Randomized clinical trial protocol: Acceptability and feasibility of combination treatment for cervical precancer among South African women living with HIV (ACT 2).

Our primary objective was to determine the acceptability and feasibility (safety, tolerability, adherence, and retention) of combination treatment of CIN2/3 (i.e., surgical excision followed by adjuvant 5FU) among WLWH. Our secondary objectives were related to the efficacy of combination treatment for CIN2/3. At week 24, we assessed (a) regression of cervical disease to cervical intraepithelial neoplasia grade 1 (CIN1) or normal histology and (b) clearance of the high-risk HPV (hrHPV) genotype(s) detected in baseline cervical ThinPrep samples. ACT 2 was a double-blind, randomized, placebo-controlled feasibility trial. Briefly, WLWH with CIN2/3 confirmed on LEEP were randomly assigned (1:1) to receive 8 doses of intravaginal 5FU cream or placebo cream (once every 2 weeks). Participants were followed for 24 weeks to assess acceptability and feasibility (safety, tolerability, adherence, and retention) of the combined treatment approach ( Figure 1 ). The study was designed as a feasibility trial to assess acceptability, safety, adherence, and retention prior to a fully powered effectiveness trial. Although secondary efficacy outcomes were explored, the study was not powered for efficacy comparisons. Participant enrollment and follow-up have been completed. Data cleaning and statistical analyses are currently ongoing. The randomization schedule was generated by the study statistician in SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina) using a random number generator and a blocked randomization design. The block size(s) were kept concealed from the study team throughout the trial. Randomization IDs were assigned in a concealed manner using a central, web-based REDCap utility. An unblinded study pharmacist used the randomization schedule to provide the assigned 5FU or placebo treatment. This protocol was reviewed and approved by the University of Witwatersrand Human Research Ethics Committee (201122) and the University of North Carolina Institutional Review Board (20-3565). The trial was publicly registered on the South African National Clinical Trials Register and with the U.S. National Library of Medicine at ClinicalTrials.gov ( NCT05413811 ). A SPIRIT checklist is provided in the Supplementary Materials ( Supplemental File 1 ). ACT 2 was conducted in Johannesburg, South Africa. WLWH were recruited from the cervical cancer prevention program at an HIV clinic located in a large public-sector teaching hospital. South Africa is home to more than 5 million WLWH and has the world’s largest public-sector HIV treatment programs [ 21 ]. Despite substantial local and international investments in HIV, cervical cancer remains a leading cause of cancer deaths in the country [ 22 - 25 ]. Our trial opened in March 2023 under protocol version 3, which included the following inclusion criteria: women with (a) confirmed HIV-1 infection; (b) age ≥ 18 years; (c) on ART for at least 90 days prior to enrollment; and (d) biopsy-proven CIN2/3 within the preceding 120 days. To be eligible, women were also willing and able to provide written informed consent and to use dual contraception during the study. An amendment to the inclusion criteria was made in April 2023 and under protocol v4 ( Supplemental File 2 ) women who meet the following eligibility criteria were included: (a) confirmed HIV-1 infection; (b) age ≥ 18 years; (c) on ART for at least 60 days prior to enrollment; and (d) eligible for LEEP (i.e., those with a cervical biopsy demonstrating CIN2/3 or HSIL cytology on Pap smear within the preceding 12 months). Under both protocol v3 and v4, women were excluded from participation if they (a) were pregnant, breastfeeding, or intend to become pregnant within 180 days of enrollment; (b) had an active sexually transmitted infection (STI); (c) had a surgically absent cervix; (d) had a history of anogenital cancer or a biopsy suspicious for cervical cancer; or (e) had a medical comorbidity that would interfere with study participation. LEEP was provided in accordance with the standard of care. The 5FU (Efudix) cream was supplied by Mylan, South Africa and held in our research pharmacy. The placebo cream was an inert emollient mimicking the consistency and coloring of the 5FU cream. Vaginal applicators pre-filled with 2g of 5% 5FU or placebo cream were distributed to participants to self-apply the cream intravaginally once every 2 weeks for a total of 8 doses. Educational materials for the trial were co-designed with participants during our formative study [ 20 ]. These materials included step-by-step instructions on