Is Histological Investigation of Polyps Always Necessary?

• Introduction
• Patients and Methods
• Patients and Polyps
• Definitions
• Analysis
• Results
• Patient Characteristics
• Polyp Characteristics
• Predictive Values for Advanced Adenoma or Insignificant Polyp
• Combined Predictive Values
• Discussion
• Acknowledgments
• References

Background and Study Aims: To assess whether polyp histological type can be predicted by patient characteristics and endoscopic polyp findings.

Patients and Methods: 1681 polyps in 494 patients were categorized as advanced adenomas (villous component or severe dysplasia or early cancer) or insignificant polyps. Chi-squared tests were used to analyze whether polyp histological type could be predicted based on patient age (< 60 vs. ≥ 60 years), gender, family history of colon polyps or cancer, presence of anemia, polyp size (≤ 5 mm vs. > 10 mm), and location (left- vs. right-sided).

Results: Insignificant polyp histology (n = 1337) correlated with patient age < 60 years (P = 0.0026), lack of anemia (P < 0.0001), polyp size ≤ 5 mm (P < 0.0001), and right-sided location (P = 0.0058). Stepwise inclusion of these parameters demonstrated that the association of a ≤ 5 mm right-sided polyp in a patient < 60 years yielded the highest combined predictive value (96.2 %) for an insignificant polyp. Conversely, age ≥ 60 years, presence of anemia, polyp size > 10 mm, or left-sided location, as single or combined parameters, demonstrated a maximum predictive value of only 75.4 % for an advanced adenoma.

Conclusions: A small right-sided polyp in a young patient is associated with a small risk (3.8 %) for advanced adenomatous tissue, indicating that histological investigation of such a polyp might not always be necessary. However, the recent recognition of flat adenomas and “mini” de novo colon carcinomas in the European population also may limit the usefulness of small polyp diameters in the exclusion of severe polyp histology.

Introduction

The hypothesized adenoma-carcinoma sequence has led to a widespread policy of colonoscopic polypectomy for prevention of colorectal cancer [1]. As a guide to colonoscopic surveillance after polypectomy, the polyp size as well as the histological type and degree of dysplasia in an adenoma are considered to be the best indicators for the development of a metachronous adenoma or colorectal cancer. An increased risk for mortality from colorectal cancer is mainly found in individuals with advanced adenomas, such as those of size > 10 mm, with tubulovillous or villous histology, or with high-grade dysplasia or early carcinoma [2] [3] . Accordingly, recent guidelines recommend that patients without advanced adenomas do not need subsequent surveillance colonoscopy earlier than 5 or even 10 years [2] [3] .

If the histological type (advanced adenoma or insignificant polyp) of a removed polyp could be assessed by means of nonhistological data, the appropriate surveillance interval could be set without the need for a costly histopathological examination of the polyp. This study assessed whether there was a single indicator or a combination of patient characteristics and endoscopic polyp findings which would yield a sufficiently high predictive value for the polyp histological type to render histopathological examination of the polyp unnecessary.

Patients and Methods

Since 1986, all colorectal polyps removed endoscopically have been documented in the Registry of Colorectal Polyps at the University Hospital of Zuerich, Switzerland. Patients with gross colon cancer, inflammatory bowel disease, or familial polyposis, and patients without adenoma at initial colonoscopy were excluded from the registry. Retrospective analysis of the polyp registry yielded the data for the present study.

Patients and Polyps

Age, gender, family history of colorectal polyps or cancer, and the indication for the initial colonoscopy (anemia, altered bowel habits, abdominal pain, positive family history of colorectal polyps or cancer, various) had been recorded for each patient. The cut-off point for age groups was chosen at 60 years (< 60 vs. ≥ 60 years) in accordance with reports in the published literature [4] [5] .

Definitions

On the basis of the National Polyp Study [5], adenomas were classified as small (≤ 5 mm), medium (6 - 10 mm), or large (> 10 mm). In accordance with previous similar studies, the anatomical location was designated as right-sided (proximal to the splenic flexure) or left-sided colon [4]. Polyps were classified as hyperplastic, tubular, tubulovillous, or villous with no, mild, moderate or severe dysplasia according to the World Health Organization recommendation [6]. A polyp was defined as an advanced adenoma if it was an adenoma with a villous component (tubulovillous or villous), high grade dysplasia, or early cancer (carcinoma cells infiltrating the submucosa without involvement of the muscularis propria) [7] [8] . Polyps not fulfilling these criteria were defined as insignificant polyps.

