Development of a multivariable prediction model for tumor response and treatment tolerance based on body composition dynamics in rectal cancer undergoing neoadjuvant chemoradiotherapy

Introduction

Background

Although the incidence of rectal cancer (RC) has declined, it remains one of the leading causes of cancer-related mortality worldwide (1). Management of locally advanced rectal cancer (LARC) has evolved substantially over the past decades. Due to anatomical constraints within the pelvis and the proximity of the rectum to adjacent organs and the anal sphincter, achieving optimal oncologic outcomes while preserving organ function remains challenging. Preoperative radiotherapy (RT), commonly delivered as chemoradiotherapy (CRT), has become a standard treatment to improve local control and facilitate tumor downstaging. More recently, total neoadjuvant therapy (TNT), which integrates CRT with systemic chemotherapy before surgery, has been increasingly adopted to enhance tumor response and expand opportunities for organ preservation (2). However, substantial variability in treatment response to CRT and TNT remains unexplained.

Rationale and knowledge gap

Current pretreatment risk stratification largely relies on tumor-centric factors, including TNM stage and histopathological characteristics, while the host’s metabolic and body composition status is often underappreciated. Emerging evidence suggests that nutritional status plays an important role in cancer immunity, treatment tolerance, and survival outcomes (3). In particular, cancer cachexia—characterized by progressive skeletal muscle loss independent of fat loss—is common in colorectal cancer and affects up to 61% of patients (4). Although traditional indicators such as body mass index (BMI) are frequently used, they often fail to reflect actual body composition due to substantial inter-individual variability (5). Computed tomography (CT)-based body composition analysis has therefore gained attention as a reliable method for quantifying skeletal muscle and adipose tissue compartments (6). Nevertheless, most previous studies have focused on long-term outcomes such as survival or postoperative complications. The potential relationship between body composition dynamics and immediate treatment response during neoadjuvant CRT remains insufficiently investigated, as most existing studies and prediction models have primarily focused on long-term survival outcomes rather than treatment response during neoadjuvant therapy. Furthermore, existing models are largely based on tumor-centric variables, and the integration of host-related factors, particularly dynamic changes in body composition, remains limited.

Objective

Given that abdominopelvic CT is routinely performed for staging in RC, CT-derived body composition parameters can be obtained without additional patient burden. In this study, we aimed to develop and internally validate a multivariable prediction model for tumor regression and CRT-related adverse events (CRT-AEs) based on baseline body composition and its dynamic changes during neoadjuvant CRT in patients with LARC. We further explored the distinct roles of skeletal muscle and adipose tissue compartments in treatment tolerance and tumor response. We present this article in accordance with the TRIPOD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0269/rc).

Methods

Study population

This single-center retrospective cohort study included patients newly diagnosed with LARC who underwent neoadjuvant CRT between March 2020 and March 2024 at Soonchunhyang University Cheonan Hospital. LARC was defined as clinical stage II–III disease based on pre-treatment magnetic resonance imaging (MRI) and CT. Patients meeting the exclusion criteria were excluded. The exclusion criteria were as follows: (I) absence of preserved medical records or imaging data; (II) diagnosis or treatment of another malignancy within five years prior to rectal cancer diagnosis; (III) presence of metabolic or endocrine disorders (e.g., uncontrolled diabetes, thyroid disease, or Cushing’s syndrome); (IV) chronic hepatic or renal disease (cirrhosis or chronic kidney disease stage ≥3); (V) use of corticosteroids, anabolic agents, or nutritional or hormonal supplements within six months prior to CRT; (VI) psychiatric conditions requiring medication that could interfere with treatment compliance or the evaluation of treatment response. One patient discontinued CRT due to a severe adverse event and was therefore excluded from the primary analysis of treatment response but included in the analysis of adverse events.

A total of 138 patients were ultimately included in the analysis. The sample size was determined based on the number of eligible patients treated during the study period. The number of outcome events was considered adequate for multivariable logistic regression analysis. All patients received long-course RT with a total dose of 5,000–5,040 cGy delivered in 25–28 fractions. Concurrent chemotherapy was administered according to the National Comprehensive Cancer Network guidelines, consisting of either oral capecitabine (825 mg/m² twice daily on days of radiation only) or a combination of 5-fluorouracil (400 mg/m² intravenous bolus) and leucovorin (20 mg/m² intravenous bolus) administered for four days during weeks 1 and 5 of RT.

