e20547 Background: Pembrolizumab produces variable benefit in PD-L1-high metastatic NSCLC, and mechanisms of resistance remain insufficiently defined. Plasma-based epigenomic profiling could potentially identify pathway-level biology not captured by PD-L1 IHC or genomic mutations alone. An exploratory post hoc analysis was performed using comprehensive epigenomic profiling from 1 mL of plasma to characterize pathways associated with pembrolizumab response and resistance from INTR@PID Lung 037 study samples. Methods: 192 longitudinally collected plasma samples (baseline, on-treatment, and end-of-treatment [EOT]) were analyzed using Precede’s comprehensive epigenomic liquid biopsy platform. 99% of samples passed analyte-specific QC metrics and 55% samples had detectable ctDNA, consistent with previous reports. Differential promoter and enhancer activation and pathway enrichments were evaluated across PFS-stratified groups. Pathway activity dependence on ctDNA levels was used to investigate whether signals originated from tumor or stroma. Pathway activities were profiled across gain of function (GOF) and loss of function (LOF) alterations. Epigenomic activation of ADC targets was also assessed. Results: Baseline plasma from non-responders (lowest PFS tertile) showed enrichment of MYC and proliferative pathways, and an RNA-based intrinsic IFN-resistance signature, whereas responders showed reduced KRAS and EMT signalling. Despite patients being selected for PD-L1 expression ≥50%, higher epigenomic PD-L1 promoter activation trended with greater tumor shrinkage. Differential pathway activities for 86 mutations were identified, including 68 linked to immune-related programs (IFN, TNFα, JAK/STAT). NF1 and ARID1A LOF mutations exhibited immune-enriched profiles, whereas KEAP1 LOF and MDM2 GOF showed immune-depleted profiles, consistent with poor ICB response. Paired baseline-EOT samples showed acquired resistance characterized by squamous-like differentiation and keratinization pathways, consistent with lineage plasticity and immune exclusion. ADC target profiling demonstrated a subset of patients exhibiting co-activation of MUC1 and EGFR , suggesting potential suitability and combination potential for bispecific or targeted ADC strategies. Conclusions: Plasma epigenomic profiling resolved genes and pathways associated with pembrolizumab response, in PD-L1-high NSCLC, including intrinsic IFN-driven resistance and acquired squamous-like resistance. This approach also captured functional immune and tumor-associated biology not evident from PD-L1 IHC or mutation status. These findings provide a clear rationale for the use of plasma epigenomics for non-invasive monitoring, patient stratification, and future evaluation of combination and targeted strategies.

Abstract Introduction: Blood-based liquid biopsies offer potential for non-invasive cancer screening. However, detecting early-stage disease is complicated by low levels of circulating tumor biomarkers and background noise from normal cells. To address this, we developed a novel deep-learning framework to detect cancer signal at the resolution of single DNA reads. Applied to bisulfite-converted cell-free DNA (cfDNA) samples, our method significantly improves early-stage cancer sensitivity. Methods: We designed a massively parallel 2-D convolutional neural network architecture that differentiates cancer and non-cancer signal in cfDNA by learning local methylation patterns at thousands of genomic regions. The model takes next generation sequencing (NGS) data as input, encodes aligned sequences within genomic windows as images, and outputs informative feature vectors for classification. However, training is complicated by two real-world data limitations: (1) disease samples contain a mix of unlabeled fragments from normal and diseased cells, and (2) acquiring sufficient early-stage disease data is costly, burdensome, and time-intensive. To address these, we designed a novel data generation technique that (1) assigns positive labels for groups of reads via in silico spike-in of tumor biopsy reads and (2) generates large, diverse datasets via fine-tuned in silico mixing of non-cancer cfDNA reads. Our model’s architecture has key advantages: compact input encoding, interpretable saliency maps, and scalable parallel architecture: we trained on 720TB of data across 10 million genomic bases in a single day, highlighting our framework’s efficiency. Results: We validated our method with targeted bisulfite sequencing data from the CORE-HH clinical study (NCT05435066, N=1229 non-cancers, N=1118 cancers, including N=599 Stage I/II). We pretrained the model on 1.3 billion training examples generated using a held-out set of non-cancer plasma (N=174) and tumor tissue biopsies (N=505). Predictions on data from clinical samples yielded feature vectors, with saliency maps confirming the model highlights biopsy-learned patterns. In a 10x5 cross-validation, classifiers trained on these feature vectors improved overall sensitivity by 9.9 points (Stage I: +6.5 pts, II: +17.6 pts, III: +14.3 pts, IV: +9.5 pts) at 98.5% specificity, compared to classifiers trained on region-wide average methylation values. These performance improvements, coupled with the scalability of the framework, underscore its potential as a transformative tool in the early diagnosis of cancer and establish a foundation for training models on NGS data in other liquid biopsy assays. Future work will investigate the potential to incorporate per-read embeddings from DNA-based large language models, without sacrificing scalability. Citation Format: Jackson A. Killian, Kade Pettie, Kyle Gowen, Shiva Farashahi, Esther Brown, Feras Hantash, Jocelyn Charlton, Franziska Michor, Kieran I. Chacko, Dorna Kashef. Improving early cancer detection by training scalable deep neural networks to extract tumor signal from cell-free DNA [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5465.

