Telomere Length, Traditional Risk Factors, Factors Related to Human Immunodeficiency Virus (HIV) and Coronary Artery Disease Events in Swiss Persons Living With HIV

Abstract

Background

Leukocyte telomere length (TL) shortens with age and is associated with coronary artery disease (CAD) events in the general population. Persons living with human immunodeficiency virus (HIV; PLWH) may have accelerated atherosclerosis and shorter TL than the general population. It is unknown whether TL is associated with CAD in PLWH.

Methods

We measured TL by quantitative polymerase chain reaction (PCR) in white Swiss HIV Cohort Study participants. Cases had a first CAD event during 1 January 2000 to 31 December 2017. We matched 1–3 PLWH controls without CAD events on sex, age, and observation time. We obtained univariable and multivariable odds ratios (OR) for CAD from conditional logistic regression analyses.

Results

We included 333 cases (median age 54 years; 14% women; 83% with suppressed HIV RNA) and 745 controls. Median time (interquartile range) of TL measurement was 9.4 (5.9–13.8) years prior to CAD event. Compared to the 1st (shortest) TL quintile, participants in the 5th (longest) TL quintile had univariable and multivariable CAD event OR = 0.56 (95% confidence interval [CI], .35–.91) and OR = 0.54 (95% CI, .31–.96). Multivariable OR for current smoking was 1.93 (95% CI, 1.27–2.92), dyslipidemia OR = 1.92 (95% CI, 1.41–2.63), and for recent abacavir, cumulative lopinavir, indinavir, and darunavir exposure was OR = 1.82 (95% CI, 1.27–2.59), OR = 2.02 (95% CI, 1.34–3.04), OR = 3.42 (95% CI, 2.14–5.45), and OR = 1.66 (95% CI, 1.00–2.74), respectively. The TL-CAD association remained significant when adjusting only for Framingham risk score, when excluding TL outliers, and when adjusting for CMV-seropositivity, HCV-seropositivity, time spent with detectable HIV viremia, and injection drug use.

Conclusions

In PLWH, TL measured >9 years before, is independently associated with CAD events after adjusting for multiple traditional and HIV-related factors.

HIV infection, coronary artery disease, leucocyte telomere length, multivariable analysis, traditional risk factors

Persons living with human immunodeficiency virus (HIV; PLWH) may be at increased risk of aging-associated conditions compared to HIV-negative persons, including an approximately 2-fold higher incidence of coronary artery disease (CAD) [1]. There is considerable interest in early CAD prediction in PLWH by means of biomarkers [2], and coronary computed tomography (CT) angiography and calcium score [3]. Leucocyte telomere length (TL) shortens with age and may be a marker of cardiovascular aging [4]. An association between shorter TL and increased risk for CAD is now well recorded in the general population [5–8]. Although the relationship between TL and CAD is likely complex [9], genetic studies suggest a causal link [10–12], as do epidemiological studies documenting short TL many years before CAD events or carotid atherosclerosis [5, 8].

TL may be shorter in PLWH compared to HIV-negative persons [13, 14]. This has been linked to early TL shortening in the setting of HIV-seroconversion [15, 16], increased immune activation, even after suppressive antiretroviral therapy (ART) [14], and to the in vitro inhibition of telomerase activity (the key enzyme involved in TL maintenance) by certain antiretroviral agents [17]. Few longitudinal studies are available, which have suggested that TL may increase with effective HIV viral suppression after successful ART [18, 19].

The aim of the present study was to evaluate any independent association of TL with CAD events in PLWH, in the context of all relevant clinical risk factors, HIV-related factors, and adverse antiretroviral exposures.

