Abstract
215 Background: Lung cancer (LC) is the top cause of cancer-associated mortality worldwide, with a 10-year overall survival rate of 5%. Although most LCs are smoking related, in the US, 25% of non-small cell LC (NSCLC) are diagnosed in persons with little or no smoking history. Fusions involving anaplastic lymphoma kinase (ALK) are the oncogenic driver in ~3–7% of NSCLC. While inhibitors targeting the kinase domain of ALK (TKIs) have proven extremely effective, inevitably, resistance develops with limited treatment options beyond second line TKIs. Additionally, NSCLCs without identified molecular alterations have even more limited tumor specific treatment options. Methods: We developed a precision medicine-based platform (PMP) to screen patient-derived, minimally cultured, organoid material (PDM) directly from resections with curated panels of drugs. PDM collected during clinically indicated procedures is plated in 3D-culture to generate patient-derived organoids (PDOs) and screened with drugs tailored to each tumor type. PDOs are screened at therapeutically relevant doses, drawing from pharmacokinetic data for each drug. We have optimized an assay to rapidly screen for EML4-ALK fusions and can perform next-generation sequencing in real time (~7 days) to integrate with drug screening results. Our organoid cultures retain the full spectrum of tumor microenvironment present in the original sample. Results: To date, we have analyzed 83 cases, including 8 EML4-ALK NSCLC. We have demonstrated the ability to produce high quality data from low input samples (biopsies). In one EML4-ALK NSCLC we were able to collect PDM from two distinct anatomic spaces (pleural effusion and peritoneal fluid) and screen with the same panel of drugs, with nearly identical results, highlighting the reproducibility and consistency of our assay. Screening of EML4-ALK tumors which have progressed to second or higher line TKIs, demonstrate sensitivity to earlier generation ALK TKIs, a known phenomenon. Characterization of tumors with unknown clinical drivers identifies that ~1/3 tumors which have no actionable or hypothetically prioritized variants, and they exhibit particularly poor response to chemotherapies. Our results recapitulate known resistance/progression in samples previously exposed to therapy, demonstrating a strong negative predictive value. Longitudinal assessment is being tracked to robustly assess positive predictive value (PPV). Conclusions: Our PMP captures robust and reproducible results that are consistent with known clinical pathogenesis. Moving forward, we are collecting longitudinal data from enrolled patients in parallel with clinical trials to demonstrate the PPV of our PMP. We additionally strive to demonstrate reproducibility to obtain Clinical Laboratory Improvement Amendments approval and to deliver results to patients and physicians to help guide clinical care.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call