High Rate of Recurrence of Cervical Intraepithelial Neoplasia After Surgery in HIV-Positive Women
".... The present study performed in the era of HAART confirms the results of earlier reports of high recurrence rates of cervical lesions after surgery in HIV-positive women, with an observed rate of recurrence of CIN of 53.9% at 36 months and of 22.3 per 100 patient-years of follow-up. Our study strongly supports the hypothesis that recurrent CIN is the feature of cervical neoplasia most characteristic of HIV disease in the context of immunodeficiency. Thus, women with CD4 counts less than 200 cells/mm3 had a 9 times higher risk for persistence and/or recurrence of CIN, emphasizing the role of cellular immune responses in the control of the HPV-related cervical disease..... whereas HAART had a strong protective effect...."
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 39(4) 1 August 2005
Heard, Isabelle MD, PhD*; Potard, Valerie MS; Foulot, Herve MD; Chapron, Charles MD; Costagliola, Dominique PhD; Kazatchkine, Michel D MD*
From the *Institut National de la Sante et de la Recherche Medicale (INSERM) U430 and Unite d'Immunologie Clinique, Hopital Europeen Georges Pompidou, Paris, France; INSERM U720, Universite Pierre et Marie Curie, Paris, France; and Department of Gynecology, Hopital Cochin, Paris, France.
Objective: Our study investigated the rate of recurrence of cervical intraepithelial neoplasia (CIN) in HIV-positive women after surgery in the era of highly active antiretroviral therapy (HAART).
Methods: One hundred twenty-one HIV-positive women were followed-up with cytology, colposcopy, and histology after surgery for CIN. We conducted univariate and multivariate analyses to determine the relation between recurrence of CIN and risk factors using Cox proportional hazard models with left truncation.
Results: The rate of recurrence of any CIN was 22.3 per 100 patient-years and the rate of high-grade CIN was 8.6 per 100 patient-years during 166 and 279 patient-years of follow-up, respectively. In multivariate analysis, a positive margin was associated with a risk of recurrence of any CIN (relative risk [RR] = 3.5, 95% confidence interval [CI]: 1.2-9.8) and a risk of recurrence of high-grade CIN (RR = 9.0, 95% CI: 2.2-36.5). CD4 counts <200 cells/mm3 were associated with a risk of recurrence of any CIN (RR = 9.4, 95% CI: 2.7-32.7) but not with a risk of recurrence of high-grade CIN. HAART exhibited a protective effect on the recurrence of any CIN (RR = 0.3, 95% CI: 0.1-0.7) and of high-grade CIN (RR = 0.2, 95% CI: 0.1-0.7).
Conclusion: CD4 cell counts <200/mm3 and a positive margin were predictors of recurrence, whereas HAART had a strong protective effect. Although surgery is highly effective in immunocompetent patients, it seems to be effective only in preventing progression to cancer in HIV-infected women.
Women infected with HIV are at increased risk of developing cervical intraepithelial neoplasia (CIN), a cervical cancer precursor.1-3 The risk of developing CIN is associated with infection with human papillomavirus (HPV), which is detected in more than 60% of HIV-infected women,4,5 and HIV-induced immunodeficiency.6-8 In immunocompetent HIV-seronegative women, recently published recommendations for the management of CIN call for expectant management of low-grade lesions and for surgical excision or ablation of high-grade CIN.9,10 Several studies have indicated a high rate of success of current treatment strategies for CIN in immunocompetent women.9-11 In contrast, failure of surgical treatment has been reported in HIV-positive women as well as an increased risk of recurrence of CIN.12-15 Recurrence rates were reported to be greater in women with low CD4+ T-cell counts, reaching 87% at 3 years in severely immunocompromised women.14 Significant progress in HIV therapy has been achieved since these early studies were published, with the introduction and widespread use of highly active antiretroviral therapy (HAART), which has resulted in a dramatic decrease in AIDS-associated mortality and morbidity.16 The increase in life expectancy associated with HAART may provide opportunities for lesions to recur or to progress to invasive cervical carcinoma (ICC). Conversely, immune restoration associated with antiretroviral therapy could result in enhanced host immune responses and control of cervical neoplasia,17 potentially leading to excisional therapy being more effective for eradication of CIN.
