PATIENTS AND METHODS

The study comprised 186 consecutive migrant mine workers who presented in 1998 with non-vesicular genital ulcerations at the East Driefontein clinic for sexually transmitted diseases near Carletonville, Gauteng, South Africa. In each case, swab specimens were obtained directly from the ulcers and tested for Haemophilus ducreyi, herpes simplex virus (HSV), and Treponema pallidum specific DNA sequences by using a multiplex polymerase chain reaction (PCR) technique (Roche Molecular Systems, Alameda, CA, USA).² A very small proportion (not recorded) of the men had an overt urethral discharge but none of them complained of symptoms of urethritis. Nevertheless, a swab specimen was taken from the urethral meatus for the assessment of urethritis. This was diagnosed if there were five or more polymorphonuclear leucocytes (PMNLs) per high power microscope field (hpf; ×1000 magnification) in a Gram-stained urethral smear. Also, a first catch urine specimen (10–15 ml) was collected and tested for Neisseria gonorrhoeae and Chlamydia trachomatis specific DNA sequences by ligase chain reaction (LCR) assays (Abbott Laboratories, Chicago, IL, USA)^3,4 according to the manufacturer’s instructions, and for Mycoplasma genitalium by a PCR assay as described previously.^5,6 Venous blood was tested for chlamydial antibodies by using a modified microimmunofluorescence test.⁷ Broadly cross reacting antibody detected at titres of 1: = or >256 was considered diagnostic for lymphogranuloma venereum (LGV). In addition, serum was tested for HIV antibody (ELISA; Abbott Laboratories, Chicago, IL, USA) and this was confirmed either by western blot (HIV blot 2.2; Diagnostic Biotechnology, Singapore), or by an indirect immunofluorescence test (Serofluor; Virion).

RESULTS

Aetiology of genital ulcers

Of 186 men with genital ulcers, the cause was H ducreyi in 100 (53.8%), HSV type 2 in 34 (18.3%), T pallidum in 12 (6.5%), and LGV in six (3.2%); 17 (9.1%) of the men had a mixed infection, and 17 (9.1%) had ulcers, the nature of which was not diagnosed.

Occurrence and microbiology of urethritis

Of the 186 men with genital ulcers, 99 (53%) had five or more PMNLs per hpf on microscopic examination of a urethral smear and were considered, therefore, to have concomitant urethritis. Of these, 45 (45.4%) were infected with N gonorrhoeae, of whom six (13%) were also infected with C trachomatis and another five (11%) with M genitalium. Thus, 54 (54.6%) of the men were diagnosed by exclusion as having non-gonococcal urethritis (NGU); five (9%) of these men were infected with C trachomatis, another five (9%) with M genitalium, and one man with both micro-organisms. Eighty seven (47%) of the men had fewer than five PMNLs per hpf and were deemed not to have urethritis. However, two (2.3%) of these men were infected with N gonorrhoeae, another two (2.3%) with C trachomatis, and seven (8%) others with M genitalium.

Occurrence and effect of HIV infection

One hundred and twenty (64.5%) of the men were infected with HIV. As shown in table 1, the distribution of chancroidal and herpetic ulcers—that is, the majority of ulcers, was not appreciably different in HIV positive and HIV negative men.

The occurrence of urethritis in men with or without HIV infection was compared. Of the 120 HIV positive men, 32 (26.7%) had gonococcal urethritis, 37 (30.8%) had NGU, and 51 (42.5%) did not have urethritis. Of 66 HIV negative men, 13 (19.7%) had gonococcal urethritis, 17 (25.8%) had NGU, and 36 (54.5%) did not have urethritis. The difference between the occurrence of gonococcal urethritis and of NGU in the HIV positive and HIV negative groups is, however, not statistically significant (χ² test with Yates’s correction: 1.48, 1 df, p= 0.22; and 1.0, 1 df, p=0.32, respectively).

Relation between ulcers and urethritis

The proportion of men with an ulcer of known aetiology who had gonococcal urethritis, NGU, or no urethritis is shown in table 2. For men with chancroidal or herpetic ulcers, the results are shown relative to their HIV status because the numbers involved are sufficiently large. There is a trend, not statistically significant (χ² with Yates’s correction, p=0.14), for HIV positivity to favour the occurrence of gonococcal urethritis in men whose ulcers were due to H ducreyi. However, this apart, HIV positive status seemed to have no influence on the occurrence of urethritis, whether gonococcal or NGU, in men with ulcers of known cause. Overall, the occurrence of gonococcal urethritis (24% of the men), NGU (29%), and no urethritis (47%) was a distribution that was similar irrespective of ulcer aetiology. Gonococcal urethritis dominated in men with LGV, but the numbers involved are too small to determine whether this is meaningful.

