Associations between immune depression and cardiovascular events in HIV infection Caroline A. Sabina, Lene Ryomb, Stephane De Witc, Amanda Mocrofta, Andrew N. Phillipsa, Signe W. Wormb, Rainer Weberd, Antonella D’Arminio Monfortee, Peter Reissf, David Kamaraa, Wafaa El-Sadrg, Christian Pradierh, Francois Dabisi,j, M Matthew Lawk, Jens Lundgrenb,l, for the D:A:D Study Group
Objective: To consider associations between the latest/nadir CD4þ cell count, and time spent with CD4þ cell count less than 200 cells/ml (duration of immune depression), and myocardial infarction (MI), coronary heart disease (CHD), stroke, or cardiovascular disease (CVD) (CHD or stroke) in 33 301 HIV-positive individuals. Design: Longitudinal cohort study. Methods: Analyses were undertaken using Poisson regression. To investigate whether analyses of stroke were robust to the type of endpoint, we additionally included stroke- like events and rejected strokes into the stroke endpoint. Results: Participants experienced 716 MI, 1056 CHD, 303 stroke, and 1284 CVD events. Whereas there was no evidence of a higher MI/CHD risk in those with lower latest/nadir CD4þ cell counts after adjustment [current CD4þ <100 cells/ml: relative rate (95% confidence interval) 0.96 (0.62–1.50) for MI, 0.89 (0.30–2.36) for CHD; nadir CD4þ <100 cells/ml: 1.36 (0.57–3.23) for MI, 0.98 (0.45–2.16) for CHD], stroke and CVD rates were higher in those with a latest CD4þ cell count less than 100 cells/ml [2.26 (1.29–3.94) and 1.14 (0.84–1.56), respectively]. All events occurred less fre- quently in those who had not experienced immune depression, although evidence for a linear association with duration of immune depression was weak. The association between stroke risk and the latest CD4þ cell count strengthened as stroke-like and rejected strokes were included; conversely, associations with established stroke risk factors weakened. Conclusion: We do not find strong evidence that HIV-positive individuals with a low CD4þ cell count are more likely to experience MI/CHD. Although strokes appear to occur more commonly at low CD4þ cell counts, this may be partly explained by misclassifi- cation or other biases. ß 2013 Wolters Kluwer Health | Lippincott Williams & Wilkins AIDS 2013, 27:2735–2748 Keywords: bias, cardiovascular disease, CD4þ lymphocyte count, myocardial infarction, stroke aResearch Department of Infection and Population Health, UCL, London, UK, bCopenhagen HIV Programme, University of Copenhagen, Copenhagen, Denmark, cDepartment of Infectious Diseases, CHU Saint-Pierre Hospital, Brussels, Belgium, dDivision of Infectious Diseases and Hospital Epidemiology, University Hospital Zurich, University of Zurich, Zurich, Switzerland, eHospital San Paolo, University of Milan, Milan, Italy, fAcademic Medical Center, and Stichting HIV Monitoring, Amsterdam, The Netherlands, gICAP-Columbia University/Harlem Hospital, New York, USA, hDe´partement de Sante´ Publique, Centre Hospitalier Universitaire, Nice, iINSERM, Centre INSERM U897 ‘Epide´miologie et Biostatistique’, jUniversite´ Bordeaux Segalen, Institut de Sante´ Publique Epide´miologie De´veloppement (ISPED), Bordeaux, France, kKirby Institute, University of New South Wales, Sydney, Australia, and lEpidemiklinikken M5132, Copenhagen University Hospital/Rigshospitalet, Copenhagen, Denmark. Correspondence to Professor Caroline Sabin, Research Department of Infection and Population Health, UCL, Royal Free Campus, Rowland Hill Street, London NW3 2PF, UK. Tel: +44 207 7940500x34752; e-mail: [email protected] See Acknowledgements section for full listing of study group. Received: 1 May 2013; revised: 5 June 2013; accepted: 13 June 2013.
