Safety and Tolerability of Tenofovir for Preexposure Prophylaxis Among Men Who Have Sex With Men - Editorial
Download the PDF here|
Download the PDF here
"Seroconversions: Among all 400 participants, 7 seroconversions were documented. None occurred among participants taking TDF. Three occurred among participants taking placebo; 3 occurred among delayed arm participants who had not yet started drug. One occurred in a participant assigned to placebo who was HIV-1 antibody negative at screening and enrollment and then was seropositive at the 1-month visit. Confirmatory Western blot was also positive. Viral load performed on a stored specimen from the enrollment visit was 1770 copies per milliliter; thus, this participant was already infected at enrollment. No K65R mutations were noted among any seroconverting participants."
"In multivariable analyses, back pain was more likely among TDF recipients (P = 0.04); these reports were not associated with documented fractures or other objective findings. There were no grade ≥3 creatinine elevations; grades 1 and 2 creatinine increases were not associated with TDF receipt. Estimated percentage of study drug doses taken was 92% by pill count and 77% by Medication Event Monitoring System."
JAIDS Journal of Acquired Immune Deficiency Syndromes:
1 September 2013
Vermund, Sten H. MD, PhD
Vanderbilt Institute for Global Health and the Department of Pediatrics;
Vanderbilt University School of Medicine, Nashville, TN
Efforts to reduce HIV incidence among men who have sex with men (MSM) have been disappointing.1-18 The iPrEx clinical trial found that tenofovir disoproxil fumarate (TDF) and emtricitabine (FTC) prescribed as a single pill (Truvada) for daily use among MSM in North and South America were associated with a modest 44% decline in HIV seroincidence.19,20 However, the protection among men who actually adhered to the daily drug regimen was far higher, suggesting a new paradigm for HIV protection for motivated at-risk MSM.21 Safety is a paramount consideration for any prophylactic medical intervention, whether a diagnostic procedure, chemoprevention, or immunization. Despite the mixed efficacy results in clinical trials,19,22-28 the prospect of widespread and prolonged use of oral or topical (eg, microbicide) pre-exposure prophylaxis (PrEP) has generated an interest in further evaluation of safety, particularly of TDF, given U.S. Food and Drug Administration approval and Centers for Disease Control and Prevention and WHO guidelines for PrEP use by MSM using FTC/TDF.29-36
Grohskopf et al37 in this issue of JAIDS describe a large U.S. multicenter study of the clinical safety of daily TDF among HIV seronegative MSM, in anticipation of its use for PrEP. In this 4-arm randomized placebo-controlled trial, neither participants nor their evaluators knew whether they were assigned 300 mg TDF orally per day or placebo. In a clever approach to assessing whether taking the pills might be associated with changes in risk behavior over time, each of the TDF and placebo groups had an immediate or 9-month delayed dosing arm.
The authors are thorough in their evaluation of safety, and no events differed significantly between the 186 TDF and 187 placebo recipients (55% of the 679 MSM screened for the study). An impressive 87% (n = 325) of men completed the final study visit. Fully, 90% of MSM participants reported at least 1 adverse event, only 3% of whom were deemed severe. While a multivariable analysis found that back pain was significantly more likely among TDF recipients (P = 0.04), no objective evidence of back disorders could be found. Nor can we be sure that this finding is not spurious, a possible consequence of multiple comparisons of outcomes.
A principal finding is that no evidence of TDF-associated renal disease was seen, compared with the placebo group. Among 3 men who had either persistent elevation of serum creatinine ≥0.5 mg/dL over baseline or had confirmed grade 4 hypophosphatemia, all were assigned to placebo. Nonetheless, it is well known that TDF is associated with renal side effects when in widespread clinical use among HIV-infected persons.38,39 Whether the longer term and more widespread use of TDF for PrEP will result in more TDF-associate renal disease is not known.
It will be helpful in future subanalyses for the investigators to examine drug users apart from nonusers, particularly, in the context of adherence. Users of stimulants are at higher risk for seroconversion and also may be less adherent to PrEP. Alcohol use is of importance for any such study for the same reasons. Age may also interact with drug side effects in substantial ways.
