Do Infection Patterns of Human Papillomavirus Affect the Cytologic Detection of High-Grade Cervical Lesions on Papanicolaou Tests?

The 171 621 Pap tests recorded in our laboratory information system between March 2013 and June 2014 were analyzed along with results from hrHPV testing (Cobas 4800 system, Roche Molecular Diagnostics, Pleasanton, California) and follow-up biopsies, when available. The cervical lesions detected on Pap tests were categorized into 2 general groups: low-grade cervical lesions, encompassing atypical squamous cells of undetermined significance (ASC-US) and low-grade squamous intraepithelial lesions (LSILs), and high-grade cervical lesions (HGCLs), including ASC, cannot rule out HSILs; atypical glandular cells; HSILs, and carcinomas. The results of follow-up biopsies, performed within 14 months of cytology-HPV cotesting, were recorded. Cervical intraepithelial neoplasia grade 2 or worse on biopsy were considered HSILs. All cervical intraepithelial neoplasia grade 2 lesions were confirmed by immunohistochemical staining for p16 and Ki-67. Real-time histologic-cytologic correlation was performed at the time of biopsy sign-out, as per our usual clinical practice.

All Pap tests were performed with 1 of the 2 liquid-based methods, in accordance with the preference of the referring practitioner: ThinPrep (Hologic, Madison, Wisconsin) and SurePath (Becton Dickinson, Franklin Lakes, New Jersey). Of the 257 cases with biopsy-confirmed HSIL, the preceding Pap tests were performed on the ThinPrep platform in 133 cases (51.8%) and on the SurePath platform in 124 cases (48.2%). The indication for cervical biopsy was not solely based on the preceding Pap test results. Some patients were referred to colposcopy based on their HPV genotyping results or other clinical situations. The cytologic data acquired by our laboratory were benchmarked with the College of American Pathologists Laboratory Accreditation Program database. The Pap tests and biopsies were interpreted by board-certified cytopathologists or gynecologic pathologists in an academic medical center.

In this general screening population, the HSIL reporting rate on cytology was 0.24% (313 of 130 648 cases), and there was consistent cytohistologic correlation in 70% (167 of 239) of the cases. Most cases with uncorrelated data were due to tissue-sampling variances (98%; 234 of 239 cases). A total of 257 women had HSIL on follow-up biopsies, with an average age of 37.9 years (range, 22–62 years). Of the women with biopsy-confirmed HSIL, the preceding Pap test diagnoses included 20 negative results (8%) for intraepithelial lesions or malignancy; 71 cases (28%) of ASC-US; 31 (12%) ASCs, cannot rule out HSILs; 6 (2%) atypical glandular cells; 95 (37%) LSIL; and 34 (13%) HSILs (Figure 1). No carcinomas or other malignancies were identified in the group. Seventy-one cases (28%) had high-grade cytologic lesion (HGCL) index Pap test results, including ASC, cannot rule out HSILs; atypical glandular cells; and HSILs.

Cobas HPV cotesting was performed in 236 of the 257 biopsy-confirmed HSIL cases (92%). Of those, hrHPV was detected in 92% (n = 215) of the cases, and results were negative in 9% (n = 21) of the women. The biopsies that tested positive for hrHPV showed 4 unique infection patterns: (1) single-genotype HPV-16 (62 of 215; 29%) or (2) HPV 18 (6 of 215; 3%) infections, (3) non-16/18 hrHPV infections (107 of 215; 50%), and (4) mixed infection patterns with combined HPV-16, HPV-18, and/or non-16/18 hrHPV (40 of 215; 19%) genotypes (Figure 2). The Pap test results with the corresponding hrHPV genotypes detected on cotesting are provided in the Table.

