Joinpoint regression analysis identified a notable change in primary tumor resection rate in 2001; corresponding annual percentage changes (APCs) were –0.41% in 1988-2001 and –2.39% in 2001-2010 (P < .001). Annual percentage changes for median relative survival rate were 2.18% in 1988-2001 and 5.43% in 2001-2009 (P < .001) The blue line represents 1998-2001 and the black line represents 2001-2010.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Hu C, Bailey CE, You YN, et al. Time Trend Analysis of Primary Tumor Resection for Stage IV Colorectal Cancer: Less Surgery, Improved Survival. JAMA Surg. 2015;150(3):245–251. doi:10.1001/jamasurg.2014.2253
With the advent of effective modern chemotherapeutic and biologic agents, primary tumor resection for patients with stage IV colorectal cancer (CRC) may not be routinely necessary.
To evaluate the secular patterns of primary tumor resection use in stage IV CRC in the United States.
Design, Setting, and Participants
A retrospective cohort study using data from the National Cancer Institute’s Surveillance, Epidemiology, and End Results CRC registry. Demographic and clinical factors were compared for 64 157 patients diagnosed with stage IV colon or rectal cancer from January 1, 1988, through December 31, 2010, who had undergone primary tumor resection and those who had not. Rates of primary tumor resection and median relative survival were calculated for each year. Joinpoint regression analysis was used to determine when a significant change in trend in the primary tumor resection rate had occurred. Logistic regression analysis was used to assess factors associated with primary tumor resection.
Main Outcomes and Measures
Difference in primary tumor resection rates over time.
Of the 64 157 patients with stage IV CRC, 43 273 (67.4%) had undergone primary tumor resection. The annual rate of primary tumor resection decreased from 74.5% in 1988 to 57.4% in 2010 (P < .001), and a significant annual percentage change occurred between 1998-2001 and 2001-2010 (–0.41% vs –2.39%; P < .001). Factors associated with primary tumor resection were age younger than 50 years, female sex, being married, higher tumor grade, and presence of colon tumors. Median relative survival rate improved from 8.6% in 1988 to 17.8% in 2009 (P < .001); the annual percentage change was 2.18% in 1988-2001 and 5.43% in 1996-2009 (P < .001).
Conclusions and Relevance
The majority of patients with stage IV CRC had undergone primary tumor resection but, beginning in 2001, a trend toward fewer primary tumor resections was seen. Despite the decreasing primary tumor resection rate, patient survival rates improved. However, primary tumor resection may still be overused, and current treatment practices lag behind evidence-based treatment guidelines.
Colorectal cancer (CRC) is the third most commonly diagnosed cancer and third leading cause of cancer death in both men and women in the United States, and 136 830 new cases and 50 310 deaths due to the disease are expected in 2014.1 Approximately 20% of patients are diagnosed with stage IV disease at the time of initial diagnosis, and their reported 5-year relative survival (RS) rate is 12.5%.2 Treatment options include palliative systemic therapy and primary tumor resection (PTR), but evidence has suggested that PTR may not be routinely necessary for patients with unresectable stage IV CRC.3,4 For patients with resectable stage IV CRC, PTR is curative if it is performed with synchronous or staged resection of all metastatic lesions. Unfortunately, 80% to 90% of patients with stage IV disease present with unresectable metastases.5
Primary tumor resection is currently recommended for patients with unresectable stage IV disease who present with symptoms related to their primary tumor (eg, obstruction, perforation, significant bleeding). The role of PTR for asymptomatic patients remains controversial. Some physicians advocate PTR to prevent the development of symptoms associated with an intact primary tumor. Approximately 10% to 25% of patients with intact primary tumors will develop symptoms, but there is considerable morbidity (4%-30%) and mortality (2%-10%) associated with noncurative PTR.4,6,7
For the past several years, systemic chemotherapy for stage IV CRC has greatly improved beyond the regimen of fluorouracil and leucovorin. Since 2000, the US Food and Drug Administration has approved several systemic and biologic chemotherapeutic agents, including irinotecan, oxaliplatin, bevacizumab, cetuximab, capecitabine, and panitumumab. These agents can be used as first- or second-line agents alone or in combination with fluorouracil and leucovorin. Several trials have demonstrated that administration of these agents prolongs median survival, with low rates of subsequent acute primary tumor–related complications, and these agents might convert unresectable disease to resectable disease that can subsequently be treated with curative surgery.8-14 Therefore, current National Comprehensive Cancer Network guidelines recommend systemic chemotherapy without PTR in patients presenting with nonobstructive unresectable stage IV disease.15 Also of note, the US Food and Drug Administration approved both regorafenib and aflibercept in 2012 for the treatment of stage IV CRC, further adding to the available armamentarium.
