline of the cancer risk assessment method of the present invention) The outline of the cancer risk assessment method of the present invention is as follows:

[0081] First, urine samples belonging to the case group (cancer patients) and urine samples belonging to the control group (healthy individuals) are obtained, and the concentrations of specific elemental groups contained in each of these urine samples are measured. Then, statistical analysis is performed on the obtained concentration data of the elemental groups to calculate the correlation between the concentration data. Next, a discriminant formula is obtained based on the correlation calculated in this way. Then, the concentration data of each of the urine samples is applied to the discriminant formula obtained in this way to obtain an "indicator" that identifies whether or not the person (subject) who provided each of the urine samples belonging to the case group or the control group has cancer. Based on the "indicator" obtained in this way, an evaluation result related to cancer risk is obtained. Preferably, the evaluation result includes an estimate of the site (type) of cancer and an estimate of the probability (incidence rate) of having cancer.

[0082] Specifically, the inventors first performed preliminary processing as described below, then determined the optimal measurement conditions for measuring the concentrations of elements contained in the urine sample, and selected a group of elements to be used for cancer risk assessment.

[0083] (1. Determination of Optimal Measurement Conditions) First, as the optimal conditions for measuring the concentration of elemental groups contained in a urine sample, we decided to use the optimal measurement conditions used in the "cancer risk assessment method" described in Patent Document 1 mentioned above. This is because the optimal measurement conditions used in the "cancer risk assessment method" of Patent Document 1 mentioned above were found to be the optimal conditions for measuring the concentration of elemental groups contained in a blood (plasma or serum) sample, but as a result of tests conducted by the inventors, it was found that they can also be applied to measuring the concentration of elemental groups contained in a urine sample.

[0084] However, if the optimal conditions for measuring the concentration of elements contained in a urine sample are unknown, or if it is desired to find new optimal conditions, they can be found, for example, as follows. First, urine samples belonging to cancer patients and urine samples belonging to healthy individuals are obtained. These urine samples are then randomly divided into two groups based on sex, age, and cancer site (type), with one group designated as the "test urine sample" and the other as the "evaluation urine sample." Subsequently, nitric acid and hydrogen peroxide are mixed into the test urine sample and the evaluation urine sample, respectively, and heated in a suitable sealed container to oxidize and decompose substances such as proteins and phosphates to obtain a homogeneous solution. The sealed container is washed beforehand with nitric acid and ultrapure water to reduce metal contamination. After pre-treating the test urine sample and the evaluation urine sample in this manner so as not to interfere with the measurement of elemental concentrations, an internal standard solution is added to the solution and adjusted to a fixed volume using ultrapure water to obtain the processed solution. Then, the concentrations of the 75 elements contained in each of the processed solutions obtained in this way are measured using ICP mass spectrometry (ICP-MS) under varying conditions, and the optimal conditions for measuring the concentrations of the 75 elements contained in the test urine sample and the evaluation urine sample are found using the obtained measurement results.

[0085] For measuring the concentrations of various elements as in the present invention, ICP mass spectrometry is preferred. This is because ICP mass spectrometry is currently recognized as the simplest method and provides highly quantitative results. However, it goes without saying that the invention is not limited to ICP mass spectrometry. For example, inductively-coupled plasma optical emission spectroscopy (ICP-OES), atomic absorption spectrometry (AAS), and X-ray fluorescence analysis (XRF) can also be used. Furthermore, if other analytical methods (methods for measuring elemental concentrations) more suitable than ICP mass spectrometry are developed in the future, it goes without saying that analytical methods other than ICP mass spectrometry may be used.

[0086] (2. Selection of the element group for evaluation) Next, in order to select the element group to be used for cancer risk assessment (the element group for evaluation), the concentrations (contents) of the 75 elements contained in each of the test urine samples prepared as described above were measured by ICP mass spectrometry under the optimal measurement conditions determined as described above. The obtained concentration data were then statistically analyzed for the differences in elemental concentrations between the two groups: the cancer patient group (case group) and the healthy control group (control group). In this analysis, discriminant analysis and binary logistic regression analysis were performed to identify the elements involved in the differences in elemental concentrations between these two groups and to determine the risk of developing cancer.

[0087] At this time, considering the combinations of elements, the computer repeatedly searched for the combination that showed the greatest difference between the combined elements, that is, the combination of elements that could best distinguish between the two groups, the cancer patient group (case group) and the healthy control group (control group). As a result, it was found that the highest discriminative ability was obtained when using a combination of the 20 elements "Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl". In other words, these 20 element combinations, selected from the 75 element groups contained in each of the test urine samples, were selected as the "evaluation element group". These 20 elements, as a result, were all the elements for which concentration measurement was possible (concentration data was obtained) by ICP mass spectrometry. Furthermore, the combination of the 20 elements identified in this way is clearly different from the combination of the 17 elements "Na, Mg, P, S, K, Ca, Fe, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Ag, Cs" used in the "cancer risk assessment method" of Patent Document 1 mentioned above.

[0088] (3. Measurement and Analysis of the Concentrations of the Evaluation Element Group) Through the preliminary processing described above, the optimal measurement conditions for measuring the concentration of elements contained in the urine sample were determined, and the evaluation element group to be used for cancer risk assessment was selected. The concentrations of the evaluation element group (20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) contained in the urine sample belonging to the cancer patient group and the urine sample belonging to the healthy control group were measured under their respective optimal concentration measurement conditions, and concentration data (concentration values) for these 20 elements were obtained. Discriminant analysis was then performed on the obtained concentration data to obtain a discriminant formula to be used for cancer risk assessment.

[0089] Subsequently, the concentration data of the evaluation element group contained in each of the urine samples belonging to the cancer patient group and the healthy control group were applied to the discriminant formula to calculate the discriminant score. The discriminant score thus obtained can be used as an "indicator" for estimating the risk of cancer in the person (subject). For example, if the discriminant score is above or below a predetermined reference value (e.g., 0), then all persons (subjects) who provided the evaluation urine sample group corresponding to the discriminant score are determined to belong to the case group, and if the discriminant score is below or above the reference value, then they are determined to belong to the control group.

[0090] Furthermore, by performing a binary logistic regression analysis using the aforementioned discriminant score, a relationship (discriminant score-cancer incidence rate relationship) was obtained between the discriminant score and the probability that the person (subject) has cancer. Therefore, by comparing the discriminant score with the discriminant score-cancer incidence rate relationship, it becomes possible to estimate and evaluate the person's (subject's) risk of developing cancer based on its incidence rate.

[0091] In this manner, the evaluation results for the cancer risk of the person (subject) are obtained, and it is preferable that the evaluation results include information indicating the site (type) of cancer present. Furthermore, it is preferable that the evaluation results include the probability of having cancer.

[0092] Furthermore, if, in "1. Determination of Optimal Measurement Conditions" above, the urine samples obtained from the cancer patient group and the healthy control group are divided into a test urine sample group and an evaluation urine sample group, then in "3. Measurement and Analysis of the Concentration of the Evaluation Element Group," the concentration of the evaluation element group contained in the test urine sample group is measured using the test urine sample group under the optimal element concentration measurement conditions described above. Then, by performing discriminant analysis on the obtained concentration data, a discriminant formula used for cancer risk assessment can be obtained. In other words, in this case, the test urine sample group is used from the determination of the optimal measurement conditions to the generation of the discriminant formula. Subsequently, in the same manner as with the test urine sample group, the concentration of the evaluation element group is measured in the evaluation urine sample group to obtain concentration data. Then, the concentration data of the evaluation urine sample group obtained in this way is applied to the discriminant formula. In this way, the discriminant score can be calculated. From there onward, the procedure is the same as when the obtained urine samples from the cancer patient group and the healthy control group are not divided into a test urine sample group and an evaluation urine sample group.

[0093] (3-1. Details of Discriminant Analysis) Next, we will explain the discriminant analysis described above in detail.

[0094] First, for the 20 elements (Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) that were measured as the "evaluation element group" mentioned above, we performed discriminant analysis between two groups: a "healthy control group" and a "cancer patient group" as the case group.

[0095] Specifically, a t-test was first conducted to determine the difference in population means between two groups: a healthy control group and a cancer patient group. This was to investigate the extent to which the aforementioned group of 20 elements used for evaluation influenced the distinction between these two groups. The results of this test showed differences between the two groups for each element individually, but it was found that this analysis ignored the relationships between elements and had many problems for use in risk assessment of cases. Therefore, it was determined that it was necessary to solve the aforementioned problems by analyzing the data using "discriminant analysis," a type of multivariate analysis that can take into account the relationships between elements.

[0096] Therefore, a discriminant formula was derived as follows. This was done to analyze the concentration balance (correlation) among the aforementioned group of elements for evaluation. Since the concentrations of individual elements in the aforementioned group of elements for evaluation vary from person to person and are difficult to use as indicators, the correlation between the concentrations of the aforementioned group of elements for evaluation was determined.

[0097] The discriminant function can generally be expressed as shown in the following formula (1): Discriminant score (D) = Function (F) (explanatory variables 1 to n, discriminant coefficients 1 to n) (1) (where n is an integer greater than or equal to 2)

[0098] The discriminant function in equation (1) can be written as follows, equation (2), by considering the weights (degree of influence on discrimination) of each explanatory variable from 1 to n: Discriminant score (D) = (Discriminant coefficient 1) × (Explanatory variable 1) + (Discriminant coefficient 2) × (Explanatory variable 2) + ... + ... (Discriminant coefficient n) × (Explanatory variable n) + constant (2)

[0099] Therefore, by using the concentrations of the aforementioned evaluation elements (20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) selected from the results of a t-test comparing the population means of two groups, a healthy control group and a cancer patient group, as explanatory variables 1 to n, and using the discriminant coefficients 1 to n as their weights (where n=20), a discriminant formula was obtained. This discriminant formula could be easily obtained by loading the concentration values ​​(concentration data) of these 20 elements into a known discriminant analysis program (e.g., SAS, SPSS) on a known computer. The discriminant coefficients 1 to n (where n=20) of this discriminant formula were determined by the discriminant analysis program so as to maximize the correlation ratio between the actual values ​​of the presence or absence of cancer and the discriminant score (D), in other words, so as to maximize the degree of separation (discriminative ability) between the healthy control group and the cancer patient group.

[0100] Once a discriminant formula is obtained, the concentration (concentration data) of the urine sample group for evaluation is applied to the discriminant formula on the computer to obtain a discriminant score (D). If the discriminant score (D) is less than or equal to (or greater than) a reference value (e.g., 0), the subject is determined to belong to the cancer patient group (case group), and if the discriminant score (D) is greater than or equal to (or less than) the reference value, the subject is determined to belong to the healthy person group (control group).

[0101] The concentration data (concentration values) of the urine sample group used for evaluation were applied to the discriminant formula obtained by the discriminant analysis described above, and a discriminant score was obtained to determine whether the subject belonged to the cancer patient group or the healthy control group. Good prediction results were obtained.

[0102] (3-2. Details of Binary Logistic Regression Analysis) Next, we will explain the binomial logistic regression analysis described above in detail.

[0103] In addition to determining whether the aforementioned subjects belong to the cancer patient group or not, we performed a binary logistic regression analysis to determine the probability (cancer incidence rate) that the subjects belong to either the cancer patient group (case group) or the healthy control group (control group).

[0104] The incidence rate is generally given by the following formula (3), using the discriminant score (D) obtained by the discriminant analysis described above: Incidence rate = 1 / [1 + exp(-discriminant score)] (3)

[0105] Using formula (3), the incidence rate can be obtained using the discriminant score (D), so the probability that the subject will be included in the case group (cancer patient group) (cancer incidence rate) can be determined. In other words, the subject can know their current risk of developing cancer probabilistically.

[0106] Applying the discriminant score (D) obtained from the discriminant analysis between two groups, a healthy group (control group) and a cancer patient group (case group), to the above formula (3) gives the probability that the subject falls into the cancer patient group (cancer incidence rate).

[0107] When the concentration data of the aforementioned evaluation element group was applied to the discriminant formula for cancer patients to obtain the discriminant score (D), and this score was then applied to the formula (3) obtained by the binomial logistic regression analysis, the probability of developing cancer was obtained, and a good prediction result was obtained.

[0108] The cancer risk assessment method of the present invention uses 20 elements (Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) selected in the preliminary treatment described above as an "assessment element group." The concentrations of these 20 elements contained in a urine sample of an unknown subject (a person who wishes to have their cancer risk assessed) are measured under the optimal measurement conditions determined in the preliminary treatment described above. The obtained concentration data (concentration values) of the 20 elements are applied to a discriminant function to calculate the correlation between the concentrations of the assessment element group. Based on the obtained correlation, an index is generated to identify whether or not the unknown subject has cancer, and the cancer risk of the unknown subject is evaluated based on this index to create an assessment result. Based on the aforementioned correlation, once the discriminant coefficient of the discriminant function for determining whether the unknown subject belongs to the control group or the case group is determined, a discriminant formula is generated. By applying the concentration data of the evaluation element group present in the urine sample of the unknown subject to this discriminant formula, an evaluation result can be obtained as to whether or not the unknown subject has cancer.

