For panels C and D, covariates adjusted include age, sex, socioeconomic status, Charlson comorbidity score, timing of hepatic resection, mean number of lesions, mean size of largest lesion, surgical margin, and extent of surgical resection.
eFigure. Identification of patients with colorectal cancer who underwent resection of liver metastases in Ontario from 2002 to 2009.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Booth CM, Nanji S, Wei X, Mackillop WJ. Management and Outcome of Colorectal Cancer Liver Metastases in Elderly Patients: A Population-Based Study. JAMA Oncol. 2015;1(8):1111–1119. doi:10.1001/jamaoncol.2015.2943
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Surgical resection is standard treatment for patients with colorectal cancer (CRC) liver metastases (LM). Limited data describe practice and outcomes among elderly patients.
To describe management and outcomes of surgical resection of CRC LM in elderly patients in routine practice.
Design, Setting, and Participants
Population-based retrospective cohort study of routine clinical practices in the Canadian province of Ontario. All cases of CRC in Ontario who underwent resection of LM between January 1, 2002, and December 31, 2009, were identified using the population-based Ontario Cancer Registry and included in this study. Complete information about vital status in the Ontario Cancer Registry was available up to December 31, 2012; cause of death was available up to December 31, 2010. Final study analyses were performed March 13, 2015. Surgical resections of CRC LM were identified from hospital admission records. Pathology reports provided details regarding extent of disease and surgical procedure. Patients were classified into 3 age groups: younger than 65 years, 65 to 74 years, and 75 years or older. We describe volume of resected CRC LM as a ratio of incident cases per CRC LM resection. Use of perioperative chemotherapy was identified through linked electronic treatment and physician billing records. Preoperative and postoperative chemotherapy was defined as chemotherapy given within 16 weeks of surgery.
Main Outcomes and Measures
Overall survival and cancer-specific survival measured from time of LM resection.
We identified 1310 patients: 710 (54%) younger than 65 years; 414 (32%) 65 to 74 years; and 186 (14%) 75 years or older. Case volumes of CRC LM resection varied substantially across age groups. For patients younger than 65 years, there was 1 resection per 26 incident cases; 65 to 74 years, 1 per 38; and 75 years or older, 1 per 101 (P < .001). Patients less than 65 years of age had a mean of 2.3 lesions; 65 to 74 years, 2.0; and 75 years or older, 1.6 (P < .001). For patients younger than 65 years, mean size of the largest lesion was 4.0 cm; patients 65 to 74 years, 4.4 cm; and 75 years or older, 4.5 cm (P = .04). The likelihood patients younger than 65 years were to undergo a major liver resection of more than 3 segments was 65%; 65 to 74 years, 65%; and 75 years or older, 42% (P = .04). The percentage of patients younger than 65 years who underwent perioperative chemotherapy was 71% (501 of 710); 65 to 74 years, 57% (237 of 414); and 75 years or older, 41% (77 of 186) (P < .001). The incidence of 90-day mortality for patients younger than 65 years was 2% (11 of 710); 65 to 74 years, 5% (20 of 414); and 75 years or older, 8% (14 of 186) (P < .001). Cancer-specific survival at 5 years for patients younger than 65 years of age was 49%; 65 to 74 years, 47%; and 75 years or older, 35% (P < .001). Overall survival for patients younger than 65 years was 49%; 65 to 74 years, 44%; and 75 years or older, 28% (P < .001).
Conclusions and Relevance
Resection of CRC LM is associated with greater risk of postoperative mortality among elderly patients despite less aggressive treatment. Although the long-term outcomes are inferior to younger patients, a substantial proportion of elderly patients will have long-term survival.