A 2005 report of patients with stage IV CRC who had been treated during the fluorouracil and leucovorin era reported a high rate of PTR among these patients.16 The rate of PTR in the modern era of systemic chemotherapies and biologic therapies is unknown. Based on our experience, we hypothesized that the rate of PTR has decreased over time, reflecting the availability of modern systemic treatment options. The primary aim of this study was to evaluate the secular patterns of PTR use and survival in stage IV CRC in the United States using a population-based database.
A retrospective cohort study was performed using the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database. The SEER program collects cancer incidence and survival data from 18 cancer registries covering approximately 26% of the US population.17 The program also collects demographic information (eg, age, sex, race/ethnicity) and clinical information (eg, primary tumor site, tumor histology, tumor grade, tumor stage, treatment, patient survival). The SEER program does not collect data on chemotherapy treatments, which was therefore not evaluated in this study. Patients in the database had been assigned a stage at initial diagnosis; therefore, subsequent metachronous metastases do not alter the originally assigned stage.
Patients with histologically confirmed stage IV colorectal adenocarcinoma diagnosed between January 1, 1988, and December 31, 2010, were eligible to be included in the study. We used SEER historic stage A, which is derived from Extent of Disease for 1988-2003 and Collaborative Stage for 2004 and later, to identify patients who had presented with metastatic disease. Exclusion criteria included age younger than 18 years or older than 90 years, survival time of less than 30 days after confirmed diagnosis, CRC not the first and only diagnosis of malignant disease, and cancer reported from a nursing home, hospice, autopsy, or death certificate. To minimize the number of patients who had received surgery with curative intent, we also excluded patients who had undergone partial or total removal of other organs during PTR. This study was exempt from review by The University of Texas MD Anderson Cancer Center Institutional Review Board because preexisting data with no personal identifiers were used.
Demographic and tumor factors of patients with stage IV CRC who had or had not undergone PTR were compared using χ2 tests. Primary tumor resection was defined as partial, subtotal, or total colectomy; proctectomy; or total proctocolectomy without partial or total removal of other organs. The counterpart was defined as no surgery, local tumor destruction, or excision. The PTR rate was calculated for each year in our study period. We used multivariable logistic regression to identify factors associated with PTR. Covariates included age, sex, race/ethnicity, marital status, year of diagnosis, SEER region, tumor grade, and tumor location.
To identify an apparent change in the PTR rate, we used Joinpoint regression analysis to determine points in time at which a significant change in the PTR rate had occurred.18 The best-fitting point where the rate had changed significantly was chosen by means of a permutation test. The P value for each permutation test was estimated using Monte Carlo methods. We started with no joinpoint and tested whether 1 or more (up to 4) joinpoints were statistically significant and, if they were, added them into the model to reflect a significant change in PTR rate. We used annual percentage change to characterize trends for each segment as separated by each joinpoint. The annual percentage change, which indicates the rate change at a constant percentage of the rate of the previous year, was estimated by fitting a regression line to the natural logarithm of the rates.
We used median 5-year RS to measure the length of time from date of diagnosis to half of the patients being alive. Relative survival is defined as the ratio of observed to expected survival in a comparable cohort of the general population matched according to age, sex, and race/ethnicity. Relative survival is a widely recognized net survival measurement used to quantify excess mortality attributable to a disease. Using RS for survival trend analysis has 2 advantages: it does not depend on the accuracy of the reported cause of death and it accounts for life expectancy variation over time.
