The gastrointestinal tract (GIT) is the site of more cancers than any other organ system in the body. In terms of morbidity and mortality, the main GIT cancers are colorectal, gastric, pancreatic, oesophageal and hepatocellular. A notable feature of most types of GIT cancers is the enormous variation in incidence in different regions of the world [see below]. These differences in incidence appear to be due primarily to local environmental factors and not to racial or genetic factors. Thus, when populations migrate from high prevalence regions to low prevalence locations, the organ specificity of at least some cancers changes to that of the new region.

Each type of GIT cancer has its own preferential marker or group of markers. The aim of this paper is to review critically the role of markers in the main GIT cancers, and to make recommendations for their clinical use. New markers still under investigation are only mentioned briefly.

COLORECTAL CANCER

Colorectal malignancy is one of the most common cancers in the Western world, affecting approximately 1 in 20 people. In terms of incidence, it ranks second to breast cancer in women and third to prostate and lung cancers in men. In contrast to most Western countries, the frequency of colorectal cancer is considerably lower in Africa and parts of Asia.

Diagnosis and screening

The most useful and widely investigated marker for colorectal cancer is CEA. In colorectal cancer, both the proportion of patients with elevated levels and the extent of elevation are primarily dependent on disease stage. Thus in one typical study (1), high levels of CEA (i.e., >2.5 ng/ml) were found in 28% of patients with Dukes’ A, 45% with Dukes’ B, 75% with Dukes’ C and 84% with Dukes’ D stage disease. Using a cut-off point of 5 ng/ml, the proportion of patients with elevated levels were 3%, 25%, 45% and 65% for patients with Dukes’ A, B, C and D disease, respectively (1). Because of this lack of sensitivity for early disease, CEA is of little value in the detection of Dukes’ A or B colorectal cancer. A further problem with CEA is lack of specificity, which is only approximately 90% using a cut-off point of 2.5 ng/ml (2). Many benign diseases (Table 1) as well as most advanced types of adenocarcinoma can give rise to elevated CEA concentrations (for review, see ref. 2).

This poor sensitivity and specificity when combined with the low prevalence of colorectal cancer in asymptomatic populations means that CEA cannot be recommended as a screening test for colorectal cancer in unselected individuals. For similar reasons, CEA cannot be used alone for diagnosing or ruling out colorectal cancer. However, in patients with appropriate symptoms, a grossly elevated value (e.g., more than five times the upper limit of normal) should be considered highly suggestive for cancer in that individual (3).

Prognosis

While pre-operative levels of CEA are of little value in detecting early colorectal cancer, a number of investigators have shown that patients with either high pre- or post-operative concentrations of the marker have a worse outcome than those with low values (for review, see refs. 2,4). In some but not all of these studies, the prognostic value of CEA was reported to be independent of Dukes’ stage (2,4,5), the most widely used parameter for assessing prognosis in colorectal cancer.

CEA may, however, give prognostic information within the Dukes’ subgroups. Additional prognostic markers are particularly required for the Dukes’ B category. Approximately 40-50% of patients with Dukes’ B disease have aggressive disease. Furthermore, recent preliminary data suggest that adjuvant chemotherapy has a modest but detectable beneficial effect on outcome of patients in this subgroup (6). Rather than administer adjuvant chemotherapy to all patients with Dukes’ B disease, it would be desirable to have a marker capable of discriminating between patients with aggressive and indolent disease. The availability of such a marker should aid the selection of aggressive tumours that could benefit from adjuvant therapy and at the same time avoid giving therapy to patients likely to have a good outcome.

In certain studies, preoperative levels of CEA have been shown to be prognostic in patients with Dukes’ B stage disease (for review, see ref. 7). It remains to be shown, however, whether CEA measurement can identify the patients within this subgroup who could benefit from adjuvant chemotherapy. In contrast to patients with Dukes’ B disease, it was recently recommended that all patients with Dukes’ C disease should receive adjuvant chemotherapy (8). Preliminary data suggest that both serum and tissue levels of CA 19-9 may also be prognostic in colorectal cancer (9). Further studies are, however, necessary to confirm these findings.

