Colon Cancer
Predisposition to colon cancer. Certain conditions either predispose to colon cancer or are frequently associated with it. These include age over 40 years; family history of cancer; the syndromes of multiple polyposis, Gardner’s syndrome, and Peutz-Jegher’s syndrome; and ulcerative colitis present for more than 8 years. The chance of having a second colon cancer simultaneously (synchronous) with one newly discovered appears to be about 4% (range, 1.5%–18%), and the chance of developing a second carcinoma at some time after resection of the first is about 5%–10%.
As a general rule, GI carcinoma is not common in persons under age 40 years (although it can occur) and increases steadily in probability after that age, as do many other types of cancer. The major symptom of colon carcinoma is change in bowel habits, either toward chronic diarrhea or constipation. However, either upper or lower GI tract carcinoma may be relatively asymptomatic until very late.
Detection methods for these GI lesions are of three kinds: tests for occult blood in the stool, x-ray examination, and direct visualization.
X-ray. Barium enema is the standard x-ray procedure for the colon. In a barium enema study the barium is washed into the colon through a tube after all feces have been eliminated by laxatives and regular enemas. The major cause of poor barium enema studies is inadequate colon preparation. If feces remain in the colon after cleansing, obviously the barium cannot fill these areas, and small lesions may be missed. Barium enema may be performed by two techniques: regular and air contrast. Air contrast is more sensitive for colon lesions than standard barium enema, especially for polyps, but is somewhat more difficult to perform well and is even more dependent on good preparatory cleansing of the colon.
Direct visualization techniques. Direct visualization techniques include gastroscopy for stomach lesions and proctoscopy, sigmoidoscopy, and fiberoptic colonoscopy for rectal and colon lesions. Proctoscopy examines the anus and rectum. Sigmoidoscopes have two variants: the rigid type permits vision only to 25 cm, whereas the flexible type can reach as high as 40-50 cm. The fiberoptic colonoscope in experienced hands can be used to view almost the entire colon. Biopsy specimens can be obtained at the same time. Simple digital rectal examination detects many rectal and prostate cancers. For this reason, rectal examination is always included as a part of any good physical examination.
Stool occult blood. The most useful laboratory screening test at present is examination of the feces for blood. Usually this blood is occult (not grossly visible), although sometimes it is grossly visible. If it is from the upper GI tract the stool is often black (“tarry”), whereas lower GI bleeding may still show unchanged blood and color the stool red. Anemia of the chronic iron deficiency type is often present, although not always, and sometimes may be severe. Occult blood in the feces can be demonstrated by simple chemical tests for hemoglobin. These are based on peroxidase activity of hemoglobin, which is detected when it catalyzes the oxidation of a color reagent by a peroxide reagent. The most popular test agents are benzidine, guaiac (as powder, tablet, or impregnated filter paper), and orthotolidine. Many studies have been done evaluating one or more of these methods. Results have often been conflicting and, at times, completely contradictory. Nevertheless, some consensus emerges. Benzidine is the most sensitive of the reagents but yields a great number of false positive results. It is currently not available in the United States. Orthotolidine (most commonly in the form of a tablet called Hematest) has intermediate sensitivity, consistently detecting 10-15 ml of blood placed in the stomach experimentally. False positive results (in patients on an unrestricted diet) are most often reported as 20%–30%. Guaiac in powder form provides surprisingly divergent results for different investigators, but the majority report a lesser degree of sensitivity than with Hematest. Guaiac-impregnated filter paper slides have been available since 1966 under various trade names, of which the earliest and best known is Hemoccult. The guaiac-impregnated filter papers are said to be approximately 25% as sensitive as guaiac powder or orthotolidine. Limits of consistent detection (>90% sensitivity) are variously reported as 6-20 mg of hemoglobin/gm of stool. Some of the discrepancies in the literature may reflect an increase in sensitivity of the newer guaiac-impregnated filter paper tests compared to the older versions. In patients on an unrestricted diet, false positive results are reported in 1%–12% of cases, most being trace or weak reactions. In vitro tests on specimens with added blood in amounts considered to be normal (usually 2 mg of hemoglobin/gm of feces) similarly yields about 10% false positive results.
