Article Text
PDF
Abstract
Aims Screening colonoscopy has led to more colorectal carcinomas presenting at an early stage potentially curable by endoscopic resection. In this study, we examined the clinical and histological features of a contemporary series of malignant colorectal polyps (MCPs) with subsequent surgical resection.
Methods We conducted a retrospective study on a consecutive series of MCPs from 239 patients, predominantly males (57.7%) with a median age of 66 years, and assessed histological parameters associated with residual disease on the surgical specimens.
Results Median MCP size was 18.6 mm, with 23.1% polyps measuring ≤10 mm. From the 140 surgical resection specimens, residual disease was identified in 20 cases, including 12 cases with metastatic lymph nodes and/or 9 cases with residual carcinoma in the large bowel wall. Histological parameters associated with nodal metastases were greater width and greater depth of the invasive component (p=0.001 and 0.006, respectively), poor differentiation (p=0.003) and a cribriform pattern (p=0.01). The risk of nodal metastases was 23.3% if two or three of these features were identified, while it was 0% and 4.5% if none or one was present, respectively. A positive margin was not associated with nodal metastasis and might be adequately treated by local endoscopic resection.
Conclusions Surgical resection should be recommended if ≥2 of these adverse histological features are present and may be warranted if one feature is present. A positive margin may require additional local resection but not necessarily surgery if no other adverse factors are present.
- COLON
- MALIGNANT TUMOURS
- CANCER
- GASTROINTESTINAL DISEASE
Statistics from Altmetric.com
Introduction
Colorectal carcinoma is a leading cause of cancer death worldwide. The majority arise from a precursor polyp, most commonly a conventional adenoma, which can be identified and removed during colonoscopy, thereby reducing the risk of development of and death from colorectal carcinoma.1 ,2 Endoscopically removed polyps containing carcinoma, defined by invasion through the muscularis mucosae, are referred to as malignant colorectal polyps (MCPs). They account for 0.75–5.6% of all colorectal polyps and up to 10% of colorectal polyps removed in bowel cancer screening programmes.3 MCPs are of two types: invasive carcinoma arising from a pre-existing adenoma and, more rarely, polypoid carcinoma without an evident pre-existing adenoma. Further management of patients following endoscopic excision of MCP is based on the risk of residual colon wall disease, determined primarily by histological assessment of the polypectomy specimen, and the morbidity and mortality risk related to the surgical procedure for a given patient. Conventional literature supports polypectomy alone for low-risk patients when no adverse features are evident. Adverse histological features associated with high risk of residual disease include qualitative parameters (poor differentiation, vascular invasion, tumour budding) or quantitative parameters (Haggitt level, Kikuchi level, size of invasive tumour, margin involvement).3–6 These factors are cumulative to predict the risk of residual disease in the surgical resection specimen.3 ,6
With the increasing implementation of screening colonoscopy programmes and improved colonoscopic techniques, many early colorectal carcinomas are being safely treated by endoscopic excision. Moreover, recent studies have reported lower rates of residual disease compared with the widely referred study by Ueno et al.6 The aim of this study was twofold. First, we retrospectively examined the clinical and histological features of a consecutive series of 239 MCPs. Second, we examined colectomy specimens performed as subsequent treatment for high-risk MCP and assessed histological parameters associated with risk of residual disease.
Methods
Malignant colonic polyp study group
The database from Envoi Specialist Pathologists (Brisbane, Australia) was searched for the terms “polyp”, “adenocarcinoma” or “carcinoma” and “Haggitt level” between March 2007 and September 2014. Demographics data, polyp location and previous clinical history were extracted from endoscopy reports. The slides from all MCPs were reviewed for the following histological features:
Polyp configuration: pedunculated or non-pedunculated. Because the endoscopic description of the polyp configuration was available in less than half the cases, we defined a pedunculated polyp by requiring the identification of a definite stalk.
Polyp size, recorded for polyps received intact only: the maximum polyp diameter recorded at macroscopic description was used and correlated with the histological size. Where a discrepancy existed such as when a long stalk of a pedunculated polyp contributed disproportionately to the polyp dimensions, the histological size was taken to be the most accurate as has recently been reported.3
Invasive tumour size: width and depth of invasion both measured in millimetre in well-orientated sections with the aid of a graticule. For polyps removed piecemeal, the greatest dimensions of invasive tumour seen in any piece were recorded. The depth of invasion was measured as the distance between the muscularis mucosae (sometimes artificially reconstructed by joining both lateral edges if destroyed by the tumour) and the deepest invasive front (figure 1).
