High lactate dehydrogenase at time of admission for allogeneic hematopoietic transplantation associates to poor survival in acute myeloid leukemia and non-Hodgkin lymphoma

Artificial intelligence reveals the predictions of hematological indexes in children with acute leukemia

Childhood leukemia is a prevalent form of pediatric cancer, with acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) being the primary manifestations. Timely treatment has significantly enhanced survival rates for children with acute leukemia. This study aimed to develop an early and comprehensive predictor for hematologic malignancies in children by analyzing nutritional biomarkers, key leukemia indicators, and granulocytes in their blood. Using a machine learning algorithm and ten indices, the blood samples of 826 children with ALL and 255 children with AML were compared to a control group of 200 healthy children. The study revealed notable differences, including higher indicators in boys compared to girls and significant variations in most biochemical indicators between leukemia patients and healthy children. Employing a random forest model resulted in an area under the curve (AUC) of 0.950 for predicting leukemia subtypes and an AUC of 0.909 for forecasting AML. This research introduces an efficient diagnostic tool for early screening of childhood blood cancers and underscores the potential of artificial intelligence in modern healthcare.

The development of a prediction model based on random survival forest for the prognosis of non- Hodgkin lymphoma: A prospective cohort study in China

Background and objective

The pathological staging of non-Hodgkin lymphoma (NHL) is complex, the clinical manifestations are varied, and the prognosis differ considerably. To provide a useful reference for early detection and effective treatment of NHL, we developed a random survival forest (RSF) prognostic model based on machine learning (ML) algorithms using prospective cohort data collected from Chongqing Cancer Hospital from Jan 1, 2017 to Dec 31, 2019 (n = 1449) to compare with the traditional cornerstone method Cox proportional hazards (CPH) model and evaluate the predictability of the model.

Methods

Patients were randomly split into a training cohort (TC) and validation cohort (VC) based on 65/35 ratio. The least absolute shrinkage and selection operator (LASSO) regression analysis was used to extracted the important features. And the RSF was modeled to explore the prognostic factors impacting the overall survival (OS) of patients with NHLs in the TC and validated in the VC. The C-index, the Integrated Brier Score (IBS), Kaplan-Meir method, the receiver operating characteristic (ROC) curve, and the area under the ROC curve (AUC) were selected to measure performances and discriminations of the models. In addition, individual survival probability predicted for NHL patients.

Results

According to the features extracted by LASSO model and univariable Cox model, 16 variables were selected to develop the RSF model with log-rank splitting rule, which were age, ethnicity, medical insurance, Ann Arbor stage, pathology, targeted-therapy, chemo-therapy, peripheral blood neutrophil count to lymphocyte count ratio (NLR), peripheral blood platelet count to lymphocyte count ratio (PLR), serum lactate dehydrogenase (LDH), CD4/CD8, platelet (PLT), absolute neutrophil count (ANC), lymphocyte (LYM), B-symptoms, and (CPR) were important prognostic factors. Compared to the CPH model (C-index = 0.748, IBS = 0.166), the RSF model (C-index = 0.786, IBS = 0.165) is outperformed in predictability and accuracy. The AUC of the RSF model to estimate the 1-, 3-, and 5-year OS in TC were 0.847, 0.847, and 0.809, respectively; while those in the CPH were 0.816, 0.803, and 0.750, respectively.

Conclusions

To provide practical implications for the implementation of individualized therapy, the study constructed a high-performed RSF model and reveal that it outperformed the traditional model CPH. And the RSF model ranked the risk variables. In addition, we stratified the risk of NHL patients and estimated individual survival probability based on the RSF model.

Development and validation of prediction model for overall survival in patients with lymphoma: a prospective cohort study in China

OBJECTIVE

The survival of patients with lymphoma varies greatly among individuals and were affected by various factors. The aim of this study was to develop and validate a prognostic model for predicting overall survival (OS) in patients with lymphoma.

METHODS

We conducted a prospective longitudinal cohort study in China between January 2014 and December 2018 (n = 1,594). After obtaining the follow-up data, we randomly split the cohort into the training cohort (n = 1,116) and the validation cohort (n = 478). The least absolute shrinkage and selection operator (LASSO) regression analysis was used to select the predictors of the model. Cox stepwise regression analysis was used to identify independent prognostic factors, which were finally displayed as static nomogram and web-based dynamic nomogram. We calculated the concordance index(C-index) to describe how the predicted survival of objectively confirmed prognosis. The calibration plot is used to evaluate the prediction accuracy and discrimination ability of the model. Net reclassification index (NRI) and decision curve analysis (DCA) curves were also used to evaluate the prediction ability and net benefit of the model.

RESULTS

Nine variables in the training cohort were considered to be independent risk factors for patients with lymphoma in the final model: age, Ann Arbor Stage, pathologic type, B symptoms, chemotherapy, targeted therapy, lactate dehydrogenase (LDH), β2-microglobulin and C-reactive protein (CRP). The C-indices of OS were 0.749 (95% CI, 0.729-0.769) in the training cohort and 0.731 (95% CI, 0.762-0.700) in the validation cohort. A good agreement between prediction by nomogram and actual observation was shown in the calibration curve for the probability of survival in both the training cohort and validation cohorts. The areas under curve (AUC) of the area under the receiver operating characteristic (ROC) curves for 1-year, 3-year, and 5-year OS were 0.813, 0.800, and 0.762, respectively, in the training cohort, and 0.802, 0.768, and 0.721, respectively, in the validation cohort. Compared with the Ann Arbor Stage system, NRI and DCA showed that the model had a higher predictive capacity and net benefit.

CONCLUSION

The prediction models reliably estimate the outcome of patients with lymphoma. The model had high discrimination and calibration, which provided a simple and reliable tool for the survival prediction of the patients, and it might help patients benefit from personalized intervention.

Prediction of Nonrelapse Mortality in Patients With Acute Myeloid Leukemia and Acute Lymphoblastic Leukemia Receiving Allogeneic Stem Cell Transplantation With Posttransplantation Cyclophosphamide-based Graft Versus Host Disease Prophylaxis

Graft versus host disease (GVHD) prophylaxis with posttransplantation cyclophosphamide (PTCY) has been established to reduce severe GVHD, and thereby potentially reducing nonrelapse mortality (NRM) after allogeneic stem cell transplantation (alloSCT). We evaluated the predictive capacity of established NRM-risk scores in patients receiving PTCY-based GVHD prophylaxis, and subsequently developed and validated a novel PTCY-specific NRM-risk model. Adult patients (n = 1861) with AML or ALL in first complete remission who received alloSCT with PTCY-based GVHD prophylaxis were included. The PTCY-risk score was developed using multivariable Fine and Gray regression, selecting parameters from the hematopoietic cell transplantation-comorbidity index (HCT-CI) and European Group for Blood and Marrow Transplantation (EBMT) score with a subdistribution hazard ratio (SHR) of ≥1.2 for 2-year NRM in the training set (70% split), which was validated in the test set (30%). The performance of the EBMT score, HCT-CI, and integrated EBMT score was relatively poor for discriminating 2-year NRM (c-statistic 51.7%, 56.6%, and 59.2%, respectively). The PTCY-risk score included 10 variables which were collapsed in 3 risk groups estimating 2-year NRM of 11% ± 2%, 19% ± 2%, and 36% ± 3% (training set, c-statistic 64%), and 11% ± 2%, 18% ± 3%, and 31% ± 5% (test set, c-statistic 63%), which also translated into different overall survival. Collectively, we developed an NRM-risk score for acute leukemia patients receiving PTCY that better predicted 2-year NRM compared with existing models, which might be applicable to the specific toxicities of high-dose cyclophosphamide.