Measurement of body composition by DXA, BIA, Leg-to-leg BIA and near-infrared spectroscopy in breast cancer patients – comparison of the four methods

Abstract

Body composition plays a crucial role in therapy adherence and the prognosis of cancer patients. The aim of this work was to compare four measurement methods for determining body composition regarding their validity, reliability and practicability in order to be able to draft a practical recommendation as to which method is most suitable as a standard measurement method in oncology. Fat mass (FM) and fat-free mass (FFM) was estimated for 100 breast cancer patients with ages of 18-70 years during a defined 20-week inpatient and outpatient rehabilitation process after primary therapy or follow-up rehabilitation. The four methods used were dual energy X-ray absorptiometry (DXA), bioelectrical impedance analysis (BIA), leg-to-leg BIA and near-infrared spectroscopy (NIRS). At baseline (t0) and after 20 weeks (t20) the agreement between the four body composition analysis methods was quantified by pairwise method comparisons using Bland-Altman bias and limits of agreement estimates, t-tests and Lin's concordance correlation coefficients (CCCs). CCCs and Bland-Altman plots indicated that DXA and BIA, DXA and NIRS as well as BIA and NIRS showed an excellent agreement concerning FM estimation at both time points (CCC>0.9). In contrast, no methods agreed with a CCC higher than 0.9 with respect to FFM estimation. However, most estimates were also significantly different between two methods, except for BIA and NIRS which yielded comparable FFM and FM estimates at both time points, albeit with large 95% limits of agreement intervals. The agreement between DXA and BIA was best in the lowest BMI tertile and worsened as BMI increased. Significant differences were also found for FFM changes measured with DXA versus BIA (mean difference-0.4kg, p=0.0049), DXA versus to Leg-to-leg BIA (-0.6kg, p=0.00073) and for FM changes measured with DXA versus Leg-to-leg BIA (0.5kg, p=0.0011). For accurate and valid body composition estimates, Leg-to-leg BIA cannot be recommended due to its significant underestimation of FM or significant overestimation of FFM, respectively. BIA and NIRS results showed good agreement with the gold standard DXA. Therefore both measurement methods appear to be very well suitable to assess body composition of oncological patients and should be used more frequently on a routine basis to monitor the body composition of breast cancer patients.