Microbial Translocation and Gut Damage Are Associated With an Elevated Fast Score in Women Living With and Without HIV

Abstract Background Steatohepatitis is common in persons living with HIV and may be associated with gut microbial translocation (MT). However, few studies have evaluated the gut-liver axis in persons living with HIV. In the Women's Interagency HIV Study, we examined the associations of HIV and circulating biomarkers linked to MT and gut damage using the FibroScan–aspartate aminotransferase (FAST) score, a noninvasive surrogate for steatohepatitis with advanced fibrosis. Methods Among 883 women with HIV and 354 without HIV, we used multivariable regression to examine the associations of HIV and serum biomarkers linked to MT and gut damage (kynurenine and tryptophan ratio, intestinal fatty acid–binding protein, soluble CD14, and soluble CD163) with a log-transformed FAST score after adjusting for key covariates. We used a path analysis and mediation models to determine the mediating effect of each biomarker on the association of HIV with FAST. Results HIV infection was associated with a 49% higher FAST score. MT biomarker levels were higher in women with HIV than women without HIV (P < .001 for each). MT biomarkers mediated 13% to 32% of the association of HIV and FAST score. Conclusions Biomarkers linked to MT and gut damage are associated with a higher FAST score and mediate the association of HIV with a higher FAST score. Our findings suggest that MT may be an important mechanism by which HIV increases the risk of steatohepatitis with advanced fibrosis.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a major cause of liver-related morbidity and mortality, and its prevalence is enriched in people living with HIV (PLWH) [1][2][3].While MASLD encompasses a range of histologic severity, metabolic dysfunction-associated steatohepatitis (MASH) has been associated with advanced fibrosis, cirrhosis, and hepatocellular carcinoma [4,5].Prior work evaluating liver biopsies in PLWH who did not have viral hepatitis coinfection reported a relatively high prevalence of nonalcoholic steatohepatitis (NASH) and significant liver fibrosis (≥F2), with pooled estimated prevalences of 42% and 22%, respectively, but lacked a comparison group of persons without HIV [6].
We previously found that among 1309 women without viral hepatitis (928 living with HIV and 381 without HIV), HIV infection was independently associated with a 3.7-fold higher odds of having an elevated FibroScan-aspartate aminotransferase (FAST) score, a noninvasive measurement of histologic MASH with an elevated nonalcoholic fatty liver disease activity score (≥4) and significant liver fibrosis (≥F2) [7].Moreover, HIV viral suppression was associated with lower odds of an elevated FAST score [8].These findings suggest that HIV increases the risk of MASH with advanced liver disease.Yet, the underlying mechanisms by which HIV influences the pathogenesis of MASLD and MASH remain unclear.
Recent studies have elucidated an important link between the gut microbiome and chronic liver disease, often termed the gut-liver axis.Translocated microbial by-products can activate hepatic Kupffer and stellate cells, leading to macrophage Microbe Translocation, FAST Score, and HIV • OFID • 1 Open Forum Infectious Diseases M A J O R A R T I C L E activation, inflammation, and fibrosis, with certain microbial signatures observed to be associated with MASLD in patients without HIV and in murine models [9][10][11][12].HIV infection has been associated with profound dysbiosis and changes in gut permeability that persist despite antiretroviral therapy [4,[13][14][15], although gut microbial composition may be slightly modified by an antiretroviral therapy regimen [15].PLWH have gut microbiomes with more pathogenic bacteria and less commensal bacteria when compared with HIV-seronegative controls [16].HIV-associated dysbiosis has been linked to increased tryptophan catabolism, leading to an increase in catabolism by-products in the gut, most notably kynurenine.Kynurenine binds to T cells and inhibits differentiation of TH17 cells, thus leading to a reduction of interleukin 17 and 22 production.This in turn increases the permeability of the gut-mucosal barrier, allowing for potential translocation of microbes and their by-products [16].
Prior work has found an association between HIV and circulating biomarkers of microbial translocation, as well as links between microbial translocation and liver disease in people without HIV [17][18][19][20].These studies support the hypothesis that microbial translocation and the resulting inflammatory response contribute to liver disease in PLWH.However, no studies to date have directly investigated how the gut-liver axis may be associated with the occurrence of MASH in PLWH.

