Carcinogenic risk of human papillomavirus (HPV) genotypes and potential effects of HPV vaccines in Korea

This study investigated the distribution of HPV types in Korean women and evaluated the carcinogenic risk of individual HPV types and the potential effects of HPV vaccines. A total of 4,081 HPV-positive samples between 2014 and 2017 were included. The most prevalent genotypes were HPV 16, 58, 68, and 56. Among them, HPV 16 was significantly higher in high-grade squamous intraepithelial neoplasia or worse (HSIL+ ) group. In cytologically evaluating the risk for HSIL+ by individual HPV types, HPV 16 was associated with the highest risk of HSIL+ (OR = 10.82; 95% CI: 7.93–14.77), followed by HPV 33, 31, 52, 18, 58, 51, and 35, in descending order (OR = 3.50 [type 33] to 2.62 [type 35]). Among those types, HPV 16, 18, 31, 33, and 58 were also significantly associated with HSIL+ on histologic evaluation. The analysis of the HPV subgroups covered by the different vaccines revealed that the HPV types covered by the 9-valent vaccine had a high association with HSIL+ (OR = 4.09; 95% CI: 3.02–5.54). Our findings highlight the different carcinogenic risks posed by the high risk HPV genotypes and the positive potential effects of the 9-valent HPV vaccine in reducing HPV-associated cervical cancer in Korea.

because of the low prevalence of HPV infection in the cohort population. Therefore, it could be beneficial to estimate the association between individual HPV types and cervical cancer in an HPV-positive group. In this study, we evaluated the distribution of HPV types in Korean women and estimated the carcinogenic potential of individual HPV types by considering the association between specific HPV types and cervical cancer or precancerous lesions. In addition, we evaluated the potential effects of HPV vaccines in Korea.

Study population and sample evaluation. Among women who underwent HPV genotyping at Samsung
Medical Center between October 2014 and March 2017, we retrieved the anonymized records of patients with co-tested cervical cytology and HPV genotyping results. For the cytology specimens, Papanicolaou-stained liquid-based tests were performed with using a ThinPrep 5000 Processor (Hologic Inc., Malborough, MA) according to the manufacturer's protocol. Pathologists diagnosed the cytology slides according to the 2001 Bethesda system for cervicovaginal cytology: negative for intraepithelial lesions or malignancy (NILM); atypical squamous cells of undetermined significance (ASCUS); low-grade squamous intraepithelial neoplasia (LSIL); atypical squamous cells-cannot exclude high-grade squamous intraepithelial neoplasia (ASC-H); high-grade squamous intraepithelial neoplasia (HSIL); squamous cell carcinoma (SCC); and adenocarcinoma (ADC) 15 . For our analysis, we used three subgroups based on the cytologic diagnoses: NILM, LSIL (including ASCUS and LSIL), and HSIL+ (including ASC-H, HSIL, SCC, and ADC).
Of the 28,834 women with cervical cytology and HPV results, 91.5% received HPV genotyping as part of a routine health check-up, and 8.5% were referred from clinics due to abnormal cervical cytology results or other gynecologic problems. Among the 24,753 (85.8%) HPV-negative women, 0.9% had abnormal cytology results, including ASCUS (0.7%), LSIL (0.2%), ASC-H (<0.1%), and HSIL (<0.1%). Among the 4,081 HPV-positive patients, 3,242 patients had a single HPV result with a matched cytology result, and 839 patients had one or more follow-up results. We analyzed 4,081 (14.2%) HPV-positive results, including both single and initial results (Fig. 1). Among the HPV-positive patients, 596 patients (14.6%) underwent a follow-up biopsy within a year of the initial cervical pap test. The follow-up colposcopic biopsy specimens were stained with hematoxylin and eosin and diagnosed as NILM, LSIL, HSIL, SCC, or ADC. For our statistical analyses, the histologic diagnoses were also classified as NILM, LSIL, or HSIL+ (including HSIL, SCC, and ADC).

