Trends, patterns, and outcomes in the management of malignant colonic polyps in the general population of the United States

Authors

  • Nabil Wasif MD,

    Corresponding author
    1. John Wayne Cancer Institute at Saint John's Health Center, Santa Monica, California
    2. Department of Surgery, Mayo Clinic Scottsdale, Scottsdale, Arizona
    • Department of Surgery, Mayo Clinic, 5777 E. Mayo Blvd., Phoenix, AZ 85054===

    Search for more papers by this author
    • Fax: (480) 342-2866

  • David Etzioni MD,

    1. Department of Surgery, Mayo Clinic Scottsdale, Scottsdale, Arizona
    Search for more papers by this author
  • Melinda A Maggard MD,

    1. Department of Surgery, David Geffen School of Medicine at University of California at Los Angeles, Los Angeles, California
    Search for more papers by this author
  • James S Tomlinson MD,

    1. Department of Surgery, David Geffen School of Medicine at University of California at Los Angeles, Los Angeles, California
    2. Department of Surgery, VA Greater Los Angeles Healthcare System, Los Angeles, California
    Search for more papers by this author
  • Clifford Y Ko MD, MSHS

    1. Department of Surgery, David Geffen School of Medicine at University of California at Los Angeles, Los Angeles, California
    2. Department of Surgery, VA Greater Los Angeles Healthcare System, Los Angeles, California
    Search for more papers by this author

  • Presented in part at the American Society of Clinical Oncology (ASCO) Annual Meeting, May 29-June 2, 2009, Orlando, FL; and the Society of Surgical Oncology 63rd Annual Cancer Symposium, March 3-7 2010, St. Louis, MO.

Abstract

BACKGROUND:

Colonoscopic screening has increased detection of malignant colonic polyps (MPs). The goal of this study was to analyze trends and patterns in the surgical treatment of MPs and to assess outcomes as measured by lymph node retrieval and patient survival.

METHODS:

The Surveillance, Epidemiology, and End Results (SEER) registry was queried to identify 19,743 patients with MPs from 1988 to 2003.

RESULTS:

The majority of MPs were present in the sigmoid and descending colon (54%), and 38% were tubular, 40% were tubulovillous, and 22% were villous. During the study period, the incidence of MPs increased from 29% to 42% and surgical resection (SR) was increasingly used for treatment, from 54% to 70%. Predictors of SR were aged <70 years (odds ratio [OR], 1.3; 95% confidence interval [95% CI], 1.2-1.4), high-grade tumors (OR, 1.5; 95% CI, 1.2-1.9), diagnosis in 1999 through 2003 (OR, 1.7; 95% CI, 1.5-1.9), villous subtype (OR, 2.3; 95% CI, 2.0-2.6), and right colon location (OR, 8.1; 95% CI, 7.0-9.4). In patients undergoing SR, the mean and median number of lymph nodes was only 4 and 1, respectively. A significant percentage of patients (49%) had no lymph nodes retrieved and a worse overall survival at 5 years compared with patients who had at least 1 lymph node examined (64% vs 74%; P <.001).

CONCLUSIONS:

MPs appear to be increasing in incidence, and there is a trend toward increased use of SR for treatment. However, patients undergoing SR for MPs undergo suboptimal assessment of lymph nodes, with a detrimental impact on survival. Adequate lymphadenectomy and lymph node evaluation are crucial to improve outcomes in patients with these potentially curable lesions. Cancer 2011. © 2010 American Cancer Society.

The adenoma-to-carcinoma sequence is a central tenet of the pathogenesis of colon cancer. Clinical, histologic, and molecular studies support the origin of virtually all colon cancer from adenomatous polyps. In patients undergoing surgical resection for colon cancer, a third of the specimens contain a synchronous adenoma.1 With an increasing number of adenomas, the risk of cancer rises, as is starkly demonstrated by the natural history of patients with familial adenomatous polyposis (FAP), in which 100% of patients develop colon cancer if left untreated. Patients who have known adenomas and refuse removal develop colon cancer at the rate of 4% after 5 years and 14% after 10 years, which is higher than that of the general population.1 Conversely, endoscopic removal of polyps has been shown to arrest subsequent development of carcinoma.2 The progression from normal mucosa to benign adenoma, high-grade dysplasia or carcinoma in situ, and eventually invasive carcinoma is also associated with an extensively studied cascade of genetic mutations that culminate in the development of colon cancer.1

