JP2003501005A - Methods for reducing variance in genotype-based treatment studies - Google Patents

Abstract

(57) Summary The present invention provides methods, computer programs, and computerized systems for evaluating the efficacy of various treatment procedures as a function of a subject's genotype. Genetic adaptation of the treatment and control groups reduces the overall variance of the test, and tests that test the efficacy or efficacy of the treatment procedure have increased confidence values and / or accuracy or discrimination. Can be performed by a small number of subjects in an increased state. The method of the invention comprises selecting, from the treated and control populations, a treated and control subpopulation of subjects for similarity of the polymorphism profile, wherein the treated and control populations each comprise: Includes steps treated by treatment and control procedures. As an evaluation of the test procedure, a determination is made whether there is a statistically significant difference in test parameters between the treated and control populations.

Description

Detailed Description of the Invention

(Citation of Related Application) This application is US provisional application No. 60 / 110,668 (filed on December 2, 1998).
Claim the benefit of priority to. This provisional application is hereby incorporated by reference for all purposes.

[0002]   (Field of the invention)   The present invention relates to the fields of medicine, genetics and statistics.

BACKGROUND OF THE INVENTION The practice and design of studies for the study of treatment efficacy, such as clinical trials, can result from biases that can result from "random" biological effects, whether genetic or environmental, and The aim is to eliminate biases that are deliberately or otherwise introduced by the tester. One approach to reducing bias is that individuals in the two groups are genetically unrelated to each other and survive independently if both genetic and environmental impacts on the test are considered in the study. Randomizing individuals to either treatment or control groups in terms of being equilibrated in one arm. However, the direct consequence of randomization in this method is that the variance of the measured biological conditions will be greater than when each case is adapted for genetic and known environmental effects.

One method for determining genetic variables is by evaluation of polymorphic profiles. Polymorphism (sex) refers to the coexistence of multiple forms of a sequence in a population. Several different types of polymorphism have been reported. For example, restriction fragment length polymorphism (RFLP
) Means a deviation in the DNA sequence that alters the length of the restriction fragment (
For example, Botstein et al., Am. J. Hum. Genet. 32: 314
-331 (1980)). Short tandem repeats (STRs), as the name suggests, are short tandem repeats composed of tandem dinucleotide, trinucleotide, and tetranucleotide repeat motifs. Such polymorphisms are also sometimes referred to as variable number tandem repeat (VNTR) polymorphisms (eg, US Pat. No. 5,075,2).
17; Armour et al., FEBS Lett. 307: 113-115 (1
992); and Horn et al., WO 91/14003).

The most common form of polymorphism to date involves single nucleotide variation between individuals of the same species. Such polymorphisms are single nucleotide polymorphisms or simply S
It is called NP (snip). Some SNPs that are present in the protein coding region result in the expression of variants or defective proteins and may therefore be responsible for the genetic disease. However, even SNPs that occur in non-coding regions can result in defective protein expression (eg, by causing defective splicing). Other SNPs have no phenotypic effect.

SUMMARY OF THE INVENTION Certain methods of the invention are designed to provide an assessment of the efficacy of a treatment procedure.
Generally, such methods involve selecting a treated and control subpopulation from a treated and control subject population. Here, the treated population has been treated with a treatment procedure and the control population has been treated with a control procedure. Subjects in both the treated and control populations have been characterized for polymorphic profiles and they are selected because they have similar polymorphic profiles. It is then determined whether there is a statistically significant difference in test parameters between the treated and control subpopulations. In general, such statistically significant differences are between the type of treatment and one or more polymorphic forms within the polymorphic profile for which the treated and control subpopulations were selected. Indicates that there is a correlation.

In some cases, especially where no significant differences are found, the selection and determination steps are repeated one or more times. In such additional cycles, the polymorphic profile in which the treated and control groups are selected is different from the polymorphic profile selected in the previous cycle. The polymorphic profile from which the subpopulations are selected can vary with respect to the number of polymorphic forms in that profile and the degree of similarity in profile between the treated and control groups. For example, the polymorphic profile can include a single polymorphic form, but more typically includes multiple polymorphic forms (eg, 10 or 1).
Polymorphic forms up to 00 or more). In most cases, the subpopulation polymorphism profile will be at least 10%, 50%, 75% or 100%.
Have an identity of up to.

Particular methods of the invention relate to methods for conducting clinical trials. Some of these methods include first treating treated and control patient populations with the same disease with drug and control procedures, respectively (
(Eg, placebo or different amounts of drug or treatment according to different treatment schedules). A subpopulation of patients is then selected from each of the treated and control populations for similarity in polymorphism profile. The efficacy of the drug in treating the disease is then assessed by determining whether treatment with the drug correlates with the status of the disease in the subpopulation. Using these methods, the correlation also indicates that at least one or more polymorphic forms within the polymorphic profile correlate with treatment efficacy.

Some methods of the present invention are computerized methods. For example, a particular method of the invention comprises the following steps: providing a database storing: (1) designation for each member of the treated population treated according to the treatment procedure, and controls. A designation for each member of the control population, treated according to the procedure, (2) a designation for a polymorphic profile of each member of the treated and control populations, and (3) of the treated and control populations. Specification of test parameters for each member. Using this database, subpopulations from each of the treated and control populations are selected for similarity in polymorphism profile. A determination is then made to establish whether there is a statistically significant difference in the test parameters between the subpopulations. The output from this determining step is then displayed on an output device (eg, monitor or video display).

In another aspect, the present invention provides various computer systems and programs. For example, a specific computer product is provided for evaluating a treatment instrument. Some systems include program products that generally include code for providing or receiving data, where the data includes: (1
) A designation for each member of the treated population treated according to the treatment procedure,
And (2) a designation for each member of the control population treated according to the control procedure and a control procedure, and (3) a designation for a polymorphic profile for each member of the treated population and the control population, and (3) the treated population and the control population. Specification of test parameters for each member of the.
This program also has a code for selecting subpopulations from each of the treated and control populations with similar polymorphism profiles, there are statistically significant differences in test parameters between the subpopulations. Code for determining whether or not to do so, and code for displaying an output indicating whether a statistically significant difference was found between the subpopulations. This code is typically stored on a computer-readable storage medium.

The present invention further provides a computerized system for evaluating a treatment procedure. Some systems generally include memory, a system bus and a processor. The processor is operably arranged to provide or receive data. Here, the data include: (1) designation for each member of the treated population treated according to the treatment procedure, and designation for each member of the control population treated according to the control procedure, (2)
Designation of polymorphism profiles for each member of the treated and control populations, and (3) Designation of test parameters for each member of the treated and control populations. The processor is further arranged to select a subpopulation from each of the treatment and control populations with similar polymorphism profiles, and a statistically significant difference in the test parameters between the subpopulations. Is present. The microprocessor may also display an output indicating whether there is a statistically significant difference between the subpopulations.

DETAILED DESCRIPTION 1. Definitions “Treatment procedure” refers to a method or process performed in a population to be treated or a member of a treatment population. Generally, the treatment procedure is a process performed in a subject to affect some biological condition, susceptibility or resistance of the subject. Examples of treatment processes include, but are not limited to, treatment with pharmaceutical compounds or other biological agents (including, for example, recombinantly produced proteins), surgical procedures, and various behaviors. Therapy (eg, prescribed diet and / or exercise regimen). The treatment procedure can be a prophylactic procedure or a therapeutic procedure. For example, the treatment procedure can include treating a member of the treatment population with a vaccine.

Similarly, “control procedure” refers to a method performed in a member of a control population. This control procedure may differ from the treatment procedure in quantitative or qualitative terms. For example, a member of the treatment procedure may receive the pharmaceutical compound, while the control population receives a placebo (ie, not the pharmaceutical compound). In other cases, the control procedure comprises administering the drug at a different concentration than the treatment procedure, or for that treatment procedure, a different schedule for administering the pharmaceutical compound. You can

A “clinical trial” is a causative, and sometimes treatment, of a particular phenotype represented by at least one random variable. This phenotype can often be a measure of the disease state or severity of the disease. Clinical trials can be in the form of case-control studies (for discrete random variables, the groups are affected and unaffected individuals), or in single population studies (where the degree or severity of this phenotype is significant). Causes of severity are investigated (eg, quantitative studies may test blood pressure, blood glucose, etc.).

A “treatment study” is the pursuit of the effect or impact of a particular treatment procedure on a subject's biological state, biological susceptibility or biological resistance. This work can be highly structured, formal, and broad in scope, or can be relatively unstructured and limited in scope. For example,
Treatment studies can be formal clinical trials or studies conducted on a relatively large population of subjects, in which the study is conducted according to specified criteria (eg, government regulations). Be seen. However, the treatment study may also be a preclinical study, a field study of a plant population or an informal study by a scientist, veterinarian or doctor of the effect of treatment on a relatively small number of subjects. In treatment studies, subjects are divided into several (often only two) groups. These may represent different dose ranges, or may simply represent treated and untreated subjects. In this study, random variables are measured post-treatment. A random variable can also be measured prior to treatment (eg, bone mineral density or blood pressure) if it is the change in the variable over the time studied. Preferably, the subject has not received any other treatment for the pathological condition. However, if such conditions are irrational, the study should be designed so that subjects in both treated and untreated groups are receiving the same alternative treatment. Subjects in treatment studies may be of any type of organism, such as animals (including humans), plants,
Bacteria and viruses).

“Biological condition” refers to the condition, susceptibility or resistance of an organism the study seeks to determine if the treatment procedure works. Typically, the biological condition is the physical or physiological condition of the organism. For example, in some cases, the biological condition is a pathological condition (ie, a normally absent physiological condition such as a disease). The pathological conditions studied using the present invention are typically conditions associated with minimal environmental variables (eg, high cholesterol levels in serum), but this is not required. An example of a pathological state is AID
S, arteriosclerosis, cancer, diabetes, hypertension, serum high cholesterol levels or psychosis. The biological condition can be the biological sensitivity or resistance of the subject. For example, the treatment study may include analysis of the effect of a particular treatment on the susceptibility of plants to herbicides or to frost damage of plants. Alternatively, the study may relate, for example, to the biodefense response (ie resistance) to some types of injury / stroke.

A “random variable”, whether discrete or continuous, may in particular be any biological, physiological or biochemical endpoint measured or observed in a clinical trial setting. . This includes measured and observed effects of treatment, changes in observation and measurement (so-called natural development) over time, or any other intervention that may alter a characteristic, sign or symptom. Examples of random variables include, for example, pharmaceutical compounds; biopharmaceuticals (including recombinantly produced proteins); surgical techniques; dietary restrictions; and pathological conditions susceptible to treatment with behavioral therapy. For example, cholesterol serum concentration, height, and weight are all continuous random variables.
This finding can be extended to discrete variables such as the presence or absence of physical characteristics or symptoms. These include, for example, nevus of the skin, cysts of the liver, certain antibodies in serum, the degree of joint swelling, the number of joints affected, or the number of hallucinations of psychotic episodes.

A “test parameter” is a measured or observed characteristic for determining the effect or efficacy (or lack thereof) of an evaluated treatment procedure, and is utilized to utilize treatment protocols and controls. Determine if there are statistically significant differences in the protocol. The test parameter can be a random variable.

Typically, test parameters are expressed in a quantitative sense, but in some cases, test parameters may be evaluated in a qualitative sense. The nature of the test parameter will vary according to the biological condition being tested. If the biological condition is a disease, the test parameter will provide a measure for the condition of the disease. For example, if the biological condition is AIDS, the test parameter can be HIV concentration in the blood of the subject. If the biological condition is arteriosclerosis, the test parameter can be serum cholesterol concentration.

The term “variance” refers to the variation, variance, spread, or variation about the arithmetic mean. Generally, the variance is the mean of the squared deviations (Armitage,
P. , STATISTICAL METHODS IN MEDICAL RE
SEARCH, Blackwell Scientific, Oxford, U
nitted Kingdom (1971)). Large variances indicate large deviations from the arithmetic mean. For example, if cholesterol level is the test parameter being measured, the average cholesterol level is determined. The variance represents the squared deviation of the mean of all cholesterol levels relative to that mean. Other statistical measures of spread or variance about the mean may also be used. Typically, the variance of the test parameters will take the shape of a bell or a normal (Gaussian) curve. Pictorially, the present invention reduces the variance and thus narrows the bell shape of the normal curve, that is, mathematically speaking, its distribution is sharp.

