J. Pathol. 188: 378–381 (1999)

FHIT GENE IN GASTRIC CANCER: ASSOCIATIONWITH TUMOUR PROGRESSION AND PROGNOSIS

1, 1, - 1, 2 . 1*1Institute of Pathology, Heinrich-Heine-University, 40225 Düsseldorf, Germany

2Institute of Medical Statistics, Heinrich-Heine-University, 40225 Düsseldorf, Germany

SUMMARY

The FHIT (fragile histidine triad) gene has been recently identified and cloned at chromosome 3p14.2 including FRA3B, the mostcommon fragile site in the human genome. FHIT is suggested to be a candidate tumour suppressor gene in gastrointestinal tract tumours.To elucidate the role of the FHIT gene in gastric cancer, a total of 133 curatively R0-resected gastric carcinomas were investigated forloss of heterozygosity (LOH) at 3p14.2, using four polymorphic microsatellite loci (D3S1300, D3S1313, D3S1481, and D3S1234).LOH of the FHIT gene affecting at least one of the investigated loci was observed in 20 of 123 informative tumours (16·3 per cent). Thepresence of LOH was correlated neither with major prognostic factors such as pT category, pN category or vascular invasion, nor withhistological type or grade of differentiation of the tumours. In addition, there were no differences in the prognosis between patients withgastric carcinomas showing LOH at the FHIT gene and patients with tumours lacking LOH at the FHIT gene. These findings suggestthat LOH of the FHIT gene represents an event in the tumourigenesis of only a small subset of gastric carcinomas and does not correlatewith tumour progression or prognosis. Copyright ? 1999 John Wiley & Sons, Ltd.

KEY WORDS—FHIT gene; gastric cancer; prognosis

INTRODUCTION

The FHIT gene, located at chromosome region3p14.2, has been recently proposed as a tumour suppres-sor gene covering the common human fragile siteFRA3B and the t(3;8) translocation breakpoint in renalcell carcinoma.1 Encoding an Ap3A hydrolase, it mightbe involved in the regulation of the cell cycle or DNAreplication.2,3 Although its in vivo function is stillunknown, FHIT gene deletions or aberrant transcriptshave been identified in a variety of human malignancies,such as lung cancer,4–6 breast cancer,7–11 and head andneck carcinomas,12,13 as well as in tumours and tumourcell lines of the gastrointestinal tract.1,14–18 These find-ings suggest that FHIT might be a wide-ranging tumoursuppressor gene, as important as p53 for the tumouri-genesis of various human malignancies.19 However, ingastric and colorectal carcinomas, contradictory resultsconcerning the frequency of FHIT gene abnormalitieshave been reported,17,20 and the same is true for ovarianand endometrial cancer21,22 and for brain tumours.23

Alterations of the FHIT gene seem to be rare events andless important for the development of these tumours.Nevertheless, a correlation could be shown betweenLOH affecting microsatellite markers within the FHITgene and abnormal FHIT transcripts in lung or headand neck cancer,6,12,13 suggesting that loss of one FHITallele might be an early and important step in thecarcinogenesis of these tumours. Concerning gastriccancer, there are three studies1,14,16 encompassing a totalof 9, 40, and 45 tumours, respectively, showing allelic

CCC 0022–3417/99/090378–04$17.50Copyright ? 1999 John Wiley & Sons, Ltd.

deletion or aberrant transcripts in up to 50 per cent ofthe investigated gastric carcinomas. Although these datasuggest a possible role for the FHIT gene in the carcino-genesis of gastric cancer, nothing is known about theassociation with clinico-pathological parameters or itsputative prognostic role in gastric cancer, or indeed inother types of human malignancies.

In our study, therefore, the 3p14.2 region was inves-tigated in 133 curatively R0-resected gastric carcinomasfor LOH at microsatellite loci within or adjacent to theFHIT gene, with special emphasis on the comparisonwith pathological parameters and patients’ prognosis.

