Medical Device Link.

MOLECULAR DIAGNOSTICS

Figure 1. Examining the raw data from the 92-gene RT-PCR CancerType ID test by AviaraDX (San Diego).

A mantra that scientists have strived to uphold is “underpromise and overdeliver.” However, society is far from benefiting from a plethora of clinically relevant genomic tests that were supposed to transform individualized treatments. Instead, seven years after the landmark reporting of the Human Genome Project, there is a paucity of genomic tests that have any real clinical relevance, especially in oncology. The major reason has been the time and effort required to validate and prove the clinical utility for a given genomic classifier, and the uncertainty regarding the changing regulatory path.

In 2008, approximately half a million people in the United States will die from cancer. The majority of these deaths will result from solid tumors such as lung, colorectal, breast, pancreas, and prostate cancers. In the postgenomic era, translational researchers have put in an enormous effort to identify molecular-based prognostic and predictive factors that will help physicians make better decisions to individualize treatments. Given this effort, the availability of a large number of clinically useful tests would be expected. However, such availability is not the case, and the question is why.

This article examines the reasons for the delay in development, and argues that while clinically relevant genomic classifiers are indeed emerging, the development timelines have been and will continue to be longer than anticipated.

Discovering Genomic Classifiers

The invention in the 1990s of oligonucleotide, or cDNA, microarrays for gene expression analysis allowed the simultaneous measurement of tens of thousands of mRNAs, which enabled gene expression profiling of the entire human genome for any given tumor sample. In a groundbreaking study, Lander demonstrated that gene expression profiling could classify different cancer types.¹ These investigators developed a gene expression signature or genomic classifier that could discriminate or differentially classify acute myeloid leukemia from acute lymphoblastic leukemia. Since this seminal study, many other researchers have conducted whole-genome gene expression profiles of various cancers in search of diagnostic, prognostic, and drug response (predictive) classifiers.

For example, in breast cancer, Botstein identified five distinct and reproducible subtypes with differing clinical outcome: basal-like, Her2, normal breast-like, luminal A, and luminal B/C.² In addition, an extensive number of gene expression profiling studies have been conducted to classify lymphomas. Most notable are studies conducted by Staudt, which identified a genomic classifier that distinguishes Burkitt’s lymphoma from diffuse large-B-cell lymphoma (DLBCL) and a classifier that identifies three distinct types of DLBCL with different clinical outcomes: germinal center B-cell-like, activated B-cell-like, and primary mediastinal B-cell.³

In lung cancer, Meyerson demonstrated that gene expression profiling could reproducibly identify three human lung adenocarcinoma subtypes in multiple independent patient cohorts.⁴ These gene expression profiling studies and numerous others generated much optimism for the development of new oncology tests. However, such genomic cancer testing did not become a widespread clinical reality because the following three criteria were not met:

• Identification of a specific, unmet clinical need.
• Development of a robust, compatible assay.
• Clinical validation of the assay with multiple and large, outcome-based studies.

Researchers and clinicians must explore each of these criteria and understand the importance of the successful completion of each one in order to continue the successful translation of genomic assays from bench to bedside.

Translating Genomic Discoveries to Clinical Tests

Identifying the Unmet Clinical Need

Table I. (click to enlarge) Genomic tests commercially available or in development.

Successful translation of genomic discoveries to clinically validated tests is a multistep process that begins with identifying a clinical need and ends with a developed test that will change or influence treatment decisions. Table I lists many of the genomic tests currently available in the United States along with the assay types and citations for the pertinent validation studies.

The first step is to discover a genomic classifier that addresses a specific clinical utility. In breast cancer, an unmet clinical need is identifying which breast cancer patients are at risk of recurrence after surgical removal of their cancer. Clinical studies indicate that more than 20% of estrogen receptor (ER)-positive, node-negative patients that receive tamoxifen monotherapy are at risk of distant metastasis within 15 years postsurgery. By identifying this risk, physicians and patients can make a more informed treatment decision postsurgery (i.e., more- or less-aggressive treatment). Several commercially available genomic tests address this clinical need: OncotypeDX by Genomic Health Inc. (Redwood City, CA), MammaPrint by Agendia (Amsterdam), and Aviara Breast Cancer Index by bioTheranostics Inc. (San Diego).

