WO2012125051A1 - Selenoprotein genotypes and serum selenium level as markers of cancer risk - Google Patents

Abstract

A method for early detection of an increased or decreased risk of developing cancer, which comprises detecting a germline alteration in the sequence of the selenoprotein genes has been disclosed.

Description

Selenoprotein genotypes and serum selenium level as markers of cancer risk FIELD OF THE INVENTION

Method and composition for detection of an increased predisposition to cancers of breast, prostate, lung and larynx, based on the use of the detection of constitutional selenoprotein mutations in combination with the measurement of a particular selenium concentration in the same individual. Generally, the invention concerns an early diagnostic method. The subject of the invention allows the identification of genetic profiles and selenium concentrations which are correlated with an increased or decreased predisposition to cancers of the breast, prostate, lung and/or larynx, which cannot be reduced to the sum of the effects of the single factors considered, and may be modulated regulating the amount of selenium dietary intake based on accordingly labeled food products.

BACKGROUND OF THE INVENTION

Selenium (Se) is a microelement involved in carcinogenesis through the metabolic pathway of the selenoproteins. Both epidemiological studies and supplementation trials have suggested its efficacy in lowering the incidence of cancers of the liver, colon, prostate or lung, among others (Yu S.Y. et al. 1997 Biol Trace Elem Res 56: 1 17-124; Yoshizawa K. et al. 1998 J Natl Cancer Inst (Betsheda) 90: 1219-1224; Brooks J.D. et al. 2001 J Urol, 166: 2034-2038; Li et al. 2004 J Natl Cancer Inst 96: 696-703; Sanmartin C. 2008 Mini Rev Med Chem 8: 1020-1031 , Knekt P. et al. 1998 Am J Epidemiol 148:975-982) . However, the most recent data indicate that chemopreventive selenium activity can differ between individuals and may be dose- dependent (Lippman SM et al. 2009 JAMA 30:39-51 ; Jabtonska E et al. 2008 Eur J Nutr, 47:47-54; Karp E et al. 2010 ASCO, Abstract CRA 7004).

The relationship between selenium levels and breast or ovarian cancer, as well as prostate cancer, is not clear; literature data are contradictory (Lippman SM et al. 2009 JAMA 30:39-51 ; Garland et al. J. Am. Coll Nutr. 1993, 12(4):400-1 1 ; Garland et al. J. Natl. Cancer Inst. 1995, 87(7):497-505; Ghadirian et al. Cancer Detect. Prev. 2000, 24:305-13; El-Bayoumy et al. Mutat. Res. 2004, 551 (1 -2): 181 -97; Jariwalla et al. 2009 Br. J. Nutr. 101 (2): 182-9; Yamanoshita et al. Tohoku 2007 J Exp Med 212(2): 191 -8).

Analysis of literature data and results of our own studies suggest that there are three reasonable causes for these observed inconsistencies in cancer chemoprevention:

1 ) Genotype differences between individuals: distinct susceptibility of a particular person to cancer prevention dependent on Se. Several authors suggest that good candidates for identification of sub-group of patients for Se prevention trials are variants of selenoprotein genes.

2) Differences in baseline Se level: several trials suggested that cancer prevention with Se occur rather in sub-groups of individuals with low level of plasma / serum Se (Duffield-Lillico et al. 2003 BJU Int 91 :608-612). There is a very limited number of publications on association between cancer risk in subgroups of individuals identified by combined analyses of variants in selenoprotein genes and plasma / serum Se concentration (Jablo ska E. et al. 2008 Eur J Nutr 47:47-54; Steinbrecher A. et al, 2010 Cancer Epidemiol Biomarkers Prev, 19: 2958-2968; Penney K.L. et al. 2010 Cancer Prev. Res. 3: 604-610)

3) Differences in Se activity depending on its chemical form and on the activity of interacting compounds: some authors emphasize that one of the reasons for differing results between NPC and SELECT trials could be due to the use of yeast-Se for NPC and Se-methionine for SELECT (Selenius M. et al. 2010 Antioxid Redox Signal 12:867-880; Lippman SM et al. 2009 JAMA 30:39-51 ). Dependence of Se activity on the interaction with several compounds like plant compounds, vitamins or chemicals has been also reported (Whanger P.D. 2004 Br J Nutr, 91 : 1 1 -28; Davies CD. et al. 1999 J Nutr 29: 63-69; Finley J.W. et al. 2001 Biofactors 14: 191 -196; van Ommen B. et al 2008 Br J Nutr 99: 572-80 Fairweather-Tait S.J. et al. 2010 Am J Clin Nutr 91 : 1423-1429).

Techniques to measure selenium concentration are of common use such as fluorometric or microfluorometric methods (e.g. Charalabopoulos et al. Br. J. Cancer 2006, 95(6):674-6), graphite furnace atomic absorption spectrometry (GFAAS) (e.g. Lin et al. Biol. Trace Elem. Res. 1998, 64(1 -3): 133-149), or by combination of different chromatography techniques coupled with different techniques of mass spectrometry (e.g. Ando et al. Eur. J. Mass Spectrom. 2003, 9(6):619-22; Bueno et al. Talanta 2009, 78(3):759-63). These techniques for measurement of selenium levels will be considered conventional from here on.

Selenium is incorporated into the organism basically in the form of an aminoacid - selenocysteine - in so called selenoproteins or selenoenzymes. Currently there are several selenoproteins identified, however the biological function is known for just a small fraction of them (Brown KM and Arthur JR. Selenium, selenoproteins and human health: a review. Public Health Nutrition: 4(2B), 593-599). In functionally characterized selenoenzymes, selenocysteine is part of the catalytic group within their active site and is directly involved in redox reactions (Zhong L and Holmgren A. Essential role of selenium in the catalytic activities of mammalian thioredoxin reductase was revealed by characterization of recombinant enzymes with selenocysteine mutations. J Biol Chem 275: 18121-18128, 2000).

Selenoproteins are involved in different biological processes ranging from DNA synthesis to protection against oxidative stress. The latter role is assumed to be the most probable metabolic link leading to an increased cancer risk in case of altered gene expression or gene structure of some of these selenoproteins. I.e. cancer associated with selenium deficiency may be attributed to increased oxidative stress and alterations in redox signaling due to selenoprotein impairment. As an example, an animal model comprises transgenic mice overexpressing a mutant TRSP gene (Selenocystein tRNA gene) - that is a common requisite for the expression of many selenoproteins - leading to a reduced expression of most of those selenoproteins, which are then crossed with mice genetically modified to develop prostate cancer. The resulting selenoprotein-deficient mice exhibited accelerated development of lesions associated with prostate cancer progression (Diwadkar-Nawsariwala V et al. Selenoprotein deficiency accelerates prostate carcinogenesis in a transgenic model. Proc. Natl. Acad. Sci. USA 103:8179-8184, 2006). Similarly, abnormal expression levels of certain selenoproteins have been reported to be associated with an increased risk of developing cancer of different types (Kumaraswamy E et al. Structure-expression relationships of the 15-kDa selenoprotein gene: possible role of the protein in cancer etiology. J Biol Chem 275: 35540-35547, 2000; Huang KC et al. Selenium binding protein 1 in ovarian cancer. Int J Cancer 1 18: 2433-2440, 2006; He QY et al. Diverse proteomic alterations in gastric adenocarcinoma. Proteomics 4: 3276-3287, 2004; Hough CD et al. Coordinately up-regulated genes in ovarian cancer. Cancer Res 61 : 3869-3876, 2001 ).

