Significance of macrolide resistance in Streptococcus pneumoniae.

Sir, We wish to comment on the paper by Van Kerkhoven et al.,1 to dispel certain misconceptions that may arise from a cursory review of their study. The authors present a 3 year retrospective chart review of patients hospitalized with pneumococcal bacteraemia at a single hospital. Of the 136 patients identified, 14 (10.3%) and 33 (24.3%) had isolates non-susceptible to penicillin and erythromycin, respectively. Of the erythromycin-resistant isolates, 94% had an MLSB-resistance phenotype and high-level erythromycin resistance (MIC > 64 mg/L). Further analysis was limited to 12 patients who received ≥2 days of oral antibiotic therapy before admission, and were therefore considered to have breakthrough bacteraemia. Four of the 12 patients received clarithromycin before admission and eight received β-lactams (five were receiving co-amoxiclav). It should be noted that all four isolates from patients receiving clarithromycin were highly resistant to erythromycin (MIC > 256 mg/L), whereas all five isolates from patients receiving co-amoxiclav were fully susceptible to penicillin (MIC ≤ 0.016 mg/L). Once hospitalized, all patients received highdose β-lactams. Three elderly patients (one receiving clarithromycin before hospitalization, two receiving cefadroxil) died within 3 days of hospitalization. The authors concluded that breakthrough pneumococcal bacteraemia during macrolide therapy was associated with macrolide resistance, suggesting that macrolide resistance is clinically relevant. On the other hand, breakthrough bacteraemia during treatment with co-amoxiclav was not associated with penicillin resistance, a finding attributed to the inadequate pharmacokinetics of co-amoxiclav 500/125 mg taken twice daily. First of all, it is no surprise that the authors were able to find instances of breakthrough bacteraemia associated with high-level macrolide resistance. The MIC for each isolate greatly exceeded the achievable levels of clarithromycin, both in serum and in the lung. Indeed, it is more notable that only four cases could be identified at a large referral hospital over a 3 year period marked by ‘extensive use of macrolides’ and macrolide resistance rates above 30%. If for every third case of pneumococcal pneumonia treated with macrolides ‘macrolide resistance ...leads to treatment failure’, then many more clinical failures would be expected. Although the denominator of patients with macrolide-non-susceptible pneumococcal pneumonia treated with macrolides is not known, the small number of clinical failures would appear, in fact, as an endorsement of macrolide efficacy against macrolide-susceptible and, apparently, some macrolidenon-susceptible infections. It is also important to note that two or three of each of the patients failing macrolide or β-lactam therapy would have met accepted severity criteria predicting a high risk of death, thereby justifying admission and combination therapy with a β-lactam plus a macrolide or, alternatively, with a fluoroquinolone, until culture and susceptibility results were available. Although these cases were used to illustrate failure of the antibiotic, the outcomes may have been preventable with better medical decision-making. Most importantly, because macrolide resistance among pneumococci in Belgium is dominated by the erm(B)-mediated mechanism, the study does not shed light on whether efflux-mediated resistance is clinically relevant. In their discussion, the authors cite a study by Lonks et al.2 as demonstrating that both low-level (that is, effluxmediated) as well as high-level macrolide resistance is responsible for therapeutic failure. That study, however, included only one patient isolate with an MIC < 16 mg/L, a value more reflective of low-level resistance. In fact, there is no convincing evidence that such lowlevel macrolide resistance increases the risk of macrolide failure. On the other hand, recent evidence from animal models suggests that isolates with MICs up to 8 mg/L may respond to clarithromycin.3 Such evidence is supported by the concentrations of drug achievable in the alveolar epithelial lining fluid and lung parenchyma of healthy human volunteers.4 This unresolved conundrum has important implications in North America, where efflux mechanisms account for 61%–85% of macrolide-resistant isolates.5 In other words, if pneumococci with MICs ≤ 8 mg/L are actually treatable with clarithromycin, then clinically significant macrolide resistance occurs in <10% of invasive pneumococcal isolates. Finally, we are reluctant to accept that the inadequate pharmacokinetics of co-amoxiclav 500/125 mg twice daily was responsible for treatment failures in five non-elderly patients infected with fully susceptible organisms, and with little apparent co-morbidity. Even this dosage exceeds well-accepted pharmacodynamic parameters predictive of efficacy (for example, serum concentrations >MIC for >50% of the dosing interval).6 If marginal pharmacokinetics were the issue, organisms with reduced susceptibility to β-lactams should be over-represented among breakthrough infections. Whereas we are in full agreement with the authors that all antimicrobials should be prescribed for maximal effectiveness rather than convenience, the failures on co-amoxiclav therapy point out the inadequacies of anecdotal reports of treatment failure. The fact is that treatment of bacteraemic pneumococcal pneumonia fails sometimes, whether due to inherent bacterial virulence, or issues related to the host or antibiotic regimen. This possibility makes anecdotes less meaningful. As a result, the collective literature leaves us in no position to determine whether efflux-mediated macrolide resistance in pneumococci is clinically relevant.