Method for assisting prediction of visual field progression in glaucoma and method for screening pharmaceutical composition

Abstract

To provide a method for classifying glaucoma into clinically meaningful endotypes, predicting the progression of visual field impairment in glaucoma, and supporting an appropriate medical treatment for each endotype.SOLUTION: A stratification model was constructed by performing metabolomic analysis of blood samples obtained from multiple patients with primary open-angle glaucoma and classifying the analysis results into endotypes using unsupervised machine learning. Accordingly, by performing metabolomic analysis of a subject's blood sample and applying the results to the stratification model, the subject's endotype can be determined. On the basis of the classified endotype, more appropriate treatment can be provided.SELECTED DRAWING: None

Description

Technical Field

[0001] The present invention relates to a method for assisting in predicting the progression of visual field impairment in glaucoma, and further to a method for screening a pharmaceutical composition for treating and preventing glaucoma visual field progression.

Background Art

[0002] Glaucoma is a disease in which the optic nerve is damaged due to increased intraocular pressure or the like, resulting in visual field constriction or defects. It may progress without the patient noticing the symptoms, and if treatment is delayed, it can lead to blindness. Glaucoma increases with age, and it is said that approximately 76 million people worldwide are affected, and the number of affected people will increase to 95 million by 2030. In Japan, it is said that there are 4 million patients, and it is the leading cause of secondary blindness in Japan.

[0003] Glaucoma can be broadly classified into primary glaucoma with unknown cause, secondary glaucoma caused by trauma, corneal diseases, etc., other eye diseases, or steroid hormones, and pediatric glaucoma caused by congenital abnormalities in the angle. It is said that primary glaucoma accounts for 90% of glaucoma, and among primary glaucoma, primary open-angle glaucoma (POAG, hereinafter referred to as POAG for primary angle glaucoma) with gradually progressing symptoms is the most common, and it is considered that about 70 million people worldwide are affected.

[0004] POAG occurs frequently in people aged 40 and over, regardless of gender, and it is said that it occurs at a rate of one in 20 people aged 40 and over. However, since the progression is slow and there are no subjective symptoms, it is often not noticed in the early stages. Diagnosis of glaucoma requires examinations in ophthalmology such as intraocular pressure examination, fundus examination, and visual field examination, but since there are no subjective symptoms, it is often not possible to undergo the examinations. Therefore, when noticed, the visual field impairment has often progressed. The lost vision and visual field cannot be restored by drugs or surgery. Therefore, it is necessary to detect glaucoma early.

[0005] Conventionally, methods for simply examining glaucoma have been proposed. Patent Document 1 discloses a combined analysis for improving the detection of glaucoma and the evaluation of the progression rate based on a combination of structural examination and functional examination. It is disclosed that by combining and evaluating examinations with different modalities, glaucoma can be detected and its degree of progression can be evaluated. Patent Document 2 discloses a method for detecting glaucoma by measuring the amount or concentration of autoantibodies or proteins in a biological sample of a subject. However, these methods have not yet reached clinical use.

[0006] In addition, since the vulnerability of retinal ganglion cells is different, in POAG, clinical symptoms and outcomes vary depending on the site of visual field defects. The non-uniformity of the disease also hinders early detection and treatment. Various factors can be considered as the causes of the non-uniformity of the risk and progression of POAG. Although an increased intraocular pressure has been confirmed as a risk factor for POAG, an increased intraocular pressure has been reported to be affected by genetic and environmental factors (Non-Patent Documents 1 to 3), and these are considered to be part of the cause of the non-uniformity. In addition, it has been reported that glaucoma patients have variations in their genome (Non-Patent Document 4), transcriptome (Non-Patent Document 5), cytokine (Non-Patent Document 6), and metabolomics profile (Non-Patent Document 7), suggesting that these may be systemic factors inducing the non-uniformity of glaucoma.

