PROMPT DIAGNOSIS AND TREATMENT of ulcerative sexually transmitted diseases (STDs) has been shown to reduce morbidity and to decrease the risk of HIV transmission.1–3 Ideally, prompt treatment of genital ulcer disease (GUD) should be administered after etiologic diagnosis, at the time of initial clinical presentation. However, in most low-resource settings, same-day laboratory tests are often not available as a result of heavy caseloads, lack of laboratory equipment or supplies, and in some instances, limited clinical training of laboratory personnel. Consequently, syndromic criteria have provided a rational approach to ensure prompt management of GUD worldwide.4
Pathogens most frequently associated with GUD are Treponema pallidum, Haemophilus ducreyi, and herpes simplex virus (HSV). The prevalence of these pathogens should guide syndromic diagnosis; however, prevalence varies both by geographic region and over time.5,6 Until the early 1990s in Southern Africa, most GUD was attributable to H. ducreyi and T. pallidum,5–8 and the contribution of HSV was negligible. Hence, in the absence of vesicular lesions, most GUD was treated syndromically for both syphilis and chancroid.4,7 However, recent evidence suggests that the incidence of GUD has generally decreased in the region, whereas the proportion attributable to HSV has increased.5,7
We have previously studied the cause of genital ulcers in the southern African country of Malawi so as to develop country-specific STD guidelines.7,8 The current report summarizes the results of a more recent cross-sectional study conducted in STD patients at Lilongwe Central Hospital, Malawi, to monitor the prevalence of HSV, H. ducreyi, and T. pallidum infection so as to reevaluate the syndromic guidelines.
Between December 1998 and April 1999, 599 consecutive STD patients were treated syndromically at Lilongwe Central Hospital STD Clinic. One hundred forty-six patients (24%) had genital ulcers. Informed consent was obtained from all 146 participants. Nine were excluded because they had phimosis, precluding adequate examination and measurement of the ulcer (n = 3), or the ulcer specimen tubes cracked before analysis (n = 4), or participants had incomplete questionnaires (n = 2). Of the remaining 137 participants, 3 had T. pallidum detected by polymerase chain reaction (PCR) and were included in computing prevalence estimates, although they were excluded from analyses when algorithm were developed. All treatment was dispensed according to standard protocol. Before enrollment, the study protocol was approved by the Institutional Review Boards in Malawi and at the University of North Carolina at Chapel Hill.
Data and Specimen Collection
Demographic and clinical data were collected using a questionnaire as part of the clinical examination. Self-reported data collected included age, sex marital status, sexual partners, a history of ulcers, and self-treatment behaviors before consultation. During clinical examination, patients’ external genitals and inguinal region were examined, and any lymph node enlargement and the morphology of the ulcers were noted. Among women, vaginal and cervical examinations were also conducted. For patients with GUD, the lesions were cleaned with a gauze pad and material was taken from the base of the ulcer using Dacron-tipped swabs, then placed in Roche Amplicor transport media. Specimens were frozen and stored at −70°C and shipped to the University of North Carolina at Chapel Hill, North Carolina, for multiplex-polymerase chain reaction (M-PCR) (see the method described subsequently). In addition, blood was obtained for syphilis serology. Serum was tested using the MacroVue RPR Card Test (Becton Dickinson, Cockeysville, MD); RPR-positive results were confirmed by Serodia-TPPA (Fujirebio, Tokyo, Japan).
Qualitative Assessment of Herpes Simplex Virus Type 2
To aid in the presumptive diagnosis of HSV infection, the serum specimens were tested using Focus Technologies HSV-2 EIA for the presence or absence of human IgG class antibodies. The polystyrene microwells were coated with recombinant gG-2 antigen. Serum samples and controls were then diluted 1:10 and incubated in microwells to allow specific antibody present in the samples to react with the antigen. Nonspecific reactants, including unbound antibody, were removed by washing, and peroxidase-conjugated antihuman IgG was added and reacted with bound IgG. Excess conjugate was removed by washing. Enzyme substrate and chromogen were added and the color allowed to develop. After adding the stop reagent, the resultant color change was quantified by a spectrophotometric reading of optical density. Sample optical density readings were compared with reference cutoff optical density readings to determine results.