Analysis

Chi-squared tests with Yates’ correction for continuity were used to assess whether there was an association between the presence of an advanced adenoma and patient characteristics (age (< 60 vs. ≥ 60 years), gender, family history of colon polyps or cancer, presence of anemia) or endoscopic polyp findings (size [≤ 5 mm vs. > 5 mm, and ≤ 10 mm vs. > 10 mm], and polyp location [left-sided vs. right-sided]). An overall P value of less than 0.05 was considered significant. A correction for the multiple testing of data was performed using Bonferroni’s method. This takes into account the chance of error per test by dividing the desired P value (P = 0.05) for the statistical comparison as a whole by the number of tests (n = 7) being made. Therefore, the corrected P value used to assess for significance of a single test was P = 0.0071.

Predictive values (PV) for an advanced adenoma or an insignificant polyp were calculated for all patient and polyp characteristics. The strength of the association of the patient or polyp parameters with advanced adenoma or insignificant polyp was measured by the odds ratio (OR) with the 95 % confidence interval.

Finally, forward stepwise inclusion of those parameters significantly associated with the polyp histology, in order of decreasing PV, was performed to assess the combination of parameters with the greatest PV and to exclude parameters without any independent predictive value.

Results

A total of 1938 polyps from 963 colonoscopies in 532 patients were documented over a 13-year period. Complete sets of data including patient characteristics (age, gender), family history, indication for the colonoscopy, polyp size, site, and histology were available for 1681 polyps from 826 colonoscopies in 494 patients; these data were analyzed.

Patient Characteristics

The mean patient age was 62.5 years (range 20 - 90), and 1096 (65.2 %) polyps were found in patients ≥ 60 years old. There were 294 (59.5 %) men and 200 (40.5 %) women. Of these patients, 79 (16.0 %) reported a positive family history for colorectal polyps or cancer. The presence of anemia was the indication for the initial colonoscopy in 184 patients (37.2 %). Other indications were altered bowel habits in 104 (21.1 %), pain in 84 (17.0 %), a positive family history of colon polyps or cancer in 44 (8.9 %), and various in 74 (15.0 %).

Polyp Characteristics

A total of 828 (49.3 %) polyps were found at initial colonoscopy, and 853 (50.7 %) were detected at a subsequent surveillance colonoscopy. Of the polyps, 1081 (64.3 %) were small (≤ 5 mm), 451 (26.8 %) medium (6 - 10 mm), and 149 (8.9 %) large (> 10 mm). The anatomical distribution included 631 (37.5 %) right-sided polyps (176 in the cecum, 165 in the ascending colon, 90 in the hepatic flexure, and 200 in the transverse colon) and 1050 (62.5 %) left-sided polyps (62 in the splenic flexure, 121 in the descending colon, 453 in the sigmoid colon, and 414 in the rectum).

A total of 1171 polyps (69.7 %) were adenomas, 474 (28.2 %) were hyperplastic, and 36 (2.1 %) were classified as “other” (including inflammatory and juvenile polyps). The 1171 adenomas were of the following histological types: 855 (73.0 %) were tubular, 293 (25.0 %) were tubulovillous, and 23 (2.0 %) were villous. Of the adenomas, 743 (63.5 %) showed mild dysplasia, 344 (29.4 %) moderate dysplasia, 71 (6.1 %) severe dysplasia, and 13 (1.1 %) early cancer. All polyps with carcinoma infiltrating the submucosa had cancer-free margins and no lymphatic or venous invasion. In summary, 344 (20.5 %) polyps contained advanced adenomatous tissue, and 1337 (79.5 %) were insignificant polyps.

Predictive Values for Advanced Adenoma or Insignificant Polyp

Age, categorized as < 60 vs. ≥ 60 years, had a significant predictive value for advanced adenoma or insignificant polyp (15.4 % vs. 22.6 %, and 84.6 % vs. 77.4 %, respectively; OR 1.59 [95 % CI 1.25 - 2.03]; P = 0.0026) (Table [1]). Advanced adenomas (20.6 % vs. 22.5 %) as well as insignificant polyps (79.4 % vs. 77.5 %) were found with equal frequency in men and women (OR 1.09 [0.83 - 1.43]; P = 0.20). Also, family history of colorectal polyps or cancer did not influence the likelihood of advanced adenoma (21.2 % vs. 18.0 %) or of insignificant polyp (78.8 % vs. 82.0 %) (OR 1.54 [0.96 - 2.36]; P = 0.12). Advanced adenomas were found more frequently (26.2 % vs. 15.8 %) and there were fewer insignificant polyps (73.8 % vs. 84.2 %) if the indication for the initial colonoscopy had been anemia compared with the other indications (OR 1.86 [1.49 - 2.34]; P < 0.0001).

