Screening and diagnosis o f colorectal cancer

: Colorectal cancer is the second most common tumour in North Amer ican men and women. From present understanding of the pathogenesis and natural history ofl?.rge bowel cancer, theoretically at least, the prevalence rate could be sig nificantly decreased with careful application of simple screening measures and appro priately directed diagnostic tests. Until results of randomized controlled trials are al'ailable, it is important to recognize the pitfalls of mass screening or of substituting screening for proper investigative procedures. One possible approach co the diagno sis of coloreccal cancer is outlined. Can J Gastroenterol 1988;2(3 ):99-106.

C OLORECTAL CANCER IS REACHING epidemic proportions in the Western world as the second most common malignancy in both men a nd women, yet despite chis rapid increase in incidence. the five year survival race has remained at abou t 40% for the past two decades ( I).
Most colorectal cancers occur as th e reiult of malignant transformation of a benign colonic adenoma (2). T h is malignant potential is determined by a number of factors which include: size of the adenomatous polyp; histological rype of the polyp; and degree of cell ular atypia (3). Unremoved, adenomas will grow; a 5 mm tubular adenoma wi ll grow to 2.0 cm within three co five years, with a risk of malignancy of approximately 5 to 25% (2). Adenocarcinoma of the colon or rectum is staged pathologically according to Dukes' classification ( 4). which correlates closely with disease survival. Dukes' A patients have greater than 90°{, five year survival while Dukes' C have a 26% five year survival (5). Three potential points of impact on the natural h istory are recognized: improved treatmen t fo llowing diagnosis (tertiary preven tion); early detection (secondary prevention); and alteration of biological risk factors (primary prevention ) (6).
Screening, a secondary preventio n measure. seeks co detect premalign a n t lesions or d isease in asymptomatic individuals at a more favourable Dukes' stage in order to implement effective rreacment and so u ltimately decrease mortal ity. Howeve r, four important sys tematic errors or biases muse be considered when evaluating the potential effectiveness of a screening program. When applying a screening test co asymptomatic individuals (B' in Figure I   Characte ristics of the group targeted for screening scnucgie5 may in flue nce the applicabi lit y o f suc h a screening man e uvre to different populations. For example. volun teers, su ch as subjects w h o undergo mulciphasi c check-ups, may be more or less h ealthy th an chose who do not volunteer Moreover, specialized cli nics or renowned con sultants m ay attract a higher p roportion of patients with a particular disease nr characteristic; this is ca lled selection bias. The fourt h important bias of screening is the diagnosn c suspicio n bias. which permits cl inicia ns to ovcrdiagnose conditions, eith er b ecause of p rior expectation or in order to avoid missing a potentially fatal di sease Whik th e interpre tation of resu lts of a screen ing program must he con s idered in the ligh t of these fo ur major p1tfalls, the first two (th e lead tim e a n d length b iases) can be eliminated by con du cting a randomi:ed con trolled ma! which con siders mortality rate, rather than duration of surv ival. as its major o utcome Chong (7) has 1denuf1ed four funda- cers and 24"., of those 111 the rt•ctm1g mrnd . This hm ,L'ns1tivity rl'llccts till' recent tl'ntkncy tor bowel LalKL'r, H• occur mnrc p roximall y (I) and thus iden ti fies a sl'rious lim1tat1on nf digital rl'ctal l'Xill11111;\tlOl1 Dou hie-contrast b:m u 111 l'n l'ma ,ind colonoscopy a rc rt•garded ,ts 'diagnosu, tests' fo r la rge ho\\'cl lc-,1ons rather than ,ls screening test~. hl'C.l ll!->l' of their respcr tivc cost~ (9), the limited resources ,wail able to perform thL•se tests ( tra ined ph, s1c1an~. equipmen t and faciliues) and tht poor :icccprnnc(' b y panents, which 1, often annupated by the primary earl p hysician While cnlonoscop y a nd bnr ium e n ema may be employed in surveil lance o t high risk individuals w11h mfl ammatory bowel disease. prior can ccr o r adcnc>matous polyp, it is not yet appropriate to .ipply lhcm for screening l'Xccpt perh,1ps 111 clinical tr1als Twt> rl'cem st ud 1es h,1vc hee n conducted to assess subjec t acccptabil1ty and  ,howcd an increased five year survival r,ue in subjects who had colorccrnl canlt'r from 64",, in the first period of the ,mdv to 85'\, in rhe la~t five years. The ,;crond srudy denwnsrrated decreased monality from rcctosigmoiJ cancer~ in the gmup screened by annunl rigid sigmmdnscopy when compared with unscreened con trols. R1g1d sigmoidoscopy ,tch ieves a mean mscrnon of20cm and detectsnhout 12"o ,,tcancers, while newer Oexiblc fibrcopric ,1gmo1do copies (60 cm) permit more ,'xtcnsivc examination -up to 50 cm in orer 70"~ of subjects ( 14). The flexible ,,nJoscope appears to be less uncomforcable and is able to detect more lesions. However, one study has shown that up to-l0~o of polyps occur proximal to the •each of the flexible sigmnidoscopc ( I 5 ).
