Interpretation of Pneumococcal Susceptibility Tests with Cefaclor

Chiou et al. (1) have reported an alarming incidence of β-lactam resistance among Streptococcus pneumoniae isolates in Taiwan, Republic of China. There are two points relating to data interpretation in the article that should be discussed and clarified. The first point regards application of nonvalidated interpretive criteria to cefaclor MIC data for pneumococci. In Materials and Methods, the authors indicate that they interpreted the cefaclor MICs using breakpoints recommended for “other cephalosporins” (≤0.5 = susceptible [S]) because the National Committee for Clinical Laboratory Standards (NCCLS) has not established pneumococcal breakpoints for this cephalosporin (3). Such a practice is not justified. Susceptibility breakpoints are established and verified through careful consideration of the native in vitro activity of the compound, its in vivo distribution and pharmacokinetics following the recommended doses, and its clinical efficacy against indicated infections due to target pathogens having known in vitro responses to the antibiotics (2). Thus, it is not acceptable to apply interpretive criteria developed for one agent to the test results of another without first verifying that the breakpoints should be the same for both agents.

It should be noted that recent editions of the NCCLS approved standards for dilution susceptibility tests contain a table of criteria specific for pneumococci. After the appearance of significant numbers of non-penicillin-susceptible pneumococci (NPSP), criteria for cefaclor and several other β-lactam antibiotics were removed from the table because the committee had not seen clinical data to validate interpretations for these agents when tested against NPSP (3). A footnote to that table instructs the user to inform the physician that penicillin-susceptible pneumococci can be safely assumed to be susceptible to those deleted agents but that susceptibility to those agents among NPSP is unknown.

The second point pertains to apparent use of two different sets of cefaclor interpretive criteria for the same set of data. The authors seem to have applied two different sets of breakpoints to the same set of data reported in their Tables 1 and 2. Data for the susceptibilities of 584 S. pneumoniae isolates to the various tested antibiotics are given in Table 1 in the paper by Chiou et al. (1). For cefaclor, there are 318 isolates tabulated as susceptible, 6 tabulated as intermediate, and 260 tabulated as resistant. In Table 2, the data for the same 584 isolates are stratified by penicillin susceptibility. In that table, there is only one isolate (a Pen^s isolate) listed as cefaclor susceptible, 240 are listed as cefaclor intermediate, and 343 are listed as cefaclor resistant. One possible explanation for this discrepancy would be the application of different breakpoints for the two tables. It seems possible that the summary data for cefaclor in Table 1 were based on the breakpoints recommended by the NCCLS for cefaclor in 1993 (≤8 μg/ml = S) (4). It is likely that the cefaclor susceptibilities summarized in Table 2 are the result of application of the “assumed” breakpoints mentioned in the Materials and Methods (≤0.5 μg/ml = S). Considering the well-established efficacy of cefaclor against Pen^s pneumococci, it should be clear from these data that application of the breakpoints for other cephalosporins to cefaclor MIC misclassifies nearly all of the Pen^s isolates as other than susceptible to cefaclor.

In addition to contributing to confusion by reporting different degrees of susceptibility in two places in the paper, an undesirable precedent is set by the assumption of breakpoints for a compound that, for very valid reasons, has none at this time. Until the NCCLS sets validated breakpoints, all Pen^s pneumococci should be considered susceptible to cefaclor and non-penicillin-susceptible isolates should not be classified.