1219. Unfavorable Clinical Outcomes with Polymyxins Compared to Ceftolozane/Tazobactam for the Treatment of Carbapenem-Resistant Pseudomonas aeruginosa

Abstract Background Patients with carbapenem-resistant Pseudomonas aeruginosa (CRPA) have high in-hospital mortality rates. It is unknown if patients with CRPA treated with ceftolozane/tazobactam (C/T) have improved clinical outcomes compared to those treated with polymyxins. Methods The CDC-funded, Georgia Emerging Infections Program performed active population- and laboratory-based surveillance for CRPA isolated from sterile sites, urine, lower respiratory tract and wounds in metropolitan Atlanta, GA from 8/1/2016–7/31/2018. We reviewed charts of adults without cystic fibrosis who were hospitalized within 1 week of CRPA culture. Using a desirability of outcome ranking (DOOR) analysis which incorporates both benefits and risks into a single outcome, we estimated the probability that a patient treated first with C/T would have a more desirable clinical outcome at 30-days than a patient treated with polymyxins (polymyxin B or colistin). We adjusted for confounding using inverse probability of treatment weighting (IPTW) based on culture source and need for dialysis at baseline. A partial credit analysis allowed for variable weighting of DOOR ranks and calculation of differences in mean partial credit scores. Results Among 710 cases from 18 different hospitals, we identified 73 patients treated for CRPA infections with polymyxins (n=31) or C/T (n=42). Most patients were male (64%) and Black (80%), and those receiving polymyxins were more likely to have required dialysis at baseline (35% vs. 14%, p=0.03) (Table 1). At 30 days after culture, 34 (47%) were alive with no adverse events, 21 (29%) were alive with ≥ 1 adverse event, and 18 (25%) had died. Patients first treated with C/T had a lower 30-day mortality rate than those treated with polymyxins (14% vs 39%, p=0.03). Additionally, those receiving C/T had better overall clinical outcomes, with an adjusted DOOR probability of having an improved outcome of 67% (95% CI 53%–80%) compared to those receiving polymyxins (Figure 1). Partial credit analyses indicated consistent results across different patient values of survival with adverse events (Figure 2). Figure 1: Inverse probability of treatment weighting-adjusted desirability of outcome ranking (DOOR) distributions by treatment group, accounting for adverse events and survival status that occurred up to 30 days after CRPA culture. 1. Percentages are adjusted using inverse probability of treatment weighting, controlling for culture source and need for dialysis at baseline 2. Adverse events measured included: acute kidney injury, discharge to higher acuity location than previous residence, or being hospitalized 30 days after culture Figure 2: Inverse probability of treatment weighting-adjusted partial credit analysis. This displays the difference (ceftolozane/tazobactam minus polymyxin) in mean partial credit scores (black line) and associated 95% confidence bands (gray lines) as a function of the partial credit score assigned to an individual having at least one adverse event (range 0 – 100%). A score of 100% is assigned to patients alive with no adverse events and a score of 0% is assigned to patients who die. A difference in mean partial credit scores of approximately zero suggests there was no difference observed between treatment groups. Conclusion These findings support the recent Infectious Diseases Society of America guidance favoring C/T over polymyxins for treatment of CRPA infections. Disclosures David van Duin, MD, PhD, Entasis (Advisor or Review Panel member)genentech (Advisor or Review Panel member)Karius (Advisor or Review Panel member)Merck (Grant/Research Support, Advisor or Review Panel member)Pfizer (Consultant, Advisor or Review Panel member)Qpex (Advisor or Review Panel member)Shionogi (Grant/Research Support, Scientific Research Study Investigator, Advisor or Review Panel member)Utility (Advisor or Review Panel member) Scott R. Evans, PhD, Abbvie (Consultant)Advantagene (Consultant)Alexion (Consultant)Amgen (Consultant)AstraZeneca (Consultant)Atricure (Consultant)Breast International Group (Consultant)Cardinal Health (Consultant)Clover (Consultant)FHI Clinical (Consultant)Genentech (Consultant)Gilead (Consultant)Horizon (Consultant)International Drug Development Institute (Consultant)Lung Biotech (Consultant)Microbiotix (Consultant)Neovasc (Consultant)Nobel Pharma (Consultant)Novartis (Consultant)Nuvelution (Consultant)Pfizer (Consultant)Rakuten (Consultant)Roche (Consultant)Roivant (Consultant)SAB Biopharm (Consultant)Shire (Consultant)Stryker (Consultant)SVB Leerink (Consultant)Takeda (Consultant)Teva (Consultant)Tracon (Consultant)Vir (Consultant)


