Discussion

We have reported the clinical characteristics and outcomes from a nationwide VHA cohort who received cefiderocol to expand the existing published literature on cefiderocol therapy, particularly for treatment of P. aeruginosa infections. The key findings of this study include a clinical failure rate of 35.4% (18 of 48) and a 30-day microbiological failure rate of 29.2% (14 of 48). Of the 13 patients who died within 30 days of initiation of cefiderocol, 11 had transitioned to hospice or comfort care, therefore stopping antibiotics. This situation reflects the real-world nature of the use of cefiderocol and the severity of illness and comorbidity burden among patients receiving this medication. All-cause mortality, although precise, cannot elucidate whether the antibiotic given was ineffective or whether the overall patient condition or complications of the infection were the determining factors leading to death.

Of 48 patients prescribed cefiderocol in this study, 20 (41.6%) were at the extremes of renal function, including creatinine clearance ≥120 mL/minute or receiving renal replacement therapy. Therefore, they required increased administration frequency compared with standard dosing (ie, 2 g every 6 hours rather than 2 g every 8 hours) or dose adjustments based on effluent flow rate of continuous renal replacement therapy. Notably, augmented renal clearance (creatinine clearance >120 mL/minute) can be associate with critical illness but can also occur in non–critically ill patients. ^{Reference Katsube, Echols and Wajima22} This study also found high rates of cefiderocol susceptibility among initial causative pathogens, when tested (83.3%, 20 of 24). Overall, patients who received cefiderocol were medically complex and often required management by intensivists, infectious disease specialists, medical microbiologists, and clinical pharmacy specialists.

The study has several strengths that expand the literature on the effectiveness of cefiderocol. First, the larger cohort size compared with the existing published literature. ^{Reference Falcone, Tiseo and Nicastro6–Reference Bleibtreu, Dortet and Bonnin13} Second, this study captured real-world prescribing of cefiderocol after US FDA approval rather than clinical trials or compassionate use programs. Third, this study measured outcomes of a cohort of patient with P. aeruginosa as the most common pathogen, whereas the existing literature largely reports on A. baumannii infections. After the randomized controlled trials examining cefiderocol ^{Reference Bassetti, Echols and Matsunaga2–Reference Portsmouth, van Veenhuyzen and Echols4} were performed, only small studies have evaluated the effectiveness of cefiderocol. Only 2 small studies reported results in which P. aeruginosa was the most common pathogen. ^{Reference Meschiari, Volpi and Faltoni12,Reference Bleibtreu, Dortet and Bonnin13} Meschiari et al ^{Reference Meschiari, Volpi and Faltoni12} examined 17 patients with P. aeruginosa infection treated with cefiderocol through a compassionate-use protocol and reported a 30-day all-cause mortality rate of 23.5% and a 90-day all-cause mortality rate of 35.3%. Both rates are in line with the results of this study. Bleibtreu et al ^{Reference Bleibtreu, Dortet and Bonnin13} reported outcomes of 13 patients, 10 of whom had infections with P. aeruginosa, treated with cefiderocol through a compassionate-use protocol and reported a 23.1% mortality rate. ^{Reference Bleibtreu, Dortet and Bonnin13}

Another 5 studies have been conducted under real-world prescribing conditions. ^{Reference Rando, Segala and Vargas7,Reference Gavaghan, Miller and Dela-Pena9,Reference Hoellinger, Simand and Jeannot10,Reference Falcone and Tiseo14,Reference Falcone, Tiseo and Leonildi15} Falcone et al ^{Reference Falcone and Tiseo14,Reference Falcone, Tiseo and Leonildi15} reported a 28-day all-cause mortality rate of 22.2% (4 of 18) associated with cefiderocol therapy of metallo-β-lactamase–producing Enterobacterales. In a second study, the 30-day mortality rate was 34% (16 of 47) in patients with A. baumannii infections treated with cefiderocol. In a study by Gavaghan et al, ^{Reference Hoellinger, Simand and Jeannot10} the 30-day mortality rate was 42% (10 of 24) in real-world cefiderocol therapy for patients among whom pneumonia was the most common source of infection. A. baumannii was the most common organism in the Advocate Aurora Health system, which comprises hospitals in Wisconsin and Illinois. ^{Reference Gavaghan, Miller and Dela-Pena9} In a single center in France, Hoellinger et al ^{Reference Hoellinger, Simand and Jeannot10} reported a 30-day mortality rate of 60% in 10 patients, most of whom were highly immunosuppressed (eg, organ transplantation, acute hematologic malignancy, metastatic solid tumor), treated with cefiderocol for nonfermenting gram-negative bacilli infections, of which 6 of 10 were due to P. aeruginosa. In another single-center study in France, Rando et al ^{Reference Rando, Segala and Vargas7} reported a 46.1% mortality rate among 13 patients with A. baumannii pneumonia, of whom all but one also had COVID-19,. ^{Reference Rando, Segala and Vargas7} Also, 4 other studies reported outcomes with cefiderocol therapy for A. baumannii infections; they were conducted under compassionate use programs in Italy with mortality rates of 10% (1 of 10), 23.1% (3 of 13), 55% (23 of 42), and 30.8% (4 of 13). ^{Reference Falcone, Tiseo and Nicastro6,Reference Bavaro, Belati and Diella8,Reference Gatti, Bartoletti and Cojutti11,Reference Pascale, Pasquini and Bartoletti16}

