Proteomic biomarker and determination model for assisting identification of senile depression with beta-amyloid deposition

Abstract

This invention discloses a protein combination biomarker for assisting in the identification of β-amyloid deposition in elderly patients with depression, and the use of reagents for detecting this protein combination biomarker. It also provides the use of the aforementioned protein combination biomarker in constructing a diagnostic model for assisting in the identification of β-amyloid deposition in elderly patients with depression, and the diagnostic model itself. The protein combination biomarker includes ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17. This invention uses a multi-protein combination for assessment, comprehensively reflecting peripheral biological changes associated with β-amyloid deposition in elderly patients with depression, providing a new objective indicator for the auxiliary identification of β-amyloid deposition risk in elderly patients with depression.

Description

Technical Field

[0001] This invention belongs to the fields of psychiatry, neuroscience, in vitro diagnostics and biomarker detection technology, and specifically relates to a protein combination biomarker and determination model for assisting in the identification of whether elderly patients with depression have β-amyloid protein deposition. Background Technology

[0002] LLD (late-life depression) is a common mental disorder among the elderly. In addition to persistent low mood, loss of interest, and somatization symptoms, patients often experience decline in executive function, memory, and attention. Compared to depression in young and middle-aged adults, LLD is characterized by a prolonged course, multiple physical comorbidities, greater heterogeneity in drug responses, and poor cognitive outcomes. Some patients may progress to mild cognitive impairment or dementia during follow-up.

[0003] β-amyloid-β (Aβ) deposition is a crucial component of the pathological processes associated with Alzheimer's disease (dementia). Studies have suggested that some LLD patients may have Aβ deposition before meeting the clinical diagnostic criteria for dementia. These patients may exhibit different cognitive decline risks, disease progression rates, and require different intervention strategies compared to LLD patients without Aβ deposition. Therefore, it is clinically necessary to stratify LLD patients based on their Aβ deposition risk to guide further examination and intervention. Thus, identifying the presence of Aβ deposition in elderly individuals with depression is of significant importance for patient stratification management, determining follow-up frequency, providing early warning of cognitive risk, and assessing the window for subsequent anti-Aβ-related treatment.

[0004] Currently, the main technologies used to determine Aβ deposition status include Aβ-PET imaging, cerebrospinal fluid Aβ42 / 40 detection, and hematological marker detection. However, while Aβ-PET imaging has high spatial localization and pathological indication value, its high cost, strong equipment dependence, radiation exposure, and accessibility issues limit its application in routine psychiatric outpatient clinics and large-scale screening. Cerebrospinal fluid detection requires lumbar puncture, which is invasive, resulting in limited patient acceptance and its widespread adoption at the grassroots level. Existing hematological marker detection methods have improved testing convenience, but the effectiveness of a single hematological marker (such as plasma p-tau217 or Aβ42 / 40) can still be affected by factors such as age, comorbidities, sample processing, testing platform, and batch effects, making it difficult to establish a stable and reliable screening strategy.

[0005] Peripheral blood proteomics can simultaneously reflect multiple pathological processes, including inflammatory responses, immune regulation, blood-brain barrier function, energy metabolism, protein degradation, synaptic regulation, and stress responses. Data-independent acquisition (DIA) proteomics offers advantages such as high coverage, good quantitative reproducibility, and suitability for multi-sample comparisons, providing a technical basis for screening combined biomarkers from peripheral blood that reflect central Aβ-related pathological states.

[0006] Current proteomics research largely remains at the stage of differential protein discovery or mechanism explanation. Even if several key proteins can be screened and ROC curves plotted through machine learning, without fixed scoring formulas, standardized parameters, and judgment thresholds, it is difficult to translate the test results into verifiable and deployable positive or negative determinations. Especially in clinical translation or reagent kit development scenarios, it is necessary to clearly explain the specific methods for inputting subject protein test values ​​into the model, calculating probabilities, and determining thresholds.

