Identification of a gene expression profile that differentiates ischemic and nonischemic cardiomyopathy   

This application is a continuation of U.S. patent application Ser. No. 12/545,529, filed Aug. 21, 2009, which is a continuation of U.S. patent application Ser. No. 11/012,778, filed Dec. 15, 2004, issued as U.S. Pat. No. 7,592,138, which claims priority to U.S. Provisional Patent Application No. 60/529,834, filed Dec. 16, 2003, the contents of which applications are hereby incorporated by reference in their entirety.

BACKGROUND OF THE INVENTION

This invention relates to cardiomyopathy and especially to diagnosis and prognosis of ischemic and nonischemic cardiomyopathy. Most particularly, this invention relates to a diagnostic method to differentiate ischemic from nonischemic cardiomyopathy based on a gene expression profile of the heart tissue being evaluated. This invention also relates to a method of gene profiling and to a gene expression prediction profile prepared in accordance with said method.

Gene expression profiling holds great promise as a tool to refine diagnostic and prognostic accuracy in a variety of diseases. This technique has enjoyed widespread success in solid and hematologic malignancies and may soon be employed in clinical trials. (Alizadeh A A et al., Nature (2000); Lapointe J., et al., Proc Natl Acad Sci.(2004); Tibshirani R., et al., Proc Natl Acad Sci. (2002); Dhanasekaran S M, et al., Nature (2001); Pomeroy, et al., Nature (2002); Van de Vijver M J, et al., N Engl J Med. (2002); Golub T R, et al., Science (1999); Rosenwald A., et al., N Engl J Med. (2002). In contrast, while the ability to refine diagnosis, particularly with regard to ischemic etiology, and predict patient outcome is of tremendous importance in myocardial diseases, the application of gene expression profiling for this purpose is in its earliest stages. To date, small studies have demonstrated that gene expression differs between failing and nonfailing hearts, (Barrans J D., et al., Am J Pathol. (2002); Tan F L., et al., Proc Natl Acad Sci. (2002); Yung C K., et al., Genomics; Steenman M., et al., Physiol Genomics (2004)) dilated and hypertrophic cardiomyopathy, (Hwang J J, Allen P D, Tseng G C et al., Physiol Genomics (2002)) and before and after placement of a ventricular assist device. (Chen Y., et al., Physiol Genomics (2003); Hall J L., et al., Physiol Genomics (2003); Chen M M., et al., Circulation; Blaxal B C, et al., J AM Coll Cardiol (2003)). These studies focused on the identification of novel genetic pathways. The application of gene expression profiling to distinguish clinically relevant cardiomyopathic disease subtypes has not previously been performed and is considered controversial, due to the contention that, unlike tumors, there is a final common pathway independent of etiology for the progression of myocardial disease.

Ischemic cardiomyopathy is defined as evidence of myocardial infarction on histology of the explanted heart. Gene expression profiling would serve as a valuable adjunct to imaging and metabolic tools in the diagnosis of ischemic cardiomyopathy. Despite similar presentations, ischemic and nonischemic cardiomyopathy are distinct diseases. Patients with ischemic cardiomyopathy have decreased survival compared to their nonischemic counterparts (Felker G M, et al., N Engl J Med. (2000); Felker G M, et al., J AM Coll Cardiol. (2003); Dries D L, et al., J Am Coll, Cardiol. (2001)) and respond differently to therapies. (Kittleson M, et al., J Am Coll Cardiol. (2003); Doval H C, et al., Lancet (1994); Singh S N, et al., N Engl J Med. (1995); Reynolds M R, et al., Circulation (2003)). An ischemic gene expression profile would offer diagnostic insight, especially in patients with heart failure out of proportion to their coronary artery disease. The proportion of such patients is estimated to be up to 11% in one observational study (Felker G M, et al., J Am Coll Cardiol. (2002)). The ability to tailor treatments to specific patients by identifying those who would most benefit, is of critical importance in heart failure patients. (Reynolds M R, Circulation (2003))