MATERIALS AND METHODS

Patients

Specimens of 133 curatively resected gastric cancerpatients were analysed. Follow-up letters for all patientswere sent to the surgeons or local tumour registers toobtain up-to-date information on survival or death, andwhether death resulted from tumour recurrence or not.The median follow-up time was 2·3 years, with a rangefrom 2 months to 9·1 years. None of the patientsreceived adjuvant chemotherapy. To eliminate bias dueto deaths directly resulting from operation, patients whodied within 4 weeks after surgery were excluded from thesurvival analysis. Curative surgery was defined as theremoval of all gross tumour and the demonstrationof tumour-negative surgical margins by microscopicexamination of the total circumference of the gastricresection line (R0 resection according to the TNMclassification,24 UICC). Seventy-nine patients (62·7 percent) were male and 47 (37·3 per cent) female. The meanage was 57·4 (range 23–89) years.

*Correspondence to: Dr H. E. Gabbert, Institute of Pathology,Heinrich-Heine-University, Moorenstrasse 5, 40225 Dusseldorf,Germany.

Contract/grant sponsor: Charlotte-und-Alfred-Pierburg-Stiftung.

Received 24 August 1998Revised 9 November 1998

Accepted 22 February 1999

379FHIT IN GASTRIC CANCER

Pathological review

The histological type of tumour was determinedaccording to the World Health Organization25 and theLauren classifications.26 To compare the prognosticsignificance of LOH of the FHIT gene with otherknown prognostic parameters, the following morpho-logical details were recorded: depth of invasion (pTcategory), lymph node involvement (pN category),grade of tumour differentiation, and blood vesseland lymphatic vessel invasion (BVI and LVI,respectively27).

LOH analysis

Tumour DNA from 133 gastric carcinomas and DNAfrom corresponding non-neoplastic gastric mucosa wereisolated from formalin-fixed, paraffin-embedded tissuesections according to standard protocols using theQIAamp tissue kit (Qiagen Inc., Hilden, Germany). Ineach case, tumour DNA was prepared by a micro-dissection technique to minimize contamination withnormal tissue DNA, and tumour cells comprised at least70 per cent of the microdissected areas. One hundredand thirty-three paired tumour/non-tumour DNAs wereevaluated for LOH at the FHIT locus using the fourdifferent microsatellite markers D3S1234, D3S1300,D3S1313, and D3S1481.1,9,10

The polymerase chain reaction (PCR) was performedin a total volume of 50 ìl containing 200 ng of DNA,10 m Tris–HCl (pH 8·8), 1·5 m MgCl2, 50 m KCl,5 per cent DMSO, 0·1 per cent Triton X-100, 200 ìdNTP, 400 n of each primer, and 2·5 units of Taqpolymerase (Qiagen Inc., Hilden, Germany). After aninitial denaturation step at 94)C for 4 min, 35 cycleswere performed, consisting of 30 s at 94)C, 45 s at 56)C,and 30 s at 72)C. Five microlitres of the PCR productwas diluted (1:1) with a denaturating loading buffer (98per cent formamide, 0·1 per cent xylene cyanol, 0·1 per

Copyright ? 1999 John Wiley & Sons, Ltd.

cent bromophenol blue, and 10 m ethylendiamine-tetraacetic acid; pH 8·0). Products were separated in an8 per cent denaturating polyacrylamide gel (2 h at110 W) and silver-stained. For informative cases, allelicloss was scored if the signal of one allele was lost orreduced to at least 50 per cent in the tumour DNA,compared with the corresponding normal allele. The locidisplaying microsatellite instability were not evaluatedfor allelic loss.

Statistical analysis

Comparisons of different groups were performedusing the Wilcoxon–Mann–Whitney test (for twogroups) or the Kruskal–Wallis test (for more than twogroups). Analyses of survival were performed using theKaplan–Meier method28 and differences between thepatient groups were tested by the log-rank test.29

Prognostic relevance was verified by applying univariateand multivariate Cox regression analysis. p values lessthan 0·05 were considered significant.

RESULTS

Each tumour (n=133) was analysed at all four micro-satellite loci, the D3S1234 locus being informative in75·6 per cent, the D3S1300 locus in 41·4 per cent, theD3S1313 locus in 36·8 per cent, and the D3S1481 locusin 58·6 per cent. By combination of all four micro-satellite markers, 92·3 per cent of the tumours (n=123)were informative for at least one investigated marker.Cases showing microsatellite instability at the investi-gated loci could not be evaluated for LOH. Thus,microsatellite instability was detected in 8·0 per cent(n=9) at locus D3S1234, in 4·3 per cent (n=4) atD3S1300, in 12·0 per cent (n=13) at D3S1313, and in9·3 per cent (n=11) at D3S1481.

