Introduction

Sexually transmitted diseases (STD) facilitate the transmission of HIV. This interaction may be a major factor in the rapid spread of HIV in sub-Saharan Africa. A study in Tanzania demonstrated that strengthening STD management in primary healthcare centres resulted in a 42% reduction in the transmission of HIV.¹ This has given a new impetus to national STD control programmes. Control strategies focus on the syndromic management of STD with cheap but effective drugs, following standardised flow charts.^{2, 3} At the national level, it is important to obtain reliable data on the aetiology of major STD syndromes, to ensure that the flow charts perform well in the local context but also to develop strategies for patients whose symptoms have not improved despite proper use of the recommended drugs. Little information is available on the aetiology of STD syndromes in central Africa. We conducted a study of the aetiology of urethral discharge in Bangui, Central African Republic (CAR), which examined the role of genital mycoplasmas and of Trichomonas vaginalis, in addition to the more conventional gonococci and chlamydia.

Methods

From July 1996 to January 1997, all male patients presenting with a complaint of urethral discharge at the Centre National de Référence pour les Maladies Sexuellement Transmissibles (CNRMST) in Bangui were invited to participate in the study, which had been ethically approved by the Ministry of Public Health. The CNRMST is the national STD centre, and offers STD care to patients living in the Bangui area (approximately half a million inhabitants). Some patients are referred to the CNRMST by other healthcare providers, but the majority come on their own, either for the first treatment of their STD or because treatment obtained elsewhere has failed. Patients who agreed to participate were administered a questionnaire looking at sociodemographic characteristics, sexual behaviour, past STD, and current symptoms. A genital examination was performed by a doctor who also obtained urethral samples. Gram stain and culture for Neisseria gonorrhoeae (NG) were performed at the CNRMST on specimens obtained as routinely done in this clinic (patients were not systematically asked not to have urinated within the previous 4 hours). Cultures for NG were performed on a modified Thayer-Martin medium and identification was based on the Gram stain and oxidase test.

A urethral swab was also obtained for polymerase chain reaction (PCR) analyses, put into the Amplicor transport medium (Roche Diagnostic Systems), and kept refrigerated until its transport to Canada. PCR was carried out for the following potential pathogens: Neisseria gonorrhoeae, Chlamydia trachomatis (CT), Trichomonas vaginalis (TV), Mycoplasma genitalium (MG), and Ureaplasma urealyticum (UU). Specimens were prepared for PCR by adding an equal volume of Amplicor specimen diluent, and kept at least 3 days at 4°C before analysis to reduce the frequency of inhibition.⁴ Inhibition was monitored in samples negative for all pathogens by Amplicor internal control or a home brew internal control made by cloning the MG amplicon and inserting a DNA fragment.^{4, 5} For NG and CT, the Roche Amplicor Chlamydia trachomatis/Neisseria gonorrhoeae Test Kit (Roche Diagnostic Systems) was used according to the manufacturer's instructions. Modifications of the procedures of Jensen for M genitalium, of Blanchard and Willoughby for U urealyticum and of Shaio for T vaginalis were employed.^6–10 Primers used for MG, UU, and TV are shown in table 1. Each micro-organism was revealed separately using PCR (U urealyticum) or seminested PCR (M genitalium, T vaginalis) and detection of the amplicon after electrophoresis through agarose gels and staining with ethidium bromide. A prepared sample (25 μl) was added to a master mix (25 μl) composed of 2× buffer 20 mM TRIS-HCl pH 8.3, 100 mM KCl, 400 μM of dNTP, 1.25 unit of Taq polymerase, and 0.5 μM primers. The amplification procedure was 5 minutes at 95°C and 40 cycles of denaturation (95°C), annealing (62°C), and extension (72°C) for 20 seconds at each step. After holding the samples at 72°C for 15 minutes, the reactions were cooled to 4°C. After the first series, a seminested series of PCR was undertaken using 1 μl of the first series and 49 μl of a master mix to give the same final concentration of reagents as the initial series.

To investigate the association between potential pathogens and the urethral discharge syndrome, urethral swabs were also obtained from 100 adult male patients consulting in a Bangui health centre for complaints unrelated to the genital tract (mostly febrile syndromes attributed to malaria). These specimens were processed for PCR as for patients with urethral discharge, but no Gram stain and no culture for NG were carried out. Data were entered on epi-info 6.0 and analysed with epi-info and stata 5.0. Proportions were compared with the χ² test or Fisher's test when appropriate. Continuous data were compared with the Student's t test or the Wilcoxon rank sum test. For multivariate analysis, a logistic regression analysis was carried out.

Results

PCR was performed on all 410 patients with a complaint of urethral discharge but, owing to security problems in Bangui during the study period, we were able to perform cultures for NG on only 344 of them. Of these 344 patients, 223 were positive in both the culture and the PCR, 86 were negative in both, 26 were positive in NG PCR and negative in culture, and only nine were positive in culture and negative in PCR. We were able to confirm, with a 16S rRNA PCR (supplied by Roche Diagnostic Systems), 22 of the 26 samples positive with the Amplicor PCR but negative in culture. A patient was considered to have gonococcal urethritis if either the culture or the Amplicor PCR (or both) was positive for NG. However, for the unmatched case-control analysis, since gonococcal cultures were not available for asymptomatic controls, only patients with a positive NG PCR were considered to have gonococcal infection. For all other potential pathogens, a positive PCR was the only criterion to determine the presence of the agent.

