Koblin, Beryl A. PhD*; Mayer, Kenneth H. MD†; Noonan, Elizabeth MS‡; Wang, Ching-Yun PhD‡; Marmor, Michael PhD§,‖; Sanchez, Jorge MD, MPH¶; Brown, Stephen J. MD#; Robertson, Michael N. MD**; Buchbinder, Susan P. MD††,‡‡,§§
*Laboratory of Infectious Disease Prevention, New York Blood Center, New York, NY
†The Fenway Institute, Fenway Health, Boston, MA
‡Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA
Departments of §Environmental Medicine and
‖Medicine, New York University School of Medicine, New York, NY
¶Asociacion Civil IMPACTA Salud y Educacion, Lima, Peru
#AIDS Research Alliance, Los Angeles, CA
**Merck Research Laboratories, North Wales, PA
††HIV Research Section, San Francisco Department of Public Health
Departments of ‡‡Medicine
§§Epidemiology, University of California San Francisco, San Francisco, CA.
Correspondence to: Beryl A. Koblin, PhD, Laboratory of Infectious Disease Prevention, Lindsley F. Kimball Research Institute, New York Blood Center, 310 East 67th Street, New York, NY 10065 (e-mail: email@example.com).
Supported by Merck Research Laboratories; the Division of AIDS, National Institute of Allergy and Infectious Diseases, in the US National Institutes of Health (NIH); the NIH-sponsored HIV Vaccine Trials Network (HVTN); and the Center for AIDS Research Center Grant number 1P01 AI057127 to the NYU School of Medicine from the National Institute of Allergy and Infectious Diseases, National Institutes of Health.
M. N. Robertson is a paid employee of Merck, owns Merck stock, and has Merck stock options. B. A. Koblin, K. H. Mayer, J. Sanchez, M. Marmor, S. J. Brown, and S. P. Buchbinder served as investigators on Merck-funded research. E. Noonan, and C. -Y. Wang have no conflicts of interest to disclose.
Authors' contribution: S. P. Buchbinder and M. N. Robertson participated in the design of the study. S. P. Buchbinder and M. N. Robertson co-chaired the study and oversaw study implementation. B. A. Koblin, K. H. Mayer, J. Sanchez, M. Marmor, S. J. Brown, and S. P. Buchbinder participated in the conduct of the study. E. Noonan and C. -Y. Wang analyzed the study data. B. A. Koblin and S. P. Buchbinder drafted the report, and all co-authors participated in revising the report.
Received November 1, 2011
Accepted February 27, 2012
The global HIV epidemic continues unabated.1–3 Interventions to reduce HIV infection rates are promising.4–8 However, topical and oral antiretrovirals for prevention may require daily adherence, a challenge for maintaining effectiveness. Male circumcision may be of limited benefit for female partners of HIV-infected men, men who have sex with men (MSM), and injection drug users. An effective HIV vaccine avoids some of these challenges and could provide synergy to existing strategies. The HIV vaccine field experienced a setback when the Step Study unexpectedly found that a replication-defective adenovirus type 5 (Ad5) vector vaccine with subtype B HIV-1 gag/pol/nef inserts seemed to increase the risk of HIV acquisition among MSM who were uncircumcised or had neutralizing antibodies to Ad5 at enrollment.9 Additional analyses indicate that the vaccine-associated increased risk among these men was highest shortly after vaccination and then decreased after 18 months.10
A number of biological mechanisms for the increased HIV risk among these vaccinees have been proposed, such as increased trafficking of vector-specific T-cell lymphocytes to mucosal sites or boosting of CD4-positive T cells or Ad5-specific T cells.11–16 Alternatively, the findings may be explained by differences in HIV exposure between vaccine and placebo recipients, particularly among men who were uncircumcised or had prior Ad5 immunity. Randomization is designed to eliminate imbalances in baseline characteristics by treatment arms,17 and random allocation to treatment arms in the Step Study was stratified by prior Ad5 immunity but not by circumcision status.9 Moreover, randomization does not preclude postrandomization differences in risk characteristics. Blinding, or keeping participants unaware of their treatment assignment, aims to limit exposure differences by treatment arm.18 Unblinding of treatment assignment or perceptions of treatment assignment while blinded may contribute to changes in risk behaviors, leading to changes in exposure to HIV.19
We investigated whether differences in HIV exposure, as measured by reported sexual risk behaviors, and in perceived treatment assignment may have contributed to the increased risk of HIV acquisition observed among vaccine compared with placebo recipients in the Step Study. We also determined if vaccine treatment assignment remained significantly associated with an increased risk of HIV infection, after controlling for sexual risk behaviors and other potential confounders. We also report on the role of insertive anal intercourse in the risk of HIV infection among uncircumcised men.
