Figure 1. Flowchart of study participation. FHDH indicates French Hospital Database on HIV; HIV, human immunodeficiency virus; and MSM, men who have sex with men.
Figure 2. Impact of syphilis on plasma viral load (pVL) increase in various subgroups. Results are from logistic regression with random effects and are adjusted for age. A VL increase was defined as either an increase in the plasma human immunodeficiency virus RNA level of more than 0.5 log in patients with a baseline VL of at least 500 copies/mL or a value of at least 500 copies/mL in patients with a baseline VL of less than 500 copies/mL. aOR indicates adjusted odds ratio; and cART, combination antiretroviral therapy.
Figure 3. Adjusted differences in CD4 cell counts between patients with syphilis and syphilis-free patients. Results are selected from the multivariate linear mixed model and are given as mean (SD) (see also eTable 1).
Jarzebowski W, Caumes E, Dupin N, et al. Effect of early syphilis infection on plasma viral load and CD4 cell count in human immunodeficiency virus–infected men: results from the FHDHANRS CO4 cohort. Arch Intern Med.. Published online July 23, 2012. doi: 10.1001/archinternmed.2012.2706.
eAppendix. Supplementary methods.
eFigure. Definition of periods according to the index date in patients with syphilis and syphilisfree patients.
eTable 1. Impact of Syphilis on the CD4 Cell Count: Results of the Multivariate Linear Mixed Model
eTable 2. Mean CD4 Cell Count Variations at the Index Date According to Syphilis Status and According to the Occurrence of a VL Increase in the Following 6 Mo
This supplementary material has been provided by the authors to give readers additional information about their work.
Witold Jarzebowski, Eric Caumes, Nicolas Dupin, David Farhi, Anne-Sophie Lascaux, Christophe Piketty, Pierre de Truchis, Marie-Anne Bouldouyre, Ouda Derradji, Jérome Pacanowski, Dominique Costagliola, Sophie Grabar, for the FHDH-ANRS CO4 Study Team. Effect of Early Syphilis Infection on Plasma Viral Load and CD4 Cell Count in Human Immunodeficiency Virus–Infected MenResults From the FHDH-ANRS CO4 Cohort. Arch Intern Med. 2012;172(16):1237–1243. doi:10.1001/archinternmed.2012.2706
Author Affiliations: Assistance Publique-Hôpitaux de Paris, Groupe Hospitalier Cochin Broca Hôtel-Dieu, Unité de Biostatistique et Epidémiologie, Paris, France (Drs Jarzebowski and Grabar); INSERM U943, Paris (Drs Jarzebowski, Costagliola, and Grabar); Université Pierre et Marie Curie Paris 6, Paris (Drs Caumes and Costagliola); Assistance Publique-Hôpitaux de Paris, Hôpital Pitié-Salpétrière, Service de Maladie Infectieuses et tropicales, Paris (Drs Caumes and Costagliola); Université Paris Descartes, Paris (Drs Dupin, Farhi, and Grabar); Assistance Publique-Hôpitaux de Paris, Hôpital Tarnier, Service de dermatologie et vénérologie and Centre National de la Syphilis, Paris (Drs Dupin and Farhi); Assistance Publique-Hôpitaux de Paris, Hôpital Henri Mondor, Service d’immunologie clinique, Créteil, France (Drs Lascaux); Assistance Publique-Hôpitaux de Paris, Hôpital Européen Georges Pompidou, Service d’immunologie clinique, Paris (Dr Piketty); Assistance Publique-Hôpitaux de Paris, Hôpital Raymond Poincaré, Service de médecine aigue spécialisée, Garches, France (Dr de Truchis); Centre Hospitalier Intercommunal Robert Ballanger, Service de médecine interne et de maladies infectieuses, Aulnay-sous-Bois, France (Dr Bouldouyre); Assistance Publique-Hôpitaux de Paris, Hôpital Paul Brousse, Service des maladies infectieuses et tropicales, Le Kremlin-Bicêtre, France (Dr Derradji); and Assistance Publique-Hôpitaux de Paris, Hôpital St-Antoine, Service des maladies infectieuses et tropicales, Paris (Dr Pacanowski).
