Geriatric Factors Predict Severe Toxicity in Elderly Patients on First-line Treatment for Metastatic Colorectal Cancer 

Get Permission

[O]ur study is the first to prospectively demonstrate that geriatric characteristics are independent predictive factors of tolerance to chemotherapy and toxicity in [metastatic colorectal cancer].

—Thomas Aparicio, MD, PhD, and colleagues

Elderly patients differ with respect to tolerance of therapy for colorectal cancer. In a study reported in Journal of Clinical Oncology, Thomas Aparicio, MD, PhD, of the University of Paris, and colleagues analyzed geriatric factors for potential association with severe toxicity, dose-intensity reduction, and hospitalization among elderly patients receiving fluorouracil (5-FU)-based therapy alone or in combination with irinotecan as first-line treatment for metastatic colorectal cancer.1

This ancillary geriatric study was conducted in a subset of patients in the Fédération Francophone de Cancérologie Digestive 2001-02 trial, which was conducted exclusively in patients aged 75 years or older. The investigators found that cognitive impairment on the Mini-Mental State Examination (MMSE) and impaired autonomy on the Instrumental Activities of Daily Living (IADL) scale were associated with increased risk of severe toxicity, with cognitive impairment also being associated with increased risk of unexpected hospitalization. Data on progression-free survival, the primary endpoint of the trial, are not yet available.

Study Details

The FFCD 2001-02 trial included 282 patients who received 5-FU/leucovorin (as LV5FU2 or simplified LV5FU2; 5-FU group) or in combination with irinotecan (as LV5FU2-irinotecan or FOLFIRI; irinotecan group); irinotecan was given at 150 mg/m2 for two cycles and, in the absence of toxicity, at 180 mg/m2 in subsequent cycles. The current analysis involved 123 of these patients (44% of the total population; 62 in the 5FU group and 61 in the 5-FU/irinotecan group) who were treated at trial centers participating in the ancillary geriatric study (32 of 50 centers).

Patients completed the MMSE, which assesses cognitive function, the IADL, which assesses dependence, and the Geriatric Depression Scale. Lower scores on the MMSE and IADL indicate greater impairment, whereas lower score on the GDS indicates better mood rating. Patients with lower scores on the MMSE (≤ 27/30), IADL (≤ 7/8), and GDS (≤ 2) were compared with those with higher scores for frequency of grade 3 or 4 toxicity, dose-intensity reduction of more than 33%, and unexpected hospitalization (for any reason other than receipt of chemotherapy) over the first 4 months of treatment. Associations of other clinical and disease factors with these outcomes were also analyzed.

Overall, patients had a median age of 80 years, 54% were male, 80% had two or fewer metastatic sites, 80% had liver metastases, 85% had no prior adjuvant chemotherapy, and 75% had alkaline phosphatase ≤ 2 times upper limit of normal. Karnofsky performance status was 60 to 70 in 32% and 80 to 90 in 50%, Charlson comorbidity index was 0 to 1 in 75%, and mean score on a visual analog quality-of-life test was 60.

MMSE scores were > 27/30 in 43% of patients and ≤ 27/30 in 31%, with 26% of patients not being evaluated (completion of less than half of the items on the instrument). IADL scores were 8/8 in 37% and ≤ 7/8 in 34%, with 29% not being evaluated. GDS scores were > 2 in 10% and ≤ 2 in 63%, with 27% not being evaluated.

Tumor and geriatric variables were similar between patients in the 5-FU group and those in the irinotecan group except for greater proportions of patients in the irinotecan group with more than two metastatic sites (26% vs 11%, P = .04) and with comorbidities on the Charlson index (31% vs 16%, P = .06).

