Background

Kinase fusions (KFN) are well recognized as targetable drivers in some cancers, and KFN common in one disease can be found in unrelated histologies, as for BRAF. Recently, oncogenic KDD in BRAF and EGFR were reported, along with responses to tyrosine kinase inhibitors (TKI). We assessed the frequency of KDD and KFN across 114,200 advanced cancers to reveal the landscape of oncogenic KFN and non-canonical rearrangements (KRE) in a wide variety of subtypes.

Methods

CGP was performed on DNA and/or RNA from 114,200 solid tumors or heme malignancy samples for 184-406 cancer-related genes and select introns from 14-28 genes commonly rearranged in cancer. RNA sequencing for 265 genes was available for some cases. Selected genomic events were confirmed by manual inspection.

Results

KDD were observed in 598 cases (0.62%): BRAF (127), EGFR (115), FGFR3 (94), FGFR1 (40), RET (37), ERBB2 (35), PDGFRA (35), FGFR2 (28), MET (19), ROS1 (14), ALK (13), KIT (8), NTRK1 (8), FLT3 (6), FGFR4 (5), ERBB4 (4), PDGFRB (3), NTRK2 (2). KDD were seen in 2.7% of brain tumors, most often EGFR (66), BRAF (52), PDGFRA (13), and FGFR3 (26). In extracranial tumors, KDD were common for RET (13-16% of breast, lung, and thyroid KDD+ cases), MET (15-20% of uterine and brain KDD+ cases), and ALK (54% of lung KDD+ cases). KDD possibly related to TKI resistance were seen in BRAF V600E-positive melanoma and ALK-related NSCLC. Table 1 summarizes KFE and KFN for ALK, FGFR2/3, RET, and ROS1; 48-57 tumor types are affected per gene. KFN partner varied by tumor site; for ROS1, GOPC KFN predominate in gliomas and CRC, TFG KFN in sarcomas, and CD74 and EZR in NSCLC.Table:

1700O

Conclusions

KDD are enriched in brain tumors. Diverse KDD are found extracranially and may underlie acquired resistance. Index cases with clinical responses to matched TKIs suggest KDD, KFN and KRE can be targeted therapeutically in many histological subtypes. Recurrent KFN are found widely in cancer, with gene partner varying by subtype.

Legal entity responsible for the study

Foundation Medicine

Funding

Foundation Medicine

Disclosure

L.M. Gay, S. Ramkissoon, S. Daniel, J.A. Elvin, E. Severson, A.B. Schrock, V.A. Miller, P.J. Stephens, J.S. Ross, S.M. Ali: An employee of and stockholder in Foundation Medicine, Inc. D. Pavlick, J. Chung: Employee of and shareholder in Foundation Medicine, Inc. All other authors have declared no conflicts of interest.

Background

Thymic gland carcinomas include a variety of histologic subtypes with variable clinical aggressiveness and response to local and systemic therapies. We queried whether CGP could refine tumor subtypes and uncover new targeted and immunotherapy options for patients with relapsed and metastatic disease (mTC).

Methods

FFPE sections of 174 consecutive cases of mTC was sequenced using hybridization-captured, adaptor ligation-based libraries to a mean coverage depth of > 500X for up to 315 cancer-related genes plus 37 introns from 28 genes frequently rearranged in cancer. Total mutational burden (TMB) was determined on 1.1 Mb. Clinically relevant genomic alterations (CRGA) were defined as GA linked to drugs on the market or under evaluation in mechanism driven clinical trials.

Results

All mTC were clinically advanced and included 4% adenocarcinomas (TAC), 3% basaloid (TBC), 3% lymphoepitheliomatous (TLEC), 17% neuroendocrine (TNEC), 31% non-NE undifferentiated (TNOS), 40% squamous and 2% sarcomatoid (TSRC) carcinomas (Table). mTC were twice as common in men than women, had a peak incidence in late middle age, and featured an average of 4 GA/case and 0.9 CRGA/case. The most common molecular targets were KIT and PIK3CA. Other targets were PDGFRA, FGFR3, PTCH1, FBXW7, BRCA2, IDH1, ERBB2 and ERBB3. The more frequent subtypes (TNEC, TSCC and TNOS) tended to have more GA, with KIT targets in ∼ 10% of cases. Low TMB in mTC was common; only 6% of cases had >10 mut/Mb and 3% had >20 mut/Mb. Examples of mTC with responses to targeted therapies will be presented.Table:

1701O

Conclusions

mTC histologic subtypes have varying GA and TMB status. The more common TSCC, TNEC and TNOS feature more GA, and when combined with TAC have more CRGA including KIT mutations and higher TMB. CGP shows promise to guide both targeted and immunotherapy selection for patients with this rare malignancy.

