Search

103 Background: Hormone receptor–positive breast cancer (HR+ BC) is the most common cause of BC-related death. Over half of metastatic recurrences occur ≥ 5 years (y) from diagnosis. Detection of minimal residual disease (MRD) via circulating tumor DNA (ctDNA) can identify cancer recurrence months to years in advance and may be an important tool to guide therapy. Little is known about ctDNA in the late adjuvant setting. We investigated ctDNA dynamics and clinical outcomes in pts ≥ 5 y from diagnosis of high-risk early-stage HR+ BC. Methods: Patients with high-risk HR+ BC (T3-4 or N2-3 or T1 with 3+ lymph nodes, or T2N1 [and Oncotype RS ≥ 26; grade 3; or Ki-67 ≥ 20%]) with no evidence of recurrence 5 y after diagnosis were prospectively identified and consented. Plasma samples were collected at time of consent and at routine visits every 6-12 mos. Whole-exome sequencing (WES) was performed on primary tumor tissue to identify somatic mutations and design for each patient a RaDaR assay, a tumor-informed liquid biopsy test to detect plasma ctDNA. Per current practice standards, pts did not undergo regular surveillance imaging. All pts were followed for development of local and/or distant metastatic recurrence, as determined by their clinical provider. Results: Of 103 pts enrolled, 85 had sufficient tumor tissue, and 83 pts had successful WES. Personalized RaDaR assays were designed targeting 12-51 variants (median, 36), and used to test 219 plasma samples from 83 pts. The number of plasma samples per patient ranged from 1-7 (median, 2). 57 pts (68.7%) had stage 3 disease, and most (75, 90.4%) received curative-intent chemotherapy. All pts received endocrine therapy (ET). 39 (47%) remained on adjuvant ET at time of last follow up. Of 44 pts who completed adjuvant ET, 41 (93.2%) received > 5 y of treatment. Time from diagnosis to first sample ranged from 4.9-20 y (median, 8.4 y). Median (range) follow up was 10.2 (6.7-22.3) y from diagnosis and 1.8 (0-3.6) y from first sample. 5 pts (6%) developed distant metastatic recurrence and 2 pts (2.4%) had locoregional recurrence. 4/83 (5%) pts were MRD+ at study entry and 8/83 (10%) pts were MRD+ at any time point. 5/5 (100%) pts with metastatic recurrence were MRD+, with ctDNA lead times up to 37.6 mos. ctDNA was detected at tumor fractions of 0.0027-26.84% (median, 0.396%). 2/8 (25%) MRD+ pts had not had clinical recurrence at latest follow up, one with no follow up since detection and one 15.4 mos from ctDNA detection (with ctDNA levels 0.045% and 26.84%). Conclusions: Here we report—to our knowledge—the first data on ctDNA detection in late adjuvant HR+ BC. 10% of pts had MRD at ≥ 5 y from diagnosis. ctDNA analysis identified MRD in all cases of distant recurrence. ctDNA was detected in 2 pts who have not yet experienced recurrence. Longer follow up is necessary for these pts at high risk. Additional studies will determine if ctDNA-guided intervention can alter clinical outcomes.

