Cervical cancer continues to be one of the leading female genital cancers worldwide.1 The cervical cancer burden in Taiwan is still heavy with the incidence of invasive cancer being 24.5/100,000, which ranked the fifth place of cancer death in women.2 Traditionally, primary screening of cancer precursors remains based on cytological classification of cervical smears. Conventional Papanicolaou (Pap) test or monolayer cytology provided sensitivity rates ranged from 30% to 87% and specificity rates from 86% to 100%.3 Since human papillomavirus (HPV) has been established as a well-recognized etiologic agent of cervical neoplasia, HPV DNA testing is a promising alternative or complementary test to improve the efficacy of cervical screening and probably cost-effective, depending on screening interval.4, 5 Cervical cancer screening programs integrating HPV testing have been advocated in United Kingdom, United States, South Africa, Thailand and Germany.6–10
Hybrid Capture assay (HCII) (Digene Corporation, Gaithersburg, MD) is the most widely used method of HPV DNA testing. HCII utilizes liquid hybridization of RNA probes against HPV DNA followed by signal amplification, has been validated for its reproducibility.11 Methods using PCR is more sensitive and quality assays were generally consistent for the proficiency of various techniques in detection of HPV DNA in an international collaborative study with 29 laboratories.12
SPF1/GP6+ PCR followed by revert blotting with a HPV genechip (Easychip® HPV Blot, Kingcar, Taiwan) was found sensitive and reproducible for HPV DNA detection and genotyping in cervical swab specimens as well as paraffin-embedded tissues in our previous studies.13–15 The HPV sensitivity and reliability is comparable to HC II for cervical swab samples.13
Prophylactic HPV vaccines have been designated against the most common oncogenic HPV types 16 and 18.16, 17 Knowledge of baseline population-based HPV genotype distribution is important as a reference to assess the effect of prophylactic HPV vaccines on HPV infection after implementation of vaccination.18 In our previous study (n = 2,118), though HPV-16 remains the most prevalent type, HPV-58 or -52 or -33 is detected in 30.3% of stage I–IV cervical cancer of Taiwanese women.14
The primary endpoint of the study was to investigate the additional value of HPV DNA testing as a complementary method to Pap smear to improve the sensitivity in detecting a case of cervical intraepithelial neoplasia (CIN) grade 2 in a cross-sectional study of a population-based cohort. The secondary endpoint was to investigate HPV DNA prevalence and genotype distribution in females residing northern Taiwan and the subgroup(s) benefited the most from a combined screening strategy. The whole project is still ongoing to evaluate safety and cost-effectiveness of screening with Pap plus HPV testing at longer intervals in comparison with yearly Pap smears, which question can then be answered after long-term follow-up. In this report, we reported the design of the whole study and analyzed the cross-sectional results.
Material and methods
Eligibility criteria and patient recruitment
Females aged 30 years or older who were residents of Taoyuan, Taiwan, were invited. The number of subjects of each township was proportional to its population. Besides, a priority to invite those had not had Pap smears for at least 3 years (>50% of the study subjects) was required. Participants should have an intact uterus and no history of treatment for preinvasive or invasive cervical neoplasms. After signing informed consent, cervical swab samples were collected in women in a mobile examination booth or at outpatient clinic by gynecologists from Chang Gung Memorial Hospital (CGMH), local practitioners at their offices, or public health nurses at their health stations. Every participant had two cervical swabs: the first for a Pap smear and a second swab for HPV testing. A systemic questionnaire including age, menopausal status, use of hormone replacement therapy, use of oral contraceptives, history of abnormal Pap results or treatment for cervical neoplasms and interval from previous Pap smear was implemented. Only those with both adequate DNA and Pap quality were eligible for analysis. Cytologic reports were signed out by local certified cytology lab or at CGMH. Histopathologic interpretation was performed by 2 of our pathologists (S.H. and S-M.J.)
The 2001 Bethesda System was used for terminology for reporting results of cervical cytology.19 The flow diagram (Fig. 1) was as follows: women with abnormal Pap results (atypical squamous cells of undetermined significance [ASCUS] or worse) were designated Group I; women with normal cytology but positive HPV DNA were designated Group II; and Group III represented who had both tests negative. Group Ia were those had a cytology results greater than high-grade squamous intraepithelial lesion (HSIL) including adenocarcinoma in situ (AIS) and adenocarcinoma, and colposcopic examination with directed biopsy was recommended. Group Ib were those had ASCUS/low-grade squamous intraepithelial lesion (LSIL) or glandular abnormalities, for whom colposcopy was recommended but biopsy was not mandatory if the finding was normal or a diagnosis of CIN 1.