Over the past 2 decades, surgical resection of colorectal cancer (CRC) liver metastases (LM) has evolved into a standard treatment. Outcomes reported from leading centers1-4 and clinical trials5,6 suggest that a considerable portion of patients will achieve long-term survival (40%-55% overall survival at 5 years) with acceptable postoperative mortality (<3%). However, it is well known that results reported from clinical trials and high-volume centers may not be realized in the general population.7 One important factor that may contribute to a gap between efficacy and effectiveness is that older patients with greater comorbidity are underrepresented in clinical trials and institutional case series.8 Population-based outcome studies may be useful to describe practice and outcomes achieved in routine practice.7,9
Seventy percent of patients diagnosed with CRC are older than 65 years.10 Current demographic trends will lead to an increasing number of elderly patients presenting with potentially resectable CRC LM.11 A handful of population-based reports have described patterns of care and long-term outcomes of patients with CRC LM in routine practice.12-15 However, 3 of these reports were restricted to patients 65 years and older,12,13,15 and the fourth report did not describe management and outcome across age groups.14 Three other population-level reports from the United States have described short-term outcomes among patients undergoing liver resection for a variety of indications and/or diseases.16-18 These studies found greater postoperative morbidity and mortality among elderly patients. As highlighted in a recent review of this topic by Anaya et al,11 there is a critical lack of information regarding management and long-term outcome of CRC LM among elderly patients in routine clinical practice. We undertook a population-based study to describe management and outcomes of elderly and nonelderly patients with resected CRC LM in a contemporary cohort. The study objectives were to describe practice patterns (ie, surgery and/or chemotherapy) and short- and long-term outcomes among elderly patients with resected CRC LM treated in routine clinical practice.
In this population-based study we describe the management and outcomes of all patients with resected colorectal cancer (CRC) liver metastases (LM) treated in Ontario, Canada, from 2002 to 2009.
Case volumes of CRC LM resection varied substantially across age groups: for patients younger than 65 years, 1 resection per 26 incident cases; patients 65 to 74 years, 1 per 38; and patients 75 years and older, 1 per 101 (P < .001).
Older patients were less likely to undergo a major liver resection and less likely to receive perioperative chemotherapy.
Cancer-specific survival at 5 years for patients younger than 65 years was 49%; patients 65 to 74 years, 47%; and patients 75 years and older, 35% (P < .001).
Overall survival for patients younger than 65 years was 49%; patients 65 to 74 years, 44%; and patients 75 years and older, 28% (P < .001).
Older patients are less likely to have resection of CRC LM. While long-term outcomes are inferior to those in younger patients, a substantial proportion of elderly patients with resected CRC LM will have long-term survival.
This is a population-based, retrospective cohort study designed to describe management and outcomes of resected CRC LM in the Canadian province of Ontario. Ontario has a population of approximately 13.5 million people and a single-payer universal health insurance program. The study population included all patients with CRC who underwent liver resection from January 1, 2002, to December 31, 2009. To identify the study cohort we used the Ontario Cancer Registry (OCR) to identify all incident cases of CRC in Ontario diagnosed from 1996 to 2009. We then identified all of those cases who underwent liver resection from January 1, 2002, to December 31, 2009. The OCR does not capture diagnoses of all second primary cancers. As such, cases with liver resection more than 72 months after initial CRC diagnosis were excluded since those cases would likely represent recurrence of a second primary cancer. Cases with histology other than adenocarcinoma were excluded. To minimize misclassification of LM, we also excluded cases with a second primary liver, biliary, or pancreas cancer. Extent of LM was not available in the existing data sources. For this reason we obtained surgical pathology reports for all potentially eligible cases. Patients with evidence of metastatic colorectal cancer per the liver resection pathology report were included. The study was approved by the Research Ethics Board of Queen’s University.