[0007] Furthermore, retinal ganglion cell axons are unmyelinated and have the longest Ranvier nodes in the body, so they require more energy from mitochondria. In the etiology of POAG with central visual field defects, it has been suggested that metabolic changes play an important role. Metabolomics, which measures metabolites that are the end products of gene expression and environmental factors, has also been used in glaucoma research. A systematic review of 18 metabolome studies using various samples has identified multiple biomarkers closely related to biological mechanisms (Non-Patent Document 7). Also, in a large-scale case-control study within a cohort of 599 POAG cases matched with 599 controls, abnormal lipid metabolism has been observed even 10 years before diagnosis in glaucoma, suggesting its role (Non-Patent Document 8). Thus, although POAG has been suggested to be a heterogeneous disease in which metabolism may play an important role, glaucoma has not yet been classified into endotypes and utilized for treatment.

Prior Art Documents

Patent Documents

[0008]

Patent Document 1

Patent Document 2

Non-Patent Documents

[0009]

Non-Patent Document 1

Non-Patent Document 2

Non-Patent Document 3

Non-Patent Document 4

Non-Patent Document 5

Non-Patent Document 6

Non-Patent Document 7

Non-Patent Document 8

Non-Patent Document 9

Non-Patent Document 10

Summary of the Invention

Problems to be Solved by the Invention

[0010] As described above, since the progression of glaucoma varies depending on the endotype, the risks are likely to be different. Therefore, it is important to detect glaucoma at an early stage, classify it into endotypes, predict the progression in each type, and perform appropriate treatment for each endotype. However, although the invention described in Patent Document 1 may be able to analyze the test results with high accuracy, it is not a test method for classifying glaucoma into endotypes and evaluating the progression rate and the patient's future QOL. The invention described in Patent Document 2 can detect glaucoma and calculate the probability of suffering from it, but it is not a method capable of predicting the progression in glaucoma patients. In addition, genetic factors and environmental factors, which are said to be involved in the heterogeneity of glaucoma, are not associated with endotypes. In research on biomarkers and lipid metabolism disorders, metabolome analysis comparing groups of glaucoma and non-glaucoma patients has been conducted, but it does not characterize different endotypes of glaucoma and clarify the relationship with clinical outcomes.

[0011] As described above, although classifying the endotypes of POAG is important for predicting progression and maintaining QOL, at present, the biologically different endotypes of POAG have hardly been analyzed. In order to provide an appropriate treatment method, it is necessary to analyze the biologically meaningful endotypes of POAG and show the risk of progression according to the endotypes. By indicating the patient's endotype, an appropriate treatment method can be selected for each patient, and QOL can be maintained. The present method aims to provide a method for classifying POAG into endotypes and assisting in predicting the progression of glaucoma in patients. Another object is to provide a method for screening compounds effective for POAG.

Means for Solving the Problems

[0012] Specifically, the following methods for assisting in predicting the progression of glaucoma and screening methods are provided. (1) A method for assisting in predicting the progression of primary open-angle glaucoma (POAG), which comprises performing metabolome analysis on blood samples obtained from a plurality of POAG patients, creating a stratification model that classifies the metabolome analysis results into endotypes by unsupervised machine learning, performing metabolome analysis on the blood samples of a subject, applying the results to the stratification model, and determining the endotype of the subject, characterized by being a method for assisting in predicting glaucoma progression. The present inventors have performed metabolome analysis on blood samples obtained from glaucoma patients and found that glaucoma patients can be clinically meaningfully divided into endotypes by unsupervised machine learning. As a result, it becomes possible to stratify glaucoma patients and perform appropriate treatment for each endotype.

[0013] (2) The method for assisting in predicting glaucoma progression according to (1), wherein the stratification model is created using samples from healthy subjects in addition to POAG patients. By not only classifying glaucoma patients but also performing comparative analysis with healthy subject data, it becomes possible to predict the risk of subjects who do not currently have glaucoma developing glaucoma, and further, by classifying into endotypes, it becomes possible to predict glaucoma progression. As a result, prevention of glaucoma and early therapeutic intervention can be carried out.

