Cancer and the FRA3B/FHIT fragile locus: it's a HIT

The FHIT gene encompassing the most active common human chromosomal fragile region, FRA3B, was discovered in 1996 and proposed as a tumour suppressor gene for important human cancers. Seven years and more than 350 reports later, early questions concerning its tumour suppressor role have been answered. Recent studies on the role of Fhit loss in major types of human cancers report association with high proliferative and low apoptotic indices, node positivity, loss of mismatch repair protein, likelihood of progression and reduced survival.

Soon after discovery, it was shown that the FRA3B/FHIT locus frequently exhibits deletions in preneoplasias and cancers, Fhit protein expression is lost or reduced in the majority of human cancers, and the orthologous mouse Fhit locus is fragile and exquisitely sensitive to carcinogen damage (Huebner et al, 1998;Huebner and Croce, 2001).
In the last 3 years, more than 180 papers featuring Fhit have appeared, as have publications on possible cancer association of genes at other common fragile sites, and Fhit has been shown unequivocally to act as a tumour suppressor. Fhit knockout mice show increased susceptibility to spontaneous and induced tumours, and oral gene therapy using adeno and adenoassociated FHIT viruses prevents and reverses induced forestomach tumours in Fhit þ /À mice Dumon et al, 2001).
So what does this mean for diagnosis, prognosis and therapy of human cancers? Here we review the types of cancers in which Fhit loss is associated with specific clinical features, and consider the importance of further investigation of consequences of Fhit loss in these cancers ( Figure 1).

UPPER GASTROINTESTINAL TRACT
Oral, oesophageal and gastric cancers are common in some Asian countries and parts of Africa, and environmental agents, such as alcohol and tobacco use, as well as certain dietary habits, are suspected to play important roles.
In an early study using RNA extracted from oral carcinomas in South Africa, FHIT transcription was examined by reverse transcription of RNA and PCR amplification of the cDNA product (RT -PCR) using primers within the FHIT open reading frame. In a more recent study, these investigators (van Heerden et al, 2001) compared the frequency of detection of FHIT aberrant expression using the RT -PCR assay for mRNA aberration and analysis of level of Fhit protein expression by immunohistochemical analysis in tissue sections using several polyclonal anti-Fhit sera. The findings were that 71% of the oral squamous cell carcinomas (OSCCs) showed reduced or absent Fhit protein and half of the cases with reduced Fhit showed aberrant RT -PCR products. Thus, immunohistochemical analysis of Fhit expression is more sensitive for the detection of alterations in expression (see Table 1 for a summary). Tanimoto et al (2000) had also studied OSCCs, as well as premalignant lesions for aberrant FHIT RT -PCR products. Evidence of aberrant transcripts was found in 53% of OSCCs and in two of seven premalignant lesions, one of which was from a patient who developed OSCC during follow-up. In a study of OSCCs associated with betel and/or tobacco use (Chang et al, 2002), 28% of samples exhibited FHIT promoter methylation, and 36% of cancers and 50% of premalignant lesions showed aberrant RT -PCR products, suggesting that FHIT aberration could be an early event in oral carcinogenesis. Reduced Fhit expression was observed in 41% of OSCCs. To determine the role of FHIT in tongue cancers, Fhit expression was studied by immunohistochemistry in sections from 41 stage II -IV squamous cell tongue carcinomas (Lee JI et al, 2001). Lost or decreased Fhit expression was demonstrated in 68% of tumours, especially in poorly differentiated areas. Patients whose tumours showed low or no Fhit had significantly shorter disease-free survival times than those with high Fhit-expressing tumours. Thus, loss of Fhit could be an independent prognostic indicator of clinical outcome for tongue cancer patients. Consistently with these findings, Rosin et al (2002) have shown that detection of allelic loss at the FHIT locus and/or 9p21 is a simple test for predicting a second oral malignancy at previously treated oral cancer sites.
Similarly, frequent LOH at 3p has been observed in head and neck squamous cell carcinoma (HNSCC). Mineta et al (2003) examined 57 HNSCCs by immunohistochemistry, Western blot and RT -PCR amplification for alterations in Fhit expression and looked for association with clinicopathologic features. Low Fhit expression was observed in 53% of cancers and correlated with Ki-67 expression, an indicator of aggressive proliferation.
Moving down to the oesophagus, FHIT LOH and Fhit expression were studied in precarcinomatous lesions and carcinoma. Primary tumours (76%) showed LOH encompassing FHIT and 70% were   (Mori et al, 2000). Shimada et al evaluted the clinical impact of FHIT gene alterations in 149 ESCs by immunohistochemical analysis, and examined correlation with smoking history. Normal Fhit expression was observed in only 22% of cases, reduced expression in 45% and absence of Fhit in 33%. Fhit was also markedly reduced in muscle invasive tumours. This investigation did not find an association of Fhit loss with prognosis or smoking history. In a somewhat smaller study that included CIS and dysplasia in sections of 75 ESCs, Fhit protein was reduced, relative to adjacent normal mucosa, in 89% of invasive ESCs, 68% of CIS lesions and 43.5% of dysplastic lesions, so that Fhit loss was associated with progressive 'increases in severity of histopathological changes' (Kitamura et al, 2001).
We believe that these findings concerning oral and oesphageal cancers are very important. The treatment of these cancers can be disfiguring and debilitating and recurrence is common. These cancers are highly accessible for topical treatment using gene therapy approaches to prevention and may be candidates for future clinical trials, possibly using viral FHIT delivery.
The results of a number of recent reports of FHIT alterations in gastric cancer are also briefly summarised in Table 1. To determine if FHIT alterations are frequent in gastric carcinomas, Lee SH et al (2001) examined FHIT LOH, aberrant RT -PCR products and Fhit protein expression in 35 gastric adenocarcinomas. Aberrant transcripts were detected in 57% and Fhit protein reduction in 63%. Caselli et al (2001) studied preneoplastic lesions in histological samples of patients who developed gastric cancer within 2 years and did not observe reduction or loss of Fhit expression, whereas reduced or absent expression was observed in 61.5% of the cancers; complete Fhit loss was observed only in areas of low differentiation. Huiping et al observed absence or reduction of Fhit expression in 78% of gastric cancers and found an association between abnormal Fhit expression and positive node status. In summary, Fhit loss is frequent in gastric cancer, may not be an early event and its prognostic value has not yet been established.

