Bacterial Proliferation May Be the Key Component of Sepsis Mortality

LETTER

Halbach et al. (1) reported that antibiotic administration had no significant effect on mortality in a mouse model of intra-abdominal infection (IAI) using cecal ligation and puncture. Furthermore, ligation of the cecum without puncture followed by the introduction of sterile feces into the peritoneal cavity induced mortality in mice. The authors claimed that sepsis mortality is not fully related to bacterial proliferation but depends on bacterial components. These results could be due to the strategy of antibiotic administration and to the low number of mice per group. None of the 4 strategies of antibiotic administration that were used could be considered effective. First, subcutaneous administration is not optimal in terms of pharmacokinetics (2). Second, imipenem was clearly underdosed (only 1 dose of antibiotic in one strategy and 3 doses in the other three strategies). None of these strategies were in agreement with the 2016 recommendations for management of intra-abdominal infections (3). At least 3 to 5 days of antibiotics is recommended if the source control is complete, and a longer duration is recommended if the patient presents with persistent sepsis. Indeed, inappropriate antibiotic therapy is a well-described factor of mortality from IAI (4). Moreover, as the peritoneal bacterial load was not evaluated, it is impossible to affirm that mice did not die from bacterial proliferation. Furthermore, the human microbiota is different from the mouse microbiota. The lack of microbiological data limits generalization.

Moreover, the absence of difference is not the proof of the absence of difference. Indeed, the number of mice greatly impacts statistical analysis, as the P value depends directly on the number of observations. If more mice were included, an effect of antibiotic may have been present. For a same mortality curve, the P value decreases as the number of mice increases. In Fig. 1B in the paper by Halbach et al., the difference between the antibiotic group and the control group is not significant for 9 mice per group (P = 0.15) but might have been significant if more mice would have been included (for 18 and 27 mice per group, P = 0.037 and P = 0.010, respectively). Additionally, in Fig. 1C in that paper, there is no significant difference in mortality between groups. However, at 48 h, when antibiotics are stopped, a difference between groups may have been observed for at least 18 mice per group (P = 0.014). This could explain why several studies used up to 20 mice per group (5) to show a significant difference.

Finally, the difference in mortality between mice treated with feces sterilized and not sterilized by UV radiation is highly significant (see Fig. 4 in reference 1). This suggests that bacterial proliferation plays a key role in mortality. Fecal contamination during inoculation into the peritoneal cavity may have influenced the results and would explain the delayed mortality in the sterilized-by-UV group. This contamination is likely as working under sterile conditions is difficult in a biological laboratory setting. To assess the effect of contamination, an evaluation of peritoneal bacterial load was necessary.

In conclusion, additional experiments using appropriate antibiotic therapies, additional mice, and microbiological analysis may yield results different from those described in this study.