In table 2 are compared the distribution of the five potential pathogens in 410 cases with urethral discharge and in 100 asymptomatic controls. In univariate analyses NG, CT, and TV were found significantly more often in cases of urethral discharge than in asymptomatic controls; MG was also found more frequently in cases than in controls, but this did not reach statistical significance. UU was found significantly more often in controls than in cases. As controls were younger (median 23 versus 25 years, p<0.001) and had had fewer sex partners in the past 12 months than cases (median 1 versus 4 sex partners, p<0.001), factors which were themselves related to the prevalence of potential pathogens, and as there was a substantial number of multiple infections, a logistic regression analysis was carried out in which the odds ratios for each potential pathogen were adjusted for other pathogens but also for age and the number of sex partners in the past 12 months. It revealed a significant interaction between NG and MG and thus we present in table 2 results of univariate and multivariate analyses after stratification of cases according to whether NG was present or not. MG was associated with urethral discharge only in the absence of NG. For CT, the association with urethritis was stronger and statistically significant only among those cases negative for NG. For TV, the association reached statistical significance only among patients with NG. UU was clearly not associated with urethral discharge. The following analyses will examine NG, CT, MG, and TV to better delineate their contribution to urethral discharge in Bangui, but the role of UU will not be further examined.

Among the 410 patients with urethral discharge, 69% (283) had gonococcal urethritis (from now on defined as PCR and/or culture positive), 14% (56) had a chlamydial infection, 20% (84) harboured M genitalium, 16% (66) had T vaginalis recovered, while 11% (45) were negative for all four pathogens. CT was found in 17% (22/127) of patients with a non-gonococcal urethritis and in 12% (34/283) of those with a gonococcal urethritis (p=0.20). MG was found in 42% (53/127) of patients with non-gonococcal urethritis, compared with 11% (31/283) in men with gonococcal urethritis (p<0.001). TV was as frequent in patients with non-gonococcal (18%, 23/127) as in those with gonococcal (15%, 43/283) urethritis (p=0.55). Table 3 shows the co-infections in patients with urethral discharge; 153 (37%) of the 410 cases had more than one pathogen detected. While NG and MG were found on their own in, respectively, 67% and 48% of patients in whose samples these pathogens were documented, CT and TV were usually recovered along with other micro-organisms. CT, MG, and TV were equally frequent as co-infections in patients with gonococcal urethritis. TV was strongly associated with NG. Of all patients with TV recovered, two thirds also had NG, while one fifth also had MG, and only one fifth had TV as the sole “pathogen”. CT was also strongly associated with NG.

Overall, 49% of patients had sought treatment elsewhere before coming to the CNRMST: 56% of this subgroup had bought drugs from peddlers in markets, 21% in pharmacies, and 19% in health centres. Table 4 shows the prevalence of the four pathogens according to various behavioural or clinical characteristics. There was no significant difference in the prevalence of any of these four micro-organisms between patients who had received previous treatment and patients for whom the visit to the CNRMST was the first attempt to get treatment; this suggests that, despite the CNRMST being a reference institution, its population of patients was roughly representative of urethritis patients in Bangui. Attributing the current urethritis to contact with a sex worker increased the probability that NG (but not other pathogens) would be recovered. A discharge was observed during the clinical examination in 94% of patients; among those without a discharge, the gonococcus was found in a quarter and almost half had at least one pathogen recovered. Patients with a purulent discharge and patients with a discharge lasting less than 2 weeks were more likely to have NG and less likely to have MG than others. Even though 86% of men with ≥ 5 polymorphonuclear leucocytes (PMNs) per field had NG, 35% (19/55) of patients who did not fulfil this objective criterion of urethritis also had NG. Patients whose urethritis was not acquired from a sex worker, those with a discharge lasting for more than 2 weeks, without any discharge observed, without a purulent discharge or with less than 5 PMNs per field and patients without Gram negative diplococci on the urethral smear were all more likely to have no pathogen recovered. A review of the clinical characteristics of patients with a single pathogen recovered showed that those with CT, MG, or TV urethritis had identical clinical syndromes, with a less severe and longer lasting urethritis than patients among whom NG was found (data not shown).

