A Multi-institutional Phase 2 Study of Imatinib Mesylate and Gemcitabine for First-Line Treatment of Advanced Pancreatic Cancer

ABSTRACT

BACKGROUND:

The pancreatic tumor microenvironment is rich in receptors for platelet-derived growth factor (PDGFRs). Imatinib mesylate (IM) inhibits PDGFRs and decreases tumor interstitial fluid pressure, potentially improving drug access. These data and promising results in a phase 1 trial formed the rationale for a phase 2 trial combining IM and gemcitabine (GEM) in pancreatic cancer.

METHODS:

Eligibility criteria included chemotherapy-naïve, locally advanced or metastatic pancreatic cancer; ECOG (Eastern Cooperative Oncology Group) performance status ≤2; and adequate end-organ function. The primary end point was progression-free survival (PFS). Secondary end points included response rate, toxicity, and overall survival (OS). GEM was given at 1200 mg/m²/120 min on days 3 and 10. IM (400 mg) was taken orally on days 1 to 5 and 8 to 12 of a 21-day cycle. Response was assessed every 3 cycles.

RESULTS:

Forty-four patients from 7 centers were enrolled from October 2005 through July 2009. Median age was 62 years. The median number of cycles completed was 3 (range, 0–17). Common adverse effects included neutropenia, nausea, anemia, and fatigue. Half the patients required dose reductions. There were no complete responses to therapy. During treatment, 1 patient showed a partial response, 16 had stable disease, and 18 had progressive disease. The median PFS was 3.9 months (95% confidence interval, 2.1–5.1), the median OS was 6.3 months (95% confidence interval, 5.2–8.5), and the 1-year survival rate was 25.6% (95% confidence interval, 13.8–39.1).

CONCLUSION:

IM in combination with GEM is tolerated in locally advanced, metastatic, or recurrent pancreatic cancer, but does not show a statistically significant PFS or OS benefit over chemotherapy with GEM alone.

Pancreatic adenocarcinoma carries a dismal prognosis¹ and remains a significant cause of cancer morbidity and mortality. The purine analog gemcitabine (GEM)² has been considered the standard treatment for patients with advanced pancreatic cancer since Burris et al² demonstrated that it yields a modest, yet statistically significant, improvement in overall survival and a significant clinical benefit compared with 5-FU. Although a combination regimen of GEM and erlotinib has been approved by the U.S. Food and Drug Administration (FDA)³ and some aggressive combination regimens have shown promise in those patients with excellent performance status,^4–6 single-agent GEM remains the common standard of care for pancreatic cancer. In multiple trials, it has achieved median OS of only approximately 6 months,⁷ a finding that clearly indicates the need for new treatment strategies.

One possible strategy for more effective therapy is to improve the access of chemotherapeutic drugs to the interior of the tumor. It has been hypothesized that inhibition of platelet-derived growth factor receptors (PDGFRs) may decrease tumor interstitial fluid pressure (IFP), allowing better penetration of GEM. PDGF is a significant mediator in the proliferation of cancer-associated stromal fibroblasts.⁸ Imatinib mesylate (IM) is a small molecule that inhibits PDGF, as well as the tyrosine kinases associated with BCR-ABL and c-kit.⁹ IM has become the standard of care in the treatment of patients with Philadelphia chromosome–positive chronic myelogenous leukemia¹⁰ and c-kit-positive unresectable, malignant gastrointestinal stromal tumors.¹¹ However, the activity of single-agent IM has been disappointing in treating pancreatic cancer^12,13 and other solid tumors, including small cell lung cancer (SCLC), ovarian cancer, germ cell tumors, breast cancer, prostate cancer, and melanoma.^14–20

One rationale for using IM in anticancer therapy focuses on PDGFRs. In addition to their expression in several tumor types, PDGFRs are found both in tumor pericytes and tumor vasculature^21,22 and are thought to play a role in the control of IFP.¹⁸ PDGF A- and B-chain mRNA has been identified in human pancreatic adenocarcinoma cells and may contribute to the phenomenon of desmoplasia in pancreatic tumors by epithelial–mesenchymal interactions.²³ Through downregulation of PDGFR activity and thus downregulation of IFP, IM has been postulated to improve drug delivery to tumor cells, enhancing the effectiveness of chemotherapeutic agents. Several preclinical studies have demonstrated improved activity of cytotoxic chemotherapeutic agents, including GEM, when combined with IM.^24–26 As a single agent in c-kit-overexpressing pancreatic cancer cell lines, IM had an inhibitory effect on stem-cell-factor–induced proliferation and invasion,²⁷ and the inhibitory effect of IM was dose dependent in several pancreatic cell lines.²⁸ In vivo experiments in a mouse human malignant mesothelioma xenograft model showed increased efficacy of GEM when coadministered with IM, postulated to be due to increased GEM delivery.²⁹ Findings such as these imply that IM could delay disease progression and extend survival when coadministered with GEM.