The OCR is a passive, population-based cancer registry that captures diagnostic and demographic information for at least 98% of all incident cases of cancer in the province of Ontario.19 The OCR also provides information about vital status and cause of death. Records of hospitalization from the Canadian Institute for Health Information provided information about surgical interventions, and these records are known to be complete.20 Provincial physician billing records from the Ontario Health Insurance Plan, treatment records from regional cancer centers, and provincial records of chemotherapy delivery (New Drug Funding Program and Ontario Drug Benefits ) were used to identify chemotherapy use. Incident cases of CRC identified from OCR were linked to other electronic administrative health databases at the Institute of Clinical and Evaluative Sciences. A valid Institute of Clinical and Evaluative Services key number enabled linkage to Canadian Institute for Health Information, Ontario Health Insurance Plan, and Ontario Drug Benefits. An OCR-encrypted unique identifier enabled linkage to OPIS and New Drug Funding Program. Surgical pathology reports were obtained from the OCR. A team of trained data abstractors reviewed the pathology reports and entered information about extent of disease and surgical procedure into an electronic database.
Indicators of the socioeconomic status of the community in which patients resided at diagnosis were linked as described previously.21 Quintiles of the median household income were based on the household income distribution for the full province of Ontario. Quintile 1 represents the communities where the poorest 20% of the Ontario population resided. Geographic regions reflect the catchment areas for Ontario’s regional cancer centers.21 Each case in the current study was assigned to one of these geographic regions based on their Ministry of Health residence code. Comorbidity was classified using the Charlson comorbidity index and included all noncancer diagnoses recorded during any hospital admission within 5 years prior to surgery.22 Preoperative chemotherapy was defined as chemotherapy given within 16 weeks before resection of CRC LM; postoperative, within 16 weeks after surgery for CRC LM. For descriptive purposes, we describe practice and outcome of patients by age using the following groups: younger than 65 years, 65 to 74 years, and 75 years or older. Overall survival (OS) and cancer-specific survival (CSS) were determined from resection of CRC LM. To account for possible cause of death miscoding, CSS included death from any cancer. Complete information about vital status in the OCR was available up to December 31, 2012; cause of death was available up to December 31, 2010. Final study analyses were performed March 13, 2015.
Comparisons of proportions between study groups were made using the χ2 test. Temporal trends were evaluated using the Cochran-Armitage test for trend. To evaluate temporal and regional trends in surgery that account for trends in incidence, we report the ratio of resected CRC LM cases from 2002 to 2009 to incident CRC cases from 2002 to 2009. This was derived by dividing the number of incident cases diagnosed in a given geographic region or age group by the number of cases undergoing liver resection. Overall survival and CSS were determined using the Kaplan-Meier method. Factors associated with OS or CSS were evaluated using the Cox proportional hazards regression model. The following variables were considered: age, sex, comorbidity, socioeconomic status, interval between diagnosis and hepatectomy, and extent of liver resection. All variables were included in the multivariate model. Results were considered statistically significant at P < .05. All analyses were performed using SAS version 9.3 (SAS Institute).
Linked administrative data sets identified 1711 potentially eligible patients who underwent surgical resection of CRC LM from 2002 to 2009 (eFigure in the Supplement). Surgical pathology reports were available for 1443 (84%) of cases, and 9% of these reports (133) indicated that the procedure did not include resection of CRC LM. Accordingly, the study population included 1310 patients. Characteristics of the study population are shown in Table 1. Median age is 63 years (range, 20-87 years). Fifty-four percent (710) were younger than 65 years, 32% (414) were 65 to 74 years, and 14% (186) were 75 years or older. Sixty-two percent of cases were female (810), and the proportion of males is greater among patients 75 years or older.
From 2002 to 2009, the volumes of CRC LM resection varied substantially across age groups. For patients younger than 65 years, there was 1 resection for every 26 incident cases; 65 to 74 years, 1 per 38; and 75 years or older, 1 per 101 (P < .001). Over the study period there was a 69% increase in volume of resected CRC LM from 1 resection for every 54 incident cases to 1 resection for every 32 incident cases (P < .001). The increase in volume was seen in those patients younger than 65 years (61% increase, P < .001) and 65 to 74 years (107% increase, P < .001) but not among patients 75 years or older (19% increase, P = .09).
