Method and system for supporting determination of optimal treatment for liver tumors

Abstract

To provide a method and system for supporting the determination of eligibility of conversion surgery (CS), which is radical resection after aggressive preoperative chemotherapy, for colorectal liver metastases, thereby solving the problem in which involvement of multiple factors make it difficult to determine whether the CS is feasible.SOLUTION: A method for supporting the determination of CS eligibility is provided by, at the initial diagnosis, examining and scoring the presence / absence of RAS mutations, the presence / absence of conservable segments, the number of tumors, and tumor contact with hepatic veins.SELECTED DRAWING: Figure 3A

Description

[Technical field]

[0001] The present invention relates to a method and system for supporting the decision of optimal therapy for liver tumors, specifically, the indication of aggressive preoperative chemotherapy and conversion surgery. [Background technology]

[0002] Colorectal cancer is on the rise, with approximately 1.8 million new cases and 800,000 deaths reported annually worldwide. In Japan, the number of new cases of colorectal cancer exceeded 150,000 in 2017, and more than 50,000 people died from colorectal cancer in 2019. At the time of diagnosis of colorectal cancer, approximately 20% of cases have liver metastasis, and approximately 50-75% of metastatic and recurrent cases have liver metastasis. In advanced and recurrent stage IV colorectal cancer with resectable liver metastasis, resection including the primary lesion is recommended, and the survival rate of resected cases is significantly better than that of non-resected cases. In other words, colorectal liver metastases (CLM) are one of the few cancers for which long-term prognosis or complete cure can be expected by resection of metastatic lesions.

[0003] However, only about 20% of cases are suitable for resection at the time of diagnosis of liver metastasis. As a treatment strategy for CLM that is unresectable at the time of initial diagnosis, in addition to the traditional palliative systemic chemotherapy, conversion surgery (CS) has been performed, which enables the resection of liver metastases that were previously unresectable by the introduction of aggressive chemotherapy, and has attracted attention in recent years because it has a significantly better prognosis.

[0004] There are various cancer treatments, including surgery, chemotherapy (anticancer drugs), radiation therapy, and immunotherapy. Combining multiple treatments to achieve a higher therapeutic effect is called multidisciplinary therapy, and CS is one of them. CS refers to surgery performed with the aim of curative resection after a tumor is initially diagnosed as unresectable and undergoes a certain period of aggressive chemotherapy and other treatments, resulting in tumor shrinkage. In CLM, the 5-year and 10-year overall survival rates of patients who have undergone CS are 33% and 27%, respectively, and it is expected that CS will improve long-term survival rates.

[0005] Although there are an increasing number of reports on the effectiveness of CS, it cannot be said that CS is being performed sufficiently. The reason why CS is not being performed sufficiently is that it is difficult to determine whether CS will be possible in the future based on long-term prognosis. Determining whether CS will be suitable for future use requires multiple predictions, such as predicting the tumor shrinkage due to chemotherapy, and whether the tumor after shrinkage will be suitable for CS, making it difficult to judge. For example, although CS has shown good results in CLM, the judgment of its suitability is very complicated and requires a judgment based on the experience of an experienced liver surgeon, so there are cases where it is not applied even though it is suitable.

[0006] Since there are multiple effective cancer treatments, methods have been proposed for predicting a patient's response to a treatment and determining an appropriate treatment method (Patent Documents 1 to 3, Non-Patent Document 1). Patent Document 1 discloses a method for predicting the treatment response of cancer preoperative chemoradiotherapy using genes or proteins. Non-Patent Document 1 summarizes various reports on the prediction, diagnosis, treatment response prediction, recurrence prediction, and prognosis prediction of CLM using AI. There have also been several reports on prediction of treatment methods using machine learning. For example, Patent Document 2 discloses a method for predicting response to treatment, in which biological signatures such as DNA profiles before and after treatment of multiple subjects are obtained, and a computer is trained to learn the association with the outcome of treatment, and the treatment method for the patient is predicted. Patent Document 3 discloses a method for learning histological images using machine learning and using them for prognosis, diagnosis, and selection of treatment. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Special Publication No. 2023-505031 [Patent Document 2] Special Publication No. 2023-512698 [Patent Document 3] Special Publication No. 2022-549629 [Non-patent literature]