Particularly encouraging activity was seen with the combination of IM and GEM in an orthotopic nude mouse pancreatic cancer model (nude mice with human pancreatic L3.6pl cells injected into the pancreas).²⁶ This cell line has been used in other studies investigating novel therapies and to study the expression of PDGFR in the stroma in pancreatic cancer.^30,31 Hwang et al²⁶ evaluated the expression of PDGF ligands and receptors in clinical specimens of human pancreatic adenocarcinomas and determined the therapeutic effect of two different doses of IM on human pancreatic carcinoma cells growing in the pancreas and liver of nude mice. Tumors of mice treated for 4 weeks with IM alone were not significantly smaller than those in the controls. However, tumors of mice treated with a lower dose of GEM plus IM were more than 70% smaller than tumors in control mice and 36% smaller than those in mice treated with GEM only (P < .0002 and P < .04, respectively). Tumors from mice treated with the combination of IM and GEM showed decreased expression of activated PDGFR, decreased mean vessel density, decreased cell proliferation, and increased apoptosis of tumor cells. This enhancement of cytotoxicity by IM indicates activity of the PDGFR pathway in pancreatic cancer, which may also be mediated through c-kit and/or PDGF-α receptors on endothelial cells.

These data formed the basic rationale for a phase 1 trial combining IM and GEM. Two phase 1 trials of the use of IM and GEM in solid tumors have been reported. In the first study, George et al³² used a 30-minute infusion of GEM given on days 1 and 8 of a 21-day cycle with daily IM in patients who had received only minimal prior treatment. Yet toxicity was severe, prompting the study's closure. The second phase 1 study, conducted at The Cancer Institute of New Jersey, initially demonstrated high levels of toxicity as well.³³ Initially, treatment consisted of 28-day cycles of 600 mg/m² of GEM (10 mg/m²/min) on days 1, 8, and 15 and 300 mg of IM daily. Because of excessive toxicity, the schedule was altered to 21-day cycles of IM on days 1 to 5 and 8 to 12, and 420 mg/m² of GEM on days 3 and 10. Two final cohorts received 400 mg/day of IM on days 1 to 5, 8 to 12, and 15 to 19, and 1200 or 1500 mg/m² of GEM, respectively. There was a nonsignificant increase in GEM area under the curve (AUC) in the presence of IM. Clearance and volume of distribution were not different in the absence and presence of IM.

With modified intermittent doses of IM bracketing the GEM doses, dose-limiting toxicities were not observed, even at IM 400 mg daily on days 1 to 5 and 8 to 12 and GEM at a fixed dose rate (FDR) infusion of 1500 mg/m²/150 min. In this latter clinical trial, partial responses were observed, but more striking was the degree of disease stabilization (31% of patients). Specifically, 6 patients with pancreatic cancer had stable disease at 12 weeks, 1 with a minor response. This well-tolerated bracketed dose regimen was used in the current trial.

Based on the aforementioned promising results in phase 1, we studied this drug combination in a phase 2 clinical trial in patients with locally advanced, metastatic, or recurrent pancreatic cancer. We opted to use GEM at 1200 mg/m², as this is closer to the most common dose used in the treatment of pancreatic cancer³⁴ and is a dose that was tolerable in combination with IM in the phase 1 study just described.³³

MATERIALS AND METHODS

Patient Selection

Patients were screened for participation at hospitals in the CINJOG (Cancer Institute of New Jersey Oncology Group) network and at Northwestern University's Robert H. Lurie Comprehensive Cancer Center. All patients had pathologically confirmed pancreatic adenocarcinoma at enrollment.

