|Home | About | Journals | Submit | Contact Us | Français|
Background.When added to age, CD4 count and human immunodeficiency virus type 1 (HIV-1) RNA alone (Restricted Index), hemoglobin, FIB-4 Index, hepatitis C virus (HCV), and estimated glomerular filtration rate improve prediction of mortality. Weighted and combined, these 7 routine clinical variables constitute the Veterans Aging Cohort Study (VACS) Index. Because nonroutine biomarkers of inflammation (interleukin 6 [IL-6]), coagulation (D-dimer), and monocyte activation (sCD14) also predict mortality, we test the association of these indices and biomarkers with each other and with mortality.
Methods.Samples from 1302 HIV-infected veterans on antiretroviral therapy were analyzed. Indices were calculated closest to date of collection. We calculated Spearman correlations stratified by HIV-1 RNA and HCV status and measured association with mortality using C statistics and net reclassification improvement (NRI).
Results.Of 1302 subjects, 915 had HIV-1 RNA <500 copies/mL and 154 died. The VACS Index was more correlated with IL-6, D-dimer, and sCD14 than the Restricted Index (P < .001). It was also more predictive of mortality (C statistic, 0.76; 95% confidence interval [CI], .72–.80) than any biomarker (C statistic, 0.66–0.70) or the Restricted Index (C statistic, 0.71; 95% CI, .67–.75). Compared to the Restricted Index alone, NRI resulted from incremental addition of VACS Index components (10%), D-dimer (7%), and sCD14 (4%), but not from IL-6 (0%).
Conclusions.Among HIV-infected individuals, independent of CD4, HIV-1 RNA, and age, hemoglobin and markers of liver and renal injury are associated with inflammation. Addition of D-dimer and sCD14, but not IL-6, improves the predictive accuracy of the VACS Index for mortality.
We developed the Veterans Aging Cohort Study (VACS) Index because chronic human immunodeficiency virus (HIV) infection, aging, comorbidity, and treatment toxicity likely cause organ system injury beyond that reflected by CD4 cell count or HIV-1 RNA levels . The VACS Index incorporates age and 8 routinely monitored clinical tests: CD4 count; HIV-1 RNA, hemoglobin, aspartate aminotransferase, alanine aminotransferase, platelet, and creatinine levels; and hepatitis C virus (HCV) serostatus. The VACS Index predicts all-cause mortality more accurately than an does index restricted to age, CD4 count, and HIV-1 RNA among people living with HIV infection in North America and Europe [2–5]. However, the VACS Index has been limited to routinely available clinical data.
Among those with HIV infection, a strong association has been documented between biomarkers of chronic inflammation and morbidity and mortality [6, 7]. The biomarkers most consistently associated with patient outcomes include soluble interleukin 6 (IL-6; a marker of systemic inflammation), fibrin fragment D-dimer (D-dimer; a marker of procoagulant activity), and soluble CD14 (sCD14; a marker of lipopolysaccharide-mediated monocyte activation) . All 3 of these markers are elevated in untreated and, to a lesser degree, treated patients with HIV infection compared with uninfected controls [7, 8]. The mechanism for this persistent inflammation is unknown, but is likely to include T-cell dysfunction, a loss of immunoregulatory responses, excess burden of viral coinfections, and microbial translocation. We and others have postulated that chronic inflammation may contribute to development of end-organ dysfunction, which in turn can contribute to premature morbidity and mortality—as it does among the uninfected, aging population [6, 7, 9, 10].
The VACS Index was developed to capture the combined and likely interacting effects of HIV disease, comorbid disease, substance use, and treatment toxicity on outcome in HIV-infected subjects. Given that a complex association also likely exists between overall health and inflammation, we investigated the associations among 3 well-characterized biomarkers (IL-6, D-dimer, and sCD14), the VACS Index, and mortality in a large cohort of combination antiretroviral therapy (ART)–treated, HIV-infected veterans [6, 7].