The tendency of large bowel carcinomas to bleed has been used to some aJrantagc. Testing the stool fo r occult blood 111 excess of th e normal physio- Subjects ~ample their stool, completing ~ix slid es ( two on each of three consecutive daysl. while taking a high fibre Jict avoiding red meat and vegetables contaming wbstantial amounts of pcroxi-Jas~· A fal:;e-positive test may occur because of ora l iron treatment, animal hemoglobin from dietary red meat and from vegetables such as broccoli, cauliflower or turnips. Taking aspirin wil l inc rease normnl physio log1cal blood loss nnd may produce a positive reaction. A rwgative test may occur if the subject is wking vi tamin C, if the tumour is bleeding intermittently, which may occur in up to 2 5'';, of lcft-sidcd colonic cancers, or if thL· bleeding is at a rate less than 20 ml per day. This 'false-negative' rate may range from )4 to 50''., in known malignancies. Moreover, in the presence of slow intestinal transit, b lood undergoes digestion and hemoglobin peroxidase activity w ill nm be detected. While strongly positive slides remain unaffected by storage for up to 30 days, weakly positive slides may appear negative if stored for longer than four or five days. An exce llent review hy Simon ( 17) has high lighted the major Oaws in evaluation of fecal occult blood tests. Over 10 uncontrolled studies attempt to convince Cotorectot cancer t h e clinician to use this screen ing appro<1ch, but all a rc inconclusive Most studies reflect a suhstantial popul.rtion se lection bias and illustrate the variability in th e proportion of screened 111d1vidua ls who yield a p<1s1uvc test ( l. to 10'\,). Com p liance with completion nf test slides is dependent on the cl inical setting in which the study has been conducted and ranges from as low as 15''., in rural unselected populations to 90"{, in highly selected well motivated volunteers attend ing cancer screening clinics. In general, rhc predictive value of a positive test in these studies has heen low, ie , the proportion of ind ividuals with positive tests who turn out to have colorcctal cance r is less than 5''.(,. Finall y, five yem su rvival rate, which has been used as a measure of outcome, is subject to some o f the biases already discussed. At present, three randomized conrrollcd trials are in p rogress, two in North America ( IH, 19) and one in the Un ited Kingdom (20) . They will examine the effect of screening with the Hcmoccult rest, sigmoidoscopy or both on colonic cancer mortality. Each of these trials has over 10,000 patients in each arm of the study anJ preliminary reports have indicated that a ~ubsrant ial number of scree ned individuals who turn out to have cancer arc Dukes' stage A and B at diagnosis. A recent sympm,ium showed char interim ana lysis of the mortality rates of control and srudy groups in o ne of the North American trirds were comparable ( 21 ). However, the final results of these three trinls will not be available until 1989 or the early 1990s. Consequently, the re are no rel iable data on which to base advice to primary cnre p hysicians and general practitione rs.
