A systematic review of syphilis serological treatment outcomes in HIV-infected and HIV-uninfected persons: rethinking the significance of serological non-responsiveness and the serofast state after therapy

Syphilis is an ancient infectious disease, yet modern efforts to control the disease remain ineffective in many countries, especially among high risk populations like men who have sex with men (MSM) [1–3]. The World Health Organization (WHO) estimated that in 2008, there were 36 million prevalent cases of syphilis and 11 million incident cases in adults between the ages of 15 and 49 [4]. Although the majority of cases have occurred in underdeveloped regions like Sub-Saharan Africa, countries like China and Russia have witnessed alarming increases in syphilis rates attributed in part to societal and economic changes [5, 6]. A significant association exists between syphilis and increased risk of HIV acquisition, with hazard ratios ranging from 2.3 to 8.6 [7–9]. A recent review of studies conducted worldwide reported a 9.5 % prevalence of syphilis among adults living with HIV infection [10].

Syphilis has been associated with protean clinical manifestations, making reliance upon laboratory diagnosis critical for the practicing physician. However, due to our inability to cultivate the causative agent of syphilis in vitro, most testing for syphilis relies on measurement of immune responses rather than on direct tests for Treponema pallidum [11]. Direct detection methods are available for early syphilitic lesions, but polymerase chain reaction (PCR) assays have limited sensitivities for detecting spirochetes from blood, especially during latent infection [12, 13]. Testing for syphilis involves serological assays based on nontreponemal and treponemal antibody responses induced by T. pallidum. Nontreponemal tests measure IgM and IgG antibodies to lipoidal antigens, principally cardiolipin, released from damaged host cells and/or T. pallidum [11, 14]. Nontreponemal antibody tests are still primarily used for syphilis screening in the United States (U.S.) and developing countries, and to monitor serological response to treatment since titers are observed to decline after effective therapy. Treponemal assays, on the other hand, measure specific IgM and IgG antibodies to T. pallidum proteins and have been traditionally utilized to confirm reactive nontreponemal tests.

Since 1993, U.S. treatment guidelines have regarded a four-fold (or two dilution; e.g., 1:64 to 1:16) decline in nontreponemal antibody titers or seroreversion to negative as an indicator of an appropriate serological response after treatment [15, 16]. This recommendation was based on a study by Brown, et al. [17], in which the investigators generated curves demonstrating a four-fold decline in Venereal Disease Research Laboratory (VDRL) titers at three months following therapy, using data from patients who had symptom resolution after treatment of primary (PS) and secondary syphilis (SS). However, some patients in clinical practice do not follow the classical patterns of serological response to therapy, exhibiting less than a four-fold decline in nontreponemal titers and/or persistently low positive titers without evidence of treatment failure or reinfection. These conditions have been referred to as “serological failure,” “serological non-response,” “seroresistance,” “reagin-fast,” or the “serofast state.”

In addition to the lack of consensus regarding the terminology used to refer to nontreponemal antibody titers that do not meet exhibit an appropriate serological response after therapy, there is also uncertainty regarding whether these conditions indicate persistent infection or a residual immune response in the absence of viable organisms. Although retreatment may be beneficial for some of these patients, there are no specific markers that can differentiate those requiring additional therapy to prevent subsequent sequelae. Untreated or inadequately treated syphilis can result in severe neurological and cardiovascular complications, stillbirth, neonatal death, and irreversible congenital abnormalities [18, 19]. Not surprisingly, considerable controversy exists whether patients who do not demonstrate an appropriate serological treatment response should undergo lumbar punctures to evaluate for neurosyphilis or be retreated.

A large number of studies have been conducted to investigate serological responses after treatment of syphilis in HIV-infected and HIV-uninfected individuals. Therefore, we performed a systematic review to evaluate the frequency of serological outcomes following syphilis therapy and correlates associated with serological cure. This review illustrates the extensive observational data that have accrued regarding lack of serological cure after treatment, and highlights the need for more definitive studies regarding the immunological and clinical significance of persistent nontreponemal antibodies over time. We hypothesize that these persistent antibodies are associated with a failure of immune tolerance rather than lack of pathogen clearance after recommended therapies.

T. pallidum cannot be cultured in vitro, thus forcing reliance on serological tests for diagnosis and proof of cure. This practice has been in place for nearly a century due to the absence of novel technologies to confirm eradication of the organism. Unfortunately, our systematic review of 20 studies conducted worldwide among HIV-infected and HIV-uninfected patients with syphilis demonstrates that a substantial proportion (12.1 %) of patients exhibit serological non-response after treatment, which is time-dependent. The median proportion of serological non-response patients was 20.5 % at six months, which decreased to 11.2 % at ≥ 12 months for all stages of syphilis. Our findings highlight the importance of developing a consistent definition of serological non-response, which is not an absolute or definite condition but should be thought of as a continuum with serological cure and the serofast state over time. The presence of persistent nontreponemal antibodies may represent treatment failure, reinfection, or another undefined immune phenomenon; however, none of the studies in our review could distinguish among these possibilities when neither a four-fold rise in nontreponemal antibody titers nor a history of re-exposure was evident. The appropriate time to determine falling titers, and the correlates that affect the time when titers fall, remains unclear. Serological cure was less likely in older individuals, patients with lower baseline nontreponemal antibody titers, and patients with later stages of syphilis. The clinical or biological explanations for these observations have not been well established. Perhaps more importantly, there are no consistent criteria by which patients can be stratified to determine their need for additional evaluation or syphilis therapy when serological non-response exists.