In the present study, we examined the rate of persistence and/or recurrence of CIN in a group of HIV-infected women after standard surgical therapy, focusing on the impact of the modalities of surgical treatment and HAART on recurrence rates.
The present study performed in the era of HAART confirms the results of earlier reports of high recurrence rates of cervical lesions after surgery in HIV-positive women, with an observed rate of recurrence of CIN of 53.9% at 36 months and of 22.3 per 100 patient-years of follow-up. Our study strongly supports the hypothesis that recurrent CIN is the feature of cervical neoplasia most characteristic of HIV disease in the context of immunodeficiency. Thus, women with CD4 counts less than 200 cells/mm3 had a 9 times higher risk for persistence and/or recurrence of CIN, emphasizing the role of cellular immune responses in the control of the HPV-related cervical disease.
Among cone biopsies with evaluable margins, positive margins were diagnosed in 26% of histologic specimens. In a cross-sectional study of HIV-positive and HIV-negative women, Boardman et al20 reported a higher frequency of positive cone biopsy specimen margins among HIV-positive women than among HIV-negative women (odds ratio = 2.25, 95% CI: 1.06-4.79). The large size of CIN observed in HIV-positive women and the frequent involvement of endocervical glands may account for the high rate of incomplete excision.21 In univariate analysis, complete excision was found to be more frequent in the case of high-grade lesions and independent of the surgical procedure that had been used. The latter observation may relate to a larger sized tissue excision performed by the surgeon in the case of a severe lesion, independent of the size of the visualized lesion.
As previously reported by others, we observed that treatment failed to eradicate CIN in many HIV-infected women.12,14,21,22 Nevertheless, most recurrent lesions were of lower grade, and recurrence rates of high-grade disease were lower as compared with the overall rate of recurrence of CIN, suggesting persistent opportunistic HPV infection with regression of cervical disease and, possibly, a delayed occurrence of cervical cancer. Definitive management with hysterectomy has been suggested by Boardman et al20 for eradication of the disease. Using data from the HIV Epidemiology Research (HER) Study, Paramsothy et al23 recently showed that among HIV-infected women, evidence of CIN before or at hysterectomy was associated with SILs on vaginal Papanicolaou smears in 63% of the cases during follow-up. Although the study did not distinguish early recurrence of cervical disease from vaginal lesions existing before hysterectomy, the high rate of SIL vaginal cytology suggests that hysterectomy does not eradicate HPV-related anogenital disease.
Several factors may be associated with higher rates of surgical treatment failure, including high-grade CIN, large lesion size, satellite HPV-related lesions, involved margin, persistent HPV infection, and immunodepression.11,24-26 As previously described in immunocompetent women, we observed no relation between the grade of the lesions and risk of recurrence and/or persistence of CIN.12,14 Because of missing data for the 66 women treated before inclusion, we were not able to assess the relation between lesion size or anogenital HPV-related associated lesions and the persistence and/or recurrence of CIN. Among women treated after enrollment in the study, we found in a subset analysis (47 of the 75 women included in the recurrence analysis) that the risk of recurrence was highly associated with persistent HPV DNA (data not shown).
The role of immunodepression and the effect of HAART on outcomes of treatment of CIN have been assessed in previous studies with conflicting results. In some studies, recurrence was more frequent in highly immunosuppressed women,2,13,14,21 whereas others did not observe such correlation.12,15 In a retrospective chart review, the use of HAART along with standard excisional therapy was associated with a lower rate of recurrence or persistence of CIN as compared with excisional therapy alone (18% vs. 70%; P < 0.05).27 After a mean follow-up of 2 years after excisional treatment, no recurrence of CIN was reported in a small series of 47 women on HAART.28 Wright et al29 did not observe a positive impact of HAART on posttreatment recurrence, however. In the present study, we found that HAART was associated with a 3-fold lower risk of recurrence of any lesion. It may be that the surgical modalities used since the availability of HAART have increased the rate of residual disease, rendering HAART less protective than it actually is.
Our study needs to be interpreted in the light of several limitations. In our series, we were unable to show differences in outcome by treatment modality, possibly because of small numbers and short follow-up. Furthermore, our observations may not be extended to women not in intensive cervical screening and/or treatment programs. Nevertheless, few studies have addressed this issue at this time, and our results indicate a strong positive impact of HAART on the recurrence rate of CIN after surgery.