Prior treatment with GEM or IM was an exclusion criterion. All patients were required to have an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2, to be 18 years of age or older, and to have disease that was measurable in concordance with response evaluation criteria in solid tumors (RECIST) criteria.³⁵ Patients had to have been deemed ineligible for curative resection and could not have had prior chemotherapy for metastatic disease, with the exception that patients with prior surgical resection and a history of adjuvant fluorouracil chemotherapy were eligible if at least 6 months had passed between the last dose of adjuvant chemotherapy and the recurrence of pancreatic cancer. Patients with known brain metastases were excluded from the trial. Laboratory eligibility criteria included absolute neutrophil count (ANC) ≥1,500/μL, platelet count ≥100,000/μL, aspartate aminotransferase (AST) and alanine aminotransferase (ALT) ≤2.5 times the upper limit of normal, total bilirubin ≤1.5 times upper limit of normal, and serum creatinine ≤1.5 times upper limit of normal. (If greater that 1.5 times normal, then calculated creatinine clearance was ≥60 mL/min, according to the Cockcroft-Gault equation.)³⁶ Therapeutic warfarin was not allowed. Patients with a history of malignancy, with the exception of carcinoma in situ of the cervix and nonmelanoma skin cancer, within the past 5 years were excluded. All patients gave written, informed consent, in accordance with institutional and federal guidelines.

Treatment Plan and Dose Adjustments

All patients received the following therapies on a 21-day cycle: GEM was given at a fixed dose rate of 1200 mg/m²/120 min as an intravenous infusion on days 3 and 10. IM 400 mg was taken orally with a meal on days 1 to 5 and 8 to 12. No drugs were given on days 6, 7, or 13 through 21. Missed doses were not replaced. Antiemetic therapy was allowed (except for routine steroid use), as were erythropoietin and granulocyte colony-stimulating factor (G-CSF), according to American Society of Clinical Oncology (ASCO) guidelines. Use of ketoconazole, rifampin, simvastatin, and acetaminophen was discouraged, as was use of other drugs known to have significant interactions with CYP450 enzymes. No grapefruit juice was allowed.

Dose Modifications

Complete blood count with white count differential and platelets were obtained weekly. Electrolytes and liver function tests were required before the initiation of each cycle of therapy. Before they began therapy, the patients had to meet hematologic eligibility criteria. On day 1 of each subsequent cycle ANC >1,200/μL and platelet count >90,000/μL were necessary for therapy to be initiated. If counts were inadequate, the start of the cycle was delayed for up to 14 days, until the parameters were met. If during the previous cycle, nadir ANC ≤500/μL or nadir platelet ≤50,000/μL, then the day 3 dose of GEM was reduced by 25%. On day 10, ANC >1,000/μL and platelet count >75,000/μL were necessary to continue therapy. If counts were inadequate, the day 10 GEM treatment was omitted, as were the day 11 and 12 IM doses.

For subsequent cycles, any grade 3 or 4 nonhematologic toxicity observed on day 1 resulted in delay of treatment up to 14 days. If any grade 3 or 4 nonhematologic toxicity lasted more than 5 days during the previous cycle or if day 10 treatment had been omitted, the dose of GEM was reduced by 25%. Any grade 3 or 4 nonhematologic toxicity observed on day 10 resulted in omission of all further treatment with GEM and IM for the remainder of the cycle. All dose reductions were permanent. Two dose reductions were allowed before patients were removed from the study. Treatment delays up to 14 days were allowed. Further delays resulted in removal of the subject from the study.

Response and Toxicity Criteria

Tumor measurements were obtained at 4 weeks or less before the therapy start date and again after every third cycle. Patient response measurements were determined by standard RECIST 1.0 criteria for measurable disease. All patients who received one dose of protocol therapy were evaluated for toxicity according to the common toxicity criteria for adverse effects (CTCAE), version 3.0. In addition to laboratory evaluations, the patients underwent a complete history and physical examination before each cycle of therapy, to evaluate for toxicity. They were withdrawn from protocol therapy for the following: disease progression, unacceptable toxicity, or a 2-level decrement in performance status. They also could withdraw at their own request or if the treating physician thought that the constraints of the protocol were detrimental to their health.

Statistical Design and Analysis

The primary end point was PFS. Secondary end points included response rate, toxicity, OS, 1-year survival rate, and toxicity evaluation. Data from 42 patients were analyzed. The method of Kaplan-Meier estimation was used to assess OS and PFS. OS was defined as time from treatment start date up to death from any cause, and PFS was defined as time from treatment start date to identification of disease progression. For PFS, we computed times from study entry to either progression or death, whichever came first. An interim analysis was planned when 14 patients had received treatment and had been observed for 6 weeks; if all 14 patients had progression at 6 weeks, the trial was to be stopped on grounds of futility. The study was designed to detect, with 80% power, an increase in PFS from 2.2 months with GEM alone to 4 months with the combination therapy, using a 5% level.

