A shift from AIDS-related causes of morbidity and mortality to non-AIDS causes such as non-AIDS malignancy, liver cirrhosis, end-stage renal disease, and serious cardiovascular events occurred in patients with HIV nearly 1 decade ago because of the use of potent antiretroviral therapy (ART).1 A recent review examined the evidence suggesting a link between a persistent inflammatory state, even with suppressive ART, and morbidity and mortality attributed to these conditions termed serious non-AIDS events (SNAs) in HIV studies.2
As in studies in the general population,3–5 higher levels of inflammatory and coagulation markers have been associated with an increased risk of all-cause mortality,6–8 cardiovascular disease (CVD),9–12 and cancer13 in people with HIV. Furthermore, markers of systemic inflammation, such as interleukin-6 (IL-6), and markers of coagulation, such as D-dimer, are elevated with HIV infection, despite effective ART.14
In 2 recent reports,15,16 IL-6 and D-dimer were strongly related to all-cause mortality among participants with HIV. Most of the deaths in these analyses were not attributable to AIDS or serious non-AIDS conditions such as CVD and cancer. This raises the question whether serious conditions other than AIDS, CVD, and cancer that have heretofore not been studied, are a consequence of underlying increased inflammation and coagulation. In this article, we use data from 2 large international trials to describe rates of potentially life-threatening events not attributable to AIDS, CVD, or non-AIDS cancer and assess whether these events have an inflammatory component by studying their association with IL-6 and D-dimer.
This analysis includes participants from the control arms of 2 clinical trials conducted by the International Network for Strategic Initiatives in Global HIV Trials (INSIGHT): the Strategies for Management of Antiretroviral Therapy (SMART) study and the Evaluation of Subcutaneous Proleukin in a Randomized International Trial (ESPRIT).17,18 In SMART, HIV-positive adults with a baseline CD4+ cell count above 350 cells/mm3 were randomized to receive either continuous ART (control group) or episodic ART based on the CD4+ count. ESPRIT compared IL-2 plus ART versus ART alone (control) among HIV-positive adults with CD4+ counts ≥ 300 cells/mm3. As in a previous report that examined morbidity and mortality rates by age,19 the analyses in this report are restricted to HIV-positive participants in the control groups of SMART and ESPRIT who were on ART with an HIV-RNA level ≤ 500 copies/mL at their baseline visit to focus on serious outcomes that occur despite effective ART. Control arms in both studies were to receive continuous ART throughout follow-up aimed at viral suppression.
Baseline measurements that were obtained before randomization in both SMART and ESPRIT included age, sex, race, body mass index, baseline CD4+ and nadir CD4+ cell count, HIV-RNA level, duration of ART, history of an AIDS event, hepatitis B/C coinfection, diabetes, use of blood pressure (BP) and lipid-lowering medication, and type of ART. IL-6 and D-dimer were measured using stored plasma specimens collected at baseline for participants who provided written informed consent. Laboratory methods have been previously described.16 All samples were analyzed masked to treatment and event status.
Outcomes considered in this investigation were defined using the same criteria in SMART and ESPRIT. Using preestablished diagnostic criteria, an endpoint review committee (ERC) reviewed documentation provided by clinical sites following events for AIDS events, and the following SNAs: CVD (myocardial infarction, stroke, and coronary revascularization) events; cirrhosis; end-stage renal disease; and non–AIDS-defining cancers (except for basal cell or squamous cell skin cancer). Events considered to be confirmed or probable by the ERC were counted as endpoints.20 Causes of death were coded according to the CoDe system.21,22
Adverse experiences were graded on a 4-point severity scale (1–4) by site clinicians using a National Institute of Allergy and Infectious Diseases (NIAID) Division of AIDS toxicity table (http://rsc.tech-res.com/docs/default-source/safety/daids_ae_grading_table_v2_nov2014.pdf). In ESPRIT and SMART, only grade 4 events were reported; events of lesser severity were not. In general, grade 4 events were defined as symptoms, signs, or diagnoses that were potentially life threatening. For conditions which were not in the toxicity table, clinician judgment was a key factor in determining each event's severity grade. In these cases, clinicians used the following generic definition of grade 4 events: extreme limitation in activity—significant assistance required; significant medical intervention/therapy required; and hospitalization possible. Grade 4 events were reported throughout each trial irrespective of the assumed relationship with ART or whether ART had been discontinued. Events that were limited to a laboratory abnormality were not reported. AIDS events were not reported as grade 4 events. Two trained nurses coded the grade 4 events using the Medical Dictionary for Regulatory Activities (MedDRA), version 19.1. Each nurse participates in an NIAID quality control program that involves coding adverse events twice yearly. One nurse did the initial coding of grade 4 events for ESPRIT and SMART and the other nurse verified the coding. If there was a difference of opinion on the coding, the final code was adjudicated by a discussion between the 2 nurses. MedDRA System Organ Class (SOC) and Preferred Term (PT) codes are cited.
