DIAGNOSIS AND PROGNOSIS OF BREAST CANCER PATIENTS

The present invention relates to the identification of marker genes useful in the diagnosis and prognosis of breast cancer. More particularly, the invention relates to the identification of a set of marker genes associated with breast cancer, a set of marker genes differentially expressed in estrogen receptor (+) versus estrogen receptor (-) tumors, a set of marker genes differentially expressed in BRCA1 versus sporadic tumors, and a set of marker genes differentially expressed in sporadic tumors from patients with good clinical prognosis (i.e., metastasis- or disease-free >5 years) versus patients with poor clinical prognosis (i.e., metastasis- or disease-free <5 years). For each of the marker sets above, the invention further relates to methods of distinguishing the breast cancer-related conditions. Further described are methods for determining the course of treatment of a patient with breast cancer.

The increased number of cancer cases reported in the United States, and, indeed, around the world, is a major concern. Currently there are only a handful of treatments available for specific types of cancer, and these provide no guarantee of success. In order to be most effective, these treatments require not only an early detection of the malignancy, but a reliable assessment of the severity of the malignancy.

The incidence of breast cancer, a leading cause of death in women, has been gradually increasing in the United States over the last thirty years. Its cumulative risk is relatively high; 1 in 8 women are expected to develop some type of breast cancer by age 85 in the United States. In fact, breast cancer is the most common cancer in women and the second most common cause of cancer death in the United States. In 1997, it was estimated that 181,000 new cases were reported in the U.S., and that 44,000 people would die of breast cancer (Parker et a/., CA Cancer J. Clin. 47:5-27 (1997); Chu et al., J. Nat. Cancer Inst. 88:1571-1579 (1996)). While mechanism of tumorigenesis for most breast carcinomas is largely unknown, there are genetic factors that can predispose some women to developing breast cancer (Miki et al., Science, 266:66-71(1994)). The discovery and characterization of BRCA1 and BRCA2 has recently expanded our knowledge of genetic factors which can contribute to familial breast cancer. Germ-line mutations within these two loci are associated with a 50 to 85% lifetime risk of breast and/or ovarian cancer (Casey, Curr. Opin. Oncol. 9:88-93 (1997); Marcus et al., Cancer 77:697-709 (1996)). Only about 5% to 10% of breast cancers are associated with breast cancer susceptibility genes, BRCA1 and BRCA2. The cumulative lifetime risk of breast cancer for women who carry the mutant BRCA1 is predicted to be approximately 92%, while the cumulative lifetime risk for the non-canier majority is estimated to be approximately 10%. BRCA1 is a tumor suppressor gene that is involved in DNA repair anc cell cycle control, which are both important for the maintenance of genomic stability. More than 90% of all mutations reported so far result in a premature truncation of the protein product with abnormal or abolished function. The histology of breast cancer in BRCA1 mutation carriers differs from that in sporadic cases, but mutation analysis is the only way to find the carrier. Like BRCA1, BRCA2 is involved in the development of breast cancer, and like BRC41 plays a role in DNA repair. However, unlike BRCA1, it is not involved in ovarian cancer.

Molecular markers to discriminate between tumor types are known in the art. Perou CM, et al. (Nature 2000 406:747-752) describes molecular portraits of human breast tumors. A subset of genes was identified, wherein the variation in expression differed more between different tumors than between paired samples from the same tumor. Alizadeh AA, et al. (Nature 2000 403:503-511) describes two distinct forms of diffuse large B-cell lymphoma (DLBCL) identified based on gene expression pattern. A subset of genes was identified as being selectively expressed in one of the two forms of DLBCL. Perou CM, et al. (PNAS 1999 96:9212-9217) describes distinctive gene expression patterns in human mammary epithelial cells and breast cancers. In response to a set of experimental perturbations a subset of genes was identified that exhibited differential expression in human mammary epithelial cells. Khan J, et al. (Nature Medicine 2001 7:673-679) describes a method of classification and diagnostic prediction of cancers using gene expression profiling and artificial neural networks. A subset of genes was used to classify small, round blue cell tumor samples into diagnostic categories. Hedenfalk I, et al. (New Eng J Med 2001 344:539.548) describes gene expression profiles in hereditary breast cancer and identified a subset of genes, which differentiated among BRCA1 mutant tumors, BRCA2 mutant tumors, and sporadic tumors from breast cancer tissue. In contrast to these documents that disclose methods for differentiating between tumor types, the present invention provides methods for classifying individuals with breast cancer as having a good or poor prognosis.

Other genes have been linked to breast cancer, for example c-erb-2 (HER2) and p53 (Beenken et al., Ann. Surg. 233(5):630-638 (2001). Overexpression of c-erb-2 (HER2) and p53 have been correlated with poor prognosis (Rudolph et al., Hum. Pathol. 32(3):311-319 (2001), as has been aberrant expression products of mdm2 (Lukas et al., Cancer Res. 61(7):3212-3219 (2001) and cyclin 1 and p27 (Porter & Roberts, International Publication WO98/33450, published August 6, 1998). However, no other clinically useful markers consistently associated with breast cancer have been identified.

