Gut, 982, 2, 49-497 Cancer morbdty n ulceratve colts P PRIOR, S N GYDE, J C ACARNEY, H HOPSON, J A H WAERHOUSE, and R N ALLAN rom the Gastroenterology Unt, General Hosptal, Brmngham, and Cancer Epdemology Research Unt and Department ofpathology, Unversty of Brmngham, Brmngham SUARY Cancer morbdty at all stes has been studed n a seres of 676 patents wth ulceratve colts under long-term revew, of whom more than two-thrds had extensve dsease, and the level and pattern of rsk over tme examned. Age-, sex-, and ste-specfc ncdence rates were used to compute the number of cancers that mght have been expected to occur. A hghly sgnfcant excess of cancers was observed overall but the excess was due entrely to cancers of the dgestve system. In women there was no excess or defct of cancers outsde the dgestve system. In men there was a small defct of cancers of the respratory system. An overall -fold excess colorectal cancer rsk was found n the seres compared wth that n a relevant general populaton after patent-years at rsk had been corrected for surgcal resecton and patents wth colorectal cancer at ther frst referral had been excluded. When these data were expressed n an actuaral form the cumulatve probablty of developng colorectal cancer n the seres was 8% (.5-%) at 25 years, after the dagnoss of ulceratve colts had been establshed. he relatve rsk of developng colorectal cancer was hghest n those patents developng colts before the age of years, and the relatve rsk fell as the age at dagnoss of ther colts ncreased. he pattern of rsk of colorectal cancer over tme suggests that there s an assocaton between cancer and colts n susceptble ndvduals and that the level of rsk s related to age at onset of colts. Prevous studes of the cancer rsk n ulceratve colts -8 have been confned almost exclusvely to colorectal cancer, wth the excepton of the hepatoblary system, whch has receved some attenton. here s lttle nformaton about the cancer ncdence at other stes. We have therefore studed the whole spectrum of cancer morbdty n a seres of patents wth ulceratve colts. hs enabled us to examne some aspects of patent selecton (analyss of clncal seres beng subject to many elements of bas) to provde a bass from whch to approach the specfc problem of colorectal cancer. Estmates of the ncdence of colorectal cancer complcatng ulceratve colts n the early lterature were based on crude percentages. Actuaral methods are now used whch take nto account the decreasng numbers of patents at rsk for cancer over extended perods of observaton. A number of such studes have been publshed and are referenced n two recent reports.y ost studes have been nterpreted as showng a rased and ncreasng rsk Receved for publcaton 26 October 98 of colorectal cancer over tme, although only a few2 4-7 have compared the rsk n the seres wth that n a relevant general populaton. Despte these reports, the magntude and the pattern of rsk s uncertan and vares from one reported seres to another. hese varatons probably reflect dfferences n the composton of the reported seres and n the methods used to evaluate the rsk rather than real dfferences n the underlyng cancer rsk. he dfferng approaches to analyss have been dscussed both n general 6 and n ndvdual seres.7 We have attempted to dentfy and, where possble, to correct for areas of bas that can arse n analysng data from a clncal seres. he ncreasng ncdence of cancer that mght be expected to occur n any seres under long-term revew - that s, an ageng populaton - has been taken nto account when assessng the rsk whch mght otherwse be attrbuted to the presence of ulceratve colts. he effects of dfferng methods of analyss on the pattern of rsk of colorectal cancer over tme are presented. nally, the relatonshp between the rsk of colorectal cancer and age at onset of colts has been examned. 49 Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
