Sexually transmitted diseases that result in a disrupted genital epithelium, such as syphilis, chancroid, and herpes, have been associated with heterosexual transmission of HIV-1 using retrospective studies in the United States [7-10] and both retrospective [11] and prospective studies in sub-Saharan Africa [12, 13]. In New York City, an ongoing epidemic of genital ulcer disease has occurred in communities where HIV-1 infection related to injection drug use is well documented [10, 14]. The number of cases of primary and secondary syphilis reported to the New York City Department of Health increased from 2157 in 1985 to 4231 in 1990. The number of reported cases of chancroid increased from 1323 in 1985 to 2277 in 1989. Consequently, a prospective study of HIV-1 seroconversion was initiated to further characterize the relation between genital ulcer disease and HIV-1 transmission in primarily heterosexual persons in the United States.

Methods

Top Methods Results Discussion Author & Article Info References

Study Population

This study was done in 1 of the 12 inner-city, sexually transmitted disease clinics operated by the New York City Department of Health. The study site is located in an area of New York City where the cumulative incidence of AIDS in adults through 1990 was 1 per 100 persons (New York City AIDS Case Surveillance data) and illicit drug use, including crack (smokable freebase cocaine) use, is common. In 1990, this clinic provided care to 14 243 persons: 9589 (67%) were men and 4654 were women. Primary or secondary syphilis was diagnosed in 226 patients, and 113 were found to have chancroid. From 1988 to 1989, the HIV-1 prevalence in this clinic was 7.8%, estimated by a serosurvey that was done without using patient identifiers (Weisfuse I. Personal communication).

Study Population

Recruitment into the study consisted of two phases. In the first phase, all of the approximately 28 000 persons attending the clinic for diagnosis or treatment of a sexually transmitted disease during the study period were asked to participate in a study of the prevalence of HIV-1 infection and associated risk factors. Those who agreed (n = 2893) received HIV-1 pretest counseling and were given a return appointment to receive test results and post-test counseling. Those with a diagnosis of genital ulcer were recruited more intensively. Thus, approximately 700 (24%) of the 2893 participants recruited into phase 1 had genital ulcers. These 700 participants with genital ulcer disease represented more than 80% of all patients seen in the clinic with ulcers during the study period.

Study interviewers administered standardized questionnaires in either English or Spanish. Information on demographic characteristics, socioeconomic status, and risk behavior associated with HIV-1 transmission was collected, as previously described [14]. Of the 2543 participants who were HIV negative, 1679 returned for post-test counseling 3 weeks after initial study enrollment and were asked to participate in the prospective component (or phase 2) of the study. For those who agreed, an additional questionnaire was administered to identify potential HIV-1-related high-risk behavior during three periods of interest: the 6 months before the initial clinic visit, the 10 days before the symptom developed that resulted in the clinic visit, and the period while the symptoms were present. The questioning focused on intravenous drug use and the number of sexual contacts and the type of sexual activities with intravenous drug users, homosexual or bisexual men, prostitutes, and others. The regularity of condom use was determined using the following scale: always, usually (>50%), sometimes (approximately 50%), rarely (<50%), and never. Anonymous HIV-1 testing was available for patients who did not want to participate in the study or who did not have a sexually transmitted disease.

At the time of enrollment in phase 2, participants were asked to return to the clinic for a third time, approximately 3 months after the initial clinic visit. For the participants who returned for the final follow-up visit, a repeated serum sample for HIV-1 testing was obtained, and a questionnaire similar to that described above was administered to identify high-risk behavior for the period between the two HIV-1 antibody tests. Informed consent was obtained for both phases of the study, and the study was approved by the institutional review boards of the New York City Department of Health and the Centers for Disease Control and Prevention.

