Patent Application
20090111139
Cancer is often treated by determining information about a tumor, e.g. a cancer, and using that information as a diagnostic tool in an attempt to determine how to treat the cancer. Current diagnostic tools typically analyze where and how the tumor arose. For example, the “where” might be a determination of whether the tumor arose from a specified kind of tissue. This determination is made based on the rationale that some cancers from some kinds of tissues are believed more aggressive than others. Therefore, it has been believed that the determination of how the tumor arose may be useful in determining how to treat the tumor. For example, a high-grade tumor may be more difficult to diagnose, because of the difficulty in determining from where it arose.
Current diagnostic techniques hence often attempt to deduce the original site of the tumor.
The present application describes new techniques for determining information about a tumor.
An embodiment determines pathways responsible for cellular anomalies. Tumors are then characterized to determine multiple pathways that are associated with characteristics of that tumor, where the characteristics can be genes or over/under expression of the genes.
The present invention investigates cellular pathways that are activated in a tumor cell and forms a signature indicative of multiple such pathways.
The kinds of pathways being discussed herein are a group of pathways acting in concert, one after another, to activate cell division or inhibit cell division. For purposes of this application, the term pathway is the action from a surface receptor on the cell to some agent, such as molecule, protein, or the like. The path may change the shape of the protein, for example, and then modify some other protein, or form an enzyme. This in turn changes the behavior of something else in the cell. Cell signaling can be used to characterize actions that cause things to happen in the cell. For example, the signaling can represent a determination of what is causing the cell to divide when it should not be dividing.
The inventors believe that there are a small number of pathways, for example between 5 and 10 different pathways, that are responsible for most of the cellular anomalies that eventually become tumors.
This recognitions is based, at least in part, on noticing that some cancer drugs rarely completely cure cancer. For example, the Herceptin drug has a known mechanism, and a known pathway that it inhibits. Herceptin often slows down cancer, increasing the time of survival by some amount. However, it rarely actually cures the cancer. The inventors believe the only time that Herceptin actually cures a cancer is in the unusual case where the tumor was caused by only a single specific pathway activation.
The inventors recognize that two tumors that have the same pathways activated are more likely to respond to the same treatment even if those tumors have different origins.
According to an embodiment, a tumor cell is characterized to determine a group of different pathways that are activated in specific cell. A combination of all the different pathways forms a signature, here called an “oncogenic signature”. The signature represents the set of multiple different pathways that are activated in the specific tumor being investigated.
There is a known relationship between certain drugs and the pathways they inhibit. A specific y drug inhibits x pathway. The oncogenic signature represents a group of pathways. That signature can be converted to providing a group of drugs, one or more drugs for each pathway, the group of drugs collectively inhibiting each of the individual pathways. As an example, Herceptin is known to attack Her2. This single chemical, however, inhibits only a single pathway.
The present application describes finding multiple pathways that form an oncogenic signature, and thereby also finding finds a combination of drugs that can be used to treat the patient.
An initial determination of signatures may be carried out according to the illustration of
The pathway(s) 120 can be deduced from those results, for example by using known information. For example, the literature includes many different studies that associate genes with the pathways that create those genes. Based on the results 110, the “hidden layers” 120 are postulated. The pathways will tend to cluster, based on this data.
There is likely to be a mixture of pathways between the upper layer 100 and the lower layer 110 forming the hidden layers between the known sample, and the measured products (genes, proteins, etc) It is also known in the literature to associate certain genes with certain pathways. For example, “oncogenic pathway signatures in human cancers as a guide to targeted therapies” nature 439 page 353, Jan. 19, 2006 illustrate known techniques of sorting genes according to their pathways. The system in
The multiple different pathways which are found for tumor cells form an a priori set of pathways that are used to later characterize a sample.
The number of tests on the known tumor samples may be at least ten times greater than a number of tests on the unknown tumor samples.
The pathways are each presumably pathways that were identified during the analysis at 210, that is, a priori paths. However, if there is a cluster that cannot be identified, then it may be deduced as being a new path, and analyzed according to the
Once the oncogenic signature is found, the therapies are found using a rule based lookup technique or other analogous technique. A rule based technique may define a set of rules, for example, of the form, if paths 1, 3 and 5 are on and paths 2 and 4 are off, then use drug cocktail ABC.