To confirm results, the HSV immunoblot kit to identify the presence of antibodies to HSV-1, HSV-2 or both was used. The patient sera was diluted 1:10 and incubated with individual antigen strips. If antibodies to HSV type-specific antigens or HSV common antigens were present in the sera, the antibodies bound to the antigens immobilized on the nitrocellulose membranes. The blots were then incubated with alkaline phosphatase-conjugated goat antihuman IgG, substrate (BCIP/NBT) was added, and a-colored precipitate formed where the antihuman conjugate had bound. The resulting band reactivity was then interpreted according to the manufacturer’s directions.
Genital Ulcer Disease Pathogen Detection by Multiplex Polymerase Chain Reaction
Specimen transport buffer containing swab material was thawed, and 100 μL of the sample was mixed with an equal volume of lysis buffer followed by the addition of 200 μL of diluent from the Roche Amplicor specimen prep kit. DNA was extracted from these preparations with phenol:chloroform:isoamyl alcohol (25:24:1), precipitated with 3 mol/L sodium acetate and 95% ethanol and resuspended in 150 μL sterile water. PCR was performed with previously published primers for T. pallidum, HSV, and H. ducreyi9 using 50 μL of extracted DNA as a template. Amplification reactions contained 50 μmol/L each of primers KO3A (digoxigenin [DIG]-labeled), KO4 (unlabeled), KS30 (DIG-labeled), KS31 (unlabeled), KO7A (DIG-labeled) and KO8A (unlabeled), 1.25 mmol/L MgCl2, 200 nM dNTPs including dUTP in place of dTTP, 1 U/mL Amperase (uracil N-glycosylase), and 10 U Taq polymerase (GibcoBRL) in 100 μL 1× PCR buffer. Positive controls were purified T. pallidum, HSV, and H. ducreyi DNA, and sterile water was used as a negative control. PCR consisted of an initial incubation at 95°C for 5 minutes, followed by 35 cycles of denaturation at 95°C for 20 seconds, primer annealing at 62°C for 20 seconds, and product extension at 72°C for 20 seconds.
M-PCR products were detected in an enzyme-linked immunosorbent assay (ELISA) format using previously published oligonucleotide probes KO17, KS54, and KO159 labeled at the 5′ end with biotin. In separate tubes, 30 μL of PCR product was mixed with each of the probes according to the instructions for the PCR DIG ELISA detection kit (Roche Diagnostic Systems). Specimens and controls were tested in duplicate by ELISA. Specimens were scored as positive for a specific pathogen if duplicate wells yielded A450 ≥0.25. Specimens with discordant duplicates were reamplified and reanalyzed until the replicates were concordant.
Demographic and clinical data were entered into a database in EPIINFO version 6 (CDC) and then converted to SAS version 8 (Cary, NC) for analyses. To identify independent risk factors for HSV and those for H. ducreyi, contingency tables were used and liberal cutoff P values of 0.25 were used. Multivariate analyses were conducted using logistic regression to identify independent risk factors.
To develop individual risk scores, independent clinical and demographic predictor(s) of HSV or H. ducreyi was assigned a weighted score equivalent to 10 times the absolute value of the regression coefficient10–12 such that for the ith patient, a weighted risk score was estimated by Σ(10βixi), where βi is the regression coefficient associated with the factor and xi is assigned a value of 1 if the factor is present in the ith patient and 0 otherwise. Sensitivity and specificity were calculated for a variety of possible cutoff points. To compare the weighted algorithm with a simplified version, because the latter could be easier to adopt in a clinical setting, unweighted scores were also estimated by summing up all independent predictors. When the factor was present, a value of 1 was assigned, regardless of the strength of the association with etiologic diagnosis, and 0 otherwise. The performance of risk scores in predicting the etiologic diagnoses of HSV and H. ducreyi was determined by comparing the areas under the receiver operating characteristic (ROC) curves for the weighted and unweighted scores. Areas under the curve were calculated using the trapezoidal method proposed by DeLong.13
An etiologic pathogen was detected by M-PCR in 91 (66%) of the 137 patients with genital ulcers. Forty-seven (34%) patients had HSV and 41 (30%) had H. ducreyi, although 2 of these also had detectable T. pallidum. Five (4%) had T. pallidum, 2 of whom were coinfected with H. ducreyi. Only 46 specimens (34%) had no pathogen detectable by PCR. Twelve (9%) of the 137 patients were positive for syphilis by RPR and MHA/TP, and in 4, T. pallidum was also detected by M-PCR and 8 had no detectable pathogen. Only 1 patient with T. pallidum detectable by M-PCR was negative for syphilis by RPR and MHA/TP.
