Judging the carcinogenicity of rare human papillomavirus types

Persistent infection with human papillomavirus (HPV) types is accepted as the necessary cause of invasive cervical cancer (ICC).1 However, for the development of HPV screening tests and vaccines, a large number of mucosotropic HPV types need to be judged separately. Although some types are powerful carcinogens and can be found in a sizeable proportion of ICC,2 the classification of rarer types is more challenging: not only is their prevalence in ICC very low, but as all HPV types share a common mode of sexual transmission, confounding from the concurrent presence of other carcinogenic HPV types is difficult to rule out.3 In 2009, an International Agency for Research on Cancer (IARC) Working Group (IARC, 100B) classified HPV16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59 as carcinogenic (Group 1) to humans, and HPV68 as probably carcinogenic (Group 2A), according to an algorithm that included comparison of HPV type-specific prevalence in ICC with that among women with normal cytology (as a proxy for the background population prevalence of infection).1, 3 These 13 Group 1/2A types cluster together in an evolutionary clade that includes Alphapapillomavirus species groups 5, 6, 7, and 9. Other types in this clade were classified as possibly carcinogens (Group 2B) based solely on phylogenetic relatedness.1, 3 With the aim to provide an epidemiological basis for judging the carcinogenicity of rarer HPV types, we expanded our previous meta-analysis published in this journal2 that reported the prevalence of the Group 1/2A HPV types across the full spectrum of cervical diagnoses from normal cytology to ICC, to include 29 additional HPV types. As described previously,2 eligible studies met the following criteria: (i) use of well-validated broad-spectrum consensus polymerase chain reaction (PCR) assays and (ii) reporting of overall and type-specific HPV prevalence by strata of cytopathological and/or histopathological cervical diagnoses. The ratio of the prevalence of individual HPV types in ICC to that in women with normal cytology (ICC:normal prevalence ratio) was calculated as a proxy of carcinogenicity, with HPV6 considered as an archetypal noncarcinogenic type (IARC Group 3) for comparison purposes.3 Figure 1 describes ICC:normal prevalence ratios of individual HPV types in a total of 40,706 ICC cases and 263,971 women with normal cytology, from all world regions. Of note, type-specific prevalence was estimated only among studies that tested and reported data for the given HPV type, so denominators vary by type. All Group 1/2A types (in red in Fig. 1) showed ICC:normal prevalence ratios greater than 1.0, but with a strong variation from 21.1 (95%CI 21.5–22.7) for HPV16 down to 1.2 (95%CI 1.1–1.4) for HPV51. Furthermore, for none of Group 1/2A types did the 95% confidence interval of the ICC:normal prevalence ratio overlap with those for HPV6 (0.4, 95%CI 0.3–0.5). Prevalence of HPV DNA detection in women with ICC and normal cytology, and corresponding ICC:normal prevalence ratios, by HPV type. Among Group 2B types, HPV26, 30, 67, 69, 73, and 82 each showed ICC:normal prevalence ratios above 1.0 that resembled those of Group 1/2A types and which were significantly higher than that of HPV6. Indeed, other lines of evidence also support the carcinogenicity of HPV26, 30, 67, 69, 73, and 82. These types have been reported as single infections in cervical cancer,4 with a ratio of single to multiple infections in meta-analytical data that resembles that of Group 1/2A types.5 In addition, HPV82 is able to immortalize human keratinocytes6 and HPV26, 73, and 82 have been reported as the only actively transcribed type in ICC7 and/or periungual cancer.8 The 13 Group 1/2A types have been already included in all U.S. Food and Drug Administration (FDA)-approved HPV screening tests. Types HPV26, 30, 67, 69, 73, and 82 collectively account for approximately 1.6% of ICC, and may also be considered for inclusion in future HPV screening tests. Importantly, these types are reported so rarely in normal cytology (0.2–0.5% each), that they do not raise great concerns about over-referral of women at low risk for cervical cancer in screening programs. In contrast, some other types in the 2B group, for example, HPV53, 66, and 70, are indistinguishable from HPV6 and other noncarcinogenic types in terms of their ICC:normal prevalence ratio (Fig. 1), and relatively common in women with normal cytology (1.1, 0.6, and 0.8%, respectively). Many of the currently FDA-approved HPV screening tests already include HPV66, and Hybrid Capture 2 is also known to cross-react with HPV53, 66, and 70, phenomena which are expected to unfavorably impact test specificity and positive predictive value.9 Of note, some HPV types that were not classified as carcinogenic showed even lower ICC:normal prevalence ratios than HPV6. Some of these types may have a stronger tropism for the vagina than the cervix,10 and hence be more likely to contaminate cervicovaginal swabs from women with normal cytology than to be found in cervical-directed biopsies of ICC. Some limitations of cross-sectional meta-analyses of HPV type distribution have been described in our previous article in this journal2 including lack of sufficient sample size to explore the potential heterogeneity in the carcinogenicity of rare HPV types by geographical region or HPV-testing protocols. However, the adjustment of ICC:normal prevalence ratios by geographical region, PCR primers and/or publication year, did not materially affect the strength nor the significance of our present findings (data not shown). In summary, our findings suggest that some, but not all, rare Group 2B types that had not been classified as carcinogenic, or probably carcinogenic, by the IARC Working Group1 may be upgraded and reconsidered as pertinent targets for cervical cancer prevention. Jean-Damien Combes was supported by a fellowship from the Association pour la Recherche sur le Cancer (ARC, N 20111204169). Peng Guan, while at the International Agency for Research on Cancer (IARC), was supported by the IARC International Postdoctoral Fellowships. Yours sincerely, Jean-Damien Combes Peng Guan Silvia Franceschi and Gary M. Clifford