Gene expression profiling

In 2012, approximately 1.6 million people will be newly diagnosed with cancer in the United States, and about 400,000 of these patients will present with metastatic disease.¹ Despite extensive clinical work-up and histopathologic evaluation, it is estimated that primary site diagnosis remains unknown or there is a differential diagnosis in as many as 150,000 cases.^2-4 This diagnostic uncertainty can lead to suboptimal treatment, as site-directed chemotherapy regimens and a growing number of approved targeted therapies are dependent on accurate tumor classification.

Identification of the primary site remains challenging and elusive in metastatic cases due in part to inherent biological characteristics of many metastatic tumors. These tumors are often poorly differentiated or undifferentiated and may have atypical presentation.^2,3 Tumors can broadly metastasize to multiple organs and may no longer be present in the primary site. Furthermore, aberrant expression or lack of expression of organ-specific biomarkers detected by routine immunohistochemistry (IHC) may further confound a diagnosis (e.g., TTF-1-negative lung adenocarcinomas, CDX2-negative colon adenocarcinoma).^5,6

IHC is the standard of care for pathologic assessment of tumor classification, particularly in poorly or undifferentiated tumors; however, there are limitations to this technique that contribute to the difficulties of achieving diagnostic certainty in many metastatic tumors. IHC antibodies have varied sensitivities and specificities and are not applied in an objective and standardized manner in routine practice.^7,8 The interpretation of semi-quantitative IHC results is subjective and dependent on the selection of IHC panels and antibody types.^8,9 In a meta-analysis published in 2010, the accuracy of IHC was reported to be approximately 66% in metastatic cancers.¹⁰ These data confirm the unmet medical need in the identification of the primary site in metastatic tumors, and highlight that additional analytical approaches are necessary to aid in resolving indeterminate primary site diagnoses in patients with metastatic cancer.

Molecular classification by gene expression profiling

Advances in genomic technology in the past two decades have given rise to an emerging field of gene expression profiling for the classification of tumors to increase diagnostic certainty. Several gene expression profiling assays have been commercially developed for clinical application, utilizing either microarray- or quantitative real-time RT-PCR-based platforms.^2,11-13 In general, each of these assays classifies tumors by quantifying the similarity of a specific gene expression profile from a tumor specimen to gene expression profiles from a reference database of known tumor types.¹⁴ Gene expression profiling of tumors provides independent quantitative data that is objective and standardized, can be integrated with clinical findings, and can complement IHC in the primary site diagnosis of metastatic cancers.

Real-time RT-PCR and microarray platforms both allow for quantification of differential gene expression in a tumor; however, there are distinct advantages and disadvantages associated with each platform. The microarray platform allows for profiling of tens of thousands of genes simultaneously, but microarray exhibits lower sensitivity in measuring individual gene expression. Because of the limited dynamic range (~10²) of the platform, the genes most commonly selected for quantitative analysis by microarray are generally those exhibiting large degrees of change. This limitation is further amplified when analyzing formalin-fixed tumor biopsy tissue, containing a low number of tumor cells and degraded RNA. Quantitative real-time RT-PCR is a highly sensitive platform and is the gold standard for gene expression measurements. Although it is limited to measuring a few hundred genes at once, it is often used as a validation tool for confirming gene expression results obtained from microarray analysis.¹⁵ Real-time RT-PCR is a robust platform that combines broad dynamic range (~10⁶) and high sensitivity with the ability to detect single copies of RNA and small differences in gene expression, thereby making it a platform of choice for many molecular-based clinical laboratory tests.^16,17

Characteristics of a molecular cancer classifier

Commercially available molecular cancer classifiers analyze either messenger RNA (mRNA) or microRNA (miRNA) from formalin-fixed paraffin-embedded tissue, and they have distinct characteristics and capabilities in tumor classification. A useful molecular cancer classifier should include features that enable compatibility with variable quality and quantity of tissue specimen and demonstrate high accuracy within clinically relevant parameters. The assay should also integrate a process or technique to enrich for tumor specimen and ensure high purity of tumor cells.  It is also important for a classifier to be able to detect fragmented RNA and utilize a reproducible and highly analytical platform to measure gene expression. For a molecular cancer classifier to be diagnostically relevant in tumor classification, it is essential to have a reference database that covers a broad scope of tumors encompassing both common tumor types as well as rare tumor types, and includes sufficient depth of tumor samples. A molecular cancer classifier should also be validated and exhibit high accuracy within clinically relevant parameters including metastatic and poorly differentiated tumors and limited tissue specimens such as fine needle aspirates and small core biopsies.

