THERE ARE THREE KEY EVENTS in the development of cervical cancer: infection of cervical metaplastic epithelial cells with an oncogenic type of human papillomavirus (HPV), progression of a clone of persistently infected cells to precancer, and invasion.1 Although the approximately 15 oncogenic HPV genotypes cause virtually all cases of cervical cancer and precancer, these HPV types are also extremely common, transient, sexually transmitted agents typically acquired early in sexual life.2–6 In prospective studies, epidemiologists are studying the risks and rates of infection and of progression to precancer among HPV-infected women.
Longitudinal studies in young women have found cumulative risks of approximately 40% of HPV acquisition within 2 to 3 years of initiation of sexual intercourse with HPV type 16 accounting for approximately 25% of the infections.7,8 The majority of these early HPV infections become undetectable spontaneously within <1 to 2 years of first detection.3,9–11 Cytologic high-grade squamous intraepithelial lesions (HSIL) and histologic diagnoses of cervical intraepithelial neoplasia grade 3 (CIN 3, the precursor of cervical cancer) are uncommonly diagnosed among women in the first few years of sexual life, although the relative risk of CIN 3 is increased substantially among HPV 16-infected women. At its shortest, the time between the first HPV-positive test and the CIN 3 diagnosis can be just a few years.7,10 However, the median time to diagnosis of CIN 3 has not been well-defined partly because the short follow up of most of the published studies has ended before reaching the reported peak age of CIN 3 lesions.
Our population-based cross-sectional studies12,13 would suggest that, on average, ≥10 years elapse in Guanacaste between the peak of HPV incidence soon after sexual initiation and the later peak of diagnosis of CIN 3. However, the peak incidence of CIN 3 is necessarily dependent on screening sensitivity and intensity. We reasoned that more frequent, intensive surveillance of a population-based sample of initially virginal women might detect small CIN 3 lesions earlier than our cross-sectional screenings, leading to a better understanding of the true time between infection and progression.
This study was conducted in a subcohort of our Guanacaste population and included initially self-reported virgins who were followed for up to 5.8 years (median = 3.6 years) after reported first sexual intercourse (leading to their first pelvic examination). We examined the natural history of early HPV infections assessed primarily by DNA testing, with ancillary serologic testing, to understand the interrelation of longitudinal data with previously reported cross-sectional data.
Study Design and Population
A representative random sample of census segments (there are 40–60 households per segment) from Guanacaste Province in Costa Rica was selected and all women ≥18 years old living in those segments were invited to participate in the Proyecto Epidemiológico Guanacaste. This population-based cohort was established to study the natural history of HPV infection and cervical neoplasia. Between June 1993 and December 1994, 10,049 women were recruited into the study with a participation rate of 93.6%. Details have been described elsewhere.14
Of the 583 women who self-reported as virgins, 409 were 26 years old or younger and were followed at yearly intervals but switched to a 6-month follow-up schedule once they reported initiation of their sexual life. Virgins aged 27 and above (n = 174) who were considered less likely to begin having sexual relations during the follow up with progression to ≥CIN 2 were followed up for the first time 5 to 7 years after enrollment and are not considered here. In the younger virgin subcohort included in this analysis, 22 women never came for follow-up visits (12 refused to continue in the study, 8 moved out of the study area, one was seriously ill, and one had a hysterectomy performed) and 135 were excluded because they never reported first sexual intercourse (n = 131) or a pelvic examination was never done (n = 4). We further restricted the analysis to the initially virginal women who had a pelvic examination within 2 years of first sexual intercourse (n = 206, see Fig. 1). A different reason delayed first pelvic examination after sexual debut in some women, e.g., pregnancy shortly after first sexual intercourse or failure to return.
Possible cases of CIN 2 or worse were treated on an individual basis precluding further epidemiologic follow up. Specifically, women were censored during the study follow up if they presented an HSIL cytologic result or if the cervigram (see subsequently) suggested CIN 2, CIN 3, or cancer regardless of whether the final diagnosis was confirmed CIN 2 or worse. Women were also censored if visual examination during a follow-up visit by the examining nurse suggested cancer, but only if the lesion was confirmed by the gynecologist's colposcopy at referral; otherwise, the visual inspection was considered nonspecific and noncensoring. Of the 206 women, only 10 refused to continue in follow up, all of them after completing at least 1 follow-up visit.