COLON CANCER
When the FHIT locus was first discovered, we noted frequent homozygous deletions in colon cancer cell lines and aberrant RT -PCR products in primary colorectal cancers (Ohta et al, 1996). Since then a number of investigators have examined the status of the FHIT gene and protein in colon cancer (see Table 1 for a summary). Hao et al (2000) examined aberrant crypt foci (ACF), adenomas, primary colorectal carcinomas (CRCs) and metastatic lesions for Fhit protein expression by immunohistochemistry. In all, 44% of carcinomas showed marked loss or absence of expression and the fraction of tumours with reduced expression increased with decreasing differentiation and in tumours with metastases (62 vs 38% in tumours without metastases); 12 out of 13 metastatic lesions showed reduced expression. Only a small fraction of ACFs and adenomas showed reduced Fhit, but reduced expression was strongly associated with degree of dysplasia. These authors suggested that Fhit plays a role in development and progression from the premalignant stage through metastasis.
Germline mutations in mismatch repair genes, usually MLH1 and MSH2, cause hereditary nonpolyposis colon cancer (HNPCC). We had been interested in a possible connection between defective mismatch repair and loss of Fhit expression because Fhit knockout mice develop sebaceous tumours of the skin , as do HNPCC carriers with Muir -Torre syndrome, involving sebaceous tumours of the skin combined with colon tumours. We also know that Msh2-deficient mouse cell lines show increased chromosome fragility (Turner et al, 2002), suggesting that intact mismatch repair systems are important for stability of fragile loci. Consistently with this idea, several reports of increased Fhit loss in  Cancer and the FRA3B/FHIT fragile locus K Huebner and CM Croce mismatch repair-deficient colon cancers have appeared. Mori et al (2001), in an analysis of 62 CRC cases by immunohistochemistry for Fhit and Msh2 protein, found that Fhit loss correlated significantly with progression of carcinoma, as well as with lymph node metastasis, and loss of Msh2 correlated with loss of Fhit. Loss of Fhit occurred in 50% of sporadic CRCs and was more frequent in advanced cancers. It was concluded that mismatch repair protein may be important in maintaining the integrity of the common fragile locus within the FHIT gene. Similarly, Andachi et al (2002) reported that reduced Fhit expression is associated with mismatch repair deficiency in advanced colorectal carcinoma.
Interestingly, in a study of biopsies of periocular sebaceous gland carcinomas, of patients with Muir -Torre syndrome, Holbach et al (2002) determined that Fhit was detectable in the one sebaceous gland carcinoma with microsatellite instability but not in the five sebaceous carcinomas without microsatellite instability. These authors made the intriguing suggestion that loss of either Fhit or mismatch repair could contribute to the development of sebaceous gland carcinoma in MTS.
Finally, in a study employing computerised image analysis to quantitatively evaluate Fhit expression and apoptotic role in CRCs, Mady and Melhem (2002) concluded that absence or reduction of Fhit plays a role in the development of B23% of CRCs and was directly correlated with the occurrence of distant metastases and worse prognosis. Also, overexpression of Fhit was directly proportional to the apoptotic rate.
Thus, the results of each study were consistent with the conclusion that loss of Fhit protein expression, through damage to the FRA3B fragile locus, has an important role in the development of a significant fraction of colon cancers.

