Impact of KRAS Mutations on Clinical Outcomes in Pancreatic Cancer Patients Treated with First-line Gemcitabine-Based Chemotherapy

Pancreatic cancer is the eighth most common malignancy in Korea, and its incidence has increased over the past decade (1). Given the absence of an efficient therapeutic modality, most patients with metastatic pancreatic cancer suffer from debilitating symptoms and have an extremely poor prognosis (2). Although systemic chemotherapy is ineffective, gemcitabine has shown clinical benefits and has therefore become the standard chemotherapy for advanced pancreatic cancer (3, 4). During the past few years, the efficacy of combining other drugs with gemcitabine has been tested in a large number of phase III trials, but none of the combination therapies were superior to gemcitabine alone (5, 6). Erlotinib, an oral reversible inhibitor of epidermal growth factor receptor (EGFR) tyrosine kinase, was the first drug to improve survival and progression-free survival in combined therapy compared with gemcitabine alone in a large randomized trial (7). Erlotinib has thus become an important therapeutic option in addition to gemcitabine for advanced pancreatic cancer patients, and much effort has been directed toward identification of clinical and molecular markers to predict response and prolonged survival for patients treated with this agent (8, 9).

Ras signaling pathways are commonly activated in tumors and are involved in mediating the downstream effects of EGFR activation. Among human cancers, pancreatic cancer shows the highest frequency of KRAS gene mutations (8, 10). Given the evidence that KRAS mutations are associated with less efficient EGFR-directed targeted therapy, EGFR inhibitors are probably of limited use in pancreatic cancer patients harboring KRAS mutation, similar to observations in colorectal and non–small-cell lung cancers (NSCLC; refs. 11–14).

The aim of this study was therefore to understand the clinical significance of KRAS mutations for pancreatic cancer patients treated with gemcitabine-based first-line chemotherapy and investigate a predictive role of KRAS mutations especially in patients treated with gemcitabine/erlotinib combination.

A total of 136 recurrent or advanced (locally advanced/metastatic) pancreatic ductal adenocarcinoma patients with available formalin-fixed paraffin-embedded tumor blocks from 2003 to 2009 were included in the analysis. All patients received first-line gemcitabine-based chemotherapy. A complete set of clinical data including sex, age at diagnosis, smoking history, staging, treatment, and vital status were obtained from the medical records of each patient. All pathologic specimens were cut from formalin-fixed paraffin-embedded tumor blocks from surgical specimens (47 patients) or biopsy specimens (metastatic sites or pancreas, 89 patients), and hematoxylin and eosin–stained sections were reviewed by one pathologist (K.T. Jang). DNA was prepared from these sections following dissection of the tumor resulting in at least 50% tumor cell content. All patients had ductal adenocarcinoma. The grades of differentiation were categorized as well differentiated, moderately differentiated, poorly differentiated, and undifferentiated. The guidelines of the Institutional Review Board about human subjects were followed.

Mutation detection was carried out by well-established methods of direct DNA sequencing of KRAS exons 2 and 3 (8). DNA was extracted from 5 paraffin sections of 10-μm thickness containing a representative portion of each tumor block, using the QIAamp DNA Mini Kit (Qiagen). DNA (100 ng) was amplified in a 20-μL reaction solution containing 2 μL of 10× buffer (Roche), 1.7 to 2.5 mmol/L of MgCl₂, 0.3 μmol/L of each primer pair, 250 μmol/L of deoxynucleotide triphosphate, and 2.5 units of DNA polymerase (Roche). Amplifications were carried out by a 5-minute initial denaturation at 94°C, followed by 30 cycles of 1 minute at 94°C, 1 minute at 55°C, and 1 minute at 72°C, and a 10-minute final extension at 72°C. PCR products were then purified from a 2% gel with a QIAquick gel extraction kit (Qiagen). DNA templates were processed for DNA sequencing with ABI-PRISM BigDye Terminator (version 3.1; Applied Biosystems), with both forward and reverse sequence-specific primers. Twenty nanograms of purified PCR products was used in a 20-μL sequencing reaction solution containing 8 μL of BigDye Terminator v3.1 and 0.1 μmol/L of the same PCR primer. Sequencing reactions were carried out by 25 cycles of 10 seconds at 96°C, 5 seconds at 50°C, and 4 minutes at 60°C. Sequencing data were generated with the ABI PRISM 3100 DNA Analyzer (Applied Biosystems) and analyzed with sequencing software (Applied Biosystems) to compare variations. Primers used for KRAS were as follows: codons 12 and 13, forward: 5′-tta tgt gtg aca tgt tct aat, reverse: 5′-aga atg gtc ctg cac cag taa; codon 61, forward: 5′-tca agt cct ttg ccc att tt, reverse: 5′-tgc atg gca tta gca aag ac. In cases of wild-type KRAS, we repeated sequencing with different primers (codons 12 and 13, forward: 5′-ggt gga gta ttt gat agt gta tta acc, reverse: 5′-aga atg gtc ctg cac cag taa; codon 61, forward: 5′-cca gac tgt gtt tct ccc ttc, reverse: 5′-tgc atg gca tta gca aag ac) and KRAS qPCR kit (Qiagen).

