Prevalence of Fragile Histidine Triad Expression in Tumors from Saudi Arabia: A Tissue Microarray Analysis

Cancer is a genetic disease resulting from multiple, sequential genetic changes affecting oncogenes, tumor suppressor genes, and modifiers. Because of this multistep process, most human malignancies show various degrees of genetic heterogeneity even if they originate from single cells. Thus, cancer cells of the same clonal tumor mass may respond differently to chemotherapy or radiation therapy. Most of the human leukemias and lymphomas carry consistent chromosomal rearrangements, predominantly chromosomal translocations or inversions that activate specific oncogenes (1, 2) or cause loss of function of specific tumor suppressor genes (3), thereby initiating the process of malignant transformation. However, it is not known what the initiating events are for some of the most common human malignancies, the malignant epithelial tumors such as lung, breast, and prostate cancer.

Many tumor-suppressor genes, such as rb, wt1, p53, nf1, apc, smad, and pten, have been identified as tumor suppressor genes through loss of heterozygosity for polymorphic markers. Deletions in the short arm of chromosome 3 (3p loss of heterozygosity) are observed in various human cancers. At least three genes associated with the genesis of human cancer have been positioned on the short arm of chromosome 3p (4). Alterations of the von Hippel-Lindau (vhl) gene at 3p25 are frequently observed in renal cell carcinoma and pheochromocytoma (5). The mlh1 gene, one of the mismatch repair genes which, when defective, cause hereditary non–polypoid colon carcinoma, exists at 3p21.3 (6). The telomerase repressor gene, which suppresses the expression of the telomerase gene, also exists on chromosome 3p (7).

The fragile histidine triad (FHIT) gene is a candidate tumor suppressor gene located at chromosome 3p14.2, spanning the FRA3B common fragile site (8). The FHIT protein is homologous to Ap4A hydrolase from the yeast, Schizosaccharomyces pombe, and it also exhibits Ap3A activity in enzymatic assays (8, 9). FHIT protein is presumed to have tumor suppressor function independent of its hydrolase activity (8-10). However, the mechanisms through which FHIT mediates its suppressor function are not well established. Several investigators have shown that introduction of a wild-type FHIT gene suppresses tumorigenicity, and the transfection of FHIT in FHIT-deficient human cancer cells seem to induce apoptosis and inhibit cell growth (10-12). These results suggest that the suppressor activity of FHIT could be related with apoptosis and with the alteration of cell cycle regulator factors. Frequent allelic losses and homozygous deletions (8, 9, 13-24), as well as the loss of heterozygosity in microsatellites located at the FHIT gene have been described at the FHIT locus in several human solid tumors arising from epithelial cells (13, 19, 20, 25-32). Studies comparing FHIT expression with the status of the FHIT gene have shown significant concordance in a variety of malignancies. Abnormalities in the FHIT gene and/or its expression have been identified in a variety of human cancer cell lines and tumor tissues including lung (33), breast (34-36), urinary bladder (37), head and neck (38), esophageal (39, 40), gastric (41), colorectal (42), renal (43), liver cancers (44), and diffuse large B cell lymphoma (45).

To determine the role of FHIT in the pathogenesis and progression of cancer, we investigated the incidence of loss of FHIT expression by immunohistochemistry in a series of tumors in a tissue microarray format. We also correlated FHIT expression with age, sex, histology diagnosis, grade, and stage of various tumors.

A total of 1,889 specimens from the archives of the King Faisal Specialist Hospital and Research Centre (Riyadh, Saudi Arabia) were assessed, including tumors from 15 different sites and 75 tumors and subtypes. A multitumor array block comprised of 578 specimens from 13 different sites and 9 other array blocks were constructed from tumors of the breast, colon, liver, lung, lymphomas, kidney, and meningiomas (Table 1).