Cancer morbdty n ulceratve colts ethods PAIENS: COPOSIION O SERIES A consecutve clncal seres of patents wth ulceratve colts under the care of Professor B N Brooke and Dr W Cooke at the Queen Elzabeth and General Hosptals, Brmngham, comprsng both prmary and referred cases presentng between 94-76, has been studed. he survvors are under regular revew n the Gastroenterology Unt at the General Hosptal. he dagnoss of all patents wth nflammatory bowel dsease has been revewed to dstngush patents wth ulceratve colts and Crohn's dsease as accurately as possble. fty-nne cases have been excluded from the orgnal seres of patents wth ulceratve colts reported by Daly et al.2 he reasons for excluson were polyposs col (two cases), melanoss col (one case), Crohn's dsease (4 cases), and nsuffcent nformaton for analyss (5 cases), leavng a seres of 676 patents for assessment. or untraced patents (N=5) the years of survval are truncated f they exceed the approprate lfe table estmates. Of the 676 patents, 42 (2%) have ded and the vtal status at the end of the revew perod was unknown for 5 patents (2.2%). he clncal status was determned n 95.7% of those known to be alve by clncal examnaton, report from clncan, or personal communcaton. Nearly half the patents were seen n ther frst attack. he remander were referred later n the course of ther dsease. As these 'referred' cases were ncluded, the bas at entry to the seres was nclned towards severe dsease, partcularly as patents were often referred for surgcal treatment. hs accounts for the hgh proporton of patents treated by panproctocolectomy (able ) compared wth other reported seres. It was dffcult to defne able Proporton ofpatents undergong radcal surgery for ulceratve colts among reported hosptal seres Number of Patents % of patents n undergong total total seres radcal surgery seres Nefzger and Acheson (US Army) 96`s 525 4 8 Edwards and ruelove (Oxford) 9645 624 87 4 Watts etal (Leeds) 9664 24 56 27 Bonneve (Copenhagen) 974'` 2 56 7 Present seres (Brmngham) 676 48 65 an accurate date of dagnoss for some of the referred patents. If the date of dagnoss could not be establshed t was taken as the date frst seen at the Queen Elzabeth or General Hosptals. he age-dstrbuton at onset and dagnoss s shown n g.. he extent of the dsease and the frequency of surgcal resecton s shown n able 2. Extensve colts was defned as nvolvement of the large ntestne as far proxmally as the hepatc flexure. he dagnoss of colorectal cancer was confrmed n all cases by a revew of the hstopathologcal materal, and an accurate Dukes's classfcaton was made whenever possble. SAISICAL ANALYSIS Person-years at rsk he date of dagnoss was used as the startng pont for computng person-years at rsk. he analyss was based on 665 person-years at rsk (PYR), yeldng a mean follow-up of 5-8 years per patent (4.2 for men and 6.8 for women). or each patent the length of follow-up was computed from the date of dagnoss to December 976 or to death f ths occurred earler. he survval experenced by the whole seres was expressed as person-years at rsk n terms of sex, age, and nterval from dagnoss. In. 7 l 6 t5 Z4 49 g. Age dstrbuton at onset and dagnoss n the seres of676 patents wth ulceratve colts Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
