Patent Application
20060252056
This application is a continuation of International Application No. PCT/GB2004/002316, which was filed on Jun. 1, 2004, which designated the United States and was published in English, and which claims the benefit of United Kingdom applications GB0312451.8, filed 30 May 2003, GB0322636.2, filed Sep. 26, 2003, and GB0327132.7, filed Nov. 21, 2003. The above mentioned references are incorporated herein by reference.
The present invention relates to a method of personalized cancer therapy which employs a set of marker genes to predict whether a patient will respond to a chemotherapeutic agent and a kit for use in said method.
In particular, the method predicts patient response to erbB tyrosine kinase inhibitors. More particularly the method relates to those patients with cancers mediated alone or in part by erbB tyrosine kinase, especially patients with advanced Non-small Cell Lung Cancer (NSCLC), for example adenocarcinoma, using the levels of a set of marker genes having differential expression between responders and non responders to the erbB tyrosine kinase inhibitor.
Lung cancer is the leading cause of cancer death and is therefore a major health problem worldwide. In the treatment of this disease, chemotherapy is the mainstay, because the majority has locally advanced stage 3 (44%) or metastatic stage 4 (32%) disease at diagnosis [1]. Nevertheless, the findings of large meta-analysis revealed that platinum-based chemotherapy contributed to prolong the median survival time of patients with advanced non-small cell lung cancer (NSCLC) by only about 6 weeks compared with best supportive care [2].
In the last decade, many new cytotoxic agents have been developed including paclitaxel, docetaxel, gemcitabine, and vinorelbine, and have offered multiple choices for patients with advanced lung cancer. However, each regimen served only modest survival benefit compared with the cisplatin-based therapies [3], [4]. More recently, new therapeutic strategies including a number of molecular-targeted agents have been developed in an effort to overcome the limitations of conventional cytotoxic agents [5] [6].
In recent years it has been discovered that certain growth factor tyrosine kinase enzymes are important in the transmission of biochemical signals which initiate cell replication. They are large proteins which span the cell membrane and possess an extracellular binding domain for growth factors such as epidermal growth factor (EGF) and an intracellular portion which functions as a kinase to phosphorylate tyrosine amino acids in proteins and hence to influence cell proliferation.
Various classes of receptor tyrosine kinases are known (Wilks, Advances in Cancer Research, 1993, 60, 43-73) based on families of growth factors which bind to different receptor tyrosine kinases. The classification includes Class I receptor tyrosine kinases comprising the EGF family of receptor tyrosine kinases. This includes receptors for the ligands EGF, TGFα (also referred to as TGFA), amphiregulin (also referred to as AREG), betacellulin, heparin binding EGF, epiregulin and the neuregulins (including NRG-1, NRG-2, NRG-3 and NRG-4). More specifically, these receptors include those with a functional kinase domain called erbB I (EGFR), erbB2 (Neu, Her2) and erbB4 (Her 4), and erbB3 (her3), which does not), Class II receptor tyrosine kinases comprising the insulin family of receptor tyrosine kinases such as the insulin and IGFI receptors and insulin-related receptor (IRR) and Class III receptor tyrosine kinases comprising the platelet-derived growth factor (PDGF) family of receptor tyrosine kinases such as the PDGFα, PDGFβ and colony-stimulating factor 1 (CSF1) receptors.
It is known that the erbB family of receptor tyrosine kinases, which include EGFR, erbB2, erbB3 and erbB4, are frequently involved in driving the proliferation and survival of tumour cells (reviewed in Olayioye et al., EMBO J., 2000, 19, 3159). One mechanism by which this can occur is over expression of the receptor at the protein level, generally as a result of gene amplification. This has been observed in many common human cancers (reviewed in Klapper et at., Adv. Cancer Res., 2000, 77, 25) such as, non-small cell lung cancers (NSCLCs) including adenocarcinomas (Cerny et al., Brit. J. Cancer, 1986, 54, 265; Reubi et al, Int. J. Cancer 1990, 45, 269; Rusch et al, Cancer Research, 1993, 53, 2379; Brabender et al, Clin. Cancer Res., 2001, 7, 1850) as well as other cancers of the lung (Hendler et al., Cancer Cells, 1989, 7, 347.
As a consequence of the mis-regulation of one or more of these receptors, it is widely believed that many tumours become clinically more aggressive and so correlate with a poorer prognosis for the patient (Brabender et al, Clin. Cancer Res., 2001, 7, 1850; Ross et al, Cancer Investigation, 2001, 19, 554, Yu et al., Bioessays, 2000, 22.7, 673). In addition to these clinical findings, a wealth of pre-clinical information suggests that the erbB family of receptor tyrosine kinases are involved in cellular transformation. In addition to this, a number of pre-clinical studies have demonstrated that anti-proliferative effects can be induced by knocking out one or more erbB activities by small molecule inhibitors, dominant negatives or inhibitory antibodies (reviewed in Mendelsohn et al., Oncogene, 2000, 19, 6550).
Thus it has been recognised that inhibitors of these receptor tyrosine kinases should be of value as a selective inhibitor of the proliferation of mammalian cancer cells (Yaish et al. Science, 1988, 242, 933, Kolibaba et al, Biochimica et Biophysica Acta, 1997, 133, F217-F248; Al-Obeidi et al, 2000, Oncogene, 19, 5690-5701; Mendelsohn et al, 2000, Oncogene, 19, 6550-6565). In addition to this pre-clinical data, findings using inhibitory antibodies against EGFR and erbB2 (c-225 and trastuzumab respectively) have proven to be beneficial in the clinic for the treatment of selected solid tumours (reviewed in Mendelsohn et al, 2000, Oncogene, 19, 6550-6565).
A number of small molecule inhibitors of erbB family of receptor tyrosine kinases are known, particularly inhibitors of EGF and erbB2 receptor tyrosine kinases. For example European Patent Application No. 0566226 and International Patent Applications WO 96/33980 and WO 97/30034 disclose that certain quinazoline derivatives which possess an anilino substituent at the 4-position possess EGFR tyrosine kinase inhibitory activity and are inhibitors of the proliferation of cancer tissue including prostate cancer. It has been disclosed by J R Woodburn et al. in Proc. Amer. Assoc. Cancer Research, 1997, 38, 633 and Pharmacol. Ther. 1999, 82, 241-250 that the compound N-(3-chloro-4-fluorophenyl)-7-methoxy-6-(3-morpholinopropoxy)quinazolin-4-amine is a potent EGFR tyrosine kinase inhibitor. This compound is also known as Iressa (registered trade mark), gefitinib (United States Adopted Name), by way of the code number ZD1839 and Chemical Abstracts Registry Number 184475-35-2. The compound is identified hereinafter as gefitinib. Gefitinib has recently been approved in Japan for the treatment of inoperable or recurrent non-small cell lung cancer (NSCLC) and in the USA as a monotherapy for the treatment of patients with locally advanced metastatic NSCLC after failure of both platinum and docetaxel chemotherapies.
It is further known from International Patent Application WO 96/30347 that certain structurally-related quinazoline derivatives possessing an anilino substituent at the 4-position also possess EGFR tyrosine kinase inhibitory activity. It has been disclosed in WO 99/55683 that the compound N-(3-ethynylphenyl)-6,7-bis(2-methoxyethoxy)quinazolin-4-amine, or a pharmaceutically-acceptable salt thereof (linked to the code numbers CP 358774 and OSI-774, identified hereinafter by the code number OSI-774) is an EGFR TKI.
It is further known from International Patent Application WO 97/38983 that certain other structurally-related quinazoline derivatives possessing an anilino substituent at the 4-position also possess EGFR tyrosine kinase inhibitory activity. It has been disclosed in J. Med. Chem., 1999, 42,1803-1815 and WO 00/31048 that the compound 6-acrylamido-N-(3-chloro-4-fluorophenyl)-7-(3 morpholinopropoxy)quinazolin-4-amine (linked to the code numbers PD 183805 and CI 1033, identified hereinafter by the code number CI 1033) is an EGFR TKI.
It is further known from International Patent Application WO 97/02266 that certain other structurally-related heterocyclic derivatives also possess EGFR tyrosine kinase inhibitory activity. For example, the compound 4-[(1R)-1-phenylethylamino]-6-(4-hydroxyphenyl)-7H-pyrrolo[2,3-d]pyrimidine (linked to the code numbers PKI-166, CGP 75166 and CGP 59326, identified hereinafter by the code number PKI-166) is an EGFR TKI.
It is further known from European Patent Application No. 0787722 and International Patent Applications WO 98/50038, WO 99/09016 and WO 99/24037 that certain other structurally-related quinazoline derivatives possessing an anilino substituent at the 4-position also possess EGFR tyrosine kinase inhibitory activity. For example, the compound N-[4-(3-bromoanilino)quinazolin-6-yl]but-2-ynamide (linked to the code numbers CL-387785 and EKB-785, identified hereinafter by the code number CL-387785) is an EGFR TKI.
