Hybrid capture vs. PCR screening of cervical human papilloma virus infections. Cytological and histological associations in 1270 women

Large epidemiological studies performed over the last two decades have identified infection with human papilloma virus (HPV) as a necessary cause for cervical cancer[1, 2]. Furthermore, newer molecular techniques have greatly helped in identifying the association of high-risk HPV types with the development of precancerous lesions, as well as their role in cervical carcinogenesis[3, 4]. Since the vast majority of invasive cervical carcinoma cases are associated with HPV type 16 or 18 infections[5, 6], efforts have been made to develop preventive measures for infections with these high risk HPV types: two specific vaccines targeting infection with HPV types 16 and 18 have recently been introduced[7].

Specific HPV testing may not be appropriate as a primary screening due to claims that it lacks specificity[8] but it may assist in the management of women with cytological changes[9]. The association of findings from conventional cytological testing with those of newer molecular techniques is of great importance and helps to better understand the evolution of HPV infection in different epidemiological settings.

The purpose of the current study is to evaluate cytological findings from a large observational population sample in association with identification of HPV infection using newer molecular techniques, such as Hybrid Capture 2 and PCR. Furthermore, we correlate these findings with histological diagnosis.

This is a cross-sectional study performed at all specimens sent for diagnostic cytology at the Departments of Histology and Embryology, and the Department of Cytopathology of the National and Capodistrian University of Athens Medical School. Samples are addressed to these departments every day from many community and hospital-based Gynecology clinics covering the metropolitan area of Athens (roughly a population of 3.8 million people). Informed consent was obtained from all women included in this study that was performed between January 2000 and December 2003.

This study depicts the successful use of both HC2 and PCR in liquid based ThinPrep samples. Use of the same samples for both DNA tests was done to avoid bias. The HC2 kit assesses the presence of 18 low risk (i.e. 6,11,42,43 and 44) and high risk (i.e. 16, 18, 31, 33, 35, 39, 45, 51, 51, 56, 58, 59 and 68) HPV types [15, 16]. On the other hand, using the L1 consensus primers GP5⁺/GP6⁺ by PCR, we were able to screen for 22 HPV types. The concordance rate between the PCR and HC2 test results was high in our series. Both molecular tests exhibited high reproducibility measured by Cohen's kappa value of 0.691, similar to previous reported results [17, 18]. Although not all studies use the same PCR protocol, the concordance between HC2 and PCR testing across various studies is generally high, usually exceeding 80% [17‐20]. It appears that after a certain DNA concentration cutoff point, HC2 sensitivity increases, carrying a certain risk for false positive results[17]. Newer generation kits decrease rates of false positive testing[21]. Certain cutoff levels for relative light units compared to the ones generated by the positive control samples to increase the specificity of the assay without compromising sensitivity may be used[18]. One explanation for some of the discordant results between HC2 and PCR in our study is the fact that GP5+/GP6+ primers may actually miss some cases deleted in the L1 region of the virus. In addition, some of the samples exhibiting discordance may have infection with HPV types such as HPV 42, 44 or 68 not amplified by the used primers, but still detectable by the HC2 cocktails. Moreover, HC2 probe B cocktail can react with phylogenetically related HPV types not represented in the probe such as HPV 67, CP 6108, and CP8061. Finally, the prevalence of latent infection differs across studies, depending upon demographic parameters; this may also affect the sensitivity and specificity of the two tests.

Our results confirm the higher prevalence of HPV infection in women with abnormal cytology, in concordance with most studies so far published, that have observed that the increase in HPV prevalence is related to the increasing grade of squamous intraepithelial lesions [22‐26]. Rates of HPV detection in samples with normal cytology vary widely in the literature in women with normal cytology and range from 4.9% to 30.4% for the HC2 assay and from 3 to 34.3% for PCR testing [18, 19, 21, 25, 27]. This wide range is explained by the different nature of participating populations in such studies, and by technical evolutions in the diagnostic tests used[21, 28]. The higher analytical sensitivity of PCR explains the higher detection rates compared to HC2 in samples with low-grade cytological and no histological abnormalities and may be suggestive of the presence of latent HPV infection requiring a molecular test with higher analytical sensitivity.