The study protocol was reviewed and reapproved annually by the National Cancer Institute and the Costa Rican Institutional Review Boards.
Enrollment and follow-up visits were similar and were based on the sexual status of the participants assessed by an interviewer-administered standardized questionnaire. A heparinized blood sample was drawn and aliquoted for long-term storage at −80oC or in liquid nitrogen. A pelvic examination was performed on sexually active women, including collection of exfoliated cells with a Cervex brush (Rovers Medical Devices, B.V., Oss, The Netherlands) for conventional and liquid-based cytology (ThinPrep; Cytyc Corp., Boxborough, MA). Additional cells were collected with a Dacron swab and stored in ViraPap DNA transport medium (Digene Corp.) later changed to Digene's DNA standard transport medium (STM; Digene Corp.) for HPV DNA detection and typing. Finally, 2 images from the cervix were taken after a thorough rinse with 5% acetic acid (Cervigram; National Testing Laboratories, High Ridge, MO). Details have been described elsewhere.14,15
Human Papillomavirus Determination by Polymerase Chain Reaction
We detected and genotyped HPV in exfoliated cervical cells stored in specimen transport medium (Digene Corp.) using MY09/MY11 L1 consensus primer polymerase chain reaction (PCR) with AmpliTaq Gold polymerase as detailed elsewhere.13,16 In brief, after amplification, PCR products were analyzed by electrophoresis and hybridized with radiolabeled generic probes for HPV. Positive samples using the radiolabeled probe were typed by dot blot hybridization with biotinylated type-specific oligonucleotide probes (2, 6, 11, 13, 16, 18, 26, 31–35, 39, 40, 42–45, 51–59, AE9 [54v], 61, 62, 64, 66–74, AE10 [74v], 81, AE2 [82v], w13b,  83–85, and 89). We considered types 16, 18, 26, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, 73, AE2 (82v), and w13b (82) as oncogenic.17,18 Specimens negative for type-specific probes but positive for the more sensitive generic probe were classified as uncharacterized HPV DNA-positive.
Human Papillomavirus Virus-Like Particle Determination by Enzyme-Linked Immunosorbent Assay
Anti-HPV L1 antibody determination in plasma samples collected at each of the study visits was done with a virus-like particle-based enzyme-linked immunosorbent assay (ELISA) for HPV types 16, 18, 31, 45 and 53 at the Johns Hopkins Medical Institutions as described elsewhere.19,20 A stringent positive cut point was selected for this analysis defined as 5 standard deviations above the mean of the distribution of absorbance values of concurrently tested, self-reported virginal women (before initiation of intercourse). To minimize the possible effect of any false assessments, the virginal women's absorbance value distribution was obtained after excluding all absorbance values that were greater than the mean plus 2 standard deviations and repeating this iterative process until no outlying values were seen.
The cell samples obtained by the cervix collection brooms used at pelvic examination were split to make conventional Pap smears and liquid-based ThinPrep preparations. The ThinPreps were interpreted twice in the United States and for the last half of follow up in Costa Rica as well.15 Results were reported using the Bethesda System.21
The two cervigrams taken at each pelvic examination were sent to National Testing Laboratories, Worldwide for processing and interpretation by an expert evaluator.15,22
A colposcopic biopsy algorithm incorporating, age, parity, colposcopic appearance, and screening results determined whether punch biopsy or the loop electrosurgical excision procedure (LEEP) was taken. A treatment algorithm individualized subsequent LEEP or hysterectomy, if any treatment was needed, depending on all results.15
Before exiting the cohort, we strove to maximize the safety of those women with more subtle abnormalities. Those with HPV infection with the same type at enrollment and exit visit, minor cytologic or cervicographic abnormalities in any of the last two visits, or HPV 16 or 18 on the last visit were referred to colposcopy with an emphasis on thorough endocervical evaluation.
Histology Review and Final Diagnostic Group
After ending cohort follow up, a masked review of histology slides was performed in the United States by one of two pathologists (M.S., D.S.). In cases in which the diagnosis in the United States and Costa Rica disagreed, a third masked review was performed and the majority opinion constituted the final diagnosis (M.S., D.S., and Costa Rica). If the three pathologists all disagreed, a joint review in the United States yielded the final diagnosis.
