Genotypic tumor progression classifier and predictor
Actively dividing tumors appear to progress to a life threatening condition more rapidly than slowly dividing tumors. Assessing actively dividing tumors currently involves a manual assessment of the number of mitotic cells in a histological slide prepared from the tumor and assessed by a trained pathologist. Disclosed is a method for using cumulative information from a series of expressed genes to determine tumor prognosis. This cumulative information can be used to categorize tumor samples into high mitotic states or low mitotic states using a mathematical algorithm and gene expression data derived from microarrays or quantitative-Polymerase Chain Reaction (Q-PCR) data. The specific mathematical description outlines how the algorithm assesses the most informative subset of genes from the full list of genes during the assessment of each sample.
1. A method of predicting clinical tumor outcome in patients diagnosed with Stage I-III Lung Carcinoma comprising the steps of:
establishing a plurality of gene expression values in a tumor sample wherein the plurality of gene expression values are a plurality of genes identified in Table 1;
normalizing the plurality of gene expression values in the tumor sample to a reference expression;
defining at least one threshold value for the plurality of gene expressions;
establishing a vote of single-gene classifiers further comprising the steps of:
determining individual classifiers, further comprising:
comparing the gene expressions to the at least one threshold value;
selecting genes with expression levels above the at least one threshold value;
selecting genes with expression levels below the at least one threshold value;
assigning a positive value to the selected genes with expression levels above the at least one threshold value and assigning a negative value to the selected genes with expression levels below the at least one threshold value to form probeset data;
summing the probeset data to form a risk score; and
comparing the risk score to a sum of the al number of genes tested to form the majority vote classifier;
wherein the majority classifier is indicative of tumor outcome, such that the risk ratio above 0.15 is indicative of poor outcome and a risk ratio below 0.15 is indicative of good outcome;
administering treatment based on the outcome,
where patients with good prognosis are treated by resection and adjuvant chemotherapy, curative radiation therapy, or curative chemotherapy; and
where patients with poor prognosis are treated with palliative treatment.
2. The method of claim 1 , wherein the at least one threshold value consist of an upper threshold value and lower threshold value selected from the group consisting of the upper 33% of the gene expression values, the lower 33% of the gene expression values, the upper 15% of the gene expression values, the lower 15% of the gene expression values, the upper 50% of the gene expression values, and the lower 50% of gene expression values.
3. The method of claim 2 , further comprising the steps of:
assigning +1 to gene expression values above the upper threshold value;
assigning −1 to gene expression values below the lower threshold value; and
assigning 0 to gene expression values at or above the lower threshold value or at or below the upper threshold value.
4. The method of claim 2 , further comprising the steps of:
assigning +1 to gene expression values above the threshold value; and
assigning −1 to gene expression values at and below the threshold value.
5. The method of claim 1 , further comprising the step of identifying hazard ratios for the risk score, wherein a hazard ratio is calculated from the majority vote classifier and a score greater than 1 indicates poor clinical outcome.
6. The method of claim 1 , further comprising generating a probeset list, comprising:
providing a first probeset;
testing the first probeset against gene expression data for a tumor cell, wherein the gene expression data is compared to two threshold values;
utilizing probes identified as significantly related to gene expression n a second probeset test;
testing a second probeset against gene expression data for a tumor cell, wherein the gene expression data is compared to two threshold values, and wherein the second probeset includes probes identified as significantly related to gene expression in the first probeset test;
utilizing probes from the second probeset identified as significantly related to gene expression in a third probeset test; and
testing a third probeset against gene expression data for a tumor cell, wherein the gene expression data is compared to two threshold values, and wherein the third probeset includes probes identified as significantly related to gene expression in the second probeset test.
7. The method of claim 1 , further comprising:
combining at least one clinical parameter to the gene expression values, wherein the at least one clinical parameter is selected from the group consisting of tumor staging, history of prior cancers, lymph node involvement, smoking history, age, gender, operation type, last follow-up date, patient's status, CXR scan, CT scans, pulmonary function tests, time and site of recurrence, adjuvant treatment information, and combinations thereof.
8. The method of claim 1 , further comprising obtaining the gene expression values from protein, further comprising:
collecting RNA from the tumor sample
generating cDNA from collected protein by subjecting the tumor sample protein to SEQ ID 1; and
subjecting the cDNA to nucleic acid analysis.
9. The method of claim 1 , wherein the nucleic acid analysis is microarray, polymerase chain reaction, chromatin immunoprecipitation, gene array, quantitative-polymerase chain reaction, and reverse transciptase polymerase chain reaction.