4.3.0. The graphical abstract and the analysis overviews were created using BioRender. Role of funders The 2000HIV study is funded by ViiV Healthcare. ViiV Healthcare does not play a role in the study design, data collection, data analyses, interpretation, or writing of this manuscript. Results Characteristics of PLHIV CAP and/or LSM results were available of 1036 PLHIV; 1022 (98.6%) also had results from plasma proteomics. Of these, 814 PLHIV and 959 PLHIV were included in the steatosis and fibrosis analyses, respectively. Of 814 PLHIV, 286 PLHIV (35.1%) had simple steatosis (≥S1 and F0–F1) whereas 528 PLHIV (64.7%) did not have liver steatosis nor fibrosis (S0 and F0–F1). Of 959 PLHIV, 873 (91.0%) did not have liver fibrosis (F0–F1) whereas 86 PLHIV (9.0%) had liver fibrosis (≥F2), regardless of steatosis degree. The baseline characteristics are shown separately in [98]Table 1 for 2000HIV study participants included in the steatosis and fibrosis analyses. [99]Table S1 depicts the baseline characteristics by discovery and validation cohort. Table 1. Baseline table showing characteristics of 2000HIV participants included in the steatosis (left columns) and fibrosis (right columns) analyses. Variable Steatosis __________________________________________________________________ Fibrosis __________________________________________________________________ Overall, N = 814 S0, N = 528 S1 or higher, N = 286 p-value Overall, N = 959 F0–F1, N = 873 F2 or higher, N = 86 p-value Female sex 106 (13%) 75 (14%) 31 (11%) 0.173 128 (13%) 120 (14%) 8 (9.3%) 0.248 Age 52 (43, 59) 50 (40, 57) 55 (49, 61) <0.0001 52 (43, 59) 52 (43, 59) 54 (41, 62) 0.462 Ethnicity 0.107 0.451 Asian 33 (4.1%) 24 (4.6%) 9 (3.1%) 38 (4.0%) 36 (4.1%) 2 (2.3%) Black 83 (10%) 60 (11%) 23 (8.0%) 99 (10%) 94 (11%) 5 (5.8%) Hispanic 21 (2.6%) 13 (2.5%) 8 (2.8%) 23 (2.4%) 22 (2.5%) 1 (1.2%) Mixed 40 (4.9%) 32 (6.1%) 8 (2.8%) 50 (5.2%) 43 (4.9%) 7 (8.1%) Native American 3 (0.4%) 2 (0.4%) 1 (0.3%) 3 (0.3%) 3 (0.3%) 0 (0%) White 632 (78%) 395 (75%) 237 (83%) 744 (78%) 673 (77%) 71 (83%) BMI (kg/m2) 24.9 (22.6, 27.6) 23.9 (21.7, 26.1) 26.8 (24.7, 30.0) <0.0001 25.1 (22.7, 27.8) 24.9 (22.5, 27.5) 27.6 (24.3, 30.8) <0.0001 BMI classification <0.0001 0.000 Lean 407 (50%) 328 (62%) 79 (28%) 465 (49%) 439 (50%) 26 (30%) Overweight or obese 405 (50%) 198 (38%) 207 (72%) 492 (51%) 432 (50%) 60 (70%) Physical activity in general <0.0001 0.072 Low 132 (19%) 63 (14%) 69 (28%) 161 (20%) 140 (19%) 21 (30%) Average 426 (61%) 279 (62%) 147 (60%) 496 (61%) 460 (61%) 36 (51%) Above average 136 (20%) 108 (24%) 28 (11%) 161 (20%) 148 (20%) 13 (19%) Physical activity during work 0.081 0.269 Sedentary occupation 166 (24%) 107 (24%) 59 (24%) 188 (23%) 172 (23%) 16 (23%) Sedentary occupation, walk/cycle to work 73 (11%) 42 (9.4%) 31 (13%) 96 (12%) 83 (11%) 13 (19%) Often physically active during work 206 (30%) 147 (33%) 59 (24%) 245 (30%) 222 (30%) 23 (33%) Physically straining work (lifting) 66 (9.5%) 45 (10%) 21 (8.6%) 75 (9.2%) 70 (9.4%) 5 (7.1%) Unemployed or retired 182 (26%) 108 (24%) 74 (30%) 212 (26%) 199 (27%) 13 (19%) Smoker (Current) 248 (33%) 174 (36%) 74 (28%) 0.033 289 (33%) 264 (33%) 25 (32%) 0.919 Pack years 5 (0, 23) 5 (0, 22) 5 (0, 23) 0.768 5 (0, 23) 5 (0, 23) 5 (0, 19) 0.832 CMV serology (Pos. IgG) 753 (93%) 487 (93%) 266 (93%) 0.693 893 (94%) 811 (93%) 82 (95%) 0.468 CAP (dB/m) 243 (212, 281) 221 (198, 242) 295 (278, 323) <0.0001 245 (213, 284) 243 (212, 281) 286 (227, 319) <0.0001 LSM (kPa) 4.30 (3.60, 5.18) 4.10 (3.50, 4.80) 4.70 (4.00, 5.50) <0.0001 4.40 (3.70, 5.50) 4.30 (3.60, 5.20) 7.90 (7.30, 9.05) <0.0001 Hypertension 199 (24%) 104 (20%) 95 (33%) <0.0001 236 (25%) 213 (24%) 23 (27%) 0.630 T2DM 38 (4.7%) 16 (3.0%) 22 (7.7%) 0.003 51 (5.3%) 38 (4.4%) 13 (15%) 0.000 Lipid lowering therapy 159 (20%) 76 (14%) 83 (29%) <0.0001 199 (21%) 173 (20%) 26 (30%) 0.023 Metabolic syndrome 207 (25%) 82 (16%) 125 (44%) <0.0001 255 (27%) 219 (25%) 36 (42%) 0.001 Prior infection with HAV 80 (9.8%) 48 (9.1%) 32 (11%) 0.337 93 (9.7%) 