This study is the first to estimate the prevalence of oncogenic HPV among women with high grade lesions in New Zealand. The findings of this survey confirm that, as for other regions, HPV 16 is the most common HPV type among women with a high grade cytology report and in women with histologically-confirmed CIN2+. The prevalence of HPV16 in CIN 2/3 in New Zealand was broadly consistent with that in Australia and Europe (about 50%)  but was higher than that reported for North America, Asia, and South/Central America (all less than 40%) . In New Zealand, HPV 18 was observed in 12.1% of women with a histological diagnosis of CIN2/3. This was broadly consistent with reported rates in Australia and North America but more than that reported in Asia, Europe and South/Central America.
The study has also provided new more detailed information than has been reported to date on the pattern of the relative prevalence of HPV 16 and 18 infections in high grade lesions by age. For HPV type 16/18, prevalence peaked among women aged 20–29 years and fell with increasing age. Conversely, for the OHR HPV types, prevalence was lowest among women aged 20–29 years, and increased with increasing age to a peak among women aged 40–69 years. This observation is broadly consistent with patterns observed in the Guanacaste, Costa Rica cohort, which highlighted a similar pattern for HPV 16 and OHR types . It is also consistent with the findings of a study of age-specific HPV prevalence in CIN 3/AIS as registered in 3 US cancer registries (Michigan, Iowa, California) between 1994–200 , with a study of the age-specific HPV prevalence in Danish women subsequently diagnosed with CIN2/3,  and with a study in United Kingdom which examined the prevalence of HPV 16/18 in histologically-confirmed CIN3 . This age-related pattern of infection supports a proposed model of disease development that contends that HPV16 is more likely to progress to CIN3 precancerous disease within a shorter period, whereas OHR types progress slowly and less frequently to precancerous abnormalities .
An important strength of the study is that we recruited from the population-based National Cervical Screening Program register. A total of 27% of the population initially identified as potentially eligible were included in the final analysis, after taking into account exclusions, the ability to contact women, consent and HPV test success rates. However, of the women who could be approached a relatively high consent rate was obtained, and the observed distributions of age and ethnicity in the study population were very similar to that in the broader screening population of women aged 20–69 years with a reported high grade cytology result in New Zealand (Table 2). The extent of underlying high grade CIN2+ in the study population of women with high grade cytology (about 70% overall) was also very similar to that reported in the population overall for the first half of 2009 (70.2%) . A similar distribution of oncogenic HPV types in women with histologically-confirmed CIN 2+ was observed in women recruited via both recruitment channels (Additional file 1: Appendix Table 3A). These findings suggest that the overall study results are broadly representative of the wider population of women with high grade cytology.
In our study, and in the prior Australian study, the overall prevalence of oncogenic HPV in CIN 2/3 appeared to be somewhat higher than reported other regions. One explanation for this finding is improvements in PCR technology used to detect and genotype HPV infection. In the period between studies reported in the worldwide meta-analysis and the current New Zealand study, PCR testing has undergone two significant improvements in technology: the development of new oligonucleotide primers (PGMY09/11) and the inclusion of an AmpliTaqGold polymerase. PGMY09/11 primers were designed using DNA sequence homology for various HPV types indicated within the highly conserved region of the L1 region. In contrast to this method of designing primers, the original ‘manos’ MY09/11 primers used in previous prevalence studies were ‘degenerate’, which meant they did not necessarily provide reproducible estimates of HPV types detectable in samples . The improvement in design accompanying the PGMY primers translated into greater sensitivity for detecting HPV types 26, 35, 42, 45, 52, 54, 55, 59, 66, and 73 when compared with the traditional ‘manos’ MY09/11 primers . In addition to improvements in primer design, AmpliTaqGold polyermase, introduced in early 2000, improved enzymatic characteristics and subsequent sensitivity for detecting HPV type infections compared with the older AmpliTaq polymerase . Another factor that may account for the higher overall prevalence of oncogenic HPV in our study, relative to previous reports, is the method of recruitment. In this study, women were recruited following a high grade cytology report. We then assessed the prevalence of overall and type-specific oncogenic HPV in the subgroup of women with histologically-confirmed high grade disease. Our approach may have led to a more concentrated recruitment of women with ‘true’ high grade CIN. By contrast, previous studies may have been more likely to have included women with misclassified CIN 2 lesions . Our clinical approach to enriching the study population provides a potential addition or alternative to implementing a recently proposed ‘gold standard’ for diagnosing CIN2/3 histology, involving use of laser capture microdissection to characterize the molecular features of suspected pre-cancerous lesions.
In the current study it was assumed that samples positive for type 52 alone represented true type 52 infection, but that samples testing positive for 52 and also for at least one of types 33, 35 and 58 were not truly positive for type 52 (i.e. it was assumed apparent type 52 reactivity represented cross-reactivity). Dedicated probes are required for accurate detection of HPV type 52 in the presence of multiple infections . However, in practice, cross-reactivity was not a major concern in the current study since only one sample was positive for another type in addition to type 52. Because of our conservative approach to this aspect of the analysis, our finding of a high prevalence for type 52 may be, if anything, a slight underestimate.
This study provides a baseline measure of oncogenic HPV prevalence in a population of women in New Zealand, and provides a baseline for future similar surveys to assess the impact of HPV vaccination. Genotyping studies for high grade lesions have previously been included in a list of key recommendations to monitor the effect of HPV vaccination . Our findings imply that the current HPV vaccination program in New Zealand, which involves delivery of a vaccine against HPV types 16/18, could prevent up to 62% of high grade lesions (53% of CIN 2 and 66% of CIN 3+). These estimates are based on the assumption (in the best case for vaccination) that co-infection of other oncogenic types, in the presence of HPV 16 and/or 18, was not causally responsible for the development of the majority of high grade lesions in the current study; if this assumption does not hold then the proportion of vaccine-preventable high grade lesions could be considerably lower. These estimates of potential vaccine effect should be considered highly provisional, and a number of other factors will be influential, including vaccine program coverage in New Zealand, prior expose of the catch-up cohort to vaccine-included HPV types, and the potential for cross-protection against non-vaccine-included types. Further surveillance of oncogenic HPV infection in histologically-confirmed high grade disease in the post-vaccination era in New Zealand will continue to be informative.