Group Information: A list of the FHDH-ANRS CO4 Study Team members can be found at http://www.ccde.fr.
Background Concomitant syphilis and human immunodeficiency virus (HIV) infection is increasingly frequent in industrialized countries.
Methods From a large hospital cohort of HIV-infected patients followed up in the Paris area between 1998 and 2006, we examined the effect of early syphilis on plasma HIV-1 RNA levels and CD4 cell counts. We compared 282 HIV-1–infected men diagnosed as having incident primary or secondary syphilis with 1233 syphilis-free men matched for age (±5 years), sexual orientation, participating center, length of follow-up (±6 months), and immunologic and virologic status before the date of syphilis diagnosis (index date). Increase in viral load (VL) (plasma HIV-1 RNA) of at least 0.5 log or a rise to greater than 500 copies/mL in patients with previously controlled VL during the 6 months after the index date was analyzed, as were CD4 cell count variations and CD4 slope after the index date.
Results During the 6 months after the index date, VL increase was observed in 77 men with syphilis (27.3%) and in 205 syphilis-free men (16.6%) (adjusted odds ratio [aOR], 1.87; 95% CI, 1.40-2.49). Even in men with a VL of less than 500 copies/mL undergoing antiretroviral therapy, syphilis was associated with a higher risk of VL increase (aOR, 1.52; 95% CI, 1.02-2.26). The CD4 cell count decreased significantly (mean, −28/μL) compared with the syphilis-free group during the syphilis episode (P = .001) but returned to previous levels thereafter.
Conclusions In HIV-infected men, syphilis was associated with a slight and transient decrease in the CD4 cell count and with an increase in VL, which implies that syphilis may increase the risk of HIV transmission, even in patients receiving antiretroviral therapy and with a VL of less than 500 copies/mL.
Human immunodeficiency virus (HIV) and Treponema pallidum, the causative agent of syphilis, are sexually transmitted. In industrialized countries, the incidence of syphilis fell markedly in the 1990s1 because of simple preventive measures, behavioral changes, and better access to screening during the early years of the HIV pandemic. A resurgence of syphilis was noted in the 2000s in Europe and North America, mainly in men who have sex with men (MSM).2,3 In France, where mandatory notification of syphilis was abrogated in 2000, a syphilis surveillance network confirmed the resurgence of syphilis in MSM, particularly in Paris,4 up to 2007.5,6
Interactions between syphilis and HIV infection are not fully documented.7,8 Syphilis causes genital lesions that are known to increase the risk of HIV transmission.9,10 Several studies11- 16 have examined the effect of syphilis on HIV viral load (VL) and the CD4 cell count in the era of combination antiretroviral therapy (cART), but they gave conflicting results. Most of these studies were small, had limited follow-up, and did not account for the effect of cART.
Given the resurgence of sexually transmitted infections in MSM, it is important to examine the effect of syphilis on HIV infection for patient management and for public health reasons. We, therefore, studied the independent effect of syphilis on HIV VL and CD4 cell counts in HIV-infected men followed in the French Hospital Database on HIV (FHDH).
From the FHDH cohort, we selected all HIV-1–infected men followed in the Paris region (Ile-de-France) between January 1, 1998, and December 31, 2006. This region was chosen because a French syphilis outbreak that began in 2001 was concentrated in the Paris area4 and because we needed access to the patients' medical records to validate the diagnoses of syphilis.
Briefly, the FHDH is a nationwide hospital-based clinical cohort created in 1989.17 In the Paris region, 24 hospitals contribute data on HIV-infected patients. The only FHDH inclusion criteria are HIV-1 or HIV-2 infection and written informed consent. Sociodemographic, clinical, therapeutic, and laboratory data are collected at least every 6 months. The FHDH has been approved by the French computer watchdog body (Commission Nationale de l’Informatique et des Libertés).
