Section of Proctology

Transcription

1 13 Voluime 67June Section of Proctology President Basil Morson MD Meeting 5 December 1973 President's Address The Polyp-cancer Sequence in the Large Bowel by Basil Morson MD (St Mark's Hospital, City Road, London EC]) t is not sufficiently appreciated that cancer of the colon and rectum is one of the commonest types of malignant disease. Mortality statistics tell us that it is second only to lung cancer as a cause of death in the general population of England and Wales (Registrar-General 1972) but when its relatively much higher cure rate is taken into consideration then the incidence rates of lung and bowel cancer may be very similar. Recent figures from the United States have shown that large bowel cancer is becoming commoner and now has a higher incidence in the total population than any other cancer except that of the skin (Silverberg & Holleb 1973). Cancer of the colon and rectum is therefore an important public health problem. t is well recognized that the identification of precancerous conditions is the basis of cancer prevention. At the present time isolated polyps, familial polyposis and ulcerative colitis are the conditions which are known to predispose to cancer of the large bowel. The concept that most cancers of the colon and rectum evolve from isolated adenomatous polyps and villous adenomas is sometimes known as the polyp-cancer sequence. t is essential that the medical scientist should at all times be aware of what questions he is trying to answer: only by finding answers to relevant questions shall we make progress. Granted that a substantial body of evidence in support of the polyp-cancer sequence is already available (Morson & Bussey 1970) the following questions require to be answered in detail: (1) Do all adenomatous polyps and villous adenomas inevitably become cancerous? (2) Do all cancers of the colon and rectum evolve from pre-existing adenomatous polyps and villous adenomas? (3) How long does it take for the polyp-cancer sequence to evolve? (4) f some cancers do not arise from pre-existing polyps what is the alternative mechanism of histogenesis? n this Address aspects of the morphology of the polyp-cancer sequence will be described, together with some recent statistics from St Mark's Hospital about the malignant potential of isolated adenomatous polyps and villous adenomas. Other evidence will also be presented which gives some idea of the life history of the polyp-cancer sequence. Morphology ofthe Polyp-cancer Sequence t is common to see tumours in the colon and rectum which are partly benign and partly malignant. This can be obvious on macroscopic inspection. Sometimes, however, the benign or malignant component may be apparent only as a result of microscopic examination. A macroscopically benign adenomatous polyp or villous adenoma should always be carefully examined for evidence of invasive cancer, because the latter may be limited to small microscopic foci. Occasionally, what appears to be a benign adenomatous polyp is found on microscopic examination to be composed entirely of cancer (so-called polypoid carcinoma). Whatever their relative proportions, the presence of contiguous benign and malignant tissue is some evidence that the cancer arose from a previously benign tumour. n a series of 1961 malignant tumours seen at St Mark's Hospital from 1957 to 1968 there were 278 (14.2%) with evidence of contiguous benign tumour, either adenomatous polyp or villous

2 452 Proc. roy. Soc. Med. Volume 67 June 1974 Table 1 Early cancer of the rectum, St Mark's Hospital, : incidence of origin in benign tumour No. of No. arising in Extent ofspread cases benign tumour Limited to submucosa (56.6 Y.) Limited to muscle (18.3',) Extramural spread (7.6%) All cases (10.7 %) adenoma. Table 1 shows, in an equally large series of cases, how the frequency with which benign tumour is found in continuity with a cancer varies with the extent of spread of the malignant component. f the latter showed spread into the extramural tissues of the wall of the colon or rectum, benign tumour was found in a frequency of only 7 %. With spread in continuity limited to the wall of the bowel it rose to nearly 20%, but if there was invasion of only the submucosal tissues it was nearly 600%. This approach to the study of the polyp-cancer sequence suggests that at least half, and probably nearer two-thirds, of all cancers of the colon and rectum arise from previously benign adenomatous polyps or villous adenomas. t is probable that as cancers spread through the bowel wall so they expand on the mucosal surface and tend to destroy surviving benign tumour. Malignant Potential ofadenomatous Polyps and Villous Adenomas Adenomatous polyps, villous adenomas and their synonyms are names given to describe different histological types of benign neoplastic polyp. There is a histological spectrum, with the typical adenomatous polyp (tubular adenoma) at one end, and the typical villous adenoma (villous papilloma) at the other. The intermediate histological type has been called 'papillary' adenoma, 'villo-glandular' or 