Article Text


A long hard look at Dukes' B
  1. G T Williams
  1. Department of Pathology, University of Wales College of Medicine, Cardiff, CF14 4XN; williamsgt{at}

Statistics from

Identification of prognostic pathological factors in Dukes' B colon cancer

Carcinoma of the colon is somewhat of a Cinderella among colorectal cancers, and during the last few years its ugly sister, carcinoma of the rectum, has hogged the limelight. This has been largely due to the (deserved) attention that has been paid to reducing local recurrence in rectal cancer through better preoperative imaging, surgical technique, and adjuvant chemoradiotherapy. However, cancer of the colon remains a highly lethal tumour. There is little controversy that it is best treated initially by surgical resection, that patients with completely resected Dukes' A tumours of usual histological type require no more treatment, and that patients found to have Dukes' C tumours after histological examination of the resection specimen should be offered adjuvant chemotherapy.1 For the 40% or so of colonic cancers that are Dukes' B, however, there is great uncertainty surrounding the place of such chemotherapy, as has been highlighted in the recently published revised Guidelines for the management of colorectal cancer (2001) from the Association of Coloproctology of Great Britain and Ireland (ACP).2 Part of the reason for this is that the Dukes' B category encompasses a wide range of tumours, from those that have just penetrated the muscular coat of the bowel wall to widely infiltrative neoplasms that show extensive extramural local spread, and it would not be surprising that the benefit of postoperative chemotherapy might vary in parallel with this. Oncologists have often spoken informally about “good Bs” and “bad Bs” and the new ACP guidelines propose certain pathological criteria that could be used to identify a subgroup of cases that might benefit from chemotherapy.

Another issue contributing to the uncertainty is the quality of the pathological examination of resection specimens on which the staging of colon cancers and the identification of adverse prognostic features is based. If this is not uniform within and between clinical trials then outcome comparisons may be meaningless. For example, inadequate sampling of the primary tumour will understage some Dukes' B cancers as Dukes' A tumours while inadequate lymph node sampling will understage some Dukes' C tumours as Dukes' Bs. Furthermore, pathological examination of cancers in most multicentre clinical trials is undertaken by a large number of different pathologists whose interpretation of pathological features is subject to interobserver variation. It is only recently that there has been any attempt at ensuring uniformity of pathological approach and review of slides within trials, and even this does not completely guarantee the quality of specimen dissection. The situation is further confused by pooling of patients and grouping of trials in different analyses (IMPACT B2 1999, Mamounas et al 1999).3,4

Given this background, the paper by Petersen and colleagues5 from Gloucester in this issue of Gut is of immense importance to pathologists, surgeons, and oncologists, and to the designers of future clinical trials of colon cancer[see page65]. It describes a prospective study of 268 unselected Dukes' B colon cancers (from a total population of 673 colon cancers and 377 rectal cancers) treated by primary resection in a single centre that have been meticulously dissected and characterised by a single gastrointestinal histopathologist. The quality is assured by the fact that the mean number of tumour blocks examined was 5.7 and the mean lymph node harvest was 21.3. We can therefore be confident that these are genuine Dukes' B cases. Equally rigorous methods were used for assessing the various pathological features and for assessing their prognostic impact by univariate and multivariate analysis. The results clearly show that four factors, namely tumour perforation, peritoneal involvement, venous spread, and surgical margin involvement, when carefully assessed in this way, are independently highly predictive of outcome. Dukes' B tumours with none of these adverse features have a similar prognosis to Dukes' A tumours while the presence of tumour perforation or any two of the other three factors reduces the five year survival to 50%. Interestingly, these are four of the five factors that were proposed in the ACP guidelines (without any evidence being presented) as possible indicators for considering adjuvant chemotherapy in colon cancer (the fifth was poorly differentiated histology) and Petersen et al's paper adds much weight to these proposals. It would have been interesting to know whether the individual pathological features further predicted the pattern of disease relapse (for example, did tumour perforation and serosal involvement predict peritoneal carcinomatosis, did venous spread predict distant metastasis, and did margin positivity predict local recurrence?), because future refinement of adjuvant postoperative therapies might allow prevention of each of these complications to be more specifically targeted.

None of the four prognostic pathological factors comes as any great surprise to pathologists but there are three features of detail that are of particular note. Firstly, Petersen et al found that not only did extramural venous invasion have prognostic significance but submucosal venous invasion also, which is at variance with the findings of Talbot et al's classic paper in rectal cancer.6 It is difficult to conceive of an anatomical reason why things should be different between the colon and rectum, but if this finding is confirmed, pathologists will have to start paying attention to recording submucosal as well as extramural venous involvement when reporting colon cancer, and the Royal College of Pathologists' Minimum Dataset for cancer reports modified accordingly. Secondly, the Gloucester group found that not only is involvement of the non-peritonealised surgical margin (the so-called circumferential resection margin) an important prognostic factor but that acute suppurative inflammation at this margin that is in continuity with the tumour itself has similar importance. Such a situation is most likely to occur in sigmoid colon cancers that are accompanied by diverticulitis. Amazingly, no previous analysis of prognostic features in colon cancer has examined the significance of circumferential margin involvement, and although it appears to be infrequent in colon cancer, the evidence presented indicates that pathologists who have not assessed it previously in colon cancer should start doing so. The methods used are exactly the same as those that are now routinely used in rectal cancer, where the “bare” retroperitoneal or mesocolic surgical surface (which occupies about half the circumference of the bowel in the caecum, the proximal ascending colon, and the distal sigmoid colon but is barely existent in the transverse colon) is painted with a marker and the specimen serially sliced transversely to identify where the tumour or any associated suppuration is closest.7 Thirdly, Petersen et al's paper highlights the importance of meticulously searching for peritoneal involvement in colonic cancer by taking at least two blocks of the tumour where it is closest to the peritoneal surface. The findings reinforce previous studies from the Gloucester group that have highlighted the immense prognostic power of this feature8 and give further impetus to the consideration of intraperitoneal therapy for patients with such tumours.

Petersen et al's paper probably represents the best available prognostic analysis of pathological variables in Dukes' B colon cancer. It represents a labourious study of the highest quality achievable, carefully executed by a highly motivated expert pathologist. Its findings deserve wide recognition and integration into clinical practice.

Identification of prognostic pathological factors in Dukes' B colon cancer

View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.