A caveat in any PrEP study is the measure of adherence used, as well as measurement of risk behavior. Both reports are subject to social response bias and trial participants may have known that their reports could be important for policy making. Persons who know that they are not supposed to increase risk behavior may tell us investigators what we want to hear. The accuracy of self-reported adherence is poor, and use of the Medication Event Monitoring System (MEMS) bottle caps/microprocessors to estimate medication usage are perhaps just a small improvement. Far superior are assessments of plasma or tissue drug levels, and this would give a more definitive assessment of adherence.
Dual energy x-ray absorptiometry (DEXA) scans were done (at the San Francisco study site only) to assess for bone mineral density (BMD). DEXA was performed among both a subset of HIV-seronegative MSM who were screened but not enrolled (n = 210, none on TDF at the time) and those who subsequently enrolled in the prospective trial (n = 184 with ≥1 scan; 94 receiving TDF and 90 not receiving it). These data are not reported in the Grohskopf et al study because they were reported previously by Liu et al.40 At baseline, 20 participants (10%) had low BMD (Z score ≤ -2.0 at the L2-L4 spine, total hip, or femoral neck). In the clinical trial, the specifics of the BMD losses, as noted in the TDF vs. the placebo group or the group that had not yet received treatment at the femoral neck (as treated analysis), were: 1.1% decrease [95% confidence interval (CI): 0.4% to 1.9%] in mean BMD at the femoral neck; 0.8% decline at the total hip (95% CI: 0.3% to 1.3%); and 0.7% decline at the L2-L4 spine (95% CI: -0.1% to 1.5%). In the full 24-month follow-up, 13% of TDF recipients vs. 6% of other men experienced >5% BMD loss at the femoral neck (P = 0.13). Low BMD was associated strongly with drug use, either amphetamines [odds ratio (OR) = 5.9, 95% CI: 1.7 to 20.2 or inhalants (OR = 4.6, 95% CI: 1.3 to 15.8). In contrast, men who took multivitamins, calcium, or vitamin D were less likely to have low BMD at baseline (OR = 0.26, 95% CI: 0.10 to 0.71).40
The findings of Grohskopf et al37 are reassuring in that the risk of renal disease from TDF in seronegative MSM is not high in 2-years of follow-up, acknowledging that the sample size is only 187 men on TDF. That more than double the TDF recipients had a BMD loss of >5% at the femoral neck is a concern,40 although larger studies would be needed to confirm this trend. In summary, renal disease may not be as prominent a concern with PrEP as BMD loss is, although larger studies are needed to ensure that this is truly the case. The BMD data alone suggest that TDF-based PrEP is not benign. Lower dosing regimens, as with event-driven use (ie, not using PrEP when not sexually active), are being investigated in such studies as the HIV Prevention Trials Network (HPTN) 067 protocol (The ADAPT study), with HPTN 066 assessing pharmacokinetics of intermittent dosing approaches. New drugs that show promise for PrEP and that are not associated with BMD loss are also being studied. For example, HPTN 069 is assessing the safety and tolerability of maraviroc with and without FTC/TDF. Complementing the rectal microbicide studies of the Microbicides Trials Network, the HPTN focuses on oral PrEP development and testing.41-45
Assessing adherence was a major goal of the study reported by Grohskopf et al; much has been written about PrEP, which has good protective efficacy in men or women only with high adherence. The authors state that "daily oral TDF was well tolerated, with reasonable adherence." However, drug interruptions were frequent, with 178 temporary drug interruptions documented among the 373 participants [the authors do not make clear whether these are 178 persons with one (or more) interruptions each or whether this was an unspecified smaller number of persons with ≥1 interruption each]. The median length of TDF interruptions was 37 days, with a range of 2-428 days. Some men (17.6% of participants) stopped taking TDF altogether. For real world applications, a minimum estimate of those who might stop is necessary, especially under conditions in high-income countries where DEXA scanning might be provided to all.
The data on behavior change over time are not reported here by Grohskopf et al but have been published recently by Liu et al.46 But Grohskopf et al do report seroconversions during the study. Among all 400 enrolled participants including 27 persons who never were dispensed study drug or placebo, 7 seroconversions were noted, none among participants taking TDF. Viral load performed on a stored specimen from 1 man's enrollment visit showed him to be in the window period; he had 1770 viral copies per milliliter, but only seroconverted at a later date such that he was infected at enrollment.37 Data are not presented to calculate a person-time incidence rate nor do the authors speculate what the protective efficacy might have been. This is how it should be. The study was not designed as an efficacy trial, and to present it as such would do a disservice to the participants and to the science of the trial.
