The main objective of this single-arm phase II study was to determine the objective response rate of FOLFIRI3-b for patients undergoing first-line treatment of metastatic colorectal cancer. Secondary objectives were the monitoring of side effects, the assessment of median progression free and overall survival and the characterization of angiopoietin-2 prognostic value. Sixty-one patients were enrolled in two general hospitals and a university hospital. The study was conducted in accordance with the Declaration of Helsinki, the International Conference on Harmonization Guidelines for Good Clinical Practice. Each patient gave written consent before entering the study. The protocol was approved by the ethical committee CPP Est-II and registered on ClinicalTrials.gov (study NCT00544011).

Patients with first-line mCRC and measurable lesions were included. Previous adjuvant chemotherapy was allowed provided that the last administration was given at least 6 months before randomization. Other eligibility criteria were: age 18 to 80 years, Eastern Cooperative Oncology Group Performance Status 0, 1 or 2 and written informed consent. Exclusion criteria included: previous bevacizumab treatment; proteinuria exceeding 1 g/24 h; coagulopathy; inadequate hematologic function (absolute neutrophil count < 1.0 × 10⁹/L, platelets < 100 × 10⁹/L); inadequate hepatic function (total bilirubin > 1.5× upper limit of normal (ULN), serum transaminases > 2× ULN in absence of liver metastases or > 5× ULN in presence of liver metastases); inadequate renal function (clearance of creatinine < 30 ml/min); major surgery <28 days before inclusion; non-controlled cardiovascular disease; non-controlled hypertension; active hemorrhagic event; aspirin more than 325 mg/day treatment. As the main objective was to assess the tumor response rate, patients amenable to curative intent surgery were included in the study. Surgery of metastases was allowed after 6 cycles of FOLFIRI3-b and the precise timing left at the discretion of investigators.

FOLFIRI3-bevacizumab regimen consisted of bevacizumab 5 mg/kg day 1, irinotecan 100 mg/m² day 1, LV 400 mg/m² day 1 followed by a 46-h 5-FU continuous infusion (2400 mg/m²), and irinotecan 100 mg/m² at day 3 12 . Induction treatment was administrated every 2 weeks for a maximum of 12 cycles, until disease progression, unacceptable toxicities, surgical intervention, or withdrawal of consent.

Maintenance treatment consisted of bevacizumab 7.5 mg/kg intravenous infusion over 60 minutes, and capecitabine 1000 mg/m² day 1 to 14, every 3 weeks until tumor progression, unacceptable toxicity, or withdrawal of consent.

Dose reductions were required for all grade 3 or 4 toxicities attributed to study medications. Bevacizumab was not dose reduced but delayed or discontinued in patients with grade 3 or 4 toxicities.

At baseline and before each cycle, clinical and biological examinations were performed. Adverse events were evaluated continuously and graded according to National Cancer Institute Common Toxicity Criteria (NCI CTC) version 3. The response was defined according to Response Evaluation Criteria In Solid Tumors (RECIST) v1.0. Radiological response assessment was validated in a blinded fashion by an independent radiologist. Tumor responses were assessed every eight weeks by spiral computed tomography. Resectability of metastases was determined by IRFC-multidisciplinary gastrointestinal oncology team before treatment initiation and when the best response was achieved.

Blood samples were drawn at baseline and immediately processed for plasma and serum freezing at -80°C. Samples frozen more than 4 hours following venous blood collection were not included in the analysis. Enzyme-linked immunosorbent assays (ELISA) were used to measure angiopoietin-2 in serum samples and VEGF-A in plasma samples (RnD systems) according to the manufacturer’s instructions. Each sample was analyzed in duplicate.

