BUSINESS

Company Overview

      We are a life science company focused on the development and commercialization of genomic-based clinical diagnostic tests for cancer that allow physicians and patients to make individualized treatment decisions. In January 2004, we launched our first test under the brand name Onco type DX for early stage breast cancer patients. We believe that Onco type DX is the first genomic test with clinical evidence supporting its ability to predict the likelihood of cancer recurrence, the likelihood of patient survival within 10 years of diagnosis and the likelihood of chemotherapy benefit. Our initial test is focused on patients with early stage, node negative, or N-, estrogen receptor positive, or ER+, breast cancer who will be treated with tamoxifen, a frequently used hormonal therapy. Approximately half of the 230,000 patients expected to be diagnosed with breast cancer in the United States in 2005 are predicted to be early stage cancer patients that are N- and ER+. Many of the diagnostic factors currently used in connection with early stage breast cancer are subjective, have limited capability to predict future cancer recurrence and are not useful in predicting response to chemotherapy. We believe that the use of Onco type DX can provide a deeper understanding of each patient’s breast cancer and therefore should result in better informed and more appropriate treatment decisions.

      We developed Onco type DX using a multi-step approach, conducting clinical studies on tumor specimens from more than 2,600 breast cancer patients. Our technology provides quantitative gene expression information for each patient’s tumor, which we refer to as an oncotype. When an oncotype is correlated with known clinical outcomes, it can be useful in predicting the likelihood of an individual patient’s tumor behavior. Onco type DX for breast cancer utilizes a 21-gene panel whose composite gene expression profile can be represented by a single quantitative score, which we call a Recurrence Score. The higher the Recurrence Score, the more aggressive the tumor and the more likely it is to recur. The lower the Recurrence Score, the less aggressive the tumor and the less likely it is to recur. Onco type DX has been clinically validated for N-, ER+, tamoxifen-treated breast cancer patients by two, large independent studies. Moreover, we have demonstrated that the Recurrence Score correlates with chemotherapy benefit, and we are undertaking further studies to support this finding. Onco type DX is commercially available at a list price of $3,460 through our laboratory located in Redwood City, California, which is accredited under the Clinical Laboratory Improvement Amendments of 1988, or CLIA, and by the College of American Pathologists, or CAP. As of March 31, 2005, Onco type DX has been ordered by over 900 physicians throughout the United States.

Scientific Background

          Limits of Existing Approaches for Determining Cancer Treatments

      Cancer is a group of complex molecular diseases characterized by the uncontrolled growth and spread of abnormal cells resulting from genetic mutations or damage that can severely disrupt normal body functions. In 2005, approximately 1.4 million people in the United States are expected to be diagnosed with cancer. Common types of cancer include breast, prostate, lung and colon. Cancers are difficult to treat because each type responds differently to treatments depending upon the individual and the type and location of the cancer.

      To treat cancer effectively, physicians diagnose and gauge the stage of a patient’s disease to determine the best course of therapy. The most common practice used to diagnose cancer is through pathologic evaluation of tumors under a microscope. For solid tumors, tumor tissue is typically removed through surgery or needle biopsy, fixed in formalin or other fixatives to preserve the tissue and sent to a pathology laboratory, where it is embedded in paraffin wax. A pathologist places thin sections of this fixed paraffin embedded, or FPE, tissue onto glass slides so it can be studied under a microscope. In many cases, pathologists also use molecular staining techniques, including immunohistochemistry-based stains for protein expression, to improve the quality of their diagnosis. After visually examining the sample, the pathologist judges whether the biopsy contains normal or cancerous cells. The pathologist may also grade the tumor based on how aggressive the cancer cells appear under the microscope.

      Once a pathologist diagnoses cancer, the patient’s physician determines the stage of the cancer based on further analysis of the patient’s condition using a variety of clinical measures, including the pathology grade, size of the tumor, how deeply the tumor has invaded tissues at the site of origin and the extent of any invasion into surrounding organs, lymph nodes or distant sites. Patient history, physical signs, symptoms and information obtained from existing tests are also evaluated and considered.

      Physicians currently rely primarily on tumor pathology grade and stage when predicting whether a cancer will recur, which is the key determinant in treatment decisions. Because tumor pathology and staging are heavily dependent on visual assessment and human interpretation, physicians and patients make treatment decisions often using subjective and qualitative information that may not reflect the molecular nature of the patient’s cancer. As a result, many patients are misclassified as high risk when they are truly low risk for recurrence or low risk when they are high risk for recurrence, resulting in over-treatment for some and under-treatment for others.

      For many cancer patients, chemotherapy is commonly used as a treatment. Chemotherapy involves the use of highly toxic drugs to kill cancer cells. It is often given after surgery to kill remaining cancer cells that could not be physically removed to reduce the risk of disease recurrence. Chemotherapy can take months to complete and can dramatically impact a patient’s quality of life. Patients usually experience a wide range of acute toxicities, including infection, pain in the mouth and throat, weight loss, fatigue, hair loss, rashes and injection site reactions. In addition, long-term effects of chemotherapy can include cognitive impairment, cardiac tissue damage, infertility, disease of the central nervous system, chronic fatigue, secondary malignancies and personality changes. Overall benefits of chemotherapy vary significantly across cancer populations, and the benefit of treatment may not always justify the cost of the therapy or the physical and mental burden patients endure.

          Use of Genomics to Understand Cancer

      Genomics is the study of complex sets of genes, their expression and their function in a particular organism. A gene is a set of instructions or information that is embedded in the DNA of a cell. For a gene to be turned on or “expressed” by a cell, the cell must first transcribe a copy of its DNA sequence into messenger RNA, which is then translated by the cell into protein. Proteins are large molecules that control most biological processes and make up molecular pathways, which cells use to carry out their specific functions.

      Genomics can also be used to understand diseases at the molecular level. Diseases can occur when mutated or defective genes inappropriately activate or block molecular pathways that are important for normal biological function. Disease can result from inheriting mutated genes or from developing mutations in otherwise normal cells. Such mutations can be the cause of cancer. The ability to detect a mutation or its functional results and to understand the process by which the mutation contributes to disease is crucial to understanding the molecular mechanisms of a disease.

      A common form of genomic analysis is the measurement of gene expression, or the presence and amount of one or more RNA sequences in a particular cell or tissue. Mutations may change the gene expression pattern of a cell as the cell responds to an altered genetic code. Quantifying the differences in gene expression has become a common way to study the behavior of an altered cell. This method allows for the measurement of the expression of single or multiple genes. These expression levels can be correlated with disease and clinical outcomes.

      Advances in genomic technology have accelerated the rate and lowered the cost of gene expression analysis, thus providing unprecedented opportunity for clinical utility. We believe gene expression technology has the potential to improve the quality of diagnosis and treatment of disease by arming patients and physicians with an understanding of disease at a molecular level that is specific to each patient.

      Cancer results from alterations in cells caused by the molecular changes of mutated genes. The behavior of cancer is dependent on many different genes and how they interact. Cancer is complicated and it may not be possible to identify a single gene that adequately signals a more aggressive or less aggressive type of

cancer. The ability to analyze multiple genes expressed by the tumor provides more valuable information, which enables individualized cancer assessment and treatment.

      The key to utilizing genomics in cancer is identifying specific sets of genes and gene interactions that are important for diagnosing different subsets of cancers. Studies can be performed which link response to therapy or the likelihood of recurrence to the pattern of gene expression in tumors. These results can then be used to develop tests that quantify gene expression of an individual’s tumor, allowing physicians to better understand what treatments are most likely to work for an individual patient or how likely a cancer is to recur.

Our Solution

      Our genomic-based diagnostic approach correlates gene expression information to clinical outcomes and provides information designed to improve treatment decisions for cancer patients. We have optimized technology for quantitative gene expression on FPE tissue by developing methods and processes for screening hundreds of genes at a time using minimal amounts of biopsy tissue. This technology allows us to analyze archived samples of tissue, retained by hospitals for most cancer patients, to correlate gene expression with known clinical outcomes. Once we have established and validated a test, we can then analyze a patient’s tumor and correlate the result to known clinical outcomes. As a result, each tumor’s gene expression can be quantified and correlated with responsiveness to therapy or the likelihood of cancer recurrence or progression. Onco type DX, our first clinically validated product, uses this quantitative molecular pathology approach to provide an individualized analysis of each patient’s tumor.

