This article is mentioned in:

Introduction

Selenium is a trace element for which no direct sign of its being essential in human nutrition was found until 1979. In the same year, the existence of a correlation between the low concentration of selenium in the geographical area of Keshan in China and the pathology, known as Keshan disease, was identified by a research group in this country (1). Selenium is stored in human tissues in varying amounts: 30% of tissue selenium is in the liver, 15% in the kidney, 30% in muscle, 10% in the plasma and the remaining 15% throughout the other organs. Selenium is found in trace quantities in a number of dietary agents such as dairy products, meat products, poultry, fish, fruits, vegetables and cereals, rendering it part of the food chain.

Selenium is considered to be a significant antioxidant, a precursor of the antioxidant enzyme known as ‘glutathione peroxidase’ (GPX), which protects cells from free radical damage in a number of neuronal and neuromuscular disorders, such as stroke and cerebrovascular disease, Alzheimer’s disease, Parkinson’s disease, familial amyotrophic lateral sclerosis and Duchenne muscular dystrophy. The proposed mechanisms mainly invoke the function of GPX as well as that of selenoprotein P (2). The evidence linking a lack of selenium with cancer is found in epidemiological and clinical studies. Low dietary selenium levels became an accurate way of predicting future cancer rates. A previous study showed that selenium supplementation led to a 50% reduction in cancer mortality (3). In general, the mean content of selenium in serum from patients with various types of cancer was lower than that of the control groups. Therefore, the correlation between decreased levels of selenium and increased DNA damage and increased oxidative stress further indicates the significance of this trace element (4).

A number of studies, including geographic, pre-clinical animal and prospective, as well as intervention studies, have shown selenium to be involved in a number of types of gastrointestinal cancer, such as esophageal, colorectal and prostate, as well as liver cancer (5–6), and have suggested its putative role in cancer, as well as progression, and, subsequently, metastasis prevention (7). A significant positive correlation between plasma selenium levels and liver cancer was also found. A number of epidemiological studies performed on humans showed significantly lower serum, plasma and liver selenium levels in patients with liver diseases, such as chronic hepatitis and cirrhosis, and different grades of hepatocellular injury, than those measured in the healthy control groups (8–13). Hepatocellular carcinoma (HCC) develops in the liver with severe impairment of cellular antioxidant systems, since, in patients with liver metastases from various types of cancer, cellular scavenger enzymes exhibited normal activities, despite low selenium concentrations (14).

In particular, HCC is a major health problem worldwide, being the fifth most common malignancy in males and the eighth in females and the third most common cause of cancer-related mortality in the world. The incidence of HCC is on the increase, with marked variations among geographic regions and racial and ethnic groups, relative to the exposure to documented environmental risk factors (15). In particular, Southern Italy has the highest rates of HCC in Europe (16). Previously, we evaluated the expression of the human selenium binding protein-1 (SELENBP1) in tissue samples of HCC patients. SELENBP1 is a protein that incorporates exogenously administered radioactive (75Se) sodium selenite in the liver in vivo (17–18). These studies provide evidence that this protein is down-regulated in the liver tissue of HCC patients and that its gradual loss is associated with an increased malignant grade (19).

In the present study, liver selenium levels in tissue samples of Italian HCC patients were measured by atomic absorption spectrometry in order to investigate whether there was a correlation between selenium and SELENBP1 concentrations. To the best of our knowledge, this is the first study that has examined this type of correlation in human liver tissue.

Materials and methods

Tissue sample

Paraffin-embedded specimens of liver tissue from 20 HCC patients (5 females, 15 males) were analyzed. The clinicopathological characteristics of these patients are shown in Table I. The study included 5 HCC patients with grade I tumors, 9 HCC patients with grade II and 6 HCC patients with grade III tumors. These patients had underlying hepatitis C virus (HCV)-related cirrhosis and their age ranged between 57 and 82 years. All 20 patients in this study provided informed consent. Liver tissue samples were obtained from patients recruited as part of an ongoing study to investigate associations between cytokine dynamics and liver injury progression. Following standard diagnostic procedures, biopsies were obtained from the liver and were collected for research purposes.

Table I Clinicopathological characteristics of HCC patients subdivided according to grading.a

Table I

Clinicopathological characteristics of HCC patients subdivided according to grading.a

Grading	Age		Gender		Tumor invasion			Tumor size (cm)
	>70	<70	M	F	T1	T2	T3	<4	>4
I	1	4	4	1	3	2	0	2	3
II	7	2	7	2	1	3	5	3	6
III	3	3	4	2	2	2	2	3	3

a The number of patients to which each parameter refers is reported.