In the particular case of the cancer types which are subject of the present specification, it has been reported poor outcome for lung adenocarcinoma where SELENBP1 protein levels are decreased (Chen G et al. 2004 J Pathol 202: 321- 329), differential expression for HSP56 in prostate cancer cells (Yang M and Sytkowski AJ 1998 Cancer Res 58: 3150-3153), low SELENBP1 and SELENBP2 mRNA levels in mammary carcinomas (Lanfear J, Fleming J, Walker M, and Harrison P. Carcinogenesis 14: 335-340,1993), whereas, on the contrary, TRX, TRXR1 and GPX2 are overexpressed in breast carcinoma (Matsutani Y et al. Clin Cancer Res 7: 3430-3436, 2001 ; Turunen N et al. APMIS 1 12: 123-132, 2004; Esworthy RS, Baker MA, and Chu FF. Cancer Res 55: 957-962, 1995).

Much less information is available on the effects of particular mutations of those selenoproteins on cancer risk. Some reports have shown an association between somatic changes of some selenoproteins and breast and ovarian cancer. Loss of heterozygosity (LOH) of the GPX1 locus has been observed in breast cancer, among other cancer types (Hu YJ and Diamond AM 2003 Cancer Res 63: 3347-3351 ).

Some germline mutations - as are the subject of the present specification, in contrast to somatic mutations as the aforementioned - have been also suggested to have a potential relationship to breast cancer risk. A GPX1 alteration (rs1050450) was demonstrated to be less responsive to selenium supplementation in breast cancer cells in vitro (Hu YJ and Diamond AM 2003 Cancer Res 63: 3347-3351 ). And the same polymorphism was shown to increase breast cancer risk, but only among postmenopausal women (Ravn-Harn G et al. 2006 Carcinogenesis 27: 820-825). Said polymorphism was again found to be increasing the risk of breast cancer but only for a specific genotype combination with SOD2 (Cox DG et al. 2006 BMC Cancer 6:217). However, most analogous studies have failed to confirm such association between GPX1 (rs1050450) and breast cancer risk (Knight et al. 2004 Cancer Epidemiol Biomarkers Prev 13: 146-149; Ahn J et al. 2005 Cancer Epidemiol Biomarkers Prev 14: 2459-2461 , Cox DG et al. 2004 Cancer Epidemiol Biomarkers Prev 13: 1821 -1822). Two linked alterations in GPX4 (rs713041 and rs757229) are significantly associated with an increased risk of death among breast cancer patients, however, here again no direct influence on breast cancer risk was assessed (Udler M et al. J Clin Oncol 25(21 ):3015-3023, 2007).

For the case of lung cancer, four studies have focused on the potential effects of GPX1 (rs1050450). The results, however, could not be more antagonistic. Where some of the studies suggested an increased risk, others showed a decreased risk for the same GPX1 genotype (Ratnasinghe D et al. 2000 Cancer Res 60: 6381 -6383; Lee CH et al. 2006 J Prev Med Pub Health 39: 130-134; Raaschou-Nielsen O 2007 Cancer Lett 247: 293-300; Yang P et al. 2004 Carcinogenesis 25: 1935-1944). Most interestingly, the SEP15 G/A variant at 1 125 (rs5859), in linkage disequilibrium with the variant rs5845 that is object of the present invention, has been suggested to affect the chemopreventive effects of Se. Among smokers, those individuals carrying the genotype AA may benefit most from a higher Se intake, whereas in those with the GG or GA genotype, a higher Se status may increase the risk for lung cancer (Jabtonska E et al. 2008 Eur J Nutr, 47:47-54).

For the case of prostate cancer, three variants of the selenoprotein SEP15 have been suggested to be significantly associated with prostate cancer mortality (rs479341 , rs1407131 and rs561 104), additionally, rs561 104 has been suggested to modify the association of plasma selenium levels with prostate cancer survival (Penney et al. 2010 Cancer Prev Res 3:604-610), however this variant is not in linkage disequilibrium with rs5845, object of the present invention. The possible influence of some GPX1 variants on prostate cancer risk, depending on serum Se levels has also been shown in literature (Steinbrecher et al. 2010 Cancer Epidemiol Biomarkers Prev 19:2558-2568); rs1050450 non-CC significantly increments the chemopreventive effects of higher serum Se levels. The CAPS study showed an increased prostate cancer risk for carriers of the GPX4 minor allele at rs713041 and for carriers of the SEP15 rs5859 (in linkage disequilibrium with rs5845), however in the latter case only for men with high PSA values (Rayman MP 2009 Biochimica et Biophysica Acta 1790: 1533-1540).

There are no published links known to us between the risk of larynx cancer and the genetic profile of any selenoprotein.

Selenoproteins have been subject of different patents. They mostly cover the industrial production of enhanced selenoproteins for the treatment of different diseases (e.g. WO2007002163, WO2006013003, US20101 13338). Some inventions disclose the differential expression of particular selenoproteins as diagnostic markers for cancer (KR20100036243, WO2008096767, CN 1920570, CN 101493464). Some of the selenoproteins subject of the present specification - GPX1 , GPX4, TXNRD2 and SEP15 - are also affected. Glutathione peroxidase expression level has been described as a diagnostic marker for breast cancer (JP5087815). Moreover, GPX1 - knock-out transgenic mice have been patented as an animal model for cancer (US2003009776). SEP15 abnormal expression, as well as the presence of its polymorphism rs5845, has been claimed to be a marker for cancer risk (W09951637).

However, besides SEP15 rs5845, no germline polymorphism of the mentioned genes has ever been specified as a diagnostic marker for cancer of any type. And none of them have been disclosed as modulating cancer risk in conjunction with selenium levels in the body, as is the subject of the present invention. in this scenario, where cancer risk may be affected by the selenium level in the body in interaction with the genetic profile of the individual, it becomes evident the need of an optimized protocol of Se supplementation for cancer prevention,

it is generally accepted that most of dietary Se is highly bioavaiiable: >90% of Se- methionine is absorbed, Se-cysteine appears to be absorbed very well, -100% of seienate is absorbed, but a significant fraction is lost in urine, whereas >50% of seienite is absorbed and is better retained than seienate (Institute of Medicine Dietary Reference Intakes: Vitamin C, Vitamin E, selenium and carotenoids. Washington DC, National Academic Press, 2000).

The particular compound in which Se is present in the body seems to be critical for anti-cancer activity. Se-methy!seienocysteine is clearly more effective than seienite, selenomethionine or selenocysteine against mammary tumorigenesis in animals, and methylseienol could be the active form of Se against tumor formation (Ip C, et ai. 2009 Cancer Metastasis Rev 21 : 281 -289). However, some observations indicate that it may not apply for other tumor models, as for example some types of colon cancer (Whanger PD, et a!. 2004 Br J Nutr 91 ; 1 1 -28).

One of the most attractive and convincing data on human cancer chemoprevention with Se supplementation were provided by Combs et al, (Biomed Environ Sei 1997; 10: 227-234) in a trial where Se-enriched yeast was successfully applied to prevent lung, colorectal and prostate cancer. Se-enriched yeast is a mixture where the dominant part is selenomethionine, but there is also a fraction of Se- methylseienocysteiine and other selenium compounds. Those positive results from Combs et aL were not reproducible in the SELECT trial (Dunn BK, et al. 2010 Nutr Cancer 62: 896-918), Some experts hypothesize that this may be so because in the latest studies just selenomethionine or selenomethionine and vitamin E have been applied instead of a combination of compounds such as in Se-enriched yeast.

There are indications that plants convert Se to forms of different effectiveness. For example Se-enriched broccoli were reported to be more effective than seienite, se!enate or selenomethionine in the reduction of experimental colon carcinogenesis (Davis C et al. 2002 J Nutr 132: 307-309). In contrast, seienite, selenate and selenomethionine were more effective for induction of glutathione peroxidase (GPX) activity than Se-enriched broccoli (Finiey JW et al. 2000 J Nutr 130: 2384-2389). in a study by Kirby et al. (2008 J Agr Food Chem 58: 1772-1779), the plasma Se response in a supplementation trial appeared to be related to the form of Se in wheat flour biscuits: intake of selenomethionine in enriched wheat biscuits resulted in a greater increase in plasma Se after 6 months than the oxidized selenomethionine in enriched biscuits.