'tubulo-villous' adenoma. Whatever the name used, the histological structure is intermediate between adenomatous polyp and villous adenoma and will be described as such here. t will be seen from Table 2 that the adenomatous polyp (75%) is far more common than the typical villous adenoma (10 %) and that about 15% of tumours have an intermediate structure. f a series of tumours part malignant and part benign is examined for the relative frequency Table 2 Adenomatous polyps and villous adenomas: frequencies of histological types Histological type No. Y. Tubular adenoma Villous adenoma ntermediate Total with which the three different histological types of polyp are represented in the benign component, further useful information is obtained: it is found (Table 3) that they are equally represented. This can be interpreted as meaning that about onethird of cancers arise from adenomatous polyps, one-third from villous adenomas and one-third from the intermediate type. Bearing in mind the much greater frequency of adenomatous polyps, these figures suggest that the villous growth pattern has a greater potential for malignancy than the typical adenomatous polyp. Table 3 Frequencies of histological types in benign component of partly malignant tumours Histological type No. %, Tubular adenoma Villous adenoma ntermediate Total The malignancy rate for the different histological types of polyp can be assessed by an analysis of the frequency with which cancer is found in a series of tumours which are either wholly or partly benign. n this context 'cancer' is defined as invasion across the line of the muscularis mucosm. Table 4 shows that the general malignancy rate for the common adenomatous polyp is only 5 Y, compared with 40% for villous adenomas. The intermediate type of polyp, at 22%, seems to come nearer to the villous growth pattern than to the adenomatous polyp. These figures emphasize that villous adenomas have much the greatest malignant potential. How can this be explained? Table 4 Adenomatous polyps and villous adenomas: relationship of histological type to incidence of carcinoma Total Histological type cases No. %/ Tubular adenoma Villous adenoma ntermediate With malignancy 14 t has been clear for many years that there is a close relationship between the size of polyps and their malignant potential. The St Mark's Hospital experience is shown in Table 5. The malignant potential of adenomatous polyps and villous adenomas (including the intermediate type) is very low indeed for tumours under 1 cm diameter; with a diameter of 1-2 cm the malignancy rate increases to one in ten, but nearly half of all polyps with a diameter over 2 cm contained evidence of invasive cancer. These figures obviously have great practical importance for radiologists, surgeons and pathologists.

3 15 Section ofproctology 453 The common adenomatous polyp is usually small and very rarely contains invasive cancer (Muto et al. 1974). As they grow, however, so the malignant potential increases, and about one-third of those over 2 cm in size are malignant. But even small villous adenomas have a malignancy rate of about 10% and cancer is found in about 50% of those over 2 cm in diameter. The malignancy rate for the intermediate histological type approaches much closer to that for villous adenomas than to that for adenomatous polyps. The villous type of growth pattern usually presents as a larger tumour than the ordinary adenomatous polyp. Although there are important differences in the frequency and behaviour of adenomatous polyps, villous adenomas and the intermediate histological type, it must be emphasized that the cytological characteristics in all three varieties are the same. t is essential to distinguish between the histology, or tissue architecture, of these tumours and their cellular features. t is. customary to regard the latter as an epithelial dysplasia (or atypia) which can be graded as mild, moderate or severe. Many histopathologists would regard severe dysplasia as synonymous with carcinoma-in-situ. However, this is an expression best avoided in the diagnosis of tumours of the colon and rectum because of its controversial meaning. The grade of dysplasia in polyps is not always uniform and foci of severe dysplasia may be seen in a tumour which otherwise has mild or moderate dysplastic changes. The criteria for judging the degree of epithelial dysplasia include nuclear changes such as enlargement, pleomorphism, loss of polarity, stratification, and an increase in the number of mitotic figures, some of which may be abnormal forms. With increasing severity of dysplasia there is loss of differentiation and usually, but not invariably, a decreasing amount of mucin secretion. The diagnosis of malignancy is made only by observing invasion across the line of the muscularis mucosev. Once this barrier has been breached there is potential for metastasis, although a distinction must be made between true cancer and pseudocarcinomatous invasion (Muto etal. 1973). The grade of epithelial dysplasia is valuable in the study of the malignant potential of adeno- Table S Adenomatous polyps and