DOI:10.1097/01.aids.0000432457.91228.f3
ISSN 0269-9370 Q 2013 Wolters Kluwer Health | Lippincott Williams & Wilkins 2735 Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 2736 AIDS 2013, Vol 27 No 17
Introduction Europe, Australia, and the United States [5]. The primary study aim is to investigate associations between use of The evidence for a potential role of immune depression antiretroviral drugs and risk of CVD and other major (CD4þ lymphocyte depletion) in the development of disease events. Data are collected prospectively during cardiovascular disease (CVD) among individuals with routine clinic visits; the standardized dataset includes HIV infection is conflicting. Whereas several studies information on sociodemographic factors, AIDS events have reported a higher risk of CVD in HIV-positive and deaths, known risk factors for CVD, laboratory individuals with a low CD4þ cell count [1–3], others markers for monitoring HIV and CVD, antiretroviral have reported only a modest association [4] or no treatment, and treatments that influence CVD risk. The association at all [5–9]. Although such an association, if present analysis includes the 33 301 participants from present, might be indicative of an inflammatory effect of enrolment cohorts I (1999–2000) and II (2004) only. untreated HIV infection [10,11], it may also reflect bias resulting from increased monitoring frequency in those For the analyses reported herein, we considered the first with low CD4þ cell counts, residual confounding, or occurrence of each of the following centrally validated misdiagnosis of HIV events as CVD. outcomes: MI – definite, possible, or unclassifiable events; CHD – MI, sudden cardiac death (fatal cases in One possible explanation for the inconsistent findings from which the underlying cause of death could not be the different studies is the variation in the endpoint that is established as a MI, but wherein cardiovascular risks were assessed. Whereas some studies have considered endpoints present at death, and there was no evidence of other based on a well defined and prospectively validated out- noncardiovascular causes of death), or invasive procedure come [e.g. myocardial infarction (MI), ischemic stroke], (coronary artery bypass graft, carotid endarterectomy, or others have considered composite CVD or coronary heart angioplasty); stroke; or CVD – first CHD or stroke event. disease (CHD) outcomes and studies may vary in whether The approach to data collection and classification was nonfatal as well as fatal events are considered. Study designs established at the initiation of the study with steps taken to may differ in respect to the methods used to capture ensure that any CVD events would be distinguishable information on outcomes, exposures, and potential con- from advanced HIV disease. In brief, information on all founders and the study setting may also have an impact on incident cases of MI/stroke is reported to the co- the completeness and reliability of outcome ascertainment. ordinating center via a study event form, which captures Finally, HIV-positive individuals may have infections that detailed information about the event and circumstances can cause symptoms similar to MI and stroke via non- surrounding it (and, for strokes, neuroimaging and atherosclerotic pathways. The clinical presentation of some cerebrospinal fluid data, wherever available). Each event is opportunistic infections [e.g. cytomegalovirus (CMV) validated and coded using criteria applied in the WHO infection, central nervous system (CNS) infections] may MONICA Study [15] and blind to information on the mimic the symptoms of CVD (angina and stroke, respec- patient’s clinical status (e.g. antiretroviral treatment status, tively) and syphilis may lead to stroke [12]. Furthermore, CD4þ cell count, viral load). Full details are provided in the lifestyles of many HIV-positive individuals mean that [16,17]. some may present with CVD or stroke secondary to recreational drug use [13,14]. As some of these factors may Classification of myocardial infarctions be more common in those with low CD4þ cell counts, the Reported MIs are classified as definite, possible or inadvertent inclusion of such events into analyses may unclassifiable, or are rejected if there was insufficient introduce bias when considering associations with the evidence to verify that the event was an MI. The diagnosis CD4þ cell count. and subsequent classification was based on an established algorithm [15] adapted from standardized criteria that The main aim of this study was to consider associations included relevant symptoms, relevant increase and decline between the latest and nadir CD4þ cell count, and time in cardiac enzymes, ischemic changes in electrocardio- spent with a CD4þ cell count less than 200 cells/ml graphic readings and, in cases of death, autopsy results (’duration of immune depression’), and incident CVD. if available. Secondary objectives were to assess the robustness of any associations seen to the type of event included in the Classification