VACS has been well described [11, 12]. It includes HIV-infected veterans and an age/race/site–matched control group of uninfected veterans in care. Our long-term goal is to understand the role of comorbid (ie, not AIDS-defining) medical and psychiatric disease, including alcohol and other substance use and abuse, in determining clinical outcomes in HIV infection. These analyses are based on data from the prospectively enrolled and consented 8-site study .
In 2005–2006, we collected and banked blood and DNA specimens from consenting subjects. Blood samples were provided by 1532 HIV-infected participants; of these, 1302 were on ART, defined as ≥3 antiretroviral agents for at least 1 year. Blood specimens were collected using serum separator and ethylenediaminetetraacetic acid blood collection tubes. After collection, blood specimens were shipped to a central repository, the Massachusetts Veterans Epidemiology Research and Information Center in Boston, Massachusetts.
All biomarkers were analyzed under the supervision of Dr Russell Tracy, director of Clinical Biochemistry Research at the University of Vermont [6, 7]. IL-6 was measured using a chemiluminescent immunoassay (QuantiGlo IL-6 immunoassay; R&D Systems, Minneapolis, Minnesota). This assay’s detectable range was 0.4–10 000 pg/mL. We have found that this assay’s sensitivity, which falls between the standard and the high-sensitivity enzyme-linked immunosorbent (ELISA) assays for IL-6 from R&D Systems, is most appropriate for studies of HIV-positive individuals. Calibration was by the manufacturer, traceable to the National Institute for Biological Standards and Control 89/548 (IU/mL). Neither soluble IL-6 receptor nor soluble gp130 demonstrates significant interference in this assay. Four levels of controls were run with the VACS samples, with the interassay coefficient of variation (CV) ranging from 7.7% to 12.3%.
D-dimer, a terminal product of plasmin acting on a fibrin, increases during the activation states of coagulation and was measured by the STAR automated coagulation analyzer (Diagnostica Stago) using an immuno-turbidometric assay (Liatest D-DI; Diagnostica Stago, Parsippany, New Jersey). The agents used in this assay are latex particles coated with 2 different mouse antihuman monoclonal antibodies specific to D-dimer. There is no significant cross-reactivity with native fibrinogen. The detectable range is 0.01–20 ug/mL. We used 4 controls, with interassay CVs ranging from 2.8% to 14.8%.
CD14 is the cell-surface receptor for lipopolysaccharide (endotoxin) on monocytes. The soluble form (sCD14) occurs either through proteolysis or loss of the glycophosphatidylinositol anchor via phospholipase activity, and increased sCD14 indicates the presence in blood of gram-negative bacteria. The lab measured sCD14 with an ELISA (Quantikine sCD14 immunoassay, R&D Systems), with a detectable range of 40–3200 ng/ml, using a standard 200-fold sample dilution and 4 controls, with interassay CVs ranging from 7.2% to 8.1%.
Since initial publication , we have refined the VACS Index by omitting AIDS-defining illnesses, diagnoses of substance use, and hepatitis B infection. AIDS events are uncommon among those on ART  and are variably associated with mortality . Diagnoses of substance abuse and hepatitis B infection are inconsistently measured across cohorts. As a result, the VACS Index is now based on age and 8 routine clinical tests. The Restricted Index is limited to CD4 count, HIV-1 RNA, and age (Table 1). Laboratory values used were closest to the date of the blood banking. Hepatitis C was considered present after a single seropositive test until a subsequent test demonstrated viral clearance. The accuracy of this revised index for mortality among HIV-infected individuals on ART has been externally validated .
We compared the associations of the VACS and Restricted indices and each component with IL-6, D-dimer, and sCD14. Because these biomarkers are not normally distributed, we used Spearman rank correlation to measure association. To evaluate whether correlations among VACS Index components and biomarkers of inflammation varied by level of HIV-1 RNA suppression or HCV infection, we also conducted analyses stratified by HIV-1 RNA <500 and ≥500 copies/mL and HCV status. In nested Cox models, we estimated unadjusted and adjusted associations with mortality and calculated Harrell’s C statistic. Because C statistics are insensitive to improvements in discrimination, we calculated net reclassification improvement (NRI) to determine whether addition of VACS Index components to the Restricted Index or addition of inflammatory biomarkers to the VACS Index improved discrimination [15, 16]. We used a parametric model (γ) to predict mortality for these calculations. Data was complete on 99.8% (1302 of 1304) of sampled subjects, so no imputation was done.