In summary, the work of the San Francisco, Boston, and Atlanta investigators suggests that renal disease will not be common, although some persons terminated their PrEP use once renal warning signs emerged. The BMD data have been presented earlier, but suggest the need for safer options to TDF. While the rate of adverse events for PrEP has been low, monitoring of drug side effects is needed in seronegative persons on PrEP much as it is needed in persons with HIV-infection being treated for their disease. This is suggested by the reasons for discontinuation of drug from this study of Grohskopf et al. Although side effects are uncommon, they could be important, and clinical follow-up is needed. How PrEP gets translated into real-world prevention practice is a challenge given the need for clinical services for screening, drug administration, and monitoring of toxicity.29-36 It remains an open question whether enough MSM will use PrEP, adhere to it, and tolerate it to make a difference in the high HIV incidence that beleaguers the MSM community worldwide.47-54 A new paradigm for men's health services may be necessary for integrated clinical and prevention services to truly make a difference in the global epidemic.54
JAIDS Journal of Acquired Immune Deficiency Syndromes:
1 September 2013
Randomized Trial of Clinical Safety of Daily Oral Tenofovir Disoproxil Fumarate Among HIV-Uninfected Men Who Have Sex With Men in the United States
Grohskopf, Lisa A. MD, MPH*; Chillag, Kata L. PhD*; Gvetadze, Roman MD, MSPH; Liu, Albert Y. MD, MPH; Thompson, Melanie MD; Mayer, Kenneth H. MD; Collins, Brandi M. MPH; Pathak, Sonal R. MPH; O'Hara, Brandon MSPH; Ackers, Marta L. MD, MPH*; Rose, Charles E. PhD; Grant, Robert M. MD, MPH#; Paxton, Lynn A. MD, MPH*; Buchbinder, Susan P. MD
*Epidemiology Branch;Quantitative Sciences and Data Management Branch, Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, GA;Bridge HIV, San Francisco Department of Public Health, San Francisco, CA;AIDS Research Consortium of Atlanta, Atlanta, GA;Fenway Health, Beth Israel Deaconess Medical center, Harvard Medical School, Boston, MA;Northrop Grumman Corporation, Falls Church, VA; and#Gladstone Institutes; Department of Medicine, University of California, San Francisco, CA.
Objectives: To evaluate the clinical safety of daily tenofovir disoproxil fumarate (TDF) among HIV-negative men who have sex with men.
Design: Randomized, double-blind, placebo-controlled trial. Participants were randomized 1:1:1:1 to immediate or delayed study drug (TDF, 300 mg orally per day, or placebo).
Methods: Four hundred healthy HIV-uninfected men who have sex with men reporting anal sex with another man within the previous 12 months enrolled in Atlanta, Boston, and San Francisco. HIV serostatus, clinical and laboratory adverse events (AEs), adherence (pill count, Medication Event Monitoring System, and self-report), and sexual and other sociobehavioral data were assessed at 3-month intervals for 24 months. Primary outcomes were clinical safety, assessed by incidence of AEs and laboratory abnormalities.
Results: Study drug was initiated by 373 (93%) participants (186 TDF and 187 placebo), of whom 325 (87%) completed the final study visit. Of 2428 AEs reported among 334 (90%) participants, 2366 (97%) were mild or moderate in severity. Frequencies of commonly reported AEs did not differ significantly between TDF and placebo arms. In multivariable analyses, back pain was more likely among TDF recipients (P = 0.04); these reports were not associated with documented fractures or other objective findings. There were no grade ≥3 creatinine elevations; grades 1 and 2 creatinine increases were not associated with TDF receipt. Estimated percentage of study drug doses taken was 92% by pill count and 77% by Medication Event Monitoring System. Seven seroconversions occurred: 4 on placebo and 3 among delayed arm participants not yet on study drug.
Conclusions: Daily oral TDF was well tolerated, with reasonable adherence. No significant renal concerns were identified.