Colon cancer itself is the third leading cancer as well as third in cancer deaths for both sexes in the United States. In 2009, a total of 106,100 new cases of colon cancer were diagnosed and there were 49,920 deaths attributable to the disease.3 This high incidence and prevalence has led to an emphasis on prevention and screening and development of guidelines endorsed by national societies.4 Intensified screening means many of the cancers discovered are picked up in the early stages, mostly in the form of a malignant or invasive polyp. A malignant polyp is an adenoma containing a carcinoma that invades into the submucosa but not the muscular layer of the colon and constitutes up to 11% of all endoscopically removed polyps.5 Adenomas with evidence of high-grade or severe dysplasia but no invasion beyond the basement membrane are known as in situ polyps.

The risk of a colon adenoma harboring a carcinoma has been linked to several factors, including histology, size, and degree of dysplasia.6, 7 Adenomatous polyps are classified as tubular, villous, or tubulovillous. The most common type is tubular (75%-87%) and usually pedunculated with a smooth lobulated surface. Villous adenomas (5%-10%) are usually sessile and have a “shaggy rug” appearance. For adenomas to be classified as villous, >75% of the polyp must show typical villous architecture. The third group of tubulovillous polyps (8%-15%) demonstrates a mix of the previous subtypes, in which 25% to 75% of the composition is villous.6 The risk of finding a malignancy is 4.8% for tubular, 22.5% for tubulovillous, and 40.7% for villous polyps.7 Larger polyps are also more likely to be malignant; the incidence increases from 1.3% for polyps <1 cm, to 9.5% for polyps measuring 1 to 2 cm and 46% in polyps measuring >2 cm.7 As expected, both villous histology and size >2 cm are also associated with the development of severe dysplasia. In turn, polyps with evidence of severe dysplasia or carcinoma in situ are malignant in 34.5% of cases.6

Guidelines for the management of malignant polyps have been outlined by the National Comprehensive Cancer Network (NCCN) and are available on the World Wide Web.8 In patients with an adenoma (either tubular, tubulovillous, or villous) and invasive cancer (ie, a malignant polyp), no additional surgery is required if the polyp is completely removed endoscopically and has favorable histologic features. Favorable histologic features are low-grade tumors (grade 1 and 2), no lymphovascular invasion, and a negative resection margin. The presence of high tumor grade, lymphovascular invasion, or a positive resection margin should trigger definitive surgical resection for adequate therapy in most cases. Surgical resection for malignant polyps should follow standard oncologic principles and conform to national guidelines recommending removal and examination of at least12 lymph nodes.9

The primary goal of this study was to investigate trends and patterns in the treatment of malignant colonic polyps in the general population of the United States. Secondary goals were to assess the quality of surgical resection as measured by lymph node evaluation and patient survival outcomes.

MATERIALS AND METHODS

Surveillance, Epidemiology, and End Results Database

The National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) tumor registry database was used for this study. SEER contains over 3 million cases from 18 geographic sites, covering approximately 26% of the US population. The database was designed to reflect the overall characteristics of the US population and is regarded as a model population-based tumor registry. Quality control is an important component of the SEER program; the current standard for accuracy of the data in the registry is an error rate of <5%.10 SEER Program registries routinely collect data on patient demographics (age, sex, race, etc.), primary tumor characteristics (size, histology, grade, depth of invasion, etc.), lymph node staging (number of lymph nodes examined, number of lymph nodes positive, etc.), surgical therapy (polypectomy, segmental resection, etc.), vital status, and survival. Although information on radiotherapy is recorded, no information on chemotherapy is provided. The November 2008 update was used for this study, providing information from 1973 to 2006.11 As a population-based study with no patient identifiers involved, our study was exempt from Institutional Research Board review.

