In our study we investigated associations of anthropometric markers derived from an automated 3D body scanner as well as manual measurement with FFM assessed by ADP and BIA in 1,593 individuals from Northeast Germany. For 3D body scanner markers of abdominal fat, we found strong inverse associations with relative FFM, which were stronger than those observed for the manually measured waist and hip circumferences. Correlations of the body scan markers with FFM were stronger in females than in males. Especially in males, correlations of body scan markers with FFM were higher for ADP FFM than for BIA FFM.
By automatically collecting 47 different anthropometric sites via 3D body scan, we were able to access a large, comprehensive collection of data in order to narrow down the vast bandwidth of circumferences, lengths, distances, areas or volumes to a lower number of relevant measurements. Anthropometric assessment has been endorsed as an alternative to technology-based measurements of low muscle mass, e.g., imaging modalities or BIA, in settings where resources are limited [35]. We found that eleven anthropometric markers were associated with both assessment methods of FFM and, thus, could be useful in the clinical assessment of a person’s FFM. Those markers were circumferences of belly, buttock, hip, waist, and maximum belly, as well as depths of the abdomen and buttock. By contrast, calf, and upper arm circumference commonly applied to assess FFM in clinical practice showed considerably weaker associations. Considering these findings, FFM may be particularly related to fat accumulation in the waist, hip, and abdomen area of a person, regardless of sex or age. Since circumferences in this region can be taken comparatively easy by an inelastic tape measurement, the assessment will not be dependent on a large stationary measuring device. However, non-circumference anthropometric markers may not be as easy to determine using an ordinary tape measure. For the parameters "body depth" and "buttock abdomen depth", for example, the two reference points for applying the measuring tape are not on a horizontal plane. It is possible to place the test person’s back against a wall and define this as the dorsal reference point, but the deformation of the buttocks is inevitable. In addition, it is not yet clear how to map the ventral reference point from there on a horizontal plane. In the case of the "abdominal depth" value, the measurement axis lies on a horizontal plane, but the dorsal reference point for applying the measuring tape is in the lumbar spine lordosis, so that simply placing the participant against the wall may not be an option. This suggests to use circumferences as convenient anthropometric markers in order to assess FFM in a clinical setting.
Our study showed that automatically derived anthropometric markers were stronger correlated to FFM than their manually measured equivalents. Circumferences measured by scan are shown to be slightly different from those derived by inelastic tape, which may be the result of a different measuring position. Assuming that parameters such as “high hip circumference” or “middle hip circumference” serve as the equivalent of manually measured, for instance, “hip circumference”, we hypothesize that either automatic body scanners have a systematic advantage over manual anthropometry or that it's possible that the locations of the circumference measurements that we found through our search were simply more useful. Repp et al. [36] showed that the site of waist circumference measurement is important to improve the prediction of visceral adipose tissue. The exact measuring regions of aforementioned manually measured circumferences are described and defined in the ISO 7250-1:2017 standard [37].
Ng et al. [38] described that 3D scans produce reliable data for estimating, among others, FFM as determined by ADP and BIA. This provided a basis for verifying the results of our study. However, the study by Ng et al. had a small study population. Moreover, the study population consisted exclusively of healthy adults, whereas our study examined a much larger study population with a broad phenotype. Bennett et al. [39] evaluated estimates of body composition and anthropometry provided by 3D imaging compared with manual anthropometry assessments and DXA scans. They observed a strong association for FFM and FM between body scanner and DXA. Unfortunately, this study investigation also had a rather small sample size.
In our analyses, the beta-estimates for the correlations between anthropometric parameters and FFM based on ADP and BIA were not similar. This was particularly obvious in men. In addition, the correlation between BIA and ADP FFM was substantially higher in females than in males, which may explain these findings. Previous studies reported differences in FFM between BIA, ADP and DXA to be higher in obese individuals [40, 41]. In our study, the male participants had a slightly higher BMI than females. However, this difference may not explain a possible systematic bias in measuring FFM by BIA. In that respect, previous studies investigating systematic sex specific differences in FFM by BIA revealed conflicting results [42–45]. There is no clear consensus on whether BIA overestimates or underestimates FFM in males, females, or even gender-independently.
In our study, the male study population showed a higher prevalence of hypertension and type 2 diabetes compared to the female study population. The presence of chronic diseases such as hypertension or type 2 diabetes were reported to be associated with a chronic inflammation or malnutrition leading to a reduced FFM on the one hand, while these metabolic conditions were associated with accumulation of visceral fat on the other hand [46, 47]. Since chronic inflammation, malnutrition and aberrant fat distribution can cause altered hydration status and BIA equations depend on the consistency of hydration status [13, 48, 49], this could be a possible explanation for the strong differences in correlations between BIA and ADP in men. However, this should still be part of further investigations.
Though we cannot exclude a systemic difference in FFM as measured by BIA, we can conclude that independent of the measuring method of FFM, we found the same eleven anthropometric markers to be correlated most strongly with FFM. This is not a potential shortcoming of the respective measurement methods for determining FFM, but a strength of these anthropometric markers.
Besides the comprehensiveness of anthropometric markers, the large number of individuals (n = 1593), the high level of quality assurance, particularly in standardization of non-invasive examination methods and data management represent strengths of the study [34]. The techniques accessed for estimating FFM are broadly used in clinical, as well as research settings and therefore have a high degree of validity on population level [15–17, 21, 22].
However, we refrained from using the gold standard of DXA measurement in favor of radiation exposure and practicality, leaving a minor amount of study strength on the way. In addition, our study population consisted exclusively of Caucasians, which supports validity within this ethnicity but limits conclusions about other study populations. Further research is necessary to investigate the relationship between anthropometry and FFM within other ethnicities. Another limitation is that manually measured anthropometric data carries the risk to be not as accurate as automatically scanned data, since an inter-observer bias cannot be excluded entirely [50]. Distinguishing whether there is a systematic difference between methods or a measurement area adjustment is critical to the correlation with FFM in order to transfer our results to an outpatient setting may be part of future research.