Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 27
Review of the HIVNET 012 Perinatal HIV Prevention Study 3 Study Design, Treatment Assignment, and Adherence to Study Regimens The committee interpreted its charge as encompassing scientific elements of the design of HIVNET 012, including the research setting, eligibility criteria for participants in the study, drug regimens, the randomization scheme, sample size, and statistical methods. This chapter addresses these issues. This chapter also addresses the implementation of the randomization procedure, drug management procedures, and participants’ adherence to study regimens. Chapter 4 focuses on issues related to efficacy and safety data, and Chapter 5 focuses on ethical issues related to the design and conduct of the study. BACKGROUND Mother-to-child transmission of HIV-1 can occur in utero (antepartum), at the time of labor and delivery (intrapartum), or during breastfeeding (postpartum) (Working Group on Mother-To-Infant Transmission of HIV, 1995). Enrollment in HIVNET 012 began in November 1997. At that time, the standard of care for preventing mother-to-child transmission in both the United States and Europe was based on the results of AIDS Clinical Trials Group (ACTG) protocol 076. That trial showed that a three-part maternal-infant zidovudine (ZDV) regimen reduced the rate of transmission by about two-thirds, from 25.5% to 8.3%, compared to a placebo (Connor et al., 1994). The drug regimen studied in ACTG 076 consisted of 100 milligrams of ZDV given orally five times daily to HIV-1-infected pregnant women beginning at 14–34 weeks gestation, followed by
OCR for page 28
Review of the HIVNET 012 Perinatal HIV Prevention Study an intravenous infusion of ZDV during labor (2 milligrams per kilogram of body weight in the first hour, and 1 milligram per kilogram of body weight each hour until delivery). Infants received oral ZDV (2 milligrams per kilogram of body weight every 6 hours) for the first 6 weeks of life. Women enrolled in ACTG 076 had CD4+ lymphocyte counts greater than 200/µL and generally formula-fed their infants. These findings spurred the Centers for Disease Control and Prevention and other groups to recommend that all pregnant women in the United States should be offered testing for HIV-1 infection, and that ZDV should be administered to those found to be infected, as well as to their infants. Rapid implementation of these recommendations led to a prompt and dramatic decrease in rates of mother-to-child transmission (Fiscus et al., 1996). A strong public health infrastructure and federal and state commitment of additional resources made that implementation successful. Despite that success, the ACTG 076 regimen was too expensive for many resource-constrained settings, required identification of HIV-1 infection before or during pregnancy, and necessitated patients to adhere to daily multidose ZDV regimen for weeks or months. The regimen also required inpatient delivery, facilities and expertise in administering intravenous ZDV and in administering and monitoring ZDV therapy for 6 weeks in the newborn infant. The approach was clearly not practical or affordable in sub-Saharan Africa and resource-limited settings globally, where most mother-to-child transmission occurred in 1997 and still occurs today. Recognition of this fact led to a number of trials—HIVNET 012 among them—designed to evaluate simpler, less-expensive approaches for use in resource-limited settings. CHOICE OF UGANDA AS STUDY SITE Uganda has been hard hit by the HIV/AIDS epidemic. In 1997 the prevalence of HIV-1 among pregnant women in Kampala was estimated at 15% (USAID, 2005). In 2001, some 280,000 women ages 15 to 49 were living with HIV-1 infection (UNDP, 2002). Of the 170,000 infants born each year to HIV-1-infected Ugandan women, some 43,000 would be expected to acquire HIV-1 infection. Vertical transmission accounts for 15–20% of all new HIV-1 infections in Uganda (Uganda AIDS Commission Secretariat, 2002). Antiretroviral therapy was largely unavailable in Uganda while HIVNET 012 was being conducted, and there were no programs for preventing mother-to-child transmission. Poor public health, clinical, and laboratory infrastructure and capacity, as well as high rates of out-of-hospital delivery, made the ACTG 076 regimen impractical for use in Uganda.
OCR for page 29
Review of the HIVNET 012 Perinatal HIV Prevention Study With some 1,500 beds, Mulago Hospital is the largest government referral hospital in Uganda and the main teaching hospital for Makerere University School of Medicine in Kampala. The hospital provides both tertiary care to Ugandan residents and primary care to the country’s poorest residents who cannot afford care elsewhere. The Department of Obstetrics and Gynaecology provides inpatient and outpatient services including normal deliveries, emergency care for women with complicated pregnancies, care for gynecological emergencies, and postnatal care. The Department of Paediatrics and Child Health also has an acute-care unit, a neonatal intensive-care unit, specialized outpatient clinics, a nutritional unit, and the Child and Health Development Center. When HIVNET 012 was initiated, approximately 85% of hospital admissions were HIV-related (Guay et al., 2002). Thus Mulago Hospital and its staff had the capacity and experience to conduct a clinical trial of a less intensive regimen for preventing mother-to-child transmission. CHOICE OF DRUG REGIMENS Given the impracticality of the intense ACTG 076 regimen for resource-limited settings, investigators in Africa and elsewhere soon began evaluating simpler, less expensive regimens for preventing mother-to-child transmission. Nevirapine (NVP) was considered promising because of its potent and rapid antiretroviral effect (Bardsley-Elliot and Perry, 2000; Mirochnick et al., 1998). Data from a preliminary study of 10 mothers given either 100 or 200 milligrams of NVP orally at the onset of labor also produced positive results (Mirochnick et al., 1998). In that study, the median maternal serum level was 714 ng/ml. Infants whose mothers had received doses at least 2 hours before delivery maintained serum levels of NVP greater than 100 ng/ml for 3 days. This level was thought to be potentially protective against HIV-1. Appreciable levels of NVP were also found in breast milk. Hence NVP was readily absorbed and passed directly to the fetus at levels that might prevent transmission of HIV-1, and no toxicity was noted in either mother or child. On that basis, the HIVNET 012 investigators chose to study an NVP regimen consisting of 200 milligrams taken orally by pregnant women at the onset of labor, followed by 2 milligrams per kilogram of body weight for infants within 72 hours of birth. The impetus for the short-course ZDV regimen used in HIVNET 012 was the belief that a simpler, less expensive version of the ACTG 076 three-part ZDV regimen might retain most or all of its benefits, thereby enabling widespread prevention of mother-to-child transmission in resource-poor settings. The HIVNET 012 investigators chose a ZDV regimen consisting of 600 milligrams orally to the mother at the onset of labor, followed by