Fig. 1—Representative examples of LOH at different chromosomal loci in gastric carcinomas. Normal DNA(N) and tumour DNA (T) from the same patients

J. Pathol. 188: 378–381 (1999)

380 T. NOGUCHI ET AL.

LOH affecting at least one locus of the FHIT gene(Fig. 1) could be observed in 16·3 per cent (n=20) of thetumours showing marked differences in LOH fre-quencies between the four markers. Thus, LOH of theFHIT gene could be detected in 3·2 per cent (3 of 93informative cases) at D3S1234, in 9·1 per cent (5 ofinformative 55 cases) at D3S1300, in 26·5 per cent (13 of49 informative cases) at D3S1313, and in 2·6 per cent (2of 78 informative cases) at D3S1481.

Comparing the presence of LOH at the FHITlocus with histomorphological variables, there were nostatistically significant differences concerning the histo-logical type and growth pattern of the tumours accord-ing to the Lauren and WHO classifications (Table I).Regarding the depth of invasion (pT category), allelicloss of the FHIT gene could be observed in 25·0 per cent(n=10) of all early gastric carcinomas investigated, butthere were no statistically significant differences in theLOH frequency between early gastric cancer (pT1tumours) and advanced (pT2–pT4) tumours (p=0·31).In addition, no significant differences could be observedwhen LOH of the FHIT gene was compared with lymph

Copyright ? 1999 John Wiley & Sons, Ltd.

node involvement (pN category), grade of tumour dif-ferentiation, and vascular invasion (Table I). The samewas true when all four microsatellite markers wereanalysed separately.

Finally, analysis based on the log-rank test revealedno significant differences in the survival rates (Fig. 2)between patients with LOH affecting the FHIT gene andthose without LOH (p=0·29). Additionally, LOH of theFHIT gene did not influence the prognosis within differ-ent subgroups of patients according to the Laurenclassification, or the major prognostic parameters depthof invasion (pT category) and lymph node involvement(pN category).

By multivariate analysis, only the parameters pTcategory and pN category as well as blood vesselinvasion, but not allelic loss of the FHIT gene, could beverified as independent prognostic factors.

DISCUSSION

In the present study, the frequency of LOH was aboutthe same in early and advanced gastric carcinomas.Concerning the pathological profile of the tumours,there were no differences between tumours showingLOH at the FHIT gene locus and those without.Furthermore, a possible prognostic influence of FHITgene alterations in cancer patients was investigated, butno differences in prognosis could be observed betweentumours with LOH of the FHIT gene and those without.Thus, allelic loss of the FHIT gene could be correlatedneither with a more aggressive tumour type, nor with aprognostic impact. Our data suggest that FHIT geneinvolvement is an early event in tumourigenesis, but onlyin a small group of gastric carcinomas. To our knowl-edge, no comparable data are so far available concern-ing a possible prognostic impact of the FHIT gene ingastric cancer or in other types of human malignancies.

Concerning the frequency of LOH affecting the FHITgene in gastrointestinal cancer, our data are in concord-ance with the findings of Tamura et al.17 who were also

Fig. 2—Survival rates of 123 gastric cancer patients with tumoursshowing LOH at the FHIT gene (n=20) and tumours without LOH atthe FHIT gene (n=103). Log-rank test: p=0·29

Table I—LOH at the FHIT gene and correlation with differenthistomorphological variables

n

No. of LOHpositive tumours

(%) p value

Lauren classificationIntestinal 57 10 (17·5)Diffuse 54 9 (16·7) 0·73Mixed 12 1 (8·3)

WHO classificationAdenocarcinoma

Papillary 3 0 (0·0)Tubular 45 8 (17·8) 0·84Mucinous 8 1 (12·5)Signet ring cell 48 9 (18·8)Undifferentiated 19 2 (10·5)

GradingG1*/G2 19 3 (15·8)G3 82 14 (17·1) 0·93G4 22 3 (13·6)

pT categorypT1 40 10 (25·0)pT2 46 5 (10·9) 0·31pT3/4† 37 5 (13·5)

pN categoryNode negative 64 13 (20·3)Node positive 59 7 (11·9) 0·21

Blood vessel invasionNegative 100 17 (17·0)Positive 23 3 (13·0) 0·64

Lymphatic vessel invasionNegative 74 13 (17·6)Positive 49 7 (14·3) 0·63

*G1 only one tumour.†pT4 only one tumour.