Laboratory Methods

Patients with a genital ulcer had the following diagnostic tests: syphilis serology, dark-field examination of ulcer exudate for Treponema pallidum, Gram stain of ulcer exudate, microbiologic culture for Haemophilus ducreyi using blood and chocolate media, and Tzanck smear for herpes virus. Testing for syphilis was done using the rapid plasma reagin card test (Hynson, Wescott, and Dunning; Baltimore, Maryland) and the microhemagglutination assay for T. pallidum (Fujirebio Inc., Tokyo, Japan, and Ames Division, Miles Laboratory, Elkhart, Indiana). Haemophilus ducreyi was isolated using previously described methods [15]. Antibody testing for HIV-1 was done by an enzyme-linked immunosorbent assay (DuPont, Wilmington, Delaware), followed by a confirmatory Western blot analysis of all reactive samples using reagents prepared by the Laboratory of Retrovirology and Immunobiology of the New York City Department of Health [16].

Diagnosis of Genital Ulcer Disease

Primary syphilis was diagnosed when a genital ulcer was present and the ulcer exudate was dark-field positive, the rapid plasma reagin card test was positive, or the patient had recent contact with a person known to have syphilis. A diagnosis of chancroid was made when a positive culture occurred for H. ducreyi, if Gram stain of the ulcer exudate showed pleomorphic gram-negative rods, or if clinical findings suggested chancroid (tender or multiple ulcers, painful inguinal adenitis) with a negative dark-field examination as well as a negative syphilis serologic test result and a negative Tzanck smear. Genital herpes was diagnosed when the lesions were vesicular or recurrent or an ulcer had a positive Tzanck smear. In the absence of, or with negative, microbiologic and serologic data, the clinical diagnosis was made by the supervising physician in the clinic and not by study personnel.

Statistical Analysis

Analysis was done using the SAS statistical software system, version 6.06 (SAS Institute, Inc., Cary, North Carolina). The strength of the association between individual categorical variables or continuous variables grouped categorically and HIV-1 seroconversion was evaluated by the relative risk, and 95% direct precision-based CIs were obtained. Statistical relations were tested by the chi-square test or the Fisher exact test (two-tailed). Differences between continuous variables were analyzed by the Student t-test (two-tailed), the Wilcoxon rank-sum test, or the median test. The SAS LOGIST procedure was used to fit the multiple logistic regression model to the single binary outcome variable (HIV seroconversion or no seroconversion). The adjusted odds ratios obtained in this model approximated the adjusted relative risk.

Measurement of Risk Index

For heterosexual men, a summary measure was constructed of the risk for HIV-1 transmission attributable to a combination of the probability of encountering and the frequency of exposure to an HIV-1-infected partner. The sexual risk index was generated for the period from 6 months before the clinic visit through the second HIV test. Indices were computed by taking the sum of the products of the number of sexual contacts with partners in each of four risk groups (prostitutes, female intravenous drug users, women with chancroid, and women with no risk) and the estimated HIV-1 seroprevalence among the members of that group. Prevalence estimates for risk groups were determined from among our own study participants during phase 1, unblinded risk-factor serosurvey because sexual contacts were likely to occur in the local geographic area. The overall seroprevalence for the 644 women with no risk in phase 1 was 6%, for the 88 injection drug users it was 44%, for the 167 female prostitutes it was 29%, and for the 26 women with chancroid it was 19%. Using these values, risk indices were generated for four groups of male study participants: those with chancroid who did and did not seroconvert and those without chancroid who did and did not seroconvert. The risk indices were scaled proportionately from 0 to 100 for graphic presentation. Statistical comparisons were done between the Sexual Risk Index medians of the patients with chancroid and those without chancroid who seroconverted; between those without chancroid who seroconverted and those without chancroid who did not seroconvert; and between patients with chancroid who seroconverted and all patients who did not seroconvert using the median test.

Results

Top Methods Results Discussion Author & Article Info References

Of the 2543 seronegative persons identified in the cross-sectional component (phase 1) of the study, 1679 (62%) returned to the clinic to obtain their test results; 1669 (99%) of those who returned agreed to participate in the prospective study. Overall, 1325 (79%) of the 1669 patients who were HIV-1 negative and enrolled in phase 2 returned for the final interview and the second HIV-1 antibody test. Patients who completed the study and those lost to follow-up (n = 344) were similar when compared by sex, race, age, and diagnosis (data not shown).

