Persistent infection of the cervical epithelium with high-risk types of HPV plays a major role in the development, maintenance and progression of CIN.1, 2 HPV has been found in 99.7% of cervical cancers worldwide.3
Sexual intercourse is the primary mode for transmission or acquisition of HPV, and prevalence is related to many determinants involving sexual behavior characteristics.4, 5, 6 Case-control studies of male sexual partners of women with cervical cancer have shown a relation between male sexual behavior and cervical cancer.7, 8 The presence of HPV DNA in penile swabs conveyed a 5-fold risk of cervical cancer to their wives.5 Studies on the risk factors of CIN supported a protective effect of condom use in women, emphasizing the venereal nature.9, 10, 11 Coker et al.12 found that among high-risk HPV-positive women longer-duration barrier method use was associated with a reduced risk of CIN. Manhart and Koutsky13 reported in a meta-analysis that data on effects of condom use preventing HPV infection and HPV-related conditions were inconsistent. However, none of these studies was conducted explicitly to assess the effectiveness of condoms at preventing HPV infection and HPV-related conditions.
Previously, we showed that flat penile lesions are associated with the presence of HPV and that regression is dependent on the presence of HPV14 (see also Bleeker and Hogewoning, 2003, accompanying paper). Continuous transmission of HPV in sexual partners having HPV-associated genital lesions might reduce the chance of viral clearance. In a randomized clinical trial, we investigated the influence of condom use on the clinical course of HPV-associated genital lesions and HPV infection in sexual partners. In the present report, we evaluate the effects of condom use on CIN regression and HPV clearance.
Women referred to the colposcopy clinic of the Albert Schweitzer Hospital, Dordrecht, the Netherlands, from January 1995 to June 2002 were asked to bring in their male sexual partner. Eligible were women with an abnormal cervical smear (mild dysplasia or worse) and/or cCIN and/or hCIN. Women were evaluated for CIN by colposcopy and by histologic evaluation of cervical biopsy specimens at baseline. Cervical smears were taken for cytology and HPV testing by PCR. The outline of the study was verbally explained and written information given to couples who had no other sexual partners. Willing couples returned within 2 weeks to discuss the study protocol in detail, and additional questions were answered. Both the period of condom use, i.e., only during the study and for at least 3 months, and the instructions on condom use, i.e., during genital–genital contact, were discussed with participants. Latex condoms (Durex fetherlite; Netherlands LRC, Leerdam, the Netherlands) with basic lubricant (without spermicidal and/or virus-inactivating substances) were given free, to increase study compliance. Exclusion criteria were surgical treatment of the cervical lesion and regular condom use at baseline, i.e., using condoms for birth control. CIN2 and CIN3 lesions over ≥2 cervical quadrants were treated according to standard protocols. Irrespective of CIN grade at baseline, couples were randomized for condom use in blocks of 2 if conservative management of the CIN lesion was decided. Per block of 2 couples, the sequence of condom and noncondom allocation was based on a table of random numbers. Colposcopic and cervical smears were repeated after 3, 6 and 12 months and subsequently every 6 months. Condom use was verbally verified at each visit by asking the frequency of condom use failure. To minimize interference with the natural course of CIN, a biopsy specimen was taken during follow-up only when progression of CIN lesions was suspected on the basis of findings at colposcopy and/or cytology to justify the LLETZ procedure. Follow-up was ended if women were treated by LLETZ. Couples were asked to complete a questionnaire on lifestyle habits, including sexual behavior. Independently of each other and in separate rooms, questionnaires were completed. This questionnaire was introduced in 1999. The study protocol was approved by the ethics review board of the hospital (protocol CGE/95/238), and the couples signed informed consent before enrollment.
Before colposcopic evaluation, cervical smears were taken for cytologic and HPV testing. For HPV testing, a Cervex brush (Rovers Medical Devices, Oss, the Netherlands) was used to collect cells from the cervix. The brush was placed in 5 ml of PBS solution with 0.001% thimerosal (Merthiolate) (BDH, Poole, UK), and samples were sent to the laboratory and stored at 4°C until testing.15 Standard colposcopy was performed after application of 3% acetic acid solution and blinded of data from previous visits. Colposcopic characteristics such as acetowhiteness, mosaic, punctation, leukoplakia and atypical vessels were assessed and classified according to the international terminology.16 Minor changes, e.g., thin acetowhite epithelium, fine mosaic, fine punctation and thin leukoplakia, were noted as cCIN1 and major changes, e.g., dense acetowhite epithelium, coarse mosaic, coarse punctation and atypical vessels, as cCIN2 or cCIN3. Other diagnoses, e.g., squamous metaplasia, inflammatory changes, polyps or atrophic changes, were also noted. An overview of the colposcopic impression was drawn and categorized on the basis of the most atypical site: no cCIN, cCIN1, cCIN2 and cCIN3. Subsequently, colposcopic findings were documented by photographs. Photographs were reviewed by an experienced colposcopist blinded of any clinical data. In case of discrepancies (<10%), a consensus diagnosis was made. Lesions were graded before linking these data on condom use.