Case Selection

The study period was from 1988 to 2003. The International Classification of Diseases (ICD-10) histology codes 8210, 8211, 8261, and 8263 were used to identify malignant colonic polyps. Patients with adenomatous polyps in the setting of FAP (ICD-10 code 8220 and 8221) were specifically excluded, as were all polyps located in the rectum. Patients with no histologic confirmation of the diagnosis and cases identified from autopsy reports only were also excluded. Detailed information on the extent of invasion was extracted from the extent of disease codes and correlated with the 6th edition of theAmerican Joint Committee on Cancer staging system for colon cancer (T1 = submucosal). The extent of disease codes also contains information on location of the invasive tumor within the polyp itself, that is, head versus stalk.

Details of endoscopic and surgical therapy were extracted from the relevant site-specific codes. For the purposes of this study, local excision included all methods used for endoscopic removal of the tumor (ie, polypectomy, cryosurgery, electrosurgery, or laser ablation). Similarly, segmental resection, hemicolectomy, and other surgeries were all combined into the single category of surgical resection.

Statistical Analysis

Continuous variables were analyzed using independent-samples Student t tests and categorical variables using chi-square analyses. Kaplan-Meier survival curves and log-rank tests were used to identify differences in overall survival, defined as duration of survival after resection. Univariate analysis was used to identify differences in patients undergoing local excision and surgical resection. Multivariate logistic regression analyses were used to identify predictors for having undergone a surgical resection. Variables included in the model were sex, age (as a dichotomous variable aged <70 years and aged >70 years), race, site, location, polyp type, and time period of diagnosis. Predictors for survival were identified by Cox regression analysis using proportional hazards modeling. Patients with incomplete data were excluded from regression analyses. All tests were 2-tailed, significance levels were set at P <.05, and confidence intervals at 95% (95% CI). Statistical analysis was performed using SPSS 16.0 statistical software (SPSS Inc, Chicago Ill).

RESULTS

Demographics and Trends

Our study population was comprised of 19,743 patients with malignant colonic polyps, of whom 63% underwent surgical resection (Table 1). The majority of the study population was male and white, with a mean age of 70 years. The incidence of malignant polyps increased over the time period of the study, from 29% in 1988 to 1993 to 42% in 1999 to 2003, as did the proportion of patients undergoing surgical resection; from 54% to 70%. Wide variation was noted in the use of surgical resection for malignant polyps across the geographical regions included in SEER; from 42% in Hawaii to 74% in Kentucky.

Table 1. Demographic and Tumor Characteristics
VariablesAll Patients (n = 19,743)Local Excision, 7350 (37%)Surgical Resection, 12393 (63%)P
  1. SEM indicates standard error of the mean; SEER, Surveillance, Epidemiology and End Results.

Mean age ± SEM, y70.0 ± 0.069.1 ± 0.1470.0 ± 0.10<0.001
Sex
 Male10,711 (54%)4269 (58%)6442 (52%)<.001
 Female9032 (46%)3081 (42%)5951 (48%) 
Race   <.001
 White16,160 (82%)5872 (81%)10,288 (83%) 
 Black2032 (10%)721 (10%)1311 (11%) 
 Other1413 (8%)656 (9%)757 (6%) 
Time period   <.001
 1988-19935735 (29%)2613 (36%)3122 (25%) 
 1994-19985735 (29%)2298 (31%)3455 (28%) 
 1999-20038255 (42%)2439 (33%)5816 (47%) 
SEER registry   <.001
 Alaska32 (0%)14 (0%)18 (0%) 
 Atlanta890 (4%)320 (5%)570 (4%) 
 California1699 (9.0%)436 (6%)1263 (10%) 
 Connecticut2278 (12%)1024 (14%)1254 (10%) 
 Detroit2129 (11%)810 (11%)1319 (11%) 
 Hawaii685 (3%)397 (6%)288 (4%) 
 Iowa1728 (9%)627 (9%)1101 (9%) 
 Kentucky556 (3%)147 (2%)409 (3%) 
 Los Angeles3651 (18%)1342 (18%)2309 (19%) 
 Louisiana567 (3%)150 (2%)417 (3%) 
 New Jersey560 (3%)170 (2%)390 (3%) 
 New Mexico566 (3%)246 (3%)320 (3%) 
 Georgia37 (0%)8 (0%)29 (0%) 
 San Francisco1557 (8%)591 (8%)966 (8%) 
 San Jose-Monterey584 (3%)177 (2%)407 (3%) 
 Seattle1566 (8%)647 (9%)919 (7%) 
 Utah658 (3%)244 (3%)414 (3%) 
Polyp type   <.001
 Tubular7537 (38%)3406 (47%)4131 (33%) 
 Tubulovillous7976 (40%)2973 (40%)5003 (41%) 
 Villous4230 (22%)971 (13%)3259 (26%) 
Site   <.001
 Ascending colon and cecum6081 (31%)856 (12%)5225 (43%) 
 Transverse colon and flexures2688 (13%)703 (10%)1985 (16%) 
 Descending colon and sigmoid10,624 (54%)5561 (78%)5063 (41%) 
 Unknown350 (2%)   
 Mean tumor size ± SEM, cm1.7 ± 01.73 ± 0.041.75 ± 0.03.73
 Unknown13,533 (69%)5115 (70%)8418 (68%) 
Grade   <.001
 High882 (4%)123 (2%)759 (6%) 
 Low10,606 (54%)2491 (34%)8115 (66%) 
 Unknown8255 (42%)4736 (64%)3519 (28%) 