Sporadic tumors, those not currently associated with a known germline mutation, constitute the majority of breast cancers. It is also likely that other, non-genetic factors also have a significant effect on the etiology of the disease. Regardless of the cancer's origin, breast cancer morbidity and mortality increases significantly if it is not detected early in its progression. Thus, considerable effort has focused on the early detection of cellular transformation and tumor formation in breast tissue.

A marker-based approach to tumor identification and characterization promises improved diagnostic and prognostic reliability. Typically, the diagnosis of breast cancer requires histopathological proof of the presence of the tumor. In addition to diagnosis, histopathological examinations also provide information about prognosis and selection of treatment regimens. Prognosis may also be established based upon clinical parameters such as tumor size, tumor grade, the age of the patient, and lymph node metastasis.

Diagnosis and/or prognosis may be determined to varying degrees of effectiveness by direct examination of the outside of the breast, or through mammography or other X-ray imaging methods (Jatoi, Am. J. Surg. 177:518-524 (1999)). The latter approach is not without considerable cost, however. Every time a mammogram is taken, the patient incurs a small risk of having a breast tumor induced by the ionizing properties of the radiation used during the test. In addition, the process is expensive and the subjective interpretations of a technician can lead to imprecision. For example, one study showed major clinical disagreements for about one-third of a set of mammograms that were interpreted individually by a surveyed group of radiologists. Moreover, many women find that undergoing a mammogram is a painful experience. Accordingly, the National Cancer Institute has not recommended mammograms for women under fifty years of age, since this group is not as likely to develop breast cancers as are older women. It is compelling to note, however, that while only about 22% of breast cancers occur in women under fifty, data suggests that breast cancer is more aggressive in pre-menopausal women.

In clinical practice, accurate diagnosis of various subtypes of breast cancer is important because treatment options, prognosis, and the likelihood of therapeutic response all vary broadly depending on the diagnosis. Accurate prognosis, or determination of distant metastasis-free survival could allow the oncologist to tailor the administration of adjuvant chemotherapy, with women having poorer prognoses being given the most aggressive treatment. Furthermore, accurate prediction of poor prognosis would greatly impact clinical trials for new breast cancer therapies, because potential study patients could then be stratified according to prognosis. Trials could then be limited to patients having poor prognosis, in turn making it easier to discern if an experimental therapy is efficacious.

To date, no set of satisfactory predictors for prognosis based on the clinical information alone has been identified. The detection of BRCA1 or BRCA2 mutations represents a step towards the design of therapies to better control and prevent the appearance of these tumors. However, there is no equivalent means for the diagnosis of patients with sporadic tumors, the most common type of breast cancer tumor, nor is there a means of differentiating subtypes of breast cancer.

The invention provides gene marker sets that distinguish various types and subtypes of breast cancer, and methods of use therefore. The invention provides a method for classifying an individual afflicted with breast cancer as having a good prognosis or a poor prognosis, wherein said individual is a human, wherein said good prognosis indicates that said individual is expected to have no distant metastases within five years of initial diagnosis of breast cancer, and wherein said poor prognosis indicates that said individual is expected to have distant metastases within five years of initial diagnosis of breast cancer, comprising: (ia) calculating a first classifier parameter between a first expression profile and a good prognosis template, or (ib) calculating a second classifier parameter between said first expression profile and said good prognosis template and a third classifier parameter between said first expression profile and a poor prognosis template: said first expression profile comprising the expression levels of a first plurality of genes in a cell sample taken from the individual, said good prognosis template comprising for each gene in said first plurality of genes, the average expression level of said gene in a plurality of patients having no distant metastases within five years of initial diagnosis of breast cancer: and said poor prognosis template comprising, for each gene in said first plurality of genes, the average expression level of said gene in a plurality of patients having distant metastases within five years of initial diagnosis of breast cancer: said first plurality of genes consisting of at least 5 of the genes for which markers are listed in Table 5: and (iia) classifying said individual as having said good prognosis if said first classifier parameter is above a chosen threshold or if said first expression profile is more similar to said good prognosis template than to said poor prognosis template, or (iib) classifying said individual as having said poor prognosis if said first classifier parameter is below said chosen threshold or if said first expression profile is more similar to said poor porgnosis template than to said good prognosis template. In one embodiment, a method for classifying a cell sample as ER (+) or ER (-) comprising detecting a difference in the expression of a first plurality of genes relative to a control is described, said first plurality of genes consisting of at least 5 of the genes corresponding to the markers listed in Table 1. In specific embodiments, said plurality of genes consists of at least 50, 100, 200, 500, 1000, up to 2,460 of the gene markers listed in Table 1. In another specific embodiment, said plurality of genes consists of each of the genes corresponding to the 2,460 markers listed in Table 1. In another specific embodiment, said plurality consists of the 550 markers listed in Table 2. In another specific embodiment, said control comprises nucleic acids derived from a pool of tumors from individual sporadic patients. In another specific embodiment, said detecting comprises the steps of : (a) generating an ER (+) template by hybridization of nucleic acids derived from a plurality of ER (+) patients within a plurality of sporadic patients against nucleic acids derived from a pool of tumors from individual sporadic patients; (b) generating an ER (-) template by hybridization of nucleic acids derived from a plurality of ER (-) patients within said plurality of sporadic patients against nucleic acids derived from said pool of tumors from individual sporadic patients within said plurality; (c) hybridizing nucleic acids derived from an individual sample against said pool; and (d) determining the similarity of marker gene expression in the individual sample to the ER (+) template and the ER (-) template, wherein if said expression is more similar to the ER (+) template, the sample is classified as ER (+), and if said expression is more similar to the ER (-) template, the sample is classified as ER (-).

