Recent epidemiologic evidence lists cigarette smoking among the risk factors of cervical cancer (CC) and its precursor [cervical intraepithelial neoplasia (CIN)] lesions. In a pooled analysis of IARC multicentre, case-control studies, women who ever smoked were at a significantly increased risk of developing CC and CIN.1 It seems likely that the effects of smoking on cervical disease could be ascribed to potent tobacco carcinogens, present in high concentrations in the cervical mucus of women smokers.2,3 These carcinogens may contribute additional DNA damage to epithelial cells infected by oncogenic [high-risk (hr)] Human papillomavirus (HPV) types, most notably the disruption of p53 and pRb gene pathways by overexpression of the E6 and E7 viral oncogenes.4
Although never confirmed by in vivo or in vitro experimentation, there is some indication that smoking facilitates the acquisition and stabilization of the virus, thus favoring persistence of HPV infections. It is known from early studies that tobacco components reduce the number and affect the function of Langerhans cells and CD4+ lymphocytes in cervical tissues, and also causes decrease in the activity of natural killer cells.5-7 It is still unclear, however, which of these 2 mechanisms has a major impact on the increased risk of developing CIN or CC among smokers. Indeed, recent data implicate that the prevalence of hr-HPV infections is higher among current smokers, and the strength of this association seems to be dose-dependent (i.e., number of cigarettes smoked).8 Similarly, in a large multicentre screening trial in 3 New Independent States (NIS) of the former Soviet Union (Russia, Belarus, and Latvia), smoking proved to be an independent risk factor for hr-HPV infections (but not prevalence of CIN), but unfortunately, this study did not examine the effect of smoking on the incident CIN.9
Several uncontrolled factors may further complicate the analysis of the role played by smoking in the development of CC and CIN. For instance, some known epidemiologic risk factors (e.g., multiple partners)—associated with an increased probability of contracting hr-HPV infections—are more prevalent among smokers.10 As an inherent flaw carried by many epidemiologic studies, also those that have examined the relationship between smoking and CC were subject to bias and confounding, raising frequent concerns.11-13 One of these uncertainties is the variability of the HPV type-distribution across the world and even within large countries, e.g., like Brazil.14,15 There is a clear indication that different HPV types may have distinct epidemiologic and biologic properties, making results inconsistent and sometimes even conflicting.16 Another reservation repeatedly presented against these population-based HPV studies is the unequivocal fact that epidemiologic, economical, behavioral, and geographical dissimilarities of different populations preclude the generalization of the results from 1 geographical region to another.16
To enable unbiased estimation of the smoking as a risk factor of CIN and CC, study designs need to take into account the concurrent acquisition/persistence of hr-HPV infections and development of incident high-grade CIN (CIN2+). This is appropriately done in our Latin American Screening (LAMS) Study testing 8 optional screening tools in a cohort of over 12,000 women in Brazil and Argentina, and providing prospective follow-up (FU) data on disease outcome.17 In the current study, we examined the role of cigarette smoking as a predictor of 2 main outcomes; (1) acquisition of hr-HPV infections, and (2) development of CIN2+ during a prospective FU of over 1000 women of the LAMS study cohort.
Subjects and Methods
LAMS study is a multicenter population-based cohort study testing optional screening strategies and assessing risk factors of cervical disease in a cohort of over 12,000 women enrolled between January 2002 and November 2003 in Brazil and Argentina. A total of 12,114 consecutive women from the cities of Campinas, São Paulo, and Porto Alegre (Brazil) as well as Buenos Aires (Argentina) were enrolled in the cohort to undergo screening with conventional Pap smear, hr-HPV testing by Hybrid Capture 2 (HC2), visual inspection with acetic acid (VIA) or Lugol iodine (VILI), cervicography, and screening colposcopy.17 All centers performed conventional Pap smear, HC2, and VIA, whereas Porto Alegre performed VILI, Buenos Aires did screening colposcopy, and only Campinas did cervicography.18,19 The study protocol has been approved by the local Ethics Committees of all participating clinics. All enrolled women gave their agreement to participate by signing the Informed Consent Forms, written in their native language.