Nearly 80% (n = 107) of patients with GUD had serum antibodies directed against HSV-2, and 94% (n = 128) had serum antibodies directed against HSV-1. Of the 43 patients with no etiologic diagnosis by PCR, 87% had serum antibodies directed against HSV-2. We found that 34% of individuals reported a prior episode of genital ulcers and most (89%) of these had IgG to HSV-2, whereas only 5 RPR-positive cases reported this history. Among the 47 patients in whom HSV was detected by PCR, 28% had no HSV-2 antibodies, suggesting that a substantial number of patients were new HSV-2 infections. Thirty-three percent (n = 45) of patients reported a history of genital ulcers before the episode leading to consultation, 7% (n = 15) reported 2 to 3 prior episodes, and 9 (7%) reported at least 4 prior episodes of genital ulcers. In the current study, among the 45 patients reporting a history of genital ulcers, the etiologic diagnoses of H. ducreyi, HSV, and T. pallidum were 20%, 32%, and 2%, respectively.
Herpes Simplex Virus.
Table 1 presents the distribution and Table 2 presents the mutually adjusted odds ratios for the association between demographic and clinical characteristics of participants with M-PCR-based diagnoses of HSV and H. ducreyi, as well as serologic diagnosis of HSV-2, compared with those without these diagnoses. Patients diagnosed with HSV were similar to those with no PCR-based etiologic diagnosis in the reported number of sexual partners in the preceding 3 months, self-treatment before consultation, and number of ulcers present. They were also similar to those without HSV diagnosis about ulcer morphology, including whether ulcers were shelved, shiny, puritic, or friable. However, compared with patients with GUD for whom HSV could not be detected, patients with HSV were more likely to be older (odds ratio [OR], 0.46; 95% confidence interval [CI], 0.21–1.00), married (OR, 3.56; 95% CI, 1.23–10.31), and female (OR, 7.38; 95% CI, 1.98–27.2). They were also more likely to consult early (OR, 3.23; 95% CI, 1.06–9.77), to have had a recent sexual encounter (OR, 3.23; 95% CI, 1.06–9.77), and to have had GUD in the past (OR, 4.14; 95% CI, 1.05–16.25). PCR-based HSV cases were less likely to have lymph node enlargement (OR, 0.22; 95% CI, 0.05–0.87) or deep ulcers (OR, 0.10; 95% CI, 0.03–0.39). These demographic and clinical characteristics appeared to also predict HSV-2, diagnosed by serology, although, as expected, the associations were weaker (Table 2). Thus, 7 risk factors independently predicted HSV infection among these patients with GUD.
Table 1 also shows the distribution, and Table 2 the odds ratios for the association between demographic and clinical characteristics of patients with GUD and detectable H. ducreyi. The 41 patients with H. ducreyi were demographically similar to those without H. ducreyi, although those with H. ducreyi tended to be somewhat younger (OR, 2.00; 95% CI, 1.03–3.85). Patients with H. ducreyi detectable by PCR also were similar to those without this diagnosis about healthcare utilization indicators such as late consultation and self-treatment. On clinical examination, H. ducreyi ulcers were more likely to be deep (OR, 3.32; 95% CI, 1.17–9.40). Ulcers among H. ducreyi cases were no more likely than those without detectable H. ducreyi to be friable (OR, 0.99; 95% CI, 0.35–2.85) and lymphadenopathy was apparent (OR, 1.22; 95% CI, 0.45–3.33). Thus, 2 factors, younger age and deep ulcers, were independently predictive of H. ducreyi infection.