TMA construction was done as described previously (46). Briefly, all the H&E slides were screened and nonnecrotic well-fixed tumor areas were mapped with an indelible marker pen. Using these mapped slides as a reference, 0.6 mm diameter punches were obtained from the donor blocks. The tissue microarrayer (Beecher Instruments, Woodland, WI) was used. A map of the recipient block was prepared with coordinates and a number for each sample to correctly identify the tumors. The punched-out tissue cores from the donor block were inserted in the recipient block. The cores were arranged 8 mm from the edges and the distance between two cores was 0.8 mm. The array blocks were incubated at 45°C for 10 minutes to improve adhesion between cores and paraffin of the recipient block. They were cut at room temperature with a standard microtome (Thermo Shandon, Cheshire, United Kingdom) and slides were prepared using tape sectioning system (Instrumedics, Inc., St. Louis, MO).

Sections of 5 μm from all the array blocks were cut, slides were deparaffinized in xylene and rehydrated in pure ethanol. Endogenous peroxidase was blocked using 3% hydrogen peroxidase in methanol for 10 minutes. Antigen retrieval was done by placing the sides in a citrate buffer (pH 6.0) and microwaving them for 5 minutes at 750 W and for 15 minutes at 250 W.

The sections were incubated for 90 minutes in 1:900 dilutions of polyclonal rabbit antibodies reacting against FHIT protein (ZR44; Zymed, San Francisco, CA). Bound antibody was detected with biotinylated link antibody (Dako, Carpinteria, CA) and horseradish peroxidase–labeled streptavidin (Dako). Color was developed in 3,3′-diaminobenzidine with H₂O₂ as substrate (Dako). The sections were then washed in running tap water, lightly counterstained with Gill's hematoxylin, dehydrated through ascending graded alcohols, cleared in xylene, and mounted in DPX. All normal epithelia (lung, liver, urinary bladder, and colon) showed strong cytoplasmic expression of FHIT and served as an internal positive control. Separate negative controls (renal glomeruli and lung cancer) were appropriately negative for FHIT protein. In addition, no staining was observed when primary antibody was replaced by normal rabbit serum IgG. Expression was scored on a three-tiered scale for both intensity (grade 1, absent/weak; grade 2, moderate; grade 3, strong) and extent (grade 1, percentage of positive cells is <10%; grade 2, 10-50%; grade 3, >50%). The intensity and extent scores were multiplied to give a composite score (1-9) for each tumor. A score of 0 was defined as absent or lost expression, scores of 1 to 3 were defined as markedly reduced FHIT expression, and scores of 4 to 9 were considered as normal expression (47-49). Because at least three different slides for each array constructed was stained with anti-FHIT antibody, the mean of all three FHIT scores was used for statistical analyses.

Statistical analysis was done using SAS version JMP IN 5.1 software (SAS Institute, Cary, NC), and all P values reported are two-tailed. Univariate analysis of categorical variables was conducted using contingency analysis and χ² tests. The surviving fraction was estimated using the Kaplan-Meier method. In the final model, all variables were considered statistically significant at P < 0.05. Univariate analysis of FHIT expression was done with age, sex, grade, stage, and survival in breast, colon, kidney, lung, liver cancers, and diffuse large B cell lymphomas.

In normal tissues, a uniform strong cytoplasmic staining of FHIT was seen in all epithelial cells of the skin, breast, colon, liver, lung, kidney, ovary, esophagus, urinary bladder, and stomach. Similar FHIT expression was seen in normal lymphoid cells as well as in the soft tissues.

FHIT expression was found to be reduced by 19% to 100% in the various tumor types studied (Table 2). Decreased or absent FHIT expression was seen in 46.3% of the breast carcinomas. No correlation was seen between FHIT expression and age, sex, histology diagnosis (H&E diagnosis), and survival pattern. Loss of FHIT expression was correlated with higher grade (P = 0.0187) and advanced stage (P = 0.030; Table 3). Decreased or absent FHIT expression was seen in 20.7% of the colorectal adenocarcinomas and a positive correlation was observed between loss of FHIT expression and advanced stage (Dukes) of colorectal cancers (P = 0.0392). No correlation was seen between FHIT expression and age, sex, H&E diagnosis, grade, and survival pattern (Table 4).