The study design, procedures, and primary results of the trial are previously published.9
Study Design and Population
This double-blind, phase II test-of-concept trial randomized 3000 HIV-1–negative participants aged 18-45 years in North America, Caribbean, South America, and Australia to receive 3 injections of MRKAd5 HIV-1 gag/pol/nef vaccine (n = 1494) or placebo (n = 1506). Men were eligible if they reported unprotected anal intercourse with a male partner or anal intercourse with 2 or more male partners and comprise this analysis sample.9
Briefly, vaccine or placebo, in a 1:1 ratio, was given by injection at enrollment, week 4, and week 26. Randomization was prestratified by study site, sex, and baseline Ad5 antibody titer but not by circumcision status. Participants were considered to have prior Ad5 immunity at baseline if their Ad5 neutralizing titer was >18. In September 2007, the trial met prespecified futility boundaries at the first interim analysis, and all immunizations were stopped.9
Risk behaviors in the previous 6 months were assessed by a standardized interviewer-administered questionnaire at screening and every 26 weeks. For sexual risk behaviors, participants were asked about the total number of male and female sexual partners and how many they knew were definitely HIV infected, definitely not HIV infected, or not absolutely sure whether they were HIV infected. Participants then were asked about occurrence of insertive and receptive anal intercourse (protected and unprotected) by serostatus of the male partners. These variables were used to construct variables of unprotected receptive/insertive anal intercourse with positive or unknown status partners. Participants also were asked about exchanging (giving and receiving) sex for money, drugs, or services and frequency of crack cocaine, amyl nitrite, and amphetamine use from “never” to “daily” on a 7-point response scale. Finally, evidence of sexually transmitted infection was recorded, including chlamydia or nonspecific urethritis, gonorrhea, syphilis, genital sores or ulcers, or abnormal penile discharge.
Circumcision status was self-reported at the screening visit. Of 1764 MSM enrolled, 1148 (67.1%) rolled over to a follow-up protocol at the end of the Step Study and had their circumcision status assessed by physical examination. For 4 men, discrepancies between self-report and physical examination were found, and the examination data were used.
Perceived treatment assignment was determined by asking the participant whether they thought they received vaccine, placebo, or they were not sure. This question was asked at 1 study visit starting in mid 2007. Study sites implemented the question at different times, and not all participants had the opportunity to answer the question. Thus, data for this variable were available for 1370 (77.7%) of the men, and the percent of participants with complete data varied by site.
HIV-1 infection status was determined at enrollment and at weeks 12, 30, and 52 and every 26 weeks thereafter. Specimens were screened with an immunoassay (Uni-Gold Recombigen HIV test; Trinity BioTech, Jamestown, NY; or the Multispot HIV-1/HIV-2 Rapid Test; Bio-Rad, Hercules, CA). Reactive tests were confirmed with an HIV-1 Western blot and HIV-1 plasma viral RNA assay (Amplicor Monitor version 1.5; Roche, Basel, Switzerland). A blinded end point adjudication committee made the final identification of HIV-1 infection status.
All participants provided written informed consent. Human experimentation guidelines of the United States Department of Health and Human Services and those of the authors' institutions were followed. The protocol was approved by the institutional or ethics review committee at every site and is registered with ClinicalTrials.gov (number NCT00095576).
