Background: The lipoprotein insulin resistance (LP-IR) score has been shown to assess insulin resistance, predict future type 2 diabetes, and improve with regular exercise. The lipidomic profile is known to differ with insulin resistance and type 2 diabetes, but few studies have examined lipidome associations with LP-IR, particularly in response to an exercise intervention. Methods: Plasma lipids were measured using the C8-positive LC-MS method in 671 participants from the HERITAGE Family Study (56% Female, 35% Black, 35.2 yrs) before and after a 20-week exercise intervention. LP-IR, a weighted index of six lipoprotein parameters, was measured before and after training through nuclear magnetic resonance spectroscopy (Labcorp, NC). Linear mixed models were used to test the associations of 193 known plasma lipids with LP-IR before and after exercise training. All models were adjusted for age, sex, and race, while post-training models also adjusted for baseline lipid and baseline LP-IR. Results: A total of 162 lipids (84%) were associated (FDR<0.05) with LP-IR at baseline ( Figure 1 ). The top positive associations were found for TG species, while cholesterol ester species showed the top inverse associations. Following training, the change in 143 (74%) lipids were associated (FDR<0.05) with change in LP-IR. There were 129 lipid species associated with both baseline and changes in LP-IR, while 33 were only associated with baseline measures and 14 only associated with changes in LP-IR following exercise ( Figure 1 ). Conclusions: We found that most lipids were associated with LP-IR regardless of exercise training status. Importantly, we identified a subset of lipids that were only associated with changes in LP-IR, which may represent exercise responsive biomarkers of exercise induced changes in LP-IR. However, further research is needed to identify the biological mechanisms connecting the changes in these lipid species and metabolic changes following exercise training.

Background and aims: Previous studies have derived and validated an HDL apolipoproteomic score (pCAD) that predicts coronary artery disease (CAD) risk. However, the associations between pCAD and markers of cardiometabolic health in healthy adults are not known, nor are the effects of regular exercise on pCAD. Methods: A total of 641 physically inactive adults free of cardiovascular disease from the HERITAGE Family Study completed 20 weeks of exercise training. The pCAD index (range 0-100) was calculated using measurements of apolipoproteins A-I, C-I, C-II, C-III, and C-IV from ApoA-I-tagged serum (higher index = higher CAD risk). The associations between pCAD index and cardiometabolic traits at baseline and their training responses were assessed with Spearman correlation and general linear models. A Bonferroni correction of p < 8.9 x 10(-04) was used to determine statistical significance. Results: The mean +/- SD baseline pCAD index was 29 +/- 32, with 106 (16.5 %) participants classified as high CAD risk. At baseline, pCAD index was positively associated with blood pressure, systemic inflammation, and body composition. HDL size, VO2max, and HDL-C were negatively associated with pCAD index at baseline. Of those classified as high CAD risk at baseline, 52 (49 %) were reclassified as normal risk after training. Following training, pCAD index changes were inversely correlated (p < 1.4 x 10(-04)) with changes in HDL-C, HDL size, and LDL size. Conclusions: A higher pCAD index was associated with a worse cardiometabolic profile at baseline but improved with regular exercise. The results from this study highlight the potential role of HDL apolipoproteins as therapeutic targets for lifestyle interventions, particularly in high-risk individuals.

Background: Prospective cohort studies have shown plasma cholesterol ester (CE) 18:2 is inversely associated with all-cause and cardiovascular mortality. However, less is known about the association of CE18:2 with cardiometabolic risk factors and whether it is responsive to regular exercise. Methods: A total of 193 known plasma lipids, including 11 CEs, were measured using the C8-positive LC-MS method in 671 participants from the HERITAGE Family Study (56% Female, 35% Black, 35 yrs) before and after a 20-week exercise intervention. Linear mixed models were used to test the associations of all lipids with 6 cardiometabolic traits at baseline and post-training adjusting for age, sex, and race, with post-training models also adjusted for baseline lipid and trait values. Paired t-tests were used to test the difference in CE18:2 abundance before and after exercise training. Results: Out of the 193 lipids, CE18:2 was consistently among the top 2 associations with each cardiometabolic trait at baseline ( Table 1 ). Levels of CE18:2 were inversely associated with triglycerides, visceral fat, glycA, and small LDL, while positively associated with HDL-C and insulin sensitivity. Importantly, CE18:2 abundance was increased following the exercise intervention (p=0.0003). Exercise-induced change in CE18:2 was significantly (FDR<0.05) associated with concomitant changes in all traits except visceral fat in the same directions as baseline, with CE18:2 among the top 3 associations with each training response trait ( Table 1 ). Conclusions: Plasma CE18:2 levels were associated with cardiometabolic traits in a favorable direction both before and after an exercise intervention. Thus, this plasma lipid may be an exercise inducible metabolite that is indicative of improvements in cardiometabolic health. However, additional studies are needed to confirm the observed findings and determine the mechanisms underlying these beneficial associations.

