Longitudinal Copy-Number Alteration Analysis in Plasma Cell-Free DNA of Neuroendocrine Neoplasms is a Novel Specific Biomarker for Diagnosis, Prognosis, and Follow-up

Abstract

Purpose: As noninvasive biomarkers are an important unmet need for neuroendocrine neoplasms (NEN), biomarker potential of genome-wide molecular profiling of plasma cell-free DNA (cfDNA) was prospectively studied in patients with NEN.

Experimental design: Longitudinal plasma samples were collected from patients with well-differentiated, metastatic gastroenteropancreatic and lung NEN. cfDNA was subjected to shallow whole-genome sequencing to detect genome-wide copy-number alterations (CNA) and estimate circulating tumor DNA (ctDNA) fraction, and correlated to clinicopathologic and survival data. To differentiate pancreatic NENs (PNEN) from pancreatic adenocarcinomas (PAAD) using liquid biopsies, a classification model was trained using tissue-based CNAs and validated in cfDNA.

Results: One hundred and ninety-five cfDNA samples from 43 patients with NEN were compared with healthy control cfDNA (N = 100). Plasma samples from patients with PNEN (N = 21) were used for comparison with publicly available PNEN tissue (N = 98), PAAD tissue (N = 109), and PAAD cfDNA (N = 96). Thirty percent of the NEN cfDNA samples contained ctDNA and 44% of the patients had at least one ctDNA-positive (ctDNA⁺) sample. CNAs detected in cfDNA were highly specific for NENs and the classification model could distinguish PAAD and PNEN cfDNA samples with a sensitivity, specificity, and AUC of 62%, 86%, and 79%, respectively. ctDNA-positivity was associated with higher World Health Organization (WHO) grade, primary tumor location, and higher chromogranin A and neuron-specific enolase values. Overall survival was significantly worse for ctDNA⁺ patients and increased ctDNA fractions were associated with poorer progression-free survival.

Conclusions: Sequential genome-wide profiling of plasma cfDNA is a novel, noninvasive biomarker with high specificity for diagnosis, prognosis, and follow-up in metastatic NENs.

Figure 1.

OS analysis for ctDNA⁺ (blue curve) and ctDNA⁻ (red curve) patients with significance values, in the whole cohort (A) and the subset of PNENs (B). NR, not reached.

Figure 2.

Frequency plots for detected CNAs in cfDNA of patients with PNEN (N = 13; A), and tumor tissue of patients with PNEN (N = 98; B). Gains are indicated in green and losses in red.

Figure 3.

Evolution of CNA profiles determined by ichorCNA between sample at inclusion (top) and last sample (bottom) for patient P01 (A) and P05 (B). Samples from P01 were taken 8 months apart and from P05 3 weeks apart.

Figure 4.

Illustration of the effect of tumor fraction on PFS probability (red curve, with 95% CI in blue) for P01, according to our joint model, whereby all tumor fraction measurements (*) after treatment initiation were consecutively added (A–D). When a higher tumor fraction value was added to the measurements, the survival probability decreased.

Figure 5.

Tumor fraction evolution during disease course (black curve) of patients P06 (A), P17 (B), and P04 (C). All patients started with everolimus and SSA treatment on day 0. In addition, the DSUM was indicated (gray area), as well as investigator-assessed imaging results (triangle) and treatment breaks or switches (line or box). PD, progressive disease.