The primary goal of these analyses was to determine whether differences in HIV exposure, as measured by self-reported sexual risk behaviors, may have contributed to the increased risk of HIV among MSM who were uncircumcised, had prior Ad5 immunity, or were both uncircumcised and had prior Ad5 immunity. Demographics and self-reported risk behaviors at baseline and follow-up were compared between vaccine and placebo recipients among MSM who were uncircumcised, had prior Ad5 immunity, or were both uncircumcised and had prior Ad5 immunity. Perceived treatment assignment among vaccine and placebo recipients was compared among the same 3 groups of MSM. We determined if vaccine treatment assignment remained significantly associated with an increased risk of HIV infection, after controlling for sexual risk behaviors, perceived treatment assignment, and other potential confounders. Finally, we examined the role of insertive anal intercourse in the risk of HIV infection among uncircumcised men.
All analyses were based on MSM who were HIV uninfected at study entry and who received at least 1 vaccination (modified intent-to-treat cohort). χ2 tests and Fisher exact test were used to statistically test differences in categorical variables between the groups. Wilcoxon rank sum tests were used to evaluate differences between the groups in the distribution of the number of sexual partners.
The generalized estimating equation approach for logistic regression was used to determine differences in risk behavior change by treatment arm. This method accounts for within-person variation over time and the correlation between the intervention-induced changes within an individual. The P values in the table are based on a test for differences between overall trends from baseline to 18 months by treatment arm for each outcome.
We used multivariate Cox models to estimate the vaccine effect on HIV incidence. The time-to-event variable for the Cox models was defined as the time from first vaccination to the midpoint between the date of the last HIV-1 seronegative visit and the date of the first evidence of HIV-1 infection. Participants who never showed any evidence of HIV-1 infection were right censored on the date of their last study visit before October 17, 2007. The covariates included were age (≤30 vs. >30), race (white vs. otherwise), region (North America, Australia, or otherwise), baseline Ad5 titer (≤18 vs. >18), perceived treatment (vaccine, placebo, not sure, missing), circumcision status, time-dependent drug use factors, and time-dependent sexual risk behaviors. For this analysis, time-dependent variables were utilized to allow for changes in risk behaviors over time. The potential confounders were all dichotomous for simplicity, stabilizing the model fitting, and reducing the modeling assumptions.
Cox models also were constructed among uncircumcised men to examine the role of insertive anal intercourse in the risk of HIV infection. In the first model, the baseline variable for unprotected insertive anal sex variable was included in the model to reflect the potential risk around the time of vaccine administration. In the second model, the time-dependent variable was included to allow for changes in the risk behavior over the entire time of follow-up.
ROLE OF FUNDING SOURCE
The sponsors of the study were involved in the study design, data collection, data analysis, data interpretation, writing of the report, and the decision to submit for publication. The sponsors had full access to all the data in the study and shared final responsibility for the decision to submit for publication.
From December 2004 through March 2007, 1764 MSM participants were randomized. The median age of the participants was 30 years. About half of the men (51.4%) were white and 29.0% were of multiracial identity, mostly from South America. More than half of the men (56.5%) were circumcised and 56.3% were Ad5 seropositive at baseline. At baseline, the median number of male sex partners in the prior 6 months was 6 (25% and 75th percentiles: 3, 15). About one-quarter (24.4%) of men reported having a known HIV-infected partner. Unprotected receptive and insertive anal intercourse in the previous 6 months was reported by 49.1% and 59.8% of men, respectively. Amyl nitrite (“poppers”) was used by 16.2% of men and amphetamines by 6.3%. Exchange of sex for money, drugs, or other services was reported by 16.7% of men and a recent sexually transmitted infection was reported by 12.3%. Among 1764 MSM, 726 were uncircumcised (vaccine, 360; placebo, 366), 994 had prior Ad5 immunity (vaccine, 497; placebo, 497), and 563 were both uncircumcised and had prior Ad5 immunity (vaccine, 276; placebo, 287).