of strokes endpoint, and to investigate the possible impact of the Confirmed (hemorrhagic or ischemic) strokes are events inclusion of different types of events on reported in which symptoms persisted for more than 24 h, in prognostic values of established CVD risk factors. which there was a sudden symptom onset, in which there were specific neurological symptoms indicative of stroke, Methods and in which there was no evidence of any nonathero- sclerotic CNS events. Reported events that did not meet Study design these criteria were classified as either stroke-like events The D:A:D Study is an observational study of more than or rejected events: stroke-like events were those 49 000 HIV-positive patients from 11 cohorts from resembling a true stroke, but wherein there was a known
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Immune depression and risk of cardiovascular disease Sabin et al. 2737
nonatherosclerotic and nonhemorrhagic cause; rejected strokes and stroke-like events; C. confirmed strokes, events were those wherein the symptom duration was less stroke-like events, or rejected strokes. Finally, to than 24 h and/or wherein the neurological symptoms investigate whether the inclusion of nonstroke events were unspecified. Note that our definition of stroke does was likely to introduce bias, we investigated whether the not include transient ischemic attacks due to the shorter associations between the traditional stroke risk factors and length of symptom duration. each event differed in those with a high or low CD4þ cell count through the inclusion of interaction terms with Statistical methods each stroke risk factors and the latest CD4þ cell count Individuals were followed from study entry to the earliest (<200 or 200 cells/ml). of each endpoint, death, 1 February 2011, or 6 months after last clinic visit. Each individual’s follow-up was split into a series of consecutive 1-monthly periods and his/ Results her clinical, immunologic, and virologic status at the start of each period was established. Each individual’s latest/ By 1 February 2011, the individuals in the D:A:D Study nadir CD4þ cell count at the start of each month was had experienced 716 MI (over 223 242 person-years), categorized (<100; 100–199; 200–299; 300–399; 400– 1056 CHD (222 290 person-years), 303 confirmed stroke 499; 500 cells/ml) as was the duration of immune (224 581 person-years), and 1284 CVD (221 505 person- depression (none; 2; >2, 4; >4, 6; >6, 8; >8, years) events. Event rates (95% confidence interval, CI) 10; >10 years) and event rates were calculated for were 3.2 (3.0–3.4), 4.8 (4.5–5.0), 1.4 (1.2–1.5), and 5.8 each stratum. (5.5–6.1)/1000 person-years, respectively. The charac- teristics of patients at the time of experiencing each event Analyses were performed using Poisson regression with the are shown in Table 1. latest CD4þ, nadir CD4þ, and duration of immune depression included as time-updated covariates. Models In unadjusted analyses, individuals with a lower latest initially considered each measure of immune depression CD4þ cell count generally experienced higher rates of all separately. As the latest and nadir CD4þ cell counts would endpoints (Fig. 1(a)). After controlling for potential not be expected to differ substantially in untreated confounders, however, the adjusted associations did not individuals, analyses that included all three measures were support a linear association with the latest CD4þ cell considered only in patients who had been exposed to count for the MI/CHD endpoints (Table 2a). Stroke antiretroviral treatment. Multivariable models included rates, however, remained substantially higher in those adjustment for potential confounders. Because of the small with a lower latest CD4þ cell count; although there was a number of strokes, analyses of this endpoint only included weak association with the latest CD4þ cell count for the adjustment for key stroke risk factors (sex, age, previous CVD outcome, this appeared to be driven by the strong CVD, BMI, smoking status, and hypertension); analyses of association with stroke. Associations with the nadir CD4þ the other endpoints additionally included adjustment for cell count (Fig. 1(b), Table 2a) were broadly similar. All clinical cohort, mode of HIVacquisition, ethnicity, family events occurred less frequently in individuals who had historyof CVD,calendar year, cumulative exposure to each never experienced immune depression (Fig. 1(c)), specific antiretroviral drug, and recent exposure to each although evidence for a strong linear association between nucleoside reverse transcriptase inhibitor drugs. each event and duration of immune depression after adjustment was weak (Table 2a). We then investigated the apparent associations between the latest CD4þ cell count and strokes further. First, as Among patients exposed to antiretroviral therapy, the MI ongoing CMV infection has been reported to be and CHD outcomes continued to show only weak, at associated with an increased risk of CVD in the general most, linear associations