Intestinal cancer in patients with a germline mutation in the ... - Nature

Oncogene (1998) 16, 681 ± 684  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00

SHORT REPORT

Intestinal cancer in patients with a germline mutation in the down-regulated in adenoma (DRA) gene Akseli Hemminki1, Pia HoÈglund1, Eero Pukkala2, Reijo Salovaara1,3, Heikki JaÈrvinen4, Reijo Norio5 and Lauri A Aaltonen1 Departments of 1Medical Genetics, and 3Pathology, Haartman Institute, PO Box 21 (Haartmaninkatu 3), 00014 University of Helsinki, Finland; 2Finnish Cancer Registry, Liisankatu 21 B, 00171 Helsinki, Finland; 4Second Department of Surgery, Helsinki University Central Hospital, Haartmaninkatu 4, 00250 Helsinki, Finland; 5Department of Medical Genetics, the Family Federation of Finland, PO Box 849 (Kalevankatu 16), 00101 Helsinki, Finland

A recent study has revealed that germline mutations of the down-regulated in adenoma (DRA) gene are a likely cause of a recessive intestinal absorption defect, congenital chloride diarrhea. This ®nding was in accordance with previous works showing that DRA encodes a sodium independent transporter for sulfate and oxalate. Although DRA was originally reported as a candidate tumor suppressor, these studies have questioned the relevance of DRA in cancer. To evaluate whether further studies on the role of DRA in tumorigenesis are still of interest, we examined whether individuals carrying germline DRA mutations have an excess of intestinal cancer. Cancer status of 229 members of 36 Finnish congenital chloride diarrhea families (44 homozygous patients, 70 heterozygous parents, and 115 grandparents at 50% risk of being a DRA mutation carrier) was checked at the Finnish Cancer Registry and the risk of intestinal cancer was found slightly elevated (standardized incidence ratio 3.4, 95% con®dence interval 1.4 ± 7.0, P50.05). While this result does not unambiguously demonstrate an increased intestinal cancer risk in DRA mutation carriers, it should promote further studies to determine the possible role of DRA in cancer. Keywords: DRA; CLD; cancer; neoplasia; tumor suppressor

Four years ago, the down-regulated in adenoma (DRA) gene was cloned (Schweinfest et al., 1993) and consecutively mapped to 7q22-q31.1 (Taguchi et al., 1994). DRA has a great deal of structural similarity with the diastrophic dysplasia sulphate transporter gene, which led investigators to presume a functional similarity between these two genes (HaÈstbacka et al., 1994). Structural similarity with a rat liver sulfate transporter led other researchers to study the function of DRA, and the gene was shown to encode a sodium independent transporter for sulfate and oxalate (Silberg et al., 1995). The expression of DRA is limited to the mature mucosal cells of the small and large intestine (Schweinfest et al., 1993; Silberg et al., 1995; HoÈglund et al., 1996) and to the prostate (HoÈglund et al., 1996), Correspondence: LA Aaltonen Received 16 April 1997; revised 1 September 1997; accepted 1 September 1997

