Study design

The prospective, randomized, three-arm, partially blind, noninferiority trial to evaluate three techniques for treating the cervical TZ in a screen-and-treat setting was launched in August 2017 in Lusaka, Zambia. TA using a portable, battery-driven device manufactured by Liger Medical was compared to cryotherapy and LLETZ in this trial; all procedures were performed on an outpatient basis. The study was implemented in two phases: a pilot phase and an extended phase. Before launching the pilot study, the study team extensively interacted with the engineers at Liger Medical to help them improve a prototype version of the TA device. The improved device was used for the trial.

In the pilot phase, 250 eligible women were randomly recruited to each of the three treatment arms and were followed up after 6 months. The safety and efficacy data from the pilot phase were reviewed by a data safety monitoring board (DSMB). As the efficacy of TA using the new device was not inferior to the other two techniques and as TA had a safety profile similar to the other two techniques, the DSMB allowed continuation of the trial to the extended phase with the following suggestions: (1) to reduce the application of the TA probe from 45 to 30 s; (2) to extend the follow-up time from 6 to 12 months; and (3) to carry out a formal sample size estimation based on the efficacy data of the pilot phase.Sample size was recalculated to recruit additional women to each arm in the extended phase of the trial. The women recruited to the pilot study were included to achieve the required sample size. Recruitment to the extended phase was initiated on 2 January 2019.

Study target population

The study was nested in the ongoing cervical screening program in Zambia. Screening-eligible women attending primary health clinics (Chawama Level 1 Hospital, Chipata Level 1 Hospital, Kanyama Level 1 Hospital, Levy Mwanawasa Hospital, Matero Clinic) and the University Teaching Hospital in Lusaka, Zambia were the target population for our study. Many of these women attend the clinics to receive ART. All women of an eligible age attending the clinics are routinely counseled to participate in the Zambian national cervical screening program. The clinics follow the Zambian national protocol, which recommends that women aged 25–49 should be tested using VIA and that women who are VIA positive should be offered immediate treatment (screen-and-treat approach). Women aged up to 59 years who have never been screened before are also offered VIA. Both VIA screening and treatment are performed by nurses trained as part of the cervical cancer prevention 2-week training program, including didactic and hands-on clinical mentoring. As part of quality control, trained nurses are encouraged to use distance consultation of local doctors or gynecologists using the electronic transfer of digital photographs of the screened cervix to make appropriate treatment or referral decisions8.

Women willing to undergo screening are examined by a trained nurse at the clinic to perform VIA. Women who are VIA positive are further assessed for eligibility for ablative treatment using the following criteria: the TZ is type 1 (the entire squamocolumnar junction is visible at the external os or ectocervix); the acetowhite area on the TZ does not occupy more than 75% of the ectocervix; the acetowhite area does not extend to the endocervix or vagina; and there is no suspicion of invasive cervical cancer on visual examination.Women who were VIA positive and also eligible for ablative treatment were approached to participate in our study.

Selection and recruitment of study participants

Women who were VIA positive who fulfilled the criteria for suitability for ablative treatment and showed interest to participate in the study were further assessed by a trained social worker for the following inclusion and exclusion criteria before she could initiate the informed consent process. The inclusion criteria included: age between 25 and 59 years; positive VIA test; eligible for ablative treatment based on the criteria described above; additionally, the TZ could be covered by a single application of the largest cryotherapy probe. Exclusion criteria included: pregnant at the time of recruitment; and not in a position to provide voluntary informed consent because of mental illness or other medical conditions.

During the consent process, the social worker explained the objectives, treatment procedures, follow-up requirements, benefits and harms of participating in the study and responded to any queries. Each woman willing to participate provided written consent before her recruitment to the study was finalized.