Interfering substances. False positive results may be caused by ingestion of meat that has not been cooked sufficiently to inactivate its peroxidase. Plant material also may contain peroxidase. False negative results may be caused by large doses of ascorbic acid (vitamin C). This is more likely to occur with oral doses more than 500 mg/day. False negative results are frequently caused by bleeding that is intermittent rather than continuous and by blood that is not uniformly distributed within a stool specimen.
Useful precautions. Certain precautions must be taken to minimize false positive or false negative reactions and increase detection of true lesions:
1. The patient should be on a meat-free high-residue diet beginning at least 24 hours before collections of the stool. Eliminating meat decreases weak false positive reactions. In addition, some investigators advocate preparation and boiling of a fecal suspension to destroy plant peroxidases, although few laboratories do this routinely. Some protocols eliminate any vitamin C intake for at least 48 hours before the test. The diet high-residue component increases detection of significant lesions. If patients must be screened on an unrestricted diet, someone who manifests a weakly positive result could be restudied on a restricted diet.
2. At least three stool specimens should be collected, each specimen obtained at least 1 day apart but with the collection days as close together as possible. Testing should be performed on two well-separated areas from each specimen, since the blood may not be evenly distributed. A single stool specimen yields about 40%–50% positive results in colon carcinoma, whereas increasing the number of stool specimens to three or more increases sensitivity to approximately 60%–70%.
3. Stool specimens should be tested within 48 hours after collection. Conversion of positive to negative results or vice versa has been reported after storage, although data are conflicting.
Other guaiac tests. Collecting or working with stool specimens has never appealed to most persons. Compliance rates have been low in various stool guaiac programs. It is not unusual for less than 50% of specimen kits distributed to be returned to the laboratory. Several variations of the guaiac method have been devised to improve rates of patient specimen collection. One variant is a guaiac-impregnated paper that is placed in the toilet bowl water with the stool specimen. Water leaching hemoglobin from the outside portions of the stool react with the guaiac reagent in the paper to produce a blue color. Another variant obtains the specimen by wiping the anal area with a special guaiac-impregnated paper system, from which the grossly contaminated portion is discarded and the remaining area tested. Whether these techniques will substantially improve cancer detection awaits adequately controlled clinical trials.
Nonguaiac stool tests. Immunologic tests specific for hemoglobin have been reported but to date have not been widely used due to their relatively high cost and relatively long time needed for assay. An assay method specific for heme called HemoQuant has been developed based on extraction and measurement of porphyrins from hemoglobin in the patient sample. This method also has not yet been widely used, since a measured amount of specimen must first be heated in an acid reagent to extract the porphyrins, purified by extraction with another acid reagent, and then quantitated with a fluorometer.
Other considerations. Besides ulcer, polyp, or malignancy, many other conditions affecting the nasopharynx (nasal bleeding) to the anus (hemorrhoids) may produce a guaiac-positive stool. Some of the more common serious conditions include ulcerative colitis, regional enteritis, and diverticulitis.
Carcinoembryonic antigen (CEA)
CEA is a glycoprotein antigen migrating in the beta-globulin area on electrophoresis, found in gastrointestinal tract epithelium in early fetal life but not detectable in most normal persons after birth. Immunologic serum tests have been based on antibodies against CEA. The procedure was originally thought to be specific for colon adenocarcinoma, but as more experience was obtained and modifications of the original technique were developed, it became evident that although abnormal results were most frequently found in colon carcinoma, elevated serum CEA levels could be obtained in persons with malignancies in various organs, with certain benign diseases (usually involving tissue inflammation or destruction), with occasional benign tumors, and in those persons who smoke cigarettes.
