Methods and compositions for determining risk of treatment toxicity

Abstract

Methods are provided for determining whether a patient treated with an anti-proliferative agent is susceptible to toxicity. In practicing the subject methods, an expression profile for the transcriptional response to a therapy is obtained from the patient and compared to a reference profile to determine whether the patient is susceptible to toxicity. In addition, reagents and kits thereof that find use in practicing the subject methods are provided.

Description

[0001] Many anti-proliferative agents used to treat cancer; infections, etc. also have the potential to damage normal cells. Generally dosage levels are selected to preferentially affect the target, e.g. tumor cells, but some patients are particularly susceptible to toxicity, and can suffer undesirable side effects from such treatment.

[0002] For example, ionizing radiation (IR) is used to treat about 60% of cancer patients, by depositing energy that injures or destroys cells in the area being treated. Radiation injury to cells is nonspecific, with complex effects on DNA. The efficacy of therapy depends on cellular injury to cancer cells being greater than to normal cells. Radiotherapy may be used to treat every type of cancer. Some types of radiation therapy involve photons, such as X-rays or gamma rays. Another technique for delivering radiation to cancer cells is internal radiotherapy, which places radioactive implants directly in a tumor or body cavity so that the radiation dose is concentrated in a small area.

[0003] Radiotherapy may be used in combination with additional agents. Radiosensitizers make the tumor cells more likely to be damaged, and radioprotectors protect normal tissues from the effects of radiation. Hyperthermia is also being studied for its effectiveness in sensitizing tissue to radiation.

[0004] Although most patients tolerate treatment, up to 10% of patients suffer from toxicity that can lead to significant morbidity. Non-genetic risk factors for radiation toxicity include concurrent treatment with radiosensitizing drugs and anatomical variations such as congenital malformations, post-surgical adhesions, fat content, and tissue oxygenation. Toxicity is also associated with diabetes and autoimmune diseases such as lupus. However, these causes cannot account for the vast majority of adverse radiation reactions.

[0005] In a small fraction of cases, radiation sensitivity can be attributed to known genetic mutations. Diseases of IR sensitivity include ataxia telangiectasia (AT), AT-like disorder, Nijmegan Breakage Syndrome, and radiosensitivity with severe combined immunodeficiency, but these autosomal recessive diseases are rare. Heterozygosity for mutations in ATM, the gene mutated in AT, may occur in 1% of individuals and has been reported to confer moderate sensitivity to IR in tissue culture. However, relatively few adverse radiation reactions are associated with ATM mutations.

[0006] Several attempts have been made to correlate radiation toxicity with cellular responses to IR ex vivo. Survival of cultured skin fibroblasts after IR correlated with acute radiation toxicity in some studies but not others (see Johansen et al. (1996) Radiother Oncol 40:101-9; Russell et al. (1998) Int J Radiat Biol 73:661-70; Peacock et al. (2000) Radiother Oncol 55:173-8. In another study, lymphocytes from cancer patients with radiation toxicity showed less IR-induced apoptosis than lymphocytes from control patients (Crompton et al. (1999) Int J Radiat Oncol Biol Phys 45:707-714). Peripheral blood lymphocytes from breast cancer patients with severe skin reactions showed an abnormal increase in chromosome aberrations when the cells were exposed to IR (Barber et al. (2000) Radiother Oncol 55:179-86). In these latter two studies, correlations between radiation toxicity and the ex vivo assay suggested the presence of an underlying genetic defect in some radiation sensitive patients. However, there was a large overlap between radiation sensitive patients and controls in these assays, limiting their clinical usefulness. Thus, assays to predict radiation toxicity have yielded mixed results, and the vast majority of adverse reactions remain unexplained (Brock et al. (2000) Radiother Oncol 55:93-94).

[0007] To date, there is no effective way known to the inventors to predict whether or not a patient will be susceptible to toxicity following radiation therapy. A diagnostic protocol which could provide information as to whether a patient is or is not susceptible to toxicity would be desirable for a number of reasons, including avoidance of delays in alternative treatments, elimination of exposure to adverse effects and reduction of unnecessary expense. As such, there is interest in the development of a protocol that can accurately predict whether or not a patient is susceptible to toxicity from radiation therapy.

[0008] Relevant Literature

[0009] A method of analyzing the significance of changes observed in expression patterns in microarrays may be found in International Application WO 01/84139; and Tusher et al. (2001) Proc. Natl. Acad. Sci. USA 98:5116-5121. A method for analysis of shrunken centroids is described by Tibshirani et al. (2002) Proc. Natl. Acad. Sci. USA 99:6567-6572.

SUMMARY OF THE INVENTION

[0010] Methods are provided for predicting whether an individual subjected to anti-proliferative therapy, particularly therapy that results in DNA damage, e.g. radiation therapy will be susceptible to toxicity resulting from the therapy. The ability to predict susceptibility to toxicity allows optimization of treatment, and determination of whether on whether to proceed with a specific therapy, and how to optimize dose, choice of treatment, and the like. In another embodiment, methods are provided for determining whether an individual is susceptible to toxicity.

[0011] In practicing the methods, an expression profile is obtained from the subject cells in the absence and presence of the therapy, e.g. UV radiation, ionizing radiation, presence of a chemotherapeutic agent, etc. The expression profile is used to determine the difference between the exposed and non-exposed cells, and is compared to a reference profile. Reagents and kits thereof that find use in practicing the subject methods are provided.

[0012] In another embodiment of the invention, methods are provided for statistical analysis of data, such as expression profiles in response to a stimulus, e.g. treatment with drug, exposure to radiation, exposure to specific antigenic stimulus, and the like; post-translational responses, basal expression levels; etc. to determine whether a pattern of expression or response will be predictive of a phenotype of interest. The statistical analyses usually utilize a heterogeneity-associated transformation, and nearest shrunken centroids analysis to provide a set of predictive genes.

BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee. FIGS. 1A-1F. Effect of heterogeneity-associated transformation (HAT) on gene expression data. The left panels show changes in gene expression after DNA damage, x(i), for gene i. The dashed line marks {overscore (x)}c(i), the average x(i) among the controls. The right panels show data after HAT, which was more effective in separating the radiation sensitive patients from controls. The upper panels show a hypothetical gene with transcriptional responses that were blunted in some patients and enhanced in others. The middle and lower panels show actual data for two predictive genes, cyclin B and 8-oxo-dGTPase. Patient samples were arranged by predicted probability for radiation toxicity (see FIG. 3).

[0014] FIGS. 2A-2B. Effect of heterogeneity-associated transformation (HAT) on predictive power. The nearest shrunken centroid (NSC) classifier was applied to 1491 IR-responsive genes and 2114 UV-responsive genes identified by SAM. In the NSC method, the threshold parameter determines the number of genes used for prediction (shown above the bar graphs). The upper and lower panels show the number of errors with and without HAT, respectively. White bars indicate the number of false negatives, and black bars indicate the number of false positives.

[0015] FIGS. 3A-3B. Predicting radiation toxicity from transcriptional responses to IR and UV. The plots show predictions for 15 subjects with no cancer (NoCa), 15 patients with skin cancer (SkCa), 13 control cancer patients without toxicity from radiation therapy (RadC), and 14 radiation sensitive cancer patients (RadS). HAT/NSC identified 24 predictive genes represented by 25 probe sets. The IR and UV responses were used to compute the probability of toxicity for each subject. The dotted lines indicate probability of 0.5, the prospectively defined cutoff for predicting radiation toxicity. The upper panel shows probabilities for radiation toxicity calculated from the full 48-sample training set. To avoid selection bias (see Ambroise and McLachlan (2002) P.N.A.S. 99:6562-6566), the 9 NoCa subjects were excluded from the training set because these subjects were used to identify the IR and UV-responsive genes. The lower panel shows probabilities calculated from 14-fold cross-validation as described in the text. The 9 NoCa subjects were excluded from the training sets, but included for cross-validation.

[0016] FIG. 4. Hierarchical clustering of genes that predict radiation toxicity. Data are shown for the 52 top-ranked predictive genes identified by HAT/NSC. The dendrogram above the heat map shows clustering of the 57 subjects. Shaded boxes under the dendrogram indicate the classes of subjects. The dendrogram to the left of the heat map shows clustering of the 52 genes represented by 55 probe sets. The colored boxes to the right of the heat map indicate biological function of the genes. An asterisk next to the gene description indicates UV-response data. All other data are IR-response data. Accession number, symbol, and rank in our prediction protocol are listed for each gene. Three predictive genes are listed twice, since two different probe sets (specified in parentheses) for the same gene were found to be predictive. In each case, probe sets for the same gene were closely clustered. Because centered Pearson correlation was used for clustering, genes with changes in expression that varied in the same way across samples were clustered together, independently of average changes in expression. For example, CALM1 and BASP1, two genes at the top of the heat map, were clustered together even though CALM1 was generally repressed and BASP1 was generally induced. To provide a scale for the IR-response data, the upper right panel shows the distribution of average IR responses for all 12,625 probe sets in samples from 15 subjects without cancer. The distribution of UV responses was similar.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The subject invention provides a method of determining whether a patient is susceptible to toxicity resulting from anti-proliferative therapy, where the method includes (a) obtaining a transcriptional response profile for a sample from said subject in the absence or presence of said therapy; and (b) comparing said obtained profile to a reference expression profile to determine whether said subject is susceptible to said toxicity. In certain embodiments, the expression profile is for at least about 10, usually at least about 25, and may be at least 50, at least about 100, or more of said genes listed in Table 3. In certain embodiments, the expression profile is determined using a microarray. In other embodiments the expression profile is determined by quantitative PCR or other quantitative methods for measuring mRNA.

[0018] The subject invention also provides a reference expression profile for a response phenotype that is one of: (a) susceptible to toxicity; or (b) non-susceptible to toxicity; wherein said expression profile is recorded on a computer readable medium.

[0019] For quantitative PCR analysis, the subject invention provides a collection of gene specific primers, said collection comprising: gene specific primers specific for at least about 10, usually at least about 20 of the genes of Table 3, where in certain embodiments said collection comprises at least 50 gene specific primers, at least 100, or more. The subject invention also provides an array of probe nucleic acids immobilized on a solid support, said array comprising: a plurality of probe nucleic acid compositions, wherein each probe nucleic acid composition is specific for a gene whose expression profile is indicative of toxicity susceptibility phenotype, wherein at least 10 of said probe nucleic acid compositions correspond to genes listed in Table 3, where in certain embodiments said array further comprises at least one control nucleic acid composition.

[0020] The subject invention also provides a kit for use in determining the susceptibility phenotype of a source of a nucleic acid sample, said kit comprising: at least one of: (a) an array as described above; or (b) a collection of gene specific primers as described above. The kit may further comprise a software package for data analysis of expression profiles.

[0021] Before the subject invention is described further, it is to be understood that the invention is not limited to the particular embodiments of the invention described below, as variations of the particular embodiments may be made and still fall within the scope of the appended claims. It is also to be understood that the terminology employed is for the purpose of describing particular embodiments, and is not intended to be limiting. Instead, the scope of the present invention will be established by the appended claims. In this specification and the appended claims, the singular forms “a,” “an” and “the” include plural reference unless the context clearly dictates otherwise.

[0022] Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limit of that range, and any other stated or intervening value in that stated range, is encompassed within the invention. The upper a nd lower limits of these smaller ranges may independently be included in the smaller ranges, and are also encompassed within the invention, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the invention.

[0023] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood to one of ordinary skill in the art to which this invention belongs. Although any methods, devices and materials similar or equivalent to those described herein can be used in the practice or testing of the invention, the preferred methods, devices and materials are now described.

[0024] All publications mentioned herein are incorporated herein by reference for the purpose of describing and disclosing the subject components of the invention that are described in the publications, which components might be used in connection with the presently described invention.

[0025] As summarized above, the subject invention is directed to methods of determining whether a subject is susceptible to unacceptable toxicity in response to therapeutic procedures, as well as reagents and kits for use in practicing the subject methods. The methods may also determine whether a particular cancer cell is susceptible to killing by a therapy of interest, where the differential between the target cell, e.g. a cancer cell, and the normal cell, is useful in making a determination of suitable treatment.

[0026] Methods are also provided for optimizing therapy, by determining the susceptibility of a patient to toxicity induced by one or more therapies, and based on that information, selecting the appropriate therapy, dose, treatment modality, e.g. angle and screening of radiation, etc. which optimizes the differential between delivery of an anti-proliferative treatment to the undesirable target cells, while minimizing undesirable toxicity. In one embodiment of the invention, the patient sample is exposed to two or more candidate therapies or combinations of therapies, e.g. exposure to various chemotherapeutic agents. Optionally, both a normal cell sample and a tumor cell sample are tested, in order to determine the differential effect of the treatment on normal and tumor cells. The treatment is optimized by selection for a treatment that avoids treatment that has a high probability of causing undesirable toxicity, while providing for effective anti-proliferative activity.

[0027] In further describing the invention, the subject methods are described first, followed by a review of the reagents and kits for use in practicing the subject methods.

Anti-Proliferative Agents and Treatments

[0028] Anti-proliferative therapy is used therapeutically to eliminate tumor cells and other undesirable cells in a host, and includes the use of therapies such as delivery of ionizing radiation, and administration of chemotherapeutic agents. Chemotherapeutic agents of particular interest induce DNA damage, and more particularly agents of interest induce double stranded breaks in DNA, for example the topoisomerase inhibitors anthracyclines, including the compounds daunorubicin, adriamycin (doxorubicin), epirubicin, idarubicin, anamycin, MEN 10755, and the like. Other topoisomerase inhibitors include the podophyllotoxin analogues etoposide and teniposide, and the anthracenediones, mitoxantrone and amsacrine.