J. Pathol. 188: 378–381 (1999)

381FHIT IN GASTRIC CANCER

able to demonstrate LOH of the FHIT gene in only 13per cent of the informative gastric carcinomas, and thefindings of Thiagalingam et al.,20 who could detectallelic loss at the FHIT gene in only 22 per cent ofcolorectal carcinomas. In contrast, Ohta et al.1 couldfind abberrant FHIT transcripts in 50 per cent of gastriccarcinomas, but had included only nine tumours in theirstudy. Gemma et al.16 were able to demonstrate allelicloss of the FHIT gene in 42 per cent of 38 gastriccarcinomas, most of them (n=35) being advanced stagetumours with regional or distant metastasis. Unfortu-nately, no information was provided in their paperabout the frequency of microsatellite instability of theirinvestigated loci. Recently, Kastury et al.14 found LOHof the FHIT gene in 17 of 37 gastric carcinomas (46 percent). However, for most of the investigated tumours, nodata concerning tumour stage were presented andtumour samples were collected from a number ofdifferent institutions.

At the present time, the exact function of the FHITgene product is still unknown, and the question has to beanswered whether it is in fact the FHIT gene, or possiblyanother gene located close to the FHIT gene, that isaltered and involved in tumourigenesis. Furthermore, byinvestigating allelic loss, we cannot exclude gene muta-tion and transcriptional or post-transcriptional altera-tions of the FHIT gene, although Tamura et al.17 werenot able to show abnormal FHIT transcripts in any ofthe 23 primary gastric carcinomas investigated.

Summarizing the results of our study, LOH of theFHIT gene occurs in early gastric cancer at the samefrequency as in advanced gastric cancer, but only in asmall subset of gastric carcinomas. However, LOH ofthe FHIT gene is correlated neither with a definitetumour histology or growth pattern, nor with prognosis.Since other mechanisms affecting the FHIT gene (suchas point mutation, methylation or post-transcriptionalalterations) might be involved, further studies are stillnecessary to elucidate the role of the FHIT gene ingastric cancer, including gene expression at the proteinlevel and the determination of its function.

ACKNOWLEDGEMENTS

This work was supported by the Charlotte-und-Alfred-Pierburg-Stiftung.

REFERENCES1. Ohta M, Inoue H, Cotticelli MG, et al. The FHIT gene, spanning

the chromosome 3p14.2 fragile site and renal carcinoma-associated t(3;8)breakpoint, is abnormal in digestive tract cancers. Cell 1996; 84: 587–597.

2. Barnes LD, Garrison PN, Sipirashvili Z, et al. Fhit, a putative tumoursuppressor in humans, is a dinucleoside 5*,5+*-P1,P3-triphosphate hydrolase.Biochemistry 1996; 35: 11 529–11 535.

Copyright ? 1999 John Wiley & Sons, Ltd.

3. Druck T, Hadaczek P, Fu TB, et al. Structure and expression of the humanFHIT gene in normal and tumour cells. Cancer Res 1997; 57: 504–512.

4. Sozzi G, Veronese ML, Negrini M, et al. The FHIT gene 3p14.2 is abnormalin lung cancer. Cell 1996; 85: 17–26.

5. Marchetti A, Pellegrini S, Bertacca G, et al. FHIT and p53 gene abnormali-ties in bronchioloalveolar carcinomas. Correlations with clinicopathologicaldata and k-ras mutations. J Pathol 1998; 184: 240–246.

6. Sozzi G, Sard L, De Gregorio L, et al. Association between cigarettesmoking and FHIT gene alterations in lung cancer. Cancer Res 1997; 57:2121–2123.

7. Negrini M, Monaco C, Vorechovsky I, et al. The FHIT gene at 3p14.2 isabnormal in breast carcinomas. Cancer Res 1996; 56: 3173–3179.

8. Ahmadian M, Wistuba II, Fong KM, et al. Analysis of the FHIT gene andFRA3B region in sporadic breast cancer, preneoplastic lesions, and familialbreast cancer probands. Cancer Res 1997; 57: 3664–3668.