Thirty years after the first published HIV/AIDS case,1 biomedical and behavioral interventions have not adequately controlled the pandemic. Globally, more than 33 millions are estimated to be infected. Although estimated new infections have declined annually since the late 1990s, incidence remains unacceptably high, with certain geographical areas and risk groups disproportionately affected.2 In the United States, an estimated 48,600 new infections occur annually, over half among men who have sex with men (MSM).3 A troubling increase in incidence has been noted among MSM, particularly young MSM of color.4
HIV antiretroviral preexposure prophylaxis (PrEP) may become an important prevention option for some at-risk MSM. Human PrEP trials were undertaken based on considerations of biological plausibility, experiences with antiretroviral use for prevention of perinatal transmission and PrEP, availability of well-tolerated antiretrovirals, and indications of efficacy from animal models.5-8 Most trials have evaluated tenofovir disoproxil fumarate (TDF, Viread) or the fixed-dose combination of TDF and emtricitabine (TDF/FTC, Truvada). The US Food and Drug Administration (FDA) recently approved the use of TDF/FTC as PrEP in July 2012.9 Efficacy of oral TDF/FTC PrEP was demonstrated among MSM in several Latin American countries, South Africa, Thailand, and the United States,10 although less than 10% of participants were enrolled at US sites. Efficacy of this regimen was also demonstrated in young heterosexual men and women in Botswana11 and heterosexual serodiscordant couples in Kenya and Uganda.12 However, 2 studies conducted among heterosexually active women in several African countries (the Pre-exposure Prophylaxis Trial for HIV Prevention Among African Women (FEM-PrEP) trial of daily oral TDF/FTC and the Vaginal and Oral Interventions to Control the Epidemic (VOICE) trial of daily oral TDF or TDF/FTC and daily topical 1% TDF gel) have stopped one or more arms for futility after interim data reviews.13,14
TDF use in persons with HIV infection has been associated with potentially serious adverse effects, including nephrotoxicity (proximal tubule dysfunction and Fanconi syndrome), hypophosphatemia secondary to renal wasting, and reduced bone mineral density (BMD).15-26 Less serious effects, such as nausea, flatulence, and diarrhea, may adversely affect adherence to daily TDF. We describe here clinical outcomes of the first safety study among US MSM to evaluate potential use of TDF as PrEP in this population. Effects on BMD observed in this investigation have been published previously.26 This study represents the largest safety database of US participants on oral daily PrEP to date.
In this randomized, double-blind, placebo-controlled trial, participants were randomized 1:1:1:1 to receive TDF, 300 mg orally per day, or matching placebo, immediately or after a 9-month delay. The delayed arms were intended to assess changes in risk behavior associated with taking study drug (results reported separately). Randomization was performed in blocks of 8, stratified by site. It was determined that a sample size of 400 would provide 73%-85% power to detect 5%-6% difference in risk of developing an adverse event (AE) between treatment groups, assuming AE risk in placebo groups as low as 1%.
From January 2005 through July 2007, participants were recruited from 3 sites: the San Francisco Department of Public Health (SFDPH), San Francisco, CA; the AIDS Research Consortium of Atlanta, Atlanta, GA; and Fenway Health, Boston, MA. Eligible participants were healthy biological males, 18-60 years of age, who reported anal sex with another man in the preceding 12 months, were HIV-1 negative by whole blood rapid enzyme immunoassay, had a calculated Cockcroft-Gault creatinine clearance ≥70 mL/min, were hepatitis B surface antigen negative, and had normal hematologic, biochemistry, and urinalysis profiles. Exclusion criteria included active untreated syphilis; uncontrolled hypertension; mutual monogamy for ≥1 year with an HIV-uninfected partner; chronic renal disease; osteoporosis, osteomalacia, or osteopenia; BMD Z score less than -2.5 at the total spine, total hip, or femoral neck on screening [dual-energy x-ray absorptiometry scans done at SFDPH only]; current treatment for secondary causes of low BMD; participation in other longitudinal HIV studies; current antiretroviral use, current/planned therapy with potentially nephrotoxic agents; previous/expected requirements for immunosuppressive/immunomodulatory therapy; gastrointestinal malabsorption syndrome or chronic nausea/vomiting; or medical or social conditions that would interfere with study participation. Ethical approvals were obtained through the Institutional Review Boards of Centers for Disease Control and Prevention and all study sites. Each site maintained a community advisory board representative of the study population to provide input on acceptability and feasibility of study design, implementation, and recruitment issues. Participant follow-up was completed in July 2009.