RESULTS

Patient Population and Treatment

A total of 44 patients from 7 centers were enrolled from October 2005 through July 2009. The trial was opened at a single regional cancer center and then expanded to other centers to reach accrual goals within the target period. All patients with pancreatic cancer seen in the medical oncology clinic at the lead center were screened for potential participation in the trial. One patient withdrew consent before starting treatment, and 1 was determined, after initiating treatment, to have an ampullary cancer and was evaluable for toxicity but not for response. Therefore 2 additional patients were enrolled. Patient characteristics were (see Table 1) median age, 62 years (range, 45–85) and ECOG performance status 0, 12 patients; 1, 28 patients; and 2, 2 patients; the status of 1 patient was not recorded. Eight patients had locally advanced pancreatic cancer (LAPC), and 37 had metastatic pancreatic cancer (MPC) at baseline.

Table 1.

Patient demographics: enrolled from October 2005 through July 2009

Characteristic	No. of patients	%
Total	44*
Age, y
Median	62
Range	45–85
Sex, n
Female	21	49
Male	23	51
Disease extent, n
Primary/locally advanced	8	17
Metastatic	35	84
Site of first metastasis, n
Liver	32	73
Lymph nodes/retroperitoneum	3	6
Lung	2	4.5
Peritoneum/omentum	1	2
Race, n
White	36	84
Black/African-American	4	9
Asian/Pacific Islander	2	5
Native American	1	2
ECOG performance status, n
0	12	28
1	28	65
2	2	5
Not recorded	1	2

^*One male patient was determined after initiating treatment to have a metastatic ampullary cancer and was evaluable for toxicity, but not for response.

Toxicity

In total, 161 cycles of therapy were given to 40 patients. The median number of cycles completed was 3 (range, 0–17; see Table 3). Twenty-seven cycles were delayed in 15 patients, with the most delays in cycle 3 (8 patients). Hematologic toxicity was significant. Overall, 50% of patients had grade 3 or higher neutropenia and 17% had grade 3 or higher thrombocytopenia. Although these toxicities were manageable, they resulted in numerous dose reductions. Specifically, 11 patients (25%) required dose reductions in cycle 2 for hematologic toxicity (grade 3 or higher thrombocytopenia, n = 2; grade 3 or higher neutropenia, n = 10). An additional 3 patients with grade 3 or higher neutropenia required dose reductions in cycle 3. Altogether, GEM treatment was withheld 24 times and IM treatment was withheld 15 times.

Table 3.

Treatment and response

	No. of patients	%
Cycles of therapy, n	40
Median	3
Range	0–17
Reason withdrawn from study, n
Adverse event	13	30
Progression of disease or death unrelated to treatment	22*	52
Withdrawal before beginning treatment	4	9
Other	4	9
Best response, n
Partial response	1	2
Stable disease	16	38
Median duration of response	2 mo (range, 1–13)
Progressive disease	18	43
Not assessed	8†	19

^*Thirteen patients died while receiving treatment; no deaths were related to treatment.

^†One patient was not evaluable for response.

The significant nonhematologic toxicity related to treatment were grade 3 or higher dehydration in 9% of patients, grade 3 or higher rash in 9%, grade 3 or higher fatigue in 6%, grade 3 or higher nausea in 6%, and grade 3 or higher renal failure in 2%. Clinically significant grade 1 and 2 toxicities included edema in 4% of patients and rash in 6%. No deaths were associated with toxicity. All grade 3 and 4 toxicities are listed in Table 2.

Table 2.

Grade 3/4 toxicity: percentage of patients, by grade

Toxicity	Grade 3	Grade 4
Neutropenia	34	16
Leukopenia	13	0
Nausea	6	0
Fatigue (asthenia, lethargy, malaise)	4	0
Hemoglobin	4	2
Platelets	13	4
Anorexia	6	0
Limb edema	2	0
Fever (in the absence of neutropenia)	2	0
Rash	2	0
Dehydration	9	0
Mucositis/stomatitis of oral cavity	2	0
Aspartate aminotransferase	2	0
Chest pain	2	0
Abdominal pain	2	0
Confusion	2	0
Renal failure	2	0
Psychosis (hallucinations/delusions)	2	0
Bruising (in absence of grade 3 or 4 thrombocytopenia)	2	0
Hyperbilirubinemia	2	0

Efficacy

Forty-two patients were evaluable for response (Table 3). One patient withdrew before beginning treatment, 13 withdrew during the trial for an adverse event, and 22 withdrew during the trial for progression of disease. There was no complete response. One patient had a partial response, 16 patients had stable disease, 18 patients had progressive disease while on treatment, and 8 patients discontinued treatment before response was assessed. No patients remained alive at the time of this analysis.