Cases of syphilis were extracted from the FHDH, which uses International Classification of Diseases, Tenth Revision, codes. For code A53.9 (syphilis unspecified as early or late) and for 10% of other cases randomly selected with a code corresponding to primary or secondary syphilis (codes A51.0 to A51.4), trained research assistants examined the medical records to determine whether the patient had primary or secondary syphilis. As 91% of cases with codes A51.0 to A51.4 were confirmed, no further checking was organized for those cases. Of the cases coded A53.9, 42% were ascertained as primary/secondary syphilis and were included. Early neurosyphilis (n = 6) and ocular syphilis (n = 5) were not included in this analysis because their treatment differs from that of other forms of early syphilis.
The index date was the date of syphilis diagnosis. For each patient, syphilis status was divided into 3 periods (eFigure). The index period was the period of active syphilis, from the beginning of the incubation period (estimated to be 45 days before diagnosis of primary syphilis and 90 days before diagnosis of secondary syphilis18) until cure (estimated to be 2 months after diagnosis). As in other studies,11,12,14 the length of the index period had to be wide enough to include routine HIV variables but narrow enough to capture those affected by syphilis.
We studied 2 matched groups of HIV-1–infected men: patients with syphilis and syphilis-free patients (reference group). The syphilis group consisted of men diagnosed as having incident primary or secondary syphilis after enrollment in the FHDH. The reference patients were selected from the FHDH by using an incidence density sampling approach.19 Each patient with syphilis was matched with up to 5 syphilis-free patients followed in the same FHDH participating center at the index date (±6 months) according to age (±5 years), sexual orientation (MSM vs others), immunologic status (mean CD4 cell count in the year before the index period: <200, 200-349, 350-499, or ≥500/μL), and virologic status (mean plasma VL in the year before the index period: <500, 500-4999, 5000-29 999, and ≥30 000 copies/mL). Each patient was followed for 2 years before and 2 years after the index period and was required to have at least 2 available CD4 cell counts before the index period, 1 during, and 2 after.
Sociodemographic, immunologic, and virologic characteristics before the index period were compared by using univariate conditional logistic regression analysis. To assess the impact of syphilis on HIV plasma VL, baseline VL was defined as the last value obtained before the index period. A VL increase was defined as either an increase in the plasma HIV RNA level of more than 0.5 log in patients with a baseline VL of at least 500 copies/mL or a value of at least 500 copies/mL in patients with a baseline VL of less th an 500 copies/mL. We considered only the first virologic increase in the 6 months after the index date in patients with available VL in this time frame. Logistic regression with random intercepts to account for matching was used to estimate odds ratios (ORs) and 95% CIs associated with VL increase in patients with syphilis vs syphilis-free patients, overall and in various subgroups.20
Linear mixed models were used to compare patients with syphilis and syphilis-free patients who experienced a VL increase in terms of the first available VL in patients with a baseline VL of less than 500 copies/mL or the change in VL from baseline in patients with a baseline VL of at least 500 copies/mL. Time from VL increase to a VL of less than 500 copies/mL or below the baseline VL was estimated using the Turnbull estimation for interval-censored data.21 Differences between groups were assessed using a generalization of the log-rank test without stratification as the matching was no longer maintained in this subgroup. The impact of syphilis on CD4 cell counts and slopes during and after the index period was studied by analyzing longitudinal CD4 values in a multivariate linear mixed model. As the effect of cART on the CD4 cell count depends on age,22 we adjusted this analysis for age, cART status, and the interaction between age and cART status (eAppendix).
Finally, based on the results of the previous models, we evaluated the relative effects of VL increase and syphilis on the change in CD4 cell counts during the index period using linear mixed models assessing the effects of syphilis group, VL increase, and the interaction between syphilis and VL increase. All the tests were 2-sided, and P < .05 was considered statistically significant. Commercially available (SAS; SAS Institute, Inc) and free (R) software programs were used for all the analyses.