Several basic immunoassay techniques have been used. The most frequent RIA method for many years was the Hansen procedure, in which Z-gel is used as a radioactivity separation agent. More recently, many laboratories have changed to shorter and more simple nonradioactive immunoassay methods, frequently using a sandwich antigen-antibody technique. Reference values for both of these techniques are approximately 0-2.5 ng/ml using some manufacturer’s kits but not others. Even when values for two manufacturer’s kits give similar results when performed on the same specimens, there are usually a moderate number of discrepancies, sufficient that serial tests on the same patient should be performed with the same manufacturer’s kit.
Test results in colon cancer and other conditions
In colon carcinoma, different investigators have published widely divergent results; with about 75% average detection of carcinoma, but with reported detection rates ranging from 59%–97% (Table 33-11). The smaller and earlier-stage tumors are less likely to give positive results. Among noncolonic tumors, using the Hansen method, more than one investigator has reported that 70%–90% of lung, 85%–100% of pancreatic, and 45%–60% of breast and gastric carcinomas produce abnormal results. Normal persons who smoke were CEA reactive in nearly 20% of cases, and conditions associated with elevation in more than 10% of patients include pulmonary emphysema, benign rectal polyps, benign breast diseases, cholecystitis or extrahepatic biliary duct obstruction, alcoholic cirrhosis, and ulcerative colitis. At CEA levels more than 5 ng/ml (twice upper limits), abnormal results in colorectal, lung, and pancreatic carcinoma decreased to 50%–60% and decreased to 30% in breast and gastric carcinoma. Most other conditions were reduced to 5% abnormality or less, except for alcoholic cirrhosis (about 25%), common bile duct obstruction (17%), active ulcerative colitis (13%), and emphysema (20%). At CEA levels greater than 10 ng/ml 4 times the upper normal limit), abnormal results were found in 35% of colorectal, 25% of lung, 35% of pancreatic, and 15%–30% of gastric and breast carcinoma. All the benign conditions were less than 1% reactive except for emphysema (4%), active ulcerative colitis (5%), and alcoholic cirrhosis (2%).
Table 33-11 Sensitivity of tests for colon carcinoma*
These data indicate that CEA results greater than 10 ng/ml are strongly suggestive of tumor. Results in the 5-10 ng/ml range are suggestive of tumor, and results less than 5 ng/ml are either equivocal or not helpful. However, since a substantial minority of colon cancers are not detected, since a varying number of patients with other tumors are detected, and since certain benign conditions may produce elevated values, most investigators do not recommend CEA as a screening test either for colon cancer or in most circumstances for cancer in general.
Prognostic value of carcinoembryonic antigen level in colon cancer. In general, CEA titers greater than 10 ng/ml (4 times the upper reference limit) in colorectal carcinoma most often occur in more advanced tumor stages and imply a worse prognosis. However, about 5%–10% (literature range, 0%–18%) of Dukes’ A lesions (tumor confined to the colon wall) and about 50% of Dukes’ B and C (local tumor extension with or without local node involvement) have CEA titers of 10 ng/ml or more, whereas about 10% (0%–16%) of Dukes’ D (distant metastases) have normal CEA levels. Therefore, a normal CEA level does not exclude far-advanced tumor, and high CEA titers do not mean that a colorectal carcinoma is unresectable, although high CEA titers are reasonable (but not conclusive) evidence against an early Dukes’ A) tumor stage (see also Table 33-11).