[0109] (Flowchart of the cancer risk assessment method of the present invention) Next, the flowchart of the cancer risk assessment method of the present invention will be explained with reference to Figure 1.

[0110] First, urine collected from an unknown subject (a person who wishes to undergo cancer risk assessment) is placed in a container such as a test tube (container 1) to create urine sample 2. Then, urine sample 2 is analyzed using a suitable analytical device (an ICP mass spectrometer is preferred, but not limited thereto) to measure the concentration of the evaluation elements (concentrations of the 20 elements) present in urine sample 2. In this way, concentration data (concentration values) of the evaluation elements contained in urine sample 2 are obtained (step S1).

[0111] The group of elements for evaluation that are subject to concentration measurement is the combination of the 20 elements mentioned above (Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl), which is different from the combination of 17 elements (Na, Mg, P, S, K, Ca, Fe, Cu, Zn, Se, Rb, Sr, As, Mo, Cs, Co, Ag) used as the "group of elements for evaluation" in the aforementioned Patent Document 1.

[0112] Next, the concentration data of the evaluation element group obtained in step S1 is applied to one or more discriminant functions (discriminant formulas) for determining whether the unknown subject belongs to the case group (cancer patient group) or the control group (healthy person group), and the correlation between the concentration data of the evaluation element group is calculated (step S2).

[0113] Specifically, for example, on a computer on which a known discriminant analysis program (e.g., SAS, SPSS) is installed, the 20 concentration data of the evaluation element group obtained in step S1 are loaded into the discriminant analysis program. At this time, the 20 concentration data (concentration values) of the evaluation element group obtained in step S1 are applied to each of the explanatory variables 1 to 20 of the one or more discriminant functions (or discriminant formulas), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to each of the discriminant coefficients 1 to 20 of the one or more discriminant functions (or discriminant formulas).

[0114] For example, when using the discriminant function (or discriminant formula) prepared to determine whether or not a person has some kind of cancer, the 20 concentration data (concentration values) of the evaluation element group obtained in step S1 are applied to the explanatory variables 1 to 20 of the discriminant function (or discriminant formula), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to the discriminant coefficients 1 to 20 of the discriminant function (or discriminant formula). In this way, the "cancer risk" of the unknown subject, in other words, the "risk that the unknown subject has some kind of cancer" can be evaluated.

[0115] On the other hand, when using the multiple discriminant functions (or discriminant formulas) prepared to determine whether or not cancer is present in specific different sites, the 20 concentration data (concentration values) of the evaluation element group obtained in step S1 are applied to each of the explanatory variables 1 to 20 of the multiple discriminant functions (or discriminant formulas), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to each of the discriminant coefficients 1 to 20 of the multiple discriminant functions (or discriminant formulas). In this way, the "risk of cancer in specific different sites" of the unknown subject, in other words, the "risk of the unknown subject having any of the cancers in specific different sites" can be evaluated simultaneously.

[0116] Next, based on the calculation results obtained in step S2 (correlation between the 20 concentration data of the evaluation element group), an "indicator" indicating the cancer risk of the unknown subject is generated (step S3).

[0117] Specifically, for example, in step S2, one or more discriminant scores (D) are obtained by applying the 20 concentration data (concentration values) of the evaluation element group obtained in step S1 and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above to one or more discriminant functions (or discriminant formulas). The one or more discriminant scores (D) thus obtained can be used as an "indicator" to estimate the "risk of the unknown subject having some kind of cancer" or the "risk of having one of the cancers in a specific different site."

[0118] The estimation of the cancer risk can be performed, for example, by comparing each of the one or more discriminant scores (D) with a predetermined reference value (e.g., 0). That is, if each of the one or more discriminant scores (D) is equal to (or equal to) the corresponding reference value, the subject is determined to belong to the case group (cancer patient group). This means that it is possible to infer that "the unknown subject has some kind of cancer" or "the unknown subject has one of the cancers of a specific different site." On the other hand, if the one or more discriminant scores (D) are equal to (or equal to) the corresponding reference value, the subject is determined to belong to the control group (healthy person group). This means that it is possible to infer that "the unknown subject does not have cancer."

[0119] Thus, based on the indicators obtained in step S3, an "evaluation result" including the cancer risk of the unknown subject is generated (step S4).

[0120] Furthermore, in the cancer risk assessment method of the present invention, the one or more discriminant functions (or discriminant formulas) used in step S2 differ depending on whether the risk of developing any cancer is being assessed or whether the risk of developing cancer in a specific different site is being assessed. Therefore, this method has the advantage of not only determining whether the unknown subject has any cancer, but also determining the site (type) of cancer in which the unknown subject is presumed to have. Preferably, the type of cancer in which the unknown subject is presumed to have, as determined in this way, is included in the assessment results.

[0121] Furthermore, the cancer risk assessment method of the present invention has the advantage that, depending on the site (type) of cancer, the element that is significant in distinguishing between the case group (including the observer group) and the control group (healthy persons group) differs (selected from the 20 elements of the evaluation element group). Therefore, depending on which element is significant in the distinction, the site (type) of cancer that the unknown subject is presumed to have can also be determined. Preferably, the site (type) of cancer that the unknown subject is presumed to have, as determined in this way, is included in the evaluation results.

[0122] Furthermore, in the cancer risk assessment method of the present invention, a discriminant score-cancer incidence rate relationship can be prepared in advance based on the results of the binary logistic regression analysis described above (see, for example, Figures 32 and 35). In this case, by comparing the discriminant score (D) obtained as the index in step S3 with the discriminant score-cancer incidence rate relationship, it is possible to evaluate the cancer incidence risk of the unknown subject who is presumed to have cancer, based on the numerical value of the incidence rate (probability of cancer incidence). Preferably, the incidence rate of the cancer that the unknown subject is presumed to have, as determined in this way, is included in the evaluation results.

[0123] As described above, by using a computer with a known discriminant analysis program (e.g., SAS, SPSS) installed, and by pre-applying the 20 discriminant coefficient values ​​of "the discriminant function (or discriminant expression) prepared for determining the presence or absence of any kind of cancer" or "the multiple discriminant functions (or discriminant expressions) prepared for determining the presence or absence of cancer in specific different parts of the body" obtained in the discriminant analysis, one or more discriminant scores (D) as "indicators" can be obtained simply by loading the 20 concentration data of the evaluation element group for the unknown subject obtained in step S1 into the discriminant analysis program, and the cancer risk evaluation result can be obtained immediately based on the one or more discriminant scores (D).

[0124] Therefore, it becomes possible to easily and quickly assess cancer risk in response to the requests of those who wish to have their cancer risk assessed. Specifically, the assessment results regarding "whether or not the applicant has any type of cancer" or "whether or not the applicant has any cancer of a specific different site" are provided to the applicant.

[0125] Furthermore, if the applicant is determined to have "some kind of cancer" or "one of the cancers of a specific different site," it is preferable that the evaluation results, along with the probability of developing that cancer (cancer incidence rate), be provided to the unknown subject.

[0126] (Configuration of the cancer risk assessment system of the present invention) Next, the cancer risk assessment system of the present invention will be described.

[0127] Figure 2 shows the configuration of the cancer risk assessment system 10 of the present invention. This system 10 is for carrying out the cancer risk assessment method of the present invention described above.

[0128] As is clear from Figure 2, the cancer risk assessment system 10 of the present invention comprises a data storage unit 11, a calculation unit 12, and an assessment result generation unit 13.

[0129] An external urinary element concentration measurement unit 5 is provided in the cancer risk assessment system 10. Using a urine sample 2 collected from an unknown subject (a person who wishes to undergo cancer risk assessment) placed in an appropriate container 1 such as a test tube, the concentration of the evaluation element group (a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) in the urine sample 2 is measured. The concentration data (concentration values) of the evaluation element group in the urine sample 2 obtained by the urinary element concentration measurement unit 5 is supplied to and stored in the data storage unit 11 of the cancer risk assessment system 10. For example, a known ICP mass spectrometer is used as the urinary element concentration measurement unit 5, but it is not limited to this. Needless to say, ICP emission spectrometers, atomic absorption spectrometers, X-ray fluorescence spectrometers, etc. can also be used.

[0130] The data storage unit 11 is a part that stores (saves) the concentration data of the 20 elements of the evaluation group for the unknown subject (the volunteer) obtained by the urine element concentration measurement unit 5. The data storage unit 11 is usually composed of a known storage device (for example, a semiconductor memory or magnetic memory), but any storage device capable of storing the concentration data can be used.

[0131] The calculation unit 12 is the part that performs the calculations necessary to derive the correlation between the 20 concentration data of the pre-evaluation element group stored in the data storage unit 11. The calculation unit 12 is usually configured using a known discriminant analysis program, but is not limited to this. It may also be configured with a program created to include the functions of the discriminant analysis program. In short, it is sufficient as long as it can perform the necessary calculations described above, and its configuration is not important.

[0132] Specifically, the calculation unit 12 reads out the 20 concentration data of the evaluation element group stored in the data storage unit 11, applies it to one or more discriminant functions (or discriminant formulas) for determining whether the unknown subject belongs to the case group (cancer patient group) or the control group (healthy person group), and calculates the correlation between the 20 concentration data of the evaluation element group.

[0133] For example, the 20 concentration data points of the evaluation element group stored in the data storage unit 11 are loaded into the discriminant analysis program of the calculation unit 12. At this time, the 20 concentration data points (concentration values) of the evaluation element group loaded from the data storage unit 11 are applied to each of the explanatory variables 1 to 20 of the one or more discriminant functions (or discriminant formulas), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to each of the discriminant coefficients 1 to 20 of the one or more discriminant functions (or discriminant formulas).

[0134] For example, if the discriminant function (or discriminant formula) prepared to determine whether or not a person has some kind of cancer is used, the 20 concentration data (concentration values) of the evaluation element group read from the data storage unit 11 are applied to the explanatory variables 1 to 20 of the discriminant function (or discriminant formula), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to the discriminant coefficients 1 to 20 of the discriminant function (or discriminant formula). In this way, the "cancer risk" of the unknown subject, in other words, the "risk that the unknown subject has some kind of cancer" can be evaluated.

[0135] On the other hand, when the multiple discriminant functions (or discriminant formulas) prepared to determine whether or not cancer has occurred in specific different parts of the body are used, the 20 concentration data (concentration values) of the evaluation element group read from the data storage unit 11 are applied to each of the explanatory variables 1 to 20 of the multiple discriminant functions (or discriminant formulas), and the 20 discriminant coefficient values ​​obtained in the discriminant analysis described above are applied to each of the discriminant coefficients 1 to 20 of the multiple discriminant functions (or discriminant formulas). In this way, the "risk of cancer occurring in specific different parts of the body" of the unknown subject, in other words, the "risk of the unknown subject having any of the cancers in specific different parts of the body" can be evaluated simultaneously.

[0136] The calculation unit 12 uses the one or more discriminant functions (or discriminant formulas) to calculate one or more discriminant scores (D). Each of these one or more discriminant scores (D) serves as an "indicator" for identifying whether or not the unknown subject has some kind of cancer. Based on the one or more "indicators" obtained in this way, it becomes possible to estimate the cancer risk of the unknown subject.

[0137] The evaluation result generation unit 13 is a unit that generates an "evaluation result" based on the calculation result output from the calculation unit 12, that is, the one or more discrimination scores (D) as the one or more "indicators," which describe whether or not the unknown subject has any kind of cancer, and if so, which part (type) of cancer it is, and outputs this "evaluation result" to the outside of the cancer risk assessment system 10. The evaluation result generation unit 13 is usually composed of a program created to realize its function, but is not limited to this. Since the function of the evaluation result generation unit 13 is closely related to the function of the calculation unit 12, the evaluation result generation unit 13 may be composed of a program created to utilize the function of the discrimination analysis program that constitutes the calculation unit 12. In this case, the functions of the calculation unit 12 and the evaluation result generation unit 13 are realized in a single program; in other words, the calculation unit 12 and the evaluation result generation unit 13 are configured as an integrated unit. In short, it is sufficient as long as the evaluation result generation unit 13 can realize the above-described function, and the configuration is not important.

[0138] In the evaluation result generation unit 13, the risk of developing cancer is estimated by comparing each of the one or more discrimination scores (D) with a corresponding reference value (e.g., 0). That is, if each of the one or more discrimination scores (D) is greater than or equal to (or less than or equal to) the corresponding reference value, it is determined that the subject belongs to the case group (cancer patient group). This means that it is possible to infer that "the unknown subject has some kind of cancer" or "the unknown subject has one of the cancers in a specific different part of the body". On the other hand, if the discrimination score (D) is less than or equal to (or greater than or equal to) the corresponding reference value, it is determined that the subject belongs to the control group (healthy person group). This means that it is possible to infer that "the unknown subject does not have cancer".