and in dedierentiated states of mucosa, such as in embryonic tissues or in neoplasia, DRA expression is not detected (Schweinfest et al., 1993; Silberg et al., 1995; HoÈglund et al., 1996). Although the association between lost DRA expression and neoplasia is apparent, it is not known if this loss has a role in tumorigenesis, or if it is merely a consequence of the dedierentiated status of the epithelium. Congenital chloride diarrhea (CLD) is an autosomal recessive disorder characterized by voluminous watery stools containing a high concentration of chloride (McKusick, 1994). Other features include hydramnios and premature birth. Heterozygous carriers are symptomless. The majority of Finnish CLD patients originate from the Eastern part of Finland (see map in HoÈglund et al., 1995). The population of this area is approximately 1.35 million people. Strong evidence was recently presented indicating that CLD is caused by germline mutations in the DRA gene. A three base pair in-frame deletion (DV317) and a C307W change were found in all Finnish CLD chromosomes studied (n=64). However, the C307W seemed like a polymorphism because of its high population frequency. Furthermore, it was not disease-causing when present without DV317 (HoÈglund et al., 1996). The association between digestive tract cancer and homozygous defects in another anion transporter, the cystic ®brosis transmembrane conductance regulator gene, CFTR, has been recently documented (Neglia et al., 1995). Some of the molecular mechanisms that might associate gastrointestinal cancer predisposition and defects in CFTR have been recently revealed (Abraham et al., 1996). To determine whether the germline defect in DRA has relevance in intestinal cancer, we utilized two unique data sets that were available to us. First, all identi®ed patients (n=44), parents (n=70) and grandparents (n=115) of Finnish CLD families were checked for cancer at the Finnish Cancer Registry. Second, 728 patients with colorectal neoplasia, 485 of them from the high CLD frequency area, were studied for the presence of the DV317 mutation. 416 control individuals were also analysed for this mutation. As a result of the centralized health care system of Finland, diagnostic hallmarks, and well-studied genealogy of the disease (Norio et al., 1971) the ascertainment of CLD is believed to be almost complete. Altogether 36 CLD families were known to us. In 27 families the three base pair deletion DV317 has been demonstrated to cause the disease

Mutation in DRA and intestinal cancer A Hemminki et al

682

(HoÈglund et al., 1996), from the rest of the families no DNA samples were available. In all families, the patients were diagnosed by characteristic clinical features, and by demonstrating high fecal chloride content. Previous genealogical studies have identi®ed parents and grandparents of the patients (HoÈglund et al., 1995; Norio et al., 1971). Since the disease is recessive, all parents are obligate heterozygous carriers, and 50% of the grandparents carry one defective allele. One mother of a CLD patient was exluded from this study because paternal uniparental isodisomy of chromosome 7 caused the disease in the aected child (HoÈglund et al., 1994). Data was available from 44 patients, 70 parents and 115 grandparents. One parent, and 33 grandparents were lost during follow up, usually because they had died before personal identi®cation numbers necessary for computer linking were introduced in 1967. The Finnish Cancer Registry has legal status, is population based, and has almost 100% coverage since 1953 (KylloÈnen et al., 1987; Teppo et al., 1994). Follow-up for cancer started at the birth of the ®rst CLD patient in each family (for the grandparents follow-up started at the birth of the CLD patient's parent), or on the 1st of January 1967, whichever was later, and ended at death or 31st of December 1993, whichever occurred ®rst. The numbers of observed cases and person-years at risk were counted, by 5 year age groups, separately for the three calendar periods (1967 ± 1975, 1976 ± 1984 and 1985 ± 1993). The expected numbers of cases for total cancer and for speci®c cancer types were calculated by multiplying the number of person-years in each age group by the corresponding average cancer incidence in the whole Finnish population during the period of observation. To calculate the standard incidence ratio, the observed number of cases (see Table 1) was divided by the expected number. The statistical signi®cance was tested by the Mantel ± Haenzel chi-square test, on the presumption that the number of observed cases followed a Poisson distribution. The CLD patients were followed for 697 person years, parents 1250 person years, and grandparents 2430 person years. At the end of follow-up, there were no cases of cancer reported to the Cancer Registry among the CLD patients (homozygous for the genetic defect), all 32 years old or younger. Two of the parents (obligate heterozygous carriers of the defect) had altogether three primary intestinal cancers (P50.01, standardized incidence ratio (SIR) 20.0, 95% con®dence interval (CI) 4.1 ± 58.5). One had a cancer of the rectum at 39 years of age, the other had cancers of the ileum, rectum and lung at the age of 53 years. The cancer status of the latter patient was known to us beforehand. Four of the grandparents (at 50% risk of carrying the genetic defect) had colorectal cancer (SIR 2.1, 95% CI 0.6 ± 5.3). Thus seven cases of intestinal cancer were observed in the CLD families (SIR 3.4, 95% CI 1.4 ± 7.0, P50.05). There was no signi®cant overall increase in the frequency of cancer in the kindreds (SIR 1.1, 95% CI 0.7 ± 1.6). If the CLD parent whose cancer status (two intestinal tumors) was known to us beforehand was excluded from the study, then the increase in intestinal cancer risk in the CLD families would not remain signi®cant (SIR 2.4, 95% CI 0.8 ± 5.6).