      We believe that our multi-gene analysis, as opposed to single-gene analysis, provides a more powerful approach to distinguish tumors as being more or less likely to recur or progress. Furthermore, as shown in breast cancer, our approach can be used to determine whether a cancer is more or less likely to respond to therapy. This information ultimately allows the physician and patient to choose a course of treatment that is individualized for each patient.

      Our solution fits within current clinical practice and therapeutic protocols, facilitating product adoption. We analyze tissues as they are currently handled, processed and stored by clinical pathology laboratories. Once a patient is diagnosed with breast cancer and a physician orders Onco type DX, the pathology lab provides us with the tumor block or thin sections from the biopsy specimen utilized for the diagnosis. Because the specimens are formalin fixed, they require no special handling and can be sent by overnight mail to our laboratory in California. We believe this provides an advantage over tests using fresh or frozen tissue that require special handling, such as shipping frozen tissue on dry ice. We typically analyze the tissue and deliver our results to the treating physician within 10 to 14 days of receipt of the tissue sample. This is within the crucial decision window after the tumor has been surgically removed and before the patient and the treating physician discuss additional treatment options.

      We believe our solution provides information that has the following benefits:

Business Strategy

      Our goal is to improve the quality of treatment decisions for cancer patients by providing individualized information to patients and their physicians through the genomic analysis of tumor biopsies. Key elements of our strategy include:

Our Product: Onco type DX

      Onco type DX uses quantitative molecular pathology to improve cancer treatment decisions. We offer Onco type DX as a clinical laboratory service, where we analyze tumor biopsy tissue samples in our laboratory and provide physicians with genomic information specific to the patient’s tumor. Early stage breast cancer is the first patient population where we have commercialized a genomic test that has been shown clinically to predict the likelihood of cancer recurrence, the likelihood of patient survival within 10 years of diagnosis and the likelihood of chemotherapy benefit.

      Our technology provides quantitative gene expression information for each patient’s tumor, which we refer to as an oncotype. When an oncotype is correlated with known clinical outcomes, it can be useful in predicting the likelihood of an individual patient’s tumor behavior. This allows the physician and patient to address key issues such as risk of disease recurrence or progression, likelihood of long-term survival and potential response to chemotherapy or other treatments. In breast cancer, we developed our gene panel by narrowing the field of the approximately 25,000 human genes down to 250 cancer-related genes through bioinformatic analysis of existing research literature and genomic databases. We evaluated the 250 genes in three independent clinical studies to identify a 21-gene panel whose composite gene expression profile can be represented by a single quantitative score, which we call a Recurrence Score. The higher the Recurrence Score, the more aggressive the tumor and the more likely it is to recur. The lower the Recurrence Score, the less aggressive the tumor and the less likely it is to recur. Moreover, we have demonstrated that the Recurrence Score also correlates with chemotherapy benefit, and we are undertaking further studies to support this finding.

      Approximately 230,000 new cases of breast cancer are expected to be diagnosed in the United States in 2005. Following diagnosis, a physician determines the stage of the breast cancer by examining the following:

Breast cancer tumors are classified as stage I, II, III or IV. Stage I and II are generally referred to as early stage breast cancer, and stage III and IV are generally referred to as late-stage breast cancer. Standard treatment guidelines weigh the stage of the cancer and additional factors to predict cancer recurrence and determine treatment protocol such as:

      Because these diagnostic factors have limited capability to predict future recurrence and response to chemotherapy and some are subjective, a large percentage of early stage breast cancer patients are classified as high risk. As a consequence, the use of chemotherapy has become standard practice in these patients even though the benefit to this patient group as a whole is small. Most early stage breast cancer patients have N-, ER+ tumors. These patients have been demonstrated to respond well to hormonal therapy such as tamoxifen. Identifying which of these patients will further benefit from chemotherapy is a difficult decision under current guidelines. A National Surgical Adjuvant Breast and Bowel Project, or NSABP, study published in 2004 showed that after 12 years of follow-up, overall survival in N-, ER+ breast cancer patients using tamoxifen hormonal therapy alone was approximately 83% and the overall survival using tamoxifen hormonal therapy and chemotherapy was 87%. Therefore, the incremental survival benefit of chemotherapy in this study was only 4%. Our test is designed to help identify those patients with aggressive tumors who are most likely to benefit from chemotherapy and identify those patients with less aggressive tumors who may receive minimal clinical benefit from chemotherapy.

      We believe that Onco type DX is the first genomic test that has clinical evidence supporting its ability to predict the likelihood of cancer recurrence, the likelihood of patient survival within 10 years of diagnosis and

the likelihood of chemotherapy benefit in early stage, N-, ER+ breast cancer patients treated with tamoxifen. Onco type DX is currently available at a list price of $3,460. We accept orders from all 50 states through our commercial laboratory located in Redwood City, California. Our laboratory is accredited under CLIA and by CAP.

      When the treating physician places an order for Onco type DX, the local pathology laboratory sends the tumor sample to our laboratory. Once we receive the tumor sample, it is logged in and processed by our pathology department. Suitable samples then undergo a process by which RNA is extracted and purified. We then analyze the resulting material and produce a report, typically within 10 to 14 days of the receipt of the sample, that shows a Recurrence Score on a continuum between 0-100. The Recurrence Score, along with other data and tests that physicians obtain, forms the basis for the treatment decision.

      The Recurrence Score has been clinically validated to correlate with an individual’s likelihood of breast cancer recurrence within 10 years of diagnosis. The lower the Recurrence Score the less likely the tumor is to recur and the higher the Recurrence Score the more likely the tumor is to recur. A Recurrence Score range from 0-100 correlates to an actual recurrence range from about 3% recurrence to over 30% recurrence for patients in our validation study using the NSABP Study B-14 population. This continuous range of scores differentiates Onco type DX from other tests that predict only high or low risk by providing an individualized level of risk. To evaluate our clinical validation studies and compare Onco type DX to other methods of classifying risk, we defined Recurrence Score ranges for low, intermediate and high risk groups. A Recurrence Score below 18 correlates with a low likelihood of recurrence; a Recurrence Score equal to or greater than 18 but less than 31 correlates with an intermediate likelihood of recurrence; and a Recurrence Score equal to or greater than 31 correlates with a high likelihood of recurrence. Within each risk category, Onco type DX further quantifies the risk for any given patient. For example, a low risk patient may have as low as a 3% likelihood of recurrence of breast cancer within 10 years or as high as a 11% likelihood of recurrence, depending on the individual Recurrence Score. We believe this represents a substantial improvement upon existing methods for classifying patient risk.

      We developed Onco type DX using a multi-step approach, conducting clinical studies on tumor specimens from more than 2,600 breast cancer patients. First, we developed methods, using RT-PCR, to quantify the expression of hundreds of genes in RNA isolated from fixed paraffin embedded tumor tissue. We then selected 250 cancer-related genes using bioinformatic analysis of genomic databases and our knowledge of cancer pathways. Third, we performed three independent breast cancer clinical studies in a total of 447 patients with known clinical outcomes to test the relationship between the expression of the 250 cancer-related genes and recurrence. Two of these studies were conducted at Providence Saint Joseph Medical Center and Rush University Medical Center, using samples from patients with N- and N+ tumors who received tamoxifen, chemotherapy or both. A third study was conducted in our specific target population of N-, ER+ patients treated with tamoxifen.

      From these studies we selected a panel of 21 genes, comprised of 16 cancer-related genes and five reference genes, with which we developed the Recurrence Score utilizing a number of biostatistical approaches. The Recurrence Score is obtained by first normalizing the expression of the cancer-related genes against the reference genes and then applying the Recurrence Score formula to calculate a single score scaled between 0-100.

          Clinical Validation of Prediction of Recurrence and Survival in N-, ER+ Patients Treated with Tamoxifen

      Our clinical validation studies were designed to answer two questions about the utility of the Recurrence Score when N-, ER+ breast cancer patients treated with tamoxifen make additional treatment decisions. First, we wanted to test whether Onco type DX could differentiate patients with a high likelihood of recurrence from patients with a low likelihood of recurrence. Second, we wanted to expand these results with a second study to demonstrate success in predicting breast cancer survival in a community hospital setting.