[0014] (3) A trained model for predicting the progression of POAG from the metabolome analysis results of a subject's blood sample, which comprises performing metabolome analysis on blood samples obtained from a plurality of POAG patients, selecting metabolites with large variations from the obtained analysis results, clustering by unsupervised machine learning, and classifying into endotypes, characterized by being a trained model. By using the trained model created by the present inventors, it is possible to classify into endotypes from the metabolome analysis results of a subject and predict the progression of glaucoma. Since it becomes possible for the subject and the therapist to predict glaucoma progression, it becomes possible to select a more appropriate treatment.

[0015] A method for screening a compound effective in the treatment of POAG, comprising adding a candidate compound to cells, performing metabolome analysis, comparing the metabolites of the cells to which the candidate compound has been added with the metabolites of the cells to which the candidate compound has not been added, applying the learned model of (3), and selecting an effective compound. The screening of glaucoma therapeutics can use the changes in metabolites caused by candidate compounds as an index. For example, when screening for a therapeutic for the high-risk endotype B, cultured cells having a metabolome profile similar to that of endotype B can be used, and a compound whose profile changes to that of a non-endotype B or a metabolome profile of a healthy subject after adding the candidate compound can be selected. The learned model described above can be used for metabolome analysis. Furthermore, when verifying the selected compound using a model animal, the learned model can be applied to select the compound.

Brief Description of the Drawings

[0016]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Mode for Carrying Out the Invention

[0017] In this specification, glaucoma is classified into endotypes. An endotype refers to a subtype of a disease state defined by molecular biological pathologies and their causes and mechanisms. Different from phenotypes, which are mainly classified based on differences in clinical manifestations, endotypes classify the underlying mechanisms of onset and exacerbation and the responsiveness to treatment according to molecular pathologies. Classifying diseases into endotypes can lead to the elucidation of complex pathologies and contribute to the optimization of provided treatments and the development of new drugs.

[0018] As shown below, glaucoma can be divided into five endotypes by this method. Here, it is divided into endotype B, which has a rapid progression and a high risk of blindness, and non-endotype B (hereinafter referred to as non-endotype B), and further analysis is being carried out. However, it goes without saying that detailed analysis for each endotype can be useful for treatment and new drug development. There is a possibility that appropriate treatment methods and therapeutic drugs for each endotype can be provided by future analysis.

[0019] In addition, the administration method of the compound effective for the treatment of POAG obtained by the screening method described below is not particularly limited, and it can be administered by a method generally used as a pharmaceutical administration method. The administration route may be oral administration or parenteral administration. In the case of parenteral administration, examples include ocular administration, intravenous administration, intramuscular administration, intradermal administration, subcutaneous administration, transdermal administration, etc.

[0020] The dosage form of the pharmaceutical composition according to this aspect is also not particularly limited, and a dosage form commonly used as a pharmaceutical can be appropriately adopted. For example, dosage forms administered orally such as tablets, coated tablets, pills, powders, granules, capsules, solutions, suspensions, and emulsions, and dosage forms administered parenterally such as eye drops, injections, sprays, ointments, and patches can be mentioned. In addition, as a pharmaceutical carrier, known organic or inorganic carriers such as excipients, binders, and stabilizers can be used according to the dosage form.

[0021] First, while presenting data on the classification of glaucoma into endotypes, an explanation will be given, and a method for assisting in predicting the progression of glaucoma based on this will be described.

[0022] [Subject] The medical records and clinical data of glaucoma registered patients aged 20 years or older who received treatment at Tohoku University Hospital from January 2017 to January 2020 were retrospectively analyzed. After the first visit by a glaucoma specialist, the patients underwent comprehensive examinations including slit-lamp examination, gonioscopy, dilated fundus examination, intraocular pressure measurement, and static perimetry. Those with glaucomatous optic neuropathy accompanied by visual field impairment were defined as glaucoma patients. The diagnosis of POAG was based on the following criteria. (a) An open angle without debris by gonioscopy / slit-lamp examination. (b) Optic disc abnormalities (i.e., vertical Cup / Disc ratio > 0.7) or retinal nerve fiber layer defects, and reproducible glaucomatous visual field defects. For bilateral glaucoma patients, in all analyses, the data of the eye with the worse mean deviation (MD) in the visual field examination were used.