Two new tests fo r occult blood undergoing p re limin ary investigation at p re~ent arc Hemo-Qua n t and immunodctcction techniques. T he Hemo-Quant detects hemoglobin-derived porphyrin by fluorescent ch romatography and may be able co distinguish between b lood derived from the upper a.nd the lower intestinal tract ( 22). A higher sensitivity than H emoccult is c lnimed (Hemo-Quant 97% compared w ith Hcmoccult 60'\,) m the expense ofan increased number of false positives. Preliminary reports ofimmunoquantitation techniques using nntibody specific for human hemoglobin, do nor indicate any advantage over Hemoccultin tcstspecificicy (23.24). Both of these new methods need further field testing. Once results of a favourable effect of HcmocculL test mg on mortality are available. it will then be reasonable to test mea~ures designed to improve compliance, which has been identified as a ,enous prohlem by ~cvcral nuthor~ ( 7. 17.19.20). If the effectiveness of screening remains established, only then does rnst become an im portant is~ul' for as~e~s-mt·nt The measuremcntof concentratio n of carcinoembryon ic antigen (CEA) in scrum has such a low scnsi tiviry and spccifi ciry, thm it is not clinically useful ( 2 5). Flow cytometry, which detects the freq uc ncy of cellular abnormalities in colonic cell populations sampled by wash mg or biopsy, mny be a useful surveillance rest, hut is likely to have a low specificity for colorectal ca ncer. Radionuclide scanning with radiolabcllcd monoclonal antibody is a n interesting innovation hut it is presently a n experimental technique (26). Who should be screened? Many cpidcmiologirnl studies have identified distinct risk groups for colorectal cancer ( 27 ). The high risk group includes subjects with a ny of the following: a polyposis syndrome; total ulcerative colitis of longer rh,in seven years' duration; a 'cancer family syndrome'; a p rior colon ic adenoma or carcinoma; females with a prior history of breast or urogenital cancer; md ividua ls with a fam ily history of colon ic cancer; nnd individuals with a fami ly history of any malignancy. Any subjects who arc over age 40 arc considered at average risk, if they have no high risk facto rs, and the remainder of the population is considered at low risk.
No studies recommend screening low risk patients at present. However. several a uth oritative bodies such as the American Cancer Society advocate yearly screening of average risk individuals with Hcm occu lt II augmented by interval sigmoidoscopy as often as every three years. Co nsid e ring that a fami ly practitioner may sec 3000 patients per year of whom 40'}'o arc average risk, then he or she must consider instructing and screening 1000 patients per year with 102 Hemoccul t and performing 400 screening sigmo idoscopies per year to detect a single cancer and three or four polyps. This, of course, assumes lOOq;, co mpliance o n the pnrr of the doctor and the patien t. It seems sensible at present m await the results of the randomized controlled tric1ls in progress which will detcrmi ne the effectiveness of such screening in average risk individuals.
Screening high risk individuals: Several recent studies have shown that there is a threefold increase in the risk of developing colorectal cancer in individuals with a family history ofl:irge bowel ca ncer, or women with a prior history of breast or urogenital cancer. Pilot studies undertaken by Rozcn ct al ( I l) and by Adamsen (28) to combine Hemoccult with flexible sigmoidoscopy or colonoscopy in these two groups indicate that such screening is feasible. Howeve r, neither study reports the proportion of eligible individuals who participated. It would be difficult, therefore, to advocate screening such individuals, ifonly a small numberof those eligible complied. In deed. much further work ts essential before devising strategies for screening such high risk individuals.
Who should initiate sc ree ning tests? A plethora of uncontrolled stud ies ha~ been conducted by specialists, primary care physicians, occupational health nurse:-. an<l the media in collaboration with local pharmacies, television stations, etc. Clearly, if evidence is not yet ava ilable to support the value of screening, then the question of who undertakes the screening is not relevant. Individual primary care physicians or specialists may defend their compulsion to screen high risk or elderly patients. However, the present authors recommend that it is more appropriate to refer such individuals to specialist p hysicians particularly interested in high risk groups, or to those conducting clinical trials. This wi ll at least prevent some of the potentially harmfu l effects of screen ing such as the fa lse rcassurrancc of individuals who h ave negative screening tests, or doing screening tests in symptomatic patients w ho really need a full diagnostic work-up.

