Persistent infection with certain human papillomavirus (HPV) strains (types) is accepted as the necessary cause of invasive cervical cancer (ICC). However, for the development of screening tests and vaccines, the carcinogenicity of each HPV type needs to be judged separately. Whilst some types, like HPV16 and HPV18, are clear and powerful carcinogens, classification of rare HPV types has been challenging: not only are data on prevalence in ICC scarce, but as all human Alphapapillomavirus share a common mode of sexual transmission, confounding by multiple infection with other carcinogenic HPV types is difficult to rule out.
An International Agency for Research on Cancer (IARC) working group classified 12 HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59) as carcinogenic to humans (Group 1).1, 2 These 12 types cluster together in the same evolutionary branch or “high-risk clade” that includes Alphapapillomavirus species groups 5, 6, 7, 9, and 11.2, 3 Other types in the high-risk clade were classified as possible carcinogens (Group 2B) based upon their phylogenetic relatedness [with the slight exception of HPV68, which was upgraded to probable carcinogen (Group 2A)].1, 2 However, the prevalence of Group 2A/2B-classified HPV types in ICC was not distinguishable from that of HPV6, which is not considered to be a cervical carcinogen.2
In a recent meta-analysis of 30,848 ICC, the 12 most common HPV types worldwide were consistent with the 12 Group 1-classified HPV types.4 Group 2A/2B-classified types, however, were detected in <0.1–0.6% of ICC, and HPV6 and HPV11 in 0.4 and 0.5%, respectively. However, these prevalence estimates included multiple infections, which accounted for 11% of all ICC cases (a proportion that appeared to be increasing over time due to steady increases in the sensitivity of HPV testing).4
With the hope of improving the carcinogenic classification of rare HPV types, we reanalyzed a relevant subset of studies included in the above meta-analysis, focusing on the ratio between the detection of each type in single and multiple infections (S/M ratio). The subset included 12,106 ICC from 66 studies in which all 12 Group 1-classified HPV types were tested for and information was available on type-specific S/M ratios4 (Table 1). Type-specific prevalence was estimated only among studies that tested for and reported the HPV type in question, thus denominators vary by HPV type. Prevalence estimates for individual types (Table 1) were similar to respective figures based on larger numbers in the overall meta-analysis.4
Table 1 shows type-specific HPV prevalence and S/M ratios among ICC cases positive for each HPV type. There was a strong variation in S/M ratio from 6.33 (95% CI 5.91–6.78) for HPV16 down to 0.36 (95% CI 0.13–0.99) for HPV11. All Group 1-classified HPV showed S/M ratios >0.9. This was also the case for HPV67, HPV68, HPV73, and HPV82. Other Group 2A/2B-classified types were more likely to occur as multiple infections.
HPV6 was chosen as the most frequently detected HPV type in ICC for which detection as a single infection was more likely due to the missed detection of a Group 1-classified type than a causal relationship. In comparison to HPV6, the S/M ratio was significantly higher for all Group 1-classified HPV types. This was also the case for HPV67, HPV68, and HPV73 (HPV82 was of borderline significance). For other types, there was no significant difference from HPV6.
This novel analysis reveals some important findings. First, Group 1-classified HPV types were clearly distinguishable from HPV6 based upon their high S/M ratio in ICC. This suggests that such a measure may be helpful in classifying the carcinogenic potential of rare HPV types. Indeed, certain HPV types in the high-risk clade, most notably HPV67, but also types HPV68, HPV73, and HPV82, are distinguishable from HPV6 and resemble other Group 1-classified types.
Other phylogenetic and/or mechanistic evidence supports the carcinogenicity of these types. HPV67 is phylogenetically related to HPV16 and is the only type in Alphapapillomavirus species group 9 not currently categorized as carcinogenic by the IARC. HPV68 is phylogenetically related to HPV18 and sustains the immortalization of a cancer cell line (ME180). Active HPV73 transcription has been reported in periungual cancers.5 Furthermore, although HPV67, HPV68, and HPV73 were each detected in 0.6% or less of ICC, they are each found substantially more frequently in ICC than in women with normal cytology.6
The 12 Group 1-classified types plus HPV68 are included in FDA-approved HPV screening tests Hybrid Capture 2 (HC2) and Cervista (which includes also HPV66). Our results suggest that HPV67, HPV73 and possibly HPV82 might also be considered for inclusion. These types are reported so rarely in cytologically normal women6 not to raise great concerns about over-referral in screening programs. In contrast, some other types in the high-risk clade, for example HPV53 and HPV66, were not only indistinguishable from HPV6 in terms of their S/M ratio, but are also much more common in women with normal cytology.6 Of note, HC2 cross-reacts with these five types.7 Probably, therefore, negative predictive value of HC2 is increased by detecting 67, 73, and 82, and positive predictive value decreased by detecting 53 and 66.8, 9
The limitations of cross-sectional meta-analyses of HPV type distribution have been described,4 and we did not have sufficient sample size to explore potential heterogeneity in carcinogenicity by region, calendar year (87% of cases were from studies published ≥2000), or histological type of ICC for Group 2A/2B-classified HPV types. S/M ratio was, however, significantly higher [p < 0.001] in adeno/adenosquamous than squamous cell/unspecified carcinoma for both HPV18 and HPV45. While no difference by histological type was evident for HPV16. Furthermore, the proportion of multiple infections could be underestimated or overestimated depending on the HPV testing protocol used.10
Ultimate proof of carcinogenicity of rare HPV types will require large studies identifying transcriptionally active types in microdissected tumors. In the absence of such data, however, the novel analysis presented here lends support to the carcinogenicity of HPV67, HPV68, HPV73, and possibly HPV82 in addition to the 12 Group 1-classified types.