The FINE Index: A Frailty‐ and Immunonutrition‐Based Nomogram for Survival in Geriatric Breast Cancer: A Retrospective Study Authors: Xiwen Yang, Jiang Zhu Published: 2026 ABSTRACT ABSTRACTBackgroundElderly breast cancer patients generally have a poorer prognosis. This study developed and validated the FINE Index, a nomogram integrating frailty, immunonutritional, and tumor indicators to predict overall survival (OS) in elderly patients undergoing surgery.MethodsWe retrospectively analyzed 1621 elderly breast cancer patients from West China Hospital, Sichuan University, who were randomly assigned to training (n = 1132) and internal validation (n = 489) cohorts. An additional 545 patients from Chengdu Shang Jin Nan Fu Hospital were included for external validation. Candidate variables were selected using univariate analysis, LASSO regression, and multivariate Cox regression. Model performance was evaluated using the C‐index, time‐dependent ROC curves, calibration plots, and decision curve analysis (DCA).ResultsThe FINE Index incorporated five independent predictors: age, prognostic nutritional index (PNI), TNM stage, modified five item frailty index (mFI‐5), and endocrine therapy. The C‐index values were 0.767, 0.739, and 0.731 in the training, internal validation, and external validation cohorts, respectively. Time‐dependent AUCs for 3‐, 5‐, and 10‐year OS exceeded 0.7, with good calibration and greater clinical utility than TNM staging alone. Exploratory analysis suggested an association between chemotherapy and improved OS among high‐risk patients identified by the FINE Index; however, this finding should be considered hypothesis‐generating due to the retrospective design.ConclusionsThe FINE Index is a prognostic nomogram integrating frailty, immunonutritional status, and tumor characteristics to estimate OS in elderly patients with breast cancer. The model provides a more individualized and accurate prognostic assessment, offering valuable guidance for clinical decision‐making and patient management.Keywords: elderly breast cancer, frailty, immunonutritional status, nomogram, prognosis Introduction Breast cancer is the most prevalent malignancy among women, and advanced age represents a major risk factor for its onset. In China, women aged 60 years and older account for approximately one‐third of all breast cancer cases, and this proportion is projected to increase to 41.4% by 2030 [1]. However, clinical research specifically targeting elderly patients with breast cancer remains limited. Given their distinctive physiological characteristics, careful evaluation of the risks and benefits of anticancer therapies is essential. Directly extrapolating findings from studies on younger, healthier populations to elderly patients may lead to suboptimal clinical decisions. Surgical intervention remains one of the principal treatment modalities for early and locally advanced breast cancer [2]. Breast cancer in elderly individuals is a highly heterogeneous disease. Compared with younger patients, tumors in this group often display less aggressive biological behavior [3]. However, the coexistence of multiple comorbidities, impaired nutritional status, and compromised immune function frequently complicate treatment and worsen outcomes [4, 5]. As a result, the prognosis in elderly breast cancer patients is influenced by both tumor‐related and nontumour factors. The traditional TNM staging system established by the American Joint Committee on Cancer (AJCC) remains a cornerstone for prognostic evaluation and treatment planning. However, TNM staging primarily reflects the anatomical extent of the tumor and does not adequately capture the biological and physiological heterogeneity of elderly patients. In older women following surgical treatment, survival may be strongly modified by frailty status, nutritional reserve, systemic inflammatory response, and the ability to tolerate perioperative stress and postoperative treatment. Frailty is often accompanied by chronic inflammation, impaired nutritional intake, and diminished physiological reserve [6, 7, 8]. Malnutrition may aggravate immune dysfunction and frailty, whereas systemic inflammation can contribute to both nutritional decline and functional deterioration [9, 10]. Integrating these markers may offer a more comprehensive assessment of prognosis than relying on TNM staging alone does. However, the prognostic value of combining frailty, nutritional, and inflammatory indicators in elderly patients with breast cancer following surgical treatment has not been fully clarified. Existing prognostic tools for older breast cancer patients focus mainly on tumor characteristics and conventional clinical variables, whereas host‐related geriatric factors are often insufficiently incorporated. Therefore, this study aimed to develop and validate a prognostic model integrating preoperative inflammatory, nutritional, and frailty biomarkers to predict OS in elderly breast cancer patients who underwent surgery, offering a practical tool for personalized prognostic assessment and treatment optimization. Materials and Methods Study Population Breast cancer patients who underwent surgical treatment at West China Hospital from January 2010 to December 2019 were screened. The data between January 2010 and December 2019 was retrospectively analyzed on December 2025. The inclusion criteria