9579 Background: High-risk melanoma has variable prognosis. Adjuvant immuno- (IO) and targeted therapy (TT) are approved for stage III-IV resected disease. However, a significant proportion of patients (pts) are cured by local treatment alone or relapse despite adjuvant therapy. Liquid biopsy with ctDNA assays have been used to predict response to treatment and identify pts at higher risk of progression/death. Personalized ctDNA assays are a highly sensitive approach that may enhance upfront risk stratification and early detection of relapse. Methods: Serial ctDNA Monitoring as a predictive Biomarker in advanced neoplAsms (SAMBA) is a Princess Margaret prospective ctDNA kinetics study (NCT03702309) in high-risk melanoma pts. Plasma is collected pre-op (pre-local treatment, if feasible), post-op (after surgery), and every 3-6 months (m) until radiological progressive disease (rPD). Personalized amplicon based NGS assays by Inivata (RaDaR) were used to detect somatic variants in ctDNA identified through whole-exome sequencing of matched tumor tissue. Progression free survival (PFS) and overall survival (OS) from the time of surgery were estimated with the Kaplan Meier and compared with the log-rank test. Results: As of December 2021, 82 of 100 planned pts have been enrolled. A total of 191 samples from 47 pts have been analyzed. Median age was 66 years (27-87), 33 were male (70%). Seven (15%), 30 (64%) and 10 (21%) were stage II/III/IV respectively. All pts had surgery and 8 (17%) adjuvant radiation. No systemic therapy was given to 11 pts (23%); 30 (64%) had IO and 6 (13%) TT. rPD occurred in 13 pts (28%). Median follow up was 24 months. A median of 48 variants were included in the personalized ctDNA panel design (35-52). ctDNA was detected (ctDNA+) at any time point in 12/47 pts (26%), of which 5/12 (42%) were BRAF and NRAS wt on tissue. Median PFS was 4.9 months (m) for ctDNA+ pts and not reached (NR) for ctDNA- pts at post-op (HR = 2.71 CI 0.60-12.31, p = 0.179). Median OS was 23.1 m vs NR in ctDNA+ vs ctDNA- pts (HR = 8.9, CI 1.45-54.77, p = 0.004). Two ctDNA+ pts had neoadjuvant IO and became ctDNA- before surgery. One, free of disease after 12 m, had ctDNA- in 4 follow up samples. The other pt was ctDNA+ in the post-op sample and relapsed within 3 m. Four of 45 (9%) pts had ctDNA+ at post-op. Two of them, including a pt who had neoadjuvant IO, did not receive adjuvant therapy and had rPD within 3 m. The other 2 pts received adjuvant IO; ctDNA cleared and pts remain free of disease at 12 and 34 m. Three pts with rising ctDNA over time experienced rPD after a median of 4 m (2-7). Conclusions: Personalized ctDNA analysis with RaDaR may improve risk of death stratification and selection of pts who could benefit from adjuvant treatment. Detection of ctDNA may precede rPD. Follow-up will continue in pts with rising ctDNA who have not yet had rPD. Pts accrual and sample collection are ongoing, and additional data will be presented. Clinical trial information: NCT03702309.

3039 Background: Molecular profiling of tumor tissue is the gold standard for treatment decision making in advanced non-small cell lung cancer. Results may be delayed or unavailable due to insufficient tissue samples or prolonged wait times for biopsy, pathology assessment and testing. We piloted the use of plasma molecular testing as part of the initial diagnostic work-up for patients with suspected advanced lung cancer (NCT04863924). Methods: Patients with radiologic evidence of advanced lung cancer referred to the lung rapid diagnostic program underwent plasma circulating tumor DNA (ctDNA) testing using InVisionFirst-Lung, a next-generation sequencing (NGS) assay targeting 37 genes. Standard tissue testing was performed with comprehensive NGS (Oncomine). The primary endpoint was time to treatment in stage IV NSCLC patients compared to an historical pre-COVID-19 cohort (2018-9). Secondary endpoints included actionable targets identified in plasma, % of patients starting targeted therapy based on liquid biopsy and result turnaround time (TAT). Results: Between July 1 to December 31, 2021, 60 patients were enrolled. Median age was 70 years (range 33-91), 52% were female, 57% Caucasian, 48% never smokers. Of these, 73% had NSCLC, 12% small cell, 10% non-lung pathology and 5% declined tissue biopsy. Of 44 NSCLC patients, 5 (11%) had early-stage disease and underwent curative therapy. Most stage IV patients (79%) had systemic treatment. Median time to treatment initiation in the study cohort was 34 days (n = 31, range 10-90) versus 62 days (n = 101, range 13-159) in the historical cohort (p

