MARKERS FOR PREDICTING THE RESPONSE OF A PATIENT WITH ACUTE MYELOID LEUKEMIA TO ANTI-CANCER DRUGS

[Technical Field] The present invention relates to a marker for predicting the response of a patient with acute myeloid leukemia (AML) to anti-cancer drugs. More particularly, the present invention relates to a kit for predicting the response of AML patients to anticancer drugs, which comprises an agent for measuring the expression level of the marker, and a method for predicting the response of AML patients to anticancer drugs on the basis of the extent of expression of the marker.

[Background Art]

Leukemia may be classified into acute and chronic forms according to the progression rate thereof. The clinic conditions of leukemia are various according to disease type and the chracters of the affected cells. When leukemia affects the lymphoid cells, it is called lymphocytic leukemia. When myeloid cells are affected, the disease is called myeloid leukemia. Chronic myeloid leukemia outbreaks as cells in the maturity period mutate.

Acute myeloid leukemia (AML) is a kind of blood cancer characterized by the unlimited growth and suppression of differentiation of blasts in a specific stage of cell

differentiation. AML shows various genetic mutations which are known to be closely related to the response to anticancer therapy as well as to the prognosis thereof.

For these reasons, cytogenetic methods are widely used for analyzing chromosomal variations and classfying

AML. For example, AML patients with t(8;21) (translocation between the 8 ^th and 21 ^st chromosomes), t(15;17)

(translocation between the 15 ^th and 17 ^th chromosomes) or inv(16) (inversion of the 16 ^th chromosome) have been observed to exhibit a good response to anti-cancer drugs and a hopeful prognosis (Lowenberg et al., N. Engl. J. Med. 1999, 341:1051-1062; Slovak et al . , Blood 2000, 96:4075- 4083; Byrd et al., Blood 2002, 100: 4325-4336; Grimwade et al., Blood 1998, 92:2322-2333; Grimwade et al., Blood 2001, 98:1312-1320).

In addition to cytogenetic methods, molecular analysis methods are also used to examine the relationship between gene mutation, or alteration of expression, and prognosis. Patients having internal tandem duplication mutations of FLT3 (fms-like tyrosine kinase 3) or an increased expression level of EVl (ectotropic viral integration 1 site) gene are known to have a bad prognosis. In contrast, patients having a mutation of the CEBPA (CCAAT/enhancer binding protein alpha) gene have been found to have a good prognosis (Kiyo et al., Blood 1999, 93:3074- 3080; Gilliland and Griffin Blood 2002, 100:1532-1542;

Barjesteh et al., Blood 2003, 101:837-845; Barjesteh et al., Hematol. J. 4:31-40; Preudhortime et al., Blood 2002, 100:2717-2723) .

There have been efforts to reveal the relationship of gene expression information or the expression level of specific protein with response to anticancer drugs (Broxterman et al., Leukemia 2000, 14:1018-1024; Kanda et al., Cancer 2000, 88:2529-2533; Legrand et al., Blood 2000, 96: 870-877; Christiansen et al., J. Clin. Oncol. 2001, 19: 1405-1413; Okutsu et al., MoI. Cancer. Ther. 2002, 1:1035- 1042) . With advances in DNA chip analysis technology, it has become possible to analyze expression levels of as many as tens of thousands of genes in one experiment. With this background, the genes of AML patients have been analyzed on the molecular level using DNA chips (Virtaneva et al., Proc. Natl. Acad. Sci. USA. 2001, 98: 1124-1129; Armstrong et al., Nat. Genet. 2002, 30:41-47; Yeoh et al., Cancer Cell. 2002, 1:133-143; Schoch et al., Proc. Natl. Acad. Sci. USA. 2002, 99:10008-10013; Kohlmann et al . , Genes Chromosomes Cancer 2003, 37:396-405; Yagi et al., Blood 2003 102:1849-1856) .

Nonetheless, no techniques have been provided, thus far, for predicting the response of AML patients to anticancer therapy. Particularly, there has been no way to predict how AML patients having neither chromosomal abnormalities nor the above-mentioned genetic mutations

respond to anticancer therapy.

