Prognostic Significance of CDKL2 in Non-Small Cell Lung Cancer: A Favorable Association in Lung Adenocarcinoma
TL;DR
Higher CDKL2 expression showed a favorable prognostic association in LUAD, with consistent TCGA sensitivity analyses and limited but suggestive independent-cohort support, and CDKL2 should be regarded as a candidate prognostically associated marker rather than an established clinical biomarker.
Abstract
Background and Objectives: Non-small cell lung cancer (NSCLC) comprises biologically distinct histologic subtypes, including lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC). We evaluated whether the prognostic association of cyclin-dependent kinase-like 2 (CDKL2) differs by histology. Materials and Methods: TCGA CDKL2 expression was linked to harmonized clinical data. The primary analysis used a pooled multivariable Cox model with continuous standardized CDKL2 expression, histology, and a CDKL2 × histology interaction, adjusted for age, sex, pathologic stage, and smoking. Additional analyses assessed model assumptions, tumor purity, tumor-versus-normal expression, KEGG pathways, the tumor microenvironment, and single-cell RNA sequencing. External evaluation used GSE30219. Results: The primary TCGA cohort included 943 patients (471 LUAD, 472 LUSC; 375 deaths). Higher CDKL2 expression was associated with lower mortality in LUAD (HR per 1 pooled SD = 0.721, 95% CI 0.603–0.862, p < 0.001) but higher mortality in LUSC (HR = 1.237, 95% CI 1.011–1.515, p = 0.039), with a significant interaction (HR = 1.717, p < 0.001) persisting after tumor-purity adjustment. CDKL2-low tumors were enriched for cell-cycle, DNA-replication, proteasome, and DNA-repair programs. Single-cell analyses showed predominant epithelial detection and greater malignant-cell CDKL2 detection in LUAD. In GSE30219, the favorable LUAD association was supported in a median-split-adjusted analysis of 207073_at (HR = 0.473, 95% CI 0.250–0.897, p = 0.022), whereas the corresponding continuous estimate was directionally favorable but nonsignificant; 236331_at was nonsignificant in both models; neither probe supported the adverse LUSC association or histology interaction. Conclusions: Higher CDKL2 expression showed a favorable prognostic association in LUAD, with consistent TCGA sensitivity analyses and limited but suggestive independent-cohort support. The adverse LUSC association remains preliminary, and CDKL2 should be regarded as a candidate prognostically associated marker rather than an established clinical biomarker.