Categorical variables were statistically analyzed with Fisher's exact test. The response rate and disease control rate (percentage of response plus stable disease) were calculated. Overall survival was measured from the start of chemotherapy and was analyzed by Kaplan–Meier estimation and log-rank testing. The Cox proportional hazards model was used for multivariate analysis to assess the independent effects of KRAS mutations and to obtain their HR estimates. P < 0.05 values were considered significant.

Most patients were male (72.8%) and about one-third of the patients had never smoked (61.8%). One hundred twelve patients (82.4%) had locally advanced or metastatic pancreatic ductal adenocarcinoma at diagnosis, and 47 patients (34.6%) had recurrent pancreatic cancer following curative resection. Twenty-two patients (16.2%) received gemcitabine alone, and 114 patients received gemcitabine combined with either erlotinib (51.5%) or fluoropyrimidines (32.4%). Patient characteristics are summarized in Table 1.

In this study, tumor specimens from 71 of 136 pancreatic adenocarcinomas (52.2%) were found to have a point mutation of KRAS in exon 2 or 3 (Table 1). Most KRAS mutations occurred at codon 12. Observed point mutations at codon 12 in the order of frequency were: GGT-GAT (35G>A; 41 of 71 patients, 57.7%); GGT-GTT (35G>T; 19 of 71 patients, 26.8%); GGT-CGT (34G>C; 7 of 71 patients, 10.0%); and GGT-TGT (34G>T; 3 of 71 patients, 4%). There were no mutations in codon 13, and only 1 mutation was detected in codon 61. KRAS mutations were not associated with clinical and pathologic parameters such as age, sex, Eastern Cooperative Oncology Group performance status, smoking status, tumor location, histologic differentiation, tumor-node-metastasis staging, and type of first-line chemotherapy, but patients with KRAS mutations had a high level of CA19-9.

The overall response rate of first-line gemcitabine-based chemotherapy in all patients was 18.4%, with 4 complete responses and 21 partial responses. The disease control rate (responses plus stable disease) was 43.4%. Of the 71 patients who harbored KRAS mutation, 8 patients (11.3%) achieved objective response without complete response, and disease control rate was 35.2%. In contrast, of the 65 patients with wild-type KRAS, 17 patients (26.2%) showed better objective response (4 complete responses and 13 partial responses; P = 0.02) and the disease control rate was 52.3% (P = 0.06; Table 2). Among the patients treated with erlotinib, the response rate was 23.3% in the wild-type KRAS group and 7.5% in the mutant KRAS group (P = 0.09). Among the patients treated without erlotinib, the response rate was 28.0% in the wild-type KRAS group and 16.1% in the mutant KRAS group (P = 0.26).

KRAS mutation also adversely influenced survival of pancreatic cancer patients [KRAS mutation, 5.8 months (95% CI: 5.1–6.5) vs. wild-type KRAS, 8.0 months (95% CI: 5.8–10.2); P = 0.001; Fig. 1). These observations were confirmed by multivariate analysis to assess the independent effects of mutation on overall survival. Compared with patients with KRAS mutations, the KRAS wild-type group had a significantly lower risk of death (P = 0.001; HR = 0.523, 95% CI: 0.355–0.770). Multivariate analysis also revealed good prognostic factors for overall survival: moderately differentiated histology (P < 0.001; HR = 0.437, 95% CI: 0.301–0.634); locally advanced disease (P < 0.001; HR = 0.417, 95% CI: 0.255–0.681); and response to first-line chemotherapy (P = 0.003; HR = 0.482, 95% CI: 0.297–0.780; Table 3).

To clarify the role of KRAS mutation as a predictive biomarker, we further analyzed clinical outcomes according to the KRAS status for patients who received erlotinib (Table 4). Among the patients who received gemcitabine/erlotinib (P = 0.002), the median survival for patients with KRAS mutations was 5.2 months (95% CI: 3.65–6.75), whereas the median survival for patients with wild-type KRAS was 9.7 months (95% CI: 7.55–11.8). However, KRAS mutation status was not associated with clinical outcomes in patients who underwent chemotherapy with regimens other than gemcitabine/erlotinib [P = 0.21; KRAS mutation, 7.0 months (95% CI: 5.27–8.73) vs. wild-type KRAS, 7.0 months (95% CI: 5.61–8.39)]. These results showed that the observed survival advantage is derived only from the subgroup of patients treated with gemcitabine/erlotinib and not from all patents according to the KRAS mutation (Fig. 1).