were as follows: (1) age ≥ 60 years; (2) histologically confirmed invasive breast cancer; (3) surgical treatment for breast cancer; (4) no distant metastasis; and (5) complete follow‐up information and clinical data availability. The exclusion criteria were as follows: (1) patients who received chemotherapy or other types of radiotherapy before surgery; (2) patients with a history of other malignancies; (3) patients with distant metastases at diagnosis; and (4) patients with missing clinical data or follow‐up information. The screening process is depicted in Figure S1. Patients from Chengdu Shang Jin Nan Fu Hospital constituted a regional external validation cohort. Although this cohort was obtained from a separate institution and was not involved in model development, the two institutions are geographically proximate and operate within a similar regional healthcare environment. Accordingly, this cohort was used to evaluate regional reproducibility rather than broad geographic transportability. Data Collection and Definition of Variables Comprehensive baseline demographic information, clinical parameters, laboratory tests and anthropometric measurements were collected for all enrolled patients. These included age, tumor site, education level, histological grade, histologic type, estrogen receptor (ER) status, progesterone receptor (PR) status, human epidermal growth factor receptor 2 (HER2) status, Ki‐67 index, T stage, N stage, molecular subtype, TNM stage based on the latest AJCC staging system (8th edition) [11], breast and axillary surgery type, postoperative adjuvant therapy (including chemotherapy, radiotherapy, endocrine therapy, and targeted therapy), biochemical parameters and anthropometric measures such as height and weight. All the data were obtained from the electronic medical records system. Frailty was defined using the modified five‐item frailty index (mFI‐5) developed by the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) [12]. The mFI‐5 includes diabetes, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD) or pneumonia, hypertension requiring medication and dependent functional status. Biomarker calculations were as follows: The prognostic nutritional index (PNI) was calculated as albumin (g/L) + 5 × lymphocyte count, and the neutrophil‐to‐lymphocyte ratio (NLR) was calculated as neutrophil count divided by lymphocyte count. Patient Follow‐Up Follow‐up data were collected from inpatient and outpatient medical records and telephone consultations. Patients were followed up every 3 months for the first 2 years after surgery and every 6 months thereafter. The primary endpoint of this study was OS, defined as the duration from the date of surgery to all‐cause death or the last follow‐up date. Statistical Analysis Statistical analyses and plotting were performed using R 4.4.2 and SPSS. Normally distributed continuous variables are presented as x¯ ± s, while nonnormally distributed variables are presented as medians. Between‐group comparisons were conducted using the independent sample t‐test or Mann–Whitney U test, and within‐group comparisons were performed using the paired t‐test or repeated‐measures analysis of variance. Categorical variables were compared using the chi‐square test or Fisher's exact test. A two‐sided p < 0.05 was considered statistically significant. Patients with missing values for the variables required for model development or with incomplete follow‐up information were excluded before cohort formation. No missing‐value imputation was performed, and all analyses were conducted using complete cases. The numbers of patients excluded and the corresponding reasons are presented in Figure S1. Multicollinearity was assessed using the variance inflation factor (VIF), and variables with VIF > 5.0 were excluded from the final model. Univariate Cox regression was first used to calculate the hazard ratio (HR) and 95% confidence interval (CI) for each risk factor. Significant variables were then included in multivariate Cox regression analysis based on the Akaike information criterion. Least absolute shrinkage and selection operator (LASSO) regression was also performed as a sensitivity analysis for variable selection. Independent predictors identified by multivariate Cox regression were used to construct a nomogram for predicting 3‐, 5‐, and 10‐year overall survival. Kaplan–Meier survival curves were plotted. For risk stratification, the optimal cut‐off value of the FINE Index was determined only in the training cohort using the maximum Youden index for 5‐year OS. This pre‐specified cut‐off was then fixed and applied directly to the internal validation and external regional validation cohorts. Model discrimination was assessed using the C‐index and receiver operating characteristic (ROC) curves. The net reclassification index (NRI) and integrated discrimination improvement index (IDI) were used to compare the predictive performance of the nomogram with that of the traditional TNM staging system. Calibration plots were used to assess agreement between predicted and observed outcomes, and decision curve analysis was performed to evaluate clinical benefit. The nomogram was incorporated into an online calculator, available at https://elderlybcnomogram.shinyapps.io/DynNomapp/. Detailed instructions for calculator use are provided in Supporting Information (Figures S6 and