METHODS

Study Population

Eligible participants included PLWH enrolled in the Swiss HIV Cohort Study (SHCS, http://www.shcs.ch) [20]. The study was approved by the respective local ethics committees. Participants provided written informed consent. Cases had a first CAD event and controls were CAD event-free during the study period (1 January 2000 to 31 December 2017). Because all study participants/samples will be analyzed in an upcoming genome-wide association study (GWAS), and because previous CAD-GWAS in the general population were conducted in populations of predominantly European descent [21], the study was restricted to participants of self-reported European descent.

CAD Events

CAD events were defined according to the Data Collection on Adverse events of Anti-HIV Drugs (D:A:D) study and the MONICA Project of the World Health Organization [22] and included definite myocardial infarction (MI); possible MI or unstable angina pectoris; percutaneous coronary intervention (coronary angioplasty/stenting); coronary artery bypass surgery; and fatal CAD, which required evidence of CAD before death. CAD events were validated by the treating HIV physician by chart review and D:A:D checking charts.

Case-control Matching

For each case, we aimed to select 3 SHCS controls who were CAD event-free at the CAD event date of the corresponding case (=matching date) using risk-set sampling [23]. Matching was done using incidence density sampling [24], that is, controls were matched on similar observation time duration, and their observation period was at similar calendar times. This was done in order to account for differences in ART exposures (with different potential effects on CAD risk [25, 26]) in use at different times and other differences during the observation period. Matching criteria included sex, age ±4 years, and date of SHCS registration ±4 years. Observation of cases was until the matching date and for controls was until the first regular SHCS follow-up examination after the CAD event date of the corresponding case, respectively.

CAD Risk Factors

Covariables were selected a priori, based on their published association with CAD, including smoking (current, past, never), age (per 1 year older), family history of CAD, diabetes mellitus, hypertension, and dyslipidemia (defined as previously published) [3]. HIV-related covariables included HIV viremia at the matching date (HIV RNA < or ≥50 copies/mL), CD4 nadir [27, 28] and ART exposures, defined a priori, based on their CAD-association in the D:A:D study, that is,current (last 6 months) exposure to abacavir (ABC) [25], and cumulative exposure (≥1 year) to lopinavir/ritonavir (LPV/r), indinavir (IDV), or darunavir (DRV) [26] until the matching date.

Telomere Length

We measured leucocyte telomere length (TL) in stored peripheral blood mononuclear cell (PBMCs) by quantitative polymerase chain reaction (PCR), using the single copy albumin gene as control, as previously reported [14] (see also Supplementary Methods). All participants had ≥1 sample available prior to the matching date. We aimed at measuring TL twice in each participant, that is, in the first available PBMC sample after SHCS enrollment, and in the last available sample before the matching date. This was done in order to test our hypothesis that the CAD association of short TL in PLWH is captured already in the first sample, with the last sample being unlikely to add any relevant information, as reported by others [29]. The corresponding sample of controls was obtained ±1.5 years of the event date of corresponding cases.

Power Calculation

With 255 cases and 2 controls per case we would be able to detect CAD event odds ratios of ≥1.6, at an alpha =0.05 and a power =80% [30]; with 341 cases we would be able to detect OR ≥ 1.5. The calculations assume a correlation of exposure between pairs in the case-control set of 0.2, as suggested [30] if the true correlation is not known.

Statistical Analyses

Univariable and multivariable conditional logistic regression analyses were used to estimate associations of the different quintiles of TL with CAD events for each of the case-control sets. Because age is a major contributor to CAD [3], and in order to detect any residual effect of suboptimal matching, age was included in the multivariable statistical models. We checked for interaction between TL and age using a likelihood-ratio test to evaluate any potential effect modification of TL by age. Other variables included the traditional CAD risk factors and HIV-related factors including ART exposures described above. We used Stata/SE 16.0 (StataCorp, College Station, TX, USA).

