No advantage of antimicrobial prophylaxis in AML/MDS/CMML patients treated with azacitidine—a prospective multicenter study by the Polish Adult Leukemia Group

Krzysztof Mądry, Karol Lis, Elzbieta Sienkiewicz, Joanna Drozd‐Sokołowska, Przemysław Biecek, Oktawia Sośnia, Aleksandra Gołoś, Magdalena Olszewska‐Szopa, Agata Obara, Zuzanna Walkowiak, Joanna Ściesińska, Edyta Subocz, Aleksandra Butrym, Rafał Machowicz, Bożena Katarzyna Budziszewska, Grzegorz BasakFrontiers in Oncology 2024 · DOI

Abstract

The Azacitidine Infection Risk (AIR) Model effectively discriminates infection-risk patients during azacitidine treatment. However, antimicrobial prophylaxis does not decrease the infection rate in this population.

Introduction

Patients diagnosed with myelodysplastic syndromes (MDS), acute myeloid leukemia (AML), and chronic myelomonocytic leukemia (CMML) face a high risk of life-threatening infections. This vulnerability is particularly pronounced during treatment with hypomethylating agents (HMAs) like azacitidine (AZA). While antimicrobial prophylaxis—including antibacterial, antifungal, and antiviral agents—is considered a standard of care for patients undergoing intensive induction chemotherapy, its role in the context of less intensive HMA therapy is less clear.

Current clinical practice often relies on institutional protocols or individual physician discretion rather than robust prospective evidence. This study, conducted by the Polish Adult Leukemia Group (PALG), aimed to evaluate the efficacy of standard prophylaxis and validate the Azacitidine Infection Risk (AIR) model as a predictive tool for infectious complications.

Methods

This prospective, multicenter, non-interventional study followed a cohort of 307 patients across nine Polish hematology centers. The study population included patients with MDS (55%), AML (37.8%), and CMML (7.1%).

Risk Stratification and Monitoring

The researchers utilized the AIR Model to categorize patients into risk groups at the start of treatment. The model assigns points based on five clinical and laboratory factors:

  1. Red blood cell (RBC) transfusion dependency
  2. Neutropenia
  3. Low platelet count
  4. Low serum albumin levels
  5. Poor performance status (ECOG ≥2)

The primary endpoint was the occurrence of grade III or higher infectious episodes (IE) within the first three cycles of azacitidine therapy. Statistical analysis employed multivariate logistic regression to identify independent risk factors and Kaplan-Meier estimators to assess overall survival (OS).

Results and Discussion

The Prophylaxis Paradox

Initial univariate analysis suggested a counterintuitive finding: patients receiving antimicrobial prophylaxis appeared to have a higher incidence of infections. However, further investigation revealed this as a classic example of "confounding by indication." Clinicians were more likely to prescribe prophylaxis to patients they perceived as being at higher baseline risk due to comorbidities or advanced disease.

When the researchers adjusted for these baseline health factors in a multivariate model, the use of antibacterial, antifungal, or antiviral prophylaxis showed no significant impact on the rate of infectious episodes. This suggests that standard prophylactic regimens may not be sufficient to overcome the profound immunosuppression associated with these malignancies and their treatment.

Validation of the AIR Model

The study successfully validated the AIR model's predictive power. Patients identified as "high risk" by the model had a 47% probability of experiencing a severe infection, nearly double the 22.4% rate observed in the low-risk group.

Furthermore, high baseline C-reactive protein (CRP > 20 mg/L) was identified as a potent independent predictor of infection. This finding suggests that many patients may harbor subclinical or "smoldering" infections even before the initiation of chemotherapy, which then manifest as clinical crises during treatment-induced myelosuppression.

Impact on Survival

Infection risk directly correlates with patient longevity. The survival analysis demonstrated a stark divergence in outcomes based on AIR risk categorization.

Survival curves showing the stark difference in outcomes between high and low infection risk groups as defined by the AIR model.

Patients in the high-risk AIR group faced a median overall survival of just 9 months, whereas those in the low-risk group achieved a median survival of 19 months. This highlights the critical need for better management strategies for high-risk individuals.

Conclusion and Future Directions

This prospective study confirms that while the AIR model is an excellent tool for identifying patients at high risk for infection, current standard antimicrobial prophylaxis does not effectively mitigate that risk.

The study's non-randomized design is a limitation, as the decision to provide prophylaxis was not controlled. Future research should prioritize randomized controlled trials (RCTs) to definitively assess prophylaxis in high-risk subgroups. Additionally, as the standard of care evolves toward combination therapies (e.g., azacitidine plus venetoclax), the infection risk profile may shift, necessitating updated risk models and more aggressive pre-emptive strategies, particularly for patients presenting with high baseline CRP.

Summary updated

Abstract

The AIR Model effectively discriminates infection-risk patients during azacitidine treatment. Antimicrobial prophylaxis does not decrease the infection rate.
BibTeX
@inproceedings{madry2024no,
  author = {Krzysztof Mądry and Karol Lis and Elzbieta Sienkiewicz and Joanna Drozd‐Sokołowska and Przemysław Biecek and Oktawia Sośnia and Aleksandra Gołoś and Magdalena Olszewska‐Szopa and Agata Obara and Zuzanna Walkowiak and Joanna Ściesińska and Edyta Subocz and Aleksandra Butrym and Rafał Machowicz and Bożena Katarzyna Budziszewska and Grzegorz Basak},
  title = {{No advantage of antimicrobial prophylaxis in AML/MDS/CMML patients treated with azacitidine—a prospective multicenter study by the Polish Adult Leukemia Group}},
  booktitle = {Frontiers in Oncology},
  year = {2024},
  doi = {10.3389/fonc.2024.1404322},
}
Figure