The Cancer Cell Line Encyclopedia enables predictive modeling of anticancer drug sensitivity

PMID: 22460905 · DOI: 10.1038/nature11003 · Journal: Nature (2012)

TL;DR

This study describes the Cancer Cell Line Encyclopedia (CCLE), a large-scale genomic resource comprising gene expression, copy number, and massively parallel sequencing data from 947 human cancer cell lines across 36 tumor types, coupled with pharmacologic profiling of 24 anticancer drugs across 479 lines. Using elastic net regression and naive Bayes classification, the authors identified known and novel genomic predictors of drug sensitivity, including AHR expression as a biomarker for MEK inhibitor efficacy in NRAS-mutant lines and SLFN11 expression as a predictor of topoisomerase inhibitor sensitivity.

Cohort & data

• 947 human cancer cell lines spanning 36 tumor lineages, compiled as the Cancer Cell Line Encyclopedia (CCLE)
• Genomic profiling: targeted massively parallel sequencing of >1,600 genes, mass spectrometric genotyping (OncoMap) of 492 mutations in 33 cancer genes, Affymetrix SNP 6.0 copy number arrays, Affymetrix U133 Plus 2.0 expression arrays
• Pharmacologic profiling: 8-point dose-response curves for 24 anticancer compounds across 479-481 cell lines
• Predictive modeling via elastic net regression and naive Bayes classification using >50,000 genomic input features

Key findings

• Cell lines recapitulated primary tumor genomic features with strong correlations in copy number (median r = 0.77, p < 10^-15), gene expression (median r = 0.60, p < 10^-15), and point mutation frequencies (median r = 0.71, p < 10^-2 for all but 3 lineages)
• Activating mutations in BRAF and NRAS were among the top four predictors of sensitivity to the MEK inhibitor PD-0325901; KRAS mutations were also identified with lower predictive value
• Additional top predictors: EGFR mutations for erlotinib; ERBB2 amplification/overexpression for lapatinib; BRAF V600E for RAF inhibitors PLX4720 and RAF265; HGF expression and MET amplification for PF-2341066; MDM2 overexpression for nutlin-3a
• NQO1 expression was the top predictor of sensitivity to the Hsp90 inhibitor 17-AAG
• AHR expression correlated with NRAS-mutant cell line sensitivity to MEK inhibitors; shRNA knockdown of AHR suppressed growth in three NRAS-mutant cell lines with elevated AHR expression but had no effect in two lines with low AHR expression
• MEK inhibitors PD-0325901 and PD-98059 reduced endogenous CYP1A1 mRNA levels in NRAS-mutant melanoma cells (IPC-298 and SK-MEL-2) but not in neuroblastoma cells (CHP-212)
• SLFN11 expression was the top correlate of sensitivity to irinotecan and topotecan (TOP1 inhibitors); significant in 12 of 16 lineages (Pearson’s r >= 0.2)
• Multiple myeloma cell lines (12 of 14 tested) showed enhanced sensitivity to IGF-1 receptor inhibitor AEW541, with high IGF1 expression; elevated IGF1R expression also correlated with sensitivity
• Hematologic lineages were predictors of panobinostat (HDAC inhibitor) sensitivity
• Ewing sarcoma cell lines exhibited the highest SLFN11 expression among 4,103 primary tumor samples spanning 39 lineages, and all three screened lines were sensitive to irinotecan

