Small Cell Lung Cancer Screen of Oncology Drugs, Investigational Agents, and Gene and microRNA Expression

Abstract

Background: Small cell lung carcinoma (SCLC) is an aggressive, recalcitrant cancer, often metastatic at diagnosis and unresponsive to chemotherapy upon recurrence, thus it is challenging to treat.

Methods: Sixty-three human SCLC lines and three NSCLC lines were screened for response to 103 US Food and Drug Administration-approved oncology agents and 423 investigational agents. The investigational agents library was a diverse set of small molecules that included multiple compounds targeting the same molecular entity. The compounds were screened in triplicate at nine concentrations with a 96-hour exposure time using an ATP Lite endpoint. Gene expression was assessed by exon array, and microRNA expression was derived by direct digital detection. Activity across the SCLC lines was associated with molecular characteristics using pair-wise Pearson correlations.

Results: Results are presented for inhibitors of targets: BCL2, PARP1, mTOR, IGF1R, KSP/Eg5, PLK-1, AURK, and FGFR1. A relational map identified compounds with similar patterns of response. Unsupervised microRNA clustering resulted in three distinct SCLC subgroups. Associating drug response with micro-RNA expression indicated that lines most sensitive to etoposide and topotecan expressed high miR-200c-3p and low miR-140-5p and miR-9-5p. The BCL-2/BCL-XL inhibitors produced similar response patterns. Sensitivity to ABT-737 correlated with higher ASCL1 and BCL2. Several classes of compounds targeting nuclear proteins regulating mitosis produced a response pattern distinct from the etoposide response pattern.

Conclusions: Agents targeting nuclear kinases appear to be effective in SCLC lines. Confirmation of SCLC line findings in xenografts is needed. The drug and compound response, gene expression, and microRNA expression data are publicly available at http://sclccelllines.cancer.gov.

Figure 1.

Analysis of etoposide response. A) Waterfall plot showing the range (25 nM to > 10 uM) of IC50s for small cell lung cancer (SCLC) lines exposed to etoposide for 96 hourrs. Blue bars are SCLC lines developed from previously treated tumors; red bars are SCLC lines developed from treatment-naive tumors; purple bars are SCLC lines developed from tumors of unknown treatment status. B) Concentration response curves for the 63 SCLC lines exposed to etoposide or topotecan for 96 hours. C) Etoposide IC50 plotted vs the teniposide, topotecan, talazoparib, and gemcitabine IC50s showing a correlation of R = 0.89, 0.87, 0.76, and 0.68, respectively, for the SCLC lines. D) Heat map of the etoposide eight most sensitive and eight most resistant SCLC lines along with microRNA expression from counts for miR-140-5p, miR-200c-3p, and miR-9-5p. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, greeen is high, and red is low. IC50 = the inhibitory concentration producing 50% growth inhibition.

Figure 2.

Constellation relational map showing response similarity connections among the approved and investigational anticancer agents tested in the small cell lung cancer (SCLC) lines at a stringency of 0.77. The line thickness is directly proportional to the pair-wise correlation.

Figure 3.

MicroRNA small cell lung cancer (SCLC) line clustering. A) SCLC subgroups based upon unsupervised microRNA clustering. Left to right are the red , blue , and goldenrod clusters, with the green cluster being normal cell comparators. B) Box plot showing the median and association of etoposide response of the SCLC lines, with the SCLC microRNA subgroups ( P = .14). C) Box plot showing the median and association of VS-507 (salinomycin) response of the SCLC lines, with the SCLC microRNA subgroups ( P = .12). Unsupervised clustering of the SCLC lines based on microRNA expression was performed using partitioning around medoids algorithm. The number of clusters was selected by maximum width of cluster silhouette. Differences in drug response between subgroups were tested using analysis of variance, and agents with the smallest P values were evaluated. Analysis was performed using R ( http://www.R-project.org ).

Figure 4.

Bcl-2 and PARP inhibitors. A) Heat map showing the IC ₅₀ response of the small cell lung cancer (SCLC) lines, arranged by response to ABT-737, for five Bcl-2 inhibitors. The most sensitive SCLC lines are: NCI-H2107, NCI-H889, NCI-H1963, NCI-H1105, and NCI-H748; the least sensitive SCLC lines are: NCI-H378, NCI-841, NCI-H196, DMS273, and DMS114. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. The gene expression values were: ASCL1 mean = 9.96 (range = 4.97–12.67), and the correlation with response to ABT737 was R = -0.497; BCL2 mean = 7.93 (range = 5.95–10.2), and the correlation with response to ABT737 was R = -0.588. B) ABT-263 and ABT-737 log ₁₀ IC ₅₀ plotted vs log ₂ BCL2 gene expression showing a correlation of R = - 0.476 with ABT-263 response and R = -0.588 with ABT-737 response for the SCLC lines. C) Heat map showing the IC ₅₀ response of the SCLC lines, arranged by response to talazoparib, for six PARP inhibitors. The most sensitive SCLC lines are: NCI-H211, NCI-H209, NCI-H774, NCI-H526, and NCI-H1048; the least sensitive SCLC lines are: SW1271, NCI-2029, NCI-H196, NCI-H1688, and COR L88. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. The gene expression values were: SLFN11 mean = 7.15 (range = 4.43–10.04), and the correlation with response to talazoparib was R = -0.513. IC ₅₀ = the inhibitory concentration producing 50% growth inhibition.