Sensitivity Analyses

We performed a number of sensitivity analyses to test the robustness of the TL-CAD association: Replacing all risk factors by the 10-year Framingham risk score (FRS) for CAD, or 10-year FRS risk category ≥10% [31]. Because of an association of cytomegalovirus (CMV) seropositivity with shorter TL [32], and because some studies suggest increased cardiovascular risk in PLWH who are hepatitis C (HCV) coinfected or who are injection drug users [33, 34], we performed sensitivity analyses including these variables in the model. Because controls had longer observation time with unsuppressed viremia than cases, we added this variable to the model in another sensitivity analysis. Finally, we excluded TL outliers, defined as TL > 1.5 times the interquartile range (IQR) above the 5th quintile or below the 1st quintile.

RESULTS

Participants, CAD Events

Analyses are based on 1078 SHCS participants, including 333 cases and 745 controls. Their characteristics are shown in Table 1 (13.8% women, median age at CAD event date, 54 years). Among the 333 cases, there were 176 MI, 128 percutaneous coronary interventions, 21 coronary artery bypass surgeries, and 8 fatal CAD events. Cases were older, more likely to be injection drug users or to be hepatitis C coinfected, were more likely to be current smokers, to have diabetes or dyslipidemia, had lower CD4 nadir, had longer ART exposure, similar median CD4 counts and less time with detectable viremia during the observation time, and were more likely to have been exposed to ABC, IDV, LPV/r, or DRV.

Table 1.

Characteristics of Cases and Controls

		Cases (n = 333)	Controls (n = 745)
Male sex, n (%)		287 (86.2)	641 (86.0)
Age (years), median (IQR)		54 (47–62)	53 (47–62)
HIV acquisition mode, n (%)
	Heterosexual	96 (28.8)	245 (32.9)
	MSM	158 (47.5)	369 (49.5)
	IDU	67 (20.1)	107 (14.4)
	Other	12 (3.6)	24 (3.2)
Smoking, n (%)
	Current	159 (47.8)	307 (41.2)
	Past	107 (32.1)	208 (27.9)
	Never	67 (20.1)	230 (30.9)
Cocaine use	Recentb	13 (3.9)	28 (3.8)
	Ever	27 (8.1)	59 (7.9)
Family history of CAD, n (%)		57 (17.1)	84 (11.3)
Diabetes mellitus, n (%)		56 (16.8)	49 (6.6)
Hypertension, n (%)		108 (32.4)	218 (29.3)
Dyslipidemia, n (%)		225 (67.6)	350 (47.0)
Framingham risk score (10-year risk), median (IQR)
	<10%	133 (39.9)	391 (52.5)
	10–20%	127 (39.6)	229 (31.1)
	>20%	61 (19.0)	117 (15.9)
On ART, n (%)		307 (92.2)	625 (83.9)
On ART, HIV RNA < 50 copies/mL (undetectable), n (%)		269 (80.8)	588 (78.9)
Total years on ART, median (IQR)		10.9 (6.8–15.8)	5.9 (2.3–10.6)
Duration of observationa (years), median (IQR)		11.7 (8–17.2)	11.2 (7.2–16.9)
Observation time spent with HIV RNA > 50 copies/mL, median % (IQR)		28.2 (10.3–51.9)	51 (28.5–77)
Currently on abacavir, n (%)		108 (32.4)	152 (20.4)
Lopinavir/ritonavir, exposure ≥1 year, n (%)		97 (29.1)	128 (17.2)
Indinavir, exposure ≥1 year, n (%)		76 (22.8)	58 (7.8)
Darunavir, exposure ≥1 year, n (%)		49 (14.7)	70 (9.4)
CD4 at matching date, median (IQR)		510 (353–728)	524 (379–681)
CD4 during observation time, median (IQR)		448 (321–615)	466 (358–575)
CD4 nadir (cells/μL), median (IQR)		150 (57–238)	209 (130–315)
CD4 nadir <50 cells/μL, n (%)		74 (22.2)	80 (10.7)
Previous AIDS, n (%)		103 (30.9)	150 (20.1)
Hepatitis C seropositivity, n (%)		86 (25.8)	148 (19.9)
CMV seropositivity, n (%)		291 (87.4)	609 (81.7)

All data shown apply to the matching date and are number (%) of participants, unless otherwise indicated.