Group II patients were notified and referred for a colposcopic examination and directed biopsy if indicated. Additional cases of CIN 2 identified from Group II represented the improvement of sensitivity by HPV testing. Group IIa patients were those receiving an immediate colposcopy whereas Group IIb were those not receiving colposcopic examination at CGMH. The histopathology readings were masked to the cytopathology readings. However, the colposcopists were not masked with cytology or HPV findings.
Randomly selected 400 women (Group IIIa, n = 415) would be followed annually by Pap and HPV testing. Another informed consent to participate follow-up study for Group II and Group IIIa was obtained and a more detailed questionnaire including sex history was used. The frequencies of acquisition of HPV infection and emergence of cervical neoplasia could be calculated from Group IIIa, while those of Group IIIb would be extrapolated from Group IIIa data. Therefore, the frequency of newly acquired HPV infection in a low-risk group (Group III) could be determined. The actuarial cumulative incidence rates of abnormal Pap smear and cervical neoplasia (to be calculated from linkage to the national database with informed consent of all the participants) could be compared between Group IIIa and Group IIIb to justify the extrapolation of HPV incidence. With long-term follow-up, the cost-effectiveness and diagnostic efficacy of screening every 3–5 years with both tests for those who were cytology(−)/HPV(−) at enrollment could be determined.
Sample collection and DNA extraction
Cervical swab samples were collected using cytobrush and transported in a storage medium (manufactured by King Car, I-Lan, Taiwan). DNA was extracted according to the protocol for isolation of total DNA from cultured animal cells (Qiagen, Valencia, CA).13, 14 Finally, 100 μl of DNA solution was eluted and 1 μl of the aliquot was used for polymerase chain reaction (PCR) amplification.
The SPF1/GP6+ consensus primers were used to amplify a fragment of approximately 184 bp in the L1 open reading frame. The sequences were:
The protocol was detailed previously. Each PCR experiment was performed with several positive and negative controls. Routine precaution procedures were applied to avoid carrying-over or contamination.13, 14
Quality of isolated DNA was checked with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) PCR, by which a 136 bp product was amplified. Forty cycles were performed. Because co-amplification of house-keeping gene primers could reduce the efficiency of target HPV DNA amplification,13, 14 GAPDH PCR was performed when HPV-negative results were obtained at the first round.
HPV genotying by Easychip® HPV blot
Fifteen microliters of the resultant amplimers were then hybridized with an Easychip® HPV Blot (King Car, I-Lan, Taiwan) (hereafter HPV Blot) membrane. Thirty-eight types of HPV (6, 11, 16, 18, 26, 31, 32, 33, 35, 37, 39, 42, 43, 44, 45, 51, 52, 53, 54, 55, 56, 58, 59, 61, 62, 66, 67, 68, 69, 70, 71 [CP8061], 72, 74, 81 [CP8304], 82 [MM4], 83 [MM7], 84 [MM8], L1AE5) oligonucleotide probes of 20-to 30-mer with an approximately 100–200-mer poly-T tail are immobilized on a nylon membrane, which is used for reverse-blot hybridization and detects HPV DNA in a single reaction. The hybridization and detection procedures were described previously.13, 14 HPV types were divided into high-risk and low-risk groups: the leading 14 HPV types (HPV-16, 18, 58, 33, 52, 39, 45, 31, 51, 70, 56, 59, 35, 53) of cervical cancer in Taiwanese population15 were grouped as high-risk and the remaining 24 HPV types were low-risk.
The data were analyzed by the PC-SAS software (version 9.1; SAS Institute, Cary, NC). Pearson's χ2 test was used to evaluate the proportion differences between designated groups. The strength of association was calculated by odds ratio (OR) and its 95% confidence interval (CI) between categorical variables. Continuous covariates were compared between groups using both parametric and nonparametric approaches, mainly, the student t and Mann-Whitney U test when appropriate. The total true positive (TP) cases were defined by histologically confirmed CIN 2 cases found by combined test plus cases retrieved from national registry database within 9 months of accrual (the permit for inquiry of such information was addressed in the informed consent). Percent of sensitivity improvement was determined by cases disclosed from Group I and Group II minus cases from Group I and divided by TP cases. The association between factors and likelihood of HPV-positivity and benefit was expressed as OR and was examined by Mantel-Haenszel χ2 test and test for monotonic trend. Multiple logistic regression analysis was performed to calculate multivariate-adjusted odds ratios and their 95% CI with covariates selected from univariate analyses. All tests were two-sided, and a p value less than 0.05 was considered statistically significant.
Between June 2004 and June 2005, a total of 10,014 eligible women were recruited. The reasons for exclusion were invalid Pap (n = 43), after hysterectomy or treatment for CIN or cervical cancer (n = 14), age < 30 years (n = 13), and missing data of Pap, HPV test or questionnaire (n = 59) (Fig. 1). The median age of participating women was 46 (30 to 94 years). Overall, 3,478 (34.7%) were postmenopausal, among whom 431 were current user of hormone therapy. The remaining 6,536 (65.3%) were premenopausal, among whom 1,387 were previous or current user of oral contraceptives. A total of 7,567 (76.6%) had never or not received Pap smear in 3 years (Table I).
Table I. HPV and Abnormal Pap Prevalence According to Demographic Characteristics
Prevalence,% (95% CI)
Pap ≥ ASCUS
Prevalence,% (95% CI)
HPV, human papillomavirus; CI, confidence interval; OR, odds ratio; ASCUS, atypical squamous cells of undetermined significance.
The trends were similar with age-adjusted calculation.
The overall HPV prevalence was 10.8% (95% CI, 10.5%–11.4%) of all the participants. There was a significantly positive correlation of HPV prevalence with age tested by monotonic trend and statistical significance was found in women aged 51–59 and >60 years. Furthermore, women who were postmenopausal, had never received hormone replacement therapy, and were current oral contraceptives users had significantly higher HPV prevalence adjusted for age (Table I). The prevalence of HPV-positive according to age is demonstrated in Figure 2. When separating the HPV-positives into high-risk and low-risk groups, the prevalence of high-risk group doubled the low-risk group in each age category. Furthermore, both high-risk and low-risk groups presented with a positive correlation with increasing age. The prevalence of cytology ASCUS was in ‘U’ shape association with women age, significantly excessive prevalence were found in women aged 30–39 and >60. In addition, a higher risk of cytology ASCUS was found in oral contraceptives past user, however the OR of current oral contraceptives users was not significant higher as compared with nonuser because the number (n = 3) was too small (Table I). Although the Pap smear history was not significantly associated with prevalence of HPV (p = 0.10) or abnormal Pap (p = 0.09), those never had a Pap smear or last Pap ≥ 3 years had an increased risk of histology CIN3 than those with a last Pap ≤ 3 years (last Pap < 3 years vs. ≥ 3 years vs. never: OR, 1.0 vs. 3.1 [95%CI 1.2–8.0] vs. 3.9 [95%CI 1.4–10.7]).
Because of collinearity phenomenon occurred in the presence of menopausal status and hormone replacement therapy, multivariate analyses were performed excluding hormonal replacement therapy. In addition, strong correlations were noted in age and menopausal status, therefore menopausal status and age were entered alternatively in two models (Table II). There was a significant positive correlation of HPV prevalence with older age (or postmenopausal status), current-user of oral contraceptives, and never-user of hormone replacement therapy. Past users of oral contraceptives and never users of Pap were associated with significant higher risk of abnormal Pap, while age 40–49 strata had lower risk.
Table II. Multivariate Analyses of HPV Prevalence and Abnormal Pap According to Demographic Characteristics
Pap ≥ ASCUS
Adjusted OR (95% CI)
Adjusted OR (95% CI)
HPV, human papillomavirus; CI, confidence interval; OR, odds ratio; ASCUS, atypical squamous cells of undetermined significance.
Hormonal replacement therapy was not entered in multivariate analyses due to collinearity phenomenon occurred in the presence of menopausal status. Besides, strong correlations were noted in age and menopausal status, therefore age and menopausal status were entered alternatively in two models.
Of the 10,014 women, 9,816 (98.0%) had normal cytology, and 198 (2.0%) abnormal cytology. The distributions of women in the 3 groups were: 198 in Group I, 970 in Group II, and 8,846 in Group III (Fig. 1). The prevalence of HSIL (Group Ia) and ASCUS/LSIL (Group Ib) were 0.45% and 1.53%, respectively. Among the Group I subjects, 29.8% (59/198) had a pathology result of CIN 2. Nine cases with CIN2 and 23 CIN1 were found from Group IIa. The other 91 patients with ASCUS/LSIL had either normal colposcopy or a colposcopic diagnosis CIN 1 without biopsy confirmation or had not been evaluated for abnormal Pap smears. From the national database, there were 33 patients from Group III and 9 from Group II underwent cervical biopsy, 4 with CIN 2 (2 from Group IIIb and 2 from Group IIb) and 8 with CIN 1 were disclosed. The total number of TP CIN2 cases was 72. The prevalence of CIN2 was 0.7%. In addition, 11 patients with CIN2/3 were identified in Group II. There was no specific subgroup found to most benefit from combined screening strategy (data not shown).