Haggitt level: determined on pedunculated polyps received intact only, as previously defined.7
Kikuchi level: determined on non-pedunculated polyps, as previously defined.8 The sm2 and sm3 categories were merged into a single group as the deepest part of the submucosa was not present or identifiable in endoscopic resection specimens lacking muscularis propria.
Vascular space invasion (lymphatic and venous): determined by identifying definite carcinoma cells within vascular spaces. An orcein stain was performed to aid identification of venous invasion when suspicious on the H&E-stained sections (figure 1). The diagnosis of lymphatic invasion was based entirely on histological assessment (figure 2).
Tumour differentiation: poor differentiation was based on any part of an invasive carcinoma exhibiting either no gland lumen formation (ie, solid nests or sheets of cells) or abortive small gland lumina within a sheet-like background9 (figure 2). Well-differentiated adenocarcinoma required >95% gland lumen formation. All other tumours were considered moderately differentiated.
Cribriform architectural pattern: defined by the presence of rounded or ovoid formations containing punched-out spaces with sharp margins, bridging of tumour cells and frequent central comedo-type necrosis (figure 2). This pattern was similar to cribriform ductal carcinoma in situ of the breast.
Distance to deep resection margin: in intact polypectomy specimens, measured by a graticule in millimetre to the closest cut edge. A positive margin was defined as carcinoma reaching the submucosal margin or obscured by diathermy artefact such that completeness of excision could not be determined.
Presence of tumour budding and focal dedifferentiation (or poorly differentiated clusters): tumour budding was reported significant if five or more buds of fewer than five tumour cells were identified invading the stroma beyond the leading edge of the invasive glandular outlines. Focal dedifferentiation/poorly differentiated cluster was defined as solid collections of >5 tumour cells with loss of gland lumina at the leading edge of the tumour (figure 2).
Presence and type of any pre-existing lesion from which the invasive carcinoma arose.
Examples of malignant colorectal polyps at low magnification. (A and B) 5 mm flat polyp entirely consisting of a poorly invasive adenocarcinoma with venous invasion, highlighted by the orcein stain (B). (C) 5 mm polyp with an invasive moderately differentiated adenocarcinoma arising from a tubular adenoma. (D) Measurements of the maximum width (3.9 mm) and depth (1.7 mm) of an invasive moderately differentiated adenocarcinoma arising from a tubulovillous adenoma.
Adverse histological features of malignant colorectal polyps. (A and B) Poor differentiation with solid nests or sheets of carcinoma cells with rare or abortive gland lumen formations. (C) Invasive front of an adenocarcinoma with poorly differentiated clusters and tumour budding. (D) Lymphatic space invasion (arrow). (E and F) Cribriform pattern of an invasive adenocarcinoma displaying ovoid formations containing punched-out spaces, bridging of carcinoma cells and central comedo-type necrosis.
Quantitative measures of tumour dimensions, distance to margin, were assessed by one observer (ISB), while qualitative features, Haggitt level and Kikuchi level, were assessed as a consensus of two observers (ISB and CR). Immunohistochemistry for DNA mismatch repair (MMR) proteins was performed as previously described.10
Surgical resection study group
Where a subsequent colectomy was performed, the surgical specimen was reviewed by one pathologist (ISB) to determine the presence and size of any residual adenoma and/or carcinoma in the bowel wall, the number of sampled lymph nodes and the number of lymph nodes with metastatic carcinoma. The stage was defined according to the American Joint Committee on Cancer system, 7th edition.11 Since a rectal site presents unique clinical challenges with significant morbidity, the type of operation and relationship of the polyp site to the peritoneal reflection were separately examined.
Statistical analysis
Statistical analysis was performed using SPSS statistics software V.17.0 (SPSS, Chicago, Illinois, USA). Comparison between categorical variables was performed using the Pearson χ2 test or the Fisher's exact test, where appropriate. Comparison between qualitative variables was performed using the Mann–Whitney test. A two-tailed p value was used for all analyses, and values <0.05 were considered to be significant.
Results
Clinicopathological characteristics of malignant colonic polyps
A total of 239 consecutive MCPs were identified. Patients were predominantly males (57.7%) with a median age of 66 years (IQR 18) (table 1). Of all MCPs, 130 endoscopic resection specimens (54.4%) were received as intact polyps with a median size of 18 mm (IQR 12.25). Of all intact polyps, 30 (23.1%) measured ≤10 mm and 4 (2.3%) measured ≤5 mm. The distal colon and the rectum were the most common sites for MCP (76.6%). Among the 58 intact pedunculated MPCs, 19 were Haggitt 3 and 2 were Haggitt level 4. Kikuchi level sm2 or sm3 was found in 102 of 135 sessile MCPs. The majority of MCPs (91.2%) had an identifiable benign precursor polyp component and 21 cases presented as polypoid carcinomas. Tubulovillous/villous adenoma was the most common precursor polyp (60.2%).