Risk Assessment and Receiver Operating Characteristic Curves.
We developed an algorithm weighted on logistic regression coefficients of independent predictors of HSV and H. ducreyi, for syndromic management, and compared their diagnostic accuracy to an unweighted algorithm. To develop individuals’ risk scores, independent clinical and demographic predictors of HSV and H. ducreyi were assigned a weighted score equivalent to 10 times the absolute value of the regression coefficient.10Table 3 shows the relative weights for the 7 predictors of HSV infection and the 2 independent predictors of H. ducreyi infection that we used to derive the weighted score. The use of regression coefficients as weights (β) takes advantage of the equivalence of eβ and the natural log of the odds ratio (lnOR) such that each patient with GUD has a weighted score equal to Σ10β. The choice of using the regression coefficient β over the more commonly used odds ratio was based on its additive attribute.10
Table 4 summarizes patients’ weighted scores along with the sensitivity and specificity for the weighted (Σ10β) and unweighted scores at several cutoffs. Table 4 shows that using age alone to discriminate ulcers of herpetic origin from other GUDs would have identified 6.4% of cases with a specificity of 97%. A combination of 2 demographic characteristics, marital status and sex of the patient (corresponding to a score of 33), would have identified 63% of HSV cases with a specificity of 78%. Using an unweighted algorithm, in which the equivalence of each predictor is assumed, these 2 factors alone would have identified 70% of HSV cases with a specificity of 55%. When the areas under the ROC curves depicting the weighted and unweighted (plotting the true positives against the false-positives) in Table 4 were compared using a chi-square test,13 no significant difference between the 2 areas was noted (χ2df = 1 = 0.761, P value = 0.383), suggesting that the explanatory performance of the unweighted algorithm was similar to that of the weighted algorithm. The area under the ROC curves for the weighted and unweighted algorithms for H. ducreyi were too small to be of use in predicting infection, 35% and 49%, respectively.
Based on the prevalence and strength of associations between PCR-based diagnoses and factors independently predictive of HSV and H. ducreyi, we developed algorithms to guide clinicians in the management of ulcers without dependence on laboratory testing. Factors independently predictive of HSV included being married, being older than 25 years, being a woman, early consultation, having no sexual contact in the 7 days before consultation, superficial ulcers, and lack of inguinal lymph node enlargement. Some of these findings have been previously reported and the latter clinical factors are used as syndromic criteria for the diagnosis of HSV.5 Early consultation may reflect familiarity with episodic HSV infection, because a history of previous ulcers was significantly associated with HSV infection. Nonetheless, HSV treatment could incorporate acyclovir to decrease morbidity time and/or decrease viral shedding.24 However, the effect of incorporating acyclovir on cure rates is currently being evaluated in clinical trials in Malawi.
When we compared weighted and unweighted syndromic management algorithms using independent predictors of HSV infection and compared their sensitivity and specificity in the diagnosis and management of HSV, an unweighted algorithm with 2 demographic and/or clinical features can be used to diagnose HSV with a sensitivity of 70% and a specificity of 55%, whereas weighting of these parameters did not significantly change the sensitivity (63.8; 95% CI, 48.5–76.9) and the specificity (78.2; 95% CI, 67.8–86.0). In other words, when applying a decision rule requiring maintenance of a specificity of 55% or greater to the weighted score yielded a sensitivity of 70%, (95% CI, 55–82%) and a specificity of 70% (95% CI, 59–79%). When this rule is applied to the unweighted score, the sensitivity of 70% (95% CI, 55–82%) can be achieved with a specificity of 55% (95% CI, 43–65%). These findings suggest that unweighted criteria can be used for GUD management by selecting cutoff points relevant to specific settings. In settings in which specificity is of lesser importance compared with the consequences of patients going untreated, a cut point with a higher sensitivity may be adopted.
