Thirteen Years of Infection Data Reveal a Safer Path for Leukemia Induction Therapy

When adults with acute myeloid leukemia begin induction chemotherapy, the treatment deliberately wipes out their immune defenses along with the malignant cells, leaving them dangerously vulnerable to infection for weeks. For decades, oncologists assumed that powerful preventive antibiotics were the price of survival through this period. A new 13-year study from Poland, published in Annals of Hematology, now challenges that assumption with unusually rich real-world data: stopping routine fluoroquinolone prophylaxis in 2018 did not unleash a wave of deadly infections, and overall 30-day mortality actually fell by more than half in the years that followed.

The retrospective analysis, led by Aneta Nowicka and Lidia Gil at Poznan University of Medical Sciences together with statisticians Anna Sowińska and Barbara Więckowska, followed 219 adults receiving their first induction therapy for acute myeloid leukemia between 2012 and 2025. Rather than simply counting infections, the team set out to map how the entire infectious ecology of the ward shifted over time, in parallel with rising antimicrobial resistance, the arrival of the SARS-CoV-2 pandemic, changes in prophylaxis protocols, and the steady improvement of diagnostic methods. The result is one of the most detailed longitudinal portraits to date of what actually threatens patients during the most vulnerable phase of leukemia treatment.

The central finding concerns the 2018 decision to halt universal fluoroquinolone prophylaxis, a policy change many centers feared would backfire. Using an interrupted time series approach, the researchers compared the pre-2018 era, when every patient received the preventive antibiotic, with the post-2018 era, when it was discontinued. Proportional bloodstream infections fell from 40.6 percent to 34.7 percent, and the share of infections caused by multidrug-resistant organisms dropped from 77.2 percent to 68.6 percent. Crucially, there was no compensatory surge of Gram-negative bacteria, the feared consequence of removing antibiotic pressure. Meanwhile, 30-day mortality declined from 11.9 percent to 5.1 percent.

These numbers carry a technical message that goes beyond a single drug. Fluoroquinolones exert strong selective pressure on gut and environmental bacteria, and prolonged use is a well-known driver of resistance in organisms such as Escherichia coli and Pseudomonas aeruginosa. The Polish data suggest that withdrawing this pressure allowed the local microbial ecology to rebalance, with resistant strains becoming proportionally less dominant without any increase in overall infectious burden. The authors conclude that the evolving infectious etiology in acute myeloid leukemia is driven by temporal, diagnostic, and ecological transitions rather than by any novel pathogen, and that halting fluoroquinolone prophylaxis was clinically safe.

The study also tackled a question that has divided the hematology community since the arrival of targeted therapies: do the newer induction regimens, built around agents such as FLT3 and IDH inhibitors, change the infection profile compared with conventional intensive chemotherapy? The team compared 189 patients on standard induction therapy with 30 patients on novel targeted induction. The answer, at least in this cohort, was reassuringly mundane. Rates of microbiologically documented infections were nearly identical, at 53.3 percent versus 51.3 percent, bloodstream infections were comparable at 33.3 percent versus 38.1 percent, and invasive fungal disease was actually less frequent in the targeted group, at 6.7 percent versus 13.2 percent.

Two apparent differences did emerge in the targeted therapy group, but the authors are careful to disentangle them from confounding. VIM-producing Pseudomonas aeruginosa, a carbapenem-resistant Gram-negative organism carrying the VIM metallo-beta-lactamase gene, appeared in 30.8 percent of relevant infections in the targeted cohort versus 3.3 percent in the standard cohort, a statistically significant difference. SARS-CoV-2 infection likewise affected 53.3 percent of targeted therapy patients versus 33.3 percent. Yet both targeted therapy and the pandemic era, along with more sensitive molecular diagnostics, overlapped almost completely in time. The elevated rates therefore reflect when those patients were treated and how thoroughly they were tested, not the drugs themselves.

This distinction matters for how centers design supportive care. If the infection risks of targeted induction are essentially those of the era rather than of the agent, then prophylaxis and monitoring protocols should be calibrated to local epidemiology and resistance patterns rather than to the specific drug combination. The authors frame this as a call for continuously adapting supportive frameworks, an argument reinforced by the observation that diagnostic methodology itself changed over the study period, with molecular panels detecting organisms and resistance genes that earlier cultures would have missed entirely.

The overall mortality figures provide context for the stakes. Across the full 13-year cohort, 30-day mortality during induction was 8.2 percent, a figure that reflects both the inherent toxicity of myeloablative therapy and the infectious complications that follow. The halving of that rate between eras cannot be attributed to prophylaxis policy alone, since supportive care, antifungal strategies, and general hematology practice all evolved simultaneously. But the direction of change is consistent with a growing body of evidence that indiscriminate antibacterial prophylaxis buys less protection than once believed, while fueling the resistance crisis that complicates every subsequent infection.

As an exploratory, single-center retrospective study, the analysis has limits the authors acknowledge. The targeted induction cohort is small, the pandemic confounds any comparison spanning 2020 onward, and observational time series cannot prove causation. Still, the study offers rare real-world evidence on a question that trials rarely address: what happens to the microbial landscape of a hematology ward when a long-standing prophylaxis policy is reversed. For clinicians weighing antibiotic stewardship against infection fear, the Polish experience suggests that the fear may be the more outdated of the two.

Subject of Research: Evolving infectious epidemiology in acute myeloid leukemia in the contemporary era: a 13-year retrospective, single-center analysis of induction therapies

Article Title: Evolving infectious epidemiology in acute myeloid leukemia in the contemporary era: a 13-year retrospective, single-center analysis of induction therapies

Article References: Nowicka, A., Sowińska, A., Więckowska, B., & Gil, L. (2026). Evolving infectious epidemiology in acute myeloid leukemia in the contemporary era: a 13-year retrospective, single-center analysis of induction therapies. Annals of Hematology. https://doi.org/10.1007/s00277-026-07323-3

Image Credits: AI Generated

DOI: 10.1007/s00277-026-07323-3

Keywords: acute myeloid leukemia, fluoroquinolone prophylaxis, antimicrobial resistance, bloodstream infections, induction chemotherapy, targeted therapy, SARS-CoV-2, multidrug-resistant organisms, hematology, infection epidemiology, Evolving, infectious