We collected the data of patients who were diagnosed as inoperable pancreatic adenocarcinoma between January 2007 and November 2011 at the Gangnam Severance Cancer Hospital in South Korea. Patients enrollment eligibility was as follows: histologically or cytologically confirmed ductal adenocarcinoma of the pancreas; unresectable/metastatic disease; treated with one of GEM, gemcitabine plus capecitabine (GEM-X), and GEM-T as first-line treatment; measurable or evaluable lesion; age more than 18 years; Eastern Cooperative Oncology Group (ECOG) performance status (PS) 0, 1, or 2; and adequate hematologic (granulocyte count, ≥ 1,500×10⁶/L; platelet count, ≥100×10⁹/L), hepatic (bilirubin < 1.5× upper limit of normal [ULN]), and renal function (serum creatinine < 1.5× ULN) before first-line chemotherapy. Prior radiotherapy for local disease was allowed if the disease progression had been documented and if radiotherapy was completed at least 4 weeks before enrollment. Prior chemotherapy was not permitted, except for postoperative adjuvant treatment or radiosensitizers.

Chemotherapy regimen for each patient was determined by the physician in charge. For the GEM group, gemcitabine (1,000 mg/m²) was given by a 30-minute intravenous infusion on days 1, 8 and 15, every 4 weeks. For GEM-X and GEM-T groups, gemcitabine (1,000 mg/m²) was administered on days 1 and 8, in 3-week cycles. Erlotinib was taken orally at 100 mg daily. Capecitabine was administered orally at 1,700 mg/m²/day (850 mg/m² twice daily) for 2 weeks, followed by a 1-week rest. All treatments were administered until disease progression or intolerable toxicity. The chemotherapy doses could be reduced or delayed (no more than 1 cycle) to allow recovery from toxicity. Dose reduction in gemcitabine and capecitabine was in accordance to two levels, 850 mg/m² (level 1) and 700 mg/m² (level 2) for gemcitabine and 1,500 mg/m² (level 1) and 1,200 mg/m² (level 2) for capecitabine. Patients with poorly tolerated diarrhea, skin adverse drug reaction, or any other adverse events related with erlotinib, were managed with drug interruption, followed by a restart of erlotinib. This study was approved by the Institutional Review Board of the Yonsei University.

The pretreatment baseline evaluation included a complete medical history, a physical examination, vital signs, electrocardiogram, chest radiography, and routine laboratory tests. The serum carbohydrate antigen (CA) 19-9 level was assessed at the baseline. Tumor assessment was performed by a spiral computed tomography scan at the baseline, every 2 cycles. The radiologic tumor response was evaluated in accordance to the Response Evaluation Criteria in Solid Tumors (RECIST). Toxicity was assessed according to the National Cancer Institute Common Toxicity Criteria ver. 3.0. Clinical and laboratory assessments were conducted throughout the study.

The primary end point was PFS, and the secondary end points were OS and ORR. PFS was calculated from the date of initial drug administration to the date of either disease progression or death, or it was censored at the last follow up. OS was calculated from the date of initial drug administration to the date of death from any cause, or it was censored at the last follow up. Both PFS and OS estimates were calculated using the Kaplan-Meier method, and the survival curves were compared among the treatment arms using the log-rank test. The HR with 95% confidence interval (CI) was used as the primary estimate of the difference between the two arms. The effects of potential prognostic factors were assessed using the Cox’s proportional hazards model. The proportions were compared across the groups using the χ² test or Fisher’s exact test. Statistical analyses were done with PASW statistics ver. 18.0 (SPSS Inc., Chicago, IL). A p < 0.05 was considered to be statistically significant.

A total of 127 patients were enrolled in this study. Patients received one of the following treatments: GEM (n=47), GEM-T (n=44), or GEM-X (n=36). The median follow-up duration was 27.8 months (range, 1.5 to 71.9 months). Their baseline characteristics are described in Table 1. Patients in the GEM arm were statistically older than the others. Prognostic factors, such as ECOG PS, disease extent, and baseline CA 19-9 level, were not different between the three groups, except disease extent between the GEM and GEM-T arms. The proportion of locally advanced disease, which presents a better prognosis than distant and recurred diseases, was higher in the GEM arm than in the GEM-T arm (44.7 vs. 25%, p=0.049). Locally advanced disease was present in 36.2% of all cases. The pancreatic head was the most common location of the primary lesions (57.5%). The CA 19-9 level was over the upper normal limit in 107 patients (84.3%). Patients were grouped into high or low CA 19-9 group by the median level of 500 U/mL. Prior chemotherapy and radiotherapy were administered as adjuvant treatments after curative resection of pancreatic cancer.