[0007] Therefore, there is an urgent need to provide a plasma protein biomarker and its diagnostic model for elderly patients with depression, which can be used to help identify whether elderly patients with depression have β-amyloid protein deposition through peripheral blood sample testing and fixed formula calculation. This would provide an auxiliary basis for the selection of subjects for high-cost imaging examinations and improve the objectivity of early stratification and risk management of such patients. Summary of the Invention

[0008] The purpose of this invention is to provide a protein biomarker for assisting in the identification of β-amyloid protein deposition in elderly patients with depression.

[0009] The present invention also aims to provide the use of the above-mentioned protein combination biomarkers in constructing a diagnostic model for assisting in the identification of geriatric depression with β-amyloid deposition, and the diagnostic model.

[0010] To achieve the above objectives, the present invention adopts the following technical solution.

[0011] In a first aspect, the present invention provides a protein combination biomarker for assisting in the identification of senile depression with β-amyloid deposition, comprising ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17.

[0012] This invention also provides the use of reagents for detecting the above-mentioned protein combination biomarkers in the preparation of products for assisting in the identification of β-amyloid protein deposition in elderly patients with depression. The reagents may include those for detecting the above-mentioned protein combination biomarkers using methods such as DIA proteomics, PRM / MRM targeted proteomics, immunoassay, chemiluminescence, immunofluorescence, ELISA, or Simoa.

[0013] The present invention also provides a kit for assisting in the identification of geriatric depression with β-amyloid protein deposition, which includes reagents for detecting the above-mentioned protein combination markers.

[0014] Secondly, the present invention provides the use of the above-mentioned protein combination biomarkers in constructing a diagnostic model to assist in the identification of β-amyloid protein deposition in elderly patients with depression.

[0015] The present invention also provides a diagnostic model for assisting in the identification of geriatric depression with β-amyloid deposition. The indicators of the diagnostic model include standardized variables of the detection values ​​of the above-mentioned protein combination markers.

[0016] The judgment model described in this invention is constructed through the following steps: collecting elderly subjects with depression and determining the Aβ deposition status based on results such as Aβ-PET, cerebrospinal fluid markers, or blood Aβ-related markers; dividing the subject samples into a modeling set and a validation set; detecting the expression levels of the above-mentioned protein combination markers in the blood samples of the modeling set and performing quality control, handling of undetected values, and standardization; determining fixed standardized parameters, model coefficients, and judgment thresholds based on the modeling set, and establishing a judgment model; and using the validation set to verify the recognition efficiency and application performance of the fixed judgment model.

[0017] As a preferred embodiment of the present invention, the determination model of the present invention is: Pr _LLDplus =1 / (1+exp(-X)), where Pr_ LLDplus X represents the probability that a subject has geriatric depression with β-amyloid deposition, where X is a fixed score.

[0018] The fixed score value X is obtained by the following formula: X = -0.6628 - 0.3419 × Z_ ALPI +0.7019×Z_ LGI4 -0.7417×Z_ SERPINA6 +0.7406×Z_ EWSR1 +0.1765×Z_ PRR4 -0.7945×Z_ LAMTOR4 +0.0175×Z_ RANBP2 +0.5924×Z_ HAGH +0.6082×Z_ BPI -0.5691×Z_NELL2 +0.6157×Z_ MACROH2A1 -0.4325×Z_ IGKV1-17 ;

[0019] Among them, Z_ ALPI Z_ LGI4 Z_ SERPINA6 Z_ EWSR1 Z_ PRR4 Z_ LAMTOR4 Z_ RANBP2 Z_ HAGH Z_ BPI Z_ NELL2 Z_ MACROH2A1 and Z_ IGKV1-17 These are the variables obtained by imputing low values ​​and standardizing the corresponding protein detection values.

[0020] In a preferred embodiment of the present invention, when Pr_ LLDplus When the value is greater than or equal to 0.4127, or equivalently, the X value is greater than or equal to -0.3529, the subject is judged to have geriatric depression with β-amyloid deposition, denoted as LLD(Aβ+); when Pr_ LLDplus When the value is less than 0.4127, or equivalently, the X value is less than -0.3529, the subject is judged to be prone to geriatric depression without β-amyloid deposition, denoted as LLD(Aβ-).