OCR for page 30
Review of the HIVNET 012 Perinatal HIV Prevention Study 300 milligrams every 3 hours until delivery and 4 milligrams per kilogram of body weight orally twice daily to infants for 7 days after birth. The investigators did not expect either the single-dose NVP regimen or the short-course ZDV regimen selected for use in HIVNET 012 to have much of an effect on in utero HIV-1 transmission, as the first dose of both regimens was to be administered at the onset of labor. As essentially all the women were expected to breastfeed their infants, neither regimen was expected to have an effect on transmission through breast milk after the first several days of life. Finally, HIVNET 012 was solely focused on preventing mother-to-child transmission and not on treating HIV/AIDS. Enrollment to HIVNET 012 began in November 1997. However, within 3 months, in February 1998, a trial in Thailand found that a short-course regimen of ZDV during pregnancy reduced transmission from 18.9% to 9.4%, as compared with placebo. This treatment consisted of 300 milligrams given orally to pregnant women twice daily beginning at 36 weeks of gestation and every 3 hours from onset of labor until delivery was found to reduce MTCT in Thailand (CDC, UNAIDS, NIH, and ANRS, 1998; Shaffer et al., 1999). After learning of these results, the HIVNET 012 investigators immediately stopped randomizing women to the study’s placebo arms.1 They considered whether to replace the HIVNET 012 ZDV regimen with the ZDV regimen found to be effective in the Thai trial but concluded that that regimen was fairly expensive and thus would not be available to the majority of women in Uganda. The investigators also noted that the Thai regimen might be less effective in a breastfeeding population, and that the regimen did not include a postnatal dose. The investigators therefore decided to seek approval for continued open-label enrollment in the original NVP and ZDV arms.2 This request 1 At that time, 72 women had been enrolled and assigned to one of the placebo arms, and 48 had already delivered. The remaining 24 were unblinded. Seventeen of those women had been assigned to either ZDV or NVP active arms. The 7 who had been assigned to a placebo arm were contacted and offered the active regimen corresponding to the placebo arm they had been part of; if they agreed, those previously assigned to ZDV placebo were reassigned to the ZDV arm, and those previously assigned to NVP placebo were reassigned to the NVP arm (SCHARP, 2004a). 2 The original trial was double-blind with respect to both ZDV and NVP. “The investigators’ Proposed Interim Plan (HIVNET 012 Investigators, 1998) indicates that NIH and other bodies recommended that placebo arms of their sponsored trials be discontinued. The investigators also indicated that it would take several months to redesign the trial and asked to continue the trial as an open-label trial until ‘we can design and implement a revised protocol with an appropriate control arm.’ This strategy was justified in part by the argument that, if the trial were stopped while the design was being revised, HIV infected women already enrolled or screened would receive nothing” (page 2) (HIVNET 012 Investigators, 1998).
OCR for page 31
Review of the HIVNET 012 Perinatal HIV Prevention Study was reviewed and approved by the relevant Institutional Review Boards at Johns Hopkins University and in Uganda as well as the Data and Safety Monitoring Board (DSMB).3 Enrollment in the revised protocol began in April 1998. Finding: The committee finds that Uganda was a reasonable setting for evaluation of short-course regimens for preventing mother-to-child transmission of HIV-1. Moreover, the regimens chosen for study were reasonable in the context of knowledge about prevention of mother-to-child transmission at that time. The decision to stop the placebo arms of the trial and continue with the originally designed active arms was reviewed and approved appropriately. ELIGIBILITY CRITERIA The investigators sought to develop eligibility criteria that were broadly inclusive, subject to the constraints of the appropriate age and stage of gestation of the mother, her ability to participate in the study, and contraindications to the study regimens. Thus, women were eligible to participate in HIVNET 012 if they were at least 18 years old, at more than 32 weeks gestation, HIV-1-infected, and lived within 15 kilometers of the hospital. Potential participants were excluded if they were receiving antiretroviral therapy or other disallowed drugs, had active serious non-HIV-1-related infection or illness, had known hypersensitivity to benzodiazepine, were using drugs or alcohol, had uncontrolled hypertension, were participating in other therapeutic or vaccine perinatal trials, or had abnormal hemoglobin, ALT (SGPT) or creatinine levels. Finding: The committee finds that the inclusion and exclusion criteria employed in HIVNET 012 were reasonable. DESIGN AND IMPLEMENTATION OF THE RANDOMIZATION PROCEDURES The creation of comparable treatment groups through randomization is fundamental to the validity of randomized clinical trials. This section reviews the design and implementation of the randomization procedures used in HIVNET 012, as well as the comparability of the resulting treatment groups. We begin by noting concerns addressed in the 2002 Westat 3 A DSMB is an independent committee composed of statisticians, clinicians, community representatives and ethicists who examine a study and resultant data before it begins and on an ongoing basis.