It is further known from Nature Medicine, 2000, 6, 1024-1028 and U.S. Pat. No. 6,002,008 that certain other structurally-related quinoline derivatives possessing an anilino substituent at the 4-position also possess EGFR tyrosine kinase inhibitory activity. For example, the compound 4-(3-chloro-4-fluoroanilino)-3-cyano-6-(4-dimethylaminobut-2(E)-enamido)-7-ethoxyquinoline (identified hereinafter by the code number EKB-569) is an EGFR TKI.
It is also known from WO 99/35146 and WO 01/04111 that certain other quinazoline derivatives are inhibitors of one or more of the erbB receptor tyrosine kinase inhibitors. For example the compound N-{3-chloro-4-[(3-fluorobenzyl)oxy]phenyl}-6-[5-({[2-(methylsulfonyl)ethyl]amino}methyl)-2-furyl]quinazolin-4-amine (also identified as lapatinib or GW2016 identified hereinafter by the code GW2016) is thought to be an inhibitor of both EGF and erbB2 receptor tyrosine kinases. Novartis AE788 is another suitable inhibitor compound.
Inhibition of erbB receptor tyrosine kinase may also be achieved by inhibition of the extracellular ligand binding to a receptor using suitable antibodies against an erbB receptor. For example using the anti-erbB2 antibody trastuzumab [Herceptin™] and the anti-erbb1 antibody cetuximab [C225]). The use of such inhibitory antibodies have proven to be beneficial in the clinic for the treatment of selected solid tumours (reviewed in Mendelsohn et al, 2000, Oncogene, 19, 6550-6565).
As mentioned above, gefitinib is an oral active inhibitor of epidermal growth factor receptor-tyrosine kinase (EGFR-TK), which blocks signalling pathways responsible for driving proliferation, invasion, and survival of cancer cells [7]. Potent anti-tumour effects as well as rapid improvements in NSCLC-related symptoms and quality of life have been observed in clinical studies that enrolled patients with advanced NSCLC who did not respond to platinum-based chemotherapy. In the randomized double-blind phase II monotherapy trial (the IDEAL 1 trial), use of gefitinib as 2nd or 3rd line of chemotherapy to advanced NSCLC achieved tumour response rate of 18.4% (95% CI: 11.0-25.9%), and in the IDEAL 2 trial, use as 3rd or 4th line of chemotherapy achieved that of 11.8% (95% CI: 6.2-19.7%) [8],[27],[28].
Moreover in these trials, the treatment of this drug achieved high disease control rate (54.4% in IDEAL 1, 42.2% in IDEAL 2) and overall symptom improvement rate (40.3% in IDEAL 1, 43.1% in IDEAL 2).
Those results were promising when compared with responses to conventional cytotoxic agents, but the fact remained that about half of the patients enrolled in these studies received non-effective treatment with no improvement in symptoms. Moreover, the medication exposed non-responders to adverse effects, including life threatening ones such as interstitial pneumonia [11].
Patients responses to the various chemotherapy treatments differ, therefore there is a need to find methods of predicting which treatment regimes best suit a particular patient.
There is an increasing body of evidence that suggests that patients responses to numerous drugs may be related to a patients genetic profile and that determination of the genetic factors that influence, for example, response to a particular drug could be used to provide a patient with a personalised treatment regime. Such personalised treatment regimes offer the potential to maximise therapeutic benefit to the patient, whilst minimising, for example side effects that may be associated with alternative and less effective treatment regimes. There is therefore a need for methods that can predict a patients response to a drug.
It has been found that the sensitivity of certain cancers to chemotherapeutic agents can be predicted by gene expression and hence that the suitability of cancer patients for treatment with such chemotherapeutic agents can be determined by measuring the relative levels of particular genes in patient tissue.
Accordingly, the present invention provides an isolated set of marker genes comprising at least one gene identified as having differential expression as between patients who are responders and non responders to an erbB receptor tyrosine kinase inhibitor, said gene set comprising one or more genes selected from at least the group consisting of the 51 genes listed in Table 4 herein including gene-specific oligonucleotides derived from said genes. In Table 4, accession numbers are given for the genes on the GenBank database.
Sequences of these genes are described in Table 4a and Table 4b. As will be appreciated by those skilled in the art, sequences available at the given accession numbers represent only examples of sequences of the genes referred to in the table; alternative sequences, including sequences which comprise sequencing error corrections, allelic or other variations, splice mutants and the like are also included in the definition of the gene represented by the name used. In a most preferred embodiment, the sequences referred to are the sequences set forth at the accession numbers and specific sequences given and set out in detail in Table 4a.
In a further aspect the present invention provides a set of isolated marker genes comprising at least one gene identified as having differential expression as between patients who are responders and non responders to an erbB receptor tyrosine kinase inhibitor; said gene set selected from the group consisting of the 51 genes listed in Table 4 herein including gene-specific oligonucleotides derived from said genes.
The present invention permits the improved prognosis and hence quality of life of cancer patients by matching the treatments to individual patients and so making more effective use of the types of drug available.
A preferred set is at least one or more of the first 40 genes listed in Table 4 herein.
A further preferred set is at least one or more of the first 20 genes listed in Table 4 herein.
A further preferred set is at least one or more of the first 12 genes listed in Table 4 herein.
A preferred set is at least one or more of the first 5 genes listed in Table 4 herein.
An especially preferred set is the first 12 genes listed in Table 4a herein, namely FLJ22622 (e.g. GenBank NM—024829), AREG (e.g. GenBank BC009799), C0R01C (e.g. GenBank NM—014325), AVEN (e.g. GenBank BC010488), DUSP3 (e.g. GenBank NM—004090, DJ473B4 (e.g. GenBank AI026836), PHLDA2 (e.g. GenBank BU500509), RBM7 (e.g. GenBank NM—0106090), EST (GenBank BX0952512), OSMR (e.g. GenBank AI436027), GCLC (e.g. GenBank AI971137), COL4A3BP (e.g. GenBank BQ024877).
Preferably the inhibitor is selected from gefitinib, OSI-774, PKI-166, EKB-569, GW2016, CI-1033 and an anti-erbB antibody such as trastuzumab and cetuximab.
Most preferably the inhibitor is gefitinib.
The present invention is particularly suitable for use in predicting the response to the aforementioned chemotherapeutic agents in those patients or patient population with a cancer mediated alone, or in part, by an erbB tyrosine kinase. Such cancers include, for example, non-solid tumours such as leukaemia, multiple myeloma or lymphoma, and also solid tumours, for example bile duct, bone, bladder, brain/CNS, breast, colorectal, cervical, endometrial, gastric, head and neck, hepatic, lung, muscle, neuronal, oesophageal, ovarian, pancreatic, pleural/peritoneal membranes, prostate, renal, skin, testicular, thyroid, uterine and vulval tumours.
The present invention is particularly suitable for identifying those patients with NSCLC, more particularly advanced NSCLC including advanced adenocarcinoma that will respond to treatment with chemotherapeutic agents such as an erbB receptor tyrosine kinase inhibitor as hereinbefore defined.
The present invention offers considerable advantages in the treatment of cancers such as NSCLC, especially advanced NSCLC by identifying “individual cancer profiles” of NSCLC and so determining which tumours would respond to gefitinib. This includes 1st line treatment and any other treatment regimen, such as, for example chemotherapy failed patients.
The present invention is particularly useful in the treatment of patients with advanced NSCLC who have failed previous chemotherapy, such as platinum-based chemotherapy.
The present invention is also particularly useful in the treatment of patients with locally advanced (stage lIIB) or metastasized (stage IV) NSCLC who have received previous chemotherapy, such as platinum-based chemotherapy.
The present invention also provides a method of predicting the responsiveness of a patient or patient population with cancer-, for example lung cancer, to treatment with chemotherapeutic agents, especially erbB receptor tyrosine kinase inhibitors, comprising comparing the differential expression of a set of marker genes said marker genes selected from the gene sets as defined above.
Preferably the assessment of expression is performed by gene expression profiling using oligonucleotide-based arrays or cDNA-based arrays of any type; RT-PCR (reverse transcription—Polymerase Chain Reaction), real-time PCR, in-situ hybridisation, Northern blotting, Serial analysis of gene expression (SAGE) for example as described by Velculescu et al Science 270 (5235): 484-487, or differential display. Details of these and other methods can be found for example in Sambrook et al, 1989, Molecular Cloning: A Laboratory Manual). Preferably the assessment uses a microarray assay.
Alternatively, or in addition, the assessment uses an immunohistochemical assay.