The algorithm derived from this study is presented in Figure 1. This management represents a significant divergence from current practice proposed by the World Health Organization (WHO, 2000). Empiric treatment of syphilis is reduced to reflect the infrequency of this infection. The prevalence of T. pallidum (3.7%) found in this study suggests that the number of uninfected persons is sufficiently high that the likelihood of a susceptible person having sexual contact with an infected person is small. Such a model is best augmented by a surveillance system to monitor changes in the prevalence of T. pallidum and other GUD-related pathogens. It is likely that improved technology and widespread testing for HIV will lead to more routine use of the RPR. HSV and H. ducreyi therapy will be based on selected clinical and demographic features. The RPR syphilis test is simple to do and because HIV testing is becoming the standard, syphilis testing may also become the standard, like in antenatal clinics.
In this study, we also found that GUD accounted for 24% of visits to our STD clinic in 1998–1999 compared with 60% of clinic visits in 1992,7 representing a 60% decrease. T. pallidum infection decreased by 93%, whereas H. ducreyi and HSV increased 15% and 338%, respectively. The decreased prevalence of T. pallidum likely reflects an aggressive program for empiric treatment of syphilis in the last decade. Indeed, a similar decrease in T. pallidum prevalence has been observed in Blantyre, Malawi, which has a similar treatment program.15T. pallidum infection has also decreased in other countries in this region.16
The remarkable increase in HSV-2 observed in Malawi is similar to other reports worldwide and in sub-Saharan Africa.17 The exact cause of this shift is not known. The empiric treatment of bacterial ulcers might have reduced the prevalence of syphilis and chancroid, whereas changes in sexual behavior could have promoted transmission of HSV. Many investigators have found a significantly higher prevalence of HSV in HIV infected patients with GUD.5,18–20 These data have been used to suggest that HSV facilitates transmission of HIV, and there is substantial evidence to support cotransmission of HSV and HIV.19 However, HIV could also play a role by increasing the number of people persistently shedding (contagious) HSV and by increasing the concentration of HSV in genital lesions. These findings should be, however, interpreted in the context of their limitations. First, the estimate of etiology is derived from a single STD clinic and not a random sample of the general population. However, this clinic is the prime site for STD care in Lilongwe, and there have been no major changes in this clinic or its clients since our earlier study. Second, we were unable to detect an etiologic agent in the ulcer crater in 33% of patients with GUD, whereas other investigators working in this region4 detected a pathogen in a larger proportion of patients. Although it is possible that some patients may have been infected with pathogens other than the 3 for which M-PCR was performed, HSV antibodies detected in the majority of patients with ulcers of unknown cause strongly suggest that HSV may have been the responsible pathogen for these ulcers. This study cannot, however, preclude the possibility that other previously rare pathogens such as Calymmatobacterium granulomatis causing granuloma inguinale may be involved.
In summary, we have shown that in the 7 years between 1992 and 1999, there was a sizable temporal shift in the prevalence of GUD-related pathogens in Malawi, requiring a change in syndromic management. We have also shown that these syndromic criteria need not be weighted, because the difference in accuracy between weighted and unweighted algorithms is small. The advantage of an unweighted algorithm is that it uses data available in clinic notes and can be implemented easily, requiring little data manipulation and only periodic laboratory support to estimate prevalence of etiologic pathogens. Therapy for HSV has not heretofore been routinely provided in this clinic. Although acyclovir is well tolerated and is now affordable, the clinical benefits are arguably modest, because the genital ulcers are self-limited. However, acyclovir may reduce the transmission of HSV21 to sexual partners and could reduce transmission of HIV as well.22 However, the latter remains to be tested in clinical trials.
1.Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: Randomized controlled trial. Lancet 1995; 346:530–536.
2.Dickerson M, Johnson J, Delea TE, et al. The causal role of genital ulcer disease as a risk factor for transmission of human immunodeficiency virus: An application of the Bradford Hill criteria. Sex Transm Dis 1996; 23:429–440.
