Predicting the Effectiveness of 2 Chemotherapy Regimens Applied to Breast Cancer Patients by Nutritional and Inflammatory Status

Introduction

Breast cancer remains a leading cause of morbidity and mortality among women globally, necessitating continuous advancements in treatment and prediction strategies [1,2]. Systemic chemotherapy (CT), particularly in the neoadjuvant setting, is crucial for managing locally advanced or high-risk early-stage breast cancer, offering the advantage of in vivo response assessment and potential for breast-conserving surgery [3]. Achieving a pathological complete response (pCR) (the absence of residual invasive cancer after treatment) after neoadjuvant chemotherapy (NAC) is an important indicator of improved long-term outcomes, especially in aggressive subtypes like HER2-positive breast cancer [4,5]. Anthracyclines are DNA-damaging cytotoxic agents, while trastuzumab and pertuzumab are monoclonal antibodies that target the HER2 receptor.

Malnutrition is common in cancer patients, potentially reducing their tolerance and response to therapies [6]. The interplay between cancer and its treatment can disrupt metabolism and nutrient intake, leading to malnutrition and poorer survival across various cancers, including breast cancer [7]. The Prognostic Nutritional Index (PNI), calculated from serum albumin and total lymphocyte count, is a simple, non-invasive marker reflecting nutritional status [8]. Low pretreatment PNI has been associated with worse overall survival in breast cancer, suggesting its utility as a prognostic indicator [9]. The Controlling Nutritional Status (CONUT), which incorporates albumin, cholesterol, and lymphocytes, provides a more comprehensive immune–nutritional assessment [10,11]. Elevated CONUT is linked to worse survival in breast cancer patients after resection [12]. The Modified Glasgow Prognostic Score (mGPS) and Naples Prognostic Score (NPS) are novel prognostic scores for assessing nutritional status in cancer patients [13,14].

Neoadjuvant CT for HER2-positive breast cancer often includes anthracycline and taxane-based regimens combined with HER2-targeted therapies like trastuzumab and pertuzumab [15]. Despite increasing recognition of host-related factors, there is limited evidence on whether nutritional–inflammatory indices can differentiate response or survival between anthracycline-containing (ddAC-THP) and anthracycline-free (TCHP) regimens in HER2-positive breast cancer [16]. Therefore, this study aimed to assess the predictive value of 5 pretreatment indices (CAR, PNI, CONUT, NPS, and mGPS) in differentiating treatment effectiveness and survival between ddAC-THP and TCHP regimens.

Material and Methods

STUDY DESIGN AND SETTING:

This study used a retrospective cohort design, analyzing data from electronic medical records of patients diagnosed with breast cancer who received NAC at our single-center institution. It covered the period from January 2016 to December 2024, including sufficient patients treated with the specific CT regimens and adequate follow-up time. The retrospective design enabled analysis of real-world clinical data, providing insights into the predictive value of nutritional indices in routine clinical practice. The study protocol was reviewed and approved by the Non-Interventional Clinical Research Ethics Committee. All procedures adhered to institutional and/or national research standards and to the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from all participants, and patient confidentiality was maintained throughout.

The study protocol was reviewed and approved by the Non-Interventional Clinical Research Ethics Committee of Van Training and Research Hospital (Decision No: BOPAEK/2025-02-16; 28 February 2025).

POPULATION AND ACCEPTANCE CRITERIA:

