In This Issue

See article, p. 199

Preclinical studies have shown that the irreversible pan-HER kinase inhibitor afatinib blocks the reactivation of HER family members that may contribute to resistance to HER2-targeting agents such as trastuzumab in ERBB2-amplified esophagogastric cancer. To evaluate the clinical efficacy of afatinib, Sanchez-Vega and colleagues conducted a phase II trial of afatinib alone or in combination with trastuzumab in patients with trastuzumab-refractory, ERBB2-amplified esophagogastric cancer. However, treatment yielded only modest clinical responses among this heavily pretreated patient cohort. Genomic analyses of pretreatment tumor biopsies identified coamplification of ERBB2 and EGFR in the patients with the greatest clinical response, suggesting that concurrent amplification of these genes may be a molecular determinant of afatinib sensitivity. In addition, HER2-directed functional PET imaging using ⁸⁹Zr-trastuzumab noninvasively detected intertumor heterogeneity, which was associated with poor response to afatinib. To further assess the impact of tumor heterogeneity on drug response, multiple metastatic lesions were obtained from rapid autopsies of 3 patients who initially responded to afatinib but later progressed. Genomic profiling of these samples revealed differential intra- and interpatient expression of HER2, EGFR, and MET, with selection for clones with loss of EGFR amplification or gain of MET amplification post–afatinib progression. Consistent with this finding, dual treatment with afatinib and a MET inhibitor resulted in durable tumor regression in a patient-derived xenograft model generated from a postprogression sample. These findings provide insight into determinants of sensitivity and resistance to afatinib in ERBB2-amplified esophagogastric cancer and suggest tools to predict clinical response.

See article, p. 210

Patients with deleterious mutations in BRCA1 and BRCA2 that impair homologous recombination repair are sensitive to platinum-based chemotherapy and PARP inhibition, but reversion mutations that restore BRCA function can lead to acquired resistance. To determine the prevalence of reversion mutations, Lin and colleagues performed targeted next-generation sequencing of circulating cell-free DNA (cfDNA) from 112 patients with BRCA-mutant high-grade ovarian carcinoma (HGOC) before and after treatment with the PARP inhibitor rucaparib. BRCA reversion mutations were discovered in pretreatment cfDNA from 2 of 11 (18%) platinum-refractory tumors, 5 of 38 (13%) platinum-resistant tumors, and 1 of 48 (2%) platinum-sensitive tumors. Median progression-free survival after rucaparib was 1.8 months in patients with BRCA reversion mutations and 9 months in patients without BRCA reversion mutations, indicating that BRCA reversion mutations predict primary resistance to rucaparib. In 97 patients with matched pretreatment tumor biopsies and cfDNA, there was good concordance between tumor DNA and cfDNA. However, cfDNA analysis was better able to capture multiclonal heterogeneity, identifying multiple unique BRCA reversion mutations in some patients. Analysis of 78 postprogression cfDNA samples revealed eight additional patients with BRCA reversion mutations that were not detected in pretreatment cfDNA and promoted acquired rucaparib resistance. Taken together, these findings indicate that BRCA reversion mutations in cfDNA can predict primary and acquired resistance to PARP inhibition in patients with HGOC.

See article, p. 220

Chloroquine derivatives used as antitumor agents are known to act in part through inhibition of lysosomal activity and autophagy, but the molecular target of these compounds has not been identified, and clinically used chloroquine derivatives such as hydroxychloroquine lead to only modest lysosomal inhibition at their maximum tolerated doses. Rebecca, Nicastri, and colleagues developed a more potent chloroquine derivative that inhibited autophagy at lower concentrations relative to its previously reported parental compound or hydroxychloroquine and had stronger and more sustained antitumor activity than either agent due to improved cell penetration in acidic conditions. The compound was linked to a photoaffinity probe to pull down its molecular target, which was identified as the lysosomal enzyme palmitoyl-protein thioesterase 1 (PPT1). All chloroquine derivatives analyzed were also found to directly bind PPT1 and inhibit its enzymatic activity, implicating PPT1 as the common molecular target of this class of compounds. Loss of PPT1 blocked the ability of chloroquine derivatives to inhibit lysosomal function and suppress autophagy, providing further evidence that inhibition of PPT1 underlies the antitumor activity of chloroquine derivatives. High PPT1 expression is observed in several cancer types and associated with shorter overall survival, and PPT1 knockout impaired autophagy and significantly slowed tumor growth in vivo. The insight into the mechanism of action of chloroquine derivatives provided by these findings thus not only has the potential to guide development of more potent inhibitors of autophagy but suggests PPT1 itself may be an attractive therapeutic target.

See article, p. 230

Small cell lung cancer (SCLC) is mostly responsive to chemotherapy, but the response is typically short-lived and there are no approved targeted therapies. Although RB1 inactivating mutations are found in almost all SCLCs, potentially targetable dependencies caused by loss of RB1 have not yet been identified. To find such dependencies, Oser and colleagues used an SCLC isogenic cell system to perform CRISPR/Cas9 screening and identify RB1-loss synthetic lethal targets. The screening revealed multiple functionally interacting genes that regulate chromosomal segregation, such as components of condensin complexes and their upstream regulators. Aurora B kinase (AURKB) was identified as the highest scoring “druggable” hit, and an AURKB-specific inhibitor, AZD2811, selectively depleted RB1-deficient and not RB1-proficient SCLC cells. AURKB inhibition was also synthetic lethal with RB1 loss in an extended panel of SCLC, non–small cell lung cancer (NSCLC), and breast cancer cell lines. Mechanistic studies showed that RB1 loss exacerbated the mitotic abnormalities caused by AURKB inhibition. In addition, AZD2811 specifically altered expression of genes in RB1-deficient cells linked to the G2/M checkpoint and mitotic spindle. Collectively, these results suggest that RB1 and AURKB have partially redundant roles in controlling mitosis in SCLC, possibly explaining their synthetic lethal relationship. Lastly, AURKB inhibition showed efficacy against RB1-/- SCLC cell line xenografts, RB1-/- SCLC PDXs, and autochthonous RB1-/- neuroendocrine tumors. Therefore, RB1 loss may be a predictive biomarker for sensitivity to AURKB inhibitors in SCLC and perhaps other RB1-/- cancers.

