Validation of the predictive ability for recurrence and the clinical utility of the 95-gene classifier (95GC) through an integrated analysis of five studies.

Validation of the predictive ability for recurrence and the clinical utility of the 95-gene classifier (95GC) through an integrated analysis of five studies. | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Validation of the predictive ability for recurrence and the clinical utility of the 95-gene classifier (95GC) through an integrated analysis of five studies. Aya Imai, Ryo Tsunashima, Yu Hidaka, Sae Kitano, Chikage Kato, and 19 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6282865/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 25 Jun, 2025 Read the published version in Breast Cancer → Version 1 posted 5 You are reading this latest preprint version Abstract Background In recent years, multigene assays have become indispensable tools for predicting the recurrence risk of estrogen receptor (ER)-positive, human epidermal growth factor receptor 2 (HER2)-negative early-stage breast cancer and guiding adjuvant chemotherapy decisions. Curebest™ 95GC Breast (95GC), developed in 2011 as a domestically produced multigene assay for postoperative recurrence prediction, has been commercially available since 2013. Since 2021, five validation studies evaluating the predictive performance 95GC have been published. This study presents an integrated analysis of these studies to validate the prognostic utility of 95GC further. Methods The integrated analysis included 719 real-world cases of luminal-type node-negative breast cancer patients who underwent adjuvant hormone therapy alone without extended endocrine treatment. Additionally, an expanded cohort incorporating 294 cases from Western patients within the GEO public database was analyzed, resulting in a total of 1,013 cases. Results Among the 719 real-world cases, 550 (76.5%) were classified into the 95GC Low-risk group, demonstrating a significantly superior prognosis compared to the High-risk group (P < 1.00e-12). The 5-year distant recurrence-free survival (DRFS) rate in the Low-risk group was approximately 98%, with consistent findings observed in the expanded cohort. Furthermore, an analysis of 754 CEL files using 21GC (a proxy for Oncotype DX®) identified 318 cases (42.2%) as the 21GC intermediate risk. 95GC successfully stratified these cases into two prognostically distinct subgroups. Conclusions These findings underscore the clinical utility of 95GC in safely omitting chemotherapy for Low-risk patients with good prognosis and in further stratifying the 21GC Intermediate-risk cases, thereby contributing to personalized treatment strategies. luminal type breast cancer Curebest™ 95GC Breast 21GC multigene assay recurrence risk prediction Figures Figure 1 Figure 2 Figure 3 1. Introduction Multigene assays utilizing gene expression analysis, developed in recent years, are critical tools in clinical practice for predicting recurrence risk, determining indications for adjuvant chemotherapy, and selecting optimal treatment strategies, particularly for ER-positive, HER2-negative early-stage breast cancer. Several assays, including OncotypeDX® from the U.S. ( 1 – 6 ), MammaPrint® from the Netherlands ( 7 – 10 ), and Curebest™ 95GC Breast from Japan ( 11 – 20 ), have been widely implemented and in use for many years. Among these, Curebest™ 95GC Breast (abbreviated as 95GC) was developed in 2011 as the first multigene assay in Japan for postoperative recurrence prediction ( 11 ). In 2013, assay services started. During the development of 95GC, the data from 549 ER-positive, HER2-negative, and node-negative breast cancer patients who received no therapy or tamoxifen were extracted from the public Gene Expression Omnibus (GEO) database, and comprehensive gene expression analysis in primary tumors was conducted. By comprehensively comparing 167 recurrence cases with 382 non-recurrence cases, the most recurrence-associated gene sets were directly extracted from more than 20,000 human genes. Then, a set of the top 95 genes most strongly correlated with recurrence was identified using a gene lineage- and function-independent approach solely based on correlation coefficients. Furthermore, by applying a multidimensional mathematical algorithm, Between Group Analysis (BGA), scoring recurrence risk derived from these expression levels enabled stratification into the two risk groups: the High-risk and the Low-risk. Notably, the development of 95GC was characterized by a purely mathematical selection of gene sets without referencing past literature such as PubMed et al. As a result, the 95-gene set, which was identified from over 20,000 human genes, was mainly composed of proliferation-related and transcription-related genes that had rarely been studied previously ( 11 ). This 95-gene set shows minimal overlap with existing multigene assays yet was deemed mathematically the most optimal gene set. By adopting BGA, patients can be stratified into two risk groups, the High-risk, and the Low-risk, effectively overcoming the issue of the Intermediate-risk group, which has been a point for Oncotype DX®. Initially, frozen and refrigerated specimens were the main targets for analysis; however, FFPE specimens can now be used ( 14 ). Validation studies in Japan and abroad have verified the clinical utility of 95GC. Since 2021, five validation studies have been published: a study by Osaka University (Study 1) ( 15 ), a five-institution domestic study ( 16 ), a multi-institutional study in the U.S. ( 17 ), a seven-institution domestic study ( 18 ), and a study by Hokkaido University ( 19 ). Notably, in all five studies, the 5-year DRFS for the low-risk group (luminal type N0 treated with hormone therapy alone) was consistently around 98%, demonstrating the robustness of the predictive ability of 95GC across regions and populations. Based on these findings, 95GC is considered effective for identifying patients classified as Low-risk, who are deemed to have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely. This study integrated data from all five validation studies since 2021 to examine 95GC's predictive ability and clinical utility for recurrence. 2. Materials and Methods 2 − 1 Patients The integrated analysis of the five 95GC studies collected recurrence prognosis and clinicopathological data for 719 cases (real-world data) of luminal type N0 breast cancer treated with adjuvant hormone therapy alone (no extended hormone therapy) from 14 Japanese institutions and 5 U.S. institutions. In response to the demand from 95GC users for a dataset exceeding 1,000 cases, an additional 294 Western patients were included from the (publicly available) GEO database in Study 1 ( 15 ), resulting in a total of 1,013 cases for analysis. While recurrence data was available for all cases, complete clinicopathological data could not be obtained for all cases due to differences in institutional policies of Principal Investigators (PIs) and approvals from the ethical review board (ERB) (Table 1). This study analyzed all the 719 cases (real-world data) (Fig. 1a) and the 1,013 cases (Fig. 1b). A unique feature of 95GC is that CEL files containing the expression values of all 20,000 genes in the primary tumor are returned to the user after testing. Among 754 CEL files collected through the 95GC integrated analysis study, 21GC, a proxy analysis for OncotypeDX®, was conducted via Recurrence Online ( https://recurrenceonline.com/gene-array/ ) ( 21 ). In this study, the Intermediate risk was defined based on the recurrence score (RS) ranging from 11 to 25 in the TAILORx trial ( 3 , 4 ) (Fig. 3a, 3b). 2–2 Statistics analysis Summary statistics were constructed using frequencies and proportions for categorical data (Table 1). The mean and standard deviation (SD) were constructed for continuous variables. DRFS was defined as the time from surgery to the first distant recurrence. DRFS was estimated using the Kaplan-Meier method and compared between groups using the log-rank test for clinicopathological factors, 95GC, and 21GC (Figs. 1a-b). Additionally, subgroup analyses were performed in cohorts of under/over age 50, T1, T1 and Grade 1, T1 and Grade 2, T1 and Ki67 Low (Ki67 < 20%), T2, and Grade 3 (Fig. 2a-g). Multivariate analysis using the Cox proportional hazards models evaluated the 95GC's predictive ability for DRFS, summarized with hazard ratios (HRs) and 95% confidence intervals (CIs) (Table 2). Uno's C statistic ( 22 ) was used to evaluate the discriminative ability of 21GC and 95GC to predict recurrence at 3 and 5 years (Tables 3–4). The C-statistic measures the discriminative ability of a model, representing the probability that a randomly selected positive sample is assigned a higher predicted probability than a randomly selected negative sample. A C-statistic of 0.5 indicates that the model performs no better than random classification, while a value of 1.0 represents perfect discrimination. Generally, a C-statistic of 0.7–0.8 is considered to indicate a "good" model, 0.8–0.9 a "very good" model, and values exceeding 0.9 an "excellent" model. Additionally, we assessed the improvement in predictive performance when 21GC or 95GC were added as predictors to clinicopathological factors, using continuous NRI (Net Reclassification Improvement) and IDI (Integrated Discrimination Improvement) ( 22 , 23 ) (Supplementary Tables 1–3). NRI measures the net improvement in risk reclassification, while IDI quantifies the enhancement in a model's discriminative ability by assessing changes in predictive probability between models. All statistical analyses were performed using R (version 4.1.3), and p-values < 0.05 were considered statistically significant. 2–3 Ethical Approval The integrated analysis of the five 95GC studies was approved by the Kyoto Prefectural University of Medicine ERB (Approval No.: ERB-C-3120-1). 3. Results 3 − 1 Integrated Analysis Results for 95GC (All Cases) The characteristics of the background of the patients included in the analysis are summarized in Table 1. Some patient background information was not available for real-world data from the U.S. and public data. The integrated analysis of five studies (14 Japanese institutions + 5 U.S. institutions, n = 719) showed that 550 cases (76.5%) were classified as the 95GC Low-risk, with a significantly better prognosis than the High-risk group (P < 1.00e-12). The 5-year DRFS in the Low-risk group was approximately 98% (Fig. 1a). For the combined data of 1,013 cases, which included public datasets, 746 cases (73.6%) were classified as the 95GC Low-risk, also showing better prognosis compared to the High-risk group (P < 1.00e-12). The 5-year DRFS in the Low-risk group was approximately 97% (Fig. 1b). 3 − 2 Subgroup Analysis Subgroup analyses according to several clinical factors were performed on patients included in real-world data of Japanese for whom information on patient background was available. Their results showed that the 95GC Low-risk group consistently had a better prognosis than the High-risk group, including the following subgroups according to clinical factors: Under age 50 (n = 198) / Over age 50 (n = 318) : The Low-risk group showed a better prognosis compared to the High-risk group (P = 2.39e-08 / P = 9.2e-08, Fig. 2a). Additionally, the High-risk group was more prevalent in the under age 50 cohort (29.29%) compared to the over age 50 cohort (21.70%). Moreover, the under age 50 cohort exhibited a trend toward early recurrence, while the over age 50 cohort showed a trend toward late recurrence (Fig. 2a). T1 (n = 329) : The low-risk group showed a better prognosis than the high-risk group (P = 2.12e-12, Fig. 2b). T1 and Grade 1 (n = 92) : This cohort showed an excellent prognosis, with no recurrences observed (Fig. 2c). T1 and Grade 2 (n = 99) : The low-risk group showed a better prognosis than the high-risk group (P = 0.0125, Fig. 2d). T1 and Ki67 Low (n = 133) : The low-risk group showed a better prognosis than the high-risk group (P = 4.82e-05, Fig. 2e). T2 (n = 187) : The low-risk group showed a better prognosis than the high-risk group (P = 0.000293, Fig. 2f). Grade 3 (T1 + T2, n = 38) : The low-risk group showed a trend toward better prognosis than the high-risk group (P = 0.0576, Fig. 2g). 