492 able 2 Pror, Gyde, acartney, hompson, Waterhouse, and Allan Ulceratve colts: dstrbuton by sex, extent of dsease, and operatve procedure Extensve colts Localsed dsease All cases reatment Sex No. % No. % No. % Panproctocolectomy (N=92) 2 9-5 27 4-59 2.5 Colectomy + retaned rectum/ 24 5.7 8 5.6 279 4- leorectal anastomoss (N=46) 7 55-2 65 9-6 48 64-8 8 5-6 7-4 8 6- Conservatve management 5 7-5 79-7 9.2 89-49 22-28 5.2 7 25-97 4-4 267 9-5 All cases 292 4-2 7 7-49 6.5 462 68-24 -7 676 - ales: females = : -5. order to correct for excson of the large ntestne or rectum, the patent-years at rsk accrung after colectomy or panproctocolectomy were subtracted from the total patent years at rsk to gve 'colorectal-years' at rsk. Expected numbers of cancers Incdence rates for cancer at ndvdual anatomcal stes were computed from data held at the Brmngham and West dlands Regonal Cancer Regstry for the years 96-62, the md-pont of the study. he age-, sex-, and ste-specfc ncdence rates were appled to the correspondng patent-years at rsk to compute the number of cancers that mght be expected to occur n the seres durng the perod of revew. he expected numbers of colorectal cancers were computed both for total and for 'colorectal years' at rsk. Colorectal cancers observed n the seres were dvded by mode of dagnoss nto three clncal groups. hese groups reflect possble areas of bas n the selecton of patents whch could affect the estmates of cancer rsk. 'Referred' Symptoms of cancer n ths group were nstrumental n ntatng the frst referral of the patent to the unt, and, therefore, represent selecton for cancer. 2 'Non-symptomatc' he cancers ncluded n ths group were dentfed as ncdental fndngs among patents undergong panproctocolectomy for symptomatc colts shortly after referral to the unt, havng been dagnosed elsewhere. he presence of the cancer was not consdered to have ntated referral. 'Interval' hese cancers were dentfed at revew or at operaton at least one year after the date frst seen at the Unt. Cancers were allocated to these groups on revew of the clncal materal before the tabulatons of expected numbers were drawn up and none was changed subsequently. he pattern of cumulatve relatve rsk (sum of the observed/sum of the expected numbers) of colorectal cancer over tme was frst determned by usng all observed cancers wth expected numbers uncorrected for operaton. he analyss was repeated wth the expected numbers corrected for operaton. hree further analyses were carred out to show the effect that eaoh of the three clncal groups of observed cancers had on the level of rsk and the pattern of rsk over tme, usng as observed numbers: A 'Referred' + 'non-symptomatc' + 'nterval' cancers B 'Non-symptomatc' + 'nterval' cancers C 'Interval' cancers only. In each case the expected numbers were corrected for operaton. A conventonal actuaral analyss was performed usng 'non-symptomatc' + 'nterval' cancers wth 'colorectal' years at rsk. he level and pattern rsk of colorectal cancer over tme was also examned n relaton to age at dagnoss of ulceratve colts for three age-ranges (-29 years, -44 years, and 45+ years) usng (non-symptomatc' and 'nterval' cancers wth the expected numbers corrected for operaton. Observed numbers of cancers Patents who developed cancer durng the perod of revew were dentfed from the clncal records, by searches of the Brmngham Cancer Regstry records, and from death certfcates. Patents referred wth a dagnoss of cancer were categorsed separately f t was consdered that the cancer was nstrumental n ntatng the frst referral. Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