Further described are the above methods, applied to the classification of samples as BRCA1 or sporadic. The invention provides the above methods applied to classifying patients as having good prognosis or poor prognosis. For the BRCAI/sporadic gene markers, a method may be used wherein the plurality of genes is at least 5, 20, 50, 100, 200 or 300 of the BRCAl/sporadic markers listed in Table 3. In a specific embodiment, the optimum 100 markers listed in Table 4 are used. For the prognostic markers, the invention provides that at least 5, 20, 50, 100, or 200 gene markers listed in Table 5 may be used. In a specific embodiment, the optimum 70 markers listed in Table 6 are used.

The invention further provides that markers may be combined. 5 markers. In another embodiment, at least 5 markers from Table 5 are used in conjunction with at least 5 markers from Table 3. In another embodiment, at least 5 markers from Table 1 are used in conjunction with at least 5 markers from Table 5. In another embodiment, at least 5 markers from each of Tables 1, 3, and 5 are used simultaneously.

Further described is a method for classifying a sample as ER(+) or ER(-) by calculating the similarity between the expression of at least 5 of the markers listed in Table 1 in the sample to the expression of the same markers in an ER(-) nucleic acid pool and an ER(+) nucleic acid pool, comprising the steps of: (a) labeling nucleic acids derived from a sample, with a first fluorophore to obtain a first pool of fluorophore-labeled nucleic acids; (b) labeling with a second fluorophore a first pool of nucleic acids derived from two or more ER(+) samples, and a second pool of nucleic acids derived from two or more ER(-) samples; (c) contacting said first fluorophore-labeled nucleic acid and said first pool of second fluorophore-labeled nucleic acid with said first microarray under conditions such that hybridization can occur, and contacting said first fluorophore-labeled nucleic acid and said second pool of second fluorophore-labeled nucleic acid with said second microarray under conditions such that hybridization can occur, detecting at each of a plurality of discrete loci on the first microarray a first flourescent emission signal from said first fluorophore-labeled nucleic acid and a second fluorescent emission signal from said first pool of second fluorophore-labeled genetic matter that is bound to said first microarray under said conditions, and detecting at each of the marker loci on said second microarray said first fluorescent emission signal from said first fluorophore-labeled nucleic acid and a third fluorescent emission signal from said second pool of second fluorophore-labeled nucleic acid; (d) determining the similarity of the sample to the ER(-) and ER(+) pools by comparing said first fluorescence emission signals and said second fluorescence emission signals, and said first emission signals and said third fluorescence emission signals; and (e) classifying the sample as ER(+) where the first fluorescence emission signals are more similar to said second fluorescence emission signals than to said third fluorescent emission signals, and classifying the sample as ER(-) where the first fluorescence emission signals are more similar to said third fluorescence emission signals than to said second fluorescent emission signals, wherein said similarity is defined by a statistical method. The invention further describes that the other disclosed marker sets may be used in the above method to distinguish BRCA1 from sporadic tumors, and patients with poor prognosis from patients with good prognosis.

In a specific embodiment, said similarity is calculated by determining a first sum of the differences of expression levels for each marker between said first fluorophore-labeled nucleic acid and said first pool of second fluorophore-labeled nucleic acid, and a second sum of the differences of expression levels for each marker between said first fluorophore-labeled nucleic acid and said second pool of second fluorophore-labeled nucleic acid, wherein if said first sum is greater than said second sum, the sample is classified as ER(-), and if said second sum is greater than said first sum, the sample is classified as ER(+). In another specific embodiment, said similarity is calculated by computing a first classifier parameter P₁ between an ER(+) template and the expression of said markers in said sample, and a second classifier parameter P₂ between an ER(-) template and the expression of said markers in said sample, wherein said P₁ and P₂ are calculated according to the formula: Pi=z→i•y→/‖z→i‖⋅‖y→‖, wherein z₁ and z₂ are ER(-) and ER(+) templates, respectively, and are calculated by averaging said second fluorescence emission signal for each of said markers in said first pool of second fluorophore-labeled nucleic acid and said third fluorescence emission signal for each of said markers in said second pool of second fluorophore-labeled nucleic acid, respectively, and wherein y is said first fluorescence emission signal of each of said markers in the sample to be classified as ER(+) or ER(-), wherein the expression of the markers in the sample is similar to ER(+) if P₁ < P₂, and similar to ER(-) if P₁ > P_2·