S7). Results According to the inclusion and exclusion criteria, 1621 elderly breast cancer patients from West China Hospital, Sichuan University, were included and randomly divided into a training cohort (n = 1132) and an internal validation cohort (n = 489). Baseline demographic and clinicopathological characteristics are shown in Table 1. The mean age was 65.62 years. Most patients underwent mastectomy (79.09%) and axillary lymph node dissection (75.88%). Luminal B was the most common molecular subtype (37.57%), followed by luminal A (34.05%), triple‐negative breast cancer (15.55%), and HER2‐positive disease (12.83%). A total of 781 patients (36.06%) received postoperative chemotherapy. Most patients had Stage I or II disease, accounting for 40.41% and 43.80%, respectively. Regarding frailty status, 61.38% had no frailty, 27.82% had mild frailty, and 10.80% had severe frailty. No significant differences were observed between the training and internal validation cohorts (all p ≥ 0.05). An external validation cohort of 545 patients was also included. Compared with the training cohort, the external validation cohort showed significant differences in several baseline variables, including age, BMI, NLR, PNI, education level, surgical approach, histological grade, molecular subtype, and targeted therapy use (all p < 0.05). TABLE 1Demographic and clinical characteristics of elderly breast cancer patients.CharacteristicsTraining cohort (n = 1132)Internal validation cohort (n = 489)Total (n = 1621)External validation cohort (n = 545)T VS IVT VS EV p p Age, years65.72 ± 5.1365.39 ± 5.0165.62 ± 5.1066.35 ± 5.280.2240.021BMI, kg/m2 23.87 ± 3.3623.90 ± 3.1723.88 ± 3.3024.30 ± 3.380.8350.013NLR1.93 ± 0.581.95 ± 0.661.94 ± 0.611.87 ± 0.640.6870.039PNI52.28 ± 2.7952.20 ± 2.0852.92 ± 2.7752.85 ± 2.800.578< 0.001Education level, n (%)Less than high school653 (57.69%)300 (61.35%)953 (58.79%)283 (51.93%)0.2490.004High school290 (25.62%)122 (24.95%)412 (25.42%)182 (33.39%)College or higher189 (16.70%)67 (13.70%)256 (15.79%)80 (14.68%)Tumor site, n (%)0.6710.978Left626 (55.30%)276 (56.44%)902 (55.64%)301 (55.23%)Right506 (44.70%)213 (43.56%)719 (44.36%)244 (44.77%)Breast surgery type, n (%)0.6190.048Lumpectomy233 (20.58%)106 (21.68%)339 (20.91%)90 (16.51%)Mastectomy899 (79.42%)383 (78.32%)1282 (79.09%)455 (83.49%)Axillary surgery type, n (%)0.8140.112ALND863 (76.24%)367 (75.05%)1230 (75.88%)440 (80.73%)SLNB254 (22.44%)114 (23.31%)368 (22.70%)100 (18.35%)Without axillary surgery15 (1.33%)8 (1.64%)23 (1.42%)5 (0.92%)T stage, n (%)0.5120.742T1631 (55.74%)260 (53.17%)891 (54.97%)297 (54.50%)T2424 (37.46%)198 (40.49%)622 (38.37%)214 (39.27%)T377 (6.80%)31 (6.34%)108 (6.66%)34 (6.24%)N stage, n (%)0.7090.722N0787 (69.52%)327 (66.87%)1114 (68.72%)365 (66.97%)N1179 (15.81%)87 (17.79%)266 (16.41%)95 (17.43%)N299 (8.75%)43 (8.79%)142 (8.76%)53 (9.72%)N367 (5.92%)32 (6.54%)99 (6.11%)32 (5.87%)TNM stage, n (%)0.3800.463I470 (41.52%)185 (37.83%)655 (40.41%)209 (38.35%)II486 (42.93%)224 (45.81%)710 (43.80%)246 (45.14%)III176 (15.55%)80 (16.36%)256 (15.79%)90 (16.51%)mFI‐5, n (%)0.8900.295No‐frailty694 (61.31%)301 (61.55%)995 (61.38%)347 (63.67%)Mild‐frailty318 (28.09%)133 (27.20%)451 (27.82%)134 (24.59%)Severe‐frailty120 (10.60%)55 (11.25%)175 (10.80%)64 (11.74%)Histological grade, n (%)0.4440.473Grade I42 (3.71%)20 (4.09%)62 (3.82%)25 (4.59%)Grade II562 (49.65%)226 (46.22%)788 (48.61%)256 (46.97%)Grade III528 (46.64%)243 (49.69%)771 (47.56%)264 (48.44%)Histologic type, n (%)0.175< 0.001IDC1106 (97.70%)472 (96.52%)1578 (97.35%)512 (93.94%)ILC or others26 (2.30%)17 (3.48%)43 (2.65%)33 (6.06%)ER, n (%)0.1020.534Negative367 (32.42%)179 (36.61%)546 (33.68%)185 (33.94%)Positive765 (67.58%)310 (63.39%)1075 (66.32%)360 (66.06%)PR, n (%)0.5940.724Negative424 (37.46%)190 (38.85%)614 (37.88%)209 (38.35%)Positive708 (62.54%)299 (61.15%)1007 (62.12%)336 (61.65%)HER‐2, n (%)0.611< 0.001Negative831 (73.41%)353 (72.19%)1184 (73.04%)350 (64.22%)Positive301 (26.59%)136 (27.81%)437 (26.96%)195 (35.78%)Ki670.9720.838≤ 14%492 (43.46%)213 (43.56%)705 (43.49%)234 (42.94%)> 14%640 (56.54%)276 (56.44%)916 (56.51%)311 (57.06%)Molecular subtype, n (%)0.6550.016Luminal A390 (34.45%)162 (33.13%)552 (34.05%)157 (28.81%)Luminal B431 (38.07%)178 (36.40%)609 (37.57%)229 (42.02%)HER‐2139 (12.28%)69 (14.11%)208 (12.83%)88 (16.15%)TNBC172 (15.19%)80 (16.36%)252 (15.55%)71 (13.03%)Chemotherapy, n (%)0.3620.088No742 (65.55%)309 (63.19%)1051 (64.84%)334 (61.28%)Yes390 (34.45%)180 (36.81%)570 (35.16%)211 (38.72%)Endocrine therapy, n (%)0.1490.361No359 (31.71%)173 (35.38%)532 (32.82%)185 (33.94%)Yes773 (68.29%)316 (64.62%)1089 (67.18%)360 (66.06%)Radiotherapy, n (%)0.6740.373No888 (78.45%)379 (77.51%)1267 (78.16%)417 (76.51%)Yes244 (21.55%)110 (22.49%)354 (21.84%)128 (23.49%)Targeted therapy, n (%)0.942< 0.001No980 (86.57%)424 (86.71%)1404 (86.61%)505 (92.66%)Yes152 (13.43%)65 (13.29%)217 (13.39%)40 (7.34%) Demographic and clinical characteristics of elderly breast cancer patients. Prognostic Factors for Elderly Patients With Breast Cancer Multicollinearity was evaluated for all candidate variables (Table S1). Univariate analysis of the training cohort revealed that age, BMI, NLR, PNI, TNM stage, mFI‐5 score, chemotherapy, and endocrine therapy were significantly associated with OS (all p < 0.05). Subsequent multivariate Cox regression analysis demonstrated that age, PNI, TNM