Seroreduction of syphilis non- treponemal titers during pregnancy for women with and without HIV co- infection

a recent review report-ing a worldwide 9.5% co- infection rate of syphilis in those with HIV infection. 2 In addition to increased risk of pregnancy morbidity, HIV co- infection in pregnant women with syphilis presents unique diagnostic and treatment challenges. Previous studies suggest that HIV co- Abstract Objective: To evaluate the effect of HIV co- infection on non- treponemal titers during pregnancy in women with syphilis. Methods: This is a secondary analysis of pregnant women with syphilis in the prospec-tive, observational Zambian Preterm Birth Prevention Study (ZAPPS). Treponemal ( Treponema pallidum particle agglutination) and non- treponemal (rapid plasma reagin; RPR) testing were performed on serum biospecimens, resulting in 47 participants with serologically confirmed syphilis (27 HIV- positive, 20 HIV- negative). The primary out -come, achievement of RPR titer seroreduction during pregnancy, was analyzed by logistic regression. Secondary outcomes included overall titer reduction, seroreduction rate, serologic cure, and adverse pregnancy outcomes. Results: Seroreduction of RPR titer occurred in 78% (21/27) of women with HIV ver sus 45% (9/20) of women without (adjusted odds ratio 4.66; 95% confidence interval [CI] 1.14 − 19.08). Overall RPR titer reduction, rate of seroreduction per week, and the proportion achieving serologic cure each trended higher among women with HIV compared with those without HIV. There was a trend toward decreased stillbirth incidence in participants achieving seroreduction (odds ratio 0.15, 95% CI 0.01– 1.58). Conclusion: HIV co- infection in this cohort of Zambian

Despite the scientific advances of the last half century, syphilis diagnosis and treatment monitoring remain challenging. Serologic diagnosis of syphilis requires both Treponema-specific and nonspecific testing; however, only non-treponemal serologies are effective at monitoring syphilis treatment efficacy. 6,7 Serologic cure following adequate treatment is defined as a four-fold decline in non-treponemal titer (equivalent to reduction by two dilutions) or seroreversion to a negative titer. Hence, serial monitoring of nontreponemal titers is required to assess treatment efficacy, diagnose cure, and assess for re-infection.
Following treatment, the rate of titer reduction varies based on a variety of factors, including starting titer, stage of disease, and previous infection. Hence, achievement of cure may take 12-24 months, although many patients with primary, secondary, or early late-stage disease achieve cure by 6 months. 5,6 In non-pregnant patients, HIV co-infection has been associated with reduced rates of seroreduction; however, supporting data are older, limited, and not inclusive of pregnant patients. [3][4][5] Although normative values and prediction of seroreduction rates remain elusive, guidelines for serologic monitoring and diagnosis of treatment failure outside pregnancy are well established and usually occur at 6-month intervals (3-month intervals for those with HIV). In pregnant patients, however, no optimal interval has been identified, and indications for re-treatment are less clear. Recommendations for timing of repeat titers in pregnancy differ and range from monthly to every 6 months. 6,8 Even less information is available to guide obstetric providers regarding when to re-treat based on these repeat serologies, particularly for patients with HIV. Given concerns for de- In order to better inform this decision making, we sought to evaluate the effect of HIV co-infection on non-treponemal titers during pregnancy. We hypothesized that HIV co-infection in women with syphilis is associated with decreased odds of seroreduction of nontreponemal titers during pregnancy. All ZAPPS participants were screened for syphilis and HIV at enrollment. Syphilis screening was conducted using the rapid pointof-care serologic test SD Bioline Syphilis 3.0 (Standard Diagnostics, Inc.), which is a qualitative test for treponemal-specific immunoglobulin G (IgG), IgA, and IgM antibodies. Owing to resource limitations, non-treponemal titers are not routinely performed in Zambia for either diagnosis or treatment monitoring, which is supported by the World Health Organization for low-income countries. 10 In ZAPPS, participants who tested positive for syphilis were referred for presumptive treatment at WNH-UTH, which includes three weekly intramuscular injections of Penicillin G regardless of suspected stage of disease, prior infection, or HIV status.

