Br. J. Surg. 1991, Vol. 78, October, 1162-1 166
B. M. Stephenson*?, P. J. Finan*,J. Gascoyne?, F. Garbettt, V. A. Murday? and D. T. Bishop? *Department of Surgery, Leeds General Infirmary and ?Imperial Cancer Research Fund, Genetic Epidemiology Laboratory, University of Leeds, Leeds, UK Correspondence to:
Mr B. M. Stephenson, Imperial Cancer Research Fund, 3K Springfield House, Hyde Terrace, Leeds LS2 9LU, UK
Frequency of familial colorectal cancer Familial clustering of cancer is not uncommon. Thefrequency of familial colorectal cancer was estimated by taking family histories from 100 patients presenting with apparently sporadic colorectal cancer. Compared with controls, the relative risk of a positive family history for colorectal cancer was 4.6. Life-table methods were used to examine the observed to expected mortality f r o m colorectal cancer. Overall there was afourfold increase in mortality rate (P < O.OOOl), which was greatest in female relatives of patients with colonic cancer (P < 0.001). Three families with dominant inheritance of colorectal cancer and one family with Lynch type II syndrome were identijied. Nine per cent of patients had siblings who had developed colorectal cancer a median of 4 years before the diagnosis of the index patient (range 1-17 years). It is recommended that a careful family history should be obtained f r o m all patients with colorectal cancer. Where a positive history is obtained a geneticist may determine empirical risks f o r the development of colorectal cancer and the appropriate method of surveillance may be selected.
In 1988, in England and Wales, there were nearly 17000 deaths from colorectal cancer’. The natural history of the disease and the diagnostic methods fulfil many of the criteria required of a screening programme. Screening of asymptomatic individuals using faecal occult blood tests has had a disappointing compliance rate*. Strategies aimed at the early detection of colorectal cancer in high-risk groups form part of the clinical management of patients with conditions such as long-standing inflammatory bowel disease and adenomatous polyposis coli. Screening the immediate family of patients with adenomatous polyposis coli has become routine clinical practice and the identification of an affected individual mandates the offer of a prophylactic colectomy. In contrast the screening for dysplasia and cancer in patients with ulcerative colitis is part of routine clinical practice but the ‘costs’ of screening possibly outweigh the
benefit^'^. The clustering of cancer in certain families is well recognized. In 1904, Watkins4 described a family with a high incidence of colorectal cancer and pondered ‘the question of the inheritance of cancer’. Warthin5 analysed the families of 1000 patients with several types of histologically verified cancer and is credited as being the first to suggest that the incidence of malignant tumours in these families might be due to an inherited susceptibility to cancer. Wooltd analysed the death certificates of relatives of patients with colorectal cancer and concluded that there was a genetic component to the disease, noting that first-degree relatives of patients dying from cancer of the colon or rectum had a threefold increased risk of developing colorectal cancer compared with the general population. The principal aim of this study was to establish the incidence of colorectal cancer in the first-degree relatives of patients presenting with colorectal cancer to one consultant surgeon (P.J.F.)and to establish the feasibility of offering screening to the remaining first-degree relatives.
in the study. The detailed interview usually took place in the patient’s own home and was taken by B.M.S. or F.G. The age and cause of death in relatives was verified, where possible, by reference to hospital pathology records, death certificates or from regional cancer registries. Particular care was taken to avoid biased selection of cases and no referrals had occurred as a consequence of a family history of colorectal cancer. Patients with a history of adenomatous polyposis coli were excluded. In all, 124 consecutive patients were contacted of whom nine refused to give details of their family. Patients who were unable to give a full family history or who were unsure as to the age and cause of death in their first-degree relatives were excluded ( n = 15). The results of the first 100 complete family histories are presented. As a control group, family histories were taken from age- and sex-matched individuals with various non-malignant conditions. Tables from the Office of Population Censuses and Surveys (OPCS) for 1988 were used to calculate the expected number of deaths among first-degree relatives in 10-year age groups’. Each first-degree relative contributes ‘years at risk’ only to the decades through which they have lived. The expected incidence of colorectal cancer is low In the early decades and higher later in life. Only those relatives living through later decades provide ‘years at risk’ during which there is a significant incidence of colorectal cancer. This type of analysis eliminates the problem of individuals having different numbers of relatives and controls for their ages. The Poisson distribution was used to examine the significance of any difference between the observed and expected number of deaths.