6052 Background: Immuno-oncology agents (IO) have become standard-of-care in the treatment of R/M HNSCC, but only a subset of patients (pts) benefit. Highly sensitive quantification of plasma circulating tumor DNA (ctDNA) may permit real time assessment of disease under selective pressures of treatment. Methods: R/M HNSCC pts treated with platinum-based chemotherapy (CT) or IO (anti-PD1/L1 +/- second IO) underwent serial ctDNA collection pre-cycles 1/2/3 and at disease progression, corresponding to timepoints (T) 1-4. T1 was considered baseline. Whole exome sequencing of pt tumor tissue identified patient specific somatic variants which were used as targets for RaDaR, a personalized multiplexed PCR-based NGS assay. Matched buffy coat DNA was sequenced to filter germline mutations and identify confounding CHIP. RaDaR was applied at each available T, an estimated variant allele frequency (eVAF) was calculated and correlated with progression free- (PFS) and overall- survival (OS). Findings were compared against prior (ESMO 2021) data generated using a fixed 580 gene CAPP-seq (CAncer Personalized Profiling by deep Sequencing) panel designed specifically against squamous cell carcinoma. Results: A total 114 plasma samples from 38 pts were analyzed. Of 35 pts with ctDNA detected at T1 and/or T2, 26 received IO and 9 CT. Median age was 62 (20-84), 77% were male, 69% prior smokers and 26% HPV positive. Median PFS and OS, for all 35 pts, was 2.57 mo (95% CI 0.48-4.66) and 8.37 mo (95% CI 5.42-11.32) respectively. For IO treated pts, median PFS was 2.45 mo (95% CI 0-5.18) and median OS 7.38 mo (95% CI 3.84-10.93). RaDaR panels targeted a median 48 variants (17-50). ctDNA was detected in 35/38 (92%) patients at baseline, with median eVAF 0.345% (range 0.0004% - 43.37%). ctDNA abundance at baseline did not correlate with PFS or OS. A decrease in Δ eVAF from T1 to T2, by > 30%, or > 50% identified pts with improved PFS, with HR 0.45 (0.21, 0.96) p = 0.04, 0.31 (0.14, 0.70) p < 0.01, and 0.23 (0.10, 0.56) p < 0.01, respectively. Similar results were observed for the 26 IO pts, with HR 0.40 (0.16, 1.03) p = 0.06, 0.19 (0.05, 0.66) p < 0.01, and 0.06 (0.01, 0.47) p < 0.01, respectively. A similar, but non-significant, trend was seen in median OS for pts with a decrease vs. increase in Δ eVAF (T1 to 2), 8.8 mo vs 7.3 mo (HR = 0.87 (0.42, 1.79)). For 31 pts, a comparison of Δ ctDNA levels, based on personalized RaDaR vs. CAPP-seq assays, from T1 to T2 demonstrated a correlation coefficient of R = 0.57, P < 0.01. Conclusions: In pts with R/M HNSCC, a decrease in ctDNA eVAF after first treatment correlated with improved PFS. There was a significant correlation between fixed CAPP-seq and personalized RaDaR assays when comparing Δ in ctDNA levels. Clinical Trial: NCT03712566.