Nontreponemal antibodies have played a pivotal role in syphilis screening and monitoring of treatment outcomes; yet the mechanism by which they arise is poorly understood. The first serological test for syphilis was developed in 1906 as a complement fixation assay by Wasserman,et. al. [42], using antigens from human and monkey extracts rich in T. pallidum. Other investigators soon discovered that antibodies from syphilis patients reacted not just with antigens from spirochetes but also with cardiolipin found in normal nonsyphilitic tissues. Pangborn subsequently purified cardiolipin from beef heart extract [43], which in combination with lecithin and cholesterol, serves as the antigen for current nontreponemal antibody tests. The central question is whether the cardiolipin antibodies detected in syphilis patients are a response specific to the spirochete, since the lipid composition of purified T. pallidum has been identified as 13 % cardiolipin [14, 44]. However, individuals who have no other evidence for syphilis infection have been found to have anticardiolipin antibodies similar to those detected in syphilitic patients, leading to the hypothesis that this class of antibodies may be generated in response to destruction of host tissues [45]. Nontreponemal antibodies have been observed in persons with a variety of infectious and autoimmune diseases including HIV, hepatitis, and systemic lupus erythematosus [46–48], and these potentially false positive tests continue to cause problems in diagnosis and management.

The possibility that persistent nontreponemal antibodies after treatment represent persistence of T. pallidum was raised by early investigators, based on findings from animal and human studies conducted in the 1960s [49, 50]. These studies identified treponemes in lymph nodes and CSF of rabbits one year after therapy for LLS, and reported similar findings in human subjects [49]. However, subsequent investigations questioned whether these treponemes were viable, since the transfer of material from the lymph nodes of some treated syphilis patients produced no detectable disease upon inoculation into rabbits [50]. Studies of the natural history of syphilis have demonstrated that T. pallidum can disseminate to multiple organs in humans within hours of inoculation [46]. T. pallidum is then known to infect various anatomical sites, including the central nervous system and eye as “immune privileged” areas where the spirochetes can slowly replicate and evade the immune response [46]. Symptomatic and asymptomatic neurosyphilis has been reported in a case series of serofast patients previously treated for ES [51]; however, the study involved only a small number of patients and is contrary to the findings by Fiumara [22].

In order to address the question of persisting spirochetes after therapy, highly sensitive direct detection methods for T. pallidum would be essential for analyzing patient specimens before and after treatment. Unfortunately, a meta-analysis of studies evaluating the performance of T. pallidum PCR on samples taken during different stages of infection reported pooled sensitivities as low as 31.2 % and 47.4 % in blood and CSF specimens, respectively [52], illustrating their limitations in ruling out residual spirochetes in previously treated patients. In the absence of a readily available test of cure based on detection of the organism, there is no other definitive method to detect persistent T. pallidum in serological non-responders or serofast patients. This suggests that other means, such as analysis of antigen-specific immune responses that occur during infection and after therapy, may be a more pragmatic approach to investigating these conditions.

The hypothesis that a lingering immune response is responsible for persistent nontreponemal antibodies can be argued based on a number of key observations. We found that 11 % of patients still exhibit serological non-response at ≥ 12 months after therapy; yet reported tertiary syphilis cases are still relatively uncommon worldwide. Thus, the observed proportion of serological non-response is not likely to be associated with a high frequency of cardiovascular or neurological complications as might be expected, given the global rate of syphilis. PCN has been considered to be highly effective since it was first reported for treatment of syphilis [53]. T. pallidum is extremely sensitive to minute amounts of PCN in vitro and in vivo if maintained for sufficient periods of time [54, 55], and presently, there are no data to suggest that the spirochete has developed PCN resistance. Furthermore, there are no clear benefits to providing additional therapy for serological non-responders or serofast patients although they may repeatedly undergo retreatment in clinical practice. A recent study involving ES patients who had serological non-response after initial therapy and received an additional dose of benzathine PCN demonstrated that 73 % still failed to have an appropriate serological response and only 2 % seroreverted at 12 months despite retreatment [56]. These findings imply that persistent nontreponemal antibody titers may not be due to insufficient therapy, but rather an alternative process like the immune response to treponemal infection. Presently, there have been no other studies demonstrating that retreatment of patients who are serological non-responders leads to improved outcomes.

Treponemal IgM and IgG antibodies have already been demonstrated to persist in syphilis patients after therapy [57–59], indicating the presence of plasma cells that continue to produce T. pallidum specific antibodies. This kind of pathogen-specific memory response is seen after both infections and vaccinations. Long-term study of human volunteers has shown that pathogen-specific memory responses have very slow decay rates with apparent half-lives ranging from years to millennia [60]. Therefore, the persistence of treponemal antibodies is expected despite clearance of infection after treatment.