with any of the measures of population [18], we investigated a possible modifying immune depression. Linear trends were apparent between effect of the development of a CMVAIDS-defining event all three markers and the stroke outcome (Table 2b). (as a time-updated covariate), as a surrogate for recent CMV infection, on the reported associations with In our sensitivity analyses, a prior CMV AIDS-defining immune depression. Second, to investigate whether event and a prior non-CMV, AIDS-defining opportu- any association with a low CD4þ cell count and/or CMV nistic infection were both associated with an increased could be explained by bias due to increased monitoring stroke rate [unadjusted relative rates of 1.75 (0.96–3.19) frequency in those with an opportunistic infection, and 1.79 (1.40–2.28), respectively]. However, adjust- sensitivity analyses also considered adjustment for the ment for these events did not modify the association development of other, non-CMV, AIDS-defining between the latest or nadir CD4þ cell count and the risk opportunistic infections. Third, to investigate whether of stroke (data not shown). the analyses of stroke were robust to the type of endpoint included in the analyses, we compared associations using In addition to the 303 confirmed strokes, 46 patients had a three endpoints: A. confirmed strokes only; B. confirmed stroke-like event and 38 patients had a rejected stroke
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 2738 AIDS 2013, Vol 27 No 17 stroke Rejected event Stroke-like ion were assumed not to have hypertension. stroke Confirmed ranscriptase inhibitors; PIs, protease inhibitors. Invasive procedure 2.3 mmol/l, or receipt of lipid-lowering medication; individuals in whom no Characteristics of patients at the time of each event death Sudden 0.9 mmol/l, triglyceride MI All patients at D:A:D entry 3053 (9.2) 330 (46.1) 118 (39.1) 362 (57.6) 136 (44.9) 11 (23.9) 10 (26.3) 12788 (38.4) 538 (75.1) 203 (67.2) 507 (80.7) 191 (63.0) 23 (60.9) 20 (52.6) 2630 21345 (64.1) 4124 (12.4) 366 (51.1) 86 (12.0) 119 (39.4) 16 (5.3) 354 (56.4) 85 169 (13.5) (55.8) 30 22 (9.9) (47.8) 16 (42.1) 7 (15.2) 7 (18.4) (%) exposed(%) exposed(%) exposed 23983 (72.0) 19325 (58.0) 704 (98.3) 11076 (33.3) 639 (89.3) 292 (96.7) 494 (69.0) 261 (86.4) 614 (97.8) 198 (65.6) 563 (89.7) 286 (95.4) 459 (73.1) 248 (81.9) 185 40 (61.1) (87.0) 34 (73.9) 28 35 (60.9) (92.1) 27 (71.1) 23 (60.5) 1818, 26, 30 1003 (3.0) 1465 (4.4) 21 (2.9) 38 (5.3) 13 (4.3) 20 (6.6) 17 (2.7) 24 (3.8) 16 (5.3) 16 (5.3) 2 (4.4) 2 (4.4) 4 (10.5) 2 (5.3) IDUHeterosexualOther/not known 10023 2839 (30.1) (8.5) 6017 (18.1) 152 (21.2) 43 (6.0) 114 (15.9) 58 (19.2) 31 (10.3) 68 (22.5) 144 (22.9) 43 (6.9) 70 (11.2) 80 (26.4) 35 (11.6) 48 (15.8) 8 (17.4) 3 (6.5) 9 (19.6) 11 (29.0) 10 (26.3) 2 (5.3) < > > Unknown 5364 (16.1) 205 (28.6) 134 (44.4) 148 (23.6) 72 (23.8) 13 (28.3) 9 (23.7) NonwhiteNot knownEx-smokerNever smokerUnknown 4603 10835 (13.8) (32.5) 8764 6949 (26.3) (20.9) 266 41 (37.2) (5.7) 6313 (19.0) 191 (26.7) 80 (11.2) 114 (37.8) 21 (7.0) 87 (12.2) 103 (34.1) 31 225 (10.3) (35.8) 59 20 (19.5) (3.2) 168 (26.8) 102 121 (16.2) (39.9) 49 (7.8) 74 30 (24.4) (9.9) 65 16 (21.5) (34.8) 45 13 (14.9) (28.3) 14 11 3 (30.4) (29.0) (6.5) 5 4 (10.9) (10.5) 9 (23.7) 6 (15.8) 16 (42.1) n Median (range) yearsn Median (range) yearsn 3.0 (0.0, 15.9)Median (range) years 2.3 (0.0, 6.8 13.9) (0.0, 20.5) 0.9 (0.0, 4.2 14.7) (0.0, 6.4 13.3) (0.1, 20.1) 2.5 (0.1, 3.8 13.0) (0.1, 7.3 12.9) (0.0, 22.4) 2.5 (0.1, 4.6 10.9) (0.0, 6.4 13.3) (0.1, 18.6) 2.7 (0.0, 3.7 11.9) (0.1. 5.8 13.1) (0.0, 11.2) 2.4 (0.0, 3.2 10.7) (0.0, 5.4 8.8) (0.0, 16.7) 2.2 (0.0, 7.8) 3.8 (0.0, 12.1) 1.6 (0.0, 9.4) 6.2 mmol/l, high-density lipoprotein (HDL) cholesterol (%) 10783 (33.6) 424 (59.3) 161 (53.5) 408 (65.0) 187 (62.1) 17 (37.8) 14 (37.8) n (%) 2527 (7.6) 108 (15.1) 40 (13.3) 125 (19.9) 29 (9.6) 2 (4.4) 2 (5.3) l) n m (%) 962 (2.9) 103 (14.4) 47 (15.6) 106 (16.9) 42 (13.9) 5 (10.9) 3 (7.9) (%) 5058 (15.2) 71 (9.9) 34 (11.3) 40 (6.4) 27 (8.9) 7 (15.2) 2 (5.3) (%) 7736 (23.2) 278 (38.8) 123 (40.7) 226 (36.0) 117 (38.6) 26 (56.5) 17 (44.7) (%) 1572 (4.7) 30 (4.2) 14 (4.6) 22 (3.5) 16 (5.3) 0 (–) 0 (–) b a n n n n (%) MSM 14422 (43.3) 407 (56.8) 145 (48.0) 371 (59.1) 140 (46.2) 26 (56.5) 15 (39.5) (%) (%) (%) n n n (%) 1250 (3.8) 339 (47.4) 118 (39.1) 309 (49.2) 62 (20.5) 6 (13.0) 5 (13.2) n (%) 24671 (74.1) 652 (91.1) 259 (85.8) 578 (92.0) 251 (82.8) 40 (87.0) 28 (73.7) (%) White 17863 (53.6) 409 (57.1) 167 (55.3) 383 (61.0) 152 (50.2) 27 (58.7) 21 (55.3) 50 copies/ml, (%) Current 11275 (33.9) 358 (50.0) 109 (36.1) 309 (49.2) 119 (39.2) 14 (30.4) 9 (23.7) n ), 2 n n n cell count (cells/ þ Defined as reported in Methods section; individuals in whom no information was available on blood pressure measurements or antihypertensive medicat Defined as total cholesterol Table 1. Characteristics of all patients at entry in the D:A:D Study, as well as at the time of experiencing each event. Number of patientsMale sex, Age (years)HIV exposure, Ethnicity, Median (range) 