Clostridium difficile infection (CDI) is the most common cause of nosocomial diarrhoea in the developed world, with complications of the disease, including potentially life-threatening pseudomembranous colitis and toxic megacolon.¹ The total costs of CDI treatment are estimated to be as high as $2871.00 (United States dollars) to $4846.00 per patient in the United States and between $5243.00 and $8570.00 per patient in European countries,² and there is evidence to suggest that treatment failure and disease relapse rates are increasing.³ Since the turn of the century, several significant outbreaks of CDI have occurred and have been documented in North America and Europe.^4,5,6

Toxigenic C. difficile strains produce two large clostridial toxins, TcdA and TcdB, both of which show cytotoxic activity and are responsible for the majority of disease symptoms.⁷ A third binary toxin, CdtAB is also produced by some strains, such as those belonging to ribotype 027 (RT027) and RT078, and may be associated with increased disease severity in some settings.⁸ Several toxin variant strains have also been identified. The most common of these is a subset of strains that produce only one functional toxin (TcdB) and is mainly comprised of members of the RT017 group. This group is widespread in Asia⁹ and is capable of causing severe disease across diverse populations.⁶

The most common risk factor for the development of CDI is previous exposure to antibiotics, particularly clindamycin and cephalosporins and, less frequently, fluoroquinolones.¹⁰ The resulting dysbiosis allows proliferation of C. difficile and the progression of disease symptoms. Additional risk factors include advanced age,¹¹ gastric acid suppression therapy¹² (although this is sometimes debated¹³), co-morbidities such as HIV¹⁴ and exposure to long-term healthcare facilities.¹⁵ There have also been several case reports of CDI occurring in patients with tuberculosis,^16,17,18,19 possibly related to the long-term intensive antibiotic therapy that they typically receive. However, there have been very few studies to date specifically looking at CDI in this patient group. Chang et al.²⁰ examined a potential link between fluoroquinolone use and CDI in tuberculosis patients in Hong Kong and concluded that the risk of CDI was moderate for these patients, while Lee et al.²¹ detected a relatively low incidence of CDI in Korean tuberculosis patients (2.83 cases per 1000 adults). More recently, however, Legenza et al.²² identified tuberculosis as an independent risk factor for CDI in patients in Cape Town, South Africa, and there has been at least one report of a CDI outbreak occurring in a tuberculosis hospital in the Eastern Cape province,²³ suggesting the need for further research in this area.

While the overall burden of CDI in sub-Saharan Africa is largely unknown, data from the limited number of available studies suggest that the prevalence of CDI in the region is comparable to that of high income countries in Europe and North America.^24,25 In South Africa, studies carried out in the Vhembe district of Limpopo province and Cape Town in the Western Cape report a prevalence of toxigenic C. difficile of between 10% and 20% in patients with diarrhoea.^26,27,28 However, at least two of the previous studies undertaken in the country relied on diagnostic testing that detected toxin A alone, meaning that the rate of CDI in the country may be underestimated.^27,29 Additionally, epidemiological information regarding CDI and the strains responsible for disease is currently lacking, particularly for HIV-positive/tuberculosis-positive patients, who may be at an increased risk of developing CDI. Therefore, the aim of the current study was to examine CDI in patients attending two specialist tuberculosis hospitals in Cape Town and to determine the molecular epidemiology of strains isolated from these patients.

There is a clear need for further study and increased surveillance of CDI in sub-Saharan Africa. Both hospital and community-acquired CDI cases have the potential to pose significant challenges to healthcare systems and there are currently no data on the economic burden of CDI outbreaks in the region. An inclusive ‘One-Health’ approach involving researchers at multiple levels in the community and hospital environment is important. A first step in this direction is the establishment of systematic sentinel surveillance of CDI in high-risk populations to monitor the incidence of CDI, as well as the predominant strain types and circulating antibiotic resistance phenotypes. Similar programmes in high income nations have been successful in reducing both the incidence of infection and mortality due to CDI⁴⁴ and would be of benefit to countries in sub-Saharan Africa.

In the current study, CDI cases were examined in at-risk populations of patients undergoing long-term anti-tuberculosis therapy. CDI was common at both tuberculosis institutions with at least four wards experiencing spikes in the infection rates during the study period. Three of these wards housed patients undergoing DS tuberculosis treatment regimens, while one housed patients receiving MDR tuberculosis therapy. Although C. difficile outbreak definitions vary across countries, a rate of five or more cases in a single ward with at least 20 beds over a four-week period is generally used as a threshold to initiate outbreak investigations. That this occurred in four different wards during the study is concerning and suggests the need for enhanced surveillance of C. difficile in these hospitals to help to prevent further infections.

Toxin A-negative, toxin B-positive RT017 strains made up the majority of tuberculosis patient isolates. Strains belonging to this ribotype have been implicated in outbreaks in Canada,⁴⁵ China,⁴⁶ Korea,⁴⁷ Argentina,⁴⁸ Israel,⁴⁹ Japan⁵⁰ and Europe^6,51 and they are often resistant to multiple drugs.^42,52 Moreover, in previous studies RT017 strains have been shown to have a similar 30-day mortality rate of the so-called ‘hypervirulent’ RT027 strains.^5,53 RT017 strains were also commonly isolated from other hospitals in Cape Town during the same time period, suggesting that these strains are circulating more broadly in the Cape Town patient population.^30,54 It is possible that the previous use of standalone enzyme immunoassay diagnostic tests that target toxin A alone allowed RT017 strains to proliferate undetected in the region, providing a reservoir of strains in the healthcare environment with the potential to cause outbreaks in susceptible patient populations.

Several RT017 strains isolated from both hospitals were highly related by MLVA, suggesting possible patient-to-patient strain transfer. In a large scale epidemiological analysis of C. difficile strains infecting patients in England, Eyre et al.⁵⁵ defined several possible transmission relationships for patients with clonally related C. difficile strains. For hospital-acquired cases, these included ‘ward contacts’ (two or more patients occupying the same ward over a concurrent time interval that allowed for direct patient-to-patient transmission), ‘ward contamination’ (patients occupying the same ward but with an interval of 1–28 days separating the discharge or end of infectivity of the initial case and the admission of patients who subsequently developed CDI) and ‘hospital contacts’ (patients in different wards at the same hospital over a similar time interval to ward contacts). Unfortunately, detailed ward occupation data and the duration of disease symptoms were not available for patients in this study and it is, therefore, not possible to differentiate between the three different transmission relationships in the current analysis. Nevertheless, the isolation of several clonally related strains from samples submitted by patients in the same ward within a 28-day time interval suggests that ward contact or ward contamination occurred along with possible hospital contact for samples from patients occupying different wards. Interestingly, strains with identical MLVA profiles were also isolated from different patients at longer time intervals of three to seven months. This has been observed previously⁵⁵ and may be due to the formation of spores that persist in a genetically quiescent state on surfaces in the hospital environment.

Metronidazole is the recommended antibiotic for initial episodes of CDI, with vancomycin reserved for cases that do not respond to initial treatment and for cases of recurrent disease. No resistance to either of these antibiotics was observed for local isolates. However, a small number of strains exhibited a slightly elevated metronidazole MIC of 4 mg/L. This may be clinically relevant as the maximum level of metronidazole that can be maintained in the gut during oral therapy ranges between 0.25 mg/L and 9.5 mg/L.¹ Fidaxomicin, an alternative treatment for CDI with a narrow spectrum of activity, is not yet widely available for use in South Africa.

Apart from one non-RT017 isolate, all C. difficile strains isolated from patients attending specialist tuberculosis hospitals were resistant to rifampicin. This included samples from both DS tuberculosis and MDR tuberculosis patients and contrasts with a concurrent set of isolates obtained from patients attending other hospitals in the Cape Town area (37% of strains resistant to rifampicin).³⁰ It is also higher than that observed for strains isolated from the general hospital population in Europe (0% – 63.64%).⁴² Rifampicin is included as part of the standard DS tuberculosis treatment regimen and exposure to antibiotics belonging to this class has been associated with the development of resistance in C. difficile.⁵⁶ Similar results have been observed for RT046 strains isolated from a tuberculosis hospital population in Poland.¹⁶ Resistance to rifamycin in C. difficile is usually associated with mutations in the rpoB gene, which typically confer cross-resistance to multiple members of the antibiotic class and, importantly, are associated with a low fitness burden, allowing the phenotype to be stably maintained in the bacterial population.⁵⁷

Fluoroquinolone resistance was also common among C. difficile strains isolated from tuberculosis patients. Previous exposure to fluoroquinolones is a recognised risk factor for the development of CDI, and it is thought that acquisition of resistance to third and fourth generation fluoroquinolones by RT027 strains was a contributing factor to several hospital outbreaks in North America during the early 2000s.^4,58 In South Africa, moxifloxacin is included as part of the MDR tuberculosis treatment regimen, and all C. difficile strains isolated from MDR tuberculosis patients were resistant to the antibiotic. Resistance to fluoroquinolones in C. difficile is chromosomally encoded,⁵⁹ readily develops following exposure to this class of antibiotics⁶⁰ and is associated with a low fitness cost to the bacterium.⁶¹ Therefore, the development of resistance to fluoroquinolones, as well as rifampicin, is likely to be stably maintained in the C. difficile population circulating among tuberculosis patients.

One reason proposed for the relatively low rates of CDI in patients undergoing anti-tuberculosis therapy in previous studies is that rifampicin is often effective against C. difficile and may help to protect against CDI in these patients.²⁰ Since this protection would be lacking in a background of circulating rifampicin-resistant strains, the presence of rifampicin resistance among the RT017 isolates in the current study is significant. Moreover, co-resistance to fourth generation fluoroquinolones would allow the same strains to persist in patients undergoing MDR tuberculosis treatment regimens. Together, these results may help to explain the increased risk of CDI in tuberculosis patients in Cape Town²² compared to previous studies in Hong Kong²⁰ and Korea²¹ and confirm the need to monitor this patient population more carefully.