The association between cervical neoplasia and human immunodeficiency virus (HIV) infection is well established, and HIV–seropositive women are probably the group at greatest risk for cervical intraepithelial neoplasia (CIN). Prevalence of CIN in that subgroup has been estimated to be between 20% and 50%, significantly higher than in patients immunosuppressed by other means.1–5 In addition, HIV-infected women with CIN might have more aggressive features, including higher grade lesions, more extensive cervical and endocervical involvement, and multisite lower-genital-tract involvement with human papillomavirus (HPV)-associated lesions.6–8 Most important, HIV-seropositive women with CIN have high rates of persistent and recurrent disease after standard excisional and ablative therapy, which is usually successful in immunocompetent women. Recurrent cervical disease puts HIV-infected women at continued risk for progressive neoplasia, invasive cancer, and significant morbidity from repeated surgeries. Topical 5-fluorouracil (5-FU) cream has effectively treated warts and intraepithelial neoplasia of the skin, and treatment with intravaginal 5-FU cream has been considerably successful in immunosuppressed women with lower genital tract neoplasia.9,10 Side effects from local inflammation have been of concern, but prophylactic maintenance therapy with lower doses at extended intervals has been well tolerated. Local adjunctive maintenance with topical chemotherapeutic agents potentially could interrupt the cervical neoplastic disease process.
AIDS Clinical Trials Group 200 was a randomized, two-armed, phase III, unmasked, multicenter, outpatient trial designed to compare effectiveness of topical vaginal 5-FU maintenance therapy with that of observation after standard surgery for high-grade cervical dysplasia in HIV-infected women. The primary objective of the study was to determine the efficacy of intravaginal 5-FU therapy as prophylaxis for recurrent CIN in this specific population, with the time at which CIN recurred the primary end point. Secondary objectives included determining the association of CD4 and CD8 counts with time to recurrence of cervical dysplasia and assessment of adverse effects and toxicity of 5-FU in HIV-infected women at the study dosage.
Materials and Methods
Human immunodeficiency virus–seropositive women who received standard ablative therapy for CIN 2–3 were randomized 1:1 by computer to topical vaginal 5% 5-FU cream (2 g self-administered biweekly for 6 months [12 applications]) or observation only. Stratification was by entry CD4 lymphocyte count (less than 200 cells/mm3, 200–500 cells/mm3, and more than 500 cells/mm3). Included were nonpregnant women more than 13 years old who were able to give informed consent and willing to use contraception while on study, with HIV seropositivity confirmed by federally licensed enzyme-linked immunosorbent assay and confirmed by Western blot analysis; histories of CIN 2 or 3 treated by laser therapy, cryotherapy, loop excision, or cone biopsy within the previous 12 weeks; Karnofsky performance status above 60; and histologic diagnoses of CIN 2–3 confirmed by the designated AIDS Clinical Trials Group central pathologist (M. Merino) before enrollment and randomization. Subjects with histories of vaginal or vulvar intraepithelial neoplasia or warts were eligible if lesions were biopsied to prove absence of dysplasia. Exclusion criteria included untreated or persistent vaginal or vulvar dysplasia; classic indication, after colposcopy and biopsy, for diagnostic cone biopsy, when such biopsy not done; cone biopsy specimens with positive margins, CIN 1; microinvasion; adenocarcinoma in situ; history of malignancy that required cytotoxic chemotherapy within the previous 3 months; active genital ulcerative disease; use of 5-FU (systemic or topical) within 3 months before enrollment; history of allergic reaction or severe hypersensitivity to 5-FU; and history of total hysterectomy.
