Beyond Group Averages: Rethinking Body Composition Assessment in the Era of Metabolic Bariatric Surgery: A Perspective from Korea

Obesity, particularly in individuals with body mass index (BMI) ≥35 kg/m², remains one of the most challenging public health issues of the 21st century, and metabolic bariatric surgery (MBS) is an established intervention for sustained weight loss and metabolic improvement. Under the Asia-Pacific/Korean classification, BMI 30.0–34.9 kg/m² and BMI ≥35.0 kg/m² correspond to class II and class III obesity, respectively; thus, the population discussed here largely overlaps with class III obesity by Asian criteria, although the featured article described the cohort as class II–V obesity.¹ However, the rapid and substantial weight loss after MBS does not consist solely of fat mass; a considerable proportion of lean soft tissue (LST) and fat-free mass (FFM) may also be lost, with potential adverse consequences for basal metabolic rate, muscle function, drug pharmacokinetics, and long-term metabolic outcomes. Accurate monitoring of body composition is therefore essential in the perioperative and postoperative management of patients undergoing MBS.

Dual-energy X-ray absorptiometry (DXA) is widely accepted as the reference standard for assessing body composition in clinical and research settings. Yet, DXA is expensive, requires specialized equipment and trained personnel, and is often inaccessible in routine bariatric clinics—particularly in resource-limited settings. Consequently, anthropometric predictive equations, which rely on simple measurements such as weight, height, and waist and hip circumferences, remain attractive alternatives. However, most of these equations were originally developed and validated in populations with normal weight or class I obesity, raising legitimate concerns about their applicability to patients with class II obesity or higher.

In this issue of the Journal of Obesity & Metabolic Syndrome, Gómez-Bruton et al.¹ address this important clinical gap through a well-designed multicenter observational study conducted across three research centers in Spain and Portugal. The authors evaluated the accuracy of several existing predictive equations for estimating LST and FFM in 123 patients with obesity and BMI >35 kg/m² (described in the featured article as class II–V obesity), comparing anthropometric estimates against DXA measurements obtained both before and approximately 1 month after MBS. The study provides several noteworthy contributions to the field.¹

First, the multicenter design and the inclusion of patients across a broad spectrum of higher obesity classes enhance the external validity of the findings and distinguish the study from earlier single-center validations. Second, the authors employed a robust statistical framework—including paired t-tests, constant error, mean absolute error, Pearson correlation, intraclass correlation coefficients, and 95% limits of agreement—providing a comprehensive picture of both systematic bias and precision. Third, and perhaps most clinically relevant, the study did not limit itself to a single time-point comparison, but also examined the ability of these equations to track dynamic body composition changes following the rapid weight loss induced by MBS, a scenario that has rarely been investigated.¹

The results are informative and, in some respects, reassuring. At the group level, the Li et al.² equation for individuals with higher BMI values, derived from lean body mass prediction equations developed in Chinese adults,² showed the closest agreement with DXA-derived preoperative FFM (mean difference, 0.01 kg; 95% confidence interval, −0.74 to 0.74), whereas the Salamat et al.³ equation, developed for anthropometric estimation of body composition including lean mass, showed the closest agreement for preoperative LST (mean difference, 0.03 kg). For postoperative change scores, the Li et al.² equation developed for individuals with normal BMI values from the same derivation study best approximated change in FFM, whereas the Kulkarni et al.⁴ equation, developed in Indian men and women to estimate lean body mass and appendicular LST, best approximated change in LST. These findings suggest that, when the objective is population-level or epidemiological surveillance, certain equations may serve as acceptable low-cost surrogates for DXA. This has important implications for large-scale cohort studies, quality-improvement initiatives, and settings in which DXA is unavailable.¹

The authors deserve particular credit for their intellectual honesty in reporting a less encouraging observation: at the individual level, even the best-performing equations were markedly inconsistent. Only about 25% of participants fell within ±1 kg of the actual FFM change measured by DXA when the best-performing equation was used. This is a critically important message for clinicians—one that is often obscured when studies emphasize only mean differences and correlation coefficients. By explicitly cautioning against the use of these equations for individualized clinical decisions, such as drug dose calculations or personalized nutritional prescriptions, the authors provide a much-needed corrective to the growing tendency to substitute inexpensive tools for gold-standard measurements without adequate validation.¹

