Tel.: [+91-3324844345], Fax: [+91-3324751558]
Prevalence of high-risk human papillomavirus and cervical intraepithelial neoplasias in a previously unscreened population—A pooled analysis from three studies
Article first published online: 14 SEP 2012
Copyright © 2012 UICC
International Journal of Cancer
Volume 132, Issue 7, pages 1693–1699, 1 April 2013
How to Cite
Basu, P., Mittal, S., Bhaumik, S., Mandal, S. S., Samaddar, A., Ray, C., Siddiqi, M., Biswas, J. and Sankaranarayanan, R. (2013), Prevalence of high-risk human papillomavirus and cervical intraepithelial neoplasias in a previously unscreened population—A pooled analysis from three studies. Int. J. Cancer, 132: 1693–1699. doi: 10.1002/ijc.27793
- Issue published online: 28 JAN 2013
- Article first published online: 14 SEP 2012
- Accepted manuscript online: 21 AUG 2012 03:36AM EST
- Manuscript Accepted: 25 JUL 2012
- Manuscript Received: 5 JUN 2012
- Bill & Melinda Gates Foundation through Alliance for Cervical Cancer Prevention, USA
- Qiagen Inc, Gaithersburg, USA
- high-risk human papillomavirus;
- cervical intraepithelial neoplasia;
- population prevalence;
- hybrid capture 2;
- unscreened population
Population prevalence of human papillomavirus (HPV) and cervical intraepithelial neoplasias (CIN) is an important indicator to judge the disease burden in the community, to monitor the performance of cervical cancer screening program and to assess the impact of HPV vaccination program. India being a country without any cervical cancer screening program has no published data on the population prevalence of CIN and only a few large community-based studies to report the high-risk HPV prevalence. The objective of our study was to study HPV and CIN prevalence in a previously unscreened population. We pooled together the results of three research studies originally designed to assess the performance of visual inspection after acetic acid application and Hybrid Capture 2 (HC 2). Nearly 60% of the screened women had colposcopy irrespective of their screening test results. The diagnosis and grading of cervical neoplasias were based on histology. The age standardized prevalence of HPV by HC 2 test was 6.0%. Age-adjusted prevalence of CIN1 and CIN2 was 2.3% and 0.5%, respectively. The age-adjusted prevalence of CIN3 was 0.4% and that of invasive cancer was 0.2%. The prevalence of high-risk HPV was relatively low in the population we studied, which is reflected in the low prevalence of high-grade CIN. The prevalence of CIN3 remained constant across age groups due to absence of screening.
Cervical cancer is still the number one cancer among Indian women. Due to the absence of an effective cervical cancer screening program, 130,000 new cases are detected every year in the country.1 The extraordinary advances in our understanding of the pathogenesis of cervical cancer in the recent times have facilitated the development of novel human papillomavirus (HPV)-based screening strategies that have potential to overcome many of the limitations of cervical cytology. It is being hypothesized that in the near future HPV testing will be increasingly used as the primary screening test, and cervical cytology will be reserved to triage HPV-positive women for colposcopy.2 Hybrid Capture 2 (HC 2, Qiagen, Gaithersburg, MD) has been most widely evaluated among the clinical HPV detection technologies globally and in India.3 Information on the population prevalence of high-risk HPV as well as cervical intraepithelial neoplasias (CIN) is crucial to understand the burden of disease. This not only helps in planning appropriate screening strategies but also provides key inputs for resource allocation and quality assurance of the program. In the context of HPV and CIN prevalence, the predominantly unscreened population of Indian women is likely to be different from their counterparts in western or other developed Asian countries who have the benefit of regular screening. Population-based information on the burden of cervical precancers is also important to assess the impact of HPV vaccination in the future. In the present publication, we report the population prevalence of high-risk HPV and different grades of CIN estimated from 35,308 women recruited in three different cervical cancer screening studies conducted by Chittaranjan National Cancer Institute (CNCI), Kolkata, India.