Background: Excessive visceral fat is associated with metabolic alterations and is a causal risk factor for CVD. The plasma lipidome is altered in obesity and lipidome signatures of BMI and obesity have been identified. However, few studies have examined the plasma lipidome in relation to visceral fat. Methods: Plasma lipids were measured using the C8-positive LC-MS method in 671 participants from the HERITAGE Family Study (56% Female, 35% Black, 35 yrs). Visceral fat was measured using CT scans. Linear mixed models were used to test the associations of 193 known plasma lipids with visceral fat adjusting for age, sex, race, and BMI. A FDR<5% was used to determine significance. Results: In individual models, 156 lipid species were significantly associated with visceral fat, with 109 species associated after additional adjustment for BMI ( Fig 1 ). The top positively associated lipids were triglycerides, while cholesterol esters showed the strongest inverse associations with visceral fat. A LASSO regression model retained 73 lipids and explained 78.5% of the variance in visceral fat. Delta visceral fat was calculated as the difference between predicted (from LASSO) and measured visceral fat. Examining quartiles of delta visceral fat showed that discordance between predicted and actual visceral fat was associated with differing cardiometabolic profiles independent of age, sex, race, visceral fat, and BMI. Individuals with higher delta visceral fat (Q4, overpredicted) had significantly (p<1.0x10 -04 ) higher levels of TG, apoB, total cholesterol, LDL-C, and fasting insulin and lower levels of large HDL particles and LPL activity compared to those with lower delta visceral fat (Q1, underpredicted). Conclusions: The plasma lipidome is widely associated with visceral fat levels. A lipidome-based visceral fat score may provide additional information over measured visceral fat for assessment of cardiometabolic health. Further studies are needed to test and validate the clinical utility of such lipidome-based scores.

Angiopoietin-like protein (ANGPTL) com-plexes 3/8 and 4/8 are established inhibitors of LPLand novel therapeutic targets for dyslipidemia. How-ever, the effects of regular exercise on ANGPTL3/8and ANGPTL4/8 are unknown. We characterizedANGPTL3/8 and ANGPTL4/8 and their relationshipwith in vivo measurements of lipase activities andcardiometabolic traits before and after a 5-monthendurance exercise training intervention in 642adults from the HERITAGE (HEalth, RIsk factors,exercise Training And GEnetics) Family Study. Atbaseline, higher levels of both ANGPTL3/8 andANGPTL4/8 were associated with a worse lipid, lipo-protein, and cardiometabolic profile, with onlyANGPTL3/8 associated with postheparin LPL and HLactivities. ANGPTL3/8 significantly decreased withexercise training, which corresponded with increasesin LPL activity and decreases in HL activity, plasmatriglycerides, apoB, visceral fat, and fasting insulin (allP<5.1 10(-4)) pound. Exercise-induced changes inANGPTL4/8 were directly correlated to concomitantchanges in total cholesterol, LDL-C, apoB, and HDL-triglycerides and inversely related to change in insu-lin sensitivity index (allP<7.0 10(-4)) pound. In conclusion,exercise-induced decreases in ANGPTL3/8 andANGPTL4/8 were related to concomitant improve-ments in lipase activity, lipid profile, and car-diometabolic risk factors.These findings reveal theANGPTL3-4-8 model as a potential molecular mecha-nism contributing to adaptations in lipid metabolismin response to exercise training.