86 (9.9%) 7 (8.1%) 0.609 Prior infection with HBV 246 (31%) 166 (32%) 80 (28%) 0.239 290 (31%) 267 (31%) 23 (28%) 0.499 Prior infection with HCV 62 (7.6%) 42 (8.0%) 20 (7.0%) 0.622 75 (7.8%) 69 (7.9%) 6 (7.0%) 0.760 HIV duration (Years) 12 (6, 18) 11 (6, 17) 12 (7, 19) 0.097 11 (6, 17) 12 (6, 18) 10 (5, 15) 0.064 Transmission route 0.081 0.504 Blood products 2 (0.3%) 2 (0.4%) 0 (0%) 3 (0.3%) 2 (0.2%) 1 (1.3%) Congenital 7 (0.9%) 7 (1.4%) 0 (0%) 8 (0.9%) 8 (1.0%) 0 (0%) Heterosexual 178 (23%) 107 (21%) 71 (26%) 212 (23%) 195 (23%) 17 (22%) IV drug use 7 (0.9%) 6 (1.2%) 1 (0.4%) 7 (0.8%) 7 (0.8%) 0 (0%) MSM 586 (75%) 386 (76%) 200 (74%) 684 (75%) 624 (75%) 60 (77%) Nadir CD4 count (10ˆ6 cells/L) 0.27 (0.16, 0.41) 0.28 (0.17, 0.41) 0.25 (0.12, 0.41) 0.065 0.27 (0.16, 0.41) 0.27 (0.16, 0.41) 0.26 (0.16, 0.40) 0.887 CD4:CD8 ratio pre-ART 0.28 (0.17, 0.48) 0.30 (0.18, 0.50) 0.24 (0.13, 0.40) 0.001 0.28 (0.17, 0.47) 0.28 (0.17, 0.48) 0.29 (0.19, 0.37) 0.833 HIV-1 RNA zenith (copies/ml) 100,000 (39,600, 274,068) 100,000 (36,800, 225,000) 100,000 (47,625, 345,282) 0.091 100,000 (40,000, 261,000) 100,000 (40,000, 274,062) 115,000 (52,064, 222,847) 0.442 CD4 at enrollment (10ˆ6 cells/L) 0.68 (0.51, 0.89) 0.67 (0.50, 0.85) 0.71 (0.55, 0.92) 0.135 0.69 (0.52, 0.89) 0.69 (0.51, 0.90) 0.66 (0.53, 0.83) 0.600 CD8 at enrollment (10ˆ6 cells/L) 0.81 (0.60, 1.11) 0.80 (0.59, 1.08) 0.85 (0.62, 1.18) 0.142 0.81 (0.61, 1.12) 0.81 (0.60, 1.12) 0.90 (0.65, 1.15) 0.189 CD4:CD8 ratio at enrollment 0.87 (0.59, 1.17) 0.89 (0.64, 1.16) 0.84 (0.55, 1.16) 0.205 0.87 (0.59, 1.17) 0.88 (0.59, 1.17) 0.82 (0.59, 1.15) 0.426 ART duration (Years) 9 (6, 15) 9 (5, 14) 10 (6, 16) 0.018 9 (5, 15) 9 (6, 15) 8 (5, 12) 0.146 No ART at enrollment 13 (1.6%) 8 (1.5%) 5 (1.7%) 0.777 14 (1.5%) 13 (1.5%) 1 (1.2%) 1.000 Dual therapy at enrollment 86 (11%) 49 (9.3%) 37 (13%) 0.096 101 (11%) 94 (11%) 7 (8.2%) 0.460 NRTI in current regimen 775 (95%) 507 (96%) 268 (94%) 0.140 916 (96%) 833 (95%) 83 (97%) 1.000 NtRT in current regimen 512 (63%) 335 (63%) 177 (62%) 0.660 606 (63%) 549 (63%) 57 (66%) 0.534 NNRTI in current regimen 295 (36%) 194 (37%) 101 (35%) 0.686 343 (36%) 317 (36%) 26 (30%) 0.262 PI in current regimen 72 (8.8%) 41 (7.8%) 31 (11%) 0.140 84 (8.8%) 79 (9.0%) 5 (5.8%) 0.311 INSTI in current regimen 470 (58%) 305 (58%) 165 (58%) 0.984 560 (58%) 502 (58%) 58 (67%) 0.074 PNPLA3 (rs738409) 0.874 0.663 C C 334 (57%) 208 (57%) 126 (58%) 403 (59%) 363 (59%) 40 (63%) G C 216 (37%) 134 (37%) 82 (37%) 244 (36%) 224 (36%) 20 (31%) G G 33 (5.7%) 22 (6.0%) 11 (5.0%) 37 (5.4%) 33 (5.3%) 4 (6.3%) TM6SF2 (rs58542926) 0.964 0.144 C C 506 (87%) 315 (87%) 191 (87%) 586 (86%) 536 (86%) 50 (78%) T C 72 (12%) 46 (13%) 26 (12%) 93 (14%) 79 (13%) 14 (22%) T T 5 (0.9%) 3 (0.8%) 2 (0.9%) 5 (0.7%) 5 (0.8%) 0 (0%) MBOAT7 (rs641738) 0.882 0.749 C C 175 (30%) 107 (29%) 68 (31%) 210 (31%) 188 (30%) 22 (34%) T C 294 (50%) 184 (51%) 110 (50%) 339 (50%) 310 (50%) 29 (45%) T T 114 (20%) 73 (20%) 41 (19%) 135 (20%) 122 (20%) 13 (20%) GCKR (rs780094) 0.629 0.763 C C 203 (35%) 126 (35%) 77 (35%) 242 (35%) 218 (35%) 24 (38%) T C 304 (52%) 194 (53%) 110 (50%) 349 (51%) 319 (51%) 30 (47%) T T 76 (13%) 44 (12%) 32 (15%) 93 (14%) 83 (13%) 10 (16%) MTARC1 (rs2642438) 0.710 0.100 A A 53 (9.1%) 31 (8.5%) 22 (10%) 62 (9.1%) 56 (9.0%) 6 (9.4%) A G 226 (39%) 145 (40%) 81 (37%) 271 (40%) 238 (38%) 33 (52%) G G 304 (52%) 188 (52%) 116 (53%) 351 (51%) 326 (53%) 25 (39%) [100]Open in a new tab The baseline characteristics are compared between groups using Chi-squared tests, Kruskal–Wallis rank sum tests, and Fisher's exact tests. Significant differences (defined as p-values < 0.05) are shown in bold. Abbreviations: CAP, controlled attenuation parameter; LSM, liver stiffness measurement; T2DM, type 2 diabetes mellitus; LLT, lipid lowering therapy; MetS, metabolic syndrome; HepA, pas infection with hepatitis A; HepB, past infection with hepatitis B; HepC, past infection with hepatitis C; ART, antiretroviral treatment; NRTI, nucleoside reverse transcriptase