Table 1 Cancer in CLD familes

Cancer location

Cancer histology

Intestine Rectum Sigmoid colon Ascending colon Ileum Lung

adenocarcinoma adenocarcinoma adenocarcinoma adenocarcinoma

Other Stomach Prostrate Endometrium Ovary Breast Kidney Thyroid Eye (choroid epithelium) Bladder Skin Total

Number of cases Average (parents/ age at grandparents) diagnosis

adenocarcinoma adenocarcinoma adenocarcinoma cystadenocarcinoma ductal carcinoma renal cell carcinoma papillary carcinoma

7a (3/4) 3 (2/1) 2 (0/2) 1 (0/1) 1 (1/0) 8 (1/7) 4b (0/4) 2 (1/1) 2 (0/2) 18 (1/17) 2 (0/2) 2 (0/2) 1 (0/1) 1 (1/0) 1 (0/1) 1 (0/1) 1 (0/1)

58 47 69 72 53 67 69 60 70 68 74 70 62 42 79 66 73

melanoma unknown basocellular carcinoma

1 (0/1) 1 (0/1) 7 (0/7)

57 75 69

33 (5/28)

66

microcellular carcinoma adenocarcinoma unknown

Results from the Cancer Registry follow-up: cancers found in CLD patients' parents and grandparents. aNumber of cases compared with the age and sex adjusted expected value was statistically signi®cant with the Mantel-Haenzel chi-square test, P50.05. bNumber of cases compared with the age and sex adjusted expected value was statistically signi®cant with the Mantel-Haenzel chi-square test, P50.01

In the unselected patients with colorectal neoplasia, DV317 was not found in the low frequency areas (South-Central Finland) in 243 patients (223 carcinomas, 20 adenomas). In the high CLD frequency area 6/ 485 (1.2%) patients, 2/198 (1.0%) with adenoma and 4/287 (1.4%) with carcinoma, were heterozygous for DV317 (Figure 1). Four (1.0%) of the 416 controls displayed the mutation (dierence not statistically signi®cant). In three out of the four unselected patients with DV317, carcinoma DNA was also available. Loss of the wild type allele could not be detected in any of these tumors (data not shown). The same was true with four paran embedded tumors from the CLD families displaying heterozygosity for DV317. The study focused on occurrence of intestinal cancer, but incidentally as many as eight cases of lung cancer were detected during the cancer follow-up (SIR 2.0, CI 0.9 ± 4.0). Four of these eight cases were small cell carcinomas (0.6 expected, SIR 7.1, CI 1.9 ± 18.1, P50.01), two were adenocarcinomas and two were of unknown histology. The role of DRA in colorectal tumorigenesis is unclear. The initial observation that the gene is downregulated in intestinal neoplasia suggested that it might function as a tumor suppressor (Schweinfest et al., 1993). The alternative hypothesis is that the loss of DRA expression in intestinal tumors occurs merely as a consequence of dedierentation (HoÈglund et al., 1996). Mechanisms connecting DRA and tumor development have been speculated on previously. Sulfation is necessary for the normal synthesis of proteoglycans. Defective sulfate transport might disturb this process enough to alter the eect of