Major hepatic resection (≥3 segments) was less common in patients 75 years or older (56%; n = 104 of 186) compared with patients younger than 75 years (65%; n = 734 of 1124; P = .04). While older patients had fewer lesions resected (<65 years, 2.3 lesions; 65-74 years, 2.0; ≥75 years, 1.6 [P < .001]), resected LM were larger in elderly patients (<65 years, mean size, 4.0 cm;65-74 years, 4.4 cm; and ≥75 years, 4.5 cm [P = .01]). Use of any perioperative chemotherapy was less common in elderly patients (<65 years, 71% [501 of 710]; 65-74 years, 57% [237 of 414]; and ≥75 years, 41% [77 of 186] [P < .001]). While preoperative chemotherapy alone did not vary substantially across age groups (<65 years, 16% [112 of 710]; 65-74 years, 15% [62 of 414]; ≥75 years, 14% [26 of 186] [P = .82]), older patients were less likely to get only postoperative chemotherapy (<65 years, 32% [226 of 710]; 65-74 years, 25% [102 of 414]; ≥75 years, 20% [38 of 186] [P = .002]) or combined preoperative and postoperative treatment (<65 years, 23% [163 of 710]; 65-74 years, 18% [73 of 414]; and ≥75 years, 7% [13 of 186] [P < .001]).
Short- and long-term outcomes are shown in Table 2. Median length of stay increased across age groups from 7 days to 9 days (75+), (P < .001). Hospital readmission at 30 days and 90 days after surgery was higher among elderly patients (P < .001). Postoperative mortality at 30 days and 90 days increased with age (Table 3). Factors associated with greater postoperative mortality are shown in Table 3. On adjusted analyses, compared with those younger than 65 years, older patients had an increased risk of death at 30 days (P = .02) and 90 days (P < .001). Compared with minor hepatic resection, major resection was associated with greater risk of death at 30 days (OR, 2.78; 95% CI, 1.03-7.54) but not at 90 days (OR, 1.58; 95% CI, 0.76-3.28). When analyses are restricted to patients older than 65 years, the only factor that remains significantly associated with postoperative mortality is major hepatic resection (OR, 2.78; 95% CI, 1.03-7.54 at 30 days).
Cancer-specific survival and OS curves are shown in the Figure. Five-year CSS (<65 years, 49%; 65-74 years, 47%; ≥75 years, 35% [P < .001]) and OS (<65 years, 49%; 65-74 years, 44%; ≥75 years , 28% [P < .001]) decreased with advancing age. Results of the Cox model are shown in Table 4. In adjusted analyses, older patients had inferior CSS (65-74 years, hazard ratio [HR], 1.13; ≥75 years, HR 1.57 [P = .01]) and OS (65-75 years, HR 1.14; ≥75 years, HR, 1.67 [P < .001]) compared with patients younger than 65 years. Greater comorbidity was also associated with inferior CSS and OS. Burden of resected LM was associated with long-term outcome. Patients with less than 1 liver lesion had inferior survival (CSS: HR 1.41; 95% CI, 1.16-1.72; OS: HR, 1.41; 95% CI, 1.20-1.67) compared with those with a solitary metastases. Resected LM greater than or equal to 5 cm were associated with worse outcome (CSS: HR, 1.46; 95% CI, 1.18-1.80; OS: HR, 1.43; 95% CI, 1.20-1.71) compared with lesions less than 5 cm. Having liver CRC LM resected more than 2 years after the primary tumor was resected was also associated with improved CSS and OS. When multivariate analyses are restricted to patients older than 65years, the findings are unchanged except that number of LM is not associated with OS or CSS.