Abbreviations: ART, antiretroviral therapy; CAD, coronary artery disease; CMV, cytomegalovirus; HIV, human immunodeficiency virus; IDU, intravenous drug use; IQR, interquartile range; MSM, men who have sex with men.

^aFrom registration in the SHCS until the matching date, and, for controls, until first regular, twice-yearly follow-up visit after the matching date.

^bIn 6 months prior to matching date.

Traditional- and HIV-related Risk Factors and Odds Ratio (OR) for CAD, Univariable Analysis

CAD was associated with current smoking, age, family history of CAD, diabetes mellitus, and dyslipidemia (Figure 1, Supplementary Table 1). Regarding HIV-associated factors, CAD was associated with current use of ABC, cumulative exposure to LPV, IDV, or DRV/r, and CD4 nadir but not with HIV viral load or CD4 count at the matching date. CAD was not associated with cocaine use (Supplementary Table 2).

Figure 1.

CAD odds ratio according to quintiles of telomere length, traditional and HIV-related risk factors. Uni- and multivariable conditional logistic regression of associations with CAD. Results involve 333 cases and 745 controls. Multivariable models are adjusted for all variables displayed, ie, for traditional and HIV-related risk factors. Abbreviations: CAD, coronary artery disease; CI, confidence interval; HIV, human immunodeficiency virus.

Telomere Length

All participants had ≥1 PBMC sample (all cases and 461 controls had both a first and a last sample) available for TL measurement available before the matching date. The median time (interquartile range [IQR]) from the first sample to the matching date was 9.4 (5.9–13.8) years and 9.4 (6.1–13.7) years, in cases and controls, respectively. Median time (IQR) from the last sample to matching date was 0.5 (0.3–0.8) years and 0.5 (0.2–0.9) years, in cases and controls, respectively. In the first sample, median (IQR) cross-sectional relative TL was 1.07 (0.82–1.28) and 1.10 (0.85–1.43) in cases and controls; in the last sample, median (IQR) TL was 0.94 (0.72–1.29) and 0.97 (0.76–1.28) in cases and controls, respectively (Figure 2). Annualized median percent (IQR) change of relative TL from first to last cell sample was similar in cases and controls (−0.83% [−3.69% to −3.43%], and −1.04% [−3.64% to 2.85%], respectively; P = .40).

Figure 2.

Relative TL in cases and controls in the first and last available samples. The graph shows smoothed TL values obtained from Epanechnikov kernel-weighted local polynomial regression. Abbreviation: TL, telomere length.

Telomere Length and Odds Ratio for CAD, Univariable Analysis

In the first sample, TL was associated with CAD odds ratio (OR) per unit longer OR = 0.65 (95% confidence interval [CI], .48–.88). In the last sample, TL was not associated with CAD, OR = 0.92 (95% CI, .72–1.18). TL decline from first to last sample was not associated with CAD, that is, CAD OR in the quintile with the least rapid versus most rapid annualized TL decline was 1.20 (0.77–1.87). Therefore, all subsequent analyses are based on TL in the first available sample.

Effect of Age

To account for any residual effect of suboptimal matching on age, we added age to the model. As expected, age (per year older) was associated with CAD, univariable OR = 1.23 (95% CI, 1.13–1.34). However, including age in a simple bivariable model (TL plus age) did not change the estimates: TL (per unit longer) remained independently associated with CAD, OR = 0.63 (95% CI, .46–.87). In addition, there was no evidence for an interaction between TL and age (likelihood-ratio test P = .73).

