For decades, body mass index has been the default screening tool for obstructive sleep apnea. A high BMI raises suspicion; a normal BMI often redirects the clinical gaze elsewhere. Yet BMI is fundamentally a proxy. It cannot distinguish muscle from fat or say where on the body that fat lives. In East Asian populations, where obesity-related conditions emerge at lower BMIs than in Western cohorts, that blind spot may be especially costly.
A new Mendelian randomization study from Peking University Third Hospital provides genetic evidence that body fat percentage, independent of BMI, causally increases OSA risk in East Asians. And in a multivariate analysis that directly pitted the two measures against each other, body fat emerged as the dominant driver.
What They Found
Researchers led by Rui Fan conducted a bidirectional and multivariate Mendelian randomization analysis using data from three large East Asian biobanks. OSA diagnosis came from Japan’s NBDC database (473 cases, 178,337 controls), identified by ICD-10 code G47.3 with confirmation through self-reported symptoms, physical examination, and sleep recordings showing an apnea-hypopnea index of 5 or more. BMI data drew on the same Japanese cohort (158,284 participants). Body fat percentage came from the Taiwan Biobank (102,900 individuals), measured by bioelectrical impedance.
Genetic instruments were selected with standard MR stringency: genome-wide significance at P<5×10^{-8} (relaxed to P<5×10^{-6} for the reverse direction), linkage disequilibrium clumping at 10,000 kb with r2<0.001, and an F-statistic above 10 to ensure instrument strength. The primary analysis used inverse-variance weighted MR, supplemented by MR-Egger, weighted median, weighted model, and a battery of sensitivity checks including Cochran’s Q test, MR-Egger intercept test, funnel plot inspection, leave-one-out analysis, and MR-Presso.
The forward results were clear. Genetically predicted body fat percentage was significantly associated with OSA risk using 25 genetic instruments (IVW beta = 1.272, 95% CI 0.288 to 2.257, p = 0.011). BMI, tested with 59 instruments, also showed a significant association (beta = 0.817, 95% CI 0.217 to 1.417, p = 0.008). Reverse MR direction found no evidence that OSA causally increases BFP (p = 0.127) or BMI (p = 0.833), supporting a unidirectional path from higher adiposity to sleep apnea.
The Multivariate Surprise
The headline finding came from the multivariate analysis. When body fat percentage was isolated by removing BMI’s effect, it remained a significant predictor of OSA risk (beta = 0.774, 95% CI 0.158 to 1.330, p = 0.013). But when BMI was isolated by removing body fat percentage’s effect, its association became non-significant (beta = -0.251, 95% CI -0.708 to 0.206, p = 0.282).
This asymmetry suggests that BMI may capture OSA risk largely through its correlation with body fat, rather than through an independent biological pathway. The authors are careful to qualify the finding: “These patterns likely indicate causally interconnected pathways… rather than entirely independent biological effects.” That caveat matters. Adipose tissue does not work in neatly separable compartments; visceral fat, subcutaneous fat, and overall mass interact through shared endocrine and inflammatory mechanisms. Nonetheless, the multivariate asymmetry is striking.
Why It Matters
If these genetic findings replicate in clinical settings, the implications for OSA screening are direct. Current risk assessment tools that lean heavily on BMI thresholds may systematically under-flag East Asian patients with a normal BMI but elevated body fat, a body composition profile more common in Asian populations, who tend to store greater visceral adiposity at lower body weights than Europeans.
The authors are explicit about what their study does and does not prove: “Our study provides only genetic evidence suggesting higher BFP causally increases OSA risk and may be conceptually superior to BMI as a biomarker. However, our research cannot validate BFP as a clinical screening marker for OSA, as this requires validation through larger scale phenotyping studies.” That validation gap is the next step. Waist circumference, bioelectrical impedance, and DXA scans are all candidates, but none has been tested head-to-head against current BMI-based screening in a prospective OSA detection study.
Limits
The study has important constraints that should temper enthusiasm for immediate clinical uptake. All data come from East Asian populations, and the authors note that differences in fat distribution patterns and airway anatomy limit generalizability to other ancestries. The reverse MR analyses had very few genetic instruments (one or two), yielding low statistical power to detect a causal effect in the opposite direction. OSA was identified through ICD codes, which do not capture disease severity, and mild, moderate, and severe cases are all lumped together. Body fat percentage was measured by bioelectrical impedance rather than the gold-standard DXA, introducing measurement imprecision. And while the MR design avoids confounding, it tests lifelong genetic effects, not the effects of weight change in adulthood that clinicians typically manage.
Bottom Line
Body fat percentage appears to drive OSA risk in East Asians more directly than BMI does, and BMI’s apparent association may be largely a reflection of the body fat it imperfectly tracks. For clinicians, the takeaway is not to abandon BMI, but to recognize its limits in patients whose body composition falls outside the population used to calibrate BMI thresholds. The next step is clear: prospective studies that put body composition measures to the test against real-world OSA diagnosis. Until then, what the scale cannot see, body fat measurement might.
Source
Fan R, Lu WJ, Zhang H, Jiang HL, Li T, Yan Y. Beyond the Scale: Body Fat Percentage Reveals Sleep Apnea Risk That BMI Misses. World J Otorhinolaryngol Head Neck Surg. 2026 Apr 1. DOI: 10.1002/wjo2.70100. PMID: 42500104. PMCID: PMC13399181.