This article is more than 5 years old. Information may no longer be current.
Circulating tumor DNA predicts pancreatic cancer survival
Circulating tumor DNA isolated from blood samples in patients with pancreatic adenocarcinoma independently predicted survival, according to prospective study results.
“Circulating tumor DNA (ctDNA) is an attractive biomarker that could help us to define or mange our treatments in the future,” Jean-Baptiste Bachet, MD, PhD, of the gastroenterology and digestive oncology department at Sorbonne University and Centre Universitaire des Saints-Peres in Paris, told HemOnc Today.
The prognosis of patients with pancreatic adenocarcinoma remains poor despite therapeutic advances.
To determine whether ctDNA offered prognostic value in pancreatic adenocarcinoma, Bachet and colleagues collected blood samples from 135 patients with resectable (n = 31), locally advanced (n = 36) or metastatic disease (n = 68).
Researchers used next-generation sequencing to target low-allele frequency mutations, and they used picoliter droplet digital polymerase chain reaction to screen samples for several types of mutations, including the three most frequent KRAS mutations in pancreatic adenocarcinoma.
Fifty of the 104 patients (48%) with advanced disease had detectable ctDNA, with a median mutation allelic frequency of 6.1% (range, 0.1-65.4).
Median follow-up for this group was 34.2 months.
Univariate analysis showed a strong correlation between the presence of ctDNA and shorter median OS (6.5 months vs. 19 months; P < .001). Multivariate analysis that accounted for age, sex and disease stage showed an independent association between ctDNA and poor OS (HR = 1.96; 95% CI, 1.2=-3.2).
When researchers analyzed patients with advanced disease based on mutation frequency in ctDNA, they observed a significant dose–response relationship with OS (18.9 months for the lowest tertile, 7.8 months for those in the middle tertile, and 4.9 months for those in the highest tertile; P < .001).
Six of the 31 patients with resectable disease had detectable ctDNA. After median follow-up of 33 months, 23 experienced disease recurrence and 13 had died. Researchers reported longer median DFS (17.6 months vs. 4.6 months) and longer median OS (32.2 months vs. 19.3 months) among patients with no detectable ctDNA.
“Our results are important because they will help future development and assessment of such biomarkers in clinical practice, [including] differences in rate of positive ctDNA according to stage of the disease, grade of differentiation, prognostic value [and] number of mutations found through next-generation sequencing,” Bachet said.
The results should be confirmed in prospective clinical trials to “better assess the predictive value of this biomarker in light of the dynamic biological changes that occur during treatment,” Bachet added. – by Melinda Stevens
Disclosures: Bachet reports advisory board and consultant roles with Amgen, Celgene and Merck Serono.
Perspective
Back to Top
Pietrasz and colleagues report results of a thorough evaluation of circulating tumor DNA (ctDNA) — sometimes referred to as cell-free DNA — as a biomarker in pancreatic cancer. The analysis included 135 patients who initiated chemotherapy. Investigators analyzed plasma samples using polymerase chain reaction–based identification of KRAS mutations.
Results yielded three key findings:
- ctDNA is detectable in 17% to 19% of cases with resected or locally advanced pancreatic cancer, and in 65% of metastatic disease cases;
- Pretreatment levels correlate with OS in locally advanced or metastatic disease; and
- Cases of resected cancer with detectable ctDNA had worse survival compared with cases with undetectable ctDNA.
These data are encouraging in the sense that ctDNA has the potential to be a biomarker in a disease for which none exists. CA19.9 often is used, but its reliability and clinical utility are limited. ASCO guidelines for pancreatic cancer offer best-practice recommendations for its use in various settings. Therefore, there is great need for a reliable biomarker — especially for potentially curable cases — to help identify patients at high risk for relapse to target them with aggressive systemic therapies instead of upfront surgery, and to spare patients from likely ineffective surgery if biomarker response is not seen. In the incurable setting, a biomarker can be used to assess response prior to RECIST criteria imaging changes, for example, allowing early therapeutic maneuvering.
Getting there will require more work. ctDNA needs to tested prospectively. Standardized methods to detect and sequence DNA, along with consensus target mutation definitions, are required. Data from multiple cohorts to establish uniformly applicable levels and cutoffs will be needed. As the study by Pietrasz and colleagues demonstrated, correlation with — and ultimately superiority to — CA19.9 and imaging findings will have to be established to allow clinical utility. A limitation highlighted by this work is the low prevalence in the curative setting. Perhaps absence of ctDNA can be a biomarker?
This and similar contemporary reports have demonstrated our ability to find ctDNA, and provide early signals of it being a potential biomarker. We need to demonstrate clinical applicability and utility. This study has taken a good first step in that journey.
Khorana AA, et al. J Clin Oncol. 2016;doi:10.1200/JCO.2016.67.5553.
Riva F, et al. Mol Oncol. 2016;doi:10.1016/j.molonc.2016.01.006.
Sausen M, et al. Nat Commun. 2015;doi:10.1038/ncomms8686.
Sohal DP, et al. J Clin Oncol. 2016;doi:10.1200/JCO.2016.67.1412.