Over the past decade, three-dimensional (3D) cancer models have emerged as crucial tools in high-throughput screening (HTS), providing a more reliable bridge between in vitro and in vivo drug testing. These models demonstrate higher predictive accuracy compared to traditional two-dimensional (2D) cultures, although the lack of reproducibility and standardization is the main challenge for their translation into quantitative disease modeling and therapeutic response. Therefore, optimization and standardization of 3D spheroid cultivation protocols for HTS for various cancer cell types remain pertinent. This study establishes optimized protocols for cultivating four cancer cell lines (HCT116, SW620, PANC-1, and BxPc-3) as 3D spheroids for drug screening applications. Through systematic parameter optimization, we determined optimal seeding densities (500–1000 cells/well), formation times (3 days), and incubation periods (7–21 days) for each cell line. The application of both IC50 and score metrics allowed for a comprehensive assessment of drug efficacy, overcoming limitations associated with standard sigmoidal dose–response analysis. Testing a library of more than 100 potential CDK inhibitors identified five CDK7 inhibitors and one CDK8 inhibitor with nanomolar potency, with colorectal cancer spheroids showing approximately 10% greater sensitivity than pancreatic cancer spheroids in the cases of iCDK7_4 and iCDK7_5. This standardized approach improves reproducibility in 3D cancer model screening, potentially accelerating the preclinical evaluation of novel anticancer agents against these challenging malignancies.
Background: Three-dimensional (3D) cancer models are currently essential tools in high-throughput screening (HTS), serving as a bridge between in vitro and in vivo approaches during drug development. Even though spheroids offer many advantages over 2D cultures, analyzing 3D cultures with heterogeneous morphology remains challenging due to the lack of standardized visualization techniques and multiparameter analysis. Methods: In this work, an optimized CellProfiler pipeline and a Python algorithm for weighting morphological features are used to visualize, extract, and analyze morphological data from spheroids derived from colorectal and pancreatic cancer cell lines with diverse morphologies (HCT116, LoVo, PANC-1, and CFPAC-1). Results: We developed a feature weighting process that combines multiple morphological parameters into a single metric using principal component analysis (PCA). There is a strong correlation between this process and a standard Alamar Blue proliferation assay (r = 0.89, ρ = 0.91, p < 0.001). Using this method, we were able to ascertain the IC50 values of substances that did not produce results in cell lines with heterogeneous morphology (LoVo and CFPAC-1) using a standard proliferation assay. Conclusions: By removing the need for tracer dyes, the resulting methodology may lower costs while accelerating preclinical drug development through informative multiparameter analysis of compound efficacy.