Leukemia therapy is very complicated and depends on the type of leukemia. Clinically, leukemia patients who show resistance to therapy have very short survival time. Accordingly, it is very significant to form a proper treatment scheme from the various therapies, such as radiation therapy, operation therapy, chemical therapy, etc., and to apply it to patients in view of reducing expensive medical fees as well as mental and physical pain. Of the patients whose illnesses are diagnosed as the same cancer, some may respond positively to a drug whereas others may show no response to the same drug. In addition, the response to the same drug, even though positive, may differ in level from one patient to another. Thus, predictability of the response of patients to drugs endows the physician with the ability to forego unnecessary treatments, thereby increasing the efficacy and effectiveness of the treatment used.

Leading to the present invention, intensive and thorough research into the expression behavior of certain genes of AML patients, conducted by the present inventor, resulted in the finding that 67 marker genes are relevant to a response to anticancer drugs and that the expression level of the marker genes is indicative of the response. [Disclosure]

[Technical Problem]

It is therefore an purpose of the present invention to provide a kit for predicting the response of AML patients to anticancer drugs, comprising an agent capable of predicting the response to therapy using anticancer drugs .

It is another purpose of the present invention to provide a method for predicting the response of an AML patients to the therapy using anticancer drugs.

[Drawing Description]

FIG. 1 is a graph showing p values, obtained in a log rank test for examining the significance of recurrence difference between a low risk group and a high risk group, which are divided according to the risk score of AML patients.

FIG. 2 is a Kaplan-Meier plot showing the recurrence frequency in the two groups of Fig. 1, wherein 33 of the total of 55 AML patients are in a low risk group.

[Best Mode]

In one aspect, the present invention pertains to a marker for use in predicting the response of AML patients to anticancer drugs .

In another aspect, the present invention pertains to a kit for predicting the response of AML patients to anticancer drugs, comprising an agent for measuring the

expression of a marker gene capable of predicting the response of AML patients to anticancer drugs.

It is difficult to classify the response to anticancer drugs at an arbitrary time. For example, one patient exhibits resistance to anticancer therapy from start while another patient reaches an almost complete remission stage at first, but relapses into cancer a few months later. Therefore, selecting genes having different expression levels at a predetermined clinical time as a method of sorting patients into complete and incomplete remission stages is likely to lead to error.

In the present invention, cDNA is synthesized with RNA from the bone marrow of AML patients and hybridized onto a lβk human cDNA chip to detect genes which are different in expression level from those of a normal group. Subsequently, account is taken of the time that patients show .resistance to anticancer therapy, and the following 67 genes, which are found to be significantly (p=0.001) associated with the response to anticancer drugs by way of Cox's proportional hazard model, are selected in order to utilize censored data (the case where the condition of patients cannot be monitored for reasons other than the disease, or where no recurrence occurs to the end) (Cox D. R. J. Royal Stat. Soc. B. 1972 34:187-202): GENX-3414, SPINK2, PECAMl, AQP5, LAPTM5, DDX48, F2RL3, SLC3A2, GPR51, ZC3HDC8. WASFl, PSCD4, CDKNlA, AIFl, ADFP, CD164, RAB8A,

ID2, ITGB2, MGC3047, MYST3, MT3, MGATl, BC002942, EVI2B, CPVL, MXl, LGALSl, MX2, KRTl, MICA, GABBRl, TINF2, MPV17, RXRA, JARIDlD, C2orf22, DPT, NDUFS2, CCIN, MADH3, NAG, E2F5, SMC6Ll, CASPl, LILRBl, TCF4, TNPO3, DNCIl, TIE, ADM, SMUGl, PPPlCA, SALL3, IRF2, KIAA1223, NCOA3, CKS2, ZNF226, CTBS, MetRS, MGC5395, FLJ10349, CD4, EHD4, ARL4A, and DNAJClO. Of them, the four genes PECAMl, LAPTM5, AIFl, and ITGB2 were selected again in independent probing tests . This supports the assertion that the genes are not selected by chance, but that the expression thereof is associated with a response to anticancer drugs.

Through Leave-one-out cross-validation, the selected marker genes are additionally found to be useful as predictors for the therapeutic possibility of anticancer drugs in AML patients in advance of the administration thereof.