Introduction: Angiopoietin like protein (ANGPTL) complexes 3/8 and 4/8 are established inhibitors of lipoprotein lipase and modifiable by regular exercise. However, the molecular biomarkers related to their exercise responses have not been fully elucidated. The purpose of this study was to examine the associations between plasma proteins and ANGPTL3/8 and 4/8 before and after exercise training. Methods: Measurements were taken before and after 20 weeks of exercise training in 630 adults (36% Black, 56% women, mean age 35 yrs) of the HERITAGE Family Study. Meso Scale Discovery immunoassays were used to measure ANGPTL3/8 and ANGPTL4/8 complexes in serum. Plasma proteins (n=4979 aptamers) were quantified using SomaScan. Linear mixed models tested the association between plasma proteins and each trait at baseline and post- training with full covariate adjustment. Significance was set to FDR<0.05. Results: A total of 862 and 126 proteins were significantly associated with ANGPTL3/8 at baseline or post-training, respectively, with 80 proteins associated at both time points, including PCSK9 and PLTP ( Figure 1A ). Conversely, 46 proteins were only associated with ANGPTL3/8 at post-training, 23 of which whose levels significantly changed with training including NRP1, ghrelin, HHIP, and APOA1. A total of 433 and 71 proteins were significantly associated with ANGPTL4/8 at baseline or post-training, respectively, with 46 proteins associated with both time points ( Figure 1B ). A total of 25 proteins were only associated with ANGPTL4/8 at post-training, 12 of which whose levels significantly changed with training including ghrelin, COL6A2, and MMP16. Conclusions: We identified a subset of proteins uniquely associated with exercise-induced changes in ANGPTL3/8 and 4/8. Thus, the responsiveness of ANGPTL3/8 and 4/8 to regular exercise may be related to cholesterol esterification, lipoprotein remodeling, inflammatory response, and regulation of fibrinolysis and coagulation among other biological pathways.

Introduction: Ischemic stroke is a complex heritable disease with a substantial proportion of risk attributable to polygenic factors. Little is known about how polygenic risk for stroke may manifest as intermediate phenotypes. Hypothesis: Polygenic liability for ischemic stroke will be associated with cardiometabolic phenotypes even in young adults free from disease. Methods: Polygenic risk derived from a multi ancestry population (GIGASTROKE ~1.2 million SNPs) was applied to 454 White adults from the HERITAGE Family Study with imputed whole genome data available. Participants (mean age = 31.5 ±14.5 years, 51% female) underwent robust cardiometabolic profiling including deeply phenotyped cardiopulmonary exercise tests, body composition, lipid panels, inflammatory markers, and measures of glucose homeostasis. General linear models adjusted for age and sex were used to test the association between z-scored polygenic risk of stroke (SPRS) and 130 phenotypes. A false discovery rate of <5% was used to determine significance. Results: SPRS did not associate with demographic traits or measures of body composition. SPRS was positively associated with several blood pressure related phenotypes with the strongest association between SPRS and systolic blood pressure response to acute exercise (β=0.01, p=2.86e-05), which persisted after adjustment for resting systolic blood pressure. Additionally, 19 total phenotypes were nominally associated (p<0.05) with SPRS including multiple apoB related plasma lipid traits ( Table 1 ). Interestingly, SPRS was also inversely associated with hematocrit (β= -0.06, FDR=0.004). Conclusions: Genetic risk of ischemic stroke is associated with hemodynamic traits even in a cohort of relatively young adults free from overt cardiometabolic disease. Additionally, blood pressure response to acute exercise may represent an early manifestation of genetic liability to stroke.

Introduction: VO 2 max is the product of maximal cardiac output (Q) (maximal stroke volume (SVmax) x heart rate) and peak arterio-venous oxygen content difference (a-vO 2 diff). SVmax and peak a-vO 2 diff have important implications across the spectrum of human health and performance. However, the molecular mechanisms underlying SVmax and peak a-vO 2 diff remain largely unknown. We characterize SVmax and peak a-vO 2 diff using large-scale plasma proteomics and genotyping to identify circulating proteins and putative mediators of these traits. Methods: 763 physically inactive adults without overt cardiovascular disease underwent cardiopulmonary exercise testing with rigorous measurements of Q. High-throughput plasma proteomic profiling was performed to identify proteins associated with SVmax, a-vO 2 diff, and VO 2 max. cis-Mendelian Randomization (MR) analysis was conducted to assess causal relationships between protein markers and cardiovascular performance traits. Results: Proteomic profiling revealed associations between SVmax and several adiposity-related proteins, as well as secretory muscle factors (cathepsin B, α-actinin 2, and troponin I) (Figure). Proteins associated with peak a-vO2diff reflected oxygen-carrying capacity, muscle biology, and glucose metabolism (erythropoietin, myoglobin, fructose-1,6-bisphosphatase). Significant overlap was observed between proteins associated with SVmax, a-vO 2 diff, and VO 2 max. Cis-MR identified myocilin (MYOC) and caspase recruitment domain family member 9 (CARD9) as potential mediators of cardiovascular performance and VO 2 max. Conclusions: This study provides novel insights into the molecular influences of cardiovascular performance and VO 2 max. The identified protein markers associated with SVmax and a-vO 2 diff, along with the possible mediators MYOC and CARD9, merit further investigation into their role in endurance exercise capacity and cardiometabolic health.