[0021] The present invention also provides a method for assisting in the identification of β-amyloid protein deposition in elderly patients with depression, comprising the following steps: collecting blood samples from elderly patients with depression and detecting the expression levels of the above-mentioned protein combination markers; after filling in low values ​​and standardizing the obtained detection values, substituting them into the above-mentioned judgment model for calculation to determine whether the elderly patients with depression have β-amyloid protein deposition.

[0022] The blood sample may be a peripheral blood sample, a plasma sample, or a serum sample, preferably a plasma sample.

[0023] The beneficial effects of this invention are as follows:

[0024] Using a multi-protein combination rather than a single indicator for judgment can comprehensively reflect the peripheral biological changes associated with β-amyloid deposition in geriatric depression. Furthermore, by integrating complementary recognition information carried by multiple proteins, the dependence on the detection results of any single protein is reduced. Results from embodiments of this invention show that the AUC of the twelve-protein joint judgment model is 0.991 in the modeling set and 0.864 in the validation set, both numerically higher than the AUC of individual single-protein markers in the corresponding datasets. This indicates that the twelve-protein joint judgment model has better overall recognition performance compared to using any single protein marker for judgment.

[0025] This invention transforms the detection values ​​of protein combination biomarkers into verifiable, fixed score values ​​X and Pr by fixing low-value filling parameters, standardization parameters, scoring formulas, and judgment thresholds. LLDplus The probability value and positive / negative auxiliary judgment result; the present invention also has the advantages of relatively convenient sample acquisition, verifiable calculation process, and suitability for reagent kit instructions or software module deployment, providing a new objective indicator for the auxiliary identification of β-amyloid protein deposition risk in elderly patients with depression. Attached Figure Description

[0026] Figure 1 ROC curves for identifying LLD(Aβ+) using twelve single-protein biomarkers—ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17—were generated for modeling.

[0027] Figure 2 The ROC curve of LLD(Aβ+) is used to identify the joint decision model based on twelve protein combination markers in the modeling set.

[0028] Figure 3 To verify the ROC curves of the above twelve single protein biomarkers for recognizing LLD(Aβ+).

[0029] Figure 4 ROC curves for identifying LLD(Aβ+) using a centralized fixation model of twelve proteins. Detailed Implementation

[0030] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Unless otherwise specified, the reagents or materials used in the embodiments are all from commercial sources.

[0031] Example 1: Establishment of a combined 12-protein judgment model and determination of a fixed judgment threshold

[0032] This embodiment illustrates a method for establishing a subtype determination model for geriatric depression with β-amyloid deposition based on a combination of twelve protein biomarkers.

[0033] Ninety elderly participants with depression were included and divided into LLD(Aβ-) and LLD(Aβ+) groups based on their known β-amyloid deposition status. A stratified randomization method was used to divide the participants into a modeling set and a validation set. The modeling set was used to build the decision model, determine the low-value imputation parameters, standardization parameters, model coefficients, and fix the decision threshold. The validation set was used for model application and validation in Example 2. The modeling set included 37 LLD(Aβ-) cases and 26 LLD(Aβ+) cases, while the validation set included 16 LLD(Aβ-) cases and 11 LLD(Aβ+) cases. The specific sample distribution is shown in Table 1.

[0034] Table 1. Sample distribution of the modeling set and validation set

[0035]

[0036] Peripheral blood samples were collected from each subject, and plasma was separated. Proteomics methods were used to detect the expression levels of twelve protein biomarkers. The twelve protein biomarkers included ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17.

[0037] First, the recognition efficacy of twelve protein combination biomarkers for LLD(Aβ+) was evaluated in the modeling set. Using LLD(Aβ-) as the control group and LLD(Aβ+) as the target group, receiver operating characteristic (ROC) curves were plotted for each individual protein. The ROC curves for the twelve individual protein biomarkers in the modeling set are shown below. Figure 1 The ROC values ​​for each single protein are shown in Table 2. The results show that different single proteins have different recognition efficiencies for LLD(Aβ+), suggesting that a single protein contains certain subtype recognition information, but the stability and recognition efficiency of single protein indicators are limited.

