Genital Dysplasia in Women Infected with Human Immunodeficiency Virus

Methods

Two hundred HIV-positive women underwent standard colposcopic examination between October 1997 and December 2001 at an urban university hospital infectious disease clinic that is used to train Family Medicine residents in colposcopy. Women had been referred for colposcopic examination because of a recent abnormal Papanicolaou smear, self-reported abnormalities on Papanicolaou smear, current external HPV infection, or a history of genital dysplasia. At the time of colposcopy, all women were offered supplemental screening for HPV DNA by the Digene Hybrid Capture II method; 160 women accepted. Of these samples, 79 of the high-risk HPV were available for calculation of the relative light unit, a semiquantitative measure of the amount of HPV DNA. Cervical, endocervical, vaginal, and/or vulvar biopsies were performed as indicated by direct visualization during colposcopic examination. Additional data included CD4⁺ count, HIV viral loads collected within the 3 months preceding colposcopy, CD4⁺ nadir, and changes in CD4⁺ and viral load. The administration of HAART, as defined by the US Department of Health and Human Services guidelines to include at least 3 antiretroviral medications, was recorded.20 All antiretroviral medications had been prescribed by the patient’s current health care provider. Data on age, parity, risk factor for HIV infection, and race were collected. Patients who had an undetectable viral load or a 1.5 log or greater drop in viral load or a greater than 100 count increase in CD4⁺ for at least 3 months after initiation of HAART were defined as demonstrating an immune response to HAART.

Outcome variables were histology of biopsy results, categorized as no dysplasia, mild dysplasia (mild, VAIN I, VIN I), and severe dysplasia (moderate, severe, CIS, VAIN II-III, VIN II-III). Predictor variables included HAART therapy; immune response to HAART therapy; CD4⁺ count, either as a continuous variable or categorized as <200/μL, ≥200 μL; CD4⁺ count nadir; HIV viral load, either as continuous log-transformed or categorized as <10,000 copies/mL, ≥10,000 copies/mL peak viral load, log-transformed; whether viral load was ever nondetectable; and positive high-risk HPV. Other covariates evaluated were age, race, and reported route of HIV acquisition. Bivariate statistical analyses used χ² tests to compare categorical variables on proportions with none, mild, or severe dysplasia. For continuous variables, one-way analyses of variance were used to compare means of those with none, mild, and severe dysplasia. Multiple logistic regression models were developed to assess the effects of HAART therapy and other predictor variables on dysplasia, adjusted for confounding variables. Adjusted odds ratios (OR) with 95% confidence intervals (CI) were computed. Variables were included in models if they were associated with dysplasia at P < .10 or if inclusion changed the OR for HAART therapy substantially. All analyses were conducted using SAS (version 8.2; SAS Institute, Cary, NC).

Results

Table 1 presents a description of the sample of the HIV-positive women and genital histopathology results. From the subgroup of cases in which HPV DNA was obtained, high-risk DNA was noted to be associated with higher-grade dysplasia, in that it was detected in 34 of 35 (97%) cases of the severe dysplasia and 31 of 35 (86%) cases of the mild dysplasia compared with being detected in only 60 of 90 (67%) of those cases with no dysplasia (P = .0003). There was also a positive association of a quantitative increase in the amounts of detected high-risk HPV DNA, as measured by the relative light unit, with increasing degrees of dysplasia (P = .006). No associations of race, age, or route of HIV infection with the histopathological diagnosis of mild dysplasia or severe dysplasia were found. There was a significant protective effect of a CD4 count greater than 200 μL between the categories of no dysplasia (85 of 116), mild dysplasia (21 of 41), and severe dysplasia (22 of 42) (P = .008), but a nonsignificant difference between the same groups with respect to HIV viral load greater than 10,000 copies/mL (P = .12) (Table 2).

Forty-four percent (89 of 200) of the patients were deemed respondent to HAART by our criteria. Respondent patients demonstrated both significantly higher mean value of their CD4 count (368.4 vs 225.62 μL, P < .001) and lower mean log10 of their HIV viral loads (log 3.18 vs 4.29, P < .001). Patients in this category also had a borderline positive protective association with mild [relative risk (RR) = 0.61; 95% CI = 0.36, 1.02) and severe (RR = 0.50; 95% CI = 0.24, 1.05) dysplasia. If a patient had ever demonstrated a nondetectable HIV viral load while on HAART, this was found to be protective for both mild dysplasia (RR = 0.55; 95% CI = 0.32, 0.94) and severe dysplasia (RR = 0.32; 95% CI = 0.14, 0.75). The strongest predictor of genital dysplasia was the nadir of the CD4 count mean: 137 μL (±154) in those with mild dysplasia, 128 (±142) in those with severe dysplasia versus 228 μL (±185) in those without any histologic proof of dysplasia or normal examinations (P = .0006).

