Delft University of Technology A systematic review and meta-analysis of thermal coagulation compared with cryotherapy to treat precancerous cervical lesions in low- and middle-income countries

Thermal coagulation is gaining popularity for treating cervical intraepithelial neoplasia (CIN) in screening programs in low‐ and middle‐income countries (LMICs) due to unavailability of cryotherapy.

each year. 1 The burden of cervical cancer disproportionately affects LMICs, where 85% of cases occur. 1 Screening programs aim to prevent cervical cancer by more than 85% of cases timely treatment of precancerous cervical lesions. In resource-constrained settings, the WHO recommends see-and-treat screening programs where women are screened and treated in a single visit with the loop-electrosurgical excision procedure (LEEP) or cryotherapy. 2 The number of LMICs with national screening programs has increased over the years, but coverage remains low. 3 There are numerous factors influencing uptake of screening programs, including lack of skilled healthcare workers, lack of equipment, and other health system challenges. An important logistical constraint for the sustainability of see-and-treat programs is maintenance of cryotherapy devices and the lack of availability of refrigerated gas, owing to its high importation and purchase costs and large-size cylinders needed for transport to treatment sites. 4,5 This affects the availability of treatment during screening, and thus the success of screen-and-treat programs.
Thermal coagulation is an alternative ablative technique using electricity to destroy the premalignant cervical lesions by heating.
The device is small and lightweight, making it practical for use in an outpatient setting with minimal complications. These advantages are particularly important in rural and outreach settings. 4,5 In 2013 and 2014, Sauvaget et al. 6 and Dolman et al. 7 published separate meta-analyses analyzing the efficacy of cryotherapy and thermal coagulation to treat CIN lesions in mainly high-income countries, demonstrating cure proportions of 94.0% (CIN 1), 92.0% (CIN 2), and 85.0% (CIN 3) for cryotherapy, and 96% (CIN 1) and 95% (CIN 2-3) for thermal coagulation. However, data from LMICs included in previous meta-analyses of thermal coagulation were limited to one paper from Singaporeand one from India. 8,9 To assess the efficacy of thermal coagulation in LMICs, more data from these settings should be reviewed.
LMICs differ from high-income countries in terms of healthcare structures and patient population, therefore efficacy may not be equal; for example, HIV-positivity rates are higher in LMICs than in Europe and Northern America, consequently with higher prevalence and recurrence of CIN lesions. 2 The aim of the present systematic review and meta-analysis was to assess the effectiveness of thermal coagulation to treat CIN lesions compared with cryotherapy, with focus on LMICs.

| MATERIALS AND METHODS
Papers were identified using two strategies: (1) identified papers from the previous meta-analyses, 6

| Search strategy
An electronic literature search (February 2018) was performed in PubMed, Embase, Web of Science, Cochrane Library, regional databases, and Google Scholar with assistance from a medical librarian. A wide range of definitions are used in the literature, therefore different keywords were included to cover all related publications (Data S1). Papers with a publication date before 2010 were excluded to avoid overlap with the existing meta-analyses. In Google Scholar, no date limitation was used because this database had not been searched in the meta-analyses of Sauvaget et al. 6 and Dolman et al. 7

| Eligibility criteria
Titles and abstracts of all papers were reviewed by three researchers (MF, AR, RO) for relevance and presence of original data. The remaining papers were reviewed by three researchers (AR, RO, MF) and retained if the following criteria were met: cure proportion was the outcome measure and was defined by colposcopy, biopsy, cytology, and/or visual inspection with acetic acid (VIA)/visual inspection with Lugol iodine (VILI). Cytology and pathology are not always available to measure treatment outcome in LMICs; instead, screening is frequently performed by VIA or VILI. As such, papers defining cure proportion with VIA-or VILI technique were considered.
Papers not based on original data or with insufficient data on cure proportion and follow-up were excluded. Follow-up duration had to be 6 months or more after initial treatment, sample size more than 25 patients, and loss of patients attending follow-up not more than 50%. Cryotherapy for CIN 2-3 had to be provided with the doublefreeze method, and the treatment procedure had to be performed for no other reason than for treating CIN, nor be provided simultaneously with other treatment. In case of discordant results, consensus was reached among four researchers (AR, RO, MF, JB).

| Risk of bias assessment
Study quality was assessed using a component approach. 10 Unknown HIV status of screening participants and loss to follow-up of greater than 25% were considered high risk of bias. Studies using cytology or histology to assess outcome were considered low risk of bias compared with VIA-based outcomes. The eligibility criteria described above aimed to eliminate studies with very poor study quality. and fertility outcome. Cure rates were defined as a proportion with the number of women with negative VIA/VILI, negative cytology, negative colposcopy, or negative biopsy at a minimum of 6 months' follow-up duration divided by the number of women attending follow-up. Therefore, the terminology "proportion cured" or "cure proportion" was used instead of "cure rate".