[0029] In one aspect of the invention, the anti-proliferative agent interferes with microtubule assembly, e.g. the family of vinca alkaloids. Examples of vinca alkaloids include vinblastine, vincristine; vinorelbine (NAVELBINE); vindesine; vindoline; vincamine; etc.

[0030] In another embodiment of the invention, the anti-proliferative agent is a DNA-damaging agent, such as nucleotide analogs, alkylating agents, etc. Alkylating agents include nitrogen mustards, e.g. mechlorethamine, cyclophosphamide, melphalan (L-sarcolysin), etc.; and nitrosoureas, e.g. carmustine (BCNU), lomustine (CCNU), semustine (methyl-CCNU), streptozocin, chlorozotocin, etc.

[0031] Nucleotide analogs include pyrimidines, e.g. cytarabine (CYTOSAR-U), cytosine arabinoside, fluorouracil (5-FU), floxuridine (FUdR), etc.; purines, e.g. thioguanine (6-thioguanine), mercaptopurine (6-MP), pentostatin, fluorouracil (5-FU) etc.; and folic acid analogs, e.g. methotrexate, 10-propargyl-5,8-dideazafolate (PDDF, CB3717), 5,8-dideazatetrahydrofolic acid (DDATHF), leucovorin, etc.

[0032] Other chemotherapeutic agents of interest include metal complexes, e.g. cisplatin (cis-DDP), carboplatin, oxaliplatin, etc.; ureas, e.g. hydroxyurea; and hydrazines, e.g. N-methylhydrazine.

Toxicity

[0033] The use of anti-proliferative agents and treatments in therapy, e.g. in cancer therapy, depends on a differential between the effect on undesirable cancer cells and normal cells. Certain patients are less tolerant of treatment, and suffer unacceptable toxicity in normal tissues. It will be understood by those of skill in the art that some level of damage may occur in all subjects. It will also be understood that the toxic effects may be found on various tissues, i.e. skin, central nervous system, gut, etc. depending on the specific angle and dose of therapeutic radiation, compound that is delivered, etc. Criteria for grading toxic effects are known in the art, and are reproduced herein for convenience. The methods of the present invention are useful in differentiating between patients susceptible to unacceptable toxicity, i.e. having a grade of 2, 3, 4 or 5 in any tissue; and patients susceptible to acceptable toxicity of only grade 0 or 1.

[0034] The following tables provide conventional criteria for grading radiation toxicity. Other toxicities associated with other agents are known in the relevant clinical arts, and will be readily obtained by one of skill in the art. Toxicity may occur within less than about 90 days following exposure, herein termed early toxicity, or may occur after greater than about 90 days, herein termed late toxicity.

TABLE 1
Early Toxicity
	[0]	[1]	[2]	[3]	[4]
Skin	No change	Follicular, faint or	Tender or bright	Confluent, moist	Ulceration,
	over baseline	dull erythema/epilation/dry	erythema, patchy moist	desquamatiom other than	hemorrhage, necrosis
		desquamation/decreased	desquamation/	skin folds, pitting edema
		sweating	moderate edema
Mucous Membrane	No change	Injection/may experience	Patchy mucositis which	Confluent fibrinous	Ulceration, hemorrhage
	over baseline	mild pain not requiring	may produce an	mucositis/may include	or necrosis
		analgesic	inflammatory	severe pain requiring
			serosanguinitis	narcotic
			discharge/may
			experience moderate
			pain requiring analgesia
Eye	No change	Mild conjunctivitis with or	Moderate conjunctivitis	Severe keratitis with	Loss of vision
		without scleral injection/	with or without keratitis	corneal ulceration/	(unilateral or bilateral)
		increased tearing	requiring steroids &/or	objective decrease in
			antibiotics/dry eye	visual acuity or in visual
			requiring artificial tears/	fields/acute glaucoma/
			iritis with photophobia	panopthalmitis
Ear	No change	Mild external otitis with	Moderate external otitis	Severe external otitis with	Deafness
	over baseline	erythema, pruritis,	requiring topical	dischange or moist
		secondary to dry	medication/serious	desquamation/
		desquamation not	otitis medius/	symptomatic
		requiring medication.	hypoacusis on testing	hypoacusis/tinnitus, not
		Audiogram unchanged	only	drug related
		from baseline
Salivary Gland	No change	Mild mouth dryness/	Moderate to complete		Acute salivary gland
	over baseline	slightly thickened saliva/	dryness/thick, sticky		necrosis
		may have slightly altered	saliva/markedly altered
		taste such as metallic	taste
		taste
Pharynx &	No change	Mild dysphagia or	Moderate dysphagia or	Severe dysphagia or	Complete obstruction,
Esophagus	over baseline	odynophagia/may	odynophagia/may	odynophagia with	ulceration, perforation,
		require topical anesthetic	require narcotic	dehydration or weight	fistula
		or non-narcotic	analgesics/may require	loss (>15% from pretreatment
		analgesics/may require	puree or liquid diet	baseline)
		soft diet		requiring N-G feeding
				tube, I.V. fluids or
				hyperalimentation
Larynx	No change	Mild or intermittent	Persistent hoarseness	Whispered speech, throat	Marked dyspnea,
	over baseline	hoarseness/cough not	but able to vocalize/	pain or referred ear pain	stridor or hemoptysis
		requiring antitussive/	referred ear pain, sore	requiring narcotic/	with tracheostomy or
		erythema of mucosa	throat, patchy fibrinous	confluent fibrinous	intubation necessary
			exudate or mild	exudate, marked
			arytenoid edema not	arytenoid edema
			requiring narcotic/
			cough requiring
			antitussive
Upper G.I.	No change	Anorexia with <=5%	Anorexia with <=15%	Anorexia with >15%	Ileus, subacute or acute
		weight loss from	weight loss from	weight loss from	obstruction,
		pretreatment baseline/	pretreatment	pretreatment baseline or	performation, GI
		nausea not requiring	baseline/nausea &/or	requiring N-G tube or	bleeding requiring
		antiemetics/abdominal	vomiting requiring	parenteral support.	transfusion/abdominal
		discomfort not requiring	antiemetics/abdominal	Nausea &/or vomiting	pain requiring tube
		parasympatholytic drugs	pain requiring	requiring tube or	decompression or
		or analgesics	analgesics	parenteral	bowel diversion
				support/abdominal pain,
				severe despite
				medication/hematemesis
				or melena/abdominal
				distention (flat plate
				radiograph demonstrates
				distended bowel loops
Lower G.I.	No change	Increased frequency or	Diarrhea requiring	Diarrhea requiring	Acute or subacute
Including		change in quality of bowel	parasympatholytic drugs	parenteral support/	obstruction, fistula or
Pelvis		habits not requiring	(e.g., Lomotil)/mucous	severe mucous or blood	perforation; GI bleeding
		medication/rectal	discharge not	discharge necessitating	requiring transfusion;
		discomfort not requiring	necessitating sanitary	sanitary pags/abdominal	abdominal pain or
		analgesics	pads/rectal or	distention (flat plate	tenesmus requiring
			abdominal pain	radiograph demonstrates	tube decompression or
			requiring analgesics	distended bowel loops)	bowel diversion
Lung	No change	Mild symptoms of dry	Persistent cough	Severe cough	Severe respiratory
		cough or dyspnea on	requiring narcotic,	unresponsive to narcotic	insufficiency/
		exertion	antitussive agents/	antitussive agent or	continuous oxygen or
			dyspnea with minimal	dyspnea at rest/clinical	assisted ventilation
			effort but not at rest	or radiologic evidence of
				acute pneumonitis/
				intermittent oxygen or
				steroids may be required
Genitourinary	No change	Frequency of urination or	Frequency of urination	Frequency with urgency	Hematuria requiring
		nocturia twice	or nocturia which is less	and nocturia hourly or	transfusion/acute
		pretreatment habit/	frequent than every	more frequently/dysuria,	bladder obstruction not
		dysuria, urgency not	hour. Dysuria, urgency,	pelvis pain or bladder	secondary to clot
		requiring medication	bladder spasm requiring	spasm requiring regular,	passage, ulceration or
			local anesthetic (e.g.,	frequent narcotic/gross	necrosis
			Pyridium)	hematuria with/without
				clot passage
Heart	No change	Asymptomatic but	Symptomatic with EKG	Congestive heart failure,	Congestive heart
	over baseline	objective evidence of	changes and radiologic	angina pectoris,	failure, angina pectoris,
		EKG changes or	findings of congestive	pericardial disease	pericardial disease,
		pericardial abnormalities	heart failure or	responding to therapy	arrhythmias not
		without evidence of other	pericardial disease/no		responsive to non-
		heart disease	specific treatment		surgical measures
			required
Cns	No change	Fully functional status	Neurologic findings	Neurologic findings	Serious neurologic
		(i.e., able to work) with	present sufficient to	requiring hospitalization	impairment which
		minor neurologic findings,	require home case/	for initial management	includes paralysis,
		no medication needed	nursing assistance may		coma or seizures >3 per
			be required/		week despite
			medications including		medication/hospitalization
			steroids/anti-seizure		required
			agents may be required
Hematologic	>=4.0	3.0-<4.0	2.0-<3.0	1.0-<2.0	<1.0
Wbc (X 1000)
Platelets (X	>100	75-<100	50-<=75	25-<50	<25 or spontaneous
1000)					bleeding
Neutrophils	>=1.9	1.5-<1.9	1.0-<1.5	0.5-<=1.0	<=0.5 or sepsis
Hemoglobin	>11	11-9.5	<9.5-7.5	<7.5-5.0	—
(Gm %)
Hematocrit	>=32	28-<32	<=28	Packed cell transfusion	—
(%)				required

[0035]

TABLE 2
Late Toxicity
Organ Tissue	0	Grade 1	Grade 2	Grade 3	Grade 4
Skin	None	Slight atrophy	Patch atrophy;	Marked atrophy; Gross	Ulceration
		Pigmentation change	Moderate telangiectasia;	telangiectasia
		Some hair loss	Total hair loss
Subcutaneous	None	Slight induration (fibrosia)	Moderate fibrosis but	Severe induration and	Necrosis
Tissue		and loss of subcutaneous	asymptomatic Slight	loss of subcutaneous
		fat	field contracture <10%	tissue Field contracture
			linear reduction	>10% linear
				measurement
Mucous	None	Slight atrophy and	Moderate atrophy and	Marked atrophy with	Ulceration
Membrane		dryness	telangiectasia Little	complete dryness Severe
			mucous	telangiectasia
Salivary	None	Slight dryness of mouth	Moderate dryness of	Complete dryness of	Fibrosis
Glands		Good response on	mouth Poor response	mouth No response on
		stimulation	on stimulation	stimulation
Spinal Cord	None	Mild L'Hermitte's	Severe L'Hermitte's	Objective neurological	Mono, para
		syndrome	syndrome	findings at or below cord	quadraplegia
				level treated
Brain	None	Mild headache Slight	Moderate headache	Severe headaches	Seizures or paralysis
		lethargy	Great lethargy	Severe CNS dysfunction	Coma
				(partial loss of power or
				dyskinesia)
Eye	None	Asymptomatic cataract	Symptomatic cataract	Severe keratitis Severe	Panopthalmitis/
		Minor corneal ulceration	Moderate corneal	retinopathy or	Blindness
		or keratitis	ulceration Minor	detachment Severe
			retinopathy or glaucoma	glaucoma
Larynx	None	Hoarseness Slight	Moderate arytenoid	Severe edema Severe	Necrosis
		arytenoid edema	edema Chondritis	chondritis
Lung	None	Asymptomatic or mild	Moderate symptomatic	Severe symptomatic	Severe respiratory
		symptoms (dry cough)	fibrosis or pneumonitis	fibrosis or pneumonitis	insufficiency/
		Slight radiographic	(severe cough) Low	Dense radiographic	Continuous O2/
		appearances	grade fever Patchy	changes	Assisted ventilation
			radiographic
			appearances
Heart	None	Asymptomatic or mild	Moderate angina on	Severe angina Pericardial	Tamponade/Severe
		symptoms Transient T	effort Mild pericarditis	effusion Constrictive	heart failure/Severe
		wave inversion & ST	Normal heart size	pericarditis Moderate	constrictive pericarditis
		changes Sinus	Persistent abnormal T	heart failure Cardiac
		tachycardia >110 (at rest)	wave and ST changes	enlargement EKG
			Low ORS	abnormalities
Esophagus	None	Mild fibrosis Slight	Unable to take solid	Severe fibrosis Able to	Necrosis/Perforation
		difficulty in swallowing	food normally	swallow only liquids May	Fistula
		solids No pain on	Swallowing semi-solid	have pain on swallowing
		swallowing	food Dilatation may be	Dilation required
			indicated
Small/Large	None	Mild diarrhea Mild	Moderate diarrhea and	Obstruction or bleeding	Necrosis/
Intestine		cramping Bowel	colic Bowel movement	requiring surgery	PerforationFistula
		movement 5 times daily	>5 times daily
		Slight rectal discharge or	Excessive rectal mucus
		bleeding	or intermittent bleeding
Liver	None	Mild lassitude Nausea,	Moderate symptoms	Disabling hepatitic	Necrosis/Hepatic
		dyspepsia Slightly	Some abnormal liver	insufficiency Liver	coma or
		abnormal liver function	function tests Serum	function tests grossly	encephalopathy
			albumin normal	abnormal Low albumin
				Edema or ascites
Kidney	None	Transient albuminuria No	Persistent moderate	Severe albuminuria	Malignant
		hypertension Mild	albuminuria (2+)Mild	Severe hypertension	hypertension Uremic
		impairment of renal	hypertension No related	Persistent anemia	coma/Urea >100%
		function Urea 25-35 mg %	anemia Moderate	(<10 g %) Severe renal
		Creatinine 1.5-2.0 mg %	impairment of renal	failure Urea >60 mg %
		Creatinine clearance	function Urea > 36-60	Creatinine >4.0 mg %
		>75%	mg % Creatinine	Creatinine clearance
			clearance (50-74%)	<50%
Bladder	None	Slight epithelial atrophy	Moderate frequency	Severe frequency and	Necrosis/Contracted
		Minor telangiectasia	Generalized	dysuria Severe	bladder (capacity <100 cc)
		(microscopic hematuria)	telangiectasia	generalized	Severe
			Intermittent macroscopic	telangiectasia (often with	hemorrhagic cystitis
			hematuria	petechiae) Frequent
				hematuria Reduction in
				bladder capacity (<150 cc)
Bone	None	Asymptomatic No growth	Moderate pain or	Severe pain or	Necrosis/
		retardation Reduced	tenderness Growth	tenderness Complete	Spontaneous fracture
		bone density	retardation Irregular	arrest of bone growth
			bone sclerosis	Dense bone sclerosis
Joint	None	Mild joint stiffness Slight	Moderate stiffness	Severe joint stiffness Pain	Necrosis/Complete
		limitation of movement	Intermittent or moderate	with severe limitation of	fixation
			joint pain Moderate	movement
			limitation of movement
Any toxicity that causes death is graded 5.