CVD, non-AIDS cancer events, liver cirrhosis, end-stage renal disease which were reviewed by the ERC were also reported as grade 4 events if grade 4 criteria were met. CVD and cancer events that did not meet ERC criteria were frequently reported as grade 4 events.
Throughout the trial, on-site visits were conducted to review charts for grade 4 events, AIDS events, and SNAs which were not reported.
To define mutually exclusive event categories, we defined the following major outcomes for this investigation: (1) fatal or nonfatal AIDS according to the ERC; (2) fatal or nonfatal CVD meeting ERC criteria plus grade 4 events that did not meet ERC criteria considered to be CVD by one of us (DAD); (3) fatal or nonfatal non–AIDS-defining cancer meeting ERC criteria; and (4) grade 4 events and deaths other than those in the 3 aforementioned categories. Liver cirrhosis and end-stage renal disease events meeting ERC criteria are included in the fourth category because they occurred in too few participants to consider separately. The fourth category also includes some cancer events that did not meet ERC criteria (eg, basal cell or squamous cell skin cancer).
Each participant was allowed to qualify for more than 1 event type during follow-up and then was counted in all relevant categories for event-specific analyses. All events were reviewed either by the ERC or by trained nurses masked to treatment assignment, event status, and baseline characteristics, including IL-6 and D-dimer levels.
Grade 4 events in the fourth category were also divided into 2 groups according to 2 recent reports from the Multi-Ethnic Study of Atherosclerosis (MESA).23,24 In those investigations, International Classification of Diseases diagnostic codes for hospitalizations and deaths for diagnoses that were not coded as CVD, cancer, injury, acute organ failure, psychoses, substance abuse, and metabolic disorders such as diabetes were reviewed and those medically considered to have a chronic inflammatory component or to require response to a massive infection were classified as chronic inflammation–related disease (ChrIRD) conditions. Diagnostic codes and records for the following chronic and severe pathologic conditions were reviewed: infectious diseases; endocrine, nutritional, and metabolic diseases; nervous, respiratory, and digestive system diseases; skin diseases; musculoskeletal and connective tissue disorders; genitourinary diseases; and blood disorders. Two physicians independently reviewed the diagnostic codes with no information about the study participants and identified conditions which they considered to have an important inflammatory component. In these investigations, lower levels of small-plus-medium high-density lipoprotein particles and higher levels of IL-6 and D-dimer were associated with an increased risk of ChrIRD events,23 and risks of death, CVD, ChrIRD, and cancer were associated with GlycA, IL-6, and D-dimer independently.24
This finding motivated a similar classification of SMART and ESPRIT grade 4 events. In our investigation, one of us (DAD), who was also one of the 2 physicians in the MESA who classified diagnostic codes, used the classification scheme developed for the MESA to categorize MedDRA PTs for grade 4 events and causes of death as ChrIRD or non-ChrIRD. Supplemental Digital Content Table S1, http://links.lww.com/QAI/B85, provides the PTs and frequencies corresponding to the ChrIRD and non-ChrIRD events.