A prior study noted differences in gene expression in ischemic versus nonischemic cardiomyopathy samples following LVAD (left ventricle assist device) support. However, that study did not create or prospectively validate a prediction rule. (Blaxall B C, et al., J Am Coll Cardiol. (2003)) Another study compared the gene expression profiles of ischemic and nonischemic cardiomyopathy samples and found no differentially expressed genes (Steenman M, et al., Physiol Genomics. (2003)). But that study used pooled samples from only two ischemic and two nonischemic cardiomyopathy patients, and it is likely that this study did not have adequate power to detect changes in gene expression (Mukherjee S, et al., J Comput Biol. (2003)).

Another study shows that the differential gene expression between failing and nonfailing hearts has been attributed to age and gender differences, (Boheler K R, et al., Proc Natl Acad Sci USA. (2003)). However, this analysis has not been extended to ischemic and nonischemic cardiomyopathy. Other studies have also shown that failing hearts exhibit changes in gene expression following LVAD support (left ventricle assist device). (Chen Y., et al., Physiol Genomics (2003); Hall J L., et al., Physiol Genomics (2004); Chen M M., et al., Circulation (2003); Blaxal B C, et al., J AM Coll Cardiol. (2003)). In addition, gene expression analysis was considered hypothesis-generating until validated by another technique. (Cook S A, et al., Circ Res. (2002))

Our major new finding is that a gene expression-based signature accurately distinguishes between ischemic and nonischemic etiologies of cardiomyopathy. Gene expression profiles have been successfully correlated with etiology or clinical outcome in oncology (Alizadeh A A et al., Nature (2000); Lapointe J., et al., Proc Natl Acad Sci. (2004); Tibshirani R., et al., Proc Natl Acad Sci. (2002); Dhanasekaran S M, et al., Nature (2001); Pomeroy, et al., Nature (2002); Van de Vijver M J, et al., N Engl J Med. (2002); Golub T R, et al., Science (1999); Rosenwald A., et al., N Engl J Med. (2002); Hastie T, et al., Genome Biol. (2000) and renal allograft rejection, (Sarwal M, et al., N Engl J Med. (2003)). Expression profile-based prognostic tools are in clinical trials in oncology. There is an equal need to refine diagnostic and prognostic techniques in myocardial diseases. Our findings demonstrate that gene expression profiling can accurately identify disease etiology. This has substantial clinical implications and strongly supports ongoing efforts to incorporate expression-profiling based biomarkers in determining prognosis and_response to therapy.

SUMMARY

An object of the present invention is to provide a gene expression profile that can discriminate between common causes of heart failure in patients with end-stage cardiomyopathy. We have established that the methodology to achieve this end is highly generalizable to data obtained in different laboratories.

Another object of the present invention is to establish that molecular signatures can be used to refine the diagnostic evaluation and management of heart failure, where treatment and prognosis decisions may vary based on disease etiology (Felker G M, et al., N Engl J Med. (2000); Felker G M, et al., J Am Coll Cardiol (2003); Dries D L, et al., J Am Coll Cardiol. (2001); Kittleson M, et al., J Am Coll Cardiol. (2003); Doval H C, et al., Lancet (1994); Singh S N, et al., N Engl J Med. (1995); Follath F, et al., J Am Coll Cardiol. (1998)).

More specifically, the present invention is directed to a method of preparing a gene expression prediction profile for distinguishing ischemic and nonischemic cardiomyopathy, comprising the steps of:

obtaining clinical specimens from patients suffering from ischemic or nonischemic cardiomyopathy;

isolating nucleic acid sequences from at least a plurality of said patients;

obtaining a gene expression level corresponding to each individual of said nucleic acid sequence by a gene expression profiling method;

identifying genes having statistically significant difference in gene expression by comparing the gene expression level of an ischemic specimen with the gene expression level of a nonischemic specimen, and identifying a gene expression prediction profile that distinguishes ischemic and nonischemic cardiomyopathy.