In this study, we have described management of resectable CRC LM among elderly patients and outcomes achieved in routine clinical practice. A number of important findings have emerged. First, there has been a marked increase in case volumes of resected CRC LM in routine practice. However, this change in practice did not extend to patients 75 years and older. The lower rate of liver resection in older patients may reflect differential patient/physician decision-making across age groups, and/or the fact that older patients are less likely to be eligible for aggressive surgery due to more advanced metastatic disease or comorbidity. Second, the extent of resected disease varies across age groups. Older patients have fewer, yet larger, metastases resected than younger patients. They are also less likely to undergo major hepatic resection. Third, older patients are less likely to be treated with perioperative chemotherapy than younger patients. Fourth, resection of CRC LM is associated with greater risk of postoperative mortality among elderly patients despite less aggressive treatment. Fifth, although the long-term outcomes are inferior to younger patients, a substantial proportion of elderly patients will have long-term survival.
Four other population-based studies have described management and outcomes of resected CRC LM. Each of these studies provide important insights into outcomes achieved in routine practice. However, none of them provide comparative data by age for treatment and outcomes of CRC LM. Cummings et al,12 Mayo et al,15 and Robertson et al13 all used linked data from Surveillance, Epidemiology, and End Results Medicare, and as such their findings only pertain to patients who are at least 65 years old. Postoperative mortality within these studies (3%-4% at 30 days; 8%-10% at 90 days) are comparable to the subset of patients in our study 65 years or older (3.5% at 30 days; 5.7% at 90 days). Long-term outcomes (5-year OS, 26%-33%) are slightly inferior to patients older than 65 years in our study (5-year, OS 39%). Shah et al14 have reported the fourth population-based study of resected CRC LM. Although age as a continuous variable was associated with overall survival in the Cox model, the study did not report any other comparative data for younger vs older patients with respect to management or outcome. The study by Mayo et al15 was the only study to describe use of chemotherapy. From 2003 to 2006, 72% (489 of 675) of patients received perioperative chemotherapy: 16% of cases received chemotherapy within 6 months before surgery, 47% of patients received chemotherapy within 3 months after surgery, and 10% of patients received chemotherapy before and after surgery. Comparative chemotherapy use in our study population was lower. Fifty-two percent (314 of 600) of patients 65 years of age or older in our study received perioperative chemotherapy: 15% (88), preoperative; 23% (140), postoperative; and 14% (86) before and after surgery. Chemotherapy use may be lower in elderly patients due to differing comorbidity and performance status, unique patient preferences, and physician age bias. A recent study by Orcutt et al16 used the US National Inpatient Sample to describe short-term outcomes among 4026 patients who underwent liver resection for metastatic disease (ie, not just CRC) from 2005 to 2008. They found that while in-hospital mortality increased with age (<65 years, 1.3% increase; 65-74 years, 2.2%; ≥75 years, 3.3% [P = .01]), it was relatively low and within accepted ranges.
Several institution-based case series describe outcomes of elderly patients with resected CRC LM in high-volume centers. Adam et al23 assembled a cohort of 7764 patients who had resected CRC LM from 1986 to 2008 at 102 centers in 30 countries. Consistent with our results, they found that elderly patients were less likely to undergo major hepatic resection and less likely to receive perioperative chemotherapy. Mortality at 60 days after surgery was 3.8% for patients 70 years and older. Resection rates of CRC LM among patients 70 years and older increased over time in the study by Adam et al.23 In our own study, resection rates increased over time for patients younger than 65 years and between 65 and 74 years but not among the eldest cohort. This may relate to the fact that the study by Adam et al23 included a much longer time frame (22 years) compared with our study (8 years). In their comprehensive review of the literature related to CRC LM in elderly patients, Anaya et al11 describe outcomes reported in 12 studies, almost all of which are institutional case series. Across these studies, postoperative mortality among patients at least 70 years of age was approximately 4% to 8%, and 5-year OS was in the range of 16% to 47%.
While our data suggest an acceptable level of postoperative mortality among elderly patients, there are specific surgical factors that need to be considered in older patients. For instance, perhaps the most important factor in determining adequacy for surgery is the predicted functional capacity of the remnant liver postresection. Given the impairment in liver physiology with increasing age,24 elderly patients are at a higher risk of liver insufficiency following resection, with consequent increased morbidity and even mortality. The greater length of stay seen among older patients in our study may reflect longer postoperative recovery and/or the need for a longer hospital stay due to differing social circumstances and available home support networks.