Baseline and Follow-up Risk Behaviors Among Vaccine and Placebo Recipients
Demographics and self-reported risk behaviors at baseline were compared between vaccine and placebo recipients among uncircumcised men, those with prior Ad5 immunity, and uncircumcised men with prior Ad5 immunity (Table 1). Few significant differences were observed, and the differences were small in proportion. Among uncircumcised men, vaccine recipients were more likely to report unprotected insertive anal intercourse with HIV-negative partners (24.9% vs. 18.1%; P = 0.03) but were less likely to report genital sore or ulcers (0.3% vs. 2.2%; P = 0.04). Among uncircumcised men who had prior immunity to Ad5, vaccine recipients were more likely to report unprotected insertive anal intercourse with partners of unknown HIV status (46.0% vs. 37.8%; P = 0.05).
TABLE 1-a Comparison...Image Tools
Self-reported risk behaviors during follow-up were compared between vaccine and placebo recipients among uncircumcised men, those with prior Ad5 immunity, and uncircumcised men with prior Ad5 immunity (Table 2; Figs. 1A–C). Only 1 comparison reached statistical significance, and it was not in the direction to explain the increased risk of HIV infection among vaccine recipients. That is, among uncircumcised men who had prior Ad5 immunity, vaccine recipients had a greater decline in reporting the unprotected receptive anal intercourse with unknown status partners compared with placebo recipients.
Perceived Treatment Assignment Among Vaccine and Placebo Recipients
TABLE 2-b Comparison...Image Tools
Among uncircumcised men, 26.0% of vaccine recipients believed they had received vaccine compared with 19.3% of placebo recipients [odds ratio (OR) = 1.5; 95% confidence interval (CI), 1.0–2.2]. Similarly, among men with prior Ad5 immunity, 26.6% of vaccine recipients believed they had received vaccine compared with 19.3% of placebo recipients (OR = 1.5; 95% CI, 1.1–2.1). For men who were both uncircumcised men and had prior Ad5 immunity, 24.5% of men who received vaccine believed they got vaccine compared with 16.5% of those who received placebo (OR = 1.7; 95% CI, 1.0–2.6). The largest difference in perceived treatment between vaccine and placebo recipients was among circumcised men with a baseline Ad5 titer of ≤18, the group without evidence of an increased risk of HIV infection among vaccinees: 39.0% of men who received vaccine believed they got vaccine compared with 15.4% of men who received placebo (OR = 3.5; 95% CI, 2.3–5.4).
Multivariate Analysis of HIV Infection
In univariate analysis of HIV acquisition, the hazard ratio (HR) for treatment assignment was 1.5 (95% CI, 0.9–2.3) (Table 3). Other variables associated with HIV acquisition were geographic region, 20+ male sex partners, unprotected receptive or insertive anal intercourse with positive or unknown status partners, and the use of poppers or amphetamines. Controlling for geographic region, age, Ad5 titer, circumcision status, and perceived treatment, the HR for treatment assignment in the first multivariate model increased to 2.5 (95% CI, 1.1–5.7). Controlling for time-dependent sexual risk behaviors and drug use (multivariate model 2), vaccine recipients remained at higher risk of HIV infection compared with placebo recipients (HR = 2.8; 95% CI, 1.2–6.8). Including age and the number of partners as continuous variables did not change the results of the model. No significant interactions between any sexual risk behaviors and treatment assignment were found, suggesting that the risk of HIV infection among vaccine compared with placebo recipients did not differ between those reporting specific sexual risk behaviors and those who did not.
Additional analyses were conducted among uncircumcised men. If an increased risk of HIV infection due to the vaccine was related to being uncircumcised, the effect of the vaccine on HIV infection would likely have been higher among uncircumcised men reporting unprotected insertive anal sex around the time of vaccine administration compared with men not reporting unprotected insertive anal sex at that time. We found that the HR for HIV infection for vaccine compared with placebo recipients was 4.4 (95% CI, 1.2–15.3) among uncircumcised men reporting unprotected insertive anal sex at baseline, whereas the HR among uncircumcised men who reported no insertive anal sex at baseline was 1.2 (95% CI, 0.3–4.1). Taking into account time-dependent behaviors, the HR for HIV infection for vaccine compared with placebo recipients was 2.3 (95% CI, 0.6–8.8) among uncircumcised men reporting unprotected insertive anal sex, whereas the HR among uncircumcised men who reported no insertive anal sex was 2.4 (95% CI, 0.9–7.0).