In a further aspect, the present invention provides a kit for use in a method of predicting the responsiveness of a patient or patient population with cancer, to treatment with chemotherapeutic agents, especially erbB receptor tyrosine kinase inhibitors, comprising a marker gene set as defined above on a suitable support medium. Preferably the marker gene is attached to a support material or membrane such as nitrocellulose, or nylon or a plastic film or slide.
Preferably the kit comprises a microarray.
A. Different prediction scores appear when the number of discriminating genes is changed. The number of the discriminating gene sets (from 5 to 51) corresponds to the number of selected genes from the top of the rank-ordered list in table 4. A larger value of classification score (CS) indicates better separation of the two groups.
B. Hierarchical clustering of 17 “learning” cases using 51 candidate genes for gefitinib-sensitivity (left), and 12 prediction genes that were finally selected for the GRS (right). The dendrograms represent similarities in expression patterns among individual cases; longer branches indicate greater differences. The two groups were most clearly separated by the 12-gene set.
C. Schematic distinction of responder, non-responder and “test cases” verified on the basis of the GRS. Red diamonds denote prediction scores for learning PR cases and blue diamonds represent learning PD cases. A pink triangle indicates a test PR case that had not been used for establishing GRS, and blue triangles indicate test PD cases. Yellow triangles indicate test SD cases that kept the SD status throughout the 4-month observation period, and green triangles indicate test cases once judged as SD at a certain-time point of the study but showed progression of the disease within three or four months after the start of treatment.
A. Representative image of semi-quantitative RT-PCR analysis of RNAs from the PR and PD groups. OSMR and GCLC genes were over-expressed in non-responders (PD). The integrity of each cDNA template was controlled through amplification of ACTB.
B. Immunohistochemical staining of representative samples from fiberscopic transbronchial biopsy (TBB) and lymph-node (LN) biopsy from the same PD-patient (No. LC21), using anti-AREG antibody (X 200).
C. Immunohistochemical staining of representative samples from PD patients, using antibodies for other 4 prediction markers (TGFA, ADAM9, CD9, and OSMR) (X200).
A. Expression of AREG transcript examined by semi-quantitative RT-PCR in lung-adenocarcinoma cell lines PC-9, NCI-H358, and -H522.
B. PC-9 cells cultured in medium supplemented with 10% FCS, in serum-free medium, or in serum-free conditioned medium (CM) obtained from cultures of NCI-H358 or -H522 cells. Each medium was replaced once with the same medium at the 48-hour time point; 72 hours after adding gefitinib at concentrations of 0·5 or 1·0 μM, cell viability was measured by MTT assays. The experiments were done in triplicate. The Y-axis indicates the relative MTT value (MTT in the presence of 0·or 1·0 μM gefitinib/MTT in the absence of gefitinib) of the cells incubated in different media.
C. Effect of AREG, secreted in an autocrine manner, on the resistance of NSCLC cells to gefitinib. At the start of culture, PC-9 cells were inoculated into medium containing 1·0 μM gefitinib and recombinant AREG protein (final concentrations of 1-100 ng/ml); 72 hours later, cell viability was measured by triplicate MTT assays (blue bars). The Y-axis indicates the relative MTT values (MTT at individual concentrations of AREG/MTT without AREG) of the cells.
Effect of AREG on the viability of NSCLC cells in the absence of 1·0 μM gefitinib was also studied. Individual PC-9 cells were added to medium containing recombinant AREG protein but no gefitinib; 72 hours later, viability was measured by triplicate MTT assays (red bars).
The invention will be described in more detail and illustrated by the following examples which are meant to serve to assist one of ordinary skill in the art in carrying out the invention and are not intended in any way to limit the scope of the invention. Certain elements of the invention are also described in more detail below.
“Set of Isolated Marker Genes”
These are, according to the context of the embodiments described herein, a group of genes which can be used in classification or categorisation of patent response according to the invention.
“Differential Expression”
Genes that are either expressed at a higher or lower level as between groups of responders or nonresponders.
“Responders/Non Responders”
Objective tumour responses according to Union International Contre le Cancer/World Health Organization (U ICC/WHO) Criteria are categorised as follows: complete response (CR): no residual tumour in all evaluable lesions; partial response (PR): residual tumour with evidence of chemotherapy-induced 50% or greater decrease under baseline in the sum of all measurable lesions and no new lesions; stable disease (SD) residual tumour not qualified for CR; and progressive disease (PD): residual tumour with evidence of 25% or greater increase under baseline in the sum of all measurable lesions or appearance of new lesions. As defined herein, non responders are PD.
The present invention is particularly effective for determining those patients which are CR or PR
“ErbB Receptor Inhibitors Including, Without Limitation, ErbB Receptor Tyrosine Kinase Inhibitors”
This family includes EGF, erbB2 (HER), erbB3 (note that erbB3 does not have a functional kinase domain) and erbB4 as described in the background to the invention above.
“Gene-Specific Oligonucleotides”
These are intended to be unique to the respective genes so that, for example, fragments of the gene that uniquely identify the gene. Advantageously, a gene-specific oligonucleotide is between 5 and 50 nucleotides in length, preferably about 15 to 30 nucleotides, and most preferably about 23 nucleotides.
“Arrays or Microarrays”
Array technology and the various techniques and applications associated with it are described generally in numerous textbooks and documents. Gene array technology is particularly suited to the practice of the present invention. Methods for preparing microarrays are well known in the art. These include Lemieux et al., (1998), Molecular Breeding 4, 277-289, Schena and Davis. Parallel Analysis with Biological Chips. in PCR Methods Manual (eds. M. Innis, D. Gelfand, J. Sninsky), Schena and Davis, (1999), Genes, Genomes and Chips. In DNA Microarrays: A Practical Approach (ed. M. Schena), Oxford University Press, Oxford, UK, 1999), The Chipping Forecast (Nature Genetics special issue; January 1999 Supplement), Mark Schena (Ed.), Microarray Biochip Technology, (Eaton Publishing Company), Cortes, 2000, The Scientist 14[17]:25, Gwynne and Page, Microarray analysis: the next revolution in molecular biology, Science, Aug. 6, 1999; and Eakins and Chu, 1999, Trends in Biotechnology, 17, 217-218.
The technology is described in PCT/US01/10063 and US 2002 090979 and references therein.
Commercial suppliers include Affymetrix (California) and Clontech Laboratories (California).
Major applications for array technology include the identification of sequence (nucleotide sequence/nucleotide sequence mutation) and the determination of expression level (abundance) of nucleotide sequences. Gene expression profiling may make use of array technology, optionally in combination with proteomics techniques (Celis et al, 2000, FEBS Lett, 480(1):2-16; Lockhart and Winzeler, 2000, Nature 405(6788):827-836; Khan et al., 1999, 20(2):223-9). Other applications of array technology are also known in the art; for example, nucleotide sequence discovery, cancer research (Marx, 2000, Science 289: 1670-1672; Scherf, et al, 2000, Nat Genet;24(3):236-44; Ross et al, 2000, Nat Genet. March 2000; 24(3):227-35), SNP analysis (Wang et al, 1998, Science, 280(5366):1077-82), drug discovery, pharmacogenomics, disease diagnosis (for example, utilising microfluidics devices: Chemical & Engineering News, Feb. 22, 1999, 77(8):27-36), toxicology (Rockett and Dix (2000), Xenobiotica, 30(2):155-77; Afshari et al., 1999, Cancer Res1;59(19):4759-60) and toxicogenomics (a hybrid of functional genomics and molecular toxicology). The goal of toxicogenomics is to find correlations between toxic responses to toxicants and changes in the nucleotide sequencetic profiles of the objects exposed to such toxicants (Nuwaysir, et al (1999), Molecular Carcinonucleotide sequencesis, 24:153-159).
In general, any library may be arranged in an orderly manner into an array, by spatially separating the members of the library. Examples of suitable libraries for arraying include nucleic acid libraries (including DNA, nucleotide sequence, oligonucleotide, etc libraries), peptide, polypeptide and protein libraries, as well as libraries comprising any molecules, such as ligand libraries, among others. Accordingly, where reference is made to a “library” such reference includes reference to a library in the form of an array.
