Application of interpretable machine learning algorithms to predict distant metastasis in ovarian clear cell carcinoma.

BACKGROUND: Ovarian clear cell carcinoma (OCCC) represents a subtype of ovarian epithelial carcinoma (OEC) known for its limited responsiveness to chemotherapy, and the onset of distant metastasis significantly impacts patient prognoses. This study aimed to identify potential risk factors contributing to the occurrence of distant metastasis in OCCC. METHODS: Utilizing the Surveillance, Epidemiology, and End Results (SEER) database, we identified patients diagnosed with OCCC between 2004 and 2015. The most influential factors were selected through the application of Gaussian Naive Bayes (GNB) and Adaboost machine learning algorithms, employing a Venn test for further refinement. Subsequently, six machine learning (ML) techniques, namely XGBoost, LightGBM, Random Forest (RF), Adaptive Boosting (Adaboost), Support Vector Machine (SVM), and Multilayer Perceptron (MLP), were employed to construct predictive models for distant metastasis. Shapley Additive Interpretation (SHAP) analysis facilitated a visual interpretation for individual patient. Model validity was assessed using accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and the area under the receiver operating characteristic curve (AUC). RESULTS: In the realm of predicting distant metastasis, the Random Forest (RF) model outperformed the other five machine learning algorithms. The RF model demonstrated accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and AUC (95% CI) values of 0.792 (0.762-0.823), 0.904 (0.835-0.973), 0.759 (0.731-0.787), 0.221 (0.186-0.256), 0.974 (0.967-0.982), 0.353 (0.306-0.399), and 0.834 (0.696-0.967), respectively, surpassing the performance of other models. Additionally, the calibration curve's Brier Score (95%) for the RF model reached the minimum value of 0.06256 (0.05753-0.06759). SHAP analysis provided independent explanations, reaffirming the critical clinical factors associated with the risk of metastasis in OCCC patients. CONCLUSIONS: This study successfully established a precise predictive model for OCCC patient metastasis using machine learning techniques, offering valuable support to clinicians in making informed clinical decisions.

CD8 + T cell-based molecular subtypes with heterogeneous immune landscapes and clinical significance in acute myeloid leukemia.

BACKGROUND: Acute myeloid leukemia (AML) is a heterogeneous hematological malignancy. Although high-dose chemotherapy is the primary treatment option, it cannot cure the disease, and new approaches need to be developed. The tumor microenvironment (TME) plays a crucial role in tumor biology and immunotherapy. CD8 + T cells are the main anti-tumor immune effector cells, and it is essential to understand their relationship with the TME and the clinicopathological characteristics of AML. METHODS: In this study, we conducted a systematic analysis of CD8 + T cell infiltration through multi-omics data and identified molecular subtypes with significant differences in CD8 + T cell infiltration and prognosis. We aimed to provide a comprehensive evaluation of the pathological factors affecting the prognosis of AML patients and to offer theoretical support for the precise treatment of AML. RESULTS: Our results indicate that CD8 + T cell infiltration is accompanied by immunosuppression, and that there are two molecular subtypes, with the C2 subtype having a significantly worse prognosis than the C1 subtype, as well as less CD8 + T cell infiltration. We developed a signature to distinguish molecular subtypes using multiple machine learning algorithms and validated the prognostic predictive power of molecular subtypes in nine AML cohorts including 2059 AML patients. Our findings suggest that there are different immunosuppressive characteristics between the two subtypes. The C1 subtype has up-regulated expression of immune checkpoints such as CTLA4, PD-1, LAG3, and TIGITD, while the C2 subtype infiltrates more immunosuppressive cells such as Tregs and M2 macrophages. The C1 subtype is more responsive to anti-PD-1 immunotherapy and induction chemotherapy, as well as having higher immune and cancer-promoting variant-related pathway activity. Patients with the C2 subtype had a higher FLT3 mutation rate, higher WBC counts, and a higher percentage of blasts, as indicated by increased activity of signaling pathways involved in energy metabolism and cell proliferation. Analysis of data from ex vivo AML cell drug assays has identified a group of drugs that differ in therapeutic sensitivity between molecular subtypes. CONCLUSIONS: Our results suggest that the molecular subtypes we constructed have potential application value in the prognosis evaluation and treatment guidance of AML patients.

Identification of cellular senescence-related signature for predicting prognosis and therapeutic response of acute myeloid leukemia.

Cellular senescence is closely related to the occurrence, development, and immune regulation of cancer. However, the predictive value of cellular senescence-related signature in clinical outcome and treatment response in acute myeloid leukemia (AML) remains unexplored. By analyzing the expression profile of cellular senescence-related genes (CSRGs) in AML samples in the TCGA database, we found that cellular senescence is closely related to the prognosis and tumor microenvironment of AML patients, and compared with normal samples, the overall expression level of senescent inducing genes in AML samples was down-regulated, while inhibitory genes were up-regulated. The risk score model further constructed and verified based on CSRGs could be used as an independent prognostic predictor for AML patients, and the overall survival (OS) of high-risk patients was significantly shortened. The area under ROC curve (AUC) values for the prediction of 1-, 3- and 5-year OS were 0.759, 0.749, and 0.806, respectively. In addition, patients with high-risk scores are more sensitive to treatment with cytarabine and may benefit from anti-PD-1 immunotherapy. In conclusion, our results suggest that the cellular senescence-related signature is a strong biomarker of immunotherapy response and prognosis in AML.