Women were considered eligible if they met all of the following criteria: (a) age between 15 to 60 years; (b) no previous surgical procedure of the cervix or uterine corpus; (c) had no history of abnormal Pap test in the past year; (d) free of diagnosed (= prevalent) genital wart (external or cervical condyloma), CIN or CC; (e) had no sexual intercourse during the 3 days before the examination; (f) did not have any confirmed or clinically suspect immunosuppresion (HIV, or other conditions).
Study Sites and Their Demographics
The study design, features of the clinical centers, and the demographics of their female populations were described in detail recently.18,19 In brief, Campinas and São Paulo are 2 large southwestern Brazilian cities, located only 100 km apart, with equivalent standards of living. In these cities, CC is the fourth major cause of cancer death among women, accounting for 3.3% of all female cancer deaths.20 The third Brazilian study site is from Porto Alegre, located in south of the country. This region offers the best quality of life in Brazil, and there, CC is the sixth major cause of cancer death, accounting for 6.1% of all female cancer deaths.20 The Argentine study site is from Buenos Aires. The country has an overall CC mortality rate of 7.6 of 100,000 women,21 but most of the national statistics pertain to Buenos Aires city only.
All women were subjected to thorough pelvic examination in this sequence: (i) collection of the Pap smear, and (ii) VIA. For approximately 35% of the women, a HC2 sample was collected after the Pap test had been collected and before performing VIA. The test was not available for all women for (a) budgetary constraints, and (b) the small improvement in sample power that would be theoretically obtained by augmenting the percentage of women tested for HPV above 35% of the sample. Post hoc analysis determined that women who were tested for HPV had a similar epidemiologic profile compared to those that were not tested, warranting further analysis using the whole sample. In Porto Alegre, most women were subjected to VILI shortly after VIA. All women, who had at least one of these examinations abnormal, were referred for colposcopic examination. In Buenos Aires and Campinas, women were subjected to screening colposcopy even when their exams were negative. Abnormal colposcopy prompted punch biopsies of the cervix, and women with high-grade cytologic abnormalities (HSIL) were treated by conization.
Women had their second visit scheduled 1½ month (average 45 days) to become informed about their examination and/or biopsy results and to be allotted to either (a) the treatment, or (b) the FU group. Treatment was offered to all women who had high-grade lesion (CIN2+) confirmed in the cervical biopsy. Altogether, 32 cases of invasive CC, 57 of CIN 2, and 86 of CIN 3 were diagnosed during the recruitment phase and all those were treated according to each institution's standard protocols. Women who were considered as “negative” for CIN were either (i) those with all screening tests negative; (ii) with 1 or more positive, than referred for colposcopy and whose colposcopy was negative, and (iii) women subject to colposcopically targeted biopsies with pathologic assessment ruling out CIN.
Patients who did not require treatment for CIN2+ and who had (1) abnormal Pap smear, and/or (2) hr-HPV infection, and/or (3) CIN1 at baseline were selected for FU. Five percent of the women with all screening tests negative, and who attended the postscreening visit at which patients were apprised of their screening test results, were also randomly assigned to the FU group. In the original study design, 4 such FU visits were scheduled (optimally at 6-, 12-, 18-, and 24-months). At each FU-visit, women responded to a brief questionnaire addressing any relevant gynecological events and epidemiologic features changed since the previous control (e.g., sexual partners, smoking). The FU-phase of the study was concluded in September 2006, with the longest FU-times exceeding 50 months, but the bulk of the data covers approximately 24 months (median FU time = 24.4 months; 90% central range = 6.8–32.2 months).
Cervical Cytology (Pap Smear)
Conventional Pap smears were taken using the Ayre spatula and endocervical brush, fixed in 95% ethanol and stained by the modified Papanicolaou method. Final cytologic diagnoses were issued using the Bethesda System22 and were classified as normal/inflammatory, ASC, atypical glandular cells (AGC), LSIL, HSIL, or cancer.