Cytologically, low-grade cervical lesions were often associated with 2 HPV infection patterns: (1) mixed genotype (31 of 40; 77%) and non-16/18 hrHPV infection (75 of 103; 73%). However, the lower-grade cytomorphology was significantly less common in single-HPV genotype infections of HPV-16 (33 of 62; 53%) or HPV-18 (2 of 6, 33%) (P = .01). In contrast, HGCLs on Pap test were more often associated with single-genotype infection of HPV-18 (3 of 6; 50%) or HPV-16 (20 of 62; 32%), compared with non-16/18 hrHPV genotypes (25 of 103; 24%) or mixed infection pattern (8 of 40; 20%) (P = .01) (Figure 3). Among 21 women with HSIL on biopsy who tested negative for hrHPV, 19 (90%) had abnormal Pap tests, including 12 cases (57%) of HGCL and 7 cases (33%) of low-grade cervical lesion. Only 2 women (10%) with biopsy-confirmed HSIL had double-negative results on the preceding HPV-cytology cotesting (Table).

Persistent infection with hrHPV causes most cervical cancers and precancerous lesions.³  In recent years, the Cobas hrHPV test with HPV-16/18 genotyping has been increasingly used in clinical practice as an additional test, along with the Pap test, for cervical cancer screening and triaging patients with ASC-US. The effect of HPV infection patterns on cytomorphology and detection of HGCL by the Pap test is unknown. In this study, we analyzed the association of hrHPV infection patterns with cytologic diagnoses in women with biopsy-proven HSIL.

In this cohort of biopsy-confirmed HSIL, the patients were stratified into 5 groups according to the HPV test results: (1) HPV-16, (2) HPV-18, (3) non-16/18 hrHPV, (4) mixed hrHPV, and (5) negative for hrHPV. We found that the lower-grade cytomorphology (ASC-US and LSIL) was significantly associated with HPV infections of non-16/18 hrHPV and mixed hrHPV infection patterns. In contrast, single-genotype infections of HPV-16 or HPV-18 were more frequently associated with high-grade cytomorphology on Pap tests (Figure 4, A through D). Our findings suggest that non-16/18 or mixed hrHPV infections tend to produce predominantly lower-grade cytomorphology than single-genotypic infections of HPV-16 or HPV-18 do, which may result in underdiagnosis of Pap test results and delay of clinical follow-up or necessary intervention.

According to the ATHENA trial (A Trial With Dronedarone to Prevent Hospitalization or Death in Patients with Fibrillation), 12.5% of women older than 21 years had an hrHPV infection other than the HPV-16/18 genotypes.¹³  In our cohort of women with biopsy-confirmed HSIL, 45% were infected with non-16/18 hrHPV and 17% with mixed-hrHPV genotypes. Given that these women were more likely to have lower-grade lesions on their Pap tests, cotesting with HPV testing and genotyping likely helped reduce underdiagnosis and delay of treatment.

In our previous study on women diagnosed with LSIL on cytology, those with infection of a non-16/18 hrHPV genotype were more likely to have persistent disease or to progress to HGCL.¹⁴  This suggests that these women with LSIL cytology and non-16/18 HPV infection tended to harbor high-grade lesions and were more likely to be underdiagnosed because of deceptive lower-grade cytomorphology.

According to the American Society for Colposcopy and Cervical Pathology 2012 guidelines,¹⁰  a second colposcopic examination can be delayed by 1 to 3 years after a diagnosis of LSIL or ASC-US from Pap tests with negative HPV-16/18 results. In view of our data, a reduced threshold for colposcopy or other intervention may be considered for women infected with non-16/18 hrHPV or mixed-hrHPV genotypes. Women infected with these genotypes may harbor HGCL despite deceptive lower-grade cytomorphology.

Because HPV vaccination is gradually taking effect in the North American population, the predominant HPV infection patterns are expected to change. The HPV-16 and HPV-18 genotypes may be less likely to occur over time as the predominant single-genotype infectious pattern, whereas non-16/18 hrHPV and possibly other uncommon genotypes become more prominent.¹  In addition, the significant increase in human migration around the world in recent decades has caused a dramatic change in genotypic composition of HPV in North American populations.^2,15  The shifting of HPV genotypic prevalence will have significant effect on current screening practices because infections of non-16/18 hrHPV and mixed-HPV genotypes often produce deceptive lower-grade cytomorphology. Further studies are needed to establish sensitive and effective screening approaches to adapt to the growing challenge of genotypic shifting of HPV in North American populations.