Study visits occurred at 1, 3, 6, 9, 12, 15, 18, 21, and 24 months postenrollment. Delayed arm recipients had an additional visit at 10 months. Visits included AE assessment, symptom-directed physical examination, blood and urine collection, sexually transmitted infection testing, behavioral assessment via audio computer-assisted self-interview, and risk reduction and adherence counseling. Participants were followed for 24 months, even if the study drug was discontinued. Persons who became HIV infected were seen at 3, 6, 9, and 12 months after diagnosis. Visits included assessment of HIV-1 RNA, CD4 cell count, and genotypic/phenotypic resistance.
Laboratory Safety Assessments
HIV testing with an FDA-approved rapid test was performed at each visit; preliminary positives were followed by both a second FDA-approved enzyme immunoassay and either a Western blot or an immunofluorescence assay. Participants with positive rapid and confirmatory tests were further evaluated as described below. Laboratory studies included complete blood count, white blood cell differential, comprehensive metabolic panel, including creatinine, phosphorus, and liver enzymes, and urinalysis.
Grading of Clinical and Laboratory Events
Clinical AEs and laboratory abnormalities (except serum creatinine) were graded using the National Institutes of Health Division of AIDS toxicity tables (January 2004). Laboratory abnormalities were AEs if they (1) were grade 3 or 4, confirmed upon repeat testing, (2) were associated with symptoms, (3) led to discontinuation of the study drug, (4) required treatment, or (5) were otherwise judged clinically significant. Serious AEs (SAEs) were defined as events (1) that were fatal or life threatening, (2) that required or prolonged hospitalization, (3) that resulted in permanent or significant disability or incapacity, or (4) that in the judgment of the investigator might jeopardize the participant and require intervention to prevent one of the outcomes above.
Serum creatinine was graded according to the Gilead Sciences Modified National Institute of Allergy and Infectious Diseases (NIAID) Common Toxicity Grading Scale, with the additional modification that grade 1 was defined as ≥0.5 mg/dL above baseline. All graded elevations were repeated for confirmation. For confirmed grade 1 or 2 creatinine elevations (2.1-3.0 mg/dL), the study drug was withheld and restarted with return of the serum creatinine to within 0.3 mg/dL above baseline. Confirmed recurrence of grade 1 or 2 elevations led to permanent discontinuation of the study product. For confirmed grade 3 (3.1-6.0 mg/dL) or grade 4 (>6.0 mg/dL) elevations, the study drug was permanently discontinued.
Participants with positive rapid tests were immediately discontinued from study drug; those with positive confirmatory tests remained off drug. HIV-1 RNA, CD4 cell count, and HIV-1 genotype studies were obtained. Seroconverting participants were followed for an additional year and referred for appropriate clinical care.
Exposure to Drug and Adherence Measures
Three measures were collected: pill counts at each visit, bottle openings recorded by Medication Event Monitoring System (MEMS) caps, and self-report via audio computer-assisted self-interview (participants estimated the percentage of days they took drug over the preceding month on a visual analog scale). For clinical and laboratory safety analyses, pill count or MEMS data were used to estimate exposure to study drug (ESD). Pill count ESD was calculated as a ratio of number of pills taken within study visit interval over number of days in that interval. MEMS ESD was calculated as a ratio of number of recorded cap openings in a given study visit interval over number of days in that interval (including only one opening per day). Pill count and MEMS adherence estimates were similarly defined except temporary drug interruptions were excluded from the time denominator.
All participants were included in baseline analyses. For clinical and laboratory safety, ESD, and adherence analyses, a treatment emergent (TE) cohort was defined, including all participants who were dispensed study drug. Participants entered the TE cohort with first dispense and exited with the first occurrence of the following: (1) completion of follow-up (24 months), (2) 30 days after permanent drug interruption, or (3) 30 days after last recorded study visit. For delayed arm participants, time before initiation of study drug was excluded from analyses.