Ail this proves that the analysis of the effects of the different dietary Se compounds on health is a challenging task. There is additionally a deviation between the effects of artificially synthesized Se compounds and those of natural origin. While, at the same time, there are currently no methods that may reliably extract 100% of the Se from natural dietary products without potentially affecting the composition (Fairweather-Tait SJ et aL 2010 Am J Clin Nutr 91 : 1484S-1491 S). in general, it is difficult to assign specific figures for retention and bioavailability of Se for particular person. The main reason is the large amount and complexity of Se compounds as well as the genetically determined individual predisposition for absorption, retention and metabolism of Se (selenoproteins and other genes) and covariates influencing these processes (availability of other micronutrients, such as vitamins). However, as a general rule, it can be admitted that over a period of several weeks or months after increasing / lowering general amount of Se in diet, it is possible to observe corresponding changes in Se concentration in plasma or serum of the examined subject.

According to our experience, in most patients the level of Se in plasma can be increased by 10 pg/i providing a diet with additional 10 g of Se daily for 6-8 weeks. This ratio between amount of Se in diet and its plasma / serum concentration has to be finally established individually for each particular person.

Thus, it is promising that changes in Se concentration may decrease dramaticaliy the risk of common cancers. It should be possible to achieve it by changes in dietary content of Se, Clinical effects depend on the specific forms of the Se compounds and on co-factors interacting with them. However, reliable and reproducible analyses of said effects has proved to be out of our current technical possibilities as is shown by the contradictory results in recent literature listed above. In order to overcome these limitations and translate this knowledge into routine cancer prevention it seems reasonable to implement dietary changes in Se content, based on the use of a series of food products as large as possible, provided they are examined for total concentration of Se. Currently, Se concentration is defined only in selected, usually Se-enricbed, food products available in food markets.

As a recapitulation of the state of the art, we could summarize that a) selenium levels are used as prognostic markers of breast, prostate and lung cancer, whereas the optimal levels are most probably in a different range for different genetic profiles (notably, selenoprotein profiles) b) selenoproteins are the entry point for selenium in the metabolic interplay and are also used as diagnostic / prognostic markers depending both on their expression levels and their particular allelic variants. Expression levels of selenoproteins are not mere surrogate markers of the serum selenium levels, since not only selenium availability determines their functionality, but also the particular genetic variants of the affected gene (Dumitrescu AM et al. Mutations in SECISBP2 result in abnormal thyroid hormone metabolism. Nat Genet 37: 1247-1252, 2005) or of a regulator gene (Diwadkar-Nawsariwala V et al. Selenoprotein deficiency accelerates prostate carcinogenesis in a transgenic model. Proc. Natl. Acad. Sci. USA 103:8179-8184, 2006).

The combined analysis of selenium concentration and germline selenoprotein genetic variants, to determine the optimal selenium range for a particular genetically defined group, decreasing the risk of developing a cancer of the breast or the ovaries, has only once been attempted to date, for the particular genetic subgroup of BRCA1 mutation carriers affected of a high-risk of developing breast and ovarian cancer. This kind of tailored analysis depending on the subject's genetic background seems particularly relevant from the clinical point of view in a moment where the recommended diary selenium intake is already high in most countries; 55-70 pg/day in the USA (Guidelines of the US National Institute of Health; http://ods.od.nih.gov), up to 75 μg/day in UK and is being revised to even higher values (Papp LV et al. From Selenium to Selenoproteins: Synthesis, Identity, and Their Role in Human Health. Antioxidants & Redox Signaling 9(7): 775-806, 2007), whereas the optimal selenium range could be lower for particular groups and even drive to an increased instead of decreased risk of developing cancer when passing a certain threshold. OBJECTS OF THE INVENTION

The subject of the present invention is a method for predicting the risk of developing cancers of the breast, prostate, lung and/or larynx for a given subject, depending on his constitutional genetic profile of the selenoprotein genes GPX1 , GPX4, TXNRD2 and SEP15, characterized by analysis of any genetic material obtained from the patient, and depending also on his selenium concentration, characterized by the analysis of any biologic material obtained from the patient with measurable presence of selenium such as nails, hair, urine or blood, without loss of generality.

As genetic profile of the GPX1 gene it is understood the presence or absence of the minor allele at SNP rs1050450 in the genetic sequence of said gene, as well as any linked flanking mutation in its neighborhood (positional markers).

As genetic profile of the GPX4 gene it is understood the presence or absence of the minor allele at SNP rs713041 in the genetic sequence of said gene, as well as any linked flanking mutation in its neighborhood (positional markers). As genetic profile of the TXNRD2 gene it is understood the presence or absence of the minor allele at SNP rs1 139793 in the genetic sequence of said gene, as well as any linked flanking mutation in its neighborhood (positional markers).

As genetic profile of the SEP15 gene it is understood the presence or absence of the minor allele at SNP rs5845 in the genetic sequence of said gene, as well as any linked flanking mutation in its neighborhood (positional markers).

The numeration system of the genetic sequence and the terminology to denominate the mutations used in the current patent comply with the established scientific terminology in this area.

As selenium levels, it is understood the concentration of selenium measured in any biological material of the patient, but preferably those with stable concentrations with known standards and at the same time available without use of invasive methods (such as nails, hair or urine) or almost non-invasive (such as blood extraction), without loss of generality.

As techniques for measuring selenium concentration, it is understood any conventional technique such as fluorometric or microfluorometric methods, graphite furnace atomic absorption spectrometry, or by combination of chromatography coupled with mass spectrometry, without loss of generality.

The selenium level considered as a threshold indicating an increased risk, that is optimally segregating affected and unaffected subjects, varies for different cancer types and genetic profiles. For the case of lung cancer, subjects with serum selenium concentration below -70 g/l are at an increased disease risk irrespective of the genotype, while above -90 g/l the risk tends to zero.

For the case of larynx cancer, subjects with serum selenium concentration below -70 g/l are at an increased disease risk irrespective of the genotype.

For the case of prostate cancer, subjects with serum selenium concentration below -70 μ9/Ι are at an increased disease risk for the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of SNP rs1 139793 in the gene TXNRD2, or GG of the SNP rs5845 in the gene SEP15 . For the case of breast cancer, subjects with serum selenium concentration below -70 g/l are at an increased disease risk for the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, or non-GG of the SNP rs5845 in the gene SEP15, while subjects with serum selenium concentration below -60 pg/l or above -79 pg/l are at an increased disease risk for the genotype GG of the SNP rs1 139793 in the gene TXNRD2, and while subjects with serum selenium concentration higher than -70 g/l are at an increased disease risk for the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG.

Said concentration thresholds of serum selenium, may be extended, without loss of generality, to the corresponding proportions of selenium in other tissues or body secretions, since these correlate with each other, and without loss of generality, it may be extrapolated to populations with similar baseline selenium concentrations, and after validation and normalization of the thresholds, also to other populations with different baseline levels.

The risk of developing a tumor of the breast, prostate, lung or larynx may be modulated regulating the amount of selenium dietary intake based on monitored and accordingly labeled food products, and on the systematic monitoring of the changes in the subject's Se levels.

