Article Text


Survival for colorectal cancer
  1. S S CROSS
  1. Section of Oncology and Pathology
  2. Division of Genomic Medicine
  3. University of Sheffield Medical School
  4. Beech Hill Road, Sheffield S10 2RX, UK
  5. s.s.cross{at}
  1. G GATTA,
  1. Divisione di Epidemiologia
  2. Istituto Nazionale per la Cura e lo Studio dei Tumori
  3. Via Venezian, 1, 20133 Milano, Italy
  4. gatta{at}

Statistics from

Editor,—The EUROCARE study of Gattaet al (

) provides important information on survival for colorectal cancer in different parts of Europe. In particular, it relates survival to stage, a feature which was absent from earlier studies.1-3 The study also looks at procedures used in determining stage which may be important factors in accurate staging—for example, higher numbers of lymph nodes sampled will be more likely to detect lymph node metastases and consequent upstaging of tumours.4

However, the usefulness of this study is reduced by inaccuracies in the descriptions of staging and choice of staging system. The authors have chosen to use Dukes' staging system which has been in use for at least 50 years and has been well characterised by many studies with long term follow up.5-9 Unfortunately, they have given incorrect labels to some of the stages in this system. In tables 2 and 3, Dukes' stages A and B are described as being “confined to the bowel wall” which is incorrect. Dukes' stage A is invasion into, but not through, the bowel wall whereas Dukes' stage B is invasion through the bowel wall but without lymph node metastases.7 Dukes' stages A and B combined could have been correctly labelled as no lymph node metastases. In the methods section the authors state that where only the TNM stage was available, TNM categories were converted to Dukes' stages and where stage was not explicitly stated this was reconstructed from information records (in 45% of cases in the study). It is hoped that the correct descriptions of Dukes' staging were used for this encoding process.

While it is understandable that the authors have chosen to use a well established staging system, the Dukes' system does have a major flaw which is improved by the TNM system.10 The Dukes' B stage contains tumours which may be in the pT3 stage (tumour invades through the muscularis propria into the subserosa or into non-peritonealised pericolic or perirectal tissues) or pT4 stage (tumour directly invades other organs or structures and/or perforates visceral peritoneum) both with a pN0 nodal stage. The prognosis for pT3 pN0 tumours is relatively good whereas the prognosis for pT4 pN0 tumours is poor with a high risk of local recurrence.8 11

A final histopathology observation on this study concerns the threshold set for the number of lymph nodes examined. The authors state that “it is generally considered that at least 12 lymph nodes should be examined for accurate staging” but this is not a universally accepted threshold. The recommendation from the Royal College of Pathologists in the UK is that all lymph nodes identified in the resection specimen should be examined histologically but does not specify an arbitrary minimum number12 as it is recognised that the number of lymph nodes identifiable in a specimen varies with a number of factors which include the type of specimen, extent of resection, preoperative chemotherapy and/or radiotherapy, and the patient's immune response, in addition to the diligence with which the pathologist dissects the specimen. In a recent study in a teaching hospital with optimal dissection and sampling of the entirety of each lymph node the median number of lymph nodes identified was 124 so a blanket statement that “at least 12 lymph nodes” should be examined is an unhelpful pressure on hard pressed diagnostic histopathologists.

These criticisms should not obscure the valuable information that is reported in this study but they do suggest that a greater histopathology input at its inception would have been helpful.



Editor,—We thank Dr Cross for his interest in our paper (

). The issue he raised concerning the definition of stage at diagnosis is important for meaningful comparison of cancer statistics between populations. We are pleased to offer the following clarifications.


We confirm that we incorrectly referred to Dukes' A and B colorectal cases as “confined to the bowel wall.” As pointed out by Dr Cross, Dukes' B includes tumours perforating the visceral peritoneum and directly invading other organs (T4), and our category “Dukes' A and B” encompassed T1-T4 N0 M0. A better definition of this category would have been “absence of lymph node and distant metastasis.” The stage definitions of Hermaneck and Sobin1-1 were in fact used for the encoding process, as stated in the article. Furthermore, they were specified a priori in the study protocol as the reference standard, so as to avoid problems of comparison between registries.


We used Dukes' instead of TNM because in 1990 Dukes' was the most widely used staging procedure in all cancer registry areas. The other reason we decided to use Dukes' was that from the information available to us (patients' clinical and pathological notes), it was not possible to clearly separate pT3pN0 from pT4pN0 cases (all Dukes' B) which, as Dr Cross pointed out, differ markedly in prognosis. Furthermore, it was often impossible to distinguish pT2 from pT3 cases, and for this reason we considered Dukes' A and B together. The population based nature of our study implied the use of information from numerous hospitals and pathology laboratories, which had various ways of recording information. Greater standardisation of stage reporting in pathology notes would be highly desirable.

We are currently engaged in a new high resolution EUROCARE study on colorectal cancers diagnosed in 1997 when the TNM system was more widely used than in 1990. However, the distinction between pT categories is still not adequately made in a considerable number of cases.


Our statement that “at least 12 lymph nodes should be examined for accurate staging” derives from the International Documentation System, referenced by Fielding and colleagues1-2 in their article in 1991. This contains the recommendation that, “Before deeming a radical resection to be without lymph node metastasis, it is recommended that at least 12 lymph nodes be examined . . .”. Furthermore, the 1993 TNM supplement1-3 states: “histological examination of a regional lymphadenectomy specimen . . . will ordinarily include 12 regional lymph nodes,” and this number is considered “adequate for staging”.

Our analysis used four categories for number of nodes examined: 0, 1–5, 6–11, and ⩾12; choosing other division points did not change the rank of the registries. However, it is interesting to note that in the hospital study taken as an example by Dr Cross1-4, 12 or more lymph nodes were examined in 50% of cases—well above the percentages reported by our study (range 2–31%). This suggests that in 1990 the extent of resection or the thoroughness of the pathological examination (or both) would not be considered adequate by today's standards.


  1. 1-1.
  2. 1-2.
  3. 1-3.
  4. 1-4.
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.