The members of a library are generally fixed or immobilised onto a solid phase, preferably a solid substrate, to limit diffusion and admixing of the samples. In particular, the libraries may be immobilised to a substantially planar solid phase, including membranes and non-porous substrates such as plastic and glass. Furthermore, the samples are preferably arranged in such a way that indexing (i.e. reference or access to a particular sample) is facilitated. Typically the samples are applied as spots in a grid formation. Common assay systems may be adapted for this purpose. For example, an array may be immobilised on the surface of a microplate, either with multiple samples in a well, or with a single sample in each well. Furthermore, the solid substrate may be a membrane, such as a nitrocellulose or nylon membrane (for example, membranes used in blotting experiments). Alternative substrates include glass, or silica based substrates. Thus, the samples are immobilised by any suitable method known in the art, for example, by charge interactions, or by chemical coupling to the walls or bottom of the wells, or the surface of the membrane. Other means of arranging and fixing may be used, for example, pipetting, drop-touch, piezoelectric means, ink-jet and bubblejet technology, electrostatic application, etc. In the case of silicon-based chips, photolithography may be utilised to arrange and fix the samples on the chip. The samples may be arranged by being “spotted” onto the solid substrate; this may be done by hand or by making use of robotics to deposit the sample. In general, arrays may be described as macroarrays or microarrays, the difference being the size of the sample spots. Macroarrays typically contain sample spot sizes of about 300 microns or larger and may be easily imaged by existing gel and blot scanners. The sample spot sizes in microarrays are typically less than 200 microns in diameter and these arrays usually contain thousands of spots. Thus, microarrays may require specialised robotics and imaging equipment, which may need to be custom made. Instrumentation is described generally in a review by Cortese, 2000, The Scientist 14[11]:26.
Techniques for producing immobilised libraries of DNA molecules have been described in the art. Generally, most prior art methods describe how to prepare single-stranded nucleic acid molecule libraries, using for example masking techniques to build up various permutations of sequences at the various discrete positions on the solid substrate. U.S. Pat. No. 5,837,832 describes an improved method for producing DNA arrays immobilised to silicon substrates based on very large scale integration technology. In particular, U.S. Pat. No. 5,837,832 describes a strategy called “tiling” to prepare specific sets of probes at spatially-defined locations on a substrate which may be used to produced the immobilised DNA libraries of the present invention. U.S. Pat. No. 5,837,832 also provides references for earlier techniques that may also be used.
To aid detection, targets and probes may be labelled with any readily detectable reporter such as a fluorescent, bioluminescent, phosphorescent, radioactive reporter. Labelling of probes and targets is disclosed in Shalon et al., 1996, Genome Res 6(7):639-45.
The materials for use in the methods of the present invention are ideally suited for preparation of kits. A set of instructions will typically be included.
General Recombinant DNA Methodology Techniques
The present invention employs, unless otherwise indicated, conventional techniques of chemistry, molecular biology, microbiology, recombinant DNA and immunology, which are within the capabilities of a person of ordinary skill in the art. Such techniques are explained in the literature. See, for example, J. Sambrook, E. F. Fritsch, and T. Maniatis, 1989, Molecular Cloning: A Laboratory Manual, Second Edition, Books 1-3, Cold Spring Harbor Laboratory Press; Ausubel, F. M. et al. (1995 and periodic supplements; Current Protocols in Molecular Biology, ch. 9, 13, and 16, John Wiley & Sons, New York, N.Y.); B. Roe, J. Crabtree, and A. Kahn, 1996, DNA Isolation and Sequencing: Essential Techniques, John Wiley & Sons; M. J. Gait (Editor), 1984, Oligonucleotide Synthesis: A Practical Approach, Irl Press; and, D. M. J. Lilley and J. E. Dahlberg, 1992, Methods of Enzymology: DNA Structure Part A: Synthesis and Physical Analysis of DNA Methods in Enzymology, Academic Press. Each of these general texts is herein incorporated by reference.
In a specific embodiment of the invention, a cDNA microarray system representing 27, 648 genes was used to select a set of genes predicating the responsiveness to gefitinib for advanced NSCLC. Statistical analysis of the expression profiles identified dozens of genes differentially expressed between responders and non-responders to gefitinib. A drug response scoring (DRS) system based on the expression of these genes successfully predicted the response to gefitinib therapy.
Materials and Methods
Patients and Tissue Samples
A phase II clinical study was carried out comprising a multi-center trial to explore the dominant biological factors responsible for clinical anti-tumor effect, adverse drug reactions (ADR) and pharmacokinetics of ZD1839 dosed 250 mg daily in patients with advanced non-small-cell lung cancer who have failed previous chemotherapy. The primary endpoint was to clarify a gene-expression profile that could determine in advance a potential anti-tumor effect of gefitinib. At the start of the study, the sample size was estimated using studies conducted thus far as a rationale.12,13 Since the response rate for gefitinib has been less than 20% in patients with lung cancer,8-10 about 50 patients were estimated to be required to obtain learning cases estimated above. Patients whose locally advanced (stage IIIB) or metastasized (stage IV) NSCLCs were resistant to one or more regimens of conventional chemotherapy were enrolled in this trial. Inclusion criteria were (1) age greater than 20 years, (2) Performance Status (PS) 0-2, (3) adequate liver and kidney function tests. All patients were treated with 250 mg of gefitinib orally once a day at the Tokushima University or Kinki University hospitals in Japan. The treatment was continued until the patient was dropped from the study due to (1) progression of disease, (2) intolerable toxicity, or (3) withdrawal of consent.
Objective tumor responses were assessed every 4 weeks after the beginning of treatment, according to criteria outlined by the Union International Contre le Cancer/World Health Organization (UICC/WHO). Response categories were as follows: complete response (CR), no residual tumor in any evaluable lesion; partial response (PR), residual tumor with evidence of 50% or greater decrease under baseline in the sum of all measurable lesions, and no new lesions; progressive disease (PD), residual tumor with evidence of 25% or greater increase under baseline in the sum of all measurable lesions, or appearance of new lesions; and stable disease (SD), residual tumor not qualified for CR, PR, or PD. All evaluable lesions were measured bi-dimensionally (sum of products of longest diameter and its longest perpendicular of measurable lesions) using the same techniques as baseline, e.g. plain X-ray, CT, or MRI.
At the end of 4-month treatment (or withdrawal), the best overall response was evaluated for each patient based on definitions as follows: CR, patients who qualified for CR at two sequential examination points with an interval of at least 28 days between them; PR, patients judged as PR or better at two sequential examination points with an interval of at least 28 days between them; SD, patients who were SD or better at two sequential examination points at least 28 days apart but who did not qualify as CR or PR. The first judgment of an SD case must be done at or after the first tumor assessment point (28 days after randomization); PD, the patients determined as PD at or before the first tumor assessment point (28 days after randomization); Unknown, the patient does not qualify for a best response of increased disease, and all objective statuses after baseline (before randomization) and before progression are unknown.
Prior to the gefitinib treatment, tumor specimens were taken by trans-bronchial (TBB), skin, or lymph-node biopsy with written informed consent from each patient. Ethics approval was obtained from the ethics committee of the individual institutes. Biopsy samples were frozen immediately, embedded in TissueTek OCT medium (Sakura, Tokyo, Japan), and stored at −80° C. All samples were examined microscopically, and samples from 28 patients (17 learning and 11 test cases) that contained enough cancer cells for analysis of expression profiles were initially selected for further analysis. For validation of the prediction system, a blinded set of samples from 5 newly enrolled cases (4 PD and 1 SD) were also added to the 11 test cases. Clinical and histological information about these patients is summarized in Table 1-3.
Microdissection
In view of significant differences in the proportions of cancer cells and various types of parenchymal cells that are present from one tumor to another, microdissection is a necessary means of obtaining precise gene-expression profiles on cDNA microarrays. Therefore we stained 8 μm-thick frozen sections with hematoxylin and eosin and collected cancer cells selectively, using the μCUT laser-microbeam microdissection system (Molecular Machines & Industries AG, Glattbrugg, Switzerland).14 In this system tissue sections are mounted on a thin supporting polyethylene membrane that will be cut together with the target tissue; a pulsed-ultraviolet (UV) narrow-beam-focus laser cuts out cancer cells along a pre-selected track that can be observed on a video screen. The material to be extracted is never directly exposed to the laser but only circumscribed by it; unlike other LMM systems, this one allows recovery of dissected cells to proceed without radiation. Moreover, the membrane protects the tissue on the slide against cross-contamination. Using this system we were able to isolate small areas of tissue rapidly, and to isolate single cells from histological sections (
RNA Extraction and T7-Based RNA Amplification
Total RNA was extracted from individual microdissected populations of cancer cells using RNeasy mini kits and RNase-free DNase kits (QIAGEN, Hilden, Germany) according to the manufacturer's protocols. Total RNAs were subjected to T7-based RNA amplification, as described previously.15 Two rounds of amplification yielded 40-200 μg of aRNA (amplified RNA) (>100,000-fold) from each sample. As a control probe, normal human lung poly(A)+RNA (BD Biosciences Clontech, Palo Alto, Calif. and BIOCHAIN, Hayward, Calif., USA) was amplified in the same way. Aliquots (2·5 μg) of mRNA from individual samples and from the control were reversely transcribed in the presence of Cy5-dCTP and Cy3-dCTP respectively.
cDNA Microarray
Our “genome-wide” cDNA microarray system contains 27,648 cDNAs selected from the UniGene database of the National Center for Biotechnology Information.15 Fabrication of the microarray, hybridization, washing, and detection of signal intensities were described previously.15 To normalize the amount of mRNA between tumors and controls, the Cy5/Cy3 ratio for each gene's expression was adjusted so that the averaged Cy5/Cy3 ratio of 52 housekeeping genes was equal to one. We assigned a cutoff value to each microarray slide using analysis of variance, and the Cy5/Cy3 ratio of the gene was calculated as follows: (1) if Cy5 (cancer sample) was lower than the cut off level, then the Cy5/Cy3 ratio of the gene was substituted by 2-5 percentile among the Cy5/Cy3 ratios of other genes whose Cy5 and Cy3 were higher than the cut off level; (2) if Cy3 (control sample) was lower than the cut off level, then the Cy5/Cy3 ratio of the gene was substituted by 97·5 percentile among the Cy5/Cy3 ratios of other genes whose Cy5 and Cy3 were higher than the cut off level; (3) if both Cy5 and Cy3 were lower than the cut off level, then the Cy5/Cy3 ratio of the gene was left blank.