Further described is a method for identifying marker genes the expression of which is associated with a particular phenotype. Also described is method for determining a set of marker genes whose expression is associated with a particular phenotype, comprising the steps of: (a) selecting the phenotype having two or more phenotype categories; (b) identifying a plurality of genes wherein the expression of said genes is correlated or anticorrelated with one of the phenotype categories, and wherein the correlation coefficient for each gene is calculated according to the equation ρ=c→•r→/‖c→‖⋅‖r→‖ wherein c is a number representing said phenotype category and r is the logarithmic expression ratio across all the samples for each individual gene, wherein if the correlation coefficient has an absolute value of a threshold value or greater, said expression of said gene is associated with the phenotype category, and wherein said plurality of genes is a set of marker genes whose expression is associated with a particular phenotype. The threshold depends upon the number of samples used; the threshold can be calculated as 3 X 1/n-3,, where 1/n-3is the distribution width and n = the number of samples. In a specific embodiment where n = 98, said threshold value is 0.3. In a specific embodiment, said set of marker genes is validated by: (a) using a statistical method to randomize the association between said marker genes and said phenotype category, thereby creating a control correlation coefficient for each marker gene; (b) repeating step (a) one hundred or more times to develop a frequency distribution of said control correlation coefficients for each marker gene; (c) determining the number of marker genes having a control correlation coefficient of a threshold value or above, thereby creating a control marker gene set; and (d) comparing the number of control marker genes so identified to the number of marker genes, wherein if the p value of the difference between the number of marker genes and the number of control genes is less than 0.01, said set of marker genes is validated. In another specific embodiment, said set of marker genes is optimized by the method comprising: (a) rank-ordering the genes by amplitude of correlation or by significance of the correlation coefficients, and (b) selecting an arbitrary number of marker genes from the top of the rank-ordered list. The threshold value depends upon the number of samples tested.

Further described is a method for assigning a person to one of a plurality of categories in a clinical trial, comprising determining for each said person the level of expression of at least five of the prognosis markers listed in Table 6, determining therefrom whether the person has an expression pattern that correlates with a good prognosis or a poor prognosis, and assigning said person to one category in a clinical trial if said person is determined to have a good prognosis, and a different category if that person is determined to have a poor prognosis. Also described is a method for assigning a person to one of a plurality of categories in a clinical trial, where each of said categories is associated with a different phenotype, comprising determining for each said person the level of expression of at least five markers from a set of markers, wherein said set of markers includes markers associated with each of said clinical categories, determining therefrom whether the person has an expression pattern that correlates with one of the clinical categories and assigning said person to one of said categories if said person is determined to have a phenotype associated with that category.

Also described is a method of classifying a first cell or organism as having one of at least two different phenotypes, said at least two different phenotypes comprising a first phenotype and a second phenotype, said method comprising: (a) comparing the level of expression of each of a plurality of genes in a first sample from the first cell or organism to the level of expression of each of said genes, respectively, in a pooled sample from a plurality of cells or organisms, said plurality of cells or organisms comprising different cells or organisms exhibiting said at least two different phenotypes, respectively, to produce a first compared value; (b) comparing said first compared value to a second compared value, wherein said second compared value is the product of a method comprising comparing the level of expression of each of said genes in a sample from a cell or organism characterized as having said first phenotype to the level of expression of each of said genes, respectively, in said pooled sample; (c) comparing said first compared value to a third compared value, wherein said third compared value is the product of a method comprising comparing the level of expression of each of said genes in a sample from a cell or organism characterized as having said second phenotype to the level of expression of each of said genes, respectively, in said pooled sample, (d) optionally carrying out one or more times a step of comparing said first compared value to one or more additional compared values, respectively, each additional compared value being the product of a method comprising comparing the level of expression of each of said genes in a sample from a cell or organism characterized as having a phenotype different from said first and second phenotypes but included among said at least two different phenotypes, to the level of expression of each of said genes, respectively, in said pooled sample; and (e) determining to which of said second, third and, if present, one or more additional compared values, said first compared value is most similar, wherein said first cell or organism is determined to have the phenotype of the cell or organism used to produce said compared value most similar to said first compared value.