Our study had several limitations. First, this was a descriptive study, and we did not compare cefiderocol to alternative therapies. Therefore, we could not determine whether cefiderocol was associated with increased mortality compared to other available therapies. This question likely could only be adequately addressed by a randomized clinical trial even larger than the CREDIBLE-CR phase 3 clinical trial. ^{Reference Bassetti, Echols and Matsunaga2} Previously published studies used different inclusion criteria and outcomes definitions, and we sought to use the most widely recognized criteria and definitions (eg, assessing mortality), but for those without consensus in the literature, we selected definitions based on clinical relevance. For example, the duration of cefiderocol therapy for inclusion in prior studies ranged from a single dose ^{Reference Bleibtreu, Dortet and Bonnin13} to at least 5 days ^{Reference Gavaghan, Miller and Dela-Pena9} ; we chose 2 days similar to Hoellinger et al ^{Reference Hoellinger, Simand and Jeannot10} and Falcone et al. ^{Reference Falcone and Tiseo14,Reference Falcone, Tiseo and Leonildi15} When restricting our cohort only to patients who received ≥5 days cefiderocol, the rates of 30-day all-cause mortality (26.3%; 10 of 38) and 90-day all-cause mortality (44.7%; 17 of 38) were not different from the overall cohort. Second, antimicrobial susceptibility testing was not standardized, and multidrug-resistant isolates each underwent susceptibility testing with different panels of antibiotic. Also, interpretive breakpoint criteria were at the discretion of the reporting laboratory. For example, the cefiderocol susceptible breakpoint for P. aeruginosa for the US FDA, EUCAST, and CLSI are ≤1 µg/mL, ≤2 µg/mL and ≤4 µg/mL, respectively. ^23–25 Of the cultures that reported numeric MIC or zone diameters, we found high rates (44.4%; 4 of 11) of isolates where the interpretation (S/I/R) would differ based on which breakpoints (ie, CLSI, US FDA, or EUCAST) were applied. This high rate warrants further monitoring and investigation to determine whether some form of bias (eg, reporting bias) was present; for now, we believe antimicrobial stewardship teams and medical microbiologists should consider reporting the numeric cefiderocol MIC or zone diameters, the breakpoints applied, and carefully consider how to manage isolates with MIC or zone diameters where the agencies (CLSI, US FDA, and EUCAST) have differing interpretations. This investigation reflects the real-world diversity in microbiology laboratory practices and the complexities of implementing cefiderocol susceptibility testing, particularly soon after FDA approval. ^{Reference Simner and Patel26} Notably, all subjects had cefiderocol initiated prior to availability of final cefiderocol AST results. There have been reports of cefiderocol-resistant isolates in patients who never received cefiderocol ^{Reference Streling, Al Obaidi and Lainhart27} and isolates collected prior to its approval by the US FDA, ^{Reference Iovleva, Mustapha and Griffith28} but resistance rates were low in other surveillance. ^{Reference Shortridge, Streit, Mendes and Castanheira29} Third, this study was conducted within the VHA; therefore, patients were within the United States and were predominantly male.

In summary, we have reported the clinical failure rates, microbiological failure rates, and all-cause mortality in a cohort of patients with significant comorbidities who received cefiderocol. All of these failures rates and mortality rates were high, which probably reflects the underlying comorbidities of the patients and the refractory nature of their underlying infections. Patients prescribed cefiderocol frequently had complex pharmacodynamics, which is an opportunity for stewardship teams to ensure correct cefiderocol dosing. We also detected high rates of isolates with cefiderocol MIC or zone diameters, which would have different cefiderocol antimicrobial susceptibility testing interpretations depending on which agency’s breakpoints are used. We recommend that antimicrobial stewardship teams and medical microbiologists consider routinely reporting cefiderocol MIC or zone diameters and which agency’s breakpoints are used. Cefiderocol provides a new therapeutic option for patients with multidrug-resistant gram-negative infections and should be further investigated.