Event rates and 95% confidence intervals (CIs) were estimated using Poisson regression. We then used Cox models to estimate the risk of death associated with AIDS, CVD, non-AIDS cancer, and grade 4 outcomes. For these analyses, events were considered as time-dependent covariates and all-cause mortality was the outcome. Unadjusted models are stratified by the studies (ESPRIT and SMART). Adjusted models are stratified by the studies and include the following baseline covariates: age, sex, race, body mass index, log10-transformed baseline CD4+ and nadir CD4+ cell count, previous AIDS, hepatitis B or C coinfection, use of lipid-lowering medication, use of BP medication, log2-transformed IL-6, log2-transformed D-dimer, and type of ART categorized as protease inhibitor (PI) without nonnucleoside reverse transcriptase inhibitor (NNRTI), NNRTI with or without a PI, and NRTIs with neither a PI or NNRTI.
Cox regression, stratified by the studies, was also used to study predictors, including IL-6 and D-dimer, of the major outcomes previously defined. Log2-transformed IL-6 and D-dimer were considered as predictors separately. With the log2 transformation, exponentiation of the parameter estimates from the Cox model provides the increased risk of the event associated with a doubling of the biomarker level. Other baseline covariates considered in the adjusted models with IL-6 and D-dimer were the same as those considered for the risk of death analyses. Hazard ratios (HRs) and 95% CIs are cited. The homogeneity of associations of IL-6 and D-dimer with the major outcomes was assessed using the method of Wei, Lin and Weissfeld (WLW).25 We also compared associations between biomarkers and ChrIRD and non-ChrIRD grade 4 events with this approach.
We examined the proportional hazards assumption for the biomarkers by introducing an interaction term between each log2-transformed biomarker and log follow-up time. We also calculated HRs associated with a doubling of each biomarker after excluding events in the first 2 years. As a sensitivity analysis to consider whether an earlier event that might be considered a chronic inflammatory condition did not influence the rate of a subsequent event, when considering ChrIRD events, we censored follow-up if an AIDS, CVD, or non-AIDS cancer event preceded it. Similarly, when considering non-ChrIRD events, we censored follow-up at the time of an AIDS, CVD, non-AIDS cancer, or ChrIRD event.
All analyses were performed using SAS version 9.3 (Cary, NC). P-values < 0.05 were considered statistically significant.
In total, 4792 participants were enrolled in the control arms of SMART and ESPRIT (2752 and 2040, respectively). After removing participants with CD4+ levels < 300 cells/mm3 (3 participants), participants on ART with an HIV-RNA level > 500 copies/mL (756), and those not on ART (465), 3568 participants remained. Women made up 22.8% of the 3568 participants. Ages ranged from 19 to 77 years, with median [interquartile range (IQR)] age being 42 (36–49) years. The median (IQR), baseline CD4+ and nadir CD4+ cell counts were 547 (421–727) and 210 (108–314) cells/mm3, respectively. Of total, 25.6% reported an AIDS event before enrollment. The median (IQR) time on ART at baseline was 5 (3–8) years. The median (IQR) follow-up time in years was 4.3 (2.1–6.7); the total person-years of follow-up were 16,178.
The rate of grade 4 events (22.9 per 1000 person-years) was higher than that of the composite of AIDS, CVD, non-AIDS cancer, and all-cause mortality (Table 1). Two hundred fifty-two participants had at least 1 AIDS, CVD, or non-AIDS cancer event for a rate of 17.1 events per 1000 person-years. Of those who experience a grade 4 event, most participants (252 of 339) experienced 1 grade 4 event; 54 experienced 2 events; and 33 experienced 3 or more events. ChrIRD grade 4 events accounted for 49% of the participants with a grade 4 event. The most common ChrIRD events were gastroenteritis (8 participants), hepatic cirrhosis (7), acute renal failure (6), and acute pancreatitis (6). The most common non-ChrIRD grade 4 events were depression (7), back pain (5), inguinal hernia (5), and suicide attempt (5) (Supplemental Digital Content Table S1, http://links.lww.com/QAI/B85). Grade 4 events were also summarized by SOC. Of the MedDRA SOCs, the rates of infections and infestations and gastrointestinal disorders were highest (Supplemental Digital Content Table S2, http://links.lww.com/QAI/B85).