Material and Methods
Between 1999 and 2011 three cross-sectional community-based cervical cancer screening studies were carried out by Chittaranjan National Cancer Institute in the districts of West Bengal, a state in Eastern India. The study sites were villages and towns located 30–120 km away from Kolkata and were selected based on the logistic convenience of organizing cervical cancer screening, e.g., accessibility by road, availability of health centers, permission granted by local health authorities, etc. In the selected areas all the households were surveyed by trained field workers to identify the eligible women and invite them for screening. Women aged between 30 and 65 years with intact uterus, ever married, not pregnant and with no history of CIN or cervical cancer were eligible to undergo cervical cancer screening. None of these women ever had cervical cancer screening as the test facilities were not available in those districts. Out of the all eligible women invited to take part in screening, ∼65% complied.
The first two research studies were in collaboration with International Agency for Research on Cancer (IARC), Lyon and were funded by Bill & Melinda Gates Foundation through Alliance for Cervical Cancer Prevention. The third study was funded by Qiagen Inc. (Gaithersburg, MD) through their Indian operations. All the studies were approved by CNCI Research Ethics Committee and all the participants signed written informed consent. The objectives and the methodologies of the studies were different (Table 1) and are described in brief below.
The first cervical cancer screening study aimed at evaluating the performance of different cervical cancer screening techniques. Participating women underwent screening by HC 2 to detect high-risk HPV DNA and visual inspection with acetic acid (VIA). Colposcopy was performed on all screened women irrespective of their screening test results. Directed biopsies were obtained from colposcopically assessed abnormal areas of cervix. The results of this study focusing on the performance of different screening tests have already been published.4–6
In the second screening study, all recruited women underwent VIA followed by colposcopy, irrespective of the VIA test results. If colposcopy revealed any abnormality on the cervix, guided biopsies were taken. The women detected to have CIN of any grade were considered for recruitment to a randomized study aimed at evaluating the performance of single freeze versus double freeze cryotherapy. In both the studies, the colposcopists were blinded to the results of the screening tests.
In the third study, all the participating women underwent screening with VIA and HC 2. If either of the screening tests was positive, the women were subjected to colposcopy followed by guided punch biopsies from areas colposcopically suspicious for cervical neoplasia. In HC 2 positive women without any visible lesion on colposcopy, a punch biopsy was directed from close to the squamo-columnar junction of anterior lip of cervix. This study is ongoing and aims to evaluate the performance of HC 2 and VIA when used in parallel as well as sequentially for the detection of CIN3 or cervical cancers. Data generated from women recruited till the end of 2011 have been used for the present analysis.
In all the studies, VIA was performed by trained female health workers using the VIA outcome categories (negative, positive or invasive) recommended by the IARC Visual Screening Manual.7 Sample for HC 2 test were collected by a health worker first using a brush which was dipped in a container with liquid preservative. The same health worker then performed VIA. The samples for HC 2 test were transported in iceboxes to CNCI virology laboratory for further processing. HPV DNA testing was performed using the high-risk probe set of HC 2, which detects 13 carcinogenic HPV types (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59 and 68). Signal strengths in relative light units (RLU) were compared to 1 pg/mL HPV type-16 DNA positive controls (PC), and specimens with RLU/PC ratios of one or greater were judged positive.8
The prevalence analysis of high-risk HPV and cervical neoplasias was performed from the pooled data of the three studies. Histopathological grading of cervical neoplasias was considered to be the gold standard for CIN and cancer diagnosis. Overall, 933 cases of CIN1—invasive cancer were detected. The distribution of these cases by the screening test (VIA and HC 2) results are shown in Table 2. The details of the inclusion and exclusion criteria applied to identify subjects as denominators for the analysis are given in Figure 1.
HC 2 results were available for 28,309 women and all of them were considered to estimate the crude and age-standardized HPV prevalence.