Atomic absorption spectrometer studies

Nitric acid, 65%; xylene anhydrous, 99%; formaldehyde and ultra pure water were from Sigma-Aldrich (Steinheim, Germany). Hydrogen peroxide, 30% and selenium standard 1000 ppm were from Carlo Erba Reagents (Milan, Italy). Each liver tissue was fixed in formaldehyde and embedded in paraffin as part of routine histological processing. For this study, each paraffin-embedded tissue was dissected from the paraffin block to obtain 20-μm slices. The preparation of samples was performed according to a previous study (20). The wax was removed by washing (2×30 min) in xylene. The deparaffinized specimens were dried at 80°C for 24 h and digested with a mixture of HNO3 and H2O2 at a ratio of 1:2 (v/v) at a temperature of 100°C in a bath for 15–20 min.

The selenium concentration was determined by graphite furnace atomic absorption spectroscopy on a Varian SpectrAA 200 (Victoria, Australia) spectrometer with Zeeman background correction. The quantitative determinations were carried out by a calibration curve using selenium standard solutions (5–20 ppb). Digested samples were diluted with ultra pure water to yield the selenium concentration within the calibration range.

The furnace settings were as follows: for drying, ramp to 85°C (5 sec), ramp to 95°C (40 sec), and ramp to 120°C (10 sec); for ashing, ramp to 1000°C (5 sec) and hold at 1000°C (3 sec); for atomization, ramp to 2600°C (0.8 sec, read signal), hold at 2600°C (2 sec, read signal), and hold at 2600°C (2 sec, tube clean). The absorbance was determined at 196.0 nm, and the slit was 1.0 nm.

Data analysis

Selenium concentrations (expressed in μg/g) are reported as the arithmetic means ± standard deviation for each group of patients. The non-parametric Mann-Whitney U test was used to evaluate differences between selenium concentrations from patients with grades I-III HCC. The test was capable of distinguishing values where p<0.05 with one asterisk (*), with two asterisks (**) where values were p<0.01 and with three asterisks (***) where values were p<0.0001. The correlations between the selenium levels and clinical data or immunohistochemical scores related to SELENBP1 (19) were determined using the Pearson’s correlation coefficient. P<0.05 was considered to be statistically significant. The statistical program Prism 4 (GraphPad Software, San Diego, CA, USA) was employed.

The relationship degrees between selenium levels and patients were studied by means of the hierarchical agglo-merative clustering method. Cluster analyses were performed using the STATISTICA 8.0 statistical package (Statsoft Inc., Tulsa, OK, USA). In particular, the patients were assigned to clusters step-by-step. Following each step, clusters were joined to form new clusters. The course of this process of increasing the degree of clustering was exhibited in a dendrogram. At any distance from the start of the clustering process, the number of clusters formed at that point are shown in the graph (Fig. 1). The basis for a cluster analysis was formed by calculating the degree of similarity among the relevant variables of the different patients to be clustered. In this study, the Euclidean distance between patients was selected as the measure of similarity (21–22). Although this study included a relatively small number of patients, the approach of selecting these patients for high, medium and low malignant grade is a way in which to increase the statistical power.

Figure 1 Selenium concentrations (μg/g of liver) in HCC patients with grades I–III were plotted with box-and-whisker graphs. The boxes extend from the 25th to the 75th percentile, and the line in the middle is the median. The error bars extend down to the lowest value and up to the highest. **P<0.01.

Results

In a recent study, we found that the human SELENBP1 is also down-regulated in HCC (19). The present study correlates the amount of selenium in HCC patients with the liver content of this protein. The concentration of selenium in HCC patients subdivided into three groups based on malignant grades (I–III) has been evaluated by atomic absorption spectrometry (Fig. 1). The mean concentrations found in the liver were 0.823±0.083 μg/g of selenium for HCC patients with grade I, 0.638±0.066 μg/g for those with grade II and 0.505±0.076 μg/g for those with grade III. These values are lower than those found in the livers of healthy individuals (1.9±0.6 μg/g) (23). The results are also in agreement with scattered data reported in other studies (8–13).

