Patients with celiac disease (CD), particularly those whose disease is undiagnosed or who do not adhere to a gluten-free diet, may be exposed to various health risks, the most serious of which is malignancy. Many reports of patients with lymphoma in association with steatorrhea appeared in the early literature, but before 1961 malabsorption was always attributed to the presence of the tumor. In 1962, however, it was proposed that lymphoma is a complication of CD and that the mucosal lesion is premalignant. Further observations strengthened this hypothesis. Subsequently, other malignant tumors were found in association with CD and an increased incidence of gastrointestinal cancer, in general, and carcinoma of the esophagus, in particular, was reported. Carcinoma of the small bowel in association with CD is now known to be the second most common invasive malignancy after lymphoma. All of this work has been reviewed (1,2).
An impressive amount of information exists about CD and malignancy, but there is much still to explore. Can lymphoma be prevented? What type of lymphoma is it? What is the relationship of a flat small-intestinal biopsy to lymphoma? And is chronic ulceration of the bowel part of the spectrum of lymphoma? Is there a place for screening programs to identify undiagnosed patients potentially at risk, who might derive benefit from a gluten-free diet? What constitutes a gluten-free diet, and what do patients themselves wish to know about the malignant risk? All these questions require attention.
The prevalence of lymphoma as a complication of CD is unknown for a number of reasons. Thus, the prevalence of CD itself in the community is uncertain, and lymphoma may be missed if autopsies are not routinely carried out. Some patients with established lymphoma may have CD, but the diagnosis is not usually pursued in this group (3). Data coming from referral centers will not necessarily reflect the situation in the general population, and it is likely that the prevalence is changing with the wider use of a gluten-free diet. While there are clearly reservations about the data, information on prevalence is available for groups of patients who have been followed up for many years (Table 1) (4-7,9).
Patients with CD in whom lymphoma has developed may come to our attention in one of two ways. One group comprises those with established CD who initially respond to a gluten-free diet but later deteriorate because of the onset of cancer. In the second group CD and malignancy are found together or at about the same time. Whether this latter group, with flat biopsy results and malignancy, has CD has been the subject of considerable controversy. The weight of evidence indicates that they do and that the development of malignancy is the event that brings CD to light. When compared with uncomplicated cases of CD, this group has a similar HLA antigen profile (8,10), similar changes in intraepithelial subsets (11), and a similar incidence of hyposplenism. A recent immunohistochemical study of enteropathy-associated T-cell lymphoma also supports this view (12).
In some patients the presentation is acute, manifesting in, for example, intestinal perforation or obstruction, bleeding from a tumor, or an opportunistic infection. In the majority of cases, however, the illness is insidious. At the diagnosis of lymphoma, the tumor is usually wide-spread, and the prognosis is very poor.
The diagnosis of lymphoma arising in the context of CD can be difficult to make and may be delayed, because the presenting features are often nonspecific and indistinguishable from uncomplicated CD itself, either at diagnosis or in relapse. Many hematological and biochemical abnormalities are present, but no pattern emerges to permit early diagnosis. A progressive rise in serum IgA concentration may be found, but this is also seen in CD without malignancy. Increased levels of lysozyme have been noted, but they are not a reliable indicator of lymphoma. Differences in the small-intestinal mucosa of those with uncomplicated CD and those in whom lymphoma eventually develops have been sought. Lower plasma cell and higher lymphocyte counts in the lamina propria, lower lymphocyte counts in the epithelium, hypoplastic crypts, and histiocytic aggregates have all been cited as features. These observations, however, are unlikely to be helpful in individual cases.
Upper-gastrointestinal endoscopy, which is now the preferred method of diagnosing CD, allows for diagnosis of lesions in the esophagus, stomach, and proximal bowel. Bowel radiology should always be carried out. Multiple irregular narrowed segments are characteristic of small-intestinal involvement with lymphoma, but appearances can be difficult to interpret. The place of scanning techniques is still to be determined, but they will detect the presence of mesenteric lymph node cavitation, which may be mistaken for lymphoma (13). Laparotomy may be necessary when the presence of cancer is suspected and a firm diagnosis has not been made. This can be a difficult procedure in the case of a debilitated patient, and it should not be undertaken lightly. Occasionally at laparotomy an isolated lesion can be resected, with good results.