[0021] Typically, the variance affects dissimilarities to the subject that affect the biological state (eg, genetic, environmental and measurement variables) analyzed by statistical methods. Due to the effect. For example, because in most treatment studies the same method is used to measure the biological status in the entire population tested, the variance is dependent on the individual subject and the subject's survival. Due to genetic differences with the environment. Examples of environmental impacts include diet, sleep patterns, geographical location and culture.

“Polymorphism” refers to the occurrence of two or more genetically determined alternative sequences or alleles in a population, commonly referred to as occurring at a frequency of greater than 0.1%. . A polymorphic marker or site is a locus at which genetic divergence occurs. Preferred markers have at least two alleles, each occurring at a frequency of greater than 1% in the selected population. A polymorphic locus can be only one base pair. Such loci are called single nucleotide polymorphisms or simply SNPs (snips).

Polymorphic markers include fragment length polymorphisms, variable number tandem repeats (VNTR),
Hypervariable regions, minisatellites, dinucleotide repeats, trinucleotide repeats, tetranucleotide repeats, simple sequence repeats, and insertion elements (eg, Al
u). This first identified allelic form is optionally referred to as the reference (standard) form, and the other allelic forms are referred to as alternative or modified (mutant) alleles. This allelic form that occurs most often in the selected population is sometimes referred to as the wild-type form or allele, and the other forms are referred to as the mutant form or allele. Diploid organisms can be homozygous or heterozygous for allelic forms. Biallelic polymorphisms have two forms. The triallelic polymorphism has three forms.

“Single nucleotide polymorphism” occurs at a polymorphic site occupied by a single nucleotide. This is the site of variation between allelic sequences. In this part,
Usually, highly conserved sequences of the allele (eg, 1 / 100th of the population)
Or sequences varying in less than 1/1000 members) precede and follow.

Single nucleotide polymorphisms (SNPs) usually result from the substitution of one nucleotide for another at that polymorphic site. The transition is
The substitution of one purine with another purine or one pyrimidine with another pyrimidine. Transversion is the replacement of one purine with a pyrimidine or vice versa. Single nucleotide polymorphisms can also result from nucleotide deletions or nucleotide insertions relative to a reference allele.

“Polymorphic profile” refers to one or more polymorphic forms for which a subject is characterized. Polymorphic forms are characterized by identifying which nucleotides are present at the polymorphic site in a nucleic acid sample obtained from a subject. This profile includes at least one polymorphic form, and preferably
Multiple polymorphic forms (eg, at least 5, 10, 20, 30, 40, 50, 60
, 70, 80, 90 or 100 polymorphic forms or more) are included. Polymorphic profiles are similar if the compared polymorphic profiles share at least one polymorphic form at at least one polymorphic site. Typically, similar polymorphic profiles are at least 10, 20, 30, 40, 50, 60.
, 70, 100 or 500 polymorphic sites, at least 10%, 20%
, 30%, 40%, 50%, 60%, 70%, 80%, 90% or 100%
Share the polymorphic form of identity. Polymorphic forms are identical if the nucleotides at a particular polymorphic site are the same. Thus, two polymorphic profiles, each containing 10 polymorphic forms, have 5 polymorphic forms of that polymorphic form in the two profiles.
If two are the same, then they are 50% identical. If the organism is diploid, the polymorphic form at each polymorphic site is considered to be identical in the two individuals if both individuals have the same two alleles at the polymorphic site . For example, an individual having alleles a1 and a2 at polymorphic site A is considered to have the same profile as an individual having alleles a1 and a2, but with allele a
1 and a1, or alleles a2 and a2, or alleles a1 and a3, etc., are not considered to have the same profile.

The term “linkage” is inherited together as a result of its arrangement on the same chromosome,
Describe trends in genes, alleles, loci or genetic markers, and 2
It can be measured by% recombination between two genes, alleles, loci or genetic markers.

“Linkage disequilibrium” or “allelic association” refers to
Means a preferential association of a particular allele with a particular allele or genetic marker, or a genetic marker at a nearby chromosomal location that is more frequent than expected (eg, Weir, B., Genetic). Data Analysis
sis, Sinauer Associate Inc. , 1996). For example, if locus X has alleles a and b present at equal frequencies and linked locus Y has loci c and d present at equal frequencies, combine at a frequency of 0.25. It is expected that ac will exist. If ac is more frequently present, alleles a and c are in linkage disequilibrium. Chain disequilibrium is
It can result from natural selection of specific combinations of alleles, or because the alleles were introduced into the population too late to reach equilibrium with the linked alleles.

A marker that is in linkage disequilibrium, even if the marker does not cause the disease,
It may be particularly useful in detecting susceptibility (or other phenotype) to a disease. For example, a marker (X) in linkage disequilibrium with a gene (including regulatory sequences) (Y) that is not itself a causative element of the disease but a causative element of the phenotype may not have that gene Y identified. It may be absent or may be detected to indicate susceptibility to the disease in situations that may not be readily detectable.

“Haplotype” is a close physical proximity of polymorphic markers on a single chromosome, or markers that are not physically linked but related to each other and are associated with a biological property or phenotype. A group of polymorphic markers that

“Nucleic acid” is a polymer of deoxyribonucleotides or riboxynucleotides, including known analogs of natural nucleotides, unless otherwise specified, in either single- or double-stranded form. .

As used herein, the term “genetic (child) type” refers broadly to, for example,
The genetic composition of an organism, including whether a diploid organism is heterozygous or homozygous for one or more alleles of interest.

The term “primer” refers to a suitable buffer at a suitable temperature and under suitable conditions (
That is, a single strand that can act as a starting point for template-directed DNA synthesis with four different nucleoside triphosphates and an agent for polymerization (eg, in the presence of DNA polymerase or RNA polymerase or reverse transcriptase). Refers to an oligonucleotide. A suitable length of primer depends on the intended use of the primer, but typically ranges from 15 to 30 nucleotides. However, shorter or longer primers can also be used. Short primer molecules generally require cooler temperatures to form sufficiently stable hybrid complexes with the template. A primer need not reflect the exact sequence of the template, but must be sufficiently complementary to hybridize with the template. The term "primer site" refers to the region of target DNA to which a primer hybridizes. The term “primer pair” refers to a 5 ′ upstream primer that hybridizes to the 5 ′ end of the DNA sequence to be amplified, and a 3 ′ downstream primer that hybridizes to the complement of the 3 ′ end of the sequence to be amplified. Means a set of primers that include.

The term “nucleic acid probe” refers to a nucleic acid molecule that binds to a specific sequence or subsequence of another nucleic acid molecule. The probe is preferably a nucleic acid molecule that binds through complementary base pairing with the complete sequence or subsequence of the target nucleic acid. The probe may bind to a target sequence that lacks perfect complementarity with the probe sequence, depending on the stringency of the hybridization conditions. Probes are typically isotopes,
Directly labeled with a luminophore, chromogen, or biotin (
This can be labeled indirectly, such as when using a streptavidin complex (which can later bind). By assaying for the presence or absence of the probe, the presence or absence of the selected sequence or subsequence can be detected.

A “label” is a composition detectable by spectroscopic, photochemical, biochemical, immunochemical, or chemical means. For example, useful labels include ³² P, fluorescent dyes, electron density reagents, enzymes (eg, commonly used in ELISA).
, Biotin, dioxygenin or haptens, and proteins for which antisera or monoclonal antibodies are available (eg, by incorporating a radiolabel into the peptide and using it to detect antibodies that specifically react with the peptide) ). Labels often produce a measurable signal, such as radioactivity, fluorescence or enzymatic activity, which can be used to quantify the amount of bound label.

A “labeled nucleic acid probe” is a probe that binds to a label, either covalently, through a linker, or ionic, van der Waals or hydrogen bonded, so that the presence of the probe. Can be detected by detecting the presence of the label bound to the probe.

The term “selectively hybridizes to” refers to a particular nucleotide sequence under stringent hybridization conditions when present in a complex mixed (eg, total cell) DNA or RNA. Molecular binding, duplexing, or hybridization only to the nucleotide sequences of.
The term "stringent hybridization conditions" refers to conditions under which a probe will hybridize to its target subsequence but not to other sequences. Stringent conditions are sequence-dependent and will be different in different circumstances. Longer sequences hybridize specifically at higher temperatures. Extensive guidelines for nucleic acid hybridization can be found in Tijssen, Techniques.
in Biochemistry and Molecular Biolog
y-Hybridization with Nucleic Probes
, "Overview of principles of hybridiz
ation and the strategy of nucleic ac
id assays "(1993). Generally, stringent conditions are defined as the thermal melting point (T) at a defined ionic strength pH for the particular sequence.
m) is selected to be about 5-10 ° C lower. The Tm is a hybrid of 50% of the probe complementary to the target in equilibrium with the target sequence (50% of the probe in Tm is occupied in equilibrium if the target sequence is present in excess). Soybean temperature (defined ionic strength, pH, and nucleic acid concentration). Stringent conditions are salt concentrations of less than about 1.0 sodium ion at pH 7.0-8.3, typically about 0.01 to about 1.0 M sodium ion concentration (or other Salt), and its temperature is short probe (eg 10-5).
0 nucleotides) is at least about 30 ° C., and long probes (
Conditions of at least about 60 ° C. for more than 50 nucleotides). Stringent conditions may also be achieved with the addition of destabilizing agents such as formamide. If degenerate hybridization is desired, substringent conditions are required. For example, if single nucleotide mismatches are preferred, hybridization conditions are relaxed at lower temperatures and higher salt content.

The term “statistical correlation” refers to a statistical relationship between two variables or parameters, as measured by any statistical test, including, for example, Chi-square analysis, ANOVA or multivariate analysis. Refers to a relationship. A correlation between a polymorphic form of DNA and a random variable or test parameter is statistically significant if the probability that the result will occur by chance (P value) is below a certain level (eg, 0.05). Is considered to be. The term "statistically significant difference" refers to the statistical confidence level P,
This is less than 0.05, preferably less than 0.01, and most preferably 0.
It is less than 001.

“Drug” or “pharmaceutical agent” means any substance used in the prevention, diagnosis, alleviation, treatment or cure of a disease. This term includes, for example, vaccines.

The terms “subject” or “individual” typically refer to a human, which also
It also refers to mammals and other animals, multicellular organisms (eg, plants) and unicellular organisms or viruses. By "tissue" is meant any sample taken from any subject, preferably a human. Tissues include blood, saliva, urine, biopsy samples, skin or cheek (oral cavity) chips, and hair.

[0041]   The term "patient" refers to both human and veterinary subjects.

II: Summary The present invention is useful for designing treatment studies and for assessing the efficacy of various types of treatment procedures as a function of subject genotype (eg, clinical trials). Method,
Provided are computer programs and computerized systems. The methods of the invention are designed to control for the underlying genetic factors that can affect the response to treatment. The present invention is based, in part, on the insight that controlling, either directly or indirectly, the genetic factors affecting patient response to treatment can greatly increase the efficacy of clinical trials. Some methods are designed to reduce the genetic diversity of the patient population so as to increase the probability of individuals sharing the same allele in the genes involved in the response to the treatment. If polymorphisms (usually in genes) are known to be associated with or result in differences in response to their treatment, these polymorphisms can be used directly in the design of clinical trials.

For example, the invention provides methods for reducing variance in a biological state or phenotype of interest by controlling for genetic factors that affect that phenotype. In the context of clinical trials, the phenotype of interest is the response to treatment. The genetic factors can be controlled in a number of different ways, but the principle underlying the method of the invention can be illustrated by way of example (example). If the test parameters are measured in two groups, the first (which is size n) is treated and the second (size is m) is not treated, and the mean and Variance can be calculated in standard ways (Armitage and Be
rry, Statistical Methods in Medical R
esearch, Blackwell Science, 1995). Thus, for example, in one example, the mean and variance of the group to be treated is mu ₁ and s ₁ ^2, respectively, and the mean and variance of the group not treated are each mu ₂ and s ₂ ^2. Then for the difference in response between the two groups,
An appropriate confidence interval of α% (where α = 0.95, for example) is given by:

[0044]

[Equation 1] Here, Ζα _{/ 2} is a standard normal distribution value that happens to exceed α / 2%.

Therefore, any method that reduces the variance in either sample (ie, which reduces s ₁ ² or s ₂ ² ) necessarily reduces the size of the confidence interval. Alternatively, when the variance of one of the two samples decreases, the size of the confidence interval can be considered stationary for fewer patients enrolled in the study (ie, n and / or m is decreased). obtain). Therefore, reducing the variance in the response may result in greater certainty of the difference (here encapsulated by smaller confidence intervals), or reduced sample size for the same statistical force. This variance can be reduced in a number of different ways described in the sections below.

