Surgery followed by adjuvant chemotherapy has been standard treatment for stage III colon cancer since 1990. However, to date, clinical trials have not been conducted to determine the definitive outer time limit by which adjuvant chemotherapy should be received for optimal survival benefit. The objective of the current study was to assess the association between the receipt/timing of adjuvant chemotherapy and patient survival in clinical practice.
Residents of Alberta who were diagnosed with stage III colon adenocarcinoma in years 2000 to 2005 who underwent surgery were included in the study. Patients were identified from the Alberta Cancer Registry and were linked to hospital data and neighborhood-level socioeconomic data from the 2001 Canadian Census. Cox proportional hazards models were used to estimate hazard ratios of death according to the timing of chemotherapy.
There were 1053 patients in the study; 648 (61%) initiated adjuvant chemotherapy within 16 weeks of surgery. There was no difference in overall survival or colon cancer-specific survival between those who received adjuvant chemotherapy from 8 to 12 weeks postsurgery compared with those who received it within 8 weeks. However, those who received chemotherapy 12 to 16 weeks after surgery and those who either received it >16 weeks after surgery or received no treatment had a 43% and 107% greater risk of dying, respectively, than those who received chemotherapy within 8 weeks of surgery (hazard ratio, 1.43 [95% confidence interval, 0.96-2.13] and hazard ratio, 2.07 [95% confidence interval, 1.56-2.76], respectively). Analyses were controlled for age, year, and region of residence at diagnosis; sex; neighborhood-level socioeconomic factors; and number of comorbidities.
In the early 1990s, guidelines were developed recommending that patients with stage III colon cancer receive chemotherapy after their surgery (ie, adjuvant chemotherapy).1 These guidelines were based on large, randomized studies conducted in the United States that demonstrated relative risk reductions of 33% for mortality and 40% for disease recurrence in patients who received adjuvant chemotherapy compared with those who underwent surgery alone.2 A large, population-based study conducted in the United States that included patients aged ≥65 years with stage III colon cancer recently reported that adjuvant chemotherapy initiated ≥3 months after surgery was associated with a 50% increase in colon cancer-specific mortality compared with initiating chemotherapy within 1 month.3 The guideline treatment for stage III colon cancer in Alberta consists of surgery followed by adjuvant chemotherapy initiated within 12 weeks from the date of surgery.4
Regardless of the benefits and survival advantage gained by adjuvant chemotherapy, several studies have demonstrated that large proportions of patients do not receive it or experience treatment delays.5-9 The objectives of the current study were 1) to quantify the proportion of patients receiving adjuvant chemotherapy within 12 weeks after surgery, 2) to identify factors associated with the receipt of timely adjuvant chemotherapy, and 3) to assess the association between the receipt and timing of initiation of adjuvant chemotherapy and survival among patients who were diagnosed with stage III colon cancer in Alberta between 2000 and 2005.
MATERIALS AND METHODS
All residents of Alberta who were diagnosed with stage III colon adenocarcinoma (International Classification of Diseases for Oncology [ICD-O] 10th edition10 codes c18 and c18.2-c18.9) between 2000 and 2005 who underwent surgery were identified from the Alberta Cancer Registry. Patients were excluded if they died within 16 weeks of their surgery, were diagnosed with another primary cancer within 6 months before or after their colon cancer diagnosis, did not have histologically confirmed disease, or were treated outside of Alberta. Cancer staging was based on the American Joint Committee on Cancer (AJCC) AJCC Cancer Staging Manual (version 6).11
Data were linked from the Alberta Cancer Registry, the Ambulatory Care Classification System (ACCS), the Discharge Abstract Database (DAD), and the 2001 Canadian census. The Alberta Cancer Registry (a member of the North American Association of Comprehensive Cancer registries) was established in 1942 and is responsible for recording and maintaining data on all cancer cases and cancer deaths that occur in Alberta; physicians and hospitals are legally required to report all cancer cases to the Alberta Cancer Registry. Patient demographics, tumor histology and stage, postal code of residence at diagnosis, initial treatment modalities and start dates, and date of death were obtained from the Alberta Cancer Registry. Date and cause of death are received and updated monthly from the provincial vital statistics office. All death certificates are reviewed by the Alberta Cancer Registry to confirm cause of death using coding rules from version 10 of the ICD for cause of death12; corrections are made to cause of death as appropriate.
