Search strategy
PubMed, Embase and ICHUSHI (Igaku Chuo Zasshi) will be searched from inception to March 2021. Search terms will include relevant headings and keywords in the title, abstract and text, including human papillomavirus in Japan. ICHUSHI is the domestic database for the Japan Medical Abstracts Society Database. The use of this database requires the development of a Japanese language search strategy.
The search strategy will use the following general terms, expanded and appropriately modified for each database: ‘Japan’ and ‘human papillomavirus’ or ‘HPV’, and ‘cervical cancer’, and ‘genotype’, for ‘normal cytology’, and ‘cervical disease’ or ‘cervical intraepithelial neoplasia’. For this systematic review, there will be no restrictions on the date or language of articles to be reviewed. This search strategy will be constructed and performed with the assistance of a librarian. The search strategy is outlined in Table
S2.
The reference lists of identified studies will also be reviewed, evaluated and included if eligible. Grey literature shall also be considered for inclusion if the abstracts contain sufficient information to assess their eligibility. Possible sources of grey literature will include (1) identified authorities of this subject matter, (2) conference papers and (3) government documents and published guidelines. The search strategy will be developed according to Cochran Guidelines in collaboration with a librarian and subject matter expert in both Japanese and English.
Eligibility criteria
The population of interest for this review are Japanese people with a cervix residing in Japan who were screened at least once regardless of screening interval and stage of diagnosis. Males will be excluded from this study because they do not have a cervix. If identified, transgender men with a cervix will be included in this analysis. There will be no restriction on the age of participants in the studies for inclusion. Studies will be eligible if they are randomised control trials, case-control studies, case series studies, cohort studies or cross-sectional studies. Systematic reviews will not be eligible but their reference lists will be searched to identify any further eligible studies.
In order to achieve comparability to other international studies of HPV genotype prevalence [
29], the following inclusion criteria shall be used: (1) studies that assess cervical carcinoma, low-grade or high-grade cervical lesions must include a minimum of 20 cases [
21,
22,
24]; (2) studies that describe HPV infection in normal cytology must include a minimum of 100 cases [
25‐
27]; (3) studies must include at least one HPV genotype; (4) DNA or RNA polymerase chain reaction (PCR)-based assays should be used and sufficiently described; and (5) the study must include a detailed methodological description of cervical sampling techniques.
Studies must have been performed in Japan. For studies that were not conducted in Japan, or for multi-country studies, only studies containing primary data reporting HPV genotype prevalence for women resident in Japan will be included. Studies using nucleic acid testing of blood or blood components to detect HPV or reporting HPV prevalence in anatomic sites other than the cervix will be excluded from calculations of prevalence.
Selection of studies
Covidence review software will be used to screen titles and abstracts of all studies that are initially identified by two independent reviewers according to the selection criteria [
30]. The text of all potentially relevant studies will then be evaluated in detail against the eligibility criteria by two independent reviewers.
For the full-text review, the reviewers will independently classify articles as (1) included, (2) excluded or (3) maybe. A maybe status will imply that a decision to include or exclude the article is dependent on additional information being obtained from the author. Where additional information is needed, the corresponding author of the study will be contacted via email. A second email will be sent after 1 week in the event of no response to the initial email. A 2-week waiting period after the submission of the second email will be allowed for sufficient response. After which, these studies will be excluded [
31]. Articles that both reviewers classify as excluded will be removed, whereas those that both reviewers classify as included will be included. Discrepancies will be resolved through discussion with a third independent reviewer until consensus is obtained. The opinion of a subject matter expert will be sought, if necessary.
In accordance with the PRISMA guidelines, a summary of the search process, study selection and reasons for exclusion of studies will be included [
32]. A summary of all selected studies will also be included.
Outcome measures
The outcome measure of interest in this study is the HPV genotype-specific prevalence in women with a cytological or histological diagnosis of normal, low- or high-grade lesions or cervical cancers. HPV prevalence will be measured in cervical specimens from women where cytological classification is defined as normal, atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesion (LSIL) and high-grade squamous intraepithelial lesion (HSIL), and histological classification is defined as normal, cervical intraepithelial neoplasia 1 (CIN1), cervical intraepithelial neoplasia 2 (CIN2), cervical intraepithelial neoplasia 3 (CIN3), adenocarcinoma in situ (AIS), invasive cervical cancer (ICC)—unspecified, ICC—squamous cell carcinoma or ICC—adenocarcinoma.