This analysis focused on whether the increased risk of HIV infection observed among subgroups of men in the Step Study (uncircumcised men and men with prior Ad5 immunity who received vaccine) could be explained by any differences in the risk behaviors or perceptions of treatment assignment. Differences would suggest that exposure to HIV was not equal in the 2 arms of the study, potentially affecting HIV incidence rates. The differences in risk behaviors found in this analysis were few in number, and of those differences identified, the magnitude of differences was small.
We also assessed whether perception of vaccination status could explain any differences in HIV exposure. We found that uncircumcised men who had prior immunity to Ad5 and had received vaccine were 1.7 times more likely to think that they got vaccine than placebo recipients. If this group of men increased their risk due to a perception of protection from the vaccine, HIV infection rates could have been elevated in this group. However, we did not find any significant differences in risk behaviors between the vaccine and placebo recipients in this group. Furthermore, adjustment for these differences in Cox regression analysis did not eliminate the significant association of vaccine with an increased risk of HIV infection. In fact, it was the vaccine recipients who were circumcised and Ad5 negative at baseline who were over 3 times more likely to think they had gotten vaccine, an effect much stronger than among uncircumcised men. There is evidence that participants who do not have prior Ad5 immunity have greater reactogenicity to rAd5 vaccines.20 Thus, these men may not have actively unblinded themselves by HIV testing outside of the study but had greater systemic reactions, which led them to believe that they had gotten vaccine. This group, however, did not have higher HIV incidence rates than placebo controls.
We examined whether treatment assignment remained significantly associated with HIV infection controlling for circumcision status, Ad5 status, perceived treatment, and risk behaviors. The association between HIV infection and treatment assignment could not be explained by behavioral variables or other potential confounders. There were no significant interactions between behavior and treatment arm.
Finally, a biologic mechanism is suggested by our analysis of uncircumcised men, unprotected insertive anal sex, and risk of infection associated with vaccine. Men reporting unprotected insertive anal sex at baseline and thus close to the time of vaccine administration demonstrated an increased risk of infection associated with vaccine. This increased risk was not seen among men who did not report unprotected insertive anal sex, nor did we see such differences using a variable, which took into account the reporting of unprotected insertive anal sex over 18 months of the study (time-dependent variables). This finding is consistent with other analyses from the Step Study indicating that the increased risk of HIV infection associated with vaccine seemed to be highest around the time of vaccination and then decreased after 18 months.10 It is possible that any inflammation in the foreskin area related to the time of vaccine administration could have increased the risk of HIV acquisition.21
These data have limitations. Behaviors were collected in face-to-face interviews, and thus, some behaviors may have been underreported. However, because all study staff were blinded to treatment assignment, there is no reason to believe that the data collected would have been biased by treatment arm. Most of behavioral measures were limited to binary outcomes, and the number of episodes of risk behaviors was not collected. Finer gradations may have revealed more subtle effects of behavioral risk. Finally, perceived treatment assignment was not available for 22% of participants, although they were included in the analyses as “missing.”
In sum, these analyses do not support a behavioral explanation for the observed increased HIV-infection rates among uncircumcised men or men with prior Ad5 immunity in the Step Study. The results of the Step Study emphasize the critical need for effective and sensitive measures of behaviors and attention to perception of treatment assignment in biomedical trials, so unexpected findings such as those in the Step Study can be properly examined. Identifying biologic mechanisms to explain the increased HIV infection risk is a priority.
1. Hall HI, Song R, Rhodes P, et al.. Estimation of HIV incidence in the United States. JAMA. 2008;300:520–529.
2. Sanders EJ, Graham SM, Okuku HS, et al.. HIV-1 infection in high risk men who have sex with men in Mombasa, Kenya. AIDS. 2007;21:2513–2520.