[0008] [Non-Patent Document 1] Rompianesi, G. etal., World J. Gastroenterology, 2022, 28(1): 108-122, DOI: 10.3748 / wjg.v28.i1.108 [Non-Patent Document 2] Rees, M. et al., Ann Surg. 2008, Vol.247,pp.125-135. [Non-Patent Document 3] Mise Y., et al. AnnSurg Oncol. 2020, Vol.27,pp.4188-4195. [Non-Patent Document 4] Kawakatsu S., et al.JSurg Oncol.2020,Vol.122, pp.523-528. Summary of the Invention [Problem to be solved by the invention]

[0009] The inventions described in Patent Documents 1 and 2 are methods for predicting a patient's response to treatment using biomarkers such as genes and proteins, while Patent Document 3 is a method for predicting a patient's response to treatment from pathological images. Therefore, although it may be possible to predict tumor shrinkage in response to a specific treatment, it cannot be used as an index for determining whether or not a patient is suitable for curative resection thereafter. In addition, Non-Patent Document 1 discloses various methods related to CLM, but does not disclose how to determine the suitability of CS.

[0010] The judgement of whether CS is suitable is based on the predicted tumor shrinkage, as well as on information obtained from diagnostic imaging such as CT and MRI, such as the size and number of tumors, their relationship to the tumor vasculature, and their depth from the liver surface in the case of liver metastasis mentioned above. In addition, liver reserve and the allowable volume of resection must also be taken into consideration when deciding suitability. Therefore, the suitability of CS is currently determined comprehensively for each case by an experienced liver surgeon, taking all of these factors into account and based on their experience. [Means for solving the problem]

[0011] The present invention relates to a method for supporting the decision of whether or not curative resection can be performed by aggressive preoperative chemotherapy, and in particular to a method and system for screening candidates for CS in CLM, which has a large number of target patients. (1) A method to assist in determining whether or not curative resection (conversion surgery, CS) is appropriate for liver tumors after aggressive preoperative chemotherapy, by examining the presence or absence of RAS mutations, the presence or absence of tumor-sparing areas, the number of tumors, and the contact of tumors with the hepatic vein at the initial consultation, and assisting in determining whether or not CS is appropriate. (2) A method for assisting in determining the applicability of CS according to (1), characterized in that the liver tumor is a metastatic liver tumor. (3) A method for assisting in determining the applicability of CS described in (2), characterized by scoring the presence or absence of RAS mutation, the presence or absence of a salvageable area, the number of tumors, and contact of the tumor with the hepatic vein. (4) A method for assisting in determining the applicability of CS described in any one of (1) to (3), in which a score is calculated by applying different weighting to the presence or absence of the RAS mutation and the contact of the tumor with the hepatic vein, and the presence or absence of a preservable area and the number of tumors. (5) A method for assisting in determining the applicability of CS described in (4), characterized in that the area that can be preserved is one or more of the left lateral area, the left medial area, the right anterior area, or the right posterior area. (6) An image analysis device that determines the presence or absence of tumor-sparing areas, the number of tumors, and the contact of tumors with the hepatic vein from CT and / or MRI images, and a system that determines the suitability of CS for colorectal cancer liver metastases based on the input information on the presence or absence of RAS gene mutations. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram showing the method for selecting the group of colorectal cancer liver metastasis patients used in the analysis. [Diagram 2] Graph showing overall survival (OS) in the CS and non-CS groups. [Figure 3A] Figure showing TUR (technically unresectable)-CS score and the rate of patients who converted from unresectable to resectable. [Figure 3B] Figure showing the association between TUR-CS score and overall survival. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Below, we will focus on CLM, which has a large number of cases, but even for liver tumors other than CLM, such as primary liver tumors and metastatic liver tumors from cancers other than colorectal cancer, if CS is applicable through chemotherapy, the possibility of curative resection can be determined using a method similar to the analysis shown below. We will explain this while presenting data.