stage, mFI‐5 score, and endocrine therapy were independent prognostic factors. These variables were incorporated into the final model. LASSO regression yielded a similar set of predictors, further supporting the robustness and stability of the selected variables (Figure 1, Table S2). These variables were incorporated into the final model (Table 2). Features selection by LASSO. (A) Profiles of the LASSO coefficients for the 17 candidate variables. (B) Ten‐fold cross‐validation for tuning parameter selection in the LASSO model. TABLE 2Univariate and multivariate Cox analyses on variables for the prediction of OS of elderly breast cancer patients.CharacteristicUnivariate analysis p Multivariate analysis p HR95% CIHR95% CIAge, years1.086(1.052, 1.121)< 0.0011.079(1.045, 1.114)< 0.001BMI, kg/m2 1.082(1.018, 1.150)0.011NLR1.622(1.237, 2.128)0.0011.131(0.841, 1.521)0.415PNI0.913(0.869, 0.958)< 0.0010.925(0.878, 0.974)0.003Education levelLess than high school1High school0.780(0.485, 1.255)0.306College or higher0.785(0.448, 1.375)0.397Tumor siteLeft1Right1.26(0.86, 1.85)0.236Breast surgery typeLumpectomy1Mastectomy1.138(0.698, 1.854)0.604Axillary surgery typeWithout axillary surgery1SLNB0.737(0.417, 1.302)0.294ALND1.045(0.546, 2.034)0.876TNM stageI11II1.921(1.219, 3.028)0.0051.822(1.151, 2.885)0.011III3.185(1.869, 5.428)< 0.0013.115(1.815, 5.347)< 0.001mFI‐5No‐frailty11Mild‐frailty2.345(1.488, 3.696)< 0.0012.296(1.435, 3.672)< 0.001Severe‐frailty3.822(2.241, 6.518)< 0.0013.073(1.783, 5.296)< 0.001Histological gradingGrade I1Grade II1.788(0.433, 7.385)0.422Grade III2.619(0.640, 10.722)0.181Histologic typeIDC1ILC or others1.566(0.496, 4.945)0.444Molecular subtypeLuminal A1Luminal B1.441(0.903, 2.299)0.126HER‐21.310(0.683, 2.515)0.416TNBC1.442(0.804, 2.588)0.219ChemotherapyNo11Yes1.681(1.143, 2.472)0.0080.869(0.521, 1.451)0.592RadiotherapyNo1Yes1.054(0.652, 1.702)0.831Endocrine therapyNo11Yes0.458(0.310, 0.676)< 0.0010.454(0.306, 0.676)< 0.001Targeted therapyNo1Yes0.738(0.384, 1.417)0.361 Univariate and multivariate Cox analyses on variables for the prediction of OS of elderly breast cancer patients. FINE Index Construction and Validation Based on the five identified predictors, a prognostic nomogram named the FINE Index was developed. The FINE Index integrates frailty, immunonutrition, and stage information for elderly breast cancer patients. By summing the scores of the five variables, the 3‐, 5‐, and 10‐year overall survival probabilities can be estimated (Figure 2). The FINE Index showed good discrimination in all cohorts, with C‐index values of 0.767 in the training cohort, 0.739 in the internal validation cohort, and 0.731 in the external validation cohort. These values were consistently higher than those of TNM staging, indicating better predictive performance. Time‐dependent ROC analysis further supported this finding. Across all cohorts, the AUCs of the FINE Index for predicting 3‐, 5‐, and 10‐year overall survival were higher than those of TNM stage and histological grade (Figure S2), demonstrating stable and favorable discrimination. Nomogram predicting 3‐, 5‐, and 10‐year OS of elderly breast cancer patients. The FINE Index was compared with the TNM staging system using reclassification analysis. Compared with TNM staging, the FINE Index significantly improved prognostic accuracy, with consistently positive NRI and IDI values across all time points and cohorts (Table S3). ROC curve analysis showed good discriminatory performance. The 3‐, 5‐, and 10‐year AUCs were 0.874, 0.835, and 0.746 in the training cohort; 0.842, 0.785, and 0.781 in the internal validation cohort; and 0.846, 0.759, and 0.753 in the external validation cohort, respectively (Figure S2). Calibration plots showed strong agreement between predicted and observed survival probabilities in all cohorts (Figure S3). Decision curve analysis further indicated that the FINE Index provided greater net clinical benefit than TNM staging (Figure S4). Risk Stratification Based on the FINE Index Using the maximum Youden index for 5‐year OS in the training cohort, the optimal FINE Index cut‐off value was determined as 81.12. Patients with scores < 81.12 were classified as low risk, whereas those with scores ≥ 81.12 were classified as high risk. The same threshold was applied without re‐estimation to the internal validation and regional external validation cohorts. Because the cut‐off was selected to maximize discrimination in the training cohort, the apparent risk separation in that cohort may be optimistic, and the validation cohorts provide a more appropriate assessment of its reproducibility. Kaplan–Meier survival analysis demonstrated significantly different OS between the two groups across all cohorts (Figure 3). In the training cohort, the 5‐year OS rate was 96.1% in the low‐risk group compared with 75.5% in the high‐risk group, corresponding to an absolute survival difference of 20.6%. Similar survival separation was observed in the internal and external validation cohorts (Table S4). The median follow‐up duration was 2402 days (approximately 6.6 years) in the training cohort, 2508 days (approximately 6.9 years) in the internal validation cohort, and 2602 days (approximately 7.1 years) in the external validation cohort. The overall survival rates of the entire cohort were 95.8%, 92.6%, and 84.7% at 3, 5, and 10 years, respectively, indicating