Data collection

Clinical data were retrospectively extracted from electronic medical records, including age, sex, height, weight, Charlson Comorbidity Index (CCI), CRT regimen and duration, clinical and pathological staging, recurrence or metastasis status, and follow-up duration. Follow-up was defined as the time from CRT initiation to the last outpatient visit or, in the case of death, the date recorded on the death certificate. Recurrences and deaths were systematically recorded through retrospective review of electronic medical records and outpatient follow-up documentation. To assess immune-nutritional status, laboratory parameters obtained within one month prior to the initiation of RT were reviewed, including carcinoembryonic antigen (CEA), prognostic nutritional index (PNI), and neutrophil-to-lymphocyte ratio (NLR). The PNI, which has been associated with postoperative morbidity and mortality in cancer patients, was calculated using the formula proposed by Onodera et al.: PNI = 10 × serum albumin (g/dL) + 0.005 × total lymphocyte count (per mm³) (7,8). The NLR was calculated by dividing the absolute neutrophil count by the absolute lymphocyte count (9).

Body composition analysis

CT images were obtained from abdominopelvic CT scans performed within one month before CRT initiation and 8–12 weeks after CRT completion. Body composition segmentation was performed by a trained investigator using semiautomated methods in 3D Slicer software (version 5.8.1). The investigator was blinded to patients’ clinical outcomes during image analysis to minimize measurement bias.

Subcutaneous fat (SF), visceral fat (VF), skeletal muscle (SM), and psoas muscle (PM) were identified on a single axial slice at the level of the third lumbar vertebra (L3). Tissue segmentation was performed using previously validated Hounsfield unit (HU) thresholds: −190 to −30 HU for SF, −150 to −50 HU for VF, and −29 to 150 HU for SM, including the PM.

Cross-sectional areas were normalized by the square of patient height to calculate body composition indices (cm²/m²) (10,11). The visceral-to-subcutaneous fat ratio (VSR) was defined as the ratio of the visceral fat index to the subcutaneous fat index. Changes in body composition parameters(Δ) were calculated as the difference between post-CRT and pre-CRT values (i.e., Δ = post-CRT value − pre-CRT value). Accordingly, a negative Δ value indicates a reduction during CRT, whereas a less negative or positive Δ value indicates a smaller reduction or an increase.

Evaluation of treatment response and adverse events

Tumor response was assessed using the Dworak tumor regression grade (TRG) system based on postoperative histopathological findings, as recorded in medical records. TRG was classified into five grades ranging from complete regression (TRG 4) to no regression (TRG 0). For analysis, patients were categorized into two groups: high TRG (TRG 3–4) and low TRG (TRG 0–2). CRT-AEs were evaluated according to the Common Terminology Criteria for Adverse Events (CTCAE), version 5.0. Adverse events were recorded prospectively in structured treatment records at each daily outpatient visit during the CRT course. All patients underwent a scheduled outpatient evaluation approximately one week after CRT completion to assess for persisting or newly developed symptoms, and were instructed to return to the clinic if additional symptoms arose thereafter. No RT dose reductions were implemented in any patient. When multiple symptoms occurred, the adverse event with the highest severity grade was recorded. Adverse events were categorized according to the affected organ system: upper gastrointestinal (GI), lower GI, urinary, and other systemic symptoms.

Statistical analysis

Statistical analyses were performed using SPSS (version 30.0; IBM Corp., Armonk, NY, USA) and R software (version 4.5.10). There were no missing data for the variables included in the analysis. Categorical variables were analyzed using Fisher’s exact test or Pearson’s chi-square test, while continuous variables were compared using the independent samples t-test or Mann-Whitney U test depending on data distribution. Paired comparisons were performed using paired t-tests or Wilcoxon signed-rank tests.

Univariable and multivariable logistic regression analyses were conducted to identify predictors of high TRG. All candidate variables—including pre-CRT body composition parameters, their changes during CRT (Δ), and clinically relevant variables (age, sex, CCI, clinical TNM stage, BMI, NLR, PNI, and CRT-AEs)—were first evaluated in univariable logistic regression analyses. Variables with a P value <0.20 in univariable analysis, as well as variables considered clinically relevant based on prior literature, were selected as candidates for the multivariable model. Clinical TNM stage was included as a binary categorical variable, with stage II as the reference group and stage III as the comparator. Multicollinearity among candidate variables was assessed using the variance inflation factor (VIF), and the final multivariable model was developed using a stepwise selection procedure, retaining only variables with VIF <2 to ensure acceptable model stability. A total of 46 outcome events were available for model development. With 12 candidate predictors initially entered, the events-per-predictor (EPP) ratio was approximately 3.8 (46/12), which falls below the conventional threshold of 10. Therefore, to mitigate the risk of overfitting, internal validation was performed using bootstrap resampling with 1,000 iterations.