| Statistical analysis
Pooled cure proportions with 95% confidence intervals were the primary outcome. Cure proportions were pooled in a random effects model. Analyses were stratified by treatment modality (cryotherapy versus thermal coagulation). If both 6-and 12-months' follow-up data were available, 12-months' follow-up data were used as this is the recommended follow-up duration for detection of persistent disease after initial treatment. 2 Studies were stratified per CIN grade (CIN 1, CIN 2-3, or VIA/VILI outcome) and region (Europe, North America, South America, Africa, and Asia). In a sensitivity analysis, the effect of follow-up attendance on cure rates was assessed comparing studies with 50%-75% and more than 75% follow-up attendance.
An additional search in clinical trial registers and journal databases yielded no additional publications.
After reviewing the title and abstract of all papers, 28 relevant papers were identified on cryotherapy and 20 on thermal coagulation.
After full-text review, 11 papers on cryotherapy and seven on thermal coagulation were eligible for inclusion.
Since publication of the meta-analysis by Sauvaget et al., 6 no new papers on cryotherapy from North America or Europe have been published, therefore these regions were excluded from further analysis for cryotherapy. Sauvaget et al. 6 included 20 studies from Asia, Africa, or South America in their meta-analysis on cryotherapy. In the present review, five of these studies were excluded owing to sample size of less than 25 patients, recurrence not specified per CIN grade, single-freeze technique for CIN 2-3 lesions, or no original data. [11][12][13][14][15] The remaining 15 studies conducted in Africa (n=5), Asia (n=7), and South America (n=3) were included in the present review.

| Papers from previous meta-analyses
Dolman et al. 7 included 13 studies in their meta-analysis on thermal coagulation. For the present review, five of these studies were excluded owing to follow-up duration of less than 6 months, insufficient data to calculate cure rates, and cure rates not differentiated per CIN grade. [16][17][18][19][20] The remaining eight studies conducted in Asia (n=2), North America (n=1), and Europe (n=5) were included in the present review. Figure 1 provides an overview of the included papers from the literature search and previous meta-analyses. Table 1 provides the details and references of the 40 included papers. 8,9,14, An overview of the excluded studies is provided as Table S1. 11-20

| Data from included papers
In total, data from 26 studies of 14 355 patients treated with cryotherapy and 15 studies of 4864 patients treated with thermal coagulation were included. Most papers were published in the last 10 years and described treatment by cryotherapy. Table 2

| Risk of bias assessment
Of the 26 included papers on cryotherapy, 6 (23%) provided data on the HIV status of participants. Twenty (77%) papers reported a follow-up attendance of more than 75%. Tai et al. 45 defined cure at follow-up as absence of CIN 3 lesions and did not include recurrence or persistence of CIN 1 and CIN 2 lesions. Of the 15 included papers on thermal coagulation, HIV status was reported for 3 (20%) papers. Eleven (73%) papers had a follow-up attendance of more than 75%. Visual assessment of cure rates in order of sample size did not suggest publication bias.  (Table S6).

| Efficacy of treating CIN lesions
Sensitivity analysis showed higher proportions of cure for cryotherapy papers with follow-up attendance of more than 75% for both CIN 1 and CIN 2-3 lesions (Table S7).

| HIV status
For both treatment modalities, only two studies published cure proportions for HIV-positive patients specifically. Table S8 presents all studies with cure rates for HIV-positive patients. Data on outcome of HIV-positive patients specifically was too limited to allow statistical testing.

| Treatment technique and provider
In contrast to cryotherapy studies conducted in the 1970s and  In 12 out of 15 papers on thermal coagulation, treatment was provided by physicians. Cryotherapy was more frequently provided by nurses (n=5) or by nurses and physicians (n=4).

| Pain, adverse effects, and fertility
Pain during and after treatment was discussed in seven papers (47%) on thermal coagulation. Three papers (38%) reported mild cramps or   Twelve papers (46%) on cryotherapy reported pain, varying from 1% to 30% of patients complaining of mild pain and cramps during treatment to less than 1% experiencing severe pain or cramps. Vet et al. 36 reported routine use of oral analgesics after cryotherapy in Indonesia.
Adverse reactions and complications were reported inconsistently and rarely for both treatment modalities. Table S9 shows the adverse reactions reported in 6 (40%) thermal coagulation and 15 (58%) cryotherapy papers. Fertility outcomes and pregnancy outcomes were also rarely reported. For each treatment modality, three papers mentioned subsequent pregnancies in treated patients, and three of these papers reported normal outcomes.