Hybrid Capture 2
The specimens for HC2 were tested with probe B for high-risk HPV (hr-HPV) types: HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68,23 and the tests were classified positive at the relative light unit/positive control (RLU/CO) ratio of 1pg/mL or greater. These RLU/CO ratios provide a semiquantitative estimate of the amount of HPV DNA in the specimens, i.e., the viral load in the sample. Storage of reagents and specimens as well as their processing were carried out in manufacturer-certified laboratories, under the responsibility of the investigators, following the manufacturer's instructions (Digene Diagnostics Inc.). São Paulo and Buenos Aires processed their own HC2 samples in-house, whereas Campinas and Porto Alegre had their HC2 specimens processed at Campinas University hospital laboratory.
Retrieval of the Smoking Data
The information concerning patients' smoking behaviors was gathered during the baseline interview. Patients were asked the following questions: (1) Do you smoke? (2) If you smoke, how many cigarettes do you smoke on a daily basis? (3) Have you smoked in the past? (4) How many cigarettes per day did you smoke in the past? (5) How many years since you stopped smoking? Based on questions 1 and 3, patients were allotted to 3 study groups: (a) current smokers (n = 2706), (b) smokers in the past (n = 1871), and (c) never smokers (n = 7499).
Comparison of the epidemiologic variables across the 3 groups was performed with (chi-square) Test or Fisher exact test for categorical variables (2 × 2 tables required Continuity Correction) and with Kruskal-Wallis Test for continuous variables. Risk estimates for abnormal cytology (HSIL, LSIL cut-offs), hr-HPV and CIN2+ lesions were first calculated in univariate regression, with crude Odds Ratio (OR) and Confidence Intervals (95% CI). All significant risk factors were entered in a multivariate regression model (together with smoking variables), and the adjusted ORs (95% CI) were calculated for 2 separate outcome variables: hr-HPV infections and CIN2+. In all tests, the values P <0.05 were regarded statistically significant.
Univariate survival (Kaplan-Meier) analysis was used to calculate the cumulative incidence curves (in the whole cohort) for hr-HPV infections in women with negative HPV tests at baseline, and for incident CIN2+ during the FU. The curves were compared using log-rank (Mantel-Cox) statistics. HR for the risk associates were calculated with a multivariate Cox regression model, with 95% CI. All calculations were performed using the R package for statistical computing.24
Altogether, smoking data were available from 12,076 women participating in the study. Except for the number of deliveries (P = 0.094), number of abortions (P = 0.420), and history of previous CIN (P = 0.474), all other variables recorded differed significantly across the 3 groups. The most significant differences were the higher prevalence (21.7%) of hr-HPV infections among current smokers as compared to women who never smoked (16.5%) or those smoking in the past (13.5%). Current smokers also had higher HPV loads (P = 0.0004), higher prevalence of abnormal Pap smear and CIN (P <0.0001). Table 1 also describes time-related characteristics of the smoking habits: on average, current smokers had been smoking for 15.89 ± 10.33 years (M± SD), consuming an average of 11.32 ± 8.47 cigarettes per day. Past smokers recalled of having being smokers for 10.26 ± 8.09 years, but referred having abandoned the smoking habit quite recently (8.09 years ago, on average) in Table 1.
Table 2 lists the ORs for having hr-HPV infection or CIN2+ at baseline. The odds were adjusted with a multivariate regression model, with all the listed variables being entered in the model. Being current smoker was significantly (P <0.010) associated with hr-HPV detection in this model (OR = 1.6; 95% CI = 1.2–2.1). The following factors were also significantly associated with a positive hr-HPV test at baseline: age <35 years (OR = 1.9; 95% CI = 1.6–2.3), being single (OR = 1.6; 95% CI = 1.4–1.9) and having multiple sex partners (OR = 1.8; 95% CI = 1.4–2.2). Two factors were marginally associated with a positive hr-HPV test: use of oral hormonal contraception (OR = 1.2; 95% CI = 1.0–1.4; but P = 0.03) and having being pregnant before (OR = 0.8; 95% CI = 0.6–1.3; P = 0.03). Being a current smoker increased the risk of CIN2+ at baseline (OR = 1.8; 95% CI = 1.3–2.5). Being single was also disclosed as a significant predictor of CIN2 or worse at this baseline assessment (OR = 2.5; 95% CI = 1.1–2.1). Having >5 lifetime sexual partners was marginally associated with CIN2 or worse (OR = 1.5; 95% CI = 1.0–2.2).