Fifteen AEs, defined by MedDRA preferred terms, were chosen for analyses based on clinical importance, known association with TDF, and frequency of occurrence in the study: any grade 3 or 4 event, or any grade upper abdominal pain, back pain, nausea, depression, diarrhea, dizziness, fatigue, flatulence, fracture (any site), headache, increased creatinine, decreased bone density, hypophosphatemia, and proteinuria. Analyses considered each individual AE regardless of relatedness to study drug and considered both time to first reported AE and occurrence of recurrent AEs. In univariate analyses, association of AEs with treatment assignment was studied by modeling time to first reported AE using Cox proportional hazards regression. Incidence of recurrent AEs was compared between groups by modeling AE count using the generalized estimating equation formulation of Poisson regression with total follow-up time as an offset variable.27 Multivariable analysis included treatment assignment as the main predictor and the following covariates: site, immediate/delayed arm assignment, age, race, ethnicity, and ESD. Multivariable comparison of time to first AE was performed using the extended Cox model that accommodated time-dependent covariate ESD.28,29 To avoid underestimating risk, time to first event analysis was performed by selecting the lowest value of ESD between MEMS and pill count estimates calculated for each visit interval, dichotomized at 80% and 50%. The choice of thresholds was exploratory, given that minimal necessary adherence to PrEP for effectiveness is unknown. For analyses of recurrent events, a summary measure of ESD was obtained by taking the mean value of all available visit-specific estimates for each participant, and the lower MEMS and pill count summary estimates were also dichotomized at 80% and 50% level. Potential interactions between treatment assignment and each covariate were assessed. All statistical tests were 2 sided and interpreted at alpha = 0.05 level of significance. To minimize likelihood of missing safety signals, statistical test results were not adjusted for multiplicity. Analyses were performed using SAS/STAT Version 9.2, SAS System for Windows (SAS Institute Inc., Cary, NC).
Overall, 679 persons were screened (Fig. 1). Among 279 participants who were screened but did not enroll, the most common reasons for ineligibility were laboratory abnormalities [58 (20.7%)], passive refusals [persons who did not return after screening, 50 (17.9%)], and active refusals [persons who actively declined enrollment, 37 (13.2%)].
Four hundred participants were enrolled (200 in San Francisco, 121 in Atlanta, and 79 in Boston). Two hundred were randomized to each of the delayed and immediate arms; 201 were randomized to TDF and 199 to placebo. Among all enrolled participants, 331 (83%) completed the final study visit; those who did not complete this visit were similarly distributed among the study arms (P = 0.71). A higher proportion of participants completed the final study visit in San Francisco (180/200, 90%) as compared with Atlanta (91/121, 75%) and Boston (60/79, 76%) (P = 0.0006).
A total of 373 (93%) participants were dispensed study drug at least once and thus entered the TE cohort. All 200 participants in the immediate arms initiated study drug, compared with 173 (87%) of 200 delayed arm participants (P < 0.001). Among those participants who initiated study drug, 325 (87%) completed the final study visit.
Baseline Demographic and Other Characteristics
Participants were distributed similarly among the study arms in terms of age, ethnicity, education, number of male partners in previous 3 months, and number of unprotected anal sexual contacts within the previous 3 months (Table 1). Race distribution differed, with higher proportions of participants reporting African American race in the placebo arms.
Estimates of ESD and Adherence
A total of 178 temporary drug interruptions were documented among the 373 participants: 84 on TDF and 94 on placebo. The median length of temporary drug interruptions for TDF was 37 days (range 2-428 days) and for placebo was 47 days (range 1-533 days, P = 0.055).
Sixty-six permanent drug interruptions were documented (17.6% of participants). Among these, 2 were for persistent elevation of serum creatinine ≥0.5 mg/dL over baseline (both participants were assigned to placebo) and one was for confirmed grade 4 hypophosphatemia (participant was assigned to placebo). As reported previously,25 among participants from San Francisco who received baseline and at least one follow-up dual-energy x-ray absorptiometry evaluation, 11 were permanently discontinued from study drug because of decrease in BMD of less than 5% relative to baseline (8 on TDF, 3 on placebo; P = 0.13).
Overall, median ESD estimated from pill count was 92% (range 79%-98%). Estimated exposure on drug based on MEMS data was lower, at 77% (range 57%-92%). When periods of directed drug interruption were removed from this calculation to estimate adherence, results did not substantially change: overall adherence was estimated to be 93% (range 81%-98%) by pill count and 79% (range 60%-92%) by MEMS.
Overall, 2428 TE clinical and laboratory AEs were recorded to have occurred among 334 (90%) participants. Most (2366, 97%) were of mild (grade 1) or moderate (grade 2) severity: 1845 (76%) grade 1 events and 521 (21%) grade 2 events. Sixty-two grade 3 or 4 AEs occurred among 37 (9.9%) participants. Univariate analyses of selected clinical AEs are summarized in Figure 2. None occurred significantly more frequently in the TDF as compared with the placebo arms.