Polyp size, categorized as ≤ 5 mm vs. > 5 mm, had an important predictive value for advanced adenoma or insignificant polyp (8.5 % vs. 40.2 % and 91.5 % vs. 59.8 %, respectively; OR 6.37 (4.91 - 8.27); P < 0.0001) (Table [2]). Furthermore, advanced adenomas were found more frequently (71.1 % vs. 14.5 %) and there were fewer insignificant polyps (28.9 % vs. 85.5 %) if the polyp size was > 10 mm compared to ≤ 10 mm (OR 13.40 [9.16 - 19.61]; P < 0.0001). The percentage of advanced adenomas was greater (24.4 % vs. 16.3 %) in the left colon, and insignificant polyps were more frequent in the right colon (83.7 % vs. 75.6 %) (OR 1.78 [1.47 - 2.20]; P = 0.0058).

Combined Predictive Values

Forward stepwise inclusion of the parameters with a significant association with the polyp histological type demonstrated that the combination of a ≤ 5 mm right-sided polyp in a patient aged < 60 years yielded the greatest combined PV (96.2 %) for insignificant polyp histology (Figure [1]). Inclusion of other parameters did not further increase this predictive value. The 3.8 % (five polyps) of advanced adenomas which were missed by this combination included only tubulovillous adenomas with mild (n = 3) or moderate (n = 2) dysplasia. None of these polyps was a villous adenoma or contained severe dysplasia or early cancer. The combined PV of these parameters was similar for polyps found at initial colonoscopy (95.8 %) or at subsequent surveillance colonoscopies (96.7 %) (P = 0.83).

Even the most predictive combination of patient and polyp parameters (size > 10 mm plus left-sided location plus presence of anemia) yielded a PV of only 75.4 % for an advanced adenoma (Figure [2]). Of these 46 advanced adenomas, 12 contained severe dysplasia and four early cancer. In this analysis also, the combined PV was equal for polyps found at initial colonoscopy (73.5 %) or at surveillance colonoscopies (76.4 %) (P = 0.54).

Discussion

Advanced adenomas, defined as adenomas containing villous components, severe dysplasia, or early cancer, indicate an increased risk for the development of metachronous colorectal polyps and cancer [7] [8] and therefore require more frequent surveillance colonoscopies than insignificant polyps [2] [3] . The present study analyzed whether the presence of an advanced adenoma or an insignificant polyp can be predicted by nonhistological parameters including patient characteristics and endoscopic polyp findings.

Patient age ≥ 60 years, presence of anemia, polyp size > 10 mm, and left-sided polyp location demonstrated an increased risk for advanced adenoma. Forward stepwise inclusion of these parameters yielded a maximum combined predictive value of only 75 % for an advanced adenoma (Figure [2]). This indicates that the polyps associated with the above parameters include various histological types and therefore require individualized surveillance plans. These polyps always require a histopathological examination, for two main reasons: first, to detect the polyps with cancer which require surgical treatment or early endoscopic surveillance, and secondly to find the 25 % of insignificant polyps that do not require further surveillance until at least 5 years [2] [3] , which will decrease costs.

The highest predictive value (96.2 %) for insignificant polyp histology was found in case of a small (≤ 5 mm) right-sided colon polyp in a young patient (< 60 years) (Figure [1]). The costly histopathological examination of such a polyp may not be necessary as it will not yield any clinically relevant additional information. The 3.8 % miss-rate of advanced adenomas included only tubulovillous adenomas with mild or moderate dysplasia. None of these polyps was a villous adenoma or contained severe dysplasia or early cancer.

The results of the present study are only partially comparable to those in the existing literature. Previous similar analyses [4] [5] [9] of data from polyp registries assessed only confirmed adenomas rather than polyps in general. Furthermore, they used histological parameters rather than nonhistological data to assess the risk of other histological characteristics, in particular severe dysplasia [5] [9] . In 420 Australian patients with 565 adenomas, severe dysplasia was associated with size > 10 mm, peduncular shape and villous architecture [9]. Similarly, the American National Polyp Study found that patient age, adenoma size, and the extent of the villous component were independent risk factors associated with high grade dysplasia [7]. The proof of an association of villous polyp architecture with severe dysplasia has a pure descriptive value, but does not help to reduce the number of unnecessary histopathological examinations.