Assessment was conducted at 9-week intervals and, of the 17 patients assessed to have PD, PD was found at an average of 9 weeks after treatment began; however, 2 of these patients were assessed later than planned because of cycles delayed for a total of 2 weeks each.

The Kaplan-Meier method was used for the analyses of OS and PFS. Every patient had a recorded death date, and 16 had no recorded progression date. The median PFS (Figure 1) was 3.9 months with 95% confidence interval 2.06–5.14. The median OS (Figure 2) was 6.23 months with 95% confidence interval 5.15–8.46. The 1-year survival rate was 25.6% (95% confidence interval 13.8%–39.1%). There was a nonsignificant difference in OS between patients with LAPC and MPC. The median overall survival time for patients with LAPC was 9.9 months (95% confidence interval, 5.1–13.0). The median overall survival time for patients with MPC was 6.2 months (95% confidence interval, 3.8–7.2).

Figure 1.

Progression-free survival (PFS). The median PFS was 3.9 months with 95% confidence interval 2.06–5.14.

Figure 2.

Overall survival (OS). The median OS time was 6.23 months with 95% confidence interval 5.15–8.46.

DISCUSSION

A phase 1 trial at our institution demonstrated notable prolonged disease stabilization in patients with pancreatic cancer who received GEM and IM.³³ In that trial, doses of up to 1500 mg/m² of GEM were tolerated, even in a heavily pretreated population, and in fact a maximum tolerated dose (MTD) was not reached. There was no evidence of significant pharmacokinetic interaction between GEM and IM. We report results of a test of this combination in a phase 2 clinical trial in a first-line setting for patients with LAPC or MPC. Our principal finding is that the drug combination does not yet show a statistically significant improvement in PFS or OS.

The predominant toxicity observed was neutropenia (50% ≥ grade 3), followed by thrombocytopenia (17% ≥ grade 3). This level of toxicity is similar to or slightly in excess of that reported for single-agent GEM, which has been shown to result in grade 3 to 4 neutropenia 27% to 60% of the time and thrombocytopenia 10% to 33%.^2,37 Of note, GEM was given at a fixed dose rate of 1200 mg/m²/120 min in our trial, which may have increased myelosuppression when compared with toxicity in historical studies, as studies in which a 30-minute infusion of GEM was used demonstrated less myelosuppression than fixed dose rates.³⁸ Doses of IM ranging from 400 to 800 mg/day result in rates of grade 3/4 neutropenia in the 5% range and thrombocytopenia in the <2% range in patients with solid tumors, but the bracketing dose used in this trial was the MTD from a phase 1 combination trial with GEM.^11,33,39 Clinically significant dehydration was experienced more commonly in this trial than was reported in trials of single-agent GEM.^2,37

The demonstrated median PFS of 3.9 months, the primary end point of this single-arm phase 2 trial, is not better than historical controls for single-agent GEM, nor is the median OS of 6.23 months or the 1-year survival rate of 25.6%. The median PFS of 3.9 months is larger than the 2.2 months previously reported. However, 2.2 months is well within the 95% confidence interval for our estimate, and thus the change is not statistically significant. We cannot rule out, however, that our result may understate the lack of benefit because monitoring was done after every 3 cycles of therapy and among patients whose disease progressed, the time between beginning dates of consecutive cycles averaged 23.2 days, 2.2 days longer than the intended 21 days. However, differences in PFS and OS were both in the positive direction and approached significance. This marginal result leaves open the possibility that GEM with IM confer a small benefit by exerting effects on the tumor stroma that were beyond the power of this trial to quantify.

Clinical research with attention to the interaction between the tumor and stroma in pancreatic cancer continues and has recently has focused on the SPARC (secreted protein, acidic, and rich in cysteine) pathway,⁴⁰ the role of the hedgehog pathway,⁴¹ and improving transport of GEM into cancer cells.⁴² Although both the A and B isoforms of the PDGFR have been identified in pancreatic tumor lines,²³ the current trial did not include analysis of patient tumor specimens and therefore cannot address whether the target was expressed in all or some patients and whether this expression would have correlated with response. The use of IM as a means of improving drug delivery to the tumor cells within the pancreatic cancer stroma remains an attractive possibility, but one not borne out by the results of the current trial.

CONCLUSION

The combination of GEM with IM for first-line treatment in patients with LAPC or MPC was generally as well tolerated as single-agent GEM, but the survival benefit did not reach statistical significance over historical data for single-agent GEM in this setting.