Extraction of Genes for Predicting Responsiveness to Gefitinib
To discover genes that might be associated with sensitivity to gefitinib, individual measurements of about 27,648 genes were compared between the two groups of patients, one classified as responders to gefitinib (PR) and the other as non-responders (PD). To reduce the dimensionality of the number of potent genes that could discriminate between the two classes, we extracted only genes that fulfilled two criteria: 1) signal intensities were higher than the cut-off level in at least 60% of either group, and 2) 1 MEDPR−MEDPD|≦1, where MED indicates the median calculated from log-transformed relative expression ratios in each group. Then random-permutation tests were applied to estimate the ability of individual genes to distinguish between the two classes (PR and PD); mean (μ) and standard deviations (σ) were calculated from the log-transformed relative expression ratios of each gene in both groups. A discrimination score (DS) for each gene was defined as follows:
DS=(μPR−μPD)/(PR+σPD).
The samples were randomly permutated 10,000 times for each pair of groups. Since the DS dataset of each gene showed a normal distribution, we calculated a p-value for the user-defined grouping.
Calculation of Drug-Response Scores
We calculated the drug response scores for gefitinib (gefitinib response scores, or GRS) reflecting the expression levels of candidate prediction-genes according to procedures described previously.16-18 Each gene (gi) votes for either responder (PR) or non-responder (PD) depending on whether the expression level (xi) in the sample is closer to the mean expression level of one group or the other in reference samples. The magnitude of the vote (vi) reflects the deviation of the expression level in the sample from the average of the two classes:
Vi=|xi−(μPR+μPD)/2|.
We summed the votes to obtain total votes for responders (VPR) and non-responders (VPD), and calculated GRS values as follows: GRS=((VPR−VPD)/(VPR+VPD))×100, where the GRS value reflects the margin of victory in the direction of either responder or non-responder. GRS values range from −100 to 100; the higher an absolute value of GRS, the stronger the prediction.
Cross-Validation of Scores and Evaluation of the Prediction System
The prediction scores of all samples were obtained by a leave-one-out approach, in which one sample at a time was removed from the sample set; permutational p-values and mean values of the two classes were calculated for each gene using the remaining samples. The drug-response of the withheld sample was predicted by calculating the prediction score.
These procedures were repeated for each sample.16-17
To evaluate the reliability of the prediction system, we calculated a “classification score” (CS) using the GRS values of responders and non-responders in each gene set, as follows:
CS=(μGRSpr−μGRSpd)/(GRSpr+
GRSpd).17
A larger value of CS indicates better separation of the two groups by the prediction system.
Hierarchical Clustering
We used web-available software (“Cluster” and “TreeView”) written by M. Eisen (http://genome-www5.stanford.edu/MicroArray/SMD/restech.html) to create a graphic representation of the microarray data and to create a dendrogram of hierarchical clustering. Before the clustering algorithm was applied, the fluorescence ratio for each spot was first log-transformed and then the data for each sample were median-centered to remove experimental biases.
Semi-Quantitative RT-PCR Analysis
Aliquots (5·0 μg) of the same aRNA hybridized to the microarray slides from individual samples and from the normal control lung were reversely transcribed using oligo(dT)12-18 primer and SuperScript II reverse transcriptase (Invitrogen, Carlsbad, Calif., USA). Semi-quantitative RT-PCR experiments were carried out with the following sets of synthesized primers specific to the 12 top-ranked genes used for establishing a GRS or with beta-actin (ACTB)-specific primers as an internal control: FLJ22662, 5′-GCCATAAGTGGTCCCACAGT-3′ and 5′-GTCTTCTAGTCCGTCATCTCCCT-3′; Amphiregulin (AREG), 5′-CCATAGCTGCCTTTATGTCTGC-3′ and 5′-CTTTTTACCTTCGTGCACCTTT-3′, coronin, actin binding protein, IC (COROIC), 5′-TAATCTGCTGAGGACCTTTTGTC-3′ and 5′-TAATTCACTGTCCTCTTCTGGGA-3′; apoptosis, caspase activation inhibitor (AVEN), 5′-GCTCACAGCAGTAAATGCCTA-3′ and 5′-TGCTATGCTGTAAACACTGGCTA-3′; dual specificity phosphatase 3 (DUSP3), 5′-GGATCCTTTATTGGTGGTAGAGC-3′ and 5′-CCAGAGTGACCCTGAAGATAAAT-3′; DJ473B4, 5′-ACCTGATTCTCTAGGTGCAGTTT-3′ and 5′-GTCGTTTCAACCAGGTAGTTTTG-3′; pleckstrin homology-like domain, family A, member 2 (PHLDA2), 5′-GGGCGCCTTAAGTTATTGGA-3′ and 5′-GGATGGTAGAAAAGCAAACTGG-3′; RNA binding motifprotein 7 (RBM7), 5′-TGTAATGGAGATTGTACAGGTTG-3′ and 5′-AGGAACAGTACAAATGCTGTGGT-3′; BX092512 (EST), 5′-GCACTCCTTGAAGGTACACTAAC-3′ and 5′-ATTTGTATTCACTCAGCCATGC-3′; oncostatin M receptor (OSMR), 5′-ACCCAACTTCAAAACTAGGACTC-3′ and 5′-ACAGCTTGATGTCCTTTCTATGC-3′, glutamate-cysteine ligase, catalytic subunit (GCLC), 5′-TCATGAAAGGCACTGAGTTTTG-3′ and 5′-GTTAGCTGAAGCAGCTTTATTGC-3′; collagen, type IV, alpha 3 binding protein (COL4A3BP), 5′-ATATGCACAATCCTGGAAGTGA-3′ and 5′-TGCCTTACTAGCATTACCACCAT-3′; ACTB, 5′-GAGGTGATAGCATTGCTTTCG-3′ and 5′-CAAGTCAGTGTACAGGTAAGC-3′. PCR reactions were optimized for the number of cycles to ensure product intensity within the logarithmic phase of amplification. We did phosphor imager quantification analysis (Molecular Imager FX: Bio-Rad Laboratories, Hercules, Calif., USA), and RT-PCR band intensities were quantitatively compared with normalized Cy5/Cy3 ratio of gene expression from the microarray data.
RT-PCR was performed to screen the mutation at entire region of codon 709-870 (from p-loop to activation loop) of EGFR which was recently reported as a hot spot of mutation,18 using three primer sets: fragment-1,5′-TCTTACACCCAGTGGAGAAGC-3′ and 5′-GTCTTTGTGTTCCCGGACAT-3′; fragment-2,5′-ACTATGTCCGGGAACACAAA-3′ and 5′-TTCCGTCATATGGCTTGG-3′; fragment-3,5′-CGTCGCTATCAAGGAATTAAGAG-3′ and 5′-GTAGCTCCAGACATCACTCTGGT-3′. RT-PCR products from 19 NSCLC patients treated with gefitinib were analyzed by direct sequencing.
Immunohistochemical Analysis
To confirm the differential expression of AREG and transforming growth factor-alpha (TGFA) proteins, both of which encode the ligand for EGFR and other ERBB members, and other 3 candidate markers (a disintegrin and metalloproteinase domain 9 (ADAM9), D9 antigen (p24), and OSMR), which are also known to relate to the EGFR signalling, for predicting responders vs non-responders to gefitinib, we stained clinical tissue sections obtained by fiberscopic transbronchial biopsy (TBB) and lymph-node biopsy using ENVISION+ Kit/HRP (DakoCytomation, Glostrup Denmark). Briefly, after endogenous peroxidase and protein blocking reactions, anti-human AREG polyclonal antibody (Neo Markers, Fremont, Calif., USA), anti-human TGFA monoclonal antibody (Calbiochem, Darmstadt, Germany), anti-human ADAM9 monoclonal antibody (R&D Systems Inc. Minneapolis, Minn., USA), anti-human CD9 monoclonal antibody (Novocastra Laboratories Ltd, Newcastle upon Tyne, UK), or anti-human OSMR monoclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, Calif., USA), was added, and then HRP-labeled anti-rabbit or anti-mouse IgG as the secondary antibody. Substrate-chromogen was then added and the specimens were counterstained with hematoxylin.