3–3 Multivariate Analysis Results of a multivariate analysis of 276 patients included in the real-world data for which all information on the patient background was available, 95GC was shown to be the most independent predictor of recurrence compared to other clinicopathological factors (age, tumor size, pathological grade, PgR, Ki67). Models including age (HR: 0.20, 95%CI: 0.075–0.52) or excluding age (HR: 0.19, 95%CI: 0.073–0.51) consistently identified 95GC as the strongest predictor (Table 2). 3–4 Integrated Analysis Results for 21GC (Proxy Analysis for Oncotype DX®) 21GC analysis: Recurrence Online was performed on 754 CEL files from the 95GC integrated analysis as a proxy for Oncotype DX® ( 21 ). Among the 754 CEL files, 353 cases (46.8%) were classified as the 21GC Low-risk group, 318 cases (42.2%) as the Intermediate-risk group, and 83 cases (11.0%) as the High-risk group. Significant differences in recurrence prognosis among the three groups (P = 3.00e-08) were shown. The 5-year DRFS for the Low-risk group was 96.8%, 92.2% for the Intermediate-risk group, and 73.4% for the High-risk group (Fig. 3a). For the Intermediate-risk group (n = 318) classified by 21GC, 95GC successfully stratified these patients into two distinct groups with significant differences in recurrence prognosis (P = 5.86e-05). The 5-year DRFS in the 95GC Low-risk group reclassified from the 21GC Intermediate-risk group was 96.3% (Fig. 3b). 3–5 Evaluation of Discriminative Ability 3-5-1 Comparison of Discriminative Ability for Recurrence between 95GC and 21GC The discriminative ability of 21GC and 95GC, evaluated using C-statistics, is shown in Table 3. The analysis included 832 patients with 21GC and 95GC laboratory and prognostic data available. 21GC was used as a predictor for the Low- or High-risk classification, and patients classified as Intermediate-risk by 21GC were treated as low-risk. The difference in C-statistics for recurrence prediction between 21GC and 95GC was 0.07 (95% CI: -0.02, 0.15) at 3 years and 0.06 (95% CI: -0.01, 0.13) at 5 years, indicating that the predictive ability of the two assays was comparable. 3-5-2 Discriminative Ability of 95GC in the 21GC Intermediate-Risk Group The discriminative ability of 95GC in the 21GC Intermediate-risk group, evaluated using C-statistics, is shown in Table 4. The analysis included 318 patients classified as Intermediate risk by 21GC, with 21GC and 95GC laboratory and prognostic data available. The C-statistics for 95GC was consistently high at both 3 years and 5 years, exceeding or approximating 0.7, indicating that 95GC effectively distinguished between recurrent and non-recurrent patients classified as the Intermediate risk by 21GC. 3-5-3 Discriminative Ability of Models Incorporating 21GC or 95GC The discriminative ability of predictive models incorporating 21GC or 95GC in addition to clinicopathological factors is shown in Supplementary Table 1. The analysis included 108 patients with complete clinical, laboratory, and prognostic data for 21GC and 95GC. The base model included the following clinicopathological factors: age (under 50 vs. over 50 years), grade (1 vs. 2 or 3), Ki67 status (positive vs. negative), and PgR status (positive vs. negative). 21GC was used as a predictor for the Low- or High-risk classification, and patients classified as Intermediate-risk by 21GC were treated as the Low risk. The reclassification table for these models is provided in Supplementary Table 2, showing how individuals were reclassified into different risk categories based on the new models. Cells representing whose risk classification improved are highlighted in grey. At 3 and 5 years, models incorporating 21GC (Base model + 21GC) or 95GC (Base model + 95GC) and clinicopathological factors demonstrated improved discrimination and reclassification in Supplementary Table 1,2. Improvements were more pronounced at 5 years in both models, with the following results: Base model + 21GC : 3-year NRI: 0.14 (95% CI: -0.57, 0.55); IDI: 0.00 (95% CI: -0.11, 0.33) 5-year NRI: 0.31 (95% CI: -0.49, 0.70); IDI: 0.06 (95% CI: -0.08, 0.44) Base model + 95GC : 3-year NRI: 0.36 (95% CI: -0.36, 0.64); IDI: 0.03 (95% CI: -0.04, 0.23) 5-year NRI: 0.41 (95% CI: -0.32, 0.72); IDI: 0.05 (95% CI: -0.04, 0.31) The discriminative ability of these models for recurrence prediction at 3 and 5 years, evaluated using C-statistics, is shown in Supplementary Table 3. The difference in C-statistics between Base model + 21GC and Base model + 95GC was 0.005 (95% CI: -0.17, 0.18) at 3 years and − 0.02 (95% CI: -0.18, 0.15) at 5 years, suggesting that both models had comparable predictive ability for recurrence. 4. Discussion 95GC is a novel multigene assay that utilizes comprehensive gene expression analysis to identify optimal marker gene mathematically sets for recurrence prediction. It uniquely employs a multidimensional algorithm, Between Group Analysis (BGA), to categorize patients into the High- and Low-risk groups. This study presents an integrated analysis of all five validation studies published since 2021. In the analysis of 719 luminal type node negative breast cancer cases (real-world data) treated with hormone therapy alone, the 5-year DRFS in the 95GC low-risk group were approximately 98%. This finding supports the conclusion that ‘The low-risk patients classified by 95GC, who are deemed to have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely.’ This principle forms the foundation of multigene assays. Similar results were observed in the expanded analysis of 1,013 cases, including public datasets combined in Study 1 ( 15 ). Subgroup Analysis Subgroup analysis highlighted additional insights. Specifically, no recurrences were observed in the T1 and Grade 1 cohort (n = 92) (Fig. 2c), suggesting that multigene assays may have limited utility in such inherently low-risk populations. However, the sample size of the population belonging to such cohorts in this study was small, and this finding requires further investigation with larger sample sizes. In the other subgroup analyses (e.g., T1 (n = 329), T1 and Grade 2 (n = 99), T1 and Ki67 Low (n = 133), T2 (n = 187), and Grade 3 (n = 38)), the Low-risk group consistently demonstrated significantly better prognosis or a trend toward better prognosis compared to the High-risk group. These findings suggest that multigene assays are valuable tools for guiding decisions on adjuvant chemotherapy for such subgroup. 21GC (Proxy Analysis for Oncotype DX®) 95GC provides users with CEL files containing comprehensive gene expression data as CD data after the assay. While these files are intended for research purposes only, they allow the calculation of other multigene assays as proxy analyses. Communicating 21GC results obtained through such analyses to patients is strictly prohibited. In this study, a total of 754 CEL files obtained from the 95GC integrated analysis were used for a proxy analysis of 21GC. 21GC has an established track record as a proxy for OncotypeDX® ( 12 , 15 , 20 , 21 ). The result revealed significant differences in recurrence prognosis among the three groups (Fig. 3a). The 5-year DRFS for the Low-risk group was 96.8%, indicating that chemotherapy could be omitted with relative safety. Conversely, the 5-year DRFS for the Intermediate-risk group was 92.2%, slightly worse than that of the Low-risk group, leading to uncertainty regarding the appropriateness of adjuvant chemotherapy. Implications for Real-World Data (RWD) The worse prognosis of the Intermediate-risk group (RS 11–25) in this study compared to the TAILORx trial (RS 11–25) ( 3 , 4 ) is hypothesized to be due to differences in tumor size within the cohorts. In this integrated analysis study, approximately 36% of cases were classified as T2 or larger tumors (limited to cases with available data). In contrast, the TAILORx trial predominantly included T1 tumors with a median tumor size of approximately 1.5 cm. We would like to consider which cohorts more accurately reflect real-world data (RWD) in clinical practice. According to the report of 95,870 Japanese Breast Cancer by Kubo et al. ( 24 ), approximately 37% of breast cancer cases were classified as T2 or larger. This suggests that the cohort in this integrated analysis study more closely reflects RWD than the TAILORx trial. Further evidence supporting the reflection of RWD in this integrated analysis study is presented in Fig. 3a. Here, the proportion of T2 or larger tumors (approximately 37%) closely replicates the B14 cohort validation of OncotypeDX® (T2 or larger: 38% (254/668)) ( 1 ). Over the past two decades, the definition of the Intermediate-risk group has shifted from RS 18–30 to RS 11–25. As a result, this study shows a slightly improved prognosis curve for the 21GC Intermediate-risk group compared to the B14 cohort validation. Nevertheless, the prognosis for the 21GC Intermediate-risk group (RS 11–25) remains relatively worse than that of the Low-risk group, leading to challenges in determining the appropriateness of adjuvant chemotherapy. Therefore, while the results of the TAILORx trial are highly applicable to T1 cohorts, they should be interpreted with caution when applied to real-world clinical practice (RWD), where approximately 37% of cases involve T2 tumors. In other words, determining the appropriateness of adjuvant chemotherapy for the 21-gene assay (21GC) Intermediate-risk group (RS 11–25) remains challenging. Relying solely on the TAILORx trial, which predominantly included T1 cases, may lead to an excessive omission of adjuvant chemotherapy in this group. To optimize decision-making, it is beneficial to incorporate clinical and pathological factors and other relevant genetic markers. Clinical Implications of 95GC in the 21GC Intermediate-Risk Group To assess the appropriateness of adjuvant chemotherapy for the Intermediate-risk group, 95GC was applied to 318 cases classified as the 21GC Intermediate-risk group in this study. The results demonstrated that 95GC significantly stratified the recurrence prognosis of the 21GC Intermediate-risk group into two distinct groups: the High-risk and the Low-risk (Fig. 3b). The 5-year DRFS in the 95GC Low-risk group was 96.3%, indicating favorable outcomes with hormone therapy alone, without adjuvant chemotherapy. Furthermore, the 10-year DRFS in the 95GC Low-risk group was 81.4%, suggesting that extended hormone therapy may be a good option for this group. Thus, for patients in the 21GC Intermediate-risk group, where treatment decisions are challenging, 95GC appears to be a valuable tool for further stratification. Challenges in Randomized Controlled Trials (RCTs) There is public interest in conducting a randomized controlled trial (RCT) for 95GC; however, conducting an RCT in the High-risk group for multigene assays that classify patients into only the two groups (the High-risk and the Low-risk) without an Intermediate-risk category raises ethical concerns. It should be noted that OncotypeDX® conducted an RCT exclusively in the Intermediate-risk group ( 3 , 4 ). As a result, the additional benefit of adjuvant chemotherapy for High-risk patients has only been demonstrated through retrospective studies rather than RCTs. Both OncotypeDX® and 95GC have provided evidence of the additional benefit of adjuvant chemotherapy for High-risk patients based on retrospective analyses ( 2 , 18 ). Therefore, neither OncotypeDX® nor 95GC functions as a companion diagnostic marker for specific drugs. In the past, MammaPrint®, a two-group multigene assay that classifies patients into High- and Low-risk categories, was evaluated in a randomized controlled trial (RCT). In the MINDACT trial, patients were stratified based on both Clinical and Genomic risk assessments, resulting in four groups, including the Clinical High/Genomic Low and Clinical Low/Genomic High groups, to compare recurrence outcomes ( 9 ). MammaPrint® may have adopted this trial design because, unlike Oncotype DX, it does not have an Intermediate-risk group that would serve as a suitable target for an RCT. The results showed no significant difference in prognosis between patients who received chemotherapy and those who did not in the Clinical High/Genomic Low group (P = 0.27). Similarly, no significant difference was observed in the Clinical Low/Genomic High group (P = 0.66) ( 9 ). These findings remained consistent in subsequent follow-up investigations ( 10 ). In other words, neither the Clinical High/Genomic Low group nor the Clinical Low/Genomic High group demonstrated a clear additional benefit from chemotherapy. Consequently, the MINDACT trial provided insights into the complexities of designing RCTs for multigene assays within a binary (High-/Low-risk) classification framework. Therefore, there are no plans to conduct an RCT with a similar design for 95GC. In multigene assays, the core principle is that ‘the Low-risk patients have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely.’ This study demonstrated this principle through an integrated analysis of real-world data. Application of 95GC in Luminal Type Node Positive Breast Cancer Finally, the analysis of 95GC in luminal type node positive breast cancer is introduced. Matsumoto et al. conducted a computer-based simulation analysis using public datasets to simulate the OncotypeDX® RxPONDER trial. They applied 95GC to a cohort of luminal type node positive breast cancer with RS 0–25 ( 20 ). Their results identified a poor prognosis group requiring chemotherapy in postmenopausal patients and, conversely, a favorable prognosis group unlikely to benefit from chemotherapy in premenopausal patients. According to the TAILORx trial, the RS 0–25 group includes the majority (85.7%) of luminal type breast cancer cases. This indicates that the RS 0–25 cohort is a heterogeneous population comprising patients with both favorable and poor prognoses. Therefore, treatment strategies for luminal type node positive breast cancer with RS 0–25 should not rely solely on a clinical trial result based on menopausal status or age. Instead, recurrence biomarkers such as 95GC should be considered as a guide. Even for luminal type node positive breast cancer with RS 0–25, where treatment decisions are challenging, 95GC appears to be a valuable tool. Future Perspectives This study demonstrated, through the integration of multicenter data, that the prognosis of the 95GC Low-risk group is sufficiently favorable with hormone therapy alone, allowing them to omit chemotherapy safely. We aim to continue accumulating more cases to validate these findings further. Finally, Curebest™ 95GC Breast has been developed not only as a clinically useful test but also as a novel diagnostic tool with the potential to build a research database for the future. Unlike existing assays such as Oncotype DX® and MammaPrint®, which measure the expression of dozens of genes, 95GC simultaneously quantifies the expression of approximately 20,000 human genes. It provides this data to users in the form of CEL files. By utilizing these CEL files (as demonstrated in this study with 21GC), it becomes feasible to perform analyses for additional multigene assays enabling personalized medicine, such as 23GC for predicting chemotherapy sensitivity ( 25 ), 42GC for predicting late recurrence ( 26 ), and 155GC for predicting post-chemotherapy recurrence prognosis ( 27 ). In 2021, a groundbreaking research service called 'GC Note' was launched, marking the first globally available service for cloud storage of CEL files collected nationwide after testing. The comprehensive database of whole-gene expression profiles enables the measurement and development of new multigene assays through its utilization. Implementing 95GC is expected to enable optimal personalized medicine tailored to individuals and accelerate biomarker research in Japan. However, further evidence accumulation is necessary. Abbreviations 95GC Curebest™ 95GC Breast 21GC Oncotype DX® 21GC score 21GC Recurrence score ER estrogen receptor PgR progesterone receptor HER2 human epidermal growth factor receptor type2 MGA multigene assay DRFS distant recurrence-free survival Declarations Author contributions Y.N. and A.I. designed and take the lead in this experiment. Y.H. and S.M. performed data analysis. N.U. helped supervise this experiment. All the other authors performed data collection, data analysis and contributed to sample preparation. All authors read and approved the final manuscript. Funding The authors declare that no funds, grants, or other supports were received during the preparation of this manuscript. Data availability The data analyzed during the current study are available from the corresponding author on reasonable request. Conflict of interest Yasuto Naoi has received research funding from Sysmex, ONO, Daiichi-Sankyo and AstraZeneca, and honoraria from AstraZeneca, Pfizer, Eli Lilly, Daiichi-Sankyo and Chugai outside the submitted work; he holds joint patents with Sysmex including Curebest™ 95GC Breast (JP.5725274.B2). Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study was approved by the Kyoto Prefectural University of Medicine Ethics Committee（ERB-C-3120-1）. References Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M, et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med. 2004;351(27):2817–26. Paik S, Tang G, Shak S, Kim C, Baker J, Kim W, et al. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol. 2006;24(23):3726–34. Sparano JA, Gray RJ, Makower DF, Pritchard KI, Albain KS, Hayes DF, et al. Prospective validation of a 21-gene expression assay in breast cancer. N Engl J Med. 2015;373(21):2005–14. Sparano JA, Gray RJ, Makower DF, Pritchard KI, Albain KS, Hayes DF, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med. 2018;379(2):111–21. Sparano JA, Gray RJ, Makower DF, Albain KS, Saphner TJ, Badve SS, et al. Clinical outcomes in early breast cancer with a high 21-gene recurrence score of 26 to 100 assigned to adjuvant chemotherapy plus endocrine therapy: a secondary analysis of the TAILORx randomized clinical trial. JAMA Oncol. 2020;6(3):367–74. Kalinsky K, Barlow WE, Gralow JR, Meric-Bernstam F, Albain KS, Hayes DF, et al. 21-Gene assay to inform chemotherapy benefit in node-positive breast cancer. N Engl J Med. 2021;385(25):2336–47. van de Vijver MJ, He YD, van 't Veer LJ, Dai H, Hart AA, Voskuil DW, Schreiber GJ, Peterse JL, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25):1999–2009. van 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, Peterse HL, van der Kooy K, et al. Gene expression profiling predicts the clinical outcome of breast cancer. Nature. 2002;415(6871):530–6. Cardoso F, van't Veer LJ, Bogaerts J, Slaets L, Viale G, Delaloge S, et al. 70-gene signature as an aid to treatment decisions in early-stage breast cancer. N Engl J Med. 2016;375(8):717–29. Piccart M, van 't Veer LJ, Poncet C, Cardozo JM, Delaloge S, Pierga JY, et al. 70-gene signature as an aid for treatment decisions in early breast cancer: updated results of the phase 3 randomized MINDACT trial with an exploratory analysis by age. Lancet Oncol. 2021;22(4):476–88. Naoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Development of 95-gene classifier as a powerful predictor of recurrences in node-negative and ER-positive breast cancer patients. Breast Cancer Res Treat. 2011;128(3):633–41. Naoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Comparison of efficacy of 95-gene and 21-gene classifiers (Oncotype DX) for prediction of recurrence in ER-positive and node-negative breast cancer patients. Breast Cancer Res Treat. 2013;140(2):299–306. Naoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Multi-gene classifiers for prediction of recurrence in breast cancer patients. Breast cancer (Tokyo Japan). 2016;23(1):12–8. Naoi Y, Saito Y, Kishi K, Shimoda M, Kagara N, Miyake T, et al. Development of recurrence risk score using a 95-gene classifier and its application to formalin-fixed paraffin-embedded tissues in ER-positive, HER2-negative, and node-negative breast cancer. Oncol Rep. 2019;42(6):2680–5. Naoi Y, Tsunashima R, Shimazu K, Noguchi S, et al. The multigene classifiers 95GC/42GC/155GC for precision medicine in ER-positive HER2-negative early breast cancer. Cancer Sci. 2021;112(4):1369–75. Tsukamoto F, Arihiro K, Takahashi M, Ito K, Ohsumi S, Takashima S, et al. Multicenter retrospective study on the use of Curebest™ 95GC Breast for estrogen receptor-positive and node-negative early breast cancer. BMC Cancer. 2021;21(1):1077. Fujii T, Masuda H, Cheng YC, Yang F, Sahin AA, Naoi Y, et al. A 95-gene signature stratifies recurrence risk of invasive disease in ER-positive, HER2-negative, node-negative breast cancer with intermediate 21-gene signature recurrence scores. Breast Cancer Res Treat. 2021;189(2):455–61. Naoi Y, Tsunashima R, Shimazu K, Oikawa M, Imanishi S, Koyama H, et al. Validation of the prognosis of patients with ER-positive, HER2-negative, and node-negative invasive breast cancer classified as low risk by Curebest™ 95GC Breast in a multi-institutional registry study. Oncol Lett. 2023;25(5):209. Yamashita H, Hatanaka KC, Yamagishi K, Saito Y, Hamasaki K, Taniguchi M, et al. Evaluation of a 95-Gene Classifier of formalin-fixed paraffin-embedded tissues in ER-positive, HER2-negative, and node-negative breast cancer. Anticancer Res. 2023;43(2):707–11. Matsumoto S, Tsunashima R, Kitano S, Watanabe A, Kato C, Morita M, et al. Multi-gene assay 95- and 155-gene classifiers for prognosis prediction and chemotherapy omission in lymph node-positive luminal-type breast cancer. Cancer Treat Res Commun. 2023;36:100711. Gyorffy B, Hatzis C, Sanft T, Hofstatter E, Aktas B, Pusztai L. RecurrenceOnline: an online analysis tool to determine breast cancer recurrence and hormone receptor status using microarray data. Breast Cancer Res Treat. 2012;132(3):1025–34. Uno H, Cai T, Pencina MJ, D'Agostino RB, Wei LJ. On the C-statistics for evaluating overall adequacy of risk prediction procedures with censored survival data. Stat Med. 2011;30(10):1105–16. Uno H, Tian L, Cai T, Kohane IS, Wei LJ. A unified inference procedure for a class of measures to assess improvement in risk prediction systems with survival data. Stat Med. 2012;31(25):2579–89. Kubo M, Kumamaru H, Isozumi U, Miyashita M, Nagahashi M, Kadoya T et al. Annual report of the Japanese Breast Cancer Society registry for 2016. Breast Cancer. 2020;27(4):511–518. Sota Y, Naoi Y, Tsunashima R, Kagara N, Shimazu K, Maruyama N, et al. Construction of a novel immune-related signature for prediction of pathological complete response to neoadjuvant chemotherapy in human breast cancer. Ann Oncol. 2014;25(1):100–6. Tsunashima R, Naoi Y, Shimazu K, Kagara N, Shimoda M, Tanei T, et al. Construction of a novel multi-gene assay (42-gene classifier) for prediction of late recurrence in ER-positive breast cancer patients. Breast Cancer Res Treat. 