8517 Background: Liquid biopsies based on circulating tumor DNA (ctDNA) analysis are being investigated for detection of residual disease and recurrence. Conclusive evidence for utility of ctDNA in early-stage non-small cell lung cancer (NSCLC) is awaited. Due to low ctDNA levels in early-stage disease or post-treatment, effective methods require high analytical sensitivity to detect mutant allele fractions (MAF) below 0.01%. Methods: We analysed 363 plasma samples from 88 patients with NSCLC recruited to the LUng cancer CIrculating tumour DNA (LUCID) study, with disease stage I (49%), II (28%) and III (23%). 62% were adenocarcinomas. Plasma was collected before and after treatment, and at 3, 6 and 9 months after surgery (N = 69) or chemoradiotherapy (N = 19). Additional plasma was collected at disease relapse for 17 patients. Median follow-up was 3 years, and 40 patients progressed or died of any cause. We employed the RaDaR™ assay, a highly sensitive personalized assay using deep sequencing of up to 48 tumor-specific variants. Variants identified by tumor exome analysis were tested by deep sequencing of tumor tissue and buffy coat DNA to verify somatic mutations and exclude clonal hematopoiesis. The RaDaR assay demonstrated 90% sensitivity at 0.001% MAF in analytical validation studies. Results: ctDNA was detected in 26% of samples, at median MAF of 0.047% (range: 0.0007% to > 2%), and prior to treatment in 87%, 77% and 24% for disease stage III, II and I respectively. For 62 patients, plasma was collected at a landmark timepoint, between 2 weeks and 4 months after initial treatment. ctDNA detection at the landmark timepoint was strongly predictive of clinical disease relapse, with Hazard Ratio of 20.7 (CI: 7.7-55.5, p-value < 0.0001). All 11 cases with ctDNA detected at landmark had disease progression, a median of 121 days after detection, and these included all 8 patients that relapsed within 300 days of treatment. Across 27 patients whose disease progressed during the study, ctDNA was detected at any timepoint post-treatment in 17 cases, with a median lead time of 203 days, and up to 741 days prior to clinical progression. ctDNA was detected post-treatment, in 13 of the 15 patients that progressed and had ctDNA detected prior to treatment. Conclusions: Our results support an emerging paradigm shift, by demonstrating that liquid biopsies can reliably detect recurrence of NSCLC at a preclinical stage, many months before clinical progression, thereby offering the opportunity for earlier therapeutic intervention. Clinical trial information: NCT04153526.

e15560 Background: Detection of residual circulating tumour DNA (ctDNA) in patient plasma following curative intervention for localized non-small cell lung cancer (NSCLC) could identify patients who are at higher risk of relapse. These patients may benefit from adjuvant treatment, even if they have no macroscopic disease identified by radiographic imaging, which is the current standard of care. Here we evaluate the performance of the Inivata personalized sequencing assays to detect ctDNA in a cohort of 90 patients with early-stage NSCLC undergoing treatment with curative intent. Methods: The Inivata assay uses a highly sensitive next-generation sequencing platform, to identify tumor-specific variants from exome sequencing of tumor tissue and to track up to 48 patient-specific mutations in plasma specimens by multiplex PCR and ultra-high-depth next-generation sequencing. Samples from 90 patients with Stage I-III NSCLC who underwent radical treatment with curative intent, either surgery or radiotherapy ± chemotherapy, were collected as part of the LUng cancer - CIrculating tumor DNA (LUCID) study. Results: 350 plasma samples from 90 patients were analyzed using the Inivata assay, including samples collected before and after treatment and at subsequent follow-up visits. ctDNA was detected in pre-treatment samples in 38% of 32 patients (12/32) with Stage I NSCLC and in 90% of 21 patients (19/21) with Stage II/III disease, at allele fractions ranging from 6 parts per million (ppm, equivalent to 0.0006%) to over 20,000 ppm (equivalent to 2%). In plasma samples collected post-treatment, ctDNA was detected in close to 50% of cases. Conclusions: These findings highlight the Inivata assay is a sensitive method for detection of residual ctDNA and recurrence in early stage NSCLC. Initial detection rates ranged from 38% in Stage I disease to 90% for patients with Stage II/III disease prior to treatment, including detection of ctDNA to levels as low as a few parts per million. ctDNA was detected in at least one post-treatment timepoint in close to 50% patients. Together with additional data to be presented from the full 90 patient cohort, this suggests a possible route to improving treatment and designs of adjuvant trials for early stage NSCLC by detection of residual disease post-treatment and monitoring for early detection of relapse.