Frozen tissue samples from 11 patients were selected for analysis of immunohistochemistry. Positivity of immunostaining was assessed semi-quantitatively by scoring intensity as absent or positive by three independent investigators without prior knowledge of the clinical follow-up data. Cases were accepted only as positive if reviewers independently defined them thus.
ELISA
Serum was obtained from an independent set of 35 lung-ADC patients who were treated with gefitinib based on the same protocol as this clinical study at Hiroshima University hospital in Japan (5 for PR, 10 for SD, and 20 for PD). The sera of all the patients were obtained with informed consent at the time of diagnosis and every 4 weeks after the beginning of treatment, and stored at −80° C. The serum TGFA levels were measured by an ELISA using a commercially available enzyme test kits (TGF-alpha ELISA kit: Oncogene Rsearch Products, San Diego, Calif., USA).
In Vitro Gefitinib Treatment and AREG-Autocrine Assay
Human NSCLC (adenocarcinoma) cell lines PC-9, NCI-H358, and NCI-H522 were purchased from the American Type Culture Collection (ATCC; Rockville, Md., USA). To detect expression of AREG in these NSCLC cells, total RNA from each line was reverse-transcribed for single-stranded cDNAs using oligo(dT)12-18 primer and Superscript II (Invitrogen). Semi-quantitative reverse transcriptase-PCR (RT-PCR) was carried out as described previously.14 gefitinib (4-(3-chloro-4-fluoroanilino)-7-methoxy-6-(3-morpholinopropoxy)quinazoline: ZD 1839, Iressa), an inhibitor of epidermal growth factor receptor tyrosine kinase, was provided by AstraZeneca Pharmaceuticals (Macclesfield, UK). The drug was dissolved in DMSO at a concentration of 10 mM and kept at −20° C.
We performed flow-cytometry to determine the sensitivity of lung adenocarcinoma cell lines to gefitinib treatment. Cells were plated at densities of 5×10 cells/100-mm dish and treated with 1·0 μM of gefitinib in appropriate serum-free medium. The cells were trypsinized 72 hours after the treatment, collected in PBS, and fixed in 70% cold ethanol for 30 min. After treatment with 100 μg/ml RNase (Sigma-Aldrich Co., St. Louis, Mo., USA), the cells were stained with 50 μg/ml propidium iodide (Sigma-Aldrich Co.) in PBS. Flow cytometry was performed on a Becton Dickinson FACScan and analyzed by ModFit software (Verity Software House, Inc., Topsham, Me., USA). The percentages of nuclei in G0/G1, S, and G2/M phases of the cell cycle and sub-G1 population were determined from at least 20,000 ungated cells.
To investigate whether AREG functions as an autocrine anti-apoptotic factor in lung adenocarcinoma cells treated with gefitinib, we carried out the following assay. First, gefitinib-sensitive PC-9 cells, which do not express AREG, were cultured in serum-free medium for at least 8 hours prior to gefitinib treatment. These cells were then incubated with 0·5 or 1·0 μM of gefitinib for 72 hours in media that were either serum-free or supplemented with 10% FCS, or in serum-free conditioned medium collected from 72-hour cultures of AREG-expressing cells (NCI-H358 or NCI-H522). Each medium was replaced once with the same medium containing gefitinib at the 48-hour time point. To detect the response of each cell line to gefitinib, viability was evaluated by MTT assays using Cell Counting Kits (WAKO, Osaka, Japan).
To confirm the autocrine effect of AREG on the gefitinib-resistance of NSCLC cells, we cultured PC-9 cells for 72 hours in serum-free medium containing 1·0 μM of gefitinib and recombinant AREG protein (Genzyme-Techne, Minneapolis, Minn., USA) in final concentrations of 1-100 ng/ml. Cell viability was evaluated by MTT assays. A possible effect of AREG itself on the viability of NSCLC cells was evaluated also, by culturing the PC-9 cells in serum- and gefitinib-free medium containing only recombinant AREG protein. MTT assays were performed as above.
Results
Response to Gefitinib Treatment
Of the 53 patients enrolled in this trial, 46 had tumors diagnosed as adenocarcinomas (86·8%); five were squamous-cell carcinomas (9·4%); two were large cell carcinomas (3·8%). Fifteen patients achieved a PR and nobody revealed a CR; 17 patients were classified as SD, and 19 as PD. No clinical-response data were available for two of the patients. The tumor-response rate (CR+PR/CR+PR+SD+PD) for this treatment was 29·4%, and the disease control rate (CR+PR+SD/CR+PR+SD+PD) was 62·8% (table 1).
Tumor samples were collected from 43 patients. Samples from 32 of those 43 contained sufficient numbers of cancer cells for analysis of expression profiles on our cDNA microarray. The numbers of samples that were judged to be suitable for further microarray analysis, were 8 for PR, 7 for SD, and 13 for PD (table 2). 17 of the 28 samples were analyzed as learning cases (7 for PR and 10 for PD), and 11 were as test cases (1 for PR, 3 for PD, and 7 for SD) for establishing a predictive scoring system for the efficacy of gefitinib treatment. For further validation of the prediction system, another blinded set of samples from 5 newly enrolled test-cases (4 for PD and 1 for SD) were obtained and added finally to the initial 11 test cases above.
Identification of Genes Associated with Sensitivity to Gefitinib
We attempted to extract genes that were differentially expressed between tumors from seven patients in the PR group (defined as responders) and those from 10 patients in the PD group (defined as non-responders) by comparing expression levels of 27,648 genes. (tables 2, 3).
We carried out a random-permutation test to distinguish between the two subclasses defined by tumor response, and identified 51 genes whose permutational p-values were less than 0·001 (table 4). Expression levels of 40 genes were higher, and those of the other 11 were lower, in the non-responders.
Establishment of a predictive scoring system for the efficacy of gefitinib treatment Based on the expression profiles of the 51 genes selected above, we tried to establish a predictive scoring system for the efficacy of gefitinib treatment. Prediction scores, termed gefitinib response score (GRS), were calculated according to procedures described previously (see Methods). To determine the number of candidates that provided the best separation of the two groups, we ranked the 51 genes on the basis of the significance of their permutational p-values and calculated prediction scores by the leave-one-out test, in decrements of 1 starting from the bottom of the rank-ordered list (51, 50, 49, 48 etc.). We calculated a classification score (CS), a standard we had previously defined for evaluation of the ability to discriminate two classes, for each set of genes.17
As shown in
Hierarchical clustering analyses using all 51 genes, or only the top 12, classified all 17 cases into one of two groups according to the response to gefitinib (
To validate this prediction system we investigated 8 additional (“test”) NSCLC cases (1 for PR and 7 for PD) that were completely independent of the 17 “learning” cases used for establishing the system. We examined gene-expression profiles in each of those samples and then calculated GRS on the basis of the expression levels of the 12 discriminating genes. As shown in
GRS Values for Patients with SD in Tumor Response
GRS values for the eight test-SD patients were calculated according to the predictive scoring system established above. Although the values were widely distributed from −83·0 (predicted as non-responder) to 61·6 (responder), the scores of patients who retained SD status throughout the observation period were likely to be higher than those of patients who had been judged as SD at a certain time-point of the study but showed progression of the disease within three or four months after the start of treatment (
Validation of GRS with Semi-Quantitative RT-PCR Analysis
To confirm differential expression of the top 12 predictive genes between PR and PD cases, expression values derived from microarray data were correlated with values from semi-quantitative RT-PCR of RNAs from the same patients (5 PR and 7 PD) (
Immunohistochemical Validation of GRS
To validate differential expression of the predictive protein markers between PR and PD cases, we carried out immunohistochemical staining with five different antibodies for AREG, TGFA, ADAM9, CD9, and OSMR, all of which were known to be involved in the ligand-EGFRs signalling and whose permutational p-values were less than 0·01. We first stained paired tumor tissue sections obtained by TBB and lymph-node biopsy from the same patients using these 5 antibodies. No intra-patient differences on protein expression of these five markers were observed in three different patients (
Serum Levels of TGFA
To further evaluate the availability of the prediction system in routine clinical situations, we detected TGFA protein using ELISA in serum samples from 5 PR, 10 SD, and 20 PD patients that were independently collected for serological test and were not enrolled in microarray analysis. The serum levels of TGFA were 19·0±2·8 pg/ml (mean±SE) in PD patients, 13·9±1·9 pg/ml in SD patients, and 12·8±1·4 pg/ml in PR patients (
In Vitro Gefitinib Treatment and AREG-Autocrine Assay
AREG, a ligand for EGFR and other ERBB members was significantly over-expressed in non-responders but not (or hardly) detectable in responders. To investigate whether AREG protein leads to resistance of NSCLCs to gefitinib therapy when it is secreted in an autocrine manner, we performed the following biological analyses. We initially identified expression of AREG mRNA in lung-adenocarcinoma cell lines NCI-H358 and -H522, but not in PC-9, by means of RT-PCR experiments (
We then analyzed the viability of PC-9 cells, which are gefitinib-sensitive and do not express AREG, after culture in serum-free medium or in serum-free, conditioned medium obtained from NCI-H358 or -H522 cells grown in the presence or absence of 0·5 or 1·0 μM of gefitinib. As shown in