SUMMARY OF THE INVENTION

The subject of invention is the mode of detection of an increased predisposition to cancers of the breast, prostate, lung and/or larynx and the use of the detection of constitutional compound mutations of the selenoprotein genes GPX1 , GPX4, TXNRD2 and SEP15 in combination with the measure of a selenium concentration in the same individual, characterized by the analysis of biological material obtained from the patient in search for constitutional mutations of GPX1 , GPX4, TXNRD2 and SEP15 and selenium concentration, wherein a) a low serum selenium concentration is indicating an increased risk of developing cancer of the lung or of the larynx; b) a low serum selenium concentration in combination with the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of SNP rs1 139793 in the gene TXNRD2, or GG of the SNP rs5845 in the gene SEP15 is indicating an increased risk of developing cancer of the prostate; c) a low serum selenium concentration in combination with the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, or non-GG of the SNP rs5845 in the gene SEP15 is indicating an increased risk of developing cancer of the breast; d) both a low or a high serum selenium concentration in combination with the genotype GG of the SNP rs1 139793 in the gene TXNRD2 are indicating an increased risk of developing cancer of the breast; e) a high serum selenium concentration in combination with the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG are indicating an increased risk of developing cancer of the breast, where the effect is different than the sum of the effects of the mentioned factors when taken separately. Equally subject of this invention is the regulation of the risk of developing a tumor of the breast, prostate, lung or larynx by changing the selenium dietary intake based on monitored and accordingly labeled food products, and on the systematic monitoring of the changes in the subject's Se levels.

EXAMPLE

1. Introduction

Selenium (Se) has been suggested to be a trace element of relevance in human carcinogenesis. Effective cancer prophylaxis by chemoprevention with Se has been demonstrated in experimental studies on animals and in some human clinical trials. Well known examples are studies on Se chemoprevention of prostate cancer. However, there are serious inconsistencies in the results of the published studies. A protective effect for Se supplementation was originally suggested by the Nutritional Prevention of Cancer (NPC) trial (Clark et al. 1996, JAMA 276: 1957-63). However, latest results from the SELECT trial did not support this conclusion (Lippman et al. 2009, JAMA 301 : 39-51 ). Meanwhile, there is even support for the hypothesis that Se supplementation may increase the risk of skin and lung cancers (Duffield-Lillico et al. 2003, JNCI 95: 1477-81 ; Karp et al. ASCO 2010, Abstract CRA 7004).

In our center we formerly performed a double-blind trial on a sample of 1343 BRCA1 mutation carriers and, unexpectedly, during almost 3-year follow-up we observed an increased number of cancers of the breast or the ovary among women supplemented with Se (Lubinski et al. unpublished data). This motivated a following study where the potential relationship between the genetic profile of BRCA1 and several selenoproteins was analyzed in detail for the risk of cancer of the breast or the ovary. We were able to show that among BRCA1 mutation carriers, the SNP rs1050450 in the gene GPX1 combined with serum Se levels above or equal 80 pg/l increased disease risk (OR=6.24; 95%CI= 1.09-36; p=0.040), while, in general, subjects having serum Se levels above or equal 80 pg/l showed a decreased risk of being diseased when their GPX1 genotype was not considered (OR=0.24; 95%CI= 0.06-0.94; p=0.041.

In the light of these results, a broader scope analysis of the potential interaction between selenoprotein genetic profile and serum Se levels was performed in a sample of men and women with a normal BRCA1 status and for several cancer types.

In this study we tested the impact of Se levels in blood serum on the risk of breast, prostate, lung and larynx cancer depending on the genetic profile of four key selenoproteins critical for the selenium metabolism: GPX1 , GPX4, TXNRD2 and SEP15. Gender, year of birth, prior history of cancer, adnexectomy, and smoking habits were taken as main covariates. The main goals of the study were a) the assessment of common cancer risk (cancers of lung, breast, prostate or larynx) depending on genomic variants in selected selenoproteins (GPX1 , GPX4, TXRND2 and SEP15) using the model of Polish population (Slavic ethnicity with relatively high level of genetic homogeneity) characterized by low level of Se (more than 95% of individuals with plasma / serum level below 100 g/l) and b) the elaboration of a protocol of Se supplementation with the highest chance for an efficient cancer prevention in clinical trials.

2. Methods

2.1. Subjects

Breast cancer

The initial sample consisted of 131 consecutive cases of breast cancer diagnosed at our Center (Cancer Genetics Outpatient Clinics, Dept. of Radiology, Pomeranian University of Medicine, Szczecin, Poland) in the period from July 2009 to February 201 1.

Patients were excluded from the sample if they were carriers of a BRCA1 mutation and/or if they were diagnosed with any malignancy in the past and/or if they underwent adnexectomy in the past.

For each case included in the study a healthy pair was found among unaffected individuals registered in our center. Pairs were matched for: year of birth (+/- 3 years), gender, total number of breast cancers and of any other malignancies among first degree relatives, smoking habits (packs/years +/- 10%) and adnexectomy status. In this way, 108 pairs could be created (82% from the initial 131 cases). Clinical characteristics of those pairs are shown in table 1 a.

Table 1 a. Breast cancer - clinical characteristics of individuals included in the study

Prostate cancer

The initial sample consisted of 105 consecutive cases of prostate cancer diagnosed at the Urology Clinics (Pomeranian University of Medicine, Szczecin, Poland) in the period from August 2009 to February 201 1. As in the foregoing group, patients were excluded from the series if any malignancy was diagnosed in the past.

For each case included in the study a healthy pair was found among unaffected individuals registered in our center. Pairs were matched for: year of birth (+/- 3 years), gender, total number of prostate cancers and of any other malignancies among first degree relatives and smoking habits (packs/years +/- 10%). In this way, 82 pairs could be created (78% from the initial 105 cases). Clinical characteristics of those pairs are shown in table 1 b.

Table 1 b. Prostate cancer - clinical characteristics of individuals included in the study

Lung cancer

The initial sample consisted of 95 consecutive cases of lung cancer diagnosed at the Clinic of Thoracosurgery (Pomeranian University of Medicine, Szczecin, Poland) in the period from August 2009 to January 201 1. As in the foregoing groups, patients were excluded from the series if any malignancy was diagnosed in the past.

For each case included in the study a healthy pair was found among unaffected individuals registered in our center. Pairs were matched for: year of birth (+/- 3 years), gender, total number of lung cancers and of any other malignancies among first degree relatives and smoking habits (packs/years +/- 10%). In this way, 83 pairs could be created (87% from the initial 95 cases). Clinical characteristics of those pairs are shown in table 1 c.

Table 1c. Lung cancer - clinical characteristics of individuals included to studies

Laryngeal cancer

The initial sample consisted of 1 13 consecutive cases of laryngeal cancer diagnosed at the Clinic of Otorhinolaryngology (Pomeranian University of Medicine, Szczecin, Poland) in the period from July 2009 to January 201 1 . As in the foregoing groups, patients were excluded from the series if any malignancy was diagnosed in the past. For each case included in the study a healthy pair was found among unaffected individuals registered in our center. Pairs were matched for: year of birth (+/- 3 years), gender, total number of laryngeal cancers and of any other malignancies among first degree relatives and smoking habits (packs/years +/- 10%). In this way, 91 pairs could be created (81 % from the initial 1 13 cases). Clinical characteristics of those pairs are shown in table 1 d.

Table 1d. Laryngeal cancer - clinical characteristics of individuals included in the study

All patients were informed by a physician, participated voluntarily and signed a written consent before taking part in the trial.

2.2. Measurement of selenium level in blood Selenium concentration in blood serum was determined using graphite furnace atomic absorption spectrometry (GFAAS). Measures were validated using reference material (lyophilized human reference serum samples of Serenoform™ from Nycomed Pharma AS, Oslo, Norway). Each four samples, reference material was measured to keep a constant control on the quality of the measurements. The mean drift was taken a correction value for the samples, whereas a measured drift larger than 5% from the reference material caused the device to be newly calibrated and the last four samples after the last successful control, to be newly measured.

In a quality control we measured the selenium levels of 3 subjects divided in 10 aliquots of blood serum, each. The standard deviation of the repeated measures was 3.1 , 3.9 and 4.3, for mean values of 66.0, 68.6 and 72.6, respectively.