Renal cancers showed absent or reduced FHIT expression in 31% of the tumors, and there was a statistically significant inverse correlation between loss of FHIT expression and grade (P = 0.006) and stage (P = 0.023). No correlation was seen between FHIT expression and age, sex, H&E diagnosis, and survival pattern (Table 5). Decreased or absent FHIT expression was seen in 45% of the liver carcinomas and a positive correlation was observed between loss of FHIT expression and higher grade and lymph node metastasis. No correlation was seen between FHIT expression and age, sex, H&E diagnosis, and survival pattern (Table 6). FHIT expression was reduced or absent in 67.8% of the lung tumors. Tumors showing loss of FHIT expression were of a higher stage (P = 0.031), larger size (P = 0.046), and occurred more commonly in the older age group (P = 0.006). FHIT expression was also correlated with histology (P = 0.003). One hundred percent of the small cell carcinomas and 92% of squamous cell carcinomas showed a reduced or absent FHIT expression as compared with 54% of adenocarcinomas or 25% of bronchioalveolar carcinomas (Table 7).

Loss of FHIT expression was seen in 65.78% of the non–Hodgkin's diffuse large B cell lymphomas. No correlation was seen between FHIT expression and age, sex, stage, and survival patterns (Table 8).

Reduced or absent FHIT expression was seen in 64.7% of head and neck cancers, 19% of ovarian tumors, 23.5% of gastric tumors, 54% of urinary bladder tumors, 72.3% of skin tumors, 27.3% of the meningiomas, and 80% of soft tissue tumors. Reduced or absent FHIT expression was seen in 100% of the nasopharyngeal (only nine cases) as well as in all four cases of esophageal carcinomas (Fig. 1).

Chromosomal abnormalities, including homozygous deletions and loss of heterozygosity, are among the most common features of human tumors. The short arm of human chromosome 3, particularly the 3p14.2 region, is a major site of such rearrangements. The 3p14.2 region spans the most active common fragile site of the human genome (4), encompassing a familial kidney cancer–associated breakpoint (5), mismatch repair gene (6), and a telomerase repressor gene site (7).

The FHIT gene is a candidate tumor suppressor gene that was identified in this region by positional cloning (8). FHIT encompasses FRA 3B, the most common fragile site in the human genome (8). Subsequent studies have shown that FHIT is commonly the target of chromosomal aberrations involving the short arm of human chromosome 3 and is thereby inactivated in most of the common human malignant diseases, including cancers of the lung (13), esophagus (39, 40), stomach (41), breast (17, 18, 34-36), and kidney (22).

The FHIT gene and its protein product have been the focus of recent debate with regard to their potential role in tumorigenesis (50). A tumor suppressor role for FHIT has been postulated based on the ability of FHIT to eliminate or reduce the tumorigenicity of tumor cells in nude mice (51). Loss of gene expression associated with morphologic progression from normal, nonneoplastic epithelium, through stages of hyperplasia and carcinoma in situ to invasive carcinoma, has generally been accepted as evidence of the suppressor role of that gene (52, 53).

The way(s) in which FHIT functions as a tumor suppressor gene is/are unknown, but FHIT protein has a proapoptotic effect when restored to FHIT protein–deficient cell lines (54). The FHIT protein is a dinucleoside 50, 5000-P1, P3-triphosphate (Ap3A) hydrolase (55) that produces ADP and AMP, although the tumor suppressor effect seems to be more strongly linked to substrate binding than substrate hydrolysis (56). FHIT mRNA and protein expression is found in most human tissues and genetic alterations are found in many human carcinomas, including loss of heterozygosity and translocations. Point mutations seem to be less common. FHIT mRNA splice variants are common in carcinoma, but are also frequently found in nonneoplastic tissues from healthy individuals (57). The recent findings in the molecular biology of FHIT, with particular focus on the opportunities for treatment and prevention of cancer, have been previously described (58).

In an elegant and concise review, the types of cancers in which FHIT is associated with specific clinical features and the importance of further investigation of the consequences of FHIT loss in these cancers has been highlighted (59). The treatment of oral and esophageal cancers could be disfiguring and debilitating, but these sites are highly accessible to topical treatment using gene therapy approaches to prevention and might be candidates for future clinical trials using viral FHIT delivery (59). Recently, viral FHIT gene transfer successfully prevented and reversed carcinogen-induced epithelial tumor formation in the forestomachs of FHIT-deficient mice (60, 61).