villous adenomas: size of tumour and incidence of malignancy Diameter of Total With malignancy tumour (cm) cases No. Y. < > Table 6 Adenomatous polyps and villous adenomas: grade of dysplasia and incidence of carcinoma Grade of Total With malignancy dysplasia cases No. % Mild Moderate Severe matous polyps and villous adenomas. Table 6 shows that mild dysplasia is associated with a low malignant potential, but that one-third of all polyps with severe dysplasia contain invasive cancer. n other words, there is a clear correlation of increasing epithelial dysplasia with increasing malignant potential. The St Mark's figures (Muto et al. 1974) also show that small polyps very rarely show severe dysplasia, but that when they do there is a high malignant potential. As polyps get larger the influence of the degree of dysplasia on malignant potential diminishes. Size seems to have paramount importance in judging the risk of cancer. The relationship of grade of dysplasia to histological type of polyp is also interesting (Muto et al. 1974). The malignancy rate for all adenomatous polyps with mild or moderate dysplasia, which are the great majority, is low. Severe dysplasia in adenomatous polyps is uncommon but is associated with a relatively high malignant potential (about 25 %). Villous adenomas, on the other hand, often show severe dysplasia and this may be one important explanation for their high malignant potential compared with adenomatous polyps. Multiple Benign and Malignant Tumours There is abundant evidence that most cancers of the colon and rectum arise from previously benign adenomatous polyps and villous adenomas (Grinnel & Lane 1958, Welch 1964, Morson & Bussey 1970, Potet & Soullard 1971). Granted the importance of this relationship, a study of multiple benign and malignant tumours gives useful information about groups of patients who are at increased risk from neoplastic disease of the large bowel. Out of a total of 3002 patients at St Mark's Hospital, 2412 had only one tumour, and 590 (19.7 %) had multiple synchronous tumours, either benign or malignant. Thus one in five of all patients with neoplastic disease of the large bowel had more than one benign or malignant tumour somewhere in the colon or rectum. This fact has important clinical implications. t means that meticulous examination of the whole large intestine is essential for patients who have presented with a tumour in one part of it. n the same series of 3002 patients, 210 (7%) developed a second or metachronous benign or

4 454 Proc. roy. Soc. Med. Volume 67June 1974 malignant tumour during subsequent follow-up examinations. This figure is probably an underestimate because the study covers a period of time before the introduction of colonoscopy. Moreover, not all the patients had sufficiently meticulous follow-up examinations by air contrast barium enema, which is essential for proper investigation of the whole colon. t is likely that the risk of a patient developing a second tumour is at least one in ten and probably greater. This also illustrates the importance of follow-up of all patients who have had an adenomatous polyp, a villous adenoma or a cancer removed from any part of the large intestine. Studies at St Mark's have also shown that the risk of developing cancer increases with the number of benign adenomatous polyps or villous adenomas. Metachronous Cancer ofthe Colon and Rectum Patients who have had a cancer removed from one part of the large bowel are at risk from developing a further malignant tumour in any remaining large intestine (Bussey et al. 1967, Heald & Lockhart-Mummery 1972). The interval between the first and second cancer varies from two to 31 years in the St Mark's Hospital series, with an average of 13.5 years. The cumulative risk of developing a second cancer is illustrated in Fig a POLYPS N FRST SPECMEN i..o ALL CASES /, _ / 'NOPOLYPS N / - FRST SPECMEN years Time after first tumour Fig 1 Cumulative risk ofdeveloping a second primary intestinal carcinoma This increases with the length of follow-up, reaching about 5% after 25 years, but the risk varies according to whether or not adenomatous polyps or villous adenomas were present in the first operation specimen. f they were present then the risk of a second cancer rises steeply with time to reach 10% at about 25 years. n the absence of such polyps in the first specimen the curve flattens off to a metachronous cancer rate at 25 years of only about 4%. This is evidence, first that there is an important relationship between polyps and cancer, and, secondly, that the polyp-cancer sequence is probably a process which evolves over a period of many years. Life History ofthe Polyp-cancer Sequence The time required for the polyp-cancer sequence to evolve can be measured by a study of the age distribution curves of patients with polyps and cancer. Fig 2 reveals that polyps appear on the average about four years before cancers. This must be a considerable underestimate because whereas age at diagnosis of cancer may be fairly accurate, age at diagnosis of a polyp is likely to be very approximate because it is usually symptomless. Further comment about the value of age distribution curves as applied to the polyp-cancer sequence is given below, under the heading familial polyposis. 