CERVICAL CANCER
Cervical cancer was the first to be assessed for Fhit protein expression by immunohistochemistry and Fhit loss or reduction was observed in B70% of cervical cancers, a loss that correlated with the detection of aberrant FHIT RT -PCR amplification products. In a follow-up study that included preinvasive lesions, Connolly et al (2000) examined Fhit expression in 95 invasive cervical carcinomas, 33 high-grade squamous intraepithelial lesions (HSILs) associated with concurrent cancer, 38 HSILs without associated cancer and 24 low-grade squamous intraepithelial (LSILs) lesions. Normal and LSIL samples showed moderate to strong cytoplasmic staining while Fhit staining was reduced or absent in 71% of invasive cancers, 52% of HSILs associated with invasive cancer and only 21% of HSILs without associated cancer. The conclusion was that loss of Fhit in HSILs could serve as a marker of high-grade preinvasive lesions that are likely to progress to invasive carcinoma.
In a study of 59 stage II -III tumours, absent or reduced Fhit protein was observed in 66% (Krivak et al, 2001) and multivariate analysis showed that reduced Fhit expression was a poor prognostic factor. The 3-year survival rate for patients with Fhitpositive tumours was 74 vs 37% for those with low Fhit-expressing tumours. Guo et al (2001) analysed intratumoural heterogeneity of cervical cancers by studying 3p deletions and X-chromosome inactivation patterns in multiple microdissected samples from individual cancers. Allelic losses were frequently detected at 3p14.2 (FHIT), 3p21.3 -21.2 and 3p24.2 markers and had occurred in CIN lesions synchronous with invasive lesions. Although the study involved normal and lesional DNAs from only 14 cervical cancers, the authors concluded that the 'findings suggest essential roles of genes on these 3p loci, particularly the FHIT gene, in participating in clonal selection and early development of cervical cancer'. Butler et al (2002) also studied CIN and microinvasive carcinoma (MICA) for expression of Fhit protein by immuno-histochemistry and looked for association of Fhit loss with clinical parameters, including high-risk HPV infection; 50% of CIN3 and 78% of MICA lesions showed reduction or absence of Fhit protein, while CIN1 lesions showed moderate to strong Fhit expression. A significant association was observed between loss of Fhit expression and HPV16 infection in the combined CIN and MICA lesions. A number of reports have looked for a correlation of loss of Fhit expression with presence of high-risk HPV genomes; some have found such a correlation and others have not (see Table 1 for a summary). Thus, this issue will need further, larger studies for resolution.

LUNG CANCER
Owing to the connection between carcinogen exposure and Fhit loss, Fhit involvement in lung cancer was studied rather extensively early after discovery of the FHIT gene (reviewed in Huebner and Croce, 2002). In the last 3 years, there have been further reports on Fhit expression in lung cancer. Geradts et al (2000) aimed to correlate loss of Fhit expression with molecular genetic and clinical parameters in sections of 99 NSCLCs; 53% of tumours lacked Fhit staining, a lack that correlated with LOH at the FHIT locus. Fhit loss was as frequent as abnormalities of expression of p53, RB and p16 and occurred independently of most clinincal parameters and molecular abnormalities. Pavelic et al (2001) also examined the status of the FHIT gene in lung cancer and HNSCCs and compared it to expression of p21, frequency of apoptosis and proliferation. Most malignant lung and HNSCC lesions showed aberrant FHIT expression, reduced or absent p21 and increased cell proliferation. Fhit negativity tended to correlate with a worse prognosis (22.46 months median survival vs 36.04 months for Fhit-positive cases) and the trend was significant for HNSCCs (30.86 months median survival vs 64.04 months for Fhitpositive cases, Po0.05). Thus, aberrant Fhit might be a prognostic marker for lung cancer and HNSCC. Garinis et al (2001) studied 67 NSCLCs and observed FHIT LOH in 64%, for both squamous and adenocarcinomas. Allelic imbalance at FHIT was 71% in stage I cancers, showing early involvement. There was no association with kinetic parameters or ploidy of tumours, but concurrent loss of Fhit and overexpression of p53 was observed in 39%. Authors suggested that FHIT allele losses could be the outcome of tobacco-induced mutagenesis. Pylkkanen et al (2002) aimed to determine whether absent or reduced Fhit or FHIT allele loss was associated with exposure to lung carcinogens. Reduced Fhit expression was observed in 62% of cases and was common in asbestos-exposed (67%) and nonexposed cases (59%). FHIT LOH was increased in advanced disease and in poorly differentiated tumours, supporting the significance of FHIT inactivation in lung cancer development.