Introduction: Angiopoietin like protein (ANGPTL) complexes 3/8 and 4/8 are established inhibitors of lipoprotein lipase and modifiable by regular exercise. However, the molecular underpinning of these novel biomarkers has not been fully elucidated. The purpose of this study was to examine the associations between plasma metabolites and ANGPTL3/8 and 4/8 before and after exercise training. Methods: Measurements were taken before and after 20 weeks of exercise training in 630 adults (36% Black, 56% women, mean age 35 yrs) of the HERITAGE Family Study. Meso Scale Discovery immunoassays were used to measure ANGPTL3/8 and 4/8 in serum. Plasma metabolites (n=300 named) were measured using LC-MS and the HILIC-pos method. Linear mixed models tested the association between metabolites and each trait at baseline and post-training with full covariate adjustment. Significance was set to FDR<0.05. Results: A total of 111 and 93 metabolites were significantly associated with ANGPTL3/8 at baseline or post-training, respectively ( Figure 1A ). A total of 69 metabolites were associated with ANGPTL3/8 at both time points, including ketone bodies, acylcarnitines, and DMGV, while 24 metabolites were only associated at post-training including citric acid, methionine, and glycine. A total of 89 and 40 metabolites were significantly associated with ANGPTL4/8 at baseline or post-training, respectively ( Figure 1B ). A total of 13 metabolites were associated with ANGPTL4/8 at both time points including glycine and lysophospholipids, whereas 27 metabolites were associated only at post-training including hydroxyproline, N-Lignoceroyl Taurine, and methylguanidine. Conclusions: We identified several plasma metabolites associated with ANGPTL3/8 and 4/8 before and after exercise training. Our findings indicate potential metabolic pathways related to the exercise responsiveness of ANGPTL3/8 and 4/8, including glucose-alanine cycle, urea cycle, and glycine and methionine metabolism.

Background: Plasma proteins can be biomarkers of cardiovascular health status but also physiologic effectors that mediate health benefits. Cardiorespiratory fitness (CRF) is an integrative measure of cardiovascular and metabolic health and independent predictor of future cardiovascular disease (CVD) and mortality risk, however limited knowledge exists regarding its molecular transducers. We sought to expand upon prior proteomic studies of CRF using an antibody-based technology (Olink TM ). Hypothesis: Olink proteomic profiling will identify new markers and potential mediators of CRF. Methods: We measured plasma proteins (N=1,472) using Olink’s platform in a pilot study within the HERITAGE Family Study (N=209 participants, mean age=34 years, 56% female, 36% Black) before and after 20 weeks of endurance exercise training (ET). We performed multivariable linear regression to measure the association between protein levels and CRF measured by CPET (VO 2 max in ml*kg -1 *min -1 ) adjusting for age and sex. Lean body mass was measured by hydrostatic weighing. Protein changes after ET were assessed using paired Student’s t-tests. Results: We identified 70 proteins significantly associated (FDR q <0.05) with VO 2 max. Among these, 14 have not previously been measured in the context of CRF, including perilipin-1, a modulator of lipid homeostasis in adipocytes (beta=-3.8, q=2.4x10 -4 ); latent-transforming growth factor beta-binding protein 3 (beta=-2.6, q=1.3x10 -3 ), a key regulator of TGF-beta activation; and hydroxysteroid 11-beta dehydrogenase 1 (beta=3.2, q=1.2x10 -3 ), which regulates cortisol metabolism. Carbonic anhydrase (CA) XIV (CA14; beta=5.4, q=1.3x10 -8 ), a member of the CA family with extracellular activity, had the strongest positive association with VO 2 max in our primary model as well as after further adjusting VO 2 max for lean body mass. Further, CA14 levels increased significantly after ET (log2 fold change: 0.14, q=0.007). Conclusions: Antibody-based plasma proteomics profiling identified new markers of CRF, including a CA isoform that increases after ET. These findings motivate further study of CA14’s mechanistic role in CA14 and expanded proteomics profiling to investigate molecular transducers of CRF.