inhibitor; NtRTI, nucleotide reverse transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; INSTI, integrase strand-transfer inhibitor. BMI, diabetes mellitus type 2, lipid lowering therapy, and metabolic syndrome were associated with both liver steatosis and fibrosis. Additionally, age, physical activity, current smoking, hypertension, CD4: CD8 ratio pre-ART and the duration treatment with ART were associated with steatosis ([101]Table 1). Differentially expressed proteins in PLHIV with simple steatosis First, we compared protein concentrations between PLHIV with and without steatosis. A total number of 67/2367 (2.8%) ([102]Fig. 1a, [103]Table S2, Graphical Abstract), including 58 up- and 9 downregulated proteins, were DE. The ten proteins with the highest log fold change were oxytocin (OXT), immunoglobulin superfamily member 9A (IGSF9), fibroblast growth factor 21 (FGF21), leptin (LEP), glutathione S-transferase A3 (GSTA3), insulin-like growth factor-binding protein 1 (IGFBP1), glutathione S-transferase A1 (GSTA1), beta-ureidopropionase (UPB1), liver carboxylesterase 1 (CES1), and formimidoyltransferase-cyclodeaminase (FTCD) ([104]Fig. 1b). IGSF9 and growth hormone receptor (GHR) were the most statistically significant proteins. Most top DEP were involved in glucose-, lipid- and/or amino acid metabolism. FGF21, OXT, LEP, and IGFBP1 have diverse functions including regulation of glucose and lipid metabolism. CES1 is an enzyme involved in lipid and lipoprotein metabolism and FTCD is involved in metabolism of the amino acid histidine and binds and promotes binding of vimentin filaments. In addition, GSTA1 and GSTA3 protect against oxidative stress, UPB1 catalyzes a reaction in the pyrimidine degradation pathway, and IGSF9 is involved in cell-adhesion. Fig. 1. [105]Fig. 1 [106]Open in a new tab a and b Differentially expressed proteins in PLHIV with steatosis compared to PLHIV without steatosis. [107]Fig. 1a shows a volcano plot with the results of the DE analysis for the discovery cohort. For readability of the figure, only validated proteins with a logFC < −0.25 or >0.25, or FDR-adjusted p-value <0.00001 in the discovery cohort are annotated. [108]Fig. 1b shows the expression levels (in NPX) of the top 10 proteins by steatosis grade. p-values are derived from comparison by parametric ANOVA tests between each fibrosis grade with correction for multiple testing by FDR. Tissue enrichment analysis indicate that up- and downregulated proteins were mostly enriched in the liver ([109]Fig. S3 and [110]Tables S3–S6). Enriched pathways in PLHIV with steatosis include metabolic pathways such as ‘response to lipoprotein particle’ and ‘regulation of glucose import’, as well as other pathways such as ‘regulation of growth’, ‘regulation of Complement cascade’, ‘cellular response to peptide hormone stimulus’, and ‘drug metabolism’ ([111]Fig. S4 and [112]Table S7). Next, we assessed interactions between our significant proteins ([113]Fig. S5). In a protein–protein interaction network, IGSF9 was found to have a central position, being correlated to FGF21, CES1, MAMDC4, and afamin (AFM) in PLHIV with steatosis, and to AFM and GHR in those without steatosis. Differentially expressed proteins in PLHIV with fibrosis Next, we compared protein levels between PLHIV with and without liver fibrosis. A total number of 17/2367 (0.7%) proteins including 16 up- and 1 downregulated protein were DE in PLHIV with fibrosis ([114]Fig. 2a and b, [115]Table S8, Graphical Abstract). The ten proteins with the highest log fold change were IGSF9, C19orf12, cytokeratin-18 (KRT18), adhesion G-protein coupled receptor G1 (ADGRG1), aminoacylase-1 (ACY1), A disintegrin and metalloproteinase with thrombospondin motifs like 2 (ADAMTSL2), 2-iminobutanoate/2-iminopropanoate deaminase (RIDA), retinal dehydrogenase 1 (ALDH1A1), collagen alpha-1 (V) chain (COL5A1), and integrin beta-like protein (ITGBL1). Most of these top DEP play a role in extracellular matrix organization and cell adhesion: ADAMTSL2, ITGBL1, COL5A1 and