Cervical Cancer Screening and Treatment Algorithms Using Human Papillomavirus Testing—Lessons Learnt from a South African Pilot Randomized Controlled Trial

Cervical cancer is the leading cause of cancer deaths among women in Sub-Saharan Africa (1, 2), where there is also the highest prevalence of human immunodeficiency virus (HIV) infection worldwide. In comparison to HIV-uninfected women, women living with HIV (WLWH) are at higher risk of persistent human papillomavirus (HPV) infection, precancer and recurrence of precancer after treatment and, subsequently, of cervical cancer (3–6).

In 2018, the World Health Organization (WHO) called for action to eliminate cervical cancer as a public health problem, setting as global targets for 2030: 90% of girls 9 to 14 fully vaccinated against HPV, 70% of women screened with a high-performance test by ages 35 and 45, and 90% of precancer lesions treated and of advanced cancer adequately managed (7). HPV vaccination was incorporated into the National Immunization Program in South Africa (SA) in 2014 and would not have included the targeted population for screening, who therefore still need 3 to 5 years screening to achieve the WHO elimination goals. Current WHO cervical screening guidelines recommend HPV-based screening and treatment every 5 to 10 years starting at age 30 for women in the general population and every 3 to 5 years for WLWH initiating at the age of 25 (8, 9). Screening and treatment algorithms present the advantage of reducing the number of visits and thus the loss to follow-up while optimizing use of local resources (e.g., possibility of treatment without histologic diagnosis, less specialized trained staff; ref. 10). However, in women of reproductive age or in WLWH, the high HPV test positivity calls for using a triage test, such as visual inspection with acetic acid (VIA) or partial HPV genotyping, to best select women at high risk of having or developing precancerous lesions. Conversely, the high sensitivity of HPV tests may be impaired by triage tests with low specificity, such as VIA, which may increase the risk of cervical cancer by missing precancerous lesions among those with a false negative triage result. Therefore, testing different screening and management algorithms, including immediate treatment with fast and easy to perform techniques such as thermal ablation and/or cryotherapy of all HPV-positive women warrants further consideration.

The Cervical Cancer Screening and Treatment Algorithms study (CESTA) was designed as a randomized controlled trial and piloted (CESTAP) in Durban, South Africa using HPV testing, with and without VIA triage, followed by ablative treatment (AT) as recommended by WHO (8, 9). The screen-and-treat protocols (with or without triage) were administered by nurses. The CESTA pilot was an external independent study with predetermined study population and sample size, which aimed to develop and field test study procedures, measure HPV prevalence in the catchment population for sample size calculation of the full study, as well as evaluating the feasibility of implementing screening and treatment algorithms. Here, we report on the participant's pathway in the screening and treatment algorithm by randomization arm, participants’ baseline characteristics, the prevalence of HPV infection overall and by HIV status, the retention by randomization arm, the number of cases diagnosed or missed by the VIA triage, and finally, the AT side effects reported by women in the CESTA pilot conducted in Durban, South Africa.

CESTAP is a 1:1 parallel randomized control trial where 350 women ages 30–54 years old were recruited, regardless of their HIV status, from January to June 2018 at Wentworth Hospital, Durban, SA. Women were excluded if they: reported no previous sexual activity; were pregnant or less than 3 months postpartum; had a history of cervical cancer or had been treated for a precancerous lesion or dysplasia of the cervix in the last 6 months; had a hysterectomy; were unable to consent; or refused to be tested for HIV.