Randomization and masking

The recruited women were randomly assigned to one of the three treatment groups in 1:1:1 ratio. Before the study was launched, a randomization list was generated by the IARC data manager using the random function in Excel to allocate a treatment to each participant’s ID. This list was not shared with investigators in Zambia. An IARC study coordinator inserted the treatment allocation printed on the participant’s barcode sheets with a specific color code in opaque sealed envelopes with only the randomization sequence written on the envelopes. These envelopes were mailed to the site coordinator in Zambia, who arranged them serially according to the sequence number. The social worker telephoned the site coordinator after confirming participation of an eligible woman to know her treatment allocation. The social worker conveyed the treatment allocation to the treating nurse. Once allocated to a particular treatment arm, the treatment modality could not be changed. Blinding of either the participant or the treating nurse was not feasible.

Sample collection for HPV testing

The cervical sample from every woman undergoing initial VIA before application of acetic acid was collected using a cytobrush. The samples obtained from women recruited to the study were sent to the laboratory at the University Teaching Hospital, Lusaka for high-risk HPV testing. The test results were made available on a later date and did not change clinical management.

Treatment procedures

Treatment was provided on the day of randomization at the recruiting clinic by nurses trained to perform all three treatment techniques. Some of the more difficult LLETZ procedures were performed by gynecologists at the clinic. Five percent acetic acid was applied to the cervix to delineate the lesion and the TZ before treatment. The cervix was infiltrated with local anesthetic before LLETZ. TA or cryotherapy was performed without any anesthesia.

A Liger TA device was used to perform TA. The probe was heated to 100 °C and applied to the cervix for 45 s in the pilot phase to treat the initial 250 women recruited to the arm. Treatment time was reduced to 30 s for all women recruited to the TA arm in the extended phase. Overlapping applications were made to cover a large TZ.

Cryotherapy was performed using a cycle of 3-min freezing, 5-min thawing and 3 min of repeat freezing. The probe was applied only once. The refrigerant gas used was nitrous oxide.

For the LLETZ procedure, an appropriately sized loop was selected to remove the entire TZ, if possible. However, multiple passes were used when required. Bleeding after excision was controlled by cauterizing the base of the excised area with a ball diathermy electrode followed by application of Monsel’s paste.

Posttreatment procedures

Once the procedure was completed, the woman was assessed for any side effects like pain, bleeding and vaginal injury by the treating nurse. No antibiotics were given after treatment.

A social worker explained to the woman that she would be having vaginal discharge for 1–2 weeks, which might be blood-stained at times. She was advised to avoid penetrative vaginal sex for 2 months and report to the facility or telephone the study coordinator if she had any of the following symptoms within 2 weeks of treatment: excessive vaginal bleeding, especially with the passage of clots; foul-smelling excessive vaginal discharge; and pain in the lower abdomen with a temperature exceeding 100 °F.

The social worker also explained the follow-up plan. She informed the woman about a telephone interview to be conducted 2 weeks after treatment to check for any side effects and complications and to receive feedback from the woman regarding how satisfied she was with treatment. The social worker also responded to any queries the woman might have.

At the end of the visit, the social worker conducted the patient-reported outcome survey. The woman was asked to score the level of discomfort and pain she experienced during on immediately after treatment on a visual analog scale ranging from 1 (no pain at all) to 9 (pain was so severe that she wanted the treatment to be stopped). She was also asked to rate her overall satisfaction with the service and experience on a visual analog scale of 1 (highly satisfied) to 9 (least satisfied). The woman was advised to return to the clinic for a checkup after 6 months in the pilot phase and after 12 months in the extended phase.

Quality assurance of VIA and treatment

The nurses routinely collected and stored cervical images for future quality assurance after each VIA procedure. Quality assurance sessions were held routinely on a monthly basis; 10% of the cases seen by each nurse over the past month were presented by the nurse and reviewed by the entire cervical cancer prevention team of clinicians. During the sessions, each nurse was quizzed on TZ types, presence or absence of acetowhite lesions, selection of treatment (TA versus LLETZ), presence of lesions suspicious for invasive cancer and adequacy of the LLETZ procedure in excising the entire lesion. At the end of each session, the principles of VIA and the treatment of precancerous lesions were reviewed with a focus on areas where weaknesses were noted.