Participants were predominantly male (97%), black (69%), and aged 50–64 years (57%). Although samples were collected from 2005 to 2007, most were collected in 2006 (57%). We observed 154 deaths. At the time of sample collection (Table 2), one-third of subjects had CD4 counts >500 cells/mm3 (34%), and most had HIV-1 RNA levels <500 copies/mL (71%). Twelve percent had hemoglobin levels <12 g/dL. Ten percent had a fibrosis index 4 (FIB-4) consistent with fibrosis (>3.25), and 8% had some compromise in renal function (estimated glomerular filtration rate [eGFR] < 60). Nearly half (46%) of the participants were coinfected with HCV.
Measures of IL-6, D-dimer, and sCD14 were similar to prior reported values among patients with HIV infection [6, 8]. Overall, the median was 2.1 pg/mL (interquartile range [IQR], 1.4–3.4 pg/mL) for IL-6, 0.26 ug/mL (IQR, 0.15–0.48 ug/mL) for D-dimer, and 1728 ng/mL (IQR, 1462–2115 ng/mL) for sCD14. Among those with HIV-1 RNA<500 copies/mL, the median was 1.9 pg/mL (IQR, 1.3–2.1 pg/mL) for IL-6, 0.23 ug/mL (IQR, 0.15–0.39 ug/mL) for D-dimer, and 1692 ng/mL (IQR, 1435–2075 ng/mL) for sCD14.
All components of the VACS Index were associated with IL-6, D-dimer, and sCD14 (P < .005) (Table 2; Figures 1 and and2).2). VACS Index score was correlated with IL-6, D-dimer, and sCD14 (P < .0001) (Figure 1A–C), and these were of similar magnitude as correlation among biomarkers IL-6 and D-dimer, 0.45; IL-6 and sCD14, 0.41; D-dimer and sCD14, 0.24) (Figure 1D–F).
When stratified by HIV-1 RNA suppression and by HCV infection, the VACS Index remained more strongly correlated with each biomarker than the Restricted Index in all subgroups. Overall, correlations for all VACS Index components with IL-6 were strongest among those with HIV-1 RNA>500 copies/mL who were coinfected with HCV (Figure 2A–D). Differences across groups were less pronounced for D-dimer and sCD14. Correlations among non-HIV markers of organ injury and biomarkers of inflammation were of similar magnitude with correlations of CD4 count and these biomarkers in every subgroup. Hemoglobin demonstrated the strongest correlations among those with HIV-1 RNA>500 copies/mL, regardless of HCV status (Figure 2C–D). FIB-4 was relatively consistent in its correlations across groups with the exception of IL-6, which appeared more correlated with FIB-4 among those with HCV infection. Age demonstrated stronger correlations with inflammatory biomarkers among those with HIV-1 RNA<500 copies/mL who were HCV uninfected (Figure 2A).
All 3 biomarkers were associated with mortality in unadjusted analyses (P < .0001) (Table 2). However, inflammatory biomarkers alone or in combination did not discriminate mortality as well as the VACS Index (C statistic, 0.76; 95% confidence interval [CI], .72–.80). D-dimer and sCD14, but not IL-6, remained independently associated with mortality after adjustment for the VACS Index (P < .05). The VACS Index was more discriminating of mortality than the Restricted Index (C statistic, 0.71; 95% CI .67–.75). The NRI of adding VACS Index components to the Restricted Index was 10% (Figure 3). Adding D-dimer to the VACS Index resulted in an NRI of 7%, and further addition of sCD14 improved classification modestly (NRI, 4%). Adding IL-6 resulted in no improvement (NRI, 0%).