Of the 1325 patients who completed the study, 462 (35%) presented with a genital ulcer and 863 (65%) presented for diagnosis or treatment of nonulcerative disease. Of these 1325 patients, 948 were men and 377 were women. The HIV-1 risk behavior distribution for the entire group was as follows: 77 (6%) were men who had sex with men, 49 (4%) were injection drug users, and 5 (0.4%) were both injection drug users and men who had sex with men. The probable risk behavior for the remaining 1194 (90%) participants was heterosexual activity. Among those with a genital ulcer, 13 (2.8%) patients seroconverted compared with 10 (1.2%) of the patients without a genital ulcer (relative risk, 2.4; CI, 1.1 to 5.4). The median follow-up periods for the two groups were similar, 119 days compared with 126 days.

Of the 1194 male and female participants whose only risk behavior was heterosexual activity, 12 (2.8%) of 432 with a genital ulcer seroconverted compared with 6 (0.8%) of 762 without a genital ulcer (relative risk, 3.5; CI, 1.01 to 8.98). For 92 patients, including 2 men who seroconverted, the second HIV-1 test was obtained more than 1 year after study entry and was excluded from further analysis. Of the remaining 16 patients who seroconverted, whose follow-up serum sample was obtained within 1 year of study entry, and who were heterosexuals who did not use injection drugs, 14 (88%) were men. The two women who seroconverted were having sexual relationships with men enrolled in the study who seroconverted.

Risk factors in women that were HIV-1 associated could not be statistically analyzed because of the small number of women who seroconverted. Thus, all further analyses of risk factors for heterosexual transmission of HIV-1 from this study are limited to the 758 heterosexual men who did not use injection drugs.

Of these 758 men, 344 (45%) presented with a genital ulcer and 414 (55%) presented for diagnosis or treatment of nonulcerative disease. The median age was 30 years in both groups; distribution of race and ethnicity was also similar (Table 1). Those with a genital ulcer, however, were more likely not to have completed high school and to be unemployed.

Characteristics related to sexual activity and potential risk factors for acquisition of sexually transmitted diseases were examined. Men with a genital ulcer were more likely to report not being circumcised, to have rarely used condoms, to have had more than two different partners in the 3 months before the first clinic visit, and to have had sexual contact with a woman who used crack or was a prostitute (see Table 1). The median interval between HIV-1 tests was similar for men with and without a genital ulcer: 117 days compared with 122 days.

Of the 344 men with genital ulcers, 314 had a single genital ulcer diagnosis and 30 had more than one cause for their ulcer. Chancroid was diagnosed in 137 (40%) men; 47 had positive cultures for H. ducreyi, 36 had positive Gram stains, and 54 had a clinical diagnosis of chancroid. Primary syphilis was diagnosed in 123 (36%) men; 74 had positive dark-field examinations, 48 had positive rapid plasma reagin test results, and 1 had recent contact with a person who had syphilis. Herpes simplex virus infection was diagnosed in 95 men; 51 had a positive Tzanck smear and 44 were diagnosed using clinical criteria. Nineteen patients had undiagnosed genital ulcers.

Of the 344 heterosexual men who had genital ulcers, 10 seroconverted (2.9%; CI, 1.4% to 5.3%) compared with 4 of the 414 (1%; CI, 0.2% to 2.5%) persons without genital ulcers (relative risk, 3.0, P = 0.05). In addition to the diagnosis of genital ulcer, the following were all associated with HIV-1 seroconversion using univariate analysis: not being circumcised, having the specific diagnosis of chancroid, and developing a new sexually transmitted disease during the follow-up period (Table 2). Primary syphilis (positive results using dark-field examination or the rapid plasma reagin test, or both), herpes, multiple genital ulcers, nonspecific ulcer diagnoses, and urethritis were not statistically associated with HIV-1 seroconversion.