| Main findings
The present review aimed to compare the effectiveness of thermal

| Interpretation
The cure proportions were comparable to previous reviews of cryotherapy (94.0% for CIN 1, 92.0% for CIN 2, 85.0% for CIN 3 lesions, and 89.9%-91.9% for all CIN grades), although they were F I G U R E 3 Cure proportions for CIN 2-3 lesions treated with thermal coagulation grouped by region.
slightly lower compared with a previous meta-analysis of thermal coagulation (96.0% for CIN 1 and 95.0% for CIN 2-3 lesions). 6,7,60 The lower proportions of cure found for thermal coagulation can be explained by an increased number of papers from LMICs in the present review. 7 A retrospective analysis of thermal coagulation in Bangladesh, Brazil, and India by Nessa et al. 61 found cure proportions ranging from 83% to 88% for CIN 1-3 lesions. This paper was not included in the present review owing to more than 50% loss to follow-up.
Furthermore, we employed a different strategy to assess study quality and used stricter inclusion criteria, excluding studies with follow-up duration of less than 6 months and sample size smaller than 25 patients.
It is unlikely though that the stricter inclusion criteria explained the difference in pooled cure proportions. The studies excluded from the previous meta-analysis, Hussein et al. 17  America, the prevalence of HPV in women with normal cytological findings was 24% and 16.1%, respectively, compared with 14.2% in Europe. 63 The higher prevalence of HPV in the general population, in Sub-Saharan Africa especially, might lead to a lower HPV clearance and higher reinfection rate after treatment. 32,63,64 The present review included papers with cure proportion as the primary outcome and is not representative of all literature published on pain, adverse effects, fertility outcomes, and obstetric outcomes. A systematic review of the adverse effects and benefits of cryotherapy found that complications such as major bleeding and organ damage are extremely rare (RR <0.05) but reported low-quality evidence. 65 There are currently no reviews on the adverse effects of thermal coagulation. Viviano et al. 66 reported, in a study in Cameroon, a mean visual analogue score of 3.0 ± 1.6 during treatment.
A Cochrane review 67 found an increased risk of premature delivery in women with CIN lesions, with a lower relative risk for ablative T A B L E 2 Summary of the included studies (n=41).

| Limitations
We attempted to identify all papers published on cryotherapy and thermal coagulation for treatment of CIN lesions, with focus on LMICs.
However, there are limitations to the data and findings presented.
Papers published before 2010 on cryotherapy in LMICs might have been missed because Sauvaget et al. 6 used less inclusive keywords in their literature search ("cervical intraepithelial neoplasia," "CIN," and "cryotherapy") and regional databases were not included. We believe this difference will be minimal, based on our literature search with 129 unique references identified in regional databases, of which only one abstract was found to be relevant and the full article did not meet reported cure proportions. We found few studies with nonphysician clinicians as treatment providers. It is important that more data are collected from programs with nonphysician clinicians because this will be the reality for most women screened in low-resource settings.

| Future recommendations
In future, more HPV-based screening programs will be implemented in LMICs, with higher treatment rates expected due to higher sensitivity of HPV testing compared with VIA/VILI and cytology. 62 This approach will yield greater health benefits than VIA-based programs in lowresource settings where cervical cancer incidence is high. 68 A widely available, acceptable, and effective treatment method is necessary.
Thermal coagulation is a promising alternative to cryotherapy with comparable proportions of cure, which will enhance the sustainability of screening programs in LMICs and make a significant contribution to the fight against the burden of cervical cancer worldwide. We recommend that more studies including randomized controlled trials are conducted to compare thermal coagulation and cryotherapy in LMICs to assess efficacy, safety, and provider and patient experience. proportions of cure, and report pain and adverse effects consistently.

AUTHOR CONTRIBUTIONS
MF reviewed the study design, carried out the final literature search and review, conducted data extraction and analysis, and drafted the manuscript. RO initiated the study, carried out the literature search during the orientation phase of the study, carried out data extraction and analysis, and participated in writing the final manuscript. AR initiated the study, carried out the literature search during the orientation phase of the study, assisted in data extraction, and participated in writing the final manuscript. OD carried out data analysis and participated in writing the final manuscript. JB supervised the study design, data analysis and interpretation, and participated in writing the final manuscript. AW supported data interpretation and participated in writing the final manuscript. All authors reviewed and approved the final manuscript.

ACKNOWLEDGMENTS
We thank JW Schoones, medical librarian at LUMC, for his support during the electronic literature search.

CONFLICTS OF INTEREST
The authors have no conflicts of interest.

SUPPORTING INFORMATION
Additional supporting information may be found online in the supporting information section at the end of the article.
Data S1. Keywords used in the PubMed literature search. In other databases the same keywords were used, commands were adjusted to the specific database. Table S1. Overview of studies included in previous meta-analyses but excluded from the present study.      Table S7. Sensitivity analysis of cure proportions for studies with follow-up attendance of 50%-75% and greater than 75%.