Legal entity responsible for the study

Jeffrey S Ross

Funding

None

Disclosure

J.S. Ross, V.A. Miller, P.J. Stephens: Employee, leader and owns stock in Foundation Medicine. P. Vanden Borre, N. Almog, A.B. Schrock, J. Chung, J-A. Vergilio, J. Suh, S. Ramkissoon, E. Severson, S. Daniel, S.M. Ali, J.A. Elvin, L.M. Gay: Employee and owns stock in Foundation Medicine.

Background

Next-generation sequencing (NGS) of circulating tumor DNA (ctDNA) enables non-invasive profiling of solid tumor cancers. Over the past few years, research and clinical practice guidelines have highlighted a role for liquid biopsy in patient care; however, few large datasets on clinical use have been published.

Methods

Somatic genomic profiles of 35,492 plasma samples from 30,024 advanced cancer patients were determined by a ctDNA NGS test targeting up to 73 genes (Guardant360®). Accuracy of ctDNA-detected driver alterations (PPV) was assessed by comparing to available matched tissue tests for 646 patients (lung, colon, and other cancer types). A pooled response rate analysis was performed across published/in press datasets presenting response data to alterations detected by Guardant360^®.

Results

The full cohort consisted of non-small cell lung cancer (NSCLC) (39%), breast (16%), colorectal (CRC) (10%) and multiple other solid cancer types (35%), with ctDNA alterations detected in 88%, 86%, 88%, and 82%, respectively (86% overall). 19% of patients had 1 or more ctDNA alterations associated with an FDA-approved therapy. Resistance variants were identified in 18% of NSCLC, breast, CRC, prostate, melanoma and GIST patients. PPV ranged from 92-100% for EGFR L858R/E19del/E20ins (98%), ALK/RET/ROS1 fusions (92%), BRAF V600E (95%), KRAS G12/G13/Q61 (94%), and MET E14 skipping mutations (100%). Pooled response rate to 1^st line EGFR TKIs (n = 43 NSCLC): 86% [95% CI: 71-94%]; to osimertinib (n = 19 NSCLC): 94% [72-99%]; to rociletinib (n = 63 NSCLC): 54% [41-67%]; to crizotinib (n = 11 NSCLC): 82% [48-97%]; to anti-HER2 agents (n = 7 breast): 86% [49-97%]; (n = 5 gastric): 80% [37-96%].

Conclusions

Use of liquid biopsies is increasing in clinical care, providing an option of obtaining genomic information non-invasively. This dataset, derived from liquid biopsy use in clinical practice, highlights the clinical impact of identifying alterations that are targetable by drugs with regulatory approval, including emergent resistance alterations.

Legal entity responsible for the study

Guardant Health, Inc.

Funding

None

Disclosure

S.K. Pal: Honoraria from Novartis, Medivation, and Astellas Pharma, is a consultant/advisor for Pfizer, Novartis, Aveo, Myriad Pharmaceuticals, Genentech, Exelixis, Bristol-Myers Squibb, Astellas Pharma, and receives research funding from Medivation. C. Brooks, D. Chudova, J. Odegaard, S. Mortimer, K. Banks, R.J. Nagy, A. Baca, R. Lanman, H. Eltoukhy, A. Talasaz: Employee and stockholder of Guardant Health, Inc. P. Mack: Honoraria from Guardant Health, Inc. All other authors have declared no conflicts of interest.

Background

Mutations in KRAS are the most frequent RAS alterations in human cancers and the prevalent driver event in lung adenocarcinoma (LUAD). There are no effective targeted therapies for KRAS-driven LUAD and chemotherapy remains the standard of care. Small-molecule inhibitors of the MAPK pathway, one of the prominent downstream KRAS mediators, show minimal clinical activity either as single agents or in combination with chemotherapy. Recently, wild-type KRAS (KRAS^WT) was shown to enhance tumor fitness in KRAS mutant AML and CRC cell lines while concomitantly increasing sensitivity to MEK inhibition. We hypothesized that dimerization between KRAS proteins could be a key regulator for lung adenocarcinoma biology and determinant of treatment response.