Sexual exposure to an HIV-infected partner is necessary for sexual transmission of HIV to occur. Because HIV-1 infection status was unknown for sexual partners reported by the study participants, the likelihood of sexual exposure to an HIV-1-infected female partner was estimated by generating a sexual risk index (as described in the Methods section) for the period of time from 6 months before the clinic visit to the second HIV-1 test. The index was applied to persons with and without chancroid, because 6 of 10 patients with genital ulcers who seroconverted were diagnosed with chancroid, and chancroid was a major risk factor for seroconversion in the logistic regression analysis.

Figure 1 compares the aggregate sexual risk index of patients who seroconverted with those who remained HIV-1 negative; patients were stratified by chancroid status. Persons who seroconverted and who had a diagnosis other than chancroid (n = 8) had the highest sexual risk index (that is, the highest estimated likelihood of sexual exposure to HIV-1). Patients who seroconverted and had chancroid (n = 6) and those patients who remained HIV-1 negative, regardless of their clinical presentation, had similar and lower sexual risk indices. However, no statistical differences were noted between patients with chancroid who seroconverted and patients without chancroid who seroconverted; patients without chancroid who seroconverted and those without chancroid who remained HIV negative; and patients with chancroid who seroconverted and all patients who did not seroconvert (P > 0.2).

View larger version (13K): [in this window] [in a new window]   Figure 1. The sexual risk index for heterosexual men compared by diagnosis (with or without chancroid) and HIV-1 seroconversion status. The sexual risk index (SRI) score equals the summary measure of the number of sexual partners multiplied times the HIV prevalence, using a scale of 0 to 100.

Seroconversion in patients with chancroid occurred more often in those who did than in those who did not have sexual activity while their ulcers were present. Of the 137 heterosexual men with chancroid, 3 (14%) of 21 patients who had sex while the ulcer was present had seroconversion compared with 3 (3%) of 116 who abstained (relative risk, 5.6, P = 0.04). Among those without chancroid, seroconversion rates did not differ between those who did and those who did not have sex when symptoms of their sexually transmitted disease were present (1 of 90 compared with 7 of 529; relative risk, 0.8; P > 0.2).

Multiple logistic regression analyses, which included all hypothesized risk factors listed in Table 2 and the sexual risk index (log transformed), were done to determine whether any risk factors could predict seroconversion (Table 3). Chancroid (adjusted odds ratio, 3.3; CI, 1.1 to 10.1) and the development of a new sexually transmitted disease during the follow-up period (adjusted odds ratio, 3.2; CI, 1 to 10.1) were the only factors that were statistically independent predictors of seroconversion. Sexual risk index, circumcision status, frequency of condom use, and having sex while the symptom was present were also included in the original model but were not statistically different. Genital ulcers, primary syphilis, and nonspecific ulcers were not statistical predictors of seroconversion when these diagnoses were substituted for chancroid. Length of follow-up was excluded from the models because follow-up time was somewhat shorter for participants with a genital ulcer and yielded a negative relation when length of follow-up and seroconversion were compared.

Discussion

Top Methods Results Discussion Author & Article Info References

In this prospective study of patients attending a sexually transmitted disease clinic in New York City, a diagnosis of chancroid and the development of any new sexually transmitted disease during follow-up were found to be independent risk factors for HIV-1 seroconversion in heterosexual men without a history of injection drug use. Although our initial study hypothesis included other genital ulcers such as primary syphilis and herpes genitalis as potential risk factors for HIV-1 seroconversion, most of the seroconversions were in men with ulcers due to chancroid.

Two major hypotheses, either singly or in combination, could explain our observation that chancroid was associated with HIV-1 seroconversion in heterosexual men. The first is that the ulcer associated with chancroid increases the efficiency of female-to-male HIV-1 transmission. The second is that men with chancroid have a greater likelihood of being exposed to HIV-1-infected female sexual partners. To explore these hypotheses, we calculated the sexual risk index, largely based on risk behaviors defined by the Centers for Disease Control and Prevention. Among patients who seroconverted, the index was lower among those with chancroid than among those without chancroid (P > 0.2); although the sizable difference in indices was not statistically different, the number of persons who seroconverted in this comparison was small. The index in patients with chancroid who seroconverted, however, was similar to the index among those who did not seroconvert (P > 0.2). Both the similarity of the indices in patients with chancroid who seroconverted (n = 6) and in patients who did not have seroconversion (n = 744) and the difference in indices between patients with and without chancroid who seroconverted were consistent with increased efficiency of HIV-1 transmission in the presence of the chancroid lesion.