Continuous body composition variables were standardized as z-scores, and odds ratios were calculated per one standard deviation increase. Model discrimination was assessed using receiver operating characteristic (ROC) curve analysis. Decision curve analysis was used to evaluate potential clinical utility. A clinical-only comparator model incorporating CRT-AEs alone was included to assess the incremental predictive value of adding CT-derived body composition parameters over a clinically available variable alone. Predicted probabilities were calculated based on the final multivariable logistic regression model. No model updating or recalibration was performed, and risk group stratification was not conducted, as the primary aim of this study was to develop and internally validate a prediction model rather than to define clinically applicable risk categories. External validation was not performed in this study.

Ethical statement

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Institutional Review Board of Soonchunhyang University Cheonan Hospital (IRB No. 2022-06-022). Given the retrospective design of the study, the requirement for informed consent was waived by the Institutional Review Board.

Results

A total of 175 patients were initially screened, of whom 37 were excluded based on predefined criteria. The remaining 138 patients were included in the final analysis (Figure S1). Among the 138 patients, the distribution of TRG was as follows: TRG 0 (n=2, 1.45%), TRG 1 (n=34, 24.64%), TRG 2 (n=57, 41.30%), TRG 3 (n=23, 16.67%), and TRG 4 (n=22, 15.94%). Based on this, 92 patients were categorized into the low TRG group (TRG 0–2), and 46 into the high TRG group (TRG 3–4). Regarding concurrent chemotherapy, 33.33% of patients received oral capecitabine, while 66.67% received 5-fluorouracil/leucovorin.

The overall study population had a higher proportion of patients aged ≥65 years and males; however, there were no statistically significant differences in age or sex distribution between the low and high TRG groups. Due to the older age and comorbidity profile of the cohort, a larger proportion of patients had higher CCI scores. However, most patients were classified as American Society of Anesthesiologist (ASA) grade 2, indicating a generally favorable performance status across both TRG groups. Clinical TNM stage (cStage), as assessed by CT and rectal MRI, showed that both TRG groups had a high proportion of cStage III disease, with no significant difference between the groups. In addition, there were no significant differences in NLR, PNI, CEA, or BMI between the two groups. During the follow-up period, 16 patients experienced recurrence or metastasis, 15 of whom were in the low TRG group, representing a statistically significant difference between the groups (P=0.02). Although more deaths occurred in the low TRG group (n=13, 14.13%) than in the high TRG group (n=3, 6.52%), this difference was not statistically significant. The follow-up duration appeared shorter in the high TRG group (49.41 vs. 59.22 months). Given the very low mortality rate in both groups, the Kaplan-Meier survival curves did not reach the 50% threshold in either group, precluding the calculation of median survival (Table 1).

Table 2 presents the pre- and post-CRT body composition parameters, as well as their changes (Δ), stratified by TRG group. To assess within-group changes before and after CRT, paired-samples t-tests or Wilcoxon signed-rank tests were applied depending on data normality. In both TRG groups, post-CRT reductions in SF index (SFI) and VF index (VFI) were statistically significant (all P<0.05). In contrast, SM index (SMI) and PM index (PMI) increased significantly after CRT in the low TRG group. The reduction in VFI was greater than that of SFI in both groups, and VSR values decreased after CRT. Between-group comparisons revealed no statistically significant differences in any of the pre-, post-, or change (Δ) values of body composition parameters (Table 2).

To identify factors associated with TRG, univariable analyses were first performed using pre-CRT body composition parameters, their changes (Δ), and clinical variables including age, sex, BMI, CRT-AEs, CCI, cStage, NLR, and PNI (Table 3). None of the variables showed statistical significance in the univariable analyses. A total of 46 events (high TRG) were included in the model development. In the multivariable stepwise model, age, sex, and BMI were evaluated as candidate variables but were not retained in the final model, as none showed a significant association with TRG in the univariable analysis (age: OR 0.954, P=0.90; sex: OR 1.510, P=0.29; BMI: OR 1.040, P=0.44), and their inclusion did not meaningfully alter the estimates of other predictors. Four independent predictors of TRG were identified in the final multivariable model (Table 3, Figure 1A). All retained variables showed acceptable multicollinearity, with VIF values <2. The full model coefficients, including regression coefficients and intercept, are provided in Table S2. Correlation heatmap analysis further demonstrated moderate associations among body composition parameters, but no strong collinearity among predictors retained in the final multivariable model (Figure S2).