Data will be extracted into a standardised extraction template and verified independently by a second reviewer using Microsoft Excel™. For study characteristics, the data extracted will include the location of study (city, municipality, prefecture and region), study year (year), study sample type (population based, convenient or others), setting (hospital or clinic), study design (RCT, cohort, case-control or cross-sectional), sample collection method (swab, cytobrush, cervical or vaginal wash or others), sample collection (self-collection, practitioner or others), type of cervical specimen (fresh biopsy, fixed biopsy or exfoliated), cell storage medium, HPV assay, PCR primers used and HPV typing method (DNA or RNA).
In the same extraction template, the sample size of the number of women tested (N) and the number of HPV-positive women (n) will be extracted. Where available, these will be grouped by cytological (normal, ASCUS, LSIL and HSIL) and histological classification (normal, CIN1, CIN2, CIN3, AIS, ICC—unspecified, ICC—squamous cell carcinoma or ICC—adenocarcinoma).
HPV genotype-specific prevalence data will be extracted and grouped by individual HPV genotypes, presence of any HPV type, any high-risk type (16, 18, 31, 33, 35,39, 45, 51, 52, 56, 58, 59), any low-risk type (6, 11), only low-risk types and genotypes affected by cross-protection (31, 33, 45). In addition, data will also be extracted by vaccine type. This will include the bivalent, quadrivalent and nonavalent vaccines. Each of which will include the corresponding combination of constituent HPV genotypes, i.e. one or more of those types was detected in that specimen. These are bivalent (16 and 18), quadrivalent (6, 11, 16 and 18) and nonavalent (6, 11, 16, 18, 31, 33, 45, 52 and 58), accordingly. Subsequently, genotypes that are defined as probably carcinogenic (68) and possibly carcinogenic (26, 53, 66, 67, 67, 70, 73, 82, 30, 34, 69, 85, 97) will also be extracted and grouped. Finally, prevalence data will also be extracted and grouped by HPV primary screening-detectable genotypes dependent on existing detection methods. For cohort and randomised studies, only the baseline prevalence data will be extracted.
Data analysis and synthesis
Stata version 15 will be used, utilising ‘Metaprop’, a Stata command to perform a meta-analysis of binomial data to calculate pooled prevalence estimates as described below [
34]. Analysis will be performed using the Freeman-Tukey double arcsine transformation, and Der Simonian-Laird random-effects methods will be used to compute the weighted overall pooled estimates with confidence intervals (CIs) [
35].
Statistical heterogeneity will be quantified using Cochran’s
Q and the
I2 test statistic to determine the extent of variation in effect estimates that is due to heterogeneity rather than chance. Cochrane’s
χ2 Q test statistic will be performed using an
α cut-off level of 10% [
31]. The
I2 test statistic will be used to quantify statistical heterogeneity between studies: heterogeneity from 0 to 30% will be classified as might not be important; heterogeneity from 30 to 75% will be classified as may represent moderate heterogeneity; and heterogeneity from 75 to 100% will be classified as considerable heterogeneity. In this study, the random-effects model will be chosen over the fixed-effects model. If there is substantial heterogeneity above 75%, a meta-analysis will not be performed.
HPV genotype-specific prevalence estimates
The pooled genotype-specific HPV prevalence for each HPV genotype or genotype group as previously defined will be estimated independently. Where data allows, pooled estimates will also be stratified by cytological disease stage: normal, ASCUS, LSIL and HSIL, and histological classification is defined as normal, CIN1, CIN2, CIN3, AIS, ICC—unspecified, ICC—squamous cell carcinoma or ICC—adenocarcinoma.
Age-specific prevalence estimates
The age-specific prevalence of each HPV genotype or genotype group as previously defined will be calculated for 5 year age groups for the interval 20 to 69 years. Studies that do not report age-specific HPV prevalence will be excluded from age-specific estimates.
Vaccine-type prevalence estimates
The proportion of vaccine-preventable infections will be estimated for each vaccine type group as previously defined. Where data allows, pooled estimates will also be calculated and stratified by vaccine type. This will include the bivalent, quadrivalent and nonavalent vaccines.
HPV prevalence estimates by geographic location
In order to examine the geographical distribution of HPV genotypes across Japan, pooled estimates of HPV genotype prevalence will be estimated nationally and stratified by region, prefecture and municipality where data allows.