2018;171(1):33–41. Tsunashima R, Naoi Y, Kagara N, Shimoda M, Shimomura A, Maruyama N, et al. Construction of a multi-gene classifier for prediction of response to and prognosis after neoadjuvant chemotherapy for estrogen receptor-positive breast cancers. Cancer Lett. 2015;365(2):166–73. Tables Tables 1 to 4 are available in the Supplementary Files section. Supplementary Files Table1.tif Table 1: The characteristics of the background of the patients included in the analysis. Table2.tif Table 2: Multivariate analysis of DRFS in 276 cases included in the real-world data. Table3.tif Table 3: Evaluation of the discriminative ability of 21GC and 95GC using C-statistics in 832 cases. Table4.tif Table 4: Evaluation of the discriminative ability of 95GC in the 21GC Intermediate-risk group using C-statistics in 318 cases. Supp.Table1.tif Supplementary Table 1: Evaluation of the discriminative ability of incorporating 21GC or 95GC in addition to clinicopathological factors as predictors. Supp.Table2.tif Supplementary Table 2: Reclassification table based on predictive models incorporating 21GC or 95GC and clinicopathological factors. Supp.Table3.tif Supplementary Table 3: Discriminative ability of the base model + 21GC or the base model + 95GC using C-statistics to predict recurrence at 3 and 5 years. Cite Share Download PDF Status: Published Journal Publication published 25 Jun, 2025 Read the published version in Breast Cancer → Version 1 posted Editorial decision: Minor Revision 14 May, 2025 Reviewers agreed at journal 01 Apr, 2025 Reviewers invited by journal 30 Mar, 2025 Editor assigned by journal 27 Mar, 2025 First submitted to journal 24 Mar, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6282865","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":436048133,"identity":"700f7ade-3599-432e-8c10-0837064444ba","order_by":0,"name":"Aya Imai","email":"","orcid":"","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Aya","middleName":"","lastName":"Imai","suffix":""},{"id":436048134,"identity":"036d29bb-1617-4c09-944d-491753aa5be9","order_by":1,"name":"Ryo Tsunashima","email":"","orcid":"","institution":": Independent","correspondingAuthor":false,"prefix":"","firstName":"Ryo","middleName":"","lastName":"Tsunashima","suffix":""},{"id":436048135,"identity":"b3322cc9-d9f5-4b78-9018-021ac490b987","order_by":2,"name":"Yu Hidaka","email":"","orcid":"","institution":"Kyoto University: Kyoto Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Hidaka","suffix":""},{"id":436048136,"identity":"e18352cd-1152-466b-b7c5-c9dc48f8636e","order_by":3,"name":"Sae Kitano","email":"","orcid":"","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Sae","middleName":"","lastName":"Kitano","suffix":""},{"id":436048137,"identity":"c180d03a-d1b1-4b3c-8ec3-4d5178596bca","order_by":4,"name":"Chikage Kato","email":"","orcid":"","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Chikage","middleName":"","lastName":"Kato","suffix":""},{"id":436048138,"identity":"40795cc4-a74a-42fd-8d25-e677f5e734a7","order_by":5,"name":"Akira Watanabe","email":"","orcid":"","institution":"Kyoto Prefectural University: Kyoto Furitsu Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Akira","middleName":"","lastName":"Watanabe","suffix":""},{"id":436048139,"identity":"88ebf7da-a498-4d4e-a77b-b327d1e2384f","order_by":6,"name":"Midori Morita","email":"","orcid":"","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Midori","middleName":"","lastName":"Morita","suffix":""},{"id":436048140,"identity":"65dc8685-703e-4c92-9692-cd35defe5909","order_by":7,"name":"Koichi Sakaguchi","email":"","orcid":"","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Koichi","middleName":"","lastName":"Sakaguchi","suffix":""},{"id":436048141,"identity":"59b36750-62bf-4303-ab82-027be2192a4c","order_by":8,"name":"Yoshiaki Sota","email":"","orcid":"","institution":"Osaka University School of Medicine Graduate School of Medicine: Osaka Daigaku Daigakuin Igakukei Kenkyuka Igakubu","correspondingAuthor":false,"prefix":"","firstName":"Yoshiaki","middleName":"","lastName":"Sota","suffix":""},{"id":436048142,"identity":"784a3528-a965-4dbf-9cf4-925c2f4f813b","order_by":9,"name":"Masahiko Suzuki","email":"","orcid":"","institution":"Yamagata University: Yamagata Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Masahiko","middleName":"","lastName":"Suzuki","suffix":""},{"id":436048143,"identity":"9bb4774c-636f-46d5-93a6-13baf2e82169","order_by":10,"name":"Takayuki Kinoshita","email":"","orcid":"","institution":"National Hospital Organisation Tokyo Medical Center: Tokyo Iryo Center","correspondingAuthor":false,"prefix":"","firstName":"Takayuki","middleName":"","lastName":"Kinoshita","suffix":""},{"id":436048144,"identity":"8fb286b1-95b6-4b24-8d23-f047bc79e56a","order_by":11,"name":"Seiichi Imanishi","email":"","orcid":"","institution":"Osaka Rosai Hospital: Osaka Rosai Byoin","correspondingAuthor":false,"prefix":"","firstName":"Seiichi","middleName":"","lastName":"Imanishi","suffix":""},{"id":436048145,"identity":"aaecb616-6950-484c-b578-682e79f19b24","order_by":12,"name":"Masahiro Oikawa","email":"","orcid":"","institution":"Nagasaki University: Nagasaki Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Masahiro","middleName":"","lastName":"Oikawa","suffix":""},{"id":436048146,"identity":"4ef9bc85-241a-481f-a2e9-235a270f9939","order_by":13,"name":"Yoshihiko Kamada","email":"","orcid":"","institution":"Tohoku University: Tohoku Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Yoshihiko","middleName":"","lastName":"Kamada","suffix":""},{"id":436048147,"identity":"ad259e1f-87c0-4bc1-a384-aead45e35607","order_by":14,"name":"Ken-ichi Ito","email":"","orcid":"","institution":"Shinshu University: Shinshu Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Ken-ichi","middleName":"","lastName":"Ito","suffix":""},{"id":436048148,"identity":"d9fe24a5-2a95-4898-8f74-6985268b24c1","order_by":15,"name":"Takaaki Oba","email":"","orcid":"","institution":"Shinshu Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Takaaki","middleName":"","lastName":"Oba","suffix":""},{"id":436048149,"identity":"2b5166fe-44be-43a5-bc65-4df6ae49b882","order_by":16,"name":"Shin Takayama","email":"","orcid":"","institution":"National Cancer Center","correspondingAuthor":false,"prefix":"","firstName":"Shin","middleName":"","lastName":"Takayama","suffix":""},{"id":436048150,"identity":"4c5609f0-50e6-475d-bd3e-df941d2c6eda","order_by":17,"name":"Fumine Tsukamoto","email":"","orcid":"","institution":"Japan Community Health Care Organization Osaka Hospital: Dokuritsu Gyosei Hojin Chiiki Iryo Kino Suishin Kiko Osaka Byoin","correspondingAuthor":false,"prefix":"","firstName":"Fumine","middleName":"","lastName":"Tsukamoto","suffix":""},{"id":436048151,"identity":"347ba0d6-14fe-4e5b-be73-f107c1b6b10e","order_by":18,"name":"Mina Takahashi","email":"","orcid":"","institution":"National Hospital Organization Shikoku Cancer Center: Kokuritsu Byoin Shikoku Gan Center","correspondingAuthor":false,"prefix":"","firstName":"Mina","middleName":"","lastName":"Takahashi","suffix":""},{"id":436048152,"identity":"ca03d6f0-962c-43b2-a906-d8fb94ebfe94","order_by":19,"name":"Yutaka Hatanaka","email":"","orcid":"","institution":"Hokkaido University: Hokkaido Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Yutaka","middleName":"","lastName":"Hatanaka","suffix":""},{"id":436048153,"identity":"be1bcf64-b5fa-416a-8ce6-e6a8d912ce96","order_by":20,"name":"Naoto T. Ueno","email":"","orcid":"","institution":"University of Hawai'i Cancer Center","correspondingAuthor":false,"prefix":"","firstName":"Naoto","middleName":"T.","lastName":"Ueno","suffix":""},{"id":436048154,"identity":"b87590a0-783a-4d7d-ba15-3e363d61fe82","order_by":21,"name":"Kenzo Shimazu","email":"","orcid":"","institution":"Osaka University Faculty of Medicine Graduate School of Medicine: Osaka Daigaku Daigakuin Igakukei Kenkyuka Igakubu","correspondingAuthor":false,"prefix":"","firstName":"Kenzo","middleName":"","lastName":"Shimazu","suffix":""},{"id":436048155,"identity":"9ac2f7e5-dbfd-44a4-9d9a-62b98a5b3cc4","order_by":22,"name":"Satoshi Morita","email":"","orcid":"","institution":"Kyoto University: Kyoto Daigaku","correspondingAuthor":false,"prefix":"","firstName":"Satoshi","middleName":"","lastName":"Morita","suffix":""},{"id":436048156,"identity":"37458039-460f-4c0e-bb97-b3b4ddf5f168","order_by":23,"name":"Yasuto Naoi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYJCCAx8OgCg2hIgBIS0HZxxgkCBNCzMPuha8wJz98MHDNmds6vgZ2NIe8zBsY+BvP8BQXIBHi2VPWsLhnBtpEpINbMeNeRhuM0icSWAwnoFHi8GBHIPDOR8OSxgcYG+T5v13m4HhBgMDUC8eLeffGBy2gGkB2SJPUMsNoC0MN0Ba2I6BtRgQ1vIs4WDPmTTJmc1s6YZzGG7zGJ5JbMDvl/PJhz/8OGbDz8/eZvbgDcNtObnjh48Z4wsxBGCGRAzQSYxtxkTpYECKS+bHxGoZBaNgFIyCEQEAw29JV5SKdQYAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0002-6090-0137","institution":"Kyoto Prefectural University of Medicine: Kyoto Furitsu Ika Daigaku","correspondingAuthor":true,"prefix":"","firstName":"Yasuto","middleName":"","lastName":"Naoi","suffix":""}],"badges":[],"createdAt":"2025-03-22 10:03:51","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6282865/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6282865/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s12282-025-01734-2","type":"published","date":"2025-06-25T15:57:23+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":81025814,"identity":"abba74f5-1384-4381-9d67-89b81cead0f7","added_by":"auto","created_at":"2025-04-21 10:36:50","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":672826,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e1a\u003c/strong\u003e: Kaplan-Meier curves showing DRFS in 719 luminal type N0 breast cancer patients treated postoperatively with hormone therapy alone. Data was collected from 14 Japanese and 5 U.S. institutions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e1b\u003c/strong\u003e: Kaplan-Meier curves showing DRFS in a cohort of 1,013 luminal type N0 breast cancer patients, including 719 real-world data cases and 294 public dataset cases treated with hormone therapy alone.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/3732ccce781af4cfe4ddf1a9.jpg"},{"id":81026900,"identity":"931bea3e-fcb3-45b3-bcbe-f6bf46d5e0d8","added_by":"auto","created_at":"2025-04-21 10:44:50","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1768667,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e2a\u003c/strong\u003e: Subgroup analysis of DRFS by age group: cohorts under 50 years (n=198) and over 50 years (n=318).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2b\u003c/strong\u003e: Subgroup analysis of DRFS in T1 tumors (n=329).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2c\u003c/strong\u003e: Subgroup analysis of DRFS in T1 and Grade 1 tumors (n = 92).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2d\u003c/strong\u003e: Subgroup analysis of DRFS in T1 and Grade 2 tumors (n=99).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2e\u003c/strong\u003e: Subgroup analysis of DRFS in T1 tumors with Ki67 Low (n=133).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2f\u003c/strong\u003e: Subgroup analysis of DRFS in T2 tumors (n=187).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2g\u003c/strong\u003e: Subgroup analysis of DRFS in Grade 3 tumors (T1 + T2, n=38).\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/9945671619284acb4ed99d34.jpg"},{"id":81026901,"identity":"d70d021d-81b2-4823-be32-463999a944d4","added_by":"auto","created_at":"2025-04-21 10:44:50","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":513899,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e3a\u003c/strong\u003e: Kaplan-Meier curves showing DRFS in 754 cases by the 21GC group. Among the 754 cases, 353 (46.8%) were classified into the 21GC Low-risk group, 318 (42.2%) into the Intermediate-risk group, and 83 (11.0%) into the High-risk group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3b\u003c/strong\u003e: Kaplan-Meier curves showing DRFS by 95GC groups (High/Low) in 318 cases in the Intermediate risk group with 21GC.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/8211f31b4a6afd6d2e26076a.jpg"},{"id":85686125,"identity":"c18914ea-1364-4f94-b715-aae8023e411b","added_by":"auto","created_at":"2025-06-30 16:03:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3900111,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/ee01fed9-ab91-407b-afaa-105c7b570ae0.pdf"},{"id":81025813,"identity":"de82d8a0-385f-4080-89fc-fde747e651ad","added_by":"auto","created_at":"2025-04-21 10:36:50","extension":"tif","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":97128,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 1\u003c/strong\u003e: The characteristics of the background of the patients included in the analysis.\u003c/p\u003e","description":"","filename":"Table1.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/49500bc62ad7747433ce24a2.tif"},{"id":81027414,"identity":"6cd05911-f82d-4657-a779-2845bb752804","added_by":"auto","created_at":"2025-04-21 10:52:50","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":92364,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 2\u003c/strong\u003e: Multivariate analysis of DRFS in 276 cases included in the real-world data.\u003c/p\u003e","description":"","filename":"Table2.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/8484e793dffd7fcc15138bfc.tif"},{"id":81026897,"identity":"3198344e-8061-4e7a-8bb3-60a26279a2c9","added_by":"auto","created_at":"2025-04-21 10:44:50","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":63840,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 3\u003c/strong\u003e: Evaluation of the discriminative ability of 21GC and 95GC using C-statistics in 832 cases.