Adjuvant treatments do not alter the quality of life in elderly patients with colorectal cancer

A population-based study




The current study was performed to longitudinally assess the impact of adjuvant treatments on the quality of life (QoL) of elderly colorectal cancer survivors.


The Burgundy Digestive Cancer Registry was used to select all patients aged ≥75 years who were diagnosed with colorectal cancer between 2003 and 2005. A total of 209 patients were asked to complete questionnaires during the first year after diagnosis: at the time of inclusion in the study (Q0), at 3 months after the initial diagnosis (Q3), at 6 months after the initial diagnosis (Q6), and at 12 months after the initial diagnosis (Q12) using the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30. A total of 125 patients (60%) responded. Mixed model analyses of variance for repeated measurement were used to compare QoL scores according to therapeutic schemes. Interactions between time of follow-up and treatment were tested.


Patient sex, age, location of the tumor, and TNM stage of disease did not appear to differ significantly between respondents and nonrespondents. Global Health and Emotional Functioning improved for colon cancer survivors between Q0 and Q12, and were noted to improve between Q3 and Q12 for rectal cancer patients. According to French recommendations, patients who received chemotherapy for stage III colon cancer (P = .176) or radiotherapy for rectal cancer (P = .959) reported no significant changes in Global Health compared with those patients not receiving adjuvant therapies. Patients treated with chemotherapy reported better Physical Functioning than patients who did not received chemotherapy (P = .0113).


To the authors' knowledge, the current study is the first to examine trends over time with regard to the influence of adjuvant treatments for colon and rectal cancers on QoL in a general aged population. Providing evidence that adjuvant chemotherapy for colon cancer has no negative impact on the QoL of elderly patients is of great significance in encouraging clinicians to treat this population. Cancer 2008. © 2008 American Cancer Society.

Colorectal cancer is frequently encountered in the elderly; approximately 60% to 70% of cases occur in subjects aged ≥65 years, and approximately 45% of cases are diagnosed in patients aged >75 years.1 Improvement in life expectancy plus the increased size of the elderly population have led to a growing number of affected patients. Over the past 25 years, important advances have occurred in the management of colorectal cancer. Although surgery remains the primary treatment, adjuvant treatments have been developed with the aim of reducing disease recurrence and improving survival.2, 3 Many studies have emphasized that too few elderly patients have received chemotherapy or radiotherapy despite their demonstrated effectiveness.4–7 Relative survival models, by disentangling the impact of life expectancy and coexisting pathologies from that of cancer itself, have demonstrated that age has a modest effect on prognosis, suggesting that patients should not be denied treatment for cure because of age alone.8 Beyond survival, health-related quality of life (HRQoL) is now considered to be an important outcome assessment of cancer management, particularly for the elderly. It most likely could be a determining factor in refuting the misconception that the elderly are more prone to side effects and that a patient's advanced age limits the potential benefit of treatment.

The Burgundy Digestive Cancer Registry, which records accurate data regarding the treatment and prognosis of colorectal cancer at the population level, represents a valuable resource for assessing HRQoL in targeted patients. To our knowledge, there is a paucity of reports of this type in published literature, with data being for the most part provided by specialized hospital, university, or clinical trials. As such these data cannot be used as a reference because of unavoidable selection bias.

In the current study, we report the results of a longitudinal survey of HRQoL during the first year after diagnosis of colorectal carcinoma in this nonselected, community-based series of patients aged ≥75 years. The objective was to compare, after curative resection, the HRQoL of colon and rectal cancer survivors who were treated with adjuvant therapies according to the French Consensus Conference with that of survivors who did not receive adjuvant therapies.



Subjects living in a geographic area of Burgundy (Saône-et-Loire area, comprised of approximately 550,000 inhabitants) were selected and considered to be eligible if they were aged ≥75 years at the time of diagnosis and if a colon or rectal carcinoma was diagnosed for the first time between March 2003 and September 2005 (ie, 2.5 years of registration). Patients were identified each week from all pathology laboratories located in or bordering the geographic area. All colonoscopies with a biopsy that was positive for colorectal carcinoma were reported. All pathology laboratories within the area agreed to report their cases regularly to the registry. This network of laboratories has been in existence since the creation of the registry in 1982 and trained to deliver an exhaustive notification of cases. Despite this long experience of working together, we tested this new notification procedure during a 6-month time period between September 2002 and February 2003. As soon as the patient was identified, a physician affiliated with the registry contacted the patient's personal physician so that the patient could be informed about the special survey and asked to participate. Data privacy regulations do not allow the registry to contact the patients directly.