Cancer morbdty n ulceratve colts Interpretaton he statstcal sgnfcance between the observed and expected numbers of cancer was determned usng the Posson dstrbuton. he force of the assocaton between cancer and ulceratve colts has been measured n terms of a Standardsed orbdty Rato (SR) where SR = Expected rate whch, n the context of patent-years at rsk as a common denomnator, can be expressed as observed number/expected number. he more famlar term of 'relatve rsk' has been used n the text. he results are presented by the man anatomcal dvsons, wth expected numbers for colon and rectum corrected for resecton. Results A hghly sgnfcant excess of cancers overall was observed n the seres (able ). he excess was due entrely to cancers dagnosed wthn the dgestve system, for whch the relatve rsk was of the same order for both men and women. In women, no other system or ste was sgnfcantly over- or underrepresented n the observed numbers. In men, the excess overall was sgnfcant at only the 5% level because of a small defct (of margnal statstcal sgnfcance) of cancers of the respratory system. A more detaled analyss of cancers of the dgestve system s shown n able 4. he excess of cancers n the colon and rectum was hghly sgnfcant both for men and women, and the level of rsk for the colon was the same as that for the rectum. A hghly sgnfcant excess of cancers was found n lver and gall bladder. he remander of the tract showed a small defct whch was of margnal sgnfcance and was due manly to the absence of stomach cancers n ths seres. However, when those stes whch showed a sgnfcant excess of cancers had been accounted for - namely, colon, rectum, lver and gall bladder - the expected number for all other stes (29.6) was close to the number observed (27). hrty-fve colorectal cancers were observed n the seres durng the revew perod comprsng 2 referred cancers, non-symptomatc, and nterval cancers. he clncal detals of the cancers ncludng ste, Dukes's classfcaton, and outcome are summarsed n able 5. When no correcton was made to the expected numbers, the rsk of colorectal cancer relatve to the general populaton decreased wth tme from the dagnoss of ulceratve colts from 5*5-fold at fve years to 9.-fold at the end of the revew perod (g. 2, A'). Correcton for operaton resulted n a able systemt Ste All stes Buccal cavty/throat Dgestve system Respratory system Breast/reproductve system Urnary system Retculoendothelal system Skn Unknown prmary Remander Ulceratve colts: cancer morbdty by anatomcal Sex E -77 9-99 -76.4-42 82.6 4-9 7.5 4.47-96 5.4 9 8 57 4 7 4 7-7 -29-2 -69-4. -6-85 -48-22 -8 - - -62-95 -54.86-4 2 4 2 OIE -6-9.8 4.4 4-4-2 - -2 - - -4-9 -6 2-4 2-2-5-6. - -2-2.5 2- t Expected number of cancers corrected for operaton, observed numbers excludng 2 patents referred wth colorectal cancer at frst attendance. E = Expected number of cancers. = Observed number of cancers. * p<-5. ** p<-. *** p<-. ( ) Defct. 7-fold constant cumulatve relatve rsk n the revew perod from fve years after the dagnoss of ulceratve colts had been establshed (g. 2A). he effects of the prevously defned cancer groups on the estmates of cancer rsk are llustrated n g.. he constant 7-fold rsk, whch was obtaned when all cancers (referred, non-symptomatc, and nterval) were ncluded (g. A), was reduced to an -fold rsk when 'referred' cancers were excluded from the analyss (g. B). he relatve rsk stll remaned constant over tme. When the 'nterval' cancers alone were consdered, the pattern of rsk changed. here was an p (*) (*) 49 Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
494 able 4 Ulceratve colts: cancer morbdty wth partcular reference to colon, rectum, lver, and gall bladdert Ste Sex E OIE p All stes -77 9-6 * 9-99 8-9.76 57-8 Dgestve system -6 4 4-4 4-9 7 4-7-5 4-2 Colon.29 5 7-2.89 8 9. -8 - Rectum.2 94 ** -58 7 2-.9 - Lver, gall bladder. 