Characteristics of malignant colonic polyps (MCPs) treated with endoscopic resection only and with follow-up surgical resection
Pathological findings in the follow-up surgical resection specimens
Subsequent surgical resection was performed for 140 MCPs. The differences in the clinicopathological characteristics of patients with and without follow-up surgery are presented in table 1. The surgical procedures were right hemicolectomy or subtotal colectomy (n=38), left hemicolectomy or sigmoid colectomy (n=69), high anterior resection (n=10), low anterior resection (n=14), ultralow anterior resection (n=8) and abdominoperineal resection (n=1). In 20 of the surgical specimens, residual adenocarcinoma was identified either within regional lymph nodes (12 cases; 1 involved node for 7 and 2 involved nodes for 5) and/or within the colonic wall (9 cases). The number of lymph nodes retrieved from the surgical specimens was not different between cases with and without nodal metastasis (median (IQR) 12.5 (6.5) vs 12.0 (7.0) nodes, respectively; p=0.6). The characteristics of each patient with nodal metastasis are listed in table 2. Of note, one case (patient no. 1) was from an initial 5 mm polyp in the caecum, which was diagnosed as a poorly differentiated adenocarcinoma with a neuroendocrine component (confirmed by immunohistochemistry) and a positive margin on the prior polypectomy specimen.
Main clinicopathological characteristics of the 12 patients diagnosed with a malignant colonic polyp resected by endoscopy who were diagnosed with lymph node metastasis on the follow-up colectomy
To date, no patient followed conservatively has demonstrated residual disease in the colonic wall (follow-up time 1–92 months). Hence, the overall rate for residual adenocarcinoma from our series is estimated at 20/239 (8.4%) and 12/239 (5%) for nodal metastasis. In all nine cases with residual carcinoma in the colonic wall, the initial polypectomy was incomplete with a positive margin. The final stage of the carcinoma was pT1 for seven cases, pT2 for one case and pT3 for one case. None of the intact MCPs with a negative margin but <1 mm clearance had residual carcinoma in the surgical specimen. Residual adenoma was present in 13 surgical specimens, as the only residual component in nine cases as minimal residual disease (<2 mm) and associated with residual carcinoma in four cases. For these 13 cases, the polypectomy margin was positive for adenoma in nine cases, with a clearance of <0.5 mm in two cases, 2.0 mm in one case and 2.2 mm in one case.
The outcome of the 60 patients with a low rectal MCP, of whom 33 were treated by surgery, is shown in table 3. Residual disease was present in six surgical specimens (10% of patients), including two cases with residual adenocarcinoma in the rectal wall and four cases with involved lymph nodes. No significant difference was found between the rate of residual disease in the rectal MCPs and in MCP from the rest of the large bowel (10% vs 7.8%, respectively; p=0.59).
Outcome of the 33 of 60 patients with low rectal malignant polyp who were treated by surgery
Parameters associated with nodal metastasis
Compared with the group of patients with no surgery or with no nodal metastasis on surgical specimens, cases with metastatic nodes were significantly associated with younger age (p=0.042), greater width and greater depth of the invasive component (p=0.003 and 0.005, respectively), Haggitt level 3/4 or Kikuchi level sm2/sm3 (p=0.015), poor differentiation (p<0.001) and a cribriform architectural pattern (p=0.002) (table 4). No significant association was found for lymphatic invasion, tumour budding or the presence of focal dedifferentiation.
Comparison of the clinicopathological features between patients with and without lymph node metastasis
We then assessed the predictive values of previously reported and widely used pathological parameters for the risk of lymph node metastasis. Using the three qualitative risk factors previously reported by Ueno et al6 (poor differentiation, lymphovascular invasion and tumour budding), the proportion of patients with metastatic lymph nodes in this study were 2.0%, 8.2% and 12.2% when 0, 1 or ≥2 risk factors were present, respectively (table 5). A more discriminative identification of cases with lymph node metastasis was found when poor differentiation, cribriform pattern and depth of invasion >2 mm were included in the same analysis. No nodal metastasis was found when none of these factors were present, while it was found in 4.5% and 23.3% of cases when 1 and ≥2 of these factors were present respectively (table 5).