HRQoL Questionnaires: Studied Variables

QoL was measured using the European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 (version 3.0). The QLQ-C30 is comprised of 30 items with both multi-item scales and single-item measurements. The internal validation of the QLQ-C30 allowed 15 scales to be identified and 15 scores to be generated (Role Functioning, Physical Functioning [PF], Cognitive Functioning, Emotional Functioning [EF], and Social Functioning), a Global Health (GH) status scale, a Financial Difficulties scale, and 8 scores for Symptoms (fatigue, nausea/vomiting, pain, dyspnea, insomnia, appetite loss, constipation, and diarrhea). These scores vary from 0 (worst) to 100 (best) for the Functional and Global Health scales and from 0 (best) to 100 (worst) for Symptoms scales.

The EORTC QLQ-C30 questionnaires were given directly to the patient by their personal physician when possible, or were otherwise sent by mail to their home. Completed questionnaires were returned by mail to the registry. The protocol planned to process the questionnaire using assessment in 4 time-scheduled stages: 1) the baseline form completed at the time of diagnosis (ie, before surgery or before the first cancer treatment if no surgery was performed) (Q0); 2) during the third month after the initial diagnosis (Q3); 3) during the sixth month after the initial diagnosis (Q6); and 4) during the 12th month after the initial diagnosis (Q12). The HRQoL scores were computed according to the EORTC manual.9

In addition to the EORTC QLQ-C30 questionnaires, a Spitzer QoL Index was also completed by each personal physician at the same 4 time stages. The Spitzer questionnaire is comprised of a succinct, quantitative, cancer-specific index designed for use by physicians to measure the QoL of patients. It is a validated tool adapted to French sociologic conditions and uses a clinician assessment by proxy.10, 11

Relevant clinical data concerning patients and cancers were obtained by the staff of the registry according to the usual procedure. Variables registered were comprised of patient age, sex, and site of the cancer according to the International Classification of Diseases for Oncology (ICD-10),12 including the colon (C18) and rectum (C19 and C20). Anal cancer cases were not included. Curative resection was defined as the macroscopic resection of all malignant tissue with negative surgical margins. Tumor extension at the time of diagnosis was classified according to the TNM classification.13

Two trials published in 1989 and 1990 demonstrated the effectiveness of chemotherapy for stage III colon cancers (determined according to the TNM staging system).3, 14 As a result of these publications and other trials published more recently, the French Consensus Conference recommended adjuvant chemotherapy for patients with stage III colon cancer.15 The same recommendations have been made for palliative chemotherapy for patients with advanced colorectal cancer. The effect of adjuvant radiotherapy on rectal cancer recurrence and possibly on survival has been well established,2, 16 as well as the superiority of preoperative radiotherapy compared with postoperative radiotherapy,17, 18 All patients with stage III colon cancer underwent curative surgical resection (N = 46). Among the 30 respondents, 11 received adjuvant chemotherapy (ie, within 6 months after surgery). The chemotherapy was comprised of 5-fluorouracil and leucovorin in 8 patients, oxaliplatin plus 5-fluorouracil/leucovorin in 2 patients, and capecitabine in 1 patient. Among the 66 patients with rectal cancer, 34 responded to QoL questionnaires. Twenty-three of these patients underwent curative surgical resection, 11 of whom received preoperative radiotherapy (45 grays [Gy] in 25 fractions over 5 weeks).

Statistical Analysis

With regard to the management of colorectal cancer in elderly patients, we focused on the following 5 QoL scores: GH, PF, EF, Fatigue, and Pain. A 5-point difference was considered to be the minimally clinically significant difference.19 Continuous variables and QoL scores were described using the mean (standard deviation [SD]). We described QoL scores at time of diagnosis (Q0) and at each follow-up period.

We first used a logistic regression model to determine whether respondents differed from nonrespondents. To assess the evolution of QoL according to the type of treatment received, we performed mixed model analysis of variance for repeated measurement for each score.20 It provided a test for both the effect of treatment and the effect of follow-up time independently. This provided a test for both the independent effect of treatment and the independent effect of follow-up time on QoL. A P value <.05 for treatment effect reflected a significant difference in the corresponding score according to treatment. A P value <.05 for follow-up time reflected a significant evolution over time of the corresponding score. A Pvalue <.05 for the interaction between follow-up time and treatment reflected a difference according to treatment in the evolution of the score over time. This interaction was systematically tested and was never found to be significant. After testing a covariance matrix using 3 different structures (unstructured, compound symmetry, and autoregressive), the best model was retained according to the Akaike Information Criterion. Because incompleteness of data was found to be a major issue in QoL studies, we used pattern mixture models21–23 to test the impact of adjusting for monotonous or intermittent missing data on the models likelihood, but none was found to be significant. All analyses were performed using SAS V (SAS Institute, Inc, Cary, NC) or Stata V9 (StataCorp, College Station, Tex) software. A P value ≤.05 was considered to be statistically significant.



Overall, 233 patients aged ≥75 years with colon or rectal cancer were eligible and reported by the pathology laboratory. Twenty-four patients died during the first 3 months after diagnosis and were excluded from the study. A total of 128 patients returned at least 1 questionnaire. Among these, 3 patients did not answer any items on any EORTC QLQ-C30 questionnaires; therefore, the physician's Spitzer index was the only available questionnaire. These patients were therefore excluded. The 125 remaining patients (60%) responded to at least 1 of the 4 awaited HRQoL questionnaires. Among them, 35 patients, 24 patients, 25 patients, and 41 patients, respectively, replied to 1, 2, 3, and 4 questionnaires. It was possible to compute a HRQoL score for each patient so that they were considered as respondents for this study. Stated reasons for declining to participate were either related to the assessment of the personal physician before the patient was asked to participate (dementia, incapability of completing the questionnaire, or death [N = 38]) or related to the patient after being asked (lack of interest or difficulty with writing [N = 43]). The univariate comparison of the characteristics of respondents and nonrespondents is shown in Table 1. Sex, age at diagnosis, location of the tumor, and stage of disease at the time of diagnosis did not appear to differ significantly between the 2 groups. Among patients who underwent curative surgical resection, the response rate was higher than the nonresponse rate (57.5% vs 42.3%; P = .039). Using a logistic regression model adjusting for sex and localization, surgical outcomes were found to be significantly correlated with the likelihood of response (ie, patients who underwent surgical resection for cure were more often respondents, as were patients who underwent palliative surgical procedures; P = .024).

Table 1. Epidemiologic Characteristics of Patients According to Their Questionnaire Participation
  • *

    Patients who completed at least 1 of the 4 questionnaires were considered to be respondents.

  • Determined using the chi-square test.

  • Tumors were not resected without patent visceral metastases; stage of disease was unknown in 2 cases (both colon cancers).

Age, y
Stage of disease at diagnosis
 Curative resection7742.310557.7.039
Palliative resection1768.0832.0
No resection1453.81246.1
Adjuvant treatments according to French recommendations
Chemotherapy for patients with stage III colon cancer who underwent curative resection
Radiotherapy for patients with rectal cancer

The proportion of patients receiving adjuvant treatment is shown in Table 1. The proportion of patients receiving either chemotherapy for stage III colon cancer or radiotherapy for rectal cancer did not vary significantly between respondents and nonrespondents.

Descriptions of QoL Scores

Figure 1 shows scale score and symptom item distributions according to colon and rectal cancers at Q0. At the time of diagnosis, there was no significant difference noted between tumor locations for studied scores with the exception of that for pain, which tended to be initially higher for patients with rectal cancer compared with patients with colon cancer (P = .394). Variations over time for different subscales for respondents according to the tumor location are reported in Figure 2. The mean scores for GH and EF increased for patients with colon cancer between Q0 and Q12, and between Q3 and Q12 for those with rectal cancer, whereas the mean scores for symptoms such as fatigue and pain symmetrically tended to decrease but only after the third month for patients with rectal cancer. PF remained stable for colon cancer patients whereas it increased beyond Q3 for those with rectal cancer.

Figure 1.

Mean quality of life scores for studied subscales at the time of diagnosis (Questionnaire 0 [Q0]).

Figure 2.

(a) Variations over time in different subscales for respondents for colon cancer. (b) Variations over time in different subscales for respondents for rectal cancer.