Monitoring

It is generally believed that the main application of CEA is in the monitoring of patients with diagnosed colorectal cancer. In 1980, an NIH Consensus Meeting concluded that CEA was the best available non-invasive test for the follow-up of patients with colorectal malignancy (10). In 1996, a statement from the American Society of Clinical Oncology (ASCO) concluded that CEA was the marker of choice for monitoring patients with colorectal cancer (9).

For identifying recurrences in patients with previously diagnosed colorectal cancer, CEA has a sensitivity of about 80% (range 17-89%) and a specificity of approximately 70% (range 34-91%) (for review, see ref. 2). Early studies showed that serial CEA levels could detect recurrent disease many months (usually 4-10 months) in advance of clinical evidence of disease (2). CEA testing was found to be most sensitive for diagnosing hepatic or retroperitoneal disease and relatively insensitive for either local, peritoneal or pulmonary involvement (11). Some investigators have reported that a slowly rising CEA usually indicates a locoregional recurrence while rapidly increasing levels usually suggest hepatic metastasis (3).

The value of CEA in detecting local recurrence of colorectal cancer was recently evaluated in a single institution prospective randomised trial (12). In this study using 207 patients, an elevated CEA value was the most frequent indication of local recurrence in patients without clinical symptoms. The authors concluded that CEA measurement was more cost-effective than other procedures including computerised tomography (CT) for the diagnosis of local recurrence. It was also suggested that intensive follow-up with CEA might be more beneficial for patients with rectal cancer than for those with colonic cancer, as local recurrence was more frequent after rectal cancer resection (12).

In the follow-up of patients with colorectal cancer, the optimum interval between CEA measurements has not been established. In practice, most clinicians use intervals of 3 months, at least for the first 2 years after the initial diagnosis. The ASCO Panel have stated that "any benefit from postoperative screening requires serum monitoring every 2-3 months"(10).

While no large-scale prospective randomised study has been carried out to assess the clinical value of monitoring colorectal cancer patients with CEA, the results to-date from individual trials containing small numbers of patients suggest that serial marker determinations to detect early recurrences do not enhance patient outcome (2). However, by pooling data from various small-scale randomised and comparative cohort studies, Rosen et al., (13) recently showed a statistically significant difference in cumulative five-year survival rates for patients undergoing intensive follow-up compared to a control group. In this meta-analysis, intensive follow-up was defined as history, physical examination and CEA assay at least three times per year for at least 2 years. The control group had no routine follow-up, with intervention only following the development of symptoms.

In the above study by Rosen et al (13), intensive follow-up with and without CEA was not compared. However, in a further meta-analysis, Bruinvels et al (14) showed that intensive monitoring only improved five-year survival rates if CEA assays were included.

It is worth noting that the 1996 ASCO guidelines recommended that CEA monitoring be carried out only for those patients who would be willing and able to undergo hepatic resection for recurrent disease (10).

Clearly, further work is necessary to address the impact of CEA monitoring on patient survival, quality of life and cost of care. Ideally, this study should be carried out as part of a prospective randomised trial.

Although surgery remains the most effective therapy for colorectal cancer, chemotherapy (e.g., 5-fluorouracil and levamisole) is finding increasing use especially in patients with advanced disease (for review, see refs. 15,16). Administration of this therapy may however, cause transient elevations in CEA levels (17).

The ASCO guidelines in 1996 stated that existing data was not sufficient to recommend routine use of CEA alone for monitoring response to therapy (10). If, however, no alternative simple test was available to detect a response, it was suggested that CEA should be assayed at the commencement of treatment for metastatic disease and every 2-3 months during active therapy. Two levels above baseline were considered sufficient to document progressive disease even in the absence of corroborating evidence (10).

While CEA is the preferential biochemical test for colorectal cancer, a number of other markers such as CA19-9, CA242 and cytokeratins (e.g., TPA and TPS) have also been evaluated for this malignancy (18,19,20,21,22). While some of these markers have been found to complement CEA, further work will be required to see which marker is most complementary to CEA.

PANCREATIC CANCER

In the Western world, pancreatic cancer has a estimated frequency of approximately 10 cases per 100,000 population and is the fifth most common cause of cancer-related deaths. About 90% of pancreatic malignancies are adenocarcinomas with a ductal phenotype. Neuroendocrine and acinar cell cancers represent less than 5% of all pancreatic tumours.

