From January 2001 to December 2011, 399 consecutive patients with histologically proven mCRC received first-line chemotherapy at the Centre Georges-François Leclerc (Dijon, France), and were prospectively recorded in an institutional clinical database. This study was approved by our institutional review board and all data were rendered anonymous.

All patients were followed up until death or the end of data recording (December 31, 2011). The primary endpoint was OS, defined as the time from date of diagnosis until death of any cause, or December 31, 2011, whichever occurred first. Survivors were censored at last follow-up. Median follow-up with its 95%CI was calculated using the reverse Kaplan-Meier method. The association between hematological toxicity (neutropenia, thrombocytopenia, or anemia) and patient or disease characteristics were examined using the χ² test or Fisher’s exact test for qualitative variables, and the Student t or Mann-Whitney test for continuous variables, as appropriate. Hematological toxicities were only considered if they occurred during the first two lines of chemotherapy, because data obtained thereafter could be biased due to inherent longer survival of these patients, and the classical problem of immortal time bias[16].

Survival probabilities were estimated using the Kaplan-Meier method and survival curves were compared using the log-rank test. Hazard ratio (HR) and 95%CI for univariate and multivariate analyses of OS were estimated using Cox’s proportional hazards regression with a backward elimination procedure. All predictors with P < 0.10 by univariate analysis were retained in the multivariate models. To prevent colinearity, when two variables were significantly correlated, only the more informative of the two was retained according to its clinical relevance, or according to the value of the likelihood ratio. Variables included in uni- and multivariate analyses were: neutropenia, thrombocytopenia and anemia during the two first lines of chemotherapy (occurrence or no occurrence), location of the primary tumor (colon or rectum), bevacizumab use, anti-EGFR use, number of metastatic sites (1 vs≥ 2), age (< 75, vs≥ 75 years), sex, serum carcinoembryonic antigen (CEA) (< 200 vs≥ 200 ng/mL), leukocyte count, serum level of alkaline phosphatase (ALP) (< 300 vs≥ 300 IU/L) and lactate dehydrogenase (LDH) (median value used as cutoff value), WHO performance status (0 vs 1-4), primary tumor surgery, metastases resection, number of chemotherapy lines (1 or 2 vs≥ 3) and molecules (1-3). Multivariate Cox models were constructed with all predictors with P < 0.10 in univariate analysis, carried out by the backward elimination procedure. To handle missing data, we performed Cox regression using multiple imputations[17].

Routine blood counts were taken during every chemotherapy cycle, usually the day before treatment. Hematologic toxicity was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0 [in detail anemia grade 1 (hemoglobin; Hb < 10 g/dL), grade 2 (Hb: 8-10 g/dL), grade 3 (Hb < 8 g/dL), grade 4 (life-threatening consequences); thrombocytopenia grade 1 (150000-75000/mm³), grade 2 (75000-50000/mm³), grade 3 (50000-10000/mm³), grade 4 (< 10000/mm³); neutropenia grade 1 (2000-1500/mm³), grade 2 (1500-1000/mm³), grade 3 (1000-500/mm³), grade 4 (< 500/mm³)]. For each type of cytopenia, patients were divided into three categories: absent (grade 0), mild (grades 1-2), and severe (grades 3-4).

Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, NC, United States). All tests were two sided, and P < 0.05 was considered statistically significant.

The median follow-up at the time of this analysis was 6.3 years (95%CI: 5.6-7.4 years). Patient and tumor characteristics are described in Table 1. Treatment characteristics are presented in Table 2. Most patients (88%) received more than two lines of chemotherapy. Only 3.5% received one line of chemotherapy (fluropyrimidine alone), 13.5% received only fluoropyrimidine and oxaliplatinum, and 4.5% received only fluorpyrimidine and irinotecan. In total, 78.5% received the three major drugs (oxaliplatin, irinotecan and fluoropyrimidine). The median number of chemotherapy lines was four (range: 1-8). Seventy-two percent of patients received targeted therapy (35% received only anti-EGFR, 40% received only bevacizumab, and 25% received both). Thirty-eight percent of the patients had metachronous and 62% had synchronous disease. We noted that 83% of patients had liver metastases, 35% had lung metastases, and 12% had peritoneal carcinomatosis. Although 60% of patients had only one tumor, most of these had diffuse liver or lung involvement and less than half could benefit from metastases surgery.

In the whole population, neutropenia occurred in 71.2% of the patients: 36.1% experienced grade 1-2 (mild) neutropenia, and 35.1% had grade 3-4 (severe) neutropenia during the first two lines of chemotherapy. Most patients (91.7%) developed anemia. Among these, 71.3% developed grade 1-2 anemia and only 20.4% had grade 3-4 anemia during the first two lines of chemotherapy. Comparatively few patients developed thrombocytopenia during treatment (58.8%). Grade 1-2 thrombocytopenia occurred in 50.7% of patients, and only 8.1% of patients experienced grade 3-4 thrombocytopenia. Most classical prognostic variables did not differ significantly between patients who had hematological toxicity and those who did not (Table 1). However, patients who experienced thrombocytopenia were significantly younger than those who did not (P < 0.001). Similarly, patients who experienced neutropenia were more frequently younger (P < 0.0001), had better performance status (P = 0.024), and more frequently had low alkaline phosphatase serum level at diagnosis (P = 0.023), compared with those without neutropenia (Table 1). Figure 1 shows the worst grade of neutropenia, anemia and thrombocytopenia recorded at each cycle of chemotherapy during the two first lines of treatment in the 399 patients analyzed. Table 2 represents hematological toxicity in function of treatment received. Except for neutropenia, the use of anti-EGFR or bevacizumab was not associated with significantly higher hematological toxicity. Neutropenia, but not thrombocytopenia or anemia, was more frequently experienced by patients who received all three major drugs (P < 0.0001, P = 0.053, P = 0.3), as compared to patients who received one or two drugs, or fewer than two lines of chemotherapy (Table 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Worst grade of hematological toxicity (neutropenia, anemia, and thrombocytopenia) recorded during the first two lines of chemotherapy in metastatic colorectal cancer patients (n = 399).