2.3. Genotyping of Selenoproteins GPX1 , GPX4, TXNRD2 and SEP15

One of the most common SNPs in the Polish population for each of the selenoprotein genes included in the analysis (GPX1 , GPX4, TXNRD2, SEP15) are included in a RT-PCR assay for massive genotyping in the four independent groups of cancer patients and matched controls that are constitutive of this study. For GPX1 rs1050450 (C/T 200Leu/Pro) was selected, for GPX4 rs713041 (T/C 220Leu/Leu), for TXNRD2 rs1 139793 (G/A 370Thr/lle), and for SEP15 rs5845 (G/A). The primers used in the case of GPX1 are summarized in table 2 as GPX1_2 and GPX1_2 since they were the only ones custom made. The rest are identified with their corresponding Taqman® Assay identificator.

Table 2. Primer sets for analysis of GPX1 mutations with conventional PCR. Primer pairs Primer ID Function Primer for sense strand [F] Primer for antisense strand [R]

5^'->3^' 5^'->3^'

Pair 1 for GPX1_1 G CC AG TTAAAAGG AG G CG CGAGAGAGTAGCCAGACTC

Identification

Pair 2 for GPX1 2 Control CCGGTGACTCATAGAAAATC CCTCGTAGGTTTAGAGGAAAC

ΠΡΥ-Ι ov 9

GPX4 Taqman® Assay ID C_2561693_20

TXNRD2 Taqman® Assay ID C_25651 159_20

SEP15 Taqman® Assay ID C_876937_1

The RT-PCR reaction was carried out with an LightCycler® 480 (Real-Time PCR System Roche diagnostics). The mixture of substances was performed in a Mastermix containing 2.5 μΙ of ProbesMaster, 0.0625 μΙ of assay, 1.4375 μΙ of H20, 1 μΙ of DNA (25 ng) up to a total volume of 5 μΙ. The corresponding assays are C_2561693_20 for GPX4 rs713041 (T/C 220Leu/Leu), C_25651 159_20 for TXNRD2 rs1 139793 (G/A, 370Thr/lle), C_876937_1 for SEP15 (G/A) and GPX1 (custom assay, see table 2). The RT-PCR reaction itself takes place in all four cases under the following conditions:

•DNA pre-incubation at 95°C during 10 minutes

•45 cycles of amplification / quantification consisting each of

- denaturation at 95°C during 10 seconds

- primer binding at 60°C during 30 seconds

- extension at 72°C during 1 second

•Cooling of complementary DNA at 40°C during 30 seconds

•Color compensation: - denaturation at 95°C during 1 second

- cooling at 40°C during 30 seconds

- hybridization at 67°C continuously at 1 acquisition per each °C

- cooling at 40°C during 45 seconds

•Generation of melting curves:

- hybridization at 60°C during 1 second

- hybridization at 61°C continuously at 5 acquisitions per °C

•Cooling at 40°C during 30 seconds

2.4. Statistics

Considerations

The analysis was kept as simple as possible to avoid a decreased statistical power due to an increased complexity of the model, given the limited number of analyzed subjects and the high number of combinations of variables (cancer type, serum Se concentration and 3 possible genotypes for each of the four analyzed selenoproteins).

Each of the four sets of matched paired cases and controls (corresponding to breast, prostate, lung and larynx cancer respectively) was analyzed independently for each genetic profile of each of the four selenoproteins taken into account (corresponding to GPX1 , GPX4, TXNRD2 and SEP15 respectively). However, for simplicity only two possible profiles were considered for each selenoprotein, one corresponding to the major allele in homozygosity and another one corresponding either to the minor allele in homozygosity or to the heterozygous genotype. That is a profile division according to a dominant or codominant model (for simplicity the recessive model was left out of the study).

Each of the eight resulting subgroups of patients or controls for each cancer type was further divided into quartiles according to the measured Se serum levels. Those quartiles were combined in two different ways. In the first case, the first and second quartiles were merged and compared to the third and fourth quartiles, which were also merged. In the second case, the first and fourth quartiles were merged and compared to the joined second and third quartiles. Those are a quartile divisions according to a linear model (either decreasing or increasing) and to a "U" curve (or inverted "U" curve) respectively. The combination of first and third versus second and fourth joined quartiles (corresponding to an atypical chainsaw curve) was excluded from the analysis for simplicity.

Thus, for each type of cancer, we performed a risk analysis between patients and controls in altogether 16 subsets (2 possible genotypes in 4 possible selenoproteins for 2 possible quartile combinations of Se serum levels). 16 was therefore the total number of hypotheses tested on the same set of individuals. All p-values presented in the results section are also corrected taking this value for multiple testing into account for the Bonferroni correction.

Analysis of compound genotypes was deliberately kept as simple as possible, since we had limited statistical power for the given sample sizes. For this reason, only compound genotypes with at least 20 individuals were taken into account. All conclusions arising from the results presented here, are based on the assumption of lacking unexpected synergistic effects for carriers of rare compound genotypes (i.e. for those cases only the scenario of an additive model has been taken into account). Analysis

Differences in proportions of patients and matched controls were calculated for each of the 16 subsets for each of the four cancer types were analyzed comparing observed odds ratios with expected odds ratios calculated by a random permutation test (with 5000 iterations and an accuracy of four decimals in the estimation of the empirical p-value). All statistical analyses were performed on the platform "R for statistical computing" (version 2.6.2).

3. Results

3.1. Disease risk depending only on serum Se concentration

Distribution of cases and controls in quartiles of serum Se concentration (without additional stratification by genotypes) are summarized in table 3.

Table 3. Association between Se plasma concentration and risk of cancers independently on variants of selenoprotein genes

Cancer site Quartile Se concentration range (μς)Ι\) Cases [% - No]

I 26,3 - 59,1 59,3% (32/54)

II 59,2 - 68,8 50,0% (27/54)

BREAST

III 68,9 - 79,5 38,9% (21 /54)

IV 79,5 - 129,3 51 ,8% (28/54)

I 22,0 - 61 ,4 51 ,2% (21 /41 )

II 61 ,4 - 70,2 60,0% (24/41 )

PROSTATE

III 70,3 - 79,4 43,9% (18/41 )

IV 79,6 - 136,8 40,0% (16/41 ) 1 29,0 - 56,41 80,9% (34/42)

II 56,5 - 66,6 53,6% (22/41 )

LUNG

III 66,9 - 77,1 40,5% (17/42)

IV 77,2 - 1 12,7 24,4% (10/41 )

1 28,4 - 58,9 69,6% (32/46)

II 59,2 - 69,8 62,2% (28/45)

LARYNX

III 69,9 - 80,0 39,1 % (18/46)

IV 80,3 - 1 17,4 28,9% (13/45)

There are no critical differences between quartiles for breast and prostate cancers. On the contrary, very large differences have been observed for cancers of the lung and larynx. The highest proportion of cancer cases was found in the quartile with the lowest Se concentration, 81 % for lung cancer and 70% for laryngeal cancer. Gradually, the proportion of cancer cases decreases in the next quartiles achieving in the fourth quartile the frequency of 24% for lung cancer and of 29% for laryngeal cancer.

Very remarkable was the observation of extremely low frequency of cancer cases of the lung or of the larynx in persons with serum Se concentration above 90 μ9/Ι. Not a single patient with cancer of the lung had Se concentration above 90 g/l, but this Se level could be measured in 15 persons from unaffected controls. Similarly, Se level above 90 μ9 Ι was found only in 4 patients with laryngeal cancer, but in 14 persons from unaffected controls.

3.2. Disease risk depending both on serum Se concentration and selenoprotein genotypes

Breast cancer Analyses were performed on subgroups:

a) 8 genotypes after stratification using one of the most polymorphic SNPs:

- GPX1 rs1050450, CC vs non-CC

- GPX4 rs713041 . CC vs non-CC

- TXNRD2 rs1 139793, GG vs non-GG

- SEP15 rs5845 , GG vs non-GG

b) combined quartiles:

- 1 combined with II (lowest Se levels) vs. Ill and IV (highest Se levels)

- II combined with III (central Se levels) vs. I and IV (distal Se levels).