how to use the 5FU/placebo cream ( Supplemental File 3 ). WLWH who had a clinical indication for LEEP (i.e., CIN2/3 on cervical biopsy or HSIL cytology on Pap smear) were approached to participate by a member of the study team. Those wishing to participate provided written informed consent and were assessed for eligibility, which included a review of their medical history and medications, urine pregnancy testing, pelvic examination, STI testing for N. gonorrhoeae, C. trachomatis, T. vaginalis , and screening for syphilis ( Table 1 ). At their screening visit (week 0), women underwent colposcopy and LEEP to confirm eligibility for enrollment. Additional baseline specimens included a cervical ThinPrep sample for HPV testing and blood samples for CD4+ cell count and HIV-1 plasma viral load. A cervicovaginal lavage (CVL) sample and vaginal swab were also collected for future HIV viral load, cytokine, HPV, and microbiome testing as outlined below. Women received their LEEP histology results at week 2. At week 4 (enrollment visit), women with CIN2/3 confirmed on baseline LEEP histology were randomly assigned (1:1) to receive 8 doses of intravaginal 5FU or placebo once every 2 weeks. They received step-by-step instructions for 5FU/placebo cream use and a demonstration using a pelvic model. Women applied the first dose of cream in the study clinic and were observed for immediate adverse events (AEs). They were then given pre-filled vaginal applicators and asked to apply the 5FU/placebo cream at home once every 2 weeks (8 doses total, with additional at-home doses at weeks 6, 8, 10, 12, 14, 16, and 18). We provided women with dose and symptom diaries ( Supplemental File 4 ), condoms, and urine pregnancy tests to be used prior to applying the 5FU/placebo at home. Participants were also counselled to avoid sexual intercourse for 48 hours after applying the cream. Study procedures performed at week 4 included pregnancy testing, pelvic examination, and colposcopy. We also collected a ThinPrep sample for HPV testing, as well as CVL and vaginal swab samples for future testing. Regardless of study arm, women returned for additional study visits at weeks 6, 10, 18, and 24. Study procedures conducted during the visits at weeks 6, 10, and 18 included pregnancy testing, pelvic examination, and colposcopy to assess possible AEs. At each of these visits, we collected ThinPrep, CVL, and a vaginal swab for future testing. Adherence to the study intervention was reinforced at weeks 6, 10, and 18. Women also received text and phone call reminders of both their study cream dosing schedule and their study visit schedule. At week 24 (exit visit), women underwent pregnancy testing, colposcopy, cervical biopsy, and, if clinically indicated, endocervical curettage (ECC). Repeat STI testing was performed at this visit, and we also collected liquid-based cytology, ThinPrep for HPV testing, and blood samples for CD4+ and HIV-1 plasma viral load testing. As at prior visits, CVL and vaginal swab samples were stored for future testing. At each study visit, participants received reimbursement for their time and travel costs. Women with persistent CIN2/3 at endline were followed up in our routine care clinic and offered further management with repeat LEEP or, if clinically indicated, referral for hysterectomy. Our research team performed all gynecologic procedures and was blinded to study arm assignment. At each study visit, women underwent pelvic examination of the vulva, vagina, cervix, and perineum as part of the clinical assessment for STIs and AEs. During colposcopy, 5% acetic acid was applied to the cervix and digital images of the cervical transformation zone were obtained for quality assurance. The baseline LEEP (screening visit) was performed in accordance with the standard of care. After a paracervical nerve block with injected lidocaine, a wire loop was used to excise abnormal-appearing tissue in the cervical transformation zone. Excised tissue samples were sent to our clinical trials laboratory for histologic evaluation and, if clinically indicated, immunohistochemical (IHC) staining. Similarly, at week 24, all participants underwent colposcopically-directed, 4-quadrant biopsy and ECC to adequately evaluate the cervical transformation zone [ 26 , 27 ]. Once again, tissue samples were sent for histologic evaluation and IHC staining, if clinically indicated. Liquid-based cytology (LBC) specimens (Pap smears) and cervical specimens for HPV testing/storage were both collected using a Cervex-Brush (Rovers Medical Devices, Oss, Netherlands), while vaginal swab samples were collected using a polyester-tipped swab. CVL was performed by aiming a continuous stream of 10mL of phosphate-buffered saline