Another area where genomic testing has made clinical inroads is the classification of metastatic cancers in which the primary origin is uncertain. A well-accepted fact in the oncology community is that knowledge of the primary origin is critical for determining the most appropriate treatment (i.e., the optimal chemotherapeutic regimen and/or targeted agent). Abbruzzese analyzed 879 consecutive patients with suspected unknown primary tumors and demonstrated that the survival rate of patients in which the primary tumor was diagnosed was higher compared with patients with primary tumors that remained unknown.⁵ Two genomic tests commercially available in the United States address this clinical need: Aviara CancerType ID by bioTheranostics and Pathwork Tissue of Origin by Pathwork Diagnostics (Sunnyvale, CA) (see Figure 1).

Developing a Robust and Compatible Assay

Figure 2. Unlocking RNA from formalin-fixed, paraffin-embedded (FFPE) tissue starts with processing the tissue section.

In order for genomic testing of solid tumors to have an impact on clinical management, the test needs to be not only clinically useful but also compatible with formalin-fixed, paraffin-embedded (FFPE) tissue preservation (see Figure 2). Most research studies that are focused on biomarker discovery start with frozen tissue and use genomic profiling platforms such as microarrays. Consequently, genomic discoveries must be migrated to more-robust clinical platforms such as those enabling real-time polymerase chain reaction (RT-PCR).

Currently in the United States, the majority of tumor biopsies are preserved by formalin-fixation followed by paraffin embedding to maintain the cytoarchitecture of the tissue. Formalin-fixation of tissue causes RNA to be cross-linked to proteins via covalent addition of monomethylol groups to nucleic acid bases.⁶ This cross-linking not only hampers the ability to extract RNA but also decreases the conversion of mRNA to a cDNA template by inhibiting the processivity of reverse transcriptase by these methylol modifications. In addition, within an FFPE tissue block, degradation of RNA occurs over time due to oxidation.⁶

By using proteinase K to release RNA from the matrix of cross-linked macromolecules and applying high temperatures to reverse monomethylol covalent modifications, investigators can extract RNA that can be converted to cDNA. In addition, reducing the amplicon length to 65-80 base pairs is a crucial step for the successful design of an RT-PCR assay for archived samples. By using short amplicons, quantifying specific mRNA is possible using highly degraded RNA samples from archived FFPE tissue that is 10-20 years old.

The ability to robustly measure highly degraded RNA is necessary to validate genomic classifiers. Such ability is due to the need for compatibility with current FFPE samples from routine biopsies and for demonstrated correlation between genomic classifiers and 10-15-year-old patient outcomes by quantifying the classifiers from 10-15-year-old biopsies. For genomic classifiers that are incompatible with FFPE tissues, introducing them within current standard tissue processing methods will be more difficult. More importantly, validating the classifiers will be more challenging since the majority of archived samples that are associated with clinical outcomes are preserved by formalin fixation. Currently, OncotypeDX, Aviara Breast Cancer Index, and Aviara CancerType ID genomic tests are compatible with FFPE tissues.

Another area that is sometimes overlooked in developing a robust assay is sample heterogeneity. In particular, using macrodissection or laser-based microdissection can be key to obtaining a sample that is enriched with the cells of interest. For example, many of the studies examining EGFR mutations in lung cancer use laser microdissection.⁷

Clinical Validation

The roadmap for clinical validation of a genomic classifier is defined by the intended demonstration of clinical utility. The design of the initial discovery cohort lays the foundation for such future validation studies. For example, the gene expression ratio of HOXB13/IL17BR (part of the Aviara Breast Cancer Index) was discovered as a prognostic index for ER-positive, tamoxifen-treated breast cancer patients. Subsequently, the initial validation of this ratio was completed in an independent set of ER-positive, tamoxifen-treated patients.⁸

Another crucial consideration in defining the roadmap for clinical validation is obtaining successful results from the set of samples that will convince both the oncology community and third-party payers that a particular genomic classifier is valid. More than 10 years ago, Hayes proposed levels of evidence (I-V) for grading the clinical utility of tumor markers, which remains a guidance for tumor marker validation.⁹ Hayes proposed that the highest level of evidence (level I) for a tumor marker would come from a single, high-powered, prospective, randomized, and controlled study specifically designed to test a marker. Notably, currently available tests that have been endorsed by oncologists and payers have only reached level II on the Hayes scale, which is validation within a previously conducted randomized clinical trial using archived specimens.

