Fluoroquinolone Pharmacodynamics

Sir—We read with interest the paper by Schentag et al. [1], which presented the authors' perspectives on the fluoroquinolones, their pharmacodynamic relationships, and their associations with various outcome end points. A little light might be shed with advantage upon the inconsistencies of the terminology, methods, and conclusions made by the authors. First, the authors cite their own work regarding use of the ratio of the antibiotic area under the curve (AUC) to the organism's MIC (AUC/MIC) for combination therapy [2]. However, they do not cite the work of Mouton et al. [3], which demonstrates that their method of adding AUC/MIC ratios for combination therapy is less predictive than employing the true dynamically linked variables in the analyses. Second, Schentag and colleagues refer to the article by Thomas et al. [2] as a “study” ([1], p. S41). In reality, it is a compilation of multiple protocols that were well separated in time and did not have consistent end points, because some were done before the Infectious Diseases Society of America/US Food and Drug Administration guidelines on the evaluation of anti-infective drug products [4] were published, and some were done afterward. These guidelines changed the definition of outcome. We feel that combining data from cases in the early 1980s (the cefmenoxime data) and from cases from as late as the early 1990s may lead to poor analyses, because both outcome definitions and the overall care of patients changed during this time. We would hesitate to use “study” or “meta-analysis” as a descriptor, because definitions of outcome changed demonstrably. Third, when discussing the study by Preston et al. [5], Schentag and colleagues cite values for the AUIC (area under the inhibitory concentration curve). AUIC values were not referred to in this article. The values that the authors call “AUIC” were actually AUC/MIC ratios. Fourth, Schentag and colleagues cited the data of Preston et al. [5] with regard to pneumococcal pneumonia, indicating that it “did not allow definition of the low AUIC [sic] threshold because there were no well documented microbiological failures” ([1], p. S43). In fact, there were no microbiological failures at all, either well or poorly documented, for the 20 patients with cases of Streptococcus pneumoniae pneumonia (3 of which were bacteremic). We would point out to the authors that this outcome flies in the face of their “threshold” AUC/MIC ratio of 100–125. There were 3 patients with an AUC/MIC ratio of <50 (1 ratio as low as 25), and 8 others with a ratio of <100. With use of an AUC/MIC breakpoint of 100, there would be 9 treatment successes and 11 treatment failures; with use of a breakpoint of 125, there would be 8 successes and 12 failures. Treatment failed for none of the patients. If the breakpoint of 100–125 is to have any utility for pneumococcal pneumonia, it should have good sensitivity and specificity. In fact, the sensitivities for the breakpoints of 100 and 125 are 45% and 40%, respectively. The specificity is undefined (0/0)! Furthermore, we would ask from which peer-reviewed journal the authors obtained these data, because they were most assuredly not broken out in this way in the study by Preston et al. [5], and Schentag and colleagues presented no other citation. Fifth, Schentag and colleagues also refer to “lower fluoroquinolone MICs in the early 1990s” ([1], p. S43). Again, we wonder where they obtained their data; this statement is not referenced. Examination of data from 1996–1997 (9145 strains of S. pneumoniae [6]), 1997–1998 (4148 strains [7]), and 1999–2000 (1531 strains [8]) shows MIC90 values of 2.0, 1.0, and 1.0 µg/mL, respectively, and fully sensitive rates of 97.3%, 99.8%, and 99.3%, respectively. Sixth, Schentag and colleagues indicate that Preston et al. [5] “maintain that an AUIC [sic] value of 30 is sufficient, presumably on the basis of in vitro studies” ([1], p. S43), and, for the latter part, they cite the study of Lacy et al. [9]. We are confused. Two of us (S.L.P. and G.L.D.) authored the paper in question, and we cannot recall ever writing that in any draft of the manuscript, nor do we recall referring to the study of Lacy et al. [9], which was not yet published at that time. While we agree that exposure to larger doses of drugs, relative to the MIC, is better, this is a trivial argument, because drugs have toxicities. The important issue is to define the drug exposure that kills infecting organisms and prevents emergence of resistance without causing a toxic reaction in the patient. The authors cite only the study of Lacy et al. [9], but there are now multiple in vitro and animal studies [10–12], as well as additional data from clinical trials [13], that support using a lower AUC/MIC breakpoint for pneumococcal pneumonia. For instance, one of us (P.G.A.) has examined the relationship between the free-drug AUC/MIC ratio and the response of 58 adult patients enrolled in either of 2 phase III, randomized, multicenter, double-blind studies of levofloxacin or gatifloxacin for the treatment of community-acquired pneumonia or acute exacerbation of chronic bronchitis associated with S. pneumoniae [13]. This study detected a statistically significant (P = .013) relationship between microbiological response to therapy and the free-drug AUC/MIC ratio. At a free-drug AUC/MIC ratio of <33.7, the probability of a positive response was 64%, and, at a free-drug AUC/MIC ratio ⩾33.7, it was 100% (P < .01). These data simply do not support the use of an AUC/MIC breakpoint of 100–125 for treatment of S. pneumoniae infection with fluoroquinolones. In summary, there are a number of methodological problem, misstatements, and mistaken or missing citations in the article by Schentag et al. [1]. It is time that we move from the one-size-fits-all approach to breakpoints or thresholds and opt instead to accept that there are organism-specific and host-specific relationships to be considered. In the interest of advancing the science of pharmacodynamics, we would be interested in any clarification that Schentag and colleagues could provide.