Statistically significant differences between patients and controls were found for all genes (table 4a). For three genotypes (GPX1 rs1050450 CC, GPX4 rs713041 non- CC and SEP15 rs5845 non-GG) the first and second quartiles of Se concentration were associated with higher disease risk in comparison with the merged third and fourth quartiles. A remarkable exception was TXNRD2 rs1 139793 GG for which was showing a strongly decreased risk (OR 0.26, p=0.0006) of breast cancer for women in the middle range of Se concentration (second and third compared with first and fourth quartiles), thus matching better a "U"-curve saturation model.

Table 4a. Breast cancer risk by genotype profiles and serum Se concentration quartiles.

Genotype Quartiles Se range (μς/Ι) Controls/Cases p-value OR

GPX1 rs1050450 CC 40-70 vs 71-108 18/32 vs 28/21 0.019 2.35 l+ll

GPX4 rs713041 non-CC vs 29-67 vs 68-108 31/40 vs 42/28 0.022 1.93 lll+IV

SEP15 rs5845 non-GG 26-69 vs 70-129 17/28 vs 28/16 0.0068 2.85 Il+lll 60-78

TXNRD2 rs1139793 GG vs vs 33/20 vs 16/38 0.0006* 0.26 l+IV 40-59 and 79-106

^* - difference significant after Bonferroni correction

There were only 3 compound genotypes counting at least 20 individuals. The combination GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, was present in 23 individuals. For this group a significant lower risk could be detected for the combined first quartiles in contrast with an increased risk for the last two quartiles (OR=1 1.5, p=0.012), where the largest difference accounts for the comparison of the first with the fourth quartile (p=0.002). Both results passed Bonferroni correction for multiple testing.

The combination GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 GG and SEP15 rs5845 GG, was present in 22 individuals and the combination GPX1 rs1050450 CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 GG and SEP15 rs5845 non-GG, was present in 20 individuals. For these two groups no significant differences could be found at any Se concentration quartile.

Prostate cancer

Subgroups were defined in the same way as for breast cancer. Statistically significant differences between patients and controls were found in all genes (table 4b). In four genotype groups (GPX1 rs1050450 CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 GG, and SEP15 rs5845 GG) the first and second quartiles of Se concentration were associated with higher disease risk in comparison with merged third and fourth quartiles.

Table 4b. Prostate cancer risk by genotype profiles and serum Se concentration quartiles.

* - difference significant after Bonferroni correction

For prostate cancer we had no group with more than 20 individuals for any compound genotype.

Lung cancer

Subgroups were defined in the same way as for breast cancer. Statistically significant differences between patients and controls were found for all genotypes of all genes (table 4c). In all eight genotype groups the first and second quartiles of Se concentration were associated with higher disease risk in comparison with the merged third and fourth quartiles. This suggests that, for the case of lung cancer, the dominant factor is the serum Se concentration, irrespective of the genotype profile. Remarkably no single lung cancer patient could be found for serum Se levels above 90 g/I for all genotypes (data not shown).

Table 4c. Lung cancer risk by genotype profiles and serum Se concentration quartiles.

Genotype Quartiles Se range ( g/l) Controls/Cases p-value OR GPX1 rs1050450 CC 29-66 vs 67-111 11/28 vs 24/14 0.0003^* 4.27

GPX1 rs1050450 non-CC 30-66 vs 67-113 17/22 vs 26/12 0.013 2.76

GPX4 rs713041 CC 39-65 vs 66-107 7/19 vs 15/11 0.013 3.61 l+ll

GPX4 rs713041 non-CC vs 29-67 vs 68-113 21/30 vs 35/16 0.0032 3.09

TXNRD2 rs1139793 GG lll+IV 29-68 vs 69-113 16/28 vs 28/16 0.0094 3.02

TXNRD2 rs1139793 non-GG 30-66 vs 67-111 11/22 vs 23/10 0.0017^* 4.48

SEP15 rs5845 GG 29-64 vs 65-113 14/31 vs 29/16 0.0008^* 3.95

SEP15 rs5845 non-GG 43-70 vs 71-99 12/20 vs 23/9 0.0024* 4.17

^* - difference significant after Bonferroni correction

There was only 1 compound genotype counting at least 20 individuals. The combination GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, was present in 22 individuals, but no significant differences in cancer risk could be found at any Se concentration quartile. Larynx cancer

Subgroups for larynx cancer were defined in the same way as for breast cancer. Statistically significant differences between patients and controls were found for all genotypes of all genes (table 4d). In all eight genotype groups the first and second quartiles of Se concentration were associated with higher disease risk in comparison with the merged third and fourth quartiles. This suggests that, for the case of larynx cancer as for the case of lung cancer, the dominant factor is only the serum Se concentration, irrespective of the genotype profile. Table 4d. Larynx cancer risk by genotype profiles and serum Se concentration quartiles.

^* - difference significant after Bonferroni correction

There was only 1 compound genotype counting at least 20 individuals. The combination GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 GG and SEP15 rs5845 GG, was present in 22 individuals, but no significant differences in cancer risk could be found at any Se concentration quartile. 4. Conclusions

This study shows that selenium level in blood serum is a factor associated with an increased (or decreased) risk of breast, prostate, lung and/or larynx cancer risk, however this association depends on a further factor as is the genetic profile of the selenoproteins GPX1 , GPX4, TXNRD2 and SEP15. Particularly, low serum Se levels seem to increase the disease risk for the case of lung and larynx cancer (both overwhelmingly smoking-related cancer types) irrespective of the selenoprotein profile (OR range from 2.43 to 4.90). The tendency is analogous for prostate cancer, however only for four genotypes (GPX1 rs1050450 CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 GG, and SEP15 rs5845 GG). The exception is the case of breast cancer. Although for three genotypes low serum Se levels seem again to increase the disease risk (GPX1 rs1050450 CC, GPX4 rs713041 non-CC and SEP15 rs5845 non-GG), the genotype TXNRD2 rs1 139793 GG shows a different association. For the latter case, both low and high serum Se levels are increasing the disease risk, while central values (2^nd and 3^rd quartiles) are associated with a lower risk of developing breast cancer. Moreover, the effect is not spurious but particularly evident and passes Bonferroni correction for multiple testing (OR 0.26, p=0.0006).

That creates an exception to the former rule where, for a given subject, serum Se levels are either not significantly related to the disease risk (the case of genotypes GPX1 rs1050450 non-CC, GPX4 rs713041 CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 non-GG for prostate and genotypes GPX1 rs1050450 non-CC, GPX4 rs713041 CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG for breast cancer) or the subject should rather keep his serum Se levels at higher levels to decrease the disease risk (the case of genotypes GPX1 rs1050450 CC, GPX4 rs713041 non-CC and SEP15 rs5845 non-GG for breast cancer and genotypes GPX1 rs1050450 CC, GPX4 rs713041 non-CC, SEP15 rs5845 GG and TXNRD2 rs1 139793 GG for prostate cancer, or irrespective of the genotype for larynx and lung cancer). In the exceptional case of TXNRD2 rs1 139793 GG a woman may decrease her risk of breast cancer, not keeping serum Se levels as high as possible, but moderating them near to median values.

The situation is even more complex under the perspective given by compound genotype analysis. Here we have the combination GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, which shows a significant lower risk for lower Se concentrations. This is an unexpected synergistic effect, where, although the subjects carry the GPX4 rs713041 non-CC genotype (which should increase the disease risk at lower Se levels), they have nonetheless a significantly decreased disease risk for lower Se levels.

From the perspective of clinical practice, this means that to assess the risk of a given subject, it is not only critical to measure his serum Se levels, but also to know the genotype profile of the selenoproteins GPX1 , GPX4, TXNRD2 and SEP15, in order to provide an adequate advise for chemoprevention based on Se.

Whether the relationship is causative or not, cannot be elucidated from this analysis, but in any case the selenium concentration in blood serum together with the genotype profile of selenoproteins GPX1 , GPX4, TXNRD2 and SEP15 may be used as a prognostic factor to identify subjects under a particularly high risk of developing cancer.