[0023] Five hundred and forty-four glaucoma patients (280 males and 264 females) with clinical data and metabolome data were enrolled. At the time of patient enrollment, a structured interview and chart review were conducted. Trained technicians unaware of the research purpose performed a general physical examination (blood pressure, BMI, etc.) and an ophthalmic examination (axial length, central corneal thickness, etc.).

[0024] From the original cohort (n = 544), exfoliative glaucoma (n = 2), primary angle-closure glaucoma (n = 3), and other types of glaucoma including pediatric glaucoma and secondary glaucoma (n = 16) were excluded (Figure 1). For endotype analysis, 523 POAG patients in whom 45 metabolites were profiled were used. Although the data are not shown here, both the average silhouette coefficient and modularity were reasonable even when the number of endotypes was different (the range of k was 3 - 8), and k = 3 - 5. The patient characteristics used in the metabolome analysis are shown in Table 1 separately from the patient characteristics of the 173 patients used in the subsequent analysis.

[0025] [Table 1]

[0026] Note that the abbreviations etc. in the table are as follows. CCT: central corneal thickness, IOP: intraocular pressure, Mean deviation (dB): mean deviation (refers to the mean deviation of the visual field index), dROM: Diacron - Reactive Oxygen Metabolites (oxidative stress measurement test), BAP: biological antioxidant potential (antioxidant capacity test), GRS: genetic risk score

[0027] [Metabolome analysis] The metabolome analysis was performed as follows. Blood samples were collected using a vacutainer tube (Terumo Corporation) containing EDTA - 2Na and centrifuged at 2,330×g for 10 minutes at 4°C. The dispensed plasma was placed in MATRIX (trademark) 2D screw tubes (Thermo Scientific) and stored at -80°C. 200 μL of plasma per sample was processed by a standard methanol extraction procedure, and the processed sample was transferred to a 3 mm Bruker SampleJet nuclear magnetic resonance (NMR) tube. NMR was performed at 298K using a Bruker 600MHz spectrometer, and spectra were obtained with 32k complex data points. The target profiling approach of the Chenomx Profiler module of Chenomx NMR Suite (Chenomx) was used to identify and quantify 45 metabolites.

[0028] [Statistical analysis] Since it was an exploratory analysis, the analysis was performed without using statistical methods to determine the sample size in advance. The 45 metabolites with the greatest variability were selected, log2 transformation and autoscaling were performed for each metabolite, and a distance matrix was calculated using the Pearson distance. To derive biologically distinct POAG endotypes, clustering centered on the medoid was applied. The final number of endotypes was determined based on the gap statistic.

[0029] Differences between endotypes in patient characteristics and clinical symptoms were determined using analysis of variance (ANOVA), the Kruskal–Wallis test, the chi-squared test, and Fisher's exact test as appropriate. Also, although not all analysis results are shown here, the relationship between major clinical characteristics including known glaucoma progression risk factors (high intraocular pressure, aging, reduction in central corneal thickness, pattern standard deviation at baseline) and metabolomics-driven endotypes was visualized using chord diagrams, Venn diagrams, and upset plots corresponding to the Venn diagrams. The chordDiagram package was used for the chord diagrams, the VennDiagram package for the Venn diagrams, and the ComplexUpset package for the upset plots. Each cutoff point was the patient median.

[0030] SAS (version 9.4, SAS Institute) and R-4.0.3 (R Foundation for Statistical Computing) were used for the analysis. A two-sided P < 0.05 was considered statistically significant. The false discovery rate (FDR) was controlled by the Benjamini–Hochberg method to correct for multiple testing.

[0031] [Genotyping, genetic risk score (GRS) analysis] Genomic DNA extracted from whole blood samples was genotyped using Japonica Array NEO (Non-Patent Document 9). The GRS for POAG was calculated as follows.

[0032] [Number] Here, GRS i represents the GRS of subject i, and x i.j represents the number of non-reference alleles of variant j that subject i has, and w j represents the weight parameter of variant j. w j is obtained from the weight parameters of 98 out of the 127 POAG-related variants reported by Gharahkhani et al. (Non-Patent Document 4). The weight parameters are based on the parameters of the POAG-genome-wide association study in the BioBank Japan Project (Non-Patent Document 10) used in the genome-wide association study reported by Gharahkhani et al.