Our observations document that recurrent CIN is common among HIV-positive women even in the HAART era and further emphasize the need for follow-up after surgery. Effective antiretroviral therapy for HIV disease may delay the recurrence of HPV-related disease. Persistent HPV DNA remains detected in patients on HAART, however, indicating that immune restoration on HAART is insufficient to clear HPV.17 Data from the International Collaborative Group on HIV and Cancer have shown no change in the incidence of ICC since the introduction of HAART.30 It is possible that despite active screening programs, cervical cancer occurs in HIV-infected women with longer follow-up, if HPV infection drives a greater cumulative burden on cervical cancer incidence because of a longer at-risk exposure time, despite HAART delaying disease. Although excisional therapy is highly effective for CIN in immunocompetent patients, such treatment seems to be effective only in preventing progression to cancer, at least in the short term, in HIV-infected women. Longer follow-up may allow us to define the success and/or efficacy of treatment in chronically immunocompromised women.
The demographic characteristics of the 121 women are shown in Table 1. At the time of surgery, among the 93 high-grade lesions, 77 were biopsy proven, 14 were high-grade SILs associated with high-grade colposcopic abnormalities and low-grade CIN, and 2 were high-grade SILs associated with high-grade colposcopic abnormalities with at least 1 recent biopsy showing high-grade CIN. All but 2 low-grade lesions were biopsy confirmed; the remaining 2 had low-grade CIN diagnosed at the prior visit.
Among the 28 women with low-grade CIN, laser ablation was used in 14 and excisional therapy in 14. Among the 93 women with high-grade CIN, laser ablation was used in 4, excisional therapy in 88, and hysterectomy in 1. Excisional therapy was conization in 68 women and LEEP in 34 women. Care of CIN evolved during the study period, with LEEP being more frequently used for the treatment of low-grade and high-grade lesions to the detriment of knife conization and laser vaporization not being used for treatment of high-grade lesions after June 1997 (P = 0.05).
Excisional specimen margins were available in 82 of the 102 excisional treatments. Twenty-one of the 82 women with evaluable margins had negative margins (ie, no CIN or HPV-associated cytologic changes extending to the margins) on cone biopsy specimens, independent of the modalities of surgery. In univariate logistic regression, the presence of a high-grade lesion in the conization specimen was found to be associated with complete excision (relative risk [RR] = 4.5, 95% confidence interval [CI]: 1.1-18.6; P = 0.04), with no other variable predictive of complete excision, including age, parity, time of surgery, and surgical procedure.
The final pathologic findings were available in 91 patients: 1 specimen was negative, and there were 11 cases of low-grade CIN, 77 cases of high-grade CIN, and 2 cases of carcinoma in situ.
Of the 121 women, 12 underwent surgery after enrollment and had no follow-up visit and 34 underwent surgery before enrollment and already had recurrence at the time of inclusion, leaving only 75 women for assessing recurrence of CIN. The median time of follow-up of these 75 women, calculated from the time of enrollment for women who had been treated before entry in the study and from time of surgery for women treated after enrollment, was 1.7 years (ie, 166 patient-years). There was no evidence of CIN at the last visit in 9 (69.2%) of the 13 women with low-grade lesions and in 29 (46.8%) of the 62 women with high-grade lesions at the time of surgery. On follow-up, recurrent CIN was identified in 37 women. After surgery for low-grade lesions, the recurrent disease was of similar grade to the original lesion in 2 cases (15.4%) and staged as high-grade disease in 2 cases (15.4%). After surgery for high-grade CIN, the grade of recurrent lesions was lower in 22 cases (35.5%) and similar in 11 cases (17.7%). The rate of recurrence was not different based on the type of excisional procedure that was performed (P = 0.26). The overall rate of recurrence of any CIN was 22.3 (95% CI: 16.1-30.7) per 100 patient-years of follow-up. Although most recurrences occurred in the first year after surgery, CIN further recurred in the second and third years of follow-up, reaching 53.9% at 3 years. ICC was found in the excision specimen in 1 patient treated for high-grade CIN. No incident cancer developed.
In univariate analysis, a positive margin, time of surgery, and CD4 counts less than 200 cells/mm3 were significantly associated with the risk of recurrence of cervical lesions. Being on HAART was associated with a lower risk of recurrence. There were no other variables found to be predictive of recurrence.
