Fibroblast Growth Factor 21 is Elevated in HIV and Associated with Interleukin-6

Fibroblast growth factor 21 (FGF21) is a circulating peptide hormone primarily synthesized by the liver, and to a lesser extent, by skeletal muscle, adipose tissue, and pancreas.^1,2 Its physiology is complex, given its diverse functions on multiple targets and acts as an autocrine, paracrine, and endocrine factor to produce metabolic changes.² In animals and in vitro, studies demonstrate that FGF21 stimulates expression of glucose transporter protein-1 to promote glucose uptake and metabolism and long-term energy storage in adipocytes, increases fatty acid oxidation in hepatocytes, decreases hepatic de novo lipogenesis, inhibits hepatic glycogen degradation and glucagon secretion, and preserves insulin content in pancreatic islets.^1–8

Collectively, these data have established FGF21’s beneficial role on metabolic profiles and energy expenditure. Paradoxically, however, in humans with obesity, type 2 diabetes, and non-alcoholic fatty liver disease (NAFLD), serum FGF21 concentrations are elevated.^9–16 It is debated whether the increase among people with metabolic disorders represents a compensatory mechanism or a resistant state, as data outside of HIV show evidence for both.^2,17–20

Studies have demonstrated that FGF21 concentrations are also increased in people living with HIV (PLWH),²¹ particularly those on antiretroviral treatment (ART) with lipodystrophy.^22,23 It is well-established that this patient population has high rates of metabolic syndrome, NAFLD, and other cardiometabolic co-morbidities, but relatively little is known about FGF21 in HIV.^24–31 We sought to comprehensively investigate FGF21 concentrations in PLWH, including the relationship to HIV-specific variables, inflammation, lipid metabolism, insulin resistance, and body composition.

Participants ≥18 years old enrolled in other studies at our research center with available stored plasma were included in this analysis and stratified by HIV status: 119 HIV-infected/ART-treated (≥12 months cumulative duration of ART), 31 HIV-infected/ART-naïve (no prior receipt of ART), and 29 HIV-uninfected controls. Fasting plasma concentrations of FGF21 were measured in duplicate by Quantikine ELISA kits (R&D Systems). Assay range was 31.3–2,000 pg/mL. Inter- and intra-assay coefficients of variance were <11% and <4%, respectively. FGF21 values were log-transformed to achieve a normal distribution and improve the accuracy of group comparisons.

Participant characteristics and details of clinical assessments and laboratory assays can be found in Supplemental Tables 1 and 2. There was a trend toward higher unadjusted mean (standard deviation) FGF21 concentrations for ART-treated vs. ART-naïve [FGF21_ln pg/mL: 5.5 (1.0) vs. 5.0 (1.4); P=0.09]; however, after adjusting for age, smoking, and alcohol use, there was no difference [least square means (standard error of the mean) FGF21_ln pg/mL: 5.4 (0.1) vs. 5.2 (0.2); P=0.57]. Unadjusted and adjusted FGF21 concentrations for uninfected controls were 4.8 (1.0) and 4.9 (0.2), respectively. Adjusted FGF21 concentrations were significantly higher in ART-treated PLWH compared to controls (P=0.01), but the difference was not significant between ART-naïve PLWH versus controls (P=0.21). These results suggest that FGF21 concentrations are higher in ART-treated PLWH compared to people without HIV. FGF21 concentrations are also likely increased in ART-naïve, but we could not detect a statistically significant difference, likely due to the small sample size in these two groups.

Results of bivariate/sub-group analyses (Supplemental Table 3) guided the selection of variables for stepwise multivariable regression models. Independent variables associated with higher plasma FGF21 concentrations for PLWH included older age, current smoking, higher interleukin-6 (IL-6), higher triglycerides, and lower low-density lipoprotein (LDL) cholesterol (Table 1a). When limited to ART-treated, virologically-suppressed PLWH (HIV-1 RNA <200 copies/mL; N=99) and HIV-uninfected controls (Table 1b), independent factors associated with FGF21 included older age, alcohol use, increased trunk fat, and positive HIV status.

Table 1.

Multivariable Regression Models for Variables Associated with FGF21

A. All PLWH Combined
Variable*	ß	ß (SE)	P
Age	0.62	0.31	0.049
Current smoking	0.48	0.17	0.005
Current alcohol use	0.24	0.17	0.16
Triglycerides	0.48	0.16	0.003
LDL cholesterol	−0.008	0.003	0.002
IL-6	0.35	0.10	0.0008
Trunk fat	0.16	0.13	0.22
HIV-1 RNA >200 copies/mL	−0.33	0.21	0.12
	R2 = 0.40
B. ART-Treated, Virologically-Suppressed PLWH and Uninfected Controls
Variable	ß	ß (SE)	P
HIV Status**	−0.55	0.23	0.016
Age	0.88	0.35	0.014
Current alcohol use	0.52	0.18	0.004
Trunk fat	0.35	0.14	0.0097
	R2 = 0.21

N.B. In each model, the variable with the highest P value was eliminated in a stepwise fashion; variables with P<0.25 remained in final models. P-values with bolded font designate those <0.05;

^*HOMA-IR and sTNFR-I were included in the initial model but were eliminated in stepwise fashion;

^**PLWH had higher levels of FGF21 compared to controls; FGF21, fibroblast growth factor 21; PLWH, people living with HIV; LDL, low-density lipoprotein; IL-6, interleukin-6; HOMA-IR, homeostatic model assessment of insulin resistance; sTNFR-I, soluble tumor necrosis factor receptor-I.

Notably, IL-6 was highly significantly associated with FGF21 concentrations in the multivariable regression model. IL-6 is an important pro-inflammatory cytokine in HIV³² and strongly associated with AIDS- and non-AIDS-defining morbidity and mortality.^33–37 Acute HIV causes a rapid and intense release of IL-6, but even with ART, PLWH have on average 40–60% higher concentrations of IL-6 than well-matched, uninfected adults.³⁸ IL-6 is generally regarded as a pro-inflammatory cytokine, but its signaling pathways and effects are complex and broad, and, like FGF21, IL-6 has hormone-like attributes that affect vascular disease, lipid metabolism, insulin resistance, the neuroendocrine system, and much more.³⁹ Almost all cells of the immune system produce IL-6, particularly monocytes and macrophages; however, a wide variety of other cells also produce IL-6, including adipocytes.^39,40

FGF21 has been shown to have anti-inflammatory effects.^2,41 Importantly, however, in Berti, et al,⁴² supraphysiological concentrations of FGF21 applied to human preadipocytes for 18 days (mimicking chronic high exposure) markedly enhanced IL-6 release. These results, taken together with our current data, suggest that elevated circulating FGF21 concentrations could increase systemic inflammation in PLWH, further contributing to the milieu of global immune dysfunction already playing a pivotal role in the pathogenesis of comorbidities in this population.^29–31

Few other studies have investigated the relationship between FGF21 and inflammation in HIV. Gallego-Escuredo, et al⁴³ showed that age-adjusted FGF21 concentrations were significantly positively associated with IL-6, interleukin-8, and tumor necrosis factor-α, in the context of evaluating the relationship between bone homeostasis and FGF21 concentrations. Similarly, Braun, et al⁴⁴ showed a positive bivariate relationship between FGF21 and soluble CD14, while evaluating changes in FGF21 concentrations after tesamorelin treatment. Neither of these studies, however, were intended to investigate the relationship between inflammation and FGF21 in PLWH.

HIV-related factors did not appear to play a role in FGF21 concentrations, which is consistent with other in vivo studies that evaluated a limited number of HIV-specific variables.^32,44 In contrast, Moore, et al⁴⁵ showed that in vitro antiretrovirals elicit FGF21 gene expression. This discrepancy may be due to the fact that in vivo the relationship between FGF21 concentrations and metabolic parameters (and derangements) overshadow the influence of specific antiretrovirals.

Indeed, Praktiknjo, et al⁴⁶ showed that non-obese PLWH with severe fatty liver had significantly higher concentrations of FGF21 compared to those with no or mild disease, and Domingo, et al²² demonstrated that the highest FGF21 concentrations were seen in ART-treated PLWH with lipodystrophy, a syndrome characterized by metabolic abnormalities.⁴⁷ In line with this latter study (and studies outside of HIV^{12–15,48–50}), we showed strong relationships between FGF21 concentrations and metabolic risk factors, including increased trunk fat (but not overall fat or body mass index).

One hypothesis is that in metabolically-unhealthy obesity, chronically-elevated concentrations of FGF21 from fatty liver lead to an FGF21-resistant state in adipocytes.^2,42,51 In support of this, in the aforementioned Berti, et al⁴², chronic exposure to high concentrations of FGF21 markedly altered human preadipocytes’ release of classical adipokines (with adiponectin release suppressed and leptin release enhanced), which was modulated in part by FGF21’s impact on expression of respective genes. Data in HIV are sparse, but one study reported reduced expression of adipocyte FGF21 co-receptors and little or no FGF21 mRNA expression in the context of high serum FGF21 concentrations,²³ supporting the concept of FGF21 resistance and/or impaired responsiveness.

If hepatic production of FGF21 secondary to fatty liver is the primary source of elevated circulating FGF21 concentrations that promote a metabolically-unhealthy phenotype as suggested, this would be particularly relevant in PLWH. NAFLD is becoming an increasingly important problem in HIV, with prevalence rates as high as 35–40%.^52,53 In support of this hypothesis, Braun, et al⁴⁴ showed that FGF21 was strongly and positively correlated with liver fat content in PLWH. Reductions in liver fat seen with tesamorelin treatment (growth factor hormone-releasing hormone analogue) were accompanied by decreases in FGF21 concentrations, independent of decreases in visceral adipose tissue.

In our study, there was a trend toward a positive correlation between FGF21 and aspartate transaminase (but not alanine aminotransferase). This is not surprising given that most people with NAFLD have normal liver transaminases.⁵⁴ We did not assess the presence of liver fat in our current study, which would be important to do in future studies.

Interestingly, a randomized-controlled trial reported a relative reduction in FGF21 concentrations among PLWH with metabolic or lipodystrophy syndrome after 1 year of lifestyle modification.⁵⁵ A recent phase 2 trial conducted among overweight/obese subjects with NAFLD showed significant liver fat improvement after 16 weeks with either daily or weekly FGF21 analogue.⁵⁶ Given these results and its association with metabolic syndrome risk factors, FGF21 may be considered as a future biomarker and/or therapeutic target in patients with metabolic disease.^48,57

Additional study limitations worth noting include our cross-sectional design and the fact that only a few participants had an established diagnosis of metabolic co-morbidities like diabetes, making it difficult to determine individual contributions of each disease state. However, many PLWH have metabolic co-morbidities and/or risk factors, and, arguably, understanding FGF21 in this population could be the most impactful. Likewise, there were a large proportion of PLWH on a statin or other drugs known to be beneficial in metabolic disorders and/or decrease inflammation; however, despite this, PLWH still had higher circulating FGF21 concentrations that were strongly associated with IL-6 and LDL cholesterol.

Our data contribute to the body of literature showing a relevant role of FGF21 in HIV and HIV-related metabolic abnormalities. Additional studies should be conducted to elucidate mechanisms by which FGF21 affects metabolic parameters in HIV. Specifically, further research is needed to determine how IL-6 and FGF21 pathways interrelate and the subsequent impact on comorbidity development in this population.