RF12 | PSUN80 A Novel DXA Algorithm to Aid in the Diagnosis of Familial Partial Lipodystrophy

Abstract

Abstract Familial partial lipodystrophy (FPLD) is a rare and heterogenous disease without gold-standard diagnostic criteria. The heterogeneity applies to fat distribution, symptomatology, metabolic complications, and genetic etiology. Disease heterogeneity has contributed to the lack of standardized diagnostic criteria, an obstacle to understanding both the natural history of the disease and response to therapies. We have assessed whether DXA data can aid in disease diagnosis. DXA scans quantify fat and lean mass in the whole body, arms, legs, and trunk. An increased ratio of fat percentage in the trunk to the legs (termed "trunk-leg ratio", TLR) has been proposed as an objective method to detect FPLD. Using large sets of DXA data from control subjects, along with data from FPLD subjects, we sought to improve upon TLR. We hypothesized that TLR might change as a function of overall adiposity (total body fat, TF). Females were studied first because of the high female: male ratio of disease prevalence observed in FPLD. In 2713 control females with DXA measurements within the UKBiobank (UKB), TLR increased linearly as a function of TF until a TF of ∼36% and then the TLR plateaued with higher TF. We analyzed published DXA data of FPLD subjects from the USA (n=55).1 FPLD subjects had higher TLRs than controls at each level of adiposity. Due to the non-linear relationship between total adiposity and TLR, two linear fits were generated to best discriminate controls from FPLD subjects for a given adiposity. For females with body fat <36%, the algorithm predicts FPLD if TLR >TF*0.0311+0.232. For females with body fat ≥36%, the algorithm predicts FPLD if TLR >1.353. After training on this UKB data set, we tested the algorithm on independent data sets of control females from NHANES (n=2347) and FPLD subjects (n=37, from the NIH). The algorithm had a sensitivity of 92–97% and a specificity of 99.3%. This degree of specificity is still not sufficient for implementation of rare disease screening in the general population, but could be useful in a population with increased pre-test probability of having FPLD such as non-obese individuals with Type 2 diabetes and/or hypertriglyceridemia. Because race and ethnicity are possible modifiers of body fat distribution, we examined the relationship between TF and TLR across ethnic and racial groups in NHANES. Mexican-Americans (n=659), Hispanic-other (n=81), White (n=871), Black (n=612), and other/biracial (n=124) subjects from NHANES had a similar relationship between adiposity and TLR. Therefore, the same DXA criteria can be applied across these ethnicities. In summary, the relationship between adiposity and TLR may be used to aid in the diagnosis of FPLD in females. Reference Meral R, et al. Diabetes Care. 2018;41,2255–2258. Presentation: Saturday, June 11, 2022 1:06 p.m. - 1:11 p.m., Sunday, June 12, 2022 12:30 p.m. - 2:30 p.m.