[0033] [Endotype Classification] By unsupervised machine learning of metabolome analysis results, it was classified into five endotypes (Figure 2). After classification into endotypes, 173 people with reliable visual field data over a long period (the median follow-up period was 2.98 ± 1.0 years) were used for the remaining analysis. The patient characteristics were generally similar regardless of the presence or absence of visual field data, but the 350 people without long-term visual field data were older, had a lower proportion of normal-tension glaucoma, and had higher systolic blood pressure and oxidative stress (Table 1).

[0034] Table 2 shows the patient characteristics of the above 173 cases of POAG according to five endotypes. The glaucoma patients with endotype A were male-dominated (84.1%), and had a high prevalence of obesity, hypertension, sleep apnea, cardiovascular disease, and diabetes, but a low prevalence of migraine and coldness. Many of the patients with endotype C were female, and had a high prevalence of migraine, coldness, and oral contraceptive use. There were differences in ophthalmic parameters among the five endotypes. The axial length was the longest in endotype A and the shortest in endotype D. The intraocular pressure was the highest in endotype C and the lowest in endotype D. The central corneal thickness was relatively thick in endotypes A and E and the thinnest in endotype D. Furthermore, the oxidative stress level (high dROM test value and low antioxidant capacity test value) was high in endotypes B and C. The highest genetic risk score was 0.84 ± 0.41 in endotype D, while the lowest was 0.51 ± 0.39 in endotype E.

[0035]

Table 2

[0036] To examine the visual field progression rate, the slopes of MD (Mean deviation) and TD (total deviation) in each sector (see Figure 3) based on the classification by Garway-Heath et al. were compared. There was a significant difference in the slopes of the inferior TD (sector 5) and the central TD (central), and the progression rate was the fastest in endotype B (Table 3).

[0037]

Table 3

[0038] Endotype B is an endotype with a fast progression rate and a large defect in the central visual field. Therefore, it can be said that it is the most risky endotype. Therefore, in order to better capture the difference between endotype B and other endotypes (non-endotype B), the metabolome profiles for each endotype were compared (Table 4).

[0039]

Table 4

[0040] When comparing the high-risk endotype B with non-endotype B, there were 12 metabolomes with different expression levels, and the overall metabolome profiles were significantly different (change of 1.5 times or more in absolute value with FDR < 0.1; Table 4, Figures 4 and 5). In endotype B, 34 biologically meaningful pathways were identified (FDR < 0.05). These 34 pathways were found to be related to the pathways that upregulate fatty acid biosynthesis and ketone body metabolism. In addition, in the amino acid degradation pathway, the involvement of a large number of metabolites including 8 individually regulated metabolites in the regulation was observed. Although the data are not shown here, in the sensitivity analysis, when comparing endotype B and non-endotype B using a cohort with a reliable follow-up period (n = 173), a similar trend was observed. In addition, thresholds for 45 metabolites between endotype B and non-endotype B were calculated (Table 5). If the metabolome profile of a subject is obtained, it is possible to determine whether it is the high-risk endotype B or not.

[0041]

Table 5

[0042] As described above, the inventors have found that by using machine learning without a teacher for metabolome analysis, glaucoma can be classified into five clinically meaningful endotypes. By classifying into endotypes, it becomes possible to perform more appropriate treatments. In particular, since endotype B has a high risk of blindness, when the result of being endotype B is obtained by metabolome analysis, the frequency of follow-up observation can be increased, or the intraocular pressure can be controlled low and the risk can be reduced by drug therapy, laser therapy, surgery, etc., and appropriate treatments can be performed individually. Furthermore, by performing metabolome analysis of healthy individuals and comparing with glaucoma patients, it is also possible to judge the risk of glaucoma itself and lead to early detection.