A. Selective Enrollment of Patients One approach is to control for confounding or competent factors by increasing the homogeneity of the population. In the context of genetics, a set of polymorphic markers can be tested in a large population of subjects, and those with similar polymorphic profiles will be enrolled in treatment studies. By incorporating genetic factors (represented by polymorphic profiles) into inclusion / exclusion criteria for treatment studies,
The experimenter will be able to reduce the variance in the response due to the underlying genetic factors.

B. Dividing Patient Populations into Genetically Homogeneous Subsets The second approach is to categorize individuals into subsets depending on how similar their polymorphic profiles are to each other. It is to be. Within each subset, subjects are randomly assigned to treatment or control subpopulations, eg, as in standard clinical trials. This method of dividing subjects produces genetically more homogeneous subsets than random samples of the same size. This design is equivalent to conducting several small independent treatment studies, each containing patients with more similar polymorphic profiles than would be expected by chance. Many environmental variables can be the expression of underlying genetic factors. Testing direct genetic polymorphisms not only makes it possible to reduce variance due to genetic factors that are not directly observable, but also as a surrogate for the underlying genetic factors that control them. Stratification based on functional environmental factors
It is also possible to improve n). Stratification, as used herein, refers to dividing a sample into more similar subsets than one would expect by chance for a given factor.

C. Matching Patients for Their Polymorphic Profiles An alternative approach is to match subjects in treatment and control groups. That is, pairs of individuals with similar polymorphism profiles are searched and one is assigned to the treatment group and the other to the control group. In this way, differences in the response of each pair can be tested.
Here, the pair is matched for its underlying genetics. In a manner similar to that described in Section B above, matching based on genetic factors is
It can control previously unknown causes of variance due to new, genetic factors, and also provide greater discriminatory power when matched by environmental factors with underlying genetic factors.

D. Use of Genes Known to Affect Response in Clinical Trial Design If one or more known polymorphisms are known to be associated with response to treatment, directly Can be used to assign patients to treatment and control groups. In the simplest case where the subject's polymorphism profile indicates whether they respond to treatment,
This information can be used as inclusion / exclusion criteria during enrollment, thus reducing the sample size needed to observe a given level of response. Alternatively, all subjects can be enrolled in treatment studies with assigned treatments that are not random. For example, a patient known to be a non-responder due to its polymorphic profile can be treated (eg, given placebo) according to a control procedure,
On the other hand, patients who are considered responders because of their polymorphic profile can be given treatment procedures (administered drugs). This maximizes the difference in response between the treatment and control groups. Conversely, non-responders may be given the treatment, and responders may be given the treatment. In this case, the minimal difference between treated and untreated subjects can be evaluated.

E. Dose Determination Using Genes Known to Affect Responses If one or more known polymorphisms are known to be associated with response to treatment,
This information can be used to assign the most appropriate dose to subjects enrolled in treatment studies such as clinical trials. A patient's polymorphic profile can determine the extent of an individual's response to the treatment. In this way, it may be possible to assign different doses to different patients, depending on their polymorphic profile. For example, if a treatment can potentially have side effects, it is desirable to administer a minimum effective dose. This may vary for subjects with different polymorphism profiles.

F. Post-Test Analysis and Subdivision The present invention can also be used after completion of treatment studies such as clinical trials. The data obtained from such treatment studies are re-analyzed for a subset of treated and control populations selected for polymorphic profile similarities to each other. Reanalysis of the data is performed on a subset of individuals who share a similar polymorphic profile, and indicates whether the treatment has reached statistical significance for individuals with that profile. If the profile includes one or more polymorphic forms that are in some way associated with a biological condition of interest (eg, a disease), the treatment is not reached by the treatment for the initial treatment population. In some cases, statistical significance may be reached for that subpopulation. Reanalysis of the data also shows a lack of statistical significance if the profile does not include such polymorphic DNA forms.

At this point, further reanalysis is performed. In this re-analysis, a further subpopulation of individuals from the treated and control population is selected for similarity to the second polymorphism profile. Since this individual has already been characterized for polymorphic profiles, a second reanalysis can be performed in a highly automated and iterative fashion without further experimental work. Again, the second analysis will determine whether the treatment has reached statistical significance for individuals with similar polymorphic profiles due to the subpopulations selected in the second analysis. Indicates

Subsequent rounds of analysis may be performed on the same principle, without further experimental work. A properly programmed computer can perform thousands, millions, or billions of cycles of analysis. Here, individuals in different subpopulations are selected based on their similarity to different polymorphism profiles. Performing multiple tests typically requires re-evaluation of the p-value. There, the results are declared to be statistically significant in controlling the proportion of false positive results. After a thorough analysis, if statistical significance is not achieved for any of the polymorphic profiles, the treatment procedure tested (
It can be concluded with a high degree of confidence that (drug administration) does not appear to be effective in any significant part of the population, and further studies are not justified. However, at least two conclusions follow when statistical significance is achieved for a particular polymorphic DNA profile. First, in the case of clinical trials for a drug, where the drug is efficacious in at least part of its population, and further development of the drug can be justified. Second, if the part of the general population for which the drug is effective is known, that part is determined by the polymorphism profile. This profile can be used as a diagnostic factor to identify suitable patients for treatment when a decision or choice of treatment is made.

[0054]   As an example of the method of the invention, clinical trials may be performed as follows.

1. Identification and Selection of Polymorphisms A set of polymorphisms is identified that allows the patient cohort to be divided into subgroups. These polymorphisms may be known to contribute to test parameters (eg, phenotype or endpoint) that should be measured or may be randomly selected. (In the case of random selection, genetic subgroups may show identical results for the phenotype of interest. This is because the method of grouping does not reduce the variance at the end points and the population reanalyzes as a whole. Therefore, stratification by using genetic data has no detrimental effect on the experiment or test, even if it does not affect its output).

2. Genotyping the Cohort Some or all of the markers are genotyped throughout the cohort of patients enrolled in the clinical trial. These data are then used either as inclusion / exclusion criteria (see 3a below) or to divide the cohort into subgroups (see 3b below). .

3a. Inclusion / Exclusion of Patients Using Genetic Information If some or all of the polymorphisms are known to influence the test parameter to be measured, it is known a priori In some cases it may be appropriate to exclude individuals, who present a particular phenotype or endpoint. In the context of clinical trials, this indicates that information gained from the tested set of polymorphisms excludes individuals who do not respond to the treatment.

3b. Division of clinical trials into subgroups A method is used to determine the genetic similarity of patients in the cohort.
This information is used to divide the population into subgroups that have a higher genetic similarity than might be expected by chance. That is, the subgroup is genetically more homogenous than the same large random subset.

The exact method of measuring similarity depends on the number and type of markers used. In the simplest case, two individuals can use the number of markers with the same allele to determine similarity. Many other more complex methods can be used. This gives extra weight to, for example, markers that are particularly informative or known to influence the test parameters of interest.

By modifying the method of determining genetic similarity, the experimenter can control the number of subgroups that need to be formed. For N individuals, this can range from 1 (whole population) to N (each individual is in a separate subgroup). Practical and scientific reasons are considered in determining how many subgroups are optimal for a given experiment or test. Using the method of the invention, groups can be merged at a later time.

4. Allocation of Treatments within a Genetic Subgroup Several strategies are available for dividing the patient into genetic subgroups based on information from the set of polymorphisms described in 1. , Available for conducting treatment studies such as clinical trials.

One method is to randomize treatments and placebos within each subgroup. This is similar to treating each subgroup as a separate experiment or clinical trial. The results of each subgroup can be analyzed separately or can be pooled and then analyzed.

Alternatively, treatments can be assigned randomly within a subgroup. This may be appropriate, for example, if the polymorphism is known to be associated with the desired outcome or endpoint. For example, in the context of a clinical trial, there are only two subgroups, and one of the subgroups is known to include high responders to treatment, and the other is low responders to treatment. , The treatment is assigned to the first group and the placebo is assigned to the second group to maximize the difference between treated and untreated individuals. Conversely, assigning the placebo to the first group and the treatment to the second group demonstrates the minimal difference between treated and untreated individuals. Which of these approaches is most appropriate depends on the exact purpose of the experiment or clinical trial.

5. Use of Information from One Experiment in Designing Subsequent Studies The use of stratification by using a set of genetic polymorphisms will be reassessed through subsequent experiments in clinical trials. Informationless polymorphisms can be dropped and new polymorphisms added to increase the utility of the set as a whole. The use of these polymorphisms in subsequent treatment studies or clinical trials results in greater reproducibility results and in fewer subjects needing to be enrolled in repeated studies.

By identifying and correlating polymorphisms for a particular effect of a drug, and thus reducing the variance due to genetic factors, clinicians include fewer subjects and reduce confidence intervals. Clinical trials can be developed that increase the accuracy or discriminatory power of a given test. The clinician may decide which trial design or analysis of these three aspects will change while keeping the other two stationary.

Modifying the statistics of variance and then influencing the number, accuracy or power of the studies, in addition to affecting the polymorphism in the clinical trial population in a manner as disclosed herein. The use of type marker analysis allows the identification of a subset of polymorphic markers that can be correlated with either a healthy response to treatment, an unresponsiveness or hyper-response, an unwanted response to treatment, or a toxic response during analysis. And unresponsiveness may identify patients in the clinical subset that define "different" diseases. Briefly, posterior genetic analysis that correlates with a particular clinical response, such as drug response or no response, may reveal different etiological mechanisms for the disease being treated. This appears to be especially the case for racial differences between patients, where each racial group has a differential response to treatment. Finally, analysis of phenotypic markers provides insights into the genetic diversity of treated subjects and allows clinicians to modify enrollments in drug trials to accommodate most genetic diversity To This is scientifically wise.

III. Methods A. General In the methods of the present invention, members of the treated and control (untreated) populations having the biological condition of interest (eg, disease) are: It is characterized for the polymorphic profile and test parameters that are a measure of the biological status, assuming that the member is not so characterized. Members in the treated population were treated (or treated) according to the treatment procedure, while members of the control population were treated (or treated) according to the control procedure.

Subpopulations from treated and control populations were selected for similar genetic composition in order to reduce the total variance in treatment evaluations or studies.
As a result, members in the two populations have similar or identical polymorphic profiles. The polymorphic profile of the subpopulation comprises one or more polymorphic forms. Typically, the polymorphic profile will have multiple polymorphic forms, typically at least 5
, In other cases at least 10, and in still other cases at least 100 or any number in between.

In order to minimize the genetic variance between the treated and control populations, the polymorphic profiles for the two groups are chosen to be similar.
This means that at least one common polymorphic form exists between the two subpopulations, but typically more. The polymorphic profiles for the two polymorphic forms are typically at least 10% identical, in some cases at least 50% identical, and in other cases more than 75% identical. Yes, in still other cases 90% or more identical, and in still other cases 100% identical. Analysis of phenotypic markers may provide further insight into the genetic diversity of the treated subjects, and this may lead researchers to alter members in the study to accommodate most genetic polymorphisms. to enable.

Polymorphisms can be selected in three different ways. First, polymorphisms can be randomly selected. Second, only polymorphisms known or suspected to be involved in the phenotype of interest (response to treatment in the case of clinical trials) can be selected. Third, DNA polymorphism selection can be performed by identifying polymorphisms present in regions of the genome that have previously been shown to have genetic linkage to the observable trait under study. In the first case, random polymorphisms do not appear to be directly involved in the response.
However, they can be used to determine the genetic similarity of patients and thus can be used to form more genetically homogenous subgroups beyond chance. This strategy is particularly effective when a large number of (usually unknown) genes are involved in determining the individual's response to treatment. If there are polymorphisms known to be involved in the response, or associated with (probably unknown) genes involved in the response, these may be used preferentially to determine the subgroup of patients.

All polymorphisms are neither equally informative nor useful in determining subgroups. Allele frequency, whether the polymorphism encodes or does not encode a protein, whether the polymorphism is already in linkage disequilibrium with another polymorphism in the polymorphic profile used, whether the polymorphism is associated with treatment Factors such as whether or not in linkage disequilibrium with the gene or genes known to be involved in the response can all influence the utility of a given polymorphism. Note that in some cases, it may be desirable to give some polymorphisms more weight than others in forming subgroups. That is, polymorphisms known to be associated with responsiveness may be more important (and therefore more weighted) than random polymorphisms.