generally favorable long‐term outcomes in this elderly breast cancer population. Kaplan–Meier survival curves of OS stratified by the risk levels of the nomogram. Kaplan–Meier survival curves of OS stratified by the risk levels in training cohort (A). Kaplan–Meier survival curves of OS stratified by the risk levels in internal validation cohort (B). Kaplan–Meier survival curves of OS stratified by the risk levels in external validation cohort (C). We conducted an exploratory subgroup analysis to describe the association between chemotherapy and OS within the FINE Index‐defined risk groups. Among high‐risk patients, chemotherapy recipients had longer OS in the training, internal validation, and regional external validation cohorts, with log‐rank p values of 0.015, 0.031, and 0.030, respectively. No statistically significant association was observed among low‐risk patients. However, these comparisons were unadjusted, and no formal chemotherapy‐by‐risk‐group interaction test was performed. Therefore, the differences in subgroup‐specific p values should not be interpreted as evidence that chemotherapy efficacy differed between the high‐ and low‐risk groups (Figure S5). Discussion In this study, we established the FINE Index, a prognostic model integrating frailty, immunonutritional status, and tumor‐related factors on the basis of readily available clinical variables. Compared with the traditional TNM staging system, the FINE Index demonstrates superior predictive accuracy, discrimination, and clinical applicability in predicting OS for elderly patients with breast cancer. Among the independent prognostic factors, age remained significantly associated with worse OS. In elderly patients, age is not merely a chronological characteristic but also a surrogate for cumulative biological decline. With advancing age, multiple organ systems gradually lose reserve capacity, including cardiovascular, pulmonary, metabolic, and immune functions. This reduction in physiological reserve may impair the ability to recover from surgical stress, tolerate postoperative treatment, and maintain long‐term functional stability [13, 14, 15]. In addition, aging is closely associated with immunosenescence, chronic low‐grade inflammation, and an increasing burden of comorbidities, all of which may amplify cancer‐related vulnerability and increase competing noncancer mortality [16]. Therefore, the adverse prognostic effect of age in our study likely reflects a broad background of biological vulnerability rather than the effect of time itself. TNM stage remained one of the strongest predictors of OS, which is expected because it fundamentally reflects tumor burden and disease extent [17]. A more advanced TNM stage indicates a greater likelihood of nodal involvement, occult dissemination, recurrence, and cancer‐related death [6, 18]. A greater tumor burden may also increase treatment complexity and the physiological burden on the host. For an older patient with limited reserve, the clinical consequences of a more advanced stage may therefore be magnified. Inflammation and nutritional status play important roles in cancer development and progression. Chronic inflammation promotes tumorigenesis by affecting cell proliferation, angiogenesis, invasion, and metastasis through inflammatory mediators and signaling pathways [19, 20]. Malnutrition is also common in cancer patients and can impair immunity, reduce treatment tolerance, increase treatment‐related toxicity, worsen quality of life, and ultimately affect survival [21, 22]. The PNI, calculated from serum albumin and lymphocyte count, reflects both nutritional and immune status. It was originally developed to predict postoperative complications in gastrointestinal cancer patients [23]. Previous studies have shown that low serum albumin is associated with poor prognosis in several cancers, including breast cancer [24]. Malnutrition‐related immune dysfunction may also weaken lymphocyte‐mediated antitumor responses in the tumor microenvironment [10, 25]. In cancer patients, tumor‐related inflammation can increase inflammatory factors such as IL‐6 and TNF‐α, leading to reduced albumin synthesis and increased systemic protein breakdown and energy consumption. This process contributes to weight loss, skeletal muscle loss, functional decline, and cachexia. Low albumin levels may further reduce anti‐inflammatory, antioxidant, and antithrombotic capacity, thereby worsening prognosis [26]. Similarly, lymphocyte depletion may indicate impaired antitumor immunity and reduced immune surveillance. In elderly patients, these changes may be further aggravated by immunosenescence, frailty, and chronic low‐grade inflammation [27, 28]. Recent studies have shown that the PNI is closely associated with long‐term cancer prognosis, with lower PNI generally indicating poorer outcomes. A two‐centre study reported that higher PNI was associated with a higher pathological complete response rate in breast cancer patients [29]. Several retrospective studies have also identified the PNI as an independent prognostic factor for survival in breast cancer [23, 30, 31]. To our knowledge, this is the first study to evaluate the prognostic value of the PNI in elderly breast cancer patients undergoing surgery. Incorporating the PNI into prognostic models may help support individualized nutritional assessment and perioperative