Study Population
Between 2013 and 2018, 1576 women living with and without HIV were enrolled in the Liver Disease and Reproductive Aging cohort, an ancillary study of the Women's Interagency HIV Study (WIHS; now part of the Multicenter AIDS Cohort Study [MACS]/WIHS Combined Cohort Study) [21].Women without HIV (WWOH) who were demographically similar to women with HIV (WWH) and had risk factors for HIV within the past 5 years (ie, history of sexually transmitted infection, sex without a condom with ≥3 men, sex with a partner who was HIV seropositive, injection drug use) were recruited at the same time as women living with HIV.A unique strength of the WIHS cohort is its enrollment of demographically similar women who were HIV seronegative.Age, race, socioeconomic distribution, and behavioral habits at enrollment were similar to WWH in the cohort, who in turn were representative of US WWH [22].
The details of this cohort's and parent study's recruitment and inclusion/exclusion criteria have been described [21].Briefly, all Liver Disease and Reproductive Aging participants underwent vibration-controlled transient elastography (VCTE; Echosens) and blood draw within 6 months of each other.VCTE-measured liver stiffness (kilopascals), controlled attenuation parameter (decibels per meter), and aspartate aminotransferase (international units per liter) were used to calculate the FAST score [7].Women with positive hepatitis C virus antibody (anti-HCV) were excluded from the analysis, as were women with hepatitis B virus surface antigen.Each of the 13 participating institutions' institutional review boards approved study protocols and consent forms, and all study participants gave written informed consent.

Biomarker Testing
Plasma samples from specimens obtained within 6 months of VCTE measurement and stored at −70 °C were tested for the following biomarkers linked to microbial translocation and gut damage: kynurenine and tryptophan levels (to calculate a KT ratio, a marker of dysbiosis and microbial translocation), intestinal fatty acid-binding protein (I-FABP; a marker of gut epithelial integrity loss), and the immune activation markers soluble CD14 (sCD14; also a receptor for the bacterial byproduct lipopolysaccharide) and soluble CD163 (sCD163; a protein produced by proteolytic cleavage of CD163 expressed on the surfaces of monocyte/macrophage lineage cells).

Statistical Analysis
We performed chi-square and Student t tests to evaluate differences in demographics, substance use behavior, metabolic factors, and biomarkers of microbial translocation between WWH and WWOH.
FAST scores were log transformed to approximately conform to normality due to a right-skewed distribution, and we used multivariable linear regression to evaluate the association between HIV infection and log-transformed FAST score.We adjusted the model for covariates that were significantly different between WWH and WWOH in bivariate analysis, including age and menopausal status, as well as potential confounders that could influence the FAST score, including alcohol use, current tobacco use, race/ethnicity, BMI, and HOMA-IR.Regression coefficients were exponentiated to provide the percentage difference in FAST score between WWH and WWOH.After obtaining this baseline model, we added each microbial translocation biomarker to the model to determine its association with the FAST score.To compare the effect sizes across biomarkers linked to microbial translocation and gut damage with different units of measurement, we standardized units by log transforming each biomarker level and then dividing the log-transformed value by the log IQR.
Finally, we performed separate path and mediation analyses per biomarker to examine the percentage of the direct effect of the relationship between HIV and FAST score and the percentage of the mediating effect by each biomarker.We controlled for potential confounders in the mediation models: alcohol use, tobacco use, BMI, race/ethnicity, and menopause state.In addition, since the presence of HIV has been associated with insulin resistance in prior studies [26,27], we added HOMA-IR to the mediation models to determine the percentage mediating effect of insulin resistance on the relationship between HIV and FAST score.The 95% CIs for mediating effects were obtained with a bias-corrected bootstrapping method.To account for missing data, we employed the full information maximum likelihood approach [28].We used SAS version 9.4 and STATA version 17.0 to perform the analysis.

Demographics
After excluding women who did not have other known liver disease and were anti-HCV positive (n = 205), 1237 women (883 WWH and 354 WWOH) were included in a crosssectional analysis.Overall, the median age was 49 years; most women (74.5%) were non-Hispanic Black; and over half were obese with a median BMI of 30.5 kg/m 2 (Table 1).As compared with WWOH, WWH were more likely to have a greater median HOMA-IR and less likely to report being postmenopausal.In addition, WWH were more likely to report abstinence from alcohol than WWOH (52% vs 44%) and less likely to report heavy drinking (5.7% vs 12.7%, P < .001,across alcohol categories), smoking (37% vs 44%, P = .02),or a history of injection drug use (2.7% vs 5.7%, P = .01).Among WWH, the median CD4 count was 651 cells/mm 3 , and 91% reported undergoing antiretroviral therapy.