In contrast, nontreponemal antibodies decline after therapy in the majority of syphilis patients, indicating that they are down-regulated after clearance of T. pallidum. These antibodies are autoreactive since cardiolipin, the main antigenic component that binds to nontreponemal antibodies, is present in mitochondrial membranes. Disappearance of these autoreactive antibodies after syphilis therapy leading to seroreversion suggests the involvement of tolerance control. In serological non-response and the serofast state, there is an apparent failure of the expected process for down-regulation. Persisting nontreponemal antibodies after effective therapy may thus represent failure of immune tolerance rather than lack of pathogen clearance. Recent work on HIV has shown that many antibodies against the virus are autoreactive and under tolerance controls [61–67], and careful study of these antibodies has provided clues to their origin and regulation. Understanding the nature of nontreponemal antibodies in syphilis patients who do not have an appropriate serological response could shed light on their origins, and allow differentiation between antibodies indicative of persistent infection or mere antibody persistence in the absence of infection.

Our review found conflicting evidence on the relationship between demographic, clinical factors, HIV status and serological response. The associations of syphilis stage and baseline nontreponemal antibody titers with serological treatment outcomes appear to be the most consistent among the studies in this review. Only two out of nine studies reported that HIV-infected status decreased the likelihood and time to serological cure [23, 37]. Markers of immune function such as CD4 count and HIV viral load also do not appear to affect serological outcomes among HIV-infected persons with syphilis [28, 32, 39, 40]. Patients with HIV infection have well described dysfunction of antibody formation [68], and the former finding may be due to this B cell dysfunction [69] in which there is less effective immune clearance of T. pallidum in HIV-infected persons. The epidemiological connection between HIV and syphilis is well-established [70, 71], but an understanding of the biological and immunological relationships between these co-infections remains obscure.

Several limitations are evident from this systematic review. First, there was no standard definition for serological non-response that was consistent. There was heterogeneity in the classifications of serological outcomes that did not meet the criteria for serological cure, which is partly related to the lack of consensus of the U.S. and European Union guideline definitions [41, 72]. This discordance presented challenges in the overall synthesis of results, and may have affected our estimates. Secondly, there were variations in follow-up times when a patient was considered to have serological non-response after treatment. Longer follow-up periods are necessary in order to determine serological outcomes over time; however, this must be balanced by the challenges of having patients return for serological monitoring years after completion of therapy. In general, most of the studies defined ELS as syphilis acquired < 1 year prior according to U.S. and European CDC criteria rather than < 2 years [41, 72]; however, there is a potential that some LLS cases were included as early cases resulting in lower response rates due to non-discriminatory misclassification. Only a few studies in our review provided data on the serofast state, treatment failures or reinfections; therefore, our estimates about the frequency of these conditions were limited. Finally, there has only been a handful of prospective studies to assess serological outcomes following treatment of syphilis. The majority of the studies reviewed were retrospective in study design, which presents disadvantages due to the potential for selection bias and misclassification as a result of non-standardized data collection and follow-up periods after treatment.

Our results highlight a fundamental problem in the management of T. pallidum infections that has prevailed over decades. Clinicians have been taught to expect at least a four-fold decline in nontreponemal antibody titers or seroreversion to following therapy; however, serological non-response occurs in nearly one out of every eight patients with syphilis after treatment. This significant proportion and our clinical observations regarding serological non-response and the serofast state suggest the following plausible scenarios that patients may eventually exhibit serological response or serorevert over time (e.g. as in LLS); patients may be getting reinfected at a high rate; patients may not be achieving 100 % cure from PCN; or that patients with persistent nontreponemal antibody titers have neither reinfection nor persistent T. pallidum but instead, an altered immune process in which their anticardiolipin antibodies are not down-regulated. Given the diversity of data presented here, it is conceivable that each of these possibilities holds true for a subset of patients, and we presently lack sufficient information to differentiate among these or other scenarios. There have been vast improvements in methods for detection of microorganisms, and in understanding the host response to different infections. Unfortunately, modern innovative technologies have not yet been applied to the study of syphilis. Based on the data from our systematic review, we do not believe that further retrospective studies of serological response to recommended therapies will be useful or informative. Rather, coordinated investigations involving novel B cell methods, direct detection assays for T. pallidum, and correlations with nontreponemal and treponemal antibody tests in syphilis-infected patients are essential to advancing clinical management of this formidable pathogen.

Syphilis remains a global public health threat, whose clinical management is highly dependent on nontreponemal antibody titers. Unfortunately, the basis for the cardiolipin antibodies detected during syphilis infection is poorly understood, and our systematic review affirms that a significant proportion of both HIV-infected and HIV-uninfected patients will have serological non-response after therapy. These patients may have reinfection, treatment failure, or a benign immune response. In order to explore these possibilities and inform clinical practice, efforts must be directed towards designing prospective syphilis studies and utilizing modern technologies to determine the microbiological and immunological basis for these antibodies.