This polymorphism may be present in genomic DNA, RNA or cDNA. Any polymorphism may be used, but of particular importance is a polymorphism in a gene encoding a protein that directly or indirectly affects a biochemical pathway that correlates with the biological state being measured or observed. It is a type. Thus, for example, where studies include assessing the effectiveness of methods for treating patients with high blood cholesterol levels, polymorphism profiles indicate that genes known to be involved in cholesterol synthesis and metabolism. Can be modified to include polymorphisms located at

Once a suitable subpopulation has been selected so that it has a desired level of similarity in the polymorphic profile, the polymorphic profile and the efficacy (or lack thereof) of the treatment method are It is determined whether there is a correlation between them. This is by ascertaining whether there is a statistically significant difference in the test parameters between the treated and control subpopulations. Here, the test parameter is a measure or representative of the efficacy of treatment for a biological condition shared by members of the subpopulation. A polymorphic form in the polymorphic profile of the treated subpopulations correlates with its biological status if a statistically significant difference is found (eg, the polymorphic profile correlates with a particular disease) And that the treatment method under study is useful (or lacking benefit) for treating a subject with that biological condition.

As mentioned above, such correlations are of particular importance, for example, in clinical trials for drugs. In some cases, the correlation is associated with the disease 1.
It identifies a set of genetic markers and therefore has diagnostic value. In other cases, the correlation identifies markers associated with positive treatment outcomes, and thus it is important from a therapeutic perspective.

Statistically significant differences in test parameters between treated and control populations can be determined using standard methods of statistical analysis. Methods include, for example:
These include, but are not limited to: ANOVA, logistic regression, cluster analysis, non-parameter statistics, contingency test, and other standard statistical tests.

B. Method Iteration The initially selected subpopulation polymorphism profiles are often correlated with statistically significant differences in the test parameters used to measure efficacy of treatment. do not do. In such cases, the method is directed to different subpopulations produced by using alternative definitions or measures of genetic similarity, or by dividing the population into larger or smaller subpopulations. Can be repeated with. This reflects the fact that the only unique method for grouping patients rarely exists. In fact, for studies with N individuals, often 1
It is possible to form any number of subpopulations (from the entire population) to N (each individual being in its own subpopulation). Repeating this process is often an effective way to detect within a polymorphic profile which polymorphism is particularly informative with respect to the test parameters of interest. Once a correlation has been identified, additional cycles are utilized in earlier cycles to determine, for example, whether the subset may show a greater correlation with the test parameters and thus treatment efficacy. It can be repeated with one subset of polymorphic forms.

Typically, polymorphic forms in the polymorphic profile evolve over time and are responsible for a larger portion of the genetic component of their variance. However, these polymorphic forms generally do not contribute to equivalence. Some cause more dispersion than others. Markers that do not correlate with differences between treatment and control procedures are discarded from this analysis. The set of markers as a collection has a different value than the individual markers. This set has durable value for understanding genetic contributions to different biological states of interest. Individual markers can have diagnostic utility, as can their population.

Analysis of treatments or tests involving groups or subgroups is accepted for both statistical and non-parameter (non-variance method) statistical methods.

C. Treatment and Control Groups Members of the treatment and control groups all share some biological status. For that condition, the study is designed to determine if the treatment procedure has a statistically significant differential effect compared to the control procedure. Members of the treatment and control groups can be essentially any type of organism, including: humans, non-human animals, plants, bacteria and viruses.
In some cases, the member is a mammal (eg, human, primate). This is part of a clinical trial that includes, for example, testing of pharmaceutical agents or behavioral therapies. The number of members in the subgroup selected from the treatment and control groups is at least 1, but generally exceeds 1, typically at least 5, and in other cases at least 10. Yes, and in yet other cases at least 100 or more, but any number of members between these numbers can also be selected.

In some cases, members of the subpopulation are selected not only to have similar polymorphic profiles, but also other common features. Selecting on the basis of other common attributes may further reduce the total variance beyond that achieved by reducing the variance contributing to genetic factors. Thus, for example, members of the treatment and control subpopulations can also be selected to be similarly exposed to environmental factors. Examples of such environmental factors include, but are not limited to, m or less: exposure to various factors such as radiation, chemicals and sidestream smoke; geographical location; and sleep patterns, diet and exercise. Life cycle features. In some cases, it may be useful to conduct studies with subpopulations that were similarly exposed to environmental factors; such studies would select subpopulations for similar exposure to particular environmental factors. Can serve as a contrasting element to research. In addition to environmental factors, members may also be selected from the same racial group or may be selected to share common phenotypic traits (eg, visual acuity, caution, weight, physical abnormalities).

When the methods of the invention are utilized in clinical trials, typically the two groups of subjects are not otherwise treated for their pathological condition. In some cases, the study is designed so that subjects in both treated and untreated groups receive the same alternative treatment.

D. Treatment and Control Procedures The types of treatment and control procedures vary according to the biological condition to which the treatment is directed. As mentioned above, the biological condition can be any of a number of conditions, such as, for example, a pathological condition or simply biosensitivity. A variety of different procedures can be performed when the biological condition is a pathological condition. In many cases, the procedure involves administering a pharmaceutical agent, including, for example:
1) administering a pharmaceutical agent to a member of the treated population and giving members in that control population placebo or no placebo, 2
) Giving one member of the treated population one pharmaceutical agent (or combination of pharmaceutical agents) and a different pharmaceutical agent (or combination of pharmaceutical agents) to its control member; 3) an amount. Providing the pharmaceutical agent to the treated population and providing different amounts to the control population, or 4) administering the pharmaceutical agent to the treatment population and the control population according to different schedules.

Instead of administering a pharmaceutical agent, the treatment procedure may involve some type of behavioral therapy. Examples of such treatments include, but are not limited to, certain dietary regimens (eg, low-fat, low-sodium, high-protein, or calorie-restricted diets), prescribed exercise regimens (eg, 1 Exercise a certain number of times a week for a certain period of time, do low-load exercise, exercise to reach your target heart rate, treatments that affect specific muscle groups), meditation, yoga, and stress reduction techniques. Of course, the treatment procedure may also include a combination of the above procedures. Members in the control group may receive no therapy at all, or may be treated in the opposite fashion (or may already be engaged in the opposite behavior). For example, if the treatment group is fed a low-calorie diet, the control group may be fed a high-calorie diet, or simply the normal diet is already a high-calorie diet and is therefore unchanged. Can be selected for.

The treatment groups may also be directed against biosensitivity or resistance rather than pathological conditions. Thus, for example, in the case of plants, the plant may be loaded with various pesticides (eg, fertilizers or other growth stimulators, herbicides, insecticides and pH modifiers) used to affect the growth or health of the plant. It can be treated with to assess the effect of such agents on various susceptibility and resistance of plants (eg susceptibility to frost or frost damage or resistance to herbicides). Similarly, humans or other organisms have also been treated with various agents (eg, vaccines),
The effect of the drug on various susceptibility or resistance can be determined.

E. Utility The reduction in variance achieved by the methods of the present invention allows investigators to selectively optimize treatment studies. For example, as the genetic variance decreases, the confidence level of statistical analysis increases. Thus, using the methods of the present invention, the investigators can attribute differences in the effects seen between treated and control subjects to the treatments administered rather than to the result of genetic differences between patients. You can be strongly convinced that it is due. Furthermore, the differences between the control and test groups can be recognized more quickly. This allows less and less costly research to be done. This study has the same statistical capacity as a much larger study that does not fit the underlying genetics. Alternatively, studies in which patients were matched for genetic factors detected much smaller differences in response between treated and untreated individuals than studies of the same size ignoring genetic factors. You can This allows for faster marketing of pharmaceutical companies in less costly studies, faster assessment of the feasibility and desirability of further treatment studies, and ultimately, for example, in clinical trials for pharmaceutical compounds. To

This method also makes it possible to design more efficient treatment studies. For example, once a polymorphism that correlates with a pathological condition has been identified, the subject having that polymorphism and its biological state have an effect on the biological condition of interest that other treatments have. It can be identified and enrolled in further studies for analysis. Fewer subjects need to be enrolled because subjects who do not respond to the treatment are not enrolled. Alternatively, it correlates very well with the biological condition studied if it fits between the patient in the control and the test arm in the trial 1
If a set of polymorphisms appears, subsequent testing of treatment efficacy may be tested with fewer patients, regardless of response rate if the biological condition being measured has a genetic component. Furthermore, if polymorphisms associated with different responses are identified, the dose received by a particular patient may be modified to optimize their given polymorphic profile. This is especially important if there are unwanted side effects of the treatment and it is desired to give a minimal dose.

Further, as noted above, the methods of treatment described hereinabove provide for a subset of polymorphic forms that correlate with either a favorable response or non-responsiveness to treatment, or an unwanted or toxic response to that treatment. Enables identification of Clinical trials for the efficacy of particular pharmaceutical treatments can identify individuals who do not respond to treatment, and by doing so, in some cases, identify a subset of clinical patients that define "different" diseases. Bring Such correlations may also be used as prognostic and / or diagnostic tools to identify subjects who have or appear to have the disease, or based on the particular genetic composition of the subjects. , The appropriate treatment means for that subject can be selected.

Information obtained from clinical trials in which patients are genotyped for a set of polymorphic genetic markers can be used in other stages of drug discovery and development.
For example, genes that have been shown to be associated with responses via a patient's polymorphic profile may be accepted for intervention and, therefore, represent potential drug targets. In addition, the identification of treatments that exhibit low efficacy (ie, many non-responders) or have high-efficiency adverse events can be identified by examining the patient's polymorphism profile in early-stage trials. This information can then be used in a decision as to whether or not to proceed with a larger and more costly trial. For example, it may be inappropriate to use treatments in a larger trial if non-response is associated with a common polymorphic profile common to the general population.

IV. Computer System and Program FIG. 1 shows a representative computer system 10 suitable for performing the particular methods of the present invention. As shown in FIG. 1, computer system 10 typically includes a bus 12. The bus 12 includes a main subsystem (eg, the central processor 14), a system memory 16, an input / output controller 18, external devices such as a printer 23 via a parallel port 22, a display adapter 26.
Interconnected with a display screen 24, a serial port 28, a keyboard 30, a fixed disk drive 32 via a storage interface 34, and a floppy disk drive 33 operable to receive a floppy disk 33A. Has been done. The scanner 60 via the I / O controller 18,
A mouse 36 connected to the serial port 28, a CD-ROM player 40 operable to receive a CD-ROM 42, or many other devices such as a network interface 44 may be connected. The source code for carrying out the invention may be operably located in system memory 16 or may be stored on a storage medium such as fixed disk 32 or floppy disk 33A. Other devices or subsystems may be connected in a similar fashion. Not all of the devices shown in Figure 1 are required to carry out the invention. The devices and subsystems may also be interconnected in different ways than shown in FIG. Computer system 10 as shown in FIG.
The operation of is well known in the art. Therefore, the operation of the system will not be described in detail herein.

FIG. 2 is an illustration of the exemplary computer system 10 of FIG. 1 suitable for carrying out the method of the present invention. However, FIG. 2 illustrates only one example of many possible computer types or configurations that may be used with the present invention. As shown in FIG.
The computer system 10 includes a display screen 24, a cabinet 20 and a keyboard 3.
0, scanner 60, and mouse 36. The mouse 36 and the keyboard 30 are examples of “user input device”. Other example user input devices include, but are not limited to, touch screens, light pens, trackballs and data grabbing.

Mouse 36 may have one or more buttons, such as button 37. Cabinet 20 houses conventional computer components such as floppy disk drive 33, processor 14 and storage means (see FIG. 1). As used herein, "storage means" may comprise any storage device that can store data that can be used with a computer system. Examples of such devices include, but are not limited to, disk drives, magnetic tape, solid state memory and bubble memory. The cabinet 20 connects the computer system 10 to an external device (for example, the scanner 60).
, External storage, other computers or additional peripheral devices) may be provided with additional hardware such as an input / output (I / O) interface.

In some examples, the system 10 comprises a Pentium® microprocessor 14. This is Microsoft Corporation
on WINDOWS® version 3.1, WINDOWS (
It runs the ™ 95 or WINDOWS ™ 98 operating system (OS). However, the inventive method can be readily adapted to other operating systems (eg UNIX) without departing from the scope of the invention.

