Phase II Trial to Evaluate Gemcitabine and Etoposide for Locally Advanced or Metastatic Pancreatic Cancer

Introduction

Although pancreatic cancer accounts for just 3% of new cancer cases in the United States, it is the fourth leading cause of all cancer deaths (1). The American Cancer Society predicted that in 2009, ∼42,470 people in the United States would be diagnosed with pancreatic cancer and about 35,240 would succumb (1). At the time of diagnosis, 96% to 99% of the patients have incurable disease with a median survival of less than 1 year (1). Despite multiple clinical trials with new chemotherapeutic approaches and more aggressive surgery, the 5-year survival rate remains a dismal 5%, with only ∼25% of all cases being potentially surgically resectable. Of the resectable cancers, the 5-year survival rate remains low at only 5% to 20%. Treatment with single-agent gemcitabine for pancreatic cancer has shown a marginal survival benefit, but notably, has exhibited improvement in disease-related symptoms (2–4). 5-Fluorouracil (5-FU)–based therapies combined with other chemotherapeutic agents or radiation therapy have shown a potential minimal survival benefit as adjuvant therapy after surgical resection (5). A number of other trials have been conducted with variable response and overall survival rates, with none resulting in a significant, complete response rate (6–29). Recently, the Food and Drug Administration approved erlotinib in combination with gemcitabine chemotherapy for the treatment of locally advanced, inoperable, or metastatic pancreatic cancer due to its ability to improve overall survival by 23% (hazard ratio, 0.81; ref. 30). However, the long-term survival for advanced disease remains extremely poor.

It is well established that certain molecular features of neoplastic cells correlate with their sensitivity to chemotherapeutic agents and could therefore be used as a basis for drug selection and therapeutic design. In relation to the current study, Koo and colleagues observed a correlation between activating RAS mutations in cancer cell lines and enhanced sensitivity to the cytotoxic agents, cytosine cytarabine and gemcitabine (an analogue of cytarabine; ref. 31). Similarly, they found that tumor cell lines harboring RAS oncogenes are more sensitive to topoisomerase II inhibitors, particularly to epipodophyllotoxin-derived inhibitors such as etoposide and teniposide (31–33). These studies suggested that tumors harboring activating RAS mutations (such as carcinoma of the pancreas, colon, and lung) might be particularly sensitive to these chemotherapeutic agents. In support of this hypothesis, the presence of oncogenic RAS in acute myeloid leukemic cells was associated with an increased rate of complete remission, prolonged duration of complete remission, and improved overall survival of patients in response to treatment with high-dose cytarabine (34, 35).

Activating RAS mutations are one of the most common gain-of-function mutations found in human cancer, and are present in over 90% of pancreatic cancers (36). Based on the high frequency of activating RAS mutations found in pancreatic cancers and the preferential sensitivity of RAS-transformed cells to gemcitabine and etoposide, we designed a phase II trial to evaluate the response/efficacy and toxicity of gemcitabine and etoposide combination therapy in the treatment of locally advanced and metastatic pancreatic cancer. Secondary objectives included the evaluation of the duration of response (time to progression), CA 19-9 biomarker response, overall survival, and quality of life (QOL), as well as the toxicity profile of this two-drug regimen. KRAS codon 12 and 13 mutation analysis was done on a subset of patient tumors to determine any association between the presence and prevalence of KRAS mutations and tumor response.