The study included adult female patients (≥18 years old) diagnosed with histologically confirmed breast cancer (such as invasive ductal carcinoma, invasive lobular carcinoma, or other specified types) who received NAC with either dose-dense doxorubicin and cyclophosphamide followed by paclitaxel or docetaxel, trastuzumab, and pertuzumab (ddAC-THP), or docetaxel, carboplatin, trastuzumab, and pertuzumab (TCHP). Patients with early-stage (Stage I–III) and metastatic (Stage IV at initial diagnosis) breast cancer were included to reflect the variety of clinical scenarios in which these regimens are used. Exclusion criteria included patients who: (1) received NAC other than ddAC-THP or TCHP as their primary treatment; (2) had incomplete records missing key data such as pretreatment albumin, total lymphocyte count, cholesterol, CT details, or outcome data; (3) had a history of other malignancies within the past 10 years (excluding non-melanoma skin cancer or curatively treated in situ carcinoma); (4) had pre-existing conditions known to affect nutritional status (eg, severe liver disease) or immune function (eg, HIV/AIDS); (5) participated in clinical trials that could influence their treatment or outcomes; or (6) had severe hepatic or renal impairment, active inflammatory diseases, or other conditions that could significantly alter baseline inflammatory or nutritional biomarkers, in order to minimize potential confounding effects on calculated indices.

CHEMOTHERAPY REGIMENS:

Patients were categorized into 2 groups based on the NAC regimen received. The ddAC-THP regimen consists of dose-dense doxorubicin (60 mg/m²) and cyclophosphamide (600 mg/m²) administered intravenously every 2 weeks for 4 cycles, followed by paclitaxel (80 mg/m²) weekly for 12 weeks or docetaxel (75 mg/m² every 3 weeks), concurrently with trastuzumab (initial dose of 8 mg/kg, then 6 mg/kg every 3 weeks) and pertuzumab (initial dose of 840 mg, then 420 mg every 3 weeks). The TCHP regimen consists of docetaxel (75 mg/m²), carboplatin (AUC 6 mg/mL min), trastuzumab (initial dose of 8 mg/kg, then 6 mg/kg every 3 weeks), and pertuzumab (initial dose of 840 mg, then 420 mg every 3 weeks) administered intravenously every 3 weeks for 6 cycles. Details were recorded regarding the number of cycles received, dose modifications, and treatment delays. The following variables were extracted: patient demographics, disease characteristics, pretreatment laboratory values, treatment details, and outcome data. Treatment adherence was monitored through electronic medical records and nursing documentation. Any dose modifications or delays were recorded, and <5% of cycles showed deviations from the protocol-specified timing.

NUTRITIONAL INDEX AND SCORES:

Demographic data, relevant clinical characteristics, and laboratory results were retrospectively extracted from the EHR for the 112 included patients. Laboratory parameters obtained during the first assessment were used to calculate the scores. These parameters included: serum albumin (g/dL), C-reactive protein (CRP; mg/L), total cholesterol (mg/dL), total lymphocyte count (cells/μL), absolute neutrophil count (cells/μL), and absolute monocyte count (cells/μL). Based on the laboratory values collected at the index time, the following scores were calculated for each patient: Controlling Nutritional Status (CONUT) Score calculated using serum albumin, total lymphocyte count, and total cholesterol levels, according to established criteria. Points were assigned as follows: albumin (≥3.5 g/dL: 0, 3.0–3.49: 2, 2.5–2.99: 4, <2.5: 6 points); lymphocytes (≥1600 cells/μL: 0, 1200–1599: 1, 800–1199: 2, <800: 3 points); cholesterol (>180 mg/dL: 0, 140–180: 1, 100–139: 2, <100: 3 points). The total score ranged from 0 to 12. The Modified Glasgow Prognostic Score (mGPS) was determined using serum CRP and albumin levels. Score 0: CRP≤10 mg/L. Score 1: CRP >10 mg/L and albumin ≥3.5 g/dL. Score 2: CRP >10 mg/L and albumin <3.5 g/dL (range 0–2). The C-reactive protein-to-albumin ratio (CAR) was calculated by dividing the serum CRP concentration (mg/L) by the serum albumin concentration (g/dL). The Naples Prognostic Score (NPS) was calculated based on serum albumin, total cholesterol, neutrophil-to-lymphocyte ratio (NLR), and lymphocyte-to-monocyte ratio (LMR). NLR=absolute neutrophil/absolute lymphocyte count. LMR=absolute lymphocyte count/absolute monocyte count. Points were assigned for: albumin <4 g/dL (1 pt), total cholesterol <180 mg/dL (1 pt), NLR >2.96 (1 pt), LMR <4.4 (1 pt). The total score ranged from 0 to 4. The prognostic nutritional index (PNI) was calculated using the formula: PNI=[10×serum albumin (g/dL)] + [0.005×total lymphocyte count (cells/μL)].