In accordance with an embodiment, therefore, the present invention is directed to a kit for predicting the response of AML patients to anticancer drugs, comprising an agent capable of analyzing the expression of an anticancer therapy predictive marker gene selected from among the marker genes .

As used herein, the term ^λ AML (Acute Myeloid Leukemia) ' refers to a malignant blood disease, characterized by the cancerous proliferation of the white blood precursor cells of the myloid line, which accumulate

in organs and interfere with the production of normal blood cells when the balance in the bone marrow is broken and thus invade the peripheral blood or other organs.

The term ^λ marker for predicting the response to anticancer drugs' as used herein means a material which is used to determine whether or not an anticancer drug is useful in the treatment of cancer in advance of the administration thereof by way of the measurement of the expression level on the basis of which the response to the anticancer drug is predictable. The marker may include organic biomolecules such as nucleic acids, polypeptides, proteins, lipids, and saccharides. For the purpose of the invention, the marker for predicting the response to anticancer drugs therapy is selected from among nucleic acids and polypeptides, which are predictive of the response of AML patients to anticancer drugs.

The expression level of the marker can be determined by quantitatively measuring the mRNA or protein thereof, with preference for mRNA. Examples of techniques, useful in the present invention, for determining the expression level of mRNA include, but are not limited to, DNA chip, RT-PCR, competitive RT-PCR, real-time RT-PCR, RPA (RNase protection assay), and Northern blotting techniques. Preferably, a DNA chip, in which a marker gene or fragments thereof are attached at high density on a

substrate, such as glass, is used for measuring the expression level of mRNA. The DNA chip requires a nucleotide, labeled with, for example, a fluorescent constituent at its terminus or an intermediate site, for hybridization. Preferably, the nucleotide may be cDNA synthesized from mRNA prepared ^' from a sample of interest. Following the hybridization of cDNA on the DNA chip, the expression level can be readily read.

As an alternative, RT-PCR may be used to determine the expression level of mRNA. RT-PCR is a technique in which mRNA, prepared from a sample of interest, is

"reverse" transcribed into cDNA, followed by amplification of the resulting DNA using a polymerase chain reaction

(PCR) . In the amplification step, a pair of primers specific for the marker gene is used. After the electrophoresis of the RT-PCR product thus obtained, the visualized band pattern and thickness give information on the expression level of the marker gene.

From the protein level, the expression of the marker gene can be determined. In this case, the protein of the gene may be quantitatively analyzed with an antibody which specifically binds to the protein. Examples of analytical techniques using antibodies include Western blotting, ELISA (Enzyme Linked Immunosorbent Assay) , RIA (Radioimmunoassay) , radioimmunodiffusion, rocket Immunoelectrophoresis, immunohistochemical stain,

immunoprecipitation assay, complement fixation assay, FACS, protein chip, etc., but are not limited thereto.

In accordance with the present invention, the kit for use in measuring the expression of the marker gene may comprise an additional constitutional element suitable for the analysis technique used. For example, a DNA chip kit may comprise a substrate to which a gene of interest or the cDNA thereof is attached, a reagent for fluorescent labeling, and an enzyme. An RT-PCR kit may comprise a pair of primers specific for the gene. Each of the primers is a nucleotide sequence, specific for the marker gene, having a length from 7 to 50 bp. The RT-PCR kit may comprise a container, a reaction buffer, deoxyribonucleotide triphosphates (dNTPs) , Taq-polymerase, reverse transcriptase, DNAse, or an RNAse inhibitor.

The kit according to the present invention can be applied to anticancer drugs for use in the treatment of AML. In addition, naturally occurring materials and pure chemicals prepared therefrom as well as synthetic organic chemicals for general use in chemical therapy can be analyzed in advance for therapeutic response using the kit of the present invention. The anticancer drugs can be administered in combination. For example, idarubicin (IDA) may be administered in combination with 4N-bihenoyl-l-beta- D-arabino-furanosylcytosine (BHAC) , daunorubicin with BHAC, cytarabine with one selected from among doxorubicin,