Proportion of patients with nodal metastasis in relation to the number of histological risk factors from the Ueno's study and this study
Discussion
The management of patients diagnosed with carcinoma in a colonic polyp resected endoscopically is based on the risk of residual locoregional disease versus the risk (morbidity and mortality) associated with follow-up surgery. The role of the pathologist is to provide a risk assessment analysis for the presence of residual disease based on a set of histological features. It is generally accepted that greater size of invasive tumour, lymphovascular space invasion, poor differentiation and margin involvement are high-risk factors for residual disease. Surgical resection will usually be offered when one or more of these adverse features exist.
This study was undertaken to document the pathological features of a series of MCPs, diagnosed in the era of a national bowel cancer screening programme, and to investigate the outcome for patients who subsequently underwent surgical resection. The MCPs occurred more frequently in males (57.7%) and were predominantly located in the distal colon and the rectum (76.6%), in keeping with large published series.5 ,12 This could suggest that one success of screening and investigative colonoscopy is the removal of distal colorectal adenocarcinomas at an early stage. Many MCP were small with 23.1% measuring ≤10 mm and 2.3% measuring ≤5 mm. Of note, one MCP measuring 5 mm exhibited both lymphatic and venous invasion and another 5 mm MCP was associated with lymph node metastases. This is potentially important in light of recent studies and commentary proposing that polyps <5 mm in size may not require histopathological assessment.13 In addition, proponents of CT colonography argue for its use as a screening tool on the basis that colorectal polyps ≤10 mm (the lower level of sensitivity for the test) essentially never harbour invasive carcinoma.14
Most clinicopathological features of MCPs from this study concord well with those from previous pooled analysis by Hassan et al5 and the widely cited paper by Ueno et al.6 One point of difference is the lower proportion of pedunculated polyps when polyp configuration was defined by pathologists with the strict criterion of clear stalk identification (this study and ref. 6) compared with polyp configuration defined by endoscopists.5 An important aim of this study was to identify histological parameters associated with residual disease in subsequent resection specimens. We identified residual adenocarcinoma in 8.4% of patients, including 5% with nodal metastasis and 3.8%% with adenocarcinoma in the colonic wall. These figures are slightly lower than figures from a pooled analysis of MCPs reporting residual adenocarcinoma in 14.1% of cases, including 8.6% with nodal metastases5 and a recent study with similar methodology showing 11% with residual disease including 7% nodal metastases.15 All surgical specimens with residual carcinoma in the colonic wall had a positive margin defined by the presence of carcinoma cells reaching the diathermied base of intact MCPs. None of the intact MCPs with carcinoma present between 0.1 and 1 mm from the margin showed residual carcinoma in the surgical specimens, justifying our criteria for a positive margin on MCP resection specimens. Some studies have reported an increased risk of residual disease for rectal MCPs.16 The risk and benefit equation for rectal MCPs, particularly those close to the anal verge, is problematic as it may require removal of the anus and permanent colostomy. In the subgroup of patients with a rectal MCP treated by follow-up surgery, we found that 10% of rectal MCPs had residual disease, including 6.7% with nodal metastasis, a rate not significantly different than for MCPs in other parts of the large bowel.
In this study, poor differentiation, cribriform architectural pattern and size of the invasive component (depth >2 mm, width >4 mm, Haggitt level 3/4 for pedunculated polyps or Kikuchi level sm2/sm3 for non-pedunculated polyps) were the only histological features significantly associated with lymph node metastases. The risk of nodal metastases for patients with two or three of these parameters (that included depth >2 mm) was 23.3%, while it was 0% if none was present (25% and 2.6%, respectively, if width >4 mm is used instead of depth >2 mm). Poor differentiation was defined according to the United Kingdom Royal College of Pathologists definition of any area of the carcinoma displaying either no gland lumen formation (ie, solid nests or sheets of cells) or abortive small gland lumina within a sheet-like background. The proportion of carcinoma with poor differentiation varies greatly among studies, probably due to different criteria used in this assessment. In this study, 18.1% of MCPs were classified as poorly differentiated compared with 7.2% from the pooled analysis by Hassan et al5 and 29.9% by Ueno et al.6 Despite the range in the proportion of poorly differentiated MCPs, this histological feature is significantly associated with residual disease across published series and remains a reliable predictive factor for lymph node metastasis. Tumour budding or focal dedifferentiation (poorly differentiated clusters) did not prove to be a significant adverse factor in this study; however, they represent part of an epithelial mesenchymal transition programme,17 and we acknowledge some overlap between these morphological characteristics and poor tumour differentiation in general. Uniform and reproducible criteria to define high-grade colorectal adenocarcinoma based on morphology and possibly the microsatellite instability status are urgently needed.18 ,19 MMR deficiency was not significantly associated with nodal metastasis in this study, and to date no study has shown that MMR status is of prognostic value; however, identification of Lynch syndrome might alter the subsequent treatment. Our single MMR-deficient case with a nodal metastasis was a small (3 mm width) poorly differentiated adenocarcinoma of the caecum arising from a sessile serrated adenoma. Malignant transformation of sessile serrated adenoma can be encountered in small polyps (<10 mm).20 However, nodal metastasis is not frequently found.21 The adverse behaviour associated with a neuroendocrine component in invasive adenocarcinoma has been recently documented22 and is again illustrated by one case in this study (case no. 1). Pathologists should be cognisant of the possibility that poor differentiation may in fact represent neuroendocrine differentiation, and a low threshold for further investigation is warranted.