[0014] It has been shown that prognostic factors for CLM include the number and size of liver metastases and primary lymph node metastasis, and a prognostic prediction model using these factors has been reported (Non-Patent Document 2). However, since this prognostic prediction model is used to predict the prognosis of all CLM, including resectable CLM, it is predicted that most cases of unresectable liver metastasis will be included in the poor prognosis group. Therefore, this prognostic prediction model is not considered suitable for predicting CLM for which CS can be applied. In order to predict patients who can be applied to CS from those who are judged to have a poor prognosis using existing prognostic prediction models and minimize palliative chemotherapy, we analyzed the treatment profile and long-term outcome of CLM patients.

[0015] [Patient selection used to create the model] We analyzed patient characteristics, short-term and long-term outcomes of patients with CLM who underwent multidisciplinary treatment at Cancer Institute Ariake Hospital from January 2014 to March 2021. The treatment policy was decided at a conference on multidisciplinary treatment of liver cancer. The resectability of CLM was examined by reviewing CT and MRI findings obtained at the first consultation. Unresectable CLM was defined as the presence of technically unresectable intrahepatic lesions or unresectable extrahepatic lesions. Specifically, technically unresectable CLM was defined as at least one of the following cases: (1) insufficient remnant liver volume after surgery (<30%), (2) invasion into the hepatic hilum (invasion into both hepatic arteries or both portal vein branches), and (3) invasion into the roots of the three hepatic veins. Based on our previous studies (Non-Patent Documents 3 and 4), resectable extrahepatic metastases are classified into the technically unresectable category, but resectable extrahepatic metastases were excluded from the analysis here.

[0016] [Development of a treatment plan] Technically unresectable CLM was initially treated with chemotherapy, and any changes in treatment were regularly reviewed at multidisciplinary conferences. CS was indicated if chemotherapy resulted in tumor regression and complete resection of the entire CLM (stage I or II, with or without portal vein embolization) with tumor-negative margins and sufficient remnant liver volume. Liver function parameters were also assessed preoperatively by indocyanine green test and 99mTc-galactosyl sialyalbumin (GSA) scintigraphy.

[0017] In cases of multiple CLM, primary hepatectomy, including liver parenchyma-preserving or percutaneous portal vein embolization followed by hepatectomy, was considered as the first-line treatment. In particular, liver parenchyma-preserving hepatectomy was performed as the standard treatment for CLM, even in patients with unresectable CLM. Secondary hepatectomy was indicated for advanced bilateral CLM that could not be resected by primary hepatectomy, mainly due to the presence of multiple tumors in the remaining liver.

[0018] Chemotherapy was administered as a doublet or triplet regimen. Molecular targeted drugs were prescribed according to the presence or absence of rat sarcoma viral oncogene homolog (RAS) / B-Raf murine sarcoma viral oncogene homolog B1 (BRAF) gene mutations. Postoperative adjuvant chemotherapy was administered as specified (12 cycles including preoperative chemotherapy). Chemotherapy response was evaluated every 4 cycles, defined as complete response, partial response, stable disease, or disease progression according to the Response Evaluation Criteria in Solid Tumors 1.1.

[0019] Postoperative complications were assessed using the Clavien-Dindo classification. Tumor Burden Score (TBS) was calculated based on the size and number of pretreatment tumors as follows: (TBS 2 ) = [maximum tumor diameter] 2 +[Tumor count] 2 A TBS score of 9 or more was considered to be associated with a poor prognosis. Liver segments were classified according to the Brisbane 2000 Terminology of Liver Anatomy and Resections. Viable liver segments were defined as those without diffuse presence of multiple tumors or invasion of the Glissonian pedicle.