The particular Se thresholds found in this study (tables 5a-d) corresponding to each quartile still cannot be taken a general reference. Studies with larger samples should determine the final threshold for the particular population and the corresponding confidence interval. There are large differences in baseline selenium level in blood serum between countries (table 5), but it seems reasonable to expect that our data can be directly extrapolated to countries, regions or populations with similar baseline levels of selenium, and - after additional normalization and validation - also to other populations with different baseline levels of selenium.

Table 5. Baseline selenium concentration in healthy individuals of different countries

* Range (no average available)

For men with low levels of serum selenium, which are potentially affected by prostate, larynx and lung cancer, it should be advisable always to increase serum Se levels over -70 pg/l (or the equivalent median of the population), except if the subject is non-smoker. In that case, the subject is primarily affected by prostate cancer risk, where for some genotypes (GPX1 rs1050450 non-CC, GPX4 rs713041 CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 non-GG) he could not be advised to increase nor decrease the serum Se level.

For women the situation is different. In general, it would be advisable for any subject to keep the serum Se levels over -70 pg/l (or the equivalent median of the population), however, for carriers of the genotype TXNRD2 rs1 139793 GG the upper quartile is a risk quartile, so for these women the consensus would be advising to keep the serum Se levels between -60 pg/l and -78 pg/l (or the equivalent third quartile of the population). In cases of women, carriers of the genotype TXNRD2 rs1 139793 GG which would be heavy smokers, the advise may be to stop smoking or, at least, to keep unequivocally the serum Se levels above -79 pg/l (or the equivalent fourth quartile of the population), to minimize the risk of lung or larynx cancer, which is lowest for the fourth quartile (see table 3) even at the cost of increasing the breast cancer risk.

Another exception for women is the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, which shows a significant lower risk for lower Se concentrations. Therefore, for this group the only possible advice is to quit smoking, since for them it seems impossible to decrease at the same time the risk of lung or laryngeal cancer (smoking- dependent) and the risk of breast cancer. Particularly striking is the situation for lung cancer, where independently of sex or of the genetic profile, for the group with serum Se levels above -90 μ9/Ι and up to the maximum of the corresponding control group (1 10 pg/l), the risk decreases to zero. Under this new perspective, one should expect to find that particular changes in Se concentration in the body may decrease the risk of particular cancer types for particular genetic profiles. Such changes may be simply achieved by changing the dietary content of Se. However, as mentioned above, the influence on cancer prognostics depends on the specific Se compounds and on co-factors interacting with them that may be comprised in the diet. To overcome these limitations and establish a protocol for routine Se-based, personalized cancer prevention based on genetic profiling, dietary changes in Se content should base on the use of a complete series of food products (as large as possible), to assure their effective activity in the human body. To reach this goal, both an appropriate labeling of the total amount and/or concentration of Se in each food product, a genetic profile of the subject and the detailed monitoring of the patient's Se levels on a time series before and during the dietary intervention, have to be applied.

Table 6 shows, without loss of generality, the enormous variability of food Se concentration exemplary on a comparison of Se-concentration data which can be found in literature and especially carefully measured values in our center, for patients willing to change effectively their Se plasma level with help of an anticancer diet. Just on a few examples 2 to 10 fold differences can be observed.

Table 6. Variability in Se concentration of food products Food product Literature data Anticancer diet data

Mustard - 100g 50 _Mg (3) Mg

Chicken egg (~ 60g) 14 _Mg (1 ) 7,5 Mg

Turkey breast - 100 g 32 μ₉ (1 ) 9,4 g

Chicken breast - 100 g 20 _Mg (1 ) 9,8 Mg

Sirlion - 100 g 25 _Mg (2) 4 9

Literature data in parenthesis:

1 : http://ods.od.nih.gov/factsheets/selenium/

2: Driskell JA et al. J Anim Sci 1997, 75:2950-2954

3: Anticancer Diet data: http://dietaantyrakowa.pl/

Se level should be systematically measured also within each type of food product, due to the large variations in Se content determined by unstable environmental factors and/or Se supplementation through fertilizers, in the case of plant products, and animal feeding, in the case of animal products.

Continuously, probably every few months, an actual plasma / serum Se level should be drawn for each following a prophylactic anticancer dietary intervention to lower cancer risk by changing the amount of Se intake.

Claims

1. A method for early detection of an increased or decreased risk of developing cancer, which comprises detecting a germline alteration in the sequence of the selenoprotein genes GPX1 at rs1050450, GPX4 at rs713041 , TXNRD2 at rs1 139793 and SEP15 at rs5845, and determining the selenium concentration in a biological sample from the same subject, wherein among carriers of a particular genotype a particular selenium concentration is indicative of significantly increased or decreased risk of developing cancer of the breast, prostate, larynx or lung when compared to other selenium concentrations.

2. The method of claim 1 , wherein the examined structural alterations of the genes GPX1 at rs1050450, GPX4 at rs713041 , TXNRD2 at rs1 139793 and SEP15 at rs5845 are identified as the presence of at least one copy of the minor allele as compared with the presence of the major allele in homozygosity, said minor alleles are the T-allele for GPX1 at rs1050450, the T-allele for GPX4 at rs713041 , the A- allele for TXNRD2 at rs1 139793 and the A-allele for SEP15 at rs5845.

3. The method of claims 1 and 2, wherein the examined structural alterations of the genes GPX1 at rs1050450, GPX4 at rs713041 , TXNRD2 at rs1 139793 and SEP15 at rs5845 are inferred by a genetic marker or a set of genetic markers linked or sharing haplotype with the corresponding alteration.

4. The method of claims 1 to 3, wherein the mode of detection of the structural alterations of the genes GPX1 at rs1050450, GPX4 at rs713041 , TXNRD2 at rs1 139793 and SEP 15 at rs5845 is based on analysis of DNA, RNA or proteins.

5. The method according to claim 4, wherein DNA or RNA testing is performed using any conventional technique of direct mutation detection, such as sequencing, but more preferably any conventional technique of indirect mutation detection, selected among those such as ASA-, ASO-, RFLP-PCR, Real Time PCR, mass spectrometry or microarray methods.

6. The method according to claim 4, wherein the presence of the polypeptides encoded by the genes GPX1 at rs1050450, GPX4 at rs713041 , TXNRD2 at rs1 139793 and SEP15 at rs5845 with germline alteration, is detected with the use of antibodies or other substances specific for these polypeptides or fragments thereof.

8. The method of claim 1 , wherein the biological material for detection of the selenium concentration in the subject is selected from any tissue or body secretion, but preferably from scalp hair, finger nails, toe nails, urine and blood.

9. The method of claims 1 and 8, wherein selenium concentration in the subject is estimated by measuring directly selenium, or indirectly by measuring a selenium metabolite, or any other metabolite or gene product, such as a protein or RNA, with a concentration that is correlated to the selenium concentration.

10. The method of claims 1 and 8 to 9, wherein selenium concentration is determined using any conventional technique, preferably selected among those such as fluorometric or microfluorometric methods, graphite furnace atomic absorption spectrometry (GFAAS), or by combination of chromatography coupled with mass spectrometry.