Years of smoking among current and past smokers are displayed in relation to the detection of hr-HPV, CIN2+, and Pap smear abnormalities in Table 3. In this analysis, years of being smoker had no statistically significant association with the detection of hr-HPV, CIN2, HSIL, or LSIL at the baseline examination. The only borderline association was disclosed for HSIL Pap and years of smoking among current smokers, women with HSIL Pap having been smokers longer than women with no HSIL (P = 0.074). Within the group of past smokers, the time (months) since the smoking habit had ceased had no relation to the baseline detection of hr-HPV, CIN2, HSIL, or LSIL.
Table 4 summarizes the for incident hr-HPV infection and biopsy-confirmed CIN2+ in multivariate analysis including the smoking status as a cofactor. These calculations are based on a sample of women who attended at least 1 FU visit. Being a current smoker was a significant predictor of incident hr-HPV during the FU (= 1.4; 95% CI 1.0–1.9). Use of oral hormonal contraception was also associated with incident hr-HPV infection (= 1.2; 95% CI 1.2–2.2). For incident CIN2+, baseline hr-HPV+ (= 10.1; 95% CI 1.3–76.5), being a past smoker (= 3.6; 95% CI 1.6–9.8) and current smoker (= 3.6; 95% CI 1.5–8.6), and using oral contraceptives were the significant independent predictors. The incidence rates [per 1000 women months at risk (WMR) [of CIN2+ in the 3 smoking categories (n = 1011) are compared at the bottom of the table. Compared with never smokers (0.75/1000 WMR), the incidence of CIN2+ was significantly higher among past- (3.19/1000 WMR) and current smokers (3.01/1000 WMR) as calculated by the incidence rate ratio test, P = 0.01 and P <0.01, respectively, whereas no difference was found between the past and current smokers.
Univariate (Kaplan-Meier) survival analysis showing the cumulative incidence of CIN2+ among baseline hr-HPV+ women stratified according to their smoking status is shown in Figure 1. Current and past smokers had a significantly increased risk of incident CIN2+. Survival curves deviate from each other starting after the second FU visit, when smokers start rapidly accumulating incident CIN2+ (log-rank; P = 0.0105).
Because of the cross-sectional and prospective cohort design of the LAMS study,17 we were able to assess the effect of smoking on both prevalence of hr-HPV infections and CIN2+, but also on the outcome of hr-HPV infections (persistence, clearance, acquisition) as well as on the development of incident CIN2+ during the prospective FU of over 3 years. The entire cohort of 12,114 women was stratified according to their smoking status as never smokers, past smokers, and current smokers. These 3 groups were significantly different in the majority of their demographic data and epidemiologic variables that are known risk factors of HPV, CIN, and CC. Current smokers had the highest prevalence of hr-HPV infections at baseline, and they also had the highest prevalence of HSIL (LSIL, ASCUS) and biopsy-confirmed CIN lesions. Without further analysis, this would suggest that current smokers have an increased risk of hr-HPV and CIN, if not controlled for obvious confounders.