Background: Previous studies have derived and validated an HDL apolipoproteomic score (pCAD) that predicts coronary artery disease risk. However, the relationship between pCAD and the cardiometabolic profile in healthy adults, as well as the effects of regular exercise on pCAD are unknown. Methods: A total of 642 inactive but healthy adults (56% female, 36% Black, 35±13 yrs) completed 20 weeks of endurance exercise training. HDL-bound proteins were measured in APOA1-tagged serum using targeted LC-MS/MS (Quest Diagnostics). The pCAD index was calculated as a weighted sum of five HDL-associated apolipoproteins (APO A1, C1, C2, C3, C4) with scores from 0 to 100 (higher score=higher risk). Age-, sex-, and race-specific baseline pCAD quartiles were created. Results: The mean (SD) pCAD score at baseline was 29 (32). At baseline, increasing pCAD quartile was associated with a poorer cardiometabolic profile, including higher total and visceral fat, blood pressure, inflammation, and fasting glucose and lower insulin sensitivity and lipoprotein lipase activity ( Table ). In the total sample, pCAD significantly decreased with exercise training by 4.7 (25) pts or ~16%. Across pCAD quartiles, mean decreases in pCAD were 4.6 in quartile 3 and 21.3 in quartile 4, compared to mean increases of 3.6 in quartile 1 and 3.5 in quartile 2 (p<0.05 for all within group changes). The exercise-induced decrease in pCAD was largely driven by increases in HDL-associated APOA1 and APOC1 levels and inversely correlated with changes in HDL-C (r= -0.17, p<0.0001) and HDL size (r= -0.14, p=0.0006). Otherwise, changes in pCAD were not correlated with changes in other cardiometabolic traits. Conclusions: A higher pCAD index was associated with a poor cardiometabolic profile at baseline and improved with regular exercise in healthy adults with the worst profiles. These findings highlight the potential role of HDL apolipoproteins as a theranostic target for lifestyle interventions in these individuals.

Introduction. C-reactive protein (CRP) and GlycA are established biomarkers of inflammation. Regular exercise tends to decrease CRP and GlycA levels. However, the spectrum of molecules associated with the anti-inflammatory effects of regular exercise are less well understood. Hypothesis. We hypothesized that distinct metabolite signatures exist for both baseline levels and exercise responsiveness of CRP and GlycA. Methods. Measures were performed before and after 20 weeks of endurance exercise training in 652 Black and White adults from the HERITAGE Family Study. A total of 300 targeted plasma metabolites were measured using LC-MS. High-sensitivity CRP and GlycA were measured using automated assays and NMR spectroscopy (LabCorp), respectively. Linear mixed models were used to test: 1) Association of baseline metabolites with baseline hsCRP and GlycA and 2) Association of changes in metabolite with changes in hsCRP and GlycA. Models were adjusted for age, sex, race, BMI, with family membership as a random variable, with change models also adjusting for baseline trait value. Significance was determined as FDR<0.05. Results. Baseline levels and changes of hsCRP and GlycA were moderately correlated (r=0.51 and 0.31, p<0.0001 for both, respectively). At baseline, 40 and 94 metabolites were associated with hsCRP and GlycA, respectively, with 30 metabolites associated with both phenotypes. The top baseline associations for both traits included multiple species of lysophosphatidylcholine (LPC) and phosphatidylethanolamine (PE), while cortisol, biliverdin, and bilirubin were among the metabolites associated with GlycA only. The changes in only one metabolite were associated with concomitant changes in CRP, while no associations were found for change in GlycA. Conclusions. Plasma metabolite associations with baseline hsCRP overlapped with those associated with baseline GlycA levels. Several unique metabolite associations with baseline GlycA were identified, including molecules in established inflammatory pathways. Metabolite changes with exercise were not associated with changes in either measure. These findings have implications for the use of metabolites as signatures of systemic inflammation vs as targets of lifestyle interventions.