Methods of Determining Susceptibility

[0036] The subject invention provides methods of predicting whether a patient or subject exposed to anti-proliferative therapy, particularly therapy resulting in double stranded DNA damage, e.g. ionizing radiation, including X-rays, gamma radiation, etc.; treatment with topoisomerase inhibitors as described above, and the like; will be susceptible to toxicity. In practicing the subject methods, a subject or patient sample, e.g., cells or collections thereof, e.g., tissues, is assayed to determine whether the host from which the assayed sample was obtained is susceptible to toxicity. Cells of interest particularly include dividing cells, e.g. leukocytes, fibroblasts, epithelial cells, etc. Cell samples are collected by any convenient method, as known in the art. Additionally, tumor cells may be collected and tested to determine the relative effectiveness of a therapy in causing differential death between normal and diseased cells.

[0037] To test for radiation-induced toxicity, the cell sample is exposed to radiation, including at least ionizing radiation, and preferably one cell sample is exposed to ionizing radiation and a second cell sample is exposed to ultraviolet radiation. A suitable dose of ionizing radiation may range from at least about 2 Gy to not more than about 10 Gy, usually about 5 Gy. The sample may be collected from at least about 2 and not more than about 24 hours following ionizing radiation, usually around about 4 hours. A suitable dose of ultraviolet radiation may range from at least about 5 J/m2 to not more than about 50 J/m2, usually about 10 J/m2. The sample may be collected from at least about 4 and not more than about 72 hours following ultraviolet radiation, usually around about 4 hours. The radiation exposed cell sample is assayed to obtain an expression profile for a set of genes, typically including at least about 10 top ranked genes set forth in Table 3, usually including at least about 25 top ranked genes, and may include at least about 50 top ranked genes; 100 top-ranked genes, or more, up to the complete set of predictive genes.

[0038] To test for toxicity resulting from exposure to chemotherapeutic agents, the cell sample may be exposed to radiation, as described above, or may be exposed to the therapeutic agent of interest, or to an agent having a similar profile of activity. Typically a cell sample will be compared to a control sample that has not been exposed to the therapy. The dose and time period for obtaining samples following exposure will vary with the specific agent that is selected. As is known in the art, a titration of dose may be used to determine the appropriate range for testing. Generally, samples from the cells will be obtained after at least about 4 hours and not more than about 5 days following exposure.

[0039] The term expression profile is used broadly to include a genomic expression profile, e.g., an expression profile of mRNAs, or a proteomic expression profile, e.g., an expression profile of one or more different proteins. Profiles may be generated by any convenient means for determining differential gene expression between two samples, e.g. quantitative hybridization of mRNA, labeled mRNA, amplified mRNA, cRNA, etc., quantitative PCR, ELISA for protein quantitation, and the like.

[0040] Genes/proteins of interest are genes/proteins that are found to be predictive of susceptibility to toxicity include, but are not limited to, the genes/proteins provided in Table 3, below