Session: P-72. Resistance Mechanisms
Background. Patients with carbapenem-resistant Pseudomonas aeruginosa (CRPA) have high in-hospital mortality rates. It is unknown if patients with CRPA treated with ceftolozane/tazobactam (C/T) have improved clinical outcomes compared to those treated with polymyxins.
Methods. The CDC-funded, Georgia Emerging Infections Program performed active population-and laboratory-based surveillance for CRPA isolated from sterile sites, urine, lower respiratory tract and wounds in metropolitan Atlanta, GA from 8/1/2016-7/31/2018. We reviewed charts of adults without cystic fibrosis who were hospitalized within 1 week of CRPA culture. Using a desirability of outcome ranking (DOOR) analysis which incorporates both benefits and risks into a single outcome, we estimated the probability that a patient treated first with C/T would have a more desirable clinical outcome at 30-days than a patient treated with polymyxins (polymyxin B or colistin). We adjusted for confounding using inverse probability of treatment weighting (IPTW) based on culture source and need for dialysis at baseline. A partial credit analysis allowed for variable weighting of DOOR ranks and calculation of differences in mean partial credit scores.
Results. Among 710 cases from 18 different hospitals, we identified 73 patients treated for CRPA infections with polymyxins (n=31) or C/T (n=42). Most patients were male (64%) and Black (80%), and those receiving polymyxins were more likely to have required dialysis at baseline (35% vs. 14%, p=0.03) ( Table 1). At 30 days after culture, 34 (47%) were alive with no adverse events, 21 (29%) were alive with ≥ 1 adverse event, and 18 (25%) had died. Patients first treated with C/T had a lower 30-day mortality rate than those treated with polymyxins (14% vs 39%, p=0.03). Additionally, those receiving C/T had better overall clinical outcomes, with an adjusted DOOR probability of having an improved outcome of 67% (95% CI 53%-80%) compared to those receiving polymyxins ( Figure 1). Partial credit analyses indicated consistent results across different patient values of survival with adverse events (Figure 2). 1. Percentages are adjusted using inverse probability of treatment weighting, controlling for culture source and need for dialysis at baseline 2. Adverse events measured included: acute kidney injury, discharge to higher acuity location than previous residence, or being hospitalized 30 days after culture This displays the difference (ceftolozane/tazobactam minus polymyxin) in mean partial credit scores (black line) and associated 95% confidence bands (gray lines) as a function of the partial credit score assigned to an individual having at least one adverse event (range 0 -100%). A score of 100% is assigned to patients alive with no adverse events and a score of 0% is assigned to patients who die. A difference in mean partial credit scores of approximately zero suggests there was no difference observed between treatment groups.
Conclusion. These findings support the recent Infectious Diseases Society of America guidance favoring C/T over polymyxins for treatment of CRPA infections.  Background. The Clinical and Laboratory Standards Institute (CLSI) lowered MIC breakpoints for many beta-lactam antibiotics to enhance detection of resistance among Enterobacterales. This shift was also meant to eliminate the need for routine testing for extended-spectrum beta-lactamases (ESBLs). The recommended treatment for ESBL-producing Enterobacterales is carbapenems. The IDSA guidelines for MDR-GN organisms recommend using ceftriaxone (CRO) resistance as a proxy for ESBL production and thus carbapenem treatment. Under CLSI guidelines, alternative beta-lactams such as ceftazidime (CAZ) and cefepime (FEP) may still be reported as susceptible and thus used by clinicians even in light of IDSA recommendations. The aim of this project was to characterize the MIC distributions of CAZ and FEP stratified by CRO susceptibility.

Is MIC all that matters? MIC Distributions of Ceftazidime and Cefepime in
Methods. Clinical E. coli, K. pneumoniae, and K. oxytoca isolates from blood cultures in adult patients from Nov 2016-Dec 2018 that had MICs tested by the Vitek-2 automated susceptibility testing system for CRO, FEP and CAZ were identified. Descriptive statistics were used to compare MIC distributions across the antibiotics of interest (SPSS).
Results. 573 isolates were included, of these, 17.3% were CRO resistant. Most (53%) CRO-R isolates had FEP MICs ≤2 which is considered susceptible per CLSI; 19% had FEP MICs of 4-8 which would be considered S-DD by CLSI ( Figure 1A; breakpoints noted by dashed lines). Using the EUCAST breakpoint of ≤1, only 11% of CRO-R isolates would be reported as FEP-S. For CAZ, 40% of CRO-R isolates had CAZ MICs ≤4, which is considered S by CLSI. Using the EUCAST breakpoint of ≤1, only 12% of CRO-R isolates would be reported as CAZ-S ( Figure 1B).

Cefepime MIC Distribution for Ceftriaxone Resistant Isolates
Distribution of MICs for cefepime for ceftriaxone resistant isolates with the breakpoints for EUCAST and CLSI noted with a dashed line Ceftazidime MIC Distribution for Ceftriaxone Resistant Isolates Distribution of MICs for ceftazidime for ceftriaxone resistant isolates with the breakpoints for EUCAST and CLSI noted with a dashed line Conclusion. Half of CRO-R E. coli, K. pneumoniae and K. oxytoca have FEP and CAZ MICs at or below the current CLSI breakpoints. This may lead to their use for serious ESBL infections where a carbapenem is preferred. To prevent unnecessary use, laboratories should consider suppressing FEP and CAZ susceptibilities when CRO-R or adopting more the aggressive EUCAST breakpoints for these agents.