OCR for page 32
Review of the HIVNET 012 Perinatal HIV Prevention Study review, and address these by evaluating the design and implementation of the study randomization. The Westat concerns were based on what they considered to be non-random patterns in the study data based on an apparent imbalance in the treatment assignments in a subset of enrolled women listed as born between 1973 and 1974 (Chamberlin et al., 2002). The Statistical Center for HIV/AIDS Research and Prevention (SCHARP) argued that the imbalance was less substantial than suggested by the Westat review and that the probability that an imbalance of this magnitude or greater would occur by chance (without adjusting for multiple testing) was 4%. Rather than attempting to resolve whether an imbalance did exist in a subsample of enrolled women, the committee chose to review the entire randomization design and implementation for all mothers (see below). Westat also raised concerns about unexplained gaps in identification numbers for study participants which they believed should have been sequential (Chamberlin et al., 2002). In a response to the Westat report, the investigators explained that these gaps were due to the dropping of the placebo arms in February 1998 (HIVNET 012 Investigators, 2002). The committee investigated this by examining the entire process of assigning study identification numbers (see below). The original design of HIVNET 012 randomized mother/infant pairs to four treatment arms: NVP, NVP placebo, ZDV, and ZDV placebo, in proportions of 2:1:2:1. This resulted in equal allocations to NVP, ZDV, and placebo, with the intent of comparing each active drug to placebo. The trial included two placebo preparations to provide partial blinding of both active drug regimens. The randomization procedure was based on permuted blocks of 12. This algorithm randomly assigned four infants each to NVP, ZDV, and placebo after each consecutive set of 12 assignments. The randomization list was generated by SCHARP and matched to consecutive patient identification numbers (IDs) beginning with 512-0001, 512-0002, and so on. The resulting patient IDs were then sent to Johns Hopkins University to guide packaging of study drugs.4 The resulting kits were labeled with the ID but did not include any information that would reveal the assigned treatment. The last step in the process was to assign IDs, and hence drug kits, to women enrolled into the study in the chronological order of their enrollment. 4 In April 1998, a new drug-handling and -packaging procedure was implemented in which the medications packaged in the manufacturer’s bottles were shipped to Uganda. A pharmacist on-site in Uganda prepared the drug kits.
OCR for page 33
Review of the HIVNET 012 Perinatal HIV Prevention Study This randomization plan was sound. The use of a relatively large block size of 12 in the permuted blocks scheme was appropriate given the partially blinded design. Members of our committee verified that the actual randomization list did correctly construct block sizes of 12 according to the indicated proportions. (The first block intentionally had only 9 assignments to introduce an additional component of randomness [Donnell, 2004].) The procedures for preparing and labeling drug kits were also sound. In February 1998, after 72 women had been randomized into HIVNET 012, randomization was temporarily suspended because of the emerging results from the perinatal HIV prevention trial in Thailand (CDC, UNAIDS, NIH, and ANRS, 1998). With consultation from the HIVNET 012study’s Data and Safety Monitoring Board, the HIVNET 012 investigators concluded that it would be unethical to continue randomizing women to placebo, and therefore terminated the placebo arms of the study. The Institutional Review Boards at Johns Hopkins University and in Uganda reviewed and approved the revised study design and in April 1998 the study resumed enrolling patients into the remaining ZDV and NVP arms. At the time when the placebo arms were discontinued, 24 women had been enrolled and assigned to a study regimen but had not yet delivered. Seventeen of these women had been assigned either to ZDV or NVP and remained on their assigned treatment. The remaining 7 had been assigned to placebo (4 to ZDV placebo and 3 to NVP placebo). These 7 women were switched to the active drug corresponding to the placebo arm to which they had been assigned. Hence 4 mothers switched from ZDV placebo to ZDV, and 3 switched from NVP placebo to NVP (SCHARP, 2004a). The revised randomization plan assigned women to ZDV or NVP in equal proportions, using permuted blocks of 26, so that each successive group of 26 randomized women would include 13 randomized to ZDV and 13 randomized to NVP. Kits and IDs were modified to reflect this new randomization plan. Other features of the randomization procedure remained unchanged (SCHARP, 2004a). The use of a permuted block size of 26 is also appropriate because the revised design was not blinded, and because this was a single-site study. The larger block size minimized the possibility of predicting future treatment assignments, one potential source of bias in unblinded trials. Finding: The committee finds that the designs of the original and modified randomization procedures were scientifically sound and appropriate for the research setting. Randomization occurred at the time of enrollment into the study when an individual participant was assigned a study ID number. The study coordinator assigned the expectant mother the next sequential study ID number
OCR for page 34
Review of the HIVNET 012 Perinatal HIV Prevention Study and its corresponding kit. During the placebo-controlled phase of the trial, both the study staff and participants were blind to the assignment of drug or placebo. Later, in the open-label phase of the trial, the assignment was known once the kit was opened (Guay et al., 2002). Blinding of study treatment guards against two potential sources of bias: differential ancillary treatment of participants and differential assessment of study endpoints. In HIVNET 012, lack of knowledge about the efficacy of these unproven short-course regimens and the complexity of the care setting made it unlikely that ancillary care would have differed between the two treatment arms. By focusing on the “hard” efficacy endpoints of HIV-1 infection and survival status, the investigators minimized the possibility of differential assessment of efficacy endpoints. The committee independently verified the randomization procedure and its implementation by reviewing the treatment assignment lists and checking them relative to treatment assignments of individual mothers. A consultant to the committee reviewed the chronological order in which women were enrolled in HIVNET 012 to determine whether kits were assigned in consecutive order. This review established that only four women did not receive kits in the order corresponding to the chronological order of their enrollment. In two of the four instances, the woman received the treatment assignment that she would have received if she had been randomized correctly. In the other two cases, the assigned drug was different from the assignment corresponding to the intended order of randomization. In one instance, NVP was assigned instead of ZDV; in the other, ZDV was assigned instead of NVP. Comments at the time by those involved in HIVNET 012 suggest that these few errors were accidental and arose from special situations during enrollment of expectant women. The committee found no evidence of a deliberate violation of the randomization plan. Moreover, the number of assignment errors is negligible relative to the total enrollment in the study. Thus, the committee finds that analysis of the results based on the treatment actually assigned is appropriate. Moreover, because only two mothers received a different treatment assignment than if all kits had been assigned in chronological order of enrollment, reanalysis of the results using the “intended” assignment would yield virtually identical results. As shown in Table 1 of the 1999 Lancet paper, the characteristics at enrollment of women who gave birth were similar in the two treatment groups. Aside from a significant difference in mean birthweight of 100 grams, the characteristics of study infants did not differ between the two treatment groups (Fleming, 2004; HIVNET 012 Investigators, 2002). Finding: The committee finds that the HIVNET 012 randomization procedures were implemented with a high level of accuracy, achieving the scientific goal of creating two comparable treatment groups.