Clinical validation of genomic classifiers as being prognostic versus predictive requires samples from different patient cohorts. A prognostic test predicts clinical outcomes (e.g., good or poor) regardless of whether or not there is treatment. For example, MammaPrint is an FDA-cleared prognostic test using a 70-gene expression profile of fresh or frozen breast cancer tissue samples to assess a patient’s risk for distant metastasis, and is not indicated for predicting response to therapy. The validation cohort was 302 patients who did not receive adjuvant systemic treatments (i.e., no drug therapy).¹⁰

In contrast, validating a predictive genomic classifier requires examining samples from a previously conducted randomized clinical trial. This examination is required for optimal validation of the therapy benefit from, or response to, the adjuvant postsurgical treatment. For example, the recurrence score from OncotypeDX was demonstrated to predict chemotherapy benefit in ER-positive, node-negative breast cancer patients by correlating it with the outcomes from a previously conducted clinical trial. In this trial, the National Surgical Adjuvant Breast and Bowel Project (NSABP B-20), patients were randomly selected to receive either tamoxifen or tamoxifen plus chemotherapy. The study showed there was a statistically significant interaction between the recurrence score and patients who received chemotherapy.¹¹

In another example, investigators reported that the HOXB13/IL17BR index predicted endocrine benefit for ER-positive patients. This study correlated the index to the clinical outcomes from a previously conducted randomized trial in which some patients received an additional three years of tamoxifen while others did not.¹²

To validate clinically genomic classifiers requires careful consideration, including a prespecified statistical plan with predetermined cutoff points for a given genomic classifier (e.g., REMARK criteria).¹³ In particular, a critical exercise that is sometimes overlooked is conducting a power analysis to estimate whether the sample size under consideration is adequate to demonstrate statistically the prognostic or predictive utility of the genomic classifier. This type of study takes a long time, usually 1-3 years, and includes submission of a proposal, obtaining approval from a governing body, receiving the necessary samples, conducting the study and the analysis, and publishing the findings.

As mentioned above, the highest level of validation is conducting a prospective randomized trial in which the genomic classifier is used to guide treatment. In most disease states, this trial takes 5-10 years in order to obtain the necessary clinical outcome data. For example, the Trial Assigning Individualized Options for Treatment (Rx) will examine whether the OncotypeDX recurrence score can be used to determine which patients will benefit from chemotherapy. In another study, Nevins is conducting a randomized clinical trial to determine whether a prognostic genomic classifier can identify those early-stage non-small-cell lung cancer patients who typically receive surgery alone as a treatment option but may also benefit from adjuvant chemotherapy.¹⁴

Regulatory Hurdles

In 2007, FDA released a draft guidance for in vitro diagnostic multivariate index assays (IVDMIA) to address the regulation of assays that have the following characteristics:

• Combine the values of multiple variables using an interpretation function to yield a single, patient-specific result (e.g., a classification, score, index, etc.) that is intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease.
• Provide a result whose derivation is nontransparent and cannot be independently derived or verified by the end user.

Mark G. Erlander, PhD, is the chief scientific officer at bioTheranostics Inc. (San Diego). He can be reached at merlander@ aviaradx.com.

Although this document is still a draft guidance, two tests have already received clearance. This draft guidance is used for gaining approval of prognostic indications only, and it does not provide guidance for obtaining clearance for predictive genomic classifiers. The challenge going forward is bridging the gap between this guidance and the knowledge that physicians need to obtain information from predictive tests to affect treatment and patient care. In addition, payers still appear to be reimbursing only for tests that can predict treatment benefits.

Conclusion

The discovery and development of genomic classifiers for oncology-related applications are emerging and gaining acceptance by providing clinical utility beyond currently available clinical pathological information. Although the number of genomic tests currently on the market is relatively low, many more are in the pipeline. Medical researchers are finally making good on the postgenomic promise by developing tests with true clinical utility. In science as in life, patience is a virtue.