Statistical analysis was performed using Stata MP, version 11.2 (StataCorp LP). We used SEER*Stat, version 8.1.2, and the Joinpoint regression program, version 4.0.4, produced by the Surveillance Research Program of the National Cancer Institute to obtain survival statistics and perform joinpoint analyses.19
A total of 64 157 patients met the inclusion criteria. Overall, 67.4% of patients who had presented with stage IV CRC had undergone PTR. The median age at diagnosis was 66 years (interquartile range, 56-75 years). Patients younger than 50 years make up 13.7% of the cohort. At the time of presentation with stage IV CRC, patients were more likely to have been male, been white, been married, and resided in the West (Table 1). They were also more likely to have presented with poorly differentiated or undifferentiated tumors and tumors located in the left colon.
Compared with patients who had not undergone PTR, those who had undergone PTR were more likely to have been younger and to have been married. In addition, patients who had undergone PTR were more likely to have presented with well-differentiated or moderately differentiated tumors and to have had tumors located in the colon.
The PTR utilization rates for patients with stage IV CRC were 74.5% in 1988 and 57.4% in 2010 (P < .001) (Figure). We used the Joinpoint regression program to contrast the fitted segmented regression function by a series of permutation tests. The results indicated that the model with 2, 3, or 4 joinpoints as an alternative hypothesis had the same model fit as the model with 1 joinpoint (all P > .02) (Table 2). Thus, the null hypothesis was accepted and 1 joinpoint was included in the final model. The identified joinpoint implied that the PTR rate decreased precipitously in 2001. The corresponding annual percentage changes were –0.41% in 1988-2001 and –2.39% in 2000-2010 (P < .001).
The median RS rate for patients who had been diagnosed with stage IV CRC between January 1, 1988, and December 31, 2009, improved from 8.6% in 1988 to 17.8% in 2009 (P < .001) (Figure). Survival for 2010 was not reported owing to limited follow-up time. Based on our PTR analysis, we used a joinpoint of 2001 to calculate the median RS rate. The median RS was approximately 8% to 12% between 1988 and 2001; after 2001, the median RS began to rise. The corresponding annual percentage changes were 2.18% in 1988-2001 and 5.43% in 1996-2009 (P < .001).
We performed multivariable analysis of factors associated with patients with stage IV CRC who had undergone PTR at disease presentation. The odds of having undergone PTR decreased significantly with increasing age, particularly for patients 76 years or older (odds ratio, 0.67; 95% CI, 0.62-0.72, P < .001) (Table 3). Men, black patients, and those residing in the Northeast also were less likely to have undergone PTR at disease presentation. Compared with the reference of 1988-1991, the odds of PTR had decreased over time such that, in 2007-2010, the odds ratio was 0.41 (95% CI, 0.38-0.44, P < .001). Factors significantly associated with PTR at disease presentation were being married, residing in the South, having well-differentiated or moderately differentiated tumors, and having tumors located in the left colon.
Our population-based study demonstrated that the use of PTR in patients presenting with stage IV CRC has been decreasing over time, and a notable reduction in the PTR rate was observed in 2001. The use of PTR varied significantly according to both the demographic characteristics of the patients and clinical factors. Younger age, female sex, white race/ethnicity, married status, residence in the South, higher tumor grade, and tumors located in the colon were associated with an increased likelihood of having undergone PTR. Our findings also demonstrated a trend toward a lower PTR rate in patients who had presented with metastatic CRC as well as higher RS over time.
We observed that elderly patients had been less likely to undergo PTR, which is consistent with a retrospective study published in 2005.16 In that study, elderly patients were more likely to have presented with multiple comorbidities, poor general condition, and nutritional deficiencies. Because major surgery carries a risk of morbidity and mortality, surgeons are less likely to recommend PTR for asymptomatic elderly patients.
