Computational estimation of quality and clinical relevance of cancer cell lines

Abstract

Immortal cancer cell lines (CCLs) are the most widely used system for investigating cancer biology and for the preclinical development of oncology therapies. Pharmacogenomic and genome‐wide editing screenings have facilitated the discovery of clinically relevant gene–drug interactions and novel therapeutic targets via large panels of extensively characterised CCLs. However, tailoring pharmacological strategies in a precision medicine context requires bridging the existing gaps between tumours and in vitro models. Indeed, intrinsic limitations of CCLs such as misidentification, the absence of tumour microenvironment and genetic drift have highlighted the need to identify the most faithful CCLs for each primary tumour while addressing their heterogeneity, with the development of new models where necessary. Here, we discuss the most significant limitations of CCLs in representing patient features, and we review computational methods aiming at systematically evaluating the suitability of CCLs as tumour proxies and identifying the best patient representative in vitro models. Additionally, we provide an overview of the applications of these methods to more complex models and discuss future machine‐learning‐based directions that could resolve some of the arising discrepancies.

A systematic evaluation of deep learning methods for the prediction of drug synergy in cancer

“…This may limit the clinical applicability of these models, as conclusions drawn from in silico studies based on cell line screens often fail to translate to the clinic. Cell lines are not always representative of their primary cancer types [71][72][73], due to mislabeling and contamination, genetic drift, or changes arising from cell culture, among other factors [74]. In addition, the use of cell line data ignores the influence of the tumor microenvironment on drug response [74], as well as the possible adverse systemic effects of drugs.…”

“…Cell lines are not always representative of their primary cancer types [71][72][73], due to mislabeling and contamination, genetic drift, or changes arising from cell culture, among other factors [74]. In addition, the use of cell line data ignores the influence of the tumor microenvironment on drug response [74], as well as the possible adverse systemic effects of drugs. Nevertheless, cell line screens are currently the only option providing sufficient data for the development of DLbased drug synergy prediction methods.…”

Integrative ensemble modelling of cetuximab sensitivity in colorectal cancer PDXs

“…This suggests that cell-line-trained models of cetuximab response struggle to predict PDX cetuximab sensitivity, primarily due to differences in the relationship between expression features and the target variable. These transcriptional differences between cell lines and PDXs might be due to the intense selection pressure imposed during cell line establishment, which makes available 2d models only partially representative of the general patient population 15 .…”

“…This is primarily due to the intrinsic limitations of such models, encompassing genetic, epigenetic, and transcriptomic changes resulting from their selective adaptation to artificial culture conditions 12, 13 . Furthermore, cancer cell lines do not maintain the complex heterogeneity of the tumour of derivation; they often lose or gain specific subclones and might miss relevant components of the human tumour stromal microenvironment 14, 15 .…”

Opportunities for pharmacoproteomics in biomarker discovery

“…PDXs are derived from human tissue excised from a patient's tumour and transplanted into an immunodeficient mouse. All of these pre-clinical models (see Figure 2) bear many resemblances to their tissues of origin, but they are unable to sufficiently represent all elements of a human tissue sample [65][66][67][68][69][70].…”