Carcinogenic risk of individual HR HPV types.
To assess the association of individual HPV types with HSIL+, we estimated the odds ratio (OR) of having HSIL+ (compared to NILM/LSIL) according to the individual HPV types. After adjusting for age and all other HPV types, HPV 16 was associated with the highest risk of HSIL+ (OR = 11.75; 95% CI: 8.55- 16.15). In addition, HPV 33, 31, 52, 18, 58, 51, and 35, in descending order, were significantly associated with an increased risk of HSIL+ (OR ranged from 3.50 [HPV 33] to 2.62 [HPV 35]). Those data are summarized in Table 2 and visualized in Fig. 2.
Among HPV-positive patients, 596 underwent a follow-up biopsy according to the "Practice guidelines for the early detection of cervical cancer in Korea" 16 : 0.7% of NILM patients, 3.1% of ASCUS patients, 55.3% of LSIL patients, 4.8% of ASC-H patients, 89.5% of HSIL patients, and 100% of SCC and ADC patients. Among patients with HSIL+ cytology, 84.3% were histologically confirmed as HSIL+. The correlations between histology and cytology diagnoses is shown in Table 3. We evaluated the association between individual HPV types and a histology-based HSIL+ result. After adjusting for age and HPV types, HPV type 16 was the most strongly associated with HSIL+ (OR = 9.22; 95% CI: 5.46-15.56), followed by HPV 31, 59, 58, 18, and 33, in descending order (OR ranged from 6.99 [HPV 31] to 3.28 [HPV 33]). Those data are shown in Table 4 and Supplementary Fig. S1.
Potential effects of HPV vaccines. We further analyzed the HR HPV types according to vaccine coverage. HPV 16 and 18, which are covered by all HPV vaccines, were present in 12.0% of our HPV-positive patients (Table 1) www.nature.com/scientificreports www.nature.com/scientificreports/ Age-specific HPV prevalence and its association with cytologic diagnoses. Among all of the women in our study cohort, the overall HPV prevalence was highest (28.4%) in the ≤34 years age group, and   Fig. S2). We found no discrete trends in type-specific HPV prevalence according to age. In the association between age and cytologic diagnosis, the median age of LSIL patients was 49 years, and the prevalence of LSIL showed a single peak in the 45-54 years age group. Among HSIL+ patients, the median age was 45 years, with a peak at 35-44 years followed by a decrease with age. The ages of the NILM, LSIL, and HSIL+ cytologic groups differed significantly (Bonferroni's corrected p-value < 0.0167). Our logistic regression analyses showed that older patients (>45 years) had an inverse risk for HSIL+ (Table 2).
Infection by multiple HR HPV genotypes. HR HPV types were co-infected in 6.9% of HPV-positive patients, with two to six concomitant HPV types (Supplementary Table S3). The percentage of multiple infections increased significantly from 4.8% in the NILM cytologic subgroup to 11.5% in the LSIL subgroup and 15.7% in the HSIL+ subgroup (p < 0.05). HPV 58 (24.7%), 16 (21.9%), and 51 (21.6%) were frequently identified as co-infections, and HPV 16-58 and HPV 33-35 were the two most common combinations of HPV types.