If a woman missed an appointment, she was rescheduled repeatedly until she came to the clinic. To standardize time among women within the cohort for time-dependent analysis, 6-month bins were defined using the date of the first pelvic examination with valid test results as the starting point for the creation of bins. Each clinic visit was assigned to a bin.
To study the cumulative risk of acquiring a new HPV infection, as defined by our periodic HPV DNA testing, we used the Kaplan-Meier estimator. Only those who had tested negative in all previous bins were considered at risk. We excluded 2 women who did not have a valid HPV DNA test result at the time of the first pelvic examination. Time to infection was measured from the first pelvic examination until a new HPV infection with any HPV type or until the end of the observation period if the woman never tested positive for an HPV type. Missing visits after the first pelvic examination but before the first positive HPV test were assumed negative. If a woman had more than one visit within a bin (n = 22), visits were combined keeping positive results from both visits. Missed bins without subsequent attended visits were excluded from all calculations. Our calculations assume, as an obvious limitation, that the fraction of women returning within each bin acted as a random sample of all women that could have returned during that bin.
Simple proportions were used to estimate crudely the rates of progression to CIN 3 given the scant number of cases. Given how quickly most HPV infections clear, it was not possible to study the time to clearance of infections. Specifically, we could not presume that an infection remained present until the first negative test or that it disappeared at any particular point between tests. To assume that clearance occurred at the midpoint between a positive and a subsequent negative observation would have falsely lengthened the persistence time because infections have been shown to clear after a curve that is much faster than simple linear or even exponential assumptions.11 Rather than make assumptions, we provided a few percentages based on those women who returned in successive bins.
Proportions were used to estimate the rate of seroconversion and seropersistence. Women with previous negative ELISA results to all HPV types, who at a subsequent visit tested positive for a type, were considered to have seroconverted. While recognizing the limitations of such a definition, we classified any 2 positive ELISA results for a specific type as seropersistence for that type.
The 206 women included in this analysis accumulated a total of 1,106 visits after having their first sexual intercourse with a mean of 5.4 (standard deviation = 3.0; median, 6; range, 1–11) visits per woman and a mean visit gap of 10.9 ± 7.7 months and median follow-up time of 3.6 years.
The mean age at enrollment was 20.9 ± 2.2 years and the mean age of menarche was 12.8 ± 1.4 years. The mean age of first sexual intercourse (23.4 ± 2.8 years) was later than for the rest of the Guanacaste cohort (18.5 ± 4.0 years) because we preconditioned inclusion in the study on recent initiation of sexual life, and the full cohort did not include any women under age 18. The mean interval between first sexual intercourse and first pelvic examination was 13.4 ± 6.5 months and 95.6% of women reported only one lifetime sexual partner at the time of the first screening visit. No woman reported ever smoking and 152 (73.8%) reported ever having used a hormonal family planning method.
Cumulative detection of HPV DNA was common with 110 (53.4%) women testing positive for at least one infection at least once. For 105 of the 110 women, we were able to identify the specific genotype. Concurrent and sequential infections with different types were common. Cumulatively, 37 women were observed to have one type, 23 had 2 types, 18 had 3 types, 13 had 4 types, 7 had 5 types, 3 had 6 types, one had 7 types, one had 8 types, and 2 had 9 types. The proportions of women who tested positive at least once for oncogenic and nononcogenic types were 38.4% (n = 79) and 38.8% (n = 80), respectively. The overall incidence of infection women was 6.5% (95% confidence interval = 4.8–8.9) per 6-month observation bin. The cumulative incidence rates at approximately 1, 3, and 5 years were 15.7%, 34.2%, and 49.8%, respectively (Fig. 2). The low rate of infection within 1 year might reflect that fewer women (approximately 40%) were tested 6 months after the first pelvic examination than the proportion of women tested at each of the subsequent bins (approximately 50%).