A recent National Cancer Institute of Canada Clinical Trials Group (NCIC-CTG) phase III trial showed a definite survival benefit of gemcitabine plus erlotinib treatment as compared with gemcitabine alone. However, molecular biomarkers have not been successfully identified in these cohorts because of the small number of tissue specimens (181 specimens from 569 patients, or 31.8%; refs. 7, 9). The predictive and prognostic values of EGFR mutation for sensitivity to gefitinib or erlotinib are well recognized in NSCLC, but the frequency of EGFR mutation in pancreatic cancer is very low and it is unclear whether EGFR mutation is associated with sensitivity to EGFR tyrosine kinase inhibitors (8, 15). We previously analyzed the prognostic value of EGFR gene copy number, given the fact that it was an independent predictor of gefitinib sensitivity and survival in lung cancer (16, 17). However, EGFR gene copy number had a limited role as a prognostic biomarker in pancreatic cancer patients, which has been confirmed in the recent molecular biomarker study in the NCIC-CTG trial (8, 9). Because the Ras-mediated signaling pathway lies downstream of EGFR, it has been widely studied for the predictive and prognostic values of KRAS mutations, which may be present in about 35% to 45% of patients with colorectal cancer and NSCLC and have emerged as an important predictive marker of resistance to monoclonal antibodies or tyrosine kinase inhibitors targeting EGFR. In recent preclinical study using a direct xenograft pancreatic cancer model, global EGFR pathway activation predicted only response to EGFR inhibitors but the KRAS mutations were not predictive for response to EGFR inhibitors (18). In the molecular subset analysis of the patients from the NCIC-CTG trial, 78.6% (92 of 117) of patients had a KRAS mutation, yet it was not identified as a prognostic marker in patients treated with gemcitabine/erlotinib. In contrast, KRAS mutation was shown to adversely influence clinical outcomes of pancreatic cancer patients who received gemcitabine plus erlotinib in our study, despite a relatively low frequency of KRAS mutation. Furthermore, patients with a KRAS mutation had a worse overall survival when they received a gemcitabine/erlotinib combination. A recent German study also showed that 123 of 204 tumors (70%) harbored a somatic KRAS mutation and wild-type KRAS had an improved overall survival in pancreatic cancer patients who were treated with erlotinib combined with either gemcitabine or capecitabine (wild-type KRAS 8.0 months vs. mutant KRAS 6.6 months; HR = 1.62, P = 0.011), which is comparable with our study (19).

Despite its retrospective nature, our study clearly showed the unfavorable outcomes of gemcitabine plus erlotinib combination chemotherapy in patients with KRAS mutations compared with patients with wild-type KRAS receiving gemcitabine plus erlotinib combination chemotherapy, which is comparable with the prospective molecular analysis of the recent German trial (19). These results from the 2 studies may suggest that patients with KRAS mutations might not be good candidates for gemcitabine/erlotinib combination treatment, given that the results for response rate and median survival of the patients treated with gemcitabine/erlotinib combination were inferior. On the basis of our data, we might hypothesize that KRAS is a predictive biomarker in erlotinib-treated patients rather than a prognostic marker in the overall population (Table 4 and Fig. 1). It seems that the observed survival advantage is derived from the subgroup of patients treated with gemcitabine/erlotinib, whereas no survival difference based on KRAS mutation status is obvious in the other subgroup of patients treated without erlotinib. Therefore, the use of KRAS mutations as a predictive biomarker for treatment with combination of gemcitabine/erlotinib would be the first major step toward individualized treatment of patients with advanced pancreatic cancer. However, tissue availability is a potential limitation of the current retrospective analysis, which is a critical issue in pancreatic cancer research. In this study, about 35% patients underwent previous pancreatectomy, reflecting the availability and adequacy of surgical specimens for biomarker analysis. Therefore, the possible selection bias of current study may make definitive conclusions difficult and influence the prognostic and predictive results for KRAS status, which needs to be interpreted with caution and to be prospectively validated.

Our study also confirmed that point mutations at codon 12 of KRAS are the most common oncogene alterations (20). The prevalence of KRAS mutations at codons 12 and 13 seems to vary among different ethnic groups, but our results were similar to those reported by Japanese and Chinese groups; the GGT-GAT mutation at codon 12 was the dominant mutation in Asian patients with pancreatic cancer (21, 22). In contrary, European studies revealed an 80% prevalence of KRAS codon 12 mutation more evenly distributed between GGT-GAT (42%) and GGT-GTT (32%; ref. 23). Interestingly, our result showed different frequencies of KRAS mutation types compared with those of NSCLC. In our study, 62% of KRAS mutations were found in never smokers whereas KRAS mutations are more common in current or former smokers (24, 25). Moreover, never smokers were significantly more likely than former or current smokers to have a transition mutation (G → A) rather than the transversion mutations known to be smoking related (G → T or G → C) in NSCLC. However, further analyses of 71 patients harboring specific KRAS mutations revealed no differences in the incidence of transition versus transversion mutations in our study (25). Our data suggest the possibility of different pathways in subsets of pancreatic cancer patient; therefore, the relationship between smoking status and KRAS mutation needs further investigation.

Taken together, these results suggest that KRAS mutations may be an important biomarker in determining response and prognosis in pancreatic cancer patients especially treated with erlotinib combined with gemcitabine. However, these findings should be further explored in preclinical and upcoming prospective clinical trials to potentially allow the development of new treatment algorithms to identify patients who are most likely to respond to treatment and provide a rationale for personalized medicine in pancreatic cancer.

No potential conflicts of interest were disclosed.

This work was supported by a grant of the Korea Healthcare Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (A091342 to S.T. Kim) and Basic Science Research Program through the National Research Foundation of Korea NRF-2010-0025201 (J.O. Park).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.