The mechanisms of deceptively lower-grade cytomorphology observed in women infected with non-16/18 hrHPV and mixed-hrHPV genotypes are unknown. In previous studies, multigenotype HPV infections were documented in 20% to 30% of women with HGCLs.^16–18  In our study cohort, at least 17% of the patients with biopsy-proven HSIL demonstrated coinfection with 2 or more HPV genotypes. The actual percentage of multigenotype HPV infections in this cohort could be much higher because 45% of the women were infected with non-16/18 hrHPV, which could represent single or multiple genotypes in a battery of the remaining 12 hrHPV types. A host of different immunomodulatory mechanisms has been suggested for the lower-grade lesions in multigenotype HPV infections.¹⁹  Recent studies suggest viral antagonistic interaction among a number of HPV genotypes.²⁰  This result was based on observations that patients who are seropositive for multiple HPV types have a lower risk of developing cervical cancer.²¹  The mechanism of intergenotypic competition is complex and depends on all stages of infection, from binding sites/receptors to genomic expression within host cells and interaction with the immune system of the host. Further studies are needed to unravel the underlying mechanism.

The Cobas platform detects 14 hrHPV genotypes, including HPV-16 and HPV-18, in addition to HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, HPV-59, HPV-66, and HPV-68, commonly referred to as non-16/18 hrHPV types.²²  The sensitivity and specificity of the Cobas HPV for detecting hrHPVs were 91.7% and 97.0%, respectively.²³  A previous study⁹  showed 21 of 236 biopsy-proven HSIL cases (9%) with no detectable HPV on the Cobas platform. This could be due to infection with rare HPV genotypes, which are not covered by the testing panel of the Cobas platform. Other potential contributing factors include sampling variation, low viral loads, technical errors, truly non-HPV–driven lesions, as well as false-negative results because of interference from reagents.⁹  This finding raises a question about the effectiveness of the Cobas HPV testing as a primary-screening method for high-grade cervicovaginal lesions. Our data clearly demonstrated that the false-negative rate was greatly reduced by combining cytology and HPV testing, thus providing further support for cytology-HPV cotesting as the best current strategy for detection of high-grade cervicovaginal lesions.

Despite the observed trends in this study, our results should be interpreted cautiously because most of the biopsies were performed within 1 year of the Pap tests, which is shorter than the 3-year window recommended by the US Preventive Services Task Force. This limits the ability to observe the effects of delayed diagnosis because of the false-negatives on screening and the corresponding follow-up results. In this small cohort, the HSIL correlation rate between cytology and biopsy was misleadingly low because the study group did not include women with Pap test results alone or those with HSIL on a Pap test but who had the biopsy performed elsewhere. On the other hand, some women were referred to colposcopy because of positive HPV test results, regardless of their Pap test results, which may also be a contributing factor. In addition, this is a retrospective, cross-sectional study, and follow-up data regarding the long-term effect of cytologic changes in patients with single versus multiple genotype infections of HPV are not available for assessment. The focus of this study was to investigate the association of cytomorphologic changes with the pattern of HPV genotype infection, regardless of immunomodulatory factors, such as tobacco use, HIV status, sexual history, and prior history of cervical intraepithelial lesions. Our study had a relatively small cohort of biopsy-proven HSIL cases, a fact which limited our interpretation of the less-frequent HPV genotypes. Given the limited genotyping panel of the Cobas 4800, we were unable to further analyze the interaction among various HPV genotypes, in the group with multigenotype HPV infection.

In the cohort with biopsy-confirmed HSILs, the women infected with non-16/18 hrHPV or mixed-HPV genotypes were significantly associated with deceptive lower-grade cytomorphology, compared with those infected with single HPV-16 or HPV-18 genotypes, contributing to false-negative results or underdiagnosis on Pap tests. Clinically, HPV testing provides valuable information on HPV infection patterns, which may help in risk stratification. Vigilant surveillance and dynamic management protocols may be warranted for women infected with non-16/18 hrHPV or mixed-HPV genotypes. Information provided by HPV testing and genotyping likely is far greater than solely HPV-16/18. The overall infection pattern can be increasingly valuable in clinical risk management in light of the current challenges of HPV genotype shifting because of vaccination and human migration. Further studies of the mechanisms and the interactions among various HPV genotypes are needed to establish appropriate guidelines regarding follow-up and management of these patients.

We thank Helen Chifotides, PhD, for editing of the manuscript.