5074 Background: The PSMA-directed radioligand therapy, 177Lu-PSMA-617, is the most recent FDA approved therapy in mCRPC. Despite prolonging progression-free survival (PFS) and overall survival (OS) at a population level, response to therapy is heterogeneous and resistance remains poorly understood. Benchmarking molecular correlates of clinical outcomes following 177Lu-PSMA-617 could provide critical insights into predicting response and resistance to therapy. We applied a multimodal epigenomic liquid biopsy platform to plasma samples from mCRPC patients treated with 177Lu-PSMA-617 to characterize molecular features associated with treatment response. Methods: Baseline plasma samples were collected from patients with mCRPC at the time of PSMA PET imaging and initiation of 177Lu-PSMA-617 therapy. Epigenomic profiling of genome-wide signals from promoters, enhancers, and DNA methylation was performed on 1 mL of plasma (N=85, ctDNA ≥ 0.5%). Plasma epigenomic signals were analyzed to evaluate pathway activity, their association with treatment response using Cox proportional hazards model and neuroendocrine transformation. Response to 177Lu-PSMA-617 was determined by investigator-assessed clinical-radiographic (CR)-PFS. Results: We observed a significant association between predicted PSMA PET SUV mean from plasma epigenomic signals (using a previously derived model) and response to 177Lu-PSMA-617 (hazard ratio [HR] = 0.27, P<0.05). Further, unbiased analysis of plasma epigenomic signal across the genome identified FOLH1 (the gene encoding PSMA) as being significantly associated with CR-PFS (P<0.05). Low circulating tumor fraction was also independently associated with favorable CR-PFS (HR = 0.42, P<0.05). Pathway analysis identified activation of estrogen signalling and cellular plasticity to be associated with shorter CR-PFS, and immune signalling gene signatures to be associated with longer CR-PFS (all FDR<0.1). A subset of patients (n=4) exhibited increased plasma epigenomic signal at neuroendocrine genes, such as CHGA , DLL3 and SEZ6 . While too small to draw statistical conclusions, elevated neuroendocrine gene activity in plasma was associated with numerically shorter OS. Conclusions: Epigenomic profiling of plasma cfDNA enabled minimally-invasive characterization of molecular correlates of response and resistance, identifying genes and pathways associated with favorable and poor outcomes to 177Lu-PSMA-617 in mCRPC. By providing real-time insights into tumor biology and therapeutic efficacy, this platform supports precision medicine approaches for optimizing outcomes in PSMA-targeted therapies.

Supplementary Figures S1 - S10. Supplementary Figure S1. Additional information for Patient 1. Supplementary Figure S2. Additional information for Patient 2. Supplementary Figure S3. Additional information for Patient 3. Supplementary Figure S4. Additional clinical information for Patient 4. Supplementary Figure S5. Characterization of EGFR-RAD51 in NR6 cells. Supplementary Figure S6. Relative stability of EGFR-WT, -L858R, and -RAD51. Supplementary Figure S7. Structural model of EGFR-RAD51 filaments. Supplementary Figure S8. On-target inhibition of EGFR-RAD51 by EGFR TKI. Supplementary Figure S9. Cetuximab inhibits ligand-induced activation of downstream signaling pathways in cells expressing EGFR-RAD51. Supplementary Figure S10. cDNA sequence of EGFR-RAD51.

Genomic alterations in lung adenocarcinoma displayed in Figure 2A.

Supplementary Tables S1 - S3. Supplementary Table S1. Summary of EGFR alterations in NSCLC identified by FoundationOne. Supplementary Table S2. Summary of genomic coordinates for the kinase fusions identified in this study. Supplementary Table S3. Results of MTT curve fitting from Prism.

