Cancer cachexia is a multifactorial syndrome characterized by involuntary weight loss, skeletal muscle wasting, and systemic inflammation, and its evaluation has traditionally emphasized muscle-related parameters. Conventional sarcopenia definitions, such as those proposed by the European Working Group on Sarcopenia in Older People (EWGSOP) and the Asian Working Group for Sarcopenia (AWGS), identify low skeletal muscle mass, reduced muscle strength, and impaired physical performance as core diagnostic criteria [17, 18]. However, these frameworks do not incorporate adipose tissue parameters, particularly subcutaneous fat, into the diagnostic algorithms. Historically, fat has been regarded mainly as a secondary marker of nutritional status rather than a primary determinant of clinical outcomes.

This perspective has been increasingly challenged by emerging evidence. In colorectal cancer, greater SAT area and higher fat quality have been linked to improved surgical and long-term outcomes [19,20,21,22,23]. These findings highlight adipose tissue as an energy reservoir and an active metabolic and endocrine organ with prognostic relevance, underscoring the inadequacy of BMI as a sole measure. This evidence supports the integration of adipose compartments into evaluations of cancer cachexia alongside muscle parameters [1, 3]. In this context, our study provides new insights into the prognostic role of adipose tissue in pancreatic cancer. Early SAT loss within 3 months emerged as a stronger prognostic indicator than contemporaneous changes in SMI or VAT, suggesting that SAT dynamics capture disease aggressiveness at a stage when muscle-based indices remain relatively preserved. Across dose–response Cox models, Kaplan–Meier analyses, and ROC discrimination, SAT Δ% consistently demonstrated superior prognostic discrimination compared with VAT and SMI, supporting the concept of dynamic SAT change as an early, clinically accessible marker of cachexia biology. Importantly, the prognostic value of early SAT loss persisted after accounting for tumor response at 3 months. Even when RECIST response (PD) and baseline disease extent (stage IV) were included in the multivariable model, greater SAT loss remained associated with shorter landmark survival. These findings suggest that serial SAT assessment on routine CT provides prognostic information independent of RECIST.

Prior studies evaluating CT-derived body-composition changes during chemotherapy for pancreatic cancer have reported heterogeneous results, likely reflecting differences in patient populations, timing of assessment, and analytic strategy. In unresectable disease, Klassen et al. showed that overall adipose wasting (VAT + SAT) during palliative chemotherapy was associated with shorter survival even after accounting for concurrent skeletal muscle change and disease response, supporting the clinical relevance of early adipose loss [24]. In contrast, two studies focusing on compartment-specific changes reported no independent association between SAT-only change during the first 3 months of treatment and survival [7, 25]. In addition, phenotype-based analyses have suggested that fat-only wasting during chemotherapy is associated with inferior survival, underscoring the potential prognostic relevance of early adipose depletion [26]. Our findings extend this literature by demonstrating that dynamic SAT loss over a prespecified 3-month landmark remained associated with landmark survival after adjustment for RECIST response and disease stage in an Asian cohort. Several factors may explain the discrepant findings across studies, including differences in ethnicity/body habitus and baseline adipose distribution, treatment regimens and supportive care, the distribution of radiologic response, and methodological choices such as landmark definitions and covariate adjustment. By evaluating SAT, VAT, and SMI in parallel within a uniform landmark framework, our study helps refine the temporal characterization of tissue wasting and highlights early SAT loss as a pragmatic prognostic marker in unresectable pancreatic cancer.

Visceral and subcutaneous fat depots differ in anatomy, physiology, and measurement characteristics. VAT drains to the liver via the portal circulation, is more lipolytic and proinflammatory, and has been linked to adverse outcomes both at baseline and with very early VAT loss [11,12,13]. In contrast, SAT serves as a systemic energy buffer. Therefore, rapid depletion of SAT may indicate failure of this buffer under tumor-driven inflammation and catabolism, providing an early warning of clinical decline [3, 19,20,21,22,23]. Our 3-month framework likely captured this SAT-dominant phase, whereas VAT signals appear more dependent on disease stage and timing of measurement. From a technical standpoint, SAT measurement at L3 is generally more reproducible than VAT, which can be affected by bowel distension, contrast timing, and ascites. Beyond survival modeling, within-subject comparisons of percent change across compartments revealed heterogeneity, with SAT declining more steeply than VAT or SMI. This intraindividual hierarchy supports the concept of a SAT-dominant early phase of cachexia, wherein depletion of the subcutaneous “energy buffer” precedes overt skeletal muscle wasting. From a supportive care perspective, these observations are also consistent with the hypothesis that early SAT depletion reflects a negative energy balance and/or heightened tumor-driven catabolism during chemotherapy. If SAT loss identifies patients at risk of rapid nutritional decline, early supportive interventions—such as nutritional counseling, symptom control to maintain intake, and multimodal supportive care—may be warranted. However, because this was a retrospective study and detailed dietary intake data were unavailable, our findings should be interpreted as prognostic rather than causal. Prospective studies incorporating standardized nutritional assessments and supportive interventions are needed to determine whether preventing early SAT loss can improve outcomes.

To evaluate whether SAT loss mirrored chemotherapy efficacy, we performed landmark analyses stratified by RECIST response. These analyses showed that early SAT depletion was not merely a proxy for radiological response. The prognostic effect remained evident in the disease-control subgroup (SD/PR) and persisted after adjusting for RECIST status. The lack of significance in the PD group likely reflects the limited sample size and uniformly poor outcomes rather than a true interaction effect. Collectively, these results support the interpretation that short-term SAT loss represents treatment-independent progression of cachexia. We also explored the influence of serum albumin adjustment. As a negative acute-phase reactant, albumin reflects systemic inflammation and nutritional status. Incorporation of Δ-albumin into survival models attenuated the association between SAT loss and outcome. This attenuation may reflect partial mediation, wherein inflammation drives albumin decline and fat depletion, or confounding by illness severity, with both measures capturing parallel aspects of systemic stress. Accordingly, SAT loss and Δ-albumin could be considered complementary early risk markers rather than competing metrics. Prospective studies integrating serial inflammatory markers, composite immunonutritional scores, and joint longitudinal-survival modeling are needed to clarify these interrelationships.

Some limitations of this study should be considered while interpreting the results. First, this was a single-center retrospective study with a modest sample size, and the subgroup analyses were exploratory due to limited sample size. Second, the 3-month landmark design enhanced temporal interpretation but necessarily excluded early deaths, which may have introduced selection bias. Third, functional parameters such as muscle strength and physical performance, which are core elements of EWGSOP/AWGS criteria, were not evaluated. Fourth, longitudinal weight-loss data (e.g., over the preceding 6 months) were unavailable; therefore, we could not apply the international consensus staging system for pre-cachexia, cachexia, and refractory cachexia. In addition, we did not include functional measures such as grip strength, gait speed, or serial inflammatory biomarkers (e.g., IL-6), which are integral to cachexia definitions. This omission limits comprehensive characterization of the syndrome.

In conclusion, dynamic loss of SAT was strongly associated with poor prognosis in advanced pancreatic cancer, independent of muscle-based sarcopenia parameters. These findings underscore the role of adipose tissue as an active driver of cachexia and support routine assessment of SAT change on follow-up CT. When combined with simple laboratory indices such as albumin, SAT dynamics may enhance early risk stratification and inform timely supportive care and treatment decisions. From a supportive care perspective, early detection of SAT loss could specifically guide timely nutritional counseling, structured exercise programs, or anti-inflammatory interventions, aligning with multidisciplinary cachexia management strategies.