Patients with acute myeloid leukemia (AML) are at high risk for severe infections during and after chemotherapy. Early identification of those likely to progress to sepsis during induction therapy could permit intensified monitoring and supportive care to improve outcomes. This study aimed to develop a practical risk assessment tool — a nomogram — to predict early sepsis risk when applied on days 3–5 of induction chemotherapy in newly diagnosed AML patients.
This was a retrospective review of medical records from a single institution covering February 2017 to December 2023. The analysis included 192 newly diagnosed AML patients who received induction chemotherapy. Survival differences between patients who did and did not develop sepsis during induction were evaluated using Kaplan–Meier analysis. Statistical testing employed the Mann–Whitney U test, Chi-square test, and multivariable logistic regression to identify independent predictors for sepsis and to build the nomogram.
Multivariable logistic regression identified five independent risk factors that were incorporated into the nomogram:
These variables were selected from baseline and early post-induction laboratory and genetic data to allow risk stratification early in the induction course (days 3–5).
The authors constructed a point-based nomogram that maps each predictor to a points axis, allowing clinicians to sum points across variables to obtain a total score. The total score projects to a predicted probability of developing sepsis later during the same induction course. The nomogram is intended for bedside or clinical use on day 3 to 5 of induction chemotherapy: identify each variable value, translate to points, sum points, and read the corresponding sepsis probability from the total points axis.
The nomogram demonstrated strong predictive performance in internal validation. The bootstrap-corrected area under the receiver operating characteristic curve (AUC) was 0.892; the concordance index (C-index) was also reported as 0.892. Bias-corrected calibration curves were close to the ideal 45-degree line, indicating good agreement between predicted probabilities and observed sepsis rates after bootstrapping (1000 iterations as reported for resampling in figure legend). Decision curve analysis showed a net benefit for the nomogram compared with treating all or none, suggesting potential clinical value when used to guide monitoring or supportive interventions.
Patients who developed sepsis during induction had significantly worse overall survival than those who did not. The sepsis group (n = 27) showed inferior survival on Kaplan–Meier analysis compared with the non-sepsis group (n = 165) at both 1-year and 5-year landmarks (log-rank p < 0.001). The incidence of pneumonia and invasive fungal infection was higher among patients who developed sepsis.
The nomogram offers an early, individualized probability estimate of sepsis risk that could help clinicians identify AML patients requiring closer surveillance, early diagnostic evaluation for infection, or more intensive supportive measures during induction chemotherapy. Because the model uses routinely available laboratory results and TP53 mutation status, it is potentially implementable in centers with access to these data.
Limitations reported in the article abstract are limited; specifics such as details about external validation, calibration performance metrics beyond the bootstrap-corrected curve, sample size considerations for subgroups, or protocolized management changes triggered by the nomogram were not reported in the abstract and would require review of the full text. The model performance reported derives from internal bootstrap validation; external validation in independent cohorts was not described in the abstract.
The authors developed and internally validated a nomogram to predict early sepsis risk in newly diagnosed AML patients during initial induction chemotherapy. The model incorporated TP53 mutation, AST, CRP, PCT, and urea, and showed strong discrimination (AUC/C-index 0.892) and good calibration after bootstrapping. Patients who developed sepsis had significantly worse survival and higher rates of pneumonia and invasive fungal infection. The nomogram may assist clinicians in selecting patients for intensified monitoring and supportive care during early induction, though external validation and further details on implementation were not reported in the abstract.
Conflict of interest statement: the authors declared no conflicts of interest in the article abstract.