The 110 women who were ever HPV DNA-positive contributed 275 type-specific prevalent (found at their first examination) and incident HPV infections. As a result of irregular follow-up intervals that are longer than the time course of many HPV infections, we can only crudely estimate persistence and clearance. Of the 275 type-specific infections, 6 were present at the time of CIN 3 and 58 infections were detected for the first time at the woman's last screening visit and not considered subsequently. If we presume that a single, type-specific negative test represents clearance, we observed clearance during follow up for 185 (87.7%) of the 211 remaining infections; 180 cleared by 30 months. Among the remaining 26 infections that were still present at the time of last observation, 9 were observed to have lasted for 6 months, 5 for 12 months, 6 for 18 months, 2 for 24 months, 2 for 30 months, one for 36 months, and one for 48 months.
Among the prevalent and incident infections combined, HPV 16 was the most common oncogenic type with 19 (9.2%) women testing positive at some time, followed by HPV 66 and 52 with 16 (7.8%) and 15 (7.3%) cumulative incidence, respectively. Only 8 (3.9%) women tested positive for HPV 18 at some point (Fig. 3). HPV 53 infection was the most common among the nononcogenic types with 19 (9.2%) women positive; 7 other nononcogenic HPV types were present in 3.9 to 6.3% of the women (Fig. 4).
Antibodies against at least one of 5 different HPV types (HPV 16, HPV 18, HPV 31, HPV 45, and HPV 53) were detected in 99 (48.1%) women at some point during the duration of the study. However, only 62 of these women demonstrated antibodies against the same type at more than one visit. Of the 206 virginal women who became sexually active during follow up, 53 were seropositive at a visit when they were supposedly still virginal and the remaining 46 of 99 developed antibodies against one of the 5 HPV types tested on samples at some point during follow up after sexual debut. There was not a strong relationship between serology and DNA test results (data not shown).
During follow up, 13 women (6.3%) had HSIL cytology based on a conventional smear or ThinPrep result. Histologic diagnoses of CIN 3 (i.e., definite precancer) were uncommon, with only 3 (1.5%) women diagnosed as CIN 3 over the follow-up period. The proportion of women concurrently or subsequently diagnosed with CIN 3 histologic lesions among those infected at some point with an oncogenic HPV type was low (2.5%). Expressed another way, among the 21 women who were censored during follow up because of HSIL cytology, or high-grade cervicography or colposcopic impression, 15 were never diagnosed with CIN 3; some might have represented incipient precancerous lesions (one woman had a histologic result of CIN 2).
The 4 cases of CIN 3 occurred at 24, 27, and 29 years of age with times between first sexual intercourse and CIN 3 diagnosis of 30, 39, and 40 months, respectively. Both CIN 3 lesions with a preceding record of a positive HPV DNA test were associated with persistence of HPV 16, although one of them was persistently positive for HPV 66, another oncogenic type, as well. Both women had antibodies against HPV 16 persistently detected. The third woman with a CIN 3 lesion was HPV PCR-negative in samples taken at the 2 visits that occurred after her sexual debut and before the diagnosis, although antibodies against HPV 31 were detected in one of the visits (see Table 1 and Fig. 5). We retrieved the diagnostic tissue block and tested it for HPV DNA but no HPV type was detected.
These data from a small subcohort of initially virginal women from a population-based prospective study corroborate that HPV infection is extremely common soon after sexual debut and that apparently precancerous outcomes (CIN 3) can uncommonly occur within a few years of reported first infection. However, we observed no invasive cancers among these women and cannot assess the risk or timing of invasion, because we immediately treated all women with ≥CIN 2 lesions, precluding assessment of the natural history of these lesions.
This analysis confirmed previous reports, summarized in Table 2, that HPV infection, although very common in the years immediately after initiation of sexual activity, typically resolves very quickly. The overall proportion of these young women that had cervical HPV DNA detected at some point during the study (53.4%) was similar to results of other studies; however, as Table 2 shows, this apparent similarity could be the result of multiple differences in study design acting in different directions, e.g., different laboratory methods used for HPV DNA detection, frequency of sample collection, and duration of follow-up time.
Because so very few infections persisted for >2½ years, we wonder about the usual state of the HPV-infected cervix before the diagnosis of precancerous lesions. In the complete cross-sectional data set from enrollment in the Guanacaste cohort, there was a lag of >10 years between the peak prevalence of HPV infection (37.8% among women aged 18–24) and the peak prevalence of histologic CIN 3 (2.3% in the 30- to 34-year age group). The fact that most HPV infections in the age range of 18 to 24 years seem to spontaneously disappear quickly leaves unanswered questions about which women develop infections that result in CIN 3.