| MATERIAL S AND ME THODS
Inclusion criteria for this secondary analysis included women with positive syphilis testing at enrollment who had at least one additional serum specimen from later in pregnancy available for testing.
Demographics and clinical characteristics collected during the original study were available for analysis, including maternal age, marital status, education, body mass index (calculated as weight in kilograms divided by the square of height in meters; BMI), parity and obstetric history, and gestational age at enrollment. Available pregnancy outcomes included stillbirth, birth weight, delivery gestational age, basic preterm birth phenotype (spontaneous vs. providerinitiated), and pre-eclampsia.
Serum biospecimens used for this secondary analysis were tested at a single university-affiliated laboratory in March 2020. All enrollment serum biospecimens underwent two types of testing: (1) qualitative treponemal-specific testing via Treponema pallidum particle agglutination (TPPA) assay and (2) quantitative non-treponemal testing via rapid plasma reagin (RPR) titers. Final serum biospecimens for each participant also underwent RPR testing, which allowed for comparison of RPR titer between the initial (enrollment) and final specimens.
The TPPA results were reported as nonreactive or reactive. RPR results were reported as either nonreactive or reactive with a corresponding titer. Titer results were reported in the standard exponential fashion (1:1, 1:2, 1:4, 1:8, 1:16, etc.) and recorded for analysis in both this raw exponential form and in a corresponding linear form, representing the number of serial dilutions for each titer. For example, a 1:1 titer is reactive at zero dilutions, a 1:2 titer is reactive at one dilution, a 1:4 titer is reactive at two dilutions, and so on. The difference in the number of dilutions between the two samples was then calculated for each participant. Women with reactive TPPA testing but nonreactive initial RPR testing were presumed to represent previously treated infection and excluded from analysis. Women with nonreactive TPPA testing and nonreactive RPR testing were presumed to represent false positive point-of-care results and were also excluded.
Our primary outcome was achievement of seroreduction in RPR titer between the initial and final samples, defined as a reduction in the number of dilutions from the first to the final sample or by seroconversion from any reactive RPR titer to nonreactive titer.
Secondary outcomes included achievement of serologic cure (a reduction of two or more dilutions or seroconversion to nonreactive titer), RPR titer reduction (difference in dilutions between samples), rate of RPR seroreduction (dilution reduction per week), and frequency of adverse pregnancy outcomes, including stillbirth, small-for-gestational-age (birth weight less than the 10th centile), spontaneous preterm birth, and pre-eclampsia.
Demographics, clinical characteristics, and outcomes were evaluated for those with and without HIV co-infection with bivariable analysis, using χ 2 or Fisher exact test for categorical variables and Student t test or Mann-Whitney U test for continuous variables, as appropriate. Logistic regression was used to evaluate our primary outcome by HIV status, generating unadjusted and adjusted models.
Our final model was adjusted for initial RPR titer and maternal age.
We performed all analyses using STATA SE, Version 15.1 (StataCorp, College Station, TX, USA) and used a value of P less than 0.05 to determine statistical significance. In the final study population, most women were 20-34 years old, normal weight, multiparous, married, and had less than 10 years of formal education ( Table 1). The average gestational age at enrollment was 16 weeks (standard deviation [SD] 4 weeks), and the mean number of weeks between the enrollment sample and the final sample was 19 weeks (SD 6 weeks). The final sample was taken at delivery for approximately half of participants, and for the remainder, the average duration between final sample collection and delivery was 3 weeks. Baseline characteristics were similar for women with and without HIV co-infection, with the exception of maternal age, which was lower in the HIV-negative group ( Table 1).