Topical 5% 5-FU cream was supplied in 25-g tubes free of charge (Hoffman-LaRoche Pharmaceutical Co., Nutley, NJ). Beginning no later than 1 week after randomization, subjects self-administered 2 g-doses with vaginal applicators at bedtime once every 2 weeks. Women were told to bring in all tubes of cream to document compliance at each follow-up visit. Application during menses or times of other bleeding was to be avoided, and if menses was delayed or missed and pregnancy suspected, treatment was withheld until negative pregnancy test results had been obtained. Scheduled use of 5-FU was postponed for 3 days after colposcopically directed biopsy to allow time for adequate healing, and sexual intercourse was avoided for 48 hours after each application. Permissible concurrent medications or therapy for subjects on study included antiretrovirals, protease inhibitors, immunomodulators, prophylaxis or treatment for opportunistic infections, vaginal antifungal agents or other indicated medication for vaginitis (except on days of 5-FU application), all contraceptives, and acyclovir. Nonpermissible drugs included cytotoxic chemotherapeutic agents for malignancy and high-dose steroids (10 mg of prednisone or its steroid equivalent per day).
Cervical evaluations during treatment and the time of observation and posttreatment consisted of pelvic examinations with cytologic assessment, colposcopy, and biopsy if indicated by colposcopic examination at 3, 6, 10, 14, and 18 months. If cytologic findings were consistent with dysplasia and concurrent biopsy specimens were not obtained or were considered normal, repeat colposcopy with biopsies was required within 1 month. Colposcopy was not done during menses or for 3 days after 5-FU application. Absence of clinical evidence of CIN after standard ablative or excisional therapy was confirmed by colposcopy at baseline. Recurrence was defined as histologic evidence of any grade of cervical dysplasia (CIN 1, 2, or 3) in biopsy specimens obtained any time during the evaluation schedule. The central pathologist, whose interpretation was binding in cases of discrepancy, was unaware of subject treatment status.
Toxicity evaluations were required by pelvic examination six times over the 18-month follow-up. A four-grade toxicity scale was devised to evaluate potential local signs and symptoms of inflammation, such as itching, irritation, burning, ulceration, and discharge due to 5-FU use. Confirmation by examination was required for any subject who reported severe symptoms, and the next application of 5-FU was delayed until toxicity resolved. No treatment delays were required for subjects with mild or moderate toxicity, although any scheduled application of 5-FU could be delayed at the discretion of the site physician if the subject had not recovered from a previous episode of toxicity. Treatment was to be discontinued if subjects had recurrent CIN or became pregnant.
The study was designed to detect a reduction in the yearly rate of CIN recurrence from 40% per year among women treated with ablation alone (observation) to 20% per year among subjects receiving 5-FU therapy. Randomization was done with permuted blocks with a block size of 4 within strata defined by baseline CD4 count (less than 200, 200–500, and more than 500 cells/mm3). Given the fixed 18-month follow-up and a 17% rate of loss to follow-up, we planned to enroll 158 subjects, to provide 80% power with a two-sided .05-level log-rank test to detect the prescribed reduction in yearly rate of recurrence. However, the rate of accrual was less than expected and the study was terminated prematurely with a total of 101 subjects. Only 25 subjects were entered with CD4 counts above 500 cells/mm3, so final analysis was based on randomization to only two CD4 count groups: counts less than 200 cells/mm3 and counts of 200 or more cells/mm3.
In the original study design, time to recurrence of dysplasia was to be analyzed using the log-rank test. When examinations are frequent and sample sizes are large, the assumptions of standard tests (log-rank and proportional hazards) can be violated without any impact on the estimates and tests. However, many women on both arms did not undergo planned colposcopic evaluations at the scheduled frequency, resulting in long intervals between examinations. Thus, methods developed for analyzing interval-censored data became appropriate for evaluating differences in distributions of time to recurrence. Lindsey and Ryan11 presented a comprehensive review of analysis methods for interval-censored data. A Weibull parametric model was used for recurrence times, and an accelerated failure-time model was fit in SAS PROC LIFEREG (SAS Institute Inc., Cary, NC). The Weibull model also followed the Kaplan-Meier estimates closely when recurrences were assumed to occur at the midpoint between the last date a subject was known to be recurrence free and the date of documentation of a recurrence. Thus, statistical test results and plots of time to recurrence are presented by both methods of analysis of interval-censored data. Tests for time to recurrence were stratified by the two CD4 count groups as indicated previously, and P values for association of covariates with failure times were calculated with a Wald test done in SAS PROC LIFE-REG for the Weibull model and in SAS PROC PHREG with recurrence set as the midpoint between positive biopsy findings and last examination with negative findings. Standard log-rank tests and Kaplan-Meier estimates were used for other analyses of time (eg, time to study discontinuation). Associations between dichotomous variables were evaluated using the Fisher exact test, and an exact Wilcoxon mid-rank–sum test was used to compare ordered categoric variables. No efficacy interim analyses were done for the purpose of stopping the study, so all P values were evaluated with .05 two-sided tests. All analyses were conducted on an intent-to-treat basis, and all events and censoring times were calculated from the date of randomization.