[0020] [Statistical analysis] Categorical variables are expressed as n (%) and analyzed using Fisher exact test or χ as appropriate. 2 Comparisons were made using the 1-point analysis. Continuous variables were expressed as median (IQR) and compared using the Wilcoxon rank sum test (2 groups) or the Kruskal-Wallis test (3 groups). Prediction models and TBS were evaluated using receiver operating characteristic curve (ROC) analysis and area under the curve (AUC). Univariate and multivariate analyses were performed to analyze factors associated with CS. Data for all variables included in these analyses were collected at the time of initial diagnosis (before chemotherapy). Factors with P < 0.050 in univariate analysis were entered into a logistic regression model for multivariate analysis.

[0021] In constructing the prediction model, scores were defined as follows. Significant predictors obtained by multivariate analysis were assigned a score of 2, and predictors by univariate analysis were assigned a score of 1. Odds ratios (ORs) and 95% CI were calculated for each factor. A median tumor number of 15 lesions was used as the threshold. Survival curves were constructed by the Kaplan-Meier method and compared using the log-rank test. In the intention-to-treat analysis of the entire cohort, overall survival (OS) was defined as the period from the start of chemotherapy to the date of death. In the analysis within the CS cohort, OS was defined as the period from resection to death. Recurrence-free survival (RFS) was defined as the interval from resection to recurrence. Recurrence was diagnosed based on imaging findings, clinical data, or histopathological examination. Statistical analysis was performed using JMP Pro 15.0.0 software (SAS Institute, Inc.).

[0022] [Test subjects] The patients selected for the study are shown in Figure 1. Of the 892 cases of CLM examined between January 2014 and March 2021, 309 cases were diagnosed as unresectable at the initial visit, of which 169 cases with extrahepatic metastases and 10 cases with recurrent lesions after primary hepatectomy were excluded. In addition, 8 cases who received best supportive care were excluded from 130 patients with technically unresectable CLM, and the remaining 122 cases of technically unresectable CLM were included in the analysis. Of these, 61 cases (50%) received CS after chemotherapy (CS group) and 61 cases (50%) did not receive CS (non-CS group).

[0023] [Comparison between CS group and non-CS group] The baseline characteristics of the patients are shown in Table 1. RAS gene mutations were significantly more frequent in the non-CS group than in the CS group (CS group vs. non-CS group: 27.9% vs. 57.4%, p<0.001). The median number of tumors was significantly higher in the non-CS group than in the CS group (12 vs. 16, P=0.002). However, there was no difference in the maximum tumor diameter between the two groups (5.5 vs. 6.3 cm, P=0.814). The main reasons for technical unresectable lesions were (1) insufficient remnant liver volume, (2) invasion into the hepatic hilum, and (3) invasion into the roots of the three hepatic veins. The number of factors contributing to technical unresectable disease was significantly different between the CS and non-CS groups (CS group vs. non-CS group, 1 factor: 78.7% vs. 63.9%, 2 factors: 21.3% vs. 27.9%, 3 factors: 0% vs. 8.2%, P = 0.040).

[0024] [Table 1]

[0025] Regarding chemotherapy regimens, 83.6% and 88.5% of patients in the CS and non-CS groups, respectively, received a doublet regimen, and 16.4% and 11.5% received a triplet regimen, but there was no difference between the groups (P = 0.433). However, the frequency of administration of anti-EGFR monoclonal antibodies was significantly higher in the CS group than in the non-CS group (63.9% vs 29.5%, p < 0.001). The partial response (PR) rate was significantly higher in the CS group (P < 0.001). Patients treated with anti-EGFR monoclonal antibodies had a significantly better response rate than those treated with anti-VEGFR monoclonal antibodies (anti-EGFR monoclonal antibodies vs anti-VEGF monoclonal antibodies: 77.2% vs 41.0%, p < 0.001). The TBS was significantly lower in the CS group (14.4; range, 3.5-42.1) than in the non-CS group (17.6; range, 3.5-42.1). Figure 2 shows the OS in the CS and non-CS groups. The median OS was significantly (P<0.001) longer in the CS group (5.6 years) than in the non-CS group (1.8 years).