Multivariable analyses of clinical AEs, adjusted for demographic parameters and drug exposure levels of ≥50% and ≥80%, are described in Table 2. In these analyses, back pain was associated with receipt of TDF. Review of study records of participants reporting back pain revealed no recorded fractures or other objective findings.
No grade 3 or 4 creatinine elevations were recorded. Grades 1 and 2 elevations were uncommon and occurred with similar frequency in the TDF and placebo groups: 1 (0.5%) grade 1 elevation on TDF vs. 4 (2.1%) on placebo (P = 0.20) and 1 (0.5%) grade 2 elevation on TDF vs. 2 (2.1%) on placebo (P = 0.55). Confirmed elevated creatinine values of ≥0.5 mg/dL over baseline occurred in 2 participants; both were discontinued from study drug. Both were assigned to placebo.
Mild to moderate hypophosphatemia was relatively common among study participants and occurred with similar rates on TDF and placebo. Grade 1 hypophosphatemia occurred in 12 (6.5%) participants on TDF vs. 12 (6.4%) on placebo (P = 0.85), grade 2 hypophosphatemia occurred in 31 (17%) participants on TDF vs. 25 (13%) on placebo (P = 0.45), and grade 3 hypophosphatemia occurred in 1 participant on TDF vs. 4 on placebo (P = 0.20). Only one grade 4 hypophosphatemia was reported; this participant received placebo.
A total of 29 protocol-defined SAEs occurred among 18 participants (10 on TDF and 8 on placebo, P = 0.62). The most commonly reported SAE was depression (4 reports from participants on TDF and 2 on placebo). The following were reported for individual participants in the TDF arms: pulmonary edema, appendicitis, prostate cancer, atrial fibrillation, colon cancer, gastrointestinal ulcer hemorrhage, mania, poisoning, seroma, and spinal fracture. Only one SAE, the grade 4 hypophosphatemia on placebo noted above, was considered possibly related to study drug.
One SAE involved a participant death. The cause was determined to be combined opiate and alcohol intoxication and was judged to be unrelated to study drug. The participant had been taking TDF.
Among all 400 participants, 7 seroconversions were documented. None occurred among participants taking TDF. Three occurred among participants taking placebo; 3 occurred among delayed arm participants who had not yet started drug. One occurred in a participant assigned to placebo who was HIV-1 antibody negative at screening and enrollment and then was seropositive at the 1-month visit. Confirmatory Western blot was also positive. Viral load performed on a stored specimen from the enrollment visit was 1770 copies per milliliter; thus, this participant was already infected at enrollment. No K65R mutations were noted among any seroconverting participants.
In these analyses from a study of PrEP among MSM from 3 US cities, TDF was generally well tolerated. While most participants reported at least one AE, most (>90%) were of mild/moderate severity and were not associated with TDF. Severe and SAEs were uncommon and were not associated with TDF as compared with placebo. Few creatinine and phosphorous abnormalities were detected. Creatinine elevations were mild to moderate in severity and not associated with TDF.
BMD studies among the subset of SFDPH participants demonstrated that receipt of TDF was associated with small but statistically significant decreases in BMD at the femoral neck (1.1% decrease) and total hip (0.8% decrease).26 In the multivariable analyses reported here (which were not corrected for multiple comparisons), only back pain was associated with receipt of TDF. Back pain was reported by participants in earlier clinical studies of TDF30 but did not occur at higher frequency than with placebo. No vertebral fractures were documented among those reporting back pain in this study; overall bone fractures at any anatomic site were not associated with the receipt of TDF.
This study has a number of limitations. Generalizability of these findings may be limited, in that participants were required to meet relatively stringent eligibility requirements. The adherence measures used in this study to estimate time on drug-self-report, pill count, and MEMS-each have characteristic limitations.31 The level of adherence required for protection is unknown. There were few seroconversions, none of which occurred while the participant was on active drug. This prevents assessment of the contribution of single-drug PrEP with TDF on the development of drug resistance. Most AEs in our study population occurred with relatively low frequency, restricting power to determine association with TDF and to perform multivariable analyses. However, the MEMS data suggest that the participants in this study were reasonably adherent; so, the lack of findings of TDF-associated toxicity is reassuring. Finally, this study was neither powered nor designed for efficacy assessment. These data are reassuring in that all new infections occurred among men who were not using TDF chemoprophylaxis-3 were on placebo and 3 were delayed arm participants who had not yet started drug. Pooling participants not yet on drug with those on placebo would negate the placebo-controlled aspect of the study design and would break randomization (as not all who were in the delayed arm were randomized to start placebo).