growth hormones, since proteoglycans have an important role in their binding and modulation of activity (Ruoslahti and Yamaguchi, 1991). Proteoglycans have been shown to have decreased sulfation in tumors of the liver and kidney (Nakamura and Kojima, 1981; Lapis et al., 1990). Furthermore, the normal sulfomucin secretion of the colonic mucosa is reverted to a lower sulfate content sialomucous type following malignant transformation, apparently as an early event (Filipe and Branfoot, 1974; Shamsuddin et al., 1981; Yamori et al., 1987). In hepatoma cell lines, transport of inorganic anions including sulfate is greatly reduced (von Dippe and Levy, 1982), and hyposulfated proteoglycans have a lowered anity to adhesion molecules of surrounding cells (Robinson et al., 1984). Altered intercellular contacts and transmission may contribute to invasion and metastasis characteristics of malignant cell populations. This evidence suggests that sulfation mechanisms may play a role in the carcinogenesis of various tumor types. A mechanism linking DRA defects and the risk of intestinal cancer remains to be demonstrated. A similar situation is currently seen in the more obvious association between digestive tract cancer and defects in the CFTR gene, encoding a chloride transporter (Neglia et al., 1995). In a recent article the role in cancer predisposition of diering extracellular ATP levels in dierent organ systems of CFTR homozygous and heterozygous mice was proposed (Abraham et al., 1996). Loss of the wild type DRA allele was not detected in the tumors. This is not surprising considering that CLD patients are homozygous for the defect, and thrive without prominent cancer predisposition. In the past CLD was often lethal, which explains the young age (all below 32 years) of the patients available for

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

81 bp 78 bp

this study. Further follow-up will more accurately determine their risk of malignancy. The intestinal cancer risk of the CLD parents was higher than that of the grandparents. Only 50% of the grandparents have the DRA defect whereas all parents are obligate carriers. No individuals with the DV317 mutation of DRA were found outside the CLD high frequency area. Within the region, the carrier frequency of DV317 was 4/416 (1.0%) in healthy control individuals, a result in accordance with previously reported numbers of 3/252 (1.2%) (HoÈglund et al., 1996). A similar percentage of DV317 heterozygotes was found in individuals with benign colorectal tumors (2/198, 1.0%). The DV317 mutation carriership in patients with colorectal cancer was slightly, but not signi®cantly higher, 1.4% (4/287). These ®gures suggest that the impact of DV317 on cancer susceptibility is not prominent, though a low level of contribution is possible. In this study seven patients with colorectal cancer were directly demonstrated to have germline DRA defects. The data on these patients' lesions was scrutinized for possible characteristic features. Six of the lesions (86%) were located in the rectum or sigmoid colon (Table 2). All three cases of rectal cancer in the CLD families occurred at a young age (39, 50 and 53 years) (Table 1). The number of intestinal cancers in CLD families was found signi®cantly increased (SIR 3.4, 95% CI 1.4 ± 7.0, P50.05), but only if a CLD parent whose two intestinal cancers were known to us beforehand was included to the data set. The cancer registration system in Finland is virtually complete, and the computerized record linkage procedure precise (Pukkala, 1992). Therefore, technical aspects are unlikely to cause bias in the results. Almost all CLD kindreds originated from Eastern Finland, whereas the average cancer incidence of the whole Finnish population was utilized to calculate the SIR ®gures of this report. Eastern Finland is a low frequency area for colorectal cancer (Pukkala et al., 1987). Thus rather under- than overestimation of the true cancer risk SIR in CLD families is likely. An incidental observation that small cell lung cancers had occurred frequently in the studied CLD families was made (SIR 7.1, 95% CI 1.9 ± 18.1, Table 2 Patients with DRA germline mutations and colorectal carcinoma

