Portable ultrasound shows closer agreement than bioimpedance with dual-energy X-ray absorptiometry for body composition monitoring after sleeve gastrectomy.

Background

AND

Aims

Accurate body composition assessment is increasingly recognized as an important component of clinical nutrition assessment in obesity care; however, access to dual-energy X-ray absorptiometry (DXA) remains limited. Portable alternatives such as bioimpedance spectroscopy (BIS) and ultrasound may improve accessibility, but evidence supporting their ability to monitor body composition change following metabolic bariatric surgery is limited. This study compared portable ultrasound and BIS with DXA for assessing body composition before and after sleeve gastrectomy.

Methods

In this prospective cohort study, women with obesity undergoing primary sleeve gastrectomy at three Melbourne hospitals underwent body composition assessment before surgery and 12 months postoperatively using DXA (reference method), BIS and a five-site ultrasound protocol. Cohort-derived regression equations estimated DXA-derived fat mass and lean soft tissue from ultrasound measurements. Agreement was evaluated using Lin's concordance correlation coefficients, Bland-Altman analyses and mixed-effects models.

Results

Fifty women were enrolled and 29 (58%) completed paired DXA assessments at 12 months, with paired BIS (n=28) and ultrasound (n=27) data available following quality-control exclusions. Ultrasound-based estimates showed high cross-sectional concordance with DXA (fat mass CCC=0.97-0.98; lean soft tissue CCC=0.83-0.91). For longitudinal change, correlations with DXA were strong for fat mass (r=0.97, p<0.001) and moderate correlation for lean soft tissue (r=0.53, p=0.006), with minimal bias and limits of agreement of approximately ±7 kg. BIS showed lower agreement and greater variability, particularly for lean tissue change.

Conclusion

Following sleeve gastrectomy, cohort-derived ultrasound estimates showed promising agreement with DXA-derived body composition changes, particularly for fat mass, whereas BIS estimates showed greater variability, particularly for lean tissue change. Individual-level variability remained, and the ultrasound equations were developed and evaluated within the same cohort without internal validation; independent validation is therefore required. CLINICAL TRIAL REGISTRATION https://www.anzctr.org.au/ (Study ID ACTRN12625001233415).