\u003c/p\u003e","description":"","filename":"Table3.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/f778461c09e03841ddef8dad.tif"},{"id":81025816,"identity":"ff8c7b13-3b00-42d8-b702-3616d41d17c0","added_by":"auto","created_at":"2025-04-21 10:36:50","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":54554,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 4\u003c/strong\u003e: Evaluation of the discriminative ability of 95GC in the 21GC Intermediate-risk group using C-statistics in 318 cases.\u003c/p\u003e","description":"","filename":"Table4.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/6620901a3a883ab4c06ee095.tif"},{"id":81028432,"identity":"5ac4eb02-e27b-4ad3-a7c0-d5f83cfc8e7e","added_by":"auto","created_at":"2025-04-21 11:00:50","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":66988,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Table 1: \u003c/strong\u003eEvaluation of the discriminative ability of incorporating 21GC or 95GC in addition to clinicopathological factors as predictors.\u003c/p\u003e","description":"","filename":"Supp.Table1.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/f7c3a901310bf02fd5878a4b.tif"},{"id":81025826,"identity":"a7635375-535a-44dc-a363-bb13fdf5fa05","added_by":"auto","created_at":"2025-04-21 10:36:50","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":81500,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Table 2: \u003c/strong\u003eReclassification table based on predictive models incorporating 21GC or 95GC and clinicopathological factors.\u003c/p\u003e","description":"","filename":"Supp.Table2.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/958c90b69bc76837ec14000b.tif"},{"id":81025822,"identity":"6d27727e-a567-4509-b665-d7d9abd10346","added_by":"auto","created_at":"2025-04-21 10:36:50","extension":"tif","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":70724,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Table 3: \u003c/strong\u003eDiscriminative ability of the base model + 21GC or the base model + 95GC using C-statistics to predict recurrence at 3 and 5 years.\u003c/p\u003e","description":"","filename":"Supp.Table3.tif","url":"https://assets-eu.researchsquare.com/files/rs-6282865/v1/aa82aac6d0b9101e7bc2f310.tif"}],"financialInterests":"","formattedTitle":"Validation of the predictive ability for recurrence and the clinical utility of the 95-gene classifier (95GC) through an integrated analysis of five studies.","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eMultigene assays utilizing gene expression analysis, developed in recent years, are critical tools in clinical practice for predicting recurrence risk, determining indications for adjuvant chemotherapy, and selecting optimal treatment strategies, particularly for ER-positive, HER2-negative early-stage breast cancer. Several assays, including OncotypeDX\u0026reg; from the U.S. (\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e), MammaPrint\u0026reg; from the Netherlands (\u003cspan additionalcitationids=\"CR8 CR9\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e), and Curebest\u0026trade; 95GC Breast from Japan (\u003cspan additionalcitationids=\"CR12 CR13 CR14 CR15 CR16 CR17 CR18 CR19\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e), have been widely implemented and in use for many years.\u003c/p\u003e \u003cp\u003eAmong these, Curebest\u0026trade; 95GC Breast (abbreviated as 95GC) was developed in 2011 as the first multigene assay in Japan for postoperative recurrence prediction (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). In 2013, assay services started. During the development of 95GC, the data from 549 ER-positive, HER2-negative, and node-negative breast cancer patients who received no therapy or tamoxifen were extracted from the public Gene Expression Omnibus (GEO) database, and comprehensive gene expression analysis in primary tumors was conducted. By comprehensively comparing 167 recurrence cases with 382 non-recurrence cases, the most recurrence-associated gene sets were directly extracted from more than 20,000 human genes. Then, a set of the top 95 genes most strongly correlated with recurrence was identified using a gene lineage- and function-independent approach solely based on correlation coefficients.\u003c/p\u003e \u003cp\u003eFurthermore, by applying a multidimensional mathematical algorithm, Between Group Analysis (BGA), scoring recurrence risk derived from these expression levels enabled stratification into the two risk groups: the High-risk and the Low-risk. Notably, the development of 95GC was characterized by a purely mathematical selection of gene sets without referencing past literature such as PubMed et al. As a result, the 95-gene set, which was identified from over 20,000 human genes, was mainly composed of proliferation-related and transcription-related genes that had rarely been studied previously (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). This 95-gene set shows minimal overlap with existing multigene assays yet was deemed mathematically the most optimal gene set. By adopting BGA, patients can be stratified into two risk groups, the High-risk, and the Low-risk, effectively overcoming the issue of the Intermediate-risk group, which has been a point for Oncotype DX\u0026reg;. Initially, frozen and refrigerated specimens were the main targets for analysis; however, FFPE specimens can now be used (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eValidation studies in Japan and abroad have verified the clinical utility of 95GC. Since 2021, five validation studies have been published: a study by Osaka University (Study 1) (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e), a five-institution domestic study (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e), a multi-institutional study in the U.S. (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e), a seven-institution domestic study (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e), and a study by Hokkaido University (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). Notably, in all five studies, the 5-year DRFS for the low-risk group (luminal type N0 treated with hormone therapy alone) was consistently around 98%, demonstrating the robustness of the predictive ability of 95GC across regions and populations. Based on these findings, 95GC is considered effective for identifying patients classified as Low-risk, who are deemed to have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely.\u003c/p\u003e \u003cp\u003eThis study integrated data from all five validation studies since 2021 to examine 95GC's predictive ability and clinical utility for recurrence.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\n\u003ch3\u003e2 − 1 Patients\u003c/h3\u003e\n\u003cp\u003eThe integrated analysis of the five 95GC studies collected recurrence prognosis and clinicopathological data for 719 cases (real-world data) of luminal type N0 breast cancer treated with adjuvant hormone therapy alone (no extended hormone therapy) from 14 Japanese institutions and 5 U.S. institutions. In response to the demand from 95GC users for a dataset exceeding 1,000 cases, an additional 294 Western patients were included from the (publicly available) GEO database in Study 1 (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e), resulting in a total of 1,013 cases for analysis. While recurrence data was available for all cases, complete clinicopathological data could not be obtained for all cases due to differences in institutional policies of Principal Investigators (PIs) and approvals from the ethical review board (ERB) (Table\u0026nbsp;1).\u003c/p\u003e \u003cp\u003eThis study analyzed all the 719 cases (real-world data) (Fig.\u0026nbsp;1a) and the 1,013 cases (Fig.\u0026nbsp;1b). A unique feature of 95GC is that CEL files containing the expression values of all 20,000 genes in the primary tumor are returned to the user after testing. Among 754 CEL files collected through the 95GC integrated analysis study, 21GC, a proxy analysis for OncotypeDX\u0026reg;, was conducted via Recurrence Online (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://recurrenceonline.com/gene-array/\u003c/span\u003e\u003cspan address=\"https://recurrenceonline.com/gene-array/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). In this study, the Intermediate risk was defined based on the recurrence score (RS) ranging from 11 to 25 in the TAILORx trial (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) (Fig.\u0026nbsp;3a, 3b).\u003c/p\u003e\n\u003ch3\u003e2–2 Statistics analysis\u003c/h3\u003e\n\u003cp\u003eSummary statistics were constructed using frequencies and proportions for categorical data (Table\u0026nbsp;1). The mean and standard deviation (SD) were constructed for continuous variables. DRFS was defined as the time from surgery to the first distant recurrence. DRFS was estimated using the Kaplan-Meier method and compared between groups using the log-rank test for clinicopathological factors, 95GC, and 21GC (Figs.\u0026nbsp;1a-b). Additionally, subgroup analyses were performed in cohorts of under/over age 50, T1, T1 and Grade 1, T1 and Grade 2, T1 and Ki67 Low (Ki67\u0026thinsp;\u0026lt;\u0026thinsp;20%), T2, and Grade 3 (Fig.\u0026nbsp;2a-g).\u003c/p\u003e \u003cp\u003eMultivariate analysis using the Cox proportional hazards models evaluated the 95GC's predictive ability for DRFS, summarized with hazard ratios (HRs) and 95% confidence intervals (CIs) (Table\u0026nbsp;2).\u003c/p\u003e \u003cp\u003eUno's C statistic (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e) was used to evaluate the discriminative ability of 21GC and 95GC to predict recurrence at 3 and 5 years (Tables\u0026nbsp;3\u0026ndash;4). The C-statistic measures the discriminative ability of a model, representing the probability that a randomly selected positive sample is assigned a higher predicted probability than a randomly selected negative sample. A C-statistic of 0.5 indicates that the model performs no better than random classification, while a value of 1.0 represents perfect discrimination. Generally, a C-statistic of 0.7\u0026ndash;0.8 is considered to indicate a \"good\" model, 0.8\u0026ndash;0.9 a \"very good\" model, and values exceeding 0.9 an \"excellent\" model.\u003c/p\u003e \u003cp\u003eAdditionally, we assessed the improvement in predictive performance when 21GC or 95GC were added as predictors to clinicopathological factors, using continuous NRI (Net Reclassification Improvement) and IDI (Integrated Discrimination Improvement) (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e) (Supplementary Tables\u0026nbsp;1\u0026ndash;3). NRI measures the net improvement in risk reclassification, while IDI quantifies the enhancement in a model's discriminative ability by assessing changes in predictive probability between models. All statistical analyses were performed using R (version 4.1.3), and p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e\n\u003ch3\u003e2–3 Ethical Approval\u003c/h3\u003e\n\u003cp\u003eThe integrated analysis of the five 95GC studies was approved by the Kyoto Prefectural University of Medicine ERB (Approval No.: ERB-C-3120-1).\u003c/p\u003e"},{"header":"3. Results","content":"\n\u003ch3\u003e3 − 1 Integrated Analysis Results for 95GC (All Cases)\u003c/h3\u003e\n\u003cp\u003eThe characteristics of the background of the patients included in the analysis are summarized in Table\u0026nbsp;1. Some patient background information was not available for real-world data from the U.S. and public data. The integrated analysis of five studies (14 Japanese institutions\u0026thinsp;+\u0026thinsp;5 U.S. institutions, n\u0026thinsp;=\u0026thinsp;719) showed that 550 cases (76.5%) were classified as the 95GC Low-risk, with a significantly better prognosis than the High-risk group (P\u0026thinsp;\u0026lt;\u0026thinsp;1.00e-12). The 5-year DRFS in the Low-risk group was approximately 98% (Fig.\u0026nbsp;1a).