Similar to other cancers, the carcinogenesis of hepatocellular carcinoma (HCC) is a multi-factor, multi-step, multi-genes and complicated process resulting from the accumulation of sequential genetic and epigenetic alterations. Epidemicly, the main causes of HCC are hepatitis B and C virus infection, dietary exposure to aflatoxin B1 and high-level alcohol consumption. Prolonged exposure to these risk factors is thought to cause an accumulation of chromosomal aberrations and altered gene expression, and eventually results in hepatocarcinogenesis [
17‐
19]. In China, more than 80 % of HCCs develop in patients with chronic infections with HBV. Histopathologically, precancerous lesions of HCC, such as DF, including SCC, DN, and NAH described by Su et al. [
3] are often found in cirrhotic liver tissue, and develop into early HCC, which corresponds to
in situ or microinvasive carcinoma, then develop into progressive HCC through the stage of “nodule-in-nodule”-type HCC. Geneticly, it is considered that LOH regions found in a significant portion of tumors are thought to embody tumor suppressor genes (TSGs). The delineation of such genetic alterations that occur in precancerous lesions and/or early hepatocellular carcinoma (HCC) may be important for monitoring and preventing the occurrence of HCC. However, only few of LOH assays have been reported about precancerous lesions and early HCC [
15,
16,
20]. Our previous studies demonstrated that part of DN without SCC and all the DN with SCC in liver cirrhosis tissue were monoclonal hyperplasia, and neoplastic lesion. Moreover, we revealed that there were some changes in DNA copy number in four chromosomal regions in one DN with SCC applying array-CGH [
9]. These above results showed that some genetic alteration have occurred in some precancerous lesions of HCC.
Some studies have reported that allelic loss of 4q, 8p and 16q are the most frequent chromosomal alteration in various human cancers. In particular, the loss of 8p23.1-22 is an important event in the initiation or promotion of HCC [
21,
22]. We found that there was high frequent loss of chromosomal 4q, 8p and 16q according to the result of Affymetrix SNP6.0 assay. Thus, we selected 28 microsatellite markers at some genes spanning chromosomal band 4q, 8p and 16q to further elucidate the precise location of putative TSGs that might potentially be involved in the tumorigenesis of HCC. The results showed that LOH frequencies at D8S262 for HCC were found to be 72.7 %, which indicated the gene neighboring to D8S262 might be a putative TSG related to the hepatocarcinogenesis. Moreover, many evidences have confirmed this point in many cancers at various levels, including DNA and RNA level [
23]. In addition, some studies have demonstrated that LOH is detected on chromosome 8p21.3-p22 in DNs and HCC, and the frequency of LOH is 40.9 % and 42.1 %, respectively [
19]. These results suggested that at least one putative tumor suppressor gene involved in the development of HCC might be located on 8p21.3-p22. Therefore, this gene might be related to an early genetic event of hepatocarcinogenesis [
20]. We examined LOH frequency at D8S262 for DN in order to investigate further the relationship between DN and HCC. The results demonstrated that LOH frequency at this locus for DN was 51.2 %, which was higher than 40.9 %. This indicated that the event of LOH had occurred in precancerous lesions of HCC, and the gene neighboring to D8S262 might involved in the occurrence and progression of HCC. Simultaneously, it confirmed that DN was a hepatic precancerous lesion, which was coincided in our previous conclusions [
9]. The high incidence of LOH observed at an early stage of tumor development was speculated that candidate TSGs located in this region may play an important role in early HCC. CSMD1 located at the neighbor of D8S262. It encodes multiple mRNA transcripts with the largest being 14.3 Kb long. The gene spans over 2 Mb of genomic DNA and contains 71 exons, which encode a 3565 amino acid protein consisting of 14 CUB domains and 28 SUSHI domains. It is a candidate tumor suppressor gene that maps to chromosome 8p23, a region deleted in many tumor types, and has homologies to proteins implicated in carcinogenesis. Moreover, many studies have indicated CSMD1 could be a tumor suppressor gene [
23‐
29]. Thus, according to the above results, we concluded that CSMD1 might be a TSG, and observed the expression of CSMD1 in HCC, DN and the surrounding liver tissues by immunohistochemical staining methods. The results showed that the positive rate of CSMD1 increased in turns in them. This indicated that CSMD1 might be a TSG related to the occurrence of early HCC. In addition, RT-PCR results demonstrated that the expression levels of CSMD1 were down-regulated in HCC cell lines compared with human normal hypertocyte cells HL-7702. Of course, it will need to be confirmed by a great deal of studies. Interestingly, Mohamed et al. [
23] studied the expression pattern of the CSMD1 protein in invasive ductal carcinoma (IDC), and found that reduction of CSMD1 expression was significantly associated with high tumor grade. Their results supported the idea that CSMD1 was a tumor suppressor gene. Midorikawa et al. [
28] found that homozygous deletions frequently on 8p23.2, and mRNA expression of the extremely large gene CSMD1 within this region was decreased in overt HCC, suggesting that CSMD1 plays a pivotal role in liver cancer progression. Our results showed that the expression of CSMD1 in the surrounding normal liver tissues was obviously higher than that in HCC, and reached a statistic significance. Moreover, the level of CSMD1 expression was higher in the well-differentiated HCC than that in the moderate and poor differentiated HCC though there was not a significant difference. Thus, we concluded that CSMD1 might a TSG according to our results. Lu et al. [
21,
22] also found that the LOH at 8p23.2-21 was significantly more frequent in the metastatic than corresponding primary tumor lesions in individual HCC cases, and the significant difference between them suggests that deletion of 8p23.2-22 may be not only an early event in the initiation of HCC but also involved in subsequent tumor aggressiveness, especially in the metastatic progression of HCC. Moreover, they considered that some unknown genes located adjacent to the markers D8S262, D8S1819 to D8S1109 and D8S261 play an important role in HCC. According to the above results, we found the conclusions were different for the role of CSMD1 in HCC. This may be related to the difference of people population. However, we also concluded at least one critical TSG lie at the restricted minimal regions at 8p23.2-22, and notably, CSMD1 may be one of them. At present, we considered it was a TSG related to early HCC. Of course, we need identify the gene by more studies.