25, (1) 10 z S SNGLE ADENOMA SNGLE CARCNOMA 58.1 years, 62.1 years., JO ' les / S / S / St _.,.7b b 8b AGE (Years) Fig 2 Age distributionfor benign and malignant intestinal tumours t is rarely possible to make direct observations on the polyp-cancer sequence because polyps are usually removed by local excision. The fate of four patients with adenomatous polyps in the rectum who refused treatment is seen in Fig 3: n Cases 1 and 2 the diagnosis of cancer was made about five years later; in Case 3 the polyp-cancer sequence took over twelve years to evolve; Case 4 illustrates the observation that a histologically proven adenomatous polyp may remain for at least ten years without becoming malignant. t has already been shown that the malignancy rate for all adenomatous polyps is only about 5 %, which of course suggests that many, and probably most, adenomatous polyps never evolve into cancer. d' d1 O.--=_ 3? 0.._ ' O Years Fig 3 Fate offour untreatedpolyps. 0 adenomatous polyp. 0 cancer diagnosed 16

5 17 Section ofproctology 455 t is easier to make direct observations over a long period of time on villous adenomas because these are usually large tumours which are prone to recurrence after local excision. Cases 1-7 and Case 9 in Fig 4 illustrate the observation that villous tumour, proven histologically, will remain benign for up to twenty years without becoming malignant. n Case 8 benign tumour only was observed for nearly thirty years before malignant change was detected. n Case 10 the evolution of the polyp-cancer sequence took at least ten years. All these patients had repeated local excisions for benign tumour at the same site in the rectum and it is of course possible that this delayed the onset of malignancy. Although it has already been shown that villous adenomas have a high malignant potential it is likely that the polypcancer sequence, as judged by this evidence, evolves over many years and, as with adenomatous polyps, it is likely that some villous adenomas may never become cancerous R C f a 9 f 10 a, o5 15! Years Fig 4Life history of10 villous adenomas. --period of observation ofbenign tumour. 0 cancer The life history of the polyp-cancer sequence is probably very variable. The study of metachronous cancer rates, age distribution curves and these direct, if selected, observations suggests that it is never less than five years, averages years, but may even cover a normal adult life span. Further support for this opinion is obtained from a study of the polyp-cancer sequence in familial polyposis. The Polyp-cancer Sequence in FamilialPolyposis n this genetically predetermined condition the mucous membrane of the large intestine is covered by hundreds or thousands of polyps. Mostly these have the structure of tubular adenomas (adenomatous polyps) but it is not sufficiently well recognized that a minority are villous and others have an intermediate histological structure. Patients with familial polyposis will almost inevitably get cancer if left untreated and all the evidence suggests that the cancers arise from the polyps (Bussey 1970). 25., i cinlo H z 0. a - # m W POLYPOS s CANCER 27.1 years, years " / v w lvs AGE (Years) Fig S Age distributionforpolyposis and cancer A study of the life history of the polyp-cancer sequence in familial polyposis gives very useful information which is relevant to our ideas concerning the time it takes for cancer to evolve from isolated adenomatous polyps or villous adenomas. For example, the age distribution curves for polyposis and cancer illustrated in Fig 5 show that the average age at diagnosis of polyposis without cancer is about 27 years and of polyposis with cancer, about 39 years. This gives a time interval of about twelve years between the diagnosis of polyposis and the later development of cancer. As with the age distribution curves for isolated polyps and cancers, measurement of the age at onset of polyposis is inaccurate, though probably much less so than for isolated lesions. n any case the time interval of twelve years is likely to be an underestimate of the chronology of the polyp-cancer sequence in these circumstances. More information about the time it takes for cancer to develop in patients with polyposis comes from analysis of the age at onset of polyposis and cancer in patients who, for one reason or another, did not have any treatment for their disease. Mostly these patients were under care at St Mark's Hospital before the operation of total colectomy and ileorectal anastomosis was established about twenty-five years ago. t will be seen from Table 7 that, of 59 patients with polyposis observed during five years, only 7 (12%) developed cancer. There were thus 52 patients who harboured polyps for five years without malignant change. The cancer rate between five and Table 7 Familial polyposis: length of precancerous phase Period No. ofcases No. surviving Developed cancer (years) observed without cancer No. Y Over `11.