Abstract Background Rapid advances in clinical diagnostic testing are constrained by finite specimen availability, technical and financial burdens incurred by patient recruitment, and sample collection. Expanding the availability of patient-derived analyte for experimentation could accelerate assay development and regulatory submissions, while reducing demand for additional resource-intensive clinical studies. We developed the AReS (Archived Reference Sample) platform to utilize genomic libraries as templates for PCR amplification. The resulting PCR product, referred to as an AReS library, serves as an alternative sample type from which development, optimization, and validation studies can be iteratively performed. Methods PCR conditions were optimized to maximize yield while minimizing amplification bias, generating approximately 50X more mass than the starting input. To evaluate the AReS process, aliquots of bisulfite converted original DNA libraries were further amplified under optimized conditions to produce AReS libraries. Both the original and AReS libraries were then hybrid captured using Harbinger Health’s proprietary 8.4 Mb panel and sequenced to ≥100X unique median target coverage (MTC) depth. Original libraries were compared to AReS libraries across Picard sequencing metrics by quantifying methylation across our regions of interest and by classification as determined by our cancer yes/no (CYN) determining algorithm. Harbinger Health’s CYN algorithm was developed using a multi-layered logistic regression-based machine learning approach trained on a separate patient cohort and locked prior to being used in this study. In total, we generated 528 AReS libraries from 321 unique patient-derived DNA samples, of which 124 were from patients diagnosed with cancer and 197 from patients with no cancer diagnosis. An additional sub-study was performed on 16 paired original and AReS libraries containing unique molecular identifiers (UMIs). The UMIs allowed for the comparison of individual cfDNA molecules between the two sample types. Results All AReS samples, including both intra- and inter- batch replicates, were highly concordant to the original library. There was no significant difference across sequencing metrics (e.g., conversion efficiency or %CC and MTC). All AReS libraries had similar methylation values to original libraries; Pearson correlation was greater than 98%. Our data also indicated that greater than 97% of AReS libraries were concordant with the original library by our CYN algorithm classification. Read-level UMI analysis identified that approximately 77% of reads were common between the original and AReS libraries. As both the UMI and sequence insert used were identical, these common reads were derived from the same cfDNA molecule. As a frame of reference, sequencing replicates of the original library similarly shared approximately 77% of common reads. In addition, the Pearson correlation of read frequency compared between AReS and original libraries were within 4% difference of the correlation between original library sequencing replicates. These results showed no indication of AReS-derived amplification bias. Conclusion Taken together, the AReS process produces excess libraries that highly reproducible. AReS libraries are functionally and analytically identical to original libraries and can be applied to both research and clinical use.

Supplementary Methods, Supplementary References, Supplementary Table Legends, and Supplementary Figure Legends.

Co- and anti-occurrence of genomic alterations in lung adenocarcinoma.

Abstract Background Early detection of cancer has significant potential to impact human health and society by decreasing cancer-related morbidity and mortality. While previous approaches to identify cancer-informative biomarkers are predominantly statistical, Harbinger Health has utilized foundational discoveries from developmental biology to design a targeted methylation assay for early cancer detection from cell-free DNA (cfDNA) extracted from plasma. Utilizing this biologically informed approach, we developed a fixed multi-layered logistic regression-based machine learning algorithm, trained with an in-house generated dataset of 1046 samples (621 cancer, 425 non-cancer) that predicts a binary classification (yes/no) for cfDNA samples processed through our assay. We have previously reported high sensitivity for multi-cancer detection, including for early-stage disease. Methods Here, we perform a comprehensive independent analytical validation of our assay and algorithm, encompassing 69 subjects: 19 with newly diagnosed treatment-naïve cancer (8 different cancer types) and 50 individuals with no history, diagnosis, or cancer symptoms. In total, we utilized 122 replicate samples to assess reproducibility and precision, 8 non-template controls (water) to determine limit of blank (LOB), and cohorts of matched biopsy and cfDNA to determine tumor content limit of detection (LOD). Results Precision was assessed within five different sub-studies, by comparing concordance of predicted binary cancer classification between replicate samples, giving results of 0.90 (0.95 CI: 0.764–0.959) for inter-run precision, 1.00 (0.95 CI: 0.796–1.000) for intra-run precision, 1.00 (0.95 CI: 0.871–1.000) for inter-operator precision, 0.96 (0.95 CI: 0.930–0.998) for inter-instrument precision and 0.83 (0.95 CI: 0.641–0.933) for inter-day precision. To determine LOB, we carried 8 non-template controls (water) through the entire assay and detected on average ∼0.02% unique aligned reads of a true sample on the same sequencing run. Finally, to assess tumor content LOD, we developed methodology that uses methylation signal to estimate the amount of tumor-derived DNA in each cfDNA sample and validated our estimates using whole exome sequencing, an orthogonal gold-standard approach. We then assessed the relationship between tumor content and classifier sensitivity using our training data of 625 cancer cfDNA samples and determined that our tumor content LOD whereby 95% true cancer samples were correctly predicted to be 0.037%. Conclusion Our assay shows high performance and high technical reproducibility. Our previously reported high sensitivity for stage 1 and stage 2 cancers, as well as extremely low tumor content LOD reported here supports our ability to perform early-stage multi-cancer detection, where the levels of circulating tumor DNA are low.