In a specific embodiment of the above method, said compared values are each ratios of the levels of expression of each of said genes. In another specific embodiment, each of said levels of expression of each of said genes in said pooled sample are normalized prior to any of said comparing steps. In another specific embodiment, normalizing said levels of expression is carried out by dividing each of said levels of expression by the median or mean level of expression of each of said genes or dividing by the mean or median level of expression of one or more housekeeping genes in said pooled sample. In a more specific embodiment, said normalized levels of expression are subjected to a log transform and said comparing steps comprise subtracting said log transform from the log of said levels of expression of each of said genes in said sample from said cell or organism. In another specific embodiment, said at least two different phenotypes are different stages of a disease or disorder. In another specific embodiment, said at least two different phenotypes are different prognoses of a disease or disorder. In yet another specific embodiment, said levels of expression of each of said genes, respectively, in said pooled sample or said levels of expression of each of said genes in a sample from said cell or organism characterized as having said first phenotype, said second phenotype, or said phenotype different from said first and second phenotypes, respectively, are stored on a computer.

Further described are microarrays comprising the disclosed marker sets. Amicroarray is described comprising at least 5 markers derived from any one of Tables 1-6, wherein at least 50% of the probes on the microarray are present in any one of Tables 1-6. In more specific embodiments, at least 60%, 70%, 80%, 90%, 95% or 98% of the probes on said microarray are present in any one of Tables 1- 6.

Further described is a microarray for distinguishing ER (+) and ER (-) cell samples comprising a positionally-addressable array of polynucleotide probes bound to a support, said polynucleotide probes comprising a plurality of polynucleotide probes of different nucleotide sequences, each of said different nucleotide sequences comprising a sequence complementary and hybridizable to a plurality of genes, said plurality consisting of at least 5 of the genes corresponding to the markers listed in Table 1 or Table 2, wherein at least 50% of the probes on the microarray are present in any one of Table 1 or Table 2. Also described is a microarray for distinguishing Brai-type and sporadic tumor-type cell samples comprising a positionally-addressable array of polynucleotide probes bound to a support, said polynucleotide probes comprising a plurality of polynucleotide probes of different nucleotide sequences, each of said different nucleotide sequences comprising a sequence complementary and hybridizable to a plurality of genes, said plurality consisting of at least 5 of the genes corresponding to the markers listed in Table 3 or Table 4, wherein at least 50% of the probes on the microarray are present in any one of Table 3 or Table 4. Further described is a microarray for distinguishing cell samples from patients having a good prognosis and cell samples from patients having a poor prognosis comprising a positionally-addressable array of polynucleotide probes bound to a support, said polynucleotide probes comprising a plurality of polynucleotide probes of different nucleotide sequences, each of said different nucleotide sequences comprising a sequence complementary and hybridizable to a plurality of genes, said plurality consisting of at least 5 of the genes corresponding to the markers listed in Table 5 or Table 6, wherein at least 50% of the probes on the microarray are present in any one of Table 5 or Table 6. Microarrays comprising at least 5, 20, 50, 100, 200, 500, 100, 1,250, 1,500, 1,750, or 2,000 of the ER-status marker genes listed in Table 1, at least 5, 20, 50, 100, 200, or 300 of the BRCA1 sporadic marker genes listed in Table 3, or at least 5, 20, 50, 100 or 200 of the prognostic marker genes listed in Table 5, in any combination, wherein at least 50%, 60%, 70%, 80%, 90%, 95% or 98% of the probes on said microarrays are present in Table 1, Table 3 and/or Table 5 are also described.

A kit for determining the ER-status of a sample, comprising at least two microarrays each comprising at least 5 of the markers listed in Table 1, and a computer system for determining the similarity of the level of nucleic acid derived from the markers listed in Table 1 in a sample to that in an ER (-) pool and an ER (+) pool, the computer system comprising a processor, and a memory encoding one or more programs coupled to the processor, wherein the one or more programs cause the processor to perform a method comprising computing the aggregate differences in expression of each marker between the sample and ER (-) pool and the aggregate differences in expression of each marker between the sample and ER (+) pool, or a method comprising determining the correlation of expression of the markers in the sample to the expression in the ER (-) and ER (+) pools, said correlation calculated according to Equation (4) is described. Kits able to distinguish BRCA1 and sporadic tumors, and samples from patients with good prognosis from samples from patients with poor prognosis, by inclusion of the appropriate marker gene sets are also described. A kit for determining whether a sample is derived from a patient having a good prognosis or a poor prognosis, comprising at least one microarray comprising probes to at least 5 of the genes corresponding to the markers listed in Table 5, and a computer readable medium having recorded thereon one or more programs for determining the similarity of the level of nucleic acid derived from the markers listed in Table 5 in a sample to that in a pool of samples derived from individuals having a good prognosis and a pool of samples derived from individuals having a good prognosis, wherein the one or more programs cause a computer to perform a method comprising computing the aggregate differences in expression of each marker between the sample and the good prognosis pool and the aggregate differences in expression of each marker between the sample and the poor prognosis pool, or a method comprising determining the correlation of expression of the markers in the sample to the expression in the good prognosis and poor prognosis pools, said correlation calculated according to Equation (3), is also described.