Clinical validation and clinical utility in challenging cases

Validation studies have been reported for each of the commercialized molecular classification assays.^2,18,19 The largest and most diagnostically comprehensive validation study of a molecular cancer classifier to date was recently published by Kerr and colleagues in Clinical Cancer Research.²⁰ In this multi-institutional, prospectively-defined, blinded study, the performance of a 92-gene real-time RT-PCR assay, which classifies 28 main tumor types and 50 subtypes, was assessed using 790 tumor samples with a minimum of 25 cases tested per tumor type. Cases were selected at three centers of excellence (UCLA, Mayo Clinic, and Massachusetts General Hospital), adjudicated between institutions, and blinded FFPE tumor sections were submitted for analysis. The cases included 44% metastatic tumors, with the rest composed of predominantly moderately- to poorly-differentiated primary tumors. The study also assessed the performance of the 92-gene assay in limited tissue specimens such as fine needle aspirates and small biopsies (14% of the cases).

The 92-gene assay demonstrated an overall sensitivity of 87% and specificity >99% for 28 main tumor types. For tumor subtyping of 50 histologies, the overall sensitivity was 82% and specificity was >99%. Prespecified analyses demonstrated that the high level of accuracy was stable in metastatic tumors (vs. primary tumors), high-grade tumors (vs. low-grade), and in limited tissue specimens (vs. excision biopsies and large core needle biopsies). In addition, the 92-gene assay demonstrated excellent performance in distinguishing primary from metastatic tumors, exhibited by 100% PPV for identifying a primary tumor in lung, brain, and pleura/peritoneum—organs that commonly harbor metastatic lesions. The 92-gene assay also demonstrated a low false rule-out rate, with only 5% of the cases incorrectly excluded.²⁰ Taken together, these data support the clinical application of molecular classifiers as an adjunct tool for pathologists for tumor classification in difficult-to-diagnose cases.

While validation studies have been conducted to determine the overall performance of molecular classifiers in tumor classification, until recently their performance has not been directly compared with IHC. Weiss and colleagues recently reported (in abstract form) preliminary results of a blinded study of 122 samples that compared the diagnostic accuracy of IHC + morphologic review vs. molecular classification for the determination of primary tumor site in a series of difficult-to-diagnose, high-grade, primarily metastatic tumors.²¹ The 92-gene assay demonstrated a statistically significant 10% increase in overall accuracy vs. IHC.²¹ This study provides additional evidence for the role of molecular classifiers as a complementary technique to IHC in difficult-to-diagnose metastatic tumors.

The clinical indications for use of molecular classification are evolving, but evaluation of clinical experience has shed light on clinical applications. In a retrospective analysis of 300 consecutive cases submitted for molecular testing with the 92-gene assay,²  pathology reports indicated that approximately 40% of the cases were poorly differentiated or undifferentiated, and 44% were submitted to resolve a differential diagnosis. More than 50% of the cases submitted for molecular testing were from core needle biopsies or fine needle aspirates, with limited tumor sample available for testing. In 97% of these cases, the 92-gene assay provided a molecular diagnosis for tumor type and subtype, demonstrating clinical utility when there are differential diagnoses, poorly differentiated tumors, or limited biopsy specimens.²

Strategic tissue management and molecular classification

Accurate identification of primary tumor site, histological subtype, and knowledge of biomarker status are required for optimal selection of site-specific chemotherapy regimens and molecularly targeted treatments; however, availability of adequate tissue for comprehensive diagnostic testing often is a limiting factor. This is particularly true given the movement toward less-invasive and smaller biopsies. With the growing number of molecularly targeted therapies and increased availability of molecular testing to identify candidate patient populations, strategic management of the biopsy specimen is recommended not only for H&E and IHC analysis, but also for molecular characterization.¹⁴ Schnabel and Erlander (2012)¹⁴ proposed a tissue-based diagnostic algorithm to maximize tissue use from tumor specimens by adjusting diagnostic evaluation based on the amount of available biopsy material (Figure 1).

Figure 1.

The integration of molecular classification earlier in pathological evaluation may be reasonable in cases of limited biopsy specimens (e.g., fine needle aspirates and small core needle biopsies), particularly if the tumor is poorly differentiated or has atypical characteristics. Molecular classification results can help to direct confirmatory IHC and appropriate biomarker testing. If sufficient biopsy tissue is available, molecular classification can be used as an adjunctive tool when a definitive diagnosis is not reached after the initial battery of IHC analysis.¹⁴ Appropriate tissue management and routine integration of molecular classification within the standard-of-care paradigm can aid in increasing diagnostic certainty and has the potential to positively impact patient care.

In this era of personalized medicine, accurate tumor classification is crucial for optimal patient care. Standard pathologic techniques such as IHC and morphologic review render accurate diagnoses in most cases; however, studies have demonstrated limitations of IHC in metastatic disease and poorly differentiated or undifferentiated tumors. Molecular classification of tumors can provide an objective diagnosis that is complementary to current IHC and imaging methods. Responsible tissue management and appropriate integration of molecular tumor classification in a tissue-based diagnostic algorithm may increase diagnostic certainty and enable downstream ancillary biomarker testing to help guide site-specific and molecularly targeted therapies for cancer patients.