Although our study provides important insights into the management and outcome of CRC LM among elderly patients in routine clinical practice, several methodological limitations warrant comment. The study population was identified using linked administrative health databases. While the OCR and the Canadian Institute for Health Information data sets are known to be consistent and complete,19,20 it is possible that our results may be biased by misclassification. A recent study by Anaya et al25 evaluated the validity of using administrative data sources to identify resection of CRC LM within the US Veterans Affairs system and found excellent reliability. While the electronic data sources used in this study describe general aspects of disease, treatment, and outcome for all patients in the province, detailed information related to stage of primary cancer, carcinoembryonic antigen level, chemotherapy dose and/or cycles, treatment toxic effects, and performance status is not available and limits our ability to evaluate the appropriateness of case selection for surgery and chemotherapy. Moreover, we are only partially able to control for known patient variables such as comorbidity. We also do not have information regarding other local therapies such as radiofrequency ablation. While perioperative chemotherapy may be associated with CSS or OS, it was not included in the Cox model because the available data did not identify whether preoperative chemotherapy was given to downstage LM or whether it was given to eradicate micrometastases. As such, treatment choice in this context may be associated with vastly different patient subgroups and outcomes. Accordingly we did not believe it was appropriate to include chemotherapy in the Cox model. To describe temporal trends in resection of CRC LM we used a ratio of resected CRC LM cases to incident CRC. While this does provide some correction for temporal trends in incidence, it is imperfect because the ratio for any year includes CRC LM surgical cases that were (for the most part) diagnosed in earlier years. Despite these limitations, in addition to the large sample size and resulting statistical power, a major strength of the current study is the fact that, by virtue of the OCR, our study population includes all cases of CRC LM treated with surgical resection within Ontario and is therefore unselected. By including the entire population of interest, it is possible to minimize the referral and selection biases that affect traditional institution-based observational studies.7
Resection of CRC LM is associated with greater risk of postoperative mortality among elderly patients despite less aggressive treatment. Despite this, the operative mortality is within an acceptable range. Although the long-term outcomes are inferior to younger patients, a substantial proportion of elderly patients will have long-term survival. Further work is needed to better understand the optimal treatment of older patients with CRC LM.
Corresponding Author: Christopher M. Booth, MD, Division of Cancer Care and Epidemiology, Queen’s University Cancer Research Institute, 10 Stuart St, Kingston, ON K7L 3N6, Canada (firstname.lastname@example.org).
Published Online: September 10, 2015. doi:10.1001/jamaoncol.2015.2943.
Author Contributions: Dr Booth had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Booth, Nanji, Mackillop.
Acquisition, analysis, or interpretation of data: Booth, Nanji, Wei, Mackillop.
Drafting of the manuscript: Booth, Nanji.
Critical revision of the manuscript for important intellectual content: Booth, Nanji, Wei, Mackillop.
Statistical analysis: Booth, Wei, Mackillop.
Obtained funding: Booth, Nanji.
Administrative, technical, or material support: Booth, Nanji, Mackillop.
Study supervision: Booth, Nanji.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Institute for Clinical Evaluative Sciences, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. Dr Booth is supported as a Canada Research Chair in Population Cancer Care. This work was also supported by the Canada Foundation for Innovation and Queen’s University Department of Surgery.
Role of the Funder/Sponsor: The Institute for Clinical Evaluative Sciences, the Ontario Ministry of Health and Long-Term Care, the Canada Foundation for Innovation, and Queen’s University Department of Surgery had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: Parts of this material are based on data and information provided by Cancer Care Ontario. However, the analysis, conclusions, opinions, and statements expressed herein are those of the author(s) and not necessarily those of Cancer Care Ontario.
Create a personal account or sign in to: