LAUSR

Up to half of patients with solid tumors and >80% of those with hematologic malignancies develop a fever during chemotherapy-induced neutropenia [1]. Current practice includes antibacterial prophylaxis with escalation to broader-spectrum antibiotics at the onset of neutropenic fever (NF). However, because microbiological workup remains negative in most NFs [2], this practice represents an empiric antimicrobial attack rather than a mechanistic approach inspired by precisely defined pathogenesis. The fear of failing to diagnose and adequately treat a catastrophic infection in the immunocompromised patient has resulted in heavy antibiotic exposure during the period of neutropenia. Unfortunately, this approach has led to serious adverse consequences, including increased antibiotic resistance and Clostridium difficile infection. Thus, there is an urgent need to re-evaluate the prevailing antimicrobial paradigm. Bloodstream infection (BSI) accounts for 10–25% of all NFs [2]. Cytotoxic therapy impairs gut barrier integrity, facilitating bacterial translocation and increasing the risk of BSI [3]. In addition, antibiotics disrupt the gut microbiota, which normally prevents pathogen colonization [4], provides tonic stimulation to the gut barrier [5], and facilitates recovery from chemotherapy-induced injury. Therefore, it is likely that iatrogenic damage to the microbiota contributes to BSI and NF during chemotherapy. Using serum flagellin as a culture-independent surrogate for motile bacteremia, we demonstrate that specific changes in the indigenous gut microbiota precede BSI. We enrolled 20 intensively treated acute leukemia patients (Supplementary Table 1) who consented to participate in an open-label biorepository study approved by our Institutional Review Board. We defined intensive chemotherapy as any regimen requiring a planned hospitalization of ~4 weeks. Patients did not enroll to the same study upon relapse. We defined NF as a single oral temperature of ≥101 °F or a temperature of ≥100.4 °F sustained over 1 h plus an absolute neutrophil count of <0.5 × 10/L [2]. We defined BSI as the isolation of a bacterium from blood culture. At our institution, we use levofloxacin for antibacterial, acyclovir for antiviral, and an azole for antifungal prophylaxis starting with chemotherapy until neutrophil recovery, and cefepime as the initial empiric antibiotic for NF. Variations to this recommendation were allowed. Parenteral nutrition was used if oral intake was inadequate. All patients had a central venous catheter. We collected blood (n= 213) and stool (n= 207) samples thrice weekly from admission until day 28 of chemotherapy or discharge. Samples were stored at −80 C. Serum samples were analyzed in duplicate by standard enzyme-linked immunosorbent assays for flagellin (MyBioSource, San Diego, CA). The lower limit of quantification was 9.0 pg/mL. The intraand inter-assay coefficients of variation were <8% and <12%, respectively. DNA was extracted from ~250 mg of stool using the DNeasy PowerSoil kit (QIAGEN, Hilden, Germany) on the automated QIAcube platform. The V4 hypervariable region of the 16S rRNA gene was sequenced * Armin Rashidi [email protected]