PURPOSE: Identification of a robust molecular signature is a major goal in aging research. Recent studies have identified proteins that resemble a proteomic clock and can predict accelerated biological aging. Exercise is well known to mitigate physiological and molecular changes. However, it is unknown whether regular exercise affects the predicted protein age. Our goal was to understand the effects of exercise training on the proteomic aging clock. METHODS: We measured over 5,000 circulating proteins using an aptamer-affinity based platform (SomaScan) before and after 20 weeks of endurance exercise training in 647 Black (n = 230) and White (n = 417) adults from the HERITAGE Family Study. Proteomic age score was calculated by summing the weighted expression values across 360 proteins validated in previous proteomic age score studies. Delta age (or proteomic age acceleration) was quantified as the difference between predicted and chronological age. Change in delta age was calculated by subtracting baseline delta age from post-training delta age. RESULTS: The proteomic age score was very strongly correlated with chronological age (r = 0.94, p < 0.0001). Proteomic age acceleration was associated with ethnicity, generation (parent vs offspring), and their interaction, but not sex. Specifically, baseline delta age (mean (SD)) was significantly lower in parents (5.2 (4.1) yrs) compared to offspring (10.6 (4.2) yrs) and in Blacks (8.0 (4.7) yrs) compared to Whites (9.2 (5.0) yrs). Exercise training resulted in a decrease in delta age in parents only (i.e., training attenuated proteomic age acceleration), with the decrease larger in White (-13.7 (8.0) yrs) compared to Black (-7.4 (9.0) yrs) parents. Conversely, offspring of both ethnic groups showed mean increases (+6.1 (4.1) yrs) in delta age with training (i.e., proteomic age acceleration increased). CONCLUSIONS: These results indicate that an established proteomic signature of age is sensitive to exercise training, but the magnitude of response differs by subgroups of age and ethnicity thereby limiting its potential clinical utility. Further studies are needed to examine whether reduced proteomic age acceleration with exercise training is associated with concomitant improvements in cardiometabolic traits related to healthy aging.

Background: Mass spectrometry (MS) profiling has identified over 250 proteins associated with HDL that are thought to underlie the diverse atheroprotective properties of HDL particles and thus may be important biomarkers of cardiovascular disease (CVD) risk. Likewise, recent studies have identified circulating plasma proteins as biomarkers of CVD risk factors and health outcomes. However, few studies have compared the HDL proteome with the circulating plasma proteome. Purpose: The purpose of this analysis was to examine the relationship between the abundance of individual proteins measured in whole plasma and the HDL-sized plasma fraction. Methods: We examined the HDL-sized and circulating plasma proteomes in 156 Black (30%) and White men and women (61%) from the HERITAGE Family Study. HDL was isolated from plasma via gel filtration chromatography and untargeted MS analysis was performed via nano-HPLC-MS/MS. The whole plasma proteome was measured using a modified aptamer (SOMAscan) assay. The correlations between protein abundances in HDL and whole plasma were examined for 101 HDL-associated proteins present in at least 40% of the sample. Results: The abundance of 56 proteins in HDL-sized and whole plasma were significantly (5% FDR) correlated with the strongest correlation for Haptoglobin levels (r=0.83, p=1.1x10 -40 ) ( Table 1 ). The remaining significant correlations ranged from weak to moderate (r= 0.18-0.64) and were found among a mix of frequently and occasionally observed HDL proteins. Discussion: We found that protein abundance measured in HDL-sized and whole plasma were moderately to strongly correlated for several proteins, whereas 45% of protein levels showed no association between HDL-sized and whole plasma fractions. Given the inherit differences in measurement techniques and sources of proteins, it appears that the plasma HDL-sized proteome is mostly distinct from the circulating whole plasma proteome as measured by the SOMAscan assay.

Although exercise training is known to improve body composition, the molecular biomarkers and mechanisms related to these changes have not been fully elucidated. PURPOSE: The purpose of this study was to examine the associations between change in plasma proteins and change in body composition traits in response to endurance training. METHODS: Measurements were taken before and after 20 weeks of standardized, endurance training in Black and White adults of the HERITAGE Family Study (n = 652). Over 5,000 plasma proteins were measured using an aptamer-affinity based platform (SomaScan). Underwater weighing, CT scans, and anthropometric measurements were used to derive the 11 body composition traits included in this study: BMI, body surface area, fat mass, fat free mass , %fat, waist circumference, waist-to-hip ratio (WHR), body weight, and abdominal visceral, subcutaneous, and total fat. Linear mixed models were used to test the association between change in plasma proteins and change in each body composition trait adjusted for age, sex, race, baseline BMI, and baseline trait value with family membership as a random variable. Significance was set to FDR < 0.05. RESULTS: On average, subjects were 35% Black, 56% female, 35 years old, and overweight at baseline (mean BMI 26.4 (SD 5.3) kg/m2), with %fat of 27.5 (10.4). All 11 traits significantly improved in response to training. Significant associations between changes in proteins and body composition were found for all traits except WHR, with 57 unique proteins identified. Leptin was the top association (range: 0.023 < FDR p-value<4.2x10-12) for all 9 body composition traits it associated with (Table 1). CONCLUSIONS: Although dozens of proteins were associated with changes in body composition traits, 6 proteins were associated with ≥8 traits. Globally, these proteins are involved in pathways such as adipogenesis, energy balance, and cell growth, which may potentially influence body composition and fat distribution traits.