Results

Of the 40 patients accrued to this clinical trial, 7 had locally advanced disease, whereas 33 had metastatic disease. Five patients were unevaluable due to progression of disease before cycle 2, with one death occurring from an unrelated cause, thus providing 35 evaluable patients (Table 3). Median overall survival for all patients was 6.7 months (4.5–8.1), and 7.2 months (5.5–8.8) for the evaluable patients. The 1-year overall survival was 10% (3.2–21.5) for all patients and 11.4% (3.6–24) in the evaluable patients (Table 3). Of the 35 evaluable patients, 10 (28.6%) exhibited a PR, 12 (34.3%) had stable disease (SD) and 13 (37.1%) had progressive disease (PD; Table 3). Survival curves were significantly different for patients based on their radiological best-observed response by log rank test (P = 0.0012). For those evaluable patients classified as partial responders, the median time to progression was 3.1 months (1.4–9.3; Fig. 1). Within individual response categories, it was shown that survival times for PR patients were significantly improved relative to patients exhibiting PD (P = 0.0007, Fig. 1). Notably, two patients survived over 2 years, the longest surviving 32.8 months, and 2 patients survived over 1 year. The median overall survival for the locally advanced patients was 8.8 months (6–14.8), and 6.75 months (4.5–8.1) for the metastatic patients (P = 0.09). One-year survival differences between metastatic and locally advanced groups were not significantly different (P = 0.4995). No significant survival difference was observed between patients exhibiting PR and those with SD (P = 0.0748). Additionally, there was no significant difference between the number of patients achieving PR or SD between locally advanced and metastatic disease subsets (P = 0.62).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Response and survival rates for all 35 evaluable patients. Overall survival rates over time for each response category; difference in survival time of PR versus PD was statistically significant (P = 0.0007).

Using the LOGISTIC procedure in SAS/STAT 9.2, the predictive ability of CA 19-9 levels on radiologic response was investigated. CA 19-9 levels were recorded at baseline and immediately after every cycle (every 21 days). A >20% drop in CA 19-9 from baseline was used as the threshold of response (38, 39). We observed a >20% decrease in the CA 19-9 level relative to baseline in 20 (57%) evaluable patients. In general, there was a more robust CA 19-9 response in patients exhibiting either PR or SD (data not shown). We found that the overall change in CA 19-9 from baseline level to end of study was a significant predictor of radiologic response (P = 0.048). Additionally, patients showing a >20% decrease in CA 19-9 levels from baseline to end point had approximately five times higher odds of showing radiologic response than patients who did not experience such a decrease (odds ratio, 4.9). Patients who experienced a >20% drop at any point during the study relative to baseline had ∼11 times higher odds of showing radiological response (odds ratio, 11.4). This was also shown to be a significant predictor of radiologic response (P = 0.03).

QOL improved in 12 (35%) of the evaluable patients and remained stable in 3 (8.8%; Table 3). The two most commonly observed dose-limiting toxicities were hematologic toxicity and fatigue (Table 4). Grade 3/4 hematologic toxicity was seen in 26 (74%) of the evaluable patients. Grade 2/3 fatigue was observed in 21 patients (60%). Nine patients died during study treatment or within 30 days of discontinuing treatment because of PD or disease-related complications. Of these nine patients, all were discontinued due to PD, with three patients having associated complications including pleural effusion, recurring strokes, and pain control issues, respectively. Of the remaining evaluable patients, 20 patients were eventually discontinued due to PD.

The results of our trial showed an overall response rate of 28%, which compares well with response rates reported in other clinical studies ranging from 4.1% to 33% (Table 5).

To further characterize the potential association between the presence of an activating RAS mutation and clinical response to gemcitabine/etoposide, archival tissue available from 17 patients was screened for KRAS mutations by pyrosequencing as described in Materials and Methods. Fifteen patient tumors were shown to carry exclusive single base substitutions in either position 1 or position 2 of codon 12 (Supplementary Table S1). KRAS mutations were not identified in either codon 12 or 13 in two patients (both classified as nonresponders; PD). The relative percentage of mutation burden varied greatly (6–95%) in KRAS-positive tumors (Supplementary Table S1). Interestingly, although not statistically significant [P = 0.06 (PR-PD), P = 0.14 (SD-PD), P = 0.57 (SD-PR), pairwise Wilcoxon tests], there was a general positive trend between tumors harboring a higher percentage of KRAS-positive cells and radiological tumor response [mean values for PR (70.1), SD (51.2), and PD (21.8)].