Similarly to our study findings, multivariate analysis of more than 20 000 polyps in a German polyp registry demonstrated that polyp size and site enabled an assessment of the malignancy risk [6]. Left-sided polyps with a size of > 42 mm demonstrated the highest carcinoma risk of 87.6 %. This 12.4 % error rate confirms the conclusion of our study that advanced polyp histology cannot be predicted accurately enough by endoscopic findings alone to guide clinical decisions without histopathological examination.

Carcinoma and polyps are the most frequently found colonic lesions in iron-deficiency anemia [10]. Therefore, the presence of anemia was also included in the analysis. Polyps found in the course of the evaluation of anemia demonstrated a significantly increased risk of harboring advanced adenomatous tissue. The combination of anemia and the finding of a left-sided polyp of size > 10 mm yielded the highest predictive value for an advanced adenoma.

Polyp size was the single most important predictor for polyp histology in the present study, as well as in the published literature [4] [5] [9] [11] . The size of a colorectal adenoma may reflect its duration, its rate of growth, or both. Either the length of exposure of transformed cells to a carcinogenic agent or an increased susceptibility of rapidly growing transformed cells could account for the relationship of the size of adenomas to malignant transformation [7]. Furthermore, the cumulative probability of spontaneous mutations may increase with more frequent mitoses in larger adenomas.

Our findings do not contradict the recommendation of the National Polyp Study that small polyps also require removal as even small tubular adenomas may show a progressively greater risk for advanced histology with continued growth [7]. However, our data demonstrated that a small right-sided colon polyp removed in a young patient may not require histopathological examination. As the present study resulted from the analysis of conventionally defined elevated polyps, the above conclusion may not be applied to flat and depressed colonic lesions. Japanese and recent European studies using magnifying endoscopy and chromoendoscopy found that up to one-third of adenomas did not demonstrate any polypoid elevation [12] [13] . Even small such lesions frequently demonstrated early cancer, indicating that colorectal cancer may also develop directly from normal mucosa. These mini de novo carcinomas contradict the adenoma-carcinoma hypothesis and imply that the lesion size may not serve as a means of assessing the histological degree of severity. Thus, each removed lesion must be examined histopathologically, regardless of its size, if further studies confirm the existence of mini de novo colorectal carcinoma in the Western population also. Rather than size, new polyp characteristics determined by special procedures including chromoendoscopy and high resolution endoscopy may yield predictive information concerning polyp histology.

Acknowledgments

The authors thank Irene Guldenschuh MD, Rainer Hürlimann MD, Andreas Müller MD, and Beat Müllhaupt MD, for their help in collecting the polyp data.

References

• 1 Morson B. The polyp cancer sequence in the large bowel.  Proc R Soc Med. 1974;  67 451-457
  PubMedSearch in Google Scholar
• 2 Winawer S J, Fletcher R H, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale.  Gastroenterology. 1997;  112 594-642
  CrossrefPubMedSearch in Google Scholar
• 3 Hoff G, Sauar J, Hofstad B, Vatn M. The Norwegian guidelines for surveillance after polypectomy: 10-year intervals.  Scand J Gastroenterol. 1996;  31 834-836
  CrossrefPubMedSearch in Google Scholar
• 4 Nusko G, Mansmann U, Altendorf-Hofmann A, et al. Risk of invasive carcinoma in colorectal adenomas assessed by size and site.  Int J Colorect Dis. 1997;  12 267-271
  CrossrefPubMedSearch in Google Scholar
• 5 O'Brien M J, Winawer S J, Graham-Zauber A, et al. Patient and polyp characteristics associated with high-grade dysplasia in colorectal adenomas.  Gastroenterology. 1990;  98 371-379
  PubMedSearch in Google Scholar
• 6 Jass J R, Sobin L H. Histological typing of intestinal tumours. World Health Organization. 2nd edition.  Berlin; Springer, 1989
  Search in Google Scholar
• 7 Stryker S J, Wolff B G, Culp C E, et al. Natural history of untreated colonic polyps.  Gastroenterology. 1987;  93 1009-1013
  CrossrefPubMedSearch in Google Scholar
• 8 Atkin W S, Morson B C, Cuzick J. Long-term risk of colorectal cancer after excision of rectosigmoid adenoma.  N Engl J Med. 1992;  326 658-662
  CrossrefPubMedSearch in Google Scholar
• 9 Chapuis P H, Dent O F, Bokey E L, et al. Patient characteristics and pathology in colorectal adenomas removed by colonoscopic polypectomy.  Aust N Z J Surg. 1993;  63 100-104
  PubMedSearch in Google Scholar
• 10 Kepczyk T, Kadakia S C. Prospective evaluation of gastrointestinal tract in patients with iron-deficiency anemia.  Dig Dis Sci. 1995;  40 1283-1289
  CrossrefPubMedSearch in Google Scholar
• 11 Konishi F, Morson B C. Pathology of colorectal adenomas: a colonoscopic survey.  J Clin Pathol. 1982;  35 830-841
  CrossrefPubMedSearch in Google Scholar
• 12 Stolte M, Bethke B. Colorectal mini-de novo carcinoma: a reality in Germany too.  Endoscopy. 1995;  27 286-290
  PubMedSearch in Google Scholar
• 13 Rembacken B J, Fujii T, Caims A, et al. Flat and depressed colonic neoplasms: a prospective study of 1000 colonoscopies in the UK.  Lancet. 2000;  355 1211-1214
  CrossrefPubMedSearch in Google Scholar