Between May 1993 and January 1997, 101 women from 19 AIDS Clinical Trial Group sites were enrolled; 50 were assigned to adjunctive 5-FU therapy and 51 to observation. Seven sites enrolled only one or two subjects, whereas two sites (State University of New York–Health Science Center at Brooklyn and City Hospital at San Juan) together enrolled 41% of all subjects. All subjects were eligible for analysis. Protocol completion was defined as reaching a clinical end point or having full follow-up based on the subject's entry date; there was no significant difference between the two groups in terms of study completion. Among the 101 women who entered the study, six died while on study, six were lost to follow-up, and three refused further contact; there was no significant difference between the two groups in this regard.
Distribution of baseline characteristics of subjects by treatment group is shown in Table 1. The median age of subjects was 32 years. Most were relatively healthy (88% had Karnofsky scores of 90 or above) but immune deficient (median CD4 lymphocyte count 326 cells/mm3). There was no significant difference between groups in baseline characteristics or use of antiretroviral drugs (26 in the 5-FU treatment group and 25 in the observation group). Fifty subjects were antiretroviral naive, and the median number of weeks of use among the 51 with antiretroviral experience was 40 weeks. Of 51 receiving antiretroviral therapy, 39 were receiving monotherapy and 12 were receiving two-drug therapy. Only four subjects were receiving treatment with protease inhibitors. Fifty-nine had not received any prophylactic antibiotics against opportunistic infections, whereas the remaining 42 used opportunistic infection prophylaxis for a median of 66 weeks. Sixty-one percent of subjects reported having only male sexual partners, 2% reported having only female partners, 26% reported no current sexual activity, and data were not available for the remaining 11%.
In most subjects (84%), loop excision was the primary surgery before randomization, with no significant difference between 5-FU therapy (81%) and observation (86%) groups. Twelve percent of subjects had had cone biopsies, 3% had had laser treatment, and 1% had had cryotherapy. Because most had had loop excision, no differences between treatment methods with respect to rates of recurrence of CIN could be detected. The median time from primary treatment to study enrollment was 7 weeks, and the interval was more than 8 weeks in 40% of subjects. There were no differences in the time from ablation to randomization or in ablative techniques between groups.
Of the subjects assigned to 5-FU therapy, four discontinued therapy before most (20 weeks) of the treatment period had elapsed: Two women never started treatment—one felt too ill to use the cream and requested removal from the study another gave no reason (those women are included in analysis, with their time on study censored at the date of removal)—one woman was repeatedly noncompliant and relapsed into substance abuse, and one woman had a decline in mental status and could no longer comply with the study.
Compliance with the scheduled timing of colposcopies and biopsies ranged from 30% (week 72) to 73% (week 12), with most visits showing 47–67% compliance and no difference between groups. Sixty-three subjects were censored at analysis because they did not have biopsy-proven recurrence. The most recent date of cytologic evaluation, colposcopy, or biopsy was used in those subjects for the date of censoring. For calculating time to recurrence, the biopsy date was used as the end point date. Although local site assessment of cervical pathology had no effect on end point definition, there was generally good agreement between site and central pathology analysis.