It is interesting that significantly fewer participants randomized to the delayed arms initiated study drug. The reasons for this are not known. It is possible that risk reduction counseling provided at every visit may have resulted in some participants feeling less personal need for PrEP. Also, the long interval may have provided participants time to reflect further on the potential implications of taking a pill every day. The majority of delayed arm participants, 87%, did elect to start study drug as planned.
In contrast to other PrEP trials, which required monthly visits, this trial implemented quarterly clinical and laboratory assessments. This visit schedule was adequate to detect important safety events and may be more feasible than a monthly schedule. The current Centers for Disease Control and Prevention interim guidance for the use of PrEP32 recommends HIV antibody testing, risk behavior assessment, and risk reduction counseling every 2-3 months after initiation of PrEP and assessment of blood urea nitrogen and creatinine 3 months after PrEP initiation and annually thereafter. Adherence assessment and counseling are also recommended at each visit; more frequent monitoring is recommended if inconsistent adherence is identified.
This study is relatively unusual among published PrEP studies in involving a single drug rather than a 2-drug regimen. For this first safety study of PrEP in MSM, TDF alone was evaluated, as this was potentially the more toxic agent of the combination TDF/FTC regimen and permitted more straightforward assessment of drug-associated AEs. Subsequent efficacy trials evaluated TDF/FTC, although some trials included a TDF-only arm.10-12 Although efficacy was high in the TDF-only arm of PrEP in the Partners PrEP study,12 there are concerns that PrEP with TDF alone could select viruses that have the RT K65R mutation that confers broad cross-resistance to nucleoside and nucleotide RT inhibitors, whereas FTC/TDF PrEP more commonly selects the RT M184V/I mutations that cause resistance limited to FTC and 3TC and typically cause hypersusceptibility to TDF and zidovudine (AZT). Regardless of regimen, the risk of drug resistance has been limited in trials to those starting PrEP with preexisting infection, emphasizing that HIV testing using sensitive fourth generation antigen + antibody assays or RNA assays will reduce drug resistance risk during PrEP to very low levels. The single-pill, fixed-drug combination FTC/TDF provides a simple once-daily regimen and has demonstrated a favorable safety profile in several studies.10-12
Efficacy of daily PrEP with oral FTC/TDF has been demonstrated among MSM and heterosexual men and women in 3 studies to date10-12; one study was not able to demonstrate efficacy of oral FTC/TDF for heterosexual women.13 Oral TDF alone (without FTC) was efficacious among serodiscordant couples,12 a group that is thought to have high adherence, whereas an evaluation of oral TDF PrEP in individual heterosexual women was discontinued because of futility.14 The reasons for these conflicting results for heterosexual women are unclear but could represent differential penetration of the drugs into vaginal vs. rectal or penile tissues33,34 or differences in adherence.
Evidence for efficacy of daily oral FTC/TDF for MSM is clear on an intention to treat basis, and the reduction of HIV risk was high among those with detectable drug exposure, highlighting how the use of PrEP is a critical determinant of PrEP activity.10 However, important questions remain that bear on feasibility, desirability, and appropriateness of PrEP as an HIV prevention intervention. Financial cost and other barriers to access must be considered. The long-term safety of antiretroviral drugs in HIV-negative individuals is not yet known, nor is the likelihood for emergence of resistance in real-world use, with potentially less meticulous HIV monitoring than in the clinical research setting. Also uncertain is the percentage of individuals who will be able to adhere to a PrEP regimen and the necessary monitoring. The potential for behavioral risk compensation remains an open question, although risk compensation was observed neither in previous PrEP trials nor in open-label studies of postexposure prophylaxis.35 PrEP will require ongoing engagement with the health care system to monitor adherence, potential HIV exposure, and the appropriateness of PrEP for each individual over time.36
In conclusion, these data contribute important information for the development of clinical guidelines for PrEP among US MSM. This trial augments a growing body of evidence demonstrating the safety of oral PrEP and with other published reports supports the need for thorough exploration of feasibility and acceptability of this potentially promising intervention.