possibly C19orf12 are extracellular matrix proteins involved in tissue integrity and organization, KRT18 plays a role in filament organization, FTCD binds and promotes binding of vimentin filaments, and IGSF9 and ADGRG1 are involved in cell-adhesion. In addition, ACY1 and RIDA are involved in amino acid metabolism and ALDH1A1 is involved in lipid peroxidation. ‘Methylated in normal thymocytes’ (MENT) was the only downregulated protein. MENT interacts with DNA methylation through DNA methyl transferase 3B (DNMT3B). The expression levels of each protein by fibrosis grade are shown in [116]Fig. 2b: the upregulated DEP incrementally increased while downregulated DEP MENT incrementally decreased by fibrosis stage. Fig. 2. [117]Fig. 2 [118]Open in a new tab a and b Differentially expressed proteins in PLHIV with fibrosis compared to PLHIV without fibrosis. a) Volcano plot showing the results of the DE analysis in the discovery cohort with validated proteins annotated. All validated proteins (i.e. FDR-adjusted p-value <0.05 in the discovery- and raw p-value <0.05 in the validation cohort) are annotated. b) Expression levels (in NPX) of the top 10 proteins (as defined by log fold change in discovery cohort) by steatosis grade. p-values are derived from comparison by parametric ANOVA tests between each fibrosis grade with correction for multiple testing by FDR. Upregulated DEP were enriched in Leydig cells and fibroblasts, but not in any specific tissue ([119]Fig. S6 and [120]Tables S9 and 10). The number of downregulated DEP (n = 1) in PLHIV with fibrosis was not sufficient for tissue- or cell enrichment analysis. Enriched pathways in PLHIV with fibrosis include metabolic pathways (i.e. ‘Diseases of metabolism’ and ‘amino acid metabolic process’) and ‘regulation of synapse organization’ ([121]Fig. S4). Of note, IGSF9, ADAMTSL2, KRT18, and RIDA were associated with both steatosis and fibrosis. Interestingly, several proteins associated with steatosis and fibrosis were significantly correlated with markers of cholesterol metabolism (VLDL, HDL, triglycerides), inflammation (CD14, CD163, IL6), as well as cardiometabolic diseases, HIV-specific parameters, and ART ([122]Fig. 3a and b), but not with SNPs. For example, ADGRG1 and IGSF9, both upregulated in PLHIV with fibrosis, were positively correlated with T2DM, presence of carotid plaques, prior myocardial infarction, duration of HIV and ART, and prior exposure to D-drugs and INSTI ([123]Fig. 3b). Fig. 3. [124]Fig. 3 [125]Open in a new tab a and b Correlation between the proteins associated with steatosis (a) and fibrosis (b), with SNPs of interest, markers of inflammation, -immune activation, -cholesterol metabolism, HIV-characteristics, ART exposure, and cardiometabolic diseases. The colours show the direction and strength of the Spearman correlations, and asterixis refer to the FDR-adjusted p-value: ∗ = p-value 0.01–0.05; ∗∗ = p-value 0.001–0.01; ∗∗∗ = p-value <0.001. Proteomic signatures of simple steatosis differ between lean and overweight/obese PLHIV To investigate whether lean and overweight and obese PLHIV have the same protein disturbances associated with simple steatosis, we separately compared protein concentrations between unaffected PLHIV and PLHIV with simple steatosis within each BMI group. Additionally, we examined correlations between DEP and HIV specific parameters and ART within each BMI group. [126]Fig. 4 shows the results of the DEA in lean and overweight/obese PLHIV respectively. A total of 8/2367 (0.3%) proteins were DE in lean PLHIV ([127]Fig. 4a, [128]Table S11, Graphical Abstract): IGSF9, GHR, UPB1, RIDA, GSTA1, and FTCD were up- and IGFBP2 and CKB were downregulated. A few more (n = 12) proteins were DE in overweight and obese PLHIV ([129]Fig. 4b, [130]Table S12, Graphical Abstract): IGSF9, GHR, RBP5 (retinol-binding protein 5), proline-rich acidic protein 1 (PRAP1), OXT, NHL repeat-containing protein 3 (NHLRC3), isthmin-1 (ISM1), FGF21, CES1, BPI fold-containing family B member 2 (BPIFB2), AFM, and ADAMTSL2 (all upregulated). Overlap between lean and overweight/obese was limited with only two shared DEP (IGSF9 and GHR). Moreover, log fold changes of the 2367 proteins were weakly correlated (Spearman's rho = 0.26, p-value = < 0.0001) between lean and overweight/obese PLHIV in the discovery cohort. Fig. 4. [131]Fig. 4 [132]Open in a new tab a–c Differentially expressed proteins in lean and overweight/obese PLHIV with steatosis compared to PLHIV without steatosis. a) Volcano plot showing the results of the DE analysis lean PLHIV, with only validated DEP (i.e. FDR-adjusted p-value <0.05 in the discovery- and raw p-value <0.05 in the validation cohort) annotated. b) Volcano plot showing the results of the DE analysis in overweight and obese PLHIV, with only validated DEP (i.e. FDR-adjusted p-value <0.05 in the discovery- and raw p-value <0.05 in the validation cohort) annotated. c and d) Correlation between the DEPs associated with steatosis identified in lean PLHIV (left) and overweight/obese PLHIV (right) and SNPs of interest, markers of inflammation, -immune activation, -cholesterol metabolism, HIV-characteristics, ART exposure, and cardiometabolic diseases. The colours show the direction and strength of the Spearman correlations, and asterixis refer to the significance level: ∗ = p-value 0.01–0.05; ∗∗ = p-value 0.001–0.01; ∗∗∗ = p-value <0.001. Abbreviations: D-drug = dideoxynucleoside analogue drugs (includes D4T, DDI, and DDC), D4T, stavudine; cART, combination antiretroviral treatment; NtRTI, nucleotide reverse transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; VL, HIV-1 RNA viral load; immunological non resp, immunological non-responder; IDV, indinavir; RAL, raltegravir, T2DM, type 2 diabetes mellitus; DDI, didanosine; DTG, dolutegravir; INSTI, integrase strand transfer inhibitor; DDC, zalcitabine; PI, protease inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; HBV, hepatitis B virus; HAV, hepatitis A virus; HCV, hepatitis C virus. Amino acid metabolic pathways (involving RIDA, UPB1, and FTCD) were enriched in lean PLHIV ([133]Table S13), whereas no enriched pathways were identified in overweight/obese PLHIV. In both subgroups, the number of DEP was not sufficient for tissue- or cell enrichment analysis. We next assessed correlations between DEP with markers of inflammation, immune activation, cholesterol metabolism, SNPs of interest, HIV-specific characteristics, and comorbidities separately in lean and overweight/obese PLHIV ([134]Fig. 4c). While, DEP in overweight/obese PLHIV were mostly correlated with cholesterol metabolism and inflammation markers, DEP in lean PLHIV were additionally correlated to HIV-related characteristics, exposure to cART, and cardiometabolic diseases. This also accounted for the shared DEP (IGSF9 and GHR). For example, IGSF9 correlated in lean PLHIV with HIV- and cART duration, CD4 nadir, CD4:CD8 ratio pre-cART, CD8 at enrollment, current treatment with NNRTI, exposure to raltegravir, prior exposure to D-drugs, myocardial infarction, T2DM, and presence of carotid plaque, whereas associations with HIV-characteristics and ART were very limited in overweight/obese PLHIV. Steatosis signatures are largely similar between overweight and obese PLHIV and overweight and obese controls Considering the significant correlations between DEP and HIV-specific factors and ART, especially in lean PLHIV, we aimed to assess whether proteomic signatures of steatosis in PLHIV differ from participants in the overweight/obese control cohort (all with a BMI >26 kg/m^2). Likewise, a subset was created of PLHIV with a BMI >26 kg/m^2. The subset of controls (n = 254) consisted mostly of males (n = 145, 57%), with a median age of 66 (IQR: 62–70), and a median BMI of 30.0 (IQR: 28.3–31.9). The subset of PLHIV consisted of 317 (82%) males, with a median age of 53 (IQR: 45–60), and a median BMI of 28.7 (IQR: 27.2–30.9). We compared the expression of 242 inflammatory and cardiovascular proteins between A) controls with (n = 155) vs. without (n = 99) liver