Abstract Here, we report that novel epidermal growth factor receptor (EGFR) gene fusions comprising the N-terminal of EGFR linked to various fusion partners, most commonly RAD51, are recurrent in lung cancer. We describe five patients with metastatic lung cancer whose tumors harbored EGFR fusions, four of whom were treated with EGFR tyrosine kinase inhibitors (TKI) with documented antitumor responses. In vitro, EGFR–RAD51 fusions are oncogenic and can be therapeutically targeted with available EGFR TKIs and therapeutic antibodies. These results support the dependence of EGFR-rearranged tumors on EGFR-mediated signaling and suggest several therapeutic strategies for patients whose tumors harbor this novel alteration. Significance: We report for the first time the identification and therapeutic targeting of EGFR C-terminal fusions in patients with lung cancer and document responses to the EGFR inhibitor erlotinib in 4 patients whose tumors harbored EGFR fusions. Findings from these studies will be immediately translatable to the clinic, as there are already several approved EGFR inhibitors. Cancer Discov; 6(6); 601–11. ©2016 AACR. See related commentary by Paik, p. 574. This article is highlighted in the In This Issue feature, p. 561

e15035 Background: Early cancer detection has the potential to significantly improve patient outcomes and reach the Cancer Moonshot goals of reducing the death rate from cancer by at least 50 percent over the next 25 years. Harbinger Health is pioneering early cancer detection with a blood-based test that combines recent genomic and epigenomic discoveries of early cancer and biology-informed artificial intelligence. Unlike prior approaches that are purely statistical to identify informative biomarkers, Harbinger’s approach is informed by insights into specific biological events early during tumorigenesis and is therefore optimized for detecting cancer in patients with very low levels of circulating tumor DNA. Methods: Here, we present data generated using the Harbinger Health assay from 1,046 subjects, 621 with newly diagnosed, treatment-naive cancer (15 different cancer types) and 425 individuals with no history, diagnosis, or cancer symptoms. Using this sample cohort, we developed and applied a rigorous framework for training, calibration and predicting likelihood of cancer using a multi-layered logistic regression-based machine learning algorithm, generating final outputs of binary classification (cancer yes/no). Our framework involved multiple iterations of 10-fold cross-validation of the full dataset, and we report solely on samples that appear in the held-out test set in each iteration. Additionally, we developed a similar framework to predict tissue of origin (TOO) for cancer samples. Results: The overall sensitivity of cancer detection was 82% (95% confidence interval (CI): 72.7-91.0%) at 95% specificity. Notably, the sensitivity was 74% (95% CI: 54.8-92.7%) for stage 1 and 84% (95% CI: 65.3-100%) for stage 2. Furthermore, we were able to correctly predict cancer in 95% of patient samples with at least 0.037% tumor fraction. The overall sensitivity for high incident cancers were: breast (73%), prostate (82%), lung (85%), and colorectal (96%) at 95% specificity and the overall accuracy of TOO prediction was 86% when the tumor fraction was greater than 0.1% in the top 3 most prevalent cancer types (breast, colorectal and lung). Importantly, we noted that technical variability was introduced when performing assay optimization strategies, yet we observed comparable performance when assessing subsets of stably processed samples, indicating that our performance is robust. Conclusions: Overall, these results demonstrate that the Harbinger Health platform, with its biology-informed approach, has the potential for highly sensitive multi-cancer diagnostic accuracy and specifically those with early-stage cancer. The platform is now being validated in a 10,000-subject prospective clinical trial (CORE-HH/NCT05435066).