QoL Scores According to Adjuvant or Neoadjuvant Treatments

Table 2 presents the analyses of scores over time according to treatment for patients who underwent curative surgical resection.

Table 2. Quality of Life Over Time According to Adjuvant or Neoadjuvant Treatment: A Multivariate, Repeated Measurement Analysis of Variance Performed for Each Score
  Chemotherapy for Stage III Colon CancerRadiotherapy for Rectal Cancer
  P*  P*
  • CT− indicates no chemotherapy; RT−, no radiotherapy; CT+, adjuvant chemotherapy; RT+, neoradiotherapy; Q0, questionnaire at inclusion in the study; Q3, questionnaire at 3 months after the initial diagnosis; Q6, questionnaire at 6 months after the initial diagnosis; Q12, questionnaire at 12 months after the initial diagnosis.

  • *

    Repeated analysis of variance measurement (P value for effect of treatment).

  • Interpretation using the example of physical functioning (PF): in patients with stage III colon cancer, there was no significant evolution of PF noted over time (P = .747). PF was on average 22.1 points higher in patients receiving chemotherapy compared with those not receiving chemotherapy (P = .011).

Global Health
  CT− RT− 
Time from diagnosisQ0    
Q3 8.60  −0.08 
Q6 8.37  13.47 
Q12 12.18.1531 5.98.0342
Physical functioning
  CT− RT− 
Time from diagnosisQ0    
Q3 4.36  −4.19 
Q6 5.41  11.14 
Q12 1.99.7474 2.37.0208
Emotional functioning
  CT− RT− 
Time from diagnosisQ0    
Q3 12.91  −4.70 
Q6 19.89  1.58 
Q12 17.84<.001 1.84.5877
  CT− RT− 
Time from diagnosisQ0    
Q3 −1.48  −1.11 
Q6 −7.64  −15.25 
Q12 −8.47.3940 −7.15.1163
  CT− RT− 
Time from diagnosisQ0    
Q3 −3.72  2.40 
Q6 4.57  −9.25 
Q12 −2.02.4707 −6.62.4936

GH for patients with stage III colon cancer treated with adjuvant chemotherapy did not appear to vary significantly from that of patients who did not receive chemotherapy (P = .1756) (Fig. 3a). The overall mean score for PF was found to be significantly higher for patients treated with chemotherapy than for untreated patients, regardless of follow-up (P = .0113), and even without any significant evolution noted over time in the corresponding scores (Table 2). A mean difference equal to 22 points could be considered clinically significant. The mean scores for EF were found to significantly increase between Q0 and Q6 (P <.0010) and then did not appear to vary much over time. Mean scores for fatigue and pain also did not vary over time.

Figure 3.

(a) Global health (GH) among patients with stage III colon cancer aged ≥75 years who either did or did not receive adjuvant chemotherapy. (b) GH among patients with resected rectal cancer aged ≥75 years who were either treated or not treated with preoperative radiotherapy.

GH for patients with rectal cancer receiving neoadjuvant radiotherapy did not appear to vary significantly from that of patients who did not receive radiotherapy (P = .9586) (Fig. 3b). The mean scores for EF, fatigue, and pain did not vary over time whereas those for PF significantly increased between Q3 and Q6 (P = .0208).


Many studies concerning QoL measurement for patients with colorectal cancer have been published to date, but to our knowledge there is a paucity of reports concerning the impact of adjuvant treatments on HRQoL for elderly patients, and none at a population level. To our knowledge, the current study is the first to examine the influence of adjuvant chemotherapy and radiotherapy for colon and rectal cancers on the HRQoL in a general population of patients aged ≥75 years. The EORTC QLQ-C30 questionnaire is considered to be valid and reliable for colorectal cancer.24 In the current study, the focus was on 5 functional and symptoms scales that appeared to be relevant to the elderly.

The main value of the current study was to demonstrate that after potentially curative surgery for colon or rectal cancer, elderly patients who received adjuvant treatments reported no significantly different HRQoL than those not receiving adjuvant therapies. On the 1 hand, such results were unexpected because elderly patients receiving ‘aggressive’ cancer treatments could develop toxicities that negatively impact their perceived QoL. Conversely, it has already been reported that chemotherapy with related toxicities could improve QoL because of health and treatment expectations.25 From the patient's point of view, toxicity could be acceptable because it reflects the efficacy of the treatment.