Telomere Length and Odds Ratio for CAD, Multivariable Analysis

In the final, multivariable model, TL was associated with CAD (Figure 1, Supplementary Table 1). Compared to participants in the 1st TL quintile (shortest TL), participants in the 2nd, 3rd, 4th quintile had CAD OR = 0.84 (95% CI, .49–1.43), 1.22 (.72–2.06), and 1.03 (.61–1.74), respectively; participants in the 5th TL quintile (longest TL) had CAD OR = 0.54 (95% CI, .31–.96). In comparison, CAD was associated with current smoking, OR = 1.93 (95% CI, 1.27–2.92), age (per year older), OR = 1.26 (95% CI, 1.14–1.39), diabetes mellitus, OR = 3.55 (95% CI, 2.11–5.97), and dyslipidemia, OR = 1.92 (95% CI, 1.41–2.63), but not with family history of CAD, OR = 1.26 (95% CI, .81–1.94), or hypertension, OR = 1.14 (95% CI, .81–1.63). CAD was associated with current use of ABC, OR = 1.82 (95% CI, 1.27–2.59), cumulative exposure ≥1 year to LPV/r, OR = 2.02 (95% CI, 1.34–3.04), IDV, OR = 3.42 (95% CI, 2.14–5.45), DRV, OR = 1.66 (95% CI, 1.00–2.74), but not with HIV RNA ≥ 50 copies/mL, OR = 0.92 (95% CI, .57–1.47), and there remained a trend for CD4 nadir, OR = 1.47 (95% CI, .95–2.27).

Sensitivity Analyses with Framingham Risk Score

As expected, FRS was associated with CAD in univariable analysis (Supplementary Tables 3and4): per percentage point FRS increase, OR = 1.04 (95% CI, 1.01–1.06). After adjustment for FRS, rather than for all variables shown in Figure 1, participants in the 5th TL quintile had CAD OR = 0.55 (95% CI, .34–.90) compared to the 1st TL quintile (Supplementary Table 3). Also, when considering FRS as a categorical variable, participants with FRS ≥ 10% had CAD OR = 1.62 (95% CI, 1.19–2.20), compared to participants with FRS < 10%. After adjustment for FRS category (≥10% vs <10%), rather than for all variables shown in Figure 1, participants in the 5th TL quintile had CAD OR = 0.55 (95% CI, .33–.90) compared to the 1st TL quintile (Supplementary Table 4).

Sensitivity Analysis Including CMV Seropositivity

In multivariable analysis including CMV seropositivity in the model, results remained essentially unchanged (Supplementary Table 5); participants in the 5th TL quintile had CAD OR = 0.55 (95% CI, .31–.97), compared to the 1st TL quintile.

Sensitivity Analysis Including Hepatitis C Seropositivity and Injection Drug Use

In multivariable analysis including HCV seropositivity and injection drug use in the model, results remained essentially unchanged (Supplementary Table 6); participants in the 5th TL quintile had CAD OR = 0.56 (95% CI, .33–1.00) compared to the 1st TL quintile.

Sensitivity Analysis After Including Time Spent With Detectable HIV RNA

Controls were matched on sex, age, and similar observation time/observation period, but not on ART duration (Supplementary Table 7). Compared to controls, cases had longer ART exposure (P < .01) and, accordingly, spent less observation time with detectable viremia (Table 1). As a consequence, time spent with detectable viremia was negatively associated with CAD, with univariable CAD OR = 0.97 (95% CI, .97–.98) and multivariable OR = 0.97 (95% CI, .96–.98) per additional percent. In multivariable analysis including time spent with detectable viremia, results for TL remained essentially unchanged; participants in the 5th TL quintile had CAD OR = 0.53 (95% CI, .29–.97), compared to the 1st TL quintile (Supplementary Table 7).

Sensitivity Analysis After Excluding TL Outliers

After excluding 59 participants with extreme TL values, results remained essentially unchanged (Supplementary Table 8). Compared to participants in the 1st TL quintile, participants in the 5th TL quintile had univariable CAD OR = 0.60 (95% CI, .36–.98), and multivariable-adjusted CAD OR = 0.53 (95% CI, .29–.96).