References

1. Howlader N, Noone, AM, Krapcho M, et al, eds. SEER Cancer Statistics Review, 1975-2008. Bethesda, MD: National Cancer Institute. http://seer.cancer.gov/csr/1975_2008/, based on November 2010 SEER data submission. Accessed August 3, 2012.
2. Erlander MG, Ma XJ, Kesty NC, et al. Performance and clinical evaluation of the 92-gene real-time PCR assay for tumor classification. J Mol Diagn. 2011;(13):493-503.
3. Greco FA, Hainsworth JD. Cancer of unknown primary site. In: DeVita VTJ, Hellman S, Rosenberg SA, eds. Cancer: Principles and Practice of Oncology. Philadelphia, PA: Lippincott Williams & Wilkins; 2011:2033-2051.
4. Schroeder BE, Laouri M, Chen E, Erlander MG, Schnabel CA. Pathological diagnoses in cases of indeterminate or unknown primary submitted for molecular tumor profiling. Mod Pathol. 2012;25(suppl 2; abstr 429).
5. Hecht JL, Pinkus JL, Weinstein LJ, Pinkus GS. The value of thyroid transcription factor-1 in cytologic preparations as a marker for metastatic adenocarcinoma of lung origin. Am J Clin Pathol. 2001;116(4):483-488.
6. Dennis JL, Hvidsten TR, Wit EC, et al. Markers of adenocarcinoma characteristic of the site of origin: development of a diagnostic algorithm. Clin Cancer Res. 2005;11(10):3766-3772.
7. Yaziji H, Barry T. Diagnostic immunohistochemistry: what can go wrong? Adv Anat Pathol. 2006;13(5):238-246.
8. Matos LL, Trufelli DC, de Matos MG, da Silva Pinhal MA. Immunohistochemistry as an important tool in biomarkers detection and clinical practice. Biomark Insights. 2010;5:9-20.
9. Taylor CR, Levenson RM. Quantification of immunohistochemistry—issues concerning methods, utility and semiquantitative assessment II. Histopathology. 2006;49(4):411-424.
10. Anderson GG, Weiss LM. Determining tissue of origin for metastatic cancers: meta-analysis and literature review of immunohistochemistry performance. Appl Immunohistochem Mol Morphol. 2010;18(1):3-8.
11. Monzon FA, Lyons-Weiler M, Buturovic LJ, et al. Multicenter validation of a 1,550-gene expression profile for identification of tumor tissue of origin. J Clin Oncol. 2009;27(15):2503-2508.
12. Rosenwald S, Gilad S, Benjamin S, et al. Validation of a microRNA-based qRT-PCR test for accurate identification of tumor tissue origin. Mod Pathol. 2010;23(6):814-823.
13. Ma XJ, Patel R, Wang X, et al. Molecular classification of human cancers using a 92-gene real-time quantitative polymerase chain reaction assay. Arch Pathol Lab Med. 2006;130(4):465-473.
14. Schnabel CA, Erlander MG. Gene expression-based diagnostics for molecular cancer classification of difficult to diagnose tumors. Expert Opinion on Medical Diagnostics. 2012;0(0):1-13.
15. Morey JS, Ryan JC, Van Dolah FM. Microarray validation: factors influencing correlation between oligonucleotide microarrays and real-time PCR. Biol Proced Online. 2006;8:175-193.
16. Cronin M, Sangli C, Liu ML, et al. Analytical validation of the Oncotype DX genomic diagnostic test for recurrence prognosis and therapeutic response prediction in node-negative, estrogen receptor-positive breast cancer. Clin Chem. 2007;53(6):1084-1091.
17. Lyon E, Wittwer CT. LightCycler technology in molecular diagnostics. J Mol Diagn. 2009;11(2):93-101.
18. Pillai R, Deeter R, Rigl CT, et al. Validation and reproducibility of a microarray-based gene expression test for tumor identification in formalin-fixed, paraffin-embedded specimens. J Mol Diagn. 2011;13(1):48-56.
19. Meiri E, Mueller WC, Rosenwald S, et al. A second-generation microRNA-based assay for diagnosing tumor tissue origin. Oncologist. 2012;17(6):801-812.
20. Kerr SE, Schnabel CA, Sullivan PS, et al. Multisite validation study to determine performance characteristics of a 92-gene molecular cancer classifier. Clin Cancer Res. 2012;18(14):3952-3960.
21. Weiss L, Chu P, Schroeder BE, et al. Blinded comparator study of immunohistochemistry (IHC) versus a 92-gene cancer classifier in the diagnosis of primary site in metastatic tumors. J Clin Oncol. 2012;30:(suppl; abstr e21019).