FIG. 3 is a flow chart of simplified steps in one computerized method of the present invention for evaluating treatment procedures. In step 100, a database containing multiple designations for each member of the population that has been treated according to either the treatment procedure or the control procedure or shares a common biological condition, the database of which is described further below. Will be provided. Therefore, the population includes both treatment and control populations. The designation of one group is
This is to identify each member of the two populations. Also, typically, there will be separate designations for polymorphic profiles and test parameters for each member of the population. Subpopulations from the treatment and control populations are selected for similarity in polymorphic profiles in selection step 102. At decision step 104, a determination is made whether there is a statistically significant difference in test parameters between the subpopulations. A statistically significant difference indicates that the polymorphic profile of the subpopulation correlates with the biological condition and the effect of treatment. Finally, in a display step 106, the resulting output of the decision step is displayed on an output device to facilitate its analysis. In the optional decision step 108, if there is no statistically significant difference in the test parameters for the two subpopulations, the selection step 102, the decision step 104, and the display step 106.
Is repeated with a subpopulation having a polymorphic profile different from that of the early cycles.

Accordingly, the microprocessor in the computer system of the present invention is operably arranged with respect to its system memory, its system bus and its inputs / outputs, so that the functions described above are performed. For example, the processor provides or receives data including designations for each member of the treated and control populations, as well as polymorphic forms and test parameters for each member of the two populations. The microprocessor also selects subpopulations from each of the treated and control populations for similarities in polymorphic profiles, and whether there are statistically significant differences in test parameters between the subpopulations. And operatively arranged to display the resulting output.

The computer program of the present invention includes code for providing or receiving data including various designations for the identity of members of the test and control populations, their polymorphic profiles and test parameter results. The program may also include the code necessary to carry out the selection, determination and display steps described above.

V. Methods for Determining Polymorphism Profiles A. Sample Preparation Polymorphisms are detected in target nucleic acids from the individual being analyzed. Genome DN
Virtually any biological sample (other than pure red blood cells) for the A assay
Is appropriate. For example, simple tissue samples include whole blood, semen, saliva, tears, urine,
Includes fecal material, sweat, cheeks, skin and hair. For cDNA and mRNA assays, the tissue sample needs to be obtained from the organ in which the target nucleic acid is expressed. For example, a cD whose target nucleic acid encodes cytochrome P450
If NA encoded, the liver is a suitable source.

Many of the methods described below require amplification of DNA from a target sample. This can be achieved, for example, by PCR. In general, PCR TECHNOLOG
Y: PRINCIPLES AND APPLICATIONS FOR DN
A AMPLIFICATION (edited by HA Erlich et al., Freeman)
Press, NY, NY, 1992), PCR PROTOCOLS: A
GUIDE TO METHODS AND APPLICATIONS (In
nis et al., Academic Press, San Diego, CA, 19
90), Mattilla et al., Nucleic Acids Res. 1
9: 4967 (1991), PCR Methods a by Eckert et al.
nd Applications 1:17 (1991), PCR (McPhe
Rson et al., IRL Press, Oxford), and US Pat.
83,202, each incorporated herein by reference for all purposes.

Other suitable amplification methods include ligase chain reaction (LCR) (Wu and Wall
ace, Genomics 4: 560 (1989), Landegren et al.,
Science 241: 1077 (1988)), transcription amplification (Kwoh et al., Proc. Nat'l Acad. Sci. USA 86: 1173 (198).
9)), as well as autonomous sequence replication (Guatelli et al., Proc. Nat'l.
Acad. Sci. USA 87: 1874 (1990)) and nucleic acid based sequence amplification (NASBA). The latter two amplification methods include an isothermal reaction based on isothermal transcription, which results in the amplification product, single-stranded RNA (ss).
Both RNA) and double-stranded DNA (dsDNA) are produced at ratios of about 30 or 100 to 1, respectively.

B. Detection of Polymorphisms in Target DNA There are two different types of analysis depending on whether the polymorphism in question has already been characterized. The first type of analysis is sometimes referred to as new characterization. This analysis compares target sequences in different individuals to identify points of modification, ie polymorphic sites. The second type of analysis involves determining which form (s) of the characterized polymorphism is present in the individual under test. There are various suitable procedures for determining polymorphic forms, and thus polymorphic profiles, including, for example:

1. Allele-Specific Hybridization (ASH) The ASH technique is based on the stable annealing of short, single-stranded oligonucleotide probes to perfectly complementary single-stranded target nucleic acids. Hybridization is detected from the radioactive or non-radioactive label of the probe. For each polymorphism, two or more different probes have the same D except on the polymorphic nucleotide.
Designed to have NA sequence. Each probe contains one allelic sequence,
It is completely homologous and the complement of the probe can distinguish all alternative alleles. Using proper probe design and stringency conditions, a single base mismatch between the probe and target DNA interferes with hybridization. In this manner, only one of the alternative probes hybridizes to a target sample that is homozygous for the allele (the allele is
Defined by the DNA homology between the probe and target). A sample containing DNA that is heterozygous for the two alleles hybridizes to both of the two alternative probes. For details about ASH, see Sa, for example.
iki et al., Nature 324: 163-166 (1986), Datatag.
upta, EP 235,726, Saiki, WO 89/11548, and US Pat. No. 5,468,613.

(2. Restriction Fragment Length Polymorphism (RFLP)) “Restriction fragment length polymorphism” or “RFLP” is caused by an inherent difference in a restriction enzyme cleavage site (for example, a base change at a target site). ),
Alternatively, it refers to additions or deletions in the regions flanking the restriction enzyme sites that result in fragment length differences produced by cleavage with the relevant restriction enzymes. A point mutation results in a longer fragment if the mutation occurs at a restriction site and a shorter fragment if the mutation creates a restriction site. Addition and transfer elements result in longer fragments, and deletions result in shorter fragments.

RFLPs can be used as a genetic marker in the determination of allelic segregation by quantitative phenotype. In one embodiment of the invention the restriction fragment is linked to a particular phenotypic trait. More specifically, particular restriction fragments can be used to predict the prevalence of particular phenotypic traits.

3. Direct Sequencing Polymorphisms can be represented by traditional dideoxy chain termination methods or Maxam-Gilbert methods (Sa
mbrook et al., MOLECULAR CLONING, A LABORAT
ORY MANUAL (2nd edition, CSHP, New York 1989), and Zyskind et al., RECOMBINANT DNA LABORATO.
RY MANUAL (see Acad. Press, 1988)). Other nucleic acid sequencing methods that can be used include fluorescence-based techniques (
US Pat. No. 5,171,534), mass spectroscopy (US Pat. No. 5,174,96).
2), and capillary electrophoresis (US Pat. No. 5,728,282), but is not limited thereto.

4. Drug Tagging of Oligonucleotides for Electrophoresis Oligonucleotides with additional chemical moieties that account for the different mobilities in electrophoretic separation systems can be used in polymorphism analysis. The addition of molecular species increases the apparent molecular weight of the amplified DNA fragment, even in the presence of DNA having only one nucleotide polymorphism. Added seeds,
Or the drug tag is labeled (for visualization of electrophoretic bands) before
Alternatively, it can be covalently or non-covalently attached to nucleic acids, either after being labeled. Any charges associated with the added species are blocked or neutralized, so that nucleic acid mobility remains size dependent rather than charge.

Any of a number of different moieties can be attached to nucleic acids to form multiple different sized amplification products. Examples of such moieties include, but are not limited to, phosphate monomers, acrylamides, and polypeptides. Thus, for example, a phosphate monomer unit can be attached to each nucleotide monomer that can be used in a PCR reaction prior to amplification of the polymorphic stretch of DNA. For example, one phosphate monomer is dATP, two phosphate monomers are dCTP,
3 phosphate monomers are dGTP, 4 phosphate monomers are dTTP
Be added to. The resulting amplified polymorphic nucleic acid contains different amounts of phosphate monomer, depending on the nucleotide content. Thus, if the amplified products have the same number of base pairs, the different polymorphic forms may nonetheless be electrophoretic gel separated sizes due to differences in the content of the phosphate monomers. In a variation of this general approach, monomer units may be added to specific nucleotides (eg, by capillary electrophoresis), or unamplified nucleic acids, after amplification and prior to size-based separation.

5. Isoenzyme Markers Other embodiments include identification of isoenzyme markers and allele-specific hybridization. Isoenzymes are a group of enzymes that catalyze the same reaction but differ in the physical properties that result from differences in the sequence of amino acids (and thus nucleic acid sequences). Some isoenzymes are multiply linked enzymes that contain slightly different subunits. Other isoenzymes, either multimers or monomers, are cleaved from the proenzyme at different sites in the amino acid sequence. Mutations in the isoenzyme nucleic acid can be determined by hybridizing a primer flanking the mutated portion of the isoenzyme nucleic acid amino sequence to a target nucleic acid contained in a sample obtained from the organism. The variable region is amplified and sequenced. From the sequence, different isoenzymes were determined,
And associated with phenotypic traits.

6. Amplified Variable Sequences The amplified variable sequences of genomic and complementary nucleic acid probes can also be used as polymorphic markers. The term "amplification variable sequence" refers to a genomic amplification sequence that exhibits high nucleic acid residue variability among members of the same species. All organisms have variable genomic sequences, and each organism (except clones) has a different set of variable sequences. Specific variable sequences can be used to predict polymorphic traits. The variable sequence of DNA is
Amplification can be accomplished by template-dependent extension of primers (eg, utilizing the amplification techniques described above) that hybridize to flanking regions of DNA obtained from the subject. The amplified product is then sequenced.

7. Allele-Specific Primers and Hybridization Allele-specific primers hybridize to a site in the target DNA that overlaps the polymorphism and amplify the allelic form to which the primer exhibits perfect complementarity. Just prime. This primer is used with a second primer that hybridizes at the distal site. Amplification proceeds from the two primers and produces a detectable amplified product that can be characterized for the particular allelic form present in the nucleic acid sample. For example, Gibbs Nucleic A
cid Res. 17: 2427-2448 (1989) and WO93 / 2.
See 2456.

8. Single-Stranded Conformation Polymorphism Analysis Alleles of the target sequence are single-stranded conformational polymorphisms that identify base differences by alterations in the electrophoretic migration of single-stranded PCR products. Type analysis can be used to distinguish (eg, Orita et al., Proc. Nat'l Acad. Sc).
i. USA 86: 2766-2770 (1989)). Typically, the amplified PCR product is denatured (eg, according to known chemical or thermal methods) and may refold secondary structure depending in part on the base sequence of the product, or It forms a single-stranded amplification product that can form. The different electrophoretic mobilities of single-stranded amplification products may be related to differences in the base sequence strands between alleles of the target sequence.

9. Autonomous Sequence Replication Polymorphisms can also be identified by autonomous sequence replication. In this approach, the target nucleic acid sequence is under isothermal conditions using three enzymatic activities involved in retroviral replication: (1) reverse transcriptase, (2) RNase H, and (3) DNA-dependent RNA polymerase. , In vitro, exponentially amplified (replicated) (
Guatelli et al., Proc. Natl Acad. Sci. USA
87: 1874 (1990)). By mimicking the retroviral strategy of RNA replication by the cDNA intermediate, a copy of the original target cDNA and RNA
Copies are accumulated.

10. Arbitrary Fragment Length Polymorphism (AFLP) Arbitrary fragment length polymorphism (AFLP) can also be used as a polymorphism (Vo).
s et al., Nucl. Acids Res. 23: 4407 (1995)). the term"
"Arbitrary fragment length polymorphism" refers to selected restriction fragments that are amplified before or after cleavage by a restriction endonuclease. The step of amplification allows for easier detection of a particular restriction enzyme fragment as compared to determining the size of all restriction fragments and comparing the sizes to known controls.

AFLPs allow the detection of multiple polymorphic markers (see above) and have been used in genetic mapping of plants (Becker et al. Mol. Gen. Ge.
net. 249: 65 (1995), and Meksem et al., Mol. Gen.
Genet. 249: 74 (1995)), distinguishing closely related bacterial species (Huys, Int'l J. Systematic Bacteriol).
46: 572 (1996)).

11. Simple Sequence Repeats The SSR method uses high levels of dinucleotides, trinucleotides,
Or based on tandem repeats of tetranucleotides. Dinucleotide repeats that have been reported to occur in the human genome produce 5 dinucleotide repeats, with n varying from 10 to 60.
000 times (Jacob et al., Cell 67: 213 (1991)). Dinucleotide repeats are also found in higher plants (Condit & Hub.
bell, Genome 34:66 (1991)).