Processing and testing of cervical scrapings for HPV analysis were performed as described previously at the Department of Pathology, VU University Medical Center.15 Samples were centrifuged and cell pellets resuspended in 1,000 μl TRIS buffer. In PCR tests, aliquots of 10 μl were used. The quality of the specimen was tested by β-globin PCR, and 14 high-risk (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68) and 6 low-risk (6, 11, 40, 42, 43 and 44) HPV genotypes were identified by the clinically validated HPV GP5+/6+ PCR EIA. Samples negative on both the β-globin and HPV PCR tests were considered inadequate and excluded.
Definitions of CIN regression and HPV clearance
Regression of the cCIN lesion was defined as 2 consecutive “no cCIN” diagnoses at colposcopy. Women were considered to have cleared the HPV infection when 2 consecutive negative HPV tests were obtained.
Analyses for condom use were by intention to treat. Differences between the condom and the noncondom groups were assessed by χ2 tests or independent t-tests. In survival analysis, the time point of CIN regression and HPV clearance was taken as the midpoint between the last positive and the first negative results. Kaplan-Meier curves, for regression of cCIN lesions and clearance of HPV, were compared by means of 2-sided log-rank tests. Cox regression analyses were performed to calculate the HR and 95% CI, adjusted for condom use, HPV status, histologic CIN grade (either ≤hCIN1 or ≥hCIN2) and age (tertiles). HPV status was defined by the presence of HPV at baseline. HPV-positive women were stratified into HPV-16, other high-risk HPV or only low-risk HPV to assess whether CIN regression and/or HPV clearance was related to HPV type. Interaction terms were added to the Cox model to assess whether the effect of condom use was related to HPV status or the hCIN grade of the lesions. Within the condom group, whether the effect was dependent on the duration of condom use was assessed. Therefore, Kaplan-Meier analyses were performed on women who still had colposcopic CIN or HPV present at 6 months and were stratified for condom use at this time point. Statistical significance was set at the 0.05 level. For statistical analyses, SPSS (Chicago, IL) version 9.0 software was used.
Characteristics of the study population
The trial profile is given in Figure 1. Of 255 couples selected for the presence of CIN in the female partner, 17 were not willing to participate, 78 women were treated by LETTZ and 12 women were excluded due to regular condom use, leaving 148 women randomized (condom group n = 72 and noncondom group n = 76). Women were excluded because of no colposcopic CIN (n = 9), no colposcopic diagnosis (n = 4) or loss to follow-up (n = 10). The remaining 125 women were followed: 64 in the condom group and 61 in the noncondom group. Median follow-up time was 16.4 (range 3.0–85.4) months in the condom group and 12.8 (range 3.1–63.0) months in the noncondom group. Median duration of condom use was 6.0 (range 3.0–53.7) months, and failure to use condoms was reported by 7 women with a median of 2 times during the period of condom use (range 1–5). Characteristics of both groups are shown in Table I. At baseline, colposcopic diagnoses were, in the condom group, cCIN1 (n = 34), cCIN2 (n = 29) and cCIN3 (n = 1) and, in the noncondom group, cCIN1 (n = 36), cCIN2 (n = 24) and cCIN3 (n = 1). Representative histologic diagnoses of biopsy specimens were not available in 6 (9%) women of the condom group and 4 (7%) women of the noncondom group. Eight of them had cCIN1 lesions and 2 had a cCIN2 lesion. The remaining biopsies showed, in the condom group, no hCIN in 8 (14%), hCIN1 in 18 (31%), hCIN2 in 31 (53%) and hCIN3 in 1 (2%) and, in the noncondom group, no hCIN in 11 (19%), hCIN1 in 25 (44%), hCIN2 in 19 (33%) and hCIN3 in 2 (4%). Histologic evaluation of cCIN1 lesions (n = 62) showed no hCIN in 13 (21%), hCIN1 in 31 (50%), hCIN2 in 17 (27%) and hCIN3 in 1 (2%) biopsy specimens. Histologic evaluation of cCIN2 lesions (n = 51) showed no hCIN in 6 (12%), hCIN1 in 11 (22%), hCIN2 in 32 (63%) and hCIN3 in 2 (4%) biopsy specimens. The 2 women with cCIN3 were included in the study because histologic specimens showed hCIN1 and hCIN2, respectively. Three women with hCIN3 lesions were included because their lesions were small (<2 cervical quadrants). At baseline, 2/64 (3%) cervical scrapes of the condom group and 2/61 (3%) cervical scrapes of the noncondom group were inadequate for HPV testing, i.e., were negative in both the β-globin and HPV PCR tests. Of the remaining, 101 (84%) scrapes were positive for HPV: 54 (87%) of the condom group and 47 (80%) of the noncondom group. Of the HPV-positive women, HPV-16 was found in 27 (50%) of the condom group and in 22 (47%) of the noncondom group and other high-risk HPV (not HPV-16) was found in 26 (48%) and 22 (47%), respectively. Only low-risk HPV was found in 1 (2%) HPV-positive woman of the condom group and in 3 (6%) HPV-positive women of the noncondom group. Questionnaires were obtained from 51/125 (41%) women, 26/64 (41%) of the condom group and 25/61 (41%) of the noncondom group. No statistical differences were found between the condom and noncondom groups for smoking habits, age at first sexual intercourse, number of sexual partners, number of noncondom sexual partners, history of STDs, other sexual partners last year, duration of the current relation, frequency of sexual intercourse and contraceptive use (Table I).