Diagnosis

The most widely used marker for pancreatic adenocarcinoma is CA19-9 (for review, see refs. 23,24). Summarising the data from 24 different reports, Steinberg (24) calculated that CA19-9 had a mean sensitivity of 81% and a mean specificity of 90% (using a cut-off point of 37 U/ml) for pancreatic cancer. In this meta-analysis, the specificity varied from 76% to 99% and sensitivity from 69 to 93%, depending on the patient population.

Increased concentrations of CA19-9 are not however, specific for adenocarcinoma of the pancreas. High levels can also be found in other GIT malignancies (especially with advanced disease) as well as in various benign disorders (e.g., both chronic and acute pancreatitis, cirrhosis, cholangitis and hepatocellular jaundice) (23). It should be pointed out that the specificity of CA 19-9 is particularly low in the presence of jaundice. For example in a recent prospective study using a cut-off value of 40 U/ml, the specificity was 92% in non-jaundiced patients but only 69% in jaundiced subjects (25). With a cut-off point of 100 U/ml, the respective specificities for non-jaundiced and jaundiced patients were 99% and 83%.

CA19-9 can complement radiological procedures in the diagnosis of pancreatic malignancy, especially in non-jaundiced patients (25). When used as a diagnostic aid, it is important to bear in mind that only about 55% of patients with pancreatic cancers less than 3 cm in size have an elevated level of CA19-9 (i.e., >37 U/ml) (24). Furthermore, as pointed out above, patients with certain benign diseases such as jaundice and pancreatitis may present with elevated levels of CA19-9. Consequently, CA19-9 has limited value in the diagnosis of pancreatic cancers, particularly early forms of the disease.

Prognosis and monitoring

CA19-9 has the potential both to assess prognosis and monitor patients with diagnosed pancreatic adenocarcinomas (23,24,26). However, routine use of CA 19-9 for these purposes is unproven at present.

Other mucin-type markers such as CA50, CA242, CA195, DU-PAN 2, and CAM 17.1/WGA have also been described for pancreatic cancer (26,27,28). These markers have been investigated less widely than CA19-9 but from the available evidence appear to provide similar data. CA19-9 therefore remains the "gold standard marker" against which future markers for pancreatic cancer will be evaluated.

GASTRIC CANCER

Gastric cancer is one of the most common cancer world-wide. In the West, there has been a decrease in the incidence of this disease in recent decades. High rates still persist in parts of Asia with almost 100 deaths per 100,000 population. In contrast, in the US, the death rate is only about 6 per 100,000 people.

The most widely described markers for gastric cancer are CEA, CA19-9 and CA72-4. Of these 3 markers, CA 72-4 appears to be the most sensitive and specific (29,30). CEA and CA19-9 appear to have similar specificity, although CA19-9 may be more sensitive than CEA (29). None of these markers is useful in either screening or diagnosing early gastric cancer. Furthermore, in the absence of useful therapy for this malignancy, their role in the follow-up of patients with diagnosed disease remains to be evaluated.

ESOPHAGEAL CANCER

The incidence of oesophageal cancer in different parts of the world is also extremely variable. Thus, in certain parts of Asia, the frequency is as high as 50-100 cancers per 100,000 inhabitants while in most of Europe and the US, the average incidence is 2-3 per 100,000. Histologically, over 90% of oesophageal cancers are of the squamous cell type with less than 10% adenocarcinomas.

Compared to most other types of GIT cancers, biochemical markers have been little investigated in oesophageal cancer.

Preliminary data suggests that the best available markers for the squamous cell type are SCC (31) and cytokeratins [e.g., CYFRA 21-1, TPA, TPS] (32), while CA19-9 appears to be the preferred marker for the adenocarcinoma type (30). Presently, these markers are of little value in either the diagnosis or management of oesophageal cancer.

HEPATOCELLULAR CARCINOMA

World wide, primary hepatocellular carcinoma or malignant hepatoma is one of the most common malignancies. While this malignancy is found relatively infrequently in the West (i.e. approximately 5-10 cases per 100,000 people in Southern Europe and less than 5 cases per 100,000 in Northern Europe), in parts of Asia and Africa, it is the most common cancer (33). The frequency is so high in parts of China that population screening is recommended.