Among veterans with HIV infection on ART, an index that includes generic measures of organ system injury (VACS Index) is correlated with IL-6, D-dimer, and sCD14, whether or not subjects have achieved viral suppression or are coinfected with HCV. Although the mechanism for this association cannot be determined using our study design, the associations observed are consistent with the hypothesis that chronic inflammation contributes to end-organ dysfunction and/or that several chronic diseases common among those with HIV infection may be proinflammatory. Although all 3 biomarkers are associated with mortality, their associations are not as strong as those of the VACS Index. When added to the VACS Index, IL-6 was not independently associated with mortality, whereas D-dimer and sCD14 remained significant and improved risk reclassification. This suggests that effects of IL-6 on mortality are reflected by the VACS Index, whereas those of D-dimer and sCD14 are at least partially independent. Importantly, the VACS Index is more associated with biomarkers of chronic inflammation and with all-cause mortality than the Restricted Index; this establishes the independent association of hemoglobin, liver injury, and renal injury with inflammation and with mortality after adjustment for CD4 count, HIV-1 RNA level, and age.
The VACS provides a well-characterized and validated sample for these analyses. VACS data have been compared with >30 other cohorts in North America and Europe, and associations in VACS were found to be similar to those observed in other cohorts [17, 18].
Although they offer insight for understudied populations, our data may not generalize to all affected populations. When comparing VACS subjects with Centers for Disease Control and Prevention (CDC) HIV data , VACS subjects are more likely to be people of color (77% vs 64% black or Hispanic), older (62% vs 25% aged ≥50 years), and male (97% vs 75%). In both CDC and VACS data, 48% of men report having sex with men. VACS may not represent women or younger subjects, but it does offer insight regarding these biomarkers among the growing proportion of individuals aging with HIV infection, including men who have sex with men, those with HCV coinfection, and people of color. Moreover, we used a cutoff of HIV-1 RNA < 500 copies/mL when a lower threshold is reported in current assays. Among the 918 (71%) patients in our analysis with HIV-1 RNA < 500 copies/mL, 77% had HIV-1 RNA ≤75 copies/mL, and 31% had HIV-1 RNA ≤50 copies/mL. The correlations among VACS Index and inflammatory biomarkers were similar when we restricted analysis to HIV-1 RNA ≤50 copies/mL. We doubt that lower levels of detection would substantially alter our findings.
Our findings suggest that the chronic inflammation and organ system injury observed among HIV-infected individuals on ART are strongly related to each other. Given the associations demonstrated in our analyses and in prior work, this pathophysiologic process may result from synergistic effects of HIV, aging, substance use, multimorbidity, and medication toxicity [1, 20]. Prior literature has demonstrated an association between markers of inflammation and age , and recently, investigators have demonstrated that markers of inflammation are associated with CD4 count, with HIV-1 RNA, and with mortality among those with HIV infection [6, 7]. We extend these observations by demonstrating that an index with additional indicators for organ system injury is more strongly correlated with these biomarkers and with mortality. Of note, the combination of 8 variables into a single index decreases measurement error inherent in any single biomarker. Further, clinical tests such as those included in the VACS Index (and D-dimer) must comply with Clinical Laboratory Improvement Amendments regulatory standards and are less variable than basic science assays such as IL-6 or sCD14.
Because D-dimer and sCD14 reflect systemic inflammatory processes, they may provide additional insights beyond the VACS Index. For example, both D-dimer and sCD14 are likely involved in the development of vascular disease, pulmonary disease, osteoporosis, and central and peripheral nerve injury not currently included in the VACS Index. They may also reflect earlier signs of injury to the liver, kidney, and bone marrow than are reflected by FIB-4, eGFR, or hemoglobin. Because D-dimer is inexpensive and clinically available, adding it to the VACS Index might be a reasonable first step. Once D-dimer was included, the addition of sCD14 offered only modest improvement in the VACS Index.
The prevalence of anemia among HIV-infected individuals, its correlation with these biomarkers, and its independent association with mortality  underscores its importance among those aging with HIV infection. Little is known about the underlying mechanisms of anemia among those aging with HIV infection. For many patients, anemia improves on ART . Anemia that persists or develops after treatment may indicate poor adherence, emerging HIV resistance, or additional causes of chronic inflammation. The contributing roles of continued alcohol use, chronic HCV infection, other chronic viral infections, progressive liver disease, and renal disease in anemia should all be explored.