steatosis, and B) PLHIV with (n = 233) vs. without (n = 153) liver steatosis, revealing 32 DEP in both PLHIV and controls ([135]Fig. 5a and b ([136]Tables S14 and 15, Graphical Abstract). Sixteen DEP were shared between PLHIV and controls, including 12 up- (FGF21, LEP, IL1RN, CHI3L1, LDLR, SERPINE1, IL18R1, SELE, HGF, CTSZ, IDUA, and CTSD) and 4 downregulated proteins (IGFBP1, IGFBP2, KITLG, and PON3). Among the 16 proteins exclusively associated with steatosis in PLHIV, namely FABP4, CXCL6, HBEGF, PLAT, CPCP1, RARRES2, PRSS8, VEGFA, CD8A, LGALS9, SORT1, IL10RB, CD274, ADM, CD244, and GH1, a consistent trend was observed in controls ([137]Fig. 5c). Likewise, proteins exclusively linked with steatosis in controls exhibited uniform log fold change directions in PLHIV. Accordingly, log fold changes of all 242 proteins were significantly correlated (Spearman's rho = 0.55, p-value <0.0001) between in PLHIV and controls. Fig. 5. [138]Fig. 5 [139]Open in a new tab a–c DEPs associated with steatosis identified in overweight/obese PLHIV and controls. a) Volcano plot showing the results for overweight/obese controls and b) Volcano plot showing the results for controls. All proteins with an FDR-adjusted p-value <0.05 are annotated. Figure c shows the differences and similarities of the proteomic signature of steatosis between PLHIV and controls. Discussion Metabolic dysfunction-associated steatotic liver disease (MASLD) is highly prevalent in PLHIV, but proteomics studies are lacking. In the present study we comprehensively assessed the plasma proteomic signatures of liver steatosis and fibrosis in virally suppressed PLHIV using ART, compared proteomic signatures of liver steatosis between lean and overweight/obese PLHIV, and between overweight/obese PLHIV and overweight/obese controls. We identified the protein IGSF9 as a key protein associated with steatosis and fibrosis in PLHIV. In addition, we found different proteomic signatures of steatosis in lean compared to overweight/obese PLHIV. Importantly, DEP in lean PLHIV were correlated with comorbidities, HIV-specific factors, and previous and current ART regimen, while only minor correlations were found with clinical factors for DEP in overweight/obese PLHIV. Hence, our findings provide further evidence for different pathogenic mechanisms driving liver steatosis in lean compared to overweight/obese PLHIV, with more involvement of HIV-specific characteristics and ART in lean MASLD. IGSF9 was amongst the top DEP for fibrosis and steatosis across BMI categories. IGSF9 concentrations gradually increased by steatosis grade and correlated with prior myocardial infarction, presence of carotid plaques, T2DM, HIV-specific characteristics and prior and current ART. Moreover, IGSF9 has a central position in the protein–protein network of DEP in PLHIV with steatosis. Nonetheless, IGSF9 has not yet been studied in the context of MASLD. IGSF9 is a member of the immunoglobulin superfamily, a class of proteins involved in cell adhesion, binding, and recognition processes. Consequently, IGSF9 is thought to be involved in cell–cell and cell-extracellular matrix (ECM) adhesion.[140]^26 Previous research on IGSF9 mostly focused on IGSF9 expression in different malignancies. In addition, it has been shown that IGSF9 expression is increased in adipose tissue of HIV-uninfected individuals with type 2 diabetes mellitus,[141]^27 and IGSF9 gene expression was found to be upregulated in PBMCs of HIV/HCV coinfected individuals after interferon-based treatment compared to gene expression levels before treatment initiation.[142]^28 It is tempting to speculate on the mechanism linking IGSF9 to MASLD. We found that IGSF9 is significantly correlated with markers of cholesterol metabolism including VLDL and triglyceride levels, as well as markers of inflammation including IL6, CD14, and CD163. Because of the cross-sectional nature of our study, we cannot draw conclusions on causality but can hypothesise based on existing literature. First, like other adhesion molecules, IGSF9 may facilitate the adhesion of immune cells, promoting inflammation, by which hepatic stellate cells become activated and MASLD progresses.