[0139] In this way, the "evaluation results" generated by the evaluation result generation unit 13 reveal the cancer risk of the unknown subject (the applicant).

[0140] Furthermore, the cancer risk assessment system 10 of the present invention has the advantage that, since the calculation unit 12 can use the multiple different discriminant functions (or discriminant formulas) mentioned above, it can determine not only whether or not the unknown subject has some kind of cancer, but also the location (type) of cancer that the unknown subject is presumed to have. Preferably, the type of cancer that the unknown subject is presumed to have, as determined in this way, is included in the assessment results.

[0141] Furthermore, the cancer risk assessment system 10 of the present invention has the advantage that, depending on the site (type) of cancer, the element that is significant in distinguishing between the case group (including the observer group) and the control group (healthy persons group) differs (selected from the 20 elements of the evaluation element group). Therefore, depending on which element is significant in the distinction, the site (type) of cancer that the unknown subject is presumed to have can also be determined. Preferably, the site (type) of cancer that the unknown subject is presumed to have, as determined in this way, is included in the evaluation results.

[0142] Furthermore, in the cancer risk assessment system 10 of the present invention, based on the results of the binary logistic regression analysis described above, the data of the discriminant score-cancer incidence relationship can be prepared in advance and stored in the calculation unit 12, the evaluation result generation unit 13, or other memory location (not shown) (see, for example, Figures 32 and 35). In this case, by comparing the one or more discriminant scores (D) with the corresponding discriminant score-incidence relationship read from the calculation unit 12, the index generation unit 13, or other memory location, it is possible to evaluate the cancer risk of the unknown subject (the applicant) who is presumed to have cancer, based on their incidence rate (probability of developing cancer). Preferably, the incidence rate of the cancer that the unknown subject (the applicant) is presumed to have, as determined in this way, is included in the evaluation result.

[0143] The cancer risk assessment system 10 of the present invention, having the above configuration and functions, can be implemented using, for example, a known personal computer, but is not limited thereto. Needless to say, any other computer may be used. Furthermore, for example, if a computer with a known discriminant analysis program (e.g., SAS, SPSS) installed is used as the cancer risk assessment system 10, and the 20 discriminant coefficient values ​​of "the discriminant function (or discriminant expression) prepared to determine the presence or absence of some kind of cancer" or "the multiple different discriminant functions (or discriminant expressions) prepared to determine the presence or absence of cancer in specific different sites" obtained by the above-mentioned discriminant analysis are applied in advance to the discriminant coefficients 1 to 20 of each of the one or more discriminant functions (or discriminant expressions), then by simply reading the 20 concentration data of the evaluation element group for the unknown subject from the data storage unit 11 and loading them into the discriminant analysis program, one or more discriminant scores (D) can be obtained as "indicators," and the cancer risk assessment result can be obtained immediately based on the one or more discriminant scores (D).

[0144] Therefore, it becomes possible to easily and quickly assess cancer risk in response to the requests of those who wish to have their cancer risk assessed. Specifically, the assessment results regarding "whether or not the applicant has any type of cancer" or "whether or not the applicant has any cancer of a specific different site" are provided to the applicant.

[0145] Furthermore, if the aforementioned applicant is determined to have "some kind of cancer" or "one of the cancers of a specific different site," the evaluation results, which include this fact along with the probability (incidence rate) of developing that cancer, will be provided to the aforementioned unknown subject.

[0146] When the cancer risk assessment method of the present invention, as described with reference to Figure 1, is implemented using the cancer risk assessment system 10 of the present invention, as described with reference to Figure 2, or other systems, for example, the risk of developing cancer is calculated by pattern analysis of the concentration data (concentration values) of a group of evaluation elements (specifically, 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) in a urine sample 2 collected from an unknown subject, and an assessment result is submitted that probabilistically expresses the possibility of cancer based on that risk. Specifically, a predetermined amount (e.g., 0.5 cc) of urine sample 2 is collected from the urine of an unknown subject collected during a health checkup, and the concentration of the evaluation group in the urine sample 2 is measured at a testing facility. Then, based on the concentration data of the evaluation group measured at the testing facility, the risk of developing cancer is calculated at an institution such as a risk assessment center (tentative name). The calculation result of the cancer risk is then provided to the unknown subject (patient). If cancer is suspected, patients will be encouraged to undergo "current cancer screening" at an appropriate medical institution. Personal information will be encrypted or assigned sequential numbers, and the system will be designed so that personal information is not transmitted to testing institutions or risk assessment centers.

[0147] The present invention described above will be explained in more detail below based on examples.

[0148] As shown in Figure 4, the subjects who provided urine samples used in each of the following examples were healthy individuals, cancer patients suffering from one of the following 10 types of cancer (stomach cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, lung cancer, bile duct cancer, uterine cancer, ovarian cancer, and cervical cancer), and cancer patients suffering from both stomach cancer and pancreatic cancer. Of these subjects, the cancer patient group (case group) consisted of 205 men and 166 women, totaling 371 cancer patients. On the other hand, the healthy control group consisted of 78 men and 90 women, totaling 168 healthy individuals. The total number of subjects was 539, with 52.5% being male and 47.5% being female. The age distribution of these subjects, from the healthy control group to the six cancer patient groups (stomach cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer), is shown in Figure 5, with a relatively large proportion being in their 50s to 70s. Furthermore, the mean, standard deviation, minimum, and maximum ages of these subjects are shown in Figure 6.

[0149] The urine samples from the aforementioned subjects (the healthy individuals and the cancer patients) were all stored at -80°C at the Osaka International Cancer Institute and collected by the inventors with the consent of the Institute. The urine samples from the healthy individuals were collected at the Institute during health checkups and were from individuals diagnosed as not having cancer. The urine samples from the cancer patients were collected at the Institute at the time of their initial consultation after being definitively diagnosed with cancer.

[0150] The inventors obtained the desired results through a series of steps, from obtaining (collecting) urine samples from the healthy individuals and cancer patients described above to performing statistical analysis on those urine samples. The overall process is shown in Figure 3.

[0151] First, urine samples from healthy individuals and cancer patients were obtained (collected) (Step S11), and then these urine samples were pre-treated (Step S12). Specifically, a predetermined amount (e.g., 50 μL) of the urine sample was dispensed into a suitable sealed container, nitric acid and hydrogen peroxide were added, and the container was heated at a predetermined temperature for a predetermined time. This was done to oxidize and decompose substances such as proteins and phosphates contained in the urine to create a homogeneous solution. Subsequently, an internal standard solution was added to the solution, and the volume was adjusted to a predetermined amount (e.g., 2.5 mL) with ultrapure water. After that, the concentrations of "trace elements" contained in each of the urine samples were measured by inductively coupled plasma mass spectrometry (ICP-MS) (Step S13), and the concentration data (concentration values) of these "trace elements" were obtained (Step S14). The aforementioned "trace elements" refer to the "group of elements for evaluation" described above, specifically the 20 elements listed above (Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl).

[0152] The concentration data (concentration values) of the 20 elements obtained in this manner were first subjected to a natural logarithmic transformation without correcting for urinary creatinine (Step S15), and then discriminant analysis was performed on the cancer patient group (case group) and the healthy control group (control group) (Step S16). As a result, multiple discriminant formulas were obtained, elements significant in the discriminant analysis were identified, and multiple discriminant scores were calculated (Step S17). Furthermore, a distribution of discriminant scores, a cross-tabulation table, an incidence rate graph, and an ROC curve were created (Step S18).

[0153] On the other hand, the concentration data (concentration values) of the 20 elements mentioned above were corrected for urinary creatinine (step S19), and the concentration data (concentration values) after urinary creatinine correction were subjected to a natural logarithmic transformation (step S20). Then, discriminant analysis was performed on the cancer patient group (case group) and the healthy control group (control group) (step S21). As a result, multiple discriminant formulas were obtained, elements significant in the discriminant analysis were identified, and multiple discriminant scores were calculated (step S22). Furthermore, a distribution of discriminant scores, a cross-tabulation table, an incidence rate graph, and an ROC curve were created (step S23). This was done to observe the effect of urinary creatinine correction.

[0154] The "urinary creatinine correction" mentioned above refers to dividing each of the concentration data (concentration values) for the 20 elements mentioned above by the urinary creatinine (urinary Cre) value. Since the concentrations of all urinary components are affected by factors such as water intake and sweating, and change significantly with fluctuations in urine volume, "urinary creatinine correction" is performed to suppress the influence of these fluctuations in urinary component concentrations on the analysis. "Urinary creatinine correction" has the advantage of improving the normal distribution of the concentration data (concentration values) for the 20 elements, thereby improving the accuracy of the statistical analysis (steps S16-S18, S21-S23).

[0155] The following examples were conducted using combinations of two groups classified by sex: (a) healthy males (78 cases) and males with all types of cancer (205 cases), and (b) healthy females (90 cases) and females with all types of cancer (166 cases). In addition, the following combinations were conducted without classification by sex: (c) healthy males and gastric cancer patients, (d) healthy males and colorectal cancer patients, (e) healthy males and pancreatic cancer patients, (f) healthy males and esophageal cancer patients, (g) healthy males and malignant lymphoma patients, and (h) healthy males and lung cancer patients. Therefore, these examples include a total of eight combination patterns consisting of two combinations of groups classified by sex and six combinations of groups not classified by sex. Furthermore, for each of these eight combination patterns, there were cases with and without correction for urinary creatinine. Therefore, the following examples were carried out using a total of 16 combination patterns of groups.

[0156] The results of the examples described below demonstrate that the risk of developing cancer can be calculated using concentration data of 20 specific elements contained in a urine sample. The reason why it is possible to estimate the risk of developing cancer of different sites (types) using the concentration data of the 20 elements obtained from a single urine collection is that, as shown in Figures 15 to 30, the elements that have a significant correlation for discrimination differ depending on the site (type) of cancer.

[0157] Below, Examples 1 to 16, corresponding to the combination patterns of the 16 groups mentioned above, will be described in order with reference to the attached drawings. In Examples 1 to 16 below, the presence or absence of a single type of cancer is evaluated (determined), but it is also possible to combine the discriminant formulas obtained in Examples 1 to 16 as appropriate. This has the advantage of enabling evaluation (determining) whether or not a subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer.

[0158] First, let's describe Example 1. In Example 1, the risk of developing cancer was evaluated in two groups: a group of healthy men (78 cases) and a group of men with all types of cancer (205 cases). Urine creatinine correction was not performed. Here, the site (type) of cancer was not considered; the risk of the aforementioned male subjects developing some form of cancer was evaluated.

[0159] First, the concentrations of the aforementioned evaluation elements, namely the 20 elements (Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, Tl) contained in each of the 283 urine samples provided by 78 healthy men and 205 men with all types of cancer (i.e., patients suffering from any of the 10 types of cancer: stomach cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, lung cancer, bile duct cancer, uterine cancer, ovarian cancer, and cervical cancer), were measured without correcting for urinary creatinine. This concentration measurement was performed using the "optimal measurement conditions" found during the development of the "cancer risk assessment method" described in Patent Document 1, as described in "1. Determination of Optimal Measurement Conditions". Furthermore, the 20 elements designated as the evaluation element group were obtained by searching for the combination of elements that showed the greatest difference when statistically analyzing the difference in elemental concentrations between the two groups, the cancer patient group (case group) and the healthy control group (control group), using a computer. In other words, the combination of elements that best distinguishes between the two groups (cancer patient group (case group) and healthy control group (control group)) was obtained. Figure 7 shows the 20 concentration data (concentration values) obtained from this measurement transformed using the natural logarithm.

[0160] Figure 7 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 1: a group of healthy men and a group of all cancer patients. The elements that showed a large difference in mean concentration (mean concentration difference) between these two groups were Na, P, S, K, Co, Cu, Zn, Se, Rb, Cd, Cs, and Tl—12 elements in total. These 12 elements were judged to have a statistically significant difference, with p-values ​​obtained from t-tests performed on these two groups being p < 0.05. (This is also true for Examples 2 to 16 below.) The large difference in mean concentration of these elements suggests that they may be influencing the development of cancer in men, and also suggests that there is some strong correlation between them and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0161] Figure 15 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 1. From Figure 15, it can be seen that, without correction for urinary creatinine, the elements that are significant in distinguishing between the two groups, healthy males and all male cancer patients, are P (p < 0.01), K (p < 0.01), Ca (p < 0.01), and Cu (p < 0.01). Therefore, if an unknown subject (a person who wishes to have their cancer risk assessed) is male, and it is found that the elements that are significant in distinguishing between them are P, K, Ca, and Cu, it can be inferred that the unknown male subject has some kind of cancer.