[0026] [Perioperative and long-term outcomes in the CS group] Table 2 shows the perioperative and outcomes of patients in the CS group. Of the 61 patients who underwent CS, 20 (32.8%) underwent parenchyma-sparing liver resection, and 41 (67.2%) underwent major liver resection, including 26 staged resections. Of the 16 patients (26.2%) in whom CLM disappeared, 14 (87.5%) underwent anatomical resection for all CLM, including the disappeared CLM. Thus, R0 resection was achieved in 59 patients (97%). Major complications (Clavien-Dindo classification IIIA or higher) were observed in 10 patients (16.4%). One patient died due to liver failure. Pathological response was evaluated in specimens obtained from the CS group (n = 61). As a result, 30 patients (49.2%) had a minor response, 27 patients (44.3%) had a major response, and 4 patients (6.6%) had a complete response. Overall, 37 patients (60.7%) received adjuvant chemotherapy, and 51 patients (83.6%) had a recurrence. The median recurrence-free survival was 0.6 years. Of the 51 recurrences, 28 were intrahepatic, 14 were extrahepatic, and 9 were both intrahepatic and extrahepatic. OS was significantly better in patients who underwent surgical treatment. Patients who underwent surgical treatment for recurrence (n=26) had a significantly better OS than those who did not (n=25) (median survival: not reached vs. 2.8). In recurrent cases, the presence or absence of RAS gene mutations was not associated with OS (wild type: n=34, mutant type: n=17, median survival: wild type 4.6 years vs. mutant type 3.3 years).

[0027] [Table 2]

[0028] [Identification of predictive factors using univariate and multivariate analyses] Univariate and multivariate analyses were performed to identify predictors associated with the suitability of CS (Table 3). Univariate logistic regression model analysis revealed that wild-type RAS (P = 0.001), number of tumors (P = 0.009), contact with hepatic veins (0 or 1 vs. 2 or 3 hepatic veins, p = 0.028), and presence of a preservable segment (at least one of the following segments: left lateral segment, left medial segment, right anterior segment, and right posterior segment; P < 0.001) were significantly associated with the suitability of CS. Among these, wild-type RAS (OR, 8.28; 95% CI, 2.68-24.94; P < 0.001) and presence of a preservable segment (OR, 34.24; 95% CI, 8.67-135.23; P < 0.001) were found to be significant predictors of CS by multivariate analysis.

[0029] [Table 3]

[0030] [Scoring system to predict CS] We developed a prediction model, the Technical Unresectable (TUR)-CS score, to score patients who were diagnosed as having technically unresectable CLM at the first visit and underwent CS using four statistically significant predictors, namely, the presence or absence of RAS gene mutation, the number of tumors, contact with the hepatic vein, and the presence or absence of a salvageable segment (Table 4). Significant predictors obtained from multivariate analysis (the presence or absence of RAS gene mutation and the presence or absence of a salvageable segment) were assigned a score of 2, and factors from univariate analysis (the number of tumors (≥15), contact with the hepatic vein (≥2 hepatic veins)) were assigned a score of 1. The AUC of CS using the TUR-CS score was 0.889, whereas that using TBS was 0.637. This indicated that the TUR-CS score was a better predictive model. The TUR-CS score was divided into three grades: Grade A (0-2 points), B (3-4 points), and C (5-6 points) (Table 4).

[0031] [Table 4]

[0032] Based on the TUR-CS score, 47 cases were grade A, 46 were grade B, and 29 were grade C, with conversion rates of 91.5%, 32.6%, and 10.3%, respectively (P<0.001; Fig. 3A). The prognosis of OS was significantly better in patients with grade A (median survival time, 5.7 years) than in those with grade B (2.2 years) and grade C (1.6 years) (both P<0.001). Grade B patients were more likely to undergo CS and had a relatively longer OS than those with grade C, but the difference was not statistically significant (OS: P=0.072, Fig. 3B). In patients who underwent CS, there was no difference in OS among the three grades (median survival time: 5.7 years vs. 5.0 years vs. no data; P=0.828).