Tumors are usually located in the jejunum but may be found in the ileum, stomach, and colon. The primary tumor may be present in extragastrointestinal sites, such as the thyroid, liver, or peripheral lymph nodes. Occasionally at autopsy no gross lesion is found, but only microscopic foci of lymphoma. The lymphoma is derived from T lymphocytes. Attention has also focused on intraepithelial lymphocytes. A monoclonal antibody, HML-1, developed against intraepithelial lymphocytes also stains cells of lymphoma-complicating CD (14). It is possible that enteropathy-associated T-cell lymphoma derives from intraepithelial lymphocytes and that a small-cell lymphoma arising from these cells forms an intermediate stage in the development of the large-cell tumor (15).
CARCINOMA OF THE SMALL BOWEL
Until recently carcinoma of the small bowel as a complication of CD was thought to be rare, but a British collaborative study showed this to be the second most common invasive malignancy after lymphoma (10). Anemia is the most usual initial feature and is often associated with overt or occult blood loss. Weight loss, abdominal pain, and intestinal obstruction are also prominent features. If the tumor is in the proximal duodenum, it may be seen at endoscopy. A barium series will usually be helpful in locating the lesion. Long-term survival may occur if resection is carried out while the tumor is still confined to the bowel.
MALIGNANCY IN OTHER GROUPS WITH CELIAC DISEASE
Patients with dermatitis herpetiformis have a higher than normal risk of malignancy. The relative risk of lymphoma in this group has been reported as 100 (16).
Relatives of Patients with Celiac Disease
Because CD is a genetically determined and complicated by lymphoma, it might be expected that more deaths from lymphoma would be seen in families of patients with CD. In the only study to address this problem, a significant increase in deaths from malignancy was found among family members of patients with CD (17). No increase in lymphoma was observed, but there was a higher rate of esophageal carcinoma in men. Studies of this type are difficult to carry out, and the results must be interpreted with caution.
This is an uncommon and poorly understood complication of CD, in which chronic ulcers are found mainly in the jejunum and ileum but occasionally in the colon (18). Ulcerative jejuno-ileitis may bring a patient with CD to diagnosis and, as may happen with lymphoma, cause deterioration in patients with CD whose disease was controlled on a gluten-free diet. This condition can be difficult to differentiate from bowel lymphoma both clinically and radiologically. The disorders may coexist in one patient, and sometimes ulceration will be diagnosed before lymphoma becomes apparent.
EFFECT OF A GLUTEN-FREE DIET ON MALIGNANCY
At diagnosis, lymphoma is nearly always widespread, and the prognosis is therefore very poor: survival is measured in weeks to months. Earlier diagnosis of lymphoma, which would allow for a better prognosis, is not feasible at present, and predicting which CD patients will develop cancer is not possible either. Attention has therefore turned to prevention and the role that a gluten-free diet might play. Certain aspects of the small-intestinal mucosa indicate that it shows premalignant characteristics with respect to both lymphoma and carcinoma. These features include more numerous mitoses in the crypts and lymphoid cells and irregularity and increased basophilia in the surface epithelium. In addition, the mucosa in untreated CD is abnormally permeable and may also be deficient in carcinogen-detoxifying enzymes. These disturbances may allow carcinogens to penetrate the mucosa. A gluten-free diet restores the structure and function of the mucosa toward normal and hence might reduce the malignant potential.
To try to answer the question of whether a gluten-free diet reduces the risk of malignancy, a series of patients from one clinic has been kept under long-term close surveillance (19). The analysis compared cancer morbidity among the patients with that observed in the general population using the method of years at risk. The expected number of tumors was calculated and compared with the number observed (Table 2). A twofold risk of cancer was found among the patients with CD, mainly tumors of the mouth, pharynx, and esophagus and non-Hodgkin lymphoma.