Histological and HPV results according to cytology grades are presented in Table III. The HPV prevalence of according to cytology results ASCUS, LSIL, HSIL and squamous cell carcinoma were 41.2%, 76.9%, 87.5% and 66.7%, respectively. Combining Figure 1 and Table III, the sensitivity of Pap smear was 81.9% (59/72), and specificity of 98·6% (9,803/9,942) and the sensitivity of HPV was 87.5% (63/72), and specificity of 89.7% (8,921/9,942). Furthermore, the sensitivity of high-risk HPV was 79.2% (57/72), while specificity was 92.4% (9,184/9,942). Overall, the sensitivity of Pap plus HPV testing was 97.2% (70/72) and specificity of 89% (8,844/9,942), while the sensitivity of Pap plus high-risk HPV testing was 91.7% (66/72) and specificity of 91.9% (9,140/9,942). The improvement of sensitivity in Pap with additional HPV testing was 15.3% with specificity decreased of 9.6%. In contrast, the improvement of sensitivity in primary HPV testing with additional Pap was 9.7% and specificity decreased of 0.7%, and the improvement of sensitivity in primary high-risk HPV testing with additional Pap was 12.5% and the specificity decreased of 0.5%.
Table III. HPV Positivity in Histological Results ≥ CIN 2 According to Pap Smear Grades (N = 10,014)
Pap smear grade
Histology ≥ CIN2
CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; ASCUS/LSIL, atypical squamous cells with undetermined significance or low grade squamous intraepithelial lesions; HSIL, high grade squamous intraepithelial lesions; AIS, adenocarcinoma in situ; SCC, squamous carcinoma.
Include atypical glandular cells, atypical glandular cells favor neoplasm, atypical squamous cells cannot exclude HSIL, and dysplasia cannot exclude HSIL.
One case is endometrial cancer; another did not have colposcopic examination and nothing has been retrieved from the national database as yet.
The one case with HPV negative by HPV Blot was further verified by type-specific PCR showing HPV-52 positive.
HPV genotype distribution of histological diagnosis of CIN 2 according to HPV status and HPV genotype are depicted in Table IV. A total of 37 types of HPV were identified. Among the 1,084 cases with positive HPV, 80.1% were single-type and 18.9% were multiple-type. The leading 5 types in prevalence were HPV-52 (1.4%), HPV-18 (1.2%), HPV-58 (1.1%), HPV-53 (1.0%) and HPV-70 (0.7%), respectively. On the other hand, the 4 leading probability of association with CIN2 HPV types were HPV-16 (41.7%), HPV-35 (33.3%), HPV-31 (28.0%) and HPV-33 (18.9%). Excluding multiple infections among the cases with CIN2, HPV-16 (n = 11) was the most prevalent followed by HPV-52 (n = 9), HPV-58 (n = 7) and HPV-33 (n = 5) among those single infections (n = 48).
Table IV. HPV Genotype Distribution and Odds Ratios of Histological Diagnosis ≥ CIN 2 According to HPV Status and HPV Genotype (N = 10,014)
Gold standard of TP among various studies investigating value of HPV testing and cytology often involves colposcopic directed biopsy of women with positive tests. Verification by examining all test-negative cases would be impractical in a large population-based study, while examining a random sample of the population would inevitably be limited. The national insurance (with a coverage of >98% of the whole population) provides yearly Pap smear screening program for all women older than 30-year-old has started since 1994 in Taiwan, in which the policy requires reporting all such cytology and histology results to national registry to get reimbursement. Thus, the national database for cytology and histology results provides a correction of estimating TP cases. In our study, the total TP cases defined by histology-confirmed CIN 2 identified by either test plus cases retrieved from national database within 9 months of accrual is a partial remedy of verification bias.
Though Pap smear remains the typical example of an important preventive intervention, the reduction in cervical cancer relies on regular serial screening tests. Cost of supplementary tools and triage methods vary across different countries, therefore individualized strategies are required to establish an effective cervical cancer prevention program. Generally, HPV DNA testing has a higher sensitivity but a lower positive predictive value (PPV) than cervical cytology.5 In our study, the sensitivity of Pap smear was 81.9%, and specificity of 98·6%. In contrast, the sensitivity of HPV was 87.5%, and specificity of 89.7%.The improvement of sensitivity with adding HPV testing to Pap smear was 15.3%. In contrast, the improvement of sensitivity in primary HPV testing with additional Pap was 9.7%.