TABLE 3
				IR or UV
Rank	Accession	Symbol	Name	response
1	M25753	HUMCYCB	Cyclin B	UV
2	AI436567	ATP5D	ATP synthase, H+ transporting, mitochondrial F1	IR
			complex, delta subunit
3	X54942	CKS2	CDC28 protein kinase 2	UV
4	AB011126	FBP17	formin-binding protein 17	IR
5	U14971	RPS9	ribosomal protein S9	IR
6	AL022318	MDS019	phorbolin-like protein MDS019	IR
7	L08096	TNFSF7	tumor necrosis factor (ligand) superfamily, member 7	IR
8	AL080113		RNA helicase	IR
9	AI126004	SAS10	disrupter of silencing 10	IR
10	Z23090	HSPB1	heat shock 27 kD protein 1	IR
11	D21090	RAD23B	RAD23 homolog B	IR
12	U35451	CBX1	chromobox homolog 1 (HP1 beta)	IR
13	AA890010			IR
14	M65028	HNRPAB	heterogeneous nuclear ribonucleoprotein A/B	IR
15	D26600	PSMB4	proteasome (prosome, macropain) subunit, beta type, 4	IR
16	AF072810	BAZ1B	bromodomain adjacent to zinc finger domain, 1B	IR
17	U49869		ubiquitin	IR
18	D16581	NUDT1	nudix (nucleoside diphosphate linked moiety X)-type	IR
			motif 1
19	AA121509	LOC51690	U6 snRNA-associated Sm-like protein LSm7	IR
20	X81625	ETF1	eukaryotic translation termination factor 1	IR
21	Z48501	PABPC1	poly(A)-binding protein, cytoplasmic 1	IR
22	AA121509	LOC51690	U6 snRNA-associated Sm-like protein LSm7	IR
23	U12022	CALM1	calmodulin	UV
24	U52682	IRF4	interferon regulatory factor 4	IR
25	J03592	SLC25A6	solute carrier family 25 (mitochondrial carrier; adenine	IR
			nucleotide translocator), member 6
26	J03161	SRF	serum response factor (c-fos serum response	IR
			element-binding transcription factor)
27	Z11692	EEF2	eukaryotic translation elongation factor 2	IR
28	X83218	ATP5O	ATP synthase, H+ transporting, mitochondrial F1	IR
			complex, O subunit (oligomycin sensitivity conferring
			protein)
29	X51688	CCNA2	cyclin A2	UV
30	U11861	G10	maternal G10 transcript	IR
31	D44466	PSMD1	proteasome (prosome, macropain) 26S subunit, non-	IR
			ATPase, 1
32	AB019392	M9	muscle specific gene	IR
33	AI991040	DRAP1	DR1-associated protein 1 (negative cofactor 2 alpha)	IR
34	X70944	SFPQ	splicing factor proline/glutamine rich (polypyrimidine	UV
			tract-binding protein-associated)
35	M25753		Cyclin B1	UV
36	X15414	AKR1B1	aldo-keto reductase family 1, member B1 (aldose	IR
			reductase)
37	U12779	MAPKAPK2	mitogen-activated protein kinase-activated protein	IR
			kinase 2
38	Z49254	MRPL23	mitochondrial ribosomal protein L23	IR
39	J02683	SLC25A5	solute carrier family 25 (mitochondrial carrier; adenine	UV
			nucleotide translocator), member 5
40	S87759	PPM1A	protein phosphatase 1A (formerly 2C), magnesium-	IR
			dependent, alpha isoform
41	D32050	AARS	alanyl-tRNA synthetase	UV
42	X06617	RPS11	ribosomal protein S11	IR
43	AF023676	TM7SF2	transmembrane 7 superfamily member 2	IR
44	AB002368	KIAA0370	KIAA0370 protein	IR
45	AB029038	KIAA1115	KIAA1115 protein	IR
46	D45248	PSME2	proteasome (prosome, macropain) activator subunit 2	IR
			(PA28 beta)
47	D13641	KIAA0016	translocase of outer mitochondrial membrane 20	IR
			(yeast) homolog
48	M58378			IR
49	Y18418	RUVBL1	RuvB (E coli homolog)-like 1	UV
50	L20298	CBFB	core-binding factor, beta subunit	IR
51	L24804	P23	unactive progesterone receptor, 23 kD	UV
52	AF039656	BASP1	brain abundant, membrane attached signal protein 1	UV
53	AL022721	PPARD	peroxisome proliferative activated receptor, delta	IR
54	U48734	ACTN4	actinin, alpha 4	IR
55	Z49148	RPL29	ribosomal protein L29	IR
56	U68063	SFRS10	splicing factor, arginine/serine-rich (transformer	UV
			homolog) 10
57	AJ005259	EDF1	endothelial differentiation-related factor 1	IR
58	U05340	CDC20	CDC20 (cell division cycle 20 homolog)	UV
59	M72709	SFRS1	splicing factor, arginine/serine-rich 1 (splicing factor 2,	UV
			alternate splicing factor)
60	U15932	DUSP5	dual specificity phosphatase 5	UV
61	M61764	TUBG1	tubulin, gamma 1	UV
62	AI857469	TCEB2	transcription elongation factor B (SIII), polypeptide 2	IR
			(18 kD, elongin B)
63	AL022318	MDS019	phorbolin-like protein MDS019	UV
64	AB011114	KIAA0542	KIAA0542 gene product	IR
65	X71874			IR
66	L07956	GBE1	glucan (1,4-alpha-), branching enzyme 1 (glycogen	IR
			branching enzyme
67	AF053356			IR
68	L31584	EBI 1	G protein-coupled receptor	IR
69	X78992	ZFP36L2	zinc finger protein 36, C3H type-like 2	IR
70	M81757	RPS19	ribosomal protein S19	IR
71	AL031670			IR
72	W07033	GMFG	glia maturation factor, gamma	IR
73	Z98046			IR
74	U47101	NIFU	nitrogen fixation cluster-like	IR
75	L11566	RPL18	ribosomal protein L18	IR
76	U75686		polyadenylate binding protein	UV
77	M83664	HLA-DPB1	major histocompatibility complex, class II, DP beta 1	UV
78	AL050021			IR
79	M93425	PTPN12	protein tyrosine phosphatase, non-receptor type 12	IR
80	U94905	DGKZ	diacylglycerol kinase, zeta (104 kD)	UV
81	Y08614	XPO1	exportin 1 (CRM1, yeast, homolog)	IR
82	AI540957	QP-C	low molecular mass ubiquinone-binding protein	IR
			(9.5 kD)
83	Z26876	RPL38	ribosomal protein L38	IR
84	U28386	KPNA2	karyopherin alpha 2 (RAG cohort 1, importin alpha 1)	IR
85	X65550	MKI67	antigen identified by monoclonal antibody Ki-67	UV
86	S72008	CDC10	CDC10 (cell division cycle 10 homolog)	IR
87	U03398	TNFSF9	tumor necrosis factor (ligand) superfamily, member 9	IR
88	AF049910	TACC1	transforming, acidic coiled-coil containing protein 1	IR
89	D42043	KIAA0084	KIAA0084 protein	IR
90	AB002313	PLXNB2	plexin B2	UV
91	X97074	AP2S1	adaptor-related protein complex 2, sigma 1 subunit	IR
92	AB002323	DNCH1	dynein, cytoplasmic, heavy polypeptide 1	UV
93	AF047185	NDUFA2	NADH dehydrogenase (ubiquinone) 1 alpha	IR
			subcomplex, 2 (8 kD, B8)
94	AI819948	MEL	mel transforming oncogene (derived from cell line	UV
			NK14)-RAB8 homolog
95	U14970	RPS5	ribosomal protein S5	IR
96	AI375913	TOP2A	topoisomerase (DNA) II alpha (170 kD)	IR
97	AI541050	NDUFB8	NADH dehydrogenase (ubiquinone) 1 beta	IR
			subcomplex, 8 (19 kD, ASHI)
98	D86979	KIAA0226	KIAA0226 gene product	IR
99	Z36714	CCNF	cyclin F	IR
100	M30938	XRCC5	X-ray repair complementing defective repair (double-	UV
			strand-break rejoining; Ku autoantigen)
101	J03191	PFN1	profilin 1	UV
102	X65923	FAU	ribosomal protein S30	IR
103	AF035555	HADH2	hydroxyacyl-Coenzyme A dehydrogenase, type II	IR
104	X72889	SMARCA2	SWI/SNF related, matrix associated, actin dependent	IR
			regulator of chromatin, subfamily a, member 2
105	L22473	BAX	BCL2-associated X protein	UV
106	U09813	ATP5G3	ATP synthase, H+ transporting, mitochondrial F0	IR
			complex, subunit c (subunit 9) isoform 3
107	Y00371	hsc70	71 kd heat shock cognate protein	IR
108	U94855	EIF3S5	eukaryotic translation initiation factor 3, subunit 5	IR
			(epsilon, 47 kD)
109	AA808961	PSMB9	proteasome (prosome, macropain) subunit, beta type,	IR
			9 (large multifunctional protease 2)
110	AF053356			UV
111	AF005392			UV
112	L01124	RPS13	ribosomal protein S13	IR
113	X00457	HLA-DPA1	major histocompatibility complex, class II, DP alpha 1	UV
114	AI800499	AIM1	absent in melanoma 1	IR
115	Y08110	SORL1	sortilin-related receptor, L(DLR class) A repeats-	UV
			containing
116	U12472	GSTP1	glutathione 5-transferase pi	IR
117	X78992	ZFP36L2	zinc finger protein 36, C3H type-like 2	UV
118	X91257	SARS	seryl-tRNA synthetase	IR
119	M81757	RPS19	ribosomal protein S19	UV
120	AF037448	NSAP1	NS1-associated protein 1	IR
121	AL022394			UV
122	U67156	MAP3K5	mitogen-activated protein kinase kinase kinase 5	IR
123	AF087135	ATP5H	ATP synthase, H+ transporting, mitochondrial F0	IR
			complex, subunit d
124	N24355	POLR2L	polymerase (RNA) II (DNA directed) polypeptide L	IR
			(7.6 kD)
125	D78134	CIRBP	cold inducible RNA-binding protein	IR
126	X81625	ETF1	eukaryotic translation termination factor 1	UV
127	X13710	GPX1	glutathione peroxidase 1	IR
128	U18321	DAP3	death associated protein 3	IR
129	AF072810	BAZ1B	bromodomain adjacent to zinc finger domain, 1B	UV
130	X82240	TCL1A	T-cell leukemia/lymphoma 1A	IR
131	D26598	PSMB3	proteasome (prosome, macropain) subunit, beta type, 3	IR
132	X97548	TRIM28	tripartite motif-containing 28	UV
133	D49738	CKAP1	cytoskeleton-associated protein 1	IR
134	D87078	PUM2	pumilio homolog 2	IR
135	U49278	UBE2V1	ubiquitin-conjugating enzyme E2 variant 1	UV
136	U18300	DDB2	damage-specific DNA binding protein 2 (48 kD)	IR
137	X70394	ZNF146	zinc finger protein 146	IR
138	AF041259	ZNF217	zinc finger protein 217	IR
139	M94314	RPL24	ribosomal protein L24	IR
140	U09510	GARS	glycyl-tRNA synthetase	UV
141	AF042384	BC-2	putative breast adenocarcinoma marker (32 kD)	IR
142	HG1800-HT1823			IR
143	U96915	SAP18	sin3-associated polypeptide, 18 kD	IR
144	M13934		ribosomal protein S14	IR
145	Z11697	CD83	CD83 antigen (activated B lymphocytes,	IR
			immunoglobulin superfamily)
146	U19599	BAX	BCL2-associated X protein	IR
147	AA527880			IR
148	U48734	ACTN4	actinin, alpha 4	UV
149	U14972	RPS10	ribosomal protein S10	IR
150	D00760	PSMA2	proteasome (prosome, macropain) subunit, alpha	IR
			type, 2
151	M86667	NAP1L1	nucleosome assembly protein 1-like 1	UV
152	AF057557	TOSO	regulator of Fas-induced apoptosis	IR
153	U59309	FH	fumarate hydratase	UV
154	AL049701	KIAA0471	KIAA0471 gene product	UV
155	AB029014	KIAA1091	KIAA1091 protein	UV
156	D23661	RPL37	ribosomal protein L37	IR
157	U03106	CDKN1A	cyclin-dependent kinase inhibitor 1A (p21, Cip1)	UV
158	AC004770			UV
159	AF037643	RPL12	ribosomal protein L12	IR
160	U07424	FARSL	phenylalanine-tRNA synthetase-like	UV
161	AA806768		Homo sapiens phorbolin I protein (PBI) mRNA,	UV
			complete cds
162	L49380	ZNF162	zinc finger protein 162	UV
163	AL050366	OGT	O-linked N-acetylglucosamine (GlcNAc) transferase	IR
			(UDP-N-acetylglucosamine: polypeptide-N-
			acetylglucosaminyl transferase)
164	L12723	HSPA4	heat shock 70 kD protein 4	IR
165	M13932	RPS17	ribosomal protein S17	IR
166	U51004	HINT	histidine triad nucleotide-binding protein	IR
167	M64716	RPS25	ribosomal protein S25	IR
168	Z11697	CD83	CD83 antigen (activated B lymphocytes,	UV
			immunoglobulin superfamily)
169	N98670			IR
170	U14966	RPL5	ribosomal protein L5	IR
171	D13643	DHCR24	24-dehydrocholesterol reductase	UV
172	D21262	NOLC1	nucleolar and coiled-body phosphprotein 1	IR
173	AC005943			UV
174	AF044671	GABARAP	GABA(A) receptor-associated protein	IR
175	U54559	EIF3S3	eukaryotic translation initiation factor 3, subunit 3	IR
			(gamma, 40 kD)
176	J04130	SCYA4	small inducible cytokine A4 (homologous to mouse	IR
			Mip-1b)
177	U19599	BAX	BCL2-associated X protein	UV
178	X57206	ITPKB	inositol 1,4,5-trisphosphate 3-kinase B	UV
179	D87446	KIAA0257	KIAA0257 protein	UV
180	T58471	UQCR	ubiquinol-cytochrome c reductase (6.4 kD) subunit	IR
181	U02570	ARHGAP1	Rho GTPase activating protein 1	UV
182	X51688	CCNA2	cyclin A2	UV
183	D31885	ARL6IP	ADP-ribosylation factor-like 6 interacting protein	UV
184	AI541336	NDUFS5	NADH dehydrogenase (ubiquinone) Fe-S protein 5	IR
			(15 kD) (NADH-coenzyme Q reductase)
185	V00567	B2M	beta-2-microglobulin	IR
186	M86737	SSRP1	structure specific recognition protein 1	UV
187	D80005	C9orf10	C9orf10 protein	UV
188	AF017789	TAF2S	TATA box binding protein (TBP)-associated factor,	IR
			RNA polymerase II, S, 150 kD
189	AB014458	USP1	ubiquitin specific protease 1	UV
190	X63469	GTF2E2	general transcription factor IIE, polypeptide 2 (beta	IR
			subunit, 34 kD)
191	M55914	ENO1	enolase 1, (alpha)	IR
192	Y00451	ALAS1	aminolevulinate, delta-, synthase 1	UV
193	AF046001	ZNF207	zinc finger protein 207	UV
			dolichyl-diphosphooligosaccharide-protein
194	D29643	DDOST	glycosyltransferase	IR
195	U29344	FASN	fatty acid synthase	UV
196	L13848	DDX9	DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 9	UV
			(RNA helicase A, nuclear DNA helicase II;
			leukophysin)
197	J00314	TUBB	tubulin, beta polypeptide	IR
198	X71874			UV
199	D90070	PMAIP1	phorbol-12-myristate-13-acetate-induced protein 1	IR
200	X64330	ACLY	ATP citrate lyase	UV
201	M94362	LMNB2	lamin B2	IR
202	M23114	ATP2A2	ATPase, Ca++ transporting, cardiac muscle, slow	UV
			twitch 2
203	J03040	SPARC	secreted protein, acidic, cysteine-rich (osteonectin)	IR
204	X64229	DEK	DEK oncogene (DNA binding)	IR
205	J03826	FDXR	ferredoxin reductase	UV
206	U51698	DED	apoptosis antagonizing transcription factor	UV
207	Z37166	BAT1	HLA-B associated transcript 1	IR
208	X62744	HLA-DMA	major histocompatibility complex, class II, DM alpha	IR
209	U28686	RBM3	RNA binding motif protein 3	UV
210	D00860	PRPS1	phosphoribosyl pyrophosphate synthetase 1	UV
211	L76200	GUK1	guanylate kinase 1	IR
212	AB011118	KIAA0546	KIAA0546 protein	IR
213	L08895	MEF2C	MADS box transcription enhancer factor 2,	IR
			polypeptide C (myocyte enhancer factor 2C)
214	D38551	RAD21	RAD21 homolog	IR
215	M32578	HLA-DRB1	major histocompatibility complex, class II, DR beta 1	UV
216	X66079	SPIB	Spi-B transcription factor (Spi-1/PU.1 related)	IR
217	U03398	TNFSF9	tumor necrosis factor (ligand) superfamily, member 9	UV
218	Y13936	PPM1G	protein phosphatase 1G (formerly 2C), magnesium-	IR
			dependent, gamma isoform
219	X15940	RPL31	ribosomal protein L31	IR
220	J04031	MTHFD1	methylenetetrahydrofolate dehydrogenase (NADP+	UV
			dependent), methenyltetrahydrofolate cyclohydrolase,
			formyltetrahydrofolate synthetase
221	AI032612	SNRPF	small nuclear ribonucleoprotein polypeptide F	IR
222	AJ245416	LSM2	U6 snRNA-associated Sm-like protein	IR
223	L25931	LBR	lamin B receptor	UV
224	J05614			IR
225	AL050265	TARDBP	TAR DNA binding protein	UV
226	X04366	CAPN1	calpain 1, (mu/l) large subunit	UV
227	AL050161			IR
228	D42084	METAP1	methionyl aminopeptidase 1	IR
229	U90878	PDLIM1	PDZ and LIM domain 1 (elfin)	IR
230	AL080109	KIAA0618	KIAA0618 gene product	IR
231	U94319	PSIP2	PC4 and SFRS1 interacting protein 2	IR
232	L15189	HSPA9B	heat shock 70 kD protein 9B (mortalin-2)	UV
233	X80199	MLN51	MLN51 protein	IR
234	AL050060	DKFZP566H073	DKFZP566H073 protein	UV
235	X59543	RRM1	ribonucleotide reductase M1 polypeptide	UV
236	AB019987	SMC4L1	SMC4 (structural maintenance of chromosomes 4)-	UV
			like 1
237	J04977	XRCC5	X-ray repair complementing defective repair (double-	UV
			strand-break rejoining; Ku autoantigen, 80 kD)
238	Y07969	SSP29	acidic protein rich in leucines	UV
239	U37690	POLR2L	polymerase (RNA) II (DNA directed) polypeptide L	IR
			(7.6 kD)
240	AB018328	ALTE	Ac-like transposable element	IR
241	AI540925	COX6A1	cytochrome c oxidase subunit VIa polypeptide 1	IR
242	HG1515-HT1515	Btf3b	Transcription Factor Btf3b	IR
243	U87947	EMP3	epithelial membrane protein 3	UV
244	AB028990	KIAA1067	KIAA1067 protein	IR
245	X55954	RPL23	ribosomal protein L23	IR
246	X02994	ADA	adenosine deaminase	UV
247	AB029038	KIAA1115	KIAA1115 protein	UV
248	L29254			IR
249	U05040		Homo sapiens far upstream element (FUSE) binding	UV
			protein 1 (FUBP1), mRNA
250	AF007140	ILF3	interleukin enhancer binding factor 3, 90 kD	UV
251	X59303	VARS2	valyl-tRNA synthetase 2	UV
252	AI345944	NDUFB1	NADH dehydrogenase (ubiquinone) 1 beta	IR
			subcomplex, 1 (7 kD, MNLL)
253	U21689	GSTP1	glutathione S-transferase pi	IR
254	Z24459			IR
255	U45878	BIRC3	baculoviral IAP repeat-containing 3	UV
256	AF081280	NPM3	nucleophosmin/nucleoplasmin 3	UV
257	Z25535	NUP153	nucleoporin 153 kD	IR
258	D26579	ADAM8	a disintegrin and metalloproteinase domain 8	IR
259	AF063308	DEEPEST	mitotic spindle coiled-coil related protein	UV
260	S57212	MEF2C	MADS box transcription enhancer factor 2,	IR
			polypeptide C (myocyte enhancer factor 2C)
261	Y00971	PRPS2	phosphoribosyl pyrophosphate synthetase 2	UV
262	AF067656	ZWINT	ZW10 interactor	UV
263	M91196	ICSBP1	interferon consensus sequence binding protein 1	IR
264	AI033692	BCRP1	Breakpoint cluster region protein, uterine leiomyoma,	UV
			1; barrier to autointegration factor
265	AL022326	SYNGR1	synaptogyrin 1	IR
266	AF032885	FOXO1A	forkhead box O1A (rhabdomyosarcoma)	UV
267	U03911	MSH2	mutS homolog 2 (colon cancer, nonpolyposis type 1)	UV
268	AL021154			IR
269	AB011116	KIAA0544	KIAA0544 protein	IR
270	X17644	GSPT1	G1 to S phase transition 1	UV
271	AI565760	GABARAPL2	GABA(A) receptor-associated protein-like 2	IR
272	D87735	RPL14	ribosomal protein L14	IR
273	U52112	IRAK1	interleukin-1 receptor-associated kinase 1	UV
274	X04803		ubiquitin	IR
275	AI525834	NPC2	Niemann-Pick disease, type C2 gene	IR
276	M14333	FYN	FYN oncogene related to SRC, FGR, YES	UV
277	Z97054	UREB1	upstream regulatory element binding protein 1	UV
278	AB014609	KIAA0709	endocytic receptor (macrophage mannose receptor	UV
			family)
279	AI653621	TXN	thioredoxin	UV
280	U24266	ALDH4A1	aldehyde dehydrogenase 4 family, member A1	UV
281	M37583	H2AFZ	H2A histone family, member Z	UV
282	J03805	PPP2CB	protein phosphatase 2 (formerly 2A), catalytic subunit,	UV
			beta isoform
283	U51127	IRF5	interferon regulatory factor 5	UV
284	M22806	P4HB	prolyl 4-hydroxylase beta-subunit and disulfide	UV
			isomerase
285	D11086	IL2RG	interleukin 2 receptor, gamma (severe combined	UV
			immunodeficiency)
286	AF000982	DDX3	DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide 3	UV
287	U86602	EBNA1BP2	EBNA1-binding protein 2	UV
288	AF000231	RAB11A	RAB11A, member RAS oncogene family	UV
289	L23959	TFDP1	transcription factor Dp-1	UV
290	AB020713	KIAA0906	KIAA0906 protein	UV
291	X59871	TCF7	transcription factor 7 (T-cell specific, HMG-box)	UV
292	AA310786		Homo sapiens cDNA: FLJ23602 fis, clone LNG15735	IR
293	U15085	HLA-DMB	major histocompatibility complex, class II, DM beta	IR
294	D80001	KIAA0179	KIAA0179 protein	IR
295	HG4074-HT4344	Rad2	Rad2	UV
296	AA648295	CBX3	chromobox homolog 3 (HP1 gamma)	UV
297	Y13936	PPM1G	protein phosphatase IG (formerly 2C), magnesium-	UV
			dependent, gamma isoform
298	D49489	P5	protein disulfide isomerase-related protein	UV
299	AJ012590	H6PD	hexose-6-phosphate dehydrogenase (glucose 1-	IR
			dehydrogenase)
300	D16431	HDGF	hepatoma-derived growth factor (high-mobility group	IR
			protein 1-like)
301	AA527880			IR
302	AI525665	COX8	cytochrome c oxidase subunit VIII	IR
303	U19765	ZNF9	zinc finger protein 9 (a cellular retroviral nucleic acid	UV
			binding protein)
304	M74491	ARF3	ADP-ribosylation factor 3	UV
305	AF039397			UV
306	X67951	PRDX1	peroxiredoxin 1	IR
307	AB005047	SH3BP5	SH3-domain binding protein 5 (BTK-associated)	IR
308	S75463	TUFM	Tu translation elongation factor, mitochondrial	UV
309	M63904	GNA15	guanine nucleotide binding protein (G protein), alpha	UV
			15 (Gq class)
310	D42084	METAP1	methionyl aminopeptidase 1	UV
311	W28979	FLJ20452	hypothetical protein FLJ20452	IR
312	M59465	TNFAIP3	tumor necrosis factor, alpha-induced protein 3	IR
313	M26004	CR2	complement component, receptor 2	IR
314	X04106	CAPNS1	calpain, small subunit 1	IR
315	Z14000	RING1	ring finger protein 1	UV
316	AF044671	GABARAP	GABA(A) receptor-associated protein	UV
317	D13627	CCT8	chaperonin containing TCP1, subunit 8 (theta)	UV
318	D21853	KIAA0111	KIAA0111 gene product	UV
319	HG662-HT662		Small Rna-Associated Protein	IR
320	AI087268	SNRPC	small nuclear ribonucleoprotein polypeptide C	IR
321	D80000	SMC1L1	SMC1 (structural maintenance of chromosomes 1)-	UV
			like 1
322	L31584	EBI 1	G protein-coupled receptor	UV
323	M33336	PRKAR1A	protein kinase, cAMP-dependent, regulatory, type I,
			alpha (tissue specific extinguisher 1)	UV
324	D14812	KIAA0026	MORF-related gene X	UV
325	D11139	TIMP1	tissue inhibitor of metalloproteinase 1 (erythroid	UV
			potentiating activity, collagenase inhibitor)
326	M65028	HNRPAB	heterogeneous nuclear ribonucleoprotein A/B	UV
327	AB023154	KIAA0937	KIAA0937 protein	UV
328	AA149486	COX17	COX17 homolog, cytochrome c oxidase assembly	IR
			protein
329	Y00371	hsc70	71 kd heat shock cognate protein	UV
330	X95808	ZNF261	zinc finger protein 261	IR
331	M64595	RAC2	ras-related C3 botulinum toxin substrate 2 (rho family,	IR
			small GTP binding protein Rac2)
332	D50405	HDAC1	histone deacetylase 1	UV
333	X95384	UK114	translational inhibitor protein p14.5	UV
334	M93311	MT3	metallothionein 3 (growth inhibitory factor	IR
			(neurotrophic))
335	M13792	ADA	adenosine deaminase	UV
336	D90070	PMAIP1	phorbol-12-myristate-13-acetate-induced protein 1	UV
337	AF047436	ATP5J2	ATP synthase, H+ transporting, mitochondrial F0	UV
			complex, subunit f, isoform 2
338	U24152	PAK1	p21/Cdc42/Rac1-activated kinase 1 (yeast Ste20-	UV
			related)
339	U46692		cystatin B	IR