We compared the selenium levels in three groups using the Mann-Whitney U test. The differences between the patients with grade I and II or with grade II and III are statistically significant with p=0.0017 and p=0.0018, respectively. This finding is also confirmed by the hierarchical clustering analysis of patients based on the selenium concentrations. The analysis provides evidence that the patients were placed into three groups, corresponding to the three different malignant grades (Fig. 2). Moreover, the data highlight that selenium concentrations decrease when the malignant grade increases with a correlation coefficient equal to −0.86 (Fig. 3A), whereas no correlation has been observed between selenium levels and patient characteristics such as age/gender, tumor invasion and tumor size. Finally, we have correlated the selenium concentrations of patients with grades I–III HCC and the expression of SELENBP1, which was recently evaluated in the same patients, by immunohistochemical studies (19). The rationale of this evaluation is based on the fact that this protein incorporates exogenously administered sodium selenite in the liver. Our studies provided evidence that SELENBP1 was also down-regulated in the liver tissues of HCC patients and its gradual loss was associated with an increased malignant grade (19). As reported in Fig. 3B, a significant correlation between SELENBP1 expression and selenium levels with a correlation coefficient equal to 0.83 was detected. Thus, a decrease in selenium concentration in the liver corresponds to lower values of SELENBP1. This is a notable result since we have shown a clear correlation with SELENBP1 in liver cancer, while data published in a recent study (24) report no correlation between serum or prostate tissue selenium and GPX activity in prostate cancer patients. It is noteworthy that prostate cancer also shows SELENBP1 down-regulation (25). Therefore, in general, these data appear to suggest that the selenoenzymes containing endogenous selenium as selenocysteine are not involved in those mechanisms related to the low selenium levels in HCC tissues.

Figure 2 Dendrogram of the cluster analysis. At a distance level of 10, three clusters are distinguished; the horizontal axis is scaled in terms of percentages.

Figure 3 (A) Correlation between tissue selenium concentrations (μg/g) and grading or (B) immunohistochemical scores of SELENBP1.

Discussion

Therapy for HCC remains poor; therefore, it is essential to identify novel HCC markers for early detection of this disease, and tumor-specific proteins as potential therapeutic targets. Despite the various studies conducted to understand the clinicopathological characteristics of the disease and to improve patient treatment, conventional clinicopathological characteristics have a limited predictive capacity. In 1960, it was suggested that α fetoprotein (AFP) may be a useful tumor marker for HCC (26). Although AFP is widely used as a marker, its drawbacks include a limited sensitivity and the fact that AFP is not specific to HCC (27). Thus, the diagnosis of HCC cannot depend mainly on the measurement of AFP. Consequently, molecular biomarkers associated with HCC should be identified. Different proteomics have been applied to the study of HCC, resulting in panels of altered proteins that are proposed as markers with key roles in the progression of the disease in certain cases (28–35). However, the protein repertoires reported in these studies are barely coincident and, consequently, their value as biomarkers for HCC is limited. At present, only five biomarkers for HCC are available for clinical use, from which only AFP partially fulfills the requirements of an ideal tumor marker (36).

Selenium is capable of being tightly bound by proteins, known as selenium-binding proteins, to distinguish them from real selenoproteins (37). The exact role of selenium-binding proteins is unknown, although certain evidence suggests that these proteins are involved in intra-Golgi protein transport. A homolog in Lotus japonicus has been found to be involved in the modulation of the oxidation/reduction status of target proteins in vesicular Golgi transport (38–39). Moreover, selenium has been found to be involved in the protection from ROS-induced cell damage, which is evident in a number of diseases, and its serum and tissue levels have been found to be significantly lower in patients with certain types of cancer in respect to healthy controls (5–6). Therefore, a functional correlation may be suspected between the antique protein of Lotus and that in humans. The Golgi apparatus has the crucial cellular function of protein release following protein biosynthesis on ribosomes; therefore, a functional protection against oxidative phenomena always present in the eukaryotic cell may be necessary.

In this study, we have evaluated liver selenium levels in tissue samples of HCC patients subdivided on the basis of grading by atomic absorption spectrometry. The mean selenium concentrations in the liver were lower in patients with the highest malignant grade in the following order: grade I> grade II>grade III. Moreover, these values correlated with the immunohistochemical scores related to SELENBP1 evaluated in a recent study (19). Therefore, the results suggest that i) the evaluation of selenium and SELENBP1 concentrations are particularly useful for prognosis improvement when clinical considerations suggest a hepatic biopsy in cirrhotic patients, and ii) the intake of selenium and its derivatives in cirrhotic patients should be considered as potential drugs against HCC, which is in agreement with a number of studies on human hepatoma cells and on animal models of liver cancer that showed the cytotoxic and antioxidant effects of selenium (40). To the best of our knowledge, no study has investigated the correlation between selenium and SELENBP1 concentrations in the liver tissue of patients affected by HCC, being a potentially significant target tissue for a chemopreventive effect of selenium. Although there is no other human tissue study, a number of studies investigated circulating concentrations of selenium, which were used as a surrogate for tissue measurements. Our findings are specific for HCC, however we suggest that this correlation should be confirmed for other types of cancer in which down-regulation of SELENBP1 has been found.

References