Monitoring of adverse events

Every participant was advised to attend the clinic or call the dedicated telephone number of the study coordinator if she had any of the symptoms mentioned earlier or if she required medical consultation or hospitalization because of any reason within 2 weeks after treatment. The study coordinator telephoned every participant 2 weeks after treatment to check their well-being and verify if any medical consultation was required in the intervening period. After the first call, the coordinator called the woman every 3 months until her follow-up clinic visit to check if any major illness or hospitalization occurred in the intervening period. The phone number of the partner or husband or a close relative was obtained from each study participant. This number was used if the participant could not be reached on her mobile phone.

All adverse events were recorded in the project database. Serious adverse events were notified to IARC within 24 h of receiving the information.

Posttreatment follow-up

The social worker conducted a telephone interview with each participant 2 weeks after treatment to repeat the patient-reported outcome survey and check if the woman had any clinic visit or hospitalization in the intervening period. Women were contacted over the telephone approximately 1 month before the scheduled follow-up visit to remind them about the checkup.

At the follow-up visit (which was 6 months after treatment for the pilot phase and 12 months after treatment for the extended phase), a nurse collected a cervical swab for high-risk HPV detection testing and proceeded to perform the VIA. The results of HPV testing were communicated to the women over the telephone and women who were HPV positive were recalled to the clinic. Women with type-specific persistence of infection or positive VIA findings were offered LLETZ. Women not willing to undergo LLETZ were offered ablative treatment. Women negative on the HPV test and VIA were advised to continue with routine screening.

Detection of high-risk HPV

All baseline cervical samples were tested in the laboratory for high-risk HPV infection; only women with a positive baseline HPV result had their follow-up cervical samples tested. All cervical samples were analyzed using the Xpert HPV test (Cepheid), which detects 14 high-risk HPV types and provides genotype information in three separate channels for HPV16, HPV18 and HPV45 combined, and the remaining 11 HPV types combined (HPV31, 33, 35, 39, 51, 52, 56, 58, 59, 66 and 68). Testing was performed at the University Teaching Hospital, Lusaka.

Outcomes

The primary outcome of interest was treatment success. As per a prespecified analysis plan, treatment success at the 12-month follow-up was defined as either HPV type-specific clearance in participants positive for high-risk HPV at baseline, or negative VIA testing if the baseline HPV test was negative. Secondary outcomes included treatment side effects and complications, and patient-reported outcomes (pain and discomfort and satisfaction level) immediately after treatment and 2 weeks later. As an exploratory outcome, we analyzed the histopathology results of women undergoing LLETZ to estimate overtreatment in a screen-and-treat setting. We carried out post hoc subgroup analyses stratified according to HIV status and the baseline results of HPV testing for the pilot versus extended phase.

Sample size estimation

In the pilot phase of the study, 250 women were allocated to one of the treatment arms. Successful completion of the pilot was a requirement specified by the funding agency to support the extended phase of the study. Sample size was calculated before the extended phase using data from the pilot phase. To demonstrate noninferiority in the treatment success rates, we used the following assumptions: the treatment success rate in the cryotherapy (comparator) arm to be 55% (treatment success rate in the cryotherapy arm at the 6 months follow-up at the end of the pilot phase was 55%). We used this as a conservative treatment success rate at 12 months; difference in treatment success rates between TA and cryotherapy to be 5% (at the end of the UH2 pilot phase, a 5% difference in treatment success rates at the 6-month follow-up was observed, with a 60% treatment success rate in the TA arm); a noninferiority margin of 4%; an annual follow-up default rate of 30%; an 80% power; and a 2.5% significance level. To account for the three-arm multiple comparisons and for the planned subanalyses stratified according to HIV status, a 0.00417 (0.025/3/2) significance level was used in the calculations.We estimated that a total sample size of 3,123 women who were VIA screen-positive would be required to confidently answer the primary questions posed in this study. This was equally randomized to the three arms, with 998 women per arm. The pilot study recruited 750 participants. Based on the sample size estimation, we planned to recruit an additional 2,373 participants in the extended phase of the RCT.