Overall, 38 (38%) of 101 subjects had recurrent CIN: 14 (28%) of the 50 who received 5-FU therapy and 24 (47%) of the 51 in the observation cohort. Recurrence rates were particularly high at State University of New York at Brooklyn and City Hospital at San Juan, with 14 and 11 of the 38 recurrences occurring at those institutions, respectively. Table 2 lists the P values for comparison of treatments and potential covariates for treatment failure stratified by baseline CD4 count. In a univariate analysis (Figure 1), treatment with 5-FU was significantly associated with prolonged time to recurrence of CIN when analysis was stratified by entry CD4 count (P = .04). In a stratified multicovariate proportional hazards analysis, the significance of treatment was slightly reduced (P = .08), the institution at which enrollment occurred remained highly significant (P = .002), and baseline use of antiretrovirals was not significant. The two institutions with the greatest accrual also had the highest recurrence rates (State University of New York at Brooklyn, 64%; City Hospital at San Juan, 58%) compared with the other institutions combined. Subjects enrolled by City Hospital at San Juan had lower CD4 counts than did subjects at other sites (16% at City Hospital at San Juan had more than 500 cells/mm3, compared with 27% at other institutions), whereas subjects entered at State University of New York at Brooklyn had significantly lower Karnofsky scores (23% had scores of 100: 65% at City Hospital at San Juan and 100% at other locations). Failure rates in the observation group at all sites were higher than with 5-FU therapy, and failure rates at State University of New York at Brooklyn and City Hospital at San Juan were essentially identical.
The grade of recurrent CIN also was affected by treatment approach. Overall, 18 subjects were diagnosed as having recurrent CIN 1 (ten in the 5-FU therapy group, eight in the observation group), 13 had CIN 2 (three in the 5-FU therapy group, ten in the observation group), and seven had CIN 3 (one in the 5-FU therapy group, six in the observation group). Observation subjects were more likely to have higher grade recurrence, with 31% of subjects developing CIN 2–3 compared with 8% who received 5-FU therapy (P = 0.014), and disease recurred more quickly in observation subjects as well. However, that does not account for time to CIN recurrence or the fact that subjects were removed from study at the first detection of CIN, so subjects with CIN 1 could not develop CIN 2 on study. To evaluate those data, statistical tests appropriate for data with competing risks were used, allowing comparison of cumulative incidence of recurrent disease. There was no apparent difference in the cumulative incidence of recurrent CIN 1, but there was a significantly higher cumulative incidence of recurrent CIN 2 and 3 in the observation group.
Baseline CD4 counts were significantly related to rate of recurrence. The median CD4 count of women who developed recurrent CIN was 242 cells/mm3, compared with 350 cells/mm3 in patients in whom disease did not recur, and 46% of women with CD4 counts less than 200 cells/mm3 developed recurrent disease, compared with 33% with CD4 counts at least 200 cells/mm3. The effect of CD4 count on time to recurrence is shown in Figure 2, which illustrates the more favorable outcome for women with better immune function. Differences were statistically significant (P = .04). Time to recurrence was almost identical in subjects with CD4 counts of 200–500 cells/mm3 and in subjects with counts above 500 cells/mm3, and those categories were combined for stratified analysis. When time to CIN recurrence was analyzed, with antiretroviral-naive women (n = 50) compared with antiretroviral-experienced women (Figure 3), a statistically significant difference was found (P = .05 by stratified Weibull modified Wald test, P = .07 by proportional hazards test), favoring the antiretroviral-experienced group.
Compliance with 5-FU therapy was monitored at each clinic visit. Thirty-one women reported at least one late or missed application during the treatment period, for a total of 65 reported cases of treatment-application delay. Twenty-five of those episodes were due to occurrence of menses, when women were not supposed to be applying the cream, and two episodes of missed application were ascribed to toxicity in one subject. The remainder were due to forgotting doses or true non-compliance. There was no significant difference in time to study discontinuation between treatment groups.