Use of carcinoembryonic antigen to detect recurrent tumor. The major currently recommended use for CEA is to follow the results of tumor therapy. For this, a pretherapy baseline assay is needed to determine if the CEA level is elevated. After surgical treatment, at least 4 weeks should elapse before follow-up CEA assay is performed (some cases have been reported in which CEA levels did not return to normal for 2-6 months). Thereafter, assay every 2 months for 2 years appears to be the most widely used protocol. Another problem associated with interpretation of CEA response to therapy is a transient elevation above baseline after the start of chemotherapy or radiation therapy in some patients, probably related to the destructive effect of therapy on the tumor. Return of CEA titer to the reference range or nondetectable level is fairly reliable evidence that most of the tumor has been removed. This does not guarantee that all tumor has been eliminated. Further elevation of CEA levels suggests recurrence (either local disease or metastasis). This may develop as much as 6 months before clinical evidence of recurrence. Some investigators report that CEA is more sensitive in detecting metastatic colon carcinoma to the liver than is alkaline phosphatase. However, numerous reports indicate that sporadic nonsustained CEA elevations that are not due to neoplasia may occur; sometimes they are associated with nonmalignant illnesses, but often there is no apparent cause. These elevations are usually less than 5 ng/ml. The most reliable indicator of recurrence is a sustained (on at least two occasions) elevation of at least 5 ng/ml. Even this conservative criterion is associated with about a 10% false positive rate, based on “second-look” operations. However, in some cases where tumor was not found, recurrences developed later. Another published criterion is persistent elevation in three consecutive specimens obtained during 6 weeks.
There are several problems that may confuse CEA interpretation.
1. CEA in its current format uses two antibodies in a so-called sandwich technique, one of which is a monoclonal antibody derived from mouse (murine) spleen cells. Occasionally persons have antibodies that react (or cross-react) with mouse-derived antibody (HAMA) that can falsely elevate the CEA level and various other immunoassays using similar antibody methods. However, the elevated antibody level tends to remain the same over time.
2. Twenty percent or more patients with colon carcinoma will not have a rising CEA level.
3. Colon carcinoma located in the pelvis often fails to cause CEA level elevation.
4. Smoking can elevate the CEA level (generally only mildly).
5. Small but significant fluctuations may occur (mentioned previously) because of technical reasons, acute infection or inflammation, or unknown cause.
One report indicates that the CEA level is more likely to be elevated in recurrent colorectal carcinoma with liver or retroperitoneal metastasis (about 75%) and substantially less often elevated with lung, peritoneal, or nonhepatic local or single metastases (about 45%). These investigators had very few cures on reoperation prompted by reelevation of CEA levels, therefore raising the question of the cost-effectiveness of posttherapy monitoring.
In summary, a positive CEA result does not differentiate colon tumors from those of other primary sites. There is considerable overlap between malignancy and various benign conditions in the area up to 4 times the upper reference limit (2.5-10 mg/ml, normal being 0-2.5 ng/ml), especially in the area up to twice the upper reference limit (2.5-5 ng/ml). At present, CEA use in colon cancer is limited mostly to follow-up of patients after therapy. A significant and sustained increase in titer suggests recurrence or metastasis. Since the test requires technical expertise and considerable attention to technical details, repeat assay is suggested if one individual value deviates significantly from previous values and major decisions would be based on that value.
Sensitivity of tests for colon cancer
It is important to have some idea of the sensitivity of the various tests available to detect colon cancer, since one must make decisions on the basis of these test results. This information was presented in Table 33-11. In some cases it is difficult to compile accurate data, since a technique (e.g., barium enema) may be carried out by different methods that are not specified in the report. In the case of stool guaiac (as noted previously) there is significant improvement in detection rate when more than one specimen is obtained; this is not adequately reflected in the overall statistics provided by the studies summarized in Table 33-11. Nevertheless, most reports indicate that 20%–30% of colon cancers will be missed with Hemoccult using multiple stool specimens. In fact, at least one investigator concluded that a careful history, emphasizing symptoms commonly associated with colon carcinoma (change in bowel habits toward diarrhea, constipation, or narrowing of the stool; vague abdominal pain; increased flatus or mucous discharge) was at least as sensitive as, if not more than, stool testing for occult blood in raising suspicion of colon cancer.
In summary, rectal examination and stool tests for occult blood are the best simple screening procedures for GI tumor. If these are positive or arouse strong clinical suspicion, one can proceed to x-ray studies of the area indicated. When possible, direct visualization techniques are extremely helpful. Fiberoptic colonoscopy can detect lesions throughout the colon.