[0038] Table 2 Summary of ROC analysis results for single protein and combined models in the modeling pool

[0039]

[0040] Furthermore, a joint judgment model for twelve proteins is established in the modeling set. Since some proteins may be below the detection limit or not stably detected in proteomics testing, undetected values ​​are not treated as true zero expression, but are processed using low-value imputation parameters determined by the modeling set. In a specific implementation, for each protein, the 1st percentile of its non-missing detection values ​​in the modeling set is used as a fixed low-value imputation value. This low-value imputation value is determined only by the modeling set and is consistently used in the validation set and subsequent test samples.

[0041] After undervalue imputation, each protein detection value Mi is standardized according to the following formula: Zi = (Mi - μi) / σi; where μi is the mean of the corresponding protein in the modeling set after undervalue imputation, and σi is the standard deviation of the corresponding protein in the modeling set after undervalue imputation. Once the modeling set is determined, μi and σi are used as fixed standardization parameters for subsequent model calculations on validation set samples or new test samples, and are not recalculated based on the validation set or test samples.

[0042] A binary logistic regression model was established using twelve standardized protein variables as input variables. The model output value is denoted as Pr_. LLDplus , representing the probability that a subject belongs to LLD(Aβ+). The determination model formula is as follows: Pr_ LLDplus =1 / (1+exp(-X)), where X is a fixed score value.

[0043] In this embodiment, the formula for calculating the fixed score value X is as follows: X = -0.6628 - 0.3419 × Z_ ALPI +0.7019×Z_ LGI4 -0.7417×Z_ SERPINA6 +0.7406×Z_ EWSR1 +0.1765×Z_ PRR4 -0.7945×Z_ LAMTOR4 +0.0175×Z_ RANBP2 +0.5924×Z_ HAGH +0.6082×Z_ BPI -0.5691×Z_ NELL2 +0.6157×Z_ MACROH2A1 -0.4325×Z_ IGKV1-17 Among them, Z_ ALPI Z_ LGI4 Z_ SERPINA6 Z_ EWSR1 Z_ PRR4 Z_ LAMTOR4 Z_ RANBP2 Z_ HAGH Z_ BPI Z_ NELL2 Z_ MACROH2A1 and Z_ IGKV1-17 These are the variables obtained by imputing low values ​​and standardizing the detection values ​​of the corresponding labeling proteins, such as Z_ IGKV1-17 This represents the variable after the protein IGKV1-17 detection value has been imputed and standardized.

[0044] Table 3 shows the detection column names, low-value imputation parameters, standardization parameters μi, σi, and model coefficients βi for the twelve proteins.

[0045] Table 3. Low-value imputation parameters, standardized parameters, and model coefficients of the twelve-protein joint determination model.

[0046]

[0047] Model output Pr_ based on model set samples LLDplus To plot the ROC curve of the joint model, see... Figure 2 The results showed that the twelve-protein joint decision model could effectively distinguish between LLD(Aβ-) and LLD(Aβ+) in the modeling set. The model's AUC in the modeling set was 0.991, with a 95% confidence interval of 0.976 to 1.000. The Youden index was used to determine the fixed decision protein Pr_. LLDplus The probability cutoff is 0.4127, corresponding to a fixed score value x threshold of -0.3529. When using this threshold for judgment, the modeling set has a sensitivity of 1.000, a specificity of 0.919, an accuracy of 0.952, a positive predictive value of 0.897, and a negative predictive value of 1.000. The performance of the joint model and the fixed judgment threshold are shown in Tables 2 and 4.

[0048] Table 4. Joint model recognition performance and fixed decision threshold of the modeling set

[0049]

[0050] Based on the above results, the determination rules of the twelve-protein joint determination model of the present invention are determined as follows: when Pr_ LLDplus When Pr_ is ≥0.4127, or equivalently X ≥ -0.3529, the subject is judged to be inclined towards LLD (Aβ+); when Pr_ LLDplus When X < 0.4127, or equivalently X < -0.3529, the subject is judged to be prone to LLD (Aβ-).

[0051] The above results demonstrate that the twelve-protein biomarker, consisting of ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17, can be used to establish a fixed judgment model and can convert the subjects' plasma protein test values ​​into verifiable fixed score values ​​X and Pr_. LLDplus Probability values ​​and LLD(Aβ+) or LLD(Aβ-) auxiliary judgment results.