1 1 . The method of claims 1 to 10, wherein for a woman a significantly increased risk of developing a smoking-sensitive cancer, such as cancer of the lung and cancer of the larynx, can be predicted whenever the selenium concentration in blood serum or plasma is below 70 g/l irrespective of the genetic profile; whereas for a cancer of the breast an increased risk of developing the disease for a woman with selenium concentration in blood serum or plasma below 70 pg/l is significant only for the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, or non-GG of the SNP rs5845 in the gene SEP15, while a woman with selenium concentration in blood serum or plasma lower than 60 μ9 Ι or higher than 79 μ9/Ι is at an increased disease risk for the genotype GG of the SNP rs1 139793 in the gene TXNRD2, and while a woman with selenium concentration in blood serum or plasma higher than 70 pg/l is at an increased disease risk for the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, all the thresholds having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

12. The method of claims 1 to 10, wherein for a man a significantly increased risk of developing a smoking-sensitive cancer, such as cancer of the lung and cancer of the larynx, can be predicted whenever the selenium concentration is below 70 μ9/Ι irrespective of the genetic profile; whereas for cancer of the prostate a man with selenium concentration in blood serum or plasma below 70 μ9 Ι is at a significantly increased disease risk only for the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of SNP rs1 139793 in the gene TXNRD2, or GG of the SNP rs5845 in the gene SEP15, the threshold having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

13. The method of claims 1 to 10, wherein for any subject the risk of developing a smoking-sensitive cancer, such as cancer of the lung and cancer of the larynx, can be predicted to be near to zero whenever the selenium concentration is between 90 g/l and 1 10 μ9/Ι irrespective of the genetic profile, both thresholds having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

14. The method of claims 1 to 10, wherein for any subject the risk of developing a breast cancer can be predicted to be near to zero whenever the selenium concentration is between 40 μ9/Ι and 59 μ9/Ι for compound carriers of the genotypes GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, both thresholds having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

15. The use of the identification of the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of SNP rs1 139793 in the gene TXNRD2, or GG of the SNP rs5845 in the gene SEP15, or polynucleotides thereof, combined with a concentration of selenium in blood serum or plasma below 70 g/l in a biological sample from the analyzed subject to predict a significantly increased risk of developing cancer of the prostate, compared to carriers of the same mutation genotypes with a concentration of selenium above 70 g l, the threshold having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

16. The use of the identification of the genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of the SNP rs5845 in the gene SEP15, or polynucleotides thereof, combined with a concentration of selenium in blood serum or plasma below 70 μ9/Ι in a biological sample from the analyzed subject; and the identification of the genotype non-GG of the SNP rs1 139793 in the gene TXNRD2, or polynucleotides thereof, combined with a selenium concentration in blood serum or plasma lower than 60 ig/\ or higher than 79 μ9/Ι in a biological sample from the analyzed subject; and the identification of the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, or polynucleotides thereof, combined with a selenium concentration in blood serum or plasma higher than 70 g/l in a biological sample from the analyzed subject, to predict a significantly increased risk of developing cancer of the breast compared to carriers of the same mutation genotypes with a selenium concentration outside of the mentioned ranges, the thresholds having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

17. The use of the identification of a concentration of selenium in blood serum or plasma below 70 μς Ι in a biological sample from the analyzed subject to predict a significantly increased risk of developing cancer of the lung or of the larynx compared to subjects with a selenium concentration above 70 pg/l, the threshold having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

18. The use of the identification of a concentration of selenium in blood serum or plasma between 90 pg/l and 1 10 g/l in a biological sample from the analyzed subject to predict a risk of developing cancer of the lung, which can be predicted to be near to zero, both thresholds having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

19. The use of the identification of a concentration of selenium in blood serum or plasma between 40 pg/l and 59 pg/l in a biological sample from an analyzed subject compound carrier of the genotypes GPX1 rs1050450 non-CC, GPX4 rs713041 non- CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, to predict a risk of developing cancer of the breast, which can be predicted to be near to zero, both thresholds having a standard deviation of 5, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

20. The use according to claims 15 to 19, wherein said polynucleotides comprise the whole sequence of the genes GPX1 , GPX4, TXNRD2 and SEP15, or a fragment thereof including the examined alteration.

21 . The use according to claims 15 and 20, wherein the constitutional alterations rs1050450 in the gene GPX1 , rs713041 in the gene GPX4, rs1 139793 in the gene TXNRD2, and rs5845 in the gene SEP15 are detected by a method chosen among any technique of direct mutation detection or indirect mutation detection at DNA, RNA or protein level, but preferably selected among of ASA-, ASO-, RFPL-PCR, Taqman Real Time PCR, mass spectrometry or microarrays.

22. The use according to claims 15 to 21 , wherein selenium concentration in the subject is estimated by measuring directly selenium, or indirectly by measuring a selenium metabolite, or any other metabolite or gene product, such as a protein or RNA, with a concentration that is correlated to the selenium concentration.

23. The use according to claims 15 to 22, wherein selenium concentration is determined using any conventional technique, preferably selected among those such as fluorometric or microfluorometric methods, graphite furnace atomic absorption spectrometry (GFAAS), or by combination of chromatography coupled with mass spectrometry.

24. The use of a specific labeling for food products informing about the total amount of selenium comprised in the product alongside with the selenium concentration expressed as the common use total amount per 100g of the food product, wherein the label information is of a particular relevance for the costumers depending on their sex, smoking habits, selenium concentration in the body and genetic profile of selenoproteins according to claims 15 to 23.

25. Compound composition for prediction of increased breast cancer risk among carriers of the constitutional genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, or GG of the SNP rs5845 in the gene SEP15, combined with a concentration of selenium in a biological sample from the analyzed subject below 70 g/l ; among carriers of the genotype GG of the SNP rs1 139793 in the gene TXNRD2, combined with a selenium concentration lower than 60 μ9/Ι or higher than 79 g/l in a biological sample from the analyzed subject; and among carriers of the compound genotype GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, combined with a selenium concentration higher than 70 μ9 Ι in a biological sample from the analyzed subject, comprising at least two different oligonucleotides allowing the amplification of a region of the genome of said human subject compound carrier of mentioned constitutional mutations, or another alteration linked or sharing haplotype with the former ones, and allowing its identification by conventional methods, as well as the corresponding reagents for preprocessing at least one type of tissue or body secretion allowing the direct or indirect measurement of selenium concentration by conventional methods, all the thresholds having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

26. Compound composition for prediction of increased prostate cancer risk among carriers of the constitutional genotypes CC of the SNP rs1050450 in the gene GPX1 , non-CC of SNP rs713041 in the gene GPX4, GG of the SNP rs1 139793 in the gene TXNRD2, or GG of the SNP rs5845 in the gene SEP15, combined with a concentration of selenium in a biological sample from the analyzed subject below 70 pg/l, comprising at least two different oligonucleotides allowing the amplification of a region of the genome of said human subject compound carrier of mentioned constitutional mutations, or another alteration linked or sharing haplotype with the former ones, and allowing its identification by conventional methods, as well as the corresponding reagents for preprocessing at least one type of tissue or body secretion allowing the direct or indirect measurement of selenium concentration by conventional methods, the threshold having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

27. Compound composition for prediction of increased lung or larynx cancer risk among subjects with a concentration of selenium in a biological sample from the analyzed subject below 70 μ9/Ι, comprising the corresponding reagents for preprocessing at least one type of tissue or body secretion allowing the direct or indirect measurement of selenium concentration by conventional methods, the threshold having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

28. Compound composition for prediction of a risk of lung cancer which is near to zero among subjects with a concentration of selenium in a biological sample from the analyzed subject between 90 μg/l and 1 10 μg/l, comprising the corresponding reagents for preprocessing at least one type of tissue or body secretion allowing the direct or indirect measurement of selenium concentration by conventional methods, the threshold having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.

29. Compound composition for prediction of a risk of breast cancer which is near to zero among subjects with a concentration of selenium in a biological sample from analyzed subjects between 40 μ9 Ι and 59 μg/l that are also compound carriers of the genotypes GPX1 rs1050450 non-CC, GPX4 rs713041 non-CC, TXNRD2 rs1 139793 non-GG and SEP15 rs5845 GG, comprising the corresponding reagents for preprocessing at least one type of tissue or body secretion allowing the direct or indirect measurement of selenium concentration by conventional methods, the threshold having a standard deviation of 5, in blood serum or plasma, or the corresponding proportional values of selenium concentration in whatever other tissue or body secretion analyzed.