5 45-5 -2 2 95 * -2 7 2-9 Remander of 2-44 -4 dgestve system 2-5 - 4-95 2 (*) All other stes excludng -5 6-5 colon, rectum, 8-2 - lver, gall bladder 29-6 27-9 t Expected number of cancers corrected for operaton, observed numbers excludng 2 patents referred wth colorectal cancer at frst attendance. E = Expected number of cancers. = Observed number of cancers. * p<5 ** p<-. p<o. ()Defct. W. 5 2 5 5 2 25 5 Years from dagnoss of ulceratve colts r. 2 Cumulatve relatve rsk of colorectal cancers: A - Observed number = all colorectal cancers (5); expected number unadjusted for operaton (all age groups). A Observed number - = all colorectal cancers; expected number adjusted for operaton (all age groups). Pror, Gyde, acartney, hompson, Waterhouse, and Allan 25-25- W a5 2-,- u 5 5 -a - 5 ::...B --------------------------- c 5 a a a - 5 5 2 25 Years from dagnoss of ulceratve colts g. Cumulatve relatve rsk ofcolorectal cancers: A - Observed number = all colorectal cancers; expected number adjusted for operaton (all age groups). B - Observed number = 'non-symptomatc' and 'nterval' cancers (2); expected number adjusted for operaton (all age groups). C - Observed number = 'nterval cancers' (); expected number adjusted for operaton. ncreasng relatve rsk of developng cancer up to 5 years from the dagnoss of ulceratve colts wth a constant sx-fold rsk thereafter (g. C). hese results emphasse the dsparty between estmates of rsk that can arse as a result of arbtrary selecton of data. It s generally, although not nvarably,'6 accepted that 'referred' cancers should be excluded from the analyses, as they are selectvely drawn by ther symptoms of cancer from an unknown group of patents wth ulceratve colts n the general populaton. In the remanng analyses, therefore, these referred cancers were excluded and 'nonsymptomatc' and 'nterval' cancers only were consdered. he actuaral analyss showed that the probablty of developng colorectal cancers n the seres ncreased exponentally wth tme (g. 4). Despte the sze of the ntal seres the 95% confdence lmts for the estmaton became ncreasngly wde wth tme. he probablty at 25 years was 8% (.5-%) and 2% (4.5-6%) at years but the fgure at years s based on the addton of only one further cancer. As the expected cancer rsk ncreased at a smlar rate the cumulatve relatve rsk (O/E) over tme remaned constant and at the same level as shown n g. B. he relatve rsk of developng colorectal cancer was 8-fold overall n patents aged between -29 years when ther ulceratve colts was dagnosed and t fell as the age at dagnoss ncreased (able 6, 5 A Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
Cancer morbdty n ulceratve colts4 able 5 Clncal data: cancer morbdty: (A) colorectal cancer Age at onset Intervalfrom onset of Age at death or ofsymptoms Age at symptoms of ulceratve length ofsurvval of ulceratve dagnoss of colts to dagnoss from dagnoss of Dukes's No. Sex colts (yr) cancer (yr) of cancer (yr) Ste of cancer cancer (yr) classfcaton 428 42 54 2 Sgmod colon 54 D 685 29 5 2 ultple colon 52 B 56 4 ransverse colon Survval 4 yr A 57 6 7 Descendng colon Survval 8 yr A* 276 9 54 5 ransverse colon 55 B* 96 2 5 29 ransverse colon 5 D 25 5 ransverse colon 5 D 7 49 2 Descendng colon 49 C* 28 42 54 2 ultple colon Survval 2 yr IC + 6At 69 2 9 8 Sgmod colon Survval yr B* 282 24 49 25 ransverse colon 49 B 66 4 45 Ascendng colon Survval 25 yr A 4 22 44 22 ultple sgmod colon 45 IC + IAt 8 7 2 25 Rectum 4 -t 529 4 55 2 Rectum 56 C* 292 25 74 49 Rectum 78 A 542 25 4 6 Rectum 5 D 56 28 5 25 Rectosgmod 5 D 559 7 52 5 Rectum 52 D 622 44 2 Rectum Survval 2 yr t 624 55 24 Rectum 56 C* 46 27 57 Rectosgmod Survval 2 yr C 684 29 4 4 Rectosgmod 4 D 52 7 4 27 Sgmod 5 D 44 4 49 5 Caecum 52 C 569 24 55 Caecum Survval yr t 575 24 9 4 Sgmod 4 B 8 4 4 6 Hepatc flexure Survval 9 yr B 26 24 4 6 Hepatc flexure Survval 2 yr A* 222 27 4 4 ransverse colon Survval 2 yr A 66 2 47 24 Rectum Survval -5 yr 66 24 46 22 Rectum 46 -t 627 46 54 8 Rectum 54 C 259 24 4 7 Rectum Survval 2 yr -t 686 5 52 7 ultfocal Survval 7 yr A sgmod-rectum * Dukes's classfcaton: assessment made from orgnal