There are 2 alternative hypotheses that are being widely discussed. In the first hypothesis, viral persistence and progression to precancer are viewed as sequential transitional events. In this view, oncogenic infections that persist in association with a low-grade lesion sometimes give rise to subclones with genetic changes making them precancerous. Our group has published this view,26 although we are increasingly skeptical that it can explain available data. In the second hypothesis,27 undetectable precancerous lesions develop very early at the internal margin of a low-grade viral lesion(s). The precancerous lesions grow out to a diagnosable size as the low-grade lesions around them regress.
We are attempting to reconcile the following observations. First, this longitudinal analysis did not reveal a sizable pool of persistent HPV infections that could later progress to CIN 3; in fact, vanishingly few infections persisted past 2 to 3 years, except for those in the context of already diagnosed CIN 3 or at least a lesion that was suspicious for precancer. (Of course, censoring after abnormal screening results may have limited the number of such cases.) Second, based on other prospective studies, we believe that the many young women in this subcohort who quickly cleared infections remained at low risk of CIN 3 and cancer in the subsequent decade.28,29 In fact, recurrent type-specific infection occurred rarely during the 7-year follow up of >8,000 women in the main cohort from which this subgroup was drawn.28
Therefore, the CIN 3 lesions found in these young women (median age of 27 years at the last visit) might represent the bulk, not just the early rise, of the expected later peak of CIN 3 emerging from this cohort of women. Each cohort of women initiating sexual intercourse would of course contribute, in an overlapping fashion, to produce the cross-sectional age-specific prevalence of CIN 3 that peaks between 25 and 35 years of age depending on study. We hypothesize, therefore, that the median age of CIN 3 in other carefully followed cohorts should be much younger than in cross-sectional data from the same populations before careful screening. However we cannot rule out the possibility that infections acquired at later ages have a greater potential for progression in women who have accumulated more years of exposure to known progression cofactors. We will be investigating the relative fate of incident infections acquired at different ages in the full analysis of the main cohort.
As we attempt to define the transition between infection and precancer, the precise definition of precancer will be increasingly important. Focal CIN 3 lesions, among younger women, may represent an imperfect surrogate of cancer risk. If cervical cancer development is analogous to other carcinomas, a long period of in situ growth may occur before invasion. The tiny lesions found with intensive screening might have higher regressive potential in addition to a longer latency before cancer development.30 A larger, thorough case study of the host and the lesion from visual and molecular perspectives among early and later CIN 3 lesions is necessary to address this point.
1. Wright TC, Schiffman M. Adding a test for human papillomavirus DNA to cervical-cancer screening. N Engl J Med 2003; 348:489–490.
2. Bosch FX, Lorincz A, Muñoz N, et al. The causal relation between human papillomavirus and cervical cancer. J Clin Pathol 2002; 55:244–265.
3. Ho G, Bierman R, Beardsley L, et al. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med 1998; 338:423–428.
4. Krüger-Kjaer S, Chackerian B, van den Brule AJC, et al. High-risk human papillomavirus is sexually transmitted: Evidence from a follow-up study of virgins starting sexual activity (intercourse). Cancer Epidemiol Biomarkers Prev 2001; 10:101–106.
5. Collins S, Mazloomzadeh S, Winter H, et al. High incidence of cervical human papillomavirus infection in women during their first sexual relationship. BJOG 2002; 109:96–98.
6. Richardson H, Kelsall G, Tellier P, et al. The natural history of type-specific human papillomavirus infections in female university students. Cancer Epidemiol Biomarkers Prev 2003; 12:485–490.
7. Woodman CBJ, Collins S, Winter H, et al. Natural history of cervical human papillomavirus infection in young women: A longitudinal cohort study. Lancet 2001; 357:1831–1836.
8. Winer RL, Lee SK, Hughes JP, et al. Genital human papillomavirus infection: incidence and risk factors in a cohort of female university students. Am J Epidemiol 2003; 157:218–226.
9. Moscicki AB, Shiboski S, Hills NK, et al. Regression of low-grade squamous intra-epithelial lesions in young women. Lancet 2004; 364:1678–1683.
10. Winer RL, Kiviat NB, Hughes JP, et al. Development and duration of human papillomavirus lesions, after initial infection. J Infect Dis 2005; 191:731–738.