Similar to other observations, both hybrid capture and PCR HPV DNA testing showed a strong correlation with the diagnosis of CIN lesions or squamous cell carcinoma [23, 29]. The extremely high negative predictive values of both molecular tests underlie their importance in screening for high grade lesions and are in concordance with previous publications[18]. In high grade cytological and histological lesions, HPV detection rates appeared to be a little higher with PCR than with HC2 testing; for example HC2 did not identify high risk viruses in seven instances of CIN2+ histological lesions, in which cases PCR was positive. However positive and negative likelihood ratios were similar for positive HC2 (any result) or PCR - HPV positive testing (any result) both for CIN2+ or for CIN3+ histology. Actually, HC2 had a better PLR while PCR had a better NLR in both histological categories. This means that with a positive HC2 there was a greater likelihood of disease (≥ CIN2) than with PCR, while with a negative HPV PCR there was a lesser likelihood of disease (≥ CIN2) than with HC2. Nevertheless, in essence both tests gave similar results. Likelihood ratios can be used to calculate the odds of post test probabilities if multiplied by the odds of the prevalence of the disease. For example, a patient in our study had a post test probability of disease (CIN2+) of only 5.1% (using the Bayes nomogram) if she had a negative HC2 test (according to our study she had a pre-test probability of 20.4% of having a positive test). Similarly a negative PCR for high risk types 16, 18, 31 and 33 has a post test probability of disease (CIN2+) of 1.93% (Table 4). The limitation of a population that is not systematically screened, but is consecutively enrolled for deriving this data is of course recognized. However, both molecular HPV tests did not outperform cytology that had comparable if not better results, especially if the HSIL+ cytological lesion cutoff point was used. In our study all HSIL lesions by cytology were either CIN 2 or CIN 3 by histology and all cytological diagnoses of invasive carcinomas were subsequently confirmed by histology. Possible explanations for this observation include the following: a) the experience of all doctors involved; b) the increased alertness of both the referring gynecologist and the cytopathologist examining the slides; c) the fact that the Department of Cytopathology is accredited and therefore undergoes continuous quality assessment/quality control evaluation.

Although none of the patients with normal cytology and an HPV infection had a biopsy proven lesion higher than CIN 1 in our study, these women should not be considered false-positive but as having a real risk for progression to abnormal cytological findings and cervical neoplasia [30‐33]. It is known that development of precancerous lesions may shortly follow infection with HPV, despite the belief that long-term infection is a prerequisite for such an event [34]. Thus, these women should be prospectively followed by their gynecologist and submitted to cytology and other testing, as appropriate[35, 36]. The use of HPV testing has been recommended for women with ASCUS[37] and it has been shown that approximately one third of women with HSIL are subsequently identified from an initial ASCUS diagnosis [25]. These women, after positive HPV DNA testing should be referred for colposcopy[38]; our study underlines this argument, since one woman with ASCUS, had a single infection with HPV type 16 and biopsy disclosed squamous cell carcinoma. Nowadays, in the revised Bethesda 2001 classification system this case would probably be diagnosed as ASC-H. A molecular technique, more sensitive and specific than HC2 may be more appropriate for such patients.

The current observations further confirm the association of high risk HPV types with cytological detection of HSIL or invasive carcinoma but also with histologically confirmed premalignant or malignant lesions [39‐41]. All cases of CIN 2, CIN 3 or carcinomas harbored single or multiple HPV high risk oncogenic type infections. In our study single infections with HPV types 16 or 18 in cases of invasive carcinoma were strong evidence for this association and support the strong rationale for preventive HPV vaccination in our population.

The rate of invasive carcinoma discovered in this study is indicative of a largely unscreened population. The reasons and possible solutions for this observation are an important public health issue and should be further investigated. Moreover the effect of the introduction of newer techniques in population screening is a matter of intense research. Despite the excellent results with cytology in this study, it is well known that screening for cytological changes may have limited sensitivity and findings are not always reproducible [8]. In a country where screening is largely opportunistic and based on self-referral results of cytological testing are expected to be much worse than the ones presented here. Moreover, molecular HPV testing should not be introduced without careful planning; results of such testing should be communicated and explained appropriately in the context of prevalence of the disease.

In conclusion, the present study was more of a technical nature and focused on strengths and weakness of current screening methods in correlation with histological findings. HPV rates were high in high grade lesions in accordance to published literature however HPV was also identified in a large fraction of samples with normal cytology or in samples with normal biopsies. The best approach for such samples will be guided after careful identification of the local epidemiology of the transmission and risk factors for HPV acquisition. Newer methods assessing the integration of the virus may be more appropriate in the workup of such cases. In our study in an referral center with experienced doctors, thin prep cytology and molecular HPV tests performed equally well as screening tests in all scenarios and sometimes were complementary to each other. As newer methods to detect HPV continue to evolve it is important to recognize the clinicopathological correlates of such testing. This will lead to more appropriate screening strategies in the post-vaccine era.