38 (3, 92) 33301 49 (24, 92) 716 52 (24, 93) 52 (29, 90) 302 53 (25, 91) 51 628 (30, 74)a 55 (24, 81) b 303 46 38 HCV coinfection, LatestNadirHIV RNA Median (range) Median (range) 408 (0, 2670) 211 (0, 2580) 441 (1, 1814) 125 (0, 1020) 360 (1, 2484) 122 (0, 950) 462 (2, 1884) 136 (0, 1464) 431 (3, 2484) 132 (0, 1464) 274 (0, 1010) 101 (0, 294) 273 (5, 1120) 90 (0, 727) Hypertension, information was available on lipid measurements or use of lipid-lowering drugs were assumed not to have dyslipidemia. Smoking, BMI (kg/m Diabetes mellitus, Prior AIDS event, HBV coinfection, Prior CVD, Family history of CVD, Dyslipidemia, Use of ART NRTIs PIs NNRTIs ART, antiretroviral therapy; CVD, cardiovascular disease; NNRTIs, nonnucleoside reverse transcriptase inhibitors; NRTIs, nucleoside reverse t CD4
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Immune depression and risk of cardiovascular disease Sabin et al. 2739
(a) HIV through the MSM route, more likely to have Latest CD4+ count unknown ethnicity, more likely to be a current smoker or 25 2 MI ex-smoker, and to have normal ( 18, 26 kg/m )BMI. CHD They were also more likely to exhibit traditional CVD risk 20 Stroke factors, including a prior personal or family history of CVD CVD, dyslipidemia, or hypertension. CD4þ cell counts 15 (current and nadir) were higher in those with confirmed
10 strokes and patients were more likely to be on antiretroviral therapy with a suppressed viral load. In contrast, differences
5 between patients with confirmed strokes and stroke-like
Rate (/1000 person-years) events were less marked, although there were differences in 0 gender, mode of HIV acquisition, ethnicity, and current <100 100–199 200–299 300–399 400–499 ≥500 smoking. Latest CD4+ count (cells/µl) þ (b) Associations between the latest CD4 cell count, Nadir CD4+ count established stroke risk factors, and the three stroke 25 þ MI endpoints are shown in Table 3. The CD4 cell count CHD gradient appeared to strengthen with the inclusion of 20 Stroke stroke-like and rejected stroke events. Similarly, the CVD association with a low BMI also appeared to increase after 15 inclusion of stroke-like and rejected stroke events.
10 In contrast, associations with current smoking and hypertension were both attenuated as the endpoint 5 was broadened to include the stroke-like and rejected
Rate (/1000 person-years) events. 0 <100 100–199 200–299 300–399 400–499 ≥500 When only confirmed strokes were included in the + µ Nadir CD4 count (cells/ l) endpoint, there was no evidence that any of the (c) associations with stroke risk factors differed significantly Duration of immune suppression between those with low or high CD4þ cell counts 25 MI (Table 4). However, when stroke-like events were CHD included, there was some evidence that the impact of 20 Stroke hypertension (P value for interaction: 0.01) and a previous CVD 15 CVD event (P ¼ 0.08) were both reduced in those with low CD4þ cell counts. When the endpoint was 10 expanded to additionally include rejected events, the prognostic values of age (P value for interaction: 0.08), 5 a previous CVD event (P ¼ 0.05), and hypertension Rate (/1000 person-years) þ 0 (P ¼ 0.002) were all reduced in those with low CD4 None ≤2 >2,≤4 >4,≤6 >6,≤8 >8,≤10 >10 cell counts. Duration of immunosuppression (years)
Fig. 1. Crude event rates (/1000 person-years of follow-up) for myocardial infarction, coronary heart disease, stroke, R and cardiovascular disease stratified by the (a) latest CD4 Discussion R cell count, (b) nadir CD4 count, and (c) duration of immune þ depression. CHD, coronary heart disease; CVD, cardiovas- Although it appears that individuals with a low CD4 cell cular disease; MI, myocardial infarction. count were more likely to experience a new MI or CHD event over follow-up, this association is largely explained by the higher prevalence of CVD risk factors in these individuals. In contrast, stroke events appear to occur event. Three-hundred and forty-two patients met the more commonly in those with low CD4þ cell counts, criteria for endpoint B [event rate: 1.52 (1.37–1.68)/ and it is this association that appears to drive a weak 1000 person-years], and 366 for endpoint C [1.63 (1.46– association with the CVD endpoint. However, the 1.80)/1000 person-years]. Established stroke risk factors reduced association of events occurring at low CD4þ cell at the time of an event among individuals experiencing counts with the traditional stroke risk factors suggests each event are shown in Table 1. Compared with those that, even with extremely careful central review with rejected stroke events, patients with confirmed stroke and endpoint validation, a proportion of strokes that were more likely to be men, more likely to be infected with occur in individuals with low CD4þ cell counts may be
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 2740 AIDS 2013, Vol 27 No 17
Table 2. Association between various markers of immune depression and myocardial infarction, coronary heart disease, stroke, and cardiovascular disease events.