DIAGNOSIS
Colo recta l d isease is commonly encoun tered in fami ly and spcciali~t practice a nd symptoms may include a cha nge in bowel habit. with diarrhea or constipation. or an alternat ion of the two, abdominal pam or recta l b leC'ding (291. While no ns pecific symptoms such as abdominnl pain or change in bowel habit should prompt in vestiga ti on. the d il emma fo r the clinician is that rectal b leeding may be due to common bcrngn local anorectal conditions such as hem· orrhoids. anal fissure or fistu la, but can· not be ignored as an important symr· tom of colorect;i l disensc. Also, the frequency and character of the bleed· ing do not necessari ly predict the sou m ( 30). Approximate ly one-quarter ot patien ts wi th b leeding will have cli n ical!, important disease wch as carcinoma. adenomarous polyps, inflammatory bowel d isease or diverticular disease. A fu r ther 2 5'X, wi ll have anorectal disea,e with additional colonic pathology, stre,,· ing the importan ce. particu larly in patients aged over 40, of not accepting a diagnosis of perianal disease withou t a complete exam in ation of th e colon (27.28,30-3 3). Finally. symptoms of iron ddiciency anemia or the mcidcnrnl fin d· ing of anemia at:, routine health check may be associated with an occu lt neoplastic lesion 111 the cecum or right colon It is dangerous to assume that the anc· mia is necessarily due to known pre· existing conditions. such as menorrha· gia or hiatu:, hernia.
The traditional approach to patien ts with colorcctal symptom~ has been a combination of sigmoidoscopy and bar· ium enema. Since the introduction of the air-contrast barium enema in 1921 there has been considerable improve· ment in the qua lity of radiographs Improved bowel preparation a nd high q uality imaging equipment permits an excellent diagnostic proced ure in most instances. Although a few rad iologists still favour the single contrast tech n ique ( 34 ), more recently the weight of radio· logical opi n ion has favoured the use of the air-contrast barium enema (35, 36).
The introduction o f fibreoptic endoscor1 in the early 1970s has led to increasi ng use of fibrcoptic colonoscopy and sub· sequently of the flexible fibreoptic s1g· moidoscope al though the conventional rigid proctosigmoidoscope is still ex tensively used. Ini tia ll y, endoscopic a nd radiological imaging of the large bowel appeared to be complimentary procedures (37). and colonoscopy wa:, widely used in chose patien ts in whom nn inndcquate or technically poor harium enema had been obtained or in patien ts whose 1ymptoms persisted in the presence of a normal rigid proctosigmo1doscopy and banum enema examination:..
In rhe past decade there have been numerous studies claimi ng to compare the diagnostic accuracy of bariu m enema and colonoscopy. Many of these were undertaken du ring th e earl y years of colonoscopy when co lonoscopes were less versatile, and the referring clinicia n:, considerably mo re reluctant to proceed tocolonoscopic investigation. O nly seven srudies were prospective ( lSJUS-42), and while most suggest that colonoscopy is superior, there are serious limitations m their study design wh ich prevent firm conclusions being reached Many physicians, apart from gastroenterologisrs and gastroi n testi nal surgeons, still consider barium enema and sigmoidoscopy ro be less invasive, more easily tolerated by patie nts and technically easier than colonoscopy, yet able to provide equa ll y good visualization of the large bowel. A high quality a1r·contrasc barium enema, however. is not so well suited tO the elderly or debi luared patient who muse be sufficie ntl y mobile to move rapid ly on a hard x-ray table. to provide these high quality fil ms. Th1ssame population has a higher prevalence of colonic pathology and d iagno-11s may be confou nded by redu ndant bowel loops, the presence of divenicu lar disease or an inadequate preparation of rhccolon (43.44).
Patients undergoing colonoscopy are usually sedated with diazepa m a nd mepcridine and a well trained colono-11:opist can perform the diagnostic procedure in 10 co IS mins ( 45 ). Under these circumstances, colonoscopy appears to be well tolerated and is more sensiti ve rhan the bariu m enema for detection of small adenomacous polyps, early in fla mmatory bowel d isease or vascular abnormalities (46-48).
The clinician may have several reasons for favou ring colonoscopy over bariu m enema as the best investiga tion ( 49). By doing t he procedure personall y, th e doctor/patient re lationshi p may improve. The specialise may be more confident of personal d iagnostic capabili ty than chat of others, particularly with a detailed knowledge of the patient's h istory. There may be a need to decrease the de lay before a diagnosis is reached and co mi n imi:e the exposure to ionizing radiation , especia ll y in young people of reproductive age. Lasrly, the ahility to remove an ad cnomatou~ polyp when found at colono~cnpy provides the cndo~copist wi th a positive therapeutic mancuvre.
