Unveiling the limitations of drug sensitivity analyses involving cancer cell lines

Human cancer cell lines have become one of the most widely used tools in cancer research, playing a crucial role in preclinical drug discovery and development. Today, there are numerous publicly available datasets on the genetic profiles and drug sensitivities of these cell lines. This enables scientists to identify potential drug candidates tailored to specific cancer genotypes through data mining techniques.

While this strategy holds promise in reducing costs and time spent on trial-and-error, there is no guarantee that cancer cells in real tumors will respond to drugs in the same way as cancer cell lines do in vitro. For example, tumor cells with homologous recombination deficiency (HRD), indicating a dysfunctional DNA repair pathway, have shown sensitivity to therapies involving platinum agents and PARP inhibitors (a type of targeted cancer drug) in vivo. However, it remains unclear whether there is any association between HRD and sensitivity to these drugs in cancer cell lines.

In a recent study published in the journal Scientific Data on 6 February 2024, a research team from Japan, led by Professor Noriomi Matsumura from Kindai University, aimed to address these knowledge gaps. Their paper was co-authored by Dr. Kosuke Murakami from Kindai University and Dr. Shiro Takamatsu from Kyoto University Graduate School of Medicine.

The team conducted a comprehensive analysis of genomic profiles and drug response data of cancer cell lines within large-scale publicly available datasets. Their aim was to shed light on whether HRD status is indeed a predictor of higher efficacy for PARP inhibitors and platinum-based drugs.

The data for the analysis were sourced from a combination of the Cancer Cell Line Encyclopedia, COSMIC-Cell Line Project, and Gene Expression Omnibus databases. Initially, the researchers focused on BRCA1 and BRCA2 gene alterations, using the “area under the drug response curve (AUC)” as the measure for drug efficacy. Put simply, a higher AUC value indicates greater resistance to a given drug.

The results were quite surprising, as Prof. Matsumura highlights: “Among 10 platinum agent assays, including Oxaliplatin 1 and Cisplatin 1, higher AUCs were noted in cell lines with BRCA alterations. Moreover, among the 14 PARP inhibitor assays considered, none showed a significant difference between the BRCA mutants and the control group.” Subsequent analyses focusing on HRD scores, Mutational Signature 3, and other gene mutations related to the homologous recombination repair pathway yielded similar results.

These findings are counterintuitive as one would expect HRD cells to be particularly sensitive to drugs that induce DNA damage or affect DNA repair, such as platinum agents and PARP inhibitors. This, in turn, suggests that assumptions about drug efficacy based solely on in vitro experiments involving cancer cell lines could lead to incorrect conclusions. Clinical tests have shown both types of drugs to be effective in treating HRD cells.

“Together, these findings highlight the limitations of relying solely on existing cell line databases for drug and biomarker discovery. It is crucial that we approach the data from these sources with caution,” remarks Prof. Matsumura. “Moreover, these unexpected findings in clinical oncology may also apply to other targeted drugs beyond PARP inhibitors, calling for extra caution among scientists using these databases for future cancer research.”

Hopefully, this study will encourage cancer researchers to reassess their research methods, leading to the development of clinically relevant drugs that can significantly improve the prognosis of cancer patients.