Microbial Translocation Mediates the Association of HIV on FAST Score
Since we found that increased microbial translocation biomarker levels were associated with HIV and higher FAST score and that there was attenuation of the HIV-FAST association with the addition of biomarkers to the models, we performed a path analysis to estimate the effect attributable to microbial translocation and insulin resistance in the relationship between HIV and FAST score.Bootstrap bias-corrected confidence intervals for each mediation effect revealed that the addition of 3 of the 4 biomarkers (KT ratio, sCD14, and sCD163) as mediators yielded a statistically significant attenuation in the association of HIV with FAST score.After adjusting for age, BMI, race, HOMA-IR, alcohol use, current tobacco smoking, and menopause state, the percentage of the HIV association with FAST score attributable to microbial translocation ranged from 13% (I-FABP) to 32% (sCD163; Figure 2).The total HIV effect on FAST score attributable to HOMA-IR ranged from 1.3% to 3.6% in each of the 4 path analysis models and was not statistically significant for any of the models.

DISCUSSION
In this study of 1237 women with and without HIV, HIV infection was associated with higher circulating biomarkers linked to microbial translocation and gut damage.HIV infection was also independently associated with higher FAST score after adjustment for demographic, behavioral, and metabolic factors.A notable finding is that markers linked to microbial translocation and gut damage were strongly associated with higher FAST scores and mediated the association of HIV with higher FAST.
Our finding that HIV was associated with higher FAST scores is consistent with prior work from our group, which found significantly higher odds of elevated FAST score in WWH as compared with WWOH [8].Our current study expands on the prior study by evaluating microbial translocation and gut damage as potential contributors to the higher prevalence of NASH in WWH.A noteworthy finding was the strong associations of all 4 biomarkers with higher FAST scores.This finding is consistent with mounting evidence that microbial translocation and MASLD are closely linked.In murine models, fecal microbiota transplant from mice with metabolic syndrome can induce metabolic syndrome and fatty liver disease in previously healthy mice [29], and human studies have shown that insulin function (which is strongly associated with MASLD) improved in participants who were obese after receiving a fecal microbiota transplant from donors who were lean [30].
Our study is consistent with established literature demonstrating a strong association between HIV infection and microbial translocation, as all biomarkers linked to microbial   FAST score was log transformed and then exponentiated so that differences between regression coefficients on the log(e) scale can be expressed as proportions.The HIV effect (reference group: seronegative) on FAST score is reported via the baseline model (fully adjusted) and then with the addition of each biomarker to the model.
a Fully adjusted model includes body mass index, race, insulin resistance, alcohol use, tobacco, and menopause state.
In addition, it is consistent with work demonstrating an association between liver fibrosis and microbial translocation in PLWH [17,18].In our cohort, we found an association of microbial translocation biomarkers with higher FAST score regardless of HIV serostatus, although I-FABP did not have a statistically significant association with FAST score in WWOH, possibly due to lower power.Overall, our findings are consistent with our hypothesis that microbial translocation may be a driver of MASLD-related inflammation in WWH.
In addition, since WWH had higher levels of each biomarker, we hypothesized that microbial translocation may mediate the relationship between HIV and higher FAST score (Supplementary Figure 1).Indeed, the association of HIV with FAST score was attenuated after adjusting for each biomarker.Moreover, we observed that microbial translocation mediated up to 30% of the observed association of HIV with FAST.
Our findings were consistent regardless of the microbial translocation biomarker used, which is unsurprising given that each biomarker is representative of the same overall pathophysiology (translocation).It must be noted that several biomarkers linked to translocation and gut damage (KT ratio, sCD14, and sCD163) are also associated with immune activation in HIV.As such, they may be measuring multiple co-occurring pathways.It is therefore important to discuss the notable differences among the biomarkers.The tryptophan catabolism pathway (measured by KT ratio) and sCD163 affect immunoregulation and are associated with decreased T-reg differentiation, which may reflect a deeper level of injury to the liver from microbial translocation [34,35].Indeed, a prior study in WIHS found an association between higher KT ratio and increased liver fibrosis, as measured by FIB-4, among women living with HIV in the presence or absence of hepatitis C virus (HCV) infection [17].In addition, sCD163 has been associated with indirect measures of liver fibrosis in WWH and HCV [36], as well as insulin resistance in PLWH [26].In our study, sCD163 was associated with the highest indirect effect on FAST score of all the biomarkers.It may be that despite our adjustments for BMI and HOMA-IR, sCD163 may modulate the higher FAST score via microbial translocation and metabolic alterations, though notably our mediation analysis demonstrated an overall small indirect effect of insulin resistance (ranging from 1% to 3.6%) on the association between HIV and FAST score.This could point to alternate pathways that are mediated by sCD163, such as macrophage abundance, polarization, and activation and that affect MASH.In contrast, sCD14 is more directly tied with microbial translocation alone: it rises with macrophage activation and is the receptor for circulating lipopolysaccharide, a bacterial component [18,19].Elevated sCD14 has been associated with HCV-associated fibrosis in multiple studies in PLWH [18,19,37].Notably, sCD14 is secreted from the liver and may be increased in liver disease independent from microbial translocation.The last The mediation effect of insulin resistance was also calculated by HOMA-IR and ranged from 1.3% to 3.6% (data not shown).Models are adjusted for HIV status, age, body mass index, race, insulin resistance (HOMA-IR), alcohol use, tobacco, and menopause state.The 95% CIs for mediating effects were obtained by a bias-corrected bootstrapping method.FAST, FibroScan-AST; HOMA-IR, homeostatic model assessment for insulin resistance; KT ratio, kynurenine and tryptophan ratio; I-FABP, intestinal fatty acid binding protein; sCD14, soluble CD14; sCD163, soluble CD163.
biomarker that we evaluated, I-FABP, is a marker of gut turnover and can be a more specific marker of translocation as it is released with intestinal epithelial damage.I-FABP has been shown to correlate with progression of fibrosis to cirrhosis in HIV/HCV coinfection [18,31].
There are important limitations to our study.Given the crosssectional design, it is not possible to completely elucidate a causal path by which microbial translocation may mediate the effect of HIV on higher FAST score.Furthermore, although prior literature has established each biomarker as being strongly correlated with microbial translocation, the complex physiology and immunoregulation of each marker make it challenging to state with certainty that they are each representative of only microbial translocation, and there may be unknown or unmeasured confounders.The use of 4 separate proxies for translocation does allow us greater confidence in our conclusions, as they performed similarly despite being evaluated independently.In addition, while the FAST score has been validated in large cohorts excluding PLWH, it has not been validated as a surrogate for histologic MASH in PLWH.However, an elevated FAST score predicts liver-related outcomes among PLWH [38].In addition, persistent inflammation in PLWH may contribute to liver and multiorgan fibrosis and may be driving elevated FAST scores.While microbial translocation may play a role in this persistent inflammation, further studies are necessary to better elucidate this.Furthermore, WWH had lower alcohol consumption than WWOH in our study population.Since alcohol use can increase the FAST score in the absence of MASH, some WWOH may have had falsely elevated FAST scores.However, we controlled for this by adjusting for alcohol consumption in the base model.In addition to this, a limitation of our mediation analysis is that it relies on several key assumptions and control of confounders.While we adjusted mediation models for known confounders, there may be unmeasured confounding that is not being taken into account.Finally, our study may not be generalizable to men living with HIV: there are data demonstrating that sex and menopause state influence the natural history of chronic liver disease, including hepatitis C and MASLD [39,40].
In conclusion, our findings suggest that microbial translocation may be an important mechanism by which HIV increases the risk of MASH with advanced fibrosis.Prospective studies and further microbiome analyses, including stool genomic sequencing, will be critical to explore this potential pathway.Ultimately, the clinical implications of these findings and future studies could include manipulation of the microbiome to help improve MASLD outcomes.