In consideri ng whi ch procedure is mnst approprime for any particular clinical situation. the decision must be made agai nst a background of other important factors: the reliability of the investigation; how com piece an examination can be obtained: the prevalence of the disease ~uspectcd, and the safety. cost and patient'$ prefere n ce for a given proced u re.

COLORECTAL NEOPLASIA
Coloreccal rn rcinoma and adenomatous polyps are among the mos t common malignancies in men and women ( 50). lt is generally acccpred chat adenomacous polyps have the pocennal to progress to carcinoma and that removal of such polyps will prevent this process ( 51 ).
Adenomamus polyps commonly recur. In one study recurren t polyps were documented in 37":, of patients over a 3. 5 yea r period ( 15 ). Moreover, the risk chat patients with a previous colon cancer may develop a subsequen t cancer rn nges from l.3°i, to 7.6°(, ( 52 ).
The patient ar hi gh risk fo r the recurrence of polyps or development of cancer has a positive fami ly history, previ-ou~ mu ltiple adenomacous polyps, one or more large index polyps or a polypcontaini ng focal carcinoma ( 3 ). In these patients, a nn ual colonoscopy is ad vised un til no fu rther lesion is present because of the risk of missing lesions at the original colonoscopy and the development of metachronous polyps. Subseque ntly. colonoscopy may be u nder taken annually or alternated with a barium enema on a yearly basis. and the frequency of Colorecta l cancer examination decreased to two-yearly after five years of negative exami nations.
A sim ila r fo ll ow-up may be undertaken in patien ts wi th previous carcinoma. It is, however. importa nt that the colon be exa min ed carefully in t he perisu rgical period to exclude the p rese nce of sy nchronous lesio ns. Colonoscopy may he u ndertaken at the time of diagnosis co exclu de synchrono us a<lcnomacous polyps or cancer which may occu r in 20 to 25'\, of patients (53, 54). Colonoscopy may not always be possible bccau:.e of the presence of a secnosing lesion. In this instance colonscopy should be done approximately three months after surgery when the suture line may be inspected, and the remainder of the colon delcarcd free from all polyps. The detectio n of metachronous lc~ions may be undertaken on a schedule simil ar to th.:t t for the high risk polyp patient.

INFLAMMATORY BOWEL
DISEASE T h ere are seve ral indicatio n s for colonoscopy in ulcerative colitis. In th e context of colorectal ncop lasia, it is imperative co examine and obtain b iopsies and cymlogy from a colonic stricture or to evaluate a polypoid lesion or mucosa! excrescence, especially in chose with a h istory of ulcerative colitis or Croh n's d isease longer than seven yea rs. Biopsies may be obtained for histological evaluation co exclude malignancy in patien ts with long stand ing total ulcerative colitis after seven years of d isease, and in lcfbidcd colitb. after 12 co IS yea rs of d isease (55 ).
The recognition of colonic epithelial dysplasia as a pred ictor of colonic carcinoma has focused atten tion on the long term follow-up of patients with colitis. The cell ular natu re of th is ind icato r precludes the use of the bari um enema for cancer surveillance, as biopsy is always necessary. At a nnual colonoscopy, careful exam ination is made to detect any macroscopic abnormality whi ch may appear as a sma ll area of velvery-looking mucosa represen ting vi llo us change. In addition , the dysplasia associated lesion or mass h as a high risk poten tia l ( 56). lf no macroscopic abnormality is observed, mul tiple biopsies a re taken from each segment of the colon Once a surveillance program has been started, the frequency of inspection and biopsy is in pan determined by the presence and the degree of dysplasia. If high grade dysplasia 1s detected. then colectomy is recom m ended. If low grade dysp las ia is found. repeat colo noscopy and biopsy should be done in six months. If dysplasta is intermediate, endoscopy and repeat biopsy shou ld be repeated in less than th ree month s. In the absence of dysplasia, annual colonoscopy has been recomme nded (57).