When this was done using multivariate models with baseline hr-HPV and baseline CIN2+ as outcomes, the results are interesting. Indeed, there were several covariates that were associated with prevalent hr-HPV at least as strong as smoking, e.g., marital status, >5 sexual partners, age <35 years. Importantly, smoking was an independent risk factor of prevalent CIN2+ (OR = 1.8, 95% CI 1.3–2.5). These results are deviate in part from the data reported in the NIS cohort, where smoking was not an independent predictor of CIN2+, although smoking had been an independent predictor of hr-HPV in that study, exactly as in the present cohort.10
Epidemiologic data on the role of smoking as a risk factor of CC and on the acquisition and persistence of HPV infections have emerged only recently.9,10 The early studies examining the development of CIN among smokers were limited by methodologic approach because the analyses were invariably restricted to HPV-positive women.4,25,26 This limitation precluded the assessment of the smoking on HPV acquisition, which has been addressed in more detail only recently. The pooled analysis on HPV Prevalence Surveys, undertaken by IARC, casts further light on the possible mechanisms responsible for the increased prevalence of CIN and CC among smokers.1 In this pooled analysis, there was a consistent association between smoking and HPV infection, and this association seems to be dependent on the smoking intensity: the OR adjusted for several other known epidemiologic cofactors, was 1.2 (95% CI 0.9–1.5) for women who smoked <5 cigarettes per day, but 2.0 (95% CI 1.3–3.1) for those smoking 15 cigarettes on a daily basis.9
The IARC pooled analysis included studies from Vietnam, Thailand, Spain, Mexico, Argentina, Chile, Colombia, and Nigeria, thereby covering a wide spectrum of epidemiologic differences in HPV type and prevalence, with socio-economic variability. In the LAMS study, the cohort is composed predominantly of urban populations from Brazil and Argentina, currently undertaking the demographic transition toward more developed living standards, but still facing economic and health-related constraints. In this predominantly urban population, the OR for current smokers to test hr-HPV positive at baseline (after adjustment for several well-known epidemiologic risk factors) was 1.6 (95% CI 1.2–2.1), which is fairly close to the OR reported in the IARC pooled analysis.9 This similar risk across different geographic regions suggests that the effects of smoking on the acquisition of HPV may not be restricted to HPV type distribution (varying from one region to another), nor be positively or negatively affected by ethnical, nutritional, or behavioral characteristics of the women. Further confirmation is provided by another major cohort study (n = 3187 women) from New Independent States of the former Soviet Union,10 in which current smokers had OR = 1.5 (95% CI 1.1–2.1) for testing hr-HPV positive at baseline. Together with the present results and the IARC survey, these data implicate that cigarette smoking has a universal and independent effect on the risk of contracting hr-HPV.
This possibility that the effect of smoking on the risk for hr-HPV and CIN might be masked by the sexual behavior of the women has been addressed to some extent recently. In their analysis, Vaccarella et al.27 showed that the number of lifetime sexual partners was positively associated to either the prevalence of HPV and the number of cigarettes smoked. In the current study, both factors were entered in the multivariate model to control whether this association is confounded or not. We observed that only the number of sexual partners (but not number of cigarettes) was significantly associated with hr-HPV (but not CIN2+) at baseline. It must be pointed out, however, that in this respect, the statistical approach might be insufficient to overcome the effect of (i) reporting (recall) bias. There is room for this type of systematic bias because even though subjects were unaware of their disease status at the time of interview (baseline), smoking is a widely recognized as a health detractor and is therefore always subject to underreporting. Moreover, memory certainly plays a role: on average, in our study, women had quit smoking 7.4 years before being interviewed. Concerning the sexual behavior, the study sample is derived from mostly a catholic society, and multiple partners are regarded by many as an unfit behavior. For instance, if women underreport their number of lifetime partners (thereby masking part of their risk of having HPV infection), the regression coefficients generated by the adjustment with the number of cigarettes will be compromised. The reverse is, of course, also true. It is known from several other epidemiologic studies that reporting and recall bias are particularly strong concerning sexual and smoking habits.1,4,9 This makes it important to regard the relationship between the number of cigarettes and number of sexual partners with caution. Equally important is to consider the confounding because of hr-HPV in all studies assessing the role of smoking as a risk factor for CIN and CC.1
We also assessed whether there was a dose-response relationship of smoking to detection of hr-HPV, Pap smear abnormalities, or biopsy-confirmed CIN lesions at baseline, by measuring the years being smoker (current or past). There was no difference in the years of smoking as related to detection of hr-HPV+, CIN2+, HSIL, or LSIL abnormalities. In the current study, we were unable to adequately measure the influence (if any) of time passed since the cessation of smoking on the detection of hr-HPV, because of the potential for reporting bias and the relatively high mean time period since cessation of smoking (7.4 years). However, it sounds sensible to infer that there must be a time-related decline in the risk of having an HPV infection after ceasing smoking, but a large cohort is likely to be necessary to evaluate this trend and determine the time period after which the smoking effect on the prevalence of HPV drops to nonsmoker level. The major studies on the subject are cross-sectional and therefore do not provide this information.9
It was of importance to assess whether smoking status is an independent risk factor of incident hr-HPV infections or incident CIN2+ during the FU. Indeed, smoking status (being a current smoker, but not being past smokers) did increase the risk of incident hr-HPV, albeit not dramatically (OR = 1.4). On the other hand, smoking increased the risk of incident CIN2+, in alignment with the fact that it was an independent predictor of baseline CIN2+. However, the baseline hr-HPV was a far more powerful predictor of incident CIN2+ than being past or current smoker. When calculated per 1000 woman months at risk, the incidence rate of CIN2+ among past and current smokers is 3-fold higher than among never smokers. This might implicate that smoke components speed up the transformation process of epithelial cells, once infected by oncogenic HPV that remains persistent.