STATISTICAL ANALYSIS:

All statistical analyses were conducted using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables were tested for normality with the Shapiro-Wilk test. Data that followed a normal distribution are reported as mean±standard deviation, while non-normally distributed data are shown as median with interquartile range. Missing data were less than 3% for all variables. Pairwise deletion was applied, and sensitivity analyses confirmed the consistency of the results. Data were independently verified by 2 oncologists to ensure accuracy. Baseline characteristics and scores were compared between specified groups (patients with versus without a particular outcome) using the t test or Mann-Whitney U test for continuous variables, and the chi-squared or Fisher’s exact test for categorical variables, as appropriate. Correlations between scores were evaluated using Spearman’s rank correlation coefficient. Univariable and multivariable logistic regression or Cox proportional hazards models for time-to-event data were used to analyze the relationship between baseline scores and the specific outcome. Overall survival was examined with Kaplan-Meier curves and log-rank tests. Multivariable Cox regression was used to identify independent predictors, with Schoenfeld residuals used to confirm the proportional hazards assumption. Variables incorporated into the multivariable models were chosen based on their relevance and univariable results (P<0.10). ROC curves and the area under the curve (AUC) were generated using SPSS’s nonparametric ROC procedure. AUC values were derived through nonparametric analysis, with 95% confidence intervals estimated by the method built into SPSS. ROC analyses were performed separately for each index (univariable ROC). The ROC was used to assess the ability of pretreatment nutritional–inflammatory indices (CONUT, CAR, mGPS, NPS, PNI, NLR, LMR) to differentiate between patients treated with ddAC-THP and those treated with TCHP. This analysis aimed to quantify how well these indices can distinguish between treatment groups based on baseline host status, providing an intuitive measure of their discriminative power regarding chemotherapy regimen selection.

Results

PATIENT CHARACTERISTICS AND TREATMENT OUTCOMES:

Baseline demographics, as shown in Tables 1 and 2, and clinical and histopathological characteristics were similar between the ddAC-THP and TCHP NAC groups, with no significant differences observed (all P>0.05). Notably, vital status was significantly higher in the TCHP group (97.5% alive) compared to the ddAC-THP group (84.7% alive; P=0.021). Additionally, survival rates favored the TCHP group (97.5%) over the ddAC-THP (84.7%; P=0.021). A positive correlation between NPS and NAC protocol (r=0.196, P=0.039) indicated that higher NPS were associated with specific CT protocols. The median follow-up duration was 38 months (range, 8–72). No patients were lost to follow-up, as survival data were confirmed from registries.

PREDICTIVE VALUE OF NUTRITIONAL AND INFLAMMATORY INDICES:

Treatment outcomes between patients receiving ddAC-THP and TCHP regimens are presented in Table 3. No differences were identified regarding neoadjuvant treatment duration (206.2±44.1 days vs 209.1±69.7 days, P=0.802) or interval between completion of neoadjuvant therapy and surgery (52±11.2 days vs 54.4±13.1 days, P=0.784). Similarly, both the number of retrieved lymph nodes (8.1±6.3 vs 8.2±5.8, P=0.766) and the number of metastatic lymph nodes (1.3±3 vs 1.5±2.4, P=0.522) did not differ. PCR rates were comparable between groups (55.6 vs 50%, P=0.713). Tumor size distributions also did not differ (P=0.359); however, the ddAC-THP had a higher rate of tumors sized 2–5 cm (9.7 vs 2.5%). Adjuvant radiotherapy utilization was similar (88.9 vs 90%, P=0.855). While recurrence or progression rates did not differ between the 2 treatment protocols (22.2 vs 15%, P=0.501), the ddAC-THP experienced higher rates of combined locoregional and metastatic recurrences (8.3 vs 0%). Survival analysis indicated a lower overall vital status in the ddAC-THP group (alive status: 84.7 vs 97.5%, P=0.021).

