High Levels of KRAS in ctDNA of Pancreatic Cancer May Mean Lower OS

January 21, 2015
Dennis Bittner, PhD

During a retrospective study of 182 patients with nonresectable pancreatic cancer, a negative association was found between baseline circulating tumor (ct)DNA KRAS counts in plasma and overall survival (OS), which indicates that patients with lower KRAS burden in ctDNA survive longer.

During a retrospective study of 182 patients with nonresectable pancreatic cancer, a negative association was found between baseline circulating tumor (ct)DNAKRAScounts in plasma and overall survival (OS), which indicates that patients with lowerKRASburden in ctDNA survive longer.1

The study was performed jointly by the University of Copenhagen, the University of California at San Francisco, and Trovagene, a US company based in San Diego, California. The source of samples for the study was plasma collected previously from patients in the Danish BIOPAC study with locally advanced or metastatic, nonresectable pancreatic cancer undergoing palliative treatment with gemcitabine or FOLFIRINOX.

Of the 182 patients, 176 had evaluable baseline samples, and 143 (79%) had more than 1 copy of mutantKRAS. The amount ofKRASctDNA in the 176 baseline samples was compared with length of patient survival to explore the potential clinical utility of using ctDNAKRASburden to predict patient outcomes.

In addition to baseline samples, longitudinal samples were also collected before cycle 2 of chemotherapy, and subsequent samples were collected every 2 to 3 months until death. A total of 640 longitudinal samples were analyzed at Trovagene using a quantitative KRAS ctDNA assay consisting of PCR based enrichment of mutant amplicons followed by deep DNA sequencing. Of the total archived samples, 617 were evaluable.

A Cox proportional hazards model used to analyze theKRASmutational load for association with overall survival (OS) revealed a statistically-significant negative association between baselineKRAScounts and OS, using either log-transformed or rank-transformedKRAScounts (P<.0001). Patients could be categorized into 3 groups of baselineKRAScounts based on median OS rates; those with zero baselineKRAScounts (n = 58), those with baselineKRAScounts >0 but ≤300 (n = 65), and those with counts >300 (n = 53;P<.0001).

The hazard ratio (HR) was 1.624 (95% CI, 1.097-2.404) between the ≤300 and >300 groups. In terms of longitudinal effects observed across the 3 groups, theKRASmutation burden remained low with treatment in the zero KRAS baseline group, while KRAS burden initially decreased in patients with ≤300 KRAS count at baseline, then later rose with treatment. For patients in the >300 KRAS count group, KRAS burden remained high with treatment. Longitudinal monitoring of KRAS ctDNA burden over time indicated that patients with longest survival were usually associated with a continued low copy number of ctDNAKRASmutations over time.

Trovagene presented data demonstrating single-copy sensitivity for their assay in spike/dilution experiments using a separate dataset, DNA blends from cell lines, and experimental design based on a Poisson distribution, and stated a lower limit of detection (LLOD) of 0.0055% mutant DNA in a background of wild-type DNA.

The assay uses a 31-basepair blocking oligonucleotide and selectively-designed binding kinetics to enrich and amplify mutant DNA fragments while suppressing wild-type sequence amplification. Barcoded adapter primers are added for compatibility with deep next-generation sequencing. Following sequencing, a proprietary analysis algorithm quantitates the level of mutant DNA, with results reported in number of copies detected per 105 genome equivalents (GE).

“We have developed an assay for detection of KRAS codon 12/13 mutations in highly-fragmented ctDNA that is based on mutation enrichment with PCR and next-generation sequencing,” Vlada Melnikova, MD, PhD, vice president of research & development, Trovagene, said, “This highly sensitive assay can detect 1 mutant DNA molecule in over 18,000 wild-type copies.”

Regarding the potential of the assay for assisting the treatment of pancreatic cancer, Melnikova said, “Stratification of nonresectable pancreatic cancer patients at baseline based on measurement of their ctDNAKRASmutation burden, as was done in this study, may be clinically useful for prognosis. The fact that the present research uses samples archived for up to 6 years also demonstrates the ability of the assay to monitorKRASctDNA burden over time.”


  1. Johansen JS, Vibat CR, Calatayud D, et al. Comparative circulating tumor DNA levels for KRAS mutations in non-resectable pancreatic cancer patients. Presented at: 2015 ASCO Gastrointestinal Cancers Symposium, January 15-17, San Francisco, CA. Session 288.


Back to the Gastrointestinal Cancers Symposium conference page