Table I. Characteristics of the Study Population
Condom group Number (%)
Noncondom group Number (%)
Not be determined due to samples being inadequate for HPV testing. CIN grade (at colposcopy and histology) refers to diagnoses made at baseline.
Age at first sexual intercourse (years, mean, range)
History of STD
Lifetime sexual partners
Overall (mean, range)
Noncondom partners (mean, range)
Other sexual partner last year
Type of relationship
Duration relation in years (mean/range)
Frequency sexual Intercourse
Frequency per month (mean, range)
Frequency last month (mean, range)
Regression of CIN
The 2-year cumulative regression rate of cCIN was 53% in the condom group (n = 57) vs. 35% in the noncondom group (n = 51, log rank p = 0.03; HR = 3.1, 95% CI 1.4–7.1) (Fig. 1, Table II). Since regression was defined as 2 consecutive “no cCIN” diagnoses, 17 women were excluded due to lack of follow-up data after they showed a normal colposcopy once. Age was not related to regression of CIN in our study population. Less regression of cCIN was seen in HPV-positive women (HR = 0.2, 95% CI 0.1–0.5) and in women with ≥hCIN2 at baseline (HR = 0.3, 95% CI 0.1–0.7). In HPV-positive women, regression of CIN was not related to HPV type when stratified into HPV-16, other high-risk HPV or only low-risk HPV. The effect of condom use was not related to HPV status (p = 0.4) or hCIN grade at baseline (p = 0.8). Since the outcome variable for regression of cCIN was defined by colposcopy, analyses were also performed with inclusion of the cCIN grade instead of the hCIN grade defined at baseline. Less regression of cCIN was observed in women with ≥cCIN2 compared to women with cCIN1 at baseline (HR = 0.3, 95% CI 0.1–0.6, p = 0.003). In women who still had cCIN lesions at 6 months, no different regression rates were found during follow-up to 24 months in those who used condoms longer than 6 months (n = 14) compared to those who used condoms for less than 6 months (n = 15; log-rank p = 0.6).
Table II. Prognostic Factors for CIN Regression and HPV Clearance
HR (95% CI)
HR (95% CI)
Adjustments were made for all other factors in the table. NA, not applicable.
Could not be calculated since no clearance was observed in this category.
The 2-year cumulative rate of HPV clearance was 23% in the condom group (n = 53) vs. 4% in the noncondom group (n = 38, log rank p = 0.02; HR = 12.1, 95% CI 1.5–97.2) (Fig. 2, Table II). Of the 125 women, 34 were excluded in the HPV clearance analyses since they were HPV-negative at baseline (n = 20), they had cervical scrapes that were inadequate for HPV testing at baseline and/or during follow-up (n = 7) or there was no available follow-up data on HPV after they were HPV-negative once (n = 7). No relation was found between clearance of HPV and age. Clearance of HPV was related to hCIN grade; i.e., less HPV clearance was found in women with ≥hCIN2 compared to women with ≤hCIN1 at baseline (HR = 0.2, 95% CI 0.04–0.8). However, HPV clearance was not related to cCIN grade at baseline (HR = 0.4, 95% CI 0.1–1.7). In HPV-positive women, HPV clearance was not related to HPV type when stratified into HPV-16, other high-risk HPV or only low-risk HPV. hCIN grade at baseline histology was not related to the effect of condom use on HPV clearance (p = 0.9). In women who still were HPV-positive after 6 months, no different clearance rates were found in those who used condoms longer than 6 months (n = 17) compared to those who used condoms for less than 6 months (n = 14, log-rank p = 0.6).