The discriminative performance of inflammatory and nutritional indices in predicting the use of ddAC-THP versus TCHP regimens was evaluated through receiver operating characteristic curve analysis (Figure 1). Among the markers assessed, CAR exhibited the strongest discrimination between CT protocols, with AUC of 0.76 (P<0.001). In contrast, NPS demonstrated an inverse association, with an AUC below the reference line (0.25; P<0.001). The remaining indices did not show meaningful discriminative power: mGPS (AUC 0.553; P=0.356), CONUT (AUC 0.516; P=0.775), NLR (AUC 0.453; P=0.407), LMR (AUC 0.505; P=0.935), and PNI (AUC 0.451; P=0.389) (Table 4).

SURVIVAL ANALYSIS:

Although the difference was marginal, the Kaplan-Meier survival analysis showed a difference in overall survival between patients treated with ddAC-THP and TCHP (Figure 2). Patients receiving ddAC-THP had a lower cumulative survival rate (84.7%) than the TCHP group (97.5%), with mean survival times of 8.69 months (95% CI: 8.3–9.09) and 17.13 months (95% CI: 15.2–19.06), respectively. However, overall comparisons by the log-rank test demonstrated no difference in survival distributions (P=0.065). Similarly, when evaluating the interval between NAC and surgery, survival analyses yielded a mean survival time of 7.04 months (95% CI: 3.68–10.40) for the ddAC-THP group and 11.58 months (95% CI: 8.09–15.07) for the TCHP, with borderline non-significant results (log-rank test; P=0.062).

Discussion

SUMMARY OF KEY FINDINGS:

This study investigated the potential of pretreatment nutritional and inflammatory indices to predict the effectiveness of 2 standard NAC regimens – ddAC-THP and TCHP – in a cohort of patients with HER₂-positive breast cancer encompassing both early-stage and metastatic disease. In this head-to-head comparison of 2 HER₂-positive NAC regimens, we found no difference in the primary efficacy outcomes between dose-dense AC→THP and TCHP. However, there was a trend toward improved survival with the anthracycline-free TCHP regimen. Our principal finding is that while established nutritional scores such as PNI and CONUT generally have prognostic value in breast cancer, they did not predict differential pCR or recurrence risk between the 2 regimens in our cohort. However, we observed a survival advantage and a trend toward improved survival with TCHP compared to ddAC-THP, alongside intriguing findings regarding the CAR and NPS.

INTERPRETATION IN THE CONTEXT OF EXISTING LITERATURE:

A novel aspect of our study is the exploration of host nutritional and inflammatory indices – the CAR and NPS – as potential guides for tailoring neoadjuvant treatment. We found that these indices, measured before therapy, stratified patients by response and survival outcomes, suggesting they could inform regimen selection. Qu et al reported that PNI was validated as an independent predictor of complete pathologic response in breast cancer patients receiving NAC [17]. In that extensive 2-center study, a low pretreatment PNI was associated with a lower probability of achieving pCR, and a nomogram incorporating PNI accurately stratified patients, underscoring that baseline nutritional immune competence can modulate CT efficacy. Similarly, Qui et al recently showed that the NPS predicts long-term survival and treatment benefit in breast cancer [12]. A high NPS was associated with shorter survival and inferior treatment efficacy in triple-negative breast cancer. Our results extend this concept to HER2-positive disease: patients with unfavorable indices (elevated CAR or NPS) tended to have worse therapy responses. It is intriguing to speculate that such patients, who are identifiable by objectively measurable blood markers, might derive disproportionate benefit from the less intensive TCHP (better tolerated) or might require adjunctive interventions, such as aggressive nutritional support, to optimize outcomes. In contrast, patients with robust dietary status and low inflammation (low CAR/NPS) may tolerate and benefit equally from either regimen, or may be candidates for trials of further de-escalation.