The primary endpoint of the trial was to determine the objective response rate to the FOLFIRI3 sans espace bevacizumab regimen in patients with first-line mCRC cancers. The study was designed as an exploratory, pilot, single-arm, multicenter, phase II study. A Simon two-stage (optimal) design was used. At least 11 out of the first 21 patients needed to achieve an objective response in stage 1 for recruitment to continue until a target of 45 fully evaluable patients had been reached. Based on standards available at the time of study design, we considered the standard objective response rate (H0 hypothesis) achieved with FOLFIRI and bevacizumab as 50%. Based on an α and β error probabilities of 0.1 respectively, we considered that if an objective response (complete and partial response) was observed in at least 26 out of 45 patients with metastatic colorectal cancers, the regimen would merit further evaluation in prospective subsequent trials. Moreover, regarding the high number of patients who underwent a curative intent surgery, the cohort was extended to 61 patients to include a total of 40 patients with non-resectable disease. Survival data were computed according to Kaplan-Meier method and analyses were performed on an intent-to-treat population including all registered patients. Secondary endpoints were PFS, OS, toxicity profile, metastases resection rate, and biologic analysis of possible predictive factors of efficacy and toxicity. PFS was defined as the time from study enrolment to the first documentation of progressive disease (PD) or death from any cause. Overall survival was defined as the time from study enrolment to death from any cause.

Sixty-one patients were enrolled in the study between October 2007 and July 2009. The median age of the patients was 63.6 years (range, 38–81). Fifty-seven patients (93%) had an Eastern Cooperative Oncology Group Performance Status (ECOG PS) 0–1. Thirty-five patients (57%) presented only one metastatic site and 7 patients (12%) had 3 or more visceral sites involved by metastases. Twenty-five patients (41%) had liver-limited metastases. Twenty-two patients (36%) displayed metachronous metastases at inclusion. Eighteen patients (30%) had received prior adjuvant chemotherapy, including an oxaliplatin-based chemotherapy for 15 patients (25% of the whole cohort). Primary tumor was resected in 3 out of the 39 synchronous metastatic patients. Two primary tumor resections were performed before enrolment due to a symptomatic disease and one resection was performed after enrolment due to an occlusion (Table  1).

All patients accomplished at least one cycle of treatment. Patients received a median of 10 cycles of induction chemotherapy (range 2–14) and 40 patients completed all planned cycles of FOLFIRI3-b. Twenty-one patients (34%) discontinued their induction treatment: 6 for progression, 3 for treatment-related toxicities and 12 for programmed metastasis surgery. The relative dose intensity (calculated from FOLFIRI3-b initiation to the last course of chemotherapy delivered before toxicity, disease progression, surgery, or the planned 12^th cycle of FOLFIRI3-b) was 88% for Irinotecan, 85% for 5-FU, and 87% for bevacizumab. Median dose administered for each cycle was 91% for irinotecan, 95% for 5-FU, and 100% for bevacizumab.

Toxicity analysis was performed in the intent-to-treat population (61 patients, Table  2). Twenty patients (33%) suffered ≥ grade III treatment-related side effects. One patient (1.6%) died during induction regimen due to treatment-related toxicity (colitis and sepsis). Moreover, one patient with synchronous metastases exhibited a colon cancer bleeding requiring surgery. Post-operative recovery was achieved within 7 days and chemotherapy was reintroduced at day 28 post-surgery. Most common grade III-IV toxicities were diarrhea (15%), neutropenia (13%), asthenia (10%), and infections (4%). Hypertension-related medications needed to be increased in 3 patients.

*One catheter related deep vein thrombosis.

Grade IV toxicity: the only grade IV toxicity observed was neutropenia.

Among the 60 assessable patients, ORR was 66.7% (95% CI: 55–79) including 58.4% of partial responses and 8.3% of complete responses (CR). Stable disease (SD) was observed in 25% of patients conferring to this strategy a disease control rate of 91.7%. An independent review of the radiological responses was performed and showed only one difference: one patient was considered to be in CR by our investigators and in non-measurable SD by the independent committee. However, later liver resection confirmed histological CR.

After a median time of follow-up of 46.7 months (data base up-dated in December 2012), 56 patients (92%) had progressed. The median PFS was 12.72 months (95% CI: 9–16 months). So far, 42 patients (68.8%) have died and median OS was 24.5 months (95% CI: 10.6-38.3 months). Of note, none of the patient’s characteristics influenced the clinical results achieved with FOLFIRI3-bevacizumab treatment. Particularly, the ORR was 58.3% in patients above 70 year-old compared to 68.8% for patients less than 70 year old at the inclusion (p = 0.51). Moreover, FOLFIRI3-bevacizumab was also effective in patients treated with oxaliplatin-based adjuvant chemotherapy (ORR of 60%, p = 0.54).