\u003c/p\u003e \u003cp\u003eFor the combined data of 1,013 cases, which included public datasets, 746 cases (73.6%) were classified as the 95GC Low-risk, also showing better prognosis compared to the High-risk group (P\u0026thinsp;\u0026lt;\u0026thinsp;1.00e-12). The 5-year DRFS in the Low-risk group was approximately 97% (Fig.\u0026nbsp;1b).\u003c/p\u003e\n\u003ch3\u003e3 − 2 Subgroup Analysis\u003c/h3\u003e\n\u003cp\u003eSubgroup analyses according to several clinical factors were performed on patients included in real-world data of Japanese for whom information on patient background was available. Their results showed that the 95GC Low-risk group consistently had a better prognosis than the High-risk group, including the following subgroups according to clinical factors:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eUnder age 50 (n\u0026thinsp;=\u0026thinsp;198) / Over age 50 (n\u0026thinsp;=\u0026thinsp;318)\u003c/b\u003e: The Low-risk group showed a better prognosis compared to the High-risk group (P\u0026thinsp;=\u0026thinsp;2.39e-08 / P\u0026thinsp;=\u0026thinsp;9.2e-08, Fig.\u0026nbsp;2a). Additionally, the High-risk group was more prevalent in the under age 50 cohort (29.29%) compared to the over age 50 cohort (21.70%). Moreover, the under age 50 cohort exhibited a trend toward early recurrence, while the over age 50 cohort showed a trend toward late recurrence (Fig.\u0026nbsp;2a).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eT1 (n\u0026thinsp;=\u0026thinsp;329)\u003c/b\u003e: The low-risk group showed a better prognosis than the high-risk group (P\u0026thinsp;=\u0026thinsp;2.12e-12, Fig.\u0026nbsp;2b).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eT1 and Grade 1 (n\u0026thinsp;=\u0026thinsp;92)\u003c/b\u003e: This cohort showed an excellent prognosis, with no recurrences observed (Fig.\u0026nbsp;2c).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eT1 and Grade 2 (n\u0026thinsp;=\u0026thinsp;99)\u003c/b\u003e: The low-risk group showed a better prognosis than the high-risk group (P\u0026thinsp;=\u0026thinsp;0.0125, Fig.\u0026nbsp;2d).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eT1 and Ki67 Low (n\u0026thinsp;=\u0026thinsp;133)\u003c/b\u003e: The low-risk group showed a better prognosis than the high-risk group (P\u0026thinsp;=\u0026thinsp;4.82e-05, Fig.\u0026nbsp;2e).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eT2 (n\u0026thinsp;=\u0026thinsp;187)\u003c/b\u003e: The low-risk group showed a better prognosis than the high-risk group (P\u0026thinsp;=\u0026thinsp;0.000293, Fig.\u0026nbsp;2f).\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eGrade 3 (T1\u0026thinsp;+\u0026thinsp;T2, n\u0026thinsp;=\u0026thinsp;38)\u003c/b\u003e: The low-risk group showed a trend toward better prognosis than the high-risk group (P\u0026thinsp;=\u0026thinsp;0.0576, Fig.\u0026nbsp;2g).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e\n\u003ch3\u003e3–3 Multivariate Analysis\u003c/h3\u003e\n\u003cp\u003eResults of a multivariate analysis of 276 patients included in the real-world data for which all information on the patient background was available, 95GC was shown to be the most independent predictor of recurrence compared to other clinicopathological factors (age, tumor size, pathological grade, PgR, Ki67). Models including age (HR: 0.20, 95%CI: 0.075\u0026ndash;0.52) or excluding age (HR: 0.19, 95%CI: 0.073\u0026ndash;0.51) consistently identified 95GC as the strongest predictor (Table\u0026nbsp;2).\u003c/p\u003e\n\u003ch3\u003e3–4 Integrated Analysis Results for 21GC (Proxy Analysis for Oncotype DX®)\u003c/h3\u003e\n\u003cp\u003e21GC analysis: Recurrence Online was performed on 754 CEL files from the 95GC integrated analysis as a proxy for Oncotype DX\u0026reg; (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). Among the 754 CEL files, 353 cases (46.8%) were classified as the 21GC Low-risk group, 318 cases (42.2%) as the Intermediate-risk group, and 83 cases (11.0%) as the High-risk group. Significant differences in recurrence prognosis among the three groups (P\u0026thinsp;=\u0026thinsp;3.00e-08) were shown. The 5-year DRFS for the Low-risk group was 96.8%, 92.2% for the Intermediate-risk group, and 73.4% for the High-risk group (Fig.\u0026nbsp;3a).\u003c/p\u003e \u003cp\u003eFor the Intermediate-risk group (n\u0026thinsp;=\u0026thinsp;318) classified by 21GC, 95GC successfully stratified these patients into two distinct groups with significant differences in recurrence prognosis (P\u0026thinsp;=\u0026thinsp;5.86e-05). The 5-year DRFS in the 95GC Low-risk group reclassified from the 21GC Intermediate-risk group was 96.3% (Fig.\u0026nbsp;3b).\u003c/p\u003e\n\u003ch3\u003e3–5 Evaluation of Discriminative Ability\u003c/h3\u003e\n\n\u003ch3\u003e3-5-1 Comparison of Discriminative Ability for Recurrence between 95GC and 21GC\u003c/h3\u003e\n\u003cp\u003eThe discriminative ability of 21GC and 95GC, evaluated using C-statistics, is shown in Table\u0026nbsp;3. The analysis included 832 patients with 21GC and 95GC laboratory and prognostic data available. 21GC was used as a predictor for the Low- or High-risk classification, and patients classified as Intermediate-risk by 21GC were treated as low-risk.\u003c/p\u003e \u003cp\u003eThe difference in C-statistics for recurrence prediction between 21GC and 95GC was 0.07 (95% CI: -0.02, 0.15) at 3 years and 0.06 (95% CI: -0.01, 0.13) at 5 years, indicating that the predictive ability of the two assays was comparable.\u003c/p\u003e\n\u003ch3\u003e3-5-2 Discriminative Ability of 95GC in the 21GC Intermediate-Risk Group\u003c/h3\u003e\n\u003cp\u003eThe discriminative ability of 95GC in the 21GC Intermediate-risk group, evaluated using C-statistics, is shown in Table\u0026nbsp;4.\u003c/p\u003e \u003cp\u003eThe analysis included 318 patients classified as Intermediate risk by 21GC, with 21GC and 95GC laboratory and prognostic data available.\u003c/p\u003e \u003cp\u003eThe C-statistics for 95GC was consistently high at both 3 years and 5 years, exceeding or approximating 0.7, indicating that 95GC effectively distinguished between recurrent and non-recurrent patients classified as the Intermediate risk by 21GC.\u003c/p\u003e\n\u003ch3\u003e3-5-3 Discriminative Ability of Models Incorporating 21GC or 95GC\u003c/h3\u003e\n\u003cp\u003eThe discriminative ability of predictive models incorporating 21GC or 95GC in addition to clinicopathological factors is shown in Supplementary Table\u0026nbsp;1. The analysis included 108 patients with complete clinical, laboratory, and prognostic data for 21GC and 95GC. The base model included the following clinicopathological factors: age (under 50 vs. over 50 years), grade (1 vs. 2 or 3), Ki67 status (positive vs. negative), and PgR status (positive vs. negative). 21GC was used as a predictor for the Low- or High-risk classification, and patients classified as Intermediate-risk by 21GC were treated as the Low risk.\u003c/p\u003e \u003cp\u003eThe reclassification table for these models is provided in Supplementary Table\u0026nbsp;2, showing how individuals were reclassified into different risk categories based on the new models. Cells representing whose risk classification improved are highlighted in grey.\u003c/p\u003e \u003cp\u003eAt 3 and 5 years, models incorporating 21GC (Base model\u0026thinsp;+\u0026thinsp;21GC) or 95GC (Base model\u0026thinsp;+\u0026thinsp;95GC) and clinicopathological factors demonstrated improved discrimination and reclassification in Supplementary Table\u0026nbsp;1,2. Improvements were more pronounced at 5 years in both models, with the following results:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eBase model\u0026thinsp;+\u0026thinsp;21GC\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e3-year NRI: 0.14 (95% CI: -0.57, 0.55); IDI: 0.00 (95% CI: -0.11, 0.33)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e5-year NRI: 0.31 (95% CI: -0.49, 0.70); IDI: 0.06 (95% CI: -0.08, 0.44)\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e \u003cb\u003eBase model\u0026thinsp;+\u0026thinsp;95GC\u003c/b\u003e:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003e3-year NRI: 0.36 (95% CI: -0.36, 0.64); IDI: 0.03 (95% CI: -0.04, 0.23)\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003e5-year NRI: 0.41 (95% CI: -0.32, 0.72); IDI: 0.05 (95% CI: -0.04, 0.31)\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eThe discriminative ability of these models for recurrence prediction at 3 and 5 years, evaluated using C-statistics, is shown in Supplementary Table\u0026nbsp;3. The difference in C-statistics between Base model\u0026thinsp;+\u0026thinsp;21GC and Base model\u0026thinsp;+\u0026thinsp;95GC was 0.005 (95% CI: -0.17, 0.18) at 3 years and \u0026minus;\u0026thinsp;0.02 (95% CI: -0.18, 0.15) at 5 years, suggesting that both models had comparable predictive ability for recurrence.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003e95GC is a novel multigene assay that utilizes comprehensive gene expression analysis to identify optimal marker gene mathematically sets for recurrence prediction. It uniquely employs a multidimensional algorithm, Between Group Analysis (BGA), to categorize patients into the High- and Low-risk groups. This study presents an integrated analysis of all five validation studies published since 2021.\u003c/p\u003e \u003cp\u003eIn the analysis of 719 luminal type node negative breast cancer cases (real-world data) treated with hormone therapy alone, the 5-year DRFS in the 95GC low-risk group were approximately 98%. This finding supports the conclusion that \u0026lsquo;The low-risk patients classified by 95GC, who are deemed to have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely.\u0026rsquo; This principle forms the foundation of multigene assays. Similar results were observed in the expanded analysis of 1,013 cases, including public datasets combined in Study 1 (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cb\u003eSubgroup Analysis\u003c/b\u003e \u003c/p\u003e \u003cp\u003eSubgroup analysis highlighted additional insights. Specifically, no recurrences were observed in the T1 and Grade 1 cohort (n\u0026thinsp;=\u0026thinsp;92) (Fig.\u0026nbsp;2c), suggesting that multigene assays may have limited utility in such inherently low-risk populations. However, the sample size of the population belonging to such cohorts in this study was small, and this finding requires further investigation with larger sample sizes.\u003c/p\u003e \u003cp\u003eIn the other subgroup analyses (e.g., T1 (n\u0026thinsp;=\u0026thinsp;329), T1 and Grade 2 (n\u0026thinsp;=\u0026thinsp;99), T1 and Ki67 Low (n\u0026thinsp;=\u0026thinsp;133), T2 (n\u0026thinsp;=\u0026thinsp;187), and Grade 3 (n\u0026thinsp;=\u0026thinsp;38)), the Low-risk group consistently demonstrated significantly better prognosis or a trend toward better prognosis compared to the High-risk group. These findings suggest that multigene assays are valuable tools for guiding decisions on adjuvant chemotherapy for such subgroup.\u003c/p\u003e\n\u003ch3\u003e21GC (Proxy Analysis for Oncotype DX®)\u003c/h3\u003e\n\u003cp\u003e95GC provides users with CEL files containing comprehensive gene expression data as CD data after the assay. While these files are intended for research purposes only, they allow the calculation of other multigene assays as proxy analyses. Communicating 21GC results obtained through such analyses to patients is strictly prohibited.\u003c/p\u003e \u003cp\u003eIn this study, a total of 754 CEL files obtained from the 95GC integrated analysis were used for a proxy analysis of 21GC. 21GC has an established track record as a proxy for OncotypeDX\u0026reg; (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). The result revealed significant differences in recurrence prognosis among the three groups (Fig.\u0026nbsp;3a). The 5-year DRFS for the Low-risk group was 96.8%, indicating that chemotherapy could be omitted with relative safety. Conversely, the 5-year DRFS for the Intermediate-risk group was 92.2%, slightly worse than that of the Low-risk group, leading to uncertainty regarding the appropriateness of adjuvant chemotherapy.