In our study, the overall HPV prevalence was 10.8% and the leading 3 were HPV-52, -18 and -58 among a representative sample of general population (age ≥ 30) in northern Taiwan. Lin et al. reported a hospital-based study in southern Taiwan (n = 4,383) that the age-specific curves of HPV prevalence was U-shape and prevalence for age ≤ 30 (32%) was higher than that of age > 30 (17.2%).20
In an Italian randomized controlled trial (n = 45,307), the experimental arm (liquid-based cytology plus HPV testing) was compared with the conventional cytology arm.21 For both arms of the age 20–34 group, women were referred to colposcopy if cytology showed ASCUS+, while in the experimental arm repeat testing at 1 year was suggested for HPV-positive cases if cytology were normal. The experimental arm had significantly higher sensitivity but lower PPV than conventional arm. Additional 12 cases of CIN 2 were found in the experimental arm.21 In the 35–60 years age group, mostly all are with LSIL and 72% of women with ASCUS were directly referred to colposcopy for both arms. For the experimental arm, women with HPV-positive were referred to immediate colposcopy. HPV alone was more sensitive than conventional cytology. The prevalence of HPV-positive but normal cytology was 5.4%.22 In the Dutch Population Based Screening Study Amsterdam (POBASCAM) trial (n = 44,102), the presence of high-risk HPV was age-dependent, decreasing from 12% at age 29–33 years to 24% at 59–61 years.23 A higher HPV prevalence (9.9%) was found among women with normal cytology in our study than these 2 trials. A population-based study from China of 685 women (age 15–59) noted an overall HPV prevalence of 16.8% and 13.6% among those without cervical abnormalities.24
In a population-based study conducted by The National Health and Nutrition Examination Survey of the US (age 15–59, n = 1,921), HPV prevalence was 26.8% and the leading 3 types were HPV-62, -84 and -53. This study found that HPV infection was age-dependent. There was a significant trend of increasing prevalence with each year of age from 14 to 24 (44.8% at age 20–24 years), followed by a gradual decrease through 59 years.25 A population-based study from Thailand (n = 1,741) found that prevalence of high-risk HPV types peaked at age 25–34 and declined with increasing age, while the age-specific prevalence curve of low-risk HPV types was U-shape.9 There are substantial differences observed in age-specific curves of HPV prevalence between populations.26
A systemic review of 19 epidemiologic studies concluded that no definite association of oral contraceptive use and increased risk of HPV infection could be found due to heterogeneity and possible bias and confounding factors.27 An IARC HPV prevalence survey (n = 15,145) revealed that long-term oral contraceptive use, high parity, early age at first full-term pregnancy were not associated with HPV prevalence.28 Smith et al. noted that current or past users of hormone were not different in HPV prevalence when compared with the never users, but past users had a significantly higher risk using combination regimens with increasing duration.29 A significant positive correlation of HPV prevalence with postmenopausal status (or older age), and current users of oral contraceptives was noted in multivariate analyses of our study. Past users of oral contraceptives and never users of Pap were associated with higher risk of abnormal Pap, while age 40–49 strata had lower risk. Increased HPV prevalence was noted in oral contraceptive current users but increased risk of Pap ASCUS occurred in past users could be explained by time lag of acquisition of HPV infection to occurrence of abnormal cytology.
Identifying a subset of women who benefit the most with combined screening program is important for cost-effectiveness. In a study from Switzerland (age 13–96 years), the best balance between sensitivity and specificity for HSIL using combined screening was found in older age group (age ≥ 50 years), in which population the prevalence of abnormal Pap smear was 3.2% (cytology HSIL 0.3%).30 In our study, the prevalence of abnormal cytology was 2% and cytology HSIL was 0.45%. Unfortunately, no specific subgroup was found to benefit most from the combined testing. In this study, a priority to invite those who not had Pap smears for at least 3 years might attain a better yield of abnormal cytology but could cause a potential bias towards HPV distribution to high-risk types.
Although sensitivity of combined Pap and HPV testing is excellent, the lower specificity of HPV testing poses high potential for anxiety, and increased costs of surveillance in the individual woman. Strategies to enforce the knowledge of HPV transmission, prevention, detection as well as risk and management of cervical neoplasia are necessary for all societies.31 Our cross-sectional results showed that the improvement of sensitivity with additional HPV testing was only 15.3%. The whole project is still ongoing to evaluate safety and cost-effectiveness of screening with Pap plus HPV testing at longer intervals, which can be answered after long-term follow-up.
The authors thank The Bureau of Health Promotion, Department of Health, Taiwan, providing the national registry data in Pap smears and histological results of the participants.