Patients

Eligible patients had histologically confirmed unresectable locally advanced or metastatic colorectal cancer, were ≥ 70 years old with an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of 0 to 2 and had not received any prior chemotherapy or biologic agent for metastatic colorectal cancer; adjuvant chemotherapy with a fluoropyrimidine was allowed if 6 months or more had elapsed since its completion. Other eligibility criteria included measurable disease by Response Evaluation Criteria in Solid Tumors (RECIST) 1.0; [16] adequate bone marrow, liver and kidney function defined as Glomerular Filtration Rate (GFR) as calculated using the Cockcroft-Gault formula ≥30 ml/min; absence of brain metastases; a life expectancy of at least three months at enrollment; frail patients as determined by the Comprehensive Geriatric Assessment (CGA) were excluded [17]. Other exclusion criteria were the daily use of aspirin (>325 mg per day), other non-steroidal anti-inflammatory drugs or anticoagulants at a therapeutic dose; and significant cardiovascular disease such as uncontrolled hypertension, coronary heart disease or previous stroke. The study has been approved by the ethics and scientific committees of the participating centers and all patients gave their written informed consent in order to enter the study (Scientific Committees and Committees for Bioethics and Medical Ethics, University Hospital of Herakleion; University General Hospital of Alexandroupolis; IASO General Hospital of Athens; Air Forces Military Hospital of Athens; Laikon General Hospital; State General Hospital of Larissa; Theagenion Anticancer Hospital of Thessaloniki).

Treatment

We used a modified XELOX regimen, administered capecitabine in one week on one week of schedule, since this is associated with less gastrointestinal adverse events and hand and foot syndrome [18]. Capecitabine was given orally at a dose of 750 mg/m² twice a day during days 1–7, a regimen that was found to be effective at a phase II study previously conducted by our group, albeit with a lower dose due to increased toxicity concerns when combined with both oxaliplatin and bevacizumab [18]. The dose was lowered to 625 mg/m² twice a day if the Glomerular Filtration Rate (GFR), as calculated using the Cockcroft-Gault formula, was ≤60 ml/min. Oxaliplatin was given intravenously at a dose of 85 mg/m² over 240 minutes on day 1. Bevacizumab was given intravenously at a dose of 5 mg/kg over 60–90 minutes according to investigator discretion, on day 1. Treatment was administered every two weeks for a total of 12 cycles of chemotherapy except for cases of disease progression, unacceptable toxicity or consent withdrawal. Treatment was postponed for 1 week if on day 1 of each cycle neutrophils were <1500/μL, platelets < 100000/μL and if there were other > grade 2 toxicities. If a patient had not recovered after delaying treatment for 2 weeks, he was removed from the study.

The prophylactic use of Granulocyte Colony Stimulating Factor (G-CSF) was not allowed except for the secondary prevention of neutropenic fever. Erythropoiesis stimulating agents were allowed in cases of grade 2 or worse anemia and were discontinued when haemoglobin levels exceeded 12gr/dL. Pre-specified dose modifications were allowed for capecitabine and oxaliplatin for grade 3–4 haemotological adverse events and grade 2–4 non-haematological adverse events. The dose of bevacizumab was reduced by 50% in cases of uncontrollable arterial hypertension despite antihypertensive treatment and it was discontinued in cases of severe haemorrhagic events.

Patient evaluation

Pre-treatment evaluation included a detailed medical history and physical examination, a complete blood cell count (CBC) with differential and platelet count, blood chemistry, serum levels of carcinoembryonic antigen (CEA) and computed tomography scans (CT) of the chest and abdomen. Patients were clinically assessed before each cycle, CBC was performed weekly and blood chemistry every two weeks before treatment administration.

Responses were evaluated with CT scans of the chest and abdomen and by measuring CEA levels every three months. Bone scintigraphy and brain CT were performed based on clinical indication. The Response Evaluation Criteria in Solid Tumors (RECIST) were used to assess tumor responses. The duration of response was measured from the first documentation of response to disease progression. The progression free survival (PFS) was determined by the interval between the initiation of treatment and the date when disease progression was first documented or the date of death from any cause. Overall survival (OS) was measured from the date of treatment initiation to the date of death. The follow up time was measured from the day of first treatment administration to the time of the present analysis (for patients still alive) or death for deceased patients.

Toxicities were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) 3.0. All patients underwent a Comprehensive Geriatric Assessment (CGA) at study enrollment, after the 6th and after the 12th cycle. The CGA includes questionnaires of daily (ADL) and instrumental (IADL) activities, the assessment of performance status, nutritional, mental (Mini Mental State Examination) and emotional status (Geriatric Depression Scale), as well as the evaluation for the presence of geriatric syndromes, the Charlson’s comorbidities index and the evaluation of concomitant medications. Patients were classified as fit if they were independent during ADL and IADL and there were no geriatric syndromes or comorbidities. Patients were classified as vulnerable if they were dependent on some IADL but not ADL, there were minor comorbidities but no geriatric syndromes.