2015 Background: Genomic DNA profiles are mandatory in advanced, treatment naive non-small cell lung cancer (NSCLC) patients (pts) and strongly recommended at progression (PD) on personalized treatment. In pts with PD limited to central nervous system (CNS), tissue biopsy is difficult and the performance of ctDNA is unknown. Methods: Clinical, molecular, imaging data of NSCLC pts included in 2 prospective studies from 01.2016 to 11.2018 at Gustave Roussy were collected. Inclusion criteria were: stage IV disease and any known tissue genomic alteration (GA) EGFR, ALK, BRAF, KRAS, HER2, ROS1, MET, PIK3CA, TP53. Plasma ctDNA collected at baseline/PD were analyzed by next-generation sequencing (NGS-InVisionFirst™-Lung) in 3 groups: pts with isolated CNS (iCNS), extra-CNS only (noCNS) or both combined (cCNS) disease. iCNS was defined as any PD to CNS, while stable/absent extra-CNS metastases (mts). ctDNA was considered positive if ≥1 GA was found. ctDNA in cerebrospinal fluid (CSF) were also collected. Results: Out of 245 pts with ≥1 ctDNA: 56 had iCNS (66 samples), 97 noCNS (127 samples) and 92 cCNS (107 samples). In this cohort, 60% were female, median age 60 years, 47% smokers; 92% had adenocarcinoma. The median number of mts sites was 3 in noCNS/cCNS groups. Proportions of tissue GA at baseline were (iCNS vs noCNS/cCNS): EGFR (50% vs 44%), ALK (30% vs 11%), BRAF (4% vs 12%), KRAS (5% vs 15%), HER2 (2% vs 5%), ROS1 (5% vs 4%). Tyrosine kinase inhibitors were used in 73% iCNS vs 61% noCNS/cCNS. Local brain treatments were performed in 43% (n = 24) vs 32% (n = 29) and leptomeningeal mts (LM) detected in 34% (n = 19) vs 8% (n = 9), in iCNS and cCNS, respectively. CtDNA was positive (+) in 52% in iCNS vs 84% in noCNS and 92% in cCNS (p < 0.0001). In iCNS, there was a non-significantly higher proportion of + ctDNA in pts with LM vs only brain disease (59 vs 48%, P = 0.44). 12/56 pts of iCNS group had serial ctDNA, being collected also at time of cCNS. In 25% of cases, a negative ctDNA at time of iCNS shifted to + at time of cCNS. In 12 iCNS pts, ctDNA was + in 6 (50%) plasma and in 10 (83%) paired CSF (p = 0.193). Conclusions: Detection of GA by plasma ctDNA is lower in NSCLC pts with isolated CNS PD. Alternative strategies (as CSF analysis) should be explored.

3055 Background: Acquired ALK resistance mutations (mut.) are the main mechanism of tyrosine kinase inhibitor (TKI) resistance (30-50%). While next-generation TKIs are more active on mut. than earlier TKIs, compound ALK resistance are associated with failure to next-generation TKIs. We evaluated the clinical utility of detecting ALK resistance mutations in blood to predict TKI efficacy. Methods: ALK positive advanced NSCLC pts were prospectively enrolled between Oct. 2015 and Aug. 2018 in 8 French institutions. Prospective samples were collected; ctDNA was analyzed by amplicon-based Inivata InVisionFirst-Lung. Results: A total of 101 pts with advanced ALK positive NSCLC were enrolled and 328 samples collected. In samples collected at TKI failure (N=74), we detected 9 single and 7 complex (≥2) ALK resistance mut. (22%), associated with EML4-ALK variant 3 (38%) vs. variant 2 (13%) vs. variant 1 (none); 30% had other somatic mut. (mainly TP53 and KRAS, PI3KCA, MET, etc.). No mutations were detected in 48% of samples (ctDNA neg). ALK mut. were more frequent after 2nd/3rd generation TKI (43% post-lorlatinib (7), 29% post-2nd gen. (31), 11% post-crizotinib (36)). ALKG1202R was the most common, as single (n=3) or complex mut. (n=4). The median overall survival (mOS) was 100.4 mo. (95% CI 41.9-158.9) and the median progression free-survival (mPFS) to subsequent line was 2.8 mo. (0.7-4.9). Patients with ctDNA neg had mOS of 105 mo. (39.3-172.1) vs. 58.5 mo. (33.1-84.0) if ≥1 ALK mut. vs. 44.1 mo. (20.0-68.2) if others ( P=0.001). Pts with the complex ALK mut. had worse OS compared to singles ALK mut. (mOS 26.9 mo. vs. 58.5 mo., P=0.001); ALK complex mut. were associated with poor efficacy to subsequent therapy (PFS 6 mo.) vs. single mut., with longer PFS (PFS >6 mo. in 56%). Detectable ALKG1202R mut. were associated with shorter median OS (58.3 mo.; 7.9-109.1) vs. overall population; 86% of cases developed rapid PD (PFS 6mo.). Conclusions: The absence of ctDNA mutations at TKI failure was associated with prolonged OS, whereas complex ALK mutations at TKI failure may predict resistance to subsequent therapy. Larger and specifically designed studies should be performed to validate these findings.