Methods

To study the role of wild-type KRAS in the context of KRAS-driven cancer cells, we used genetically inducible models of KRAS loss of heterozigosity (LOH). We developed an isogenic KRAS^MUT inducible system that lacks endogenous HRas/NRas but harbors conditional CRE^ERT2-controlled KRas^lox alleles. Furthermore, we reconstituted KRAS^WT and dimerization-deficient KRAS^D154Q in KRAS-driven murine and human LUAD cell lines lacking the wild-type KRAS allele and evaluated the in vitro and in vivo impact on tumor progression and response to MEK inhibition.

Results

KRAS^WT decreased in vitro and in vivo fitness of human and murine KRAS mutant LUAD tumor cells. However, this phenotype was reverted upon MEK inhibition, with KRAS LOH cells being more sensitive than KRAS^WT expressing cells. Interestingly, both effects were dependent on wild-type/mutant KRAS dimerization and not observed with the dimerization-deficient KRAS^D154Q. We provide a mechanistic model of the ambivalent function of KRAS^WT, linking its tumor suppressor function with increased MEK inhibitor resistance through dimerization with mutant KRAS.

Conclusions

• KRAS^WT affects cellular fitness in KRAS-driven LUAD • KRAS^WT impairs response to MEK inhibitors in KRAS-driven LUAD • KRAS^WT inhibitory effect is dependent on dimerization with mutant KRAS • Impaired wild-type/mutant KRAS dimerization restores sensitivity to MEK inhibitors in vivo.

Legal entity responsible for the study

Dana Farber Cancer Institute

Funding

Stand Up To Cancer - American Cancer Society Lung Cancer Dream Team Translational Research Grant (Grant Number: SU2C-AACR-DT17-15)

Disclosure

J. Kohler: Received consultant honoraria from Boehringer Ingelheim for writing and publishing two review articles on afatinib and travel grants from Roche, Amgen and Lilly. K.D. Westover: Reports receiving a commercial research grant from Astellas Pharmaceuticals. P.A. Jänne: Stock Ownership: Gatekeeper Pharmaceuticals Advisory Role: AstraZeneca, Boehringer Ingelheim, Pfizer, Chugai Pharmaceutical, ARIAD Pharmaceuticals, Merrimack Pharmaceuticals, Roche, Genentech, Loxo Oncology, Ignyta Research Funding: Astellas Pharma, AstraZeneca. All other authors have declared no conflicts of interest.

Background

Elevated levels of CUB domain-containing protein 1 (CDCP1) have been reported to be associated with poor prognosis in several human malignancies, including prostate cancer. However, its oncogenic role remains unexploited. Taking an advantage of multiple cross species genetic models, we demonstrate that CDCP1 per se is an oncogene. Particularly in mice, overexpression of CDCP1 in prostatic epithelial cells initiates hyperplasia, which eventually develops high-grade prostatic intraepithelial neoplasia. The functional importance of CDCP1 in tumuorigenesis was further fortified based on the evident display of invasive and metastatic prostate tumors upon its overexpression concomitantly with loss of Pten in mouse prostates. Mechanistically, we demonstrate that CDCP1 leads to the activation of SRC that further enhances the level of the transcription factor c-Myc in vivo. In turn, enhanced Myc triggers transcriptional activation of Cyclin D1 and COUP transcription factor II (COUPTF-II) that bypasses the TGF-β-dependent checkpoint and senescence barrier driven by Pten-loss. Following on, we demonstrate that targeting CDCP1 antagonizes c-Myc expression to block tumourigenesis by reactivating cellular senescence in human prostate cancer cells.

Methods

CDCP1, transgenic mouse model, PTEN, prostate cancer.

Results

-Conditional overexpression of CDCP1 promotes tumourigenesis in different transgenic model systems -CDCP1 cooperates with Pten-loss in driving full malignant prostate tumourigenesis -Overexpression of CDCP1 overcomes Pten- loss induced cellular senescence by activating Myc-Targeting CDCP1 induces cellular senescence and growth arrest by downregulating Myc.

Conclusions

In sum, our findings highlight a crucial role for CDCP1 in 1) driving tumourigeneis in several transgenic models and 2) inducing full malignant progression of Pten-null prostate tumourigenesis by eliciting an oncogenic and tumour suppressive network that results in senescence evasion and metastasis. As CDCP1 is a cell surface protein which can be targeted by currently available monoclonal antibodies, our study also put forth a novel therapeutic strategy to target SRC and MYC tumours in metastatic human prostate cancer.

Legal entity responsible for the study

IOR-Bellinzona- Andrea Alimonti

Funding

None

Disclosure

All authors have declared no conflicts of interest.