BREAST CANCER
The cytogenetics of breast cancer has recently been reviewed for 322 karyotypically abnormal samples from 256 patients (Teixeira et al, 2002). 'Interstitial deletions of the short arm of chromosome 3, with 3p14 as a minimally common deleted region, were described as a recurrent change in breast carcinoma by Pandis et aly. and have turned out to be the single most common structural rearrangement'y. Also, it occurs usually as the sole clonal abnormality and in benign proliferative disease of the breast. Molecular studies have confirmed loss of 3p in breast cancers, with loss at 3p13 -14 having been reported in more than 40% of breast carcinomas (Teixeira et al, 2002). 'A candidate tumour suppressor gene in this region is FHIT and some studies have shown that this gene is involved in breast cancery'.
Thus, although breast cancer cytogenetics is notoriously complex, loss of 3p material, including the FHIT region, is an early event in a fraction of breast cancers, and homozygous loss of at least a portion of the FHIT gene was observed in a case of bilaterial breast cancer, as were aberrant FHIT RT -PCR products and reduced Fhit expression in several series of sporadic breast cancers (reviewed in Huebner et al, 1998).
More recently, Kollias et al (2000) tested for concordant changes in left and right breast cancers of young women with bilateral cancer. Microsatellite markers were used to test for LOH at candidate genes TP53, BRCA1, BRCA2, ATM and FHIT. Four cases showed concordant loss of BRCA1 alleles in left and right cancers, four for BRCA2, seven for ATM and four cases for FHIT, suggesting possible roles for these tumour suppressor genes.
Several early studies of alterations at the FHIT locus in breast cancer reported reduced expression of Fhit in B40 -60% of mammary carcinomas and an elevated frequency of loss in BRCA2-linked breast carcinomas (Huebner and Croce, 2001). The higher frequency of alteration of FRA3B and reduced expression of Fhit in BRCA2-linked cancers was consistent with the idea that loss of BRCA2 function affects stability of the FRA3B/ FHIT locus. This theme was pursued in a study of BRCA1-linked breast cancers (Turner et al, 2002); breast tumours with deleterious BRCA1 mutations were analysed for Fhit expression by immunohistochemistry. Loss of Fhit expression was significantly more frequent in the BRCA1 cancers compared with sporadic breast tumours (9% Fhit positive vs 68% Fhit positive), suggesting that the BRCA1 pathway is also important in protecting the FRA3B/FHIT locus from damage. To further investigate the relation between repair gene deficiencies and induction of fragile sites in vitro, the frequency of aphidicolin-induced chromosome gaps and breaks was analysed in PMS2-, BRCA1-, MSH2-, MLH1-, FHIT-and TP53-deficient cell lines. Each repair-deficient cell line showed elevated chromosome gaps and breaks, consistent with the proposal that proteins involved in repair are important in maintaining the integrity of common fragile regions (Turner et al, 2002). Correspondingly, genes at common fragile sites may sustain elevated levels of DNA damage in cells with deficient DNA repair proteins such as those mutated in several familial cancer syndromes. Ingvarsson et al (2001) have examined the relation between FHIT LOH and breast tumour progression. FHIT gene markers were typed in 239 breast tumours and paired normal tissue, and results assessed relative to clinicopathologic factors and LOH at other regions. This study found that FHIT LOH was associated with oestrogen and progesterone negativity, high S-phase fraction, reduced patient survival and LOH at other chromosome regions. Perhaps most interestingly, FHIT LOH resulted in a 60% increased relative risk of death. The overall conclusion was that FHIT LOH results in growth advantage of breast tumour cells, is associated with an unstable genome and may be of prognostic value.
Similarly, Yang et al (2001), in a study of more than 160 breast cancers occurring in Asian women, found reduction of Fhit expression in 42%, and compared expression levels with clinicopathological profiles and expression of other biological markers. Reduced Fhit expression was significantly associated with histological grade, high tumour proliferation, negative ER status and p53 overexpression. Patients with tumours with loss of Fhit expression tended to have poor survival. Results of these recent breast cancer studies are summarised in Table 1.

CONCLUSION AND PERSPECTIVE
The data summarised indicate that the tumour suppressor gene, FHIT, is altered in many human tumours, particularly in those caused by environmental carcinogens, such as those present in tobacco smoke. In many of these tumours, particularly in those induced by tobacco or other environmental carcinogens, alterations of FHIT occur very early during the multistep process of carcinogenesis.
We have also shown that Fhit-negative cancer cells are very sensitive to the expression of FHIT; for example, infection with FHIT recombinant viruses (Huebner and Croce, 2001) can cause regression and prevention of tumours in experimental animals. Thus, it is logical to predict the development of a gene therapy approach for the treatment and prevention of Fhit-negative human cancers.