9095Background: Circulating tumor DNA (ctDNA), is a surrogate material for somatic mutation (mut) detection, such as EGFR in NSCLC patients (pts). However, the applicability for the detection of AL...

e21031Background: Therapeutic response in ALK+ NSCLC is driven by heterogeneous mechanisms of resistance. Liquid biopsy (LB) offers a non-invasive approach to study the tumor genomic landscape unde...

e23060 Background: The presence of a targetable driver mutation in nearly 50% non-small cell lung cancer (NSCLC) patients has enabled tailoring therapy regimens to improve survival. Serial repeat biopsies can offer an instrumental indication into the longitudinal evolution of cancer. However, tissue biopsies are invasive and can provide insufficient material for molecular testing. Mutation detection in plasma DNA as a “liquid biopsy” has been suggested as non-invasive approach to monitor tumor dynamics over time. Methods: We established an institutional protocol (NCT01511288) for the collection of liquid biopsies from stage IIIB/IV NSCLC patients either untreated, under therapy or progressive on therapy with a TKI. This protocol has included 134 NSCLC patients for whom clinical, pathological and genomic information is collected prospectively. Analyses were performed by Inivata using InVision (enhanced tagged-amplicon sequencing). Results: So far, samples from 50 patients have been analyzed. InVision allowed the detection of driver mutations in 20 plasma samples obtained at diagnosis. Tissue was unavailable for molecular analysis in 8/20 samples. We observed a concordance rate in mached plasma and tissue samples, of 92,3% (n = 12). In plasma samples from patients that relapsed under erlotinib or gefitinib we evidenced the EGFR T790M mutation in 57% of patients, with a concordance rate of 90,9%. Interestingly, analysis of serial samples collected from 3 patients under EGFR-targeted therapy showed the emergence of an EGFR T790M mutation 11 weeks before the radiographic confirmation of progression (P1); differential dynamics in the allelic fractions of mutated clones that reflected the pattern of dissociated tumor response to treatment (P2) and the presence of concomitant EGFR activating and T790M mutations, together with an EGFR C797G, BRAF V600E and KRAS G12D in a patient who progressed under osimertinib (P3). Conclusions: Our preliminary results provide further evidence on the use of liquid biopsies for monitoring disease response, resistance to treatment and tumor heterogeneity. Subsequently, we will evaluate the utility of liquid biopsies in the clinical setting to understand the dynamics of mutant clones over time.