Discussion

RAS mutations are found in >90% of pancreatic cancers (36). Koo et al. (31, 32) observed a striking correlation between activating RAS mutations in human tumor cells and enhanced sensitivity to deoxycytidine analogues as well as to topoisomerase II inhibitors. Thus, human cell lines harboring activated RAS oncogenes display enhanced cytotoxicity to deoxycytidine analogues, including 1-β-d-arabinofuranosylcytosine (cytarabine) and gemcitabine, and especially to topoisomerase II inhibitors.

The presence of RAS oncogenes in acute myeloid leukemia is associated with increased complete remission rate, longer complete remission duration, and improved overall survival in patients treated with cytarabine or the combination of cytarabine and a topoisomerase II inhibitor (34, 40). Collectively, the Koo et al. (31, 32) and Neubauer et al. (34, 35) studies provide a rationale for the combination therapy of gemcitabine and etoposide in the pancreatic cancer study that we describe here. We tested the efficacy and toxicity of this combination in patients who presented with nonresectable pancreatic cancer (41) and are particularly encouraged, both by the trend in patient response (Supplementary Table S1), as well as by the studies of Neubauer et al. (34, 35), showing a more favorable response in patients with acute myeloid leukemia harboring RAS-activating mutations. This combination has shown significant activity against non–small cell lung cancers (42), which also displays frequent RAS mutations (41). None of the studies thus far provide an explanation for why oncogenic RAS enhances sensitivity to cytarabine/gemcitabine and anthracycline/etoposide. In the pancreatic cancer patient group for which tissue was available for KRAS mutation analysis, a trend was found between radiologic response and degree of KRAS expression. Moreover, in this group, the longest survivor (28 months) showed the highest degree of KRAS mutation (95%). These findings provide intriguing support for the earlier data of Koo et al. (32), linking mutant KRAS expression with increased sensitivity to gemcitabine and etoposide in vitro. It further suggests that future clinical investigation in this area should include, in addition to more detailed study of KRAS expression as it relates to clinical response and survival, a comparison of this regimen with standard gemcitabine. It seems possible that greater KRAS expression could identify patient subpopulations with significant sensitivity to this regimen. It would be desirable to include a gene-targeting drug such as erlotinib, which was recently approved for pancreatic cancer (30), with gemcitabine plus an additional agent such as etoposide or cisplatin, anticipating that the loss of receptor tyrosine kinase activity would enhance the tumor sensitivity to chemotherapy agents.

Our clinical trial showed a response rate of 28%, which compares well with the response rates reported in other clinical studies ranging from 4.1% to 33% (Table 5). In our clinical trial, the 1-year survival rate was 11.4% (Table 3). Notably, two patients survived for more than 2 years and a total of four patients (11.4%) survived for more than 1 year. As in most other studies, dose-limiting toxicities were primarily hematologic toxicity and fatigue, but were manageable (Table 4). On the basis of observed toxicities, especially neutropenia, growth factor support was added at the discretion of the clinician and should be strongly considered when this combination therapy is used. QOL was generally described as favorable, with improvements reported in 35% of evaluable patients. This compares favorably with early studies of gemcitabine that showed improvement in clinical benefit response in 28% of patients treated with gemcitabine alone (43). Thus, the combination of gemcitabine and etoposide is generally well-tolerated and exhibits a response rate similar to other published studies (Table 5). Based on the results of this study, follow-up phase II clinical trials comparing or combining this combination with new agents targeting molecular vulnerabilities identified in pancreatic cancer (44) are warranted. An interesting finding in this study is the remarkable overall survival seen in four patients, with two surviving more than 2 years, suggesting that specific molecular subsets might exist that are particularly sensitive to this regimen. Thus, follow-up trials should include correlative studies to determine whether the molecular traits of individual patients and/or their tumors, including the RAS mutation status, could be associated with differential therapeutic responses and outcomes.

Disclosure of Potential Conflicts of Interest

G.F. Vande Woude: Consultant/advisory board. No other potential conflicts of interest were disclosed.