[0162] Figure 11(a) shows the discriminant formula obtained from the two-group discriminant analysis in Example 1. In this discriminant formula, the group of healthy men and the group of all cancer patients in men are used as dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy men group and the group of all cancer patients in men are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0163] The results of the discriminant analysis of the two groups in Example 1 are shown in Figures 31 to 33.

[0164] Figure 31(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 1. From this figure, it can be seen that the majority of the discriminant scores of the healthy male group are distributed in the left region of the figure, and the majority of the discriminant scores of the male cancer patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided to the left and right near the point (location) where the discriminant score is -0.5. It was also found that in the region close to the point (location) where the discriminant score is -0.5, the discriminant scores of the case group (all male cancer patients) and the control group (healthy males) are mixed.

[0165] Figure 31(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 1. From this figure, in the case group (all male cancer patients), 196 out of 205 samples were correctly identified (sensitivity = 196 / 205 = 95.61%), and in the control group (healthy males), 70 out of 78 samples were correctly identified (specificity = 70 / 78 = 89.74%). Therefore, 266 out of 283 samples targeted for discriminant analysis were correctly identified, resulting in an accuracy of 93.99% (266 / 283). Since this accuracy is sufficiently high, it can be concluded that the discriminant analysis in the two-group analysis in Example 1 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of cancer in men can be expected.

[0166] Figure 32 is a graph showing the relationship between the discriminant score obtained in the two-group discriminant analysis in Example 1 and the probability that the subject has some kind of cancer (discriminant score - overall cancer incidence rate relationship). This graph was obtained by performing a binary logistic regression analysis using the discriminant score obtained in the discriminant analysis. According to the graph, the larger the positive value of the discriminant score, the higher the probability (risk) of developing cancer. For example, if the discriminant score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have cancer with a probability of 90% or more." Conversely, for example, if the discriminant score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have cancer."

[0167] Figure 33(a) shows the ROC (Receiver Operating Characteristic) curve obtained from the two-group discriminant analysis in Example 1. This ROC curve was obtained by sequentially changing the discriminant score obtained from the two-group discriminant analysis in Example 1 from the lowest value to the highest value, and calculating the sensitivity and specificity from the number of urine samples classified as the control group (healthy males) and the case group (all male cancer patients). The area below the curve can be used to evaluate the goodness of the discriminant (prediction) fit.

[0168] Using the ROC curve in Figure 33(a), the Area Under the Curve (AUC) was calculated, yielding a high value of 0.9592, as shown in Figure 33(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9592 obtained in Example 1 indicates that the discrimination (prediction) between the two groups in Example 1 was very accurate. Therefore, it was found that the method of Example 1 is sufficiently effective as a tool for diagnosing the risk of cancer in men.

[0169] Next, Example 2 will be described. This example is the same as Example 1 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here as well, the site (type) of cancer was not considered, and the risk of the male subjects having some kind of cancer was evaluated.

[0170] In other words, in this Example 2, similar to Example 1 described above, the risk of developing cancer in men was evaluated using a combination of two groups: a group of healthy men (78 cases) and a group of men with all types of cancer (205 cases). However, unlike Example 1 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the group of men with all types of cancer (case group) and the group of healthy men (control group).

[0171] Figure 8 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0172] Figure 8 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 2: a group of healthy men and a group of all cancer patients. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, and Ba—18 elements in total. The large difference in mean concentration of these elements suggests that they may be influencing the development of cancer in men, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0173] Figure 16 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 2. From Figure 16, it can be seen that, with urinary creatinine correction, the elements that are significant in distinguishing between the two groups, healthy males and all male cancer patients, are P (p < 0.01), K (p < 0.01), Ca (p < 0.01), and Cu (p < 0.01). This is the same as in Example 1 above, without urinary creatinine correction. Therefore, if an unknown subject (a person who wishes to have their cancer risk assessed) is male, and it is found that the elements that are significant in distinguishing between them are P, K, Ca, and Cu, it can be inferred that the unknown male subject has some kind of cancer. Comparing Example 2 with Example 1 above, it can be seen that, in the case of all male cancers, the elements that are significant in distinguishing between them do not change depending on whether or not urinary creatinine correction is applied.

[0174] Figure 13(a) shows the discriminant formula obtained from the two-group discriminant analysis in Example 2. Similar to Example 1, this discriminant formula also uses the healthy male group and the all-cancer male group as the dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy male group and the all-cancer male group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the elements with discriminant coefficients that have a large absolute value have a greater influence on the discriminant score (D).

[0175] The results of the discriminant analysis of the two groups in Example 2 are shown in Figures 34 to 36.

[0176] Figure 34(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 2. From this figure, it can be seen that the majority of the discriminant scores of the healthy male group are distributed in the left region of the figure, and the majority of the discriminant scores of the all-cancer male group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is -0.5. It was also found that in the region close to the point (location) where the discriminant score is -0.5, the discriminant scores of the case group (all-cancer male group) and the control group (healthy male group) are mixed.

[0177] Figure 34(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 2. From this figure, in the case group (all male cancer patients), 194 out of 205 samples were correctly identified (sensitivity = 194 / 205 = 94.63%), and in the control group (healthy males), 70 out of 78 samples were correctly identified (specificity = 70 / 78 = 89.74%). Therefore, 264 out of 283 samples targeted for discriminant analysis were correctly identified, and the accuracy was 93.29% (264 / 283). Since this accuracy is sufficiently high, it can be concluded that the discriminant analysis in the two-group analysis in Example 2 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of cancer in men can be expected.

[0178] Figure 35 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 2 and the probability that the subject has some kind of cancer (discrimination score - overall cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject does not have cancer."

[0179] Figure 36(a) shows the ROC curve obtained from the discriminant analysis of the two groups in Example 2.

[0180] Using the ROC curve in Figure 36(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9598, as shown in Figure 36(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9598 obtained in Example 2 indicates that the discrimination (prediction) between the two groups in Example 2 was very accurate. Therefore, it was found that the method of Example 2 is sufficiently effective as a tool for diagnosing the risk of cancer in men.

[0181] Next, we will describe Example 3. In this example, the risk of developing cancer was evaluated in two groups: a group of healthy women (90 cases) and a group of women with all types of cancer (166 cases). Urine creatinine correction was not performed. Here, the site (type) of cancer was not considered, and the risk of the aforementioned female subjects developing some form of cancer was evaluated.

[0182] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 256 urine samples provided by 90 healthy women and 166 women with cancer were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 7.

[0183] Figure 7 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 3: a group of healthy women and a group of women with all types of cancer. The elements that showed a large difference in mean concentration between these two groups were B, Na, Mg, P, K, Co, Cu, Zn, As, Rb, Cd, Cs, and Tl—13 elements in total. The large difference in mean concentration of these elements suggests that they may be influencing the development of cancer in women, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0184] Figure 17 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 3. From Figure 17, it can be seen that, without correction for urinary creatinine, the elements that are significant in distinguishing between the two groups, healthy women and women with all types of cancer, are B (p < 0.01), P (p < 0.001), S (p < 0.001), K (p < 0.001), Co (p < 0.05), Cu (p < 0.05), Mo (p < 0.05), Cd (p < 0.01), Cs (p < 0.05), and Tl (p < 0.05). Therefore, if the unknown subject (the person who requested cancer risk assessment) is female, and it is found that the 10 elements B, P, S, K, Co, Cu, Mo, Cd, Cs, and Tl are significant in the identification, it becomes possible to infer that the unknown female subject has some form of cancer.

[0185] Figure 11(b) shows the discriminant formula obtained from the two-group discriminant analysis in Example 3. In this discriminant formula, the group of healthy women and the group of women with all cancers are used as the dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy women group and the group of women with all cancers are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0186] The results of the discriminant analysis of the two groups in Example 3 are shown in Figures 37 to 39.

[0187] Figure 37(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 3. From this figure, it can be seen that the majority of the discriminant scores of the healthy female group are distributed in the left region of the figure, and the majority of the discriminant scores of the female all-cancer patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided into left and right sides near the point (location) where the discriminant score is -0.5. It was also found that in the region close to the point (location) where the discriminant score is -0.5, the discriminant scores of the case group (female all-cancer patient group) and the control group (female healthy female group) are mixed together.

[0188] Figure 37(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 3. From this figure, in the case group (all female cancer patients), 161 out of 166 samples were correctly identified (sensitivity = 161 / 166 = 96.99%), and in the control group (healthy females), 84 out of 90 samples were correctly identified (specificity = 84 / 90 = 93.33%). Therefore, 245 out of 256 samples targeted for discriminant analysis were correctly identified, and the accuracy was 96.70% (245 / 256). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 3 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of cancer in women can be expected.

[0189] Figure 38 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 3 and the probability that the subject has some kind of cancer (discrimination score - overall cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject does not have cancer."

[0190] Figure 39(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 3.

[0191] Using the ROC curve in Figure 39(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9881, as shown in Figure 39(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9881 obtained in Example 3 indicates that the discrimination (prediction) between the two groups in Example 3 was very accurate. Therefore, it was found that the method of Example 3 is sufficiently effective as a tool for diagnosing the risk of cancer in women.

[0192] Next, Example 4 will be described. This example is the same as Example 3 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here as well, the site (type) of cancer in the subjects was not considered, and the risk of female subjects having some kind of cancer was evaluated.

[0193] In other words, in this Example 4, similar to Example 3 described above, the risk of cancer in women was evaluated using a combination of two groups: a group of healthy women (90 cases) and a group of women with all types of cancer (166 cases). However, unlike Example 3 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the group of women with all types of cancer (case group) and the group of healthy women (control group).

[0194] Figure 8 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0195] Figure 8 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 4: a group of healthy women and a group of women with all types of cancer. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, and Ba—17 elements in total. The large difference in mean concentration of these elements suggests that they may be influencing the development of cancer in women, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0196] Figure 18 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 4. From Figure 18, it can be seen that, with urinary creatinine correction, the seven elements that are significant in distinguishing between the two groups, healthy women and women with all types of cancer, are B (p < 0.01), P (p < 0.001), S (p < 0.001), K (p < 0.001), Mo (p < 0.05), Cd (p < 0.01), and Cs (p < 0.05). This differs from Example 3, which did not use urinary creatinine correction. Therefore, with urinary creatinine correction, if an unknown subject (a person who wishes to have their cancer risk assessed) is female, and it is found that the seven elements B, P, S, K, Mo, Cd, and Cs are significant in the discrimination, it becomes possible to infer that the unknown female subject has some kind of cancer. Comparing this Example 4 with Example 3 described above, it can be seen that in the case of all cancers in women, the elements that are significant in the discrimination change depending on whether or not urinary creatinine correction is performed.

[0197] Figure 13(b) shows the discriminant formula obtained from the two-group discriminant analysis in Example 4. Similar to Example 3, this discriminant formula also uses the group of healthy women and the group of women with all cancers as the dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy women group and the group of women with all cancers as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the elements with discriminant coefficients that have a large absolute value have a greater influence on the discriminant score (D).

[0198] The results of the discriminant analysis of the two groups in Example 4 are shown in Figures 40 to 42.

[0199] Figure 40(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 4. From this figure, it can be seen that the majority of the discriminant scores of the healthy female group are distributed in the left region of the figure, and the majority of the discriminant scores of the female all-cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is -0.5. It was also found that in the region close to the point (location) where the discriminant score is -0.5, the discriminant scores of the case group (female all-cancer patient group) and the control group (female healthy female group) are mixed.

[0200] Figure 40(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 4. From this figure, in the case group (all female cancer patients), 159 out of 166 samples were correctly identified (sensitivity = 159 / 166 = 95.78%), and in the control group (healthy females), 85 out of 90 samples were correctly identified (specificity = 85 / 90 = 94.44%). Therefore, 244 out of 256 samples targeted for discriminant analysis were correctly identified, and the accuracy was 95.31% (244 / 256). Since this accuracy is sufficiently high, it can be concluded that the discriminant analysis in the two-group analysis in Example 4 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of cancer in women can be expected.

[0201] Figure 41 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 4 and the probability that the subject has some kind of cancer (discrimination score - overall cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a 90% or higher probability that the subject has cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject does not have cancer."

[0202] Figure 42(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 4.

[0203] Using the ROC curve in Figure 42(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9886, as shown in Figure 42(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9886 obtained in Example 4 indicates that the distinction (prediction) between the two groups in Example 4 was performed very accurately. Therefore, it was found that the method of Example 4 is sufficiently effective as a tool for diagnosing the risk of cancer in women.

[0204] Next, Example 5 will be described. In this example, the risk of developing cancer was evaluated for two groups: a healthy control group (168 cases) and a gastric cancer patient group (118 cases) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "gastric cancer," and the risk of the subjects developing gastric cancer was evaluated.

