Syphilis co-infection does not affect HIV disease progression

INTRODUCTION

Several studies have demonstrated that CD4 cell counts decrease and viral loads increase in HIV-infected persons co-infected with syphilis. ^1–4 These changes usually reverse after appropriate syphilis treatment; however, in some studies the CD4 count does not increase back to baseline ² and the viral load remains elevated 3–9 months after syphilis treatment. ¹ The mechanism of these changes is thought to be increased immune activation and cytokine release. ¹ The clinical significance of these transient changes remains unclear. We sought to evaluate the effect of syphilis co-infection on mortality and time to AIDS diagnosis in a large cohort of prospectively followed HIV-infected persons with estimated dates of seroconversion and equal access to health care.

METHODS

A total of 2239 HIV-infected persons who were enrolled in the US Military HIV Natural History Study (Department of Defense [DoD] HIV NHS) between 1985 and 2006 and who had documented negative HIV tests prior to documented positive HIV tests were included in this analysis. The details of this cohort have been described elsewhere, ⁵ but in brief, all consenting subjects were prospectively followed at one of several military treatment facilities with study visits approximately every six months. Demographic characteristics, markers of HIV disease, laboratory tests, antiretroviral medication use and clinical events (including AIDS defining illnesses according to the 1993 Centers for Disease Control and Prevention definition ⁶ ) were recorded at each visit. A central Institutional Review Board reviewed and approved this analysis.

Confirmed syphilis was defined as a concurrently positive non-treponemal and treponemal assay. The probable syphilis was defined as having received treatment for a clinical diagnosis of syphilis. Subjects who had a positive non-treponemal assay but a negative treponemal one and who did not receive therapy for syphilis were considered to have possible syphilis. Those who did not meet one of the above definitions were considered not to have syphilis. Cox regression analyses were performed to evaluate time to death or AIDS in those with and without syphilis adjusting for age at seroconversion, ethnicity, gender, hepatitis B or C co-infection, and CD4 count and viral load at HIV diagnosis. The stage of syphilis at treatment was unknown for many participants, so was not included in the analysis. Because many subjects in this cohort eventually initiated highly active antiretroviral therapy (HAART), several Cox models were analysed. In model A, the regression analysis was censored at HAART initiation or last follow-up visit (for those who did not initiate HAART). Model B was censored at HAART initiation, last follow-up visit or 1 January 1996 (the approximate date HAART became available for our cohort), whichever came first. Lastly, Model C was censored at last follow-up visit and HAART was treated as a time-dependent covariate.

RESULTS

Of the 2239 HIV-seroconverters (mean age [SD] = 28 [6.9] years, 94% men, 44% European American, 46% African American), 205 (9.2%) had confirmed syphilis, 66 (2.9%) had probable syphilis and 48 (2.1%) had possible syphilis over the course of their follow-up (mean follow-up from HIV diagnosis [SD] = 3.5 [3.2] years). In comparison with those without syphilis, those with confirmed syphilis were more likely to be African American (74% versus 42%, P < 0.0001), men (99% versus 93%, P = 0.0239), co-infected with hepatitis B (8% versus 5%, P = 0.016) or hepatitis C (9% versus 3%, P < 0.0001), and had a higher average number of other sexually transmitted diseases (including gonorrhea, chlamydia and herpes; 1.11 versus 0.68, P < 0.0001).

There was no difference (P > 0.05) between those without syphilis and those with confirmed syphilis in duration of time between estimated seroconversion date (the midpoint between the last negative HIV test and the first positive HIV test) and date of HIV diagnosis (319 ± 271 days versus 325 ± 283 days). There was also no difference between the two groups in rank (74% enlisted versus 85%), CD4 count at HIV diagnosis (555 ± 262 versus 570 ± 237 cells/mm³) or viral load at HIV diagnosis (4.8 log10 versus 4.7 log10 copies/mL). For those who started HAART, there was no difference between the two groups in the median year of HAART start (1999 versus 1998), CD4 count at HAART start (364 ± 212 versus 382 ±197 cells/mm³) or viral load at HAART start (4.9 log10 versus 4.7 log10 copies/mL).

Syphilis infection (confirmed + probable) was not significantly associated with time to death or AIDS in any of the multivariate models (Model A: hazard ratio [HR] = 1.03, 95% CI 0.76–1.39, P = 0.85; Model B: HR = 1.06, 95% CI 0.73–1.53, P = 0.77; Model C: HR = 0.98, 95% CI 0.75–1.30, P = 0.90). Because viral load assays were not routinely available until approximately 1997, many subjects included in this analysis did not have baseline viral load results available. Therefore, viral loads were not included in the multivariate models presented. When viral loads were entered into the models, the sample size significantly decreased but the effects of syphilis on time to death or AIDS remained unchanged. Similar models of time to death alone also did not show any significant association with syphilis co-infection (data not shown). All analyses were repeated limiting syphilis to confirmed cases only and the results did not significantly change (data not shown).

DISCUSSION

Syphilis rates continue to increase in the US with the majority of cases occurring in men who have sex with men. ⁷ In HIV-infected subjects enrolled in the DoD HIV NHS, the incidence of syphilis has significantly increased over time from 0.5 per 100 person-years in 1994–1999 to 1.7 per 100 person-years in 2000–2005. ⁸ Given the increasing numbers of HIV and syphilis co-infected persons, understanding the clinical significance of co-infection is crucial. The effects of HIV infection on syphilis presentation, diagnosis and management have been studied; however, there is less information on the effects of syphilis on HIV disease progression. Here, we show that in a cohort of 2239 HIV-infected persons prospectively studied with 7827 person-years of follow-up, syphilis co-infection did not affect time to death or AIDS defining condition.

A limitation of the study is that we do not know the stage of syphilis at diagnosis for all cases, although in the DoD syphilis testing is a routine part of HIV care and most cases are diagnosed early after infection. We also did not consider response to syphilis treatment in this analysis and it is possible that inadequately treated syphilis infection could affect HIV disease progression. Strengths of the study include the large size of the cohort examined, the fact that all subjects had estimated dates of seroconversion allowing for accurate time to event analyses and the fact that all persons in the study had equal access to the health-care system including free medications. We also considered various definitions of syphilis since the serological diagnosis in individuals with HIV infection can be difficult.

Although syphilis does not affect time to AIDS or death, given that other studies have shown it to be associated with increased risk of HIV transmission, there needs to be a continued emphasis on syphilis prevention, early diagnosis and treatment in the HIV-infected population.