Type of event
MI CHD Stroke CVD
(a) Unadjusted for each other marker of immune depression Latest CD4þ cell count (cells/ml) <100 1.07 (0.71, 1.64) 0.96 (0.68, 1.37) 2.02 (1.20, 3.38) 1.19 (0.89, 1.60) 100–199 0.98 (0.71, 1.36) 1.00 (0.78, 1.30) 1.49 (0.96, 2.31) 1.04 (0.83, 1.32) 200–299 Ref. Ref. Ref. Ref. 300–399 1.03 (0.79, 1.34) 0.91 (0.73, 1.13) 0.65 (0.42, 1.00) 0.89 (0.73, 1.08) 400–499 0.88 (0.66, 1.16) 0.80 (0.64, 1.01) 0.74 (0.74, 1.13) 0.84 (0.68, 1.03) 500 0.90 (0.71, 1.14) 0.83 (0.68, 1.00) 0.75 (0.53, 1.06) 0.84 (0.71, 1.00) Nadir CD4þ count (cells/ml) <100 1.15 (0.93, 1.43) 1.13 (0.94, 1.34) 1.10 (0.81, 1.50) 1.09 (0.93, 1.28) 100–199 0.96 (0.76, 1.21) 0.96 (0.80, 1.16) 1.09 (0.79, 1.52) 0.99 (0.83, 1.17) 200–299 Ref. Ref. Ref. Ref. 300–399 0.79 (0.57, 1.09) 0.81 (0.63, 1.05) 0.63 (0.38, 1.04) 0.77 (0.61, 0.98) 400–499 0.62 (0.38, 1.01) 0.61 (0.41, 0.92) 0.51 (0.25, 1.08) 0.62 (0.43, 0.88) 500 1.11 (0.75, 1.63) 1.20 (0.88, 1.64) 0.58 (0.29, 1.17) 1.01 (0.75, 1.36) Duration of immune depression (years) None Ref. Ref. Ref. Ref. 2 1.09 (0.91, 1.31) 1.06 (0.91, 1.24) 1.27 (0.98, 1.66) 1.09 (0.95, 1.25) >2, 4 1.33 (1.04, 1.70) 1.29 (1.05, 1.58) 1.44 (1.00, 2.07) 1.29 (1.07, 1.55) >4, 6 1.33 (0.98, 1.81) 1.34 (1.04, 1.73) 1.16 (0.70, 1.92) 1.25 (0.99, 1.58) >6, 8 0.86 (0.54, 1.35) 1.04 (0.74, 1.48) 1.23 (0.64, 2.37) 1.01 (0.73, 1.39) >8, 10 1.61 (1.01, 2.57) 1.27 (0.84, 1.93) 0.91 (0.34, 2.49) 1.20 (0.81, 1.78) >10 1.13 (0.61, 2.13) 1.38 (0.87, 2.19) 2.50 (1.29, 4.82) 1.64 (1.11, 2.42) (b) Mutually adjusted for each other marker of immune depression (patients exposed to antiretroviral therapy only) Latest CD4þ cell count (cells/ml) <100 0.96 (0.62, 1.50) 0.89 (0.30, 2.63) 2.26 (1.29, 3.94) 1.14 (0.84, 1.56) 100–199 0.95 (0.68, 1.33) 0.99 (0.76, 1.29) 1.63 (1.03, 2.59) 1.04 (0.82, 1.33) 200–299 Ref. Ref. Ref. Ref. 300–399 1.09 (0.83, 1.43) 0.93 (0.75, 1.17) 0.69 (0.44, 1.09) 0.92 (0.75, 1.13) 400–499 0.96 (0.72, 1.29) 0.85 (0.67, 1.07) 0.80 (0.52, 1.25) 0.89 (0.72, 1.10) 500 1.01 (0.78, 1.32) 0.88 (0.71, 1.09) 0.83 (0.56, 1.21) 0.91 (0.75, 1.10) Nadir CD4þ count (cells/ml) <100 1.36 (0.57, 3.23) 0.98 (0.45, 2.16) 1.36 (0.54, 3.43) 1.20 (0.65, 2.23) 100–199 1.17 (0.50, 2.74) 0.89 (0.41, 1.93) 1.54 (0.63, 3.79) 1.16 (0.64, 2.13) 200–299 Ref. Ref. Ref. Ref. 300–399 0.78 (0.56, 1.09) 0.86 (0.66, 1.12) 0.66 (0.39, 1.13) 0.81 (0.64, 1.04) 400–499 0.73 (0.45, 1.19) 0.73 (0.48, 1.10) 0.62 (0.28, 1.36) 0.73 (0.51, 1.06) 500 1.27 (0.85, 1.92) 1.34 (0.96, 1.89) 0.80 (0.38, 1.69) 1.18 (0.85, 1.63) Duration of immune depression (years) None Ref. Ref. Ref. Ref. 2 0.79 (0.34, 1.82) 1.05 (0.49, 2.26) 0.67 (0.28, 1.57) 0.83 (0.46, 1.49) >2, 4 0.92 (0.39, 2.16) 1.20 (0.55, 2.62) 0.69 (0.28, 1.69) 0.94 (0.51, 1.72) >4, 6 0.93 (0.39, 2.23) 1.23 (0.56, 2.72) 0.54 (0.21, 1.39) 0.90 (0.48, 1.67) >6, 8 0.60 (0.24, 1.52) 0.95 (0.42, 2.17) 0.52 (0.18, 1.47) 0.71 (0.37, 1.37) >8, 10 1.13 (0.44, 2.90) 1.14 (0.48, 2.69) 0.35 (0.10, 1.27) 0.83 (0.42, 1.68) >10 0.81 (0.29, 2.26) 1.24 (0.51, 2.99) 0.86 (0.30, 2.42) 1.10 (0.55, 2.21)
