No abstract

Background: Tumor cell-free DNA (cfDNA) has the potential to provide minimally invasive patient specific biomarkers to monitor tumor burden. Gains and losses of chromosomal regions - as a hallmark of cancer - have been detected in plasma as copy number aberrations (CNAs), and for several cancers a relation to tumor size has been reported. Longitudinal observations during anti-cancer therapy have been mostly anecdotal. We measured CNAs changes during treatment by computing a genomic copy number instability index (CNI) of cfDNA to evaluate its potential to predict cytotoxic chemotherapy (chemo) response. Methods: 24 patients (pts) with advanced esophageal cancer (EC; n = 2), colorectal cancer (CRC; n = 3), non-Hodgkin lymphoma (NHL; n = 3), pancreatic ductal adenocarcinomas (PDAC; n = 4), and non-small cell lung cancer (NSCLC; n = 12) were included and assessable for response. DNA was extracted from pre-treatment plasma samples at baseline and sequentially for up to six cycles of chemotherapy. Copy-numbers were called from shotgun sequencing (Illumina) after mapping and quality filtering reads were counted in 5.5MBp windows (sliding) yielding a read coverage of 24,000-fold per bin. The read counts were transformed into log2 ratios and converted into a score based on Gaussian transformations. Concentration of total cfDNA was determined by digital PCR. Treatment response by imaging was recorded by RECIST 1.1 or EORTC PET/CT criteria. For pts with baseline CNI>40 (representing the 99.99% level for non-cancer healthy controls), CNI change was considered predictive of response or stable disease when there was a reduction of CNI ≥50% relative to baseline. Disease progression was demonstrated when: 1) CNI was ≥50% over nadir and <80% reduction relative to baseline; OR, 2) CNI was ≥5 fold the nadir; OR, 3) CNI demonstrated an absolute increase by 1,000 above baseline CNI. For pts with baseline CNI≤40, CNI change was considered predictive of response or stable disease when CNI remained ≤40, and PD when CNI≥2X above the baseline CNI. CNI was an early predictor of response if these results were observed by ∼3-8 weeks prior to imaging or a major change in pt's clinical status. Results: CNI was measured in 124 samples. Median baseline cfDNA was 7,238 cp/mL (range 2,319-91,978) and CNI was 171 (range 26-10,170). In general, NHL and tumors with high lymph node disease burden had highest baseline cfDNA and CNI, whereas overall the CNI was not correlated to the cfDNA content. 22 of 24 pts (91.7%) had CNI changes that were predictive of treatment response. For at least 15 pts, CNI change predicted response ∼3-8 weeks prior to scan results demonstrating response or PD. Conclusions: CNI change may serve as predictor (potentially early predictor) of therapeutic response to standard chemo for the investigated cancer types. Highest baseline cfDNA and CNI appear to be present in lymph node predominant disease, suggesting more readily shed DNA into the circulation. Citation Format: Glen J. Weiss, Julia Beck, Donald P. Braun, Kirsten Bornemann-Kolatzki, Heather Barilla, Rhiannon Cubello, Ashish Sangal, Robert P. Whitehead, Madappa Kundranda, Vivek Khemka, Howard B. Urnovitz, Ekkehard Schutz. Changes in tumor cell-free DNA copy number instability (CNI) predict therapeutic response in metastatic cancers. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3138.

No abstract

e14524 Background: Humoral tumor markers are used clinically for real-time assessment of therapeutic efficacy. In pancreatic ductal adenocarcinoma (PDAC) the predominant marker is CA19-9, which is not expressed by 10 to 30% of patients depending on race. We compared plasma cell-free DNA (cfDNA) copy number based assay with changes in serum CA19-9 levels and radiological responses to predict responses to systemic therapy. Methods: In a laboratory blinded, prospective multicenter pilot study, 40 non-resectable PDAC patients, treated with (m)FOLFIRINOX, CAPIRI, or gemcitabine +/- nab-paclitaxel) are currently enrolled. CA19-9 was determined in the local center’s laboratory. Tumor cfDNA was measured with a copy-number instability (CNI) scoring assay, determined by next generation sequencing in a centralized laboratory. The CNI score assesses the amount of cfDNA with somatic macro-alterations originating from malignant neoplasms. The difference of the values before commencing therapy (baseline) and prior to cycle 2 (either rising or falling) was calculated as a predictor of standardized radiological evaluation of chemotherapeutic efficacy. Results: 37 patients (3 drop-outs) had data for baseline and cycle 2, of which CA19-9 was elevated and evaluable in 29 patients. The direction from baseline to cycle 2 of CA19-9 and CNI scores were in agreement in 18/29 patients. 9 of 11 cases with discordant CNI score and CA19-9 had treatment response data, and CNI correlated with 7/9 (78%); in contrast 7/9 had rising CA19-9, when response was stable disease or better (22% concordance). In the 27 patients with available imaging, CNI predicted better (n = 18) than CA19-9 (n = 10) (p = 0.03 Fisher’s exact). Conclusions: This comparative study on cfDNA versus CA19-9 suggest that cfDNA CNI quantitation is a potentially more reliable blood based marker for early real-time assessment of efficacy in systemic PDAC therapy than CA19-9, compared to standard of care imaging. The better prediction after the first cycle might be due to the very short in vivo half-life of cfDNA ( < 1hr) compared to about one week for CA19-9. These results justify a larger prospective validation trial.

6055 Background: Copy number instability (CNI) signatures of cancers can be readily detected by Next Generation Sequencing of plasma cell-free DNA (cfDNA). HPV detected in oropharyngeal carcinomas is currently the only prognostic biomarker available. We report here CNI scores for disease monitoring of Head and Neck Cancers (HNC) with potential predictive value for personalized therapeutic options. Methods: A total of 132 plasma samples were collected from 54 HNC patients under informed consent and IRB approval. cfDNA was extracted from plasma, ~20M paired-end NGS mappable reads (reference: HG19) per sample were generated and CNI scores were calculated by read counting statistics. After unblinding CNI scores were evaluated as diagnostic parameter for association with disease characteristics and progression. Survival analysis was conducted after dichotomization of baseline CNI scores at a value of 31 corresponding to the 97.5thpercentile of a normal reference group (RG, n = 141). Results: CNI scores above the 97.5th RG percentile were detected in 40 out of 54 (74%) treatment naïve baseline samples. 29 patients with tumors ≤ T3 (62%, n = 42) and 11 out of 12 (92%) with T4 tumors had CNI scores > 31, with significantly higher CNI scores (p = 0.04) seen for T4 tumors. Higher CNI scores were also found in patients with tumor lymph node invasion (n = 37; median: 381, Q25-Q75: 57-1573) compared to negative lymph nodes (pN0, n = 17; 27, 19-64, p= 0.0004). A steep decline of CNI scores was detected after surgical resection, with increasing CNI scores in later disease progression. The pre-operative CNI scores were a stronger predictor of time to recurrence (p = 7*10-5) than the pN status (p = 0.05) (Cox regression). High baseline CNIs ( > 31) strongly correlated with time to recurrence (Kaplan-Meier log-rank p = 0.018) with a median of 20 months and median overall survival of 30 months in the high CNI group, neither reached in the low CNI-score group (60 m follow-up). Conclusions: Chromosomal instability within HNC was quantified from plasma cfDNA as CNI score. The CNI score may serve as better predictor for the time to recurrence interval than pN status. The data suggests that cfDNA analysis as CNI score may serve as real-time marker of treatment efficacy and outcome.