Figure 1 Screening for DV317. The arrow indicates the 3 base pair deletion observed in a normal tissue sample derived from one of the unselected colorectal cancer patients (lane 13). Each lane (1 ± 20) represents one patient. DNA samples from 728 unselected patients with colorectal neoplasia (510 carcinomas, 218 adenomas) were available to us. These samples emanated from an ongoing project in which the frequency of mismatch repair gene mutations in Finland is being studied. 485 of these (287 carcinomas, 198 adenomas) were derived from individuals living in the high CLD frequency area in Eastern Finland (HoÈglund et al., 1995). The mutation analysis was performed using normal tissue DNA as described previously (HoÈglund et al., 1996). Brie¯y: genomic primers were used to amplify the deletion region, and PCR products were electrophoresed in a 6% polyacrylamide gel to reveal the presence or absence of the 3 base pair deletion. If DV317 was detected, loss of the wild type allele was searched for in the tumor tissue when available. 416 anonymous randomly chosen blood donors from the high CLD frequency area served as controls. Mutation status was determined as shown above

Patienta Sex Age 1 2 3 4 5 6 7

F F M M M M M

Site of cancer

CLD family history

79 ascending colon none 81 rectum none 71 rectum none 64 rectum none 39 rectum father of patient 53 rectum father of patient 75 sigmoid grandfather of colon patient

Histologyb Dukesc adenoca. adenoca. adenoca. adenoca. adenoca. adenoca. adenoca.

B B B A A±B D A±B

Features of the patients with colorectal cancer and the DRA mutation in their germline. All patients are heterozygous for the mutation. a1 ± 4 are patients with sporadic colorectal carcinomas, 5 ± 7 represent CLD family members. bAdenoca. is an abbreviation for adenocarcinoma. cDukes refers to the Dukes classi®cation of colorectal cancers according to degree of invasion of surrounding tissues and metastasis. DNA was not available from all patients reported in Table 1

683


684

P50.01). In the absence of DRA expression in the lung tissue (Schweinfest et al., 1993; Hemminki, unpublished), the association between DRA defects and small cell lung cancer remains highly speculative, although DRA germline mutations may have some systemic eects. All the CLD patients and relatives were identi®ed during previous works that were not related to cancer research. This makes a selection bias favouring individuals with cancer in the CLD cohort unlikely. One may argue, whether the family of the individual with two intestinal cancers (known to us before the present work) should have been excluded from the study. However, this family was identi®ed as a CLD kindred more than 25 years ago; before any of the tumors were diagnosed (Norio et al., 1971). As individuals that were not alive after year 1966 were excluded, some grandparents with cancer might have remained unidenti®ed. If the relative risk is at the highest in young age groups, exclusion of follow-up before 1967 leads to underestimation of the true risk. At present DRA does not ful®ll the de®nition of a tumor suppressor gene (Haber and Harlow, 1997). Further studies are needed to clarify whether DRA mutations contribute to hereditary cancer predisposition, and to sporadic tumor initiation and progression. These questions can be approached by studying the cancer risk of CLD families with dierent DRA

mutations in dierent populations, and by continuing molecular studies on the possible role of DRA in tumorigenesis.

Acknowledgements We thank Dr Albert de la Chapelle for support, Drs Juha Kere and Christer Holmberg for sharing unpublished data, and Dr Anu Moisio for access to control samples. The unselected tumor samples were derived from a study conducted by the Finnish HNPCC Clinical Consortium, whose members are: Jukka-Pekka Mecklin, Heikki JaÈ rvinen, Heikki Ahtola, Matti Eskelinen, Niilo HaÈrkoÈnen, Risto Julkunen, Eero Kangas, Seppo Ojala, Jukka Tulikoura, and Erkki Valkamo. We thank Minna Veini, Marilotta Turunen, Siv Lindroos, Sinikka Lindh, Tuula Lehtinen, and Kirsi Hopponen for technical assistance. This study was aided by grants from the Academy of Finland, the Federation of Finnish Insurance Companies, the Foundation for Paediatric Research, the Ulla Hjelt Fund, the Finnish Cancer Society, the Sigrid Juselius Foundation, the Finnish Medical Society Duodecim, the Ida Montin Foundation, the Jalmari and Rauha Ahokas Foundation, the 350th Anniversary Foundation of the University of Helsinki, the National Cancer Institute (CA67941), and the Helsinki University Central Hospital. Part of the study was conducted at the FolkhaÈlsan Institute of Genetics.