daunorubicin, mitoxanthrone and thioguanine, mercaptopurin with methotrexate, mitoxanthrone with etoposide, asparaginase with vincristine, daunorubicin and prednisone, cyclophosphamide with vincristine, cytarabine and prednisone, cyclophosphamide with vincristine and thioguanine, daunorubicin with cytarabine and thioguanine and daunorubicin with vincristine and prednisone. Preferable is a combination of idarubicin with BHAC or a combination of dauborubicin with BHAC. In accordance with another aspect, the present invention provides a method for predicting the response of AML patients to anticancer drugs, comprising the steps of 1) determining the expression level of a marker gene for predicting the response to an anticancer drug in a bio- specimen of an AML patient, the marker gene being selected from a group consisting of GENX-3414, SPINK2, PECAMl, AQP5, LAPTM5, DDX48, F2RL3, SLC3A2, GPR51, ZC3HDC8. WASFl, PSCD4, CDKNlA, AIFl, ADFP, CD164, RAB8A, ID2, ITGB2, MGC3047, MYST3, MT3, MGATl, BC002942, EVI2B, CPVL, MXl, LGALSl, MX2, KRTl, MICA, GABBRl, TINF2, MPV17, RXRA, JARIDlD, C2orf22, DPT, NDUFS2, CCIN, MADH3, NAG, E2F5, SMC6L1, CASPl, LILRBl, TCF4, TNPO3, DNCIl, TIE, ADM, SMUGl, PPPlCA, SALL3, IRF2, KIAA1223, NCOA3, CKS2, ZNF226, CTBS, MetRS, MGC5395, FLJ10349, CD4, EHD4, ARL4A and DNAJClO, and 2) statistically analyzing the expression level of the gene to estimate risk scores R.

The bio-specimen useful for measuring the expression level of the marker gene for predicting the response to anticancer drugs in step 1) may be exemplified by bone marrow, lymph nodes, spleens, blood and lymph fluid, with preference for bone marrow.

Optionally, the method may further comprise the step of conforming the risk score R with reference to a database in which risk scores R of AML patients have been accumulated. The risk score R can be calculated using the HR (Hazard ratio) or β coefficient derived from Cox's proportional hazard modeling. As used herein, the term ^λ HR' means the increment of the risk score R when a coefficient increases by 1, and corresponds to e ^p . HR values of 67 genes, selected at significance p=0.001, are given in Table 1, below.

The risk score R of a patient can be calculated using the following mathematical formula 1, which expresses the sum of the individual products of the extent of expression of each gene and the coefficient β for each patient.

[Mathematics Formula l]

Wherein R stands for risk score, β± for the beta coefficient of gene i, and G± for the extent of expression of gene i.

In Table 1, HR (Hazard ratio) values of marker genes

for predicting the response to anticancer drugs, obtained from Cox's proportional hazard modeling, are summarized.

[TABLE Il

a: four genes selected in duplicate are represented licate, b: Hazard ratio, the increasing hazard when

gene expression is increased two-fold. An HR of 2 means that the hazard level doubles when gene expression is increased two-fold. c: mean deviation of gene expression, d: GenBank accession No. e: UniGene Cluster ID. A database of the risks scores (R) of AML patients is suggested in Table 4, below.

In the method for predicting the response to anticancer drugs in accordance with the present invention, both the method for measuring the expression rate of a marker gene and the anticancer drugs described above can be used.

A better understanding of the present invention may be obtained in light of the following examples which are set forth to illustrate, but are not to be construed to limit the present invention.

[Examples For the Invention ]

EXAMPLE 1: Assay for Gene Expression in the Bone Marrow Sample from AML Patients Using DNA Chip

<1-1> RNA isolation from marrow cells

RNA was isolated from bone marrow samples of 55 AML patients. For this, first, 1 ml of a bone marrow sample was mixed with 5 ml of a TriZol reagent (InVitrogen, Cat. No. 15596-018), followed by rupturing the cells with a tissue homogenizer. The subsequent RNA isolation procedure was

conducted according to the protocol provided by the manufacturer of the TriZol reagent. The RNA thus obtained was further purified using an RNeasy kit (Qiagen, Cat. No. 74106) according to the protocol provided by the manufacturer thereof.

<l-2> Quantitative analysis of isolated RNA The RNA was quantified by measuring absorbance at 260 nm using a spectrophotometer.