A total of 628 men followed in the Paris region and diagnosed as having primary or secondary syphilis between January 1, 1998, and December 31, 2006, were identified in the FHDH (Figure 1). Of these, 230 men had already been diagnosed as having syphilis at inclusion in the FHDH, 95 had too few available CD4 cells, and the primary/secondary nature of syphilis could not be validated in 21 cases because the medical records could not be retrieved. Therefore, 282 men with an incident diagnosis of primary (n = 64) or secondary (n = 218) syphilis fulfilled the inclusion criteria. The matched reference group consisted of 1233 HIV-1–infected men without syphilis.
The main characteristics of the population are given in Table 1. Most patients were MSM (89%), and their median age was 38 years. Before the index date, 85% of the patients were undergoing cART. The median CD4 cell count was 481/μL, and the VL was less than 500 copies/mL in 58% of patients. The syphilis and syphilis-free groups were well matched, except for age, and all the models were, therefore, adjusted for age as a continuous variable. Patients with secondary syphilis had slightly more advanced HIV disease than did those with primary syphilis: 19% and 8% of patients, respectively, had already had an AIDS-defining event when diagnosed as having syphilis (P = .03).
Eighteen patients in the reference group and 3 in the syphilis group had no available VL values in the 6 months after the index date and were not included in this analysis. A VL increase occurred during the first 6 months after the index date in 77 patients (27.3%) in the syphilis group and in 205 patients (16.6%) in the syphilis-free group. Regardless of the group, three-quarters of the VL increases occurred less than 2 months after the index date (P = .84).
Syphilis was significantly associated with VL increase, overall (adjusted OR [aOR], 1.87; 95% CI, 1.40-2.49) and in the different subgroups (Figure 2). The risk of VL increase was 1.56 (95% CI, 1.06-2.30) in patients with a VL of less than 500 copies/mL and 1.52 (95% CI, 1.02-2.26) in cART-treated patients with a VL of less than 500 copies/mL. The risk of VL increase associated with syphilis was higher in patients who were not taking cART at the index date than in patients receiving cART (P = .04): the aORs were 3.61 (95% CI, 1.85-7.06) and 1.65 (95% CI, 1.19-2.28), respectively. The risk of VL increase did not differ significantly between patients with primary and secondary syphilis (P = .34).
Table 2 describes the VL increase. Among patients who experienced VL increase, the median VL at the time of the increase was 54 000 copies/mL in the syphilis group and 11 318 copies/mL in the syphilis-free group (P = .01). The median duration of VL increase was not significantly different between the syphilis and syphilis-free groups, but in patients with a baseline VL of less than 500 copies/mL, it was shorter in the syphilis group (1.6 and 4.8 months, respectively; P = .01).
Results of the multivariate mixed model are shown in Figure 3, and the full results are given in eTable 1. Patient age and the duration of cART were significantly associated with changes in the CD4 cell count or slope. No period effect was found in the syphilis-free group. Compared with the syphilis-free group, the CD4 cell count decreased significantly (mean [SD]: −28/μL [8/μL]) during the syphilis episode (P = .001) and returned to previous levels (mean [SD]: 25/μL [10/μL]) after the episode, but the difference was no longer significant (−3/μL, P = .78). No significant difference was noted between the groups regarding the slope of the CD4 cell count after the episode (P = .56). The effect of syphilis was similar regardless of the CD4 cell count before syphilis diagnosis (<350/μL or ≥350/μL), the primary or secondary nature of syphilis, the baseline VL (<500 or ≥500 copies/mL), and the study period (before or after 2004) (data not shown).
As syphilis was associated with a decline in CD4 cells and an increase in VL, we tested the interaction between syphilis and VL increase with respect to the CD4 cell count (eTable 2). Compared with patients who did not experience VL increase, the decrease in the CD4 cell count at the index date was larger in patients who experienced VL increase in the first 6 months after the episode (7/μL vs −35/μL, P < .001). In patients who experienced VL increase, the decrease in the CD4 cell count at the index date was larger in patients with syphilis (−70/μL) than in syphilis-free patients (−22/μL) (P = .02). No significant interaction was found between syphilis and VL increase (P = .34).