      Oncotype DX Predicts the Likelihood of Recurrence.  Our initial validation study was performed in 2003 in collaboration with the NSABP to determine whether Onco type DX predicts the likelihood of breast cancer recurrence. This study, which was reported at the San Antonio Breast Cancer Conference in December 2003 and published in The New England Journal of Medicine in December 2004, evaluated the ability of Onco type DX to quantify the likelihood of breast cancer recurrence over 10 years. The study involved 668 patients who were enrolled in the NSABP Study B-14 between 1982 and 1988. Each patient sample was analyzed in a blinded fashion and the results provided back to the NSABP through a neutral party at the University of Pittsburgh for analysis. The Recurrence Score was used to prospectively define the following three risk groups based on our clinical development studies described above:

The Recurrence Score was able to assign patients into high and low risk groups (p<0.001) and, when the Recurrence Score was examined together with age and tumor size in a multivariate analysis, only the Recurrence Score remained a significant predictor of patient outcome (p<0.001). A p-value is a mathematical test used to determine the validity of test results. Statistical significance is usually defined as a p-value of equal to or less than 0.05 with decreasing values providing increasing probability of validity. In this study we also demonstrated that the likelihood of distant recurrence at 10 years increased continuously as the Recurrence Score increased, with a range from about 3% recurrence for a Recurrence Score of zero to greater than 30% recurrence for patients in the high Recurrence Score category. In addition, in subgroup analysis of various ages, tumor sizes and pathology grade, the Recurrence Score remained a consistent predictor of distant recurrence.

      Oncotype DX Predicts the Likelihood of Breast Cancer Survival in a Community Hospital Setting.  In collaboration with Northern California Kaiser Permanente, we conducted a large, case-control, epidemiological study of breast cancer patients diagnosed from 1985 to 1994 at 14 Northern California Kaiser hospitals. This study was initially reported at the San Antonio Breast Cancer Conference in December 2004 and further detailed results were presented at the annual meeting of the American Society of Clinical Oncology, or ASCO, in May 2005. Patients who died of breast cancer, or the cases, were matched with up to three controls based on each case’s age, race and ethnicity, tamoxifen treatment, facility and diagnosis year. Controls had to be alive at the time of the corresponding case’s death in order to compare outcomes and availability of follow-up for those patients alive at time of each case death. To be eligible, patients had to be N-, less than 75 years old and not have received adjuvant chemotherapy. Among a potentially eligible population of 4,964 patients, we identified 220 eligible cases and 570 matched controls. Approximately one-third of the study patients were treated with tamoxifen. This study was performed to confirm that the Recurrence Score predicts breast cancer survival at 10 years in ER+ patients treated with tamoxifen. With respect to the group of ER+ patients treated with tamoxifen, the absolute risk of breast cancer death at 10 years in the pre-specified risk groups was 2.8% for the low risk group, 10.7% for the intermediate risk group and 15.5% for the high risk group. Additionally, in the larger population of ER+ patients untreated with tamoxifen, the Recurrence Score was also statistically significantly associated with breast cancer death at 10 years (p<0.025). This study, conducted in a community hospital setting, demonstrates that the Recurrence Score is independently associated with risk of breast cancer death and is able to identify subgroups of patients according to low, intermediate and high risk of death at 10 years.

      Additional studies were conducted to investigate three clinical and scientific questions:

      Oncotype DX Predicts the Likelihood of Chemotherapy Benefit.  We conducted a study in 2004 with the NSABP to determine whether Onco type DX is predictive of the likelihood of chemotherapy benefit. This study, which was reported initially at the San Antonio Breast Cancer Conference in December 2004 and further detailed results were presented at ASCO’s annual meeting in May 2005, included 651 patients from the NSABP Study B-20 with N-, ER+ breast cancer enrolled from 1988 to 1993. Of these patients, 227 were treated with tamoxifen alone and 424 were treated with tamoxifen plus chemotherapy. The results of this study demonstrated that low risk patients, as defined by the Recurrence Score, had a 96% recurrence-free survival rate at 10 years without chemotherapy compared with a 95% survival rate with chemotherapy, and intermediate risk patients as defined by the Recurrence Score had a 90% survival rate without chemotherapy compared with an 89% rate with chemotherapy. High risk patients as defined by the Recurrence Score had a 60% survival rate without chemotherapy compared with an 88% rate with chemotherapy (p<0.001). These results demonstrate that Onco type DX not only quantifies recurrence and survival risk but also correlates with the likelihood of chemotherapy benefit in early stage N-, ER+ breast cancer patients.

      Oncotype DX Predicts Likelihood of Recurrence Because it Predicts both Prognosis and Tamoxifen Benefit.  In 2004, we conducted an expanded study with the NSABP Study B-14 population to determine whether Onco type DX captures information regarding likelihood of distant recurrence, response to tamoxifen, or both. This study’s conclusions were reported at the San Antonio Breast Cancer Conference in December 2004 and further detailed results were presented at ASCO’s annual meeting in May 2005. The study included 645 patients with N-, ER+ breast cancer enrolled from 1982 to 1988, 355 of whom were given placebos and 290 of whom where treated with tamoxifen. The results of this study demonstrated that Onco type DX predicts the likelihood of distant disease recurrence in tamoxifen-treated patients with N-, ER+ breast cancer because it captures both prognosis and response to tamoxifen. Furthermore, this study of Onco type DX demonstrates that low and intermediate risk patients as defined by the Recurrence Score had the largest benefit of tamoxifen and high risk patients as defined by the Recurrence Score had minimal benefit of tamoxifen. The quantitative levels of ER, as defined by Onco type DX, varied by over two-hundred fold within the ER+ population and increasing levels of quantitative ER gene expression correlated with increasing response to tamoxifen. Finally, Onco type DX was able to discriminate between high and low risk patients in a subset of patients not treated with tamoxifen.

      Results Were Inconclusive as to Whether Oncotype DX Predicts Likelihood of Recurrence in a Mixed Population of N-, Untreated Patients.  In 2003, we conducted a trial with The M.D. Anderson Cancer Center to test the predictive power of Onco type DX in untreated breast cancer patients who were either ER- or ER+. This study was first reported at the San Antonio Breast Cancer Conference in December 2003 and published in Clinical Cancer Research in May 2005. Out of a pool of over 4,000 N- patient tissue samples, only 149 patients were untreated and had a sufficient tissue sample and RNA available to make them eligible for the study. The study population differed significantly from the NSABP Study B-14 treatment arm used for our initial validation study in that none of the patients were treated with tamoxifen, and the population included ER- and ER+ patients. This study did not demonstrate a significant predictive power for Onco type DX in untreated N- patients. Importantly, it also did not demonstrate the expected predictive power for other known predictive factors. For example, tumor grade inversely correlated with expected outcomes. Subsequent evaluations of Onco type DX in the NSABP Study B-14 placebo arm using samples from untreated ER+ patients and in the Kaiser Permanente population-based study using samples from untreated ER+ and ER- patients demonstrated a correlation between the Recurrence Score and recurrence and survival. These results

were reported at the San Antonio Breast Cancer Conference in December 2004 and ASCO’s annual meeting in May 2005.

      We are sponsoring third-party studies conducted by researchers affiliated with academic institutions to examine the health economic implications of Onco type DX. One such study, which was published in The American Journal of Managed Care , analyzed data from patients in the NSABP Study B-14 multi-center clinical trial to compare risk classification based on guideline criteria from the National Comprehensive Cancer Network, or NCCN, to risk classification by Onco type DX. Of the 668 patients in the NSABP study population, NCCN guidelines classified 615, or 92%, as high risk and 53, or 8%, as low risk. Of the 615 patients classified as high risk by NCCN, Onco type DX classified 49% as low risk, 22% as intermediate risk and 29% as high risk. Of the 53 patients that NCCN classified as low risk, Onco type DX classified 6% as high risk, 22% as intermediate risk and 72% as low risk. In each case, Onco type DX provided a more accurate classification of risk than the NCCN guidelines as measured by 10 year distant recurrence free survival.

