To our knowledge, this is the first prospective study of associations between cervical ectopy, squamous metaplastic activity, and incident HPV16 infection. The observed overall decrease in ectopy from baseline to study end was not surprising because cervical maturation over time would be expected in young women. The mean metaplastic rate (4.2% per year) was also consistent with our previously published results [20
]. Among healthy young women, evidence of squamous metaplastic change significantly increased the risk for subsequent incident detection of HPV16. More specifically, each 1% change in metaplastic tissue (rate of metaplasia) contributed a 17% increased risk of HPV16 detection in the subsequent 4–8 months. In contrast, the sheer extent of ectopy was not found to be a risk factor, whether measured concurrently or up to several months prior to the incident HPV16. These findings were unchanged when adjusted for other factors thought to be associated with ectopy, squamous metaplastic activity, and/or HPV acquisition, including age, years since first sex, recent new sexual partners, hormonal contraception use, condom use, pregnancy, smoking, other genital infections, serology, and concurrent other HPV types.
The prospective study design and the quantitative documentation of the cervical epithelia were major strengths of this study. The prospective design allowed us to measure the predictors and covariates at a wide spectrum of timepoints in relation to the incident HPV16. Study visits were relatively frequent, conducted every 4 months typically, allowing us to better estimate the timepoint of incident infection. However, because biological acquisition of HPV is not immediately detectable due to the inherent limits of clinical HPV detection, consideration of the predictors at several timepoints was advantageous from this standpoint. The use of colpophotography and the quantitative approach to the ectopy measurement were additional strengths that minimized misclassification of the cervical epithelium. The proportional measure of “percent-ectopy” allowed comparison of colpophotographs taken at different visits and allowed a rare look at change in the ectocervix over time. Previous studies of cervical ectopy often relied on only qualitative nonmagnified visual inspection without acetic acid application, which is prone to errors related to subjectivity and obscuring factors such as friability.
An interesting finding was that the salient time interval of metaplasia that was associated with HPV16 acquisition was the 4–8 months prior to HPV16 detection, whereas metaplastic activity at the 0–4 month interval leading up to HPV16 detection was not significant. Given our understanding of the natural history of HPV infection, the significance of the 4–8 month interval is biologically plausible because an incubation or latency period of several weeks to months is estimated to follow the initial infection at the basal epithelial cells [36
], and HPV may not be detected by our current technologies until several months after the initial basal cell infection. The HPV life cycle and viral replication are dependent on host cell replication and differentiation [13
], supporting the notion that metaplastic activity would be a risk factor for infection. The lack of significance of metaplastic activity from the 0–4 month interval appears surprising and may be spurious, requiring replication in other studies. However, our observations of time intervals for risk associated with metaplasia appear consistent with a prior study of sociodemographic and behavioral risk factors among female college students [8
]. Winer et al found that new sex partners during the 5–8 month and 8–12 month intervals prior to incident HPV16 detection was a strong risk factor for incident HPV16, whereas new partners at 0–4 months before detection was not significant. The strongest association was found for partners at the 5–8 month interval, paralleling our findings regarding metaplastic activity. Also, as in our study, new partners was a significant factor when measured at approximately 4–12 months prior to the incident HPV16 detection rather than closer to the incident detection. Consistent results regarding the salient time interval for exposure to new partners were seen in another analysis of repeated HPV infections with new HPV types among our larger cohort as well [37
]. Furthermore, our findings parallel our prior case-control study of low-grade squamous intraepithelial lesions (LSIL), a benign manifestation of HPV infection. Incident LSIL was significantly associated with a more rapid rate of metaplastic change (also measured quantitatively) that preceded the LSIL event but was not associated with ectopy measured at study baseline [38
]. Taken together, these studies suggest that active squamous metaplasia may be an important contributor to both the establishment of HPV16 infection and the development of a clinical cervical lesion.
In contrast with our study, in a cross-sectional population-based study, Castle et al examined tissue affinity of HPV types and reported that a greater extent of ectopy was associated with trends of increasing positivity of several oncogenic types (grouped as α-9 types 16, 31, 33, 35, 52, 58, 67) and decreasing positivity of nononcogenic types (grouped as α-3/15 types 61, 71, 72, 81, 83, 84, 89) [15
]. However, comparison with our study is limited for several reasons. Given the cross-sectional design of the Castle et al study, it was impossible to distinguish between persistent and incident HPV. We also focused on HPV16, whereas Castle et al evaluated high-risk and low-risk HPV groups and did not report on HPV16 alone. Furthermore, Castle and colleagues sampled an older adult population. Older women are less likely to have “active” rates of squamous metaplasia and more likely to have persistent infections. Another difference was our evaluation of ectopy as a proportional measure of the total cervical face, in contrast with measurement of an absolute pixel count to represent ectopy by Castle et al. Comparison of our findings with the two other available cross-sectional studies of ectopy and HPV infection would be inappropriate because these studies focused on ectopy in older women who presented with genital symptoms (discharge or bleeding) rather than healthy young women [16
Limitations of our study include the specific focus on HPV16 and the inherent limitations of available methods for HPV sampling and colpophotography. Future studies with larger sample sizes should be conducted to investigate the role of squamous metaplasia and ectopy in other HPV types since findings regarding biological risks for HPV16 cannot be assumed to apply to all other genital HPV types. We were unable to conduct an extensive investigation of the other types due to their lower incidence rates and the limited follow-up time of our cohort. The HPV sampling by directed cervical lavage was another possible limitation. Currently available methods for clinical cervical HPV detection include cervical lavage and cervical swabs. A prior pilot study by our group demonstrated excellent correlation (κ
coefficient = 0.87; CI, .76–.98) between lavage and swab results [39
]. However, the most definitive and direct approach to diagnose HPV infection of the ectocervical tissue would be an analysis of tissue samples obtained by cervical biopsy. An important limitation of our clinical HPV DNA testing is the inability to identify the initial HPV infection at the cellular level. Although our clinical measurement of metaplastic activity preceded the clinical HPV detection, it is possible that undetected HPV acquisition at the cellular level could have actually preceded the clinically apparent metaplastic activity. In that case, the apparent metaplasia could have represented cellular proliferation mediated by HPV16. These issues would be difficult if not impossible to evaluate in a cohort study because timing and exact location of infection would be impossible to pinpoint and repeated biopsy in women with normal cytology would be likely unethical. Finally, although colpophotography is more accurate than unaided visual exams in detecting ectopy, it has several limitations. First, colpophotography itself is inherently limited in that the optimal angle of view and quality of image capture from visit to visit varies clinically for any given woman. We attempted to address this issue by using the proportional measure of percent-ectopy. Second, for those with metaplasia restricted to the endocervix, we were unable to measure the metaplastic rate because this area is not visible on colpophotography. Third, the use of colpophotography and computerized planimetry to measure the columnar and metaplastic epithelium remains limited by the subjectivity of colposcopic impression itself. However, we believe it is currently the best available methodology for quantitative measurement.
The relationships between cervical ectopy, squamous metaplastic activity, and HPV infection in young women are likely complex. However, our study demonstrated that active squamous metaplasia rather than the sheer extent of ectopy on the ectocervix appears to increase the risk for subsequent detection of HPV16. This in vivo observation is consistent with the life cycle of HPV, which is dependent on host cell replication and differentiation for its own replication and survival.