The results of this study of a large cohort of HIV-1–infected men in France indicate that early syphilis is associated with a nearly 2-fold increase in the risk of VL elevation in the months after diagnosis, even in patients receiving effective cART, and also with a moderate and transient decrease in the CD4 cell count.
These results confirm the increased risk of VL elevation in patients with syphilis (aOR, 1.9) observed in some studies11,12,16 but not in others.13,14 However, previous studies included fewer patients (38-118 patients) and had less robust designs, sometimes with no control group. The strength of this cohort study is that it enabled us to study time-specific changes related to syphilis infection by comparing matched patients with and without syphilis. This design helps circumvent the usual bias of observational studies regarding causal inferences. The specific relationship between syphilis and VL increase is suggested by the following results. First, VL increase was more frequent in patients with syphilis than in syphilis-free controls. Second, in patients with a baseline VL of less than 500 copies/mL, the duration of VL increase was shorter in patients with syphilis, suggesting that it was directly related to the syphilis episode, which was rapidly cured with the use of antibiotic agents that rapidly clear T pallidum. Third, the independent association between CD4 cell count decline and VL increase suggests that syphilis, similar to other infections, may activate viral replication, which, in turn, may affect CD4 cell production or turnover.
Although the risk of VL increase associated with syphilis was higher in patients without cART (aOR, 3.61) than in patients receiving cART (aOR, 1.65), note that syphilis had an effect on VL even in cART-treated patients with a VL of less than 500 copies/mL before diagnosis (aOR, 1.52). These results have important implications for HIV transmission, particularly regarding the recent concept of “treatment as prevention,” which may inappropriately lead patients to relax their sexual precautions when they are treated and have a controlled VL. Indeed, these data indicate that syphilis increases HIV VL and may, therefore, increase the risk of transmission, even when VL was previously controlled by cART. Semen HIV RNA levels correlate with plasma HIV RNA levels, and even a small increase in plasma HIV RNA levels increases the risk of HIV transmission.23 According to the results of a model-based study,24 a plasma HIV RNA level of 3600 copies/mL, as observed herein during the syphilis episode in patients with controlled VL, is associated with a rate of HIV transmission to sexual partners of approximately 15 per 100 person-years.
The present study population consisted mainly of MSM infected by syphilis, which is often used as a proxy for at-risk sexual behavior. This population is not covered by the Swiss guideline regarding HIV transmission during effective ART,25 which concerns couples in stable relationships and who have no other sexually transmitted infections. The present results indicate that populations with high-risk sexual behavior, in whom syphilis reinfection is relatively common,26,27 must be warned of the risk of HIV transmission and be advised to use condoms. Physicians should also bear in mind the possibility of syphilis when a patient receiving cART has an unexplained increase in VL or a decrease in the CD4 cell count. Because syphilis is often clinically misdiagnosed, its early diagnosis and, thereby, early treatment rely on physicians and patients8 who should be aware of its effect on HIV variables.
We found that the impact of syphilis on VL was transient, as half of the episodes resolved within 5 months, suggesting a modest individual impact regarding its public health effect. In patients with a VL of less than 500 copies/mL, the VL increase duration was shorter in patients with syphilis than in syphilis-free patients. This might be due to early effective treatment of syphilis,28 whereas in syphilis-free patients, the cause of the VL increase may be less easy to eradicate. We observed a clear temporal relationship between syphilis infection and VL elevation, followed by rapid VL normalization compatible with effective treatment of syphilis; however, the impact of syphilis treatment on the subsequent decline in VL was not addressed herein as it is not systematically recorded in the FHDH.
Regarding the impact of syphilis on the CD4 cell count, a moderate decrease (approximately −30/μL) was observed during the syphilis episode, followed by a return to baseline counts. The decrease was similar regardless of the syphilis, the CD4 cell count, or the VL. Similar results have been obtained in other studies.13,14 Two major mechanisms could contribute to this CD4 cell count decline: upregulation of HIV replication29 and the immunologic response to syphilis.30,31 These latter 2 studies showed lower CD4 cell counts and higher CD8 cell counts in patients with syphilis than in patients without syphilis. Fan et al30 attributed the decline in CD4 cell counts to Fas-mediated apoptosis. Other researchers32 have observed significant increases in circulating memory and memory/effector T cells, consistent with an exuberant cellular response to Treponema antigens during the systemic phase of syphilis.