      Based on these results, a model was designed to forecast quality-adjusted survival and expected costs, or the net present value of all costs of treatment until death, if Onco type DX was used in patients classified as low risk or high risk by NCCN guidelines. The model, when applied to a hypothetical population of 100 patients with the demographic and disease characteristics of the patients entered in the NSABP Study B-14, demonstrated an increase to quality-adjusted survival in this population of 8.6 years and a reduction in projected aggregate costs of approximately $200,000. Furthermore, the model showed that as the expected costs and anticipated toxicity of chemotherapy regimens increase, the use of the Recurrence Score to identify which patients would benefit from chemotherapy should lead to larger reductions in projected overall costs. According to this study, if all early stage breast cancer patients and their physicians used Onco type DX and acted on the information provided by the Recurrence Score, there would be significant economic benefit to the healthcare system.

New Product Development

      We developed Onco type DX using the following multi-phased clinical development platform that we intend to use in developing future products for breast and other cancers:

      Over 550,000 treatment decisions are expected to be made in the United States in 2005 for patients diagnosed with early stages of breast, colon, prostate, renal cell and lung cancers and melanoma. Early stage cancers are often treated with adjuvant treatments that are administered in conjunction with primary therapy, such as surgery and radiation, intended to prevent the recurrence of a particular cancer. The early stage patient population is generally the larger treatment population for most cancers. While our products under development focus on early stage disease, we consider product opportunities in late-stage disease when appropriate.

          Oncotype DX

      We are conducting clinical studies with Onco type DX to expand the market and improve certain product features. Approximately 65,000 patients are expected to be diagnosed in the United States in 2005 with N+ breast cancer and many may not benefit from chemotherapy or may have other health issues that increase the risk of chemotherapy treatment. Our early clinical research studies with Rush University Medical Center and Providence Saint Joseph Medical Center support further investigation of Onco type DX for this patient population. Additional studies are in the planning stage to investigate whether Onco type DX predicts the likelihood of recurrence in patients who use aromatase inhibitor-based therapies instead of tamoxifen. Aromatase inhibitors are a new class of hormonal therapy for ER+ breast cancer that has demonstrated a modest treatment benefit over tamoxifen. Additionally, we believe that individual gene scores which contribute to the Recurrence Score may have additional utility in predicting outcomes for specific therapies or disease subtypes. For example, a quantitative ER score may be a clinically useful predictor of response to tamoxifen based on our studies of the NSABP Study B-14 population.

          Second Generation Oncotype DX

      We are in the clinical research phase to investigate additional genes and gene combinations that may add to the predictive power of Onco type DX. A second generation product, if successful, could further refine and improve the classification of patients and result in better information for treatment decisions. We have identified multiple genes through research and development studies that, in varying combinations, may provide improved prediction of recurrence risk and likelihood of response to chemotherapy.

          Recurrence and Response Test for N-, ER- Breast Cancer

      We are in the clinical research phase to develop a product to predict the likelihood of recurrence and response to chemotherapy in N-, ER- breast cancer patients. This population is expected to represent approximately 30,000 patients in the United States in 2005. To date, we have conducted several clinical research studies that included N-, ER- breast cancer patients, and we plan to continue to explore opportunities in this population, including tests to better define ER- patients based on quantitative molecular pathology.

          Taxane Response Test

      We are also in the clinical research phase to develop a product to predict the response of patients to taxanes. Taxanes are a relatively new class of compounds that are used in addition to traditional chemotherapy regimens in some patients but have numerous side effects and are most often used in patients with aggressive or later stage tumors. The potential population for this product includes the estimated 65,000 N+ breast cancer patients as well as N-, ER-patients at high risk and N- patients at high risk in the United States in 2005. We have developed a number of hypotheses and selected a gene panel to investigate this product opportunity further.

          Product Development Opportunities in Other Cancers

      The following table describes our new products in development for cancers other than breast cancer:

          Colon Cancer Recurrence and Response Test

      We are in the clinical research phase of developing a test to predict the likelihood of recurrence and response to chemotherapy in patients with early stage colon cancer. Colon cancer is expected to affect approximately 120,000 individuals in the United States in 2005, of which approximately 65,000 early stage patients will need to decide whether or not to use chemotherapy for their cancer as well as which chemotherapy to use. Only a small percentage of colon cancer patients are expected to have a survival benefit from additional treatment after surgery. We have developed an investigational 758-gene panel for colon cancer and have established collaborative agreements with academic groups to access clinical samples correlated with outcome data.

          Prostate Cancer Progression and Recurrence Test

      We are in the clinical research phase of developing a test to predict the likelihood of progression and recurrence of prostate cancer in early stage patients. Approximately 250,000 men are expected to be diagnosed with prostate cancer in the United States in 2005, approximately 195,000 of whom will need to make critical decisions on whether or not to undergo surgery or radiation and on whether or not to have additional treatment after radiation. Because the side effects of surgery and local radiation therapy can be serious, a need exists for a reliable test to determine the likelihood of progression. There is also a need for a reliable test to determine the likelihood of recurrence after local treatment, because hormonal therapy and chemotherapy have significant side effects as well. We are in the process of defining our prostate cancer gene panel and have established a collaborative agreement with an academic group to conduct initial feasibility studies and to access clinical samples correlated with outcome data in prostate cancer.

          Renal Cell Cancer Recurrence and Response Test

      We are in the clinical research phase of developing a test to predict the likelihood of recurrence and response to therapy in renal cell cancer. Approximately 40,000 individuals are expected to be diagnosed with renal cell cancer in the United States in 2005. Recently reported studies suggest that some of these patients may respond to new treatments. We have conducted initial feasibility studies to extract RNA from renal cell cancer specimens and are currently working to define potential products for patients with renal cell cancer under a collaborative agreement with an academic group that has access to clinical samples correlated to outcome data.

          Non-small Cell Lung Cancer Recurrence and Response Test

      We are in the clinical research phase of developing a test to predict the likelihood of response to chemotherapy in early stage, non-small cell lung cancer. Approximately 155,000 individuals are expected to be diagnosed with non-small cell lung cancer in the United States in 2005, of which approximately 25,000 of those patients are expected to be diagnosed before the cancer spreads and will need to make chemotherapy treatment decisions. Recent clinical studies suggest that at least some of those early stage patients will benefit from chemotherapy. The use of chemotherapy in early stage non-small cell lung cancer is relatively recent and is likely to accelerate. We have completed initial feasibility studies in lung tissues as a part of our EGFR inhibitor program described below and are in the process of defining our lung cancer gene panel. We have a collaborative agreement with an academic group that has access to clinical samples correlated to outcome data.

          Melanoma Recurrence and Response Test

      We are in the research phase of developing a test to predict the likelihood of recurrence and response to therapy for patients with melanoma. Approximately 70,000 individuals are expected to be diagnosed with melanoma in the United States in 2005. Recently reported studies suggest that some of these patients may respond to new treatments. We are currently working to define potential products for melanoma under a collaborative agreement with an academic group that has access to clinical samples correlated to outcome data.

          Product Development Opportunities for Targeted Therapeutics

      New anti-cancer drugs in clinical development are designed to provide more targeted treatment which should improve efficacy and reduce side effects. A need exists to identify those patients who, based on the genomic profile of their tumors, are most likely to benefit from these therapies. We believe our individualized genomic analysis has the potential to improve patient selection for these therapies. We have had a number of discussions with pharmaceutical companies regarding the use of Onco type DX or our clinical development platform to identify subsets of patients more likely to respond to a particular therapy. We have completed several studies with different companies to evaluate our technology, and we are marketing our clinical development platform to pharmaceutical companies for exploratory clinical studies.

           Epidermal Growth Factor Receptor, or EGFR, Inhibitor Response Test

      We are in the clinical research phase to develop a test to predict the likelihood of response to the EGFR inhibitor class of drugs. The market opportunity of this test will initially be limited to metastatic disease in lung and colon cancer, with an estimated 60,000 patients in the United States in 2005, where such drugs are currently approved. We have conducted three small clinical correlation studies in lung cancer, colon cancer and head and neck cancer which allowed us to identify and file patent applications on a number of genes which may predict the response to EGFR inhibitors. Further clinical development may require partnerships with pharmaceutical companies that have access to appropriate clinical trial specimens.

      We cannot assure you that any of the above product opportunities or products in development will ever be commercialized or, if commercialized, will ever be successful.