SSR data is generated by hybridizing primers to conserved regions of the genome that flank the SSR region. The dinucleotide repeats between the primers are amplified by PCR. The resulting amplified sequences are then electrophoresed to determine size and thus the number of dinucleotide, trinucleotide, or tetranucleotide repeats.

12. Denaturing Gradient Gel Electrophoresis Amplification products produced using polymorphic chain reactions can be analyzed using denaturing gradient gel electrophoresis. Different alleles are identified based on different sequence-dependent melting properties and electrophoretic migration of DNA in solution. Erlich, PCR TEC
HNOLOGY, PRINCIPLES AND APPLICATIONS
FOR DNA AMPLIFICATION, (WH Freema
n and Co, New York, 1992), Chapter 7.

(13. Single-base extension method) Polymorphism can also be detected by single-base extension. A primer is designed to hybridize to a target sequence, thereby
'The ends are immediately adjacent to the polymorphic site but do not overlap. The target sequence is then
Contact with a primer and at least one nucleotide (typically labeled) that is complementary to the bases occupying the polymorphic site in one allele. If the allele is present, the primer is extended and labeled. In some methods, by including two differentially labeled dideoxynucleotides, each complementary to a base occupying the polymorphic site in the first and second alleles of the target, Allelic polymorphic sites are analyzed. Analysis of the label present on the extended primer indicates whether one or both allelic forms are present in the target sample.

C. High Throughput Screening In some examples, identification of polymorphisms is performed by high throughput screening. In one embodiment, high throughput screens include RFLP, AFL
Providing a library of polymorphic forms of DNA containing variable sequences including P, isoenzymes, specific alleles, and SSRs. Such a "library" is then screened against genomic DNA from the subject in a treatment trial. Once a subject's polymorphic allele is identified, the linkage between the polymorphic DNA and the treatment effect can be determined through statistical association.

Such high throughput screening can be done in many different formats. For example, these methods include hybridization reactions, where the hybridization is in a 96-, 324-, or 1024-well format.
Alternatively, it can be done in a matrix on a silicon chip. In a well-based format, a dot blot apparatus is used to deposit a sample of fragmented and denatured genomic DNA onto a nylon or nitrocellulose membrane. After cross-linking the nucleic acid to the membrane, the membrane is incubated with the labeled hybridization probe by exposure to UV light if a nylon membrane is used and by heating if nitrocellulose is used. The membrane is washed extensively to remove unhybridized probe and the presence of label on the probe is determined.

The label is incorporated into the nucleic acid probe by any of a number of methods well known to those of skill in the art. In some examples, the label is simultaneously incorporated during the amplification procedure in the preparation of nucleic acid probes. Thus, for example, polymerase chain reaction (PCR) with labeled primers or labeled nucleic acids provides labeled amplification products. In other embodiments, transcription amplification with labeled nucleotides (eg, fluorescein-labeled UTP and / m or CTP) incorporates the label into the transcribed nucleic acid probe.

Detectable labels suitable for use in the present invention are by spectroscopic, radioisotope, photochemical, biochemical, immunochemical, electrical, optical, or chemical means. Includes any composition that is detectable. Labels useful in the present invention include biotin for staining with a labeled streptavidin conjugate, magnetic beads, fluorescent dyes (eg fluorescein, Texas red, rhodamine, green protein etc.), radioisotope labels (eg. , ³ H, ¹²⁵ I, ³⁵ S, ¹⁴ C, or ³² P
), Enzymes (eg, horseradish peroxidase, alkaline phosphatase, and others commonly used in ELISA), and beads such as colloidal gold or colored glass or plastic (eg, polystyrene, polypropylene, latex, etc.). A colorimetric label is included. Patents teaching the use of such labels include US Pat. Nos. 3,817837, 3,850,7.
52, No. 3,939,350, No. 3,996,345, 4,277,43
Nos. 7, 4,275,149, and 4,366,241 are included.

Methods of detecting such labels are also well known to those of skill in the art. Therefore,
For example, radioisotope labels are detected using photographic film or scintillation counters and fluorescent markers are detected using photodetectors to detect emitted light. Enzyme labels are typically detected by providing a substrate for the enzyme and by detecting the reaction product produced by the action of the enzyme, and a colorimetric label is a simple labeling of the colored label. Detected by visualizing into.

Well-known robotic systems have been developed, especially for high throughput screening in 96-well format. These systems include automated synthesizers developed by Takeda Pharmaceutical Co., Ltd. (Osaka, Japan), as well as many robotic systems (Z) that utilize robotic arms that mimic manual synthetic actions performed by chemists.
Ymate II, Zymark Corporation, Hopkinto
n, Mass. Orca, Hewlett-Packard, Palo Al
to, Calif. ) Include automatic workstations. Any of the above devices are suitable for use with the present invention. Modifications of these devices, if any, made to operate as described herein will be apparent to those skilled in the art in their nature and implementation.

In addition, high throughput screening systems are commercially available per se (eg, Zymark Corp., Hopkinton, MA, Air Tec).
hnical Industries, Mentor, OH, Beckman
Instruments, Inc. , Fullerton, CA, Preci
sion Systems, Inc. , Natick, MA, etc.). These systems typically run all samples and reagents in a pipetting step,
All procedures are automated, including liquid dispensing, incubations run at specified times, and final reading of microplates or membranes in detectors appropriate for the assay. These configurable systems offer high throughput and fast start-up, as well as a high degree of flexibility and customization. Manufacturers of such systems provide various high throughput detailed protocols.

D. Solid Phase Arrays Polymorphic forms of DNA can also be identified by hybridization to nucleic acid arrays, some examples of which are described in WO 95/11995 (for all purposes herein). Incorporated by reference in). Modification 1 of the present invention
In one, the solid phase array is adapted for rapid, specific detection of multiple polymorphic nucleic acids. Typically, the nucleic acid probe is linked to a solid support and the target nucleic acid is
It hybridizes to the probe. Either the probe or the target, or both, can be labeled, typically with a fluorophore. If the target is labeled, it is detected by detecting the bound fluorescence. When the probe is labeled, hybridization is typically detected by quenching the label with bound nucleic acid. When both the probe and its target are labeled, detection of hybridization is typically done by monitoring the color shift caused by the proximity of the two bound labels.

The construction and use of solid phase nucleic acid arrays to detect target nucleic acids has been extensively described in the literature. Fodor et al., Science 251: 767 (
1991), Sheldon et al., Clin. Chem. 39 (4): 718 (
1993), Kozal et al., Nature Medicine 2 (7): 75.
3 (1996), and Hubbell, US Pat. No. 5,571,639. See also Pinkel et al., PCT / US95 / 16155 (WO 96
/ 17958). Design probe arrays using available technology,
In addition to being capable of being built and used, those skilled in the art can order custom arrays and array readout devices from manufacturers specializing in array manufacturing. For example, Affym of Santa Clara, CA
etrix manufactures DNA VLSIP ^(TM) arrays.

It will be appreciated that the probe design will be influenced by the intended application. For example, if several probe-target interactions are detected in one assay, eg one DNA chip, it is desirable to have similar melting points for all purposes. Therefore, the length of the probes is adjusted so that the melting points for all probes on the array are very similar (if different probes have different GC contents, different lengths for different probes will be different). May be required to achieve Tm). Melting point is a major consideration in probe design, but other factors are optionally used to further regulate probe assembly.

E. Capillary and Microchannel Plate Electrophoresis As mentioned above, certain methods of identifying polymorphisms include size-based separation (
For example, RFLP and SSR) are included. In such cases, the capillary electrophoresis method can be used to analyze the polymorphism. Such a technique is disclosed in US Pat.
, 534,123 and 5,728,282.
These patents are incorporated herein by reference. Briefly, a capillary electrophoresis tube is filled with a separation matrix. The separation matrix comprises hydroxyethyl cellulose, urea, and optionally formaldehyde.
A sample of RFLP or SSR is loaded into this capillary tube and electrophoresed. The run time is very short due to the small amount of sample and separation matrix required for capillary electrophoresis. Molecular size, and thus the number of nucleotides present in a nucleic acid sample, is determined by the techniques described herein.

Electrophoresis can also be performed in microchannel plates. These plates have channels less than 100μ in diameter etched into a solid substrate. The small size allows for faster separation of nucleic acids within the separation matrix. With these etched plates, samples can be evaluated in high throughput format.

In another high throughput format, multiple capillary tubes are placed in a capillary electrophoresis apparatus. The sample is loaded into the tube and electrophoresis of the sample is performed simultaneously. For example, Mathies and Huang,
See Nature 359: 167 (1992). The separation matrix is less viscous so that after each run the capillary tube can be emptied and reused.

It should be understood that the following examples are provided to further illustrate certain aspects of the invention and are not intended to limit the scope of the invention.

Example 1 Effect of Gene Matching on Sample Size and Reliability The present invention is demonstrated by an example of a test for serum cholesterol and the effect that a drug may have on such biological conditions. Can be done. 8 of serum cholesterol dispersion
It has been demonstrated that up to 0% can be due to heredity (eg RA Kin.
g, J. I. Rotter and A.D. G. Motulsky, THE GENE
TICS BASIS OF COMMON DISEASE, Oxford
(University Press, 1992).

The utility of locus-matching patients is that the reliability, sample size, or discriminatory power of a given test can be positively affected by gene matching.
For example, a test may detect 80% at a 5% level to detect a difference of 20 mg / dl.
, And za = 1.645 (more than 5% of cases representing values from a standard normal distribution) and zb = 0.842 (values of a standard normal distribution exceeding 80% of cases). ), The minimum sample size is calculated as:

[0133]

[Equation 2] If the genetic contribution to the variance of the variable x is 80% and the variance is 1600 (sample standard deviation squared with respect to cholesterol), then the sample size is

[0134]

[Equation 3] Is.

[0135] Therefore, 50 patients per study arm are needed to study the 20 mg / dl cholesterol difference. If the patients are genetically matched (ie, the 80% contribution to variance is eliminated), the number of patients is reduced to 10.

[0136]

[Equation 4] The power of genetic matching can be achieved in two other ways. For example, with the same number of patients (50) in each department, the difference solved by the same force by genetic matching falls from 20 to 8.8. Similarly, the test power increases from 0.8 to greater than 0.99, assuming genetic fit and 20 differences. If the genetic match is not completely achieved, even a certain amount of match can positively impact such tests. This is shown in Table 1 below.

[0137]

[Table 1] Example 2 Use of Genetic Analysis to Reduce Required Sample Size in Clinical Trials The following example demonstrates the underlying genetics that influence treatment response and the use of this information in the design of clinical trials. A method of identifying a physical factor is illustrated.

A. Genetic Subpopulations In the example of two different genetic subpopulations A and B, respectively associated with low and high response to treatment, treated from the first subpopulation A. The individual response has mean μ _A and variance σ ² _A. In the second subpopulation, the response mean and variance are μ _B and σ ² _B , respectively. No assumptions are made about the shape of any distribution (ie, it need not be normal). In untreated control individuals (given placebo instead), the mean and variance of the responses are for subpopulations A and B, respectively.

[0139]

[Equation 5] Is. When samples are taken from these populations, the distribution of sample means is
It is normally distributed. For example, if individuals of size N of treated individuals are taken from subpopulation A, the sample mean is the distribution Norm [μ _A , σ ² _A / N].

When ignoring the genetic background of the subpopulations, both the treated (case) and untreated (control) populations contain a mixture of individuals sampled from the two distributions. If the probability of individuals selected from genetic subpopulation A is p, and the probability of individuals selected from genetic subpopulation B is q = 1-p, then The distribution of response mean values can be explained.

For example, N is the total population size, and {x ₁ , x ₂ ,. ． ． , X _i ,. ． ． x _N } is a random set of variables, each describing an individual in the sample,
The expected value of the average response value is calculated by the following formula.

[0142]

[Equation 6] Where E [x with probability p_i] = Μ_AAnd E [x with probability q = 1-p _i ] = Μ_BIs. Therefore, the average of the distribution of sample mean values is

[0143]

[Equation 7] Is.

The variance of the response mean values in such a case depends on both the variance of each of the two distributions (A and B) and the difference between these distribution mean values. This variance can be represented by the sum of the individual variances.