The decrease in the CD4 cell count observed at syphilis diagnosis was independently associated with the increase in plasma HIV RNA levels and with syphilis infection. Higher HIV VL may be a consequence of immune activation and subsequent amplification of HIV replication in the blood.11,12 Cell activation may enhance the surface expression of HIV receptors and coreceptors (CD4, CCR5, and CXCR4), facilitating HIV entry. In addition, transcription of viral genes may be upregulated by cytokines expressed in response to coinfection.
Patients with syphilis exhibited a transient decline in their CD4 cell count. In patients with CD4 cell counts close to 200/μL, the decline associated with syphilis could precipitate the occurrence of an AIDS event. In contexts of limited follow-up and access to ART, in countries where syphilis diagnosis and treatment can be problematic, the impact of syphilis on the course of HIV disease is likely to be more severe than that observed in this study.
This study was limited to patients with primary and secondary syphilis and excluded those with early ocular and neurosyphilis. However, it is likely that the effects observed herein would be even more marked in patients with more severe forms of early syphilis.33,34 We did not check for concomitant sexually transmitted infections that might also have explained the VL increase.35 Another limitation is that some syphilis-free patients might have contracted syphilis without being diagnosed or recorded in the database. However, such a classification error would tend to bias the results toward a reduced effect of syphilis on HIV variables.
In conclusion, these findings indicate that early syphilis in HIV-infected patients is associated with an increase in plasma HIV RNA levels, even in patients with a plasma VL of fewer than 500 copies/mL while receiving cART. This increase may raise the risk of HIV transmission. Syphilis was also associated with a transient decrease in the CD4 cell count. These results call for continuing efforts to promote safer sex.
Correspondence: Sophie Grabar, MD, PhD, Unité de Biostatistique et Epidémiologie, Groupe Hospitalier Cochin Broca Hôtel-Dieu, Hôtel-Dieu, 1 place du Parvis Notre-Dame, 75181 Paris CEDEX 4, France (email@example.com).
Accepted for Publication: May 5, 2012.
Published Online: July 23, 2012. doi: 10.1001/archinternmed.2012.2706
Author Contributions: Drs Jarzebowski, Caumes, Costagliola, and Grabar had full access to all the data of the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Costagliola and Grabar. Acquisition of data: Jarzebowski, Dupin, Farhi, Lascaux, Piketty, de Truchis, Bouldouyre, Derradji, and Pacanowski. Analysis and interpretation of data: Jarzebowski, Caumes, Dupin, Costagliola, and Grabar. Drafting of the manuscript: Jarzebowski, Caumes, Costagliola, and Grabar. Critical revision of the manuscript for important intellectual content: Jarzebowski, Dupin, Farhi, Lascaux, Piketty, de Truchis, Bouldouyre, Derradji, Pacanowski, Costagliola, and Grabar. Statistical analysis: Jarzebowski, Costagliola, and Grabar. Administrative, technical, and material support: Costagliola. Study supervision: Caumes, Costagliola, and Grabar.
Financial Disclosure: None reported.
Funding/Support: The FHDH-ANRS CO4 is supported by Agence Nationale de Recherches sur le SIDA et les Hépatites, Institut National de la Santé et de la Recherche Médicale, and the French Ministry of Health.
Role of the Sponsors: The study sponsors had no role in the study design, data collection, data analysis, data interpretation, or writing of the article.
Previous Presentation: This original work was presented in part as a poster at the 10th International Congress on HIV and Drug Therapy in HIV Infection; November 7-11, 2010; Glasgow, Scotland.
Additional Contributions: We thank all the FHDH participants and research assistants, in particular Lydie Beniguel, PhD, Sandra Firmin, MSc, Mohamed Medjkane, MD, Selma Trabelsi, MSc, and Sara William-Faltaos, PharmD, PhD.