          Research and Development Expenses

      Our research and development expenses were $7.1 million, $9.1 million and $10.0 million for the years ended December 31, 2002, 2003 and 2004, respectively, and $2.2 million for the three months ended March 31, 2005.

Technology

      We utilize existing technologies such as RT-PCR and information technologies and optimize and integrate them into new processes. We expect to continue to extend the capabilities of the various components of our process to develop effective products. Our technology allows us to:

          Extract RNA from FPE-tumor Biopsies

      Our product development requires that we be able to quantify the relative amounts of RNA in patients’ FPE tissue specimens. We have developed proprietary technology, intellectual property and know-how for optimized and automated methods for extraction and analysis of RNA from FPE tissue. Although others can extract RNA from FPE tissue, to our knowledge the process has not been optimized and scaled up for high-throughput clinical testing and large-scale clinical development studies involving large numbers of genes. Our process uses commercially available reagents and instruments with our own proprietary process and automation protocols, which results in RNA extraction from the range of tissues used in our clinical development studies and our commercial laboratory test.

          Amplify and Detect Diminished Amounts of RNA Consistently

      We use a well-established technology that we license from Roche called RT-PCR as the basis for our quantitative molecular pathology assays. This technology uses PCR along with fluorescent detection methods to quantify the relative amount of RNA in a biological specimen. Our technology platform has the following advantages:

          Analyze Hundreds of Genes

      Historically, RT-PCR has been used to screen one or, at most, a few genes at a time. The methods and know-how we have developed allow us to expand RT-PCR technology to a scale that enables screening of hundreds of genes at a time while using minimal amounts of biopsy tissue. During our initial years of operation, we typically screened 48 to 96 genes from a standard FPE tissue sample using RNA from three 10 micron sections of tissue. By 2003, we routinely screened 192 genes from each sample and, by 2004, we screened 384 genes per sample. Today, we have the capability to screen up to 768 different genes per sample without sacrificing the sensitivity, specificity and reproducibility of RT-PCR. With continued investment in miniaturization and automation, we believe that our technology will be capable of continued increases in throughput.

          Employ Advanced Information Technology, Bioinformatics and Biostatistics

      We have developed computer programs to automate our RT-PCR assay process. We have also developed a laboratory information management system to track our gene-specific reagents, instruments, assay processes and the data generated. Similarly, we have automated data analysis, storage and process quality control. We use statistical methods to optimize and monitor assay performance and to analyze data from our clinical research and development studies.

Reimbursement

      Revenues for clinical laboratory testing services may come from several sources, including commercial third-party payors, such as insurance companies and health maintenance organizations, government payors, such as Medicare and Medicaid, and patients.

      To gain broad reimbursement coverage, we are focusing on educating payors on the following Onco type DX attributes:

      As a relatively new test, Onco type DX may be considered investigational by payors and not covered under their reimbursement policies. Consequently, we have pursued case-by-case reimbursement and expect the test will continue to be reviewed on this basis until policy decisions have been made by individual payors.

We are also working with public and private payors and health plans to secure coverage for Onco type DX based upon clinical evidence showing the utility of the test. We currently have contracts in place with commercial third-party payors covering over nine million lives, or approximately 3.1% of the United States population. We believe that it may take one or more years to achieve successful reimbursement with a majority of payors. However, we cannot predict whether, or under what circumstances, payors will reimburse our products. Payment amounts can also vary across individual policies. Denial of coverage by payors, or reimbursement at inadequate levels, would have a material adverse impact on market acceptance of our products. The current situation with our primary payors is as follows:

      Commercial Third-party Payors and Patient Pay.  Where there is a payor policy in place, we bill the payor and the patient in accordance with the established policy. Where there is no payor policy in place, we pursue reimbursement on behalf of each patient on a case-by-case basis. We request that physicians have a billing conversation with patients prior to a test being submitted to discuss the patient’s responsibility should their policy not cover the test. We also request that the physician inform the patient that we will take on the primary responsibility for obtaining third-party reimbursement on behalf of patients, including appeals for initial denials, prior to billing a patient. With this practice established, we believe that most patients receiving the Onco type DX test have agreed to the test knowing that they may be responsible for all or some portion of the cost of the test should their medical insurer deny or limit coverage. Our efforts on behalf of patients will take a substantial amount of time, and bills may not be paid for many months, if at all. Furthermore, if a third-party payor denies coverage after final appeal, it may take a substantial amount of time to collect from the patient, and we may not be successful.

      In early 2005, the Medical Advisory Panel of the Blue Cross and Blue Shield Association’s Technology Evaluation Center, or BCBSA, a technology assessment group, concluded that the existing clinical data in support of Onco type DX does not meet the panel’s technology criteria for clinical effectiveness and appropriateness. This assessment is provided for informational purposes to members of BCBSA and can be used by third-party payors and health care providers such as Blue Cross and Blue Shield, which provide healthcare coverage for nearly one-third of all Americans, as grounds to deny coverage for Onco type DX.

      Medicare and Medicaid.  In December 2004, the Medicare contractor with responsibility for processing and paying claims submitted by us announced that it would not provide coverage for Onco type DX for Medicare beneficiaries. It also indicated that there could be some questions concerning whether the hospital must bill Medicare or we can bill Medicare directly. Finally, it questioned which Medicare contractor has jurisdiction to determine coverage for Medicare claims for our test. To date, there is no national coverage determination on Onco type DX by Medicare, which means coverage is left to the discretion of the local Medicare contractor. Since the local Medicare contractor responsible for processing claims submitted by us has announced that it would not provide coverage for Onco type DX, this has resulted in the denial of Medicare claims. We are appealing these denials through established processes and working with Medicare to establish coverage for our test. During this time, we are not billing Medicare patients directly. However, if we are not successful in establishing coverage through Medicare, we may change this policy and bill future Medicare patients directly, subject to obtaining from patients their advance written consent. We believe that as much as 20% of our future market for Onco type DX may be derived from patients covered by Medicare.

Selling and Marketing

      Our selling and marketing strategy targets the oncology community, primarily medical and surgical oncologists. Our direct sales approach focuses on the clinical and economic benefits of Onco type DX and the scientific validation supporting our product. As of May 31, 2005, our selling and marketing team consisted of 34 employees. Our field staff has significant clinical oncology selling and marketing experience from leading biopharmaceutical, pharmaceutical and specialty reference laboratory companies.

      Our marketing strategy focuses on educating physicians, laboratory personnel and other healthcare professionals regarding the development of new genomic technologies and the value of the quantitative information Onco type DX provides. We also work closely with national and regional patient advocacy organizations that are focused on breast cancer care. Our customer service representatives are trained to

handle inquiries from physicians, patients and other healthcare-providers. We also utilize the Internet for communicating with external constituencies, and our web site contains clinical information for healthcare professionals and educational information for breast cancer patients.

      We promote our product through marketing channels commonly used by the biopharmaceutical and pharmaceutical industries, such as sponsored continuing medical education, medical meeting participation and broad-based publication of our scientific and economic data.

Competition

      We believe that we compete primarily on the basis of:

      We believe that we compete favorably with respect to these factors, although we cannot assure you that we will be able to continue to do so in the future or that new products that perform better than Onco type DX will not be introduced. We believe that our continued success depends on our ability to:

      Currently, our principal competition comes from existing diagnostic methods utilized by pathologists and oncologists, which generally involve assessing and evaluating the grade and stage of cancerous tumors when determining risk of recurrence. These methods, which have been used for many years and are therefore difficult to change or supplement, are typically accomplished in a short period of time without much expense. In addition, companies offering capital equipment and inexpensive kits or reagents to local pathology laboratories represent another source of potential competition. These kits are used directly by the pathologist, which facilitates adoption more readily than tests like Onco type DX that are performed outside the pathology laboratory. In addition, few diagnostic methods are as expensive as Onco type DX and others may develop lower-priced, less complex tests that could be viewed as the equivalent of ours.

      We also face competition from companies such as Agendia B.V. which offer products or have conducted research to profile gene expression in breast cancer using fresh or frozen tissue. Commercial laboratories with strong distribution networks for diagnostic tests, such as Genzyme Corporation, Laboratory Corporation of

America Holdings and Quest Diagnostics Incorporated, may become competitors. Other potential competitors include companies that develop diagnostic tests such as Bayer Healthcare LLC, Celera Genomics, a division of Applera Corporation, Roche Diagnostics, a division of F. Hoffmann-La Roche Ltd, and Veridex LLC, a Johnson & Johnson company, as well as academic and research institutions.

      Many of our present and potential competitors have widespread brand recognition and substantially greater financial and technical resources and development, production and marketing capabilities than we do. Competition among these entities to recruit and retain highly qualified scientific, technical and professional personnel and consultants is also intense. If we are unable to compete successfully against current or future competitors, we may be unable to increase market acceptance for and sales of our test, which could prevent us from increasing or sustaining our revenues, or achieving or sustaining profitability.

Regulation

          Medicare and Medicaid Coverage

      In determining whether or not Medicare will pay for a service, the Centers for Medicare and Medicaid Services, or CMS, which oversees Medicare, can permit the contractors, who process and pay Medicare claims, to make that determination or it can make a national coverage determination, which will bind all Medicare contractors. To date, CMS has not issued a national coverage determination on Onco type DX. As a result, whether or not Medicare will cover the test is the decision of NHIC California, the current local Medicare carrier for California and the contractor with jurisdiction to process claims submitted by us. In December 2004, NHIC published an article on its website stating that it would not provide coverage for Onco type DX for Medicare beneficiaries and that there could be some questions concerning which provider must bill Medicare for the test and which contractor had jurisdiction to determine coverage for Medicare claims for our test.

      In addition, in February 2005, CMS issued a notice that would affect how the date of service for a laboratory service is determined. Based on this notice, CMS could determine that the date of service for the test is the date on which the patient’s initial surgery was performed. As a result, Onco type DX could be considered a hospital service, which would mean that we would be required to bill and be paid by the hospital and that the hospital would be required to bill Medicare for the test as part of its prospective payment for the admission or encounter during which the surgery was performed. This could also result in lower reimbursement rates in the event coverage is provided.

      We intend to work with Medicare to establish coverage for Medicare beneficiaries, and we are also working with NHIC California and CMS to resolve the issue concerning which entity is required to bill Medicare for our tests. In the meantime, we intend to appeal denials received on a case-by-case basis. We cannot provide any assurance, however, that coverage will be provided in the future by Medicare, that our appeals will ultimately be successful or that other issues will be favorably resolved. In addition, each state Medicaid program, which pays for services furnished to the eligible medically indigent, will usually make its own decision whether or not to cover Onco type DX. To date, no state Medicaid program has decided to pay for the test.

          Payment

      Clinical laboratory testing services, when covered by third-party payors, are paid under various methodologies, including prospective payment systems and fee schedules. Under Medicare, payment is generally made under the Clinical Laboratory Fee Schedule with amounts assigned to specific procedure billing codes. Each Medicare carrier jurisdiction has a fee schedule that establishes the price for each specific laboratory billing code. The Social Security Act establishes that these fee schedule amounts are to be increased annually by the percentage increase in the consumer price index, or CPI, for the prior year. Congress has frequently legislated that the CPI increase not be implemented. In the Medicare Prescription Drug, Improvement and Modernization Act, or MMA, Congress eliminated the CPI update through 2008. In addition, the National Limitation Amount, or NLA, which acts as a ceiling on Medicare reimbursement, is set at a percentage of the median of all the carrier fee schedule amounts for each test code. In the past, Congress

has frequently lowered the percentage of the median used to calculate the NLA in order to achieve budget savings. Currently, the NLA ceiling is set at 74% of the medians for established tests and 100% of the median for diagnostic tests for which no limitation amount was established prior to 2001. Thus, no carrier can pay more than the NLA amount for any specific code.

      At the present time, there is no specific code to report Onco type DX. Therefore, the test must be reported under a non-specific unlisted procedure code, which is subject to manual review of each claim. Furthermore, because there is no specific code for this test, there is also no set Medicare payment amount. We are moving forward with plans to obtain procedure coding, but there is no assurance that a specific procedure code will be adopted or that adequate payment will be assigned if and when a code is adopted.

      Several provisions of the Medicare Prescription Drug, Improvement and Modernization Act of 2003 may affect future payments for clinical laboratory testing services, including Onco type DX. First, the Clinical Laboratory Fee Schedule payments under Medicare are frozen through 2008 with zero-percent annual adjustment. This would affect Medicare and Medicaid payments for Onco type DX if a specific procedure code and Clinical Laboratory Fee Schedule payment are assigned to the test. Second, Congress authorized the Medicare program to conduct a demonstration project on applying competitive bidding to certain clinical laboratory tests. It is not clear whether competitive bidding will be applied more broadly to clinical laboratory services under Medicare at some point in the future and, if so, whether this would impact payment for Onco type DX, which is provided solely by us. Third, Medicare is reforming the local contractor process to replace current contracts with fiscal intermediaries, which are billed by hospitals and other institutional providers, and carriers, which are billed by physicians, independent laboratories and other suppliers, with new contracts. These reforms may result in a change in the contractors to whom we send Medicare claims, which may affect coverage for Onco type DX. Finally, on several occasions, including in 2003 during the negotiations over the MMA, Congress has considered imposing a 20% co-insurance amount on clinical laboratory services, which would require beneficiaries to pay a portion of the cost of their clinical laboratory testing. Although that requirement has not been enacted at this time, Congress could decide to impose such an obligation at some point in the future. If so, it could make it more difficult for us to collect payment for Onco type DX.

          Clinical Laboratory Improvement Amendments of 1988

      As a clinical laboratory, we are required to hold certain federal, state and local licenses, certifications and permits to conduct our business. Under CLIA, we are required to hold a certificate applicable to the type of work we perform and to comply with standards covering personnel, facilities administration, quality systems and proficiency testing.

      We have a certificate of accreditation under CLIA to perform testing and are accredited by CAP. To renew our CLIA certificate, we are subject to survey and inspection every two years to assess compliance with program standards. The standards applicable to the testing which we perform may change over time. We cannot assure you that we will be able to operate profitably should regulatory compliance requirements become substantially more costly in the future.

      If our laboratory is out of compliance with CLIA requirements, we may be subject to sanctions such as suspension, limitation or revocation of our CLIA certificate, as well as directed plan of correction, state on-site monitoring, civil money penalties, civil injunctive suit or criminal penalties. We must maintain CLIA compliance and certification to be eligible to bill for services provided to Medicare beneficiaries. If we were to be found out of compliance with CLIA program requirements and subjected to sanction, our business could be harmed.

          Food and Drug Administration

      The U.S. Food and Drug Administration, or the FDA, regulates the sale or distribution, in interstate commerce, of medical devices, including in vitro diagnostic test kits. Devices subject to FDA regulation must undergo premarket review prior to commercialization unless the device is of a type exempted from such review. In addition, manufacturers of medical devices must comply with various regulatory requirements

under the Federal Food, Drug and Cosmetic Act and regulations promulgated under that Act, including quality system review regulations, unless exempted from those requirements for particular types of devices. Entities that fail to comply with FDA requirements can be liable for criminal or civil penalties, such as recalls, detentions, orders to cease manufacturing and restrictions on labeling and promotion.

      Clinical laboratory services are not subject to FDA regulation, but in vitro diagnostic test kits and reagents and equipment used by these laboratories may be subject to FDA regulation. Clinical laboratory tests that are developed and validated by a laboratory for use in examinations the laboratory performs itself are called “home brew” tests. Most home brew tests currently are not subject to premarket review by FDA although analyte-specific reagents or software provided to us by third parties and used by us to perform home brew tests may be subject to review by the FDA prior to marketing. We believe that Onco type DX is a type of home brew test. We believe that Onco type DX is not subject to regulation under current FDA policies and have communicated this conclusion to FDA staff. We believe that the container we provide for collection and transport of tumor samples from a pathology laboratory to our laboratory is a medical device subject to FDA regulation but exempt from premarket review. At this time, we believe the FDA is reviewing the regulatory status of Onco type DX. We cannot provide any assurance that FDA regulation, including premarket review, will not be required in the future for Onco type DX. If premarket review is required, this would adversely affect our business until such review is completed and approval or clearance to market is obtained. If premarket review is required by the FDA, there can be no assurance that our test will be cleared or approved on a timely basis, if at all. Ongoing compliance with FDA regulations would increase the cost of conducting our business, subject us to inspection by the FDA and to the requirements of the FDA and penalties for failure to comply with the requirements of the FDA. Should any of the clinical laboratory device reagents obtained by us from vendors and used in conducting our home brew test be affected by future regulatory actions, we could be adversely affected by those actions, including increased cost of testing or delay, limitation or prohibition on the purchase of reagents necessary to perform testing.

          Health Insurance Portability and Accountability Act

      Under the federal Health Insurance Portability and Accountability Act of 1996, or HIPAA, the U.S. Department of Health and Human Services, or HHS, has issued regulations to protect the privacy and security of protected health information used or disclosed by health care providers, such as us. HIPAA also regulates standardization of data content, codes and formats used in health care transactions and standardization of identifiers for health plans and providers. Penalties for violations of HIPAA regulations include civil and criminal penalties.

      We have developed policies and procedures to comply with these regulations by the respective compliance enforcement dates. The requirements under these regulations may change periodically and could have an effect on our business operations if compliance becomes substantially more costly than under current requirements.

      In addition to federal privacy regulations, there are a number of state laws governing confidentiality of health information that are applicable to our operations. New laws governing privacy may be adopted in the future as well. We have taken steps to comply with health information privacy requirements to which we are aware that we are subject. However, we can provide no assurance that we are or will remain in compliance with diverse privacy requirements in all of the jurisdictions in which we do business. Failure to comply with privacy requirements could result in civil or criminal penalties, which could have a materially adverse impact on our business.

          Federal and State Self-referral Prohibitions

      We are subject to the federal self-referral prohibitions commonly known as the Stark Law, and to similar restrictions under California’s Physician Ownership and Referral Act, commonly known as PORA. Together these restrictions generally prohibit us from billing a patient or any governmental or private payor for any test when the physician ordering the test, or any member of such physician’s immediate family, has an investment

interest in, or compensation arrangement with, us, unless the arrangement meets an exception to the prohibition.

      Both the Stark Law and PORA contain an exception for referrals made by physicians who hold investment interests in a publicly traded company that has stockholders’ equity of $75 million at the end of its most recent fiscal year or on average during the previous three fiscal years, and which satisfies certain other requirements. In addition, both the Stark Law and PORA contain an exception for compensation paid to a physician for personal services rendered by the physician. We have compensation arrangements with a number of physicians for personal services, such as speaking engagements and specimen tissue preparation. We have structured these arrangements with terms intended to comply with the requirements of the personal services exception to Stark and PORA. However, we can not be certain that regulators would find these arrangements to be in compliance with Stark, PORA or similar state laws. We would be required to refund any payments we receive pursuant to a referral prohibited by these laws to the patient, the payor or the Medicare program, as applicable.

      Sanctions for a violation of the Stark Law include the following:

These prohibitions apply regardless of the reasons for the financial relationship and the referral. No finding of intent to violate the Stark Law is required for a violation. In addition, under an emerging legal theory, knowing violations of the Stark Law may also serve as the basis for liability under the Federal False Claims Act.

      Further, a violation of PORA is a misdemeanor and could result in civil penalties and criminal fines. Finally, other states have self-referral restrictions with which we have to comply that differ from those imposed by federal and California law. While we have attempted to comply with the Stark Law, PORA and similar laws of other states, it is possible that some of our financial arrangements with physicians could be subject to regulatory scrutiny at some point in the future, and we cannot provide an assurance that we will be found to be in compliance with these laws following any such regulatory review.

          Federal and State Anti-kickback Laws

      Federal Anti-kickback Law makes it a felony for a provider or supplier, including a laboratory, to knowingly and willfully offer, pay, solicit or receive remuneration, directly or indirectly, in order to induce business that is reimbursable under any federal health care program. A violation of the Anti-kickback Law may result in imprisonment for up to five years and fines of up to $250,000 in the case of individuals and $500,000 in the case of organizations. Convictions under the Anti-kickback Law result in mandatory exclusion from federal health care programs for a minimum of five years. In addition, HHS has the authority to impose civil assessments and fines and to exclude health care providers and others engaged in prohibited activities from the Medicare, Medicaid and other federal health care programs.

      Actions which violate the Anti-kickback Law or similar laws may also involve liability under the Federal False Claims Act, which prohibits the knowing presentation of a false, fictitious or fraudulent claim for payment to the U.S. Government. Actions under the False Claims Act may be brought by the Department of Justice or by a private individual in the name of the government.

      Although the Anti-kickback Law applies only to federal health care programs, a number of states, including California, have passed statutes substantially similar to the Anti-kickback Law pursuant to which similar types of prohibitions are made applicable to all other health plans and third-party payors. California’s

anti-kickback statute, commonly referred to as Section 650, has been interpreted by the California Attorney General and California courts in substantially the same way as the HHS and the courts have interpreted the Anti-kickback Law. A violation of Section 650 is punishable by imprisonment and fines of up to $50,000.

      Federal and state law enforcement authorities scrutinize arrangements between health care providers and potential referral sources to ensure that the arrangements are not designed as a mechanism to induce patient care referrals and opportunities. The law enforcement authorities, the courts and Congress have also demonstrated a willingness to look behind the formalities of a transaction to determine the underlying purpose of payments between health care providers and actual or potential referral sources. Generally, courts have taken a broad interpretation of the scope of the Anti-kickback Law, holding that the statute may be violated if merely one purpose of a payment arrangement is to induce future referrals.

      In addition to statutory exceptions to the Anti-kickback Law, regulations provide for a number of safe harbors. If an arrangement meets the provisions of a safe harbor, it is deemed not to violate the Anti-kickback Law. An arrangement must fully comply with each element of an applicable safe harbor in order to qualify for protection. There are no safe harbors to California’s Section 650.

      Among the safe harbors that may be relevant to us is the discount safe harbor. The discount safe harbor applies to discounts provided by providers and suppliers, including laboratories, to clients with respect to Medicare, Medicaid, private pay or HMO patients, where the referring physician bills the payor for the test, not when the service provider bills the payor directly. If the terms of the discount safe harbor are met, the discounts will not be considered prohibited remuneration under the Anti-kickback Law.

      California does not have a discount safe harbor. However, certain licensees, such as hospitals or physicians, may only mark-up laboratory tests purchased by those licensees from a laboratory if certain disclosures are made to patients and third party payors regarding the mark-up. Therefore, if and when we elect to offer discounts to California customers, including any hospital or physician, such discounts would not likely be viewed by regulators as prohibited under Section 650 because the mark-up would be disclosed by the customer to its buyer under California’s mark-up laws. In contrast, any such discounts provided by us to our non-California customers would have to be analyzed under California’s Section 650.

      The personal services safe harbor to the Anti-kickback Law provides that remuneration paid to a referral source for personal services will not violate the Anti-kickback Law provided all of the elements of that safe harbor are met. One element is that, if the agreement is intended to provide for the services of the physician on a periodic, sporadic or part-time basis, rather than on a full-time basis for the term of the agreement, the agreement specifies exactly the schedule of such intervals, their precise length, and the exact charge for such intervals. Our personal services arrangements with some physicians did not meet the specific requirement of this safe harbor that the agreement specify exactly the schedule of the intervals of time to be spent on the services because the nature of the services, for example, speaking engagements, does not lend itself to exact scheduling and therefore meeting this element of the personal services safe harbor is impractical. Failure to meet the terms of the safe harbor does not render an arrangement illegal. Rather, an arrangement would not have the protections of the safe harbor if challenged by a regulator and, if necessary, the parties might be required to demonstrate why the arrangement does not violate the Anti-kickback Law.

      While we believe that we are in compliance with the Anti-kickback Law and Section 650, there can be no assurance that our relationships with physicians, hospitals and other customers will not be subject to investigation or a successful challenge under such laws. If imposed for any reason, sanctions under the Anti-kickback Law and Section 650 could have a negative effect on our business.

          Other Federal Fraud and Abuse Laws

      In addition to the requirements that are discussed above, there are several other health care fraud and abuse laws that could have an impact on our business. For example, provisions of the Social Security Act permit Medicare and Medicaid to exclude an entity that charges the federal health care programs substantially in excess of its usual charges for its services. The terms “usual charge” and “substantially in excess” are ambiguous and subject to varying interpretations.

      Further, the Federal False Claims Act prohibits a person from knowingly submitting a claims or making a false record or statement in order to secure payment by the federal government. In addition to actions initiated by the government itself, the statute authorizes actions to be brought on behalf of the federal government by a private party having knowledge of the alleged fraud. Because the complaint is initially filed under seal, the action may be pending for some time before the defendant is even aware of the action. If the government is ultimately successful in obtaining redress in the matter or if the plaintiff succeeds in obtaining redress without the government’s involvement, then the plaintiff will receive a percentage of the recovery. Finally, the Social Security Act includes its own provisions that prohibit the filing of false claims or submitting false statements in order to obtain payment. Violation of these provisions may result in fines, imprisonment or both, and possible exclusion from Medicare or Medicaid.

          California Laboratory Licensing

      In addition to federal certification requirements of laboratories under CLIA, licensure is required and maintained for our laboratory under California law. Such laws establish standards for the day-to-day operation of a clinical laboratory, including the training and skills required of personnel and quality control. In addition, California laws mandate proficiency testing, which involves testing of specimens that have been specifically prepared for the laboratory.

      If our laboratory is out of compliance with California standards, the California Department of Health Services, or DHS, may suspend, restrict or revoke our license to operate our laboratory; assess substantial civil money penalties; or impose specific corrective action plans. Any such actions could materially affect our business. We maintain a current license in good standing with DHS. However, we cannot provide assurance that DHS will at all times in the future find us to be in compliance with all such laws.

          New York Laboratory Licensing

      Because we receive specimens from New York state, our clinical laboratory is required to be licensed by New York. We maintain such licensure for our laboratory under New York state laws and regulations, which establish standards for:

      New York law also mandates proficiency testing for laboratories licensed under New York state law, regardless of whether or not such laboratories are located in New York. If a laboratory is out of compliance with New York statutory or regulatory standards, the New York State Department of Health, or the DOH, may suspend, restrict or revoke the laboratory’s New York license or assess civil money penalties. Statutory or regulatory noncompliance may result in a laboratory’s being found guilty of a misdemeanor under New York law. Should we be found out of compliance with New York laboratory requirements, we could be subject to such sanctions, which could harm our business. We maintain a current license in good standing with the DOH. However, we cannot provide assurance that the DOH will at all times find us to be in compliance with all such laws.

          Other States’ Laboratory Testing

      New Jersey may require out-of-state laboratories which accept specimens from New Jersey to be licensed in New Jersey. We have spoken with regulators at the New Jersey Department of Health and Senior Services, Clinical Laboratory Improvement Services, and are following their instructions with respect to compliance with New Jersey law. In addition, Florida, Maryland, Pennsylvania and Rhode Island require out-of-state

laboratories which accept specimens from those states to be licensed. We have obtained licenses in those four states and believe we are in compliance with applicable licensing laws.

      From time to time, we may become aware of other states that require out of state laboratories to obtain licensure in order to accept specimens from the state, and it is possible that other states do have such requirements or will have such requirements in the future. If we identify any other state with such requirements or if we are contacted by any other state advising us of such requirements, we intend to follow instructions from the state regulators as to how we should comply with such requirements.

Patents and Proprietary Technology

      In order to remain competitive, we must develop and maintain protection on the proprietary aspects of our technologies. We rely on a combination of patents, copyrights, trademarks, trade secret laws and confidentiality, material data transfer agreements, licenses and invention assignment agreements to protect our intellectual property rights. We also rely upon unpatented trade secrets and improvements, unpatented know-how and continuing technological innovation to develop and maintain our competitive position. We generally protect this information with reasonable security measures.

      As of May 31, 2005, we had 16 pending U.S. patent applications, including provisional and non-provisional filings. 10 of these U.S. patent applications also have corresponding pending applications under the Patent Cooperation Treaty. We have filed one of our patent applications nationally in Canada, Europe and Japan. We have filed a second patent application in Canada, Australia, Europe and Japan.

      In these patent applications, we have either sole or joint ownership positions. In those cases where joint ownership positions were created, we have negotiated contractual provisions providing us with the opportunity to acquire exclusive rights under the patent applications. Under three patent applications, we have elected to allow exclusive options to lapse without exercising the option. The joint ownership agreements generally are in the form of material data transfer agreements that were executed at the onset of our collaborations with third parties.

      Our patent applications relate to two main areas: gene expression technology methods, and gene markers for cancer recurrence and drug response in certain forms of cancer. We intend to file additional patent applications in the United States and abroad to strengthen our intellectual property rights. Our patent applications may not result in issued patents, and we cannot assure you that any patents that might issue will protect our technology. Any patents issued to us in the future may be challenged by third parties as being invalid or unenforceable, or third parties may independently develop similar or competing technology that are not covered by our patents. We cannot be certain that the steps we have taken will prevent the misappropriation of our intellectual property, particularly in foreign countries where the laws may not protect our proprietary rights as fully as in the United States.

      From time to time, we may receive, notices of claims of infringement, misappropriation or misuse of other parties’ proprietary rights. Some of these claims may lead to litigation. We cannot assure you that we will prevail in these actions, or that other actions alleging misappropriation or misuse by us of third-party trade secrets, infringement by us of third-party patents and trademarks or the validity of patents issued to us in the future, will not be asserted or prosecuted against us, or that any assertions of misappropriation, infringement or misuse or prosecutions seeking to establish the validity of our patents will not materially or adversely affect our business, financial condition and results of operations.

      An adverse determination in litigation or interference proceedings to which we may become a party relating to any patents issued to us in the future or any patents owned by third parties could subject us to significant liabilities to third parties or require us to seek licenses from third parties. Furthermore, if we are found to willfully infringe these patents, we could, in addition to other penalties, be required to pay treble damages. Although patent and intellectual property disputes in this area have often been settled through licensing or similar arrangements, costs associated with such arrangements may be substantial and could include ongoing royalties. We may be unable to obtain necessary licenses on satisfactory or commercially feasible terms, if at all. If we do not obtain necessary licenses, we may not be able to redesign Onco type DX

or other of our tests to avoid infringement, or such redesign may take considerable time, and force us to reassess our business plans. Adverse determinations in a judicial or administrative proceeding or failure to obtain necessary licenses could prevent us from manufacturing and selling Onco type DX or other of our tests, which would have a significant adverse impact on our business.

      All employees and technical consultants working for us are required to execute confidentiality agreements in connection with their employment and consulting relationships with us. Confidentiality agreements provide that all confidential information developed or made known to others during the course of the employment, consulting or business relationship shall be kept confidential except in specified circumstances. Agreements with employees provide that all inventions conceived by the individual while employed by us are our exclusive property. We cannot provide any assurance that employees and consultants will abide by the confidentiality or assignment terms of these agreements. Despite measures taken to protect our intellectual property, unauthorized parties might copy aspects of our technology or obtain and use information that we regard as proprietary.

Roche License

      We have obtained from Roche Molecular Systems, Inc. a non-exclusive license under certain U.S. patents claiming certain nucleic acid amplification processes known as polymerase chain reaction, or PCR, homogeneous polymerase chain reaction and reverse transcription polymerase chain reaction, or RT-PCR. The Roche license is limited to the performance of clinical laboratory services within the United States and Puerto Rico. The Roche license does not include the right to make or sell products using the patented processes. The term of the Roche license is coextensive with the patent rights licensed and is subject to termination by Roche under limited circumstances, including a change in the ownership or control of our company and our material uncured breach of the agreement. Under the license, we pay Roche a royalty based on our net revenues.

Employees

      As of May 31, 2005, we had 101 full-time employees and three part-time employees, including 18 in research and development, 16 in commercial operations, 11 in medical, three in regulatory, 34 in selling and marketing, 11 in information technology and 11 in administrative functions. None of our employees are covered by collective bargaining arrangements, and our management considers its relationships with our employees to be good.

Properties

      We currently lease approximately 25,000 square feet of laboratory and office space in Redwood City, California under leases that expire in February 2006. We believe that our existing facilities are adequate to meet our business requirements for the near-term and that additional space will be available on commercially reasonable terms, if required.

Legal Proceedings

      From time to time, we may be party to lawsuits in the ordinary course of business. We are currently not a party to any legal matters.