In a randomized clinical study, performed in a non-STD clinic, 125 women with cCIN were followed by colposcopy and HPV testing. We found that condom use promotes regression of cCIN and clearance of HPV. The 2-year cumulative regression rate for cCIN was 53% in the condom group vs. 35% in the noncondom group (p = 0.03), and 2-year cumulative rates of HPV clearance were 23% vs. 4%, respectively (p = 0.02). These beneficial effects of condom usage were independent of hCIN grade.
One limitation of our study was that regression of CIN was defined by colposcopy and not verified by histology. Only baseline biopsy specimens were taken for histopathologic diagnoses since repeated biopsies would interfere too much with the natural course of the CIN lesion. To minimize this potential weakness, regression of cCIN was defined as disappearance of the total cCIN lesion and events were scored only in women with 2 consecutive “no cCIN” diagnoses. Our findings on concordance between the colposcopic and histopathologic diagnoses are in agreement with the positive predictive rates reported by Hopman et al.17 They reported positive predictive rates of the colposcopic impression to be 62% for no CIN, 43% for CIN1, 59% for CIN2 and 78% for CIN3.
Women with a negative HPV test at baseline showed more regression of cCIN compared to women who were HPV-positive. This is in agreement with findings on the relation between HPV at baseline and cytologic regression in women with abnormal smears.18, 19 In analogy, a shorter regression time of flat penile lesions was found in the absence of HPV, as reported in the accompanying paper on male sexual partners of women with CIN.
Another limitation of our study is that the number of women included is relatively small and women were not randomized by CIN grade. Compared to the noncondom group, a relatively high number of women with high-grade CIN were included in the condom group, which might result in underestimation of the effect of condoms as the grade of the CIN lesion is an important prognostic factor for the regression of CIN.20 Analyses on the effects of condoms, however, were adjusted for the most important risk factors (i.e., CIN grade and HPV status).
As characteristics of sexual behavior were collected in only 41% of women, adjustments for these factors could not be done properly. However, no differences on sex-related factors between groups were found, as one would expect given the randomized design of our study. Moreover, as data collection was time-selected, the women who filled out the questionnaires comprised a representative sample of the whole population. Therefore, we conclude that these factors did not influence our findings on condom use. Current sex-related factors, especially frequency of sexual intercourse, might be relevant to assess the effect of condom use, though no differences were found among the couples analyzed. However, failure to use condoms, as was found in the condom branch, might result in a less favorable effect of condom use on CIN regression and HPV clearance. Beneficial effects of condom use were found after a relatively short duration of use (median 6.0 months). We used a minimal duration of 3 months' condom use before randomization, though the maximum duration of condom use was not defined at baseline. An attractive, though hypothetical, explanation for our findings might be that condoms interfere with continuous transmission of shed HPV between sexual partners. As a consequence, in individuals who benefit from this interference, the viral load would remain under a certain critical threshold, allowing effective virus elimination by the immune system, thereby accelerating regression of the lesion. Noninterference in these individuals would, however, result in viral load levels that are sufficiently high to safeguard viral persistence. That viral load may be an important determinant in the observed phenomena is substantiated by another study, which demonstrated that women with abnormal cytology and relatively low HPV-16 loads had an increased chance of viral clearance and cytologic regression.21 Since the viral load appears to be proportional to the size and severity of lesions, a more marked effect of condom use is expected for smaller, less severe lesions that consequently have lower viral loads.22 Indeed, the effect was much more pronounced for penile than for cervical lesions. The former are generally smaller and less severe and have viral loads that are 5- to 10-fold lower than in their cervical counterparts (data not shown). In contrast, individuals with high viral loads are unlikely to respond to condom use since maintenance of the lesion would not depend on exposure to virus of the partner. Because women with large and severe lesions (≥cCIN2 on >2 quadrants) were treated for ethical reasons, only women with relatively small lesions, having most likely lower viral load values, were selected for study. Therefore, an attractive subject of future studies may be to determine the HPV load by quantitative PCR methods, to further understand the mechanisms underlying the effect of condom use on male and female HPV-associated lesions.
In conclusion, although this is a relative small study, it demonstrates that intervention with condom use for at least 3 months can promote CIN regression and HPV clearance. Our results imply that condom use should be considered as an alternative strategy in the management of women with CIN and that the need for aggressive treatments may be reduced. Condom use should also be advised in pregnant women with CIN, in whom treatment is difficult. Whether condom use might be helpful in decreasing the number of latent HPV infections in HPV-positive cytologically normal women is currently under investigation.
We thank Mr. R.E. van Andel and Mrs. M.C.G.T. Verkuyten for technical assistance on HPV testing. We thank Dr. N. Muñoz of the IARC for critical reading of the manuscript and critical remarks.