\u003c/p\u003e \u003cp\u003e \u003cb\u003eImplications for Real-World Data (RWD)\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe worse prognosis of the Intermediate-risk group (RS 11\u0026ndash;25) in this study compared to the TAILORx trial (RS 11\u0026ndash;25) (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) is hypothesized to be due to differences in tumor size within the cohorts.\u003c/p\u003e \u003cp\u003eIn this integrated analysis study, approximately 36% of cases were classified as T2 or larger tumors (limited to cases with available data). In contrast, the TAILORx trial predominantly included T1 tumors with a median tumor size of approximately 1.5 cm. We would like to consider which cohorts more accurately reflect real-world data (RWD) in clinical practice.\u003c/p\u003e \u003cp\u003eAccording to the report of 95,870 Japanese Breast Cancer by Kubo et al. (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e), approximately 37% of breast cancer cases were classified as T2 or larger. This suggests that the cohort in this integrated analysis study more closely reflects RWD than the TAILORx trial.\u003c/p\u003e \u003cp\u003eFurther evidence supporting the reflection of RWD in this integrated analysis study is presented in Fig.\u0026nbsp;3a. Here, the proportion of T2 or larger tumors (approximately 37%) closely replicates the B14 cohort validation of OncotypeDX\u0026reg; (T2 or larger: 38% (254/668)) (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). Over the past two decades, the definition of the Intermediate-risk group has shifted from RS 18\u0026ndash;30 to RS 11\u0026ndash;25. As a result, this study shows a slightly improved prognosis curve for the 21GC Intermediate-risk group compared to the B14 cohort validation. Nevertheless, the prognosis for the 21GC Intermediate-risk group (RS 11\u0026ndash;25) remains relatively worse than that of the Low-risk group, leading to challenges in determining the appropriateness of adjuvant chemotherapy.\u003c/p\u003e \u003cp\u003eTherefore, while the results of the TAILORx trial are highly applicable to T1 cohorts, they should be interpreted with caution when applied to real-world clinical practice (RWD), where approximately 37% of cases involve T2 tumors.\u003c/p\u003e \u003cp\u003eIn other words, determining the appropriateness of adjuvant chemotherapy for the 21-gene assay (21GC) Intermediate-risk group (RS 11\u0026ndash;25) remains challenging. Relying solely on the TAILORx trial, which predominantly included T1 cases, may lead to an excessive omission of adjuvant chemotherapy in this group. To optimize decision-making, it is beneficial to incorporate clinical and pathological factors and other relevant genetic markers.\u003c/p\u003e \u003cp\u003e \u003cb\u003eClinical Implications of 95GC in the 21GC Intermediate-Risk Group\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo assess the appropriateness of adjuvant chemotherapy for the Intermediate-risk group, 95GC was applied to 318 cases classified as the 21GC Intermediate-risk group in this study. The results demonstrated that 95GC significantly stratified the recurrence prognosis of the 21GC Intermediate-risk group into two distinct groups: the High-risk and the Low-risk (Fig.\u0026nbsp;3b). The 5-year DRFS in the 95GC Low-risk group was 96.3%, indicating favorable outcomes with hormone therapy alone, without adjuvant chemotherapy. Furthermore, the 10-year DRFS in the 95GC Low-risk group was 81.4%, suggesting that extended hormone therapy may be a good option for this group. Thus, for patients in the 21GC Intermediate-risk group, where treatment decisions are challenging, 95GC appears to be a valuable tool for further stratification.\u003c/p\u003e \u003cp\u003e \u003cb\u003eChallenges in Randomized Controlled Trials (RCTs)\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThere is public interest in conducting a randomized controlled trial (RCT) for 95GC; however, conducting an RCT in the High-risk group for multigene assays that classify patients into only the two groups (the High-risk and the Low-risk) without an Intermediate-risk category raises ethical concerns.\u003c/p\u003e \u003cp\u003eIt should be noted that OncotypeDX\u0026reg; conducted an RCT exclusively in the Intermediate-risk group (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). As a result, the additional benefit of adjuvant chemotherapy for High-risk patients has only been demonstrated through retrospective studies rather than RCTs. Both OncotypeDX\u0026reg; and 95GC have provided evidence of the additional benefit of adjuvant chemotherapy for High-risk patients based on retrospective analyses (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). Therefore, neither OncotypeDX\u0026reg; nor 95GC functions as a companion diagnostic marker for specific drugs.\u003c/p\u003e \u003cp\u003eIn the past, MammaPrint\u0026reg;, a two-group multigene assay that classifies patients into High- and Low-risk categories, was evaluated in a randomized controlled trial (RCT). In the MINDACT trial, patients were stratified based on both Clinical and Genomic risk assessments, resulting in four groups, including the Clinical High/Genomic Low and Clinical Low/Genomic High groups, to compare recurrence outcomes (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). MammaPrint\u0026reg; may have adopted this trial design because, unlike Oncotype DX, it does not have an Intermediate-risk group that would serve as a suitable target for an RCT.\u003c/p\u003e \u003cp\u003eThe results showed no significant difference in prognosis between patients who received chemotherapy and those who did not in the Clinical High/Genomic Low group (P\u0026thinsp;=\u0026thinsp;0.27). Similarly, no significant difference was observed in the Clinical Low/Genomic High group (P\u0026thinsp;=\u0026thinsp;0.66) (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). These findings remained consistent in subsequent follow-up investigations (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn other words, neither the Clinical High/Genomic Low group nor the Clinical Low/Genomic High group demonstrated a clear additional benefit from chemotherapy. Consequently, the MINDACT trial provided insights into the complexities of designing RCTs for multigene assays within a binary (High-/Low-risk) classification framework.\u003c/p\u003e \u003cp\u003eTherefore, there are no plans to conduct an RCT with a similar design for 95GC. In multigene assays, the core principle is that \u0026lsquo;the Low-risk patients have a sufficiently favorable prognosis with hormone therapy alone, allowing them to omit chemotherapy safely.\u0026rsquo; This study demonstrated this principle through an integrated analysis of real-world data.\u003c/p\u003e \u003cp\u003e \u003cb\u003eApplication of 95GC in Luminal Type Node Positive Breast Cancer\u003c/b\u003e \u003c/p\u003e \u003cp\u003eFinally, the analysis of 95GC in luminal type node positive breast cancer is introduced. Matsumoto et al. conducted a computer-based simulation analysis using public datasets to simulate the OncotypeDX\u0026reg; RxPONDER trial. They applied 95GC to a cohort of luminal type node positive breast cancer with RS 0\u0026ndash;25 (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). Their results identified a poor prognosis group requiring chemotherapy in postmenopausal patients and, conversely, a favorable prognosis group unlikely to benefit from chemotherapy in premenopausal patients.\u003c/p\u003e \u003cp\u003eAccording to the TAILORx trial, the RS 0\u0026ndash;25 group includes the majority (85.7%) of luminal type breast cancer cases. This indicates that the RS 0\u0026ndash;25 cohort is a heterogeneous population comprising patients with both favorable and poor prognoses.\u003c/p\u003e \u003cp\u003eTherefore, treatment strategies for luminal type node positive breast cancer with RS 0\u0026ndash;25 should not rely solely on a clinical trial result based on menopausal status or age. Instead, recurrence biomarkers such as 95GC should be considered as a guide. Even for luminal type node positive breast cancer with RS 0\u0026ndash;25, where treatment decisions are challenging, 95GC appears to be a valuable tool.\u003c/p\u003e \u003cp\u003e \u003cb\u003eFuture Perspectives\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThis study demonstrated, through the integration of multicenter data, that the prognosis of the 95GC Low-risk group is sufficiently favorable with hormone therapy alone, allowing them to omit chemotherapy safely. We aim to continue accumulating more cases to validate these findings further.\u003c/p\u003e \u003cp\u003eFinally, Curebest\u0026trade; 95GC Breast has been developed not only as a clinically useful test but also as a novel diagnostic tool with the potential to build a research database for the future. Unlike existing assays such as Oncotype DX\u0026reg; and MammaPrint\u0026reg;, which measure the expression of dozens of genes, 95GC simultaneously quantifies the expression of approximately 20,000 human genes. It provides this data to users in the form of CEL files.\u003c/p\u003e \u003cp\u003eBy utilizing these CEL files (as demonstrated in this study with 21GC), it becomes feasible to perform analyses for additional multigene assays enabling personalized medicine, such as 23GC for predicting chemotherapy sensitivity (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e), 42GC for predicting late recurrence (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e), and 155GC for predicting post-chemotherapy recurrence prognosis (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn 2021, a groundbreaking research service called 'GC Note' was launched, marking the first globally available service for cloud storage of CEL files collected nationwide after testing. The comprehensive database of whole-gene expression profiles enables the measurement and development of new multigene assays through its utilization.\u003c/p\u003e \u003cp\u003eImplementing 95GC is expected to enable optimal personalized medicine tailored to individuals and accelerate biomarker research in Japan. However, further evidence accumulation is necessary.\u003c/p\u003e "},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e95GC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eCurebest\u0026trade; 95GC Breast\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e21GC\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eOncotype DX\u0026reg;\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003e21GC score\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003e21GC Recurrence score\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eER\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eestrogen receptor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePgR\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eprogesterone receptor\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eHER2\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ehuman epidermal growth factor receptor type2\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eMGA\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003emultigene assay\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eDRFS\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003edistant recurrence-free survival\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eY.N. and A.I. designed and take the lead in this experiment.\u003c/p\u003e\n\u003cp\u003eY.H. and S.M. performed data analysis. N.U. helped supervise this experiment.\u003c/p\u003e\n\u003cp\u003eAll the other authors performed data collection, data analysis and contributed to sample preparation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eFunding\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe authors declare that no funds, grants, or other supports were received during the preparation of this manuscript.\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe data analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003eConflict of interest\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eYasuto Naoi\u003c/strong\u003e has received research funding from Sysmex, ONO, Daiichi-Sankyo and AstraZeneca, and honoraria from AstraZeneca, Pfizer, Eli Lilly, Daiichi-Sankyo and Chugai outside the submitted work; he holds joint patents with Sysmex including Curebest\u0026trade; 95GC Breast (JP.5725274.B2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eEthical approval\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Kyoto Prefectural University of Medicine Ethics Committee（ERB-C-3120-1）.