Univariate Analysis

On univariate analysis, grade 3 or 4 toxicity was significantly associated with absence of prior adjuvant therapy, irinotecan treatment group, MMSE score ≤ 27/30, and IADL score ≤ 7/8. On multivariate analysis, significant predictors of severe toxicity were irinotecan treatment group (odds ratio [OR] = 5.03, P = .006), MMSE score ≤ 27/30 (OR = 3.84, P = .019), and IADL score ≤ 7/8 (OR = 4.67, P  = .011). The presence of either lower MMSE score or lower IADL score was associated with an odds ratio of 5.43 (P = .001).

Significant predictors of dose-intensity reduction of more than 33% on univariate analysis were presence of more than two metastatic sites, alkaline phosphatase > 2 times the upper limit of normal, and irinotecan treatment group. On multivariate analysis, significant predictors were alkaline phosphatase level > 2 times the upper limit of normal (OR = 4.16, P = .047) and irinotecan treatment group (OR = 6.86, P = .004). Exploratory analyses of MMSE and IADL scores as continuous variables indicated that a linear decrease in MMSE was an independent prognostic factor (OR = 24.33 for every 10% decrease in MMSE score, P = .035).

On univariate analysis, MMSE ≤ 27/30 was a significant predictor of unexpected hospitalization and the effect of female sex approached significance. On multivariate analysis, significant predictors were MMSE ≤ 27/30 (OR = 4.56, P = .005) and, surprisingly, better mood on the GDS (≤ 2; OR = 5.52, P = .035).

Analysis by Treatment Group

When the predictive effects of MMSE and IADL for severe toxicity and the effect of MMSE for hospitalization were analyzed by treatment group, there were no significant statistical interactions, indicating similar predictable effects in both groups. However, there were numerical differences between groups with regard to these risks.

Among patients with MMSE ≤ 27/30, 89% in the irinotecan group vs 50% in the 5-FU group had grade 3 or 4 toxicity; among those with MMSE > 27/30, 58% vs 30% had severe toxicity. Among patients with impaired IADL, 89% in the irinotecan group vs 55% in the 5-FU group had severe toxicity; among those with normal IADL, severe toxicity occurred in 56% vs 23%.

In patients with impairment on both MMSE and IADL, severe toxicity occurred in 89% of the irinotecan group vs 50% of the 5-FU group, and in those without impairment on both measures, severe toxicity occurred in 44% vs 13%. Unexpected hospitalization occurred in 61% of 5-FU patients vs 67% of irinotecan patients with MMSE ≤ 27/30 and in 35% vs 36% of those with MMSE > 27/30.

The authors noted that nearly 90% of patients with impaired cognitive function and autonomy treated with irinotecan experienced severe toxicity. The data also suggested that patients without impairment in these domains can benefit from the addition of irinotecan without having toxicity risk greater than that observed in patients with impairment in both domains treated with 5-FU alone.

Cautionary Notes

With regard to the finding that better mood on the GDS was significantly predictive of unexpected hospitalization, the investigators stated that the unexpected result suggests that the GDS score should be used with caution in the metastatic setting. Consistent with other findings, depression was not associated with toxicity in the current analysis.

The authors also noted that the trial was not powered for the geriatric study, and that some of the findings in the study may have been affected by the inadequate statistical power.

The authors concluded, “[O]ur study is the first to prospectively demonstrate that geriatric characteristics are independent predictive factors of tolerance to chemotherapy and toxicity in [metastatic colorectal cancer]. Intensive chemotherapy should be used with caution in patients who have cognitive impairment or dependency. Larger studies are needed to confirm our results.” ■

Disclosure: For full disclosures of the study authors, visit


1. Aparicio T, Jouve J-L, Teillet L, et al: Geriatric factors predict chemotherapy feasibility: Ancillary results of FFCD 2002-02 phase III study in first-line chemotherapy for metastatic colorectal cancer in elderly patients. J Clin Oncol 31:1464-1470, 2013.

Related Articles

Commentary: Improving the Care and Study of Older Patients with Cancer

Predicting toxicity in older patients receiving chemotherapy is an important clinical problem. It has been shown that standard oncology measures such as performance status are not adequate to predict outcomes in the elderly. Clinical measures that are important in geriatric oncology are not...