The present invention is also directed to a method of diagnosis for differentiating ischemic and nonischemic cardiomyopathy, comprising the steps of:

obtaining a clinical specimen from a patient having cardiomyopathy;

isolating nucleic acid sequences from said specimen;

obtaining a gene expression level corresponding to said nucleic acid sequence by a gene expression profiling method;

comparing the gene expression level of said specimen with a gene expression prediction profile prepared in accordance with the method described above to determine ischemic or nonischemic cardiomyopathy by performing a prediction analysis.

The present invention is further directed to a gene expression prediction profile prepared in accordance with a method comprising the steps of:

obtaining clinical specimens from patients suffering from ischemic or non ischemic cardiomyopathy;

isolating nucleic acid sequences from at least a plurality of said patients;

obtaining a gene expression level corresponding to each individual of said nucleic acid sequence by a gene expression profiling method;

identifying genes having differences, preferably statistically significant differences in gene expression by comparing the gene expression, level of an ischemic specimen with the gene expression level of a nonischemic specimen, and

identifying a gene expression prediction profile comprising genes that distinguishes ischemic and nonischemic cardiomyopathy.

The present invention is further directed to a method of treating ischemic or nonischemic cardiomyopathy, comprising the step of diagnosing for differentiating ischemic and nonischemic cardiomyopathy. The diagnosis comprises the steps of:

obtaining a clinical specimen from a patient having cardiomyopathy;

isolating nucleic acid sequences from said specimen;

obtaining a gene expression level corresponding to said nucleic acid sequence by a gene expression profiling method;

comparing the gene expression level of said specimen with a gene expression prediction profile prepared in accordance with the method of claim 1 to determine ischemic or nonischemic cardiomyopathy by performing a prediction analysis.

Other objects and features of the present invention will become apparent from the following detailed description considered in conjunction with the accompanying drawings. It is to be understood, however, that the drawings are designed solely for purposes of illustration and not as a definition of the limits of the invention, for which reference should be made to the appended claims. It should be further understood that the drawings are not necessarily drawn to scale and that, unless otherwise indicated, they are merely intended to conceptually illustrate the structures and procedures described herein.

BRIEF DESCRIPTION OF THE DRAWINGS

In the drawings:

FIG. 1 illustrates the separation of end-stage cardiomyopathy samples into a training set (used to identify the gene expression prediction profile), a test set (used to assess the accuracy of the prediction profile), and post-remodeling samples. The overall predictive accuracy was assessed by examining 210 combinations of training and test set samples.

FIG. 2 is a bargraph showing the number of genes up- and down-regulated in ischemic hearts relative to nonischemic hearts classified by functional group (www.geneontology.org).

FIG. 3 is a hierarchical clustering of 90 genes in 48 samples based on similarity in gene expression and relatedness of samples. Each row represents a gene labeled with the gene symbol and each column represents a sample. The color in each cell reflects the level of expression of the corresponding gene in the corresponding sample, relative to its mean level of expression in the entire set of samples. Expression levels greater than the mean are shaded in blue, and those below the mean are shaded in red. The samples form two distinct clusters based on etiology. Arrows denote samples that do not appear in their etiology cluster. ICM denotes ischemic cardiomyopathy and NICM denotes nonischemic cardiomyopathy.

DETAILED DESCRIPTION

As used herein, the term gene expression prediction profile or molecular signature or gene expression-based signature means a known expression profile of a set of genes to which an unknown gene expression profile of a new set of genes can be compared or evaluated.

The term clinical specimens mean samples obtained from human heart muscle in various ways.

As used herein, the term “nucleic acid” refers to polynucleotides such as deoxyribonucleic acid (DNA), and, where appropriate, ribonucleic acid (RNA). The term should also be understood to include, as equivalents, analogs of either RNA or DNA made from nucleotide analogs, and, as applicable to the embodiment being described, single (sense or antisense) and double-stranded polynucleotides. ESTs, chromosomes, cDNAs, mRNAs, and rRNAs are representative examples of molecules that may be referred to as nucleic acids.