[0033] As shown above, the inventors have clarified that the presence or absence of RAS gene mutation, the presence or absence of a preservable area, the number of tumors (median or more), and the contact of the tumor with the hepatic vein (two or more hepatic veins) are effective factors for determining the possibility of CS adaptation for CLM liver metastasis. From the grade of the TUR-CS score calculated using these four factors, even if the tumor is technically unresectable at the time of diagnosis, it is possible to determine the possibility of CS being applicable by aggressive preoperative chemotherapy. As shown by the results of comparing the OS of the CS group and the non-CS group (see Figure 2), if CS is applicable, the long-term prognosis improves, so even if the tumor is unresectable at the initial consultation, it is desirable to perform aggressive preoperative chemotherapy if the TUR-CS score indicates that CS is likely to be applicable. Conventionally, the judgment was made based on the doctor's experience, but in the future, the possibility of CS adaptation can be objectively determined by using the TUR-CS.

[0034] By accumulating more data and combining a variety of information, including not only the above four factors but also the presence or absence of extrahepatic metastasis and lymph node metastasis, tumor marker values, progression of the primary tumor (colon cancer), the presence or absence of complications and comorbidities, age, nutritional status, sarcopenia and physical activity, gene mutations other than RAS, and information obtained from CT and MRI images, it will be possible to predict the applicability of CS with greater accuracy. In addition, although we have explained in detail about liver metastasis of colon cancer here, the method can also be applied to liver tumors and liver metastases from primary tumors other than colon cancer. Furthermore, by using the same analysis method, a judgment model for CS applicability can be established for metastatic cancer in other organs.

[0035] In addition to judging CS suitability, it is also possible to estimate the recurrence pattern after resection and the re-resection rate after resection, display the possibility of CS, and also present the recommended treatment method (aggressive chemotherapy or palliative treatment).With this method, even in cases where surgery would have been given up in the past, it is now possible to select appropriate preoperative treatment and surgery at the appropriate time, which can lead to improved prognosis for colorectal cancer and colorectal cancer liver metastasis.

[0036] The system for determining CS suitability determines the number of tumors and the contact of the tumor with the hepatic vein from CT and MRI images, as well as the area that can be preserved. These results are judged by a computer attached to the CT or MRI device, scored, and sent to the CS suitability determination system. It is also possible to have AI learn the CT and MRI images at the time of the first consultation and when CS is applied, and predict the tumor shrinkage caused by aggressive chemotherapy from the images at the first consultation. The predicted results can be used to select appropriate chemotherapy. The presence or absence of mutations in the RAS gene obtained by genetic testing is also sent to the CS suitability determination system. The CS suitability determination system displays the results of these scores, as well as the possibility of resection at the current stage, and if resection is not possible, the possibility of conversion to surgery after chemotherapy. The CS suitability determination system can be developed as part of the CT or MRI device, or as an independent system.

Claims

1. In liver tumors, A method for supporting the determination of suitability for conversion surgery (CS) after aggressive preoperative chemotherapy, comprising: At the initial consultation, the presence or absence of RAS mutation, the presence or absence of a salvageable area, the number of tumors, and the contact of the tumor with the hepatic vein were examined. A method for assisting in determining applicability of CS.

2. The method for supporting the determination of suitability of CS according to claim 1, characterized in that the liver tumor is a metastatic liver tumor.

3. The method for assisting in determining suitability of CS as described in claim 2, characterized in that the presence or absence of RAS mutation, the presence or absence of a salvageable area, the number of tumors, and the contact of the tumor with the hepatic vein are scored.

4. The presence or absence of the RAS mutation and the contact of the tumor with the hepatic vein, the presence or absence of a preservable area, and the number of tumors The method for supporting judgment of applicability of a CS according to any one of claims 1 to 3, wherein the score is calculated by applying different weights.

5. The method for assisting in determining the suitability of CS as described in claim 4, characterized in that the preservable area is one or more of the left lateral area, the left medial area, the right anterior area, or the right posterior area.

6. an image analyzer for determining the presence or absence of a sparing area, the number of tumors, and the contact of the tumor with the hepatic vein from the CT and / or MRI images; A system that determines the suitability of colorectal cancer liver metastasis for CS based on input information on the presence or absence of RAS gene mutations.

Images