Of the 210 patients in the series, 108 had followed a strict gluten-free diet, while the remainder were on a normal or reduced gluten diet. For those who had followed a strict gluten-free diet for ≥5 consecutive years, the overall risk of malignancy was not significantly increased when compared with that of the general population. For those ingesting gluten, however, there was a significantly increased risk (Table 2). No differences could be detected for individual sites or types (mouth, pharynx, or esophagus and lymphoma), but when these were combined, the difference between the relative risks (6.7 for those on a strict gluten-free diet and 38.7 for those ingesting gluten) was significant (p < 0.05). Excess morbidity was also calculated and was clearly related to the amount of gluten ingested (Table 3). The decreasing trend in higher morbidity rates with increasing adherence to a gluten-free diet was significant (p < 0.01). These results, combined with anecdotal evidence from other CD clinics, indicate that a gluten-free diet does protect against malignancy and is another reason to advise all patients to adhere to a strict gluten-free diet for life.
WHAT CONSTITUTES A GLUTEN-FREE DIET?
A gluten-free diet based on cornstarch contains very small amounts of gluten. Within the context of malignancy, it is not known whether these small quantities are harmful to patients, although it could be argued from the evidence that a diet completely free of gluten should be pursued. Wheat-free products that are palatable and competitively priced are available, and it is likely that an increasing demand for them will come from patients themselves.
Screening for Celiac Disease
Screening specific groups of people who might be at high risk of CD or even large general populations is possible. It is evident from these studies that patients who are symptomatic and can benefit from a gluten-free diet can be identified. On the other hand, many patients are without symptoms. If a gluten-free diet reduces the risk of malignancy, then identifying patients with CD and starting them on such a diet should be beneficial. Clearly there are both medical and ethical issues to consider here.
Approach to Patients
Patients are becoming increasingly aware of a link between CD and malignancy and wish for further information. It is now possible to adopt an encouraging and positive attitude and advise patients that on current evidence, the best insurance against the development of this complication is strict adherence to a gluten-free diet. This is a sensitive area, and a careful approach to patients is required so that they do not become unduly worried.
1. Cooke WT, Holmes GKT. Malignancy. In: Cooke WT, Holmes GKT eds. Coeliac disease
. Edinburgh: Churchill Livingstone, 1984:172-96.
2. Holmes GKT, Thompson H. Malignancy as a complication of coeliac disease. In: Marsh MN, ed. Coeliac disease
. London: Blackwell, 1992:105-35.
3. Freeman HJ, Weinstein WM, Shnitka TK, Piercey JRA, Wensel RH. Primary abdominal lymphoma: presenting manifestations of celiac sprue, or complicating dermatitis herpetiformis? Am J Med
4. Harris OD, Cooke WT, Thompson H, Waterhouse JAH. Malignancy in adult coeliac disease and idiopathic steatorrhoea. Am J Med
5. Holmes GKT, Stokes PL, Sorahan TM, Prior P, Waterhouse JAH, Cooke WT. Coeliac disease, gluten-free diet and malignancy. Gut
6. Brandt L, Hagander B, Norden A, Stenstam M. Lymphoma of the small intestine in adult coeliac disease. Acta Med Scand
7. Selby WS, Gallagher ND. Malignancy in a 19-year experience of adult coeliac disease. Dig Dis Sci
8. O'Driscoll BRC, Stevens FM, O'Gorman TA, et al. HLA type of patients with coeliac disease and malignancy in the West of Ireland. Gut
9. Nielsen OH, Jacobsen O, Pedersen ER, et al. Non-tropical sprue: malignant disease and mortality rate. Scand J Gastroenterol
10. Swinson CM, Slavin G, Coles EC, Booth CC. Coeliac disease and malignancy. Lancet
11. Spencer J, MacDonald TT, Diss TC, Walker-Smith JA, Ciclitira PJ, Isaacson PG. Changes in intraepithelial lymphocyte subpopulations in coeliac disease and enteropathy associated T cell lymphoma (malignant histiocytosis of the small intestine). Gut
12. Murray A, Cuevas EC, Jones DB, Wright DH. Study of the immunohistochemistry and T cell clonality of enteropathy-associated T cell lymphoma. Am J Pathol
13. Holmes GKT. Mesenteric lymph node cavitation in coeliac disease. Gut
14. Spencer J, Cerf-Bensussan N, Jarry A, et al. Enteropathy-associated T cell lymphoma (malignant histiocytosis of the intestine) is recognized by a monoclonal antibody (HML-1) that defines a membrane molecule on human mucosal lymphocytes. Am J Pathol
15. Alfsen GC, Beiske K, Bell H, Marton PF. Low-grade intestinal lymphoma of intraepithelial T lymphocytes with concomitant enteropathy-associated T cell lymphoma: case report suggesting a possible histogenic relationship. Hum Pathol
16. Leonard JN, Tucker WFG, Fry JS, et al. Increased incidence of malignancy in dermatitis herpetiformis. Br J Med
17. Stokes PL, Prior P, Sorahan TM, McWalter RJ, Waterhouse JAH, Cooke WT. Malignancy in relatives of patients with coeliac disease. Br J Prev Soc Med
18. Holmes GKT. Ulcerative jejuno-ileitis. In: Allan RN, Keighley MRB, Alexander-Williams J, Fazio VW, Hanauer S, Rhodes JM, eds. Inflammatory bowel disease,
3rd ed. Edinburgh: Churchill Livingstone, 1997:431-5.