As used herein, the term expression profiling method is a method of detecting the level of gene expression based on technologies such as DNA microarray, Spotted array, cDNA array, and reverse transcription polymerase chain reaction (RT-PCR).

As used herein, the term random partitioning of the clinical specimens or samples refers to a method of grouping and matching the samples to obtain all possible outcomes resulting from the grouping and matching.

The term prediction analysis generally refers to an analytical method for identifying a gene expression prediction profile. Specifically, the term refers to obtaining a set of genes (also described as a “molecular signature”) from a new and unknown sample, that, by comparing the expression level of the genes in this set in the new sample with the gene expression of a known gene expression prediction profile, allows one to determine the group to which the new and unknown sample belongs. The expression level of the genes in the set are sufficiently and consistently different within the groups so as to allow distinguishing to which group a new sample belongs.

The present invention employs a variety of methodologies in connection with establishment of a gene expression profile. While the methodologies employed in the present invention, such as clinical sample collection, nucleic acid sample preparation, DNA microarray technologies, and statistical analysis associated with gene profile analysis are generally available, diagnosis and treatment of ischemic or nonischemic cardiomyopathy based on gene expression profiling were not considered feasible until a group of genes were isolated and identified to accurately discriminate ischemic from nonischemic heart failure.

The invention generally includes the steps as described here below.

To generate a gene expression prediction profile that can provide general prediction, diagnosis and/or prognosis, and treatment based on such diagnosis, clinical specimens are collected from the myocardial tissues of patients who have experienced ischemic or nonischemic cardiomyopathy.

In a preferred embodiment, all patients from whom the myocardial tissues were obtained that had ischemic cardiomyopathy exhibited severe coronary artery disease (>75% stenosis of the left anterior descending artery and at least one other epicardial coronary artery) and/or a documented history of a myocardial infarction. (Hare J M, et al., J Am Coll Cardiol. (1992); Felker G M, et al., J Am Coll Cardiol. (2003)) Nonischemic patients had no history of myocardial infarction, revascularization, or coronary artery disease

Preferably the myocardial tissues from surgery are immediately frozen in liquid nitrogen and stored at −80° C. tissues can also be stored with other methods.

To establish an expression profile of myocardial genes, a DNA microarray may be used. Myocardial RNA may be isolated from the frozen samples using the Trizol reagent and Qiagen RNeasy columns. Double-stranded cDNA may be synthesized from 5 pg RNA using the SuperScript Choice system (Invitrogen Corp, Carlsbad, Calif.). Each double-stranded cDNA may be subsequently used as a template to make biotin-labeled cRNA. 15 μg of fragmented, biotin-labeled cRNA from each sample was hybridized to an Affymetrix U133A microarray (Affymetrix, Santa Clara, Calif.). Affymetrix chip processing was performed. The U133A microarray allows detection of 21,722 transcripts (15,713 full length, 4,534 non-expressed sequence tags (ESTs) and 1,475 ESTs). The quality of array hybridization may be assessed by the 3′ to 5′ probe signal ratio of GAPDH and β-actin. A ratio of 1-1.2, indicates an acceptable RNA preparation.

While a DNA microarray for obtaining a gene expression profile is preferred, other expression methods known to a person of ordinary skill in the art, such as Spotted array, cDNA array, and RT-PCR, may also be used to obtain substantially the same results.