[0041] In certain embodiments, any one or more of the genes/proteins in the prepared expression profile are from Table 3, above, where the expression profile may include expression data for 5, 10, 20, 25, 50, 100 or more of, including all of, the genes/proteins listed in Table 3, above.

[0042] In certain embodiments, the expression profile obtained is a genomic or nucleic acid expression profile, where the amount or level of one or more nucleic acids in the sample is determined. In these embodiments, the sample that is assayed to generate the expression profile employed in the diagnostic methods is one that is a nucleic acid sample. The nucleic acid sample includes a plurality or population of distinct nucleic acids that includes the expression information of the phenotype determinative genes of interest of the cell or tissue being diagnosed. The nucleic acid may include RNA or DNA nucleic acids, e.g., mRNA, cRNA, cDNA etc., so long as the sample retains the expression information of the host cell or tissue from which it is obtained.

[0043] The sample may be prepared in a number of different ways, as is known in the art, e.g., by mRNA isolation from a cell, where the isolated mRNA is used as is, amplified, employed to prepare cDNA, cRNA, etc., as is known in the differential expression art. The sample is typically prepared from a cell or tissue harvested from a subject to be diagnosed, e.g., via blood drawing, biopsy of tissue, using standard protocols, where cell types or tissues from which such nucleic acids may be generated include any tissue in which the expression pattern of the to be determined phenotype exists. Cells may be cultured prior to analysis.

[0044] The expression profile may be generated from the initial nucleic acid sample using any convenient protocol. While a variety of different manners of generating expression profiles are known, such as those employed in the field of differential gene expression analysis, one representative and convenient type of protocol for generating expression profiles is array based gene expression profile generation protocols. Such applications are hybridization assays in which a nucleic acid that displays “probe” nucleic acids for each of the genes to be assayed/profiled in the profile to be generated is employed. In these assays, a sample of target nucleic acids is first prepared from the initial nucleic acid sample being assayed, where preparation may include labeling of the target nucleic acids with a label, e.g., a member of signal producing system. Following target nucleic acid sample preparation, the sample is contacted with the array under hybridization conditions, whereby complexes are formed between target nucleic acids that are complementary to probe sequences attached to the array surface. The presence of hybridized complexes is then detected, either qualitatively or quantitatively.

[0045] Specific hybridization technology which may be practiced to generate the expression profiles employed in the subject methods includes the technology described in U.S. Pat. Nos. 5,143,854; 5,288,644; 5,324,633; 5,432,049; 5,470,710; 5,492,806; 5,503,980; 5,510,270; 5,525,464; 5,547,839; 5,580,732; 5,661,028; 5,800,992; the disclosures of which are herein incorporated by reference; as well as WO 95/21265; WO 96/31622; WO 97/10365; WO 97/27317; EP 373 203; and EP 785 280. In these methods, an array of “probe” nucleic acids that includes a probe for each of the phenotype determinative genes whose expression is being assayed is contacted with target nucleic acids as described above. Contact is carried out under hybridization conditions, e.g., stringent hybridization conditions as described above, and unbound nucleic acid is then removed. The resultant pattern of hybridized nucleic acid provides information regarding expression for each of the genes that have been probed, where the expression information is in terms of whether or not the gene is expressed and, typically, at what level, where the expression data, i.e., expression profile, may be both qualitative and quantitative.

[0046] Alternatively, non-array based methods for quantitating the levels of one or more nucleic acids in a sample may be employed, including quantitative PCR, and the like.

[0047] Where the expression profile is a protein expression profile, any convenient protein quantitation protocol may be employed, where the levels of one or more proteins in the assayed sample are determined. Representative methods include, but are not limited to; proteomic arrays, flow cytometry, standard immunoassays, etc.

[0048] Following obtainment of the expression profile from the sample being assayed, the expression profile is compared with a reference or control profile to make a diagnosis regarding the radiation toxicity susceptibility phenotype of the cell or tissue from which the sample was obtained/derived. Typically a comparison is made with a set of cells from the same source, which has not been exposed to radiation. Additionally, a reference or control profile may be a profile that is obtained from a cell/tissue known to have the susceptible phenotype, and therefore may be a positive reference or control profile. In addition, a reference/control profile may be from a cell/tissue known to not have the susceptibility phenotype, and therefore be a negative reference/control profile.

[0049] In certain embodiments, the obtained expression profile is compared to a single reference/control profile to obtain information regarding the phenotype of the cell/tissue being assayed. In yet other embodiments, the obtained expression profile is compared to two or more different reference/control profiles to obtain more in depth information regarding the phenotype of the assayed cell/tissue. For example, the obtained expression profile may be compared to a positive and negative reference profile to obtain confirmed information regarding whether the cell/tissue has the phenotype of interest.

[0050] The difference values, i.e. the difference in expression in the presence and absence of radiation may be performed using any convenient methodology, where a variety of methodologies are known to those of skill in the array art, e.g., by comparing digital images of the expression profiles, by comparing databases of expression data, etc. Patents describing ways of comparing expression profiles include, but are not limited to, U.S. Pat. Nos. 6,308,170 and 6,228,575, the disclosures of which are herein incorporated by reference. Methods of comparing expression profiles are also described above.

[0051] A statistical analysis step is then performed to obtain the weighted contribution of the set of predictive genes. Nearest shrunken centroids analysis, is applied as described in Tibshirani et al. (2002) P.N.A.S. 99:6567-6572 to compute the centroid for each class, then compute the average squared distance between a given expression profile and each centroid, normalized by the within-class standard deviation.

[0052] To perform a shrunken centroids analysis, let xik be the expression for genes i=1, 2, . . . p and samples j=1, 2, . . . n. Classes are 1, 2, . . . K, and Ck is indices of the nk samples in class k. The ith component of the centroid for class k is {overscore (x)}ik=ΣjεCkxijnk/nk the mean expression value in class k for gene i; the ith component of the overall centroid is {overscore (x)}i=Σj=1xii/nn. In words, one shrinks the class centroids toward the overall centroids after standardizing by the within-class standard deviation for each gene. This standardization has the effect of giving higher weight to genes whose expression is stable within samples of the same class. 1$\begin{array}{cc}{d}_{\mathrm{ik}}=\frac{{\stackrel{\_}{x}}_{\mathrm{ik}}-{\stackrel{\_}{x}}_i}{m_k·\left(s_i+s_o\right)},& \left[1\right]\end{array}$

[0053] where si is the pooled within-class standard deviation for gene i: 2$\begin{array}{cc}{s}_i^2=\frac{1}{n-K}\sum _k\sum _{j\in C_k}{\left(x_{\mathrm{ij}}-{\stackrel{\_}{x}}_{\mathrm{ik}}\right)}^2& \left[2\right]\end{array}$

[0054] and mk={square root}{square root over (1/nk+1/n)} makes mk·si equal to the estimated standard error of the numerator in dik. In the denominator, the value so is a positive constant (with the same value for all genes), included to guard against the possibility of large dik values arising by chance from genes with low expression levels. s0 is set to be equal to the median value of the si over the set of genes.

[0055] Thus dik is a t statistic for gene i, comparing class k to the overall centroid. Eq. 1 can be rewritten as

{overscore (x)} ik ={overscore (x)} i +m k (si+so)dik  [3]

[0056] This method shrinks each dik toward zero, giving dik and yielding shrunken centroids or prototypes

{overscore (x)}′ ik ={overscore (x)} i +m k (si+s0)d′ik  [4]

[0057] The shrinkage is called soft thresholding: each dik is reduced by an amount Δ in absolute value and is set to zero if its absolute value is less than zero. Algebraically, soft thresholding is defined by

d′ ik =sign)(dik)(|dik|−Δ)+  [5]

[0058] where + means positive part (t+=t if t>0 and zero otherwise). Because many of the {overscore (x)}ik values will be noisy and close to the overall mean {overscore (x)}i, soft thresholding produces more reliable estimates of the true means. This method has the desirable property that many of the components (genes) are eliminated from the class prediction as the shrinkage parameter Δ is increased. Specifically, if for a gene i, dik is shrunken to zero for all classes k, then the centroid for gene i is {overscore (x)}i, the same for all classes. Thus gene i does not contribute to the nearest-centroid computation.

[0059] Depending on the type and nature of the reference/control profile(s) to which the obtained expression profile is compared, the above comparison step yields information as to whether a patient is susceptible to toxicity after exposure to antiproliferative therapy. As such, the above comparison step can yield a positive/negative determination of a susceptible phenotype of an assayed cell/tissue.

[0060] The prediction of susceptibility is probabilistically defined, where the cut-off for predicted susceptibility may be empirically derived, for example as shown in FIG. 3. In one embodiment of the invention, a probability of about 0.4 may be used to distinguish between susceptible and non-susceptible patients, more usually a probability of about 0.5, and may utilize a probability of about 0.6 or higher. A “high” probability may be at least about 0.75, at least about 0.7, at least about 0.6, or at least about 0.5. A “low” probability may be not more than about 0.25, not more than 0.3, or not more than 0.4. In many embodiments, the above-obtained information about the cell/tissue being assayed is employed to predict whether a host, subject or patient is treated with a therapy of interest, e.g. treatment with ionizing radiation, exposure to a chemotherapeutic agent etc., and to optimize the dose therein.

Databases of Expression Profiles

[0061] Also provided are databases of expression profiles of phenotype determinative genes. Such databases will typically comprise expression profiles of various cells/tissues having susceptible phenotypes, negative expression profiles, etc., where such profiles are further described below.

[0062] The expression profiles and databases thereof may be provided in a variety of media to facilitate their use. “Media” refers to a manufacture that contains the expression profile information of the present invention. The databases of the present invention can be recorded on computer readable media, e.g. any medium that can be read and accessed directly by a computer. Such media include, but are not limited to: magnetic storage media, such as floppy discs, hard disc storage medium, and magnetic tape; optical storage media such as CD-ROM; electrical storage media such as RAM and ROM; and hybrids of these categories such as magnetic/optical storage media. One of skill in the art can readily appreciate how any of the presently known computer readable mediums can be used to create a manufacture comprising a recording of the present database information. “Recorded” refers to a process for storing information on computer readable medium, using any such methods as known in the art. Any convenient data storage structure may be chosen, based on the means used to access the stored information. A variety of data processor programs and formats can be used for storage, e.g. word processing text file, database format, etc.

