Clinical Features and Prognostic Nomogram for Therapy-Related Acute Myeloid Leukemia after Allogeneic Hematopoietic Stem Cell Transplantation

Introduction:

Therapy-related acute myeloid leukemia (t-AML) constitutes 10%-30% of newly diagnosed AML cases, is more resistant to conventional cytotoxic therapies, and has a greater relapse rate and poorer prognosis compared with de novo AML. Research has shown that allogeneic hematopoietic stem cell transplantation (allo-HSCT) is an effective curative therapeutic approach for treating t-AML. However, the applicability of conventional pretransplant scoring systems in patients with t-AML remains uncertain and lacks specificity. To address these gaps and validate the utility of classical scoring systems in t-AML patients, this study retrospectively examined the clinical characteristics of t-AML patients who underwent allo-HSCT and developed a prognostic nomogram. We carried out this study to develop a prognostic nomogram specifically for patients with t-AML receiving allo-HSCT to estimate survival outcomes and make treatment decisions.

Method:

In this retrospective study, patients who received allo-HSCT for t-AML from January 1, 2010, to December 30, 2023. Univariate analysis and multivariate Cox regression were used to examine the effects of covariates on survival. Variables with p < 0.05 in the univariate analysis were included in the multivariate analysis. A prognostic nomogram was developed from the final Cox proportional hazards regression model. Risk points were assigned on the basis of regression coefficients from the multivariable model for the final prognostic nomogram. Nomogram performance was evaluated via the bootstrap method. Additionally, we evaluated the predictive value of this nomogram in terms of discriminative power, calibration power and net benefit.

Results:

A total of 154 patients with t-AML receiving HSCT were included to develop and validate a prognostic nomogram, and all patients were derivation cohort. The median ages at diagnosis of t-AML and at allo-HSCT were 42 and 43 years (range, 18–73; 18–75 years), respectively. The median percentage of bone marrow (BM) blasts at the time of AML diagnosis was 45% (range, 19–95.5%). The most common prior diseases were breast cancer (24.7%), lymphoma (16.9%) and acute leukemia/MDS (13.6%). The median interval from cytotoxic therapy or radiotherapy to the diagnosis of AML was 3.5 years. The 1- and 3-year overall survival (OS) rates were 85.5% and 69.3%, respectively. There was no significant difference in the clinical baseline characteristics between the two cohorts. In the OS analysis, the leukocyte count and hemoglobin level at AML diagnosis; poor cytogenetics; genetic abnormalities and MRD positivity before HSCT; BM or circulation blasts before HSCT; leukocyte and platelet counts; hemoglobin level before HSCT; neutrophil and platelet engraftment days; interphase; age at prior disease; AHD-AML; and relapse of the prior disease were identified as potential prognostic factors. According to the multivariate Cox analysis, leukocyte count at AML diagnosis ≥ 7×10⁹/L (p=0.016), genetic abnormalities before HSCT (p<0.001), platelet engraftment ≥ 28 days (p<0.001), age at prior disease ≥ 45 years (p=0.010) and relapse of the prior disease (p=0.006) were identified as independent prognostic factors. Based on regression coefficients from the multivariate analysis, we constructed a prognostic nomogram termed the LGPAR (59 points for leukocytes, 83 points for genetic abnormalities, 100 points for platelet engraftment, 62 points for age at prior disease and 63 points for relapse of prior disease). The LGPAR prognostic nomogram is available via a free browser-based online calculator at https://limenglin.shinyapps.io/LGPAR-nomogram/. Internal validation was carried out with 1000 bootstrap replicates in the derivation cohort, and the areas under the receiver operating characteristic curves (AUCs) of 1-, 3-, and 5-year OS were 0.901, 0.876, and 0.884, respectively. According to the calibration plot, the estimated and actual mortality risks were highly consistent. Moreover, the DCA curve confirmed that this model contributed to substantial net benefits.

Conclusions:

In this study, we developed a prognostic nomogram for patients with t-AML after receiving allo-HSCT. This nomogram can effectively improve the survival and prognosis of patients with t-AML after receiving allo-HSCT and accelerate the early identification of patients at high risk of death.