DISCUSSION

Our findings suggest that Swiss PLWH of European descent with the longest telomeres have approximately half the odds of a future CAD event, compared to those with the shortest telomeres. To our knowledge, this study is the first to associate TL with CAD events in PLWH. Although accelerated atherosclerosis and accelerated aging in PLWH remain unproven notions, we now provide evidence that short TL in PLWH may have relevant clinical implications, that is, increased cardiovascular risk [1, 13–17]. The association of TL with CAD events in our study is consistent with previous studies in the general population [5, 6, 8] and with the reference meta-analysis [7]. The TL-CAD association appears robust, because in multivariable analyses and in sensitivity analyses, TL remained independently associated with CAD events after adjusting for multiple traditional and HIV-associated risk factors including dyslipidemia, smoking, potentially adverse ART exposures, CMV status, HCV status, injection drug use, duration of time spent with detectable viremia, and CD4 nadir [25, 26].

The pathophysiological link between TL and CAD events remains incompletely understood. We exploited clinical, laboratory, and HIV-related data from >1000 PLWH prospectively followed at regular intervals in the well-established Swiss HIV Cohort Study. This allowed the consideration of all relevant risk factors and comorbidities associated with both TL and CAD events [9]. Some variables, for example, smoking [35, 36], may affect both CAD risk and TL, which might lead to spurious TL-CAD associations [9, 37]. However, a causal link between TL and CAD is now well accepted, based on genetic data [10–12], Mendelian randomization studies [38], and prospective documentation of short TL many years before CAD events or carotid atherosclerosis [5, 8]. Therefore, the independent TL-CAD association we show in PLWH seems to be consistent with findings in the general population that short TL may be a CAD-predisposing factor rather than simply a coincidental surrogate marker [32]. Additional strengths of our study include physician validation of all CAD events, rigorous case and control selection based on incidence density sampling to ensure similar observation duration during similar calendar periods [24], inclusion of only covariables with established CAD association, consistent results after excluding participants with the most extreme TL values, and evaluation of a potential TL-CAD association at 2 time points, including TL many years before each patient’s CAD event date, and consideration of the annualized rate of TL attrition.

The effect size of TL on CAD risk was similar in magnitude to well-established CAD risk factors. For illustrative purposes, Supplementary Figure 1 shows the same multivariable adjusted Figure 1, but with the 5th (longest) TL quintile as reference: The CAD association seen in PLWH in the 1st (shortest) TL quintile (OR = 1.92) was comparable to the effect of current smoking, dyslipidemia, or lopinavir exposure.

Our results are limited to individuals of European descent. Because of the relatively small number of women and persons >65 years included in our study, the results should be extrapolated to these populations with caution. A decreased OR for CAD was only detected for the 5th TL quintile, although no CAD association was observed for 2nd, 3rd, and 4th TL quintiles, likely because of limited statistical power [7]. Additional limitations include no differentiation of type 1 versus type 2 myocardial infarction in our study population, no information on diet or physical activity, which may affect TL [39], and limited numbers of participants treated with newer ART agents including integrase inhibitors and tenofovir alafenamide. An imbalance in observation time spent with detectable HIV viremia between cases and controls was addressed by including this variable in the model, with essentially unchanged results. Future investigation should include detailed characterization of any TL association with other aging-associated conditions in PLWH, and of CAD associations with genetic background, including gene variants associated with TL, metabolite profiles, and plasma biomarkers of inflammation [2].