<l-3> Preparation of reference RNA

Together with the RNA isolated from marrow cells, reference RNA was also used for hybridization on a DNA chip. The reference RNA was isolated from cell lines based on blood cells. In detail, the seven cell lines HL-60, K-

562, CCRF-CEM, CCRF-HSB-2, CEM-CM3, Molt-4, and THP-I were used to prepare RNA in the same manner as in Example <1-1>.

After being quantified as in Example <l-2>, the isolated

RNA fragments were mixed in equal amounts to give the reference RNA.

<l-4> DNA Chip

The DNA chip for use in the assay was a 16K human cDNA chip comprising 15,972 cDNA probes (Digitalgenomics, Korea). The DNA chip was prepared as follows. First, a plasmid carrying the cDNA was isolated from a stock of

bacteria, and probe sequences were amplified from the plasmid using PCR. The cDNA thus obtained was purified with a PCR Clean-Up kit and dissolved at a concentration from 100 to 200 ng/μl in a spotting solution containing 50% DMSO. After the cDNA solution was spotted on GAPS II slides

(Corning, Cat. No. 40006), irradiation with a suitable dose of UV light fixed the cDNA thereon.

<l-5> DNA chip assay and quantification of gene expression

The RNA isolated from 10 μg of a marrow sample and the reference RNA were reverse transcribed into cDNA in the presence of aminoallyl-dUTP, and the cDNA was then labeled with Cy5 and Cy3 through reaction respectively with monoester-Cy5 and monoester-Cy3. The labeled samples were purified using a PCR Clean-Up kit and subjected to hybridization for 16 hours or longer on the DNA chip. Then, the DNA chip was washed with a washing solution of SSC so as to remove non-specific hybrids. The washed DNA chip was scanned with a confocal laser scanner (Perkin Elmer, Scanarray Lite) to collect fluorescence data from each spot. The data were stored as TIFF files which were quantitatively analyzed using GenePix 3.0 (Axon Instruments) . The fluorescence value at each spot, quantified using GenePix 3.0, was corrected with the help of the ' lowess ' function set in an S-plus statistical

package (Insightful) according to the Yang method (Nucleic Acids Res 2002, 30:el5).

From the HR calculated by applying the gene expression level having the corrected value to Cox' s proportional hazard model, 67 marker genes capable of predicting the response of AML patients to anticancer drugs were selected. The results are given in Table 1.

EXAMPLE 2: Data about Gene Expression of AML Patient and Response to Anticancer Drugs

<2-l> Gene expression results of patients tested Bone marrow samples were taken from 55 AML patients and assayed for the expression levels of the 67 marker genes listed in Table 1, using the DNA chip in the same manner as in Example 1.

<2-2> Clinical information on patients to be tested For remission induction chemotherapy, some of the 55 AML patients were administered with BHAC for 7 or 10 days and with idarubicin for 3 or 5 days. The other patients were subjected to remission induction chemotherapy through the administration of BHAC for 3 or 5 days and daunorubicin for 3 or 5 days. Significant clinical information on the patients is summarized in Table 2, below.

[TABLE 2] Clinical Information on Patients

a: F stands for female and M for male. b: 1 was marked for the patients whose conditions could not be monitored for reasons other than the disease or who did not show disease recurrence before the final monitoring time, and 0 for the patients who suffered from disease recurrence. c: The period from anticancer therapy to recurrence is indicated as months, d: subtype of AML

according to French-American-British classification.

<2-3> Calculation of the risk score of patients

A bone marrow samples taken from the AML55 patients were measured for the expression level of each of the marker genes for predicting the response to anticancer drugs. The risk score R of the AML 55 patients was calculated using the hazard ratio or β coefficient obtained from the Cox' s proportional hazard model . The term "hazard ratio" has the same value as e ^p and refers to the increase proportion of risk when a variable is increased by 1. The hazard score R of a patient was calculated according to the following mathematic formula 1,

The risk score R of a patient can be calculated using the following mathematical formula 1, which expresses the sum of the individual products of the extent of expression of each gene and the coefficient β .

[Mathematics Formula l] R = σ βiGi Wherein R stands for risk score, βι for the beta coefficient of gene i, and G± for the extent of expression of gene i .