See article, p. 248

RB1 is mutated in some of the most aggressive and hard-to-treat malignancies, including small cell lung cancer (SCLC), and triple-negative breast cancer. The RB1 protein controls the G1-S transition in the cell cycle. Drugs that act on the cell cycle or cell-cycle regulation pathways have the potential to be synthetic lethal with mutated RB1, but no such drugs have been discovered thus far. Gong, Du, Parsons, and colleagues performed a gene–drug screen for RB1 synthetic lethality using a collection of cell-cycle inhibitors on a large panel of RB1-mutant cancer cell lines. Inhibitors of Aurora kinase A (AURKA) or Aurora kinase B (AURKB) were identified with the most pronounced cytotoxic effect toward RB1-mutant versus RB1–wild-type cell lines. Because specific AURKA inhibition was predicted to cause less of a myelosuppression effect than AURKB inhibition based on previous clinical observations, a specific AURKA inhibitor, LY3295668, was developed. Treatment with LY3295668 promoted more apoptosis in RB1-null versus RB1–wild-type lung and breast cancer cell lines with minimal effects on normal cells. RB1-null SCLC tumor xenografts regressed in response to LY3295668 at concentrations that were well tolerated in mice. A genome-scale shRNA drug suppressor screen to identify genes critical for LY3295668 cytotoxicity revealed several mitotic checkpoint complex genes, suggesting that RB1-negative cells are dependent on AURKA activity to exit mitosis. Together, these results propose that specific AURKA inhibition may provide a better therapeutic window than classic cytotoxic agents for the treatment of RB1-mutant malignancies.

See article, p. 264

Papillary carcinomas are most commonly driven by mutually exclusive mutations or fusions involving BRAF, RAS, RET, or NTRK. Other lower-frequency driver alterations have been recently described, including mutations in EIF1AX, a component of the translation preinitiation complex (PIC), but these mutations have not been well characterized. Krishnamoorthy and colleagues analyzed genomics data from 246 patients with advanced thyroid cancer and found that EIF1AX mutations occurred in 26 (11%) patients, with RAS mutations co-occurring in 25 of those. Seventeen of the EIF1AX mutations occurred in a hotspot splice acceptor site in the C-terminal tail (A113spl) and resulted in two alternatively spliced transcripts: c'splice, which results in exclusion of 12 amino acids through use of a cryptic site, and t'splice, which results in a truncated protein. The EIF1AX-A113spl mutations enhanced oncogenic transformation, accelerating colony formation in soft agar in both KRAS-mutant and wild-type cells, an effect that required the presence of c'splice. In vivo, EIF1AX-c'splice cooperated with oncogenic RAS to drive thyroid tumorigenesis. Mechanistically, the EIF1AX-A113spl mutations enhanced the affinity of EIF1AX for components of the translation PIC, resulting in stabilization of the PIC and induction of the cellular stress sensor ATF4, which suppressed EIF2α phosphorylation, globally increasing protein synthesis. RAS stabilizes MYC, and EIF1AX-A113spl further enhanced this effect, thereby sensitizing cells to MEK or BRD4 inhibition in vivo. Similarly, RAS and EIF1AX cooperated to activate mTOR, conferring sensitivity to mTOR inhibition. Collectively, these findings elucidate a mechanism by which EIF1AX mutations promote thyroid cancer and reveal potential therapeutic vulnerabilities.

See article, p. 282

Pancreatic ductal adenocarcinoma (PDAC) is characterized by a dense desmoplastic stroma comprised of cellular components, predominantly cancer-associated fibroblasts (CAF) and extracellular matrix (ECM). Having recently identified distinct myofibroblastic (mCAF) and inflammatory (iCAF) CAF subtypes in PDAC, Biffi and colleagues sought to elucidate the mechanisms underlying the development of these CAF subtypes and ascertain their roles in PDAC tumorigenesis. iCAFs, but not mCAFs or pancreatic stellate cells (PSC), exhibited increased expression of genes related to NF-κB signaling, including the NF-κB–activating cytokine IL1, which was shown to be critical for inducing NF-κB signaling in iCAFs. IL1α-deficient human PDAC organoids formed smaller tumors and harbored decreased iCAF levels in vivo compared to controls. Moreover, TGFβ signaling was activated in mouse mCAFs, but not iCAFs or PSCs, and in human PDAC; further, PSCs grown in media supplemented with TGFβ exhibited increased expression of mCAF markers and decreased expression of iCAF markers. TGFβ treatment repressed JAK/STAT3 signaling and subsequently inhibited JAK/STAT3–mediated expression of ILR1. Mechanistically, IL1-mediated induction of autocrine LIF in PSCs activates JAK/STAT signaling to promote the acquisition of the iCAF phenotype in PSCs and is antagonized by TGFβ to drive the acquisition of the mCAF phenotype. Similarly, JAK inhibitor treatment resulted in decreased tumor growth and induced the acquisition of a myofibroblastic phenotype by iCAFs in an autochthonous mouse model of PDAC. These results provide insights into the mechanisms underlying CAF heterogeneity in PDAC and provide insight into the development of potential therapeutic strategies.

Note: In This Issue is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details.