To analyze the prevalence of CIN1 and CIN2, all women with disease status confirmed by histology and women with normal colposcopy, even if they did not have biopsy, were included. The second group of women was assumed to be negative for cervical neoplasias based on the colposcopic diagnosis. All women who were positive on either VIA or HC 2 or both but did not have colposcopy were excluded. Women negative on both VIA and HC 2, were excluded if they did not have colposcopic confirmation of disease status. The reason being, in our first study (in which both screen positive and screen negative women had colposcopy) a substantial number of CIN2 (N = 24; 0.21%) and CIN1 cases (N = 111; 1.08%) were detected in this dual test negative group (N = 10,272). Women with colposcopy diagnosis of low-grade neoplasias or worse without any histology report were also excluded from analysis.
To determine the prevalence of CIN3 and cervical cancer, all women in whom the final disease status was assessed by histopathology were included. Women with normal or low-grade abnormalities on colposcopy even if they did not have a cervical biopsy were considered to be negative for CIN3 or cancer and were included in the analysis. Women negative on both VIA and HC 2 were included in the analysis and were assumed to be negative for CIN3 or cancer even if they did not have colposcopy. This is based on our data from the first study where only one cancer (0.009%) and three CIN3 (0.02%) were detected among the VIA negative and HC 2 negative women evaluated by colposcopy (N = 10,272). VIA and/or HC 2 positive women not having colposcopy or VIA negative women who did not undergo HC 2 test or colposcopy were excluded from the analysis. Women with colposcopic diagnosis of high-grade CIN or cancer but without a conclusive biopsy report were also excluded from the analysis.
The age-adjusted prevalence was obtained by calculating age-specific prevalence estimates for the age groups 30–34, 35–39, 40–44, 45–49, 50–54, 55–59 and 60–64 years by using WHO World Population Standard from the year 2000 to 2025 as the reference.9
Prevalence of high-risk HPV
The crude prevalence of high-risk HPV detected by HC 2 test in women aged between 30 and 65 years was 5.8% (95% C.I. 5.5–6.1%). The age-adjusted prevalence was 6.0%. The mean age of the HPV negative women was 38.2 ± 8.2 years and that of the HPV positive women was 38.4 ± 8.5 years. The difference was not statistically significant (p = 0.34). As shown in Figure 2, the prevalence of high-risk HPV infection was similar across the age groups (X2 = 5.94, df = 6, p = 0.43). In the studied population, women get married at early ages and for majority of them sexual life starts after marriage. There was no significant difference in the mean age at marriage between the HPV negative women and the HPV positive women (18.9 ± 12.4 years vs. 18.6 ± 12.3 years; p = 0.34). However, due to socio-cultural reasons it was not feasible to enquire about the age of sexual debut of individual woman.
The risk factors identified on univariate analysis for being HPV positive by HC 2 test were young age, early marriage, low literacy and multiple child birth. On multivariate analysis the risk of being HPV positive was higher for the women between 30 and 39 years of age compared to those between 40 and 65 years [OR = 1.13; 95% C.I. 0.96–1.33]; the women getting married before 15 years compared to those getting married at higher age [OR = 1.20; 95% C.I. 1.01–1.44]; the women with no education or only primary education compared to women having higher level of education [OR = 1.29; C.I. 1.08–1.55] and the women having more than three pregnancies than those having less number of pregnancies [OR = 1.10; C.I. 0.93–1.30].
Prevalence of CIN1 and CIN2
Among the 21,031 women eligible for calculation of prevalence of CIN1 and CIN2, there were 611 cases of biopsy proved CIN1 and 130 cases of biopsy proved CIN2. The crude prevalence of CIN1 and CIN2 were 2.9% (95% C. I. 2.7–3.1) and 0.6% (95% C.I. 0.5–0.7%), respectively. The age-adjusted prevalence of CIN1 was 2.3% and of CIN2 was 0.5%. Prevalence of CIN1 and CIN2 stratified by age group is shown in Figure 2.
