Background: Management of neutropenic fever in high-risk haematology patients is challenging; there are often few localising clinical features, and diagnostic tests have poor sensitivity and specificity. We aimed to compare how [18F]flurodeoxyglucose ([18F]FDG)-PET-CT scans and conventional CT scans affected the guidance of antimicrobial management and the outcomes of patients with persistent or recurrent neutropenic fever.

Methods: We did a multicentre, open-label, phase 3, randomised, controlled trial in two tertiary referral hospitals in Australia. We recruited adults aged 18 years or older who were receiving conditioning chemotherapy for haematopoietic stem-cell transplantation or chemotherapy for acute leukaemia and had persistent (>72 h) or recurrent (new fever beyond 72 h of initial onset interspersed with >48 h defervescence) neutropenic fever. Exclusion criteria were pregnancy, allergy to iodinated contrast, or estimated glomerular filtration rate of less than 30 mL/min. Patients were randomly assigned by computer-generated randomisation chart (1:1) to [18F]FDG-PET-CT or conventional CT. Masking was not possible because of the nature of the investigation. Scans were done within 3 days of random assignment. The primary endpoint was a composite of starting, stopping, or changing the spectrum (broadening or narrowing) of antimicrobial therapy-referred to here as antimicrobial rationalisation-within 96 h of the assigned scan, analysed per protocol. This trial is registered with clinicaltrials.gov, NCT03429387, and is complete.

Findings: Between Jan 8, 2018, and July 23, 2020, we assessed 316 patients for eligibility. 169 patients were excluded and 147 patients were randomly assigned to either [18F]FDG-PET-CT (n=73) or CT (n=74). Nine patients did not receive a scan per protocol, and two participants in each group were excluded for repeat entry into the study. 65 patients received [18F]FDG-PET-CT (38 [58%] male; 53 [82%] White) and 69 patients received CT (50 [72%] male; 58 [84%] White) per protocol. Median follow up was 6 months (IQR 6-6). Antimicrobial rationalisation occurred in 53 (82%) of 65 patients in the [18F]FDG-PET-CT group and 45 (65%) of 69 patients in the CT group (OR 2·36, 95% CI 1·06-5·24; p=0·033). The most frequent component of antimicrobial rationalisation was narrowing spectrum of therapy, in 28 (43%) of 65 patients in the [18F]FDG-PET-CT group compared with 17 (25%) of 69 patients in the CT group (OR 2·31, 95% CI 1·11-4·83; p=0·024).

Interpretation: [18F]FDG-PET-CT was associated with more frequent antimicrobial rationalisation than conventional CT. [18F]FDG-PET-CT can support decision making regarding antimicrobial cessation or de-escalation and should be considered in the management of patients with haematological diseases and persistent or recurrent high-risk neutropenic fever after chemotherapy or transplant conditioning.

Funding: National Health and Medical Research Council Centre of Research Excellence (APP1116876), Melbourne Health foundation, Gilead Research Fellowship grants supported this study.