Figure 1 Model performance and independent predictors of tumor regression after CRT. (A) Forest plot of variables retained in the multivariable logistic regression model, including pre-CRT body composition parameters, their changes (Δ), and treatment-related factors. (B) ROC curve of the multivariable logistic regression model. AEs, adverse events; AUC, area under the curve; CI, confidence interval; CRT, chemoradiotherapy; cTNM, clinical tumor-node-metastasis; OR, odds ratio; PMI, psoas muscle index; ROC, receiver operating characteristic; Se, sensitivity; SFI, subcutaneous fat index; Sp, specificity; thr, threshold; TRG, tumor regression grade; VFI, visceral fat index; VSR, visceral-to-subcutaneous fat ratio; Δ, change (post-CRT minus pre-CRT).

The absence of CRT-AEs was significantly associated with a higher likelihood of high TRG (OR 0.388; 95% CI: 0.169 to 0.889; P=0.02). Higher pre-CRT SFI (OR 1.790; 95% CI: 1.139 to 2.814; P=0.01) and greater ΔVFI (OR 1.769; 95% CI: 1.089 to 2.871; P=0.02; indicating a smaller reduction in visceral fat index during CRT, as Δ = post-CRT − pre-CRT) were also associated with improved tumor regression, whereas lower pre-CRT PMI (OR 0.568; 95% CI: 0.352 to 0.916; P=0.02) showed a significant negative association. Pre-CRT VSR (OR 1.597; 95% CI: 0.994 to 2.567; P=0.053) and ΔPMI (OR 0.646; 95% CI: 0.412 to 1.013; P=0.057) demonstrated borderline significance, while other variables were not significantly associated with TRG (P>0.05).

In the multivariable analysis, the reduced model—comprising CRT-AEs, pre-CRT PMI, pre-CRT SFI, pre-CRT VSR, and ΔVFI—showed the best overall performance with the lowest Akaike Information Criterion (AIC) (ΔAIC =0.11) and the smallest Brier score (0.21). No significant multi-collinearity was observed among these predictors, with VIF values ranging from 1.13 to 1.60, ensuring the stability of the model. The ROC curve (Figure 1B) demonstrated moderate discriminative power with an area under the curve (AUC) of 0.707 (95% CI: 0.619–0.794), sensitivity of 0.96, and specificity of 0.40. Internal validation using 1,000-bootstrap resampling yielded an optimism-corrected calibration slope close to 1.0, and the bias-corrected calibration curve (Figure S3A) closely followed the ideal 45° line, indicating good calibration. Decision curve analysis (Figure S3B) showed that both the full and reduced models provided greater net clinical benefit across a wide range of threshold probabilities compared to the CRT-AEs only comparator model and the treat-all or treat-none strategies, supporting the incremental value of incorporating body composition parameters. A nomogram was subsequently constructed as an exploratory visualization tool to illustrate the relative contributions of each predictor in the final model (Figure 2). The predicted probability can be estimated by summing the points assigned to each predictor. However, given the modest sample size and the absence of external validation, the nomogram should be interpreted as exploratory, and its clinical application requires further validation in larger, independent cohorts.

Figure 2 Exploratory nomogram for estimating the probability of high TRG after neoadjuvant chemoradiotherapy. This nomogram is intended for exploratory purposes only and requires external validation before clinical use. AEs, adverse events; CRT, chemoradiotherapy; PMI, psoas muscle index; Pr, probability; SD, standard deviation; SFI, subcutaneous fat index; TRG, tumor regression grade; VFI, visceral fat index; VSR, visceral-to-subcutaneous fat ratio; Δ, change (post-CRT minus pre-CRT).

CRT-related adverse events were observed in 64 out of 138 patients. For patients who experienced multiple symptoms, the adverse event was classified based on the symptom with the highest grade (Table 4). Among the 64 patients with CRT-AEs, 47 were in the low TRG group and 17 were in the high TRG group. In the comparison between the low and high TRG groups, the incidence of CRT-AEs tended to be higher in the low TRG group, although the difference was not statistically significant (Fisher’s exact test, OR =0.563, 95% CI: 0.253–1.227; P=0.15). According to the CTCAE-based classification, GI disorders were the most common type of adverse event in both groups. Specifically, lower GI symptoms such as proctitis, diarrhea, and rectal hemorrhage accounted for the majority, followed by upper GI symptoms and urinary disorders. No adverse events exceeding grade 2 were recorded. Only one patient discontinued treatment due to CRT-AEs—specifically, grade 2 anorexia—and this decision was made at the patient’s own request due to poor compliance, after consultation with the attending physician. To further explore whether body-composition status affected treatment tolerance, additional logistic regression analyses were performed using the occurrence of CRT-AEs as the dependent variable. In the multivariable regression model that included clinical variables and excluded VFI due to multicollinearity, TRG—which was correlated with treatment outcomes—along with age, pre-CRT SMI, and the change in SMI (ΔSMI), showed statistically significant associations (Table S1, Figure S4). In contrast, fat indices that were significant in the TRG analysis did not show any statistical significance in relation to CRT-AEs (all P>0.05).