[0205] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 286 urine samples provided by 168 healthy individuals and 118 gastric cancer patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data points (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0206] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 5: a healthy control group and a gastric cancer patient group. The elements that showed a large difference in mean concentration (mean concentration difference) between these two groups were P, S, K, Co, Cu, Zn, Se, Rb, Cd, Cs, and Tl—11 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of gastric cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0207] Figure 19 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained in the discriminant analysis of the two groups in Example 5. From Figure 19, it can be seen that, without correction for urinary creatinine, the elements that are significant in distinguishing between the healthy control group and the gastric cancer patient group are the following 10 elements: B (p < 0.01), P (p < 0.001), Ca (p < 0.001), Cu (p < 0.01), Zn (p < 0.05), Rb (p < 0.001), Sr (p < 0.01), Mo (p < 0.05), Cd (p < 0.05), and Tl (p < 0.05). Therefore, if it is determined that the ten elements B, P, Ca, Cu, Zn, Rb, Sr, Mo, Cd, and Tl are significant in distinguishing an unknown subject (a person who has requested cancer risk assessment), it becomes possible to infer that the subject has gastric cancer.

[0208] Figure 11(c) shows the discriminant formula obtained from the two-group discriminant analysis in Example 5. In this discriminant formula, the healthy control group and the gastric cancer patient group are used as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the gastric cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0209] The results of the discriminant analysis of the two groups in Example 5 are shown in Figures 43 to 45.

[0210] Figure 43(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 5. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the gastric cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +0.5. It was also found that in the region close to the point (location) where the discriminant score is +0.5, the discriminant scores of the case group (gastric cancer patient group) and the control group (healthy control group) are mixed.

[0211] Figure 43(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 5. From this figure, in the case group (gastric cancer patient group), 114 out of 118 samples were correctly identified (sensitivity = 114 / 118 = 96.61%), and in the control group (healthy individuals group), 155 out of 168 samples were correctly identified (specificity = 155 / 168 = 92.26%). Therefore, 269 out of 286 samples targeted for discriminant analysis were correctly identified, and the accuracy was 94.06% (269 / 286). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 5 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of gastric cancer can be expected.

[0212] Figure 44 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 5 and the probability that the subject has gastric cancer (discrimination score - gastric cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has gastric cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have gastric cancer."

[0213] Figure 45(a) shows the ROC curve obtained from the discriminant analysis of the two groups in Example 5.

[0214] Using the ROC curve in Figure 45(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9828, as shown in Figure 45(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9828 obtained in Example 5 indicates that the discrimination (prediction) between the two groups in Example 5 was very accurate. Therefore, it was found that the method of Example 5 is sufficiently effective as a tool for diagnosing the risk of gastric cancer.

[0215] Next, Example 6 will be described. This example is the same as Example 5 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here as well, the gender of the subjects was not considered, and the site (type) of cancer was limited to "gastric cancer," and the risk of the subjects having gastric cancer was evaluated.

[0216] In other words, in this Example 6, similar to Example 5 described above, the risk of developing gastric cancer was evaluated using a combination of two groups: a healthy control group (168 cases) and a gastric cancer patient group (118 cases). However, unlike Example 5 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the gastric cancer patient group (case group) and the healthy control group (control group).

[0217] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0218] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 6: a healthy control group and a gastric cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Ba, and Tl—19 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of gastric cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0219] Figure 20 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained in the discriminant analysis of the two groups in Example 6. From Figure 20, it can be seen that, with urinary creatinine correction, the eight elements that are significant in distinguishing between the healthy control group and the gastric cancer patient group are B (p < 0.01), P (p < 0.001), Ca (p < 0.001), Rb (p < 0.01), Sr (p < 0.001), Mo (p < 0.01), Cd (p < 0.01), and Tl (p < 0.05). This differs from Example 5, which was performed without urinary creatinine correction. Therefore, if it is found that the eight elements B, P, Ca, Rb, Sr, Mo, Cd, and Tl are significant in distinguishing an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the subject has gastric cancer. Comparing this Example 6 with Example 5 described above, it can be seen that in the case of gastric cancer, the elements that are significant in the discrimination change depending on whether or not urinary creatinine correction is performed.

[0220] Figure 13(c) shows the discriminant formula obtained from the two-group discriminant analysis in Example 6. In this discriminant formula as well, the healthy control group and the gastric cancer patient group are used as dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the gastric cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the elements with discriminant coefficients that have a large absolute value have a greater influence on the discriminant score (D).

[0221] The results of the discriminant analysis of the two groups in Example 6 are shown in Figures 46 to 48.

[0222] Figure 46(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 6. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the gastric cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +0.5. It was also found that in the region close to the point (location) where the discriminant score is +0.5, the discriminant scores of the case group (gastric cancer patient group) and the control group (healthy control group) are mixed.

[0223] Figure 46(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 6. From this figure, in the case group (gastric cancer patient group), 115 out of 118 samples were correctly identified (sensitivity = 115 / 118 = 97.46%), and in the control group (healthy individuals group), 157 out of 168 samples were correctly identified (specificity = 157 / 168 = 93.45%). Therefore, 272 out of 286 samples targeted for discriminant analysis were correctly identified, and the accuracy was 95.10% (272 / 286). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 6 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of gastric cancer can be expected.

[0224] Figure 47 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 6 and the probability that the subject has gastric cancer (discrimination score - gastric cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has gastric cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have gastric cancer."

[0225] Figure 48(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 6.

[0226] Using the ROC curve in Figure 48(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9835, as shown in Figure 48(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9835 obtained in Example 6 indicates that the distinction (prediction) between the two groups in Example 6 was performed very accurately. Therefore, it was found that the method of Example 6 is sufficiently effective as a tool for diagnosing the risk of gastric cancer.

[0227] Next, Example 7 will be described. In this example, the risk of developing colorectal cancer was evaluated for two groups: a healthy control group (168 cases) and a colorectal cancer patient group (70 cases) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "colorectal cancer," and the risk of the subjects developing colorectal cancer was evaluated.

[0228] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 238 urine samples provided by 168 healthy individuals and 70 colorectal cancer patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data points (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0229] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 7: a healthy control group and a colorectal cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Na, P, K, Co, Cu, Zn, Rb, Cd, Cs, Ba, and Tl—11 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of colorectal cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0230] Figure 21 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 7. From Figure 21, it can be seen that the seven elements that are significant in distinguishing between the healthy control group and the colorectal cancer patient group are Na (p < 0.01), P (p < 0.001), K (p < 0.01), Cu (p < 0.001), Mo (p < 0.01), Ba (p < 0.05), and Tl (p < 0.05). Therefore, if it is found that the seven elements that are significant in distinguishing between an unknown subject (a person who wishes to have their cancer risk assessed) are Na, P, K, Cu, Mo, Ba, and Tl, it becomes possible to infer that the unknown subject has colorectal cancer.

[0231] Figure 11(d) shows the discriminant formula obtained from the two-group discriminant analysis in Example 7. In this discriminant formula, the healthy control group and the colorectal cancer patient group are used as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the colorectal cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0232] The results of the discriminant analysis of the two groups in Example 7 are shown in Figures 49 to 51.

[0233] Figure 49(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 7. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the colorectal cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +0.5. It was also found that in the region close to the point (location) where the discriminant score is +0.5, the discriminant scores of the case group (colorectal cancer patient group) and the control group (healthy control group) are mixed.

[0234] Figure 49(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 7. From this figure, in the case group (colorectal cancer patient group), 64 out of 70 samples were correctly identified (sensitivity = 64 / 70 = 91.43%), and in the control group (healthy individuals group), 158 out of 168 samples were correctly identified (specificity = 158 / 168 = 94.05%). Therefore, 222 out of 238 samples included in the discriminant analysis were correctly identified, resulting in an accuracy of 93.28% (222 / 238). Since this accuracy is sufficiently high, it can be concluded that the discriminant analysis in the two-group analysis in Example 7 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of colorectal cancer can be expected.

[0235] Figure 50 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 7 and the probability that the subject has colorectal cancer (discrimination score - colorectal cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing colorectal cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have colorectal cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing colorectal cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have colorectal cancer."

[0236] Figure 51(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 7.

[0237] Using the ROC curve in Figure 51(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9845, as shown in Figure 51(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9845 obtained in Example 7 indicates that the distinction (prediction) between the two groups in Example 7 was performed very accurately. Therefore, it was found that the method of Example 7 is sufficiently effective as a tool for diagnosing the risk of colorectal cancer.

[0238] Next, Example 8 will be described. This example is the same as Example 7 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here again, the gender of the subjects was not considered, and the site (type) of cancer was limited to "colorectal cancer," and the risk of the subjects having colorectal cancer was evaluated.

[0239] In other words, in this Example 8, similar to Example 7 described above, the risk of developing colorectal cancer was evaluated using a combination of two groups: a healthy control group (168 cases) and a colorectal cancer patient group (70 cases). However, unlike Example 7 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the colorectal cancer patient group (case group) and the healthy control group (control group).

[0240] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0241] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 8: a healthy control group and a colorectal cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, and Cd—17 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of colorectal cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0242] Figure 22 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 8. From Figure 22, it can be seen that the seven elements that are significant in discriminating between the healthy control group and the colorectal cancer patient group are Na (p < 0.01), P (p < 0.001), K (p < 0.01), Cu (p < 0.01), Mo (p < 0.01), Ba (p < 0.05), and Tl (p < 0.05). This is the same as in Example 7 without correction for urinary creatinine. Therefore, if it is found that the seven elements Na, P, K, Cu, Mo, Ba, and Tl are significant in discriminating against an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the unknown subject has colorectal cancer. Comparing this Example 8 with Example 7 described above, it can be seen that in the case of colorectal cancer, the elements that are significant in the discrimination process do not change depending on whether or not urinary creatinine correction is performed.

[0243] Figure 13(d) shows the discriminant formula obtained from the two-group discriminant analysis in Example 8. Similar to Example 7, this discriminant formula also uses the healthy control group and the colorectal cancer patient group as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the colorectal cancer patient group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0244] The results of the discriminant analysis of the two groups in Example 8 are shown in Figures 52 to 54.

[0245] Figure 52(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 8. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the colorectal cancer patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided into left and right sides near the point (location) where the discriminant score is +0.5. It was also found that in the region close to the point (location) where the discriminant score is +0.5, the discriminant scores of the case group (colorectal cancer patient group) and the control group (healthy control group) are mixed.

[0246] Figure 52(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 8. From this figure, in the case group (colorectal cancer patient group), 63 out of 70 samples were correctly identified (sensitivity = 63 / 70 = 90.00%), and in the control group (healthy individuals group), 159 out of 168 samples were correctly identified (specificity = 159 / 168 = 94.64%). Therefore, 222 out of 238 samples included in the discriminant analysis were correctly identified, resulting in an accuracy of 93.28% (222 / 238). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 8 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of colorectal cancer can be expected.

[0247] Figure 53 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 8 and the probability that the subject has colorectal cancer (discrimination score - colorectal cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have colorectal cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing colorectal cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have colorectal cancer."

[0248] Figure 54(a) shows the ROC curve obtained from the discriminant analysis of the two groups in Example 8.

[0249] Using the ROC curve in Figure 54(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9845, as shown in Figure 54(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9845 obtained in Example 8 indicates that the discrimination (prediction) between the two groups in Example 8 was very accurate. Therefore, it was found that the method of Example 8 is sufficiently effective as a tool for diagnosing the risk of colorectal cancer.

[0250] Next, Example 9 will be described. In this example, the risk of developing pancreatic cancer in the subjects was evaluated using a combination of two groups: a healthy control group (168 subjects) and a pancreatic cancer patient group (47 subjects) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "pancreatic cancer," and the risk of the subjects developing pancreatic cancer was evaluated.

[0251] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 215 urine samples provided by 168 healthy individuals and 47 pancreatic cancer patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data points (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0252] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 9: a healthy control group and a pancreatic cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Na, S, K, Co, Cu, Zn, Se, Rb, Sr, Cd, Cs, and Tl—12 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of pancreatic cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0253] Figure 23 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 9. From Figure 23, it can be seen that the six elements that are significant in distinguishing between the healthy control group and the colorectal cancer patient group are Na (p < 0.01), P (p < 0.001), K (p < 0.01), Ca (p < 0.05), Cu (p < 0.01), and Mo (p < 0.05). Therefore, if it is found that the six elements Na, P, K, Ca, Cu, and Mo are significant in distinguishing an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the unknown subject has pancreatic cancer.

[0254] Figure 12(a) shows the discriminant formula obtained from the two-group discriminant analysis in Example 9. In this discriminant formula, the healthy control group and the pancreatic cancer patient group are used as the dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the pancreatic cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0255] The results of the discriminant analysis of the two groups in Example 9 are shown in Figures 55 to 57.