Results shown are relative rate and 95% confidence interval, and include adjustment for potential confounders as described in the Methods section. ART, antiretroviral therapy; CHD, coronary heart disease; CVD, cardiovascular disease; MI, myocardial infarction.
caused by HIV-associated CNS disorders and not by consider stroke as an outcome (or as part of a composite traditional atherosclerotic pathophysiological processes. outcome). Such misclassification will most likely occur at low CD4þ cell counts (when these other disorders are more Concern about the type of endpoint that is included prevalent) and will likely be greater in studies that do not within a broad definition of MI has been raised by Crane apply a similar level of critical review to reported et al. [19] who reported that of 271 definite/probable MIs endpoints. Although inclusion of these events into reported to five sites in the Centers for AIDS Research analyses may result in an increased number of events, (CFAR) Network of Integrated Clinical Systems, 127 and thus greater statistical power, their inclusion may also (47%) were secondary or type 2 MI events (i.e. they lead to misleading conclusions regarding the role of risk occurred due to myocardial supply/demand mismatch factors for stroke. These findings may at least partly in the setting of sepsis or hypoxia). Such events might explain some of the inconsistencies between studies that be expected to demonstrate weaker associations with
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. Immune depression and risk of cardiovascular disease Sabin et al. 2741
R Table 3. Association between established stroke risk factors, the latest CD4 cell count, and the three stroke endpoints.
Endpoint A Endpoint B Endpoint C
Relative rate Relative rate Relative rate (95% CI) P value (95% CI) P value (95% CI) P value
Age Per 5 years older 1.41 (1.35, 1.49) 0.0001 1.41 (1.35, 1.48) 0.0001 1.41 (1.35, 1.48) 0.0001 Male sex 1.19 (0.88, 1.62) 0.26 1.25 (0.93, 1.67) 0.13 1.21 (0.91, 1.60) 0.18 Smoking status Current smoker 1.52 (1.12, 2.06) 0.008 1.42 (1.07, 1.89) 0.02 1.38 (1.05, 1.81) 0.02 Ex-smoker 1.04 (0.74, 1.45) 0.82 1.01 (0.74, 1.38) 0.95 0.96 (0.71, 1.29) 0.78 Never smoker 1 – 1 – 1 – Unknown 1.59 (1.07, 2.37) 0.02 1.34 (0.92, 1.96) 0.12 1.44 (1.01, 2.05) 0.04 BMI (kg/m2) <18 2.06 (1.32, 3.24) 0.002 2.02 (1.32, 3.08) 0.0001 2.07 (1.39, 3.09) 0.0004 18, 261–1–1– >26, 30 0.77 (0.55, 1.08) 0.14 0.75 (0.54, 1.03) 0.08 0.79 (0.58, 1.07) 0.13 >30 1.02 (0.64, 1.62) 0.95 1.02 (0.66, 1.59) 0.92 1.00 (0.65, 1.55) 0.99 Unknown 0.68 (0.39, 1.19) 0.18 0.73 (0.44, 1.22) 0.23 0.71 (0.43, 1.17) 0.18 Previous CVD event 1.40 (1.02, 1.90) 0.03 1.35 (1.00, 1.81) 0.05 1.36 (1.02, 1.81) 0.04 Hypertension 2.14 (1.66, 2.75) 0.0001 1.96 (1.54, 2.49) 0.0001 1.91 (1.51, 2.41) 0.0001 Latest CD4þ cell count Per doubling 0.81 (0.74, 0.89) 0.0001 0.77 (0.71, 0.84) 0.0001 0.75 (0.70, 0.81) 0.0001
Three endpoints are as follows: A. confirmed strokes only; B. confirmed strokes þ stroke-like events; and C. confirmed strokes, stroke-like events, and nonstroke events. CI, confidence interval; CVD, cardiovascular disease.