In the final multivariate model, risk factors that were found to be associated with recurrence were a positive margin, CD4 counts <200 cells/mm3, and no HAART. Highly immunodepressed women were found to have a 9-fold higher recurrence than women with less CD4 counts cells/mm3 (RR = 9.4, 95% CI: 2.7-32.7; P = 0.002). Controlling for CD4 cell count, time of surgery, and margin status, HAART demonstrated a protective effect, as shown by a significantly lower risk of recurrence of CIN (RR = 0.3, 95% CI: 0.1-0.7; P = 0.007).
Because HAART and CD4 level may interact, the risk of recurrence was then calculated in 2 different analyses using a Cox regression model. As compared with women with CD4 cell counts greater than 500 cells/mm3, the RR of recurrence among women not receiving HAART with CD4 cell counts between 200 and 500 cells/mm3 was 4.1 (95% CI: 0.9-18.8), and in women with CD4 counts <200 cells/mm3, the RR was 14.9 (95% CI: 2.8-78.0; P = 0.003). As compared with women with CD4 counts greater than 500 cells/mm3, the RR of recurrence among women receiving HAART with CD4 counts between 200 and 500 cells/mm3 was 3.5 (95% CI: 0.6-19.4), and in women with CD4 counts <200 cells/mm3, it was 5.0 (95% CI: 0.8-30.1; P = 0.21). This suggests that the CD4 count contributes to recurrence outcome among nontreated women, whereas it has no impact in women on HAART.
We assessed recurrence of high-grade disease in 100 women, including the 75 women selected for the study of recurrence of any CIN and 25 women who had undergone surgery before enrollment and presented with low-grade CIN at inclusion. Recurrent high-grade CIN was identified in 24 patients. The overall rate of recurrence for high-grade disease was of 8.6 (95% CI: 5.8-12.8) per 100 patient-years of follow-up during a total of 279 patient-years. In the multivariate model, we found the same impact of a positive margin on recurrence of high-grade CIN as on recurrence of any CIN (RR = 9.0, 95% CI: 2.2-36.5) and a lower impact of CD4 count <200 cells/mm3 (RR = 4.6, 95% CI: 1.0-20.6). The protective effect of HAART remained significant (RR = 0.2, 95% CI: 0.1-0.7; P = 0.01).
PATIENTS AND METHODS
All women attending the gynecology outpatient clinic of the HIV departments of Hopital Europeen Georges Pompidou and Hopital Cochin, Paris, from June 1993 until November 2003 were offered participation in a prospective gynecologic survey. Eligibility criteria for enrollment in the study included documented HIV-positive status and consenting to the protocol. Among the 1004 eligible women, 999 decided to enroll in the study and 5 declined because of fear of loss of confidentiality by participating in a cohort of HIV-infected patients. The protocol was approved by the French National Agency for AIDS Research (ANRS).
The cohort has been described elsewhere.6 Briefly, the study included a structured questionnaire and gynecologic examination, including a Papanicolaou smear and a standardized colposcopic examination of the cervix at enrollment and every 6 months. All patients with major colposcopic abnormalities or high-grade squamous intraepithelial lesions (SILs) underwent a biopsy unless the women refused or were not compliant for follow-up. Biopsies were also performed in most women with low-grade SILs. Regular follow-up was proposed to women with low-grade CIN; unless they preferred surgical treatment, efforts to bring women in for treatment were made for patients with high-grade CIN. The most recent values of CD4+ T-cell counts and plasma HIV RNA levels within 6 months of the gynecologic examinations were obtained from clinical records. Smears and biopsies were read by the same pathologist (C. Bergeron, Laboratoire Pasteur Cerba, Cergy-Pontoise, France). Cytologic categories were normal, atypical squamous cells of undetermined significance (ASCUS), low-grade SILs, and high-grade SILs, according to the Bethesda System.18 Histologic categories were normal, metaplasia, low-grade CIN, and high-grade CIN.19
We used the clinical classification of cervical disease proposed by Wright et al.1 CIN was thus defined as high grade on the basis of histologically proven high-grade CIN or low-grade CIN at histology associated with high-grade colposcopic abnormalities and a smear showing a high-grade SIL or of high-grade colposcopic abnormalities and a smear showing a high-grade SIL in the absence of biopsy. CIN was defined as low grade on the basis of histologically proven low-grade CIN or of metaplasia at histology associated with a low-grade SIL and colposcopic abnormalities. During follow-up, a normal smear and colposcopy, minor colposcopic changes associated with normal cytology and histology, and ASCUS associated with normal colposcopy were considered as no evidence of CIN.
