Allelic Ratio of KRAS Mutations in Pancreatic Cancer

Abstract

KRAS is not simply mutated or wild type in pancreatic ductal adenocarcinoma (PDAC); actually, it has never been. More than 90% of PDACs carry mutated KRAS alleles; however, the impact on PDAC biology may vary with the tumor-specific allelic ratio and dosage of mutated KRAS.

Despite therapeutic advances, pancreatic ductal adenocarcinoma (PDAC) remains the deadliest cancer. As beautifully outlined in the recent review by Chiu et al. [1], genomic medicine has delivered tremendous molecular pieces; however, solving the PDAC puzzle remains difficult. To exemplify the daunting complexity, we explored the frequency of somatic KRAS mutations in PDAC and came to a conclusion with practical relevance.

As pointed out by Chiu et al. [1], numerous studies using conventional sequencing technologies reported that ∼70%–80% of PDAC samples carry an activating KRAS mutation [2, 3]; however, more recent data indicate a frequency of ∼93% [4]. The discrepancy is caused by different sensitivities in the sequencing technologies. Although conventional approaches can detect ∼10% mutant alleles (depending on preanalytical variables and the applied technology) [5], next-generation high-coverage sequencing (i.e., deep sequencing) can potentially detect mutant alleles far below 5% [6]. At first glance, the more sensitive next-generation data simply replace the legacy data [4]; however, the statement that >90% of PDACs harbor KRAS mutations [4] does not take into account that some of these cases harbor only a small fraction of mutant KRAS alleles (e.g., allelic ratio of mutant to wild-type allele of 1.2%). Consequently, the statement is a generalization that implies a shared KRAS biology without considering the underlying reasons for the spectrum of allelic ratios.

One readily apparent (and microscopically visible) reason for variation in allelic ratios is the presence of nontumor cells in tumor samples. PDAC nearly always contains stromal and inflammatory components that introduce a variable fraction of wild-type alleles. We accounted for tumor purity by calculating the corrected allelic ratio (equal to allelic ratio divided by cellularity [7]) (Fig. 1A). Notably, despite correction for purity, allelic ratios span the full range from wild type to ≥100% mutant alleles. The apparent spectrum has several implications because both the lower and higher ends of the spectrum suggests heterogeneity within the cancer population (clonality) or variations in DNA content (ploidy), respectively. With respect to cases with low corrected allelic ratios, namely, those revealed by deep sequencing, it is tempting to speculate that the functional consequences of mutant KRAS at 1.2% are different from those of tumors that harbor >80% mutant KRAS alleles. It is too early to gauge the full biological and clinical impact of low-frequency mutations; however, overall survival in the subgroups is not identical (Fig. 1B). Specifically, in the data set from Biankin et al. [4], we noted a trend for PDAC with a corrected allelic ratio of ≥10% to be associated with shorter overall survival. Interestingly, several studies using conventional sequencing technologies with detection thresholds of ∼10% mutated alleles have previously shown that the presence of a KRAS mutation is a marker of poor prognosis (briefly reviewed by Boeck et al. [8]). Viewed in conjunction, these data indicate that KRAS mutated tumors, which compose ∼93% of all PDAC cases, are heterogeneous and that prior data should not simply be disregarded in light of newer, more sensitive technologies. Moreover, these considerations demonstrate that the same mutated gene in the same type of cancer may have different biological, prognostic, and—with ongoing trials assessing the efficacy of novel KRAS inhibitors [1, 2] —possibly therapeutic implications [1]. The mutation status is only one element of a more difficult equation revealed by new sequencing technologies. Computational methods for precise determination of tumor content can overcome limitations of histology-based approaches and take at least purity, clonality, and ploidy into account [9]. With these tools readily available for implementation in routine molecular diagnostics, we propose that the tumor-specific allelic ratio of somatically mutated genes should be an integral component of a comprehensive molecular diagnostic report for personalized cancer care [1, 10].

Figure 1.

KRAS mutations in PDAC: allelic ratio and overall survival. (A): Mutation frequencies differ between deep sequencing studies and traditional Sanger sequencing due to different detection thresholds. Samples are plotted by cellularity-corrected allelic ratios and span the entire spectrum from homozygous wild-type to ≥100% mutated KRAS. Deep sequencing identified 10 wild-type cases (∼7%), whereas application of a traditional Sanger sequencing threshold revealed that 17 cases (∼12%) would have been called wild type. Thus one cannot directly compare KRAS mutation frequencies between Sanger and deep sequencing studies. The asterisk (∗) indicates cases with copy number gain in which the number of mutant KRAS alleles reached ratios of >100% when compared with the normal biallelic situation. (B): In the data set from Biankin et al. [4], overall survival in cases with low allelic ratios (i.e., Sanger threshold in Figure 1A) was not identical to those with higher allelic ratios of KRAS mutations. Although the difference did not reach statistical significance (p = .10, univariate log-rank test), the presence of KRAS mutations as a marker of poor prognosis has been shown in previous studies using conventional sequencing methods [3, 8]. Consequently, >90% of PDAC cases harbor somatic KRAS mutations, and low allelic ratios may have different clinical implications.

Abbreviation: PDAC, pancreatic ductal adenocarcinoma.

In conclusion, KRAS is not simply mutated or wild type in PDAC; actually, it has never been. More than 90% of PDACs carry mutated KRAS alleles; however, the impact on PDAC biology may vary with the tumor-specific allelic ratio and dosage of mutated KRAS.

Disclosures

The authors indicated no financial relationships.