In the present analysis, different predictors were disclosed for baseline hr-HPV and baseline CIN2+ and the same was true with incident hr-HPV and incident CIN2+. The number of incident CIN2+ cases (n = 30) was not particularly large, however, and one needs to consider the possibility that the different sets of predictors for hr-HPV and CIN2+ could be related to the lower prevalence of CIN2+ as compared to that of hr-HPV. This has been encountered in other population-based studies, suggesting that the much lower prevalence of CIN and CC makes hr-HPV more amenable to epidemiologic risk assessments. Indeed, this has prompted most studies to limit their analyses to HPV-positive women,1,4,25 which hampers controlling for the confounding effect of HPV. Similarly, studies analyzing the effect of smoking in speeding up the progression of low-grade CIN to high-grade CIN or cancer are few; actually, there is only one consistent study on this subject, and their results have never been reproduced.28
The present results implicate that being current smoker increases the prevalence of baseline hr-HPV infections. Past, current and never smokers differ in the majority of key epidemiologic risk factors of HPV, CIN, and CC, most notably those associated with risk sexual behavior. Together with several of these risk factors, current smoking is an independent predictor of baseline hr-HPV infections and for baseline CIN2+. It is known that persistent hr-HPV infections increase the risk of cervical abnormalities regardless of the smoking status, and hr-HPV is also the single most powerful predictor of incident CIN2+. Being past or current smoker are the other 2 independent predictors of incident CIN2+, the incidence among these women being 3-fold higher than among never smokers. The most feasible explanation of these data implicates that smoking-related high-risk sexual behavior increases the risk of contracting hr-HPV infection, making current smokers with persistent hr-HPV+ true high-risk patients for cervical disease. Of note, it seems after analyzing our data, and contrasting our results to the most recent and solid epidemiologic studies on the subject, that besides increasing the risk of acquiring a hr-HPV infection, smoking also modifies the effect of a persistent infection, by further increasing the risk of CIN development. Interestingly, it may also be derived from our data that this increased risk associated with cigarette smoking persists several years after cessation of smoking, for the for incident CIN2+ in past and current smokers were very similar.
1. Plummer M, Herrero R, Franceschi S, et al. Smoking and cervical cancer: pooled analysis of the IARC multi-centric case-control study. Cancer Causes Control 2003; 14:805–814.
2. Procopzyk B, Cox JE, Hoffman D, et al. Identification of tobacco-specific carcinogen in the cervical mucus of smokers and nonsmokers. J Natl Cancer Inst 1997; 89:868–873.
3. Assieh A, Melikian A, Sunb Peng, et al. Identification of benzo[a]pyrene metabolites in cervical mucus and DNA adducts in cervical tissues in humans by gas chromatography-mass spectrometry. Cancer Lett 1999; 146:127–134.
4. Tolstrup J, Munk C, Thomsen BL, et al. The role of smoking and alcohol intake in the development of high-grade squamous intraepithelial lesions among high-risk HPV-positive women. Acta Obstet Gynecol Scand 2006; 85:1114–1119.
5. Poppe WA, Ide PS, Drijkoningen MP, et al. Tobacco smoking impairs the local immunossurveillance in the uterine cervix. Gynecol Obstet Invest 1995; 39:34–38.
6. Poppe WA, Peters R, Drijkoningen MP, et al. Cervical cotinine and macrophage-Langerhans cell density in the normal uterine cervix. Gynecol Obstet Invest 1996; 41:253–259.
7. Ferson M, Edwards A, Lind A, et al. Low natural killer-cell activity and immunoglobulin levels associated with smoking in human subjects. Int J Cancer 1979; 23:603–609.