11532 Background: Molecular profiling is limited by access to sufficient tumor tissue for comprehensive analysis and due to tumor heterogeneity, the complete range of tumor DNA abnormalities may not be represented nor accurately reflect the clinical evolution of disease. Circulating tumor DNA (ctDNA) can be used as a minimally-invasive liquid biopsy for the detection, quantification and monitoring of molecular abnormalities for personalized treatment strategies. Methods: In a prospectively designed program, to date, we have recruited 227 advanced NSCLC patients having received prior therapy, with unknown molecular profile at time of blood collection. Blood collections (10ml K2-EDTA) were performed to assess molecular profile prior to or at time of relapse. Repeat samples were performed on patients initiated on treatment and followed for up to 18months. Patient samples were analyzed with InVision (enhanced tagged-amplicon sequencing) using a 34 gene panel. Interim analysis performed with full descriptive summary statistical analyses to be presented at conference. Results: ctDNA profiling detected somatic mutations in 182pts (80.2%), predominantly located in TP53 (46%), EGFR (28%), KRAS (11%) and STK11 (7%, half of which had concurrent KRAS). Of note, clinically actionable mutations were detected: T790M (25pts, median 1.4% AF), ERBB2 (8 pts), MET (8pts) and BRAF (4pts) providing eligibility for new therapy options. 20pts including 12 EGFR/T790M+ve were evaluated for ctDNA monitoring up to 18 months (median 10m); correlation between dynamic change in mutation allele fraction and clinical response was observed, especially predictive of relapse to treatment. 10 patients demonstrated SD/PR response to osimertinib treatment with T790M detection at low allele fraction (7/10 < 1% AF with 1pt at 0.08% AF). Conclusions: ctDNA can be used as a non-invasive ‘liquid biopsy’ for molecular profiling of NSCLC patients to detect clinically relevant and actionable mutations when tissue biopsy is unavailable. Liquid biopsies can be repeated as needed where tissue is not feasible, providing real-time information to support personalised treatment.

11540 Background: Molecular profiling is limited by tumour heterogeneity and access to sufficient tissue for comprehensive analysis. Circulating tumour DNA (ctDNA) can be used as a minimally-invasive liquid biopsy for mutation detection, quantification and monitoring for personalised treatment strategies. Methods: We recruited 110 patients into a prospectively-designed study for Stage III/IV NSCLC patients intended to initiate 1stline platinum-based chemotherapy. Blood collections (10ml K2-EDTA) were performed prior to treatment and analysed by InVision (enhanced tagged-amplicon sequencing) using a 34-gene panel. Tissue biopsies, when available, were analysed by NGS (Ion-Torrent, Sanger) for concordance analysis. To evaluate correlation with outcome, repeat blood collections were performed in selected patients. Results: 110 NSCLC pts were included (61% male, 14% never-smoker, and 70% adenocarcinoma). ctDNA profiling detected mutations in 83 pts (79%). TP53 (44%), KRAS (17%), STK11 (18%; 11/19 with KRAS/STK11) and EGFR (10%) were the commonest abnormalities detected. Additionally, MET (6%), ERBB2 (6%), PIKC3A (6%) and BRAF (4%) mutations and EGFR, MET, ERBB2 amplifications were detected in 2% of patients, respectively. 20% of the mutations detected in ctDNA were observed at < 0.5% allele fraction, with 6% between 0.03%-0.25% AF. Tissue was available in 44 pts; somatic mutations were detected in 73%. Tissue & liquid concordance was 92.3%. 10 pts (23%) reported as tissue negative had a positive liquid biopsy. 33 advanced NSCLC patients were evaluated for longitudinal serial ctDNA monitoring up to cycle 4 of chemotherapy; the ratio of mutated molecules between D1 and D42 was significantly correlated with change in RECIST 1.1 measurement at D42 (p-value = 0.002625 CI 95% 0.298, 0.875). Conclusions: ctDNA can be used as a ‘liquid biopsy’ for molecular profiling of NSCLC patients to detect clinically relevant and actionable mutations when tissue biopsy is unavailable. Liquid biopsies can be used longitudinally and may provide an early surrogate for response evaluation by radiographic RECIST.

11533Background: Effective and non-invasive methods for actionable molecular alterations in NSCLC patients (pts) are urgently needed. Molecular profiling is often limited by access to sufficient tu...