6 456 Proc. roy. Soc. Med. Volume 67 June 1974 ten years increased to 25 %, but there were still 29 patients with polyps which remained benign for ten years. At years the cancer rate increased to just over 30%, and there were still 13 patients who had had polyps for fifteen years without any malignant change. The cancer rate at years was over 50 %, but 4 patients had polyposis which remained benign for twenty years. Lastly, there Were 3 patients in which the polyp-cancer sequence took over twenty years. These figures lend support to the concept that the evolution of cancer of the colon and rectum from adenomatous polyps and villous adenomas takes at least five years and may be more than twenty years, but on the average lies between ten and fifteen years. t must also be remembered that only one or a few cancers develop in polyposis and that most of the benign tumours in this condition do not become malignant during a normal life span. t has already been shown that the malignancy rate for adenomatous polyps is only about 5 %, and for villous adenomas about 40%. This must mean that many adenomatous polyps and villous adenomas never become malignant. For many years it has been customary at St Mark's Hospital to treat patients with familial polyposis by total colectomy and ileorectal anastomosis. The polyps in the rectum are removed by fulguration and recurrences of benign tumour at subsequent follow-up examinations are also treated by this method. n Fig 6 the risk of cancer in the rectal stump after colectomy and ileorectal anastomosis is given for 86 patients followed for up to 25 years. So far only 2 patients have developed cancer of the rectum, at two and NUMBER OF PATENTS 40 AT RSK Years TME AFTER OPERATON Fig 6 Risk ofrectal cancer after colectomy and ileorectal anastomosisfor polyposis 18 at seven years after colectomy. The longest follow-up is for 11 patients who have remained free of cancer for between twenty and twenty-five years. This low incidence of cancer of the rectum after colectomy and ileorectal anastomosis at St Mark's Hospital is in great contrast to that reported from the Mayo Clinic (Moertel et al. 1970), and requires some explanation. t may be significant that the rectal stump was kept free of polyps by fulguration in all the St Mark's patients. n other words, removal of polyps may mean prevention of cancer. t is also possible that the operation of total colectomy and ileorectal anastomosis so changes the environment of the rectal mucosa that neoplastic activity is inhibited. Hence, perhaps, the observation of regression of polyps in the rectum after this operation. The high cancer rate in the rectum in the Mayo Clinic series may, however, be explained by different methods of selection and management of patients with polyposis. Conclusions At the beginning of this Address four questions were asked for which answers can now be attempted. First, do all adenomatous polyps and villous adenomas inevitably become cancerous? The answer must be 'no', if only because the prevalence of these tumours is so much greater than the prevalence of cancer. Evidence has been presented which shows that the risk of cancer varies with the size of the polyp, with its histological type and the degree of epithelial dysplasia. Generally speaking, the risk of cancer is very low indeed for tumours under 1 cm in diameter but increases to 50 % when the polyp is over 2 cm in diameter. Villous adenomas are usually larger than adenomatous polyps. Villous adenomas have considerably greater malignant potential than adenomatous polyps. Apart from the important issue of size, it has been shown that the cancer rate increases with the degree of epithelial dysplasia, and severe dysplasia is more common in villous adenomas than in adenomatous polyps. However, the evidence also suggests that many adenomatous polyps and villous adenomas never become malignant. Second, do all cancers of the colon and rectum evolve from pre-existing adenomatous polyps and villous adenomas? The evidence suggests that the great majority of cancers of the colon and rectum have evolved through the polyp-cancer sequence. The answer to the fourth question can be conveniently answered now. f some cancers do not arise from adenomatous polyps and villous adenomas what is the alternative mechanism of