A dedicated community health worker mobilized the community in the catchment area and gave clinical appointments to women. On presentation to the clinic (i.e., Visit 1: screening visit), trained study nurses enrolled women into the study after obtaining signed informed consent. Women willing to participate were asked about their HIV status; for those who did not know, an HIV rapid test was offered after counselling. Confirmed HIV-uninfected women and self-reported HIV positives were enrolled in the study while newly diagnosed HIV-positive women were referred to HIV care and excluded from the study because they would not have been comparable to other HIV-positive women already on antiretroviral therapy (ART) for several months/years. Self-report from HIV-positive participants, or medical records from antiretroviral (ARV) clinics when available, were used to obtain HIV clinical history and treatment using a standardized questionnaire. Women unwilling to participate were advised to use routine cervical cancer screening (i.e., cytology screening every 3 years). Figure 1 depicts the recruitment and screening and treatment algorithms.

Participants were randomized into one of the two study arms: VIA triage or no triage and, depending on their initial screening results, could attend up to four visits (Fig. 1).

• (i) VIA triage arm: At Visit 1, women self-collected a vaginal sample, and a cervical sample was collected by a nurse and received an appointment for 2 weeks later. Those testing negative for HPV in both samples were explained the meaning of this negative result and exited the study. Women testing HPV positive in either sample continued with the study (Visit 2: triage and treatment visit). At Visit 2, trained nurses performed VIA by applying dilute acetic acid on the cervix for 1 minute. Results were either: negative (no acetowhite lesions observed), positive (acetowhite lesions observed), inadequate [squamocolumnar junction (SCJ) not fully visualized], or suspicious for cancer. Nurses collected two to four biopsies from observed acetowhite lesions in women with positive VIA, or two biopsies at 12 and 6 o'clock at the SCJ for women with negative VIA. Nurses collected an endocervical sample for cytology from women with inadequate VIA. These women and those with suspected cancer were referred to colposcopy (Visit 3). Women with positive VIA were also assessed to see whether they were eligible for AT (i.e., the lesion did not extend into the endocervix, was not larger than 75% of the ectocervix, and there was no suspicion of cervical cancer); women eligible for AT were randomized to receive either cryotherapy or thermal ablation, which was offered on the same visit (Visit 2). Women not eligible for AT were referred to colposcopy (Visit 3). Once histology results were available, VIA-negative women with cervical intraepithelial neoplasia (CIN)2+ were also referred to colposcopy and treatment with large loop excision of the transformation zone (LLETZ) or other as appropriate.

• (ii) No triage arm: At Visit 1 (screening visit), the nurse collected a cervical sample for HPV testing. Self-sampling was not offered in this arm. Women who tested negative for HPV exited the study and those positive continued with a second visit (Visit 2: AT visit). At Visit 2, after evaluation with dilute acetic acid to determine eligibility for AT (same criteria as for women in the VIA triage arm), eligible women received thermal ablation. Those not eligible for AT were referred to colposcopy (Visit 3) and treatment was offered as appropriate. The no triage arm was designed as the simplest possible as it would be implemented in real-life setting, that is, without histology and self-sample collection and AT by thermal ablation only.

All HPV-positive women were recalled a year later (Visit 4: 1-year follow-up) for HPV testing: those testing positive underwent colposcopy with biopsy of observed lesions and treatment as appropriate; those testing negative were referred for routine screening.

In the VIA triage arm, after receiving instructions from a nurse, women self-collected at the clinic a sample with a swab (first 99 women) or a conical endocervical brush (CareHPV brush, Qiagen, next 76 women) that was washed into 20 mL of Preservcyt medium (Hologic Inc.). In both arms, cervical samples were taken by a nurse using the Cervex brush (Rovers; Unimar) and washed into a 20 mL of Preservcyt medium. Samples were transported to a central lab at the University of KwaZulu-Natal (Durban, South Africa) where HPV DNA testing was done using GeneXpert HPV assay (Xpert HPV Assay, Cepheid). The Xpert HPV Assay is a qualitative in vitro test for the detection of 14 HPV types in five groups: HPV16, HPV18/45, HPV31/33/35/52/58, HPV51/59, and HPV39/56/66/68 (11). Invalid and error samples were tested again only if both self- and nurse-samples were invalid or errors (i.e., without positive or negative result).