The findings of Gómez-Bruton et al.¹ are particularly relevant to the Korean metabolic and bariatric community. In Korea, National Health Insurance Service reimbursement for MBS was introduced in 2019 for patients with BMI ≥35 kg/m² or BMI ≥30 kg/m² with obesity-related comorbidities, with partial reimbursement also available for selected patients with inadequately controlled type 2 diabetes mellitus and BMI 27.5–30 kg/m². Procedure volume increased markedly after reimbursement was introduced and, although the uptake of anti-obesity medications such as glucagon-like peptide-1 receptor agonists may have altered subsequent growth patterns, MBS remains an important treatment option and postoperative body composition monitoring remains a routine clinical concern.^5,6 However, whether Iberian-derived predictive equations can be directly applied to Korean patients remains uncertain.

Several lines of evidence suggest caution. Koreans and other East Asians generally exhibit lower FFM and higher body fat percentages at a given BMI than Caucasians, and the distribution of appendicular skeletal muscle mass also differs meaningfully. In addition, a recent meta-analysis that included Korean and other Asian cohorts reported that FFM decreases significantly after MBS even when muscle strength is relatively preserved, underscoring the complexity of postoperative body composition assessment in Asian populations.⁷ Moreover, several Korean investigators have already developed and cross-validated ethnicity-specific predictive equations using Korea National Health and Nutrition Examination Survey data, supporting the view that body composition equations should ideally be tailored to the populations in which they are applied.^8,9 The Korean Sarcopenic Obesity Study has also highlighted the bidirectional relationship between visceral fat and skeletal muscle in Korean adults.¹⁰ Taken together, these observations imply that even if the Li et al.² or Salamat et al.³ equations perform well at the group level in Iberian patients with obesity and BMI >35 kg/m², their direct clinical adoption in Korean MBS practice would be premature.

Beyond ethnicity-related concerns, several additional limitations merit consideration. The follow-up period of approximately 1 month after MBS is relatively short. Given that changes in FFM and LST continue to evolve over 6 to 24 months after surgery, and that muscle loss during this later phase may carry greater clinical consequences, it remains uncertain whether the tested equations would perform similarly over longer follow-up intervals. In addition, two different DXA models were used across the participating centers without formal cross-calibration, and waist and hip circumference protocols were not fully standardized. Although these factors are unlikely to overturn the main conclusions, they may have introduced measurement heterogeneity. Finally, the study evaluated existing equations rather than developing a new predictive model specifically for patients with class II obesity or higher. In light of the individual-level inaccuracy observed, the field would benefit from equations derived and validated specifically in cohorts with obesity and BMI ≥35 kg/m², potentially incorporating additional predictors such as age, sex, ethnicity, and bioelectrical impedance-derived parameters.

Despite these limitations, the work by Gómez-Bruton et al.¹ represents an important step toward evidence-based body composition monitoring in metabolic bariatric care. It reinforces a critical principle that clinicians and researchers must not overlook: group-level accuracy does not guarantee individual-level reliability, and the choice of assessment tool must be aligned with the clinical question at hand. For epidemiological research and program evaluation, well-selected anthropometric equations may suffice; for individualized clinical decision-making in patients undergoing MBS, DXA—or comparably accurate methods—should remain the standard whenever feasible.

As the role of MBS continues to evolve within a broader obesity-treatment landscape that now includes increasingly effective pharmacotherapy, the need for accurate, accessible, and validated body composition tools remains substantial. The findings of Gómez-Bruton et al.¹ should encourage the Korean metabolic and bariatric community to initiate multicenter validation studies of existing equations in Korean patients with obesity and BMI ≥35 kg/m², develop and cross-validate ethnicity- and BMI-specific predictive equations tailored to the postoperative MBS setting, and pursue longer-term studies that capture the full trajectory of postoperative body composition change.