A critical review of major surveillance programs wh ich fo llow the above recommendatio ns has ill ustrated that none are controlled and all arc subject to lead time bias ( 58). Only 3 7'';. of patients with significant dysplasia have been fou nd to have cancer and 20qt, of ca ncers were detected outside of surveillance programs. Furthermore, the cost-benefit ratio appears to be high and patient comp lia nce with su rveillance itself and subsequent colectomy may be highly variable. A careful prospective random ized surveillance program is considered a high priority.
A lthough increased risk of colorectal c:rncer has been identified in subjects with colonic Crohn's disease, surveillance strategics and nsk profiles are not so well defined as those with ulcerative colitis. Further studies arc necessar y to determine the most appropriate strategy for follow-up.

STRATEGY OF INVESTIGATION
A plan of investigation for pmients with coloreccal symptoms can be defined based on analysts of the published data. W hen the principal concern is the diagnosis of neoplastic disease, patients may be stra tified according to age and whether they have rectal bleeding (Figure 3). In patien ts aged u n der 40 who have no history of rectal b leeding. the probability of adenomamu s polyps 0r carcinoma is low Diverticular disease, which may make high qual ity barium enemas more difficult and conceal other lesions, is uncommon. Young patients are also mobile and ba rium enema studies arc usually of high quality. On this basis, patients may undergo flexible sigmoidoscopy and double contrast barium enema.
In patients over 40 years of age. without a history of bleeding, between 10 and 20'';, will have one or more adenomatous polyps. Subjects considered to be at h igh risk for coloreccal neop lasia could u ndergo immediate colonoscopy while the remainder may be investigated by a combination of flexible sigmoidoscopy after full bowel preparation. combined with barium enema. lf an adenomatous polyp is detected, colonoscopy will have to be undertaken in order to determine the presence of any further polyps and (or polypectomy to be performed In pattencs in whom the flexible sigmoidoscopy is normal, harium enema should be done. Patients who arc bleeding and arc aged under 40, shou ld undergo a digital rec· tal exnmination. anoscopy and flcxihk s1gmoidoscopy which will determine the next stage of investigation. If a polyp is found, colonoscopy should be under· taken whi le the presence of inflammatory disease wt!! lead to a barium enema Local pcrianal disease.:: such as hemor· rhotds or fissures in this age group will usually not need any further invcstiga· tion unless the patient is in a high risk group for ncoplas1a, that is tn say hns a family history of bowel, breast or urogen ital cancer, or a family history of colonic adenoma or polyposis coli Flex· ible sigmoidoscopy 1s preferable to rigid sigmoidoscopy because of the higher sen· sitivity for ncoplasia and inflammatory bowel disease, although no reports hnvc examined the relative sensitivity of each procedure based upon age (Table I ) When perianal disease is present flexible sigmoidoscopy is still warranted as concomitant colonic disease is present in over 2 5"'o of subjects ( 59 ). lflocal exam ination and the flexible sigmoidoscop, do not provide the diagnosis, and the patient is in a high risk group, he or she should have a barium enema If bleeding recurs, colonoscopy will be needed. even if a prior barium enema was negative.
In patients wit h bleeding or iron deficiency anemia who arc aged over 40. the literature strongly suggests that colonoscopy may more frequently provide a diagnosis, and th at it will often sholl' neoplastic disease or vascular cctas1a (48.60.61 ).

CONCLUSION
Colonscopy. o r ba ri um enema and flexible sigmoidoscopy remain the appropriate tests for diagnosing colorectal cancer in subiects who have symptoms or signs of colonic disease These arc the tests best suned for the investigation of a positive fecal occult hlood test or iron deficiency anemia and for the surveillance of specific high risk groups such as patients with prior polyps The digiral rectal examination is a poor screening test for large bowel cancer Hemoccult or sigmoidoscopy (or in comhination), while having low sensitivirv. may be reasonable ~crcening test~ in some subgroups. Symptomatic individuals need full in vestigation and not screening tests. At prese nt there is insufficient evidence to recomme nd mass screening, o r screening of average risk mdividu::ils un til further data are avai lable from the randomized con trolled trials. Finally. high risk individuals could be enrolled in pilot studies or screened by specialists with a particular interest m these populations.