[0043] The ability to determine glaucoma, particularly endotype B glaucoma with a high risk of blindness, from a blood sample is very important for early detection of glaucoma and reduction of the blindness risk. As described above, glaucoma has no symptoms and requires an examination in ophthalmology for diagnosis, so it is currently difficult to detect it early without undergoing an examination. As shown here, since this method enables classification into endotypes of glaucoma using a blood sample, it becomes possible to perform metabolome analysis during a health check or the like and lead to early detection and reduction of the blindness risk. For example, by applying this method to a health check for people aged 40 or older, where the proportion of glaucoma patients is increasing, analyzing the blood sample, and recommending an ophthalmology visit for those suspected of having glaucoma, it becomes possible to perform early detection and early treatment intervention.

[0044] In addition, by classifying into biologically meaningful endotypes, it is possible to develop therapeutic drugs for each endotype. Endotype B with a high risk has a high oxidative stress level and an increase in fatty acid biosynthesis and ketone body metabolism. Therefore, compounds that reduce oxidative stress and affect fatty acid synthesis and ketone metabolism can be used as candidate compounds. As a screening method, for example, for a therapeutic drug for endotype B, a candidate compound is added to the culture medium of cultured cells, such as SH-SY5Y (human neuroblastoma), and oxidative stress, fatty acid synthesis, and ketone metabolism are analyzed, and a compound that changes the metabolic profile to that of a normal or low-risk non-endotype B can be selected. The effect of the selected compound can be confirmed using a known glaucoma animal model. For the analysis, the learned model obtained by the above-mentioned metabolome analysis can be applied for the analysis. Specifically, a compound that changes the metabolome profile of cultured cells or model animals classified as endotype B to a metabolome profile close to that of non-endotype B or a healthy person can be selected.

[0045] Since the classification of endotypes was performed by metabolome analysis using blood samples, the administration route of the therapeutic drug may not only be eye drops but also oral administration. In addition, not only as a therapeutic drug, but also as a preventive drug or supplement that can be continuously ingested by those at high risk of developing glaucoma and those at high risk of transitioning to endotype B before the onset of glaucoma, and can be taken prophylactically.

[0046] As shown above, by developing a method for classifying glaucoma into endotypes, it is not only possible to lead to early detection and early therapeutic intervention of patients, but also to develop therapeutic drugs for each endotype.

Claims

**Claim 1** A method for assisting in predicting the progression of primary open-angle glaucoma (POAG), comprising: performing metabolome analysis on blood samples obtained from a plurality of POAG patients; creating a stratification model that classifies the metabolome analysis results into endotypes by unsupervised machine learning; performing metabolome analysis on a blood sample of a subject; applying the results to the stratification model; and determining the endotype of the subject, wherein the method for assisting in predicting glaucoma progression is characterized by the above steps. **Claim 2** The method for assisting in predicting glaucoma progression according to claim 1, wherein the metabolome analysis is performed using NMR. **Claim 3** The method for assisting in predicting glaucoma progression according to claim 1, wherein the blood sample is plasma. **Claim 4** The stratification model in the method for assisting in predicting glaucoma progression according to claim 1 is created using samples from healthy subjects in addition to samples from POAG patients. **Claim 5** The method for assisting in predicting glaucoma progression according to claim 1, wherein the stratification model classifies glaucoma patients into five endotypes. **Claim 6** The method for assisting in predicting glaucoma progression according to claim 4, wherein the stratification model stratifies healthy subjects and glaucoma patients, and further classifies glaucoma patients into five endotypes. **Claim 7** A trained model for predicting the progression of POAG from the metabolome analysis results of a subject's blood sample, comprising: performing metabolome analysis on blood samples obtained from a plurality of POAG patients; selecting metabolites with large variations from the obtained analysis results; clustering by unsupervised machine learning and classifying into endotypes, wherein the trained model is characterized by the above steps. **Claim 8** A method for screening a compound effective for the treatment of POAG, comprising: adding a candidate compound to cells; performing metabolome analysis; comparing the metabolites of the cells to which the candidate compound is added with the metabolites of the cells to which the candidate compound is not added; applying the trained model of claim 7 and selecting an effective compound, wherein the screening method is characterized by the above steps.

Images