3.Wasserheit J. Epidemiologic synergy: interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis 1991; 19:62–64.
4.World Health Organization. Management of patients with sexually transmitted diseases. Report of a WHO study group. World Health Organization Technical Report 1991; 810:1–103.
5.Morse SA, Trees DL, Htun Y, et al. Comparison of clinical diagnosis and standard laboratory and molecular methods for the diagnosis of genital ulcer disease in Lesotho: Association with human immunodeficiency virus infection. J Infect Dis 1997; 175:583–589.
6.Trees DL, Morse SA. Chancroid and Haemophilus ducreyi—An update. Clin Microbiol Rev 1995; 8:357–375.
7.Beherts FM, Liomba G, Lule G, et al. Sexually transmitted disease and human immunodeficiency virus control in Malawi: A field study of genital ulcer disease. J Infect Dis 1995; 171:451–455.
8.Bogaerts J, Ricart CA, Van Dyck E, et al. The etiology of genital ulceration in Rwanda. Sex Transm Dis 1989; 16:123–126.
9.Orle KA, Gates CA, Martin DH, Body BA, Weiss JB. Simultaneous PCR detection of Haemophilus ducreyi, Treponema pallidum, and herpes simplex virus types 1 and 2 from genital ulcers. J Clin Microbiol 1996; 34:49–54.
10.Harrel F. Predictive outcomes: Applied Survival Analysis and Logistic Regression. Monograph. Charlottesville, VA: University of Virginia, 1997.
11.Mayaud P, ka-Gina G, Cornelissen J, et al. Validation of WHO algorithm with risk assessment for clinical management of vaginal discharge in Mwanza, Tanzania. Sex. Transm Infect 1998; 74(suppl 1):S77–S84.
12.O’Farrell N, Hoosen AA, Coetzee KD, van den Ende J. Genital ulcer disease: Accuracy of clinical diagnosis and strategies to improve control in Durban, South Africa. Genitourin Med 1994; 70:7–11.
13.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: A non-parametric approach. Biometrics 1988; 44:837–845.
14.Taha TE, Dallabetta GA, Hoover DR, et al. Trends of HIV-1 and sexually transmitted diseases among pregnant and postpartum women in urban Malawi. AIDS 1998; 12:197–203.
15.Dyer JR, Eron JJ, Hoffman IF, et al. Association of CD4 cell depletion and elevated blood and seminal plasma human immunodeficiency virus type 1 (HIV-1) RNA concentrations with genital ulcer disease in HIV-1-infected men in Malawi. J Infect Dis 1998; 177:224–227.
16.Htun Y, Morse SA, Danger Y, et al. Comparison of clinically directed disease specific and syndromic protocols for the management of genital ulcer disease in Lesotho. Sex Transm Infect 1998; 74(suppl 1):S23–S28.
17.Chen CY, Ballard RC, Beck-Sague CM, et al. Human immunodeficiency virus infection and genital ulcer disease in South African—The herpetic connection. Sex Transm Dis 2000; 1:21–29.
18.Bosu WK. Syndromic management of sexually transmitted diseases: Is it rational or scientific? Trop Med Int Health 1999; 4:114–119.
19.Reynolds SJ, Risbud AR, Sherperd ME, et al. Recent herpes simplex virus type 2 infection and the risk of human immunodeficiency virus type 1 acquisition in India. J Infect Dis 2003; 187:1513–1515.
20.McFarland W, Gwanzura L, Bassett MT, et al. Prevalence and incidence of herpes simplex virus type 2 infection among male Zimbabwean factory workers. J Infect Dis 1999; 180:1459–1465.
21.Watts DH, Brown ZA, Money D, et al. A double-blind, randomized, placebo-controlled trial of acyclovir in late pregnancy for the reduction of herpes simplex virus shedding and cesarean delivery. Am J Obstet Gynecol 2003; 188:836–843.
22.Galvin SR, Cohen MS. The role of sexually transmitted diseases in HIV transmission. Nat Rev Microbiol 2004; 2:33–42.