[0063] As used herein, “a computer-based system” refers to the hardware means, software means, and data storage means used to analyze the information of the present invention. The minimum hardware of the computer-based systems of the present invention comprises a central processing unit (CPU), input means, output means, and data storage means. A skilled artisan can readily appreciate that any one of the currently available computer-based system are suitable for use in the present invention. The data storage means may comprise any manufacture comprising a recording of the present information as described above, or a memory access means that can access such a manufacture.

[0064] A variety of structural formats for the input and output means can be used to input and output the information in the computer-based systems of the present invention. Such presentation provides a skilled artisan with a ranking of similarities and identifies the degree of similarity contained in the test expression profile.

Reagents and Kits

[0065] Also provided are reagents and kits thereof for practicing one or more of the above-described methods. The subject reagents and kits thereof may vary greatly. Reagents of interest include reagents specifically designed for use in production of the above described expression profiles of phenotype determinative genes.

[0066] One type of such reagent is an array of probe nucleic acids in which the phenotype determinative genes of interest are represented. A variety of different array formats are known in the art, with a wide variety of different probe structures, substrate compositions and attachment technologies. Representative array structures of interest include those described in U.S. Pat. Nos. 5,143,854; 5,288,644; 5,324,633; 5,432,049; 5,470,710; 5,492,806; 5,503,980; 5,510,270; 5,525,464; 5,547,839; 5,580,732; 5,661,028; 5,800,992; the disclosures of which are herein incorporated by reference; as well as WO 95/21265; WO 96/31622; WO 97/10365; WO 97/27317; EP 373 203; and EP 785 280. In certain embodiments, the number of genes that are from Table 3 that is represented on the array is at least 10, usually at least 25, and may be at least 50, 100, up to including all of the genes listed in Table 3, preferably utilizing the top ranked set of genes. The subject arrays may include only those genes that are listed in Table 3, or they may include additional genes that are not listed in Table 3. Where the subject arrays include probes for such additional genes, in certain embodiments the number % of additional genes that are represented does not exceed about 50%, usually does not exceed about 25%. In many embodiments where additional “non-Table 3” genes are included, a great majority of genes in the collection are phenotype determinative genes, where by great majority is meant at least about 75%, usually at least about 80% and sometimes at least about 85, 90, 95% or higher, including embodiments where 100% of the genes in the collection are predictive genes.

[0067] Another type of reagent that is specifically tailored for generating expression profiles of phenotype determinative genes is a collection of gene specific primers that is designed to selectively amplify such genes, for use in quantitative PCR and other quantitation methods. Gene specific primers and methods for using the same are described in U.S. Pat. No. 5,994,076, the disclosure of which is herein incorporated by reference. Of particular interest are collections of gene specific primers that have primers for at least 10 of the genes listed in Table 3, above, often a plurality of these genes, e.g., at least 25, and may be 50, 100 or more to include all of the genes listed in Table 3. The subject gene specific primer collections may include only those genes that are listed in Table 3, or they may include primers for additional genes that are not listed in Table 3. Where the subject gene specific primer collections include primers for such additional genes, in certain embodiments the number % of additional genes that are represented does not exceed about 50%, usually does not exceed about 25%. In many embodiments where additional “non-Table 3” genes are included, a great majority of genes in the collection are phenotype determinative genes, where by great majority is meant at least about 75%, usually at least about 80% and sometimes at least about 85, 90, 95% or higher, including embodiments where 100% of the genes in the collection are predictive genes.

[0068] The kits of the subject invention may include the above described arrays and/or gene specific primer collections. The kits may further include a software package for statistical analysis of one or more phenotypes, and may include a reference database for calculating the probability of susceptibility. The kit may include reagents employed in the various methods, such as primers for generating target nucleic acids, dNTPs and/or rNTPs, which may be either premixed or separate, one or more uniquely labeled dNTPs and/or rNTPs, such as biotinylated or Cy3 or Cy5 tagged dNTPs, gold or silver particles with different scattering spectra, or other post synthesis labeling reagent, such as chemically active derivatives of fluorescent dyes, enzymes, such as reverse transcriptases, DNA polymerases, RNA polymerases, and the like, various buffer mediums, e.g. hybridization and washing buffers, prefabricated probe arrays, labeled probe purification reagents and components, like spin columns, etc., signal generation and detection reagents, e.g. streptavidin-alkaline phosphatase conjugate, chemifluorescent or chemiluminescent substrate, and the like.

[0069] In addition to the above components, the subject kits will further include instructions for practicing the subject methods. These instructions may be present in the subject kits in a variety of forms, one or more of which may be present in the kit. One form in which these instructions may be present is as printed information on a suitable medium or substrate, e.g., a piece or pieces of paper on which the information is printed, in the packaging of the kit, in a package insert, etc. Yet another means would be a computer readable medium, e.g., diskette, CD, etc., on which the information has been recorded. Yet another means that may be present is a website address which may be used via the internet to access the information at a removed site. Any convenient means may be present in the kits.

Method of Analyzing Genes for Predictive Value

[0070] In another aspect of the invention, methods are provided for identifying genes and proteins that are predictive of a phenotype of interest. Such analytical methods provide a set of molecules whose pattern of expression yields information about a phenotype of interest. The molecules may be transcriptional responses, expression of a protein, post-translational protein modification, e.g. cleavage, phosphorylation and dephosporylation, glycosylation, etc.

[0071] The pattern of expression may be basal levels of expression in a target cell type, e.g. expression of a gene in a cancer cell, differential expression of a gene in a normal v. a cancer cell, expression of a gene during a specific developmental stage, basal phosphorylation of a protein in a cell, and the like. The pattern of expression may also be in response to a treatment of interest, e.g. exposure to radiation, exposure to a therapeutic agent, exposure to cytokines, response of cells in a mixed lymphocyte reaction, and the like. The shrunken centroid analysis described above may be used to determine an expression profile for any phenotype of interest.

[0072] The phenotype of interest may be susceptibility to toxicity, response to a therapeutic regimen or agent, development of autoimmune disease, development of graft rejection, development of graft v. host disease, distinction of heterogeneity in an early stage of cancer, e.g. prediction of probable course of disease, and the like.

[0073] To obtain the set of predictive genes, initially cohorts are gathered for the phenotype of interest, e.g. patients suffering from a disease of interest, responders and non-responders to a treatment of interest, and the like. One or more cohorts are gathered for the phenotype of interest, and one or more for a control, preferably a matched control group, according to methods known in the art.

[0074] An expression profile for the trait to be examined is made. Convenient methods for examining large groups of genes include hybridization to microarrays, as discussed above and in the examples. Alternatively, proteomics arrays may be used to determine protein profiles, antibody array can be used to detect the presence of epitopes of interest in a sample, various methods known in the art for quantitative hybridization of a nucleic acid may be used, and the like. As discussed above, the basal expression level may be taken, or a response to a particular stimulus. In many cases it is desirable to determine a difference in expression between a control and a test sample. The expression may be normalized a control, to expression of a housekeeping gene or genes, etc., as known in the art.

[0075] Many phenotypes of interest are actually the result of different underlying genotypes, where a heterogeneous response over a patient population can make analysis difficult. To address the problem of heterogeneity, the following heterogeneity-associated transformation (HAT) is performed, using the following equation: 3$\begin{array}{cc}{x}^{\prime }\left(i\right)={\left[x\left(i\right)-{\stackrel{\_}{x}}_c\left(i\right)\right]}^2& \left[6\right]\end{array}$

[0076] where x(i) is the change in expression for gene i, and {overscore (x)}c(i) is the average change in expression for gene i among the control samples. HAT generates equivalent values for changes in gene expression that are blunted in some cases and enhanced in others, and hence can capture heterogeneous abnormalities among the radiation sensitive patients. Genes with divergent transcriptional responses might be overlooked by comparing the average response of controls to the average response, but are successfully identified after transforming the data.

[0077] After transforming the data, nearest shrunken centroid analysis is performed, as described above and in Tibshirani et al. (2002), supra. The centroid of gene expression for a class of samples is defined as a multi-component vector, in which each component is the expression of a gene averaged over the samples. Samples are then classified by proximity to the nearest centroid. In order to verify the prediction, it is desirable to test profiles against an independent set of samples, or with cross-validation.

[0078] The probability of a specific outcome is then calculated. The cut-off for a particular diagnosis will be determined empirically, based on the specific set of data, and may be modeled to include the weighted probability for rare events.

[0079] The above-described analytical methods may be embodied as a program of instructions executable by computer to perform the different aspects of the invention. Any of the techniques described above may be performed by means of software components loaded into a computer or other information appliance or digital device. When so enabled, the computer, appliance or device may then perform the above-described techniques to assist the analysis of sets of values associated with a plurality of genes in the manner described above, or for comparing such associated values. The software component may be loaded from a fixed media or accessed through a communication medium such as the internet or other type of computer network. The above features are embodied in one or more computer programs may be performed by one or more computers running such programs.

[0080] The following examples are offered by way of illustration and not by way of limitation.

EXPERIMENTAL

[0081] Toxicity from radiation therapy is a grave problem for cancer patients, and methods are needed for predicting its occurrence. Microarrays were used to analyze abnormal transcriptional responses to DNA damage in cultured lymphocytes. A transformation of the data was devised to account for the possibility that toxicity can arise from defects in different pathways. The risk of toxicity was then computed for each patient using nearest shrunken centroids, a method that identifies predictive genes. Transcriptional responses in 24 genes predicted radiation toxicity in 9 of 14 patients with no false positives among 43 controls. Some patients had defective responses to ionizing radiation, while others had defective responses to both ultraviolet and ionizing radiation. This approach has the potential to predict toxicity from ionizing radiation and other anticancer agents, enabling physicians to design a safe treatment plan for each patient.

[0082] Materials and Methods

[0083] Patient cell lines. Subjects were enrolled with informed consent between 1997 and 2002 in accordance with Stanford regulations for human subjects research. Radiation toxicity was graded according to the RTOG Acute and Late Radiation Morbidity Scoring Criteria. Radiation therapy patients donated peripheral blood samples at least 2 months following completion of treatment and resolution of any toxicity. Lymphoblastoid cell lines were established by immortalization of peripheral blood B-lymphocytes with Epstein-Barr virus from the B95-8 monkey cell line. Cells were grown in RPMI 1640 (Gibco) with 15% heat inactivated fetal bovine serum, 1% penicillin/streptomycin, and 2 mM glutamine and stored in liquid nitrogen.

[0084] Treatment of cells with UV and IR. Lymphoblastoid cells were subjected to mock, UV, and IR treatment. For UV treatment, 5×107 cells were suspended in PBS at 6×105 cells/ml to ensure uniform exposure to UV. Cells subjected to mock and IR treatment were also suspended in PBS during this period to ensure similar treatment. For UV treatment, cells were exposed for 15 sec to a germicidal lamp at a fluence of 0.67 J/m2/sec to deliver a 10 J/m2 dose, seeded at 3×105 cells/ml in fresh media, and harvested for RNA 24 hrs later. For IR treatment, 4×107 cells were exposed to 5 Gy IR 20 hrs after the PBS wash and harvested for RNA 4 hrs later.

[0085] Microarray hybridization. Total RNA was labeled with biotin and hybridized to a U95A_v2 GeneChip® microarray, according to manufacturer's protocols (Affymetrix, Santa Clara, Calif.). The expression level for each gene was calculated by Affymetrix GeneChip Microarray Analysis Suite software version 4.0. To account for differences in hybridization between different chips, data from hybridizations were scaled to the average of all data sets, as described by Tusher et al. (2001) Proc. Natl. Acad. Sci. USA; 98:5116-5121.

[0086] Analysis of microarray data. The data was in the form of change in gene expression, computed for each individual as the difference in expression before and after exposure to UV or IR. Analyses were based on changes in gene expression, because this was less sensitive to variation among different individuals than the basal or induced levels of expression. Thus, we used the paired data option in Significance Analysis of Microarrays (SAM), which ranks genes by change in expression relative to the standard deviation in multiple samples. IR-responsive and UV-responsive genes were identified using data from 9 normal individuals.” The false discovery rate (FDR) is the percentage of genes falsely called significant when the change in gene expression for each individual is randomly chosen to be left unaltered or multiplied by −1. Responsive genes were obtained by choosing a threshold corresponding to an FDR of 10%.

[0087] The nearest shrunken centroid (NSC) classifier was applied to the radiation toxicity and control classes (Tibshirani et al. (2002) Proc. Natl. Acad. Sci. USA 99:6567-6572). The centroid for a class of samples was defined as a multi-component vector, in which each component was the expression of a predictive gene averaged over the samples in that class. NSC shrinks the class centroids towards the overall centroid after normalizing by the within-class standard deviation for each gene. The probability for radiation toxicity associated with an expression profile was computed from its distances to the radiation toxicity and control centroids.

[0088] The accuracy of a supervised classifier such as NSC may appear to be high when applied to the training samples, i.e., the samples used to define the centroids. However, this is not statistically valid. The number of genes is much greater than the number of samples in microarray experiments, providing many opportunities to find genes with expression patterns that correlate with the class of interest. Thus, supervised classifiers are susceptible to overfitting, and their accuracy must be tested by cross-validation on samples not used for training Ambroise and McLachlan (2002) Proc Natl Acad Sci USA; 99:6562-6566.

[0089] We subjected NSC to 14-fold cross-validation by dividing the samples into 14 subsets. Each subset contained one radiation sensitive patient plus 2 or 3 controls selected from the radiation controls, skin cancer patients, and non-cancer controls. We withheld one subset and trained NSC on the remaining samples to identify a set of predictive genes, which defined a radiation sensitive centroid and a control centroid. Each sample from the withheld subset was classified by its proximity to the nearest centroid. This protocol was repeated for each of the 14 subsets until every sample was classified. To avoid biasing our predictions, samples from the 9 subjects analyzed by SAM were excluded as training samples for NSC, but were assigned probabilities for radiation toxicity.