OCR for page 35
Review of the HIVNET 012 Perinatal HIV Prevention Study SAMPLE SIZE As originally designed, HIVNET 012 had four treatment arms: ZDV, ZDV placebo, NVP, and NVP placebo. These were to be analyzed as a three-arm study comparing the safety and efficacy of two short-course regimens of NVP and ZDV with the combined placebo arms. The two placebo arms were created to provide partial blinding of treatment assignments for both active regimens. The primary endpoint for determining sample size was the rate of HIV-1 positivity in infants up to 18 months of age (Jackson et al., 1997). The resulting sample size was also used to assess power for the endpoint of HIV-1 infection-free survival through 18 months, and safety and tolerance of the drug. The original trial was designed to yield 90% power to detect an absolute 33% reduction in the primary endpoint, at the one-sided 0.025 significance level favoring the active drugs. To achieve these design characteristics, a target sample size of approximately 500 mother/infant pairs was to be enrolled in each of the three treatment regimens, for a total sample size of 1,500 randomized mothers (Fleming, 2004). As noted, the placebo arms were discontinued and the study interrupted in February 1998. In April 1998, when enrollment into the ZDV and NVP arms resumed, the investigators considered whether and how the study design should be modified. The intermediate goals of the study became the comparison of the ZDV and NVP arms, in a two-arm open-label Phase IIB screening trial, aimed at selecting one arm to advance to a future phase III trial with an anchor comparison arm. The anchor was expected at that time to be a regimen selected from another African trial (Petra Study Team, 2002). The two-arm trial was designed as a non-inferiority comparison of the NVP regimen to the ZDV regimen, owing to the ease of administration and lower cost of NVP. The HIVNET 012 Proposed Interim Plan (HIVNET 012 Investigators, 1998) outlines the design characteristics for a total sample size of either 250 or 500 randomized mothers. This approach would result in one of three possible outcomes: (1) NVP alone to be selected if the estimated transmission rate was no more than 3% higher in the NVP group than in the ZDV group; (2) both ZDV and NVP to be selected if the estimated transmission rate difference between the two groups was 3–5%; and (3) ZDV alone to be selected if the estimated transmission rate difference exceeded 5%, favoring ZDV. Assuming that 500 mother/infant pairs were to be enrolled, this new design had more than an 80% probability of selecting NVP alone, and less than a 10% probability of selecting ZDV alone, if the true transmission rates for the NVP and ZDV regimens were identical. The efficacy of a two-stage approach, continuing with a phase IIB screening trial, followed by a
OCR for page 36
Review of the HIVNET 012 Perinatal HIV Prevention Study Phase III randomized controlled trial after the PETRA5 results were known, motivated the decision to reinitiate enrollment with a new target sample size of approximately 400–600 eligible HIV-1 infected pregnant women. In fact, the trial achieved a total sample size of 626 randomized mothers, equally balanced with 313 receiving NVP and 313 receiving ZDV. Thus, for an absolute difference in transmission of 12%, this design realized more than the planned 80% probability of correctly choosing NVP alone as the preferred regimen for further study in a Phase III randomized controlled trial. Finding: The committee finds that the original and revised sample size targets for the HIVNET 012 trial were sufficient to achieve the study goals. STATISTICAL METHODS Guay and colleagues (1999) used standard failure time methods for assessing HIV infection and HIV-free survival. Specifically, they used the method of Kaplan and Meier (1958) to estimate the distribution of time to HIV positivity and HIV-free survival time. From these time-to-event distributions, they estimated the probabilities at weeks 6–8 and 14–16 for each treatment arm, using standard methods for survival data to compare the estimated probabilities in the two treatment groups. While HIV infection was evaluated only periodically, the approach used by the authors is both valid and efficient because infants in the ZDV and NVP arms had the same evaluation schedule, there were very low rates of incomplete information, and the treatment groups were evaluated at the times of scheduled PCR6 evaluations. The use of 2-sided p values is also appropriate, and the reporting of nominal p values, without adjustment for interim analyses, is appropriate because of the conservative spending function used during the interim analysis of the study results. Finding: The committee finds that the statistical methods employed in HIVNET 012 and described in the publications were appropriate. The results obtained from the analyses of HIV infection and HIV-free survival were properly interpreted by the study authors. Additional analyses of efficacy in the Results and Discussion sections of Guay et al. 5 The PETRA study was begun in 1998 in South Africa, Tanzania, and Uganda under the auspices of UNAIDS, and investigated the combination of ZDV and lamivudine (3TC) (in a 4-arm, randomized, placebo-controlled trial) for the prevention of perinatal HIV transmission (Peiperl, 2002). 6 PCR stands for HIV-1 RNA polymerase chain reaction, a test used to determine HIV infection status.
OCR for page 37
Review of the HIVNET 012 Perinatal HIV Prevention Study (1999) and Jackson et al. (2003) were presented in a balanced manner and with appropriate qualifications. DRUG MANAGEMENT The original HIVNET 012 protocol stipulated that on-site investigators would maintain complete records of all study drugs received from Boehringer Ingelheim and Glaxo Wellcome, including how drugs were dispensed. Unused drugs were to be returned to the manufacturer or destroyed once the study was complete (Jackson et al., 1997).We review the major components of the drug management system briefly below.7 Drug Packaging and Handling Before Enrollment The study relied on two sets of procedures for packaging and handling of study drugs prior to enrollment. Initially, manufacturers sent the drugs directly to the Johns Hopkins University pharmacy, where active and placebo preparations were packaged into either small white bottles for syrups or single-dose cellophane/foil strips for tablets that were identical in appearance. Outer package labeling was identical for all arms and contained no information that could be used to identify the study arm (Jackson and Guay, 2004). The bottles were labeled with the study identification number (given at randomization) and secured with tamper-proof seals, to verify that they had not been opened before the mother enrolled and was given a bottle. This drug-packaging system was used for the first two batches of study products sent to Uganda on October 7, 1997, and April 17, 1998 (see Table 3.1 for details).8 After the site received the first two batches of drugs, and after enrollment was suspended in February 1998 to allow the redesign of the study without placebo arms,9 enrollment was stopped for a second time on May 7 For a more detailed discussion of drug packaging, handling, and dosing, please see the 2002 Description of Study Procedures (Guay et al., 2002). Facts were confirmed in email communication from L. Guay, March 29, 2005 (Guay, 2005d). 8 The first batch of drug kits (study ID number 512-0001 to 512-0150) along with replacement study product was brought by Dr. Jackson in checked baggage to the study site on October 7, 1997. A second shipment of prepackaged study product was sent from Johns Hopkins University by courier and received on the site on April 21, 1998 (see Table 3-1). (Guay et al., 2002). 9 On February 18, 1998, the placebo arms were discontinued. SCHARP unblinded the 24 enrolled, randomized mothers who had not yet delivered. Seven of the 24 women had been assigned either NVP or AZT placebo. The placebo kits were removed and replaced with the matching active agent. The remaining placebo kits on site that had not been assigned to a woman were removed from the supply and destroyed at the end of the study (Guay et al., 2002).
OCR for page 38
Review of the HIVNET 012 Perinatal HIV Prevention Study 25, 1998, and resumed on July 7, 1998, after the packaging procedure had been changed. Under the new procedure, the Division of AIDS Clinical Research Products Management Center (CRPMC) sent the drugs (packaged in their original manufacturer’s bottles) and pre-labeled empty drug kits directly to Uganda for on-site kit preparation by a pharmacist at the time of enrollment.10 These changes were introduced in response to concerns raised by Division of AIDS (DAIDS) about lack of information on the stability of the drug in small bottles (Guay et al., 2002; Jackson and Guay, 2004). SCHARP generated a new randomization schedule for use in packaging and labeling with a new series of ID numbers beginning with 512-0401 through 512-0888. CRPMC made three drug shipments on July 6, 1998, November 2, 1998, and March 15, 1999 (see Table 3.1). Each weekday morning, the pharmacist prepared drug kits according to the number of women expected for enrollment that day, based on a note in the pharmacy log from the study coordinator, beginning with the next sequentially numbered empty kit. The pharmacist signed the note once he completed making the kits, but these notes are no longer available. The information was recorded in the pharmacy log though (Guay et al., 2002; Guay, 2005a). All study staff members except the on-site pharmacist were blinded to the drug assignment schedule until after enrollment was complete. Dr. Guay, one of the HIVNET 012 co-investigators, periodically reconciled the quantities of drugs received and dispensed and recorded this information in the pharmacy log (Guay et al., 2002). Drug Handling After Enrollment and Dosing Women enrolled in the study at antenatal clinics. Each woman received an identification number after study staff had verified her eligibility and obtained informed consent. When notified that a woman scheduled for enrollment had arrived at the clinic, the study coordinator collected the next sequentially numbered set of source files from the Mulago University-Johns Hopkins University data center and corresponding study drug kits from the storage room. A staff member delivered the appropriate files and kits to the antenatal clinic and gave them to the study midwife in charge. After the placebo arm was dropped, the staff could determine whether the regimen was NVP or ZDV once the kit was opened and documented this on the enrollment form (Guay et al., 1999). 10 Family Health International contracted with McKesson Bioservices to prepare and label the empty individual drug kits. Individual drug containers inside the kit were labeled with the specific study regimen. Labeling on the outside of the kit was identical for the two study arms (Guay et al., 2002)
OCR for page 39
Review of the HIVNET 012 Perinatal HIV Prevention Study Women were instructed to take their initial dose at the onset of labor. When a participant arrived at the labor ward, she gave her estimated time of labor onset and time of initial dose to the nurse. For mothers assigned to ZDV treatment, study staff administered doses of ZDV every 3 hours. A drug dosing chart in the source file recorded the drugs, doses, their dates and times, and the person who administered them. The hospital chart did not include records of drug dosing (Guay et al., 2002). Study staff gave the infant dose of NVP at the time of discharge. The majority of infants were dosed within 24 hours. For women who delivered at home or arrived at the labor ward later, the protocol allowed infant dosing up to day 7. Women who delivered at home were encouraged to bring their infants to the research site at Mulago Hospital as soon as possible, and infants were administered medication when they arrived (Guay et al., 2002). The infants in the ZDV arm were given the study drug twice a day. The staff initially administered the ZDV exactly 12 hours after delivery, but later changed the procedure and gave dosings at 8:00 a.m. and 8:00 p.m. to make it easier for mothers when they returned home. The doses given by hospital staff were documented on the chart. Mothers were instructed to come back for the 7-day visit, told when to stop the drug, and asked to return their vial. The pharmacy return log recorded the number of unused tablets and syrup returned from the labor ward and participant along with a final accounting of the amount received, used, and returned. It appears that all unused and returned study products were destroyed in January 2002 except for eight kits which were transported back to the United States in July 1998 for quality control checks and which were later destroyed (Guay et al., 2002). Reviews of Pharmacy Procedures The Westat Site Visit Report (Chamberlin et al., 2002) expressed several concerns including that the study site did not use a subject treatment assignment list, several types of dosing errors occurred, and the temperature was not monitored in the rooms where the study products were stored. (Subsequent temperature monitoring under similar conditions by the investigators was in the acceptable range for the study products [Jackson et al., 2003a]). In a Summary of the Westat Debriefing, DAIDS noted that the Westat report “was unable to establish the full extent of such [dosing] errors or their significance for the validity of the study conclusions” and that those areas identified would require “more detailed review” (DAIDS, NIAID, 2002). DAIDS also noted that the HIVNET 012 statisticians provided analyses
OCR for page 40
Review of the HIVNET 012 Perinatal HIV Prevention Study TABLE 3.1 Shipments of Study Drug Batch / Shipment Date Received On Site Shipment Method Number of Drug Kits Shipped ID of Drug Kits Shipped Additional Study Drug Shipped Dates Kits Were Useda For shipments 1 and 2, drug kits were prepared and filled by Johns Hopkins University (JHU) and sent to Uganda. 1 10/7/97 B. Jackson; checked baggage on commercial airline 142 (8 kits removed for quality control check at JHU) 512-0001–512-0150b 6 bottles of study tablets; 6 bottles of study syrup (sent as replacement stock) 11/3/97–4/21/98 2 4/21/98 Shipped from JHU via TNT Express 72c 512-0151–512-0257 None 4/22/98–5/25/98 For shipments 3, 4, and 5, drugs and prelabeled empty kits were sent to Uganda for packaging by an on-site pharmacist. 3 7/6/98 Shipped from CRPMC via World Courier 400 (200 ZDV and 200 NVP) 512-0401–512-0801d 30 bottles of study tablets; 40 bottles of study suspension 7/7/98–2/23/99e 4 11/2/98 Shipped from CRPMC via World Courier 0 N/A 10 bottles study tablets; 20 bottles of study syrup N/A 5 3/15/99 Shipped from CRPMC via World Courier 199 (100 NVP and 99 ZDV) 512-0799, 512-0802–512-0999f 25 bottles tablets; 9 bottles of study syrup 3/16/99–4/30/99
OCR for page 41
Review of the HIVNET 012 Perinatal HIV Prevention Study aOf the 813 drug kits, 652 were used for study participants. 161 were not used and were later destroyed (Guay, 2005c). bThe following 8 kits were randomly selected by the statistical center and removed from the batch prior to shipment for quality control: 512-0011; 512-0037; 512-0081; 512-0109; 512-0120; 512-0129; 512-0144; 512-0150. cOf 107 kits in Batch #2, 35 kits were assigned to placebo and were never made or shipped to the site because the placebo arm was dropped. Patients were not assigned these ID numbers (Guay, 2005b). dSCHARP generated a new randomization schedule for use in packaging and labeling with a new series of ID numbers beginning with 512-0401. Kit #512-0799 was inadvertently skipped and not included in Batch #3 (Guay, 2005d; Guay et al., 2002). This was not formally documented at the time of receipt, but a note was sent to the study monitor (FHI) regarding the missing drug kit. This kit was later shipped as part of Batch #5. The last kit used was #512-0888 (Guay et al., 2002). eKits received from CRPMC (shipments 3 and 5) were used from July 7, 1998, to April 30, 1999 (the end of enrollment) (Guay et al., 2002). fAn additional drug kit (512-0889) was prepared in anticipation of another enrollment, but no participant was assigned to this kit, and it was returned to the pharmacy and disposed of with other unused products (Guay et al., 2002). SOURCE: Guay (2005d); Guay et al. (2002).
OCR for page 42
Review of the HIVNET 012 Perinatal HIV Prevention Study showing that errors reported by Westat would “have little or no impact on the validity of the study’s core conclusions” (DAIDS, NIAID, 2002). The study statisticians and the HIVNET 012 investigators themselves described non-adherence as reflective of real-world conditions, and attributed most of the dosing errors cited in the Westat report to participant non-adherence (HIVNET 012 Investigators, 2002). The investigators also corrected several inaccuracies in the Westat report’s description of procedures related to drug storage and handling (HIVNET 012 Investigators, 2002). The DAIDS Pharmaceutical Affairs Branch also visited Uganda in July 2002 to review the drug-management procedures at the site. They examined study drug treatment assignment, drug kit preparation by the pharmacist, drug distribution, documentation of dispensing procedures, drug return, chain of custody, and destruction of unused drugs. The DAIDS remonitoring team expressed concerns about some inadequate documentation and dosing errors as well. They noted various ways in which more detailed information in the pharmacy records and source documents could have more fully documented the pathway from packaging of study drugs to dosing of the mothers and infants, thereby providing more robust evidence regarding the management of study drugs and level of adherence to study regimens (DAIDS, NIAID, 2003). HIVNET 012 investigators acknowledged some deficiencies and missing documentation relating to drug management, but noted that the study achieved a very high level of adherence to treatment regimens (Jackson and Guay, 2004), as described in The Lancet articles (Guay et al., 1999; Jackson et al., 2003a). After reviewing all relevant information, the committee concluded that a number of sources document the distribution and administration of assigned study drugs to study mothers and their infants in accordance with their treatment assignments. In both the first and second method of drug packaging, the proper packaging of study drug in accordance with the randomization schedule is well documented. Information available in the source documents indicates that drug kits were consistently provided to the appropriate mothers and caregivers. Thus, the information available to the committee indicates that the potential deficiencies in documenting drug management noted by Westat and DAIDS were, in fact, deficiencies of documentation, not drug delivery, and that study drugs were consistently and appropriately provided to study participants. ADHERENCE The committee also reviewed the available information on participants’ adherence to study treatments. HIVNET 012 measured such adherence both directly and indirectly. All patients were asked about their adherence
OCR for page 43
Review of the HIVNET 012 Perinatal HIV Prevention Study to the medications and answers were recorded on the case report form. In addition, the nursing and labor and delivery staff were queried regarding adherence. In the ZDV arm, 17 of 308 women delivered outside of Mulago Hospital. In the NVP arm, 20 out of 311 women delivered at a site other than Mulago Hospital (SCHARP, 2004b). In the ZDV arm, 50 women were redosed (for reasons such as false labor or vomiting), 3 of whom had delivered outside of Mulago Hospitals. Of the 244 women in the ZDV arm who delivered at the hospital and who were not redosed, 142 women took a dose prior to arrival, 101 received a dose after arrival at the maternity ward, and for 1 woman, the study database did not state where the dose was taken. In the NVP arm, 12 women were redosed, 2 of whom had delivered outside of Mulago Hospitals. Of the 281 women in the NVP arm who delivered at the hospital and who were not redosed, 166 women took their dose prior to arrival, 111 received their dose after arrival at maternity ward, 3 were not dosed, and for 1 woman no information was available (SCHARP, 2004c). Cord blood was obtained from the infants in both arms and specimens were frozen for later analysis.11 In the ZDV arm, specimens were obtained for 278 of 308 infants (90%), and NVP was detected in only 1 specimen (<1%). In the NVP arm, specimens were available for 275 of the 311 infants (88%). Of these, 3 were never dosed and had no detectable NVP. Of the remaining 272 specimens, 256 (94%) had detectable blood concentrations of NVP. In the 16 specimens obtained from the 18 patients12 randomized to placebo (89%), no NVP was detected. Only NVP concentration was measured as NVP has a long half-life (Jackson et al., 2003b). The investigators did not attempt to measure ZDV in the cord blood due to its short half-life which would make findings difficult to interpret. For example, if one tested for ZDV but none was found, one would not be able to determine whether the drug was not taken or if it was taken several hours prior to the blood draw. Without careful timing of the intake of ZDV dose in relation to sampling time, interpretation of ZDV blood levels would be meaningless. Studies have shown that the majority of women who received a single dose of NVP had detectable NVP levels 2 weeks after a single dose (Cressey et al., 2005; Jackson et al., 2005; Muro et al., 2004). Since ZDV has a short half-life of 1.1 hours in non-pregnant women and clearance is increased by 11 Cord blood was not available for all subjects for several reasons: some women delivered at home or elsewhere outside of the hospital, some cord blood samples were not obtained (especially in instances of emergency C-section), and cord blood was clotted before it was drawn. The investigators did tests on all available cord blood specimens (Jackson, 2005). 12 Of the 19 patients who were enrolled to the placebo arm, one was lost to follow-up. As a result, there are only 18 deliveries recorded (Jackson, 2005; Jackson et al., 2003a).