[0145]

[Equation 8] If the random variable Y is Y = 0, then V [x _i | Y = 0] = σ ² _A , and if Y = 1, V [xi | Y = 1] = σ ² _B And the expected value of this variance is

[0146]

[Equation 9] Is. Further, if E [x _i | Y = 0] = μ _A and E [x _i | Y = 1] = μ _B , then

[0147]

[Equation 10] Is. By standard theory, V [x _i ] = E [V [x _i | Y]] + V [E [x _i |
Y]],

[0148]

[Equation 11] Is. That is, when ignoring a subpopulation, the variance is always greater than the variance of one of the subpopulations. This can be intuitively understood from the form of Equation 6. The variance is
It is a weighted sum of two population-specific variances (σ ² _A , σ ² _B ) and a term representing the difference between the ^two population mean values (μ _A −μ _B ) ² . Therefore, the variance consists of both within-population and inter-population variance.

The variance of the sample response means (if the ratio of individuals from subpopulations A and B is p: q) is normal (by the Central Limit Theorem), and the mean and variance

[0150]

[Equation 12] Have.

B. Marker Information Determining which of the two genetic subpopulations (A and B) an individual belongs to can be done by examining genetic markers. In general, the more markers genotyped, the greater the probability of assigning an individual to the correct group. In the absence of genetic information, the sample is a ratio p: q of individuals from subpopulations A and B.
Represents a random mixture of the two populations. If the two genetic backgrounds have the same frequency, then p = q = 0.5, and the sample mean is characterized as:

[0152]

[Equation 13] In some examples, all individuals from the subpopulation are genotyped for k equally informative markers. This is often the case when the markers are randomly selected (ie nothing is known about the genes involved in the response). The additional marker usually provides less information (ie, the k + 1 marker increases the probability of correctly assigning an individual to a subpopulation, but provides less information than the kth marker). This does not have to be the case, but often this is the case. For example, if one has prior knowledge of the genes involved in the response, this is typically tested first. Alternatively, if there is no information about the genes underlying the response, genetic matching is based on relationships (ie, the overall degree of genetic similarity in the genome), and thus the first few markers may With decreasing information from each,
Have more information.

If the k markers are genotyped, consider a simple model in which the probability of assigning an individual to the correct genetic subpopulation is determined by the formula:

[0154]

[Equation 14] Since k tends to be infinite, the probability of exact assignment is asymptotic to one. This probability can be used in the above equation to determine the mixture of sampled populations. The expected mean and variance of the sample mean values for k → ∞ and p → 1 in equations 2 and 6 are (μ _A , σ ² _A / N) for population A and (μ _{A for} population B). _B , σ ² _B / N) (to set p = 0 with this information and select non-responders).

C. Power of Clinical Trials by Gene Matching In a clinical trial consisting of a cohort of patients, half were treated and half were given placebo, where the two genetic subpopulations were of equal frequency. If there is
The response in the treated samples is normally distributed with mean and variance according to the formula:

[0156]

[Equation 15] In the control (placebo) sample, the mean distribution is again normal and has the mean and variance according to the formula:

[0157]

[Equation 16] For simplicity, the two samples are chosen to be the same size, but
It does not have to be so. For large N (> 30), the standard theorem of normal distribution was used to find the sample size needed to detect the difference in the response of the treated and placebo groups at the α% level, force β.

[0158]

[Equation 17] Is. However, N is the number in each division of the test, and the total number of individuals is 2N. From this equation, it is clear that the sample size increases as the difference between the mean values of the cases and controls decreases (ie the mean values are the same,

[0159]

[Equation 18] , Then the sample size is infinite). Also, the sample size increases as the subpopulation variance increases.

[0160]   (D. Examples of sample sizes for matched and non-matched populations)   In one example, the two genetic subpopulations are not treated

[0161]

[Formula 19] And if the mean response to treatment of individuals from group A is represented by μ _A = 5, σ _A = 8, they have the same response characteristics. The response to treatment of individuals from group B was μ _B = 0.
, Σ _B = 8, the same as the response for the placebo (ie, the non-responder). Furthermore, in this example, a significant level of 5% (α = 0.05) was used and the sample size represents the minimum number of individuals required for 80% force (β = 0.8). In Table 2 below, the number of markers (k), the probability of individuals selected from Group A (ie, correctly identified as responders) (p), the variance of the sample means of the treated population.

[0162]

[Equation 20] And the sample size (N) required for each section of the study is shown.

[0163]

[Table 2] In this example, the variance within the population is fixed at 64 for both genetic subpopulations, as well as in treated and untreated samples. Table 2 shows that the variance is 70 when the marker is not genotyped. This increased variance is entirely due to the difference in response due to the underlying genotype. If this is the cause (p → 1), the variance reverts to the expected value 64. This expansion of the dispersion has a significant effect on the required sample size (in Equation 11, the sample size does not increase linearly with increasing dispersion, but rather with the square of the fraction). If 83 individuals are needed in each department of the study when genetic information is not available, then using that polymorphic profile to correctly assign individuals as responders, only 20 individuals will be present. Needed.

Example 3 Effect of Linkage Disequilibrium and Marker Allele Frequency on the Power of Clinical Trials In this example, two genetic subpopulations A and B, and 1 present in both subpopulations. Suppose there are two polyallelic SNPs (single nucleotide polymorphism). One of these alleles labeled with g has a frequency p [g | A] = p _A within subpopulation _A.
And have a frequency p [g | B] = p _B in subpopulation B. In this particular example, the two subpopulations have equal frequencies (ie, p (A) = p (B) = 0.5). Then, in this condition, the overall population frequency of rare alleles (ie, paired alleles with lower frequencies in the general population) is p (g) =
(P _A + p _B ) / 2. In this example, the allele g was shown to be associated with an increased response to that treatment. The increased response may be due to the function of the allele itself, but more commonly is due to the allele being in linkage disparallel with any (unknown) genetic factor or mutation. Is. The size d of the linkage disequilibrium is d _A = p (A & g) -p (A) p (g) for subpopulation A,
Then, for subpopulation B, it is defined that d _B = p (B & g) −p (B) p (g).
The factors responsible for the increased response are more frequent in subpopulation A and S
If the NP marker is in the vicinity, it is expected that d _A > d _B. In this sense, this subpopulation is associated with a high response of unknown genes (A
) Alleles and hyporesponsive (B) alleles can be considered to have.

[0165]   In some instances, SNPs carry genes involved in response to treatment
Existing within a region known to
To which the subpopulation B showed a low response. Then this
In this example, the SNP can be used to determine which population the individual is from
Or, that is, they may be enrolled in clinical trials. For example, p _A = 0.6 and p_B= 0.2, this has the g allele
The individual that is derived from the sub-population A is derived from the individual from the sub-population B.
It means that it seems to be three times more than what you do. This marker is an individual
Respond well to the treatment (ie, they are from subpopulation A)
Or not). If the two subpopulations have the same frequency, then p (A & g)
= 0.3 and p (B & g) = 0.1, thus d_A= P (A & g) -p (A) p (g) = 0.3-0.5 * 0.4 = 0.1 (1
) And d_B= P (B & g) -p (B) p (g) = 0.1-0.5 × 0.4 = -0.1 (
2) Becomes

That is, there is a positive relationship between the g allele and the individuals belonging to subpopulation A.
Conversely, there is a negative relationship between the g allele and subpopulation B. Using these measures of linkage disequilibrium, when an individual has the g allele (from subpopulation A)
It is possible to calculate the conditional probability of an individual who has a high response. This calculation is as follows:

[0167]

[Equation 21] Note that by substituting for d _A and d _B , these conditional probabilities can also be expressed (using Bayes' theorem) as

[0168]

[Equation 22] That is, the strength of the relationship is summarized by the difference between the frequencies of that allele within the subpopulation.

Using information about this increase in SNP, the probability of accurately selecting individuals with high responses is between 50% (assuming the two subpopulations have the same frequency) to 75% null. . This probability is a function of the frequency of the associated marker allele, which affects the response between the two alleles and the strength of linkage disequilibrium. Then, as described in the previous section, with values of p = 0.5 indicating random sampling of individuals from two populations and p = 0.75 indicating sampling with information from genetic markers. First, the required sample size can be calculated.

Note that in the above example it is assumed that the allele frequencies in the two populations are known. This is often the case when the polymorphism is already known to be involved in the response but may not be generally known. However, it is also often possible that previous tests could be used to estimate the frequency of alleles in responding and non-responding individuals.

This method can of course be extended to multiple markers. In this case, G ∈ {g ₁ ,
． ． ． , G _k } have frequencies {p ₁ ,. ． ． , P _k } and subpopulation B
At frequency {q ₁ ,. ． ． , Q _k } shows the genotype of the individual at a set of multi-allelic markers. Then, Bayes' theorem gives the following equation.

[0172]

[Equation 23] Here, subpopulation A and subpopulation B have the same frequency. Similarly, p [B | G] can be expressed as follows.

[0173]

[Equation 24] These values can then be used as in the case of one locus to determine the probability of assignment to an exact subpopulation of individuals, and thus calculate the sample size.

For example, assume that five markers are genotyped, and for all of them, the rare allele has a frequency of 0.4 in subpopulation A and a frequency of 0.5 in subpopulation B. The study is conducted on patients from subpopulation A and individuals from subpopulation B.
It has been found that the latter individual does not respond to treatment. Patients are enrolled based on the results of 5 markers. Using Equations 7 and 8, n out of 5 markers
Table 3 shows the probabilities of individuals belonging to the subpopulation A when individuals have a rare allele in individuals.

[0175]

[Table 3] Table 3 shows that the probability of correctly assigning individuals to the correct subpopulation increases rapidly with the number of genotyped markers. In this example, the marker was randomly selected (not known to be associated with a particular gene) and 0.1 was selected as the difference in the frequency of rare alleles between the two subpopulations. The information of the type shown in Table 3 may be used directly to classify individuals into subpopulations or as criteria for enrollment in studies. For example, individuals who carry the rare allele at 1 or 0 polymorphisms out of 5 may be included in clinical trials. This increases the probability that the test will include individuals who respond to the treatment, where subpopulation A responds to treatment and subpopulation B does not respond to treatment. For individuals without the rare allele, the probability of individuals belonging to subpopulation A is 0.713, and for individuals with one rare allele, the probability of individuals belonging to subpopulation A is 0.620. is there. This selection criterion by itself greatly increases the proportion of individuals from subpopulation A that are enrolled in the clinical trial. This simple method of selection is only given as an example and depends on the number of polymorphisms and their respective allele frequencies, many other more complex procedures (eg cluster analysis).
Can be used.

Example 4 Effect of Number of Markers and Their Allele Frequency on the Power of Polymorphism Profiles to Distinguish Separate Patient Groups In this example, markers with common alleles (ie Two alleles have similar frequencies), and two types of markers with rare alleles. For the first marker type, one allele had a frequency of p ₁ = 0.5 in subpopulation A (responders) and q ₁ = 0.4 in subpopulation B (non-responders). Have a frequency. For the second marker type, one allele has a frequency of p ₂ = 0.1 in subpopulation A and a frequency of q ₂ = 0.08 in subpopulation B. In both cases, the rare allele has a 20% lower frequency in subpopulation B compared to its frequency in subpopulation A. The population of markers (total number K) is genotyped in a sample of 2000 patients known to belong to either subpopulation A or subpopulation B (from previous clinical studies). For each of these individuals, there are k observations x ₁ , x ₂ , x ₃ , with a value of 0 if a rare allele is present and a value of 1 if none is present.
． ． ． , X _k exist. These individuals are y if from subpopulation A.
= 0, and if it belongs to the sub-population B, it can be classified into the sub-population where y = 1. Using such training data, 2000 individuals were generated, and

[0177]

[Equation 25] A linear logistic model of the form (Christensen's "Log Li
near Models and Logistic Regression "
(Springer Verlag, New York, 1997)) is used to assign to one subpopulation or the other. Other statistical methods (as described fragmentarily in Example 3) can also be used. This linear logistic model was chosen to illustrate another classification method.

Table 4 shows the probabilities of assigning individuals to the correct subpopulations for 2, 5, 10, 20, and 50 markers. Values are shown for both types of markers, and mixtures of the two. In this example, all markers are assumed to be independent of each other. If not, other more effective statistical methods (eg, the classification tree method (see Breiman et al., Classify).
"Cation and Repression Trees" CRC Pres
S., 1984) can be applied.