References Abraham EH, Vos P, Kahn J, Grubman SA, Jeerson DM, Ding I and Okunie P. (1996). Nature Med., 2, 593 ± 596. Filipe MI and Branfoot AC. (1974). Cancer, 34, 282 ± 290. Haber D and Harlow E. (1997). Nature Genet., 16, 320 ± 322. HaÈstbacka J, de la Chapelle A, Mahtani MM, Clines G, Reeve-Daly M-P, Daly M, Hamilton BA, Kusumi K, Trivedi B, Weaver A, Coloma A, Lovett M, Buckler A, Kaitila I and Lander ES. (1994). Cell, 78, 1073 ± 1087. HoÈglund P, Haila S, Socha J, Tomaszewski L, SaarialhoKere U, Karjalainen-Lindsberg ML, Airola K, Holmberg C, de la Chapelle A and Kere J. (1996). Nature Genet., 14, 316 ± 319. HoÈglund P, Holmberg C, de la Chapelle A and Kere J. (1994). Am. J. Hum. Genet., 55, 747 ± 752. HoÈglund P, Sistonen P, Norio R, Holmberg C, Dimberg A, Gustavson KH, de la Chapelle A and Kere J. (1995). Am. J. Hum. Genet., 57, 95 ± 102. KylloÈnen LEJ, Teppo L and Lehtonen M. (1987). Ann. Chir. Gynaecol., 76, 185 ± 190. Lapis K, Kavalsky I, Jeney A, Pogany G, Molnar G, Repassy D, Szecseny A and Karacsonyi S. (1990). Tokai J. Exp. Clin. Med., 15, 155 ± 165. McKusick VA. (1994). Mendelian Inheritance in Man: a catalog of human genes and genetic disorders, 11th ed. Johns Hopkins University Press: Baltimore and London. Nakamura N and Kojima J. (1981). Cancer Res., 41, 278 ± 283. Neglia JP, FitzSimmons SC, Maisonneuve P, SchoÈni MH, SchoÈni-Aolter F, Corey M, Lowenfelts AB and the Cystic ®brosis and cancer study group. (1995). N. Engl. J. Med., 332, 494 ± 499.

Norio R, Perheentupa J, Launiala K and Hallman N. (1971). Clin. Genet., 2, 182 ± 192. Pukkala E, Gustavsson N and Teppo L. (1987). Atlas of cancer incidence in Finland 1953 ± 82. Cancer society of Finland publication No. 37: Helsinki. Pukkala E. (1992). Use of Record Linkage in Small-area Studies: Geographical and Environmental Epidemiology. Elliott P, Guzick J, English D, Stern R (eds.). Oxford university press: Oxford, pp. 125 ± 131. Robinson J, Viti M and HoÈoÈk M. (1984). J. Cell. Biol., 98, 946 ± 953. Ruoslahti E and Yamaguchi Y. (1991). Cell, 64, 867 ± 869. Schweinfest CW, Henderson KW, Suster S, Kondoh N and Papas TS. (1993). Proc. Natl. Acad. Sci. USA, 90, 4166 ± 4170. Shamsuddin AKM, Weiss L, Phelps PC and Trump BF. (1981). J. Natl. Cancer Inst., 66, 413 ± 419. Silberg DG, Wang W, Moseley RH and Traber PG. (1995). J. Biol. Chem., 270, 11897 ± 11902. Taguchi T, Testa JR, Papas TS and Schweinfest C. (1994). Genomics, 20, 146 ± 147. Teppo L, Pukkala E and Lehtonen M. (1994). Acta Oncol., 33, 365 ± 369. von Dippe P and Levy D. (1982). J. Biol. Chem., 257, 4381 ± 4385. Yamori T, Kimura H, Stewart K, Ota DM, Cleary KR and Irimura T. (1987). Cancer Res., 47, 2741 ± 2747.