[0052] Example 2: Application of the twelve-protein joint decision model in the validation set

[0053] This embodiment illustrates the application of the fixed twelve-protein joint determination model established in Example 1 in samples that were not involved in modeling.

[0054] The validation set reserved in Example 1 was used as the application sample set. The validation set includes 16 LLD(Aβ-) cases and 11 LLD(Aβ+) cases. The validation set samples are not used in model training, nor are they used in determining the low-value imputation parameters, standardized parameters, model coefficients, or fixed decision thresholds. They are only used to evaluate the application performance of the decision model obtained in Example 1.

[0055] First, the recognition efficacy of twelve single-protein biomarkers for LLD(Aβ+) was evaluated in the validation set. The results showed differences in recognition efficacy among the individual proteins in the validation set, indicating that single-protein indicators are significantly affected by sample differences. The ROC curves of the twelve single-protein biomarkers in the validation set are shown below. Figure 3 The ROC values ​​for each single protein and the combined model are shown in Table 5.

[0056] Table 5. Summary of ROC analysis results for single proteins and combined models in the validation set.

[0057]

[0058] Subsequently, the twelve protein detection values ​​of each subject in the validation set were substituted into the judgment model determined in Example 1 for calculation. Specifically, for undetected values ​​in the validation set samples, the fixed low-value imputation parameters determined by the modeling set in Example 1 were used for processing; then, the fixed μi and σi determined in Example 1 were used for standardization to obtain twelve standardized variables Zi; then, the twelve Zi were substituted into the judgment model formula determined in Example 1 to calculate the fixed score X for each subject, and Pr_ was further calculated. LLDplus =1 / (1+exp(-X)).

[0059] According to the decision rules of the decision model, when Pr_ LLDplus When Pr_ is ≥0.4127, or X ≥ -0.3529, it is determined to be LLD(Aβ+); when Pr_ LLDplus When X < 0.4127, or X < -0.3529, it is judged as LLD (Aβ-). The fixed score X (X-score), Pr_LLDplus, model judgment results, and reference group results for each subject in the validation set are shown in Table 6.

[0060] Table 6. Model calculation and judgment results for each subject in the validation set.

[0061]

[0062] Furthermore, Pr_ based on the validation set samples LLDplus ROC curves of the fixed twelve-protein joint model were plotted to evaluate its recognition performance in samples not involved in the modeling. The ROC curves of the validation set fixed twelve-protein joint model are shown below. Figure 4 The performance of the fixed joint model on the validation set is shown in Table 7. The results show that the AUC of the fixed joint model on the validation set is 0.864, with a 95% confidence interval of 0.697 to 1.000. Using the same fixed decision probability threshold of 0.4127 on the modeling set, the validation set sensitivity is 0.818, specificity is 0.750, accuracy is 0.778, positive predictive value is 0.692, and negative predictive value is 0.857. These validation results indicate that the decision model of this invention can serve as an auxiliary screening and stratification tool for the risk of β-amyloid deposition in elderly patients with depression, providing a reference for the selection of subjects for further examination.

[0063] Table 7. Joint model recognition performance and fixed decision threshold of the modeling set

[0064]

[0065] The above results demonstrate that the twelve-protein joint determination model established in this invention not only exhibits high recognition efficiency in the modeling set but also distinguishes between LLD(Aβ-) and LLD(Aβ+) in the validation set where no modeling was involved. Since the low-value imputation parameters, normalization parameters, model coefficients, and determination thresholds are all fixedly determined by the modeling set, subsequent new samples only need to detect the aforementioned twelve proteins, and calculate fixed score values ​​X and Pr_ according to the fixed processing steps. LLDplus This allows for the output of auxiliary determination results for LLD(Aβ+) or LLD(Aβ-) based on a fixed threshold. Therefore, the twelve protein combination biomarkers and their determination model described in this invention can be used for auxiliary identification, risk stratification, and subsequent examination decisions for β-amyloid deposition-related subtypes in elderly patients with depression.