[0090] Hierarchical clustering (Eisen et al. (1998) Proc. Natl. Acad. Sci. USA; 95:14863-14868) used centered Pearson correlation and complete linkage clustering, and was displayed with TreeView. Biological functions were assigned from the literature and the SOURCE database.

[0091] Results

[0092] Radiation sensitive patients and controls. Fourteen radiation therapy patients were enrolled after suffering unusual levels of radiation toxicity within one month of treatment, as judged by a faculty member in the Department of Radiation Oncology at Stanford. Toxicity was severe enough so that 11 of these 14 patients required interruption or early termination of treatment. These interventions helped limit the reported toxicities to grades 2 and 3. Thirteen patients with radiation toxicity limited to grades 0 or 1 were recruited as controls. We attempted to match this patient group to the radiation sensitive group by radiation field and dose, tumor type, gender, and concurrent chemotherapy (Table 4). The average age of the radiation control patients was 59 years ±13 years, while the average age of the radiation sensitive patients was 51 years ±11 years. Since the risk of radiation toxicity increases with age (Turesson et al. (1996) Int J Radiat Oncol Biol Phys; 36:1065-75), the younger age of the radiation sensitive patients was protective and should enhance the validity of our results. This study incorporated significant heterogeneity in radiation treatments. Importantly, the radiation sensitive group was matched to the radiation control group. This facilitated our goal to find genes that predicted acute toxicity, independently of the underlying tumor or site of treatment.

TABLE 4
Clinical characteristics of radiation therapy patients
Age/gender/diagnosis		Patient	Reaction	Grade	Radiation/concurrent chemotherapy
Radiation sensitive patients
37F	breast cancer	RadS4*	skin	3‡	45 Gy to breast
49F	breast cancer	RadS14	skin	2‡	50 Gy to breast, 10 Gy boost/cytoxan, 5-
					FU
53F	breast cancer	RadS12	skin	2‡	55 Gy to breast
65F	breast cancer	RadS1	*skin	3‡	45 Gy to breast
37F	Hodgkin's disease	RadS10	skin; breast	3‡	40 Gy mantle field, 10 Gy neck boost
			cancer 20y
			later
50M	Hodgkin's disease	RadS6	skin; stroke	3	44 Gy mantle field
			8y later
67M	Hodgkin's disease	RadS8	pneumonitis	2‡	43 Gy mantle field, 36 Gy spade field
57M	low grade lymphoma	RadS7	mucositis;	3‡	50 Gy to mandible & neck, 45 Gy to hip, hip
			osteonecrosis		& jaw, cystitis 10y later
			of
60M	low grade lymphoma	RadS2	*skin	3†	31 Gy to lacrimal glands in both orbits
41M	cancer of tongue	RadS3	*mucositis	3‡	70 Gy to tongue/tpz, cisplatin, 5-FU
45M	salivary gland cancer	RadS9	skin,	3‡	40 Gy to oral cavity, 48 Gy to neck, 12 Gy
			mucositis		to tongue/cisplatin, 5-FU
67F	endometrial cancer	RadS13	diarrhea	3†	42 Gy to pelvis
52F	orbital pseudotumor	RadS11	orbital edema	2	31 Gy to orbit
33F	brainstem AVM	RadS5	*cerebral	3	18 Gy stereotactic radiation to brainstem
			edema
Radiation control patients
45F	breast cancer	RadC8	skin	1	50 Gy to chest wall
59F	breast cancer	RadC7	skin	1	50 Gy to breast, 10 Gy boost
65F	breast cancer	RadC9	skin	1	50 Gy to breast, 10 Gy boost
73F	breast cancer	RadC12	skin	1	50 Gy to breast
78F	breast cancer	RadC13	skin	1	50 Gy to breast, 10 Gy boost
39F	Hodgkin's disease	RadC1	none	0	44 Gy total lymphoid irradiation
49F	Hodgkin's disease	RadC4	none	0	44 Gy mantle field
46M	mixed cell lymphoma	RadC2	none	0	36 Gy to para-aortic & inguinal nodes, 31 Gy
					to orbital recurrence
63M	large cell lymphoma	RadC3	none	0	36 Gy to parotid gland
50F	salivary gland cancer	RadC5	skin	1	56 Gy to oropharynx
56M	cancer of tonsil	RadC10	skin,	1	70 Gy to oropharynx/cisplatin, 5-FU
			mouth
			dryness
70F	cancer of oropharynx	RadC6	skin,	1	66 Gy to oropharynx
			mouth
			dryness
76M	cancer of tongue	RadC11	skin,	1	70 Gy to oropharynx/tpz, cisplatin, 5-FU
			mouth
			dryness
*patient misclassified by NSC/HAT analysis of UV and IR responses
‡dose involved interruption of treatment
†dose involved early termination of treatment
Patients with reactions limited to grade 0 or 1 were included radiation controls (RadC). Patients with acute reactions (RadS) were enrolled as described in the text. Patients RadS6, RadS7, and RadS10 also suffered from grade 4 late reactions 8, 10, and 20 years following radiation therapy. Patients are numbered in the order in which they appear in FIGS. 1 and 3 from left to right.
Abbreviations:
AVM = arteriovenous malformation;
5-FU = 5-fluorouracil;
tpz = tirapazamine

[0093] Cells were exposed to UV as well as IR to determine whether some radiation sensitive patients have a general defect in responding to DNA damage. Because skin cancer is associated with UV exposure, we enrolled 15 patients diagnosed with skin cancer before age 40 to serve as additional controls. A successful classification method should not assign a high risk for radiation toxicity to the skin cancer patients. Fifteen subjects without cancer were matched to the skin cancer patients for age, gender, and race. Because we recruited patients with early skin cancer, their average age was 38 years ±8 years, and the average age of the normal individuals was 31 years ±5 years, which were significantly younger than the age of the radiation sensitive patients. A total of 57 subjects were recruited for study.

[0094] Analysis by SAM and nearest shrunken centroids. To identify genes normally induced or repressed by IR or UV, we applied SAM to data from 9 subjects without a history of cancer. SAM identified 1491 IR-responsive genes and 2114 UV-responsive genes. We previously developed an enhancement of nearest centroids, nearest shrunken centroids (NSC), which successfully identified small sets of highly predictive genes for other classification problems. However, when we applied NSC to these IR and UV-responsive genes, classification required 1831 genes while generating 10 errors.

[0095] Heterogeneity-associated transformation. A new approach was needed to identify predictive genes. Radiation toxicity can arise from several different underlying genetic defects, generating divergent transcriptional responses. For example, one subset of radiation sensitive patients could have a defect in signaling through ATM, leading to a failure to activate p53 after IR and a blunted response in p53-induced genes. Another subset could have a defect in DNA repair, leading to prolonged activation of ATM and enhanced transcription of p53-induced genes.

[0096] To address the problem of heterogeneity, we performed the following heterogeneity-associated transformation (HAT) 4$\begin{array}{cc}{x}^{\prime }\left(i\right)={\left[x\left(i\right)-{\stackrel{\_}{x}}_c\left(i\right)\right]}^2& \mathrm{Equation}1\end{array}$

[0097] where x(i) is the change in expression for gene i, and {overscore (x)}c(i) is the average change in expression for gene i among the control samples. HAT generates similar values from changes in gene expression that are blunted in some cases or enhanced in others, and hence can capture heterogeneous abnormalities among the radiation sensitive patients. Simulations of microarray data demonstrated that NSC/HAT is more efficient than NSC alone in identifying genes with heterogeneous responses, but less efficient in identifying genes with homogeneous responses.

[0098] Genes with heterogeneous transcriptional responses were successfully identified after transforming the data with HAT. FIG. 1 shows the effect of HAT on two predictive genes, cyclin B and 8-oxo-dGTPase. When x′(i) replaced x(i) for the set of 1491 IR-responsive genes and 2114 UV-responsive genes, NSC identified a subset of 24 genes that predicted radiation toxicity, with 5 false negatives and no false positives (FIG. 2). The low error rate occurred for a wide range of threshold values for the nearest shrunken centroid classifier. Thus, HAT enhanced the power of NSC, suggesting that the radiation sensitive patients constitute a heterogeneous group.

[0099] Prediction of radiation toxicity. Of the 24 predictive genes, 20 were IR-responsive, and 4 were UV-responsive. NSC/HAT used these responses to compute a probability of radiation toxicity for each subject in the 48-sample training set (FIG. 3, upper panel). The separation between the radiation sensitive patients and controls indicated a strong correlation between the responses of the 24 genes and radiation toxicity. This correlation was confirmed by 14-fold cross-validation, which predicted radiation toxicity in 9 of 14 patients, with no false positives among 43 controls, which included the 9 subjects previously used to identify the damage response genes, p=2.2×10−7 by Fisher's two-tailed exact test (FIG. 3, lower panel).

[0100] The genes identified during cross-validation were essentially the same as the genes identified from the full 48-sample training set. Among the 24 genes identified for each of the 14 cross-validation trials, 80% were among the 24 top-ranked genes from the 48-sample training set, and 99% were among the 52 top-ranked genes from that set (FIG. 4). To test the stability of the cross-validation protocol, we performed 10 new trials of 14-fold cross-validation by withholding different subsets of patients. All 10 trials successfully predicted toxicity in the same 9 of 14 patients with no false positives among the controls.

[0101] Delayed toxicity in the form of progressive damage after completion of treatment is a grave problem. Three patients (RadS6, RadS7, and RadS10) suffered grade 4 delayed toxicity, and all were predicted successfully (Table 4). Toxicity from non-genetic factors cannot be predicted by our approach. Of the 5 patients with radiation toxicity not predicted by NSC/HAT, at least 2 (RadS3 and RadS5) were at high risk for toxicity from non-genetic factors. Patient RadS3 suffered grade 3 mucositis from an experimental protocol that included high dose radiation plus tirapazamine, cisplatin, and 5-FU. Subsequent review of patients treated by this protocol revealed that 28 of 62 (45%) suffered mucositis of grade 3 or higher. Patient RadS5 had an arteriovenous malformation that was treated with stereotactic guidance of a single 18 Gy dose to a 1.8 cm3 volume in the midbrain and pons. A statistical model indicates that the midbrain and pons region has the highest probability for permanent symptomatic injury, with a 40% to 45% probability for the dose and volume delivered to RadS5. To determine whether RadS3 and RadS5 had an effect on the results, we excluded them and repeated the analysis. Despite the decreased number of samples available for training, NSC/HAT successfully predicted toxicity in 9 of the remaining 12 cases, with no false positives among 43 controls.

[0102] Ruling out confounding variables. The enormous number of genes analyzed by microarrays offers great opportunity for discovery. However, transcriptional responses that appear to be predictive might instead be due to a confounding variable. Here, the confounding variable could be some other difference between the radiation sensitive patients and the control subjects. The subjects with no cancer or skin cancer were younger than the subjects with radiation toxicity. They were also free of cancers of the internal organs, which might be associated with an abnormal response to DNA damage. Furthermore, they were never treated with IR, and 5% to 10% might be at risk for toxicity. To address this problem, we omitted the 30 subjects with no cancer or skin cancer and analyzed the 27 radiation therapy patients. T his restricted analysis was also successful despite the fewer samples available for training. A set of 13 genes yielded the same 5 false negatives reported above, with no false positives among the 13 controls. When tested on the 30 omitted subjects, these 13 genes predicted only 3 positives, consistent with the expected low risk for toxicity in the general population. The set of predictive genes was stable in the face of restricted analysis. Nine of the 13 genes were among the 24 top-ranked genes identified with the 48-sample training set, and 20 of the 24 predictive genes from the 48-sample training set were among the top 81 ranked genes in the restricted analysis.

[0103] Heterogeneity among the radiation sensitive patients. The 57 subjects and 52 top-ranked predictive genes identified by HAT/NSC were organized by hierarchical clustering (FIG. 4). The 52 genes were obtained from the 48-sample training set and included 40 IR-responsive genes and 12 UV-responsive genes. The radiation sensitive patients did not form a single cluster, suggesting that radiation toxicity arises from more than one type of underlying defect. Four radiation sensitive patients clustered loosely on the left side of the heat map. Cells from these patients had abnormal responses in many of the 52 genes, including the cluster of 9 UV-responsive genes at the bottom of the heat map. These patients may have a general defect in responding to DNA damage. Five radiation sensitive patients clustered on the right side of the heat map. These patients had a relatively normal response in the UV-response gene cluster, but had prominent defects in IR-response genes.

[0104] Genes with transcriptional responses that predict radiation toxicity. No single gene predicted radiation toxicity. Instead, the response of several genes provided a signature for toxicity. The 52 top-ranked predictive genes are involved in several different cellular processes (FIG. 4).

[0105] Four genes had roles in DNA repair. XPC-complementing protein (RAD23 homolog B) is involved in nucleotide excision repair. Its response to IR was abnormal in many radiation sensitive samples. The 8-oxo-dGTPase gene product (NUDT1) hydrolyzes 8-oxo-dGTP to 8-oxo-dGMP, which is then converted to the nucleoside, 8-oxo-dG, thus preventing misincorporation of 8-oxo-dGTP into DNA. Urinary 8-oxo-dG is a biomarker for oxidative DNA damage, and decreased levels correlated with acute radiosensitivity in breast cancer patients. These results may be explained by the abnormal IR-suppressed expression of 8-oxo-dGTPase we observed in several radiation sensitive patients (FIG. 1). IR-induced DNA double-strand breaks are repaired by homologous recombination (HR) or nonhomologous end-joining.