Prevalence of CIN3 and invasive cancer
A total of 34,882 women were eligible for the estimation of prevalence of CIN3 and invasive cancer. In this group, there were 126 CIN3 and 66 invasive cancers confirmed by histology. The estimated crude prevalence of CIN3 was 0.4% (95% C.I. 0.3–0.4) and that of invasive cancer was 0.2% (95% C. I. 0.1–0.2). The age-adjusted prevalence was 0.4% and 0.2% for CIN3 and cancer, respectively. The mean age of the women with CIN3 diagnosis was 40.0 ± 8.4 years and that of the women with invasive cancer was 46.0 ± 9.3 years. The prevalence of CIN3 and invasive cancer in different age groups is shown in Figure 3. When stratified by age, the prevalence of CIN3 was generally constant after the age of 40 years, whereas the prevalence of invasive cancer increased gradually with age and peaked in 60–65 years age group. Adenocarcinoma was detected only in four cases, two of them were aged between 35 and 40 years and two between 40 and 45 years.
Relation between HPV viral load and grade of cervical neoplasia
The RLU/PC values were available for 16,259 women screened by HC 2 in the third study. The relationships between different ranges of RLU/PC values and the cervical neoplasias detected are shown in Table 3. With increasing RLU/PC values the risk of detection of CIN3+ disease increased significantly.
Prevalence of high-risk HPV
The crude high-risk HPV prevalence of 5.8% observed in our study is considerably less than the prevalence reported by other studies in India. A community-based study from Dindigul, a rural district in South India observed 12.5% prevalence of high-risk HPV.10 The higher prevalence in the Dindigul study can be explained by the facts that the target age was different (16–59 years) and a Polymerase Chain Reaction (PCR) based method was used for the detection of HPV. Even then, the difference may reflect the true disparity in the population burden of HPV and the consequential risk of cervical cancer between the two regions. We compared the age-adjusted incidence rate of cervical cancer reported by the population-based cancer registries of Kolkata (city close to our study area) and of Chennai (city close to Dindigul). The incidence of cervical cancer in Kolkata (19.3/100,000 population) was much lower than that in Chennai (38.9/100,000 population).11 Another large community-based study from India that reported HPV prevalence was conducted at Barshi, a rural district in west-central India.12 The women in that study were tested by HC 2 and had almost similar age group (30–59 years) as ours. Even then, the Barshi study reported much higher HPV prevalence of 10.3%. This high-HPV prevalence may account for the high age-adjusted incidence rate of cervical cancer (27.4/100,000) reported by Barshi cancer registry.11 This is commensurate with the observation by Bosch and de Sanjose that without the protective effect of screening, high incidence rates of cervical cancer are observed and can be predicted for populations with high HPV-DNA background prevalence.13
Prevalence of CIN and cervical cancer
Information on population prevalence of cervical intraepithelial neoplasia from an unscreened population as in India is relatively scarce. A recent global review of 103 studies reported population prevalence of CIN involving 12,400,000 women of which only 14% were from Asia and none from India.16 In the review, the age-specific prevalence data were limited even among the few studies from Asia.
We reported the age-adjusted prevalence of CIN1, CIN2 and CIN3 to be 2.3%, 0.5% and 0.4%, respectively, among women between 30 and 65 years who have never been screened. Across the globe CIN1 is more frequently observed in young women, is commonly associated with transient HPV infection and does not predict any meaningful risk of CIN3.17, 18 We also observed a high prevalence of CIN1 in 30–35 years age group that came down to less than 2% in the women beyond 45 years and remained almost static after that. Out of the 458 CIN1 cases with HC 2 test results only 21.0% were HPV positive. The mean age of the HPV positive CIN1 cases was significantly greater than the mean age of the HPV negative CIN1 cases (37.0 ± 7.6 years vs. 35.0 ± 6.2 years; p = 0.008). This observation may signify that the HPV positive CIN1 encountered in relatively older women have higher possibility of persistence/progression though there is no evidence to substantiate this hypothesis.