11. Plummer M, Schiffman M, Castle PE, et al. A two-year prospective study of clearance versus persistence of individual types of cervical human papillomavirus infection among women with ASCUS or LSIL cytology. J Infect Dis. In press.
12. Herrero R, Hildesheim A, Bratti C, et al. Population-based study of human papillomavirus infection and cervical neoplasia in rural Costa Rica. J Natl Cancer Inst 2000; 92:464–474.
13. Herrero R, Castle PE, Schiffman M, et al. Epidemiologic profile of type-specific human papillomavirus infection and cervical neoplasia in Guanacaste, Costa Rica. J Infect Dis 2005; 191:1796–1807.
14. Herrero R, Schiffman M, Bratti C, et al. Design and methods of a population-based natural history study of cervical neoplasia in a rural province of Costa Rica: The Guanacaste Project. Rev Panam Salud Publica 1997; 1:362–375.
15. Bratti MC, Rodríguez AC, Schiffman M, et al. Description of a seven-year prospective study of human papillomavirus infection and cervical neoplasia among 10,000 women in Guanacaste, Costa Rica. Rev Panam Salud Publica 2004; 15:75–89.
16. Castle PE, Schiffman M, Gravitt PE, et al. Comparisons of HPV DNA detection by MY09/11 PCR methods. J Med Virol 2002; 68:417–423.
17. Muñoz N, Bosch FX, de Sanjose S, et al. Epidemiologic classification of human papillomavirus types associated with cervical cancer. N Engl J Med 2003; 348:518–527.
18. Cogliano V, Baan R, Straif K, et al. Carcinogenicity of human papillomaviruses. Lancet Oncol 2005; 6:204.
19. Wang SS, Schiffman M, Shields TS, et al. Seroprevalence of human papillomavirus-16, -18, -31, and -45 in a population-based cohort of 10 000 women in Costa Rica. Br J Cancer 2003; 89:1248–1254.
20. Viscidi RP, Schiffman M, Hildesheim A, et al. Seroreactivity to human papillomavirus (HPV) types 16, 18, or 31 and risk of subsequent HPV infection: Results from a population-based study in Costa Rica. Cancer Epidemiol Biomarkers Prev 2004; 13:324–327.
21. The 1988 Bethesda System for reporting cervical/vaginal cytologic diagnoses. National Cancer Institute Workshop. JAMA 1989; 262:931–934.
22. Schneider D, Herrero R, Bratti C, et al. Cervicography screening for cervical cancer among 8460 women in a high-risk population. Am J Obstet Gynecol 1999; 180:290–298.
23. Giuliano AR, Harris R, Sedjo RL, et al. Incidence, prevalence, and clearance of type-specific human papillomavirus infections: The young women's health study. J Infect Dis 2002; 186:462–469.
24. Muñoz N, Mendez F, Posso H, et al. Incidence, duration and determinants of cervical human papillomavirus infection in a cohort of Colombian women with normal cytological results. J Infect Dis 2004; 190:2077–2087.
25. Brown DR, Shew ML, Qadadri B, et al. A longitudinal study of genital human papillomavirus infection in a cohort of closely followed adolescent women. J Infect Dis 2005; 191:182–192.
26. Schiffman MH, Brinton LA. The epidemiology of cervical carcinogenesis. Cancer 1995; 76(suppl):1888–1901.
27. Kiviat NB, Critchlow CW, Kurman RJ. Reassessment of the morphological continuum of cervical intraepithelial lesions: Does it reflect different stages in the progression to cervical carcinoma? IARC Sci Publ 1992; 119:59–66.
28. Schiffman M, Herrero R, DeSalle R, et al. The carcinogenicity of human papillomavirus types reflects viral evolution. Virology 2005; 324:17–27.
29. Sherman ME, Lorincz AT, Scott DR, et al. Baseline cytology, human papillomavirus testing, and risk of cervical neoplasia: A 10-year cohort analysis. J Natl Cancer Inst 2003; 95:46–52.
30. Myers ER, McCrory DC, Nanda K, et al. Mathematical model for the natural history of human papillomavirus infection and cervical carcinogenesis. Am J Epidemiol 2000; 151:1158–1171.
© Copyright 2007 American Sexually Transmitted Diseases Association
This article has been cited