Results
During an outpatient visit all patients who had undergone treatment for primary colorectal cancer were invited to give a family history to identify first-degree relatives (parents, siblings and children) for screening. The family history of a deceased index case was obtained from the spouse and confirmed by a first-degree relative before inclusion
The 100 index patients were aged from 36 to 89 years (mean 71 years) with an equal male:female ratio. Eighty-six were elective patients and 14 presented as emergencies. The most frequent site of cancer was the rectum and rectosigmoid (45 patients). There were 23 caecal and right colonic tumours, a distribution expected in an unselected population. Twenty-three of the index cases gave a history of colorectal cancer in a first-degree relative compared with only five in the control group (relative risk for a positive family history of 4.6). There was no difference in the incidence of any other common type of malignancy (Table I). There was no significant difference in the age, site or the Dukes’ classification of the cancer in the patients with a family history compared with the remaining patients with no family history. Of the 23 index cases with a positive family history, the death in first-degree relatives was confirmed in only 20 (relative
1162
0007-1323/91/101162-05
Patients and methods
(c 1991 Butterworth-Heinemann
Ltd
Frequency of familial colorectal cancer: B. M. Stephenson et al. I
111
b
I
Ill
C
Ill
d
111
Figure 1 Family rrees uf patients wirh ‘sporadic’ colorectal cancer. a< are examples o f fanlilies with colorectal cancer inherited in a dominant ,firshion (Lynch t.vpe I ) und d is an esample o f Lynch type I I syndrome. In e the sister of the index case died.from colorectal cancer 3 years before [he diagnosis in our case. JJ, Index case; @, deceased; 0, male; 0, femule; 0 . colorectul cancer: @, hreasr cancer; 0 , carcinomatosis
e risk for a positive family history of 4).In two patients the cause of death of the parent alleged to have died from colorectal cancer between 30 and 50 years ago could not be traced, and one index patient’s brother died from gastric cancer when it was alleged that he had died from colorectal cancer. The 20 confirmed families had 23 first-degree relatives with confirmed colorectal cancer. There were three families with colorectal cancer inherited in a dominant fashion (defined as at least three first-degree relatives with colorectal cancer) (Figure 1 a-c) and one family with Lynch type I1 syndrome (three first-degree relatives with colorectal cancer and other adenocarcinomas (breast, ovarian and endometrial)) (Figure Id). Nine patients with a confirmed family history of colorectal cancer had siblings who had developed colorectal cancer a median of 4 years before the diagnosis of the index patient (range 1-17 years), as illustrated by Figure l e . As the mortality statistics from the OPCS rely upon the diagnosis of colorectal cancer appearing on the death certificate, only those cases confirmed by death certificates were valid when used in the life-table analysis. Of the 23 first-degree relatives from the 20 families, 18 were confirmed by death certificates. The 100 index patients had 840 first-degree relatives from which calculation of the number of years at risk, in 10-year age groups,
Br. J. Surg..Vol. 78, No. IO,Octoberl991
was made. This allowed the analysis of the observed to expected incidence of colorectal cancer from the OPCS statistics. There were 18 observed deaths from colorectal cancer in the firstdegree relatives compared with an expected incidence of 4.1 1 (0bserved:expected ratio of 4.38; P
B. M. Stephenson*?, P. J. Finan*,J. Gascoyne?, F. Garbettt, V. A. Murday? and D. T. Bishop? *Department of Surgery, Leeds General Infirmary and ?Imperial Cancer Research Fund, Genetic Epidemiology Laboratory, University of Leeds, Leeds, UK Correspondence to:
Mr B. M. Stephenson, Imperial Cancer Research Fund, 3K Springfield House, Hyde Terrace, Leeds LS2 9LU, UK
Frequency of familial colorectal cancer Familial clustering of cancer is not uncommon. Thefrequency of familial colorectal cancer was estimated by taking family histories from 100 patients presenting with apparently sporadic colorectal cancer. Compared with controls, the relative risk of a positive family history for colorectal cancer was 4.6. Life-table methods were used to examine the observed to expected mortality f r o m colorectal cancer. Overall there was afourfold increase in mortality rate (P < O.OOOl), which was greatest in female relatives of patients with colonic cancer (P < 0.001). Three families with dominant inheritance of colorectal cancer and one family with Lynch type II syndrome were identijied. Nine per cent of patients had siblings who had developed colorectal cancer a median of 4 years before the diagnosis of the index patient (range 1-17 years). It is recommended that a careful family history should be obtained f r o m all patients with colorectal cancer. Where a positive history is obtained a geneticist may determine empirical risks f o r the development of colorectal cancer and the appropriate method of surveillance may be selected.
In 1988, in England and Wales, there were nearly 17000 deaths from colorectal cancer’. The natural history of the disease and the diagnostic methods fulfil many of the criteria required of a screening programme. Screening of asymptomatic individuals using faecal occult blood tests has had a disappointing compliance rate*. Strategies aimed at the early detection of colorectal cancer in high-risk groups form part of the clinical management of patients with conditions such as long-standing inflammatory bowel disease and adenomatous polyposis coli. Screening the immediate family of patients with adenomatous polyposis coli has become routine clinical practice and the identification of an affected individual mandates the offer of a prophylactic colectomy. In contrast the screening for dysplasia and cancer in patients with ulcerative colitis is part of routine clinical practice but the ‘costs’ of screening possibly outweigh the
benefit^'^. The clustering of cancer in certain families is well recognized. In 1904, Watkins4 described a family with a high incidence of colorectal cancer and pondered ‘the question of the inheritance of cancer’. Warthin5 analysed the families of 1000 patients with several types of histologically verified cancer and is credited as being the first to suggest that the incidence of malignant tumours in these families might be due to an inherited susceptibility to cancer. Wooltd analysed the death certificates of relatives of patients with colorectal cancer and concluded that there was a genetic component to the disease, noting that first-degree relatives of patients dying from cancer of the colon or rectum had a threefold increased risk of developing colorectal cancer compared with the general population. The principal aim of this study was to establish the incidence of colorectal cancer in the first-degree relatives of patients presenting with colorectal cancer to one consultant surgeon (P.J.F.)and to establish the feasibility of offering screening to the remaining first-degree relatives.