After the protocol-specified treatment, 69 patients received further treatments. A second-line chemotherapy was administered to 31.9%, 52.3%, and 27.7% of patients in the GEM, GEM-T, and GEM-X arms, respectively. The selection of a second-line chemotherapy was up to the individual investigator, and 5-fluorouracil or platinum-based regimens were mostly used.

One hundred twenty-four patients were assessed for response. One patient was lost to follow-up right after the first dose of chemotherapy, and two patients did not undergo the first follow-up tumor assessment due to their refusal. Table 2 presents a summary of the response results. Patients in the GEM-X arm had an improved partial response rate over GEM and GEM-T (21.2% vs. 12.7% and 15.9%, respectively) and improved the overall disease control rate (partial response plus stable disease; 72.7% vs. 63.8% and 59.1%, respectively); however, there was no statistical significance. The treatment duration was significantly extended in the GEM-X group. The median duration of gemcitabine treatment was 17, 10.4, and 22.8 weeks in the GEM, GEM-T, and GEM-X groups, respectively.

PFS was significantly prolonged in the GEM-X group compared to the GEM and GEM-T groups (GEM-X vs. GEM; median, 8.9 vs. 5.2 months; HR, 0.46; 95% CI, 0.27 to 0.79; p=0.035, and GEM-X vs. GEM-T; median, 8.9 vs. 3.9 months; HR, 0.36; 95% CI, 0.2 to 0.64; p=0.001). GEM-X significantly reduced the hazard of progression compared with GEM and GEM-T (Fig. 1A). The OS was also significantly improved with GEM-X compared to GEM (median OS, 12.1 vs. 10.4 months; HR, 0.52; 95% CI, 0.28 to 0.94; p=0.033) or to GEM-T (median OS, 12.1 vs. 9.9 months; HR, 0.52; 95% CI, 0.28 to 0.96; p=0.037) (Fig. 1B). The 1-year survival rates were 44.4% for GEM-X, 25.5% for GEM, and 22.7% for GEM-T. In the three arms, both partial response and stable disease extended PFS and OS compared to the progressive disease. However, there was no significant difference in clinical benefit between partial response and stable disease (all p > 0.05 in PFS and OS).

In the analysis of GEM-X compared to GEM-T, the results of the subgroup analyses of survival by the baseline characteristic and clinical factor are displayed in Table 3. GEM-X significantly reduced the hazard of progression compared with GEM-T in all cases, except high CA 19-9 patients. A statistically significant benefit of GEM-X in OS was confined to patients with age less than 65.

All 127 patients received at least one dose of medication and were available for an assessment of its toxicity. The median number of cycles was 4 (range, 1 to 14), 4 (range, 1 to 21), and 7 (range, 1 to 20) cycles in the GEM, GEM-T, and GEM-X groups, respectively. A total of 239 cycles of GEM, 236 cycles of GEM-T, and 272 cycles of GEM-X were administered. Adverse events are summarized in Table 4. The treatments were generally well tolerated in all arms. The relative dose intensity of gemcitabine was 93% for GEM, 98% for GEM-T, and 94% for GEM-X. The relative dose intensity of erlotinib was 96% and of capecitabine was 78%. Patients in the GEM-T arm experienced higher frequencies of rash and diarrhea, but these were generally grade 1 or 2. The GEM-X arm was related to more frequent neutropenia and grade 1 or 2 hand-foot syndrome. GEM arm showed more anemia and thrombocytopenia. Infection was more common in the GEM arm, and the main cause of infection was biliary obstruction (80%). Due to older age and ECOG PS 2 patients having received GEM, it is suggested that anemia, thrombocytopenia, and infection may be related more with poor patients’ baseline condition rather than chemotherapeutic regimen. Of the patients receiving GEM or GEM-T, 27% had underwent at least one gemcitabine dose reduction. The main cause of dose modification was hematologic toxicity. A higher proportion (33%) of patients had a dose reduction in the GEM-X arm than in other arms. The dose of capecitabine was reduced in 25% of GEM-X patients (Table 4).