Biopsies and LLETZ were interpreted at a local laboratory and reinterpreted by an independent pathology expert. Discordant cases were reviewed by both pathologists together to obtain a consensus on the diagnosis. The most severe histologic, agreed upon diagnosis, between biopsies and excisional treatment (i.e., LLETZ), is reported here.

Cryotherapy was performed using a 3-minute freeze, 5-minute thaw, 3-minute freeze application of cryoprobe on the transformation zone (TZ). Thermal ablation was performed using Liger machines and thermoprobe was applied on the TZ during 45 seconds at 100°C from one to up to five times.

At Visit 1 in the VIA triage arm, women were asked about their sampling preference (self-sample, nurse-collected, or no preference).

In both arms, details regarding treatment side effects were collected during face-to-face interviews immediately after AT (e.g., pain, crumps, vaginal burns, bleeding), and posttreatment side effects by telephone 1 week later (e.g., watery discharge, unexpected significant discomfort, or heavy bleeding), as well as AT acceptability. Women who had side effects related to treatment other than watery discharge were recalled for clinical evaluation.

At the end of recruitment, discussions among investigators and nurses were carried out to gather their perception of the feasibility of implementing each of the algorithms (HPV testing, triage, ablative treatment).

This article reports on: (i) HPV prevalence in the catchment population as well as the hierarchical partial genotyping among HPV-positive women, overall and by HIV status; (ii) retention to screening and treatment algorithms and the percentage of women treated; (iii) number of histologically confirmed cervical intraepithelial lesions grade 2 or worse (CIN2+) detected or missed by VIA triage among HPV-positive women in the VIA triage arm; and (iv) agreement between self- and nurse-collected samples by HIV status in the VIA triage arm.

In addition, the women's acceptability of AT and their side effects, and the nurses’ perception of performing the screening and treatment algorithms (see above) were also evaluated.

A convenience sample size of 350 consecutive eligible women were recruited, regardless of their HIV status, and randomized into the two study arms.

Sealed envelopes, prepared by and brought to the local clinic by the International Agency for Research on Cancer (IARC) data manager, were used to allocate women to each screening and treatment arm and to either cryotherapy or thermal ablation in the VIA triage arm. Randomized allocation was done using the Soares big stick randomization design algorithm using R software. This method assigned sequentially an arm without exceeding the defined degree of imbalance between the two arms (12, 13).

Baseline participants’ characteristics (age, ethnicity, marital status, education, HIV, and HPV status) were tabulated by randomization arm. Regardless of randomization arm, the prevalence of HPV infections (overall and by hierarchical partial genotype group among HPV positive), by HIV status, and the percentage of women completing the screening and treatment algorithm by age were calculated with 95% confidence intervals (95% CI), as appropriate. Detection of CIN2+ was also computed in the triage arm according to VIA results and HIV status. Reported treatment side effects (immediately after treatment and 1 week later) for both cryotherapy and thermal ablation were also tabulated. Medians and interquartile ranges of women's reported pain (score ranging from 0 = “no pain at all” to 10 = “too painful to complete treatment”) immediately after treatment and women's treatment acceptability (score ranging from 0 = “very unacceptable” to 10 = “highly acceptable”) were calculated. χ² tests were used for group comparisons. All analyses were performed with STATA 14.2 and the CONsolidated Standards Of Reporting Trials (CONSORT) framework was used to report CESTA RCT.

The CESTA study (NCT04794660) follows the declaration of Helsinki ethical guidelines and was approved by the Biomedical Research Ethics Committee of the University of KwaZulu-Natal (Durban, South Africa) and IARC ethical committee.

The data generated in this study are available upon request from the corresponding author.

Figure 2 presents the CONSORT diagram of the study (14); 369 women presented to the clinic, and 19 could not be included to the study because of ineligibility reason such as newly HIV positive diagnosed at screening (n = 11); 350 eligible women were included and randomized for HPV testing followed by VIA triage (n = 175) or immediate treatment (no triage, n = 175).