Abstract Background: Non-colorectal MSI-H tumors are increasingly identified by CGP. Rare types such as MSI-H BC remain poorly defined with an evidence gap on how to optimally sequence or combine with standard of care treatment. MSI can be measured by either IHC, PCR, or CGP and can be caused by both sporadic and germline variants within different tumor types. Prior studies in BC have shown evidence of dMMR by IHC cases MSS based on PCR. This could be due to intra-tumor heterogeneity, specific microsatellite loci evaluated, or penetrance of germline, somatic, or epigenetic alterations. Published data suggests carriers of germline pathogenic MMR variants have a BC risk equivalent to the normal population and currently germline testing is recommended only for BRCA. Currently in advanced BC, standard tumor biomarker testing includes IHC, PCR, and FISH; however, with increasing use of CGP we demonstrate additional actionable biomarkers as well as potential germline variants in MSI-H BC. Methods: DNA was extracted and hybrid capture CGP was performed on 29,160 BC cases. TMB was determined on 0.8-1.2 Mb of DNA and MSI status on 95-114 loci. Genomic LOH was also evaluated. Comparative analysis was done with 101 MSI-H BC, 841 MSS BC and 4,988 non-breast MSI-H cancers. Histological subtype was obtained from the pathology along with orthogonal testing for ER/PR/HER2 status. Somatic-germline-zygosity (SGZ) status was predicted using a published research use algorithm. Select case reports with clinical outcomes will be presented. Results: We identified 101 (0.35% of total) MSI-H BC cases: 29 ER+/HER2-, 5 HER2+, 29 TNBC, and 28 unknown. Amongst BC cases with known subtype, TNBC was enriched for MSI-H vs MSS (53.4 vs 35.8%, p=0.005). The median TMB in MSI-H BC (26.1 mut/Mb, IQR 17.4;42.8) was significantly lower than that of MSI-H colon (46.1mut/MB) and higher than that of MSI-H uterine tumors (22.6mut/Mb) in our comparison group (p<0.001 for both, Kruskal-Wallis test). Pathogenic variants in an MMR gene were found in 61.4% of MSI-H BC with MLH1 loss being the most common (13.6%) and much higher vs. the non-breast MSI-H cohort (2.4%, p<0.0001). Germline mutations in MMR genes in BC are rare yet 5/52 MMR short variants identified in 101 MSI-H BCs were predicted to be germline, 34 somatic, and 13 could not be determined. We identified 21 MSI-H BC patients with a total of 25 pathogenic BRCA1/2 alterations of which 4 were likely germline, 10 were homozygous, and were enriched in TNBC. These were mainly frameshift mutations, including BRCA2 T3033fs* in 5/18 (28%) cases; however, 7/25 were deletions, rearrangements, or nonsense mutations. Median gLOH was significantly higher in BRCA altered (19.7%) compared to BRCA wild-type MSI-H BC cases (9.6%) (p=0.007, Wilcox test). Additional potentially targetable biomarkers included 26 CDx eligible PIK3CA mutations, 11 ERBB2 activating point mutations in the TKD or ECD domain, 1 FGFR2 rearrangement, and 6 AKT1 E17K mutations. Four cases also had concurrent (CD274) PD-L1 amplifications. Conclusion: MSI-H BC is rare but CGP can identify additional therapeutic options for rational combination with targeted therapies such as PI3K, PARP, and HER2 inhibitors. BRCA alterations may be of germline or somatic origin and they may be targetable, as demonstrated by gLOH, rather than passenger mutations. Further characterization of these tumors and comparison to both MSS BC and non-breast MSI-H tumor types, combined with treatment outcomes, can provide insights on rationale combinations and/or sequencing of therapeutic agents. Citation Format: Kimberly McGregor, Natalie Danzinger, Jeffrey S. Ross, Kyle Gowen, Alexa B. Schrock, Garrett M. Frampton, Dean C. Pavlick, Jan W. Davis, Carl R. Gray, Jeffrey M. Venstrom. Therapeutic considerations in microsatellite instability high (MSI- H) breast cancers (BC) identified by comprehensive genomic profiling (CGP) [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS5-04.

504 Background: Microsatellite instability (MSI) testing has become critically important in clinical cancer care of patients with cancer given the recent pan-tumor FDA approval of pembrolizumab for use in patients with MSI-High (MSI-H) tumors. We previously demonstrated the robustness of a novel proprietary algorithm for determination of MSI status via NGS from solid tumor biopsy specimens (J Clin Oncol 34, 2016 (suppl; abstr 1523)). Traditional MSI tests such as PCR or IHC are impractical for pan-tumor adoption, as MSI-H prevalence outside of gastrointestinal and endometrial cancers is usually < 1%. NGS-based ctDNA profiling provides an opportunity for both MSI and actionable alteration testing in patients in whom tissue-based biopsy is not available. Methods: Genomic DNA (gDNA) from five previously characterized MSI-H cell lines: (DLD1, 22Rv1, LNCap, RL952, CL188), and one MSS cell line (SCC9) was enzymatically-fragmented to simulate ctDNA and titrated to various dilution levels with DNA from a healthy hapmap subject (NA12878). Samples were screened with a 70-gene panel, FoundationOne Liquid, that includes 180 mononucleotide repeat sequences (8-26bp long in the human reference genome). Length variability in the 180 repeat loci was utilized to generate an overall MSI score via principal components analysis. The NGS based MSI algorithm was applied to all the samples. Results: Assessment of these six cell lines, targeting five dilution levels confirmed by SNP mixing ratios, show that our NGS based MSI test for liquid biopsies has 96% sensitivity at > 2% tumor fraction with 100% PPV. The regression intercept of the MSI-H dilution samples with the pre-established MSI-H calling threshold shows our method has a LOD of 1.03% tumor fraction. MSI-H prevalence data from liquid biopsies of gastrointestinal tumors obtained during clinical care will also be presented. Conclusions: These data demonstrate the feasibility of using NGS-based liquid biopsy assays for MSI testing. This ctDNA-based approach will allow for increased access to checkpoint inhibitors in a pan-tumor setting, which would be especially relevant for cancers where routine MSI testing is impractical or when tissue is not available.