Importantly, TL was associated with CAD in our participants when measured >9 years before the CAD event, suggesting that shorter TL predated and was not a consequence of CAD [5]. Around the CAD event date, however, TL was not associated with CAD risk. This remains unexplained but is consistent with previous reports: In the general population, TL shortening over time added no CAD-predictive value to a single cross-sectional TL measurement [29], and short TL but not the TL shortening rate over time was associated with carotid plaque [40]. Finally, the association of advancing age with TL was relatively weak in the reference TL-CAD meta-analysis [7]. In PLWH, TL measured several years before a first CAD event may robustly summarize the events affecting TL over an individual’s lifetime, most notably the significant TL shortening that may occur in the context of HIV seroconversion [15, 16]. Of note, no solid data suggest accelerated TL shortening in PLWH after HIV seroconversion. Indeed, recent longitudinal studies suggest TL increase with modern suppressive ART regimens [18, 19].

In conclusion, TL was independently associated with CAD events in PLWH in our study. The exact mechanisms of shorter TL in PLWH compared to the general population and the potential for accelerated TL shortening in PLWH remain incompletely understood. Our results suggest that short TL in PLWH matters, by documenting a relevant TL-CAD link in PLWH, and by putting the TL association into perspective with other relevant traditional and HIV-related risk factors. TL might therefore provide CAD risk information that current knowledge on CAD risk stratification in PLWH does not yet capture (eg, TL shortening around the time of HIV seroconversion). As a clinical consequence, documenting short TL in a given patient might further emphasize the need to optimize management of CAD risk factors, including dyslipidemia, diabetes, and selection of appropriate ART. The clinical value of TL testing, however, will rely on demonstration of improved CAD risk stratification in other populations of PLWH and in prospective studies. This was beyond the scope of the case-control design of our study.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions.Study design: T. E., M.R., N. A. K., P. R., J. F., R. D. K., H. F. G., B. L., P. E. T. Patient recruitment: B. H., C. H., K. W., J. A. R., E. B., K. E. A. D., A. C., H. F. G., P. E. T. Data acquisition: N. A. K., P. R., J. F., H. F. G., B. L., P. E. T. Data analysis: T. E., M. R., I. C. S., N. A. K., P. R., B. L., P. E. T. Drafting of the manuscript: T. E., I. C. S., P. E. T. Critical review and revision of the manuscript: all authors.

Acknowledgments.SHCS data are gathered by the Five Swiss University Hospitals, 2 Cantonal Hospitals, 15 affiliated hospitals, and 36 private physicians (listed inhttp://www.shcs.ch/180-health-care-providers). The authors acknowledge the effort and commitment of SHCS participants, investigators, study nurses, laboratory personnel, and administrative assistance by the SHCS coordination and data center.

Financial support.Swiss HIV Cohort Study (project 836), Swiss National Science Foundation(grant number 177499), Swiss HIV Cohort Study research foundation. The funders had no role in study design, study management, data collection, data analysis, data interpretation, and writing of the manuscript.

Potential conflicts of interest.P. R.’s institution, outside of the scope of the current study, has received independent scientific grant support from Gilead, Janssen, Merck, and ViiV, and P. R. has served on scientific advisory boards for Gilead, ViiV, Merck, Teva, for which honoraria were all paid to his institution. A. C. reports grants from AbbVie, MSD, Gilead, and ViiV, outside the submitted work. R. D. K. reports personal fees from Gilead, outside the submitted work. H. F. G., outside of this study, received unrestricted research grants from Gilead, Roche, and Yvonne Jacob Foundation; fees for data and safety monitoring board membership from Merck; consulting/advisory board membership fees from Merck, Teva, ViiV, Gilead, Sandoz, and Mepha. B. L. reports personal fees from Gilead and ViiV, outside the submitted work, and consultancy and statistical analyses fees from Kantonsspital Baselland. P. E. T.’s institution received grants and advisory fees from Gilead and ViiV. E. B. reports advisory board fees and travel grants paid to his institution from Gilead, MSD, ViiV, Abbott, Pfizer, and Sandoz, outside the submitted work. All other authors report no potential 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.

Author notes

T. E., M. R., and I. C. S. contributed equally to the manuscript.