in the study. The detailed interview usually took place in the patient’s own home and was taken by B.M.S. or F.G. The age and cause of death in relatives was verified, where possible, by reference to hospital pathology records, death certificates or from regional cancer registries. Particular care was taken to avoid biased selection of cases and no referrals had occurred as a consequence of a family history of colorectal cancer. Patients with a history of adenomatous polyposis coli were excluded. In all, 124 consecutive patients were contacted of whom nine refused to give details of their family. Patients who were unable to give a full family history or who were unsure as to the age and cause of death in their first-degree relatives were excluded ( n = 15). The results of the first 100 complete family histories are presented. As a control group, family histories were taken from age- and sex-matched individuals with various non-malignant conditions. Tables from the Office of Population Censuses and Surveys (OPCS) for 1988 were used to calculate the expected number of deaths among first-degree relatives in 10-year age groups’. Each first-degree relative contributes ‘years at risk’ only to the decades through which they have lived. The expected incidence of colorectal cancer is low In the early decades and higher later in life. Only those relatives living through later decades provide ‘years at risk’ during which there is a significant incidence of colorectal cancer. This type of analysis eliminates the problem of individuals having different numbers of relatives and controls for their ages. The Poisson distribution was used to examine the significance of any difference between the observed and expected number of deaths.
Results
During an outpatient visit all patients who had undergone treatment for primary colorectal cancer were invited to give a family history to identify first-degree relatives (parents, siblings and children) for screening. The family history of a deceased index case was obtained from the spouse and confirmed by a first-degree relative before inclusion
The 100 index patients were aged from 36 to 89 years (mean 71 years) with an equal male:female ratio. Eighty-six were elective patients and 14 presented as emergencies. The most frequent site of cancer was the rectum and rectosigmoid (45 patients). There were 23 caecal and right colonic tumours, a distribution expected in an unselected population. Twenty-three of the index cases gave a history of colorectal cancer in a first-degree relative compared with only five in the control group (relative risk for a positive family history of 4.6). There was no difference in the incidence of any other common type of malignancy (Table I). There was no significant difference in the age, site or the Dukes’ classification of the cancer in the patients with a family history compared with the remaining patients with no family history. Of the 23 index cases with a positive family history, the death in first-degree relatives was confirmed in only 20 (relative
1162
0007-1323/91/101162-05
Patients and methods
(c 1991 Butterworth-Heinemann
Ltd
Frequency of familial colorectal cancer: B. M. Stephenson et al. I
111
b
I
Ill
C
Ill
d
111
Figure 1 Family rrees uf patients wirh ‘sporadic’ colorectal cancer. a< are examples o f fanlilies with colorectal cancer inherited in a dominant ,firshion (Lynch t.vpe I ) und d is an esample o f Lynch type I I syndrome. In e the sister of the index case died.from colorectal cancer 3 years before [he diagnosis in our case. JJ, Index case; @, deceased; 0, male; 0, femule; 0 . colorectul cancer: @, hreasr cancer; 0 , carcinomatosis
e risk for a positive family history of 4).In two patients the cause of death of the parent alleged to have died from colorectal cancer between 30 and 50 years ago could not be traced, and one index patient’s brother died from gastric cancer when it was alleged that he had died from colorectal cancer. The 20 confirmed families had 23 first-degree relatives with confirmed colorectal cancer. There were three families with colorectal cancer inherited in a dominant fashion (defined as at least three first-degree relatives with colorectal cancer) (Figure 1 a-c) and one family with Lynch type I1 syndrome (three first-degree relatives with colorectal cancer and other adenocarcinomas (breast, ovarian and endometrial)) (Figure Id). Nine patients with a confirmed family history of colorectal cancer had siblings who had developed colorectal cancer a median of 4 years before the diagnosis of the index patient (range 1-17 years), as illustrated by Figure l e . As the mortality statistics from the OPCS rely upon the diagnosis of colorectal cancer appearing on the death certificate, only those cases confirmed by death certificates were valid when used in the life-table analysis. Of the 23 first-degree relatives from the 20 families, 18 were confirmed by death certificates. The 100 index patients had 840 first-degree relatives from which calculation of the number of years at risk, in 10-year age groups,
Br. J. Surg..Vol. 78, No. IO,Octoberl991
was made. This allowed the analysis of the observed to expected incidence of colorectal cancer from the OPCS statistics. There were 18 observed deaths from colorectal cancer in the firstdegree relatives compared with an expected incidence of 4.1 1 (0bserved:expected ratio of 4.38; P