Tumor Characteristics

The distribution of polyp types in the current study population was 38% tubular, 40% tubulovillous, and 22% villous. The majority of polyps in our study were present in the descending colon and sigmoid (55%), followed by the ascending colon and cecum (31%) and the transverse colon and flexures (13%) (Table 1). The mean tumor size was 1.7 cm, although a significant proportion of patients had no tumor size recorded (69%).

Local Excision Versus Surgical Resection

Comparison of patients who underwent surgical resection versus local excision revealed several important differences. Patients undergoing surgical resection were more likely to be female, and to have villous tumors, right-sided tumors, and high-grade tumors (Table 1). No significant difference was observed with regard to tumor size between patients undergoing local excision compared with those treated with surgical resection. On multivariate analysis, age <70 years, more recent time period of diagnosis, high tumor grade, right-sided tumor location, and villous characteristics were all found to be predictive of surgical resection (Table 2).

Table 2. Predictors of Surgical Resection{TC}
N = 11,254aORP95% CI
  • OR indicates odds ratio; 95% CI, 95% confidence interval.

  • a

    Number entering regression.

  • b

    Referent category.

Age >70 yb1.00  
Age <70 y1.30<.0011.18-1.43
Time period 1988-1993b1.00  
1994-19981.19.0051.05-1.35
1999-20031.71<.0011.52-1.92
Low tumor gradeb1.00  
High tumor grade1.54<.0011.24-1.91
Tubular typeb1.00  
Tubulovillous1.13.0291.01-1.25
Villous2.29<.0011.98-2.64
Located in left colonb1.00  
Transverse colon3.41<.0012.91-3.98
Right colon8.08<.0016.97-9.37

Lymph Node Assessment After Surgery for Malignant Polyps

We were also interested in studying the quality of surgical resection in patients who underwent surgical resection (n = 12,393), as measured by lymph node counts. A significant proportion of patients undergoing surgical resection had no lymph nodes observed in the specimen (49%). The mean and median number of lymph nodes retrieved was 4 and 1, respectively. As seen in Figure 1, patients who had at least 1 lymph node examined had better overall survival at 5 years compared with patients who had no lymph nodes examined (74% vs 64%; P <.001). The corresponding disease-specific survival rate at 5 years was 92% versus 89% (P <.001). In patients who underwent surgical resection and had at least 1 lymph node examined (n = 6184), the rate of lymph node positivity was 7.1% (437 of 6184 patients).

Figure 1.

Survival difference between patients with no lymph nodes examined (NNE) versus those with at least 1 lymph node examined (NNE >1) is shown.

The mean lymph node count was significantly higher for the most recent time period of the study (1999-2003) compared with the time periods 1988 to 1993 and 1994 to 1998 (4.9 vs 3.5 and 3.8, respectively; P <.001 for both comparisons).

