Acute myeloid leukemia (AML) is characterized by aggressive clonal expansion, marrow failure, and recurrent molecular alterations. Standard surveillance—serial bone marrow biopsy, multiparameter flow cytometry, and next-generation sequencing—can be invasive, costly, and limited in frequency. The authors evaluated whether a minimally invasive panel of routinely available peripheral hematology, targeted molecular expression and protein quantification, plus patient medication adherence, could provide complementary prognostic information for dynamic monitoring of disease activity.
The study frames a pragmatic, low-burden approach intended to supplement conventional surveillance by leveraging peripheral biomarkers and behavioral metrics to identify early transition to critical deterioration or disease-related mortality.
This longitudinal cohort included 66 patients observed over 272 discrete follow-up encounters. Serial assessments were scheduled at days 0, 14, 41, 71, and 78. The primary clinical endpoint was longitudinal progression to critical deterioration or disease-related death. The study reports IRB approval from Shahid Beheshti University of Medical Science (IRB Protocol #00000432) and confirms that necessary participant consent procedures were followed.
The investigators measured routine peripheral complete blood count indices—notably white blood cell count (WBC) and hemoglobin (Hb)—together with targeted molecular transcript levels for FLT3, NPM1, and DNMT3A. They also quantified soluble FLT3 protein in peripheral samples. Medication adherence was captured as a compliance metric and incorporated into models as a predictor variable.
Specific assay methods, limits of detection, or laboratory platforms for the transcript and soluble protein measurements were not detailed in the abstract; those methodological details are not reported in the source abstract.
To account for repeated measures within subjects, generalized linear mixed-effects models (GLMMs) with patient-specific random intercepts were constructed. Predictors were standardized and effects were reported per 1–standard-deviation (1-SD) increment where applicable. Model fit and comparative performance were assessed with the Akaike Information Criterion (AIC) and discrimination was evaluated using Harrell’s concordance index (C-index).
The authors compared single-domain models (for example, hematology only) to integrative multivariable models that combined hematologic indices, molecular markers, and adherence.
An inverse correlation was observed between leukocyte count and hemoglobin concentration (Pearson r = -0.877, p < 0.001), a pattern consistent with marrow infiltration and secondary erythroid suppression during disease expansion.
In multivariable mixed-effects analyses, two variables had independent associations with the primary endpoint:
Targeted molecular measures, including FLT3 transcript levels and soluble FLT3 protein quantification, provided additional prognostic information when integrated with peripheral blood indices and adherence metrics.
The integrative model that combined routine hematologic indices, targeted molecular quantification (including FLT3 measures), and adherence outperformed single-domain models. Improvement in model fit was substantial (reported ΔAIC ≥ 116). Discrimination for the integrated model was reported as Harrell’s C-index = 0.81, indicating good ability to distinguish patients who progressed to critical deterioration or disease-related mortality from those who did not.
Exact cross-validation procedures, calibration plots, or external validation results are not reported in the abstract; such details were not provided in the source summary.
The authors declare no competing interests and report institutional review board approval and participant consent. The abstract states that all data produced in the study are available upon reasonable request from the corresponding author.
Limitations explicitly described in the source abstract are limited. The report is a preprint and has not undergone peer review; the abstract does not provide comprehensive methodological details such as assay platforms, adherence measurement instruments, the handling of missing data, or external validation—these details were not reported in the abstract.
Within this cohort, readily accessible peripheral hematologic indices (notably WBC and Hb), targeted molecular quantification (including FLT3 transcripts and soluble FLT3 protein), and medication adherence offered complementary prognostic information about disease activity. The authors propose that integrating these low-burden parameters into clinical workflows could enable a more efficient, scalable paradigm for early risk stratification in hematologic malignancies.
Given this is a preprint, the findings should not be used to guide clinical practice until peer review and further validation (including external cohorts and detailed methodological reporting) are available. The study suggests next steps would include complete methodological reporting, validation of the model in independent datasets, and assessment of how integration into clinical workflows affects decision making and patient outcomes.