Discussion

Key findings

This study investigated the relationship between body composition dynamics and treatment outcomes in patients with LARC undergoing neoadjuvant CRT. Three principal findings emerged. First, SMI was primarily associated with treatment tolerance, as muscle depletion was significantly related to the occurrence of CRT-AEs. Second, adipose tissue parameters—particularly higher baseline subcutaneous fat and a smaller reduction in visceral fat during CRT—were independently associated with high TRG. Third, a multivariable model integrating these host-related factors demonstrated moderate predictive performance for tumor regression. These findings suggest that body composition compartments may have distinct clinical roles during CRT. While skeletal muscle appears to reflect the host’s capacity to tolerate treatment, adipose tissue dynamics may be more closely linked to tumor response. The model showed moderate discriminative performance (AUC =0.707) after internal validation using bootstrap resampling, which is consistent with the exploratory nature of this study and the relatively small sample size. With respect to survival outcomes, the apparently shorter follow-up duration in the high TRG group is likely attributable to a higher proportion of recently enrolled patients toward the latter part of the study period, rather than reflecting inferior survival. Consistent with the established prognostic value of tumor regression in LARC, recurrence or metastasis was significantly more frequent in the low TRG group (16.30% vs. 2.17%; P=0.02).

Strengths and limitations

This study has several strengths. First, we evaluated both baseline body composition and longitudinal changes during CRT using standardized CT-based measurements obtained from routine staging imaging. Second, the analysis incorporated treatment-related adverse events, allowing the investigation of treatment tolerance as a potential mediator between host metabolic status and therapeutic outcomes. Third, internal validation using bootstrap resampling enhanced the robustness of the predictive model.

However, several limitations should be acknowledged. This was a retrospective single-center study, which may limit the generalizability of the findings. Furthermore, although the predefined exclusion criteria were designed to minimize confounding from conditions known to independently affect body composition, they may have introduced a degree of selection bias. In particular, the exclusion of patients with metabolic comorbidities or prior malignancies may mean that our study population does not fully represent the broader spectrum of LARC patients encountered in routine clinical practice. The study period was defined from March 2020 onward, as treatment protocols and pretreatment imaging schedules at our institution were not fully standardized prior to this period; inclusion of earlier patients may have introduced systematic bias into body composition measurements. The sample size was relatively modest, and external validation was not performed; therefore, the generalizability of the prediction model should be interpreted with caution, and validation in larger, multicenter cohorts will be essential before clinical application.

Regarding dietary factors, systematic assessment of individual dietary intake during CRT was not performed in this study. As patients received long-course CRT on an outpatient basis, precise quantification of daily caloric and protein consumption was not feasible in a standardized manner—a recognized limitation shared across retrospective body composition studies in oncology. Although patients at our institution routinely receive structured dietary counseling following colorectal cancer diagnosis and throughout treatment, individual dietary adherence during CRT was not formally recorded and may have independently contributed to the observed body composition changes. Future prospective studies should incorporate standardized nutritional assessments alongside CT-based body composition analysis to better disentangle the effects of dietary intake, treatment toxicity, and tumor biology. Physical activity during CRT was similarly not assessed. Finally, although recurrence, mortality, and follow-up duration were recorded, the study was not designed to formally evaluate long-term oncologic outcomes such as disease-free or overall survival, and median survival could not be estimated given the low mortality rate in both groups.