We were not surprised to find that only 51.5% of patients with tumors in the rectosigmoid or rectum but 76.8% of patients with tumors in the colon had undergone PTR. Several factors might have led to such an observation, including a higher infection rate after rectal surgery, the complexity of rectal surgery, patient fear of a permanent colostomy if anal function could not be preserved, and improved local control with the use of multimodality therapy.20-22 We also observed that approximately two-thirds of patients with stage IV CRC had still undergone noncurative PTR despite the high risk for morbidity and mortality associated with PTR.4,6,7 Little is known about the effect of PTR on survival or quality of life. A survival benefit has been observed in retrospective studies, but those studies are subject to potential selection bias.23,24 The results of 2 ongoing randomized trials of PTR in patients with stage IV disease may shed some light on the issues of survival and quality of life.25,26
Currently, the recommended treatment for stage IV CRC is systemic therapy. Systemic chemotherapy and biologic therapy can control symptoms, prolong survival, and possibly convert unresectable disease to resectable disease. Fortunately, owing to advances in chemotherapeutic and biologic agents, physicians have better systemic treatment options available. Several trials have demonstrated improved survival in patients with stage IV CRC who had been treated with irinotecan, oxaliplatin, bevacizumab, cetuximab, or panitumumab as single agents or in combination with fluorouracil and leucovorin or capecitabine.8-14 In our study, despite a trend toward decreased rates of PTR beginning in 2001, we observed improvements in the RS rate (a surrogate for disease-specific survival). Both of these findings coincided with advances in chemotherapeutic agents that started in 2000. Despite the proven benefit of chemotherapy, more than half of the patients in our cohort with stage IV CRC had undergone PTR in 2010. It appears that current treatment practice may still lag behind evidence-based treatment guidelines and that there is still work needed to translate evidence on the effectiveness of health care decisions into clinical practice.
Our study has several limitations. First, the decreasing rate of PTR could have been primarily the result of more effective systemic therapy or of greater reluctance among surgeons to operate on patients with asymptomatic stage IV CRC. The extent to which these factors contributed to the observed decreasing PTR rate is unknown. The SEER database does not provide information regarding chemotherapy, so we were unable to confirm that the accelerated reduction in PTR was a result of modern systemic chemotherapy. However, the remarkable reduction in the PTR rate after 2001 coincided with the advent of modern chemotherapeutic agents, and a recent study demonstrated decreasing PTR rates with the adoption of modern systemic chemotherapy.27 Second, SEER provides no information regarding the intent of PTR. Patients could have undergone palliative resection or PTR and resection of metastasis with curative intent. While the association with left-sided tumors and PTR suggests that concern for obstruction may have been a factor, the association with married status and race/ethnicity suggests other health care access factors may be at play. Finally, we did not perform a primary comparative analysis of the effect of PTR on survival. Such analysis within SEER is subject to considerable selection bias that cannot be mitigated using standard multivariable regression techniques and would lead to potential for misinterpretation. However, there is considerable controversy around this issue, highlighting the need for randomized study.3
Our study demonstrated a decreasing use of PTR and improved survival for patients presenting with stage IV CRC. A trend toward nonsurgical management was noted beginning in 2001, which coincided with the availability of new chemotherapeutic and biologic treatment agents. Despite the availability of more effective chemotherapeutic options, a considerable number of patients with stage IV CRC continue to undergo PTR. Our findings indicate potential overuse of PTR among these patients and highlight a need to better understand the clinical decisions and outcomes associated with that treatment.
Accepted for Publication: June 11, 2104.
Corresponding Author: George J. Chang, MD, MS, Department of Surgical Oncology, The University of Texas MD Anderson Cancer Center, T. Boone Pickens Academic Tower, 1515 Holcombe Blvd, Unit 1484, Houston, TX 77030 (firstname.lastname@example.org).
Published Online: January 14, 2015. doi:10.1001/jamasurg.2014.2253.
Author Contributions: Dr Chang had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Hu and Bailey contributed equally to this article.
Study concept and design: Hu, You, Chang.
Acquisition, analysis, or interpretation of data: Hu, Bailey, Skibber, Rodriguez-Bigas, Feig, Chang.
Drafting of the manuscript: Hu, Bailey, Chang.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Hu, Bailey, Chang.
Obtained funding: Chang.
Administrative, technical, or material support: You, Chang.
Study supervision: Chang.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported in part by grants T32CA009599 (Dr Bailey) and K07-CA133187 (Dr Chang) from the National Institutes of Health/National Cancer Institute and an American Society of Clinical Oncology Foundation Career Development Award (Dr Chang).
Role of the Funder/Sponsor: The funding sources 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.
Create a personal account or sign in to: