TECHNICAL FIELD

The technical field of the invention concerns analysis of genetic biomarkers to enable early detection of metastatic hepatic cancer stemming from colorectal tumors and to determine a treatment regime based on analysis of said biomarkers

The invention solves the problem of detecting hidden signs of hepatic metastasis in colorectal cancer patients in order to provide more effective treatment before objective clinical symptoms otherwise become apparent. A rule-based method and system of the invention indicate metastasis or lack thereof based on divergent liver prometastatic gene expression levels of specific genes that occur in the tumor microenvironment of the patient’s liver. Alternatively, protein signatures of involved genes rather than the genes themselves may be used to indicate metastasis where such protein signatures are derived from an analysis of the patient’s blood serum or plasma.

BACKGROUND

The liver is a major metastasis-susceptible site in the human body and a majority of patients with hepatic metastases die from the disease regardless of treatment. Hepatic metastasis is conventionally detected by imaging techniques, which typically cannot detect cancer lesions less than about five to seven millimeters in diameter. By the time the lesion reaches that size, however, millions or even billion of cancer cells have already spread throughout the patient’s body and little if anything can be done to abate the disease. Thus, the average CRC (ColoRectal Cancer) patient dies within two to five years, more or less, of conventional cancer detection.

A focal liver lesion in the liver, for example, more likely represents a metastatic tumor than a primary malignancy. In addition, a majority of patients develop multiple liver metastases in both lobes that vary in diameter suggests that cancer cell seeding and growth occur in independent and separate episodes. Numerous experimental and clinical studies have focused on factors that regulate metastasis recurrence in the liver. At present, however, genetic and phenotypic properties of specific cancer cells able to implant and grow in the liver have not yet been established for any primary tumor type. Neither the contribution of the patient’s genetic expressions nor the patient’s physiologic background to the incidence and progression of hepatic metastases is presently understood.

Liver metastasis development is promoted by a broad range of organ-specific

prometastatic factors, including cancer cell growth-stimulating factors, tumor stromal cell-stimulating factors, tumor angiogenesis-stimulating factors, and hepatic immune suppressant factors, among others. The experimental identification of some of these factors made it possible to understand certain hepatic metastasis development inhibition. However, it is not clear if these diverse factors have a control role during human liver metastasis disease. Neither is it clear if such factors have already occurred prior to CRC development (as a constitutive predisposition to liver metastasis), if they were induced by certain comorbidities and therapies, and/or if they were induced remotely by CRC cells endowed with this prometastatic feature.

Therefore, it is plausible that the liver might acquire a prometastatic condition

concomitant with CRC progression and that such condition might be activated by either tumor-dependent or tumor-independent factors. Either way, these factors may activate remotely a“Liver Prometastatic Reaction” (LPR) favorable for the hepatic colonization of circulating cancer cells, and they should be designated as LPR-stimulating factors (LPR-SF), irrespective of their nature.

Production of LPR-SF and their delivery into the mesenteric vein circulation may be upregulated in CRC cells (including tumor and non-tumor stromal cell lineages) by tumor site-dependent factors (as for example, colonic inflammation, tumor hypoxia and mechanical stress, diet, gut microflora-derived bacterial factors, etc.), but also by factors from other intraperitoneal organs whose venous blood is draining into the mesenteric veins (spleen, pancreas, visceral fat, etc ). In addition, they may also be activated by systemic factors reaching the liver through the hepatic artery.

Once developed, the LPR may in turn lead to the hepatic cell production of Metastasis- Stimulating Factors (LPR-derived MSF) of potential interest as targets for anti metastatic therapy. Their specific hepatic cell origin and their nature and effects on both cancer and stromal cells are now being recently understood. For example, LPR-derived MSF upregulated CRC cell expression of certain liver metastasis-specific genes, not expressed at primary CRC, suggesting they may also represent liver metastasis-specific molecular targets for therapy.

Therefore, LPR-specific genes and proteins may represent clinically-valuable hepatic biomarkers for predicting a risk level and/or detecting development of hepatic CRC metastasis. In addition, LPR-derived soluble factors should leave the liver through the suprahepatic vein and therefore they should be detectable in the peripheral blood, alerting on the occurrence of LPR in a given cancer patient.

The possibility that LPR-derived MSP can regulate some of liver metastasis-associated genes suggests that the CRC prometastatic phenotype includes both liver-independent and liver-dependent metastasis-associated genes, the first occurring at the primary tumor and the second only at metastatic sites, activated by the LPR-derived MSF.

Therefore, liver-independent metastasis-associated CRC genes may have diagnostic value as prometastatic detectors or predictors at the primary tumor, while liver- dependent metastasis-associated CRC genes, which should be detectable at metastatic, but not at primary sites, may be valuable as targets for therapy. In addition, liver-independent metastasis-associated CRC genes may be involved in the CRC production of LPR-SF, which in turn would induce LPR-derived MSF further supporting hepatic metastasis development.

The inventor hereof has discovered that development of hepatic metastases is associated with an aberrant tissue-reconstitution process that results from bidirectional reciprocal effects between cancer cells and resident hepatic cells. On one hand, cancer cells and their soluble and exosomal proteins regulate gene expression in hepatic cells residing in, or infiltrating into, various sites of metastases. At these sites, cancer cells exert selective pressures on hepatic cells thereby shaping their functional phenotypes. Conversely, constituents of the liver microenvironment may also regulate gene expressions in the cancer cells thereby controlling their fate and determining their ability to progress towards metastatic formation.

Additionally, there are pathophysiological processes such as aberrant hepatic regeneration, inflammation and fibrosis that change the hepatic microenvironment and notably affect development of metastases. Therefore, tumor microenvironment regulating hepatic metastasis in a given patient consists of structural and functional factors resulting from both hepatic-cancer cell interactions and previous or concurrent hepatic diseases.

Neoplasms from right and left colon and rectum frequently metastasize to the liver. At a transcriptional level, hepatic metastasis development is in part associated with marked changes in gene expression of colorectal cancer cells that may originate in a primary tumor. Other prometastatic changes occur in the liver and are regulated by hepatic cells, which represent a new microenvironment for metastatic colon cancer cells. In addition, hepatic parenchymal and non-parenchymal cell functions are also affected by both cancer cell-derived factors and various systemic pathophysiological factors of a patient having CRC.

Liver and gastrointestinal tract physiology and pathology are interrelated. For example, gallstones (cholelithiasis) and cholecystectomy are related to digestive system cancer through inflammation, altered bile flux, and changes in metabolic hormone levels. More importantly, it has been established that a statistically significant risk of colorectal cancer follows cholelithiasis (Lee et al, 2016; Gosavi et al, 2017). Similarly, fatty liver, which is a hepatic manifestation of metabolic syndrome, is a well-known risk factor for CRC (Barbois et al, 2017). If hepatic gene expression disorders precede CRC occurrence, early biomarkers of CRC risk and development may be assessed.

In the past two decades, a growing amount of data has been reported suggesting that carcinomas of the right and left colon should be considered as different tumor entities. Right-sided colon cancers (RCC) and left-sided colon cancers (LCC) have different embryological origins, and various differences exist between them. Tumor location is associated with prognosis in colorectal cancer patients, and those with RCC have a significantly worse prognosis than those with LCC (Yahigi et al 2016). RCC should be treated distinctively from LCC (Zhao et al, 2017), and the establishment of standardized management for colon cancer by tumor location is needed.

Characterization of genes that are differentially expressed in tu origenesis is an important step in identifying those that are intimately involved in the details of a cell's transformation from normal to cancerous, and from non-metastatic to metastatic cells. However, little is known about molecular changes that occur in key organs (as for example the liver) during the course of cancer development and its metastatic disease. While changes in the expression level of individual genes have been reported, investigation of gene expression changes that occur in the liver of patients with cancer and without cancer as provided by the present invention has not been previously known or documented.

In brief summary, there exists a need in the art for the identification of new CRC disease-associated hepatic genes and resultant proteins as molecular biomarkers to, among other things, to (i) monitor and assess the pathogenic contribution of liver to CRC and hepatic CRC metastasis development; (ii) identify and/or screen candidate cancer patients suitable for liver metastasis-specific therapies at the cancer

microgenesis stage rather than by imaging; and (iii) discover and/or screen

pharmaceutical cellular and molecular compositions targeting those liver genes with CRC and CRC metastasis-stimulating activities in patients with colorectal cancer or CRC risk thereof.

These and other needs are met by various aspects of the present invention.

SUMMARY OF THE INVENTION Technical Problem

The invention addresses problems of (i) identifying colorectal cancer patients having a risk of developing metastatic hepatic cancer years before clinical symptoms arise, (ii) detecting occult metastatic hepatic cancer before objective clinical symptoms become known using conventional techniques, e.g., by imaging and (iii) determining an appropriate therapeutic treatment of colorectal cancer according to primary tumor location.

Solution to the Problem

According to an aspect of the invention, there is provided a rule-based method (or a manually operated or digital device utilizing said method) for early detection of metastatic cancer in the tumor microenvironment of a patient having a colorectal tumor, where the method comprises (a) obtaining patient data comprising a plurality of genetic expression levels of genes from tumor unaffected hepatic tissue of the patient where the genes include selected ones of genes from group 1 genes (PRDX4, CRP, ID1 , MT 1 E, TNFSF14, MRC1 , ICAM1 , IL18, IL10, TFN) and group 2 genes (NGF, EPHA1 ,

ERBB2IP, SDC1 , COL18A1 , KNG1 , ADH1 B, CYP2E1 ), (b) obtaining references values for said group 1 and group 2 genes wherein the reference values respectively indicate expression levels of corresponding genes of a person free of colorectal tumors, (c) analyzing the patient data and reference values to interpret the significance of overexpression of selected group 1 patient data genes relative to group 1 reference values and/or significance of underexpression of selected group 2 patient data genes relative to group 2 reference values; and (d) producing an indication confirming or dispelling hepatic metastasis according to interpretation of significance of overexpression and/or underexpression of group 1 and/or group 2 patient data genes.

Other aspects of the invention include wherein a processor utilizing said method assigns a weight to respective genes according to predetermined significance of indication of hepatic metastasis; wherein the processor utilizes protein signatures of said genes to indicate overexpressions or underexpressions thereof; wherein the processor utilizes correlation, clustering and/or heatmaps to interpret significance of said gene expressions; wherein the processor additionally utilizes selected ones of group 3 genes (HP, VTN, RPS27, RPL23, GAPDH, TXN, VEGFA, CEACAM1 , IGF1 , TGFB1 , DDR2, NOS2, and BMP7) to validate significance of gene expressions; wherein selected ones of genes comprise statistically significant ones of group 1 , group 2 and group 3 genes; wherein the processor comprises a digital microprocessor instead of a manually operated device; and wherein the patient data and reference values comprise cycle count information derived from polymerase chain reactions,

A further aspect of the invention includes wherein a processor comprises a series of software modules to execute program instructions to perform at least two of (i) a partial least squares-discrim inant analysis of said selected genes of a patient with and a patient without CRC, (ii) a clustering analysis of selected genes in a patient with and a patient without CRC, (iii) a Spearman’s correlation analysis of selected genes to assess new and lost correlations of selected genes in respective categories in a patient with and a patient without CRC, (iv) a hierarchical clustering analysis of selected genes in a patient with and a patient without CRC, (v) distribution analysis of selected gene expression levels for genes in respective functional categories in a patient with and a patient without CRC, and (vi) determining high-low expression levels of selected genes in functional categories indicative of location of primary tumors in a patient with and a patient without CRC.

A yet further aspect of the invention includes wherein a processor detects (A) a left-side colon location of a CRC tumor according to overexpressed levels of statistically significant ones of (i) proinflammatory genes IL18, ID1 , TNF, TNFSF14, AND ADH1 B,

(ii) immune regulation genes ICAM1 , MRC1 , KNG1 , and SDC1 , and/or (iii) metabolic bioprotection genes PRXD4, MTE1 , P, NOS2 and CRP; (B) a rectal location of said CRC tumor according to underexpressed levels of statistically significant ones of (i)

IL18, ID1 , VEGFA, TNFSF14, ADHI B and CYP2E1 proinflammatory genes, (ii) ICAM1 , KNG1 , SDC1 AND BMP7 immuno regulation genes, and (iii) GAPDH, TXN, MTE1 , HP, CR and ERBB2IP metabolic bioprotection genes; and (C) a right side colon location of said CRC tumor according to (i) high expression level of at least one of ID1 and TNF proinflammatory genes, (ii) low expression level of at least one of ADH18 and CYPE1 proinflammatory genes, (iii) high expression level of at least one of immune regulation genes IL10, MRC1 and BMP7, (iv) low expression level of at least one of immune regulation genes KNG1 and SDC1 , and (v) low expression level of at least one of VTN and NGF fibrogenic and regeneration genes.

The invention further comprise a kit utilizing said method to produce said indication confirming or dispelling metastatic cancer where the kit comprises (a) instructions for performing gene assessments and (b) either a low density genetic expression array of reagents for PCR replication/detection or a low density protein array of antibodies for hybridization with patient’s blood serum/plasma. The kit may comprise a low density genetic expression array for measuring genetic expressions of group 1 genes (PRDX4, CRP, ID1 , MT1 E, TNFSF14, MRC1 , ICAM1 , IL18, IL10, TFN), group 2 genes (NGF, EPHA1 , ERBB2IP, SDC1 , COL18A1 , KNG1 , ADH1 B, CYP2E1 ), and group 3 genes (HP, VTN, RPS27, RPL23, GAPDH, TXN, VEGFA, CEACAM1 , IGF1 , TGFB1 , DDR2, NOS2, and BMP7) and instructions for using the genetic expression array. The kit may also comprise low density protein array for measuring protein signatures of group 1 genes (PRDX4, CRP, ID1 , MT1 E, TNFSF14, MRC1 , ICAM1 , IL18, IL10, TFN), group 2 genes (NGF, EPHA1 , ERBB2IP, SDC1 , COL18A1 , KNG1 , ADH1 B, CYP2E1 ), and group 3 genes (HP, VTN, RPS27, RPL23, GAPDH, TXN, VEGFA, CEACAM1 , IGF1 , TGFB1 , DDR2, NOS2, and BMP7) and (b) instructions for using the protein array. The kit either includes a low density genetic expression array of reagents for PCR

replication/detection or a low density protein array of antibodies for hybridization with patient’s blood serum/plasma.

Advantages of Effects of Invention

The invention enables accurate and reliable detection of prometastatic hepatic cancer in the tumor microenvironment at a very early stage, which enables administration of therapeutic remedies prior to development of hard-to-treat invasive metastatic cancer.

The invention also enables a determination of the most appropriate treatment according to location of primary colorectal tumor. BRIEF DESCRIPTION OF THE DRAWINGS

Figs. 1 A and 1 B show comparative transcriptomic analysis among metastatic CRC tissue, tumor-unaffected hepatic tissue and peripheral blood mononuclear cells from stage IV patients with CRC having systemic metastasis disease where Fig. 1A illustrates hybridization between RNA from metastatic CRC tissue and tumor-unaffected hepatic tissue and Fig. 1 B illustrates hybridization between RNA from metastatic CRC tissue and peripheral blood mononuclear cells.

Fig. 1 C shows a combination of Figs. 1A and 1 B in a single analysis diagram.

Fig. 1 D is a Venn diagram showing overlapping sets of the number of genes for each of the three samples where CMN indicates mononuclear cells, M represent metastatic cells, and H represents tumor unaffected hepatic tissue.

Fig. 2 shows a logarithmic scale representation of the relative quantification (RQ) values of the liver prometastatic gene expressions in CRC patients with respect to same values in patients without CRC.

Fig. 3 shows divergent liver prometastatic gene expression patterns of involved genes In patients with and without CRC.

Figs. 4A-4D respectively show distribution of liver prometastatic gene high-expressing patients by functional categories (proinflammatory genes (Fig. 4A), immune-regulation genes (Fig. 4B), metabolic bioprotection genes (Fig. 4C) and fibrogenic/regeneration genes (Fig. 4D)) relative to tumor location (i.e., rectum, left side, and right side) in patients with CRC.

Figs. 5A-5H show comparisons between proinflammatory gene expression levels in liver from patients with and without CRC.

Figs. 6A-8I show comparisons between immuno-regulation gene expression levels in liver from patients with and without CRC.

Figs. 7A-7H show comparisons between metabolic bioprotection gene expression levels in liver from patients with and without CRC.

Figs. 8A-8F show comparisons between Fibrogenic and Regeneration gene expression levels in liver from patients with and without CRC. Fig. 9 show comparison of overall expression profiles across samples from patients with and without CRC via a Principal Component Analyses (PCA) of gene expressions in patients (P1 -P51 ) with and in patients (C52-C72) without CRC where a first principal component (Dim 1 ) reveals a 27.18% of the variation and a second (Dim 2) reveals a 15.65% variation that separated most of patients with CRC from patients without CRC.

Figs. 10A and 10B show Partial Least Squares-Discriminant Analysis (PLS-DA) intended to discriminate patients with and without CRC based on their hepatic expression level of liver prometastatic genes (in patients (P1 -P51 ) with and in patients (C52-C72) without CRC) where elliptical shapes of Fig. 10A adopted by lines define the position coordinates of included patients (C indicates patients without CRC and P indicates patients with CRC). In this case, the discriminatory capacity was associated with the first component in the analysis. In Fig. 10B, position coordinates of liver prometastatic genes are plotted in correlation circles whose diameters define the influence of genes in the prediction of the class of patient where in this case metabolic bioprotection and fibrogenic/regeneration genes are in the smaller circle indicating that their expression levels had less ability to predict the patient's class than immune protection and proinflammatory genes, mainly located in the large correlation circle, and therefore had a greater predictive capacity to discriminate patients with and without CRC. Figs. 11 A and 11 B show heatmaps of clustering data for patients with and without CRC according their liver prometastatic gene expression patterns based on AACt ratios (where the heat map indicates Euclidean Distance, Clustering, Average Linkage in a Globa Map Type)

Fig. 12 shows Spearman’s Correlation patterns among liver prometastatic genes in patients with and without CRC where PI indicates proinflammatory genes, MB indicates Metabolic Bioprotection genes, IR indicates Immune Regulation genes, and FR indicates Fibrogenic and Regeneration genes.

Figs. 13A-13D show hierarchical clustering was performed based on Pearson’s correlation of Euclidean Distance among the genes and gene dusters, and the results presented as a dendrogram plot in order to define the transcriptional structure of prometastatic genes in hepatic biopsies from patients without and with CRC (Figs. 13A and 13C) and where a cluster including PRX4, SDC1 , VEGFA, ID1 and CRP genes defined a main change in the hepatic transcriptional structure between patients without and with CRC (Figs 13B and 13D).

Fig. 14 is an exemplary functional block diagram of an apparatus to automate detection of hepatic metastasis based on inputs of gene expression or protein signatures thereof supplied by a user or by measurement/testing equipment where functions thereof may be carried out by programming instructions of a general purpose processor.

DESCRIPTION OF EMBODIMENTS

Disclosed herein are procedures and a device to detect occult CRC and liver metastasis and recurrence (i.e., a complementary diagnostic test) to identify candidate patients reasonably suitable to receive liver metastasis-specific therapies (a companion diagnostic test). The device uses, among other things, a series of mathematical, correlation and statistical analysis techniques to examine, compare and analyze relationships between and among expression levels of uniquely identified genes of hepatic tissues of patients with and without CRC. The invention includes utilization of a data processing device to automate gene analyses presented herein in order to provide a computer-determined output or result for diagnostic and/or treatment guidance to health care practitioners.

Fig. 1 A shows a comparative transcriptomic analysis between metastatic CRC tissue and tumor-unaffected hepatic tissue from Stage IV patients with CRC having systemic metastasis disease. Fig. 1 B shows a comparative transcriptomic analysis between metastatic CRC tissue and peripheral blood mononuclear cells from Stage IV patients with CRC having systemic metastasis disease. Fig. 1 C combines that results of Figs.

1 A and 1 B, while Fig. 1 D shows the data in set and subset relation. These relationships help identify specific prometastatic genes used in the analysis.

More specifically, Fig. 1A shows hybridization between RNA from metastatic CRC tissue and tumor-unaffected hepatic tissue; Fig. 1 B shows hybridization between RNA from netastatic CRC tissue and peripheral blood mononuclear ceils; Fig. 1 C shows a combination of Figs 1 A and 1 B; and Fig 1 D shows the number of genes for each sample compared and illustrated using a Venn diagram where“H” represents hepatic tissue,“M” represents metastatic tissue, and“CMN” represents mononuclear blood cells. According to the analysis described in connection with Figs. 1A, 1 B, 1 C and 1 D, Table 1 below lists 122 genes whose expression levels were more than two-fold-upregu!ated in tumor-unaffected hepatic tissue compared to the expression levels in metastatic tissue and peripheral blood mononuclear cells from Stage IV patients with CRC having systemic metastasis disease.

In bold are twenty-one genes whose expression levels were upregulated in liver parenchymal and non-parenchyma! sinusoidal cells given the conditioned medium from cultured CRC cells (HT-29 CRC cell line). This gene subset was selected for further analysis.

Table 1

Table 2 below lists twenty-eight genes whose expression levels were more than two- fold-downreguiated in tumor-unaffected hepatic tissue compared to the expression in metastatic tissue and peripheral blood mononuclear cells from Stage IV patients with CRC having systemic metastasis disease. In bold are ten genes whose expression levels were downregulated in liver parenchymal and non-parenchymai sinusoidal cells given the conditioned medium from cultured CRC cells (HT-29 CRC ceil line). This gene subset was also selected for further analysis. Table 2

Table 3 below shows liver prometastatic gene families (Inflammatory, Immune

Regulation, Metabolic Bioprotection, and Fibrogenic Regeneration) of the thirty-one, two-fold upregu!ated and two-fo!d down-regulated genes of Tables 1 and 2 whose altered expression level in tumor-unaffected hepatic tissue is associated with liver metastasis growth in patients with CRC. The functional gene classification activity was performed manually by accessing the Gene Ontology and PubMed databases and is based on known biopathological functions assigned individually to studied genes. Below in Table 3 are listed and sorted by functional categories these 31 liver prometastatic genes.

Table 3

A first teaching use for the present invention concerns identifying metastasis-associated genes in the tumor-unaffected hepatic tissue of Stage-IV cancer patients with metastatic CRC. As discussed in connection with Fig. 1 , comparative transcriptome profiling using RNA from hepatic CRC metastases, tumor-unaffected hepatic tissue, and peripheral mononuclear blood cells uncovered approximately 122 genes specifically over expressed and approximately 28 genes specifically under-expressed, each group being more than two-fold overexpressed or under-expressed in tumor-unaffected hepatic tissue from Stage-!V cancer patients with metastatic CRC. These genes are identified in Tables 1 and 2 above. Upregulated and downregulated gene sets were selected for further analysis. Transcriptome profiling was obtained from surgically removed liver specimens and archived biopsies of patient tissue. Table 3 shows a subset of these liver-associated genes (over-expressed and under-expressed genes) isolated according to their association with cancer-related cellular functions of inflammation, immune regulation, metabolic bioprotection and regeneration, i.e. , functional categories. Further laboratory tests were performed on this subset of liver-associated genes to categorize them according to additional prometastatic criteria including (1 ) altered expression level in tumor-unaffected hepatic tissue associated with liver metastasis growth in patients with CRC, (2) altered expression in cultured liver parenchymal and non-parenchymal cells exposed to soluble factors from cultured CRC cells, and (3) altered expression associated with experimental hepatic colonization and growth of circulating CRC ceils in animal models of CRC.

Table 4 below shows actual clinical data taken from forty-five patients (29 patients with CRC and 16 without CRC) that were included in the study on the expression pattern of liver prometastatic genes in hepatic biopsies from patients with and without CRC where TNM indicates tumor node metastasis stage.

Table 4

C! icat Parameters Patients with CRC Patients without CRC

No. % No. %

Gender

Female 10 34 8 50

Male 19 66 8 50

Average Age 58 57 Metabolic Syndrome 14 48 7 43

Cholelithiasis 0 0 16 100

Tumor localization

Left-sided Colon 11 39 0 0 Right-sided Colon 34 0 0 Rectum 20 0 0

Others (gastric, duodenum) 2 7 0 0

Table 5 below shows measurement data indicative of the thirty-one two-fold plus upregulated and down-regulated liver prometastatic gene expression levels under investigation in patients with and without CRC. The data shown therein represents average normalized (Ct ratio of studied gene/Gt of constitutive gene) Ct (cycle threshold) values ±SD (standard deviation) as well as mean probability values“p- values.”

Fig. 2 depicts a logarithmic scale representation of the relative quantification (RQ) values of liver prometastatic gene expressions in CRC patients with respect to same values in patients without CRC Error bars indicate maximum and minimum RQ values indicates statistically significant values where probability p<Q.05 (i.e., five percent).

Fig 3 depicts liver prometastatic gene expression patterns (color-coded in provisional application according to functional category but here, s indicates proinflam matory genes, A indicates immune-regulation genes, # indicates metabolic bioprotection genes, and Vindicates fibrogenic and regeneration genes) in patients with and without CRC. Differences between patients with and without CRC were statistically significant according to a U-Man Whitney test (p<0.05). Statistically significant genes are identified by denoted in the lower legend of Fig. 3. Advantageously, by examining the expression levels of one or more statistically significant genes in group 1 and/or group 2 of patients with CRC relative to a control or reference (i.e., gene expression levels of persons free of CRC), rather than waiting for clinical signs to become apparent by imaging or non-biochemicai changes in the microenvironment, one may detect metastatic cancer in the hepatic biochemical microenvironment to enable very early treatment and potential eradication of metastatic cancer cells. An aspect of the invention includes a complementary diagnostic test to detect“liver prometastatic reaction level and class” in patients with CRC without metastatic disease. Expression of liver prometastatic genes in hepatic tissue selected above was next studied in twenty-nine patients with CRC (at stages III and IV) and sixteen patients without CRC used as controls. Table 4 details clinical information about the patients involved in the study. Based on normalized Ct values (i.e. , cycle counts during the PCR process), Table 5 shows average values of the gene expression levels for the 31 genes involved for the 29 patients with CRC. As reflected in Fig. 3, expression levels were significantly (probability p<0.05) increased for ten liver prometastatic genes (group 1 ) and decreased in eight genes (group 2), with non-statistically significant (i.e.,

insignificant) changes in twelve to thirteen genes, when comparing patients with (29 patients) and without (16 patients) CRC. The vertical axis of Fig. 3 reflects Ct data relative to a control (i.e., reference values), or ratio of the Ct cycle count (i.e., Ct value) of the sample under examination relative to the Ct cycle count of a control or reference Ct value. Regarding group 1 genes of patients with CRC, i.e., the dotted trace, detection of an expression level signal at a lower cycle count Ct in the PCR process indicates a higher gene expression level. Detection of an expression level signal for each of the group 2 genes of patients with CRC occurred at a higher cycle count Ct as reflected in the middle portion of the dotted trace of Fig. 3. However, no statistically significant changes in gene expression levels of the latter group of twelve to thirteen genes (group 3) were detected when comparing CRC patients with and without hepatic metastases (or between CRC patients and CRC-free patients as controls) suggesting that detected liver prometastatic gene expression changes in tumor unaffected hepatic tissue nevertheless occurred in the liver of CRC patients irrespective of having or not having metastases. In addition, correlation of respective gene expression levels of group 3 genes was deemed required to validate the efficacy of the relative expression levels of group 1 and group 2 genes. In other words, without congruence of the expression levels of patients with and without CRC among group 3 genes, the conditions denoted for group 1 and group 2 genes would not be valid. According to an aspect of the present invention, an output device of a rule-based system with the aid of programming instructions may produce a plot of Ct values of genes denoted in Fig. 3, either on a graphic display device or a printed chart, in order to guide medical professional in advising patients about metastatic cancer. To produce a result, the rule-based system or device may give weight to the respective gene expression levels and/or provide selection of only certain ones of the genes deemed significant to take into consideration.

Figs. 4A-4D and Table 6 further indicate that expression levels of the various genes differ according to the anatomical location of the patient’s primary CRC, i.e., in the rectum, left-side colon or the right-side colon. Therefore, a liver prometastatic reaction occurs in the liver of patients with CRC prior to metastasis development and, in accordance with another aspect of the invention, by scoring number and intensity of gene changes according to the relationships shown in Figs. 4A-4D, the inventive system or device may provide an indication of the genesis of prometastatic hepatic cancer. It is also noted that the majority of proinflammatory (seven out of eight) and immune regulation (six out of nine) liver prometastatic genes, but only a minority of fibro- regenerative (one out of five) and metabolic bio-protective (three out genes eight) were significantly (p<0.05) changed in patients with CRC versus patients without CRC (Table 3, Table 6, and Figs. 4A-4D). This suggests that in addition to the number of changed genes, the kind of changed genes in functional terms defines the Liver Prometastatic Reaction Class in the liver of patients with CRC. Both the number and functional categories of liver prometastatic genes changed in patients with CRC may serve as a complementary diagnostic test for the quantitative assessment of liver metastasis risk and recurrence in patients with CRC, and thus, as a precursor, may form the basis of a method of detecting occult CRC subclinically in patients having no clinical symptoms of CRC at all. A processing system to receive data inputs and appropriate program instructions may be utilized to automatically output this determination on a display device or other output.

Fig. 5 shows proinf!ammatory gene expressions in liver from patients with and without CRC. Data are represented as increasing distribution of mean values. Data express normalized Ct values (Ct Ratio of studied gene/Ct of constitutive gene). The discontinuous line marks the intermediate point between the minimum and maximum ratios obtained for each gene (Y axis) for total number of analyzed samples (X axis) from patients with and without CCR.

Fig. 6 depicts immuno-regu!ation gene expressions in liver from patients with and without CRC. Data are represented as increasing distribution of mean values. Data express normalized Ct values (Ct Ratio of studied gene/Ct of constitutive gene). The discontinuous line marks the intermediate point between the minimum and maximum ratios obtained for each gene (Y axis) for total number of analyzed samples (X axis) from patients with and without CCR. Fig. 7 depicts metabolic bioprotection gene expressions in liver from patients with and without CRC. Data are represented as increasing distribution of mean values. Data express normalized Ct values (Ct Ratio of studied gene/Ct of constitutive gene). The discontinuous line marks the intermediate point between the minimum and maximum ratios obtained for each gene (Y axis) for total number of analyzed samples (X axis) from patients with and without CRC.

Fig. 8 depicts fibrogenic and regeneration gene expressions in liver from patients with and without CRC. Data are represented as increasing distribution of mean values. Data express normalized Ct values (Ct Ratio of studied gene/Ct of constitutive gene). The discontinuous line marks the intermediate point between the minimum and maximum ratios obtained for each gene (Y axis) for total number of analyzed samples (X axis) from patients with and without CRC.

Based on analyses illustrated in Figs. 5-8, another aspect of the invention includes the device or apparatus performing an analysis of expression level data to provide a complementary diagnostic test to provide an alert of possible occult CRC in patients without clinical evidence of CRC but with other digestive system diseases that increase CRC risk, such as but not limited to cholelithiasis and metabolic syndrome.

Comparative distribution of gene expression levels of selected genes of studied CRC patients and their controls without CRC, by their expression of liver prometastatic genes (as indicated by the analyses shown in Figs. 5-8) demonstrate that those genes best contributing to the segregation of patients with and without CRC are Metabolic bioprotection genes PRDX4, MT 1 E, CRP and NOS2; Immune regulation genes ICAM1 , IL10 and MRC1 ; and Proinflammatory genes ID1 , TNF-a, IL18 and TNFSF14. All of these genes remarkably increased their expression levels in patients with CRC while decreased their expression levels in patients without CRC. On the contrary, immune- regulation genes SDC1 , COL18A1 and KNG1 , Proinflammatory genes EPHA1 ,

CYP2E1 , ADH1 B, and fibrogenic/regeneration gene !MGF increased their expression levels in patients without CRC while decreased their expression levels in patients with CRC, as indicated in Figs. 5-8.

Fig. 9 shows a comparison of overall expression profiles across samples from patients with and without CRC in respective Scoring Value and Loading value charts, which also may be generated and analyzed by the invention device or apparatus. Scoring chart of Fig. 9 shows results for a Principal Component Analyses (PCA) of the data, which is used to emphasize variations and reveal data patterns of gene expressions in patients with (P1 -P51 ) and without (C52-C72) CRC. The first principal component (Dimension 1 ) sets forth a 27.18% variation, whereas the second Dimension 2 sets forth a 15.65% variation. It is seen that Dimension 2 separated most of patients with CRC from patients without CRC.

A principal component analysis (PCA), multivariate regression analysis used to distinguish samples with multiple measurements was conducted, the results of which are shown in Fig. 9. Supervised discriminant analysis showed that liver prometastatic Immune regulation and proinflammatory genes were the most discriminative for patients with and without CRC.

Figs. 10A and 10B show a Partial Least Squares-Discriminant Analysis (PLS-DA) of gene expression data in respective Scoring and Loading plots, which are intended to discriminate between patients with and without CRC based on their hepatic expression levels of liver prometastatic genes. Fig. 10A depicts elliptical shapes adopted by lines that define the position coordinates of included patients (patients C52-72 without CRC and patients P1 -51 with CRC). In this case, the discriminant capacity was associated with the first component in the analysis. Fig. 10B depicts position coordinates of liver prometastatic genes plotted in correlation circles whose diameters define influence of the genes in the prediction of the class of patient. In this case, metabolic bioprotection and fibrogenic/regeneration genes are in the smaller circle, indicating that their expression levels had less ability to predict the patient's class than Immune regulation and proinflammatory genes mainly located in the large correlation circle, which indicates a greater predictive capacity of the patient class. The inventive device or apparatus may also perform these analyses and provide a responsive output for use by a health care professional.

Figs. 10A and 10B show results of a supervised discriminant analysis (i.e. , a Partial Least Squares-Discriminant Analysis, PLS-DA) to classify genes and patients by their correlation and ability to predict patients with and without CRC. The elliptical shapes adopted by lines in Fig 10A define position coordinates of included patients and show that the discriminant capacity was associated with the first component in the analysis. Next, position coordinates of studied liver prometastatic genes were plotted in correlation circles, whose diameters define the influence of the genes in the prediction of the class of patient (Fig. 10B). Studied genes were distributed in correlation circles according to their functional category and once again, metabolic bioprotection and fibrogenic/regeneration genes were in the smaller circle, indicating that their expression levels had less ability to predict the patient's class, while Immune regulation and proinflammatory genes were mainly located in the large correlation circle, indicating their greater predictive capacity of the patient class.

Figs. 11 A and 11 B shows heatmaps for clustering patients with and without CRC according their liver prometastatic gene expression patterns based on AACt ratio. Fig. 11 A shows four subgroups of patients with distinct gene expression patterns, two of them being enriched by patients with CRC and the two others by patients without CRC. Fig. 10B shows two subgroups with distinct gene expression patterns (genes with significant averages difference and significant RQ (relative quantification), enriched by either patients with or without CRC that were generated using the most discriminating genes. Some patients (noted in color green in the first and third groups) without CRC are seen to be grouped with patients with CRC suggesting they may have occult CRC (which was later confirmed by colonoscopy), while other patients with CRC (as noted in the second and fourth groups) were grouped with patients without CRC. Interesting, none of these ectopic CRC patients had hepatic metastases. Thus, according to yet another aspect of the present invention, manifestations of clustering provide a basis for early subciinical detection and pretreatment of occult CRC in patients lacking clinical symptoms.

An unsupervised hierarchical cluster analysis was performed to determine whether aggregation of genes by their expression similarity level per patient contributed to segregation of patients with and without CRC. Application of Euclidean distances between studied genes resulted in the appearance of clusters allowing the distribution of patients according to their transcriptional similarity levels. As shown in Figure 11 A, the heatmap outlined four mixed subgroups of patients with distinct gene expression patterns, two of them enriched by patients with CRC and two others by patients without CRC. A new heatmap (Fig 11 B) was constructed using genes with the highest predictive power of the class of patient, as evidenced in the PLS-DA analysis. In this case, the power of discrimination was comparable to that obtained in the previous heatmap, but in this case there was a segregation in two large mixed subgroups rather than four, both of which being enriched either in patients with CRC or without CRC. Some patients without CRC are grouped with patients with CRC suggesting they may have occult CRC (which was later confirmed by colonoscopy), while some patients with CRC are grouped with patients without CRC (none of CRC patients had hepatic metastases). Figs. 12A and 12B show Spearman’s correlation of expression levels among liver prometastatic genes in patients with and without CRC. Only statistically significant (p<0.05 or higher) correlations with coefficient Rho equal to or greater than 0.7 were considered in this analysis. Nine of the ten correlations in patients with CRC involved five bioprotective genes (HP, ERBB2IP, GAPDH, CRP, PDRX4); four of these correlation gains were produced among bioprotective genes (ERBB2IP-GAPDH;

ERBB2IP-PDRX4; CRP-GAPDH; CRP-HP), three among metabolic bioprotection and proinflammatory genes (CRP-TNFSF14; HP-TNFSF14; GAPDH-ID1 ) (CRP-NGF), immune-regulation genes (PDRX4-CEACAM1 ), as well as between proinflammatory and immune-regulation genes (TNFSF14-COL18A1 ). In contrast, eight out of the fourteen lost gene correlations in patients with CRC occurred in immune-regulation gene group (involving CEACAM1 , MRC1 , ICAM1 , IL10, BMP7 genes), of which four were lost between immune-regulation and fibrogenic/regeneration genes (ICAM1 - TGFB1 ; IL-10-NGF, CEACAM1 -NGF, MRC1 -NGF), whereas only two were lost between immune-regulation and proinflammatory genes (BMP7-TNF, CEACAM1 -TNF) and another two among immune-regulation genes (CEACAM1 -BMP7; MRC1 -BMP7). There was also a striking loss of seven correlations between proinflammatory genes and other functional categories of genes (metabolic bioprotection,

fibrogenic/regeneration and immune-regulation genes). Thus, in yet another aspect of the present invention, a statistically significant manifestation of a number and/or functional category of lost and new correlations of gene expression levels relative to such correlations in patients without CRD provides an additional rule-based system (the multigenic inter- and intra-functional group transcriptional relationships) and/or methodology to predict and treat occult CRC in patients lacking evidence of clinical symptoms.

Figs. 13, Charts A-D, show hierarchical clustering performed based on Pearson’s correlation Euclidean distance among the genes and gene clusters, and the results presented as a dendrogram plot in order to define the transcriptional structure of prometastatic genes in hepatic biopsies from patients without (Chart A) and with (Chart C) CRC. A cluster primarily including PRX4, SDC1 , VEGFA, ID1 and CRP genes define the main change in the hepatic transcriptional structure between patients without and with CRC, as shown in Charts B and D. This analysis may be automated using data processing device or equipment. Thus, according to another aspect of the invention, hierarchical clustering of PRX4, SDC1 , VEGFA, ID1 and CRP genes may form a subclinicai parameter or indicator that is utilized by a data processing device to systematically automate prediction of CRC cancer risk and provide an alert of possible occult CRC in patients without clinical evidence of CRC but with other diseases that increase CRC risk. Spearman’s correlation analysis was used to study the structure of transcriptional associations among liver prometastatic genes in patients with and without CRC, and to identify those gene correlations changing between patients with and without CRC. As shown in Fig. 12, correlations among genes from patients with CRC were strengthened in the metabolic bioprotection gene group, while they were lost among genes in the immune-regulation gene group. A hierarchical clustering (Fig. 13) was performed based on Pearson’s correlation Euclidean distances among liver prometastatic genes and their gene clusters, and represented as dendrogram plots in order to define the

transcriptional structure of prometastatic genes in hepatic biopsies from patients with and without CRC (Fig. 13, Charts A and C ). A cluster including PRX4, SDC1 , VEGFA, ID1 and CRP genes defined the main change in the hepatic transcriptional structure between patients with and without CRC (Charts C and D). Therefore, an additional feature contributing to identifying CRC-dependent gene expression changes in patients without clinical evidence of CRC is the correlation pattern among liver prometastatic genes. It was also revealed that the relationship between and among gene expression levels within functional categories differ according to location of the primary tumor in patients having CRC cancer According to another aspect of the present invention, the inventive device may determine and use this information to identify or direct a type of treatment administered to a patient. Figs. 4A, 4B, 4C and 4D, for example, show differences in distributions of liver prometastatic genes in high-expressing patients according to whether the primary tumor is located in the rectum (dotted trace), right-side (dot-dash trace) colon and left-side (dashed trace) colon. In particular, Fig. 4A shows a first relational distribution of liver prometastatic gene expressions in high-expressing patients for Proinfla matory genes (IL18, ID1 , TNF, VEFA, EPHA1 , TSFSF14, CYP2E1 and ADH1 B) according to primary tumor location in patients with CRC. Fig. 4B shows a second relational distribution of liver prometastatic gene expressions in high-expressing patients for Immune regulation genes (I CAM 1 , IL10, MRC1 , KIMG1 , SDI1 , GOL18A1 , IGF1 and MR 7) according to primary tumor location in patients with CRC. Fig. 4C shows a third relational distribution of liver prometastatic gene expressions in high- expressing patients for Metabolic Bioprotection genes (GAPDH, PRDX4, TXN, MT1 E, HP, NOS2, CRP, and ERBB2IP) according to primary tumor location in patients with CRC. Fig 4D shows a fourth relational distribution of liver prometastatic gene expressions in high-expressing patients for Fibrogenic/Regeneration genes (RPL23, DDR2, TGFB1 , VTN and IMGF) according to primary tumor location in patients with CRC.

A further aspect of the inventive apparatus includes a complementary diagnostic test to indicate a possible anatomical location of an occult CRC in patients without clinical evidence of CRC, but with other digestive system diseases increasing CRC risk, such as cholelithiasis and metabolic syndrome. As shown in Figs. 13 and 4A-4D, patients with tumors of localization in the left-side colonic area (including splenic flexure, descending colon, sigmoid colon or recto sigmoid junction) were the ones that most frequently increased the expression of liver prometastatic genes, followed by patients with right-sided tumors (including cecum, ascending colon, hepatic flexure or transverse colon), whereas patients with rectal tumors were those more frequently decreasing the expression level. The anatomical location of CRC determined the liver prometastatic gene expression pattern and the percentage of patients with high and low expression of these genes. Therefore, in accordance with this further aspect of the invention, the inventive device may develop and use these patterns suggest the possible anatomical location of an occult CRC in patients without clinical evidence of CRC, but with other digestive system diseases increasing CRC risk, such as cholelithiasis and metabolic syndrome in order to provide a basis to direct and determine a best possible treatment regime Table 6 shows distribution of liver prometastatic genes by functional categories and tumor location. Rectal Tumor Pattern is indicate by Low hepatic expression of genes from the four prometastatic gene functional categories with bigh-IL10, MRC1 and NOS2 gene expression, which suggest Immunotolerance / immunosuppression without inflammatory background and possible beneficial effects of immunotherapy in metastasis prevention. Left-sided colonic Tumor Pattern (including CRC within the splenic flexure, descending colon, sigmoid colon or recto sigmoid junction) is indicated by High hepatic expression of proinflammatory, immune regulation and metabolic bioprotection genes, with drop in BMP7 and NGF gene expression, which suggests very high-risk prometastatic microenvironment and possible beneficial effects of anti inflammatory therapies in metastasis prevention. Right-sided colonic Tumor Pattern (including primary CRC in the cecum, ascending colon, hepatic flexure or transverse colon) is indicated by a slight increase of proinflammatory and immune regulation gene expression with ADH1 B, SDC1 and VT gene expression decrease, which suggests slight immunotolerance / immunosuppression under inflammatory conditions and possible beneficial effect of anti-inflammatory therapies in metastasis prevention.

According to yet another aspect of the invention, the apparatus may perform an analysis to determine a treatment regime in accordance with high-low gene expression levels of genes within respective functional categories and anatomic location of the tumor along the colonic tract. Such a processing device or apparatus may provide such

determination in an automated diagnostic and treatment system.

Personalized treatment of patients based on a multiplex of molecular biomarkers defining precise functional features of cancer that may strongly increase the efficacy of the chosen therapies. In this study, the analysis of liver prometastatic gene functional categories by anatomical location of the CRC identified three distinct functional patterns with therapeutic implications (Table 6 and Fig. 4). Rectal Tumor Pattern was indicated by Low hepatic expression of genes from the four prometastatic gene functional categories with high-IL10, MRC1 and NOS2 gene expression, which suggest

Immunotolerance / immunosuppression without inflammatory background and possible beneficial effects of immunotherapy in metastasis prevention. Left-sided colonic Tumor Pattern is indicated by High hepatic expression of proinflammatory, immune regulation and metabolic bioprotection genes, with drop in BMP7 and NGF gene expression, which suggests very high-risk prometastatic microenvironment and possible beneficial effects of anti-inflammatory therapies in metastasis prevention. Right-sided colonic Tumor Pattern was indicated by Slight increase of proinflammatory and immune regulation gene expression with ADH1 B, SDC1 and VT gene expression decrease, which suggests slight Immunotolerance / immunosuppression under inflammatory conditions and possible beneficial effect of anti-inflammatory therapies in metastasis prevention.

Fig. 14 shows an exemplary apparatus that processes gene expression data to determine CRC risks and/or to determine a proposed treatment according to the present invention where input 10 receives gene expression levels (i.e., Ct values is as control levels) of patients without CRC and input 12 receives gene expression level of a patient under examination who may or may not have a CRC risk. Module 14 performs clustering analysis, module 16 performs a partial least squares analysis, module 18 performs a Spearman’s or other correlation analysis, module 20 assess over-expressed and under-expressed gene expression levels (i.e , high-low express levels), module 22 performs a distribution analysis, and module 24 performs other analysis indicative of a divergent gene expression levels of the patient under examination. The rule-based device processes gene express levels in a series of program modules 14, 16, 18, 20, 22 and 23 according to the respective analysis. Module 14, for example, executes program instructions to perform a clustering analysis of expression levels of the patient under examination from input 12 against control levels supplied from input 10.

An initial series of computational results of expression level processing from respective modules 14-24 may be weighted by predetermined weighting factors 30-40 according to their significance. For example, if different weights are to be assigned to computational results of the respective modules, results of each module may be assigned a weighting factor between 0.5 and 1.0 according to their importance. Such weighting factors may be assigned by a testing laboratory, researcher, medical practitioner, or may simply predetermined as a fixed value. Control or reference values against which patient data is compared may be fixed, or determined impromptu when obtaining patient data. A clustering analysis may be assigned a weighting factor higher or lower than other that other modules. After applying any such weighting factor, the results from modules 14-24 are then supplied to a rule-based engine 50 to produce final results taking info consideration all gene expression data, which then products an output at 60

deterministic of occult CRC, CRC risk factors, tumor location, etc.

The written description, drawing figures, tables and charts presented herein are not intended to limit the scope of the invention but merely provide an illustration of the core concepts and embodiments that may be implemented to carry out the teachings set forth herein. Based on these teachings, persons skilled in the art may devise alternative embodiments or modify the illustrated embodiments without departing from the scope of the invention. Accordingly, the scope of invention is defined by the appended claims rather than by the description or illustrated embodiments.