Several models were evaluated to determine the most parsimonious model. When included in a model with HAART therapy, neither age, race, route of infection, current HIV viral load, nor current CD4⁺ count was found to be associated with dysplasia. High-risk HPV, which was nearly perfectly associated with dysplasia, was not included in the models. Both the nadir of CD4⁺ count and nondetectability of HIV at any previous time were highly associated with HAART therapy itself. HAART therapy in turn was more strongly protective against dysplasia when nadir of the CD4⁺ count and nondetectability of the HIV virus were included in the models. (OR = 0.34; 95% CI, 0.17, 0.72).

Discussion

Before HAART, the natural course of HIV infection led to a progressive decline in a patient’s innate immunity. With the decrease of the patients’ CD4⁺ count, a reciprocal increased susceptibility to various opportunistic infections occurred. Previous studies have reported a similar increase in the incidence of HPV-induced dysplasia with declining CD4⁺ counts in patients that are not treated with antiretrovirals.8,10 With the introduction of HAART in the United States, the incidence of most AIDS-defining illnesses decreased as patients experienced a numerical restoration of their CD4⁺ lymphocyte count.21 A perplexing exception to this trend was shown in a study in which the incidence of invasive cervical cancer actually increased in the HAART era.22 Thus, those markers of the immune system that represent a successful immunorestoration against opportunistic diseases in patients infected with HIV are not yet fully defined.

In this cohort of women infected with HIV, we identified both a total CD4⁺ count below 200 and a low CD4 count nadir as immune-based markers for the risk of genital dysplasia.

Recent data showing that prolonged periods of CD4⁺ lymphopenia in patients infected with HIV resulted in defects in T cell proliferation regardless of the current CD4⁺ count or HIV viral load could explain the fact that the CD4⁺ nadir was a strong predictor of dysplasia.23 This phenomenon could also result in periods of HPV proliferation and/or activation. It is possible that there is an incomplete immune restoration with HAART, and the ability to recognize and respond to HPV specific antigens is lost at lower CD+ nadirs, which results in persistent HPV infection. This is consistent with our study, in which the CD4⁺ nadir of those without high risk HPV DNA was much higher than in those with detectable HPV DNA. (238 vs 161, P = .02).

We also detected a trend toward a protective effect of HAART against HPV-related genital dysplasia at the time of colposcopic examination. Of concern is that this study did not detect any direct effects of HAART, CD4, or HIV virus load on HPV quantity, and that 60 of 90 (67%) of the patients without dysplasia displayed detectable high-risk HPV DNA, which may in turn forebode future dysplasia.24–26

Weaknesses in this study are that we did not account for the nadir CD4 count at the time of HAART initiation, and many patients had antiretroviral therapy started at a lower CD4 level than is currently recommended. The timing of HAART therapy was related to both a patient’s status at the time of diagnosis and/or to her willingness to start antiretroviral therapy. Whether the protective response of HAART against genital dysplasia was an effect of a partial immunorestoration or from the direct suppression of HIV or HPV could not be ascertained from these data, because there were positive influences from HAART on both of these parameters, and either one could effect the expression of HPV.

There is an important limitation to observational studies such as ours that prevents the evaluation of long-term changes in immune status or on outcomes. Another limitation of this study is that we do not have exact data on the duration of HPV infection or HIV infection

Our findings may be an important incentive for women to take HAART because recent studies have shown that women have a greater reluctance to start HAART compared with their male counterparts.27,28 Because women survive longer with HIV infection, guidelines that provide for optimal care of genital dysplasia need to be maximized. The screening guidelines for women who have experienced a low nadir CD4 count may need to be modified to encourage closer monitoring. Because many women infected with HIV receive gynecological care from their primary care provider, it will be important for them to know the natural course of HPV-related diseases. Of utmost importance is that many women infected with HIV are unaware of their status. Prompt identification of these patients is paramount, because if a woman’s immune function can be preserved through earlier control of HIV infection, there may be a resultant overall decrease in the incidence of genital dysplasia and subsequent cancer.