• C trachomatis
• chlamydia infection
• mycoplasma
• NGU
• non-specific urethritis

A test of cure following treatment for uncomplicated cervical or urethral Chlamydia trachomatis infection with either single dose azithromycin (1 g) or doxycyline (100 mg twice daily for 7 days) is currently not recommended. Earlier trials indicated that both treatments are more than 95% effective.1–3 However, recent evidence strongly suggests that treatment failure may occur in more than 5% patients. This was the subject of a recent editorial by Handsfield4 and a late breaker symposium at the recent ISSTDR meeting in Quebec. Handsfield has argued persuasively that this apparent increase in treatment failure with azithromycin is probably not real. Tissue culture, which is less sensitive than nucleic acid amplification tests, was predominantly used in the original treatment trials and would not have been able to detect small numbers of persistent of C trachomatis bacteria.1 4 This article reviews the evidence for treatment failure, considers whether we need to modify current treatment regimes and suggests possible topics for future research.

It has always been assumed that individuals retesting positive for chlamydia after a full course of treatment may be due to re-infection.2 But, azithromycin treatment failures at levels >5% where re-infection has been excluded have been documented in women, men with non-gonococcal urethritis (NGU) and in men with rectal chlamydia. Two studies in women, not at risk of re-infection, have observed treatment failure rates of approximately 8%.5–7 A high failure rate (23%) was also recently observed in men with non-gonococcal urethritis treated with single dose azithromycin who were advised to abstain from sexual intercourse.8 Although re-infection could not be excluded, the doxycycline (100 mg twice daily for 7 days) group which did not differ in terms of sexual behaviour following treatment had a significantly lower failure rate (5.2%).4 8 In a retrospective study of rectal chlamydia infection in men, a treatment failure rate of 6%9 was observed in men not at risk of re-infection. Two other studies in men with non-gonococcal urethritis are also consistent with this but in one study it is unclear what proportion were treated with azithromycin and in the other it was not possible to estimate the effect of re-infection.10 11 Following up treated patients for longer periods (up to about 1 year) show more treated individuals retest chlamydia-positive as a result of re-infection, either from an untreated partner(s) or a new sexual partner(s).12 Not being able to distinguish treatment failure from re-infection using nucleic acid amplification tests or culture has undoubtedly contributed to the delay in recognising that treatment failure with azithromycin is a problem.12

There are three biologically plausible explanations for azithromycin treatment failure: first, a phenomenon called heterotypic resistance and second the consequence of single dose therapy with a bacteriostatic antibiotic. These are not mutually exclusive. A third possibility, although unlikely, is the emergence of homotypic (genetically inherited) resistance in C trachomatis to azithromycin.

Heterotypic antimicrobial resistance refers to the replication of a heterogeneous population of resistant and susceptible bacteria from a subculture of a single resistant organism propagated on antimicrobial-containing medium, that is, it is not genetically inherited. Chlamydia exhibits this phenomenon at high infectious loads.7 13 14 Other bacteria also exhibit similar phenomena called phenotypic switching. Populations of Escherichia coli exhibit pre-existing heterogeneity with normally growing cells and a small subpopulation of ‘persister’ cells, which have reduced growth rates. These persister cells are less susceptible to antimicrobial therapy, a characteristic that is not genetically inherited.15 It is possible that C trachomatis may possess a similar mechanism at high bacterial loads, with the development of a subpopulation of persister forms that are less susceptible to antimicrobial therapy.15 Patients with urethritis have high loads and there is a wide variation in chlamydia load in the female genital tract, with higher loads in younger women, and in the rectum.16–18

Given the complexities of the chlamydia life cycle19 it is possible that the persister form(s) involved in heterotypic resistance may take a number of weeks to re-establish detectable replicative infection following single dose azithromycin treatment. It is relevant that the newer studies reporting higher failure rates have longer follow-up periods in excess of 5 weeks and so are more likely to identify treatment failures as a result of heterotypic resistance. Thus, heterotypic resistance may be playing a role in treatment failure in vivo. It may however be more complex than this as a recent study using real time quantitative PCR indicates that quinolones, macrolides and tetracyclines are bacteriostatic and not bactericidal when exposed to the bacterium for <48 h.20 This could facilitate the development of heterotypic resistance by failing to eradicate the slow growing population of persister forms. The authors conclude that the clearance of chlamydia depends not only on the activity of the antibiotic but also the capacity of the host to eliminate the remaining bacteria.20

Nevertheless, there is evidence to suggest that prolonged exposure to azithromycin is likely to be chlamydicidal in vivo. Chlamydial antimicrobial assays are complex, non-standardised and interpretation is difficult.13 14 21 The majority of studies only expose chlamydia-infected cells to antimicrobials for 48 h and assess chlamydicidal activity after 48 h.13 14 21 Mpiga and Ravaoarinoro22 assessed the ability of the L2 strain of chlamydia (the LGV serovar) to produce infectious progeny up to 10 days after a range of exposure times to erythromycin, a macrolide antibiotic as is azithromycin,23 given at its minimal chlamydicidal concentration, as determined above. Infectious progeny could not be detected after 10 days of exposure. Although the authors did not examine for heterotypic resistance, the infectious load of chlamydia used was high enough to ensure this would have occurred.13 The work of Mpiga and Ravaoarinoro22 suggests that prolonged courses of bacteriostatic antibiotics, including doxycycline, may indeed be chlamydicidal. It is possible that a shorter duration of exposure would be effective against the occulogential tract serovars D-K, as serovar L2 in vivo requires a longer duration of therapy than serovars D-K.2 Azithromycin is concentrated in the tissues and has a half life of 68 h.23 Data from respiratory tract infection studies using a 1.5 g dose of azithromycin administered over 3–5 days show therapeutic levels of azithromycin in target tissues for up to 10 days.23 24 Given its half life, increasing the dose of azithromycin to 3 g (1 g single dose then 500 mg once daily for 4 days) total would be likely to maintain tissue levels for over 12 days. Azithromycin is now off patent and so the additional cost of such a regimen is likely to be minimal.

Doxycycline (100 mg twice daily for 7 days) appears to be an effective treatment for urethral and cervical chlamydia infection with a failure rate of probably <5%.1 8 However, studies using doxycycline often involve individuals with acute infection.1 8 It is possible that doxycycline may be more efficacious than azithromycin in individuals with acute disease, but not in those chronically infected. Chlamydia has a biphasic life cycle involving the extracellular infectious elementary bodies and intracellular replicative reticulate bodies.19 C trachomatis has an intracellular replication cycle of about 48 h.13 19 It can however be induced to exist in a chronic persistent state (aberrant reticulate bodies) without the production of infectious elementary bodies: latent infection.19 Doxycycline, in vitro, has a lower minimal chlamydicidal concentration in cells acutely infected compared with cells latently infected.21 Although it is unknown whether chlamydia can establish latent infection in vivo,19 if it does occur, this is more likely to be present during chronic rather than acute infection. Thus, it is possible that doxycycline may have a higher failure rate in individuals chronically infected but there is no in vivo evidence to support this hypothesis at present.

Finally, although there have not been any published cases of resistant isolates with known mutations conferring azithromycin resistance in vivo, we should not be complacent. This may simply be because so few centres offer chlamydia antimicrobial resistance assays. Azithromycin resistance can be selected for in vitro but this seems to be associated with reduced fitness, that is, a slower replication rate, which is likely to mitigate against rapid dissemination within a community.25 There is one report of women with recurrent infection having reduced antimicrobial sensitivity, but unfortunately, macrolide resistance genotyping has not been performed in these women.26 This work needs confirmation before we can conclude that azithromycin genotypic resistance can occur in vivo without concomitant reduced fitness, but it is a cause for concern. One approach would be for individuals with apparent azithromycin treatment failure who have not been at risk of re-infection to be assessed for antimicrobial resistance and then genotyped if positive. Such work is ongoing at the Sexually Transmitted Bacterial Reference Laboratory, Health Protection Agency, London.

Treatment failure and/or re-infection facilitates ongoing transmission and is likely to put women at increased risk of developing long term chronic sequelae.27 28 In the short term, we should seriously consider repeat testing following treatment with single dose azithromycin as suggested by Handsfield.4 False positive results may occur up to 3 weeks post treatment due to persistent DNA29 and as persistence may take a number of weeks to emerge, follow-up testing should be deferred for at least 5 weeks.3 Research into improving our understanding of chlamydia antimicrobial resistance in vivo and in vitro is urgently needed and should include prospective studies of chlamydia load prior to treatment in vivo.7 Azithromycin should not be discounted in the management of chlamydia as it has excellent in vitro activity13 21 and performs better than doxycycline in animal studies.7 30 Evidence from trachoma (serovars A-C) suggests that increasing the dose of ‘single dose’ treatment may not be effective31 and would increase side effects.32 In vitro data suggest that prolonged exposure of C trachomatis to an antimicrobial is required for optimum efficacy which is consistent with animal studies.7 22 30 Perhaps this is not surprising given the complex life cycle of C trachomatis. Randomised controlled studies of short course (1.5–3.0 g over 3–5 days) versus single dose azithromycin therapy are now indicated.