3. van GF, de Lind van Wijngaarden JW. A review of the epidemiology of HIV infection and prevention responses among MSM in Asia. AIDS. 2010;24(suppl 3):S30–S40.
4. Abdool KQ, bdool Karim SS, Frohlich JA, et al.. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science. 2010;329:1168–1174.
5. Grant RM, Lama JR, Anderson PL, et al.. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med. 2010;363:2587–2599.
6. Templeton DJ. Male circumcision to reduce sexual transmission of HIV. Curr Opin HIV AIDS. 2010;5:344–349.
7. Rerks-Ngarm S, Pitisuttithum P, Nitayaphan S, et al.. Vaccination with ALVAC and AIDSVAX to prevent HIV-1 infection in Thailand. N Engl J Med. 2009;361:2209–2220.
8. Cohen MS, Chen YQ, McCauley M, et al.. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365:493–505.
9. Buchbinder SP, Mehrotra DV, Duerr A, et al.. Efficacy assessment of a cell-mediated immunity HIV-1 vaccine (the Step Study): a double-blind, randomised, placebo-controlled, test-of-concept trial. Lancet. 2008;372:1881–1893.
10. Duerr A, Huang Y, Buchbinder S, et al.. Extended follow-up confirms early vaccine-enhanced risk of HIV acquisition and demonstrates waning effect over time among participants in a randomized trial of recombinant adenovirus HIV vaccine (Step Study). J Infect Dis. 2012. In press.
11. Gray G, Buchbinder S, Duerr A. Overview of STEP and Phambili trial results: two phase IIb test-of-concept studies investigating the efficacy of MRK adenovirus type 5 gag/pol/nef subtype B HIV vaccine. Curr Opin HIV AIDS. 2010;5:357–361.
12. Masek-Hammerman K, Li H, Liu J, et al.. Mucosal trafficking of vector-specific CD4+ T lymphocytes following vaccination of rhesus monkeys with adenovirus serotype 5. J Virol. 2010;84:9810–9816.
13. Chakupurakal G, Onion D, Cobbold M, et al.. Adenovirus vector-specific T cells demonstrate a unique memory phenotype with high proliferative potential and coexpression of CCR5 and integrin alpha4beta7. AIDS. 2010;24:205–210.
14. Hutnick NA, Carnathan DG, Dubey SA, et al.. Baseline Ad5 serostatus does not predict Ad5 HIV vaccine-induced expansion of adenovirus-specific CD4+ T cells. Nat Med. 2009;15:876–878.
15. Benlahrech A, Harris J, Meiser A, et al.. Adenovirus vector vaccination induces expansion of memory CD4 T cells with a mucosal homing phenotype that are readily susceptible to HIV-1. Proc Natl Acad Sci U S A. 2009;106:19940–19945.
16. O'Brien KL, Liu J, King SL, et al.. Adenovirus-specific immunity after immunization with an Ad5 HIV-1 vaccine candidate in humans. Nat Med. 2009;15:873–875.
17. Schulz KF, Grimes DA. Generation of allocation sequences in randomised trials: chance, not choice. Lancet. 2002;359:515–519.
18. Schulz KF, Grimes DA. Blinding in randomised trials: hiding who got what. Lancet. 2002;359:696–700.
19. Bartholow BN, Buchbinder S, Celum C, et al.. HIV sexual risk behavior over 36 months of follow-up in the world's first HIV vaccine efficacy trial. J Acquir Immune Defic Syndr. 2005;39:90–101.
20. Priddy FH, Brown D, Kublin J, et al.. Safety and immunogenicity of a replication-incompetent adenovirus type 5 HIV-1 clade B gag/pol/nef vaccine in healthy adults. Clin Infect Dis. 2008;46:1769–1781.
21. Mayer KH, Venkatesh KK. Interactions of HIV, other sexually transmitted diseases, and genital tract inflammation facilitating local pathogen transmission and acquisition. Am J Reprod Immunol. 2011;65:308–316.
© 2012 Lippincott Williams & Wilkins, Inc.