[0066] In one specific embodiment, the judgment model of the present invention is used for preliminary screening and stratification of β-amyloid deposition risk in subjects clinically diagnosed with geriatric depression. For subjects judged by the judgment model to be prone to LLD (Aβ+), their Aβ deposition status can be evaluated by combining clinical symptoms, cognitive function assessment results, and other risk factors, preferably using Aβ-PET imaging, cerebrospinal fluid Aβ-related marker detection, or other further examinations. For subjects judged by the judgment model to be prone to LLD (Aβ-), continuous follow-up can be conducted in conjunction with clinical conditions, and further examinations can be performed when cognitive decline or other related clinical manifestations occur. That is, the results of the judgment model of the present invention are used to provide auxiliary screening and risk stratification information; in practical applications, the output results of the judgment model can be combined with clinical symptoms, cognitive function assessment, imaging examinations, or other body fluid marker detection results.

Claims

1. A protein biomarker for assisting in the identification of β-amyloid deposition in elderly patients with depression, characterized in that, Includes ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1 and IGKV1-17.

2. Use of the reagent for detecting the protein combination markers of claim 1 in the preparation of products for assisting in the identification of geriatric depression with β-amyloid deposition.

3. A kit for assisting in the identification of geriatric depression with β-amyloid deposition, characterized in that, Including reagents for detecting the protein combination markers of claim 1.

4. The use of the protein combination biomarker of claim 1 in constructing a diagnostic model to assist in the identification of β-amyloid deposition in elderly patients with depression.

5. A diagnostic model for assisting in the identification of β-amyloid deposition in elderly patients with depression, characterized in that, The indicators of the judgment model include standardized variables of the detection values ​​of the protein combination markers; the protein combination markers include ALPI, LGI4, SERPINA6, EWSR1, PRR4, LAMTOR4, RANBP2, HAGH, BPI, NELL2, MACROH2A1, and IGKV1-17.

6. The determination model according to claim 5, characterized in that, The model was constructed through the following steps: elderly subjects with depression were collected, and the Aβ deposition status was determined based on results such as Aβ-PET, cerebrospinal fluid markers, or blood Aβ-related markers; the subject samples were divided into a modeling set and a validation set; the expression levels of the protein combination markers in the blood samples of the modeling set were detected and quality controlled, undetected values ​​were processed, and the modeling set was standardized; fixed standardized parameters, model coefficients, and decision thresholds were determined based on the modeling set, and a decision model was established; the recognition efficiency and application performance of the fixed decision model were verified using the validation set.

7. The determination model according to claim 5, characterized in that, The determination model is as follows: Pr_ LLDplus = 1 / (1 + exp(-X)), where Pr_ LLDplus is the probability that the subject belongs to the elderly depression with beta-amyloid deposition, and X is the fixed score value; The fixed score value X is obtained by the following formula: X = -0.6628 - 0.3419 × Z_ ALPI +0.7019×Z_ LGI4 -0.7417×Z_ SERPINA6 +0.7406×Z_ EWSR1 +0.1765×Z_ PRR4 -0.7945×Z_ LAMTOR4 +0.0175×Z_ RANBP2 +0.5924×Z_ HAGH +0.6082×Z_ BPI -0.5691×Z_ NELL2 +0.6157×Z_ MACROH2A1 -0.4325×Z_ IGKV1-17 ; Among them, Z_ ALPI Z_ LGI4 Z_ SERPINA6 Z_ EWSR1 Z_ PRR4 Z_ LAMTOR4 Z_ RANBP2 Z_ HAGH Z_ BPI Z_ NELL2 Z_ MACROH2A1 and Z_ IGKV1-17 These are the variables obtained by imputing low values ​​and standardizing the corresponding protein detection values.

8. The determination model according to claim 7, characterized in that, When Pr_ LLDplus A value greater than or equal to 0.4127, or an X value greater than or equal to -0.3529, indicates that the subject is diagnosed with geriatric depression accompanied by β-amyloid deposition; when Pr_ LLDplus When the value is less than 0.4127, or the X value is less than -0.3529, the subject is judged to be prone to geriatric depression without β-amyloid deposition.

Images