[143]^29 Second, the relationship between IGSF9 and MASLD may rely on the effects of molecular pathways that are modulated by Focal Adhesion Kinase (FAK). FAK, a protein tyrosine kinase, has been shown to play an important role in the progression of liver fibrosis.[144]^30 FAK becomes activated when ECM proteins bind to integrins. The activation of FAK may be further enhanced upon binding of growth factors to growth factor receptors. [145]^31 Notably, our protein–protein network revealed a strong correlation between IGSF9 and growth hormone receptor (GHR). Upon activation, FAK regulates the development of liver fibrosis by several mechanisms, including the activation of hepatic stellate cells (HSC), differentiation of myofibroblasts, cell migration and survival and expression of ECM proteins.[146]^30 On the other hand, FAK signalling may also attenuate liver fibrosis. It has been shown that mice with FAK-deficient hepatocytes exhibited increased fibrogenesis, which was linked to the hedgehog/smoothened pathway amongst other pro-fibrotic pathways.[147]^32 In patients with breast cancer, it was shown that IGSF9 interacts with FAK, which led to inhibition of downstream effects of the FAK/AKT signalling pathway including epithelial mesenchymal transition.[148]^31 Both scenarios suggest that the associations between IGSF9 and cholesterol and triglycerides are indirect and reflect the effects of liver steatosis on lipid metabolism, whereas the link between IGSF9 and inflammation may be more direct and causative of liver steatosis. The link between IGSF9 and liver steatosis should be further investigated. The second main finding of our study concerns the differences between lean and overweight/obese PLHIV. This is particularly interesting since the pathophysiology of lean MASLD is poorly understood but represents a significant burden in PLHIV.[149]^3 First, we found that of eight DEP in lean PLHIV with steatosis, and twelve DEP in overweight/obese PLHIV with steatosis, two DEP were shared. This limited overlap indicates distinct pathophysiology between lean and overweight/obese PLHIV. In lean PLHIV with liver steatosis, DEP are involved in insulin sensitivity (i.e. IGFBP-2), lipid metabolism (i.e GHR and CKB), and inflammation and detoxification processes (i.e. GSTA1 and FTCD). In overweight/obese PLHIV with liver steatosis, DEP are involved in lipid metabolism (i.e. RBP5 and FGF21), inflammation and detoxification processes (i.e. PRAP1, CES1, BPIFB2, and AFM), fibrosis (i.e. ADAMTSL2), and hormonal regulation (i.e. OXT). The unique proteomic signatures of lean and overweight/obese PLHIV with liver steatosis suggest differences in multiple mechanisms that may contribute to liver steatosis. Second, associations of DEP with comorbidities, HIV characteristics, and exposure to ART differed between lean and overweight/obese PLHIV. While DEP in overweight/obese PLHIV with liver steatosis were linked to inflammation and cholesterol metabolism with no apparent association with HIV or ART, DEP in lean PLHIV with liver steatosis were associated not only with inflammation and cholesterol metabolism, but also with comorbidities, HIV-characteristics, and exposure to ART. This accounted as well for the shared DEP. Third, in a comparison with overweight/obese controls recruited from the general Dutch population, proteomic signatures of steatosis were largely similar between overweight/obese PLHIV and controls. Altogether, our observations of similar proteomic signatures of steatosis among overweight individuals with or without HIV, with minor correlations with HIV-related factors and ART, in contrast with a unique proteomic signature in lean PLHIV with steatosis with marked correlations with HIV-related factors (i.e. immunological non-responders, CD4:CD8 ratio pre-ART) and exposure to ART (i.e. dideoxynucleosides and INSTI) suggest that mechanisms related to HIV and ART contribute to liver steatosis in lean PLHIV. We cannot explain all associations using the current literature, but we may speculate about some associations: immunological non-responders exhibit persistent inflammation and immune activation, and (prior) treatment with D-drugs such as stavudine may impair mitochondrial function causing decreased beta-oxidation of fatty acids, as well as insulin resistance.[150]^33^,[151]^34 Inflammation, insulin resistance, and alterations in fatty acid metabolism are well-known contributors to MASLD.[152]^35 This study has a few limitations. First, the cross-sectional nature hinders us to draw conclusions on causality. Second, liver biopsies were not available. Hence, we do not have information on the frequency of steatohepatitis in our population and its proteomic signature. Third, we measured the circulating proteomic signatures and were not able to relate this to protein abundance in the liver itself. Fourth, this study included a low proportion of females. We corrected for the effects of sex to avoid any confounding effect. Fifth, the validation cohort included a higher proportion of participants with liver fibrosis than the discovery cohort. This may have increased statistical power on one hand, but also reduced generalizability on the other. Finally, few limitations specifically concern our comparison with the control population. First, we were only able to analyse a limited number of proteins. Second, HIV testing was not done in the overweight/obese controls. However, they were recruited from the general population where HIV prevalence rates are below 0.2% and subjects had no sign of immune deficiency. Third, steatosis was assessed in controls using a different modality, namely magnetic resonance spectroscopy. Direct comparison of the two cohorts was not a possibility. Despite these limitations of using the 300-OB cohort as a control cohort, there are some benefits as well. The relation between liver steatosis, as diagnosed by validated methods, and levels of 242 proteins measured by the same methods across both cohorts provides an initial indication of overlapping proteomic signatures between overweight or obese PLHIV and controls. The main strength of our study is the large proteomics panel consisting of 2367 proteins that we employed in more than 1000 well-characterised PLHIV with transient elastography measurements. Such a study has not been performed previously and has yielded relevant insights into the pathophysiology of MASLD in lean and overweight or obese PLHIV, and possibly interesting biomarkers. In conclusion, we identified specific proteomic signatures for PLHIV with simple steatosis and for those with fibrosis, including the key protein IGSF9. Proteomic signatures of lean and overweight/obese PLHIV differed with more involvement of HIV-specific factors and ART in lean PLHIV. These insights may be relevant to screening and treatment of MASLD in PLHIV. Contributors LvE, QdM, ET, and LJ contributed to the study design. LvE, EM, AN, AG, MB, and WV performed the investigation. LvE drafted the manuscript. QdM, EM, AN, AG, MB, WV, NV, JDS, JR, NR, JvL, GW, MN, AvdV, ET, and LJ revised and approved the manuscript. LvE, AG, MB, WV, and NV accessed and verified the data. All authors read and approved the final version of the manuscript and had final responsibility for the decision to submit the manuscript for publication. Data sharing statement Data used in this study will be made available with restricted access either via Radboudumc Research or a data repository that is most appropriate for the data when all foreseen manuscripts from the 2000HIV study have been published as it is a collaborative effort in which we re-use the same datasets. Further information and request for data resources should be directed to 2000HIV study principal investigator André J.A.M. van der Ven and project data manager Vasiliki Matzaraki. Declaration of interests The authors have nothing to disclose. Acknowledgements