| RE SULTS
Two-thirds of women with HIV co-infection were on antiretroviral therapy before pregnancy (18/27, 67%), and all but two of these had an undetectable viral load. Among the 11 women with detectable viral loads (11/27, 41%), the RNA copies/ml of plasma ranged from 126 copies/ml to 89 366 copies/mL. No HIV-negative women seroconverted during pregnancy.
The only clinical characteristic that differed between women who achieved a reduction in RPR titer and those who did not was HIV status ( Table 2). The mean starting titer was higher among women who achieved seroreduction than those who did not (4.1 versus 3.2), but this association did not achieve statistical significance. Among HIV-infected women, use of preconceptional antiretroviral therapy was not associated with achieving a reduction in titer.
Our primary outcome, reduction of RPR titer, occurred in 78%  Table 3). After adjusting for initial raw RPR titer and maternal age, this remained statistically significant (adjusted OR 4.66; 95% CI 1.14-19.08) ( Table 3). Change in RPR titer is shown for individual participants in Figure 2, grouped by HIV status.
Analysis of our secondary outcomes revealed that the proportion of women achieving serologic cure was higher in women with HIV co-infection (8/27, 30%) than those without (3/20, 15%), but this did not achieve statistical significance ( Table 3) HIV compared with those without, but these comparisons achieved only borderline statistical significance. There were no significant associations between adverse pregnancy outcomes and achievement of RPR seroreduction, although there was a trend toward decreased incidence of stillbirth in those achieving seroreduction (OR 0.15, 95% CI 0.01-1.58) ( Table 4). HIV co-infection was associated with higher incidence of small for gestational age, which was statistically significant and remained so after excluding stillbirths ( Table 4).

| DISCUSS ION
Our hypothesis that seroreduction of RPR titers during pregnancy is less likely to occur in pregnant women with HIV co-infection than in those without was not supported by our results. Instead, HIV co-infection in this cohort of Zambian women was associated with greater odds of titer seroreduction in pregnancy.
There are several possible explanations for these unexpected results. First, there may be differences between the two groups in stage of disease, although there was no significant difference in starting titer (which has been shown to be associated with disease stage), and results remained significant even when adjusting for starting titer. Second, adherence to syphilis treatment may have differed between groups. Although the same clinical management was used for all participants, this management included referral for treatment at a partnering clinic, and confirmation of this treatment was not captured in the ZAPPS study data. Additionally, clinicians were not blinded to HIV status during the study, and it is possible that staff or provider bias exists in ensuring treatment follow up for HIVpositive participants, who are often perceived as being at higher risk.
Even taking these possible confounders into consideration, our results should prompt re-evaluation of how HIV co-infection affects syphilis treatment response. Although our results should be interpreted with caution, they suggest that syphilis-infected pregnant women with HIV co-infection may not be at increased risk for delayed treatment response as supported by previous literature.
Our study has several strengths. The original ZAPPS study from which this secondary analysis was performed collected clinical data and biospecimens prospectively. This contemporary cohort of pregnant women with syphilis undergoing serial serologic testing is one of only a few available that contains a substantial number of patients with HIV and is by far the largest. The mean gestational age between initial and final samples (19 weeks) is a longer pregnancy duration than many of the available studies that report non-treponemal titers in pregnancy over time. Additionally, this analysis is strengthened by the use of three separate serologic tests performed for each participant to confirm diagnosis.
Our study also has limitations. These include missing clinical information, such as stage of disease, patient symptoms, confirmation of treatment, and partner treatment status. This cohort is  relationship between HIV co-infection and treatment failure should be revisited, especially in the context of an older body of literature and the improvements in HIV care over the last two decades.
Additionally, the stillbirth incidence in our final study population (9%) is notable and is more than twice that of the larger ZAPPS cohort (4%). 11 Our results show a trend toward higher

ACK N OWLED G M ENTS
The research presented in this article was funded by the Bill and

AUTH O R CO NTR I B UTI O N S
This secondary analysis was conceived by CMW, JSAS, and EMS.
Conception, data collection, and analysis of the primary study were performed by MPK, JTP, EMS, BV, and JSAS. Secondary data collection and analysis were performed by CMW with data interpretation by CMW and JTP and statistical analysis review by CAW. The manuscript was drafted by CMW, and all authors have critically appraised the manuscript and approved the final version.

CO N FLI C T S O F I NTE R E S T
The authors have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Research data are not shared. c One participant in the HIV-negative group was lost to follow up and did not have delivery data collected. d All stillbirths in this sub-study cohort occurred antepartum (none occurred intrapartum).

O RCI D
e Pre-eclampsia outcome was only available for the denominator listed.