NCCN Guidelines

To test the current NCCN guidelines for management of malignant polyps with local excision, we selected the subset of our study population with low-grade tumors confined to the head or stalk of the polyp who underwent surgical resection (and hence a negative resection margin). In these patients (n = 2034), the rate of lymph node positivity was 5.5%.

Cox Regression Analyses

On univariate analysis, tumor grade, lymph node status, and number of lymph nodes retrieved were significantly related to survival. The type of resection performed (local excision vs surgical resection), sex, race, and location of polyp were not found to be significant. On multivariate analysis, lymph node positivity was found to have an adverse impact on survival (Table 3). Furthermore, patients who had at least 1 lymph node examined had a survival benefit compared with patients who had no lymph nodes examined (hazard ratio, 0.68; 95% CI, 0.64-0.73).

Table 3. Predictors of Survival After Surgical Resection{TC}
N = 10,935aHRP95% CI
  • HR indicates hazard ratio; 95% CI, 95% confidence interval; NNE, number of lymph nodes examined.

  • a

    Number entering regression analysis.

  • b

    Referent category

Lymph node negativeb1.00  
Lymph node positive1.25.0111.05-1.47
NNE = 0b1.00  
NNE >10.68<.0010.64-0.73

DISCUSSION

In the current study, we attempted to explore trends and patterns in the treatment of malignant colonic polyps in the general population of the United States. Both the incidence of malignant polyps as well as the use of surgical resection for treatment increased over the duration of the study period. Wide variation was noted in the use of local excision versus surgical resection among the geographic regions in the SEER database. We also identified suboptimal evaluation of lymph nodes in patients who did undergo surgical resection as well as a survival disadvantage for those patients with no lymph nodes in the specimen compared with those with at least 1 lymph node examined.

The increasing incidence of malignant polyps over the time period of the study may be attributed to rising awareness of and compliance with screening recommendations for colon cancer.12 A detailed exploration of this was not possible because the denominator of total number of patients undergoing screening is not available in SEER. The increase in the use of surgical resection from 54% in 1988 to 1993 to 70% in 1999 to 2003 is most likely multifactorial in nature. During this time, there was an evolution in both endoscopic polypectomy techniques as well as in minimally invasive or laparoscopic surgery. This may have resulted in a lower threshold for surgical resection of colon polyps due to the perceived advantages of minimally invasive surgical resection over traditional open surgery and the increasing application of this technique by surgeons.13, 14 Improvements in endoscopic techniques may have led to an increasing willingness to remove multiple polyps as well as larger polyps, both of which increase the chances of finding a malignant polyp and subsequent referral for surgery.15 Finally, national guidelines dictating management of malignant polyps evolved and may have helped in standardization of care.16

Hassan et al conducted a pooled analysis of 1900 patients with a diagnosis of malignant polyps collected from published literature on the subject, many of whom underwent surgical resection (53.7%).5 In their analysis, the distribution of malignant polyps demonstrated left colon predominance (71%), with 9% present in the right colon and 21% in the rectum, similar to the preponderance of left-sided polyps noted in our data (55%). Pedunculated polyps comprised 65.7%, and the remaining 34.3% were sessile. Data from the current study indicated a spilt of 38% tubular (most often pedunculated), 40% tubulovillous, and 22% villous (most often sessile). This shift in distribution of subtypes toward greater villous characteristics can be attributed to the unique nature of our data set in that only patients with malignant polyps were studied. The incidence of malignancy is higher in sessile polyps; hence, these patients are disproportionately represented in our data. Finally, they report a similar incidence of high-grade tumors (7.2%) as well as lymph node metastases (8.6%) as observed in our study (4% and 7.1%, respectively).