directly at and into the cervical opening to bathe the endocervix and ectocervix. The fluid pool was then aspirated from the posterior fornix. All histologic and cytologic specimens were processed and reviewed at our clinical trials laboratory in Johannesburg, the Bio-Analytical Research Corporation (BARC). Central review of both histology and cytology specimens was performed by pathologists blinded to the study arm assignment. HPV testing was performed using the GeneXpert system (Cepheid, Sunnyvale, California), which employs a cartridge-based real-time PCR system and reports five separate results: (a) HPV16, (b) HPV18/45, (c) HPV 31/33/35/52/58, (d) HPV51/59, (e) HPV39/68/56/66 [ 28 , 29 ]. STI testing for N. gonorrhoeae , C. trachomatis , and T. vaginalis using vaginal samples was also performed using the GeneXpert system [ 30 - 33 ]. Rapid testing for syphilis was performed using Syphilis BD Macro-Vue (Abbott Laboratories, Abbott Park, Illinois), a screening assay for T. pallidum antibodies [ 34 ]. Women diagnosed with STIs received appropriate treatment, and baseline LEEP procedures were not delayed in accordance with the local standard of care. Their partners also received empiric STI treatment. CD4+ T-cell counts were determined by flow cytometry performed on whole blood samples using the FACSCanto Clinical Flow Cytometry System (BD Biosciences, Franklin Lakes, New Jersey). Plasma HIV viral load was measured using the Abbott RealTime HIV-1 Assay (Abbott m2000 system, Abbott Laboratories, Abbott Park, Illinois). Residual plasma and ThinPrep samples were aliquoted and stored at −70°C (or below) for future HPV, DNA methylation, microbiome, and related biomarker analyses. CVL samples were aliquoted and stored at −70°C for future HIV viral load and cytokine testing. Vaginal samples were also aliquoted and stored at −70°C for future HPV and microbiome analyses. Lastly, FFPE tissue specimens were stored at ambient temperature for future HPV, DNA methylation, and related biomarker analyses. Participants used dose and symptom diaries to self-report data on cream use and side effects, respectively. To allow comparison to our previously published trial [ 19 ], symptom diaries were designed to be open ended and participants were asked to record symptoms as they occurred rather than completing daily entries. During study visits at weeks 4, 6, 10, 18, and 24, participants also underwent pelvic examination and colposcopy. All AEs noted during these visits were graded using NCI’s Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 and the DAIDS Female Genital Grading Table for Use in Microbicide Studies v1.0. Dose-limiting toxicities (DLT) were defined as (a) any Grade 2 or greater AE that was possibly, probably, or definitely related to the study drug and resulted in a reduction to the number of doses or discontinuation of the study drug, or (b) any Grade 1 genital AE (e.g., blisters, ulcerations, or pustules) that is possibly, probably, or definitely related to the study drug and resulted in a reduction to the number of doses or discontinuation of the study drug. Women were asked to return all used and unused 5FU/placebo applicators at follow-up visits, and the applicators were counted and tested for evidence of vaginal insertion using an ultraviolet light. The combination of applicator counts [ 35 ], ultraviolet inspection (UVI) [ 36 ], and self-report [ 37 ] was intended to limit the potential for recall and social desirability bias often introduced when adherence data collection occurs at study visits rather than in real time [ 38 - 40 ]. Our primary outcomes were acceptability and feasibility. Acceptability was measured at weeks 10 and 24 using a questionnaire with 7 Likert scale items. Feasibility (safety, tolerability, adherence, and retention) was assessed between weeks 4 and 24. Our safety endpoint is the number and percentage of women in each study arm experiencing a grade 2 or higher AE, or any grade 1 genital AE (e.g., blisters, ulcerations, or pustules), that is possibly, probably, or definitely related to the study intervention. Our tolerability endpoint is the number and percentage of women in each study arm unable or unwilling to apply at least 4 of 8 doses (50%) of the 5FU/placebo cream due to a DLT. For our adherence endpoint, 3 sources of adherence data (applicator counts, UVI, and self-report) were collected, and the corresponding adherence rates will be reported. Intra-class correlation (ICC) will be used to estimate within-woman agreement between the three data sources [ 41 ]. UVI will be considered the primary adherence measure. Finally, our retention endpoint is the number and percentage of women in each study arm who are retained in study follow-up over the 24-week study duration. Our secondary outcomes were related to the efficacy of combination treatment for CIN2/3. At week 24, we will assess (a) regression of cervical disease to CIN1 or normal histology and (b) clearance of the hrHPV genotype(s) detected at baseline (week 0). Descriptive statistics will be used to (a) delineate the overall study sample that was recruited from a population of WLWH who are on ART and have a clinical indication for LEEP, and (b) assess for clinically meaningful imbalances between the randomly assigned study arms (see Supplemental File 5 for the Statistical Analysis Plan). For the primary acceptability and feasibility endpoints, our main analyses will include all participants who are eligible, enrolled, randomized, and who receive at least one dose of 5FU or placebo (i.e., a modified intention-to-treat [ITT] approach). Additionally, per-protocol and ITT analyses will be conducted as detailed in Supplemental File 5 . We will describe the responses to each Likert scale item individually. To compare 5FU vs. placebo, an acceptability summary score will be calculated as a percentage (0 to 100%) of 0 to 28 possible Likert scale points, and the distribution of scores will be compared between randomization arms at weeks 10 and 24, separately, using a Wilcoxon rank-sum test and descriptive statistics. We will then estimate the proportion of women who reported 80% or higher acceptability summary score in each randomization arm, and a corresponding binomial 95% confidence interval (CI) will be constructed for each study arm. Lastly, we will estimate the difference in proportions (5FU vs. Placebo) with a corresponding 95% CI. We will report the number and percentage of women in each randomization arm experiencing a safety event. A Kaplan Meier (KM) estimate of safety event probability and a corresponding 95% CI will be constructed for each randomization arm at study week 24, and safety will be compared between the randomization arms using an estimated risk difference (5FU - placebo) with a corresponding 95% CI. The time origin is randomization (week 4). We will report the number and percentage of women in each randomization arm unable or unwilling to apply at least 4 of 8 doses of the 5FU/placebo cream due to a DLT. A KM estimate of tolerability event probability and a corresponding 95% CI will be constructed for each randomization arm at study week 18, and tolerability will be compared between the randomization arms using a risk difference (5FU - Placebo) with a corresponding 95% CI. For each of the three adherence measures, we will derive a binary endpoint classifying each woman as adherent if she administers at least 6 of 8 doses of 5FU/placebo. Within each randomization arm, we will report the number and percentage of women achieving this adherence threshold with a corresponding binomial 95% CI. The percentage of women who achieve adherent dosing will be compared between the randomization arms using an estimated difference in proportions with a corresponding 95% CI. We will report the number and percentage of women in each randomization arm who are retained in study follow-up at each visit over the 24-week study duration. Week 24 retention will be estimated within each randomization arm with a corresponding binomial 95% CI. Week 24 retention will be compared between the 5FU and placebo arms using an estimated difference in proportions and corresponding 95% CI [ 42 ]. All eligible participants will have CIN2/3 (confirmed by LEEP histology) measured at study week 0 (baseline for this analysis). We will report the number and percentage of women in each randomization arm who regress (improve) to CIN1 or normal at week 24 with a corresponding binomial 95% CI. Regression of cervical disease will be compared between the randomization arms using an estimated difference in proportions and corresponding 95% CI. We will report the number and percentage of women in each randomization arm who demonstrate genotype-specific hrHPV clearance, defined as having none of the hrHPV types at week 24 that were originally present at baseline (week 0). A corresponding binomial 95% CI will be constructed. This composite measure of hrHPV type-specific clearance will be compared between the randomization arms using an estimated difference in proportions and corresponding 95% CI. Because the GeneXpert HPV assay reports grouped hrHPV genotypes rather than individual types (except for HPV16), clearance will be assessed at the level of these genotype groups. We enrolled 180 participants for 1:1 randomization to 5FU or placebo cream. A priori , we anticipate grade 2 or higher safety AEs and tolerability events to be uncommon [ 19 ], thus our sample size calculations first addressed the acceptability and adherence endpoints; then we calculated anticipated precision for the safety and tolerability endpoints. Precision and power calculations for secondary outcomes are included in the Supplemental File 4 . We assumed ≥90% of women will be retained in study follow-up, thus throughout our precision and power calculations we assume 10% attrition over the course of the study. To estimate the proportions of women reporting an 80% or higher acceptability score, overall and in each arm, we calculated a binomial 95% CI. We assumed 80% of women will achieve this acceptability threshold at weeks 10 and 24 and thus enrolling 180 women will provide an anticipated margin of error of ±8.7% within arm ( Table 2 ). Likewise, to estimate the proportions achieving adherence overall and within each arm, we calculated a binomial 95% CI. We assumed that 80% of women will achieve adequate adherence (i.e., 6 of 8 doses). To measure this proportion with a margin of error <10% (anticipated ±8.7%) within arm, we will enroll 180 women. If we observe 90% retention over 24 weeks, the 95% CI precision for retention will be (0.85, 0.94) pooled over arms and (0.82, 0.95) within arm. Analogously, we calculated approximate precision and binomial 95% CIs for safety and tolerability with a low proportion of women expected to experience these outcomes. For example, for an observed proportion of 10% experiencing a safety event, we have approximate precision of ±6.5%, separately within each arm. If the observed proportion of women experiencing a tolerability event is 5%, we have approximate precision of ±4.7%, within each arm. Regarding statistical power, if the true probability of experiencing a primary safety event is 10% in the placebo arm, then 90 enrolled women per arm will provide us with 90% power to detect a difference in probabilities of 20% or greater ( Figure 2 ). Participation in the trial was voluntary. Prior to screening, study personnel trained in Good Clinical Practice (GCP) obtained written informed consent from all participants. The study procedures, risks, and benefits were discussed, and women could ask questions prior to providing consent. The informed consent forms were available in English, Sesotho, and Zulu (the three most common languages in Johannesburg). This was a moderate-risk trial. Mild side effects of intravaginal 5FU include symptoms such as burning, redness, or vaginal discharge. Serious side effects include pain, bleeding, or vaginal ulcers. The treatment regimen (5FU once every 2 weeks) has been previously studied in other settings where women reported mild, grade 1 AEs related to the cream use (e.g., irritation, spotting). However, none of the women enrolled in previous studies using 5FU once every 2 weeks reported grade 2 or higher AEs. Further, none of the AEs experienced by women in prior studies were reported to interfere with usual activities [ 19 ]. Additional risks associated with study participation include the known risks of clinical procedures such as cervical biopsy and LEEP. Cervical biopsies are minimal risk, but women may experience brief cramping and discomfort, and, occasionally, a small amount of bleeding. LEEPs require the administration of local anesthetic and may also include cramping and discomfort during the procedure. Bleeding from the LEEP site is controlled with either Monsel’s solution or silver nitrate. LEEPs are also associated with a small risk of fever, infection, pelvic pain, and possible preterm delivery in future pregnancies. All LEEPs performed during the study are clinically indicated. Participants did not incur any additional risks beyond standard clinical care. Collecting lower genital tract samples, such as cytology, cervical swabs, vaginal swabs, CVL, and blood are all minimal risk procedures. Finally, the loss of confidentiality is associated with data collection and storage. At each step in the study, we protected participant privacy and confidentiality to reduce these risks. All study procedures were conducted in private. Data were stored in secured, locked cabinets and on password-protected computers. There were no direct benefits of participation in this trial. However, women may have benefited from enhanced health education and close clinical monitoring. Additionally, the knowledge obtained may help to guide future research on and implementation of combination treatment for CIN2/3 in WLWH. The trial was externally monitored and reviewed annually by an independent Data Safety and Monitoring Committee (DSMC) convened by the Lineberger Comprehensive Cancer Center at the University of North Carolina at Chapel Hill. The DSMC assessed recruitment, accrual, retention, and safety.

Although the introduction of ART has reduced the incidence of some cancers (e.g., Kaposi’s sarcoma, non-Hodgkin lymphoma) among individuals living with HIV, the role of ART in reducing the incidence of cervical cancer and its precursors remains unclear [ 43 , 44 ]. As increasing numbers of WLWH initiate ART in LMICs such as South Africa, longer life expectancies are anticipated to lead to a rising burden of chronic diseases – including cervical cancer – in this population [ 45 , 46 ]. WLWH in LMICs face additional burdens associated with limited access to cervical cancer prevention measures, such as HPV vaccination, screening, and surgical treatment of cervical precancer. In preparation for a large-scale effectiveness trial of LEEP combined with intravaginal 5FU for CIN2/3 in WLWH, we conducted a feasibility trial of this combination treatment approach. Given the low cure rates achieved with surgery alone among WLWH with CIN2/3 [ 12 - 14 ], a treatment strategy that combines surgery (excision or ablation) with an adjuvant therapy, such as a self-administered intravaginal medication, would be highly impactful on cervical disease persistence/recurrence and improve long-term cancer risk among WLWH.

Cervical cancer remains the second most common cancer among women worldwide, and over 85% of the global burden of this disease occurs in the developing world. Women living with human immunodeficiency virus (HIV; WLWH) are at increased risk of human papillomavirus (HPV) infection, with the prevalence of HPV in this group as high as 45-90% [ 1 - 3 ]. Although cervical cancer is preventable, cytologic abnormalities, precancer (high-grade cervical intraepithelial neoplasia [CIN2/3]), and invasive cancer also occur more frequently in WLWH [ 2 , 4 ]. Current management of CIN2/3 is based on surgical therapy alone, which can be excisional (e.g., cold knife conization, loop electrosurgical excision procedure [LEEP]) or ablative (e.g., cryotherapy, thermocoagulation, laser ablation) [ 5 ]. Although surgical therapy is highly successful in women without HIV [ 6 , 7 ], cure rates for CIN2/3 are significantly lower in WLWH [ 8 - 11 ]. Three African randomized trials confirm a high risk (19-33%) of persistent/recurrent disease following surgical treatment for CIN2/3 in WLWH, including those virologically suppressed on antiretroviral therapy (ART) [ 12 - 14 ]. These findings are consistent with a prior U.S. study in which treatment failure occurred in 55% of WLWH with CIN2/3 over a 6-year period [ 9 ]. By comparison, LEEP treatment failure occurs in only 5–10% of women without HIV [ 6 ]. WLWH could therefore benefit from effective adjuvant therapy following surgery for CIN2/3. We hypothesized that topical 5-fluorouracil, a widely available generic drug, can be repurposed as an adjuvant treatment for CIN2/3 and self-administered after surgical excision or ablation. Multiple observational studies support the efficacy of topical 5-fluorouracil (5FU) for treatment of HPV-related diseases (e.g., genital warts, vulvar and vaginal precancer) [ 15 - 17 ]. Two prior U.S. randomized trials also support the use of intravaginal 5FU as adjuvant therapy for cervical disease in WLWH [ 18 ] and primary treatment for CIN2 in women without HIV [ 19 ]. As adjuvant 5FU has not previously been studied in a low- and middle-income country (LMIC) setting, this project was conducted in two phases: a formative study [ 20 ] followed by a feasibility trial. Here, we outline the protocol for our feasibility trial known as “ACT 2.”