The purpose of data normalization is to convert probe-set data from the microarray hybridization (the raw data obtained from the microarray) to gene expression values. The microarray contains multiple probes for each given transcript, the intensity of hybridization to each of these probes must be combined to create a single quantitative value for the expression of each transcript. In addition, normalization allows for correction for variation within chips and across samples so that data from different chips can be simultaneously analyzed. The robust multi-array analysis (RMA) algorithm, which is described in references (Irizarry R A, et al., Biostatistics (2003) and Irizarry R A, et al., Nucleic Acids Res. (2003)), may be used to pre-process the Affymetrix probe set data into gene expression levels for all samples. The contents of Irizarry R A, et al., Biostatistics (2003); Irizarry R A, et al., Nucleic Acids Res. (2003) are incorporated by reference in their entirety. Although other methods may be used to normalize the data, such as using Affymetrix\'s default preprocessing algorithm (MAS 5.0), RMA is preferred, which results in classifiers with better predictive power. (Irizarry R A, et al., Nucleic Acids Res. (2003))

In order to create the gene expression prediction profile using genes that are differentially expressed in ischemic versus nonischemic samples, a statistical analysis for identifying genes that exhibit changes in gene expression, preferably statistically significant changes in gene expression, between ischemic and nonischemic samples was performed. For this purpose, Significance Analysis of Microarrays (SAM) is preferred. Reference (Tusher V G, et al., Proc Nat/ Acad Sci USA. (2001)) provides details of SAM analysis, the content of which is incorporated by reference in its entirety. SAM identifies genes with changes, preferably statistically significant changes in expression by assimilating a set of gene-specific tests (similar to the t-test) which we will refer to as the SAM-statistics. For any given threshold, a resampling procedure is used to estimate false discovery rates (FDR) of lists of genes for which the SAM-statistic is bigger than this threshold. At a FDR of 0.1%, there were 3332 differentially expressed genes between ischemic and nonischemic hearts. These 3332 genes were then subject to further analysis. Other statistical methods known to a person of ordinary skill in the art may also be used to accomplish the same objective.

To test consistency between an expression profile relative to ischemic or nonischemic cardiomyopathy, a classification algorithm based on the methodology used by the Prediction Analysis of Microarrays software PAM (Tibshirani R, et al., Proc Natl Acad Sci USA. (2002)) was employed. By doing so, a gene expression profile that distinguishes ischemic from nonischemic cardiomyopathy samples is identified. While other known methods may be used for the same purpose, PAM is preferred. PAM is a supervised classification method that defines a score for each gene, representative of its contribution to predictive power. Given a set of genes, PAM defines a prediction rule based on classification of the training set that is then applied to the test set. Details about PAM are provided in reference Tibshirani R, et al., Proc Natl Acad Sci USA. (2002), the content of which is incorporated by reference in its entirety.

To assess if the accuracy of the etiology prediction profile is affected by baseline clinical covariates (including age, gender, systolic function, and medication use) as well as differences in etiology, individuals from which the clinical specimens are obtained were stratified, based on these covariates, and the predictive accuracy was assessed.

Continuous variables may be summarized by the median and quartiles and groups may be compared using the Wilcoxon rank sum test. Categorical variables may be summarized by proportions and compared using Fisher\'s exact test.

Prediction accuracy is determined based on the sensitivity and specificity of the prediction, where sensitivity is the proportion of ischemic cardiomyopathy samples correctly classified by gene expression profiling, and specificity is the proportion of nonischemic cardiomyopathy samples correctly classified.

The present invention yields a prediction tool that was generalizable to samples from different laboratories, and for ischemic non-ischemic cardiomyopathy, the prediction tool was independent of disease severity.

To determine if the etiology prediction profile was affected by differences in clinical characteristics between ischemic and nonischemic cardiomyopathy patients, we stratified our analysis based on clinical covariates mentioned in Table 1 below and found that the sensitivity and specificity of our analysis was not affected. This supports the idea that the excellent predictive accuracy of our method is not an artifact of differences in baseline characteristics.