Severity as Measured by Risk of Death for Grade 4 and Other Events
In both the unadjusted and covariate adjusted models using time-dependent indicators for grade 4 events, the risk of death was significantly greater (P-value < 0.001) for those who experienced a grade 4 event as compared to those who did not (Table 2). For those who experienced 2 or 3 grade 4 events, the risk of death was further increased compared with those who did not. Adjusted HRs associated with 1, 2, and 3 or more events were 5.0 (95% CI: 2.8–9.0; P-value = <0.001), 7.8 (95% CI: 3.0–20.7; P-value = <0.001), and 27.9 (95% CI: 12.5–62.0; P-value = <0.001), respectively. The HR for grade 4 events was similar to that of AIDS events but was smaller than those of CVD and non-AIDS cancer events. The risk of death was greatest for those with non–AIDS-defining cancers (26.8; 95% CI: 15.7–45.7).
In separate models, adjusted HRs for death after ChrIRD and non-ChrIRD grade 4 events were 12.4 (95% CI: 7.4–20.8) and 5.2 (95% CI: 3.0–8.9), respectively. Additional models considered adjusted and unadjusted risks of death including time-dependent covariates for both ChrIRD and non-ChrIRD grade 4 events in the same model. The estimated HRs and 95% CIs for the unadjusted and adjusted models, respectively, were 9.5 (6.1–14.9) and 10.5 (6.2–17.8) for ChrIRD events and 4.1 (2.6–6.6) and 3.8 (2.2–6.7) for non-ChrIRD events. The risk of death associated with ChrIRD events was significantly greater than the risk of death associated with non-ChrIRD grade 4 events in the unadjusted (P-value = 0.03) and adjusted (P-value = 0.02) models.
Associations With IL-6 and D-dimer
Higher baseline levels of IL-6 and D-dimer were associated with an increased risk with each of the major outcomes in unadjusted models (Table 3). In the covariate adjusted analysis, HRs were generally reduced but remained significant for most associations, including both biomarkers for ChrIRD events and D-dimer for non-ChrIRD events (Table 4). The association of D-dimer with CVD events that met ERC criteria was similar to the associations with CVD events that met grade 4 but not ERC criteria, whereas IL-6 was more associated with CVD events that met ERC criteria than the grade 4 CVD events that did not meet ERC criteria (Table 4).
There was no evidence that the proportional hazards assumption was violated for any of the biomarker–outcome associations. When events occurring in the first 2 years were excluded (128 of 339 events) from the analysis, HRs associated with a doubling of IL-6 for all grade 4, ChrIRD grade 4, and non-ChrIRD grade 4 events were 1.14 (95% CI: 0.96–1.36), 1.23 (95% CI: 0.97–1.56), and 1.15 (95% CI: 0.91–1.44), respectively. Excluding events in the first 2 years, HRs associated with a doubling of D-dimer for all grade 4, ChrIRD grade 4, and non-ChrIRD grade 4 events were 1.28 (95% CI: 1.10–1.49), 1.39 (95% CI: 1.13–1.70), and 1.14 (95% CI: 0.94–1.40), respectively.
We also performed a sensitivity analysis for the ChrIRD events in which follow-up was censored if an AIDS, CVD, and non-AIDS cancer event occurred before the ChrIRD event. This had a very little effect on the HRs for the 2 biomarkers (Supplemental Digital Content Table S3, http://links.lww.com/QAI/B85). Likewise, similar analyses for the non-ChrIRD events that censored follow-up for these events or a ChrIRD event also yielded similar HRs (Supplemental Digital Content Table S3, http://links.lww.com/QAI/B85).
We assessed the homogeneity of IL-6 and D-dimer associations with the major outcomes. The effect of D-dimer was homogenous across all major outcomes (P = 0.50), whereas the effect of IL-6 varied across the major outcomes (P < 0.001), largely reflecting the stronger associations with all-cause mortality as compared to other events.