Observations on the occurrence of C trachomatis and M genitalium

As shown in table 3, C trachomatis and M genitalium occurred almost independently of each other. When HIV status is not considered, C trachomatis was found more often in men with any urethritis than in those without (11 (12) of 99 v 2 of 87; Fisher’s exact test, p=0.02), in men with gonococcal urethritis than in those without (6 of 45 v 2 of 87; p=0.02), but the difference did not reach significance in men with NGU (5 of 54 v 2 of 87; p=0.11, or 6 of 54 v 2 of 87; p=0.054).When HIV status is considered, C trachomatis was detected in 11 of 120 HIV positive men compared to two (three if the dual infection with M genitalium is included) of 66 HIV negative men (Fisher’s exact test; p=0.196). Among HIV positive men, C trachomatis was detected more often in men with any urethritis than in those without urethritis (10 of 69 v 1 of 51; p=0.02), more often in those with gonococcal urethritis than in those without (5 of 32 v 1 of 51; p=0.03) but not significantly more often in men with NGU than in those without urethritis (5 of 37 v 1 of 51; p=0.08). Further, among HIV negative men, C trachomatis was not found more often in men with gonococcal urethritis, NGU, or any urethritis than in men without urethritis. When HIV status is not considered, M genitalium was found no more frequently in men with any urethritis, gonococcal urethritis or NGU, than in men without urethritis. When HIV status is considered, M genitalium was detected in 13 of 120 HIV positive men compared to four (five if the dual infection with C trachomatis is included) of 66 HIV negative men (Fisher’s exact test; p=0.193). Among HIV positive men, M genitalium was detected no more frequently in men with any urethritis, gonococcal urethritis, or NGU than in those without urethritis. Among HIV negative men, again M genitalium was detected no more frequently in any of the groups than in men without urethritis.

DISCUSSION

An infectious cause for the genital ulcers was defined in more than 90% of the men recruited into this study, the majority having ulcers due to H ducreyi. These men were migrant mine workers who were served by a relatively small, but stable, group of sex workers, which accounts for the relatively high prevalence of chancroid recorded here. Since this study, the relative prevalence of chancroid has decreased as a result of the application of periodic preventive treatment with azithromycin for the sex workers.⁸ Surprisingly, more than half of all the men also had objective evidence of concomitant urethritis, of which 45% was gonococcal urethritis and 55% was NGU. A few of the men had phimosis so that PMNLs seen in a Gram stained smear may have come from the subpreputial sac rather than the urethra, thus leading to a slight overestimation of the prevalence of NGU but not, of course, of gonococcal urethritis.

There would seem to be no biological sense in an ulcer of whatever cause influencing the occurrence of urethritis of a particular aetiology. However, defining the relation between ulcers and urethritis, if any, was complicated by the fact that approximately two thirds of the men were also infected by HIV. It seems that HIV infection did not enhance the occurrence of any type of ulcer, nor did it increase the likelihood of gonococcal urethritis or NGU occurring. Although, for reasons unknown, C trachomatis was associated significantly with gonococcal urethritis in HIV positive but not HIV negative men, neither C trachomatis nor M genitalium was associated significantly with NGU in either HIV positive or HIV negative men. This is of interest because others⁹ have reported that they detected M genitalium in the urethra of men who had AIDS more often than of men without AIDS. Unfortunately, there are no CD4 data available from the current study to help to determine whether the severity of immunodeficiency might influence microbial detection, but the lack of any association with HIV positivity would seem to make this unlikely. While it is important to define how HIV infection affects the behaviour of C trachomatis, M genitalium, and other micro-organisms in the male urethra, it seems that a setting in which HIV infection is so rife is unlikely to be optimal for defining the pathogenicity of M genitalium and could give a distorted view in comparison with studies undertaken mainly in developed countries¹⁰ where HIV infection is far less frequent. This apart, the failure to clearly associate C trachomatis or M genitalium with NGU is, perhaps, at least partially attributable to the fact that the urethritis was probably chronic in most cases and asymptomatic, the patients being seen not because they were complaining of urethral symptoms, but because they had genital ulcers. It is noteworthy that in another study¹¹ acute symptomatic NGU was associated with these micro-organisms, whereas asymptomatic disease was not. Furthermore, the diagnosis of NGU in a few men may have been spurious because of the inability to distinguish between urethral inflammation and “contaminating” subpreputial inflammatory cells.

Syndromic management is meant to address symptomatic disease only. However, irrespective of the detailed microbial aetiology of urethritis, it is clear that men presenting with genital ulcers in the population that we have studied, and probably in those in other developing countries, have asymptomatic urethritis so frequently as to make syndromic management, based on their ulcers alone, inappropriate. Ideally, it would be useful to have a non-invasive screening test, based on symptoms and/or signs, that was predictive of urethritis, a positive result being confirmed by a Gram stain test. Clearly, this approach is not feasible in this population and probably not in others too. Alternatively, a definitive, but simple screen for urethritis, based on examination of a Gram stained urethral smear, could be made and patients with positive results treated accordingly; or because there is more than a 50% chance of gonococcal urethritis or NGU existing, all patients, irrespective of such definitive diagnoses, could be treated to cover for urethritis, as part of an expanded syndromic treatment programme for genital ulcers. In this population, where chancroid and syphilis were found to be the most frequent causes of curable genital ulceration, an ideal antibiotic combination would be a single intramuscular injection of benzathine penicillin (2.4 U), to cover for syphilis, followed by a single oral dose of 500 mg of ciprofloxacin, to cover for both gonorrhoea and chancroid, and 1 g of azithromycin, to cover for NGU caused by C trachomatis or M genitalium, and for chancroid. However, in resource-poor settings, azithromycin could be substituted by 500 mg of erythromycin given orally four times daily for 7 days, or possibly by a tetracycline. Furthermore, in geographical locations where the aetiology of genital ulcer disease is different from that described here, other antimicrobial combinations may be more appropriate.

CONTRIBUTORS

RCB was responsible for the overall management of the study and contributed to writing the manuscript; HGF, YH, and FR were responsible for the collection of specimens and for laboratory testing and advised on aspects of the manuscript; JSJ examined specimens for M genitalium and advised on aspects of the manuscript; DT-R initiated work on M genitalium and was largely responsible for writing the manuscript.