nutritional interventions to improve treatment outcomes. Frailty is a common geriatric syndrome characterized by reduced physiological reserve, decreased resistance to stressors, and increased vulnerability to environmental factors [14]. Previous studies have reported that frailty affects up to 43% of breast cancer patients and is strongly associated with postoperative complications, poor treatment tolerance, and increased mortality [32]. Therefore, frailty assessment and intervention are important for improving outcomes in elderly breast cancer patients. However, traditional frailty indices, such as the CCI, mFI‐11, and CSHA‐FI, are difficult to calculate in routine clinical practice and are often limited by unavailable or unreliable variables in large datasets. In contrast, the mFI‐5 is a simpler and more practical tool that has been widely used in orthopedics, neurosurgery, and urology, where it has been shown to independently predict postoperative complications and prognosis [33, 34, 35]. The mFI‐5 reflects biological rather than chronological aging by capturing comorbidity burden and reduced functional reserve. In elderly breast cancer patients, frailty may worsen prognosis through several mechanisms. Frail patients are more likely to experience postoperative decompensation, delayed recovery, complications, and incomplete functional recovery [14, 36, 37]. Frailty is also commonly associated with reduced mobility, sarcopenia, chronic inflammation, and functional dependence, which may reduce tolerance to adjuvant therapy and long‐term resilience [8, 14]. In addition, frailty may increase noncancer‐related mortality independent of tumor progression [38]. Previous studies have shown that the mFI‐5 can predict complications after breast cancer surgery [36, 37] and is associated with postoperative quality of life after autologous breast reconstruction [39]. However, to our knowledge, the mFI‐5 has not previously been used to predict prognosis in elderly breast cancer patients undergoing surgery. This study is the first to incorporate the mFI‐5 into a prognostic model, which may help identify high‐risk patients, optimize perioperative management, improve individualized risk communication, and expand the clinical application of the mFI‐5 in elderly breast cancer patients. Despite these advantages, the mFI‐5 should not be interpreted as a comprehensive frailty assessment. It is primarily a comorbidity‐ and functional‐dependence‐based index and does not directly capture several geriatric domains that are highly relevant to surgical oncology, including cognitive function, gait speed, grip strength, sarcopenia, nutritional intake, polypharmacy, psychological status, and social vulnerability. As a result, the FINE Index may underestimate vulnerability in older patients who have few recorded comorbidities but substantial functional decline, cognitive impairment, or muscle loss. The FINE Index should be considered a practical risk‐stratification tool based on routinely available clinical data, rather than a substitute for comprehensive geriatric assessment. Moreover, dependent functional status is only one component of the mFI‐5 and cannot replace direct assessment of activities of daily living, instrumental activities of daily living, mobility, cognition, or physical performance. Future studies should combine the FINE Index with comprehensive geriatric assessment and objective measurements of sarcopenia, such as skeletal muscle index derived from computed tomography or validated physical‐performance tests. Adjuvant endocrine therapy is a key treatment for hormone receptor‐positive breast cancer and has been shown to reduce recurrence and improve long‐term outcomes. After surgery, its benefit mainly comes from continuous suppression of estrogen receptor signaling, which inhibits hormone‐dependent tumor cell growth and reduces the progression of residual microscopic disease [40, 41]. However, the prognostic association between endocrine therapy and improved OS in this retrospective model should be interpreted cautiously. Receipt and continuation of endocrine therapy may also reflect better functional reserve, fewer competing health burdens, greater treatment tolerance, better adherence, and more regular follow‐up. In contrast, frailer patients or those with substantial comorbidities may be less likely to initiate, tolerate, or persist with long‐term endocrine therapy [13, 42, 43, 44]. Therefore, endocrine therapy in the FINE Index may represent both antitumour treatment exposure and broader patient‐level treatment feasibility, rather than a purely direct causal effect. In summary, the five independent predictors in our model reflect a multidimensional prognostic pathway involving tumor burden, biological aging, host vulnerability, and treatment feasibility. This may explain why models integrating both tumor‐related and host‐related factors provide better prognostic stratification than TNM staging alone. To our knowledge, this is the first nomogram to predict prognosis in elderly breast cancer patients by combining frailty and nutritional indicators. Although the PREDICT tool can accurately estimate 5‐year overall survival in elderly breast cancer patients, it performs less well for 10‐year survival and does not include comorbidities [45]. Another model based on the CCI was developed in 2231 elderly breast cancer