We also assessed the homogeneity of IL-6 and D-dimer with ChrIRD grade 4 events and non-ChrIRD grade 4 events. The association with D-dimer was similar for ChrIRD and non-ChrIRD events (P = 0.66). The association with IL-6 was stronger for ChrIRD events than non-ChrIRD events (P = 0.003).
Given the positive association between the 2 markers and each event, we considered a model that included both biomarkers after log transformation and other baseline covariates with the composite outcome of AIDS, CVD, non-AIDS cancer, grade 4 events, or death from any cause (n = 564 participants with at least 1 event), to study the effect of each marker independent of the other. Based on this model, a doubling of IL-6 and D-dimer yielded HRs of 1.21 (95% CI: 1.10–1.33; P < 0.001) and 1.23 (95% CI: 1.12–1.34; P < 0.001), respectively.
Other Factors Associated With Grade 4 Events and With Those Classified as ChrIRD and Non-ChrIRD Grade 4 Events
Grade 4 events were significantly associated with black race, use of BP-lowering treatment, and coinfection with hepatitis in addition to IL-6 and D-dimer (Supplemental Digital Content Table S4, http://links.lww.com/QAI/B85). Associations between baseline factors and ChrIRD and non-ChrIRD events were similar (Supplemental Digital Content Tables S5, http://links.lww.com/QAI/B85 and S6, http://links.lww.com/QAI/B85). The associations with black race and coinfection with hepatitis were significant for both types of grade 4 events.
Consistent with 2 other reports, which included participants with lower CD4+ counts, we show that grade 4 events are a major source of morbidity among participants with HIV.26,27 Among the participants in our cohort, all of whom had CD4+ counts ≥ 300 cells/mm3 at study entry, the rate of grade 4 events was 3–6 times higher than AIDS, CVD (expanded to include less serious events and CVD events that did not meet ERC criteria), or non-AIDS cancer considered separately and was higher than the rate of these 3 outcomes considered as a single composite outcome.
Everyone in our investigation was taking suppressive ART. Thus, we can only speculate whether the grade 4 events are due to underlying HIV disease or to ART. In the Strategic Timing of AntiRetroviral Treatment (START) trial, which compared immediate versus deferred ART in participants with much higher nadir CD4+ counts, rates of grade 4 events and unscheduled hospitalizations were small and not significantly different between the 2 ART strategies.28 This suggests that among participants with higher nadir CD4+ counts, grade 4 events are more likely due to underlying HIV disease and other comorbidities such as hepatitis coinfection, demographics (age and race), and lifestyle factors (eg, hypertension) than ART.
We found that baseline measurements of IL-6 and D-dimer, often measured several years before the event, were strongly related to the risk of grade 4 events. The biomarker associations with grade 4 events were similar in strength to those for CVD and non-AIDS cancer, 2 non-AIDS conditions which have been widely studied in HIV.1 This adds some support to our hypothesis that the consequences of underlying inflammation are broader than the end-organ diseases referred to as serious non-AIDS conditions.
Grade 4 events represent a variety of potentially life-threatening conditions. We used findings from a recent study in the general population to categorize grade 4 events as those likely to have an inflammatory component as a predominant pathology. ChrIRD events represented 49% of the participants with grade 4 events, and these events were more strongly associated with baseline IL-6 than non-ChrIRD events. However, the associations of baseline D-dimer with ChrIRD and non-ChrIRD events were similar—both significant and of a magnitude similar to the association with CVD. Given the positive, but moderate, correlation between IL-6 and D-dimer,16,29 in the setting of HIV, there may be larger number of grade 4 events due to HIV-induced activation of inflammatory and coagulation pathways than we classified as ChrIRD. Although more research on the extent of morbidity and mortality due to inflammation is needed, it seems clear that there are a number of conditions associated with an increased risk of death other than AIDS and those that we and others have considered “serious non-AIDS conditions” that are influencing the health of patients with HIV and that could be targets of intervention trials that aim to reduce inflammation.
It is possible that the associations of IL-6 and D-dimer with grade 4 events may be different among patients with higher nadir CD4+ cell counts. IL-6 levels are inversely related to nadir CD4+ counts30 but D-dimer levels are positively associated with nadir CD4+ cell counts.29 In a recent review article,2 it was suggested that the drivers of inflammation may be different in those who initiate ART early after infection compared with later as in this investigation. This result poses the question of whether the associations of IL-6 and D-dimer with grade 4 events depend on the nadir CD4+ count.
The findings from our investigation have implications for the design of trials of novel interventions that target inflammation in HIV. Such trials might be more efficiently performed with composite primary or secondary outcomes that include not only AIDS, CVD, and non-AIDS cancer events but also other inflammatory conditions such as all or a subset of grade 4 events.
Strengths of our study include the size of the cohort and the systematic collection of AIDS, CVD, non-AIDS cancer, and grade 4 events over a median follow-up of 4.3 years.
Weaknesses include the absence of data on the occurrence of grade 4 events before enrollment. Grade 4 events occurring during follow-up may not be incident events (eg, they could be a worsening of a preexisting condition). Also, although events reviewed by the ERC had extensive documentation, documentation of grade 4 events typically consisted of 2 or 3 terms that defined the medical event. Thus, classification of grade 4 events as CVD and as ChrIRD or non-ChrIRD is imperfect. In addition, some grade 4 events may have been scheduled hospitalizations for procedures/conditions which were not life threatening. Finally, although the median baseline CD4+ count was 547 cells/mm3 for our cohort, the median nadir CD4+ count was 210 cells/mm3. As previously suggested, it will be important to understand the relative rate of grade 4, AIDS, CVD, and non-AIDS cancer and associations of these events with IL-6 and D-dimer among individuals with higher nadirs.
In conclusion, we found that IL-6 and D-dimer were strongly associated with grade 4 events. These events are associated with an increased risk of death, and among participants on suppressive ART with CD4+ counts ≥ 300 cells/mm3, occur at a much higher rate than AIDS, CVD, and non-AIDS cancer events. Our findings showed that potentially life-threatening conditions associated with the activation of inflammatory and coagulation pathways due to HIV are more extensive than AIDS and SNAs which heretofore have been considered.
The authors acknowledge the SMART and ESPRIT participants and investigators (see N Engl J Med 2006;355:2283–2296 for the complete list of SMART investigators and N Engl J Med 2009;361:1548–1559 for the ESPRIT investigators).
1. Phillips AN, Neaton J, Lundgren JD. The role of HIV in serious diseases other than AIDS
2. Hunt PW, Lee SA, Siedner MJ. Immunologic biomarkers, morbidity, and mortality in treated HIV infection. J Infect Dis. 2016;214(suppl 2):S44–S50.
3. Folsom AR, Delaney JA, Lutsey PL, et al. Associations of factor VIIIc, D-dimer
and plasmin-antiplasmin with incident cardiovascular disease and all-cause mortality. Am J Hematol. 2009;84:349.
4. Danesh J, Kaptoge S, Mann AG, et al. Long-term interleukin-6
levels and subsequent risk of coronary heart disease: two new prospective studies and a systematic review. PLoS Med. 2008;5:e78.
5. Ridker PM, Rifai N, Stampfer MJ, et al. Plasma concentration of interleukin-6
and the risk of future myocardial infarction among apparently healthy men. Circulation. 2000;101:1767–1772.
6. Kuller LH, Tracy R, Belloso W, et al. Inflammatory and coagulation biomarkers and mortality in patients with HIV infection. PLoS Med. 2008;5:e203.
7. Tien PC, Choi AI, Zolopa AR, et al. Inflammation and mortality in HIV-infected adults: analysis of the FRAM study cohort. J Acquir Immune Defic Syndr. 2010;55:316.
8. Ledwaba L, Tavel JA, Khabo P, et al. Pre-ART levels of inflammation and coagulation markers are strong predictors of death in a South African cohort with advanced HIV disease. PLoS One. 2012;7:e24243.
9. Triant VA, Meigs JB, Grinspoon SK. Association of C-reactive protein and HIV infection with acute myocardial infarction. J Acquir Immune Defic Syndr. 2009;51:268.
10. Duprez DA, Neuhaus J, Kuller LH, et al. Inflammation, coagulation and cardiovascular disease in HIV-infected individuals. PLoS One. 2012;7:e44454.
11. Ford ES, Greenwald JH, Richterman AG, et al. Traditional risk factors and D-dimer
predict incident cardiovascular disease events in chronic HIV infection. AIDS
12. Nordell AD, McKenna M, Borges ÁH, et al. Severity of cardiovascular disease outcomes among patients with HIV is related to markers of inflammation and coagulation. J Am Heart Assoc. 2014;3:e000844.
13. Borges ÁH, Silverberg MJ, Wentworth D, et al. Predicting risk of cancer during HIV infection: the role of inflammatory and coagulation biomarkers. AIDS
14. Neuhaus J, Jacobs DR Jr, Baker JV, et al. Markers of inflammation, coagulation, and renal function are elevated in adults with HIV infection. J Infect Dis. 2010;201:1788–1795.
15. Borges ÁH, O'Connor JL, Phillips AN, et al. Interleukin 6 is a stronger predictor of clinical events than high-sensitivity C-Reactive protein or D-dimer
during HIV infection. J Infect Dis. 2016;214:408–416.
16. Grund B, Baker JV, Deeks SG, et al. Relevance of Interleukin-6
for serious non-AIDS
morbidity and death among HIV-positive adults on suppressive antiretroviral therapy. PLoS One. 2016;11:e0155100.
17. SMART Study Group. CD4+ count-guided interruption of antiretroviral treatment. N Engl J Med. 2006;355:2283–2296.
18. The INSIGHT-ESPRIT Study Group and SILCAAT Scientific Committee. Interleukin-2 therapy in patients with HIV infection. N Engl J Med. 2009;361:1548–1559.
19. Miller CJ, Baker JV, Bormann AM, et al. Adjudicated morbidity and mortality outcomes by age among individuals with HIV infection on suppressive antiretroviral therapy. PLoS One. 2014;9:e95061.
20. Lifson AR, Rhame FS, Belloso WH, et al. Reporting and evaluation of HIV-related clinical endpoints in two multicenter international clinical trials. HIV Clin Trials. 2006;7:125–141.
21. Kowalska JD, Friis-Møller N, Kirk O, et al. The coding causes of death in HIV (CoDe) project: initial results and evaluation of methodology. Epidemiology. 2011;22:516–523.
22. Lifson AR, Belloso WH, Carey C, et al. Determination of the underlying cause of death in three multicenter international HIV clinical trials. HIV Clin Trials. 2008;9:177–185.
23. Duprez DA, Otvos J, Tracy RP, et al. High-density lipoprotein subclasses and noncardiovascular, noncancer chronic inflammatory-related events versus cardiovascular events: the multi-ethnic study of Atherosclerosis. J Am Heart Assoc. 2015;4:e002295.
24. Duprez DA, Otvos J, Sanchez OA, et al. Comparison of the predictive value of GlycA and other biomarkers of inflammation for total death, incident cardiovascular events, noncardiovascular and noncancer inflammatory-related events, and total cancer events. Clin Chem. 2016;62:1020–1031.
25. Wei L, Lin DY, Weissfeld L. Regression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073.
26. Reisler RB, Han C, Burman WJ, et al. Grade 4 events
are as important as AIDS
events in the era of HAART. J Acquir Immune Defic Syndr. 2003;34:379–386.
27. Maduna PH, Dolan M, Kondlo L, et al. Morbidity and mortality according to latest CD4+ cell count among HIV positive individuals in South Africa who enrolled in project Phidisa. PLoS One. 2015;10:e0121843.
28. INSIGHT START Study Group. Initiation of antiretroviral therapy in early asymptomatic HIV infection. N Engl J Med. 2015;373:795–807.
29. Borges AH, O'Connor JL, Phillips AN, et al. Factors associated with D-dimer
levels in HIV-infected individuals. PLoS One. 2014;9:e90978.
30. Borges AH, O'Connor JL, Phillips AN, et al. Factors associated with plasma IL-6 levels during HIV infection. J Infect Dis. 2015;212:585–595.