The ACCS and DAD databases contain diagnosis and procedure codes on all outpatient and inpatient hospital visits in the province of Alberta. All hospital visits that occurred in the year before the patient's cancer diagnosis were used to identify comorbidities using an enhancement to the Charleson Comorbidity Index.13 Comorbidity scores were categorized into 3 groups: no serious comorbidity, 1 serious comorbidity, and ≥2 serious comorbidities.
The 2001 Canadian census was used to obtain socioeconomic indicators at the geographic level for each patient, also called the dissemination area (a neighborhood with approximately 600 households). Four variables were used as measures of the neighborhood socioeconomic status: 1) median income; 2) proportion of employment; 3) proportion separated, divorced, or widowed; and 4) proportion not graduated from high school.
Exploratory data analysis was performed to determine cutoff levels for categorical variables. Descriptive statistics were calculated for the overall cohort with respect to patient and clinical characteristics. Chi-square or Fisher exact tests, as appropriate, were used to assess associations between patient/clinical characteristics and the timing of receiving chemotherapy. The patient and clinical characteristics that were evaluated included sex, age at diagnosis, region of residence at diagnosis, neighborhood socioeconomic factors, number of comorbidities, and year of diagnosis.
Time from the date of surgery to the date of first chemotherapy session was calculated, and patients were categorized into the following 4 groups for survival analysis: received chemotherapy 1) within 8 weeks after surgery, 2) 8 to 12 weeks after surgery, 3) 12 to 16 weeks after surgery, or 4) no treatment within 16 weeks. The last group originally was separated into “received chemotherapy >16 weeks” and “never received chemotherapy.” In the final analysis, these 2 groups were combined, because the hazard ratios (HRs) for these 2 groups relative to the reference group (<8 weeks) were very similar.
Kaplan-Meier curves were used to describe the patient survival stratified by time from surgery to adjuvant chemotherapy. Cumulative incidence curves were used to describe the cumulative mortality from colon cancer-specific deaths, treating the other causes of death as competing risk.14 The Kaplan-Meier and cumulative incidence curves were started at 16 weeks after surgery. This starting time was the earliest time point that allowed us to define all “time from surgery to adjuvant chemotherapy” groups before the starting time: Deaths before this starting time were not included in the analysis.
Cox proportional hazards models were used to estimate the adjusted colon cancer-specific and overall mortality HRs according to the time from surgery to adjuvant chemotherapy (time-dependent covariate) starting at 16 weeks after surgery. The following covariates were included in the Cox regression models to adjust for relevant factors related to survival to produce adjusted HR estimates for the time-to-adjuvant-treatment variable: sex, age at diagnosis, region of residence at diagnosis, number of comorbidities, year of diagnosis, and neighborhood-level socioeconomic factors. To closely adjust for age at diagnosis, a natural cubic spline of age at diagnosis was used with 4 knots.15P values for the time to adjuvant chemotherapy and year of diagnosis were calculated based on a trend test with 1 degree of freedom. Patients were followed to the earlier of death or March 31, 2009. All statistical analyses were conducted using SAS statistical software (version 9.2; SAS Institute, Cary, NC) and R (version 2.9; R Foundation for Statistical Computing, Vienna, Austria).
There were 1211 residents of Alberta diagnosed with stage III colon cancer between 2000 and 2005. The following patients were excluded from the study: Eighty-two patients died within 16 weeks of their surgery, 52 patients were diagnosed with another cancer within 6 months before or after their diagnosis, 6 patients did not have histologic confirmation of their disease, 14 patients had a histology other than adenocarcinoma, 2 patients did not undergo surgery, and 2 patients were treated outside of Alberta. The remaining 1053 patients were included in the study.