To investigate whether AREG, secreted in an autocrine manner, inhibits apoptosis of NSCLC cells treated with gefitinib, we cultured PC-9 cells in serum-free medium containing recombinant AREG protein at final concentrations of 1-100 ng/ml, in the presence or absence of 1·0 μM gefitinib. The viability of PC-9 cells incubated with both AREG and 1·0 μM gefitinib was increased in comparison to cells incubated with 1·0 μM gefitinib only, in an AREG-dose-dependent manner (
Discussion
A large body of evidence supports the view that molecules in the EGFR autocrine pathway are involved in a number of processes important to cancer formation and progression, including cell proliferation, angiogenesis, and metastatic spread.5 Therapeutic blockade of specific signalling, therefore, could be a promising strategy for cancer treatment. Gefitinib, a synthetic anilinoquinazoline, inhibits the tyrosine kinase activity of EGFR by competing with adenosine triphosphate for a binding site on the intracellular domain of the receptor.7 In phase II trials (IDEAL 1 and IDEAL 2), use of gefitinib as a 2nd-, 3rd-, or 4th-line monotherapy for advanced NSCLC achieved tumor-response rates of nearly 20%,8-10 which were superior to those achieved with conventional cytotoxic agents. Multivariate analysis of patients in the IDEAL 1 study suggested that the response rate in females might be higher than in males, and higher in patients with adenocarcinomas than in patients with squamous-cell carcinomas (odds ratios 2·7 and 3·5 respectively).9 Recent study suggested that individuals in whom gefitinib is efficacious are more likely to have adenocarcinomas of the bronchioloalveolar subtype and to be never smokers (odds ratios 13·5 and 4·2 respectively).19 The higher tumor-response rate (29·4%) documented in the clinical trial reported here might reflect a higher proportion of patients with adenocarcinoma (46 adenocarcinomas, five squamous-cell carcinomas and two large-cell carcinomas) than has been the case in other studies. The clinicopathological determinants of gefitinib sensitivity including bronchioloalveolar carcinoma (BAC) features are predictve to a certan extent,9,10,19,20 however, previous reports and our observations obviously suggest that no factors can perfectly predict the response of NSCLC to gefitinib treatment. Therefore novel methods to discriminate responders from non-responders in advance could allow a more focused use of gefitinib in clinical settings.
By statistical analysis of gene-expression profiles of advanced NSCLCs obtained on cDNA microarrays, we identified dozens of genes associated with sensitivity to gefitinib. We introduced a prediction-scoring system based on expression of the 12 genes that had shown the most significant differences in expression levels between responder (PR) and non-responder (PD) groups. This set of genes was selected from expression profiles of lung adenocarcinomas; however, the GRS system successfully classified all eight of our “test” PR and PD cases in accord with their clinical responses to gefitinib, and one of them was a squamous-cell carcinoma. Moreover, this system was likely to separate intermediate tumor responses (SD) into two groups, one representing patients who succeeded in maintaining the tumor-static effect for a long period and the other representing patients who failed to do so.
In practical terms, we need to predict the chemosensitivity of individual tumors using the minimally invasive techniques available at every hospital, because patients with advanced NSCLCs are rarely candidates for surgical resection of their tumors. Therefore we have tried to establish a prediction system that requires only the amount of cancerous tissue that can be obtained by, for example, flexible bronchofiberscopy. By verifying individual steps of the method, we were able to precisely profile gene expression in biopsy specimens as small as 1 mm. Relevant microarray results were confirmed by semi-quantitative RT-PCR for 12 genes that showed the most significant differences to establish a GRS system. Furthermore, we validated the effectiveness of antibodies for 5 different biomarkers (AREG, TGFA, ADAM9, CD9, and OSMR), all of which were reported to be involved in the ligand-EGFR signalling, for discriminating potential responders from non-responders, in both TBB and lymph-node biopsy samples. Moreover, we were able to detect serum TGFA proteins in lung-ADC patients by ELISA. Further evaluation of these markers for clinical use are necessary, however, the limited number of genes required for prediction should eventually enable laboratories to diagnose in advance the efficacy of gefitinib treatment for an NSCLC patient, using routine procedures such as serological examinations of blood, PCR experiments, or immunohistochemical analysis of biopsy specimens.
To our knowledge, this is the first report about gene-expression profiles of unresectable “advanced” lung cancers, although profiles of surgically resected specimens of “early” lung cancers have been reported.21,22 However, about 70% of tumors in patients diagnosed with NCSLC are already locally advanced or metastatic, which generally renders them resistant to conventional therapeutic modalities. Therefore the genes listed here should be useful for disclosing molecular mechanisms of lung-cancer progression and may be potential targets for drug development.
Gefitinib was developed as a “selective” inhibitor of EGFR-TK; however, no clear association between the level of EGFR activation and response to gefitinib has been found in vitro or in vivo.7,23 In clinical trials, gefitinib has been more effective against adenocarcinomas than against squamous-cell carcinomas,9,10 although over-expression of EGFR is less frequent in adenocarcinomas.24 Therefore, it is important to identify which individual tumors are good targets for this treatment. In our analysis using clinical samples, the difference in EGFR protein expression between responders and non-responders were not statistically significant. On the other hand, amphiregulin (AREG) and transforming growth factor alpha (TGFA), both of which encode the ligand for EGFR and other ERBB members, were significantly over-expressed in non-responders but not (or hardly) detectable in responders p=0·0000000000093 and 0·0095 respectively; table 4).
The significance of the ligands and the EGFR autocrine loop in growth and survival of lung-cancer cells is indisputable,24-26 but the role of AREG in formation and progression of cancers is poorly understood. However, several lines of evidence suggest that over-expression of AREG is associated with shortened survival of patients with NSCLC.24 Moreover, anti-apoptotic activity of AREG in human lung-adenocarcinoma cells was reported recently.25 To investigate whether the anti-apoptotic activity of AREG leads to resistance of NSCLC cells to gefitinib therapy, we performed a biological assay using a gefitinib-sensitive but AREG-non-expressing NSCLC cell line, PC-9. We found that the anti-tumor activity of gefitinib on PC-9 cells was dramatically decreased by autocrine secretion of AREG. This evidence strongly suggests that although growth-factor signalling by the EGFR is markedly complicated at every step because of the multiplicity of ligands, dimerization partners, effectors, and downstream pathways,26 AREG might be a principal activator of the ligands-receptor autocrine growth pathway that leads to cancer progression and resistance to gefitinib.
Several elements associated with the EGFR-TK pathway are present on our list of differentially-expressed genes. For example, genes encoding dual specificity phosphatase 3 (DUSP3), ADAM9, CD9, and OSMR were expressed predominantly in non-responders (p=0·00000000094, 0·01, 0·000022, and 0.0000011, respectively). DUSP3 gene modulates EGFR signalling by dephosphorylating mitogen activated protein kinase (MAPK), a key mediator of signal transduction,27 and ADAM9 is involved in activation of EGFR signalling by shedding the ectodomain of proHB-EGF (pro Heparin-binding epidermal growth factor-like growth factor).28 CD9 physically interacts with transmembrane TGFA. CD9 expression strongly decreases the growth factor- and PMA-induced proteolytic conversions of transmembrane to soluble TGFA and strongly enhances the TGFA-induced EGFR activation.29 OSMR is reported to be constitutively associated with ERBB2 in breast cancer cells.30 Although other target molecules for gefitinib have been suggested, our results suggest that EGFR signalling is at least one of the important processes involved in response to this drug.
Since gefitinib can induce apoptosis of some cancer cells in vivo, other molecules with anti-apoptotic activity, as well as AREG, may contribute to a tumor's resistance to the drug. AVEN (apoptosis, caspase-activation inhibitor), which was specifically expressed in our non-responders (p=0·00000000042), is known to enhance the anti-apoptotic activity of Bcl-xL and to suppress Apaf-1-mediated caspase activation.31 On the other hand, mechanisms regulating drug transport should also affect drug resistance. GCLC (glutamate-cysteine ligase, catalytic subunit), which plays an important role in cellular detoxification of anticancer drugs such as cisplatin, etoposide and doxorubicin,32 was over-expressed in our group of non-responders (p=0·00000012). As these genes correlated negatively with responses to chemotherapy in our panel of tumors (i.e. the higher the expression of these genes, the greater the resistance to gefitinib), they might be involved in the mechanism(s) leading to that resistance. It should be noted also that the functions of nearly half of our candidate prediction-genes are unknown. Therefore further investigations will be needed to reveal more clearly the biological events underlying responses of NSCLCs to gefitinib.
In summary, we identified 51 genes whose expression differed significantly between responders and non-responders to gefitinib among human lung carcinomas, and established a numerical scoring system, based on expression patterns of 12 of those genes, to predict the response of individual tumors to this drug. Although further validation using a larger set of clinical cases will be necessary, the data presented here may yield valuable insights into the molecular events underlying signal-suppressing strategies and provide important information about gefitinib treatment for individual NSCLC patients by testing a set of genes with high predictive values.
1. Fossella, F. V., et al., Randomized phase III trial of docetaxel versus vinorelbine or ifosfamide in patients with advanced non-small-cell lung cancer previously treated with platinum-containing chemotherapy regimens. The TAX 320 Non-Small Cell Lung Cancer Study Group. J Clin Oncol, 2000. 18(12): p. 2354-62.
2. Non-small Cell Lung Cancer Collaborative Group. Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomised clinical trials. Bmj, 1995. 311(7010): p. 899-909.
3. Schiller, J. H., et al., Comparison of four chemotherapy regimens for advanced non-small-cell lung cancer. N Engl J Med, 2002. 346(2): p. 92-8.
4. Kelly, K., et al., Randomized phase III trial of paclitaxel plus carboplatin versus vinorelbine plus cisplatin in the treatment of patients with advanced non-small-cell lung cancer: a Southwest Oncology Group trial. J Clin Oncol, 2001. 19(13): p. 3210-8.
5. Baselga, J., Why the epidermal growth factor receptor? The rationale for cancer therapy. Oncologist, 2002. 7 Suppl 4: p. 2-8.
6. Traxler, P., Tyrosine kinases as targets in cancer therapy—successes and failures. Expert Opin Ther Targets, 2003. 7(2): p. 215-34.
7. Wakeling, A. E., et al., ZD1839 (Iressa): an orally active inhibitor of epidermal growth factor signaling with potential for cancer therapy. Cancer Res, 2002. 62(20): p. 5749-54.
8. Herbst, R. S., Dose-comparative monotherapy trials of ZD1839 in previously treated non-small cell lung cancer patients. Semin Oncol, 2003. 30(1 Suppl 1): p. 30-8.
9. Fukuoka, M., et al., Final results from a phase □ trial of ZD1839 (‘Iressa’) for patients with advanced non-small cell lung cancer (IDEAL 1). Pro Am Soc Clin Oncol 2002. 21;298a(A1188).
10. Kris, M G., et al., A phase II trial of ZD1839 (‘Iressa’) in advanced non-small cell lung cancer (NSCLC) patients who had failed platinum-and docetaxel-based regimens (IDEAL 2). Pro Am Soc Clin Oncol 2002. 21;292a(A1166).
11. Inoue, A., et al., Severe acute interstitial pneumonia and gefitinib. Lancet, 2003. 361(9352): p. 137-9.
12. Bohm, M., et al., Microbeam MOMeNT: non-contact laser microdissection of membrane-mounted native tissue. Am J Pathol, 1997. 151(1): p. 63-7.
13. Okabe, H., et al., Genome-wide analysis of gene expression in human hepatocellular carcinomas using cDNA microarray: identification of genes involved in viral carcinogenesis and tumor progression. Cancer Res, 2001. 61(5): p. 2129-37.
14. Kitahara, O., et al., Alterations of gene expression during colorectal carcinogenesis revealed by cDNA microarrays after laser-capture microdissection of tumor tissues and normal epithelia. Cancer Res, 2001. 61(9): p. 3544-9.
15. Golub, T. R., et al., Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science, 1999. 286(5439): p. 531-7.
16. MacDonald, T. J., et al., Expression profiling of medulloblastoma: PDGFRA and the RAS/MAPK pathway as therapeutic targets for metastatic disease. Nat Genet, 2001. 29(2): p. 143-52.
17. Kaneta. Y.,et al., Prediction of sensitivity to STI571 among chronic myeloid leukemia patients by genome-wide cDNA microarray analysis. Jpn J Cancer Res 2002. 93, p. 849-856.
18. Pavelic, K., et al., Evidence for a role of EGF receptor in the progression of human lung carcinoma. Anticancer Res, 1993. 13(4): p. 1133-7.
19. Kikuchi, T., et al., Expression profiles of non-small cell lung cancers on cDNA microarrays: Identification of genes for prediction of lymph-node metastasis and sensitivity to anti-cancer drugs. Oncogene, 2003. 22(14): p. 2192-205.
20. Heighway, J., et al., Expression profiling of primary non-small cell lung cancer for target identification. Oncogene, 2002. 21(50): p. 7749-63.
21. Beer, D. G., et al., Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med, 2002. 8(8): p. 816-24.
22. Miura, K., et al., Laser capture microdissection and microarray expression analysis of lung adenocarcinoma reveals tobacco smoking-and prognosis-related molecular profiles. Cancer Res, 2002. 62(11): p. 3244-50.
23. Moasser, M. M., et al., The tyrosine kinase inhibitor ZD1839 (“Iressa”) inhibits HER2-driven signaling and suppresses the growth of HER2-overexpressing tumor cells. Cancer Res, 2001. 61(19): p. 7184-8.
24. Rusch, V., et al., Overexpression of the epidermal growth factor receptor and its ligand transforming growth factor alpha is frequent in resectable non-small cell lung cancer but does not predict tumor progression. Clin Cancer Res, 1997. 3(4): p. 515-22.
25. Fontanini, G., et al., Evaluation of epidermal growth factor-related growth factors and receptors and of neoangiogenesis in completely resected stage I-IIIA non-small-cell lung cancer: amphiregulin and microvessel count are independent prognostic indicators of survival. Clin Cancer Res, 1998. 4(1): p. 241-9.
26. Brundage, M. D., D. Davies, and W. J. Mackillop, Prognostic factors in non-small cell lung cancer: a decade of progress. Chest, 2002. 122(3): p. 1037-57.
27. Hurbin, A., et al., Inhibition of apoptosis by amphiregulin via an insulin-like growth factor-1 receptor-dependent pathway in non-small cell lung cancer cell lines. J Biol Chem, 2002. 277(51): p. 49127-33.
28. Yarden, Y. and M. X. Sliwkowski, Untangling the ErbB signalling network. Nat Rev Mol Cell Biol, 2001. 2(2): p. 127-37.
29. Prenzel, N., et al., EGF receptor transactivation by G-protein-coupled receptors requires metalloproteinase cleavage of proHB-EGF. Nature 1999. 402(6764):884-8
30. Nelson, Chau, et al., Aven, a novel inhibitor of caspase activation, binds Bcl-xL and Apaf-1. Molec Cell 2000.6: p. 31-41.
31. Tipnis, S R., et al., Overexpression of the regulatory subunit of r-glutamylcysteine synthetase in Hela cells increases r-glutamylcysteine synthetase activity and confers drug resistance. Biochem J 1999. 337, p. 559-566.
(1) Learning cases were used for developing the GRS, whereas test cases were used for validation of the
(2) Another blinded set of samples from 5 newly enrolled cases (4 PD and 1 SD) were also added to these 11 test cases later.
(1) ADC, adenocarcinoma; SCC, squamous-cell carcinoma.
(2) TNM clinical classification and stage grouping were assessed based on the UICC/WHO classification.
(3) Mutation at codon position 709-870 (from p-loop to activation loop) of EGFR (GenBank Accession No. NM005228).
(4) Objective Tumor Response to Gefitinib was assessed every 4 weeks after the start of treatment using UICC/WHO Criteria. PR, partial response; SD, stable disease; PD, progressive disease
(5) Overall Best Response was evaluated based on the definitions as mentioned in materials and methods.
(6) learning, samples used for developing the GRS; test, samples used for validation of the GRS.
(7) GRS: gefitinib response score determined by prediction system
(*) For further validation of the GRS, another blinded set of samples from 5 newly enrolled cases (4 PD and 1 SD) were also added to these 28 cases later.
rotein zeta 2 (zygin II)
(*) The 12 and 51 gene sets were listed as the rank-order of permutational p-values that were less than 0.001.
Correlations positive for all 12 genes and significantly positive for 7 of 12 ge
| Number | Date | Country | Kind |
|---|---|---|---|
| 0312451.8 | May 2003 | GB | national |
| 0322636.2 | Sep 2003 | GB | national |
| 0327132.7 | Nov 2003 | GB | national |
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/GB04/02316 | Jun 2004 | US |
| Child | 11290173 | Nov 2005 | US |