There were no grade 3 (severe) or grade 4 (potentially life-threatening) adverse effects or toxicities related to study treatment, and none of the comparisons of overall incidence between groups regarding signs and symptoms, hematologic abnormalities, or local reproductive effects yielded statistically significant results. All high-grade events in both arms occurred after 6 months and were related to HIV infection. Sixteen women in the 5-FU treatment group and ten in the observation group had at least one local reproductive symptom, with 10 5-FU treatment subjects and five observation subjects reporting at least one episode of bleeding, irritation, or discharge. Only one subject, in the 5-FU treatment group, had any local reproductive signs or symptoms that definitely could be attributed to the study regimen. Negative pregnancy test results were required at baseline except in cases of previous bilateral tubal ligation or lack of sexual activity with men. Two pregnancies occurred, both in observation subjects. One woman carried the pregnancy to term, and the other had a uterine curettage after spontaneous abortion during week 14.
Ten deaths were reported among the 101 women (six while on study): seven among subjects assigned to 5-FU therapy and three in the observation group. The last follow-up for survival was the date off study. In none of the cases in which death occurred was cervical intraepithelial neoplasia or study treatment listed as the primary or contributing cause, and all deaths but one (a suicide) apparently were due to AIDS-related illness. All deaths in the 5-FU therapy group occurred after the treatment period, and three occurred after subjects were off study at weeks 78, 84, and 103. Two deaths in the observation group occurred at weeks 20 (from Pneumocystis carinii pneumonia) and 26 (from HIV wasting syndrome). In the analysis, deaths were censored because they were not attributable to treatment or CIN.
The results of our study confirmed the high rate of CIN recurrence in HIV-positive women and were consistent with previous findings.12–15 The overall rate of recurrence in our study was 38%, with a maximum follow-up of 18 months. Maiman et al12 reported a rate of 39% at 1 year, compared with 9% in HIV-seronegative controls, whereas Petry et al13 reported a 60% recurrence rate in HIV-positive women after cone biopsy or laser ablation, compared with 10% in an HIV-negative control group. With a longer follow-up of 3 years, Fruchter et al14 reported a 62% recurrence rate. Because loop excision was the primary surgery before randomization in most of our subjects, our data are probably most comparable to those of Wright et al,15 who reported a failure rate of 56% for seropositive women, all of whom had loop excision. Our study also confirmed the reported negative effect of greater immunosuppression on rate of recurrence,16–18 evident in both groups, and the effect of CD4 count on time to recurrence. Petry et al13 noted that women with CD4 counts less than 400 cells/mm3 or immunosuppression for more than 3 years were more likely to have progressive lesions, whereas Maiman et al12 found that women with CD4 counts less than 500 cells/mm3 were at extremely high risk for recurrence compared with women with counts greater than 500 cells/mm3, whose risk was twice that among controls. Fruchter et al14 reported that 90% of HIV-infected women with CD4 counts less than 200 cells/mm3 developed recurrent dysplasia at 3 years and that women who developed recurrence had a median CD4 count of 193 cells/mm3, compared with 316 cells/mm3 for the total CIN cohort, figures remarkably similar to our findings.
The most important findings of our study were the difference in recurrence rate between the 5-FU therapy group (28%) and the observation group (47%) and the statistically significant differences in time to recurrence between the two groups. The recurrence rate in the 5-FU treatment group was superior to those reported in nearly all studies of HIV-infected women who received standard therapy only, especially in light of the 18-month follow-up. The marked difference in rates of recurrence of high-grade dysplasia between groups was also important. It is clear that all high-grade CIN, primary or recurrent, must be treated aggressively in HIV-positive women, but it is somewhat controversial whether CIN 1 needs to be treated similarly or whether posttreatment CIN truly represents recurrence in women with previously diagnosed CIN 2–3 or evidence of HPV exposure. If we take the more conservative approach toward the nature of CIN 1, the difference between the two groups becomes more impressive. The positive results of this study show that local adjunctive vaginal maintenance therapy with a topical chemotherapeutic agent can be prophylactic against cervical neoplastic disease in HIV-infected women successfully and that systemic approaches are not invariably necessary.