Table 1 lists the demographic, clinical, and neighborhood characteristics of the 1053 patients who were included in the study stratified by time from surgery to the receipt of adjuvant chemotherapy. It also indicates the proportion of patients that received treatment and the proportion that died during the study follow-up. Overall, 648 patients (61%) with stage III colon cancer received adjuvant chemotherapy within 16 weeks after undergoing surgery, and 45% died during the study follow-up. The minimum time to chemotherapy after surgery was 24 days, and the maximum was 260 days. Twenty patients received adjuvant chemotherapy >16 weeks postsurgery; the primary reason for delays in this group was because of time needed to evaluate the possibility of metastases (eg, a delay in obtaining a computed tomography scan). Patients aged ≥65 years and patients with comorbidities were less likely to receive adjuvant chemotherapy. Patients who received chemotherapy >12 weeks postsurgery were more likely to live in neighborhoods with a high percentage of divorced, separated, or widowed; a low employment rate; and a low median household income. In addition, this group had more comorbidities compared with patients who received adjuvant chemotherapy within 12 weeks of surgery. The Edmonton area had the highest proportion of patients who received adjuvant chemotherapy within 12 weeks after surgery.
Table 1. Characteristics of Patients With Stage III Colon Cancer in Alberta, Canada With Respect to Receiving Adjuvant Chemotherapy and Their Mortality
Figures 1 and 2 provide the Kaplan-Meier and cumulative incidence curves for overall survival and colon cancer-specific mortality, respectively, according to the time from surgery to adjuvant chemotherapy. The risk of death was greater for patients who received adjuvant chemotherapy >12 weeks postsurgery compared with those who received it within 12 weeks postsurgery for both overall and colon cancer-specific mortality.
Table 2 provides the adjusted overall and colon cancer-specific mortality HRs and corresponding 95% confidence intervals (CIs) for patients with stage III colon cancer according to the timing of adjuvant chemotherapy. There was no difference in the overall or colon cancer-specific mortality hazard for patients who received chemotherapy 8 to 12 weeks after surgery compared with those who received it within 8 weeks after surgery. Patients who received adjuvant chemotherapy 12 to 16 weeks after surgery had a 1.43 times higher mortality hazard compared with those who received the treatment within 8 weeks (HR, 1.43; 95% CI, 0.96-2.13). The same treatment group was associated with an 18% increase in the hazard of colon cancer-specific mortality (HR, 1.18; 95% CI, 0.75-1.85), although the hazard was not statistically significant: However, the test for trend was highly statistically significant (P < .001). Patients who did not receive adjuvant chemotherapy within 16 weeks postsurgery had more than a 2-fold hazard of death compared with those who received it within 8 weeks (HR, 2.07; 95% CI, 1.56-2.76) and had a 76% increase in the hazard of colon cancer-specific mortality (HR, 1.76; 95% CI, 1.28-2.42).