In the VIA triage arm, 53/175 (30.3%) women were HPV positive. Among the 50 HPV-positive women attending Visit 2 (triage/treatment visit), 42 women had adequate VIA: 18 (43%) were VIA negative and 24 (57%) were VIA positive. Of those who were VIA positive, 19 of 24 were eligible for and were treated with AT on the same day; this included 38% of those in the triage arm that were HPV positive). Five VIA-positive women ineligible for AT were referred to colposcopy (2 women were treated with LLETZ and 3 had negative colposcopy); an additional 8 of the 53 women in the VIA triage arm were referred directly to a colposcopy/gynecologic clinic for management: 6 had inadequate VIA (SCJ non-fully visible), 1 had suspected cancer, and 1 needed fistula surgery. The 1-year follow-up visit was attended by 36 (68%) HPV positives at enrolment. At this visit, 24 (67%) women were HPV negative and 12 (33%) were HPV positive. All HPV positives, but one, attended colposcopy (6 with abnormal results were treated with LLETZ and 5 who had negative colposcopy returned to screening routine).

In the no triage arm, 45/175 (26%) women were HPV positive, all attended the ablative treatment visit (Visit 2), 27 (60%) were eligible for and treated with thermal ablation on the same day. The remaining 18 (40%) were not eligible for AT (5 for suspicion of cancer, 10 for a SCJ non-fully visible, two lesions too big or too much into the canal, and 1 had too many polyps) and were referred to colposcopy (Visit 3): 7 of them were treated with LLETZ, 9 had negative colposcopy, 1 woman was referred for surgery for reason not related to cervical neoplasia, and 1 woman was lost to follow-up. The 1-year follow-up visit (Visit 3) was attended by 38 (84%) HPV positives at enrolment. At this visit, 23 (61%) women were HPV negative and 15 (39%) were HPV positive. All HPV positives, but 4, attended colposcopy; 3 were treated (2 LLETZ, 1 hysterectomy) and 8 had negative colposcopy.

The mean (±SD) age for enrolled women was 42 (±7) years. Most participants were African or mixed ethnicity (89%), 68% had completed secondary school and 55% were employed. Among the 350 women included in the study, 75 (21%) self-reported being HIV positive and being under treatment (60% with undetectable viral load). No statistical differences in women characteristics were observed between arms (Table 1).

Table 2 shows the prevalence of HPV infection as well as the hierarchical positivity of HPV16, HPV18/45, and other high-risk HPV (all other GeneXpert groups together) among HPV-positive women, by HIV status. The overall HPV prevalence was 28% (95% CI: 23–33). WLWH had higher overall HPV prevalence compared with HIV-uninfected women (52%, 95% CI: 40–64, 21% 95% CI: 17–27, P < 0.05, respectively), HPV16 being more present in WLWH than in HIV-uninfected women (36% vs. 12% respectively, P < 0.05); HPV18/45 was slightly more frequent, although not significant, in WLWH than in HIV-uninfected women (21% vs. 14%, respectively, P = 0.362). Infections with HPV types other than HPV16, HPV18, or HPV45, were more common in HIV-uninfected women than in WLWH (75% vs. 44%, P < 0.05).

Among the 175 women allocated to the VIA triage arm, 169 completed the screening and treatment algorithm, including 23 women treated (19 with ablative treatment, 4 with excisional treatment) representing 13% of the women randomized and 43% of the HPV-positive women (Table 3).

Among the 175 women allocated to the no triage arm, 173 completed the screening and treatment algorithm, including 34 women treated (27 with ablative treatment, 7 excisional treatment) representing 45% of the women randomized and 76% of the HPV-positive women.