19. Holmes GKT, Prior P, Lane MR, Pope D, Allan RN. Malignancy and coeliac disease: effect of a gluten free diet. Gut
Collin: In my hospital (Tampere) we followed 450 CD patients for about 4 to 5 years, and in only one of them did lymphoma develop. That doesn't mean that lymphoma has disappeared from Tampere. During the past 5 years we have found two more patients with CD and lymphoma. They came to the hospital because of symptoms of the malignancy. On the other hand, among 300 patients with CD we found that 30% had silent CD. These patients were found by CD screening.
Peña: Dr. Holmes, do you think that aggressive therapy, for example, immunotherapy, could improve the prognosis of the patients with CD who have intestinal ulceration or lymphoma?
Holmes: The two patients that I discussed who had intestinal ulceration had straightforward CD and responded to a gluten-free diet. However, they subsequently deteriorated with the onset of intestinal ulceration. Additional small-bowel biopsies showed flat mucosa. In one of these patients the serum albumin concentration fell to the very low level of 9 g/L, and she was treated with parenteral nutrition, steroids, and azathioprine and is well 6 years later. The other one was treated just with steroids and also improved. I have other patients with ulceration, none of whom were treated with aggressive therapy. I have one patient who showed signs of lymphoma 9 years after the onset of intestinal ulceration. Patients with lymphoma should be treated aggressively with chemotherapy if it is deemed advisable. There is accumulating evidence that this treatment prolongs survival in some cases.
Lamers: We have all seen similar patients. We have had patients who had lymphomas long after the onset of ulcerative jejunitis, while other patients have ulcerative jejuno-ileitis without any signs of lymphoma.
Holmes: These patients are difficult to treat, as is evident in the publication of Baer and colleagues in Gastroenterology in 1980 (Baer et al. Gastroenterology 1980;79:754-65. They found 40 such patients, and 29 were known to have died. Today survival may be better with more aggressive treatment. In the past it is likely that some of these patients with ulcerative jejunitis did not survive long enough to manifest lymphoma.
Lamers: It is difficult to decide how to treat these patients with ulcerative jejunitis; some of them are clinically fine. At times their complaints recur; I then treat them with steroids and metronidazol.
Vandenbroucke: Dr. Holmes, to come back to the point of the risk of lymphoma among people with CD, are they clinically recognized as patients with CD or not? You mentioned studies on series of patients with lymphoma. In some of the cases CD was recognized only after the diagnosis of lymphoma. Do you know of data on patients with lymphoma with an earlier clinical diagnosis of CD? This might be a starting point for risk calculation.
Holmes: In the publication of Freeman and colleagues in 1977, it is reported that some patients were first found to have lymphoma, which then led to the diagnosis of CD (Freeman et al. Am J Med 1977;63:585-94). In most other instances lymphoma and CD are diagnosed at about the same time, or else lymphoma complicates CD after a considerable period of time.