[0106] Human RuvB-like protein 1 (RUVBL1) is homologous to bacterial RuvB, a DNA helicase that catalyses branch migration of Holliday junctions during HR. RuvB-like proteins are also components of the yeast INO80 complex, which remodels chromatin, and confers resistance to DNA damaging agents. PTB-associated splicing factor (PSF) may be involved in HR by promoting DNA strand invasion. Interestingly, RUVBL1 and PSF responded abnormally to UV but not IR in many radiation sensitive patients. None of the 52 top-ranked predictive genes was involved in nonhomologous end-joining. However, this pathway does not respond to IR transcriptionally, but rather involves activation of a DNA-dependent protein kinase.

[0107] Five predictive genes are involved in the general stress response. Cells from radiation sensitive patients showed abnormal IR responses in genes encoding c-fos, MAP kinase-activated protein kinase 2 (MAPKAP2), heat shock protein 27 (HSPB1), which is a substrate of MAPKAP2 phosphorylation, and protein phosphatase 1A (PPM1A), which inhibits stress-activated protein kinase cascades. Abnormal UV responses were observed for calmodulin (CALM1).

[0108] Four predictive genes are involved in the ubiquitin/proteasome protein degradation pathway, which is induced by oxidative stress. Abnormal IR responses were observed for ubiquitin B (UBB), proteasome activator subunit (PSME2), and two subunits of the 26S proteasome, β subunit 4 (PSMB4) and the non-ATPase subunit 1 (PSMD1).

[0109] Three cell cycle genes responded abnormally to UV in some radiation sensitive patients: cyclin B1 (CCNB1), cyclin A2 (CCNA2), and CDC28 protein kinase 2 (CKS2), which negatively regulates CDK-cyclin complexes.

[0110] Apoptosis genes included tumor necrosis factor (TNFSF7), core binding factor (CBFB), and the mitochondrial adenine nucleotide transporter (ANT). ANT regulates mitochondrial membrane permeability during apoptosis. The fibroblast isoform of ANT (SLC25A6) responded abnormally to IR, and the liver isoform (SLC25A5) responded abnormally to UV in most radiation sensitive patients. Four predictive genes were involved in RNA processing, and the remaining 18 predictive genes were involved in a diverse set of pathways.

[0111] Many cases of radiation toxicity are associated with abnormal transcriptional responses to DNA damage. To identify a subset of highly predictive genes, we subjected the transcriptional responses to a heterogeneity-associated transformation (HAT). Classification by nearest shrunken centroids (NSC) with HAT predicted 9 of 14 cases of radiation toxicity with no false positives among 43 controls. Notably, the false positive rate was very low with a 95% confidence interval of 0% to 7%. Toxicity was successfully predicted in 64% of the radiation sensitive patients with a 95% confidence interval of 42% to 87% by the exact binomial distribution. Even the lower limit of this confidence interval suggests that a significant number of adverse radiation reactions are associated with abnormal transcriptional responses. Furthermore, 2 of the 5 patients not predicted by NSC/HAT were at high risk for radiation toxicity from non-genetic factors and may have been properly classified in terms of transcriptional responses.

[0112] These results are valid for several reasons. First, to guard against the identification of genes that later fail when tested on an independent set of samples, our results were subjected to cross-validation. We used 14-fold cross-validation, which is more robust than the commonly used “leave-one-out” approach. Second, we imposed the additional test of restricted analysis to rule out confounding variables; when we restricted the training set to the 27 radiation therapy patients, there was little effect on prediction error or on the identity of predictive genes. Third, we applied nearest centroids with HAT to the IR responses of all 12,625 probe sets on the microarray. On cross-validation, we successfully predicted 8 of 14 cases of radiation toxicity (RadS5, RadS7, and RadS9-14) with only 2 false positives (RadC8 and RadC9) among the 43 controls. Thus, our results were not an artifact of gene selection bias.

[0113] Finally, our protocol for predicting radiation toxicity used a plausible biological endpoint, the transcriptional response to DNA damage. Appropriately, 20 of the 24 top-ranked genes contributed IR responses, and only 4 genes contributed UV responses. When we attempted to predict radiation toxicity from the less plausible endpoint of basal gene expression, we obtained a low error rate after cross-validation. However, basal expression failed our additional test of restricting analysis to the radiation therapy patients; the prediction error rate increased significantly, and the set of predictive genes changed markedly, indicating the presence of confounding variables that affected basal gene expression.

[0114] The mechanisms of radiation toxicity are heterogeneous. Some radiation sensitive patients had abnormal transcriptional responses to both UV and IR, and others had abnormal responses only to IR. The abnormal responses involved genes from a diverse set of pathways with functions in DNA repair, response to stress, protein degradation, cell cycle regulation, apoptosis, and RNA processing. The genes with abnormal responses may not be mutated, but rather reflect an abnormality in some other gene. For example, abnormal responses in both UV and IR could arise from mutations affecting p53 or ATR. In patients with abnormal responses restricted primarily to IR, the underlying mutations could be in the ATM-dependent signaling pathway or a DNA double-strand break repair pathway. Radiation toxicity may also arise from the combined effect of polymorphisms in several genes.

[0115] It is evident that subject invention provides a convenient and effective way of determining whether a patient will be responsive to therapy. The subject methods will provide a number of benefits, including avoidance of delays in alternative treatments, elimination of exposure to adverse effects of therapeutic antibodies and reduction of unnecessary expense. As such, the subject invention represents a significant contribution to the art.

[0116] All publications and patent applications cited in this specification are herein incorporated by reference as if each individual publication or patent application were specifically and individually indicated to be incorporated by reference. The citation of any publication is for its disclosure prior to the filing date and should not be construed as an admission that the present invention is not entitled to antedate such publication by virtue of prior invention.

[0117] Although the foregoing invention has been described in some detail by way of illustration and example for purposes of clarity of understanding, it is readily apparent to those of ordinary skill in the art in light of the teachings of this invention that certain changes and modifications may be made thereto without departing from the spirit or scope of the appended claims.

Claims

1. A method of predicting whether a subject will be susceptible to undesirable toxicity resulting from treatment with an anti-proliferative therapy, said method comprising: (a) obtaining an expression profile for any gene from Table3 for a response to said anti-proliferative therapy in a sample from said subject; and (b) comparing said obtained expression profile to a reference expression profile to predict whether said subject is susceptible to undesirable toxicity.

2. The method according to claim 1, wherein said anti-proliferative therapy comprises administration of ionizing radiation.

3. The method according to claim 1, wherein said anti-proliferative therapy comprises administration of a chemotherapeutic agent that results in DNA damage.

4. The method according to claim 3, wherein said DNA damage comprises double-stranded breaks in DNA.

5. A method of determining whether a subject is susceptible to undesirable toxicity resulting from treatment with radiation therapy, said method comprising: (a) obtaining an expression profile for the response to radiation for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine whether said subject is susceptible to undesirable radiation toxicity.

6. The method according to claim 5, wherein expression profile is a transcriptional profile.

7. The method according to claim 5, wherein said expression profile comprises at least 10 sequences from Table3.

8. The method according to claim 5, wherein said expression profile comprises at least 50 sequences from Table 3.

9. The method according to claim 5, wherein said undesirable toxicity is at least a grade 2 toxicity.

10. A method of determining whether a subject is susceptible to undesirable toxicity resulting from treatment with administration of a chemotherapeutic agent that induces double-stranded breaks in DNA, said method comprising: (a) obtaining an expression profile for the response to said chemotherapeutic agent for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine whether said subject is susceptible to undesirable toxicity.

11. A method of predicting whether a subject will be susceptible to undesirable toxicity resulting from treatment with radiation therapy, said method comprising: (a) obtaining an expression profile for the response to radiation for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine the probability that said subject is susceptible to undesirable radiation toxicity.

12. The method according to claim 11, wherein expression profile is a transcriptional profile.

13. The method according to claim 11, wherein said expression profile comprises at least 10 sequences from Table 3.

14. The method according to claim 11, wherein said expression profile comprises at least 50 sequences from Table 3.

15. The method according to claim 11, wherein said undesirable toxicity is at least a grade 2 toxicity.

16. A method of predicting whether a subject will be susceptible to undesirable toxicity resulting from treatment with administration of a chemotherapeutic agent that induces double-stranded breaks in DNA, said method comprising: (a) obtaining an expression profile for the response to said chemotherapeutic agent for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine the probability that said subject is susceptible to undesirable toxicity.

17. A method of determining the suitability of a patient for radiation therapy, the method comprising: predicting whether a subject will be susceptible to undesirable toxicity resulting from treatment with radiation therapy, said method comprising: (a) obtaining an expression profile for the response to radiation for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine the probability that said patient is susceptible to undesirable radiation toxicity; wherein a patient that is predicted to have a high probability of undesirable radiation toxicity is less suitable for radiation therapy.

18. The method according to claim 17, wherein expression profile is a transcriptional profile.

19. The method according to claim 17, wherein said expression profile comprises at least 10 sequences from Table 3.

20. The method according to claim 17, wherein said expression profile comprises at least 50 sequences from Table 3.

21. The method according to claim 17, wherein said undesirable toxicity is at least a grade 2 toxicity.

22. A method of determining the suitability of a patient for treatment with an anti-proliferative chemotherapeutic agent that induces double-stranded breaks in DNA, the method comprising: predicting whether a subject will be susceptible to undesirable toxicity resulting from treatment with said chemotherapeutic agent, said method comprising: (a) obtaining an expression profile for the response to said chemotherapeutic agent for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine the probability that said patient is susceptible to undesirable toxicity; wherein a patient that is predicted to have a high probability of undesirable toxicity is less suitable for said treatment with an anti-proliferative chemotherapeutic agent.

23. A method of optimizing anti-proliferative therapy for a patient, the method comprising: (a) obtaining an expression profile for the response to said anti-proliferative therapy for a sample for any gene from Table3 from said subject; and (b) comparing said obtained expression profile to a reference expression profile to determine the probability that said patient is susceptible to undesirable toxicity; wherein a dose of said anti-proliferative therapy is selected to minimize to undesirable toxicity, while providing for effective anti-proliferative activity.

24. The method according to claim 23, further comprising obtaining an expression profile for a response to one or more additional anti-proliferative therapies; comparing said expression profiles to determine which therapy minimizes undesirable toxicity while providing for effective anti-proliferative activity.

25. The method according to claim 23, further comprising obtaining an expression profile for the response to said anti-proliferative therapy for (i) a normal cell sample for any gene from Table3 from said subject and (ii) a tumor cell sample for any gene from Table3 from said subject; comparing said expression profiles from said normal cell and said tumor cell to determine which therapy minimizes undesirable toxicity while providing for effective anti-proliferative activity.

26. A method of obtaining an expression profile for the transcriptional response to radiation, the method comprising: exposing a cell sample from an individual to radiation; extracting mRNA from said cell; quantitating the level of mRNA corresponding to a sequence in Table 3; comparing said level of mRNA to the level of said mRNA present in a cell sample from said individual not exposed to radiation.

27. The method according to claim 26, wherein said exposing to radiation comprises exposes said cell to a dose of ionizing radiation of from about 2 to about 10 Gy.

28. The method according to claim 27, wherein said mRNA is extracted after at least about 2 and not more than about 24 hours following said exposure.

29. The method according to claim 27, further comprising exposing a cell sample from said individual to ultraviolet radiation at a dose of at least about 5 J/m2 and not more than about 50 J/m2.

30. The method according to claim 29, wherein said mRNA is extracted after at least about 4 and not more than about 72 hours following said exposure.

31. The method according to claim 26, wherein said comparing step comprises a nearest shrunken centroid analysis step.

32. A method of obtaining an expression profile for the transcriptional response in a phenotype of interest, the method comprising: exposing a cell sample from an individual to said anti-proliferative therapy; extracting mRNA from said cell; quantitating the level of mRNA corresponding to a sequence of interest; comparing by nearest shrunken centroid analysis said level of mRNA to the level of said mRNA present in a cell sample from said individual not exposed to said anti-proliferative therapy.

33. The method according to claim 32, wherein said phenotype of interest comprises anti-proliferative therapy.

34. A kit for determining susceptibility to undesirable toxicity, the kit comprising: a set of primers specific for at least 10 genes as set forth in Table 3; and instructions for use.

35. The kit according to claim 34, further comprising a set of primers specific for at least 25 sequences set forth in Table 3.

36. The kit according to claim 34, further comprising a set of primers specific for at least 50 sequences set forth in Table 3.

37. The kit according to claim 34, further comprising a software package for statistical analysis of expression profiles.

38. A kit for determining susceptibility to undesirable toxicity, the kit comprising: a microarray comprising probes specific for at least 10 genes as set forth in Table 3; and instructions for use.

39. The kit according to claim 38, further comprising probes specific for at least 25 sequences set forth in Table 3.

40. The kit according to claim 38, further comprising probes specific for at least 50 sequences set forth in Table 3.

41. The kit according to claim 38, further comprising a software package for statistical analysis of expression profiles.

42. A method for determining a set of sequences whose expression is predictive of a phenotype of interest, the method comprising: obtaining an expression profile for a set of candidate sequences from a cohort group having said phenotype of interest and from a control group; subjecting said expression profiles to heterogeneity associated transformation; analyzing transformed or untransformed expression profile data by nearest shrunken centroids; and calculating the probability that a specific sequence is predictive of said phenotype of interest.

43. The method according to claim 42, wherein said sequence is a nucleotide sequence expressed in a target cell.

44. The method according to claim 42, wherein said sequence is a protein sequence expressed in a target cell.

45. The method according to claim 42, wherein said sequence is a protein sequence post-translationally modified in a target cell.

46. The method according to claim 42, wherein said obtaining an expression profile for the transcriptional response to radiation comprises: exposing a cell sample from an individual to stimulus; extracting mRNA or protein from said cell; quantitating the level of mRNA or protein; comparing said level of mRNA or protein to the level present in a cell sample from said individual not exposed to said stimulus.