patients to predict overall survival, but it did not include immune or nutritional biomarkers and lacked external validation, which may limit its generalizability [46]. Importantly, frailty and malnutrition are dynamic and potentially reversible. Therefore, preoperative assessment of nutritional status and frailty, together with interventions such as prehabilitation, optimized perioperative care, and close postoperative monitoring, may help improve recovery and prognosis. Although most patients in our cohort had Stage I–II disease and relatively low frailty burden, this reflects the real‐world characteristics of elderly breast cancer patients undergoing surgery, for whom anatomical tumor burden alone may not fully capture survival heterogeneity. Among patients with similar early‐stage disease, differences in physiological reserve, nutritional status, and treatment tolerance may substantially influence long‐term outcomes. Therefore, the primary value of the FINE Index is not to identify patients with advanced disease, but to refine risk stratification among elderly patients who appear clinically favorable according to conventional staging. In exploratory analyses, chemotherapy was associated with longer OS among high‐risk patients, whereas no statistically significant association was observed among low‐risk patients. These findings should be interpreted with substantial caution. The analysis was retrospective, unadjusted, and not based on a prespecified treatment‐effect hypothesis. In addition, no formal interaction test was performed; therefore, the presence of a statistically significant association in one risk group and the absence of significance in another do not establish heterogeneity of chemotherapy effect. Confounding by indication, treatment selection, tumor biology, functional status, comorbidity burden, and treatment tolerance may have influenced both chemotherapy receipt and survival. Accordingly, the FINE Index should currently be regarded as a prognostic model rather than a predictive biomarker for chemotherapy benefit. Prospective studies using prespecified interaction analyses, appropriate adjustment for confounding, or randomized treatment comparisons are required before the model can be considered for chemotherapy decision‐making. Furthermore, these findings should not be interpreted as replacing established chemotherapy decision‐making frameworks, including genomic assays and clinicopathological assessment. Instead, the FINE Index may provide complementary information by incorporating host vulnerability and treatment tolerance‐related factors that are not captured by tumor‐based models alone. Several limitations should be acknowledged. First, the retrospective design is subject to selection bias, residual confounding, and incomplete capture of treatment adherence and care delivered outside the study institutions. Patients with incomplete clinical or follow‐up data were excluded, and complete‐case analysis may have introduced additional selection bias. Second, although the validation cohort was obtained from a separate institution, the two centres were geographically proximate and shared a similar regional healthcare environment. The present analysis should therefore be regarded as regional external validation, and broader validation in geographically and ethnically diverse populations remains necessary. Third, the mFI‐5 is a practical but incomplete frailty measure and does not directly assess cognition, activities of daily living, gait speed, grip strength, sarcopenia, polypharmacy, psychological status, or social vulnerability. Fourth, cause‐specific mortality was not reliably available; therefore, competing‐risk analyses of breast cancer and noncancer mortality could not be performed. This limitation is particularly relevant in older patients, among whom comorbidities, frailty, and age‐related conditions may account for a substantial proportion of deaths. Fifth, the risk threshold was selected by maximizing the Youden index in the training cohort, which may have produced optimistic risk separation despite application of the fixed threshold in the validation cohorts. Sixth, the chemotherapy subgroup analysis was exploratory and unadjusted, and no formal interaction analysis was performed. These findings cannot establish chemotherapy benefit or support treatment selection. Prospective, multicentre studies incorporating comprehensive geriatric assessment, standardized cause‐of‐death information, and prespecified treatment‐effect analyses are required. Conclusions The FINE Index integrates frailty, immunonutritional status, and tumor characteristics to provide individualized estimates of overall survival in older patients with breast cancer undergoing surgery. At present, the FINE Index should be considered a prognostic risk‐stratification tool rather than a predictor of chemotherapy benefit. Prospective validation in independent, geographically diverse cohorts, together with comprehensive geriatric assessment and prespecified treatment‐effect analyses, is required before clinical implementation or use in treatment‐selection decisions. Author Contributions Xiwen Yang: writing – original draft, software, methodology. Jiang Zhu: conceptualization, writing – review and editing, validation, funding acquisition. Funding The study was supported by grants from the National Natural Sciences Foundation of China (82605979), the Natural Science Foundation of Sichuan Science and Technology Program (2025ZNSFSC1900), and the Postdoctor Research Fund of West China Hospital, Sichuan University (2025HXBH162). Disclosure The authors have nothing to report. Ethics Statement This study was approved by the institutional review boards of West China Hospital and conducted in accordance with the Declaration of Helsinki (ID: 2023‐2243). Written informed consent for the use of clinical data was obtained from all participants, and all personal identifiers were anonymized prior to analysis. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table S1: Results of multicollinearity test. Table S2: The coefficients of Lasso regression analysis. Table S3: The C‐index, NRI and IDI of the nomogram and the TNM staging system alone in survival prediction for elderly breast cancer patients. Table S4: Overall survival rates according to FINE Index risk stratification in the training cohort, internal validation cohort and external validation cohort. Figure S1: The flowchart of including and dividing patients. Figure S2: ROC curves of the nomogram compared with TNM stage and histological grade. The ROC curves of nomogram at 3‐, 5‐ and 10‐year in the training cohort (A), internal validation cohort (B) and external validation cohort (C). The ROC curves of TNM stage at 3‐, 5‐ and 10‐year in the training cohort (D), internal validation cohort (E) and external validation cohort (F). The ROC curves of histological grade at 3‐, 5‐ and 10‐year in the training cohort (G), internal validation cohort (H) and external validation cohort (I). Figure S3: The calibration curves for predicting elderly breast cancer patients OS at 3‐, 5‐ and 10‐year in the training cohort (A), internal validation cohort (B) and external validation cohort (C). Figure S4: DCA of the nomogram for predicting the 3‐year OS in the training cohort (A), internal validation cohort (B) and external validation cohort (C); DCA of the nomogram for predicting the 5‐year OS in the training cohort (D), internal validation cohort (E) and external validation cohort (F); DCA of the nomogram for predicting the 10‐year OS in the training cohort (G), internal validation cohort (H) and external validation cohort (I). Figure S5: Kaplan—Meier survival curves of OS stratified by chemotherapy. Kaplan—Meier survival curves of OS stratified by chemotherapy in training cohort (A), internal validation cohort (B) and external validation cohort (C) in high‐risk patients. Kaplan—Meier survival curves of OS stratified by chemotherapy in training cohort (D), internal validation cohort (E) and external validation cohort (F) in low‐risk patients. Figure S6: Example Patient Prediction K‐M Plot. Figure S7: Example Patient Numerical Summary data. Acknowledgments The authors gratefully acknowledge each editor and reviewer for their efforts on this study. Data Availability Statement Data sharing not applicable to this article as no datasets were generated or analysed during the current study. References Associated Data Supplementary Materials Table S1: Results of multicollinearity test. Table S2: The coefficients of Lasso regression analysis. Table S3: The C‐index, NRI and IDI of the nomogram and the TNM staging system alone in survival prediction for elderly breast cancer patients. Table S4: Overall survival rates according to FINE Index risk stratification in the training cohort, internal validation cohort and external validation cohort. Figure S1: The flowchart of including and dividing patients. Figure S2: ROC curves of the nomogram compared with TNM stage and histological grade. The ROC curves of nomogram at 3‐, 5‐ and 10‐year in the training cohort (A), internal validation cohort (B) and external validation cohort (C). The ROC curves of TNM stage at 3‐, 5‐ and 10‐year in the training cohort (D), internal validation cohort (E) and external validation cohort (F). The ROC curves of histological grade at 3‐, 5‐ and 10‐year in the training cohort (G), internal validation cohort (H) and external validation cohort (I). Figure S3: The calibration curves for predicting elderly breast cancer patients OS at 3‐, 5‐ and 10‐year in the training cohort (A), internal validation cohort (B) and external validation cohort (C). Figure S4: DCA of the nomogram for predicting the 3‐year OS in the training cohort (A), internal validation cohort (B) and external validation cohort (C); DCA of the nomogram for predicting the 5‐year OS in the training cohort (D), internal validation cohort (E) and external validation cohort (F); DCA of the nomogram for predicting the 10‐year OS in the training cohort (G), internal validation cohort (H) and external validation cohort (I). Figure S5: Kaplan—Meier survival curves of OS stratified by chemotherapy. Kaplan—Meier survival curves of OS stratified by chemotherapy in training cohort (A), internal validation cohort (B) and external validation cohort (C) in high‐risk patients. Kaplan—Meier survival curves of OS stratified by chemotherapy in training cohort (D), internal validation cohort (E) and external validation cohort (F) in low‐risk patients. Figure S6: Example Patient Prediction K‐M Plot. Figure S7: Example Patient Numerical Summary data. Data Availability Statement Data sharing not applicable to this article as no datasets were generated or analysed during the current study.