The other major finding of clinical importance in our study was that vaginal 5-FU at the dosage and intervals of this protocol was extremely well tolerated, with minimal morbidity in HIV-infected women with CIN. Vaginal 5-FU therapy was used with success in immunocompromised women with lower-genital-tract neoplasia but more commonly has been used for vaginal and vulvar disease, not as prophylaxis against cervical neoplasia. Sillman et al,19 using the technique of 5-FU chemosurgery, found that of 44 women with neoplasia at 90 sites, 43 went into remission, for a mean time of 38 months. In a randomized controlled study involving 90 women with HPV-associated lesions of vagina and vulva, Krebs et al9 found that after ablative therapy, women using prophylactic topical 5–FU at a dosage of 2 g biweekly were significantly less likely to develop recurrent lesions than were controls. Krebs et al20 also reported that after topical 5-FU treatment for vaginal HPV-associated lesions, 8.2% of subjects had troublesome chronic mucosal ulcers that were dose and duration dependent. However, those women used 5-FU weekly and continuously over 5–7 days, or periodically once a week for 10 weeks. Although longer follow-up might be needed for the women in our study, it is probable that the less aggressive biweekly regimen was responsible for the superior toxicity profile.
Although this study was not designed specifically to assess the effect of antiretroviral therapy on recurrence of CIN, the fact that the antiretroviral-experienced group in our study had more favorable outcomes regarding time to CIN recurrence than did the antiretroviral-naive group was interesting. We expected that that group might have higher rates of recurrence because of more severe immunosuppression, necessitating use of antiretrovirals, but there was only a median of 40 weeks of antiretroviral use in our 51 treated subjects, so it is difficult definitely to attribute the superior outcome to drug effect. Few women in our study used protease inhibitors, and given the superior virologic and immunologic response among many patients receiving combination therapy, data are needed to assess whether high rates of recurrent CIN among previous cohorts of HIV-infected women can be decreased by more highly active antiretroviral therapy.
A potential source of bias in this study relates to the fact that 41% of all subjects were enrolled at two AIDS Clinical Trials Group sites (State University of New York at Brooklyn and City Hospital at San Juan) and that recurrence rates were significantly higher at those institutions than at the other 17 sites. Those differences can be explained, in part, by the more severe immunosuppression (at City Hospital at San Juan) and the significantly lower Karnofsky scores (at State University of New York at Brooklyn) of women at those sites. An explanation might be that gynecologic and colposcopic expertise was more prevalent at those institutions, leading to higher rates of detection of CIN. This explanation is feasible because the difficulty in incorporating adequate gynecologic personnel at the study sites affected the protocol in other ways, in the form of a prolonged time to complete the study, a suboptimal total accrual compared with original estimates, and suboptimal compliance with scheduled colposcopic evaluations. Rates of recurrent CIN in observation women were higher than in the 5-FU treatment group across all sites, and more meticulous follow-up most likely would increase rates of detection in both groups, so the importance of committed gynecologic expertise in clinical trial programs for HIV-infected women cannot be overemphasized.
The management of CIN in HIV-seropositive women continues to present an enormous challenge for women's health specialists caring for this unique group. As medical therapy has improved for HIV-infected women CIN has become a chronic, persistent, and potentially aggressive complication. Adjunctive maintenance 5-FU therapy should be considered a safe and effective option to reduce recurrence of CIN after standard treatment for high-grade cervical dysplasia in HIV-infected women. Future investigations should be performed to determine whether 5-FU therapy can be extended safely with continued efficacy and what its effect may be on lower-grade dysplasia in HIV-positive women, CIN in women immunosuppressed by other causes, and recurrent and persistent CIN in immunocompetent women.
1. Maiman M, Tarricone N, Viera J, Suarez J, Serur E, Boyce JG. Colposcopic evaluation of human immunodeficiency virus-seropositive women. Obstet Gynecol 1991;78:84–8.
2. Schrager LK, Friedland GH, Maude D, Schreiber K, Adachi A, Pizzuti DJ, et al. Cervical and vaginal squamous cell abnormalities in women infected with human immunodeficiency virus. J Acquir Immune Defic Syndr 1989;2:570–5.