hstology report - sldes of orgnal hstology not avalable for revew. t Dukes's classfcaton could not be establshed from avalable hstology when revewed. t ultple colon cancers. All cases of colorectal carcnoma had extensve colts at dagnoss of carnoma. All deaths related to colorectal cancer except patent (postoperatve death) and patent 66 (cerebrovascular accdent). Survval to.2.76. g. 5). here was a two-fold rsk among patents dagnosed wth ulceratve colts after 45 years of age (g. 5) but ths dd not reach statstcal sgnfcance (able 6). he cancers n the seres were dagnosed at a relatvely early age when compared wth those expected n the general populaton (g. 6). he age-specfc relatve rsk decreased wth age even though a substantal number of person-years at rsk (4.7 at 6+ years) was avalable for analyss. he rsk of cancer n the colon and rectum when analysed separately was smlar wth an -fold rsk (able 7). Dscusson 495 A hghly sgnfcant excess of cancers of the dgestve system was found n the seres, the ncreased rsk beng confned to cancers of the colon, rectum, and gall bladder. he fact that the numbers of cancers observed at other stes were close to the numbers expected suggests that there was no undue selecton for cancers of the dgestve tract n preference to other stes. he ncreased colorectal cancer rsk n the seres when compared wth that n the general populaton s lkely therefore to be real rather than due to selecton factors operatng n the seres. Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
496 -c L- W - (n U U) - L. u_. *.I.,. 5 5 2 25 Years from dagnoss Pror, Gyde, acartney, hompson, Waterhouse, and Allan -2-8 5 4-2 a c- a (D. - g. 4 Rsk ofcolorectal cancers durng revew perod. Observed cumulatve rsk (%) ofcolorectal cancers adjusted for operaton and based on 'non-symptomatc' and 'nterval' cancers (- ) wth 95% confdence lmted (... ). Expected cumulatve rsk (%) (. -. - ). Cumulatve relatve rsk *-*-*). Observed number = 'nonsymptomatc' and 'nterval' cancers, expected number adjusted for operaton. x.x In.ch al a, E C-) l........................... m able 6 Colorectal cancers n relaton to age at dagnoss ofulceratve coltst Non-symptomatc + nterval Age at cancers Interval cancers dagnoss (yr) E OIE p OIE p -29-29 7-9 8 27-6 -44.8 2 *.7 * 45+.98 2 2- - 2 2 - OAL 2.8 2 * 6. * t Expected numbers adjusted for operaton; rsk = patent-years per cancer * p<.5. *** p<o.ool. When patents referred wth symptomatc cancer on ther frst vst ('referred' cancers) were excluded from the analyss (removng ths obvous bas towards selecton for cancer) and correctons made to the patent-years at rsk for surgcal resecton an -fold rsk was found n the seres (the 95% confdence lmts for ths rato are 7.-6.7). A comparson of our results wth other seres analysed by comparable methods shows that the relatve rsk n our seres s smlar to that reported by Edwards and rueloves and also by acdougall6 (able 8). Glat et a7 report a much lower cancer rsk from el n 6. - 4. - 2. - Q ua- 6- c 4 - a - 2- - 29 years U-.2-44 years 4... 45 + years 5 5 2 25 5 Years from dognoss of ukeratve colts g. 5 Cumulatve relatve rsk ofcolorectal cancers by age at dagnoss of ulceratve colts;.29 years ( ); -44 years (. ); 45+ years (...). Observed numbers = 'non-symptomatc' and 'nterval' cancers, expected number adjusted for operaton. z l....i. % 4.%.\........ X. %. 'X O 2 4o 5 6 7 8 9 Age at dagnoss of colorecta! cancer -55-4 - -2 ~- - o. _rn m g. 6 Observed ( ) and expected numbers ( -*-) ofcolorectal cancers by age at dagnoss of cancer wth correspondng relatve rsk: observed/expected number (... ). Observed number = 'non-symptomatc' and 'nterval' cancers, expected number adjusted for operaton. % -. Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.