[0256] Figure 55(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 9. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the pancreatic cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +1.5. It was also found that in the region close to the point (location) where the discriminant score is +1.5, the discriminant scores of the case group (pancreatic cancer patient group) and the control group (healthy control group) are mixed.

[0257] Figure 55(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 9. From this figure, in the case group (pancreatic cancer patient group), 46 out of 47 samples were correctly identified (sensitivity = 46 / 47 = 97.87%), and in the control group (healthy individuals group), 161 out of 168 samples were correctly identified (specificity = 161 / 168 = 95.83%). Therefore, 207 out of 215 samples targeted for discriminant analysis were correctly identified, and the accuracy was 96.28% (207 / 215). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 9 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of pancreatic cancer can be expected.

[0258] Figure 56 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 9 and the probability that the subject has pancreatic cancer (discrimination score - pancreatic cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing pancreatic cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have pancreatic cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing the disease is 10% or less, so it can be evaluated that "the subject is presumed not to have pancreatic cancer."

[0259] Figure 57(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 9.

[0260] Using the ROC curve in Figure 57(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9927, as shown in Figure 57(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9927 obtained in Example 9 indicates that the distinction (prediction) between the two groups in Example 9 was performed very accurately. Therefore, it was found that the method of Example 9 is sufficiently effective as a tool for risk diagnosis of pancreatic cancer.

[0261] Next, Example 10 will be described. This example is the same as Example 9 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here again, the gender of the subjects was not considered, and the site (type) of cancer was limited to "pancreatic cancer," and the risk of the subjects having pancreatic cancer was evaluated.

[0262] In other words, in this Example 10, similar to Example 9 described above, the risk of developing pancreatic cancer was evaluated using a combination of two groups: a healthy control group (168 cases) and a pancreatic cancer patient group (47 cases). However, unlike Example 9 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the pancreatic cancer patient group (case group) and the healthy control group (control group).

[0263] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0264] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 10: a healthy control group and a pancreatic cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, and Ba—17 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of pancreatic cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0265] Figure 24 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 10. From Figure 24, it can be seen that the six elements that are significant in distinguishing between the healthy control group and the pancreatic cancer patient group are Na (p < 0.01), P (p < 0.001), K (p < 0.01), Ca (p < 0.05), Cu (p < 0.001), and Mo (p < 0.05). This is the same as in Example 9 above, without correction for urinary creatinine. Therefore, if it is found that the six elements that are significant in distinguishing between an unknown subject (a person who wishes to have their cancer risk assessed) are Na, P, K, Ca, Cu, and Mo, it can be inferred that the unknown subject has pancreatic cancer. Comparing Example 10 with Example 9 above, it can be seen that in the case of pancreatic cancer, the elements that are significant in distinguishing between the two groups do not change depending on whether or not correction for urinary creatinine is performed.

[0266] Figure 14(a) shows the discriminant formula obtained from the two-group discriminant analysis in Example 10. Similar to Example 9, this discriminant formula also uses the healthy control group and the pancreatic cancer patient group as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the pancreatic cancer patient group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0267] The results of the discriminant analysis of the two groups in Example 10 are shown in Figures 58 to 60.

[0268] Figure 58(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 10. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the pancreatic cancer patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided into left and right sides near the point (location) where the discriminant score is +2.0. It was also found that in the region close to the point (location) where the discriminant score is +2.0, the discriminant scores of the case group (pancreatic cancer patient group) and the control group (healthy control group) are mixed together.

[0269] Figure 58(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 10. From this figure, in the case group (pancreatic cancer patient group), 46 out of 47 samples were correctly identified (sensitivity = 46 / 47 = 97.87%), and in the control group (healthy individuals group), 159 out of 168 samples were correctly identified (specificity = 159 / 168 = 94.64%). Therefore, 205 out of 215 samples targeted for discriminant analysis were correctly identified, and the accuracy was 95.35% (205 / 215). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 10 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of pancreatic cancer can be expected.

[0270] Figure 59 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 10 and the probability that the subject has pancreatic cancer (discrimination score - pancreatic cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have pancreatic cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have pancreatic cancer."

[0271] Figure 60(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 10.

[0272] Using the ROC curve in Figure 60(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9924, as shown in Figure 60(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9924 obtained in Example 10 indicates that the discrimination (prediction) between the two groups in Example 10 was very accurate. Therefore, it was found that the method of Example 10 is sufficiently effective as a tool for diagnosing the risk of pancreatic cancer.

[0273] Next, Example 11 will be described. In this example, the risk of developing esophageal cancer was evaluated in two groups: a healthy control group (168 cases) and an esophageal cancer patient group (41 cases) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "esophageal cancer" to evaluate the risk of the subjects developing esophageal cancer.

[0274] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 20 urine samples provided by 168 healthy individuals and 41 esophageal cancer patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data points (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0275] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 11: a healthy control group and an esophageal cancer patient group. The elements that showed a large difference in mean concentration (mean concentration difference) between these two groups were B, S, K, Co, Cu, Zn, Se, Rb, Cd, Cs, and Tl—11 elements in total. The large difference in the mean concentration of these elements suggests that they may have an influence on the development of esophageal cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0276] Figure 25 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 11. From Figure 25, it can be seen that the seven elements that are significant in distinguishing between the healthy control group and the colorectal cancer patient group are Li (p < 0.05), B (p < 0.001), P (p < 0.01), Cu (p < 0.001), As (p < 0.05), Mo (p < 0.05), and Cd (p < 0.001). Therefore, if it is found that the seven elements Li, B, P, Cu, As, Mo, and Cd are significant in distinguishing an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the unknown subject has esophageal cancer.

[0277] Figure 12(b) shows the discriminant formula obtained from the two-group discriminant analysis in Example 11. In this discriminant formula, the healthy control group and the esophageal cancer patient group are used as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the esophageal cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0278] The results of the discriminant analysis of the two groups in Example 11 are shown in Figures 61 to 63.

[0279] Figure 61(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 11. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the esophageal cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +1.5. It was also found that in the region close to the point (location) where the discriminant score is +1.5, the discriminant scores of the case group (esophageal cancer patient group) and the control group (healthy control group) are mixed.

[0280] Figure 61(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 11. From this figure, in the case group (esophageal cancer patient group), 39 out of 41 samples were correctly identified (sensitivity = 39 / 41 = 95.12%), and in the control group (healthy individuals group), 161 out of 168 samples were correctly identified (specificity = 161 / 168 = 95.83). Therefore, 200 out of 209 samples targeted for discriminant analysis were correctly identified, and the accuracy was 95.69% (200 / 209). Since this accuracy is a sufficiently high value, it can be concluded that the discrimination in the two-group discriminant analysis in Example 11 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of esophageal cancer can be expected.

[0281] Figure 62 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 11 and the probability that the subject has esophageal cancer (discrimination score - esophageal cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing esophageal cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have esophageal cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing esophageal cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have esophageal cancer."

[0282] Figure 63(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 11.

[0283] Using the ROC curve in Figure 63(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9922, as shown in Figure 63(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9922 obtained in Example 11 indicates that the discrimination (prediction) between the two groups in Example 11 was very accurate. Therefore, it was found that the method of Example 11 is sufficiently effective as a tool for esophageal cancer risk diagnosis.

[0284] Next, Example 12 will be described. This example is the same as Example 11 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here again, the gender of the subjects was not considered, and the site (type) of cancer was limited to "esophageal cancer," and the risk of the subjects having esophageal cancer was evaluated.

[0285] In other words, in this Example 12, similar to Example 11 described above, the risk of developing esophageal cancer was evaluated using a combination of two groups: a healthy control group (168 cases) and an esophageal cancer patient group (41 cases). However, unlike Example 11 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the esophageal cancer patient group (case group) and the healthy control group (control group).

[0286] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0287] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 12: a healthy control group and an esophageal cancer patient group. The elements that showed a large difference in mean concentration between these two groups were Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, and Ba—19 elements in total. The large difference in mean concentration of these elements suggests that they may have an influence on the development of esophageal cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0288] Figure 26 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 12. From Figure 26, it can be seen that the six elements that are significant in discriminating between the healthy control group and the esophageal cancer patient group are Li (p < 0.05), B (p < 0.001), P (p < 0.01), Cu (p < 0.01), Mo (p < 0.05), and Cd (p < 0.001). This differs from Example 11, which did not use urinary creatinine correction. Therefore, if it is found that the six elements Li, B, P, Cu, Mo, and Cd are significant in discriminating against an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the unknown subject has esophageal cancer. Comparing Example 12 with Example 11 described above, it can be seen that in the case of esophageal cancer, the elements that are significant in discriminating change depending on whether or not urinary creatinine correction is used.

[0289] Figure 14(b) shows the discriminant formula obtained from the two-group discriminant analysis in Example 12. Similar to Example 11, this discriminant formula also uses the healthy control group and the esophageal cancer patient group as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the esophageal cancer patient group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0290] The results of the discriminant analysis of the two groups in Example 12 are shown in Figures 64 to 66.

[0291] Figure 64(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 12. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the esophageal cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +1.5. It was also found that in the region close to the point (location) where the discriminant score is +1.5, the discriminant scores of the case group (esophageal cancer patient group) and the control group (healthy control group) are mixed.

[0292] Figure 64(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 12. From this figure, in the case group (esophageal cancer patient group), 38 out of 41 samples were correctly identified (sensitivity = 38 / 41 = 92.68%), and in the control group (healthy individuals group), 162 out of 168 samples were correctly identified (specificity = 162 / 168 = 96.43%). Therefore, 200 out of all 209 samples targeted for discriminant analysis were correctly identified, and the accuracy was 95.69% (200 / 209). Since this accuracy is sufficiently high, it can be concluded that the discriminant analysis in the two-group analysis in Example 12 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of esophageal cancer can be expected.

[0293] Figure 65 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 12 and the probability that the subject has esophageal cancer (discrimination score - esophageal cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have esophageal cancer with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have esophageal cancer."

[0294] Figure 66(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 12.

[0295] Using the ROC curve in Figure 66(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9911, as shown in Figure 66(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9911 obtained in Example 12 indicates that the distinction (prediction) between the two groups in Example 12 was performed very accurately. Therefore, it was found that the method of Example 12 is sufficiently effective as a tool for esophageal cancer risk diagnosis.

[0296] Next, Example 13 will be described. In this example, the risk of developing malignant lymphoma was evaluated in two groups: a healthy control group (168 cases) and a malignant lymphoma patient group (27 cases) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "malignant lymphoma," and the risk of the subjects developing malignant lymphoma was evaluated.

[0297] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 195 urine samples provided by 168 healthy individuals and 27 malignant lymphoma patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0298] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 13: a healthy control group and a malignant lymphoma patient group. The nine elements that showed a large difference in mean concentration between these two groups were P, K, Co, Cu, Zn, Rb, Cd, Cs, and Tl. The large difference in mean concentration of these elements suggests that they may be influencing the development of malignant lymphoma, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0299] Figure 27 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 13. From Figure 27, it can be seen that the four elements that are significant in discriminating between the healthy control group and the malignant lymphoma patient group are B (p < 0.01), P (p < 0.001), Cu (p < 0.001), and Mo (p < 0.05). Therefore, if it is found that the four elements B, P, Cu, and Mo are significant in discriminating against an unknown subject (a person who wishes to have their cancer risk assessed), it becomes possible to infer that the unknown subject has malignant lymphoma.

[0300] Figure 12(c) shows the discriminant formula obtained from the two-group discriminant analysis in Example 13. In this discriminant formula, the healthy control group and the malignant lymphoma patient group are used as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the malignant lymphoma patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0301] The results of the discriminant analysis of the two groups in Example 13 are shown in Figures 67 to 69.

[0302] Figure 67(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 13. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the malignant lymphoma patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided to the left and right near the point (location) where the discriminant score is +2.0. It was also found that in the region close to the point (location) where the discriminant score is +2.0, the discriminant scores of the case group (malignant lymphoma patient group) and the control group (healthy control group) are mixed together.

[0303] Figure 67(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 13. From this figure, in the case group (malignant lymphoma patient group), 26 out of 27 samples were correctly identified (sensitivity = 26 / 27 = 96.30%), and in the control group (healthy individuals group), 159 out of 168 samples were correctly identified (specificity = 159 / 168 = 94.64). Therefore, 185 out of 195 samples included in the discriminant analysis were correctly identified, and the accuracy was 94.87% (185 / 195). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 13 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of malignant lymphoma can be expected.

[0304] Figure 68 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 13 and the probability that the subject has malignant lymphoma (discrimination score - malignant lymphoma incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing malignant lymphoma. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have malignant lymphoma with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing malignant lymphoma is 10% or less, so it can be evaluated that "the subject is presumed not to have malignant lymphoma."

