Human papillomavirus genotype distribution in cervical samples collected in routine clinical practice at the Nantes University Hospital, France

Based on a large number of subjects, our study reports the HPV overall prevalence and HPV genotype distribution in cytological and histological specimens collected in routine clinical practice in a hospital-based gynaecologic department.

Among all 1,255 women, the observed proportion of those with abnormal smears (22%) is much higher than that previously reported by Bergeron et al. in France (3–4%) [22, 23]. This discrepancy may result from differences in the recruitment of patients. In each Bergeron study, approximately 200,000 smears were collected from public or private structures whereas in the present study, the recruitment was hospital-based with as a result, inclusion of more severe cases. Since it is hospital-based, our study population is thus not representative of the general population. However, our results depict university hospital routine clinical practice. Noteworthy, the diagnosis distribution among abnormal cytological samples is rather similar to Bergeron’s data with approximately 50% of ASCUS, 36% of LSIL, and 13% of HSIL.

As expected, the prevalence of HPV positive samples increases with increasing diagnosis severity. In cytological samples, the overall HPV prevalence varies indeed from 13% in normal samples to 70–75% in LSIL and HSIL. In ASCUS cases, the prevalence is in-between at approximately 30%. In histological samples, the overall HPV prevalence increases with the severity of the lesion reaching a maximum of 89% in CIN3 cases. The HPV prevalence observed in normal histological specimens (33%) is higher than that in normal cytological samples (13%). This may be explained by the fact that biopsy was performed in case of abnormal smear, clinical sign, or presence of oncogenic HPV type.

The overall HPV prevalence in women with normal cytological diagnosis observed in our study is twice higher than that reported in the US by Castle et al. in about 800,000 smears collected from women above 30 years of age [24]. The authors observed indeed that 4% of women with normal smears were carcinogenic HPV positives whereas the prevalence of high-risk HPV genotypes in this subgroup reached 8.2% in our cohort. This higher prevalence could be partly explained by the hospital-based recruitment in our study and by the different age distribution between cohorts. In France, a previous study based on 2,145 smears without lesion [25] collected from women above 30 years who underwent their biennial or triennial routine screening in the university hospital of Reims, reported that high-risk HPV genotypes were found in 10.8% of cases which is close to our findings. Another French study on 3,640 women (above 20 years of age) with normal cytology enrolled either from the university hospital of Amiens or from private gynaecologists, reported a high-risk HPV prevalence of 13.5% [26]. In a meta-analysis based on 157,879 women with normal cervical cytology [27], the overall HPV prevalence was estimated to be 10.4% (8% in Europe). The high-risk HPV prevalence observed in our subgroup of women with normal cytological diagnosis appears thus similar to that reported earlier. In the present study, presence of high-risk genotypes in women with normal smears is highly associated with age with a prevalence decreasing from 24% to 6.5% before and after 25 years of age, respectively. Based on a rather large number of women with normal cytological diagnosis (n = 980), our study stresses the fact that a substantial proportion of these women is infected by a high-risk HPV, this proportion exceeding 20% in women below 25 years of age. Such a high prevalence of HPV in younger women with normal cytology was already reported by De Sanjosé et al. [27] in the above mentioned meta-analysis.

On the other hand, the overall HPV prevalence in women with abnormal cytological or histological diagnosis is somehow lower in our study than what was previously reported in the literature. In CIN2/3 specimens for example, the observed prevalence of 77% is much lower than the 98% reported by Pretet et al. [17]. The same is true for the HPV prevalence in LSIL (69 versus 98%) [16]. A possible explanation is that in the present study, cytological or histological diagnosis and genotyping were possibly made on different samples and at different time points. These two parameters were associated with each other by linkage of the hospital genotype and diagnosis databases. Even if HPV infection is known to last for about 13–19 months [9], one can not exclude that HPV clearance occurred between diagnosis and genotyping. However, this hypothesis is unlikely since the mean delay between cytological diagnosis and genotyping (15 days) or between histological diagnosis and genotyping (11 days) is rather short. The HPV negative status of some cervical cell suspension specimens with severe dysplasia could instead result from a low HPV viral load or from the fact that some genotypes are not matched by the MY09/MY11 primers. A reasonable explanation for these discrepancies in HPV prevalence is indeed the use of HPV detection methods of different sensitivity [18]. Such differences in sensitivity and characteristics of PCR systems for the detection of HPV within clinical samples have been previously described and there is good evidence that some genotypes are better matched to the consensus primer pools than others [28‐31]. To cope with these differences in sensitivity, we also provided genotype-specific prevalence data among HPV positive women.

HPV 16 was by far the most commonly encountered genotype in cervical smears and histological specimens even when the diagnosis was normal. Figures 3 and 4 clearly show that the prevalence of HPV 16 increases with increasing diagnosis severity hereby confirming that HPV 16 tends to persist longer and is more aggressive than other genotypes [32]. Since the overall HPV prevalence in women with abnormal smears in our study is lower than previously reported, we considered HPV positive samples to calculate the prevalence of each specific genotype. Doing so, the prevalence of HPV 16 in LSIL, CIN2 and CIN3 becomes of the same magnitude as what was previously observed [16, 17]. HPV 16 was indeed found in 29% of our LSIL cases which is very close to results reported in the meta-analysis by Clifford et al. (27%) [33]. To illustrate the high pathogenic role of HPV 16, we calculated the ratio of HPV prevalence in CIN3 versus normal histology and versus CIN1. These ratios indicate that HPV 16 is almost four-fold more prevalent in CIN3 than in CIN1 and about nine-fold more prevalent in CIN3 than in normal histological samples hereby confirming that this genotype is associated with lesions presenting a high probability of progression to severe forms. Although the number of cases is limited, HPV 53 appears as the second most common genotype in normal cytological samples and in LSIL (11 and 13%, respectively) and as the third most common in ASCUS (12%). However, the prevalence decreases in HSIL to less than 4%, suggesting that samples infected with HPV 53 are at low risk of progression towards severe forms although some few cases of cervical cancer following HPV 53 monoinfection have been described [18]. The same phenomenon is observed for biopsies in which HPV 53 is frequently encountered in normal and CIN1 specimens (16%) and much less prevalent in CIN2 and CIN3 (below 4%). Alone or in association, HPV 6, 11, 16 and/or 18 are found in over 50% of HPV positive cytological and histological specimens hereby suggesting that a substantial proportion of these lesions could be avoided by a quadrivalent HPV vaccination.

Despite some methodological limitations, our study provides accurate French data on HPV genotype distribution in samples collected during routine hospital-based gynaecologic follow-up and brings further evidence of the potential impact of a quadrivalent vaccination.