Microsatellite instability (MSI) is an important biomarker for predicting response to immune checkpoint inhibitor therapy, as emphasized by the recent checkpoint inhibitor approval for MSI-high (MSI-H) solid tumors. Herein, we describe and validate a novel method for determining MSI status from a next-generation sequencing comprehensive genomic profiling assay using formalin-fixed, paraffin-embedded samples. This method is 97% (65/67) concordant with current standards, PCR and immunohistochemistry. We further apply this method to >67,000 patient tumor samples to identify genes and pathways that are enriched in MSI-stable or MSI-H tumor groups. Data show that although rare in tumors other than colorectal and endometrial carcinomas, MSI-H samples are present in many tumor types. Furthermore, the large sample set revealed that MSI-H tumors selectively share alterations in genes across multiple common pathways, including WNT, phosphatidylinositol 3-kinase, and NOTCH. Last, MSI is sufficient, but not necessary, for a tumor to have elevated tumor mutation burden. Therefore, MSI can be determined from comprehensive genomic profiling with high accuracy, allowing for efficient MSI-H detection across all tumor types, especially those in which routine use of immunohistochemistry or PCR-based assays would be impractical because of a rare incidence of MSI. MSI-H tumors are enriched in alterations in specific signaling pathways, providing a rationale for investigating directed immune checkpoint inhibitor therapies in combination with pathway-targeted therapies.

Chromosomal rearrangements involving ALK receptor tyrosine kinase gene (ALK) are detected in lung cancer and other malignancies. The most common ALK fusion partner in NSCLC is echinoderm microtubule associated protein like 4 gene (EML4). Multiple ALK fusion partners have been described.1Shaw A.T. Hsu P.P. Awad M.M. Engelman J.A. Tyrosine kinase gene rearrangements in epithelial malignancies.Nat Rev Cancer. 2013; 13: 772-787Crossref PubMed Scopus (237) Google Scholar Patients with alternate variants of ALK fusions may have differential responses to ALK tyrosine kinase inhibitors (TKIs).2Yoshida T. Oya Y. Tanaka K. et al.Differential crizotinib response duration among ALK fusion variants in ALK-positive non-small-cell lung cancer.J Clin Oncol. 2016; 34: 3383-3389Crossref PubMed Scopus (211) Google Scholar We discuss the initial report of a myelin transcription factor 1 like gene (MYT1L)-ALK fusion identified in the tumor of a patient with metastatic NSCLC, and the tumor's unique response to treatment. A 41-year-old male never-smoker presented with spinal cord compression. Pathologic examination of a sample from epidural tumor resection showed thyroid transcription factor 1–positive adenocarcinoma. Imaging demonstrated a left upper lobe mass with liver and bone metastases, indicating T2bNxM1b disease. Polymerase chain reaction genomic testing revealed EGFR and KRAS wild type. Fluorescence in situ hybridization demonstrated no evidence of ALK or ROS1 rearrangements. The patient received carboplatin and pemetrexed, which resulted in a reduction in his lung lesion, but progression of his liver metastases. Follow-up genomic profiling by hybrid capture–based next-generation sequencing (NGS) assay revealed a 27.5-Mbp duplication of chromosome 2 (ALK kinase domain included),3Frampton G.M. Fichtenholtz A. Otto G.A. et al.Development and validation of a clinical cancer genomic profiling test based on massively parallel DNA sequencing.Nat Biotechnol. 2013; 31: 1023-1031Crossref PubMed Scopus (1450) Google Scholar resulting in a novel MYT1L-ALK fusion (Fig. 1). This fusion retained exons 1 to 14 of MYT1L that, like other oncogenic ALK fusions, were fused to exons 20 to 29 of ALK, coding for the tyrosine kinase domain. The patient began taking crizotinib, which resulted in a 4-month response until his lung and liver tumors progressed and new brain metastases appeared. He received whole brain radiation and began nivolumab. The liver and lung tumors progressed, with stable brain metastases. The patient began taking ceritinib while continuing nivolumab for 2 months, followed by ceritinib alone for a total of 5 months. Brain magnetic resonance imaging showed a response, but lumbar and thoracic spine magnetic resonance imaging showed progression. Computed tomography of the patient's chest, abdomen, and pelvis showed new thickening of the atrial wall at the right lateral heart border and unchanged lung lesion and liver metastases. The patient began taking alectinib, and after 2 months, imaging demonstrated a partial response and disappearance of the right heart border abnormality. One month later, his brain, lung, and liver tumors progressed. He discontinued alectinib therapy and underwent a left frontal craniotomy for tumor resection. He did not receive additional therapy owing to clinical deterioration; additional biopsies and genomic testing were not feasible. During the patient's treatment course, his tumor had variable responses to multiple ALK TKIs (Fig. 2). In this case, NGS revealed a large duplication event in chromosome 2, resulting in a novel MYT1L-ALK fusion. Myelin transcription factor 1-like protein is located on the cytoplasmic surface of the Golgi membranes and the endoplasmic reticulum and is an important factor in the cell cycle (specifically, in checkpoint recovery).4Villeneuve J. Scarpa M. Ortega-Bellido M. Malhotra V. MEK1 inactivates Myt1 to regulate Golgi membrane fragmentation and mitotic entry in mammalian cells.EMBO J. 2013; 32: 72-85Crossref PubMed Scopus (28) Google Scholar Because ALK fusion partners may be a determinant of response to treatment and its possible future resistance type,5Lin J.J. Zhu V.W. Yoda S. et al.Impact of EML4-ALK variant on resistance mechanisms and clinical outcomes in ALK-positive lung cancer.J Clin Oncol. 2018; 36: 1199-1206Crossref PubMed Scopus (218) Google Scholar detection of these fusions is crucial. Fluorescence in situ hybridization testing for ALK failed to detect the fusion in either the primary lung cancer and the liver metastases. The ability of NGS to detect ALK fusions and their fusion partners may increase patients’ therapeutic options. Identifying novel events such as the MYT1L-ALK fusion may be important to predicting a tumor’s response to currently available ALK TKIs and matching patients to appropriate targeted therapies. “Teachable Moment” Interventions in Lung Cancer: Why Action MattersJournal of Thoracic OncologyVol. 13Issue 5PreviewThe commentary by Friedrich Stiefel, MD, and Céline Bourquin, PhD,1 published in the February issue of The Journal of Thoracic Oncology, has raised a serious concern related to evidence-based practice in the care of cancer patients. In brief, Stiefel and Bourquin, from the Psychiatric Liaison Service at Lausanne University Hospital, suggest that using a cancer diagnosis as a “teachable moment” to encourage a patient to quit smoking may be a misguided and detrimental intervention. The authors state, “to target this population with teachable moment interventions may … increase the perceived or enacted stigma associated with lung cancer” and argue that such an approach, towards a patient who continues to smoke after a diagnosis of cancer is “not sane.” Full-Text PDF Open Archive