Discussion
Our aim in this study was to investigate the distribution of HPV types in Korean women and evaluate the association between individual HPV types and the risk of precancerous lesions and cervical cancer. We also evaluated the potential effects of current HPV vaccines.
In HPV-positive patients, the most frequent HPV type was HPV 16, followed by HPV 58 and 68. In patients with HSIL+, HPV 16, 58, and 52 were frequently identified, with HPV 16 as the most prevalent type. Although previous studies in Korean populations have reported inconsistent results regarding the HPV type-specific distribution [17][18][19] , those studies did agree that, compared with the worldwide prevalence 20,21 , HPV 58 is one of the most prevalent types in Korea, with HPV 16 and 18 being relatively less prevalent. This finding implies that the 9-valent vaccine, which covers five HR HPV types in addition HPV 16 and 18, would be beneficial to use in Korea.
In this study, the proportion of HR HPV types increased significantly from the NILM to the LSIL and HSIL+ cytology groups. However, not all HR HPV types were associated with HSIL+ . The analysis of cervical cytology revealed that HPV 16,18,31,33,35,51,52, and 58 were significantly associated with HSIL+. Of those, HPV 16,18,31,33, and 58 were also associated with HSIL+ in histologically confirmed cases. Although HPV 59 showed a significant association with HSIL+ in the histologic evaluations, only six HPV 59 positive samples were included in the histologic analysis, so that the result should be interpreted with caution. Previous studies traced the disease progression of HPV-positive NILM or LSIL cytology patients and found that HPV 16,18,31,33,35,52,and 58 were associated with the risk for HSIL+ [22][23][24][25] , which is consistent with our findings. Because HR HPV genotypes confer significantly different prognostic information, it might be helpful to develop guidelines for specific HPV types to effectively manage patients. HPV 16,18,31,33, and 58, which all had high carcinogenic potential in both the cytology and histology results, are covered by the 9-valent vaccine. When the risk for HSIL+ was evaluated in the HR HPV subgroups according to vaccine coverage, the HPV types (HPV 31/33/45/52/58) covered by the 9-valent vaccine were associated with twice the risk for HSIL+ than the HR HPV types (HPV 35/39/51/56/59/66/68) not covered by the vaccine. Thus, the 9-valent vaccine could be effective at preventing cervical cancer in the Korean population. In Korea, the 2-valent and 4-valent vaccines are currently included in the national immunization program as a 2-dose schedule for 12-year old girls. The 9-valent vaccine was approved by the Ministry of Food and Drug Safety in January 2016 and can be given as a non-national immunization program vaccine. Huh et al. reported that the 9-valent vaccine showed efficacy against HPV 31, 33, 45, 52, and 58 and similar efficacy toward HPV 6, 11, 16, and 18 as the 4-valent vaccine 26 . In addition, 2-doses of the 9-valent vaccine showed non-inferior immunogenicity to the standard 3-doses in 10-14-year-old females 27 . Although factors such as cost should certainly be considered  Table 5. Multivariate logistic regression analyses of HR HPV vaccine subgroups according to HSIL+ versus NILM/LSIL in the cytology diagnoses, adjusted for age. CI = confidence interval, HR = high risk, HSIL+ = high-grade squamous intraepithelial neoplasia or worse, LSIL = low-grade squamous intraepithelial neoplasia, NILM = negative for intraepithelial lesions or malignancy, OR = odds ratio.  Table 6. Multivariate logistic regression analyses of HR HPV vaccine subgroups according to HSIL+ versus NILM/LSIL in the histologic results, adjusted for age. CI = confidence interval, HR = high risk, HSIL+ = highgrade squamous intraepithelial neoplasia or worse, LSIL = low-grade squamous intraepithelial neoplasia, NILM = negative for intraepithelial lesions or malignancy, OR = odds ratio. www.nature.com/scientificreports www.nature.com/scientificreports/ when determining a national immunization program, implementation of the 9-valent vaccine might produce a substantial decrease in cervical cancer in Korea.
Among HPV-positive patients, the median age was youngest in the HSIL+ group, gradually increasing in the LSIL and NILM cytology groups. The prevalence of HSIL+ peaked in the 35-44 years group and decreased with age. These findings are in line with other recent studies in Korea 18,28,29 . Because women start being infected with HR HPV when they become sexually active, and HR HPV can cause HSIL in approximately 7 years, HSIL most commonly happens to women in their early 30s 30 . In addition, the age-specific prevalence of CIN is attributable to the coverage of cervical cancer screening 31,32 . Because cervical cancer screening is registered as a national cancer screening program in Korea, the screened population could receive timely therapeutic intervention, allowing the prevalence of HSIL+ to decrease in the older age groups.
This study has some limitations. First, it is a retrospective cross-sectional study. In addition, the HPV genotyping was performed with two different methods-the HPV 9G DNA Chip Test and the GeneFinder in the first and second halves of the study period, respectively. Although both methods have been validated [33][34][35] , no direct comparison of these two tests has been performed, so there could be discordance between them. Instead, we compared the results of the study periods covered by the different HPV genotyping methods. Of the total 28,834 cytology/HPV co-tested women, 13,697 patients (47.5%) were tested using the DNA chip and 15,137 patients (52.5%) were tested using the GeneFinder. The prevalence of HPV was 17.5% in the first period and slightly but not significantly increased to 18.3% in the second period. The prevalence of HSIL+ was 6.6% in the first period and 5.6% in the second period, while HPV 16 was identified in 1.8% and 1.5% in the first and second periods, respectively. Although the prevalence of HSIL+ and individual HPV types did differ in the two periods, the genotyping tests were performed in different populations, making it difficult to identify any significant differences between the two methods. Nonetheless, this large-scale study identified the distribution of individual HPV types in Korean women and used risk stratification to assess the carcinogenic potential of specific HPV types. In addition, we found a high correlation between cytology and histology diagnoses, indicating that our results are reliable. This study thus provides guidance for improving the HPV-related national immunization program and management guidelines for individual HPV types. The HPV 9G DNA chip detects HPV genotypes via microarray, and the test was performed according to the manufacturer's protocol 34 . Briefly, HPV genomic DNA was extracted from cervicovaginal cytology samples and amplified via polymerase chain reaction (PCR). The PCR product was mixed with hybridization solution and loaded on the HPV 9G DNA chip. The DNA chip was then hybridized and washed at room temperature (25 °C). The final product was automatically scanned and analyzed.

Methods
The GeneFinder is a microsphere bead-based test that is highly specific and sensitive for detecting HPV genotypes 36 . After HPV genomic DNA was extracted, the specimen was mixed with HPV type-specific polystyrene microspheres dyed with specific intensities of fluorophores. After hybridization and PCR amplification, the microspheres were streamed using Luminex technology (Luminex, Austin, TX). HPV genotypes were detected by a Luminex analyzer based on a flow cytometry method.
The  11 , are regarded as HR HPV, whereas HPV 6, 11, 34, 40, and 42 are regarded as LR HPV. If patients had an HPV type identified by one method but not the other, the HPV type was categorized as "other. " Statistical analysis. Patient ages were stratified into five subgroups: ≤34 years, 35-44 years, 45-54 years, 55-64 years, and ≥65 years. A χ 2 test or Fisher's exact test was used to evaluate the statistical significance of differences in categorical variables among groups. Post-hoc test was performed using Tukey's Honestly Significant Difference test, and p-values from multiple comparisons were corrected by the Bonferroni method. Age was additionally analyzed as a continuous variable using the Kruskal-Wallis test followed by Dunn's post-hoc test. To assess the association of HPV subtypes with HSIL+, we performed logistic regressions and calculated the OR and 95% CI. For HPV types, the OR was defined as the ratio between the odds of being in the HSIL+ group in the presence of an individual HPV type and the odds of being in the HSIL+ group in the absence of an individual HPV type. The results were visualized by forest plot using the SGPLOT procedure. For all analyses, p < 0.05 was considered statistically significant, and all analyses were two-sided. Data were analyzed using SAS version 9.4 (SAS Institute Inc., Cary, NC).