3. Feingold AR, Vermund S, Kelley KF, Schreiber LK, Munk G, Friedland GH. Cervical cytological abnormalities and papillomavirus in women infected with HIV. J Acquir Immune Defic Syndr 1990;3:896–903.
4. Korn A, Autry M, DeRemer P. Sensitivity of the Papanicolaou smear in human immunodeficiency virus-infected women. Obstet Gynecol 1994;83:401–4.
5. Wright TC Jr, Ellerbrock TV, Chiasson MA, Van Devanter N, Sun XW. Cervical intraepithelial neoplasia in women infected with human immunodeficiency virus: Prevalence, risk factors and validity of Papanicolaou smears. New York Cervical Disease Study. Obstet Gynecol 1994;84:591–7.
6. Maiman M, Fruchter RG, Serur E, Remy JC, Feuer G, Boyce JG. Human immunodeficiency virus and cervical neoplasia. Gynecol Oncol 1990;38:377–82.
7. Fruchter RG, Maiman M, Sillman FH, Camilien L, Webber CA, Kim DS. Characteristics of cervical intraepithelial neoplasia in women infected with the immunodeficiency virus. Am J Obstet Gynecol 1994;171:531–7.
8. Conti M, Agarossi A, Parazzini F, Muggiasca ML, Bochini A, Negri E. HPV, HIV infection, and risk of cervical intraepithelial neoplasia in former intravenous drug abusers. Gynecol Oncol 1993;49:344–8.
9. Krebs HB. Prophylactic topical 5-fluorouracil following treatment of human papillomavirus-associated lesions of the vulva and the vagina. Obstet Gynecol 1986;68:837–41.
10. Sillman FH, Sedlis A. Anogenital papillomavirus infection and neoplasia in immunodeficient women: An update. Dermatol Clin 1991;9:353–69.
11. Lindsey JC, Ryan LM. Tutorial in biostatistics: Methods for interval-censored data. Stat Med 1998;17:219–38.
12. Maiman M, Fruchter RG, Serur E, Levine PL, Arrastia C, Sedlis A. Recurrent cervical intraepithelial neoplasia in human immunodeficiency virus-seropositive women. Obstet Gynecol 1993;82:170–4.
13. Petry KU, Scheffel D, Bode U. Cellular immunodeficiency enhances the progression of human papillomavirus-associated cervical lesions. Int J Cancer 1994;57:836–40.
14. Fruchter RG, Maiman M, Sedlis A, Bartley L, Camilien L, Arrastia CD. Multiple recurrences of cervical intraepithelial neoplasia in women with the human immunodeficiency virus. Obstet Gynecol 1996;87:338–44.
15. Wright TC, Koulos J, Schnoll F, Swanback J, Ellerbrock TV, Chiasson MA. Cervical intraepithelial neoplasia in women infected with the human immunodeficiency virus: Outcome after loop electrosurgical excision. Gynecol Oncol 1994;55:253–8.
16. Johnson J, Burnett A, Willet G, Young MA, Doniger J. High frequency of latent and clinical human papillomavirus cervical infections in immunocompromised human deficiency virus-infected women. Obstet Gynecol 1992;9:321–7.
17. Schafer A, Friedman W, Mielke M, Schwartlander B, Koch MA. The increased frequency of cervical dysplasia-neoplasia in women infected with the human immunodeficiency virus is related to the degree of immunosuppression. Obstet Gynecol 1991;164:593–9.
18. Spinillo A, Tenti P, Zappatore R. Prevalence, diagnosis and treatment of lower genital neoplasia in women with human immunodeficiency virus infection. Eur J Obstet Gynecol Reprod Biol 1992;43:235–41.
19. Sillman FH, Sedlis A, Boyce JG. A review of lower genital intraepithelial neoplasia and the use of topical 5-fluorouracil. Obstet Gynecol Surv 1985;40:190–220.
© 1999 The American College of Obstetricians and Gynecologists
20. Krebs H, Helmkamp BF. Chronic ulcerations following topical therapy with 5-fluorouracil for vaginal human papillomavirus-associated lesions. Obstet Gynecol 1991;78:205–8.