Ishii et al. have lucidly illustrated and explained that FHIT may be altered at different levels in different types of cancer (62). In the same article, they have hypothesized a schema, illustrating tumor cell propagation from possible ancestral cancerous or tumor stem cells with self-renewal potential to daughter cells, in which FHIT may be altered at different stages of precancer. They divided tumors into three classes:

• Class A: In some tumors, FHIT is inactivated in a precursor cell, which results in loss of tumor suppressor function and leads to the expansion of tumor cells. In tumors of this category, aberration of the FHIT gene and FHIT protein reduction are associated with increased tumor proliferation, decreased apoptosis, and poorer survival of patients, as shown in lung and head and neck cancers (23, 26-28). The restoration of FHIT expression would be effective in the regression of such tumors, which are thus candidates for FHIT gene therapy, as supported by in vitro studies and animal experiments (60, 61).

• Class B: In tumors in which the biological behavior seems to be unrelated to FHIT status, other factors could drive tumor progression. Several studies showed that although FHIT is altered in an early stage of cervical carcinogenesis, association of FHIT loss with histopathologic grades or clinicopathologic variables is not necessarily observed (63, 64). This may be because other factors, such as human papilloma virus infection in cervical cancer, drive tumor progression.

• Class C: In some hematopoietic disorders, such as chronic myelogenous leukemia, tumor stem cells may exist far upstream from the FHIT-negative cells, as shown in case C. A relatively large study with a total of 195 Philadelphia chromosome–positive chronic myelogenous leukemias showed that lack of FHIT protein expression was detected in 4% of cases, and reduced FHIT expression was not associated with progression, response to therapy, or with prognosis in chronic myelogenous leukemia (65).

The TMA approach is optimally suited to identify those samples with frequent alterations of a specific gene (66, 67). They serve as an excellent tool to develop and compare immunohistochemical analysis. Hundreds of tumors can be immunostained under standardized conditions on one TMA slide. The small diameter of each arrayed tissue sample limits the comparison to a very small tissue area with a minimal likelihood of genetic, tissue processing, or immunostaining heterogeneity. Although the total number of abnormal cases detected by TMA may be low due to regional heterogeneity of immunostaining, the perfect standardization of staining more than compensates for it.

In our study, we analyzed FHIT by immunohistochemistry on three TMA blocks of breast, colon, lung, liver, and kidney tumors. The mean of the three FHIT scores obtained was taken as the final FHIT score for that specimen. Those specimens showing heterogeneous staining in the three array slides were reviewed again. Those specimens which showed one TMA slide with normal FHIT staining (score >3) and the other showing reduced (score <3) or absent (score = 0) were labeled as heterogeneous or discordant staining. These constituted a very small percentage (0-12.5%).

Reduced or absent FHIT expression was seen in only 46.5% of the breast cancers studied. Around 10% of the cases showed heterogeneous immunohistochemistry staining in the three TMA slides and were indicative of the incidence of heterogeneous FHIT staining. This is in contrast to earlier studies which have found reduced FHIT expression in 69% to 76% of the breast cancers (34-36). However, a study done in a Japanese population showed that the loss of FHIT expression was seen in 42.2% of the cases studied (68). FHIT alteration may not be an early event in breast carcinogenesis in Saudi Arabia, as the prevalence of loss of FHIT expression is much lower. Earlier and ongoing studies have found that the behavior of breast carcinomas from Saudi Arabia is different compared with Western countries (69). Further studies involving the analysis of the FHIT gene to observe allelic loss and abnormal transcripts are needed and are being done in our set-up to study and confirm these differences.

FHIT expression was reduced or absent in only 20.7% of the colorectal carcinomas. An earlier study showed loss or reduced FHIT expression in 44% of the colorectal cancers (42). No correlation was seen between FHIT expression and age, sex, H&E diagnosis, grade, tumor size, lymph node metastasis, and survival time. A positive correlation was observed between loss of FHIT expression and advanced stage (Dukes) of colorectal cancers (P = 0.0392). Loss of FHIT expression was seen in the earlier studies in the range of 23% to 50% using immunohistochemistry (70-72). Loss of FHIT expression was correlated with higher grade, Dukes stage, distant metastasis, and worse prognosis. Loss of mismatch repair protein was correlated with loss of FHIT, and these FHIT-deficient cases showed a significant correlation with progression of carcinoma, as well as lymph node metastasis (70, 71). These findings suggest that mismatch repair protein may be important in maintaining the integrity of the common fragile locus within the FHIT gene, and FHIT plays a role in a small but significant fraction of colorectal cancers.