Overall, there was no difference in retention to treatment (Visits 2 and 3) between arms. However, the referral rate to colposcopy was almost double in the no triage arm than in the VIA triage arm, [13/50 (23%) vs. 18/45 (40%) 40%, P = 0.76]. The main reasons for colposcopy referral, in both arms, was partial or nonvisibility of the SCJ and suspicion of cancer, followed by other ablative treatment ineligibility such as lesion too big or lesion extending into endocervical canal. No differences were observed by HPV infection type or HIV status; however, women over 45 years of age were more often referred to colposcopy (P < 0.05) mostly because of the non-fully visibility of the SCJ.

Similarly, there was no significant difference in the retention to the 1-year follow-up (Visit 4) but a slight increase in attendance was observed among women allocated to the no triage arm.

Altogether 10 women were treated with cryotherapy (VIA triage arm) and 36 with thermal ablation (9 in the VIA triage arm, 27 in the no triage arm). Immediate side-effects were mild for both treatments. One week after treatment, discharge (watery or brown) was the most reported side effect. Acceptability was moderate to high (score 6 of a scale 0–10) for both ATs (Supplementary Table S1).

Nurses understood that it was important to treat HPV-positive women and had no concern to first triage them with VIA or treat them without triage, but they noticed more worrying/stress in patients treated with cryotherapy than with thermal ablation.

Among the 175 women in the VIA triage arm who had both samples, 45 (26%) reported preferring the nurse-collected sampling, 35 (20%) the self-collected sampling and 95 (54%) had no preference.

The concordance for HPV positivity between the vaginal (swabs and brushes) and cervical samples from 173 women in the VIA triage arm was 88% (kappa: 0.7, 95% CI: 0.6–0.8). There was no statistical difference in the HPV positivity based on the sampling method (Supplementary Table S2). One nurse-collected cervical brush and one self-collected vaginal swab had invalid test results.

Figure 3 reports on the VIA triage performance according to the histologic diagnosis of women in the VIA triage arm. Among 35 women with histologic diagnosis, 18 were VIA positive including 4 with CIN2+ (1 WLWH, 3 HIV-uninfected women); all 18 VIA-positive women were immediately treated. Among 17 VIA negative (who had random biopsy per protocol), 3 were diagnosed with CIN2+ (1 WLWH, 2 HIV-uninfected women); these 3 women were later treated when biopsy results were available. On the basis of VIA triage results, 78% (14/18) women with positive VIA triage had no evidence of precancer on biopsy. In addition, 3 out of 17 (18%) with negative VIA had biopsy-proven precancer and would have been untreated if biopsies were not collected as part of the study protocol.

Screening and treatment algorithms, when scaled-up, should have major impact in reducing cervical cancer incidence and mortality, especially among WLWH who are at high risk of developing cervical cancer (3, 15). Screening and treatment algorithms have the potential advantage of reducing the number of visits therefore the loss to follow-up (16, 17). In this study, completion of the algorithms was high in both randomization arms and no difference was found while using a triage method (VIA triage arm) or not (no triage arm). However, although different (but nonstatically), the colposcopy referral was high in both arms, especially in older women due to their SCJ not being fully visible (18–20). This natural evolution of the cervix may be a strong barrier for screening and treatment algorithms for older women as non-fully visible SCJ prevents to do VIA triage, assessment of AT eligibility and AT itself (21); hence, further research is needed to overcome these anatomic difficulties. Women referred to colposcopy represent a proportion of women that may not receive the appropriate care in real settings, especially in Sub-Saharan Africa where there is an important lack of trained specialists and colposcopy equipment limiting care delivery.