9. Hayashi S, Tanimoto K, Hajiro-Nakanishi K, et al. Abnormal FHITtranscripts in human breast carcinomas: a clinicopathological and epi-demiological analysis of 61 Japanese cases. Cancer Res 1997; 57: 1981–1985.

10. Man S, Ellis IO, Sibbering M, Blamey RW, Brook JD. High levels of alleleloss at the FHIT and ATM genes in non-comedo ductal carcinoma in situand grade I tubular invasive breast cancers. Cancer Res 1996; 56:5484–5489.

11. Panagopoulos I, Pandis N, Thelin S, et al. The FHIT and PTPRG genes aredeleted in benign proliferative breast disease associated with familial breastcancer and cytogenetic rearrangements of chromosome band 3p14. CancerRes 1996; 56: 4871–4875.

12. Virgilio L, Shuster M, Gollin SM, et al. FHIT gene alterations in head andneck squamous cell carcinomas. Proc Natl Acad Sci U S A 1996; 93:9770–9775.

13. Mao L, Fan YH, Lotan R, Hong WK. Frequent abnormalities of FHIT, acandidate tumour suppressor gene, in head and neck cancer cell lines.Cancer Res 1996; 56: 5128–5131.

14. Kastury K, Baffa R, Druck T, et al. Potential gastrointestinal tumoursuppressor locus at the 3p14.2 FRA3B site identified by homozygousdeletions in tumour cell lines. Cancer Res 1996; 56: 978–983.

15. Shridhar R, Shridhar V, Wang X, et al. Frequent breakpoints in the 3p14.2fragile site, FRA3B, in pancreatic tumours. Cancer Res 1996; 56: 4347–4350.

16. Gemma A, Hagiwara K, Ke Y, et al. FHIT mutations in human primarygastric cancer. Cancer Res 1997; 57: 1435–1437.

17. Tamura G, Sakata K, Nishizuka S, et al. Analysis of the fragile histidinetriad gene in primary gastric carcinomas and gastric carcinoma cell lines.Genes Chromosomes Cancer 1997; 20: 98–102.

18. Zou TT, Lei J, Shi YQ, et al. FHIT gene alterations in esophageal cancerand ulcerative colitis (UC). Oncogene 1997; 15: 101–105.

19. Lisitsyn NA, Lisitsina NM, Dalbagni G, et al. Comparative genomicanalysis of tumours: detection of DNA losses and amplification. Proc NatlAcad Sci U S A 1995; 92: 151–155.

20. Thiagalingam S, Lisitsyn NA, Hamaguchi M, et al. Evaluation of the FHITgene in colorectal cancers. Cancer Res 1996; 56: 2936–2939.

21. Hendricks DT, Taylor R, Reed M, Birrer MJ. FHIT expression in humanovarian, endometrial, and cervical cancer cell lines. Cancer Res 1997; 57:2112–2115.

22. Greenspan DL, Connolly DC, Wu R, et al. Loss of FHIT expression incervical carcinoma cell lines and primary tumours. Cancer Res 1997; 57:4692–4698.

23. Frank S, Muller J, Plaschke J, et al. The putative tumour suppressor geneFHIT at 3p14.2 is rarely affected by loss of heterozygosity in primary humanbrain tumours. Cancer Res 1997; 57: 2638–2641.

24. International Union Against Cancer. TNM Classification of MalignantTumours, 4th edn. Berlin: Springer-Verlag, 1987; 43–46.

25. Oota K, Sobin LH. Histological Typing of Gastric. and OesophagealTumours. International Histological Classification of Tumours, 18. Geneva:WHO, 1990.

26. Lauren P. The two histological main types of gastric carcinomas: diffuse andso-called intestinal-type carcinoma. Acta Pathol Microbiol Immunol Scand1965; 64: 31–49.

27. Gabbert HE, Meier S, Gerharz CD, Hommel G. Incidence and prognosticsignificance of vascular invasion in 529 gastric cancer patients. Int J Cancer1991; 49: 203–207.

28. Kaplan EL, Meier P. Nonparametric estimation from incompleteobservations. J Am Stat Assoc 1958; 53: 457–481.

29. Kalbfleisch JD, Prentice RL. The Statistical Analysis of Failure DataNew York: Wiley, 1980.

J. Pathol. 188: 378–381 (1999)