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003ePaik S, Shak S, Tang G, Kim C, Baker J, Cronin M, et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med. 2004;351(27):2817\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaik S, Tang G, Shak S, Kim C, Baker J, Kim W, et al. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol. 2006;24(23):3726\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSparano JA, Gray RJ, Makower DF, Pritchard KI, Albain KS, Hayes DF, et al. Prospective validation of a 21-gene expression assay in breast cancer. N Engl J Med. 2015;373(21):2005\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSparano JA, Gray RJ, Makower DF, Pritchard KI, Albain KS, Hayes DF, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. N Engl J Med. 2018;379(2):111\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSparano JA, Gray RJ, Makower DF, Albain KS, Saphner TJ, Badve SS, et al. Clinical outcomes in early breast cancer with a high 21-gene recurrence score of 26 to 100 assigned to adjuvant chemotherapy plus endocrine therapy: a secondary analysis of the TAILORx randomized clinical trial. JAMA Oncol. 2020;6(3):367\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalinsky K, Barlow WE, Gralow JR, Meric-Bernstam F, Albain KS, Hayes DF, et al. 21-Gene assay to inform chemotherapy benefit in node-positive breast cancer. N Engl J Med. 2021;385(25):2336\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan de Vijver MJ, He YD, van 't Veer LJ, Dai H, Hart AA, Voskuil DW, Schreiber GJ, Peterse JL, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347(25):1999\u0026ndash;2009.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan 't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, Peterse HL, van der Kooy K, et al. Gene expression profiling predicts the clinical outcome of breast cancer. Nature. 2002;415(6871):530\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCardoso F, van't Veer LJ, Bogaerts J, Slaets L, Viale G, Delaloge S, et al. 70-gene signature as an aid to treatment decisions in early-stage breast cancer. N Engl J Med. 2016;375(8):717\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePiccart M, van 't Veer LJ, Poncet C, Cardozo JM, Delaloge S, Pierga JY, et al. 70-gene signature as an aid for treatment decisions in early breast cancer: updated results of the phase 3 randomized MINDACT trial with an exploratory analysis by age. Lancet Oncol. 2021;22(4):476\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Development of 95-gene classifier as a powerful predictor of recurrences in node-negative and ER-positive breast cancer patients. Breast Cancer Res Treat. 2011;128(3):633\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Comparison of efficacy of 95-gene and 21-gene classifiers (Oncotype DX) for prediction of recurrence in ER-positive and node-negative breast cancer patients. Breast Cancer Res Treat. 2013;140(2):299\u0026ndash;306.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Kishi K, Tsunashima R, Nakayama T, Maruyama N, Shimazu K, et al. Multi-gene classifiers for prediction of recurrence in breast cancer patients. Breast cancer (Tokyo Japan). 2016;23(1):12\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Saito Y, Kishi K, Shimoda M, Kagara N, Miyake T, et al. Development of recurrence risk score using a 95-gene classifier and its application to formalin-fixed paraffin-embedded tissues in ER-positive, HER2-negative, and node-negative breast cancer. Oncol Rep. 2019;42(6):2680\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Tsunashima R, Shimazu K, Noguchi S, et al. The multigene classifiers 95GC/42GC/155GC for precision medicine in ER-positive HER2-negative early breast cancer. Cancer Sci. 2021;112(4):1369\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTsukamoto F, Arihiro K, Takahashi M, Ito K, Ohsumi S, Takashima S, et al. Multicenter retrospective study on the use of Curebest\u0026trade; 95GC Breast for estrogen receptor-positive and node-negative early breast cancer. BMC Cancer. 2021;21(1):1077.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFujii T, Masuda H, Cheng YC, Yang F, Sahin AA, Naoi Y, et al. A 95-gene signature stratifies recurrence risk of invasive disease in ER-positive, HER2-negative, node-negative breast cancer with intermediate 21-gene signature recurrence scores. Breast Cancer Res Treat. 2021;189(2):455\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNaoi Y, Tsunashima R, Shimazu K, Oikawa M, Imanishi S, Koyama H, et al. Validation of the prognosis of patients with ER-positive, HER2-negative, and node-negative invasive breast cancer classified as low risk by Curebest\u0026trade; 95GC Breast in a multi-institutional registry study. Oncol Lett. 2023;25(5):209.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYamashita H, Hatanaka KC, Yamagishi K, Saito Y, Hamasaki K, Taniguchi M, et al. Evaluation of a 95-Gene Classifier of formalin-fixed paraffin-embedded tissues in ER-positive, HER2-negative, and node-negative breast cancer. Anticancer Res. 2023;43(2):707\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMatsumoto S, Tsunashima R, Kitano S, Watanabe A, Kato C, Morita M, et al. Multi-gene assay 95- and 155-gene classifiers for prognosis prediction and chemotherapy omission in lymph node-positive luminal-type breast cancer. Cancer Treat Res Commun. 2023;36:100711.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGyorffy B, Hatzis C, Sanft T, Hofstatter E, Aktas B, Pusztai L. RecurrenceOnline: an online analysis tool to determine breast cancer recurrence and hormone receptor status using microarray data. Breast Cancer Res Treat. 2012;132(3):1025\u0026ndash;34.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUno H, Cai T, Pencina MJ, D'Agostino RB, Wei LJ. On the C-statistics for evaluating overall adequacy of risk prediction procedures with censored survival data. Stat Med. 2011;30(10):1105\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUno H, Tian L, Cai T, Kohane IS, Wei LJ. A unified inference procedure for a class of measures to assess improvement in risk prediction systems with survival data. Stat Med. 2012;31(25):2579\u0026ndash;89.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKubo M, Kumamaru H, Isozumi U, Miyashita M, Nagahashi M, Kadoya T et al. Annual report of the Japanese Breast Cancer Society registry for 2016. Breast Cancer. 2020;27(4):511\u0026ndash;518.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSota Y, Naoi Y, Tsunashima R, Kagara N, Shimazu K, Maruyama N, et al. Construction of a novel immune-related signature for prediction of pathological complete response to neoadjuvant chemotherapy in human breast cancer. Ann Oncol. 2014;25(1):100\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTsunashima R, Naoi Y, Shimazu K, Kagara N, Shimoda M, Tanei T, et al. Construction of a novel multi-gene assay (42-gene classifier) for prediction of late recurrence in ER-positive breast cancer patients. Breast Cancer Res Treat. 2018;171(1):33\u0026ndash;41.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTsunashima R, Naoi Y, Kagara N, Shimoda M, Shimomura A, Maruyama N, et al. Construction of a multi-gene classifier for prediction of response to and prognosis after neoadjuvant chemotherapy for estrogen receptor-positive breast cancers. Cancer Lett. 2015;365(2):166\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 4 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"breast-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"brca","sideBox":"Learn more about [Breast Cancer](http://link.springer.com/journal/12282)","snPcode":"12282","submissionUrl":"https://www.editorialmanager.com/brca/default2.aspx","title":"Breast Cancer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"luminal type breast cancer, Curebest™ 95GC Breast, 21GC, multigene assay, recurrence risk prediction","lastPublishedDoi":"10.21203/rs.3.rs-6282865/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6282865/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eIn recent years, multigene assays have become indispensable tools for predicting the recurrence risk of estrogen receptor (ER)-positive, human epidermal growth factor receptor 2 (HER2)-negative early-stage breast cancer and guiding adjuvant chemotherapy decisions. Curebest\u0026trade; 95GC Breast (95GC), developed in 2011 as a domestically produced multigene assay for postoperative recurrence prediction, has been commercially available since 2013. Since 2021, five validation studies evaluating the predictive performance 95GC have been published. This study presents an integrated analysis of these studies to validate the prognostic utility of 95GC further.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThe integrated analysis included 719 real-world cases of luminal-type node-negative breast cancer patients who underwent adjuvant hormone therapy alone without extended endocrine treatment. Additionally, an expanded cohort incorporating 294 cases from Western patients within the GEO public database was analyzed, resulting in a total of 1,013 cases.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eAmong the 719 real-world cases, 550 (76.5%) were classified into the 95GC Low-risk group, demonstrating a significantly superior prognosis compared to the High-risk group (P\u0026thinsp;\u0026lt;\u0026thinsp;1.00e-12). The 5-year distant recurrence-free survival (DRFS) rate in the Low-risk group was approximately 98%, with consistent findings observed in the expanded cohort. Furthermore, an analysis of 754 CEL files using 21GC (a proxy for Oncotype DX\u0026reg;) identified 318 cases (42.2%) as the 21GC intermediate risk. 95GC successfully stratified these cases into two prognostically distinct subgroups.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eThese findings underscore the clinical utility of 95GC in safely omitting chemotherapy for Low-risk patients with good prognosis and in further stratifying the 21GC Intermediate-risk cases, thereby contributing to personalized treatment strategies.\u003c/p\u003e","manuscriptTitle":"Validation of the predictive ability for recurrence and the clinical utility of the 95-gene classifier (95GC) through an integrated analysis of five studies.","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-21 10:36:45","doi":"10.21203/rs.3.rs-6282865/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Minor Revision","date":"2025-05-14T07:27:17+00:00","index":"","fulltext":""},{"type":"reviewerAgreed","content":"","date":"2025-04-01T10:45:18+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-03-31T01:35:33+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-03-27T06:09:19+00:00","index":"","fulltext":""},{"type":"submitted","content":"Breast Cancer","date":"2025-03-24T06:42:57+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"breast-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"brca","sideBox":"Learn more about [Breast Cancer](http://link.springer.com/journal/12282)","snPcode":"12282","submissionUrl":"https://www.editorialmanager.com/brca/default2.aspx","title":"Breast Cancer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"73dc6892-1bfe-456c-a82c-3d8b4350527d","owner":[],"postedDate":"April 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-06-30T16:00:01+00:00","versionOfRecord":{"articleIdentity":"rs-6282865","link":"https://doi.org/10.1007/s12282-025-01734-2","journal":{"identity":"breast-cancer","isVorOnly":false,"title":"Breast Cancer"},"publishedOn":"2025-06-25 15:57:23","publishedOnDateReadable":"June 25th, 2025"},"versionCreatedAt":"2025-04-21 10:36:45","video":"","vorDoi":"10.1007/s12282-025-01734-2","vorDoiUrl":"https://doi.org/10.1007/s12282-025-01734-2","workflowStages":[]},"version":"v1","identity":"rs-6282865","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6282865","identity":"rs-6282865","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}