Renal cancers showed absent or reduced FHIT expression in 31.3% of the tumors, and there was a statistically significant correlation between loss of FHIT expression and increasing grade (P = 0.006) and more advanced stage (P = 0.023). An earlier study found that loss of FHIT expression was significantly less pronounced in poorly differentiated renal call carcinoma or advanced tumor stage (43). In this study, all the clear cell carcinomas studied showed a reduced FHIT expression, which was not observed in our cases in which normal FHIT staining was seen in 69% of the cases. We observed a moderate to strong intensity of FHIT staining in the kidney tumors, and in some cases, this matched the staining intensity of normal kidney tubules which were used as controls. In another study done on collecting duct carcinomas, FHIT expression was reduced in only 3 of the total 11 cases (73). Hadaczek et al. described a correlation between reduced FHIT expression and 3p allelic loss in renal carcinomas (74). Although in our study, FHIT inactivation does not seem to be a common event in renal cell carcinomas, the involvement of the FHIT gene in the tumorigenesis of this tumor cannot be ruled out.

Reduced or absent FHIT expression was seen in 45% of the liver cancers in this cohort from Saudi Arabia, where exposure to hepatic carcinogens is low. Loss of FHIT expression was more commonly observed in poorly differentiated hepatic tumors (P = 0.027) and associated with lymph node metastasis (P = 0.0392). Reduced or absent expression was seen in 75% of the grade 3 hepatocellular carcinomas as compared with only 22% in the grade 1 tumors. A positive, although not statistically significant, correlation was also noted between reduced FHIT expression and advanced stage. There was no difference in the survival analyses between the two groups (P = 0.492).

An earlier study on a cohort from the U.S., where the exposure to hepatic carcinogens is low, has showed a reduced expression of FHIT in 15% of the hepatic tumors (44). Marked reduction or absence of FHIT protein by immunohistochemistry staining has been reported in 65% of the 83 hepatocellular carcinomas examined from China, where loss was associated with increasing tumor size and stage (75). FHIT inactivation is probably a later event associated with hepatic carcinogenesis and might play only a minor role in hepatic carcinogenesis in Saudi Arabia.

FHIT expression was reduced or absent in 67.6% of the lung tumors, which is in concordance with earlier studies (28, 33, 76, 77). Tumors showing loss of FHIT expression were of a higher stage (P = 0.031), larger size (P = 0.046), and occurred more commonly in the older age group (P = 0.006). FHIT expression was also correlated with histology (P < 0.006). One hundred percent of the small cell carcinomas and 92% of the squamous cell carcinomas showed a reduced or absent FHIT expression as compared with 54% of adenocarcinomas or 25% of bronchioalveolar carcinomas. There was a positive correlation between loss of FHIT expression and older age, which was, however, not statistically significant (P = 0.061). Similarly, there was no difference in the survival patterns (P = 0.199) between the two groups (normal versus reduced/absent FHIT expression). Extensive studies have been carried out due to the loss of FHIT gene following exposure to various carcinogens (76). Earlier studies have shown a correlation between loss of FHIT expression and a higher proliferation index and a lower apoptotic index; as well as an inverse correlation between loss of FHIT expression and patient survival (33). The same study stressed the important role of FHIT in carcinogenesis, especially squamous cell carcinomas, in association with smoking. In another study, there was no correlation between FHIT expression and a variety of clinical variables including survival and abnormal immunohistochemical expression of p53, rb, and p16 (77). The same authors concluded that loss of FHIT expression is an extremely common and independent genetic abnormality that occurs independently of the metastatic state and other molecular abnormalities.

Loss of FHIT expression was seen in 65.8% of the non–Hodgkin's diffuse large B cell lymphoma. No correlation was seen between FHIT expression and age, sex, stage, and survival patterns. There is only one earlier study of FHIT expression by immunohistochemistry in diffuse large B cell lymphoma. In that study, polyclonal rabbit IgG anti-FHIT antibody was used at a dilution of 1:200 at an incubation time of only 10 minutes (45). This study was done on 31 patients and showed that decreased FHIT expression indicates a significantly bad prognosis in diffuse large B cell lymphomas.