OCR for page 44
Review of the HIVNET 012 Perinatal HIV Prevention Study 47 to 65% in pregnant women (Mirochnick et al., 1998), inability to detect ZDV in the cord blood could indicate either that the mother had not taken the drug or that she took her ZDV dose more than 3 hours prior to the blood draw. As a result, the investigators did not attempt to measure ZDV in the cord blood. Two other measures of adherence were available to the committee. In the group of mothers assigned to NVP, HIV RNA concentration decreased by approximately one log one week after dosing, returning to baseline by the 6-week blood sample. There was no change in HIV RNA concentration in the mothers who took ZDV, which is to be expected given the short half-life of ZDV. Finally, HIV resistant mutations to NVP were only seen in the mothers who took NVP (Jackson and Guay, 2004). This finding is consistent with the DAIDS remonitoring evaluation which concluded that the assigned drugs were given to the appropriate participants (DAIDS, NIAID, 2003). Finding: The committee finds that the HIVNET 012 investigators used appropriate practices for packaging and distributing study drugs, so that the assigned drug was consistently provided to the appropriate mothers and their infants. Evidence from cord blood specimens indicates that participants achieved a high level of adherence to the NVP regimen. Though no direct evidence is available on blood levels of ZDV, the maternal reports of high levels of adherence to the treatment regimen, the fact that hospital personnel administered a substantial fraction of the ZDV regimen, and the absence of detectable levels of NVP in the blood of participants in the ZDV arm suggest that high levels of adherence were also achieved in the ZDV arm. The high level of adherence to study regimens indicates that the treatment arms formed an appropriate basis for assessing the efficacy and safety of the study regimens. REFERENCES Bardsley-Elliot A, Perry CM. 2000. Nevirapine: A review of its use in the prevention and treatment of paediatric HIV infection. Paediatric Drugs 2(5):373-407. CDC (Centers for Disease Control and Prevention), UNAIDS (Joint United Nations Programme on HIV/AIDS), NIH (National Institutes of Health), and ANRS (Agence Nationale de Recherche sur le SIDA). 1998. Joint Statement by the Centers for Disease Control and Prevention (CDC), the Joint United Nations Programme on HIV/AIDS (UNAIDS), the National Institutes of Health (NIH) and the Agence Nationale de Recherche sur le SIDA (ANRS), Following the Announcement by the Ministry of Public Health of Thailand and the CDC of Results from their Mother-to-Child Transmission Trial, February 18, 1998. Chamberlin J, Gustavson SA, Hensley M, Lander. 2002. Site Visit Report, Kampala, Uganda, February 18-28, 2002. Westat Corporation, MD.
OCR for page 45
Review of the HIVNET 012 Perinatal HIV Prevention Study Connor EM, Sperling RS, Gelber R, Kiselev P, Scott G, O’Sullivan MJ, VanDyke R, Bey M, Shearer W, Jacobson RL, Jimenez E, O’Neill E, Bazin B, Delfraissy J, Culnane M, Coombs R, Elkins M, Moye J, Stratton P, Balsley J, for The Pediatric AIDS Clinical Trials Gropu Protocol 076 Study Group. 1994. Reduction of maternal-infant transmission of human immunodeficiency virus type 1 with zidovudine treatment. Pediatric AIDS Clinical Trials Group Protocol 076 Study Group. New England Journal of Medicine 331(18): 1173-1180. Cressey TR, Jourdain G, Lallemant MJ, Kunkeaw S, Jackson JB, Musoke P, Capparelli E, Mirochnick M. 2005. Persistence of nevirapine exposure during the postpartum period after intrapartum single-dose nevirapine in addition to zidovudine prophylaxis for the prevention of mother-to-child transmission of HIV-1. Journal of Acquired Immune Deficiency Syndromes 38(3):283-288. DAIDS, NIAID (Division of AIDS, National Institute of Allergy and Infectious Diseases). 2002. DAIDS Summary of March 19 Westat Debriefing. DAIDS, NIAID. 2003. HIVNET 012 Monitoring Report. Donnell D. 2004, November 24. Mem to Alicia Gable: Listings of randomization scheme. Fiscus SA, Adimora AA, Schoenbach VJ, Lim W, McKinney R, Rupar D, Kenny J, Woods C, Wilfert C. 1996. Perinatal HIV infection and the effect of zidovudine therapy on transmission in rural and urban counties. Journal of the American Medical Association 275(19):1483-1488. Fleming T. 2004. Powerpoint Presentation to Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study on September 30, 2004: HIVNET 012 Institute of Medicine Review. Guay LA, Musoke P, Fleming T, Bagenda D, Allen M, Nakabiito C, Sherman J, Bakaki P, Ducar C, Deseyve M, Emel L, Mirochnick M, Fowler MG, Mofenson L, Miotti P, Dransfield K, Bray D, Mmiro F, Jackson JB. 1999. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 354(9181): 795-802. Guay L, Mmiro F, Musoke P, Fanning M. 2002. HIVNET 012: Phase IIB Trial to Evaluate the Efficacy of Oral Nevirapine and the Efficacy of Oral AZT in Infants Born to HIV-Infected Mothers in Uganda for Prevention of Vertical HIV Transmission. Description of Study Procedures. Guay L. March 27, 2005a. Re: Pharmacy logs. E-mail to Guay L, Gable A on behalf of the IOM’s Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study. Guay L. March 28, 2005b. Re: Minor question. E-mail to Jackson JB, Gable A on behalf of the IOM’s Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study. Guay L. March 29, 2005c. Re: Additional question. E-mail to Guay L, Gable A on behalf of the IOM’s Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study. Guay L. March 29, 2005d. Re: Verification on drug management procedures. E-mail to Guay L, Gable A on behalf of the IOM’s Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study. HIVNET 012 Investigators. 1998. HIVNET 012 Proposed Interim Plan. HIVNET 012 Investigators. 2002. Summary Sheet of Major Inaccuracies in the Westat Report. Jackson B. March 28, 2005. Re: 012 cord blood nevirapine levels. E-mail to Alicia Gable. Jackson B, Guay L. 2004. Powerpoint Presentation to Committee on Reviewing the HIVNET 012 Perinatal HIV Prevention Study on September 30, 2004: HIVNET 012: Safety and Efficacy at 18 months.