Comparison with similar research

Previous studies evaluating body composition in colorectal cancer have mainly focused on long-term outcomes such as survival or postoperative complications (12,13). For example, Liu et al. (14) reported that body composition parameters were associated with treatment outcomes after neoadjuvant therapy; however, their study primarily evaluated postoperative complications rather than tumor regression itself. Furthermore, CRT-AEs were not incorporated into their analysis. In contrast, the present study evaluated both tumor regression and treatment tolerance simultaneously, allowing a more integrated assessment of host-related factors during neoadjuvant therapy. These findings support the possibility that different body composition compartments contribute to treatment response through distinct biological pathways. Most previous studies evaluated baseline body composition alone, whereas our study additionally assessed dynamic changes during CRT, which may better reflect treatment-related metabolic adaptation. Supporting the broader clinical relevance of adipose tissue in LARC, a recent study by Kolenda et al. (15) reported that low visceral and subcutaneous adipose tissue indices were significantly associated with worse overall survival in LARC patients undergoing multimodal treatment. Notably, skeletal muscle parameters were not independently associated with survival in their cohort, which aligns with the compartment-specific roles of body composition observed in our study, where skeletal muscle dynamics were more closely linked to treatment tolerance rather than tumor regression.

Explanations of findings

One possible explanation for these observations lies in the systemic metabolic alterations that occur during cancer treatment. Cancer cachexia is characterized by progressive skeletal muscle wasting mediated by catabolic pathways such as the ubiquitin-proteasome system and the autophagy-lysosome pathway (16,17). In cancer patients, disruption of the balance between anabolic and catabolic signaling promotes muscle degradation and reduces physiological reserve. Tumor downsizing induced by CRT may partially alleviate cachexia-related metabolic stress (18,19). In this context, skeletal muscle status may reflect the host’s physiological capacity to tolerate treatment-related stress. Consistent with this interpretation, our analysis showed that SMI was significantly associated with the occurrence of CRT-AEs, supporting the concept that muscle status is closely linked to treatment tolerance.

Interestingly, lower baseline PMI was associated with high TRG in our cohort. However, this observation should be interpreted cautiously and does not necessarily indicate that reduced muscle mass directly improves treatment response. Instead, it may reflect differences in body composition distribution or metabolic status among patients. For example, patients with relatively lower psoas muscle index may have had higher relative adiposity, which could partly explain the observed association between adipose tissue dynamics and tumor regression. In addition, because the psoas muscle represents only a limited component of total skeletal muscle mass, PMI may not fully capture whole-body muscle status (20,21). Therefore, this observation warrants further investigation in larger prospective studies.

In contrast to skeletal muscle, adipose tissue demonstrates complex metabolic and immunologic functions in cancer. Adipose tissue is not only an energy storage organ but also an endocrine tissue that secretes adipokines, cytokines, and inflammatory mediators capable of modulating systemic metabolism and immune responses. Visceral adipose tissue (VAT) in particular has been shown to exhibit greater metabolic activity and stronger inflammatory signaling compared with subcutaneous fat. In our study, smaller reductions in visceral fat during CRT were associated with high TRG. Rather than suggesting a direct causal relationship, this observation may reflect differences in systemic metabolic adaptation during treatment, whereas excessive depletion of VAT may indicate metabolic stress or inflammatory activation. Supporting this concept, experimental studies have shown that factors derived from VAT can influence tumor-associated inflammatory pathways and modulate cellular responses to radiation (22,23). Adipokines such as adiponectin have also been reported to exert antitumor effects in experimental models.

RT-related tissue responses may also help explain these observations. Experimental studies have demonstrated that irradiated adipose tissue can undergo structural changes, including adipocyte atrophy and inflammatory cell infiltration (24), while irradiated skeletal muscle may develop fibrosis and inflammatory responses (25,26). Unlike studies focusing on the direct local effects of radiation within the pelvis, our analysis assessed body composition at the level of the third lumbar vertebra, providing a systemic measure of host metabolic status during treatment. This systemic perspective may help contextualize the observed association between visceral fat dynamics and tumor regression in our cohort.

CRT-AEs may act as a potential mediator linking host body composition to effective treatment delivery. Neoadjuvant CRT for rectal cancer primarily relies on RT for local tumor control, with chemotherapy administered mainly as a radiosensitizer (27). Treatment-related toxicity may therefore compromise treatment delivery and overall therapeutic effectiveness. Previous studies examining systemic chemotherapy have shown that reduced skeletal muscle mass is associated with increased toxicity and dose-limiting adverse events (28,29). In our analysis, SMI was significantly associated with CRT-AEs, suggesting that muscle status may influence treatment tolerance and indirectly affect therapeutic outcomes during CRT.

Implications and actions needed

From a clinical perspective, these findings may have several implications. CT-based body composition analysis can be readily obtained from routine staging imaging and therefore does not impose additional burden on patients. Importantly, the clinical utility of body composition parameters should be interpreted differently depending on their timing. Baseline measures—particularly pre-CRT subcutaneous fat index and pre-CRT psoas muscle index—are available before treatment initiation and may serve as practical pretreatment indicators for risk stratification. Identifying patients with unfavorable baseline body composition profiles prior to CRT could support more individualized treatment planning and early initiation of supportive interventions, such as nutritional optimization or prehabilitation. In contrast, dynamic changes in body composition during CRT cannot be assessed in advance and are therefore not directly applicable to pretreatment decision-making. However, these changes may provide valuable real-time insight into treatment-related metabolic adaptation. Monitoring body composition dynamics during CRT could help guide supportive care strategies—such as nutritional supplementation or physical rehabilitation—rather than modifying the treatment protocol itself. Furthermore, in the evolving era of TNT, where induction chemotherapy precedes CRT, there exists a potential window for longitudinal assessment. Serial body composition measurements between treatment phases may offer an opportunity for adaptive clinical decision-making, such as optimizing supportive care or informing subsequent treatment intensity and surgical planning. Incorporating CT-based body composition assessment into routine clinical evaluation may therefore facilitate more personalized treatment strategies and supportive care in the evolving era of neoadjuvant therapy for rectal cancer.

Conclusions

This retrospective study suggests that body composition dynamics are associated with tumor regression and treatment tolerance in patients with LARC undergoing neoadjuvant CRT. High TRG was associated with the absence of CRT-AEs, higher baseline subcutaneous fat, and a smaller reduction in visceral fat during treatment. SMI was primarily related to treatment tolerance, whereas adipose tissue parameters were more closely associated with tumor response, suggesting compartment-specific roles of body composition during CRT. Although the predictive model showed moderate discriminative performance (AUC =0.707), CT-based body composition analysis may provide a practical, noninvasive tool to complement clinical assessment and support more individualized treatment strategies in the evolving era of TNT. The nomogram presented in this study should be regarded as exploratory, and external validation in larger, multicenter cohorts will be essential before clinical implementation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0269/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0269/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0269/prf

Funding: This study was supported by the Soonchunhyang University Research Fund.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2026-0269/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Institutional Review Board of Soonchunhyang University Cheonan Hospital (IRB No. 2022-06-022). Given the retrospective design of the study, the requirement for informed consent was waived by the Institutional Review Board.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Roshandel G, Ghasemi-Kebria F, Malekzadeh R. Colorectal Cancer: Epidemiology, Risk Factors, and Prevention. Cancers (Basel) 2024;16:1530. [Crossref] [PubMed]
2. Boublikova L, Novakova A, Simsa J, et al. Total neoadjuvant therapy in rectal cancer: the evidence and expectations. Crit Rev Oncol Hematol 2023;192:104196. [Crossref] [PubMed]
3. Soldati L, Di Renzo L, Jirillo E, et al. The influence of diet on anti-cancer immune responsiveness. J Transl Med 2018;16:75. [Crossref] [PubMed]
4. Ni J, Zhang L. Cancer Cachexia: Definition, Staging, and Emerging Treatments. Cancer Manag Res 2020;12:5597-605. [Crossref] [PubMed]
5. Holmes CJ, Racette SB. The Utility of Body Composition Assessment in Nutrition and Clinical Practice: An Overview of Current Methodology. Nutrients 2021;13:2493. [Crossref] [PubMed]
6. Gibson RS. Principles of nutritional assessment. 2nd ed. New York: Oxford University Press; 2005.
7. Nogueiro J, Santos-Sousa H, Pereira A, et al. The impact of the prognostic nutritional index (PNI) in gastric cancer. Langenbecks Arch Surg 2022;407:2703-14. [Crossref] [PubMed]
8. Onodera T, Goseki N, Kosaki G. Prognostic nutritional index in gastrointestinal surgery of malnourished cancer patients. Nihon Geka Gakkai Zasshi 1984;85:1001-5.
9. Siqueira JM, Soares JDP, Borges TC, et al. High neutrophil to lymphocytes ratio is associated with nutritional risk in hospitalised, unselected cancer patients: a cross-sectional study. Sci Rep 2021;11:17120. [Crossref] [PubMed]
10. Yaacobi Peretz S, Kessner R, Bar Y, et al. Body composition metrics as a determinant of trastuzumab deruxtecan related toxicity and response. NPJ Breast Cancer 2025;11:38. [Crossref] [PubMed]
11. Fujiwara N, Nakagawa H, Kudo Y, et al. Sarcopenia, intramuscular fat deposition, and visceral adiposity independently predict the outcomes of hepatocellular carcinoma. J Hepatol 2015;63:131-40. [Crossref] [PubMed]
12. De Nardi P, Giani A, Maggi G, et al. Relation between skeletal muscle volume and prognosis in rectal cancer patients undergoing neoadjuvant therapy. World J Gastrointest Oncol 2022;14:423-33. [Crossref] [PubMed]
13. Miranda AL, da Costa Pereira JP, de Sousa IM, et al. Impact of body composition and muscle health phenotypes on survival outcomes in colorectal cancer: a multicenter cohort. Sci Rep 2024;14:31816. [Crossref] [PubMed]
14. Liu Z, Lu S, Wang Y, et al. Impact of Body Composition During Neoadjuvant Chemoradiotherapy on Complications, Survival and Tumor Response in Patients With Locally Advanced Rectal Cancer. Front Nutr 2022;9:796601. [Crossref] [PubMed]
15. Kolenda P, Mardas M, Radomyski P, et al. Body Composition and Survival in Locally Advanced Rectal Cancer Patients Treated with Neoadjuvant Radiochemotherapy. Nutrients 2025;17:3309. [Crossref] [PubMed]
16. Setiawan T, Sari IN, Wijaya YT, et al. Cancer cachexia: molecular mechanisms and treatment strategies. J Hematol Oncol 2023;16:54. [Crossref] [PubMed]
17. Fearon K, Strasser F, Anker SD, et al. Definition and classification of cancer cachexia: an international consensus. Lancet Oncol 2011;12:489-95. [Crossref] [PubMed]
18. Albrecht HC, Wagner S, Sandbrink C, et al. Downsizing of rectal cancer following neoadjuvant radiotherapy (5 × 5 Gy) and long interval surgery evaluated using MRI semiautomated volumetric measurements, a retrospective study. Front Surg 2023;10:1106177. [Crossref] [PubMed]
19. Van den Begin R, Kleijnen JP, Engels B, et al. Tumor volume regression during preoperative chemoradiotherapy for rectal cancer: a prospective observational study with weekly MRI. Acta Oncol 2018;57:723-7. [Crossref] [PubMed]
20. Pigneur F, Di Palma M, Raynard B, et al. Psoas muscle index is not representative of skeletal muscle index for evaluating cancer sarcopenia. J Cachexia Sarcopenia Muscle 2023;14:1613-20. [Crossref] [PubMed]
21. Rutten IJG, Ubachs J, Kruitwagen RFPM, et al. Psoas muscle area is not representative of total skeletal muscle area in the assessment of sarcopenia in ovarian cancer. J Cachexia Sarcopenia Muscle 2017;8:630-8. [Crossref] [PubMed]
22. Chen M, Huang J. The expanded role of fatty acid metabolism in cancer: new aspects and targets. Precis Clin Med 2019;2:183-91. [Crossref] [PubMed]
23. Gyamfi J, Kim J, Choi J. Cancer as a Metabolic Disorder. Int J Mol Sci 2022;23:1155. [Crossref] [PubMed]
24. Kosmacek EA, Oberley-Deegan RE. Adipocytes protect fibroblasts from radiation-induced damage by adiponectin secretion. Sci Rep 2020;10:12616. [Crossref] [PubMed]
25. Wallner C, Drysch M, Hahn SA, et al. Alterations in pectoralis muscle cell characteristics after radiation of the human breast in situ. J Radiat Res 2019;60:825-30. [Crossref] [PubMed]
26. D'Souza D, Roubos S, Larkin J, et al. The Late Effects of Radiation Therapy on Skeletal Muscle Morphology and Progenitor Cell Content are Influenced by Diet-Induced Obesity and Exercise Training in Male Mice. Sci Rep 2019;9:6691. [Crossref] [PubMed]
27. Li Y, Wang J, Ma X, et al. A Review of Neoadjuvant Chemoradiotherapy for Locally Advanced Rectal Cancer. Int J Biol Sci 2016;12:1022-31. [Crossref] [PubMed]
28. Choi CS, Kin K, Cao K, et al. The association of body composition on chemotherapy toxicities in non-metastatic colorectal cancer patients: a systematic review. ANZ J Surg 2024;94:327-34. [Crossref] [PubMed]
29. Chiloiro G, Cintoni M, Palombaro M, et al. Impact of body composition parameters on radiation therapy compliance in locally advanced rectal cancer: A retrospective observational analysis. Clin Transl Radiat Oncol 2024;47:100789. [Crossref] [PubMed]