We now briefly discuss the prevalence of FHIT expression in some tumors which were arrayed in the multitumor array block. Complete loss or reduced FHIT expression was seen in 24% of the total 34 gastric carcinomas in our study. Loss of FHIT is possibly an early event in gastric carcinoma and has been seen in 49% of the 55 gastric adenocarcinomas studied (41).

Loss of FHIT expression was seen in 54% of the bladder tumors. An earlier study had shown a loss of FHIT expression in 61% of the bladder cancers as well as a significant correlation between reduced FHIT expression and advanced stage of the disease. As for other neoplasms caused by environmental carcinogens, FHIT inactivation could play an important role as a late event in the development of bladder tumors (37). Loss of FHIT expression was seen in 63% of the skin tumors. Reduced FHIT expression was seen in 73% of the basal cell carcinomas and 70% of the squamous cell carcinomas. This suggests an early loss of FHIT gene in the progression of skin cancers. To the best of our knowledge, this is the first such study using immunohistochemistry to study FHIT expression in skin cancers. An earlier study had found a high frequency of FHIT gene abnormalities in Merkel cell carcinomas (78). Abnormal FHIT transcripts were seen in 57% of the cases.

Other authors studied some non–melanoma skin cancers (basal cell carcinoma, squamous cell carcinoma, and actinic keratosis) and concluded that the FHIT gene is not a very common target in skin cancers (79).

The meningiomas also showed reduced or absent expression in 15% of the cases. Normal FHIT expression was seen in 72% of the cases. Staining heterogeneity was seen in 12.5% of the cases in the fibroblastic, meningothelial, and transitional cell subtypes. To the best of our knowledge, no previous studies have been done to see FHIT gene abnormalities in meningiomas.

Loss of FHIT expression was seen in 80% of the soft tissue sarcomas. Surprisingly, all 17 cases of dermatofibrosarcoma protuberans, a benign tumor, showed complete loss of FHIT expression. In an earlier study by reverse transcription-PCR analysis of the FHIT gene, normal and abnormal FHIT transcripts were found in 11 (69%) of 16 osteosarcomas, and in 3 (27%) of 11 Ewing sarcomas (80).

Loss of FHIT expression was seen in 19.04% of the ovarian tumors. Earlier studies have shown the absence of FHIT protein by immunohistochemistry in 34% of the ovarian carcinomas (81), and concluded that FHIT probably plays a role in a small proportion of ovarian cancers (82). FHIT expression was reduced or absent in nasopharyngeal carcinomas (all 9 cases), in esophageal carcinoma (all 4 cases), and 64.7% of the head and neck cancers studied (17 cases). In an earlier study, low FHIT expression was seen in 53% of the head and neck squamous cell carcinomas and correlated with Ki-67 expression (83). In an earlier study, primary esophageal tumor (76%) showed loss of heterozygosity encompassing FHIT, and 70% were negative for FHIT protein (84). In this study, tumors from patients who were heavy users of tobacco and alcohol showed significantly higher frequencies of loss of FHIT expression. Noncancerous squamous epithelia were mostly positive for FHIT, but five samples from heavy tobacco/alcohol users were FHIT-negative. In addition, most carcinomas in situ, 50% of severe and moderate dysplasias, and 33% of mild dysplasia were FHIT-negative, suggesting that FHIT loss is an early event in esophageal squamous cell carcinoma development (84).

The data summarized indicates that the tumor suppressor gene, FHIT, is altered in almost all human tumors, particularly those caused by environmental carcinogens. In some of these tumors, such as lung cancer, in which FHIT loss occurs early in the carcinogenesis pathway, there is a correlation between FHIT loss and the clinical variables. However, in some tumors, loss of FHIT expression is associated with other factors in some tumors and the role of FHIT is still to be elucidated. Finally, there is a subgroup of tumors in which FHIT loss is seen only in a small fraction of the tumors and FHIT probably plays a very minor role in carcinogenesis (Fig. 2). We have also highlighted that ethnic differences exist in the loss of FHIT expression in tumors from Saudi Arabia, especially in breast carcinomas and liver carcinomas.