Abstract Introduction: Increase in targeted therapies has resulted in the need for a single assay capable of detecting diverse biomarkers indicated for these agents. Comprehensive genomic profiling (CGP) provides such a solution, but due to the complexity and number of assays available today, standardization of validation has become critically important. We present FoundationOne CDx, the first NGS-based comprehensive companion diagnostics (CDx) platform developed and performed in compliance with FDA 21 CFR part 820. The assay interrogates 324 genes, and has CDx indications in five tumor types associated with 17 targeted therapies (Table 1). The versatile assay design will facilitate streamlined development of future CDx indications. Methods: DNA extracted from FFPE tumor tissue underwent whole-genome shotgun library construction and hybridization-based capture, followed by sequencing using Illumina HiSeq 4000. Sequence data were processed using a proprietary analysis pipeline designed to detect base substitutions, indels, copy number alterations, rearrangements, microsatellite instability (MSI), and tumor mutational burden (TMB). Results: Clinical validity was established such that the concordance between CGP and approved CDx were statistically non-inferior to that of two runs of approved CDx. For analytical validity, limit of detection (LoD) was at allele frequency 4% for known substitutions and indels. LoD was 16% tumor content for copy number amplifications, 30% for homozygous deletions, 11% for rearrangements, 12% for MSI, and 20% for TMB. Concordance with an orthogonal NGS platform was 94.6% for substitutions and indels. Within-assay reproducibility had PPA 99.4%. Conclusion: Rapid expansion of targeted therapies and CDx has necessitated a new approach and urgency to defining performance standards. We developed a comprehensive CDx assay and demonstrated clinical and analytical validity to support and accelerate using CGP for routine clinical care. Table 1. Companion Diagnostic IndicationsIndicationBiomarkerTherapyNon-small cell lung cancer (NSCLC)EGFR exon 19 deletions and EGFR exon 21 L858R alterationsafatinib, gefitinib, or erlotinibEGFR exon 20 T790M alterationsosimertinibALK rearrangementsalectinib, crizotinib, or ceritinibBRAF V600Edabrafenib in combination with trametinibMelanomaBRAF V600Edabrafenib, vemurafenibBRAF V600E and V600Ktrametinib, cobimetinib, in combination with vemurafenibBreast cancerERBB2 (HER2) amplificationtrastuzumab, ado-trastuzumab-emtansine, or pertuzumabColorectal cancerKRAS wild-type (absence of mutations in codons 12 and 13)cetuximabKRAS and NRAS wild-type (absence of mutations in exons 2, 3, and 4)panitumumabOvarian cancerBRCA1/2 alterationsrucaparib Citation Format: James X. Sun, Yali Li, Coren Milbury, Joel Skoletsky, Christine Burns, Wai-ki Yip, Jun Luo, Ninad Dewal, Adrienne Johnson, Kyle Gowen, Jing Tong, Yuting He, Jie He, Pei Ma, Jared White, Steve Roels, John Truesdell, Eric Peters, Houston Gilbert, Charlie Wu, Erica Schleifman, Johannes Noe, Carl Barrett, Kenneth Thress, Suzanne Jenkins, Julia Elvin, Geoff Otto, Doron Lipson, Jeffrey Ross, Vincent Miller, Philip Stephens, Michael Doherty, Christine Vietz. A clinically validated comprehensive companion diagnostic platform for care of patients with advanced cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4757.