We created a gene expression profile in end-stage cardiomyopathy samples and tested the profile in samples of comparable stage. We also tested the profile in post-LVAD samples of nonischemic hearts where the prediction profile performed perfectly in classifying ischemic or nonischemic cardiomyopathy, although only one of three ischemic post-LVAD samples was correctly classified. This suggests that ischemic hearts exhibit more extensive changes in gene expression following LVAD support than nonischemic hearts. While this seems to be in contrast to a recent study which determined that nonischemic cardiomyopathy patients exhibited greater changes in gene expression. (Blaxall B C, et al., J Am Coll Cardiol. (2003)), the duration of LVAD support in that study was relatively short (mean (±SD) of 57±15 days), compared with our present study (190±151 days), and this may have affected changes in gene expression.

Unlike the majority of studies in cardiology, where microarray analysis is concentrated on the discovery of novel genetic pathways, our analysis is focused on clinical prediction. Thus, our validation involved application of the identified gene expression prediction profile to classify independent samples. Using this approach, well-validated in the cancer literature, (Tibshirani R, et al., Proc Natl Acad Sci USA (2002); Van de Vijver M J, et al., N Engl J Med. (2002); Golub T R, et al., Science. (1999)) we have determined the etiology of independent samples with excellent accuracy over a wide range of combinations of test set samples.

To the best of the inventor\'s knowledge this study is the first proof that microarray analysis can contribute substantially to improving clinical diagnosis and optimizing therapy based on gene expression profiling in heart tissues. The present study also forms a basis for future studies using molecular profiling to differentiate heart failure by clinically relevant parameters, including prognosis and response to therapy.

The invention may be further illustrated by the following examples, which are not limitations to the present invention.

The study sample comprised 41 samples from 27 patients with cardiomyopathy. Myocardial tissue was obtained from patients with different stages: 1) 25 end-stage tissue obtained at time of left ventricular assist device (LVAD) placement or cardiac transplantation, and 2) 16 post reverse-remodeling: following the removals of LVAD support (average duration: 190±151 days). Twenty-eight of the samples were paired; i.e., obtained from one patient at LVAD implantation and at LVAD removal during transplantation. Samples were from two institutions: 1) Johns Hopkins Hospital in Baltimore, Md. (n=20 patients, n=27 samples) and 2) University of Minnesota in Minneapolis, Minn. (n=7 patients, n=14 samples). Samples from the latter institution were collected and prepared independently, (Chen Y, et al., Physiol. Genomics. (2003)) and gene expression data files were kindly provided. The subsequent description applies to the 27 samples collected from patients at the Johns Hopkins Hospital.

All patients had ischemic (n=11) or nonischemic (n=16) end-stage cardiomyopathy with severely reduced ejection fraction, left ventricular dilation, elevated pulmonary arterial and wedge pressures, and reduced cardiac index (Table 1). Importantly, these hemodynamic and remodeling measures were similar between groups. Ischemic cardiomyopathy patients were older, all male, more likely to be on angiotensin-converting enzyme inhibitors (ACEI), and less likely to be on intravenous inotropic therapy.

TABLE 1 clinical characteristics of patients* Ischemic Nonischemic Clinical Characteristic (11 subjects) (16 subjects) Age, y 57.5 (54-60) 46 (37-52)† Male 100% 67%‡ Left ventricular ejection 18.8 (15.0-25.0) 15.0 (10.0-20.0) fraction, % Left ventricular end-diastolic 6.8 (6.4-7.3) 7.4 (6.8-8.3) diameter, cm Pulmonary artery pressure, mmHg Systolic 49 (35-64) 50 (45-57) Diastolic 25 (18-33) 30 (24-30) Pulmonary capillary wedge 27 (14-31) 25 (20-30) pressure, mmHg Cardiac Index, L · min1 · m2 2.2 (1.5-2.4) 1.5 (1.3-1.9) Medications Beta Antagonists  70% 39%  Ace inhibitors or 100% 62%† Angiotensin receptor blockers Diuretics 100% 69%  Intravenous inotropic therapy§

Download full PDF for full patent description/claims.




You can also Monitor Keywords and Search for tracking patents relating to this Identification of a gene expression profile that differentiates ischemic and nonischemic cardiomyopathy patent application.
###


Other recent patent applications listed under the agent :