8. Vaccarella S, Herrero R, Snidijers PJ, et al. Smoking and papillomavirus infection: pooled analysis of the International Agency for Research on Cancer HPV Cancer Prevalence Surveys. Int J Epidemiol 2008:1–11.
9. Syrjänen K, Shabalova I, Petrovichev N, et al. Smoking is an independent risk factor for oncogenic human papillomavirus (HPV) infections but not for high-grade CIN. Eur J Epidemiol 2007; 22:723–735.
10. Herrero R, Brinton LA, Reeves WC, et al. Invasive cervical cancer and smoking in Latin America. J Natl Cencer Inst 1989; 8:205–211.
11. Trottier H, Franco EL. The epidemiology of genital human papillomavirus infection. Vaccine 2006; (suppl 1):S1/4–S1/15.
12. Castellsague X, Munoz N. Chapter 3: Cofactors in human papillomavirus carcinogenesis–role of parity, oral contraceptives, and tobacco smoking. J Natl Cancer Inst Monogr 2003; 31:20–28.
13. Campo S, ed. Papillomavirus Research: From Natural History to Vaccines and Beyond. Norwich, UK: Caister Academic Press, 2006:1–424.
14. Cavalcanti SM, Zardo LG, Passos MR, et al. Epidemiological aspects of human papillomavirus infection and cervical cancer in Brazil. J Infect 2000; 40:80–87.
15. Reeves WC, Rawls WE, Brinton LA. Epidemiology of genital papillomaviruses and cervical cancer. Rev Infect Dis 1989; 11:426–439.
16. Bosch X, de Sanjosé S. The epidemiology of human papillomavirus infection and cervical cancer. Dis Markers 2007; 23:213–227. Review.
17. Syrjänen K, Naud P, Derchain S. Comparing PAP smear cytology, aided visual inspection, screening colposcopy, cervicography, and HPV testing as optional screening tools in Latin America. Study design and baseline data of the LAMS study. Anticancer Res 2005; 25:3469–3480.
18. Longatto-Filho A, Erzen M, Branca M, et al. Human papillomavirus testing as an optional screening tool in low-resource settings of Latin America: experience from the Latin Am Screening study. Int J Gynecol Cancer 2006;16: 955–962.
19. Sarian LO, Derchain SF, Naud P, et al. Evaluation of visual inspection with acetic acid (VIA), Lugol′s Iodine (VILI), cervical cytology, and HPV testing as cervical screening tools in Latin America This report refers to partial results from the LAMS study. J Med Screen 2005; 12:142–149.
21. Argentina, Ministry of Health. Estadísticas vitales. Información básica año 2005. Buenos Aires: Dirección de Estadísticas e formación de Salud, 2006.
22. Solomon D, Davey D, Kurman R, et al. Forum Group Members; Bethesda 2001 Workshop. The 2001 Bethesda System: terminology for reporting results of cervical cytology. JAMA 2002; 287:2114–2119.
23. Lörincz AT, Castle PE, Sherman ME, et al. Viral load of human papillomavirus and risk of CIN 3 or cervical cancer. Lancet 2002; 360:288–289.
24. R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3–900051-07–0, Available at: http://www.R-Project.org
25. McIntyre-Seltman K, Castle PE, Guido R, et al. Smoking is a risk factor for cervical intraepithelial neoplasia grade 3 among oncogenic human papillomavirus DNA-positive women with equivocal or mildly abnormal cytology. Cancer Epidemiol Biomarkers Prev 2005; 14:1165–1170.
26. Matos A, Moutinho J, Pinto D, et al. The influence of smoking and other cofactors on the time to onset to cervical cancer in a southern European population. Eur J Cancer Prev 2005; 14:485–491.
27. Vaccarella S, Franceschi S, Herrero R, et al. Sexual behaviour, condom use, and human papillomavirus infection: pooled analysis of the IARC human papillomavirus prevalence surveys. Cancer Epidemiol Biomarkers Prev 2006; 15:326–333.
28. Ho GY, Kadish AS, Burk RD, et al. HPV 16 and cigarette smoking as risk factors for high-grade cervical intra-epithelial neoplasia. Int J Cancer 1998; 78:281–285.