Among the 999 women enrolled in the cohort, 5 with a history of hysterectomy and 25 who had no follow-up evaluation after enrollment were excluded from the analysis. CIN was detected at least once in 529 of the 969 women. Among the latter, 121 underwent surgical treatment of cervical disease: 66 women had been treated before inclusion (24 with low-grade CIN and 42 with high-grade CIN), and 55 women were treated during follow-up (4 with low-grade CIN and 51 with high-grade CIN). The end points of the present analysis were the presence of CIN and that of high-grade CIN. Recurrence after surgery was defined as histologically proven CIN diagnosed within 3 months after surgery or, in the absence of biopsy, by at least 2 consecutive abnormal Papanicolaou smears associated with abnormal colposcopic findings. Follow-up was ended at the last visit for those who did not show recurrent lesions.
HAART included at least 2 nucleoside reverse transcriptase (RT) inhibitors in combination with a protease inhibitor, a nonnucleoside RT inhibitor, or a third RT inhibitor. Women who were receiving fewer than 3 antiretroviral drugs were not considered as being on HAART.
Treatment included ablative (laser vaporization) and excisional (loop electrosurgical excision procedure [LEEP] and conization) modalities. For women treated after enrollment, initial routine colposcopy was performed using 3.0% acetic acid and Lugol iodine to outline the extent of the lesion. Based on the surgeon's decision, patients with high-grade lesions were then usually treated with excisional therapy and patients with small-sized low-grade lesions limited to the exocervix and seen in their entirety were treated by laser vaporization. CO2 laser vaporization was performed to a depth of 5 to 7 mm involving the entire transformation zone and extending 0.5 cm beyond the visible lesion. Patients with a small cervix size and lesions limited to the exocervix were treated with LEEP. For LEEP, an ERBE electrosurgical generator unit (ERBOTOM ICC350) was used set to 120 W for cutting and 75 W for coagulation. Conization was performed in case of (1) large lesions extending into the exocervix, (2) lesions extending into the endocervical canal, or (3) poor visualization of the transformation zone. Electrosurgery was performed with the cervical crater being left open. Hemostatic coagulation was performed when necessary.
Statistical analyses were performed using SAS software (version 8; SAS Institute, Cary, NC). Univariate logistic regression models were used for analysis of factors associated with negative cone biopsy specimen margins at surgery. Parameters investigated included patient age at surgery, parity, grade of the lesion, time of surgery, and surgical procedure. Odds ratios were calculated among the women with an evaluable margin.
Analysis of the recurrence of any CIN was restricted to the women with normal Papanicolaou test results at baseline among the women treated before inclusion in the cohort and to the women with at least 1 follow-up visit after surgery for women treated during follow-up. Analysis of the recurrence of high-grade CIN was restricted to the women with normal Papanicolaou test results or low-grade lesions at baseline among the women treated before inclusion in the cohort and to the women with at least 1 follow-up visit after surgery for women treated during follow-up. A multiple Cox regression model with left truncation was used to identify risk factors for recurrence of any CIN and for recurrence of high-grade CIN. This corrects for the bias that would be attributable to recurrence of any grade of lesion for the first end point and of high-grade CIN for the second end point occurring before inclusion for women with surgery before inclusion. The delayed entry of patients into the risk set was calculated as the time between prior surgery and inclusion in the cohort. Hormonal contraception, HAART, and CD4 cell counts were included as time-dependent variables. The following factors were assessed: age, race, obstetric history, group of transmission, total number of sexual partners, hormonal contraception, tobacco smoking, grade of the lesion, treatment modalities, margin status, time of surgery (before or after June 1997), HAART, and CD4 cell count. The time at risk was the time between the first surgery and the recurrence of any disease for the first end point and of high-grade disease for the second end point or the last follow-up, which ever occurred first. Univariate analysis was used to screen for an association with recurrence. Variables with P < 0.10 were considered for inclusion in the final multivariate model in which the usual P < 0.05 was considered as significant.