Screening

The marker of choice for hepatocellular carcinoma is a -fetoprotein (AFP). In parts of China and sub-Saharan Africa where this malignancy is endemic, screening using AFP has been carried out. In these areas, screening can be justified because of the high disease prevalence and the recent finding that earlier diagnosis led to both decreased tumor size and increased patient survival (34). It should however, be pointed out that the value of AFP in screening for hepatocellular carcinoma has not been determined in prospective randomised trials.

Because of the relatively low incidence in most of the Western world, screening for hepatocellular carcinoma is not warranted. However, in 1986, an NIH Consensus Conference recommended that subjects who are positive for hepatitis B surface antigen and who have chronic active hepatitis or cirrhosis, should be screened every 3 months using AFP and every 4-6 months using ultrasound (35). Hepatitis B carriers who do not have liver disease can be screened less frequently.

Persistent infection with hepatitis C virus is also an important risk factor for hepatocellular carcinoma (36). The relative risk of hepatocellular carcinoma in subjects with hepatitis C virus infection and cirrhosis is approximately 100-fold the risk in uninfected persons (36). Patients with chronic hepatitis C infection should thus be also monitored with AFP and ultrasound (36).

Diagnosis

One of the problems in using AFP for differentiating between hepatocellular carcinoma and both hepatitis and cirrhosis is that elevated levels may also occur in these benign conditions (35). Thus, a number of approaches have been attempted to enhance the differential diagnostic capacity of AFP. One of these is based on the different extents of glycosylation of AFP produced by malignant and non-malignant liver cells, i.e., AFP synthesised by the former tends to be more fucosylated than that produced by the latter (for review, see ref.37). This fucosylated AFP can be separated from "normal" AFP by reactivity with lectins such as Lens culinaris agglutinin A (AFP L3) and erythroagglutinating phytohaemagglutinin (AFP P4 and P5) (37,38). Measurement of the L3, P4 and P5 fractions of AFP therefore helps differentiate hepatocellular carcinoma from cirrhosis (37,38).

A second approach to enhance the specificity of AFP is based on the finding that in some benign conditions, AFP elevations may be transient, whereas in malignancy concentrations remain high or even increase (35). Assay of AFP every two to three weeks may therefore eliminate false-raised values.

A relatively new marker undergoing evaluation for hepatocellular carcinoma is des-gamma-carboxy prothrombin (DCP), also known as PIVKA II (protein induced by vitamin K absence or antagonist II). In a recent report in which 60 patients with hepatocellular carcinoma, 60 patients with cirrhosis and 273 normal subjects were studied, the sensitivity, specificity and accuracy of a new assay for DCP were found to be 60%, 92.3% and 81.4%, respectively (39). When these results were combined with AFP measurement, 86% of the hepatocellular carcinoma patients and 78.3% of those with solitary malignancy were positive for at least one of the markers (39).

CONCLUSION

The general problems with GIT cancer markers are lack of sensitivity for early disease and lack of specificity for malignancy. These twin problems, especially when combined with the relatively low prevalence of GIT cancers in the general population, prevent the use of available markers for screening asymptomatic populations. A possible exception however, is the use of AFP in screening for hepatocellular cancer in certain high risk areas. Lack of sensitivity and specificity also limits the diagnostic ability of these markers although CA19-9 may be useful in differentiating between benign and malignant pancreatic disorders in non-jaundiced patients.

Many of the markers discussed above are presently finding application in the follow-up of patients with diagnosed cancers. CEA, in particular, is frequently assayed for monitoring patients with colorectal cancer and recent meta-analyses have concluded that intensive follow-up using this marker enhanced patient outcome. While a clear benefit for serial CEA assays has not yet been shown in prospective randomised trials, it is our view that this marker may be of use in the surveillance of certain patients with diagnosed colorectal cancer, i.e., to detect asymptomatic recurrences that can be operated on for cure (13). Furthermore, in the follow-up of patients, the use of markers is cheaper and more convenient for the patient than radiological procedures.

The following caveats should, however, be borne in mind when using CEA in the surveillance of patients with diagnosed colorectal cancer: a) elevations usually only occur in patients with advanced disease, b) not all patients with recurrent colorectal cancer will exhibit increased levels, c) high levels may occur in conditions unrelated to recurrent colorectal cancer and d) certain cytotoxic therapies may cause transient increased concentrations. For patients with either gastric or oesophageal cancer, markers currently contribute little to either diagnosis or follow-up. Clearly, better markers for GIT cancers are needed. Development of simple assays for mutated c-oncogenes and suppressor genes may be a worthwhile approach here.