Figure 2 .
Figure 2. Path analysis of HIV, microbial translocation and gut damage, and FAST score.Path analysis shows multivariable adjusted effects of HIV infection status on steatohepatitis, as determined by FAST score.Standardized β coefficients for each mediator are shown, as well as percentages of the total effect attributable to microbial translocation and gut damage (microbial translocation as demonstrated by each biomarker) and non-microbial translocation pathways (other mechanisms of infection status, not including insulin resistance).The mediation effect of insulin resistance was also calculated by HOMA-IR and ranged from 1.3% to 3.6% (data not shown).Models are adjusted for HIV status, age, body mass index, race, insulin resistance (HOMA-IR), alcohol use, tobacco, and menopause state.The 95% CIs for mediating effects were obtained by a bias-corrected bootstrapping method.FAST, FibroScan-AST; HOMA-IR, homeostatic model assessment for insulin resistance; KT ratio, kynurenine and tryptophan ratio; I-FABP, intestinal fatty acid binding protein; sCD14, soluble CD14; sCD163, soluble CD163.

Table 1 . Baseline Cohort Data by HIV Serostatus
Number of missing values: HOMA-IR, n = 203; waist circumference, n = 62; age, n = 2; body mass index, n = 2; CD4 nadir, n = 1.Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAP, controlled attenuation parameter; HOMA-IR, homeostatic model assessment for insulin resistance; INSTI, integrase strand transfer inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor.a P values express the difference between women living with HIV and without HIV.P values of continuous variables were calculated by unpaired t test.P values of categorical variables were calculated by a χ 2 test.b Insulin resistance is defined as HOMA-IR >2.0.