Table 2. Adjusted Overall and Colon Cancer-Specific Mortality Hazard Ratios for Patients With Stage III Colon Cancera
The objectives of this study were to identify how the timing of adjuvant chemotherapy after surgery was distributed, to identify factors associated with the timely receipt of adjuvant chemotherapy, and to determine whether the receipt and timing of adjuvant chemotherapy were associated with survival in patients who were diagnosed with stage III colon cancer in Alberta after adjusting for other factors. Seventy-four percent of patients aged ≥75 years and 76% of patients with ≥2 serious comorbidities did not receive adjuvant chemotherapy within 16 weeks postsurgery. This is a surprising finding, because the current study was under a publicly funded healthcare system, and standard treatment guidelines had been in existence for 10 to 15 years at the time these patients were diagnosed. Clinical trials and observational studies have indicated that older patients and those with comorbidities benefit from adjuvant chemotherapy.16-19 Gross et al. observed that, based on 5-year survival, the benefit of adjuvant chemotherapy does not change regardless of the number of chronic health conditions in an individual patient.20 Another study that surveyed a nationally representative sample of 1000 general surgeons and 1000 oncologists in the United States, however, indicated that both types of physicians hesitate to recommend adjuvant chemotherapy to patients aged >72 years or with comorbidities who have stage III colon cancer.21 Efforts are need to address treatment disparities in these patient populations. A study similar to ours that used National Cancer Institute Surveillance, Epidemiology, and End Results cancer registry data also indicated that the receipt of chemotherapy 12 weeks after surgery is the outer limit at which the maximum survival benefit of adjuvant chemotherapy is reached.3
In Alberta, chemotherapy is organized and managed centrally; before receiving chemotherapy, patients must have a consultation with an oncologist. Treatment guidelines are created by the appropriate provincial tumor group using the best available knowledge. The treatment guidelines for patients with stage III colon cancer state that adjuvant chemotherapy should begin within 12 weeks after surgery.4 Our results are consistent with those guidelines. However, in the current study, a large portion of patients (46%) did not receive adjuvant chemotherapy or received delayed treatment. We reported this finding previously and identified the factors related to not receiving adjuvant chemotherapy.6 A significant finding was that 20% of all patients who were diagnosed with stage III colon cancer were not referred to an oncologist postsurgery; thus, potentially clinically eligible patients did not have the opportunity to receive adjuvant chemotherapy.6 In addition, in the current study, of the 20 patients who received adjuvant chemotherapy >16 weeks after surgery, the delay for 8 of those patients (40%) was because of system delays (eg, delays in obtaining postsurgical computed tomography scans to confirm the absence of metastatic disease).
Further investigations to identify the reasons for the large proportion of patients who do not receive adjuvant chemotherapy, however, still are needed to optimize patient outcomes. The physician may delay chemotherapy because of slow recovery from surgery, postsurgery complications, or possibly a change in the patient's decision to receive adjuvant chemotherapy. Alternatively, delays may occur because of inefficiencies in the healthcare system or shortages of resources to deliver care. The variation in treatment by geographic region and the findings from our initial investigations are consistent with the latter possibilities. Currently, we are investigating referral patterns to oncologists and treatment patterns of oncologists to determine whether there is practice variation that could be addressed. We also are conducting a chart review to clarify and quantify reasons why patients do not receive chemotherapy if they have a consultation with an oncologist. We are engaging the surgical and oncology communities in these initiatives to positively affect change with respect to both the receipt and the timely receipt of adjuvant chemotherapy, and we are investigating the unexpected finding of an increasing trend in adjusted mortality hazards over years.
The strengths of this study are that it is population-based and includes all patients who were diagnosed in the province of Alberta over a 6-year period with 4 to 9 years of follow-up. These strengths make the findings robust and generalizable. Only a clinical trial, however, can directly evaluate the effect timing of treatment has on survival. The intent of the study, however, was not to replace a clinical trial but, rather, to assess the association between treatment (and its timing) and survival in a real clinical practice setting and in a population-based manner. Other limitations of the study are that we did not have treatment details, such as completeness of the regimen or the specific chemotherapy regimens received. Clinical factors that were unavailable and also may affect patient survival were surgical complications, the number of positive lymph nodes, grade, and patient's functional status.
In conclusion, the results from our study are consistent with current guidelines for the treatment of stage III colon cancer in Alberta: surgery plus adjuvant chemotherapy that begins within 12 weeks after surgery. Forty-six percent of the patients who were diagnosed between 2000 and 2005, however, did not receive chemotherapy or did not receive it within 12 weeks after surgery. Efforts are needed to improve the uptake and initiation of adjuvant chemotherapy within 12 weeks after surgery to maximize the survival of patients with stage III colon cancer.
CONFLICT OF INTEREST DISCLOSURES
This research was made possible by grants received from the Canadian Institute for Health Research, Canadian Cancer Society, and the Alberta Cancer Foundation. Yutaka Yasui is supported by Alberta Innovates-Health Solutions and the Canada Research Chair Program. Isac S. F. Lima is supported by Alberta Ingenuity Center for Machine Learning.