REFERENCES

1. Wanebo HJ, Rao B, Pinsky CM et al: Preoperative carcinoembryonic antigen level as a prognostic indicator in colorectal cancer. N Eng J Med 229: 448-451, 1978.
2. Fletcher RH: Carcinoembryonic antigen. Ann Int Med 104: 66-73, 1986.
3. Begent RHJ: The value of carcinoembryonic antigen measurement in clinical practice. Ann Clin Biochem 21: 231-238, 1984.
4. Grem J: The prognostic importance of tumor markers in adenocarcinomas of the gastrointestinal tract. Current Opinion Oncol 9: 380-387, 1997.
5. Carpelan-Holstrom M, Haglund C, Lundin J, Jarvinen H, Roberts P: Pre-operative serum levels of CA242 and CEA predict outcome in colorectal cancer. Eur J Cancer 32A: 1156-1161, 1996.
6. Erlichmann C, Marsoni S, Seitz JF et al.: Event free and overall survival is increased by FUFA in resected B colon cancer: a pooled analysis of 5 randomised trials. Proc Am Soc Clin Oncol 16: Abstract 991, 1997.
7. Kronborg O: Staging and surgery for colorectal cancer. Eur J Cancer 29A: 575-583, 1993.
8. NIH Consensus Panel: Adjuvant therapy for patients with colon and rectal cancer. JAMA 264: 1444, 1990.
9. American Society of Clinical Oncology: Clinical practice guidelines for the use of tumor markers in breast and colorectal cancer. J Clin Oncol. 14: 2843-2877, 1996.
10. Goldenberg DM, Neville AM, Carter AC, Go VLW, Holyoke ED, Isselbacher KJ et al: CEA (carcinoembryonic antigen): its role as a marker in the management of cancer. J Cancer Res Clin Oncol 101: 239-242, 1981.
11. Moertel CG, Fleming TR, Macdonald J, Haller DG, Laurie JA, Tangen C: An evaluation of the carcinoembryonic antigen (CEA) test for monitoring patients with resected colon cancer. JAMA 270: 943-947, 1993.
12. Pietra N, Sarli L, Costi R, Ouchemi C, Grattarola M, Peracchia A: Role of follow-up in the management of local recurrence of colorectal cancer. Dis Colon Rectum 41: 1127-1133, 1998.
13. Rosen M, Chan L, Beart RW, Vukasin P, Anthone G: Follow-up of colorectal cancer, a meta-analysis. Disease Colon Rectum 41: 1116-1126, 1998.
14. Bruinvels DJ, Stiggelbout AM, Kievit J, van Houwelingen HC, Habbema DF, van de Velde CH: Follow-up of patients with colorectal cancer, a meta-analysis. Ann Surg 219: 174-182, 1994.
15. Machover D: A comprehensive review of 5-fluorouracil and leucovorin in patients with metastatic colorectal cancer. Cancer 80: 1179-1187, 1997.
16. Midgley R, Kerr D: Colorectal cancer. Lancet 353: 391-399, 1999.
17. Moertel CG, Fleming TR, Macdonald JS, Haller DG, Laurie JA: Hepatic toxicity associated with fluorouracil plus levamisole adjuvant therapy. J Clin Oncol 11: 2386-2390, 1993.
18. Quentmeier A, Schlag P, Giesen HP, Schmidt-Gayk H. Evaluation of CA 125 as a tumor marker for gastric and colorectal cancer in comparison with CEA and CA19-9. Eur J Surg Oncol 13: 197-201, 1987.
19. Hall NR, Finan PJ, Stephenson BM, Purves DA, Cooper EH: The role of CA242 and CEA in surveillance following curative resection for colorectal cancer. Br J Cancer 70: 549-553, 1994.
20. Carpelan-Holmstrom M, Haglund C, Kuusela P, Jarvinen H, Roberts PJ: Pre-operative serum levels of CEA and CA 242 in colorectal cancer. Br J Cancer 71: 868-872, 1995.