Genes & alterations

• BRAF — V600E mutation predicts sensitivity to MEK inhibitors (PD-0325901) and RAF inhibitors (PLX4720, RAF265)
• NRAS — activating mutations predict MEK inhibitor sensitivity; AHR expression further stratifies response within NRAS-mutant lines
• KRAS — mutations identified as predictor of MEK inhibitor sensitivity with lower predictive value than BRAF/NRAS
• EGFR — mutations predict erlotinib sensitivity
• ERBB2 — amplification/overexpression predicts lapatinib sensitivity
• MET — amplification predicts sensitivity to PF-2341066 (MET/ALK inhibitor)
• HGF — expression predicts sensitivity to PF-2341066
• MDM2 — overexpression predicts nutlin-3a sensitivity
• AHR — elevated expression correlates with MEK inhibitor sensitivity in NRAS-mutant lines; functional dependency confirmed by shRNA knockdown
• SLFN11 — expression predicts sensitivity to topoisomerase I inhibitors (irinotecan, topotecan) across multiple lineages
• NQO1 — expression is the top predictor of 17-AAG (Hsp90 inhibitor) sensitivity
• IGF1 / IGF1R — high expression in multiple myeloma correlates with IGF-1R inhibitor (AEW541) sensitivity
• PTEN, PTPN5, SPRY2 — expression identified as additional predictive features for MEK inhibition
• MITF / SOX10 — regulated genes overexpressed in melanoma-specific model predicting RAF inhibitor sensitivity
• EXT2 — variants significantly correlated with erlotinib sensitivity
• TOP1 — target enzyme of irinotecan and topotecan
• CYP1A1 — transcriptional target of AHR; expression reduced by MEK inhibitors in NRAS-mutant melanoma cells
• TP53 — highly prevalent mutation across tumor types; excluded from some analyses to avoid confounding

Clinical implications

• SLFN11 expression may serve as a biomarker to stratify patients for topoisomerase inhibitor treatment, particularly in Ewing sarcoma where SLFN11 expression is highest among solid tumors
• Multiple myeloma may be a promising indication for IGF-1 receptor inhibitor clinical trials based on high IGF1/IGF1R expression and enhanced AEW541 sensitivity
• AHR expression may serve as a mechanistic biomarker for enhanced MEK inhibitor sensitivity in NRAS-mutant cancers
• NQO1 expression may function as a predictive biomarker for Hsp90 inhibitor response
• Lineage-specific drug sensitivities (e.g., hematologic cancers to HDAC inhibitors) support tissue-of-origin-informed treatment selection
• Large annotated cell line collections can enable preclinical stratification schemata that may inform cancer clinical trial design

Limitations & open questions

• In vitro cell line drug sensitivity may not fully recapitulate in vivo tumor response due to absence of microenvironment, immune interactions, and pharmacokinetic factors
• Lineage can confound predictive analyses; lineage-specific models sometimes outperformed global models (demonstrated for melanoma with MEK inhibitors)
• SLFN11 knockdown did not affect steady-state growth, leaving the mechanistic basis for its predictive value unresolved
• The AHR-MEK inhibitor interaction was validated only in NRAS-mutant cell lines and awaits clinical confirmation
• Only 24 compounds were profiled; broader pharmacologic interrogation and additional data types (whole genome sequencing, epigenetics, proteomics) could improve predictive power
• Clinical validation of identified biomarkers (SLFN11 for topoisomerase inhibitors, AHR for MEK inhibitors, IGF1/IGF1R for IGF-1R inhibitors) remains to be performed

Citations from this paper used in the wiki

• “SLFN11 expression [was] the top correlate of sensitivity to irinotecan … SLFN11 expression also emerged as the top predictor of topotecan sensitivity” PMID:22460905
• “AHR expression was associated with MEK inhibitor efficacy in NRAS-mutant lines” PMID:22460905
• “most multiple myeloma cell lines (12 of 14 lines tested) exhibited enhanced sensitivity to the IGF-1 receptor inhibitor AEW541” PMID:22460905
• “NQO1 expression was identified as the top predictive feature for sensitivity to the Hsp90 inhibitor 17-AAG” PMID:22460905
• “activating mutations in BRAF and NRAS were among the top four predictors of sensitivity … for the MEK inhibitor PD-0325901” PMID:22460905
• “Ewing’s sarcomas also exhibited the highest SLFN11 expression among 4,103 primary tumor samples spanning 39 lineages” PMID:22460905

This page was processed by paper-compiler on 2026-05-06.