[0305] Figure 69(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 13.

[0306] Using the ROC curve in Figure 69(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9857, as shown in Figure 69(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9857 obtained in Example 13 indicates that the discrimination (prediction) between the two groups in Example 13 was very accurate. Therefore, it was found that the method of Example 13 is sufficiently effective as a tool for diagnosing the risk of malignant lymphoma.

[0307] Next, Example 14 will be described. This example is the same as Example 13 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here again, the gender of the subjects was not considered, and the site (type) of cancer was limited to "malignant lymphoma," and the risk of the subjects suffering from malignant lymphoma was evaluated.

[0308] In other words, in this Example 14, similar to Example 13 described above, the risk of developing malignant lymphoma was evaluated using a combination of two groups: a healthy control group (168 cases) and a malignant lymphoma patient group (27 cases). However, unlike Example 13 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the malignant lymphoma patient group (case group) and the healthy control group (control group).

[0309] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0310] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 14: a healthy control group and a malignant lymphoma patient group. The elements that showed a large difference in mean concentration (mean concentration difference) between these two groups were Li, B, Na, Mg, P, S, K, Ca, Cu, Zn, As, Se, Rb, Sr, Mo, and Cd—16 elements in total. The large difference in mean concentration of these elements suggests that they may be influencing the development of malignant lymphoma, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0311] Figure 28 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 14. From Figure 28, it can be seen that the elements that are significant in discriminating between the healthy control group and the malignant lymphoma patient group are B (p < 0.01), P (p < 0.001), Cu (p < 0.001), and Mo (p < 0.05). This is the same as in Example 13 above, without correction for urinary creatinine. Therefore, if it is found that the elements that are significant in discriminating against an unknown subject (a person who wishes to have their cancer risk assessed) are B, P, Cu, and Mo, it can be inferred that the unknown subject has malignant lymphoma. Comparing Example 14 with Example 13 described above, it can be seen that in the case of malignant lymphoma, the elements that are significant in discriminating do not change whether or not correction for urinary creatinine is performed.

[0312] Figure 14(c) shows the discriminant formula obtained from the two-group discriminant analysis in Example 14. Similar to Example 13, this discriminant formula also uses the healthy control group and the malignant lymphoma patient group as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the malignant lymphoma patient group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0313] The results of the discriminant analysis of the two groups in Example 14 are shown in Figures 70 to 72.

[0314] Figure 70(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 14. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the malignant lymphoma patient group are distributed in the right region of the figure. In other words, it can be seen that the data is divided to the left and right near the point (location) where the discriminant score is +2.0. It was also found that in the region close to the point (location) where the discriminant score is +2.0, the discriminant scores of the case group (malignant lymphoma patient group) and the control group (healthy control group) are mixed together.

[0315] Figure 70(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 14. From this figure, in the case group (malignant lymphoma patient group), 26 out of 27 samples were correctly identified (sensitivity = 26 / 27 = 96.30%), and in the control group (healthy individuals group), 158 out of 168 samples were correctly identified (specificity = 158 / 168 = 94.05%). Therefore, 184 out of 195 samples included in the discriminant analysis were correctly identified, and the accuracy was 94.36% (184 / 195). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 14 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of malignant lymphoma can be expected.

[0316] Figure 71 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 14 and the probability that the subject has malignant lymphoma (discrimination score - malignant lymphoma incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing malignant lymphoma. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "the subject is presumed to have malignant lymphoma with a probability of 90% or more." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing malignant lymphoma is 10% or less, so it can be evaluated that "the subject is presumed not to have malignant lymphoma."

[0317] Figure 72(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 14.

[0318] Using the ROC curve in Figure 72(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9852, as shown in Figure 72(b). Since an AUC value closer to 1 indicates the goodness of discriminative fit, the value of 0.9852 obtained in Example 14 indicates that the discrimination (prediction) between the two groups in Example 14 was very accurate. Therefore, it was found that the method of Example 14 is sufficiently effective as a tool for diagnosing the risk of malignant lymphoma.

[0319] Next, Example 15 will be described. In this example, the risk of developing lung cancer was evaluated for two groups: a healthy control group (168 cases) and a lung cancer patient group (25 cases) (without correction for urinary creatinine). Here, the gender of the subjects was not considered, and the site (type) of cancer was limited to "lung cancer," and the risk of the subjects developing lung cancer was evaluated.

[0320] First, the concentrations of the aforementioned evaluation elements (the 20 elements mentioned above) contained in each of the 193 urine samples provided by 168 healthy individuals and 25 lung cancer patients were measured in the same manner as in Example 1 described above, without correcting for urinary creatinine. The 20 concentration data (concentration values) obtained from this measurement were transformed using the natural logarithm, as shown in Figure 9.

[0321] Figure 9 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 15: a healthy control group and a lung cancer patient group. Ten elements showed a significant difference in mean concentration between these two groups: Na, Mg, P, K, Cu, Zn, Rb, Cd, Cs, and Ba. The large differences in the mean concentrations of these elements suggest that they may be influencing the development of lung cancer, and also suggest that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0322] Figure 29 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 15. From Figure 29, it can be seen that the eight elements that are significant in distinguishing between the healthy control group and the lung cancer patient group are Na (p < 0.01), Mg (p < 0.05), P (p < 0.05), K (p < 0.001), Cu (p < 0.01), Rb (p < 0.05), Ba (p < 0.05), and Tl (p < 0.01). Therefore, if it is found that the eight elements that are significant in distinguishing between an unknown subject (a person who wishes to have their cancer risk assessed) are Na, Mg, P, K, Cu, Rb, Ba, and Tl, it becomes possible to infer that the unknown subject has lung cancer.

[0323] Figure 12(d) shows the discriminant formula obtained from the two-group discriminant analysis in Example 15. In this discriminant formula, the healthy control group and the lung cancer patient group are used as dependent variables, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the lung cancer patient group are used as independent variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0324] The results of the discriminant analysis of the two groups in Example 15 are shown in Figures 73 to 75.

[0325] Figure 73(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 15. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the lung cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +2.0. It was also found that in the region close to the point (location) where the discriminant score is +2.0, the discriminant scores of the case group (lung cancer patient group) and the control group (healthy control group) are mixed.

[0326] Figure 73(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 15. From this figure, in the case group (lung cancer patient group), 21 out of 25 samples were correctly identified (sensitivity = 21 / 25 = 84.00%), and in the control group (healthy individuals group), 155 out of 168 samples were correctly identified (specificity = 155 / 168 = 92.26). Therefore, 176 out of 193 samples included in the discriminant analysis were correctly identified, and the accuracy was 91.19% (176 / 193). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 15 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of lung cancer can be expected.

[0327] Figure 74 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 15 and the probability that the subject has lung cancer (discrimination score - lung cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing lung cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has lung cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing lung cancer is 10% or less, so it can be evaluated that "the subject is presumed not to have lung cancer."

[0328] Figure 75(a) shows the ROC curve obtained from the discriminant analysis of the two groups in Example 15.

[0329] Using the ROC curve in Figure 75(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9600, as shown in Figure 75(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9600 obtained in Example 15 indicates that the distinction (prediction) between the two groups in Example 15 was performed very accurately. Therefore, it was found that the method of Example 15 is sufficiently effective as a tool for lung cancer risk diagnosis.

[0330] Finally, Example 16 will be described. This example is the same as Example 15 described above, except that urine creatinine correction was performed on the concentration data (concentration values) of the 20 elements before discriminant analysis. Here again, the gender of the subjects was not considered, and the site (type) of cancer was limited to "lung cancer," and the risk of the subjects having lung cancer was evaluated.

[0331] In other words, in this Example 16, similar to Example 15 described above, the risk of developing lung cancer was evaluated using a combination of two groups: a healthy control group (168 cases) and a lung cancer patient group (25 cases). However, unlike Example 15 described above, "urinary creatinine correction" was performed on the concentration data (concentration values) of the 20 elements. Subsequently, a natural logarithmic transformation was applied to the concentration data (concentration values) after urinary creatinine correction, and discriminant analysis was performed on the lung cancer patient group (case group) and the healthy control group (control group).

[0332] Figure 10 shows the 20 concentration data points (concentration values) obtained from this measurement, after being transformed using the natural logarithm.

[0333] Figure 10 shows the mean and standard deviation of the concentrations of the evaluation elements (the 20 elements) present in urine samples from two groups in Example 16: a healthy control group and a lung cancer patient group. The elements that showed a large difference in mean concentration (mean concentration difference) between these two groups were Li, Na, Mg, P, K, Cu, Zn, Rb, Cd, and Ba—10 elements in total. The large difference in the mean concentration of these elements suggests that they may have an influence on the development of lung cancer, and also suggests that there is some strong correlation between these elements and the factors that caused changes in the homeostasis (steady state) of trace element concentrations in urine samples.

[0334] Figure 30 shows the discriminant coefficients and their significance (p-values) for each of the evaluation element group (the 20 elements) obtained from the discriminant analysis of the two groups in Example 16. From Figure 30, it can be seen that the eight elements that are significant in distinguishing between the healthy control group and the lung cancer patient group are Na (p < 0.01), Mg (p < 0.05), P (p < 0.05), K (p < 0.01), Cu (p < 0.01), Rb (p < 0.05), Ba (p < 0.05), and Tl (p < 0.01). This is the same as in Example 15 without correction for urinary creatinine. Therefore, if it is found that the eight elements that are significant in distinguishing between an unknown subject (a person who wishes to have their cancer risk assessed) are Na, Mg, P, K, Cu, Rb, Ba, and Tl, it becomes possible to infer that the unknown subject has lung cancer. Comparing this Example 16 with the above-described Example 15, it can be seen that in the case of lung cancer, the elements that are significant in the discrimination do not change depending on whether or not urinary creatinine correction is performed.

[0335] Figure 14(d) shows the discriminant formula obtained from the two-group discriminant analysis in Example 16. Similar to Example 15, this discriminant formula also uses the healthy control group and the lung cancer patient group as the dependent variable, and the 20 concentration data of the evaluation element group (the 20 elements) obtained from the urine samples of the healthy control group and the lung cancer patient group as explanatory variables 1 to 20. Among the discriminant coefficients 1 to 20 of the 20 elements, the element with the largest absolute value of the discriminant coefficient has a greater influence on the discriminant score (D).

[0336] The results of the discriminant analysis of the two groups in Example 16 are shown in Figures 76 to 78.

[0337] Figure 76(a) is a histogram showing the frequency distribution of the discriminant scores of the two groups in Example 16. From this figure, it can be seen that the majority of the discriminant scores of the healthy control group are distributed in the left region of the figure, and the majority of the discriminant scores of the lung cancer patient group are distributed in the right region of the figure. In other words, the data is divided into left and right sides near the point (location) where the discriminant score is +2.0. It was also found that in the region close to the point (location) where the discriminant score is +2.0, the discriminant scores of the case group (lung cancer patient group) and the control group (healthy control group) are mixed.

[0338] Figure 76(b) is a table showing the sensitivity, specificity, and accuracy obtained from the two-group discriminant analysis in Example 16. From this figure, in the case group (lung cancer patient group), 21 out of 25 samples were correctly identified (sensitivity = 21 / 25 = 84.00%), and in the control group (healthy individuals group), 155 out of 168 samples were correctly identified (specificity = 155 / 168 = 92.26%). Therefore, 176 out of 193 samples included in the discriminant analysis were correctly identified, resulting in an accuracy of 91.19% (1764 / 193). Since this accuracy is sufficiently high, it can be concluded that the discrimination in the two-group discriminant analysis in Example 16 was performed appropriately, and therefore, a highly reliable risk assessment (prediction) of lung cancer can be expected.

[0339] Figure 77 is a graph showing the relationship between the discrimination score obtained in the two-group discriminant analysis in Example 16 and the probability that the subject has lung cancer (discrimination score - lung cancer incidence rate relationship). According to this graph, the larger the positive value of the discrimination score, the higher the probability (risk) of developing lung cancer. For example, if the discrimination score is approximately +2.2 or higher, it can be evaluated that "there is a probability of 90% or more that the subject has lung cancer." Conversely, for example, if the discrimination score is approximately -2.2 or lower, the probability of developing lung cancer is 10% or less, so it can be evaluated that "the subject does not have lung cancer."

[0340] Figure 78(a) shows the ROC curve obtained from the two-group discriminant analysis in Example 16.

[0341] Using the ROC curve in Figure 78(a), the area under the curve (AUC) was calculated, yielding a high value of 0.9567, as shown in Figure 78(b). Since an AUC value closer to 1 indicates the accuracy of the judgment, the value of 0.9567 obtained in Example 16 indicates that the distinction (prediction) between the two groups in Example 16 was performed very accurately. Therefore, it was found that the method of Example 16 is sufficiently effective as a tool for lung cancer risk diagnosis.