D. Külling,M.D. 

Division of Gastroenterology
Dept. of Internal Medicine
University Hospital of Zürich

Rämistrasse 100
8091 Zurich
Switzerland


Fax: Fax:+ 41-1-2554503

Email: E-mail:kulling@gmx.ch

References

• 1 Morson B. The polyp cancer sequence in the large bowel.  Proc R Soc Med. 1974;  67 451-457
  PubMedSearch in Google Scholar
• 2 Winawer S J, Fletcher R H, Miller L, et al. Colorectal cancer screening: clinical guidelines and rationale.  Gastroenterology. 1997;  112 594-642
  CrossrefPubMedSearch in Google Scholar
• 3 Hoff G, Sauar J, Hofstad B, Vatn M. The Norwegian guidelines for surveillance after polypectomy: 10-year intervals.  Scand J Gastroenterol. 1996;  31 834-836
  CrossrefPubMedSearch in Google Scholar
• 4 Nusko G, Mansmann U, Altendorf-Hofmann A, et al. Risk of invasive carcinoma in colorectal adenomas assessed by size and site.  Int J Colorect Dis. 1997;  12 267-271
  CrossrefPubMedSearch in Google Scholar
• 5 O'Brien M J, Winawer S J, Graham-Zauber A, et al. Patient and polyp characteristics associated with high-grade dysplasia in colorectal adenomas.  Gastroenterology. 1990;  98 371-379
  PubMedSearch in Google Scholar
• 6 Jass J R, Sobin L H. Histological typing of intestinal tumours. World Health Organization. 2nd edition.  Berlin; Springer, 1989
  Search in Google Scholar
• 7 Stryker S J, Wolff B G, Culp C E, et al. Natural history of untreated colonic polyps.  Gastroenterology. 1987;  93 1009-1013
  CrossrefPubMedSearch in Google Scholar
• 8 Atkin W S, Morson B C, Cuzick J. Long-term risk of colorectal cancer after excision of rectosigmoid adenoma.  N Engl J Med. 1992;  326 658-662
  CrossrefPubMedSearch in Google Scholar
• 9 Chapuis P H, Dent O F, Bokey E L, et al. Patient characteristics and pathology in colorectal adenomas removed by colonoscopic polypectomy.  Aust N Z J Surg. 1993;  63 100-104
  PubMedSearch in Google Scholar
• 10 Kepczyk T, Kadakia S C. Prospective evaluation of gastrointestinal tract in patients with iron-deficiency anemia.  Dig Dis Sci. 1995;  40 1283-1289
  CrossrefPubMedSearch in Google Scholar
• 11 Konishi F, Morson B C. Pathology of colorectal adenomas: a colonoscopic survey.  J Clin Pathol. 1982;  35 830-841
  CrossrefPubMedSearch in Google Scholar
• 12 Stolte M, Bethke B. Colorectal mini-de novo carcinoma: a reality in Germany too.  Endoscopy. 1995;  27 286-290
  PubMedSearch in Google Scholar
• 13 Rembacken B J, Fujii T, Caims A, et al. Flat and depressed colonic neoplasms: a prospective study of 1000 colonoscopies in the UK.  Lancet. 2000;  355 1211-1214
  CrossrefPubMedSearch in Google Scholar

D. Külling,M.D. 

Division of Gastroenterology
Dept. of Internal Medicine
University Hospital of Zürich

Rämistrasse 100
8091 Zurich
Switzerland


Fax: Fax:+ 41-1-2554503

Email: E-mail:kulling@gmx.ch