traditional MI risk factors; patients with secondary MIs Associations were reported between the CVD rate and tended to have lower CD4þ cell counts than those with both the baseline and latest CD4þ cell count. As the primary MIs, suggesting that an association with a low authors found no association with antiretroviral treat- CD4þ cell count may be induced if such events are ment, analyses were not adjusted for treatment use. The included in endpoints. French APROCO-COPILOTE study reported a strong association between a CD4þ cell count less than There is increasing evidence that HIV, or its associated 200 cells/ml and a lower risk of a major coronary or inflammation and/or thrombosis, may lead to an arterial disease event (MI, stroke, coronary and peripheral increased risk of CVD [20–22]. However, findings from arterial disease, or cardiovascular surgery), although only the literature surrounding a possible association between 49 events occurred in 1154 participants [24]. cardiovascular/cerebrovascular events (CVEs) and CD4þ cell count are conflicting. Triant et al. [1] considered the In the Strategies for Management of Antiretroviral risk of acute MI (ICD-9-CM code 410.xx [23]) in HIV- Therapy (SMART) trial [7], no association was found diagnosed patients in two hospitals in Boston. Whereas between the baseline CD4þ cell count and the risk of a 41.4% of the 273 patients with acute MI had a latest major CVD event (death from CVD, clinical MI, silent CD4þ cell count less than 200 cells/ml, only 25.4% of MI, nonfatal stroke, or coronary artery disease requiring 6244 controls had a CD4þ cell count in the same range. surgery or invasive procedure), although only 72 events The association with the latest CD4þ cell count remained occurred in the trial, limiting statistical power. All trial after adjustment for several demographic and cardiovas- endpoints were captured using a detailed clinical case cular risk factors, although an association with the nadir definition, and were reviewed and classified centrally. CD4þ cell count was less supported. Of note, only 59% of Ford et al. [9] considered patients participating those with acute MI had a CD4þ cell count available for in National Institutes of Allergy and Infectious analysis in comparison to 70% of those without acute MI, Diseases (NIAID) clinical protocols; 52 patients were raising the possibility that some of the association may be a identified from electronic medical records as having consequence of unmeasured confounding. In a study CVD (acute MI, silent MI, coronary revascularization, from the French Hospitals Database on HIV [8], whereas acute coronary syndrome, cardiovascular accident, lower nadir CD4þ counts were slightly lower in individuals extremity revascularization, or sudden cardiovascular with an MI than in matched controls without an MI, death). There was no significant difference in the CD4þ cell counts at the event did not differ significantly. nadir CD4þ count between cases and matched controls, Although a standard definition of MI was used, events nor in the latest CD4þ cell count, either measured within were classified up to 6 years retrospectively and may, the 4 months prior to the event or 2 years prior to therefore, be subject to ascertainment bias. Lichtenstein the event, with traditional CVD risk factors being the et al. [3] followed patients attending clinics participating only factors that predicted an individual’s case/control in the HIV Outpatient Study (HOPS). Signs, symptoms, status. and diagnoses were extracted from medical charts and were reviewed centrally for quality. CVD was defined as Other related outcomes have also been considered, an MI, nonembolic/nonhemorrhagic stroke, coronary including studies of the metabolic syndrome [25–27] and artery disease, angina, or peripheral arterial disease. diabetes [28,29], few of which have reported strong
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited. 2742 AIDS 2013, Vol 27 No 17
M þ
3 associations with the CD4 cell count. There is some
value evidence from other studies of an association between a
P þ
l low CD4 cell count and coronary artery calcification m score [30], dyslipidemia [31], arterial stiffness [32], carotid intima–media thickness [33], brachial artery flow- mediated dilation [34], and carotid lesions [35], although 200 cells/