Liver disease is a leading cause of death among those aging with HIV infection. Prior work has demonstrated that HCV infection progresses more rapidly to liver failure and death among those with HIV infection [24–27], but few have explored liver injury among those with HIV infection not coinfected with HCV. Further, although there is growing evidence that FIB-4 is an excellent indicator of liver fibrosis among those with HIV and/or HCV infection [28, 29], its association with inflammation and all-cause mortality has not been previously reported.
Of organ systems included in the VACS Index, renal disease demonstrated the weakest associations with IL-6, D-dimer, and sCD14. This may reflect a true difference in association or limitations of eGFR as an indicator of renal disease. Although single-site studies suggest that among those with eGFRs <90 mL/minute/1.73 m the performance of eGFR is close to a 24-hour urine creatinine clearance [30, 31], it may not be as relevant as proteinuria or other less routinely assessed measures.
We believe that the VACS Index (with or without the addition of D-dimer) will prove more clinically useful than any single biomarker. Although our unadjusted hazard ratio for the highest quartile vs the lowest of of IL-6, D-dimer, and sCD14 on mortality were close to those reported among other populations with HIV infection [6, 7], the VACS Index was more strongly predictive of all-cause mortality than any biomarker and is readily quantified using standardized and widely available clinical tests. D-dimer and sCD14 improved classification when added to the VACS Index, but not as much as the addition of the VACS Index components improved the Restricted Index.
To date, we have demonstrated that the VACS Index is highly predictive of all-cause mortality and strongly associated with biomarkers of inflammation. We have yet to prove that changes in the VACS Index due to intervention correspond to changes in risk of mortality. Nothing short of a randomized trial can definitely establish the utility of the VACS Index as a surrogate outcome. Nevertheless, we have made 3 observations that support this assumption. First, we showed biologic plausibility in this analysis in that the VACS Index is more correlated with biomarkers of inflammation (IL-6), microbial translocation (D-dimer), and hypercoagulability (sCD14) than the Restricted Index . Second, in prior work we have shown that hemoglobin, FIB-4, and eGFR, as well as CD4 count and HIV-1 RNA, change substantially in response to ART initiation  and that the VACS Index is more responsive to ART initiation (and to differences in ART adherence) than the Restricted Index . Finally, also in prior work, we have demonstrated that the prognostic discrimination of the VACS Index is consistent when applied at any point over the first 5 years of ART . Taken together, these findings suggest that the VACS Index may provide a superior means of tracking disease burden over time among HIV-infected individuals on ART.
The VACS Index provides a stronger indication of inflammation (ie, stronger correlation with IL-6, D-dimer, and sCD14) and overall mortality than an index restricted to CD4 count, HIV-1 RNA, and age. The superior prognostic information available in the VACS Index can be realized without additional cost because required laboratory tests are already routinely monitored in clinical care. Whether the additional information offered by incorporating D-dimer (or sCD14) into the VACS Index justifies the added cost and complexity remains to be determined. Use of the VACS Index may help clinicians better gauge response to treatment so that modest improvement in a single test (eg, CD4 count) is not overshadowed by a decline in overall score.
We acknowledge the veterans who participate in the Veterans Aging Cohort Study and the study coordinators and staff at each of our sites and at the West Haven Coordinating Center. Without the commitment and care of these individuals, this research would not be possible. We would also like to acknowledge the substantial in-kind support we receive from the Veterans Affairs Healthcare System.
This work was supported by the National Institutes of Health: National Institute on Alcohol Abuse and Alcoholism (U10-AA13566), National Heart, Lung, and Blood Institute (R01-HL095136; R01-HL090342; RCI-HL100347), and National Institute on Aging (R01-AG029154; K23 AG024896). J. P. T. was supported by the Training Program in Environmental Epidemiology (T32 ES07069).
All authors: No reported conflicts.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.