CIN2 diagnosis can be equivocal due to misclassification of disease and can be caused by non-high-risk HPV genotypes.19, 20 It has been observed that CIN2 diagnosed following an high grade squamous intra-epithelial lesion (HSIL) cytology and CIN2 associated with HPV 16 are most strongly linked with CIN3 and least regressive. They probably have different natural history than the other CIN2 cases.21 In our study, out of the 96 CIN2 cases having HC 2 results, only 43.7% were high-risk HPV positive. The mean age of the HPV positive CIN2 cases was higher than the HPV negative CIN2 cases (37.0 ± 7.2 vs. 36.0 ± 7.4 years) though the difference was not statistically significant (p = 0.5). Compared to other studies evaluating HC 2, the HC 2 positivity for CIN2 in our study was significantly lower. All these HC 2 negative CIN cases were detected when biopsies were obtained from VIA positive women with colposcopically visible lesions. The natural history of these HC 2 negative CIN2 cases detected on VIA and colposcopy is unknown and requires further study. Misclassification of CIN2 on histology is also a possibility. The observed prevalence of CIN2 was constant across the age groups.
Unlike CIN1 or CIN2, CIN3 are the true cervical cancer precursors. Given adequate follow-up time and if left untreated 30–50% of them will progress to invasive disease.22 The age-specific prevalence of CIN3 in our study gradually increased from 30 years of age and remained almost constant after the age of 40 years showing a slight decline in the sixth decade of life. This is in contrast to the CIN3 prevalence observed in a screened population where the prevalence peaks among women aged 35–39 years and then comes down sharply and very few cases are detected beyond 60 years.23 We observed that 17.0% of the CIN3 cases were high-risk HPV negative. HC 2 positive CIN3 cases had significantly higher mean age compared to the HC 2 negative CIN3 cases (41.0 ± 8.6 years vs. 36.0 ± 6.2 years; p = 0.03).
Relation between HPV viral load and detection of CIN
HC 2 assesses the viral load of high-risk HPV semi-quantitatively. We have demonstrated that with higher HPV viral load, the detection rates of CIN3+ disease increase significantly. The true clinical relevance of viral load observed by HC 2 testing is not yet established. In a French prospective study, baseline HPV positivity was associated with an increased risk of incident cytologic abnormalities, especially when the load was high (≥100 pg/mL, RR 8.7). HR-HPV load was demonstrated to have a significant influence on subsequent CIN2/3 development in that study. The authors concluded that high viral load could be used as a short-term marker of progression toward precancerous lesions, although lower load did not inevitably exclude progressive disease.24 Other studies have failed to demonstrate such predictive value of high viral load detected by HC 2.25, 26
The choice of 1.0 RLU/PC positive cut point was based on a population sample of disease at the time of validating the HC 2 test.27 Our study provided the opportunity to re-examine the optimal HC 2 cut point in a large sample of women from India. It has been postulated that raising the RLU/PC cutoff for the HC 2 test positivity beyond the presently accepted value of 1 can increase clinical specificity of this assay but there is a risk of loss of sensitivity.28 In our cohort of women screened by HC 2, if we increased the RLU/PC cutoff for test positivity to 2, the test specificity would have improved slightly since additional 185 women with normal or CIN1 histology would have tested negative. But in doing so we would have missed one CIN2 and one invasive cancer. On the other hand if we reduced the cut point to 0.5, sensitivity would not have improved much as we could detect one additional CIN2 only and no additional CIN3+ lesions. Specificity would have dropped appreciably since 747 additional histology proved normal or CIN1 cases would have tested positive due to reduced threshold.
Strengths and weaknesses of the study
Our study has certain limitations. The study sites have been selected on the basis of logistic convenience rather than by systematic sampling of the population and even in those sites the compliance of target women was suboptimal. However, a sample survey performed earlier by us among the non-compliant women observed that their age distribution and socio-economic status was not significantly different from the participating women.29 Hence the prevalence rates are unlikely to be significantly affected by the low compliance of the target women. A more rigorous workup to establish true disease status of the screened women, like colposcopy and biopsy for all, could increase the detection rate of cervical neoplasias. A cross-sectional study in China observed that nearly 20% of all CIN2 and CIN3 had normal colposcopy findings and the disease could be detected by four quadrant biopsies only.30 However, such an approach would not have been acceptable to our study subjects and could not be pursued due to ethical reasons.