One of the main strengths of the current study was the high overall response rate, especially for a longitudinal study with 1-year time iterative questionnaires targeting an elderly population. In circumscribing the population to patients aged ≥75 years who were considered suitable for treatment according to French recommendations (ie, adjuvant chemotherapy for stage III colon cancer and adjuvant radiotherapy for rectal cancer), we limited selection bias because of the heterogeneity of the studied patients. Therefore, in excluding the most severely ill patients, there was no significant difference noted, particularly with regard to age and stage of disease, between respondents and nonrespondents. Nevertheless, HRQoL did appear to influence the likelihood of responding to the questionnaire and we cannot exclude the possibility that nonrespondents may have a lower HRQoL than respondents.

After the effectiveness of adjuvant chemotherapy for stage III colon cancer was reported in 1990, there was a dramatic increase in the use of this treatment.3, 14 Nevertheless, data from Europe suggest that in the year 2000, only 20% to 25% of elderly patients received adjuvant chemotherapy.5, 26, 27 When studied, older patients have indicated that the primary determinant of their decision regarding chemotherapy was their physician's advice.28 The physician's attitude may explain the low utilization of adjuvant chemotherapy. Providing evidence in a longitudinal representative study that adjuvant chemotherapy for colon cancer has no negative impact on the HRQoL of elderly patients is of great importance in encouraging clinicians to treat this population. In the current study, the treated group reported having a GH that was similar to that of the untreated group. Surprisingly, the global PF score was higher for treated patients compared with untreated patients, and did not appear to significantly change during the first year after surgery. This result could partly reflect a clinical selection bias, in which chemotherapy tends to be offered to the more healthy patients. It could also be explained by the ‘response shift’ phenomenon, which leads patients faced with cancer and its strong treatment to adopt a more pragmatic attitude regarding their everyday life experiences and therefore to reduce their initial expectations regarding QoL and health, resulting in improvement of perceived QoL.29 Neither global EF, fatigue, nor pain were found to vary significantly between the 2 groups. When compared with the baseline values, the EF score improved over the 12 postoperative months for all patients. This may be linked to a gradual attenuation of the ‘announcement’ effect of cancer, adaptation around cancer, and perhaps a positive effect of patients' contact with the healthcare system.

In what to our knowledge are the few available population-based data regarding the practice of radiotherapy in the elderly published to date, it would appear elderly patients are being undertreated compared with younger patients (35% vs 79% in Burgundy over the period 1994–1996).4 To our knowledge, little is known regarding the burden of radiotherapy for rectal cancer on HRQoL, with the majority of studies having focused predominantly on younger patients, sexual disorders, or surgical rectal resection with or without a stoma. The results of the current study failed to demonstrate a decreasing global HRQoL for elderly patients receiving radiotherapy. The small number of cases did not allow for the presence or absence of a stoma to be taken into account in analyses. Nevertheless, there was no significant difference noted in the proportion of stomas between patients who received radiotherapy and those who did not. To the best of our knowledge, there are no available data in the literature with which to compare our results. A German study conducted with younger patients reported a significant decrease in PF for patients after surgery, but this study mixed patients with both early and advanced stages of disease.30 In particular, the need for transportation between the home and the few radiotherapy centers does not appear to be perceived as decreasing the HRQoL for elderly patients.

The small number of patients in the current study warrants caution in interpreting the results. Yet we think that the lack of HRQOL differences is not due to inadequate statistical power. GH was found to be similar in patients receiving radiotherapy and in patients not receiving radiotherapy over the follow-up period. Patients receiving chemotherapy were found to start with better GH (which may be in part an indication bias), but this difference remained the same over the follow-up period. This finding reinforces the idea that the lack of significant interaction between the administration of chemotherapy and the follow-up period is not because of a lack of power.

It is well known that as many patients as possible should be included in therapeutic trials. The results of the current study encourage the inclusion of the elderly into treatment studies for colorectal carcinoma.


We thank all private and public pathologists who agreed to collaborate in the rapid designation of incident cases (Dr. Pasquiou-Mège, Dr. Morel, and Dr. Jacquot). We also thank E. Gauthier, S. Normand, and C. Deville for their contribution to this work