[0342] The present invention is widely applicable to fields where it is desirable to quickly and easily predict whether or not a person (or animal) has cancer and to assess the risk of developing it.

[0343] 1 Container 2 Urine sample 5 Urine element concentration measurement unit 10 Cancer risk assessment system 11 Data storage unit 12 Calculation unit 13 Evaluation result generation unit

Claims

The steps include obtaining concentration data of evaluation elements contained in urine samples collected from subjects, The steps include applying the concentration data of the evaluation element group to a discriminant function for determining whether the subject belongs to the control group or the case group, and calculating the correlation between the concentrations of the evaluation element group, Based on the aforementioned correlation, the step of generating an index for identifying whether or not the subject has cancer, The system includes the step of evaluating the cancer risk of the subject based on the aforementioned indicators and creating an evaluation result, A method for evaluating cancer risk, characterized in that the group of elements for evaluation is a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl. 　 As the aforementioned index, the discriminant score calculated by applying the concentration data to the discriminant formula generated based on the correlation is used. By comparing the aforementioned discrimination score with a predetermined reference value, it is determined whether the subject belongs to the control group or the case group. The cancer risk assessment method according to claim 1, wherein the assessment result includes an estimation of whether or not the subject has cancer, based on the result of the aforementioned determination. 　 By comparing the aforementioned discrimination score of the subject with the relationship between the discrimination score and the probability of developing cancer, the probability of developing cancer for the subject is estimated. The cancer risk assessment method according to claim 2, wherein the estimated probability of cancer incidence of the subject is included in the assessment result. 　 As the aforementioned discriminant formula, a formula generated to evaluate whether or not a male subject has some kind of cancer is used, without specifying the site of the cancer. The cancer risk assessment method according to claim 2 or 3, wherein the assessment result includes a determination of whether or not the male subject has any type of cancer. 　 As the aforementioned discriminant formula, a formula generated to evaluate whether or not the female subject has some kind of cancer is used, without specifying the site of the cancer. The cancer risk assessment method according to claim 2 or 3, wherein the assessment result includes a determination of whether or not the female subject has any type of cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has gastric cancer without specifying their gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the evaluation result includes a determination result of whether or not the subject has gastric cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has colorectal cancer without specifying the gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the evaluation result includes the determination result of whether or not the subject has colorectal cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has pancreatic cancer without specifying their gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the assessment result includes a determination of whether or not the subject has pancreatic cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has esophageal cancer without specifying their gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the evaluation result includes a determination result of whether or not the subject has esophageal cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has malignant lymphoma without specifying the gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the evaluation result includes a determination result of whether or not the subject is suffering from malignant lymphoma. 　 As the discriminant formula, a formula generated to evaluate whether or not the subject has lung cancer without specifying gender is used. The cancer risk assessment method according to claim 2 or 3, wherein the assessment result includes a determination of whether or not the subject has lung cancer. 　 The aforementioned discriminant formula is: Along with a first discriminant formula generated to assess whether the aforementioned male subject has any type of cancer without specifying the site of the cancer, A second discriminant formula was generated to evaluate whether the subject has gastric cancer, A third discriminant formula was generated to evaluate whether the subject has colorectal cancer, A fourth discriminant formula generated to evaluate whether or not the subject has pancreatic cancer, A fifth discriminant formula was generated to evaluate whether the subject has esophageal cancer, A sixth discriminant formula was generated to evaluate whether the subject has malignant lymphoma, A seventh discriminant formula was generated to evaluate whether the subject has lung cancer and At least one discriminant selected from the group consisting of is used. The cancer risk assessment method according to claim 2 or 3, wherein, if the first discriminant formula determines that the male subject has some kind of cancer, the evaluation result includes the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, according to one or more discriminant formulas selected from the group. 　 The aforementioned discriminant formula is: Along with a first discriminant formula generated to assess whether the aforementioned female subjects have any type of cancer without specifying the site of the cancer, A second discriminant formula was generated to evaluate whether the subject has gastric cancer, A third discriminant formula was generated to evaluate whether the subject has colorectal cancer, A fourth discriminant formula generated to evaluate whether or not the subject has pancreatic cancer, A fifth discriminant formula was generated to evaluate whether the subject has esophageal cancer, A sixth discriminant formula was generated to evaluate whether the subject has malignant lymphoma, A seventh discriminant formula was generated to evaluate whether the subject has lung cancer and At least one discriminant selected from the group consisting of is used. The cancer risk assessment method according to claim 2 or 3, wherein, if the first discriminant formula determines that the female subject has some kind of cancer, the evaluation result includes the determination that the subject has one or more cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on one or more discriminant formulas selected from the group. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation, four elements P, K, Ca, and Cu selected from the 20 elements used as the evaluation element group are significant in the determination, and if the subject is male, the assessment result includes the inference that the subject has some kind of cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation described above, (a) without correction for urinary creatinine in the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Co, Cu, Mo, Cd, Cs, and Tl—are significant in the discrimination, or (b) with correction for urinary creatinine in the urine sample, 7 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Mo, Cd, and Cs—are significant in the discrimination, and when the subject is female, the assessment result includes the inference that the subject has some kind of cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation described above, (a) without correction for urinary creatinine in the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Cu, Zn, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, or (b) with correction for urinary creatinine in the urine sample, 8 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, the assessment result includes the inference that the subject has gastric cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation, if seven elements selected from the 20 elements used as the evaluation element group—Na, P, K, Cu, Mo, Ba, and Tl—are significant in the determination, the assessment result includes the inference that the subject has colorectal cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation, if six elements selected from the 20 elements used as the evaluation element group—Na, P, K, Ca, Cu, and Mo—are significant in the determination, the assessment result includes the inference that the subject has pancreatic cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation described above, if (a) seven elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, As, Mo, and Cd—are significant in the determination without correction for urinary creatinine in the urine sample, or (b) six elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, Mo, and Cd—are significant in the determination with correction for urinary creatinine in the urine sample, the assessment result includes the inference that the subject has esophageal cancer. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation, if four elements B, P, Cu, and Mo selected from the 20 elements used as the evaluation element group are significant in the determination, the assessment result includes the inference that the subject has malignant lymphoma. 　 A cancer risk assessment method according to any one of claims 1 to 3, wherein, based on the correlation, if eight elements selected from the 20 elements used as the evaluation element group—Na, Mg, P, K, Cu, Rb, Ba, and Tl—are significant in the determination, the assessment result includes the inference that the subject has lung cancer. 　 A data storage unit that stores concentration data of evaluation elements contained in urine samples collected from subjects, A calculation unit that applies the concentration data of the subject stored in the data storage unit to a discriminant function for determining whether the subject belongs to the control group or the case group, and calculates the correlation between the concentrations of the evaluation element group, The system includes an evaluation result generation unit that generates an index for identifying whether or not the subject has cancer based on the correlation calculated by the calculation unit, and an evaluation result generation unit that evaluates the subject's risk of developing cancer based on the index and generates an evaluation result. A cancer risk assessment system characterized in that the evaluation element group is a combination of 20 elements: Li, B, Na, Mg, P, S, K, Ca, Co, Cu, Zn, As, Se, Rb, Sr, Mo, Cd, Cs, Ba, and Tl. 　 As the aforementioned index, the discriminant score calculated by applying the concentration data to the discriminant formula generated based on the correlation is used. By comparing the aforementioned discrimination score with a predetermined reference value, it is determined whether the subject belongs to the control group or the case group. The cancer risk assessment system according to claim 22, wherein the assessment result includes an inference as to whether or not the subject has cancer, based on the result of the aforementioned determination. 　 By comparing the aforementioned discrimination score of the subject with the relationship between the discrimination score and the probability that the subject has cancer, the probability of the subject developing cancer is estimated. The cancer risk assessment system according to claim 23, wherein the estimated probability of the subject developing cancer is included in the assessment result. 　 As the aforementioned discriminant formula, a formula generated to evaluate whether or not a male subject has some kind of cancer is used, without specifying the site of the cancer. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the male subject has any type of cancer. 　 As the aforementioned discriminant formula, a formula generated to evaluate whether or not the female subject has some kind of cancer is used, without specifying the site of the cancer. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the female subject has any type of cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has gastric cancer without specifying their gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has gastric cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has colorectal cancer without specifying the gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has colorectal cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has pancreatic cancer without specifying their gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has pancreatic cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has esophageal cancer without specifying their gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has esophageal cancer. 　 As the aforementioned discriminant formula, a discriminant formula generated to evaluate whether or not the subject has malignant lymphoma without specifying the gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has malignant lymphoma. 　 As the discriminant formula, a formula generated to evaluate whether or not the subject has lung cancer without specifying gender is used. The cancer risk assessment system according to claim 23 or 24, wherein the evaluation results include a determination of whether or not the subject has lung cancer. 　 The aforementioned discriminant formula is: Along with a first discriminant formula generated to assess whether the aforementioned male subject has any type of cancer without specifying the site of the cancer, A second discriminant formula was generated to evaluate whether the subject has gastric cancer, A third discriminant formula was generated to evaluate whether the subject has colorectal cancer, A fourth discriminant formula generated to evaluate whether or not the subject has pancreatic cancer, A fifth discriminant formula was generated to evaluate whether the subject has esophageal cancer, A sixth discriminant formula was generated to evaluate whether the subject has malignant lymphoma, A seventh discriminant formula was generated to evaluate whether the subject has lung cancer and At least one discriminant selected from the group consisting of is used. The cancer risk assessment system according to claim 23 or 24, wherein, if the first discriminant formula determines that the male subject has some kind of cancer, the assessment result includes a determination that the subject has at least one of the following cancers selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on at least one discriminant formula selected from the group. 　 The aforementioned discriminant formula is: Along with a first discriminant formula generated to assess whether the aforementioned female subjects have any type of cancer without specifying the site of the cancer, A second discriminant formula was generated to evaluate whether the subject has gastric cancer, A third discriminant formula was generated to evaluate whether the subject has colorectal cancer, A fourth discriminant formula generated to evaluate whether or not the subject has pancreatic cancer, A fifth discriminant formula was generated to evaluate whether the subject has esophageal cancer, A sixth discriminant formula was generated to evaluate whether the subject has malignant lymphoma, A seventh discriminant formula was generated to evaluate whether the subject has lung cancer and At least one discriminant selected from the group consisting of is used. The cancer risk assessment system according to claim 23 or 24, wherein, if the first discriminant formula determines that the subject is suffering from some kind of cancer, the assessment result includes a determination that the subject is suffering from at least one of the following cancers, selected from gastric cancer, colorectal cancer, pancreatic cancer, esophageal cancer, malignant lymphoma, and lung cancer, based on at least one discriminant formula selected from the group. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, four elements P, K, Ca, and Cu selected from the 20 elements used as the evaluation element group are significant in the determination, and if the subject is male, the assessment result includes the inference that the subject has some kind of cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, (a) without correction for urinary creatinine in the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Co, Cu, Mo, Cd, Cs, and Tl—are significant in the discrimination, or (b) with correction for urinary creatinine in the urine sample, 7 elements selected from the 20 elements used as the evaluation element group—B, P, S, K, Mo, Cd, and Cs—are significant in the discrimination, and when the subject is female, the evaluation result includes the inference that the subject has some kind of cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation described above, if (a) without correction for urinary creatinine in the urine sample, 10 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Cu, Zn, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, or (b) with correction for urinary creatinine in the urine sample, 8 elements selected from the 20 elements used as the evaluation element group—B, P, Ca, Rb, Sr, Mo, Cd, and Tl—are significant in the determination, the assessment result includes the inference that the subject has gastric cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, if seven elements selected from the 20 elements used as the evaluation element group—Na, P, K, Cu, Mo, Ba, and Tl—are significant in the determination, the assessment result includes the inference that the subject has colorectal cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, if six elements selected from the 20 elements used as the evaluation element group—Na, P, K, Ca, Cu, and Mo—are significant in the determination, the assessment result includes the inference that the subject has pancreatic cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation described above, if (a) seven elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, As, Mo, and Cd—are significant in the determination without correction for urinary creatinine in the urine sample, or (b) six elements selected from the 20 elements used as the evaluation element group—Li, B, P, Cu, Mo, and Cd—are significant in the determination with correction for urinary creatinine in the urine sample, the assessment result includes the inference that the subject has esophageal cancer. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, if four elements B, P, Cu, and Mo selected from the 20 elements used as the evaluation element group are significant in the determination, the assessment result includes the inference that the subject has malignant lymphoma. 　 A cancer risk assessment system according to any one of claims 22 to 24, wherein, based on the correlation, if eight elements selected from the 20 elements used as the evaluation element group—Na, Mg, P, K, Cu, Rb, Ba, and Tl—are significant in the determination, the assessment result includes the inference that the subject has lung cancer.

Images