We have excluded 300 cases in which colposcopy diagnosis was CIN1 or worse and biopsies were either not obtained or were inconclusive. Non-inclusion of these high-risk cases could potentially reduce the prevalence of CIN2 and CIN3 in our study. To ascertain that, we estimated the predictive value of colposcopic diagnosis of CIN1+ to detect CIN2 and CIN3 in our study. Only 7.0% of the women with colposcopy diagnosis of CIN1 or worse had biopsy proved CIN2 or CIN3 and according to this estimate 21 additional cases of CIN2 and CIN3 could be detected among those 300 cases without conclusive biopsy reports. Similarly, CIN1 prevalence could also be underestimated due to non-inclusion of these women. Non-inclusion of VIA/HC 2 positive women could also be the reason for low prevalence of CIN in the study.
HC 2 is not the ideal test to estimate the prevalence of HPV. The authors acknowledge that interpretation of CIN, especially that of CIN1 and CIN2, is subjective and a formal review of all the slides could provide more validity to the prevalence results. Such blinded review was not done. However, as a routine clinical practice, the opinion of a second pathologist was sought if there was discrepancy between colposcopy and histology diagnosis. The histology diagnosis arrived at by consultation among the two pathologists was considered as final.
In spite of these limitations the study has for the first time revealed the population prevalence of CIN1–CIN3 in a community-based large cohort of unscreened women from India who have undergone reasonable number of investigations (VIA and HC 2 for screening and colposcopy for more than 60% of the screened women) before being included in prevalence analysis. Identifying high-risk HPV by HC 2 is clinically and epidemiologically more relevant since women who test negative on HC 2 and still have HPV infections detectable by PCR have very low risk of developing CIN3+ disease.31
The regional variations in high-risk HPV prevalence can be responsible for different incidence rates of cervical cancer across the regions in India. In our population with low HPV burden, the prevalence of high-grade lesions was not high but had the age distribution typically observed in an unscreened population. The observed long interval between the mean age at occurrence of CIN3 (40 years) and of the mean age at sexual initiation (18 years) in the unscreened population confirms the long latency period of HPV infection before it leads to cervical cancer.
- 1International Agency for Research on Cancer (IARC). GLOBOCAN2002: cancer. Incidence, mortality and prevalence worldwide. Lyon, France: IARC Press, 2004, Cancer Base No. 5, version 2.0., , , et al.
- 7A practical manual on visual screening for cervical neoplasia. IARC technical publication no. 41. Lyon: International Agency for Research on Cancer, 2003., .
- 9Age standardization of rates: a new WHO standard. GPE discussion paper series: No.31 EIP/GPE/EBD. World Health Organization, Geneva, 2001., , , et al.
- 11Cancer epidemiology, prevention and control. Curr Sci 2004; 86: 518–27., .
- 13Chapter 1: Human papillomavirus and cervical cancer—burden and assessment of causality. J Natl Cancer Inst Monogr 2003; 31: 3–13., .
- 16A global review of age-specific and overall prevalence of cervical lesions. Int J Gynecol Cancer 2010; 20: 1244–49., , .
- 18ASCUS-low-grade squamous intraepithelial lesion Triage Study (ALTS) Group. Prospective follow-up suggests similar risk of subsequent cervical intraepithelial neoplasia grade 2 or 3 among women with cervical intraepithelial neoplasia grade 1 or negative colposcopy and directed biopsy. Am J Obstet Gynecol 2003; 188: 1406–12., , .
- 26Recurrent human papillomavirus infection detected with the hybrid capture II assay selects women with normal cervical smears at risk for developing high grade cervical lesions: a longitudinal study of 3,091 women. Int J Cancer 2002; 102: 519–25., , , et al.