At the time of analysis, 311 patients had died (77.9%). Patients who experienced at least one episode of neutropenia had significantly better outcome compared to those with no neutropenia in terms of OS (log-rank test P < 0.0001) (Figure 2A). Similarly, patients who experienced thrombocytopenia also had significantly better OS compared to patients that did not present with thrombocytopenia during chemotherapy (log-rank test P = 0.008) (Figure 2B). In contrast, occurrence of anemia was associated with poorer outcome (log-rank test P = 0.038) (Figure 2C). Mild (grade 1-2) or severe (grade 3-4) toxicity had the same prognostic role for each hematological toxicity. Two hundred and four patients received FOLFOX (5-fluorouracil plus oxaliplatin combination) alone or with targeted therapies; 113 patients received FOLFIRI (5-fluorouracil plus irinotecan combination) alone or with targeted therapies; and 82 patients received monotherapy by fluoropyrimidine alone or with targeted therapies as first-line treatment. We observed that neutropenia was a predictor of better outcome in patients that received FOLFOX alone or with targeted therapies (mean OS of 27 mo in patients without neutropenia vs 47 mo in patients with neutropenia, P = 0.004), or in patients that received FOLFIRI alone or with targeted therapies as first-line therapy (mean OS of 28 mo in patients without neutropenia vs 41 mo in patients with neutropenia, P = 0.04). Similar results were observed in patients treated with fluoropyrimidine alone or with targeted therapies as first-line therapy (mean OS of 22 mo in patients without neutropenia vs 34 mo in patients with neutropenia, P = 0.0005).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Kaplan-Meier curves for overall survival stratified according to the occurrence or not of neutropenia (A), thrombocytopenia (panel B), and anemia (C) recorded during the first two lines of chemotherapy in metastatic colorectal cancer patients (n = 399). P values were calculated using the log-rank test.

Univariate analysis of factors associated with OS indicated that among the classical prognostic factors, age > 75 years, primitive colon (vs rectal) tumor, poor WHO status, CEA level > 200 ng/mL, multiple metastatic sites, initial anemia, high ALP or LDH serum levels, and high leukocyte count were also significantly associated with poorer OS (Table 3). As regards treatment-related variables, bevacizumab use, anti-EGFR use, resection of the primary tumor, resection of metastases, and use of the three major cytotoxic drugs were also significantly associated with better survival (Table 3).

Multivariate analysis revealed that poor WHO status (P = 0.021), anemia during the first two chemotherapy lines (P = 0.029), multiple metastases locations (P = 0.026), and high serum levels of ALP (P = 0.007), or LDH (P < 0.001), were independently associated with poorer OS. On the contrary, resection of the primary tumor (P < 0.001), resection of metastases (P < 0.001), occurrence of thrombocytopenia (P = 0.036) and neutropenia (P < 0.001) were all independently associated with better OS (Table 3). Thrombocytopenia and neutropenia were better correlated (r = 0.35) than anemia and neutropenia (r = 0.03), or anemia and thrombocytopenia (r = 0.04).

Analysis with multiple imputations confirmed that neutropenia during chemotherapy for mCRC was associated with increased survival (P = 0.004), while thrombocytopenia was no longer significant (P = 0.079). To avoid the bias of longer survival in patients with hematological toxicity, a landmark analysis using a time point of 6 mo from the diagnosis was performed because > 90% of patients developed hematological toxicity during the first 6 mo follow-up[9]. Patients who died before the landmark time were excluded from the analysis regardless of the presence or absence of hematological toxicity. Thirty-five patients were excluded from the analysis because they died before the landmark time. Using such analysis, univariate and multivariate analyses confirm that neutropenia and thrombocytopenia during chemotherapy for mCRC were associated with increased survival (data not shown).

During metastatic colorectal cancer, cytotoxic agents are the major anticancer agents. Currently, the dosage schedule of chemotherapy remains empirical. Further research is warranted to define better the optimal schedules.

Usually, the recommended dosage of chemotherapy is determined on the basis of dose-finding phase I trials. Such studies determine the toxic profiles of cytotoxic agents in a small number of patients, but do not take into account the inter-individual variability of drug metabolism. Some reports suggests that hematological toxicity may reflect the bioactive dosage of chemotherapy.

Neutropenia and thrombocytopenia occurring during the first two lines of chemotherapy were significantly associated with better overall survival (OS) (HR = 0.55, 95%CI: 0.43-0.70, P < 0.0001 and HR = 0.70, 95%CI: 0.56-0.88, P = 0.025 respectively). In contrast, anemia during chemotherapy was significantly associated with poorer OS (HR = 1.9, 95%CI: 1.22-2.97, P = 0.005).

This study underlines the possible prognostic role of neutropenia and thrombocytopenia occurring during chemotherapy for metastatic colorectal cancer (mCRC).

Hematological toxicity, namely neutropenia, anemia and thrombocytopenia, was graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 3.0. For each type of cytopenia, patients were divided into three categories: absent (grade 0), mild (grades 1-2), and severe (grades 3-4).

This is a well-written, interesting article showing an association between neutropenia and thrombocytopenia occurring during chemotherapy and better survival for mCRC. It would be of interest to know where the metastases were, if they were metachronous or synchronous, why 60% of the patients had only one metastatic site, and only 25% had surgery for the metastases.