Calculated risk scores of the AML 55 patient are given in Table 3. In Table 3, the risk score of the AML 55 patients is found to be -23.29, as calculated using 67 marker genes for predicting response to anticancer drugs.

From the expression levels of the 67 marker genes, obtained with respect to the AML55 patients, the risk scores were calculated, and the results are listed in Table 3.

[TABLE 3]

EXAMPLE 3: Assay for Prediction of Response to Anticancer Drugs Using the Expression of Marker Gene

<3-l> Significance Testing in prediction of response to anticancer therapy

Leave-one-out cross-validation was conducted to predict the response of AML patients to the anticancer drugs using the expression of the marker genes selected according to the method described in Example 2, and to evaluate the prediction.

Leave-one-out cross-validation is a statistical technique in which after the data of one patient is excluded from the total data set and marker genes are selected, the response of the patient to anticancer drugs can be predicted using the expression of the marker genes in the patient. The repetition of this procedure as many times as the total number of patients allows an independent prediction to be made for each individual patient. Using the risk scores obtained from the independent prediction, survival curves are compared between a high risk patient group and a low risk patient group such that the significance of prediction for the response of individual patients to anticancer drugs can be determined.

<3-2> Prediction of response to anticancer drugs using the marker genes selected at p=0.001

To examine whether the response of AML patients to anticancer drugs could be significantly predicted at p=0.001 in the same manner as in Example 2, ^λ leave-one-out' cross-validation was conducted.

Of the total of 55 patients samples, one sample was used as the validation data while the remaining samples were used as the training data for gene selection. The gene selection was repeated 55 times, such that each patient sample is used once as the validation data. Using the expression of the selected gene, the risk score of the one excluded patient was calculated. Risk score data and censored data are given in Table 4, below.

[TABLE 4] Independently Calculated Risk Scores of Individual Patients

Using the calculated risk score, the patients can be classified into a high risk group and a low risk group. However, it may be somewhat unreasonable to use the risk scores alone as a criterion. In order to determine how the two groups could be divided with the best significance, a Kaplan-Meier plot, which provides recurrence curves of the two groups, was generated while the number of the members in the low risk group was increased by one in order of increasing risk. A log rank test was performed to examine whether there was a significant difference between the recurrence curves of the two groups .

FIG. 1 is a graph showing p values obtained in the log rank test. When 33 patients are subject to a low risk group with the remaining 22 patients in a high risk group

on the basis of this analysis, the Kaplan-Meier plot for the two groups exhibits the most significant difference (p=0.0002) . The Kaplan-Meier plot generated at this time is shown in FIG. 2. These results demonstrate that when patients are divided into risk groups according to the risk score based on the expression of marker genes, the high risk group and the low risk group are predicted significantly (p=0.0002).

<3-3> Prediction results of response to anticancer drug depending on significance for gene selection

Genes selected at p=0.001 were used to predict the response to anticancer drugs in Example 3-2. When a different p value is selected (that is, the significance is changed) , the number of the genes selected is changed. If the p value is reduced, a lower number of genes is selected. A higher p value results in the selection of a larger number of genes . In order to determine how many genes are suitable for the prediction of response to anticancer therapy, a test for the prediction validation of response to anticancer therapy was conducted at p=0.001, 0.0005, 0.0001, and 0.00001. The results are summarized in Table 5, below.

[TABLE 5l

Results of Prediction of Response to Anticancer Therapy at Different p Values for Gene Selection

As seen in Table 5, when the p value was decreased upon gene selection, that is, when genes were selected at a higher significance, the Kaplan-Meier plot showed a larger difference between the two risk groups. Namely, although all of the genes selected at p=0.001 are useful for predicting the response to anticancer therapy, such prediction is possible only with parts of them.

[industrial Applicability]

Marker genes useful in predicting the response of AML patients to anticancer drugs can be selected in accordance with the present invention. Accordingly, when biological samples taken from AML patients are applied to a kit comprising an agent for measuring the expression of the marker genes for predicting the response of AML patients to anticancer drugs, the expression patterns of the genes can be analyzed so as to predict the response of the AML

patients to anticancer drugs .

In consequence, the present invention can give information on suitable chemotherapy for AML patients, thereby improving the therapeutic effect of the anticancer drugs used and relieving the pain and economic burden of the patients.