• FIG. 1 is a Venn-type diagram showing the overlap between the marker sets disclosed herein, including the 2,460 ER markers, the 430 BRCA1/sporadic markers, and the 231 prognosis reporters.
• FIG. 2 shows the experimental procedures for measuring differential changes in mRNA transcript abundance in breast cancer tumors used in this study. In each experiment, Cy5-labeled cRNA from one tumor X is hybridized on a 25k human microarray together with a Cy3-labeled cRNA pool made of cRNA samples from tumors 1, 2, ... N. The digital expression data were obtained by scanning and image processing. The error modeling allowed us to assign a p-value to each transcript ratio measurement.
• FIG. 3 Two-dimensional clustering reveals two distinctive types of tumors. The clustering was based on the gene expression data of 98 breast cancer tumors over 4986 significant genes. Dark gray (red) presents up-regulation, light gray (green) represents down-regulation, black indicates no change in expression, and gray indicates that data is not available. 4986 genes were selected that showed a more than two fold change in expression ratios in more than five experiments. Selected clinical data for test results of BR CA1 mutations, estrogen receptor (ER), and proestrogen receptor (PR), tumor grade, lymphocytic infiltrate, and angioinvasion are shown at right. Black denotes negative and white denotes positive. The dominant pattern in the lower part consists of 36 patients, out of which 34 are ER-negative (total 39), and 16 are BR CA1-mutation carriers (total 18).
• FIG. 4 A portion of unsupervised clustered results as shown in FIG. 3. ESR1 (the estrogen receptor gene) is coregulated with a set of genes that are strongly co-regulated to form a dominant pattern.
• FIG. 5A Histogram of correlation coefficients of significant genes between their expression ratios and estrogen-receptor (ER) status (i.e., ER level). The histogram for experimental data is shown as a gray line. The results of one Monte-Carlo trial is shown in solid black. There are 2,460 genes whose expression data correlate with ER status at a level higher than 0.3 or anti-correlated with ER status at a level lower than -0.3.
• FIG. 5B The distribution of the number of genes that satisfied the same selection criteria (amplitude of correlation above 0.3) from 10,000 Monte-Carlo runs. It is estimated that this set of 2,460 genes reports ER status at a confidence level ofp >99.99%.
• FIG. 6 Classification Type 1 and Type 2 error rates as a function of the number (out of 2,460) marker genes used in the classifier. The combined error rate is lowest when approximately 550 marker genes are used.
• FIG. 7 Classification of 98 tumor samples as ER(+) or ER(-) based on expression levels of the 550 optimal marker genes. ER(+) samples (above white line) exhibit a clearly different expression pattern that ER(-) samples (below white line).
• FIG. 8 Correlation between expression levels in samples from each patient and the average profile of the ER(-) group vs. correlation with the ER(+) group. Squares represent samples from clinically ER(-) patients; dots represent samples from clinically ER(+) patients.
• FIG. 9A Histogram of correlation coefficients of gene expression ratio of each significant gene with the BRCA1 mutation status is shown as a solid line. The dashed line indicates a frequency distribution obtained from one Monte-Carlo run. 430 genes exhibited an amplitude of correlation or anti-correlation greater than 0.35.
• FIG. 9B Frequency distribution of the number of genes that exhibit an amplitude of correlation or anti-correlation greater than 0.35 for the 10,000 Monte-Carlo run control. Mean=115.p(n>430)=0.48% and p(>430/2)=9.0%.
• FIG. 10 Classification type 1 and type 2 error rates as a function of the number of discriminating genes used in the classifier (template). The combined error rate is lowest when approximately 100 discriminating marker genes are used.
• FIG. 11A The classification of 38 tumors in the ER(-) group into two subgroups, BRCA1 and sporadic, by using the optimal set of 100 discriminating marker genes. Patients above the white line are characterized by BRCA1-related patterns.
• FIG. 11B Correlation between expression levels in samples from each ER(-) patient and the average profile of the BRCA1 group vs. correlation with the sporadic group. Squares represent samples from patients with sporadic-type tumors; dots represent samples from patients carrying the BRCA1 mutation.
• FIG. 12A Histogram of correlation coefficients of gene expression ratio of each significant gene with the prognostic category (distant metastases group and no distant metastases group) is shown as a solid line. The distribution obtained from one Monte-Carlo run is shown as a dashed line. The amplitude of correlation or anti-correlation of 231 marker genes is greater than 0.3.
• FIG. 12B Frequency distribution of the number of genes whose amplitude of correlation or anti-correlation was greater than 0.3 for 10,000 Monte-Carlo runs.