As explained above, if a known tumor shows no known pathways and/or no known drugs for inhibiting the pathways, this indicates that this must be a novel tumor which has no a priori data associated therewith. At this point, the patient's data is accessed using a microarray or other analysis device to find other genes and markers associated with the new path. In essence, a person whose tumor does not meet any of the known paths becomes a new clinical study.
Notice the significant difference between this technique and previous paradigms. The way things stand now, drugs are approved for a specific disease. With this technique, drugs would be approved for a specific pathway.
In an embodiment, the raw data from any known tumor or non-tumor sample will produce thousands of genes or gene products. As explained above, some genes will be indicative that pathway “A” has been followed. Other times combinations of genes, e.g., such as gene X in combination with gene Y will be indicative of pathway B. All of this can be based on studies or previously available literature. The raw data is used to form a data set of pathways, based on large amounts of data.
Further tests after the a priori knowledge is obtained then operates using a reduced subset of genes or gene products to find the active paths. For example, the reduced set may include hundreds of gene products, as compared with the initial determination which may analyze thousands of values. The paths are used to form the oncogenic signature for those paths that are active (410), and a prediction of a drug cocktail for the paths (420).
Models for mapping of input features such as genes and their expressions, protein, RNA, or other features to a decision of pathways may include such methods as artificial neural networks, fuzzy logic, support vector machines, hierarchical clustering, rule sets, finite state machines and hidden Markov models.
The techniques of optimization for models of features to pathways and/or signatures to drug selection can include population-based methods such as evolutionary computation, evolutionary algorithms, evolutionary programming, evolutionary strategies, genetic algorithms, genetic programming, enhanced colony optimization, particles swain optimization, differential evolution, associated evolutionary approaches that make use of variation and selection, as well as non-population-based approaches such as stimulated annealing and gradient descent based methods.
The rule sets may take any of a different number of different forms. For example, a simple set may be linearly separable, such as if there is input 1 equal to a value n; input 3 equal to a value y, then pathway x may be identified. The rule sets may be much more complex, such as if input 1*input 3/(square root of input 22)<3, then pathway x, else pathway why. The mapping may also use a linear mapping or a nonlinear mapping. For example, any other similar technique may alternatively be used, such as those disclosed in the above referenced article that show various ways in which gene expression patterns can be used to predict oncogenic pathways.
The general structure and techniques, and more specific embodiments which can be used to effect different ways of carrying out the more general goals are described herein.
Although only a few embodiments have been disclosed in detail above, other embodiments are possible and the inventors intend these to be encompassed within this specification. The specification describes specific examples to accomplish a more general goal that may be accomplished in another way. This disclosure is intended to be exemplary, and the claims are intended to cover any modification or alternative which might be predictable to a person having ordinary skill in the art. For example, other techniques of determining the a priori knowledge may be used.
Also, the inventors intend that only those claims which use the words “means for” are intended to be interpreted under 35 USC 112, sixth paragraph. Moreover, no limitations from the specification are intended to be read into any claims, unless those limitations are expressly included in the claims.
The operations and/or flowcharts described herein may be carried out on a computer, or manually. If carried out on a computer, the computer may be any kind of computer, either general purpose, or some specific purpose computer such as a workstation. The computer may be an Intel (e.g., Pentium or Core 2 duo) or AMD based computer, running Windows XP or Linux, or may be a Macintosh computer. The computer may also be a handheld computer, such as a PDA, cellphone, or laptop. Moreover, the method steps and operations described herein can be carried out on a dedicated machine that does these functions.
The programs may be written in C or Python, or Java, Brew or any other programming language. The programs may be resident on a storage medium, e.g., magnetic or optical, e.g. the computer hard drive, a removable disk or media such as a memory stick or SD media, wired or wireless network based or Bluetooth based Network Attached Storage (NAS), or other removable medium or other removable medium. The programs may also be run over a network, for example, with a server or other machine sending signals to the local machine, which allows the local machine to carry out the operations described herein.
Where a specific numerical value is mentioned herein, it should be considered that the value may be increased or decreased by 20%, while still staying within the teachings of the present application, unless some different range is specifically mentioned. Where a specified logical sense is used, the opposite logical sense is also intended to be encompassed.