21. Von Kleist S, Hesse Y, Kananeeh H: Comparative evaluation of four tumor markers, CA242, CA19-9, TPA and CEA in carcinomas of the colon. Anticancer Res 16: 2325-2332, 1996.
22. Kornek G, Schenk T, Raderer M, Djavarnmad M, Scheithauer W: Tissue polypeptide-specific antigen (TPS) in monitoring palliative treatment response of patients with gastrointestinal tumours. Br J Cancer 71: 182-185, 1995.
23. Duffy MJ: CA19-9 as a marker for gastrointestinal cancers: a review. Ann Clin Biochem 35: 364-370, 1998.
24. Steinberg W: The clinical utility of the CA 19-9 tumor associated antigen. Am J Gastroenterol 85: 350-355, 1990.
25. Ritts RE, Nagorney DM, Jacobsen DJ, Talbot RW, Zurawski Jr VR: Comparison of preoperative serum CA 19-9 levels with results of diagnostic imaging modalities in patients undergoing laparotomy for suspected pancreatic or gallbladder disease. Pancreas 9: 707-716, 1994.
26. Lundin J, Roberts PJ, Kuusela P, Haglund C: The prognostic value of preoperative serum levels of CA 19-9 and CEA in patients with pancreatic cancer. Br J Cancer 69: 515-519, 1994.
27. Moossa AR, Gemagami RA: Diagnosis and staging of pancreatic neoplasms. Surg Clin N Am. 75: 871-890, 1995.
28. Yiannakou JY, Newland P, Calder F, Kingsnorth AN, Rhodes JM: Prospective study of CAM17.1/WGA mucin assay for serological diagnosis of pancreatic cancer. Lancet 349: 389-392, 1997.
29. Guadagni F, Roselli M, Amato T, Cosimelli M, Perri P, Casale V, et al: CA72-4 measurement of tumor-associated glycoprotein 72 (TAG-72) as a serum marker in the management of gastric carcinoma. Cancer Res 52: 1222-1227, 1992.
30. Byrne DJ, Browning MCK, Cuschieri A: CA72-4: A new tumour marker for gastric cancer. Br J Cancer 77: 1010-1013, 1990.
31. Mealy K, Feely J, Reid I, McSweeney J, Walsh T, Hennessy TPJ: Tumour marker detection in oesophageal carcinoma. Eur J Surg Oncol. 22: 505-507, 1996.
32. Yamamoto K, Oka M, Hayashi H, Tangoku A, Gondo T, Suzuki T: CYFRA 21-1 is a useful marker for esophageal squamous cell carcinoma. Cancer 79: 1647-1655, 1997.
33. Akriviadis EA, Llovet JM, Efremidis SC, Shouval D, Canelos R, Ringe B, et al: Hepatocellular carcinoma. Br J Surg 85: 1319-1331, 1998.
34. Tang ZY, Yu YQ, Zhou XD, Ma ZC, Yang R, Lu JZ, et al: Surgery of small hepatocellular carcinoma. Analysis of 144 cases. Cancer 64: 536-541, 1989.
35. Di Bisceglie AM, Rustgi VK, Hoffnagle JH, Dusheiko GM, Lotze MT: Hepatocellular carcinoma. Ann Int Med 108: 390-401, 1988.
36. Ince N, Wands JR. The increasing incidence of hepatocellular carcinoma. New Engl J Med 340: 798-799, 1999.
37. Lamerz R. AFP isoforms and their clinical significance (overview). Anticancer Res 17: 2927-2930, 1997.
38. Sato Y, Nakata K, Kato Y, Shima M, Ishii N, Koji T, et al: Early recognition of hepatocellular carcinoma based on altered profiles of alpha-fetoprotein. N Engl J Med 328: 1802-1806, 1993.
39. Okuda H, Nakanishi T, Takatsu K, Saito A, Hayashi N, Watanabe K, et al: Measurement of serum levels of des-gamma-carboxy prothrombin in patients with hepatocellular carcinoma by a revised enzyme immunoassay kit with increased sensitivity. Cancer 85: 812-818, 1999.

TABLE 1. Some benign conditions that may give rise to elevated serum CEA levels. Please note that smoking may also increase CEA levels. [Data summarised from References 2 and 3]

Back