Twenty-one patients (34%) underwent curative-intent surgery of their metastases. A complete histological response was observed in 5 patients. R0 resection was achieved in twenty patients, while the resection status was considered as R1 for one patient. Four of these patients were initially considered as non-resectable, including one patient with 2 sites involved by metastases (lung and liver). Five patients (24%) among the 21 patients who have undergone metastasis surgery were still in complete remission at the time of the last analysis. The median progression free survival was 23.6 months (95% CI: 17.3-29.9 months) in patients who underwent a surgery of their metastases compared to 9.4 months (95% CI: 8.47-10.33 months) for patients with non-resected metastases (Figure  1; p < 0.001). The median overall survival was not reached in patients who underwent a metastatic surgery, while median OS was 18.33 months (95% CI: 15.15-21.5 months) in patients with unresectable disease (p < 0.001).

Twenty patients did not receive maintenance chemotherapy: 1 patient died during induction regimen, 11 patients displayed progressive diseases before capecitabine-bevacizumab initiation, 8 patients underwent curative-intent surgery and no further treatment was proposed. Then, capecitabine-bevacizumab therapy was administered to 40 eligible patients and one more patient, who refused capecitabine, received only bevacizumab.

A total of 346 cycles and a median of 8.7 cycles was administered (range 1–33) per patient at the time of analysis. Median dose of capecitabine per cycle and per patient was 83.4%.

The treatment was generally well tolerated. No grade IV toxicity was recorded and 20% of patients experienced grade III toxicities. Most frequent grade III toxicities were hand-foot syndrome (15%), and diarrhea (5%). Bevacizumab had to be stopped before capecitabine in only one case due to catheter-related deep vein thrombosis. Overall, maintenance treatment was discontinued for toxicity or withdrawal of consent in 2 patients (5%). The median duration of disease control from first cycle of maintenance therapy was 8 months (range, 1–38).

In 51 patients, plasma and serum were available at baseline for analysis. Among angiogenic factors, angiopoietin-2 (Ang-2) was recently proposed in a cohort of 34 patients, as a candidate biomarker for outcomes of mCRC patients treated with bevacizumab-containing chemotherapy 14 . Then, we decided to monitor Ang-2 and VEGF-A levels in this study. Seven patients included in this phase II clinical trial had increased VEGF-A levels compared to normal volunteers or stage II-III colorectal cancers. However, we did not observe a significant negative influence of increased VEGF-A levels on the ORR (83%), PFS (10.7 months) or OS (20.6 months). A preliminary set of experiments confirmed that Ang-2 levels remain below 5 ng/mL in all normal volunteers (n = 20) or stage II-III colorectal cancers (n = 20), in line with the results of Goede et al. 14 . Ang-2 plasma levels above 5 ng/mL at baseline were observed in nine patients (17.3%) of our cohort. ORR was 44% in patients with increased levels of Ang-2, compared to 74.4% in patients with Ang-2 levels below 5 ng/mL. The median PFS was 7.7 months (95% CI: 0–15.9 months) in patients with high Ang-2 levels compared with 13.6 months (95% CI: 10.1-17.2 months) in patients with Ang-2 levels below 5 ng/mL (Figure  2A). Furthermore, overall survival was significantly better in patients with low levels of Ang-2 (median OS: 34.7 months; 95% CI: 19.8-49.7) than in patients with high levels of Ang-2 (median OS: 7.7 months; 95% CI: 5–16.3 months; Figure  2B). In multivariate analysis, surgery of metastases and Ang-2 levels were the only independent prognostic factors for PFS and OS. Indeed, a level of Ang-2 above 5 ng/mL was confirmed as an independent prognostic factor for progression free survival (HR = 0.357; 95% CI: 0.168-0.76, p = 0.005) and overall survival (HR = 0.226; 95% CI: 0.098-0.53, p = 0.0002). Altogether, these results confirm the clinical interest of angiopoietin-2 monitoring to predict PFS and OS in mCRC patients and suggest a decreased efficacy of FOLFIRI3-b in these patients.