It appears likely that HIV can be transmitted either after or concurrent with the transmission of chancroid from women to men. Among heterosexual men with chancroid, HIV-1 seroconversion occurred more often in men who were sexually active when their ulcer was present. This finding supports the hypothesis that the ulcer in men facilitates entry of HIV. However, three patients with chancroid who seroconverted denied being sexually active when they were symptomatic. This observation also supports another hypothesis that HIV-1 may be transmitted concurrently with H. ducreyi. Data from Nairobi showing that HIV-1 can be cultured from genital ulcers in HIV-1-infected persons support this latter transmission hypothesis [17].

It is possible that chancroid is a marker for a characteristic or behavior—such as frequency of partner change or trauma during sexual activity—that was not measured and that increases the likelihood of acquiring HIV-1. The fact that the development of any new sexually transmitted disease was associated with HIV-1 seroconversion suggests that these unmeasured factors may be systematically different among those persons who do and do not seroconvert.

Several limitations of our study are noteworthy in addition to the relatively small number of patients who were identified as seroconverters. As previously mentioned, the HIV-1 serologic status of the sexual partners of our study participants was unknown and could only be estimated. The sexual risk index is a crude approximation of the risk for encountering an HIV-infected partner. Perhaps equally important is the lack of information on the relative infectivity of these sexual partners. To determine when in the course of a patient's HIV-1 infection he or she is most infectious to others will require studies that include identification and characterization of sexual partners.

Although standardized diagnostic tests for genital ulcers were used, the sensitivity of a microbiologic diagnosis for chancroid is problematic [18]. Among the 137 patients diagnosed as having chancroid in our study, 39% had a diagnosis based solely on clinical criteria. Thus, misclassification within genital ulcer disease entities probably occurred. However, the demographic data and the rates of HIV-1 seroconversion were similar in those patients who had microbiologically confirmed disease when compared with those who had a clinical diagnosis of chancroid (data not shown). Similarly, only about 60% of patients with primary syphilis had a positive result using dark-field examination. In addition, herpes simplex virus as a cause for the genital ulcer may have been underestimated because cultures were not done.

Our initial hypothesis was that genital ulcers, including chancroid, syphilis, and herpes simplex virus infection, were all likely to increase the efficiency of HIV-1 transmission. In fact, only chancroid was statistically associated with HIV-1 seroconversion. The reasons for this difference remain unclear. A clinical diagnosis of chancroid may have defined a group of patients with larger ulcers that bleed readily or have purulent exudates yet who do not have chancroid. However, chancroid is generally associated with more inflammation [19] than is syphilis or herpes simplex virus infection. Thus, chancroid may be more closely associated with HIV-1 excretion and acquisition than other "less inflamed" genital ulcers.

Although chancroid and acquisition of a new sexually transmitted disease during follow-up were risk factors for heterosexual acquisition of HIV-1, the estimated proportion of clinic attendees developing a new sexually transmitted disease (15%) was much greater than the proportion of patients who had chancroid (0.8%). Thus, although the efficiency of HIV-1 transmission may be enhanced in the presence of chancroid, it is likely that the patients with recurrent sexually transmitted diseases have a greater effect on overall heterosexual HIV-1 transmission. With high rates of sexually transmitted diseases occurring in major urban areas within the United States where high rates of HIV-1 infection occur secondary to injection drug use, heterosexual transmission of HIV-1 may become more common. Improved and more focused strategies for preventing all sexually transmitted diseases through decreasing the number of partners and consistently using condoms or having periods of abstinence could help limit the spread of HIV-1 within certain groups in the United States.

Presented in part at the Seventh International Conference on AIDS, Florence, Italy, June 1991, and the American Public Health Association Meetings, New York, New York, October 1990.