Statistical analysis

This multicenter single-arm prospective phase II study evaluated the efficacy and safety of the XELOX/BEV combination as front line treatment for elderly patients with metastatic colorectal cancer. The sample size calculation was conducted according to an optimal Simon two-step design testing: The null hypothesis was that the ORR is ≤ 30% versus the alternative hypothesis that the ORR is ≥ 40% (α = 0.05, power 80%). An interim analysis took place on the first 15 patients and as the number of responders was at least 6, then 31 additional patients were planned to be enrolled for a total of 46 patients.

The analysis of the primary endpoint was performed for the intent-to-treat population, defined as all patients who received at least one treatment cycle in the study. Patients who withdrew without a response assessment were classed as non-responders in the analysis. Chi-square test was used for the association of response with other dichotomous variables (patients’ age, sex, tumor stage, etc.). Binary logistic regression was carried out in order to evaluate which of the significant factors at the chi-square analysis had a significant influence on response. Cox’s proportional hazards multivariate analysis was used to evaluate which of the significant factors at the univariate analysis had a significant influence on PFS and OS. Statistical significance was set at p = 0.05.

Patient characteristics

Compliance with the treatment

A total of 409 chemotherapy cycles were administered (median 10, range: 1–16). A total of 41 cycles (10%) were delayed with a median delay of 9 days (range, 5–66 days). The reason for treatment delay was haematologic (n = 9 cycles), non-haematologic (n = 5 cycles) toxicity or both (n = 1 cycle), while in 26 cycles the delay was related with late admission or response evaluation. Dose reduction was required in 18 cycles (4.4%) because of haematological (n = 2 cycles) and non-haematological (n = 12 cycles) as well as for non-toxicity related reasons (n = 4 cycles). Three patients discontinued treatment because of treatment-related toxicity and two patients withdrew their consent. The mean dose intensity was 4700 mg/m²/week for capecitabine corresponding to the 89.5% of the preplanned protocol dose, 40 mg/m²/week for oxaliplatin corresponding to the 94.5% of the preplanned protocol dose and 2.3 mg/kg/week for bevacizumab corresponding to the 92% of preplanned dose. G-CSF was not used since none of the patients developed febrile neutropenia, while erethropoetin was administered in two patients (4%) for the treatment of grade II anemia. Twenty-three (47.9%) patients completed the treatment as per protocol, 1 patient was lost to follow-up, 2 patients underwent definitive liver metastasectomy, and 1 patient discontinued treatment because of an episode of acute cholecystitis.

Treatment efficacy

Forty-seven patients were evaluable for response to treatment. The study met its primary end point since in an intension to treat analysis, there were four (8.5%) complete [19] and 18 (38.3%) partial responses (PR), for an overall response rate of 46.8% (95% CI: 32.5%–61.1%). Thirteen (27.7%) additional patients had stable disease (SD), for a disease control rate (CR + PR + SD) of 74.4%. The median response duration was 7.1 months (range, 1.2–21.0). After a median follow-up of 15.6 months (range: 0.1–45.4 months), the median progression free survival was 7.9 months (95% CI: 5.9–9.8 months) and the median overall survival 20.1 months (95% CI: 15.6–25.7 months). The probability of 1- and 2-year survival rate were 76.7% and 42.1%, respectively.

On pre-planned subgroup analysis, response rate was found to be affected by baseline albumin levels. Of the 22 patients that had a baseline albumin measurement, all responses were observed in patients with an albumin levels >3.5 gr/dL (ORR 50%) compared to patients with an albumin level <3.5 gr/dL (ORR%; p = 0.030). Response to treatment was not affected by status according to comprehensive geriatric assessment (p = 0.634), IADL score (p = 0.920), MMSE score (p = 0.285), GDS score (p = 0.151), haemoglobin levels (p = 0.086) or body mass index (BMI, p = 0.457). PFS was affected by baseline albumin (10.2 months for albumin ≥3.5 gr/dL versus 1.1 months for albumin <3.5 gr/dL, p = 0.038) and baseline haemoglobin [20] levels (8.6 months for Hb ≥ 10.5 gr/dL versus 1.1 months for Hb < 10.5 gr/dL, p = 0.007) (Figures 1 and 2). CGA (p = 0.272), IADL (p = 0.805), MMSE (p = 0.413), GDS (p = 0.088) and BMI (p = 0.551) did not affect PFS. Finally, OS was only affected by Hb levels (17.4 months for Hb ≥ 10.5 gr/dL versus 1.1 months for Hb < 10.5 gr/dl, p = 0.001) (Figure 3). OS according to CGA status was 20.7 months for fit patients versus 17.4 months for vulnerable patients, a difference that could not reach statistical significance. These results were independent of patients’ age (71–80 vs. 81 or higher), PS, number of organs involved and Kohne prognostic index.

Toxicity