Extended-Spectrum Beta-Lactamase (ESBL)-Producing Escherichia coli versus Klebsiella pneumoniae: Does type of germ really matter?

To the Editor—We read with great interest the recent article by Scheuerman et al ¹ showing that extended-spectrum β-lactamase (ESBL)-producing Escherichia coli (ESBL-EC) and Klebsiella pneumoniae (ESBL-KP) bloodstream infections (BSIs) differ significantly in terms of mortality (33.7% vs 17.4%; P=.016). Because their study concerns a highly relevant and popular topic, some points should be discussed.

First, ESBL-KP–infected patients were more often hospitalized in ICU than those infected by ESBL-EC (P<.001), partly due to a septic shock, which may explain such a high rate of mortality (33.7%) for a bloodstream infection (BSI). Indeed, the observed mortality rate for ESBL-KP was similar to the average mortality for those with gram-negative BSIs in the ICU (35%) according to the prospective EUROBACT International cohort study. ² Also, ICU-acquired BSIs are associated with a 40% increase in the risk of 30-day mortality. ³ Therefore, it is hard to believe that such a difference could be accounted for in any statistical adjustment, and thus, it constitutes a selection bias.

Second, the main source of BSI was urinary tract in the ESBL-EC arm (P=.005), while it is acknowledged that the severity of urinary tract infection is not related to the presence of bacteremia. ⁴ Such data underly the hypothesis that ESBL-KP infections might have been more severe than those due to ESBL-EC. For instance, multidrug-resistant BSIs complicating respiratory tract infections have been associated with an increased mortality (odds ratio [OR], 3.26; 95% confidence interval [CI], 1.29–8.22). ⁷

Third, no information is provided about the respective antimicrobial regimens between ESBL-EC and ESBL-KP patients. However, it is currently argued that carbapenem alternatives are associated with a higher mortality rate than carbapanems for the treatment of ESBL BSI. In fact, the MERINO trial by Harris et al ⁵ was recently suspended due to an increase in mortality in the arm receiving piperacillin-tazobactam (12.3%) versus meropenem (3.7%). ⁵ Such data should have been discussed. Likewise, no information is provided on treatment duration or dose, which may have varied between the 2 groups in the present study. ¹ Both factors play a role in the outcome of treatment, especially when used against multidrug-resistant organisms. ⁶

Interestingly, we previously showed that BSI severity or mortality among spinal cord injury patients over 15 years was not related to the multidrug-resistant characteristics of the microorganism. ⁷ Although our sample size was small (n<30), a closer look at the outcome between ESBL-EC (n=26) and ESBL-KP (n=13) did not reveal any statistical difference in terms of mortality rate (7.7% in each arm). Moreover, the mortality rates were similar for other ESBL microorganisms (Enterobacter spp, Morganella spp, and Proteus spp (n=21)), ~9.5% (P=.99, data not shown).

In fact, we believe that the findings of Scheuerman et al, which showed no impact of CTX-M isolates in comparison to other ESBL genotypes, might support the idea that the type of germ does not play a major role. Indeed, mortality seems more related to patient comorbidities and severity of infection, as shown in Table 2 of the article, ¹ with significant discrepancies between the 2 groups in terms of length of stay to bacteremia (P=.017), source of infection (P=.005), ICU ward admission (P<.001) and underlying cardiovascular disease (P<.001). Moreover, in a rabbit model of sepsis induced by a multidrug-resistant Klebsiella pneumoniae, Zhou et al ⁸ showed that mortality was higher for the rabbits infected by susceptible than those infected with multidrug-resistant strains.

Overall, the impact of ESBL-KP isolates on mortality rate might have been overestimated, in the light of the severity of the patient condition.