Non-small cell lung cancer (NSCLC) has emerged as a paradigm for clinical application of precision medicine as optimal therapy is commonly chosen based on genomic biomarkers identified in a patient's tumor sample. Recurrent driver alterations are well described, however, a need to continually identify rare variants remains clinically relevant. We identified an incident case of advanced NSCLC with a PDGFR-α N848 K activation loop mutation with no other concurrent oncogenic drivers. Amino acid sequence alignment confirmed homology to the imatinib-sensitive KIT N822 K activation loop mutation observed in GIST. The patient achieved a 2-year response to single agent imatinib that is ongoing. While PDGFR-α N848 K is rare among public sequencing databases our cases strongly suggests functional relevance and highlights the importance of studying rare variants in NSCLC.

Introduction: In patients with advanced-stage ovarian cancer, FDA has recently granted approval of treatment with PARP inhibitors (PARPi) in patients harboring deleterious BRCA mutations (~25% of population). However, there is clear evidence of “BRCA-like” patients who respond to PARPi without BRCA mutations. To identify such patients, it has been shown that deficiency in homologous recombination repair leads to a common phenotype of genome-wide loss of heterozygosity (LOH). Depending on the cutoff employed, genomic LOH could identify more than twice the number of ovarian cancer patients who could benefit from PARPi than measuring BRCA alone. We present here an NGS-based platform developed and performed in compliance with FDA 21 CFR part 820. The assay provides a tumor measurement of BRCA1/2 (FDA-approved as FoundationFocus CDx BRCA) as well as genomic LOH, and is on the same platform as the comprehensive FoundationOne CDx, which interrogates 324 genes. Methods: DNA extracted from FFPE tumor tissue underwent whole-genome shotgun library construction and hybridization-based capture, followed by sequencing using Illumina HiSeq 4000. Sequence data were processed using a proprietary analysis pipeline designed to detect base substitutions, indels, copy number alterations (CNA), genomic rearrangements, microsatellite instability (MSI), and tumor mutational burden (TMB). A genome-wide LOH profile based on SNPs is measured as part of the CNA pipeline, and is summarized as the percentage of the tumor genome displaying LOH (scored from 0-100%), with ≥16% being considered LOH high based on clinical data derived from ARIEL2 Part 1, a phase II study of the PARPi rucaparib for the treatment of platinum-sensitive ovarian cancer (ARIEL2; NCT01891344) Results: For analytical validity, BRCA limit of detection (LoD) was at allele frequency 5.9% for substitutions and non-repetitive indels, and 30% tumor content for LOH. Overall percent agreement with comparator NGS assay was 97.3% for BRCA. No orthogonal platform concordance was established for LOH as no validated test exists. Within-assay reproducibility was measured with overall concordance of 100% for BRCA, and 98% for LOH. Conclusion: We developed a novel diagnostic assay (in compliance with FDA 21 CFR part 820) that can measure BRCA and genomic LOH simultaneously, and established robust analytical validation data. Citation Format: James X. Sun, Kevin Lin, Yali Li, Kyle Gowen, Yuting He, Coren Milbury, Christine Burns, Jun Luo, Steve Roels, Murtaza Mehdi, John Truesdell, Pei Ma, Lakshman Ramamurthy, Christine Vietz, Jeri Beltman, Thomas Harding, Doron Lipson, Jeffrey Ross, Vincent Miller, Philip Stephens, Michael Doherty, Julia Elvin. A validated diagnostic assay for identifying ovarian cancer patients with deleterious BRCA mutations and high genomic loss of heterozygosity (LOH) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4544.