Cancer morbdty n ulceratve colts 497 able 7 Colorectal and hepatoblary cancers n patents wth ulceratve colts Ste E OIE p All stes -76 57-8 Colon -8 - Rectum 9. Lver, gall bladder -2 7 2.9 Remander 29-6 27-9 E = expected number. = observed number. *** p<o-. Avv (able 8). A more conservatve estmate of the rsk n ths seres can be calculated by consderng only those cancers dagnosed after one or more years of follow-up ('nterval' cancers). hs elmnates any possble bas n the seres arsng from early cancers dentfed unexpectedly ('nonsymptomatc' cancers) n patents referred to the unt specfcally for early surgcal treatment. When 'nterval' cancers alone are consdered there s a sx-fold overall rsk n the seres. he pattern of ncreased colorectal cancer rsk over tme n ths seres s of nterest, as, whchever model s used, there s a constant ncreased rsk wth tme 5 years after dagnoss rather than a rsng rsk wth tme as mght have been expected (g. ). hs apples to all ages of onset of dsease (g. 5). hs constant rsk s surprsng, as a rsng relatve rsk wth tme mght be expected f nflammaton were actng as an addtonal promotonal factor for the development of cancer n the ulceratve colts group when compared wth the general populaton. he cancer rsk n the colon and rectum s smlar f the two stes are analysed separately (able 7). As the rectum s almost nvarably nvolved from onset of dsease a hgher relatve rsk of rectal cancers mght be expected f nflammaton were mplcated as a promotonal factor. able 8 Relatve rsk ofcolorectal cancer n ulceratve colts n prevously reported unselected seres Cancers Study No. of Ste of patents cancer Expected Observed OIE Edwards and ruelove Colo- 9645Oxford 624 rectal - 22 7. acdougall Colo- 9646 London 724 rectal -2 5 2-4 Glat et al 9747 el Avv 54 Colon -9 - Present seres 676 JColon 8. Rectum.9. In ths seres the age at onset of ulceratve colts affects the colorectal cancer rsk. he relatve rsk of developng colorectal cancer s sgnfcantly ncreased n patents whose ulceratve colts s dagnosed before years of age (able 6), but not n those dagnosed over 45 years. Patents dagnosed between ages and 44 years show an ntermedate level of rsk. he pattern of rsk over tme s smlar n each age group. he constant relatve rsk over tme and the marked effect of age at onset on the level of the colorectal cancer rsk may be clues to the aetology of colorectal cancer n ulceratve colts but the relatonshp s as yet unexplaned. References Slaney G, Brooke BN. Cancer n ulceratve colts. Lancet 959; 2: 694-8. 2 Lennard-Jones JE, orson BC, Rtche JK, et al. Cancer n colts: assessment of the ndvdual rsk by clncal and hstologcal crtera. Gastroenterology 977; 7: 28-9. Whelan G. Cancer n ulceratve colts: why are results n the lterature so vared? Cln Gastroenterol 98; 9: 469-76. 4 Goldgraber B, Humphreys E, Krsner JB, et al. Carcnoma and ulceratve colts, a clncal-pathologcal study. Gastroenterology 958; 4: 84-6. 5 Edwards C, ruelove SC. Course and prognoss of ulceratve colts. Part IV. Cancer of the colon. Gut 964; 5: 5-22. 6 acdougall IP. he cancer rsk n ulceratve colts. Lancet 964; 2: 655-8. 7 Glat, Zemshlany Z, Rbak J, et al. Ulceratve colts n a Jewsh populaton of el-avv Yafo. Gastroenterology 974; 67: 9-8. 8 Kewenter J, Ahlam H, Hulton L. Cancer rsk n extensve colts. Ann Surg 978; 88: 824-8. 9 Rtche JK, Allan RN, acartney J, et al. Blary tract carcnoma assocated wth ulceratve colts. Q J ed 974; 7: 26-79. Akwar OE, Heerden JA, oulk W et al. Cancer of the ble ducts assocated wth ulceratve colts. Ann Surg 975: 8: -9. Sackett DL, Whelan G. Cancer rsk n ulceratve colts. Scentfc requrements for the study of prognoss. Gastroenterology 98; 78: 62-5. 2 Daly DW. he outcome of surgery for ulceratve colts. Ann R Coll Surg Engl 968; 42: 8-57. Nefzger D, Acheson ED. Ulceratve colts n the Unted States army. Gut 96; 4: 8-92. 4 Watts JcK, DeDombal, Watknson G et al. Early course of ulteratve colts. Gut 966; 7: 6-. 5 Bonneve, Bnder V, Anthonsen P et al. he prognoss of ulceratve colts. Scand J Gastroenterol 974; 9: 8-9. 6 Greensten AJ, Sachar DB, Smth H et al. Cancer n Unversal and left-sded ulceratve colts: factors determnng rsk. Gastroenterology 979; 77: 29-4. Gut: frst publshed as.6/gut.2.6.49 on June 982. Downloaded from http://gut.bmj.com/ on 7 September 28 by guest. Protected by copyrght.