Similar to other studies, we found cribriform pattern of the invasive component to be an adverse factor of MCP6 ,23 and in all colorectal adenocarcinoma.24 Introduced in the latest WHO classification as a morphological subtype of colorectal adenocarcinoma, cribriform carcinoma of the colon resembles cribriform carcinomas of the breast or salivary glands and is often associated with comedo-type necrosis. The cribriform pattern has been found to be associated with a microsatellite stable—CpG island methylator high phenotype,25 a molecular subtype associated with adverse outcome.26 ,27
The size of the invasive component of MCPs assessed by various methods has been associated with an increased risk of nodal metastasis in previous studies.6–8 ,28 The assessment of Haggitt levels (pedunculated polyps) and Kikuchi levels (non-pedunculated polyps) was of prognostic value in our study. However, this semi-quantitative assessment of the level of invasion can be problematic in practice if there is fragmentation or poor orientation of the tissue. Moreover, Kikuchi levels cannot be accurately determined in endoscopic resection specimens that generally do not include the muscularis propria. The measurement of the invasive tumour size as the maximum width of the carcinoma and/or the distance between the muscularis mucosae and the deepest invasive front was easier to perform and is becoming a preferred option. While most reliable in intact polyps, we found invasive tumour size measurement also useful in fragmented MCP specimens since in the vast majority of cases the bulk of the invasive tumour is included in one or two fragments and will either exceed the prognostic threshold of >4 mm width or be fully contained in one piece by being <4 mm diameter. A positive margin in endoscopically resected MCPs, in the absence of other adverse histological features, was not associated with nodal metastasis. Based on our data, small tumour size (<2 mm) in fragmented polyps may be amenable to endoscopic follow-up if other adverse factors are absent.
In conclusion, the rate of residual disease in our series of 239 MCPs was 8.4%, including 9 cases with residual adenocarcinoma in the colonic wall, all showing a positive margin on the MCP, and 12 cases with nodal metastasis, including a small adenocarcinoma arising from a 5 mm sessile serrated adenoma. Histological features associated with nodal metastasis were poor differentiation, cribriform pattern and invasive tumour size defined by depth >2 mm (or width >4 mm). From our results, surgical resection should be recommended if two or three of these adverse histological features are present (23.3% risk of nodal metastasis) and may be warranted if only one feature is present (4.5% risk of nodal metastasis), depending on tumour location and patient comorbidities. No nodal metastases were found if none of these features were identified and surgery could be avoided for these patients. Cases with a positive margin defined by carcinoma cells reaching the diathermied base may require additional local resection but do not necessarily require surgery if no other adverse factors are present.
Take home messages
8.4% of malignant colorectal polyps resected at endoscopy will show residual adenocarcinoma within regional lymph nodes or the intestinal wall in subsequent surgical colorectal resection.
Increased tumour size and poor tumour differentiation are the main risk factors for lymph node metastases.
Risk for lymph node metastases is greatest when two or more adverse features are found.
Polyps measuring ≤10 mm in diameter may harbour invasive carcinoma.
References
Footnotes
Handling editor Cheok Soon Lee
Contributors Collection of cases and related data by ISB, GM and AB. Review of material by ISB and CR. Preparation of manuscript by ISb MLB and CR.
Competing interests None declared.
Ethics statement The study was conducted under approval by the ethics committee of QIMR Berghofer Medical Research Institute (P1298).
Provenance and peer review Not commissioned; externally peer reviewed.