The majority of participants in the BIENESTAR cohort were frankly IR as defined by a QUICKI value <=0.350 [44
]. However, the prevalence of IR was not significantly higher in HIV-infected participants compared with the uninfected controls, and HIV-infection was not an independent predictor of IR when adjusted for other variables in multivariate analyses. These results agree with a previous study conducted by our group in the Nutrition for Healthy Living Cohort [50
], in which the rate of IR in HIV-infected participants was 51%, a prevalence that was not significantly different from matched controls derived from the NHANES III study (47%) in analyses adjusted for known predictors of IR. Brown et al. found in the MACS cohort higher rates of elevated insulin among HIV-infected men compared to HIV-uninfected control subjects after adjusting for a variety of risk factors including antiretroviral therapy [51
]. In the present study no other HIV-related factors were associated with IR. Other studies have shown a weak association of insulin resistance with higher concurrent or nadir CD4 count [52
In addition to BMI and trigycerides, both of which are recognized risk factors for IR in the general population, predictors of IR in this population included HCV infection and the use of NRTI plus and NNRTI agents, but not PI. The ability of HCV to impact IR was apparent only in the HIV co-infected participants. In this group alone, antiretroviral therapy appeared to be an additional contributor to IR, independently of HCV. These two factors could explain why IR was more common in the HIV/HCV co-infected group compared to the HIV-mono infected group, despite the lower average BMI and triglyceride levels in the HIV/HCV co-infected group. There was no evidence that the effects of HCV infection and antiretroviral therapy act synergistically, though the absence of an effect of antiretroviral therapy in the HIV-mono-infected patients is suggestive. This study may have lacked the statistical power necessary to detect such an effect. Our results confirm previous reports of an effect of HCV on IR [24
The role of liver injury in IR, induced either by HCV or by antiretroviral therapy [53
], is not clear. In a multivariate model, both FIB-4 and HCV were independent predictors of IR. The addition of FIB-4 to the model with HCV attenuated the size of the effect estimate for HCV, but only by about 23%, and the term for HCV infection remained significant. These data support the hypothesis that HCV affects IR by a mechanism that is independent of HCV-induced liver damage [54
]. Studies conducted in a chimpanzee model suggest interplay between cellular lipid metabolism and HCV replication [56
Our results showed an independent effect of antiretroviral therapy on IR in the HIV/HCV co-infected participants although there was no association of IR with PI in this study. Since PI therapy was associated with IR early in its use [57
], it is likely that this was specific to select PI agents, such as indinavir, rather than to the class as a whole. NRTI and NNRTI class agents have also been associated with IR [50
]. The elevation of serum lactate by NRTI has been proposed as an explanatory mechanism for NRTI-related IR [59
]. Lactate levels were not collected in this study. Our data do not support an important role for peripheral lipoatrophy in IR. This result agrees with that of Meininger and colleagues [61
] who found that low extremity fat was a predictor of IR only among HIV-infected men with lipodystrophy, but not among HIV-infected men without lipodystrophy. It also agrees with a recent study by Blumer et al., showing NRTI-related IR may develop in persons showing no alterations in body composition [60
This study has numerous strengths. The participants were recruited from the community as opposed to a clinic-based sample, and therefore the data should be representative of the target population, and without reference to the presence of self-perceived lipodystrophy. A healthy control group was included in this study and evaluated by the same team using the same measures. The study was restricted to Hispanics, which gives the study high internal validity. Factors associated with HCV infection, which have been lacking in previous studies, including liver injury, HIV infection, drug use and other lifestyle factors were measured and evaluated in these analyses.
Limitations of the study include the cross-sectional design, which cannot establish the temporal relationship between HCV infection and IR. IR was measured using a surrogate and not the euglycemic hyperinsulinemic clamp technique [62
], considered the gold standard, which may have resulted in some misclassification. The extent of liver injury was not determined using liver biopsies. Finally, the generalizability of our findings may be limited given that only Hispanic subjects were included.
In conclusion, the prevalence of frank IR is high in this group of Hispanic adults, but not higher in person infected with HIV. Our findings contribute to a growing body of evidence that HCV has an important role in IR in the context of HIV. This study also suggests that the effect of HCV on IR is independent of liver injury. Antiretroviral therapy including NRTI agents appears to promote IR in HIV-infected Hispanics independently of obvious alterations in fat distribution. Further studies are needed to clarify the underlying mechanisms. Since HIV, HCV and IR disproportionately affect Hispanics in the US, physicians should routinely monitor this high-risk population, including HCV-infected patients with no apparent liver disease, and encourage exercise, weight control and other interventions that promote insulin sensitivity. Other culturally appropriate interventions, including dietary interventions, to reduce IR in this population are needed.