7 19 Section ofproctology 457 histogenesis? Apart from the special problem of cancer in ulcerative colitis no alternative has yet been described. The concept of cancer 'de novo', whatever this means in morphological terms, remains in the realm of hypothesis. How long does it take for the polyp-cancer sequence to evolve? t is not possible to give an accurate answer, but the study of patients with isolated polyps and familial polyposis suggests that the sequence evolves slowly. On the average it takes years; it may take as little as five years or as long as 25 years. Because adenomatous polyps and villous adenomas have significant potential for malignant change, every opportunity should be taken to discover and remove them. At a time when the surgical cure rate for cancer of the colon and rectum shows no sign of improvement, prevention becomes an issue of great clinical importance. The introduction of the double contrast barium enema and the colonoscope means that we can now accurately examine the entire large bowel. Patients who have had an adenomatous polyp, a villous adenoma or a cancer in one part of the large bowel have an increased risk of tumours in the remaining large intestine. This means that the discovery of a benign or malignant tumour should always be accompanied by intensive investigation of the whole large bowel by air contrast barium enema, colonoscopy or both. Patients who have had polyps or cancers removed have a significantly increased risk of producing further polyps or cancers in any remaining large intestine, and should therefore have regular follow-up examinations by air contrast barium enema or colonoscopy, whichever is clinically appropriate. Study of the life history of the polyp-cancer sequence suggests that these follow-up examinations can be performed at three-yearly intervals provided the large bowel was tumour-free at the last examination. Only by persistent follow-up and meticulous examination of the whole large bowel will a significant measure of cancer prevention be achieved. Acknowledgments: wish to record my special thanks to Dr H J R Bussey for his help and advice. Some of this work was carried out by Dr T Muto during his appointment as a World Health Organization Research Fellow in the Department of Pathology at St Mark's Hospital. REFERENCES Bussey H J R (1970) Thesis, University of London Bussey H J R, Wallace M H & Morson B C (1967) Proceedings ofthe Royal Society ofmedicine 60, Grinnel R S & Lane N (1958) nternational Abstracts ofsurgery 106, 519 Heald R J & Lockhart-Mummery H E (1972) British Journal ofsurgery 59, Moertel C G, Hill J R & Adson M A (1970) Archives ofsurgery 100, Morson B C & Bussey H J R (1970) Predisposing Causes of ntestinal Cancer. Current Problems in Surgery, Year Book Medical Publishers, Chicago Muto T, Bussey H J R & Morson B C (1973) Journal of ClinicalPathology 26, 25 (1974) (in preparation) Potet F & Soullard J (1971) Gut 12, 468 Registrar-General (1972) Statistical Review of England and Wales for the year 1970, Part 1: Tables, Medical. HMSO, London Silverberg E & Holleb A (1973) Ca - A Cancer Journalfor Clinicians 23, 2-27 Welch C E (1964) Polypoid Lesions of the Gastrointestinal Tract. W B Saunders, Philadelphia Meeting 24 October 1973 The following cases were presented: Anal Varices Mr C G Marks (for Mr T Rowntree) (St Mark's Hospital, London EC V2PS) rradiation Strictures of leum and Rectum Mr R D Rosin (for Mr A York Mason) (St Helier Hospital, Carshalton, Suirrey) Carcinoma Arising in Loop leostomy in a Patient with Ulcerative Colitis Mr W Miller & Dr R H Riddell (for Mr Todd) (St Mark's Hospital, London EC] V2PS) Anal Stricture Following Diathermy of Perianal Warts Mr R H Grace (for Mr H E Lockhart-Mummery) (St Thomas' Hospital, London SE] 7EH) Adult Hirschsprung's Disease Mr R H S Lane (for Mr A Parks) (St Mark's Hospital, London EC] V2PS) Benign Duodeno-colic Fistula Mr M M Orr (for Mr P R Hawley) (St Bartholomew's Hospital, London ECJA 7BE) Pseudomembranous Enterocolitis Dr A B Price & Mr R H Grace (for Mr A E Thomson) (St Mark's Hospital, London EC V2PS) Villous Papilloma of Rectum Associated with Multiple Colonic Polyps Removed per Colonoscope Mr B J Stoodley (for Mr M Pemberton) (Chase Farm Hospital, Enfield, Middlesex EN2 8JL)