• FIG. 13 The distant metastases group classification error rate for type 1 and type 2 as a function of the number of discriminating genes used in the classifier. The combined error rate is lowest when approximately 70 discriminating marker genes are used.
• FIG. 14 Classification of 78 sporadic tumors into two prognostic groups, distant metastases (poor prognosis) and no distant metastases (good prognosis) using the optimal set of 70 discriminating marker genes. Patients above the white line are characterized by good prognosis. Patients below the white line are characterized by poor prognosis.
• FIG. 15 Correlation between expression levels in samples from each patient and the average profile of the good prognosis group vs. correlation with the poor prognosis group. Squares represent samples from patients having a poor prognosis; dots represent samples from patients having a good prognosis. Red squares represent the 'reoccurred' patients and the blue dots represent the 'non-reoccurred'. A total of 13 out of 78 were misclassified.
• FIG. 16 The reoccurrence probability as a function of time since diagnosis. Group A and group B were predicted by using a leave-one-out method based on the optimal set of 70 discriminating marker genes. The 43 patients in group A consists of 37 patients from the no distant metastases group and 6 patients from the distant metastases group. The 35 patients in group B consists of 28 patients from the distant metastases group and 7 patients from the no distant metastases group.
• FIG. 17 The distant metastases probability as a function of time since diagnosis for ER(+) (yes) or ER(-) (no) individuals.
• FIG. 18 The distant metastases probability as a function of time since diagnosis for progesterone receptor (PR)(+) (yes) or PR(-) (no) individuals.
• FIG. 19A, B The distant metastases probability as a function of time since diagnosis. Groups were defined by the tumor grades.
• FIG. 20A Classification of 19 independent sporadic tumors into two prognostic groups, distant metastases and no distant metastases, using the 70 optimal marker genes. Patients above the white line have a good prognosis. Patients below the white line have a poor prognosis.
• FIG. 20B Correlation between expression ratios of each patient and the average expression ratio of the good prognosis group is defined by the training set versus the correlation between expression ratios of each patient and the average expression ratio of the poor prognosis training set. Of nine patients in the good prognosis group, three are from the "distant metastases group"; of ten patients in the good prognosis group, one patient is from the "no distant metastases group". This error rate of 4 out of 19 is consistent with 13 out of 78 for the initial 78 patients.
• FIG. 20C The reoccurrence probability as a function of time since diagnosis for two groups predicted based on expression of the optimal 70 marker genes.
• FIG. 21A Sensitivity vs. 1-specificity for good prognosis classification.
• FIG. 21B Sensitivity vs. 1-specificity for poor prognosis classification.
• FIG. 21C Total error rate as a function of threshold on the modeled likelihood. Six clinical parameters (ER status, PR status, tumor grade, tumor size, patient age, and presence or absence of angioinvasion) were used to perform the clinical modeling.
• FIG. 22 Comparison of the log(ratio) of individual samples using the "material sample pool" vs. mean subtracted log(intensity) using the "mathematical sample pool" for 70 reporter genes in the 78 sporadic tumor samples. The "material sample pool" was constructed from the 78 sporadic tumor samples.
• FIG. 23A Results of the "leave one out" cross validation based on single channel data. Samples are grouped according to each sample's coefficient of correlation to the average "good prognosis" profile and "poor prognosis" profile for the 70 genes examined. The white line separates samples from patients classified as having poor prognoses (below) and good prognoses (above).
• FIG. 23B Scatter plot of coefficients of correlation to the average expression in "good prognosis" samples and "poor prognosis" samples. The false positive rate (i.e., rate of incorrectly classifying a sample as being from a patient having a good prognosis as being one from a patient having a poor prognosis) was 10 out of 44, and the false negative rate is 6 out of 34.
• FIG. 24A Single-channel hybridization data for samples ranked according to the coefficients of correlation with the good prognosis classifier. Samples classified as "good prognosis" lie above the white line, and those classified as "poor prognosis" lie below.
• FIG. 24B Scatterplot of sample correlation coefficients, with three incorrectly classified samples lying to the right of the threshold correlation coefficient value. The threshold correlation value was set at 0.2727 to limit the false negatives to approximately 10% of the samples.