Cancer-related inflammation is now understood as a key determinant of tumor progression and survival [18]. Metrics such as NLR and composite scores, including GPS, CAR, and NPS, integrate host–tumor interplay into clinically accessible indices. In breast cancer, a recent meta-analysis of Dowling et al with 27 studies demonstrated that a low pretreatment NLR had nearly double the odds of achieving pCR to NAC compared to those with elevated NLR [19]. High peripheral lymphocyte counts, which is a surrogate of intact cell-mediated immunity, were likewise associated with higher pCR rates [19]. These findings suggest that an unfavorable inflammatory profile, such as high NLR or CAR, may indicate a tumor–host milieu that is less responsive to CT, possibly due to immunosuppressive factors or cytokine-driven treatment resistance. Our study used CAR, which combines an acute-phase reactant and albumin, as a candidate predictor. This mirrors prior observations in metastatic breast cancer, where elevated CRP and hypoalbuminemia, as a component of CAR, correlated with shorter survival. Yamada et al showed that mGPS equal to 2 (CRP >1 mg/dL and albumin <3.5 g/dL, equivalent to a high CAR) had markedly worse overall survival, and that a PNI <40 was an independent predictor of poor outcome on CT [20].

CLINICAL IMPLICATIONS:

Patients with poor nutritional–inflammatory status often have reduced treatment tolerance, which might limit the effective delivery of neoadjuvant chemotherapy. In our cohort, this pattern may explain why individuals with high CAR/NPS appeared less able to complete the intended ddAC→THP regimen, whereas TCHP was more consistently deliverable. Although our findings highlight a possible link between treatment tolerability and baseline nutritional–inflammatory indices, these observations remain correlational. The modest survival trend favoring TCHP also calls for caution, as our study was not powered for survival outcomes and the borderline P value limits definitive conclusions. Finally, elevated CAR/NPS may reflect comorbidity or a higher disease burden rather than an actual predictive effect. Prospective studies are needed to determine whether improving nutritional–inflammatory status can meaningfully enhance treatment tolerance.

STRENGTHS AND LIMITATIONS:

As a strength, our study contributes to the discourse by providing early clinical data to inform therapeutic direction. Given the retrospective, single-center design, residual confounding cannot be excluded. Nutritional indices may reflect comorbidities or unmeasured clinical factors. The study is underpowered for definitive comparisons of survival. Causal inference between nutritional indices and regimen effectiveness is not possible. Moreover, external validation in multiethnic, multicenter cohorts is required to confirm the generalizability of our findings. Additionally, excluding patients with hepatic or renal dysfunction may have attenuated the observed associations between nutritional indices and survival. The apparent survival difference favoring TCHP should be interpreted with particular caution. Although measured baseline characteristics were broadly similar between groups, survival outcomes in a retrospective cohort can be influenced by temporal trends, evolving supportive care, and unmeasured selection factors. For instance, TCHP has generally been adopted later than ddAC-THP in routine clinical practice and may be more often offered to fitter patients in some settings. It is therefore plausible that at least part of the observed difference reflects cohort effects rather than a true regimen-specific survival advantage.

Conclusions

CAR showed modest discriminative ability between treatment groups in this cohort. Other indices had limited utility. These associations should be validated prospectively before any clinical application is considered. The anthracycline-free TCHP regimen was associated with a higher observed proportion of survivors compared to ddAC-THP. However, the survival difference was only of borderline significance in time-to-event analysis. Prospective, adequately powered studies are needed to assess whether CAR or related indices can be integrated into multifactorial algorithms for regimen selection or supportive care planning in HER2-positive breast cancer.