OCR for page 46
Review of the HIVNET 012 Perinatal HIV Prevention Study Jackson B, Parsons T, Musoke P, Nakabiito C, Donnell D, Fleming T, Mirochnick M, Mofenson L, Fowler M, Mmiro F, Guay L, and HIVNET 012. 2005. Association of Cord Blood Nevirapine with Self-Reported Timing of Dose and HIV-1 Transmission in the HIVNET 012 Study, Abstract #71. [Online] Available at http://www.retroconference.org/org/2005/cd/Abstracts/25816.htm. Accessed March 7, 2005. Jackson JB, Musoke P, Fleming T, Guay LA, Bagenda D, Allen M, Nakabiito C, Sherman J, Bakaki P, Owor M, Ducar C, Deseyve M, Mwatha A, Emel L, Duefield C, Mirochnick M, Fowler MG, Mofenson L, Miotti P, Gigliotti M, Bray D, Mmiro F. 2003a. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: 18-month follow-up of the HIVNET 012 randomised trial. Lancet 362(9387):859-868. Jackson B, Parsons T, Musoke P, Nakabiito C, Donnell D, Fleming T, Mirochnick M, Mofenson L, Fowler MG, Mmiro F, Guay L. 2003b. Strong correlation of cord blood nevirapine level with self reported timing of dose and HIV-1 transmission in the HIVNET 012 study. Jackson B, Mmiro F, Guay L, Musoke P, Fowler MG. 1997. Protocol HIVNET 012: A Phase III Placebo-Controlled Trial to Determine the Efficacy of Oral AZT and the Efficacy of Oral Nevirapine for the Prevention of Vertical Transmission of HIV-1 Infection in Pregnant Ugandan Women and Their Neonates. Kaplan EL, Meier P. 1958. Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53(282):457-481. Mirochnick M, Fenton T, Gagnier P, Pav J, Gwynne M, Siminski S, Sperling RS, Beckerman K, Jimenez E, Yogev R, Spector SA, Sullivan JL. 1998. Pharmacokinetics of nevirapine in human immunodeficiency virus type 1-infected pregnant women and their neonates. Pediatric AIDS Clinical Trials Group Protocol 250 Team. Journal of Infectious Diseases 178(2):368-374. Muro E, Droste J, ter Hofstede H, Bosch M, Dolmans W, Burger D. 2004. Nevirapine plasma concentrations are still detectable after more than 2 weeks in the majority of women receiving single-dose NVP: Implications for intervention studies. Conference on Retroviruses & Opportunistic Infections, Abstract No. 891. Peiperl L. 2002. Antiretroviral Treatments to Reduce Mother-to-Child Transmission of HIV. HIV InSite, University of California San Francisco. [Online] Available at http://hivinsite.ucsf.edu/InSite.jsp?page-home-00-00&doc=3098/0098#19/ Accessed March 21, 2005. Petra Study Team. 2002. Efficacy of three short-course regimens of zidovudine and lamivudine in preventing early and late transmission of HIV-1 from mother to child in Tanzania, South Africa, and Uganda (Petra study): A randomised, double-blind, placebo-controlled trial. Lancet 359(9313):1178-1186. SCHARP (Statistical Center for HIV/AIDS Research and Prevention). October 25, 2004a. Memo: Implementation of randomization in HIVNET 012. SCHARP. 2004b. File: t_m_delivery.pdf. SCHARP. 2004c. File: t_m_dosing.pdf. Shaffer N, Chuachoowong R, Mock P, Bhadrakom C, Siriwasin W, Young N, Chotpitayasunondh T, Chearskul S, Roongpisuthipong A, Chinayon P, Karon J, Mastro T, Simonds R. 1999. Short-course zidovudine for perinatal HIV-1 transmission in Bangkok, Thailand: A randomised controlled trial. Bangkok Collaborative Perinatal HIV Transmission Study Group. Lancet 353(9155):773-780. Uganda AIDS Commission Secretariat. 2002. HIV/AIDS in Uganda: The Epidemic and the Response. [Online] http://www.aidsuganda.org/pdf/hiv_aids_impact.pdf. Accessed March 7, 2005.
OCR for page 47
Review of the HIVNET 012 Perinatal HIV Prevention Study UNDP (United Nations Development Programme). 2002. Human Development Report 2002. Chapter 7: Leading global health crises and challenges. USAID (United States Agency for International Development). 2005. Uganda. [Online] Available at http://www.usaid.gov/our_work/global_health/aids/Countries/africa/uganda.html. Accessed May 1, 2005. Working Group on Mother-To-Infant Transmission of HIV. 1995. Rates of mother-to-infant transmission of HIV-1 in Africa, America, and Europe: Results from 13 perinatal studies. Journal of AIDS Retrovirology 8:506-510.
Representative terms from entire chapter: