To the Editor—Treating patients with multidrug-resistant (MDR) or extensively drug-resistant (XDR) tuberculosis is long, expensive, and complicated, particularly when 4 active drugs recommended to design an effective regimen are missing [1–5]. New drugs (ie, bedaquiline and delamanid [6, 7]) and a few repurposed ones (linezolid [8, 9] and carbapenems [10–12]) are attracting interest. To date, the largest clinical study evaluating imipenem-clavulanate (IC) in the treatment of MDR tuberculosis included 10 cases [12].
The aim of our observational study was to compare the therapeutic contribution (effectiveness, safety, and tolerability profile) of IC added to an optimized background regimen (OBR) designed according to World Health Organization guidelines [4], compared with an OBR control group, in the treatment of MDR/XDR tuberculosis cases. Between 2003 and 2015, a total of 84 consecutive patients treated with IC-containing regimens were compared with 168 controls recruited by the International Carbapenems Study Group in 21 centers and 8 countries in Europe and Latin America [10, 11]. Data were analyzed from patients with culture-confirmed MDR tuberculosis with mycobacterial strains resistant to at least isoniazid and rifampicin.
An OBR regimen was administered after drug susceptibility testing carried out by externally quality-assured laboratories [10, 11]. Physicians prescribed antituberculosis drugs without any compelling criteria for experimental protocols, so blinding and randomized methods were not followed. Imipenem was administered at a dose of 500 mg 4 times a day for a median (interquartile range [IQR]) of 187 (60–428) days.
Patients treated with IC had more previous exposures to antituberculosis drugs lasting >1 month (median [IQR], 2 [1–3] vs 1 [0–2]) and higher number of resistances (8 [7–8] vs 5 [4–6]) than OBR controls. They also showed more resistance to fluoroquinolones (79.0% vs 16.8%), amikacin (50.0% vs 13.0%), kanamycin (75.8% vs 18.2%), and capreomycin (63.9% vs 13.5%) and a higher prevalence of XDR tuberculosis (67.9% vs 6.0%) than OBR controls (all P < .001).
The median (IQR) time to sputum smear conversion was similar in IC-treated patients and controls (30 [30–60] vs 30 [0–56] days; P < .001), whereas the time to culture conversion was longer in IC-treated patients, although not significantly so (60 [30–90] vs 42 [30–90] days; P = .08; Figure 1). Sputum smear and culture conversion rates were lower in IC-treated patients than in OBR controls (sputum smear, 79.7% vs 98.0%; culture conversion, 71.9% vs 100.0%; both P < .001), as were success rates (59.7% vs 85.8%; P < .001), whereas death (17.9% vs 1.8%; P < .001) and failure (6.0% vs 0.0%; P < .001) rates were higher. Adverse events (minor) due to IC were reported in only 5.4%.
Time to sputum culture conversion in patients with multidrug-resistant tuberculosis exposed or not exposed to imipenem-clavulanate (P = .77).
Time to sputum culture conversion in patients with multidrug-resistant tuberculosis exposed or not exposed to imipenem-clavulanate (P = .77).
Our findings show that IC-containing regimens achieved nondifferent or worse results than IC-sparing ones. Perhaps the severity of IC-treated cases favored the administration of a carbapenem in the absence of alternatives, so selection bias related to the retrospective, observational nature of the study cannot be excluded. Interestingly, the success rates among IC-treated patients is similar to that achieved in other major international MDR tuberculosis cohorts, confirming that IC may have a role in MDR tuberculosis treatment [1, 2, 12]. To our knowledge, this is the first large study evaluating the effectiveness, safety, and tolerability of IC-containing regimens while comparing their clinical performance with outcomes in a control group.
Note
Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
Comments