HPV-based screening and treatment algorithms need further investigation as HPV testing has high sensitivity for the detection of cervical precancerous lesions particularly CIN2+ in women (22) but has potential for overtreating women without precancerous lesions because the test positivity may indicate a transient HPV infection rather than a persistent infection associated with the development of cervical cancer precursors. Thus, the screening and treatment algorithm without triage offers the advantage of treating all HPV-positive women [with possible CIN2+ lesions] eligible for AT at the same visit with the risk of overtreatment and potential harms. However, AT has the potential to destroy HPV-infected cells at the SCJ that otherwise could later lead to new lesions (23). The question of triaging HPV-positive women remains important as it often implies an additional cost and workload for the health care provider. VIA as triage can be performed at the same visit as the treatment and at lower cost than alternative triage methods (e.g., cytology); however, an important concern remains that VIA sensitivity and specificity varies widely (24, 25), possibly jeopardizing the excellent sensitivity of the primary HPV testing and may avoid reaching the target of 90% of women with disease treated from the WHO cervical cancer elimination call (25). Our results suggest that VIA does not mitigate the potential treatment of those without precancerous lesions when no triage is used and that the addition of VIA does not identify all precancerous lesions present: 14 of 18 VIA-positive women did not have precancerous lesions on histology and in 3 out of 7 women with CIN2+, VIA was negative; assuming they were eligible for AT, these 3 women would have been left untreated and at increased risk of progression to cancer. Currently, WHO guidelines recommend using triage on WLWH who screened positive for HPV and suggest using either HPV16/18 or HPV16/18/45 partial genotyping, cytology, colposcopy, or VIA as triage method. On the basis of our results, if VIA is chosen, efforts to guarantee VIA quality assurance and training should be prioritized to ensure staff working knowledge at the highest level and efficient scaling-up if this algorithm is chosen (25).

The pilot study showed a prevalence of HPV of 28% which seems to be lower than in other cohorts reported in Southern Africa although this might be explained by differences in demographics of the study groups, especially in terms of age and HIV prevalence (2, 26). A higher prevalence of HPV in WLWH (52%) was observed, compared with those who were HIV uninfected (21%), this significantly higher HPV positivity among WLWH has also been reported elsewhere (27, 28).

Concordance between self-collected and nurse-collected specimens was high. In our study, women were asked to self-collect a sample and then had speculum examination by a nurse who collected a cervical sample. Women who preferred the nurse-collected sample (26%) thought that nurses were more reliable and would collect a better sample, while women who preferred self-sampling (20%) found it simpler, faster, less invasive, and more comfortable, but most of the women (54%) did not report any preference on the sampling method. Very few side effects after AT were reported (immediately or 1 week after treatment) for both cryotherapy and thermal ablation; however, nurses reported observing more discomfort from women who received cryotherapy due to the length of the treatment. Women reported overall high satisfaction for the screening and treatment service and adherence to both algorithms was high.

In conclusion, this pilot demonstrated that screening and treatment algorithms with or without VIA triage can be implemented in primary care settings and nurses, once trained, nurses are keen to perform all clinical procedures (from HPV sampling to treatment by AT including triaging as needed). However, the question of which test, both as primary screening and as triage is most suitable for WLWH remains unanswered and have led us to run the full CESTA study focusing on WLWH.

No disclosures were reported.

Where authors are identified as personnel of the IARC/WHO, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy, or views of the IARC/WHO.

H.M. Sebitloane: Supervision, investigation, methodology, writing–original draft, writing–review and editing. M. Forestier: Formal analysis, investigation, writing–review and editing. T.G. Ginindza: Investigation, methodology, writing–review and editing. W. Dhlomo: Investigation, methodology, writing–review and editing. E. Moodley-Govender: Investigation, writing–review and editing. T.M. Darragh: Investigation, writing–review and editing. A. Baena: Writing–review and editing. R. Herrero: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. M. Almonte: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. N. Broutet: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. H. De Vuyst: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing.

We thank the UKZN Department of Health for allowing us to implement the study. We also thank the HRP (the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction) for their technical support, principally in the writing of the protocol. We also thank the IARC, the WHO, the University of KwaZulu-Natal for all their support. We are so thankful for the UKZN Clinical Medicine laboratory support. We acknowledge Dr C. Tiloke for the assistance in article preparations. The study was funded by HRP (the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction, to H.M. Sebitloane and T.G. Ginindza).