Not surprisingly, younger patients were more likely to undergo surgical resection in the current study (odds ratio [OR], 1.3 for patients aged <70 years). Other factors that predicted surgical resection instead of local excision were high tumor grade (OR, 1.5), villous subtype (OR, 2.3), right colon location (OR, 8.1), and most recent time period of 1999 to 2003 (OR, 1.71). Both tumor grade and villous subtype are predictive of malignancy in colonic polyps, and therefore a selection bias may be in place in referral of these patients for surgical resection. Differences in the use of surgical resection for malignant polyps of the right colon (86%) compared with the lower use for descending colon and sigmoid polyps (48%) observed in the current study may be attributed to the lower morbidity and mortality for right-sided colon resections and anastamoses compared with left-sided and pelvic anastamoses.17

The poor lymph node evaluation observed after surgical resection represents a major area of concern. The mean and median number of lymph nodes removed was 4 and 1, respectively, and 49% of patients had 0 lymph nodes removed (Nx). There is now good evidence from single-institution as well as population studies to suggest that a minimum of 12 lymph nodes is necessary to adequately stage colon cancer, and this is the position of several national societies.9, 18 This has clearly permeated into practice, as reflected by the temporal trends in lymph node retrieval noted in the current study, in which the mean counts from the most recent time period of the study (1999-2003) were the highest. Although our group has previously suggested that less than 12 lymph nodes may be necessary to adequately stage T1 colon cancer, the low rate of lymph node evaluation in this study is clearly inadequate.19 This puts patients at risk of cancer recurrence and death, as evidenced by the improved survival noted in patients who had at least 1 lymph node examined compared with those who had no lymph nodes in the specimen. This also has implications for adjuvant therapy discussions as well as cancer surveillance after resection.

We can speculate on some of the reasons for this low retrieval of lymph nodes. It is possible in many cases that resection was for large polyps not amenable to endoscopic removal that were not biopsied. In other cases, even with a benign biopsy, characteristics of the polyp were such that a surgical referral for resection was made. In both these clinical scenarios, surgical resection may have been a diagnostic rather than a therapeutic procedure, and complete oncologic resection incorporating adequate lymphadenectomy was not performed. In these patients, the cancer was likely found incidentally after histologic analysis. The finding of a malignant polyp on endoscopy referred for surgical resection may also lead to subconscious downstaging on the part of the surgeon and less aggressive surgical resection than would be performed if the diagnosis were a more advanced colon cancer. Finally, the diligence of the pathologist also impacts lymph node counts and the same mental downstaging may be occurring in many cases. Nevertheless, we are not able to completely explain the dismal lymph node retrieval rate noted in this national study, and further research is needed to clarify the causes.

It is interesting to note that the results of the current study indicate that, in malignant polyps considered safe to remove endoscopically by NCCN guidelines (ie, low-grade tumors removed with negative resection margins), the incidence of lymph node metastases is 5.5%. Admittedly, the guidelines also consider no lymphovascular invasion as one of the favorable features, and we were unable to study this variable because it is not coded for in the SEER database. Nevertheless, the 5.5% incidence remains high and compares with the incidence of lymph node metastases reported in T1 breast cancer, for which routine lymph node evaluation is standard of care.

Other limitations of the current study must be acknowledged. The study is observational in nature and is vulnerable to the limitations of selection bias that is inherent to such a study design. We are unable to account for variability in pathology protocols across the time period of the study and the geographic regions included in the database. As mentioned earlier, the majority of malignant polyps had no tumor size recorded and this limits our analysis of this variable. Polyp size dictates likelihood of successful removal endoscopically and is predictive of malignancy. The large number of patients with missing information regarding polyp size suggests that many may have been removed piecemeal and therefore pathologic assessment of size was not possible. This may also explain why no difference in mean tumor size was noted between the groups treated with local excision and surgical resection.

In conclusion, the results of the current study demonstrate that the incidence of malignant polyps undergoing surgical resection has increased over the time period of the study; however, wide variability across different geographic regions remains. Surgical resection for malignant polyps is influenced by the age of the patient, tumor grade, polyp subtype, and tumor location as well as the time period of diagnosis. Lymph node evaluation is suboptimal in patients undergoing surgical resection, and this has an adverse impact on survival. This is an area of quality improvement that needs to be addressed by administrators and physicians alike for these potentially curable lesions. Finally, caution must be used in the use of endoscopic techniques for resection of malignant polyps, even those with favorable features.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.

Ancillary