Figure 5.

ASCL-1 and MYC analysis. A) Plot of the relative gene expression of ASCL-1 vs the relative gene expression of c-Myc. Each point is a small cell lung cancer (SCLC) line. In general, there was an inverse relationship between the expression of ASCL-1 and c-Myc. The correlation is R = -0.59. B) Heat map organized by expression of ASCL-1 and showing the miRs that are positively and negatively correlated with expression of ASCL-1. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. The correlation values are: miR-95 (R = 0.66), miR-141-3p (R = 0.61), miR-7-5p (R = 0.54), miR-200b-3p (R = 0.66), miR-200a-3p (R = 0.62), miR-200c-3p (R = 0.59), miR-135a-5p (R = 0.59), miR-375 (R = 0.69), miR-455-3p (R = -0.58), and miR-455-5p (R = -0.62).

Figure 6.

mTOR and IGF1R inhibitors. A) Heat map showing the IC ₅₀ response of the small cell lung cancer (SCLC) lines, arranged by response to BEZ-235 for 12 mTOR inhibitors. The most sensitive SCLC lines are: LXFS 650L, DMS 114, NCI-H1882, NCI-H719, and NCI-H1048; the least sensitive SCLC lines are: NCI-H711, NCI-1417, NCI-H378, DMS187, and NCI-H660. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. The gene expression values were: MTOR mean = 9.04 (range = 7.84–10.69), and the correlation with response to sirolimus was R = -0.23. B) Concentration response curves for the 63 SCLC lines exposed to everolimus or BEZ-235 for 96 hours. C) Heat map showing the IC ₅₀ response of the SCLC lines, arranged by response to linsitinib for eight IGF1R inhibitors. The most sensitive SCLC lines are: LXFS 605L, NCI-H187, NCI-H526, COLO 668, and NCI-H1876; the least sensitive SCLC lines are: NCI-H1048, DMS 53, DMS273, DMS114, and COR L88. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. There is no association with the expression of the gene shown or the microRNAs shown. D) Concentration response curves for the 63 SCLC lines exposed to linsitinib for 96 hours.

Figure 7.

Nuclear kinase inhibitors. A) Heat map showing the IC ₅₀ response of the small cell lung cancer (SCLC) lines, arranged by response to ARQ-621 for six KSP/Eg5 inhibitors. The most sensitive SCLC lines are: NCI-H69, NCI-H1105, NCI-H211, NCI-H2107, and NCI-H1048; the least sensitive SCLC lines are: NCI-H69/LX10, DMS187, NCI-H2029, NCI-H196, and DMS53. The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. B) Concentration response curves for the 63 SCLC lines exposed to ARQ-621 for 96 hours. C) ARQ-621 log ₁₀ IC ₅₀ plotted vs log ₂ EPAS1 and TMEM127 gene expression showing a correlation of R = 0.71 for EPAS1 with ARQ-621 response and R = 0.55 for TMEM127 with ARQ-621 response for the SCLC lines. The expression values were: EPAS1 mean = 7.58 (range = 5.72–12.72), and TMEM127 mean = 9.09 (range = 8.15–11.07). D) Heat map showing the log ₁₀ IC ₅₀ response of the SCLC lines, arranged by response to ARQ-621 for six KSP/Eg5 inhibitors, eight Polo-like kinase inhibitors, 10 aurora kinase inhibitors, and showing response to etoposide. Green indicates IC ₅₀ ≥ 10 ^-5 ; yellow indicates IC ₅₀ ≥ 10 ^-7 ; red indicates IC ₅₀ ≥ 10 ^-9 . IC ₅₀ = the inhibitory concentration producing 50% growth inhibition.

Figure 8.

An exceptional responder. A) Heat map showing the IC ₅₀ response of the small cell lung cancer (SCLC) lines response to three FGFR1 inhibitors. Green indicates log ₁₀ ≥ 10 ^-5 ; yellow indicates IC ₅₀ ≥ 10 ^-7 ; red indicates IC ₅₀ ≥ 10 ^-9 . B) Concentration response curves for the 63 SCLC lines exposed to the FGFR1 inhibitors AZD-4547 (NSC764239), PD-173074 (NSC766908), and BGJ-398 (NSC764487) for 96 hours. C) DMS114 chromosome 8 map showing copy number on the y-axis . The data were derived from the CCLE database. D) Heat map showing the IC ₅₀ response of the SCLC lines, arranged by response to AZD-4547, gene expression of FGFR1, and microRNAs miR-548v, miR-935, miR-762, miR-645, and miR-199a-5p by the SCLC lines. The most sensitive SCLC lines are: DMS114, NCI-H1688, NCI-H526, NCI-H2107, and NCI-H211.The heat map is shown as Z score to allow comparison across scales, where yellow is the mean, green is high, and red is low. IC ₅₀ = the inhibitory concentration producing 50% growth inhibition.