Discussion

Two thirds of patients seeking treatment for urethral discharge in Bangui had evidence of gonococcal infection. This is in line with reports from other African countries where N gonorrhoeae is the aetiological agent in 53–80% of cases of urethritis.¹¹ Only 12% of patients with gonococcal urethritis in Bangui were found to have concomitant CT infection. PCR should have been more sensitive than the methods (cultures, ELISA) used in previous studies on the African continent, but we cannot exclude the possibility of underdetection of CT due to inhibition of CT PCR in patients with symptomatic gonococcal infection.¹² CT was also rather uncommon as an aetiological agent of non-gonococcal urethritis (NGU): only 17% of these patients had CT documented in their samples. This paucity of C trachomatis infection is comparable with other reports from African countries where C trachomatis was involved in only 3–16% of cases of NGU and was found concomitantly in only 3–12% of cases of gonococcal urethritis.^{11, 13, 14} This contrasts with developed countries where 11–34% of male patients with gonorrhoea also have C trachomatis infection and where C trachomatis is the aetiological agent in 15–58% of cases of NGU.¹⁵

Mycoplasma genitalium has recently been identified as an agent of NGU in Western Europe.^16–18 MG was a much more common cause of NGU in Bangui than C trachomatis, being found in 42% of such patients. We thus confirmed our recent finding from seven west African countries that M genitalium is an aetiological agent of NGU in Africa,¹⁰ and indeed its contribution to this syndrome seems to be more prominent in central Africa than in west Africa. As our observations depend on a home brew PCR, we confirmed our results by digesting the amplicon with the restriction endonuclease Sau3A I and precisely sizing by polyacrylamide gel electrophoresis the two restriction fragments predicted from the sequence. This PCR does not amplify reference strains of Mycoplasma pneumoniae, U urealyticum, or Mycoplasma hominis at high concentrations, nor Canadian genital samples that were positive by culture for the latter two micro-organisms. It readily detected 10 copies of the positive control M genitalium target sequence cloned in the plasmid pUC 19 even when they were spiked into NG positive samples.

The characteristics of MG urethritis in Africa are similar to what is observed with chlamydial urethritis. MG induces a less severe urethral inflammation than NG, so that patients are less likely to have a purulent discharge and to complain of dysuria while waiting longer before seeking treatment. It seems likely that many of these patients remain sexually active during this period. Thus, MG could have some impact on HIV transmission, as recently suggested by a study in American heterosexual couples,¹⁹ even though its effect on the urethral shedding of HIV might be more modest than that of NG.²⁰ The severe inflammation induced by the gonococcus may prompt the patient to discontinue sexual activity and to find an appropriate treatment more rapidly than with MG. There is unfortunately little information on the optimal treatment of MG urethritis, but it seems likely that a 7 day course of doxycycline would be adequate for most cases.²¹ The current WHO recommended syndromic management of urethral discharge, which consists of a single dose treatment for gonorrhoea (generally with ciprofloxacin) and a 1 week course of doxycycline,² is probably effective in patients infected with MG and does not need to be reconsidered.

The contribution of TV to the urethral discharge syndrome in Bangui is unclear. It was found in 16% of cases of urethral discharge, a prevalence similar to what we and others found in west Africa (14%) and Malawi (21%).^{10, 22} TV is also found in 6–12% of asymptomatic men.^{10, 22} In west Africa, TV was found significantly more often in cases of urethral discharge than in asymptomatic controls after adjustment for other pathogens and various confounders,¹⁰ whereas in the current study this difference did not reach statistical significance. This may simply be due to a somewhat smaller sample size, as the adjusted odds ratios were similar in both studies (2.62 in west Africa, 2.81 and 1.97 in Bangui). However, as in the west African study, TV was usually recovered along with other micro-organisms and was the sole pathogen in only 13 (3%) patients with urethritis in Bangui. In cases of co-infections, further studies will be required to delineate if TV contributes to the inflammation or is merely a co-transmitted innocent bystander. In the meantime, it seems unwarranted to systematically add metronidazole as a first line drug in all patients with urethral discharge, but this drug should be considered in patients whose discharge persists after having received ciprofloxacin and doxycycline, especially since a recent study showed that patients with TV urethritis have higher levels of seminal HIV shedding than asymptomatic controls.²² Although TV is probably the most frequent sexually transmitted pathogen in women,²³ it is not clear that treating men harbouring TV in their urethra would impact on TV in women by decreasing the reservoir.

Our results show that the syndromic approach is the only rational way to deal with patients presenting with urethral discharge in sub-Saharan Africa. At least three, and probably four, pathogens are involved in the aetiology of urethral discharge, and multiple infections are common. Furthermore, there is substantial overlap between urethritis of various aetiologies with regard to the clinical presentation. Thus, the simultaneous treatment of all patients for gonorrhoea (preferably with ciprofloxacin), CT, and MG (with doxycycline) seems the only logical way to deal with this problem. Drugs potentially active against these three pathogens (azithromycin, ofloxacin) are unfortunately too expensive to be considered at the present time. Future aetiological studies should focus on patients who have failed first line therapy, to define rational second line algorithms and clarify the role of T vaginalis.

In Bangui, having received a previous treatment had no effect on the prevalence of various pathogens. This merely reflects the dismal quality of STD care in this city, especially in the informal sector where treatment with ineffective drugs are the rule rather than the exception.²⁴ Unless the syndromic approach is used not only in government health centres but in the informal sector where most STD patients seek care, the potential impact of STD control on the spread of HIV will remain a missed opportunity.