The invention relates to sets of genetic markers whose expression patterns correlate with important characteristics of breast cancer tumors. i.e., estrogen receptor (ER) status, BRCA1 status, and the likelihood of relapse (i.e., distant metastasis or poor prognosis). Sets of genetic markers that can distinguish the following three clinical conditions are described. First, the invention relates to sets of markers whose expression correlates with the ER status of a patient, and which can be used to distinguish ER (+) from ER (-) patients. ER status is a useful prognostic indicator, and an indicator of the likelihood that a patient will respond to certain therapies, such as tamoxifen.

Also, among women who are ER positive the response rate (over 50%) to hormonal therapy is much higher than the response rate (less 10%) in patients whose ER status is negative. In patients with ER positive tumors the possibility of achieving a hormonal response is directly proportional to the level ER (P. Clabresi and P. S. Schein, MEDICAL ONCOLOGY (2ND ED.), McGraw-Hill, Inc., New York (1993)). Second, the invention further relates to sets of markers whose expression correlates with the presence of BRCA1 mutations, and which can be used to distinguish BRCA1-type tumors from sporadic tumors. Third, the invention relates to genetic markers whose expression correlates with clinical prognosis, and which can be used to distinguish patients having good prognoses (i. e., no distant metastases of a tumor within five years) from poor prognoses (i. e., distant metastases of a tumor within five years). Methods are provided for use of these markers to distinguish between these patient groups, and to determine general courses of treatment. Microarrays comprising these markers are also provided, as well as methods of constructing such microarrays. Each markers correspond to a gene in the human genome, i. e., such marker is identifiable as all or a portion of a gene. Finally, because each of the above markers correlates with a certain breast cancer-related conditions, the markers, or the proteins they encode, are likely to be targets for drugs against breast cancer.

As used herein, "BRCAI tumor" means a tumor having cells containing a mutation of the BRCA1 locus.

The "absolute amplitude" of correlation expressions means the distance, either positive or negative, from a zero value; i. e., both correlation coefficients -0.35 and 0.35 have an absolute amplitude of 0.35.

"Status" means a state of gene expression of a set of genetic markers whose expression is strongly correlated with a particular phenotype. For example, "ER status" means a state of gene expression of a set of genetic markers whose expression is strongly correlated with that of ESR¹ (estrogen receptor gene), wherein the pattern of these genes' expression differs detectably between tumors expressing the receptor and tumors not expressing the receptor.

"Good prognosis" means that a patient is expected to have no distant metastases of a breast tumor within five years of initial diagnosis of breast cancer.

"Poor prognosis" means that a patient is expected to have distant metastases of a breast tumor within five years of initial diagnosis of breast cancer.

"Marker" means an entire gene, or an EST derived from that gene, the expression or level of which changes between certain conditions. Where the expression of the gene correlates with a certain condition, the gene is a marker for that condition.

"Marker-derived polynucleotides" means the RNA transcribed from a marker gene, any cDNA or cRNA produced therefrom, and any nucleic acid derived therefrom, such as synthetic nucleic acid having a sequence derived from the gene corresponding to the marker gene.

A set of 4,986 genetic markers whose expression is correlated with the existence of breast cancer by clustering analysis is described. A subset of these markers identified as useful for diagnosis or prognosis is listed as SEQ ID NOS : 1-2,699. The invention also relates to a method of using these markers to distinguish tumor types in diagnosis or prognosis.

In A set of 2,460 genetic markers that can classify breast cancer patients by estrogen receptor (ER) status ; i.e., distinguish between ER (+) and ER (-) patients or tumors derived from these patients-is also described. ER status is an important indicator of the likelihood of a patient's response to some chemotherapies (i. e., tamoxifen). These markers are listed in Table 1. The invention also relates to subsets of at least 5, 10, 25, 50, 100, 200, 300, 400, 500, 750, 1,000, 1,250, 1,500, 1,750 or 2,000 genetic markers, drawn from the set of 2,460 markers, which also distinguish ER (+) and ER (-) patients or tumors. Preferably, the number of markers is 550. Further described is a set of 550 of the 2,460 markers that are optimal for distinguishing ER status (Table 2). A method of using these markers to distinguish between ER (+) and ER (-) patients or tumors derived therefrom, is also provided.

In another embodiment, a set of 430 genetic markers that can classify ER (-) breast cancer patients by BRCA1 status; i. e., distinguish between tumors containing a BRCA1 mutation and sporadic tumors is described. These markers are listed in Table 3. Subsets of at least 5, 10, 20, 30, 40, 50, 75, 100, 150, 200, 250, 300 or 350 markers, drawn from the set of 430 markers, which also distinguish between tumors containing a BRCA1 mutation and sporadic tumors- are further provided. Preferably, the number of markers is 100. A preferred set of 100 markers is provided in Table 4. A method of using these markers to distinguish between BRCA1 and sporadic patients or tumors derived therefrom, is also described.

The invention relates to a set of 231 genetic markers that can distinguish between patients with a good breast cancer prognosis (no breast cancer tumor distant metastases within five years) and patients with a poor breast cancer prognosis (tumor distant metastases within five years). These markers are listed in Table 5. Subsets are provided of at least 5, 10, 20, 30, 40, 50, 75, 100, 150 or 200 markers, drawn from the set of 231, which also distinguish between patients with good and poor prognosis. A preferred set of 70 markers is provided in Table 6. In a specific embodiment, the set of markers consists of the twelve kinase-related markers and the seven cell division- or mitosis-related markers listed. The invention also provides a method of using the above markers to distinguish between patients with good or poor prognosis.