[0179]

[Table 4] In these simulations, markers with similar frequencies for the two alleles are more effective in determining which individuals belong to which group than markers with significantly different allele frequencies. An equivalent mixture of markers of these two classes provides intermediate results, as expected.

In many cases, data from previous clinical trials are available, but there is no information as to which of the two original subpopulations the responders and non-responders were drawn from. Such scenarios may be more acceptable for analysis by clustering methods. Data for 2000 individuals were simulated using the same allele frequencies as above. These data were analyzed using K-means clustering (K = 2) to see how well a subpopulation could be defined with markers alone. Due to the little information available, this method has little efficacy if few markers are used (k <10). Results for 10 or more markers are shown in FIG.

[0181]

[Table 5] In this example, markers that are common to both alleles work more efficiently than markers where one allele is rare. These results using cluster analysis have a lower probability of correctly assigning individuals than in Table 4. This is expected due to the little information available in advance regarding these simulations.

As illustrated in part by the foregoing examples and the above description, the present invention has numerous applications, generally for treatment studies, and in particular for clinical trials. For example, the invention includes the use of a polymorphic profile to conduct a clinical trial on a population of patients with the same disease, wherein the polymorphic profile comprises
It contains at least one polymorphic site that is either unknown or associated with a disease. The present invention also provides a re-analysis of data from clinical experiments, where statistical significance is determined with respect to the original treated and control subpopulations selected for polymorphic profile similarity. Includes the use of polymorphic profiles to do. The invention also includes using a polymorphic profile to divide a population of individuals undergoing clinical trials into multiple subsets, wherein the elements of the subsets are more distinct than those of different subsets. Also show similarities in large polymorphic profiles.

The examples and embodiments described herein are for illustration purposes only, and various modifications or alterations in its respect are suggested to those skilled in the art, and the spirit and scope of the present application, It is understood that they fall within the scope of the appended claims. All publications, patents and patent applications cited herein are referred to when each individual publication, patent or patent application is specifically and individually indicated to be incorporated by reference. Similarly, for all purposes, all of which are expressly incorporated herein by reference.

[Brief description of drawings]

FIG. 1 shows a computer system for performing the method of the present invention.

FIG. 2 is a diagram of a computer system for performing the method of the present invention.

[Figure 3]   FIG. 3 is a flow chart of a method for evaluating a treatment procedure according to the present invention.

─────────────────────────────────────────────────── ─── Continued front page    (81) Designated countries EP (AT, BE, CH, CY, DE, DK, ES, FI, FR, GB, GR, IE, I T, LU, MC, NL, PT, SE), OA (BF, BJ , CF, CG, CI, CM, GA, GN, GW, ML, MR, NE, SN, TD, TG), AP (GH, GM, K E, LS, MW, SD, SL, SZ, TZ, UG, ZW ), EA (AM, AZ, BY, KG, KZ, MD, RU, TJ, TM), AE, AL, AM, AT, AU, AZ, BA, BB, BG, BR, BY, CA, CH, CN, C R, CU, CZ, DE, DK, DM, EE, ES, FI , GB, GD, GE, GH, GM, HR, HU, ID, IL, IN, IS, JP, KE, KG, KP, KR, K Z, LC, LK, LR, LS, LT, LU, LV, MA , MD, MG, MK, MN, MW, MX, NO, NZ, PL, PT, RO, RU, SD, SE, SG, SI, S K, SL, TJ, TM, TR, TT, TZ, UA, UG , US, UZ, VN, YU, ZA, ZW (72) Inventor Jones, Highwell Bee.             United States California 94301,               Palo Alto, Webster Sutori             No. 1, 530 F-term (reference) 4B024 AA01 AA11 AA20 CA02 HA11                 4B063 QA05 QA20 QQ42 QR32 QS40                       QX10

Claims (52)

[Claims] 1. A method of evaluating a treatment procedure, comprising: (a) a subpopulation of treated subjects and a subpopulation of control subjects from a population of treated subjects and control subjects. Wherein the treated population has been treated with a treatment procedure, the control population has been treated with a control procedure, and the subjects in both the treated and control population are Characterized for a type profile, subjects in both the treated and control subpopulations are selected for polymorphic profile similarity, and (b) as an evaluation of the treatment procedure, Determining whether there is a statistically significant difference in test parameters between the treated and control subpopulations , Method comprising. 2. The method of claim 1, further comprising the step of subjecting a second treated population and a control population to a further cycle of said selecting and determining steps. 3. The treated and control subpopulations are selected for similarity to a first polymorphism profile and the second treated and control subpopulations are second polymorphisms. The method of claim 1, wherein the method is selected for similarity to profiles. 4. The method of claim 1, wherein the treatment procedure comprises administering an agent to a member of the treated population. 5. The treatment procedure comprises administering an agent to a member of the treated population, and the control procedure does not include administering the agent to a member of the control population. The method according to 1. 6. The treatment procedure comprises administering an agent to a member of the treated population and the control procedure comprises administering a placebo to a member of the control population. The method described in. 7. The treatment procedure comprises the step of administering a first agent to a member of the treated population, wherein the control procedure comprises providing the member of the control population with the first agent. Administering a different second agent,
The method of claim 1. 8. Each of the first and second agents is a combination of agents,
The method according to claim 7. 9. The treatment procedure comprises administering a first amount of an agent to a member of the treated population, the control procedure comprising a second amount different from the first amount. 2. The method of claim 1, comprising administering the agent of claim 1 to a member of the control population. 10. A second schedule, wherein the treatment procedure comprises administering a drug to a member of the treated population according to a first schedule, and the control procedure is different from the first schedule. 2. The method of claim 1, comprising administering the agent to a member of the control population according to. 11. The method of claim 1, wherein the treatment procedure comprises behavioral therapy. 12. The method of claim 11, wherein the behavioral therapy comprises diet. 13. The method of claim 11, wherein the behavioral therapy comprises exercise therapy. 14. A group in which the test population contains a plurality of plants, and the treatment means includes a step of administering a pesticide to the plurality of plants, wherein the pesticide comprises a herbicide, an insecticide, and a growth promoter. The method of claim 1, selected from: 15. The subpopulation is human, animal, or plant.
The method described in. 16. The method of claim 15, wherein the subpopulation is human. 17. The method of claim 15, wherein the subpopulation is a plant. 18. The method of claim 1, wherein the subpopulation is a bacterium. 19. The method of claim 1, wherein the subject subpopulation is selected as being similarly exposed to environmental factors. 20. The method of claim 1, wherein the subpopulations of subjects are selected as being differentially exposed to at least one environmental factor. 21. The method of claim 1, wherein the subject subpopulations are selected from the same race. 22. The method of claim 1, wherein the subpopulations of subjects are selected for a common phenotypic trait. 23. The method of claim 1, wherein the subpopulations from the treated and control population each comprise at least 5 members. 24. The method of claim 23, wherein each of the subpopulations comprises at least 10 members. 25. The method of claim 24, wherein each of the subpopulations comprises at least 100 members. 26. The method of claim 1, wherein the polymorphic profile for each of said subpopulations is a single polymorphic form. 27. The method of claim 1, wherein the polymorphic profile for each of said subpopulations comprises multiple polymorphic forms. 28. The method of claim 27, wherein the polymorphic form is in a region encoding multiple genes. 29. The plurality of genes encode enzymes in a metabolic pathway,
The method of claim 28. 30. The method of claim 29, wherein said treatment means is a possible method of treating a disease and at least one subset of said plurality of genes correlates with said disease. 31. The method of claim 29, wherein the treatment means is a possible method of treating a disease and the metabolic pathway correlates with the disease. 32. The method of claim 1, wherein the polymorphic profile for each of the subpopulations comprises at least 10 polymorphic forms, respectively. 33. The method of claim 32, wherein the polymorphic profile for each of the subpopulations comprises at least 100 polymorphic forms. 34. The method of claim 1, wherein the polymorphic profile for the subpopulations is at least 10% identical. 35. The method of claim 34, wherein the polymorphism profile for the subpopulations is at least 50% identical. 36. The method of claim 35, wherein the polymorphism profile for the subpopulations is at least 75% identical. 37. The method of claim 36, wherein the polymorphism profiles for the subpopulations are the same. 38. The method of claim 1, wherein the test parameter is a measure of disease. 39. The method of claim 38, wherein the disease is cancer. 40. The method of claim 38, wherein the disease correlates with elevated serum cholesterol levels. 41. The test and control populations comprise plants and the test parameters are selected from the group consisting of susceptibility to herbicides, susceptibility to insecticides, susceptibility to disease, and susceptibility to frost damage. The method of claim 1. 42. A method of performing a clinical trial, comprising: (a) treating a treated population of patients with a disease with a drug.
Treating a control population of patients with the disease with a control procedure, and (b) selecting a subpopulation of patients from each of the treated and control populations having similar polymorphic profiles. (C) determining whether treatment with the drug correlates with the condition of the disease in the subpopulation and assessing the efficacy of the drug in treating the disease. 43. The method of claim 42, wherein the determining step comprises determining if there is a statistically significant difference in test parameters between the subpopulations. 44. The method of claim 42, further comprising the step of determining a polymorphism profile for each patient in the treated population and the control population prior to the selecting step. 45. The method of claim 42, wherein the control procedure comprises administering a placebo to members of the control population. 46. The method of claim 42, wherein the drug has efficacy in treating a disease other than the disease in which the clinical trial is being conducted. 47. A method of assessing a treatment procedure comprising: (a) providing a database, the database comprising: (i) for each member of the treated population treated according to the treatment procedure. A designation for each member of the control population treated according to the designation and control procedure; (ii) a polymorphic profile of each member of the treated population and control population; and (iii) the treated Specifying the test parameters for each member of the population and the control population, and (b) selecting a subpopulation for similarity in polymorphic profiles from each of the treated and control populations. And (c) statistics in the test parameters between the subpopulations. And determining whether a significant difference exists in, (d) comprises the step of displaying the output of results from the determination step, the method. 48. A computer program product for evaluating a treatment procedure, comprising: (a) a code providing or receiving data, (i) designation for each member of the treated population treated according to the treatment procedure. , And a designation for each member of the control population treated according to a control procedure, (ii) a designation for a polymorphic profile of each member of the treated population and the control population, and (iii) the treated population. And a designation for the test parameters for each member of the control population, and (b) a code for selecting a subpopulation from each of the treated and control populations having similar polymorphism profiles, (C) in the test parameters between the subpopulations, A code for determining whether there is a statistically significant difference; (d) a code for displaying the output of the result from the step (c); and (e) a computer-readable storage medium holding the code. And computer program products, including. 49. A system for evaluating a treatment procedure, comprising: (a) a memory; (b) a system bus; (c) a processor; (i) providing or receiving data; Designation for each member of the treated population treated according to the procedure, and designation for each member of the control population treated according to the control procedure, and polymorphism profile of each member of the treated population and the control population And (ii) a test parameter for each member of the treated and control populations, and (ii) from each of the treated and control populations with similar polymorphism profiles, Selecting a population, and (iii) assigning to the test parameters between the subpopulations A processor operably arranged to determine whether there is a statistically significant difference and (iv) display the resulting output from the step (iii), system. 50. A method of performing a clinical trial, comprising: (a) determining a polymorphic profile for an individual in a population of patients with the same disease, wherein the polymorphic profile is at least one polymorphic profile. Including a type morphology at a polymorphic site not known to be associated with the disease; (b) selecting a subpopulation of individuals with a similar polymorphic profile from the population; (c) the subpopulation. Administering a treatment therapy to a treated group within the population and a control therapy to a control group within the subpopulation, and (d) the treatment is expected to vary in response to an effective treatment therapy. Determining test parameters in the treated and control patients, and (e) whether the parameters show a statistically significant difference between the treated and control groups. A step of determining, encompassing method. 51. A method of performing a clinical trial, comprising: (a) determining a polymorphism profile for an individual in a population of patients with the same disease; and (b) determining a subset of individuals from the population. ,, so that the individuals in the sub-sets show greater similarity in polymorphism profiles than individuals in different sub-sets, and (c) members within each sub-set of the population. Assigning to the treated and control subpopulations, each of the treated and control subpopulations receiving at least one individual from each subset; and (d) to the treated subpopulation. Administering treatment therapy and administering control therapy to the control subpopulation, and (e) the treated subpopulation expected to vary in response to effective treatment therapy. And (f) determining whether the parameter shows a statistically significant difference between the treated and control groups, and in a patient in the control subpopulation. Including the method. 52. The method of claim 51, wherein the subset is a pair of individuals.