The original study protocol did not have a plan to incorporate a pilot study. The estimated sample size for the study was 4,434 women to be distributed equally among the three arms. The funding agency (National Cancer Institute/National Institutes of Health (NIH)) advised us to split the study into a pilot and an extended phase. After completion of the pilot phase, we reestimated the required sample size using data from the pilot phase. The revised total sample size was 3,873 (including the 750 women who participated in the pilot phase). This explains the discrepancy in sample size between the protocol and the present article.

Statistical analysis

Study data were collected and managed using Research Electronic Data Capture hosted at IARC, and analyzed using STATA v.17.0 (Stata-Corp). The participant baseline characteristics and follow-up status, and primary and secondary outcomes, were shown as proportions; the comparison between the three study arms was determined using the Pearson chi-squared test.

Multiple imputation was used to cater for data missingness in the outcomes and some of the explanatory variables. The effect of the treatment arm on the primary outcome was evaluated using the protocol analysis (in which only participants who followed the study procedures as per the assigned treatment arm were included). Effect estimates were provided as relative risks together with their 95% CIs obtained from the generalized linear models for the binomial regression using the log link function. The multivariate regression model involved adjustment for age and baseline HIV status. The regression analyses were further adjusted for clustering on a combined phase–study center variable to cater for the possible correlation of responses within the study phase and center (that is, the responses may have been independent between but not necessarily within study centers). Statistical significance between the effect estimates of the TA arm compared to each of the other two treatment arms was inferred if the lower bound of the CI was greater than 0.61. This criterion was based on the assumptions for the sample size and based on the difference in treatment success and the margin of noninferiority. Post hoc subgroup analyses were performed in which the assessment of the primary outcome was stratified according to baseline HPV and HIV status, and according to the pilot and extended phases. An additional post hoc subgroup analysis was done using VIA only in the definition of treatment success.

Ethics and inclusion statement

The collaborators at the University Teaching Hospital, Zambia had major roles in study design, study implementation, data analysis and manuscript preparation. G.P.P., the principal investigator from Zambia, obtained a visiting scientist position at IARC to work with IARC scientists to plan the study, determine the statistical analysis plan and the sample size. Once the study was approved by the funding agency, a meeting was organized in Zambia to plan the details of the study implementation, to ensure the safety, well-being and data privacy of the study participants and to conduct the study according to good clinical practice. We planned to implement the study in multiple primary care clinics in Lusaka, Zambia to ensure that the study outcomes could be quickly adopted and scaled up in the ongoing VIA-based cervical screening program. Building the capacity of different healthcare professionals (nurses, gynecologists, pathologists) and social workers in Zambia was one of the major objectives of this collaborative study. Accordingly, multiple training programs were organized in Zambia. Collaborators from Zambia visited IARC from time to time and especially when the pilot and extended phase data were jointly analyzed. To ensure the safety and well-being of the trial participants, all study staff were trained in good clinical practice. A dedicated mobile phone number was made available 24/7 for participants to contact the study coordinator regarding any problem. The study coordinator maintained regular telephone contact with the participants, especially during the COVID-19 outbreak, to ensure that their follow-up took place in a safe environment. The DSMB consisted of five members, including a lay person, to monitor the study from time to time. The renowned researcher and gynecological oncologist from South Africa, L. Denny, chaired the DSMB. Because the study involved participation of a large number of women who were HIV positive, an expert in managing patients who were HIV positive was a DSMB member.

Ethics approval

The study was reviewed and approved by the research ethics committees at IARC (IEC 2016-01 approved on 4 March 2016) and UNC (MH/101/23/10/1 approved on 10 July 2017). Each study participant signed a written informed consent. An information leaflet was shared with the participants and the contents were explained by the study coordinator before obtaining consent.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.