When cancer researchers speak of body composition, they are referring to the proportions of muscle, fat, and other tissues that make up the body - not simply how much someone weighs. Sarcopenia, the abnormal loss of skeletal muscle mass, has been recognized as an important prognostic factor in many cancers because muscle serves far more functions than movement: it is a metabolic reservoir, an immune support system, and a buffer against the physical demands of treatment.
Patients with sarcopenia tolerate surgery and chemotherapy poorly, experience more complications, and tend to recover more slowly. One proposed biological mechanism involves inflammatory signaling: low muscle mass is associated with elevated levels of pro-inflammatory cytokines including IL-6 and TNF-alpha, which can stimulate tumor cell proliferation and promote resistance to chemotherapy drugs.
Body composition is most often measured from CT scans that patients already receive for staging or follow-up. By analyzing specific cross-sections of these scans - particularly at the level of the third lumbar vertebra (L3) - radiologists can calculate the skeletal muscle index (SMI), which normalizes muscle area to the patient's height. Despite this method's widespread use, a fundamental limitation is that it measures muscle at one single slice rather than across the entire relevant region of the body.
This study enrolled 385 endometrial cancer patients treated at Seoul National University Hospital between 2014 and 2018. All had pre-treatment CT scans available for analysis. The researchers used a commercially available AI software tool called DEEPCATCH v1.0.0.0 to automatically segment and measure body composition from these scans - eliminating the need for manual slice-by-slice measurement.
DEEPCATCH produced two different types of body composition measurements from the same scans. The first was the traditional L3 skeletal muscle index (L3 SMI): muscle cross-sectional area at the L3 vertebral level divided by height squared, with sarcopenia defined as L3 SMI below 39.0 cm2/m2. The second was a novel volumetric SMI: the total volume of skeletal muscle measured across the entire waist region, from L1 to L5 vertebrae, normalized to height squared. Sarcopenia by this method was defined as volumetric SMI below 206.0 cm3/m3.
The study also measured fat tissue volumes (total fat, visceral fat, subcutaneous fat) and calculated corresponding fat indices. This allowed comparison of whether muscle mass, fat distribution, or both were relevant to endometrial cancer outcomes. Kaplan-Meier survival analysis and multivariate Cox regression were used to determine which body composition variables independently predicted progression-free and overall survival.
The central finding was a sharp divergence between the two measurement approaches. L3 SMI-based sarcopenia showed no significant association with survival outcomes: progression-free survival did not differ between sarcopenic and non-sarcopenic patients (p=0.335), nor did overall survival (p=0.241). By the traditional single-slice method, muscle mass appeared irrelevant to endometrial cancer prognosis.
The volumetric approach told a completely different story. Patients with low volumetric SMI (below 206.0 cm3/m3) had substantially worse 5-year progression-free survival compared to those with normal volumetric muscle mass: 77.3% versus 88.8% (p=0.004). The difference in overall survival was even more striking: 92.8% versus 99.4% (p=0.003). In multivariate analysis that adjusted for age, stage, grade, and BMI, low volumetric SMI remained an independent predictor of worse PFS (adjusted hazard ratio 1.762, 95% CI 1.051-2.953) and worse OS (aHR 5.964, 95% CI 1.296-27.448).
The weak correlation between L3 SMI and volumetric SMI (Pearson r=0.266) explains why the two methods gave such different results. A single cross-section at L3 captures only a fraction of total waist muscle volume, and can misclassify patients whose muscle mass distribution is uneven across vertebral levels. The volumetric approach, by integrating muscle across five lumbar levels, provides a far more complete picture of the patient's actual muscle reserve.
In addition to muscle, the researchers measured total fat volume, visceral fat volume, and subcutaneous fat volume from the same CT scans. None of these fat tissue measurements were significantly associated with either progression-free or overall survival in endometrial cancer patients.
This finding may seem counterintuitive given that obesity is a well-established risk factor for developing endometrial cancer in the first place. The apparent contradiction is explained by what researchers call the obesity paradox: in patients who already have cancer, higher body fat has been observed to correlate with better - not worse - survival in some studies, possibly because adipose tissue serves as a caloric reserve during treatment and illness.
The dissociation between fat volume and muscle volume as prognostic factors supports the idea that it is specifically muscle depletion - not overall body weight or fat accumulation - that drives the adverse survival effects associated with body composition in endometrial cancer. BMI itself was also not an independent predictor of survival in multivariate analysis, further reinforcing that weight alone is a poor proxy for the metabolic and functional status of a cancer patient.
The fundamental advantage of volumetric body composition measurement is completeness. The L3 method - which has been standard in oncology body composition research for decades - was adopted as a practical compromise: L3 was identified as the single slice that best correlates with whole-body muscle mass in healthy subjects. But this correlation is far from perfect in cancer patients, who may have asymmetric muscle wasting, altered fat distribution, or structural changes from prior surgery.
By measuring the entire waist region from L1 through L5, the volumetric approach captures real biological variation that the single-slice method averages away. The study's finding that L3 SMI and volumetric SMI correlate only weakly (r=0.266) suggests that in a substantial proportion of patients, the L3 slice is genuinely unrepresentative of total muscle reserve. For these patients, single-slice classification produces incorrect results - and as the survival data showed, these misclassifications have direct clinical consequences.
The automation provided by AI-based segmentation like DEEPCATCH addresses the practical barrier that made volumetric measurement historically impractical. Manual volumetric segmentation requires radiologists to trace muscle boundaries across dozens of CT slices - a process too time-consuming for routine clinical use. AI tools that can do this automatically from standard CT scans used for cancer staging could make volumetric body composition assessment a feasible addition to existing clinical workflows without adding significant burden.
This study is the first to apply waist volumetric body composition measurement to endometrial cancer prognosis, and its results make a clear case for replacing the traditional single-slice approach with volumetric assessment. Where L3 SMI failed to identify any survival signal, volumetric SMI revealed an independent and clinically meaningful prognostic factor - one that remained significant even after adjusting for established risk factors including stage, grade, and age.
For clinicians, the practical implication is that sarcopenia should be assessed volumetrically rather than from a single CT slice when pre-treatment CT scans are available. Low volumetric SMI could flag patients who require nutritional support, resistance exercise programs, or closer monitoring before and during treatment. Given that sarcopenia is potentially modifiable through targeted interventions, identifying it accurately before treatment begins creates a window for intervention that the single-slice method would systematically miss.
Important limitations include the retrospective single-institution design with a Korean patient population, which may limit generalizability. The optimal cut-off value for volumetric SMI (206.0 cm3/m3) was derived from this cohort and requires external validation before widespread adoption. Future studies should test this approach in diverse populations and investigate whether correcting sarcopenia through prehabilitation interventions actually improves survival outcomes.