br The purpose of this study is to address this
The purpose of this study is to address this knowledge gap by thor-oughly examining causes of death in older adults with colon cancer who are assessed for consideration of adjuvant chemotherapy. We aimed to compare causes of death, including proportion of colon can-cer-related deaths, in younger and older patients and between patients who received chemotherapy and those who did not. We specifically fo-cused on patients 50 and older, given decision-making in patients youn-ger that 50 is largely unaffected by life expectancy considerations. We also aimed to assess for differences in causes of death related to colon cancer versus non colon cancer deaths, and between patients who re-ceived at least one 122320-73-4 of chemotherapy and those who did not. We performed exploratory analyses in older patients to examine predictors of mortality.
2. Materials and Methods
The study was approved by the institutional Research Ethics Board. The data were collected from the British Columbia Cancer Agency (BCCA) registry and Gastrointestinal Cancer Outcomes Unit (GICOU). During the study timeframe, the BCCA was comprised of 6 regional can-cer centers that cover different geographical catchment areas in the province. As such, the BCCA aims to provide a province-wide, popula-tion-based cancer control program to approximately 5 million residents in British Columbia. Approximately 15,000 to 20,000 new patients with cancer are referred annually for management. The GICOU collects clini-cal, pathologic and vital statistics data on patients referred to the BCCA with colorectal cancer. Date of first local, regional, and distant recur-rence is also captured within the database. The primary source of death data, including date and cause of death, is from British Columbia (BC) Vital Statistics. The GICOU supplements this information when in-formation is found that either contradicts or has not been received from BC Vital Statistics.
In addition to data collected in the database, chart review was per-formed to confirm cause of death. For patients who had been coded in the database as having died from colon cancer, but there had been no documented recurrence, charts were reviewed by a second, indepen-dent reviewer. Similarly, an independent review of cases in which a re-currence of colon cancer had been documented but cause of death was non-colon cancer was also conducted. If there were discrepancies be-tween reviewers or inadequate information was available, cause of death was coded as “unknown”. Manual chart review was performed for all patients in the older patient subgroup to collect additional clinical data parameters that may predict life expectancy. The data points col-lected reflected conditions that have been associated with life expec-tancy in non-cancer cohorts including cardio-pulmonary conditions, renal function, performance status, smoking status and body weight [15–17].
2.1. Description of the Study Population
British Columbia residents aged 50 and older, diagnosed with path-ological stage III colon cancer between January 1, 2005 and December 31, 2009, who were referred to the BCCA within 6 months of diagnosis were included. We selected this time period to allow for adequate sam-ple size for follow-up and reliable ascertainment of 5-year recurrence and survival. Vital statistics were up to date to 2015. We categorized the “younger” age group as ages 50–69, and the “older” age group as ages 70 and above. These definitions of young and old have been used frequently in prior published literature.
Patients were excluded if they had a diagnosis of rectal cancer, a pre-vious or synchronous colorectal cancer, and histologic diagnosis other
than adenocarcinoma. Patients were also excluded if they did not have definitive surgery or died within 30 days of diagnosis. Finally, we ex-cluded patients who received any form of neoadjuvant therapy.
2.2. Statistical Analyses
SAS version 9.4 was used for all statistical analyses. Descriptive sta-tistics were used to perform inter-group comparisons. Wilcoxon rank test was applied to compare medians. Chi-square test and Fisher's exact test were used to compare categorical variables.
In our multivariate analysis of predictors of death (colon cancer ver-sus non-colon cancer) was analyzed with age, smoking status, comor-bidities, Eastern Cooperative Oncology Group (ECOG) performance status and Body Mass Index (BMI) using multivariate logistic regression and forward selection. Sensitivity analyses were performed on ECOG and BMI because of the large number of unknown values (44% and 38%).
Kaplan–Meier method was used to estimate the overall survival (OS) and recurrence free survival (RFS) for young and old groups. OS and RFS between are compared using log-rank test. The hazard ratio of old to young was calculated using Cox regression. Then competing risk method was used to estimate the RFS. Non CRC death was censored by Kaplan–Meier method or Cox Regression, while it was treated as a competing event by competing risk method. Gray test was used to com-pare the incidence of relapse or CRC death for young and old groups, and Fine-Gray regression was used to compute the hazard ratio of the two groups. OS was associated with stage, age, grade, lymph nodes removed and chemo treatment for older cohort using Cox Regression. Kaplan– Meier method and log-rank test were used to compare the OS between CRC versus non CRC death for older cohort. The hazard ratio of CRC ver-sus non CRC death was computed using Cox regression. Then OS be-tween CRC versus non CRC death was adjusted to stage, age, grade, lymph nodes removed and chemo treatment using Cox regression.