A New Dimension in Biomarker Research

Cutting-edge cancer research aims to unravel a tumor’s complexities at the molecular level. There are many different ways to work toward this goal – we can investigate the cancer genome, transcriptome, or proteome. Each of these can provide unique information, but comes with its own set of capabilities and limitations (see Table 1).

For the last two decades, the transcriptome has been a focus in cancer research – which has led to the discovery of numerous promising RNA biomarkers for diagnosis, prognosis and prediction of therapy benefit. But identifying biomarkers is only half the battle; to validate them and develop them into clinical applications, we need improved methods for RNA analysis in clinical specimens.

Tools of the trade

Most interrogations of the cancer transcriptome begin with a comprehensive transcriptomic discovery program, using high-throughput approaches like microarrays or RNA sequencing (RNA-Seq). The evolution of next-generation sequencing (NGS) over the past 10 years has offered researchers the ability to look at the cancer transcriptome in greater detail than ever before. Once we define a set of transcripts whose expression differs between healthy and tumor tissue, or between clinical outcomes, we can then focus on validating their functional significance and clinical relevance. That’s the point where spatial information provided by RNA in situ hybridization (RNA ISH) and immunohistochemistry (IHC) come in. Analyzing RNA in its morphological context is highly desirable for cancer research, and the spatial resolution provided by RNA ISH adds a new dimension to gene expression analysis. For instance, it can provide us with precise localization of target RNA in single cells, allowing direct mapping of RNA biomarkers to specific cell types in the tumor tissue.

RNA ISH can be conducted using either isotopic or non-isotopic probes. Non-isotopic probes, labeled with substances like biotin or digoxigenin, present a particularly pragmatic approach, improving turnaround time, sensitivity and safety in comparison with isotopic probes (1). But because RNA targets are short and the probes can only incorporate a limited amount of label, the technique lacks sufficient sensitivity for most expressed genes. Attempts to amplify the targets before hybridization, or the signals afterward, have met with a poor signal-to-noise ratio thanks to nonspecific binding and cross-hybridization in complex tumor sections. Only with recent technological advances has the sensitivity, specificity and ease-of-use improved enough to make non-isotopic ISH a practical choice (2–6). We can even perform robust single RNA molecule detection in routine formalin-fixed, paraffin-embedded (FFPE) clinical specimens (see Figure 1), unlocking the potential of RNA for a wide range of cancer study areas.

Getting a handle on heterogeneity

Tumor molecular heterogeneity presents a significant challenge for cancer researchers and pathologists alike. Methods like RT-qPCR and RNA-Seq, which require nucleic acids to be in solution, destroy a sample’s morphological context and spatial resolution. Researchers are limited to comparing RNA expression data among heterogeneous cell populations – unless they use in situ methods (see case study “Analyzing Prostate Tumor Molecular Heterogeneity”).

Analyzing Prostate Tumor Molecular Heterogeneity (7, 8)

Researcher: Nallasivam Palanisamy, associate scientist, Henry Ford Health System; associate research professor (adjunct), University of Michigan.

Research topic: Refining approaches for molecular classification to replace morphological assessment of tumors. This involves the discovery of new molecular markers in cancer, particularly recurrent gene fusions, and understanding their role in cancer development.

Methods:
Discovery – RNA-seq
Initial transcriptome sequencing presents an unbiased characterization of a given sample, identifying biomarkers in both protein-coding and noncoding genes.

Validation – RNA ISH
For subsequent biomarker profiling, RNA detection is the only option when looking at noncoding genes; even some of the markers based on protein-coding genes do not have good antibodies. Palanisamy explained, “ETV1, ETV4 and ETV5 genes are overexpressed in a small subset of prostate cancer, and RNA screening is the method of choice. Even for the genes with good antibodies, if the protein level is variable or always too low for detection, supporting protein analysis with information on RNA expression forms an unequivocal assessment.”

Approach
RNA ISH followed by IHC is performed on the same slide in a sequential manner. Given the limited availability of tissue from a small biopsy, it is important to develop methods to detect more than one type of marker on the same slide. Palanisamy commented, “Development of combined protein and RNA detection methods may overcome many concerns for accurate detection of biomarkers, and I can see this being the standard practice in future molecular cancer profiling.”

No protein? No problem!

New classes of long noncoding RNAs (lncRNAs) are highly valuable as biomarkers capable of uncovering a specific biological trait or measurable change directly associated with a physiological condition or disease status (9). But because these RNAs are not ultimately translated into protein counterparts, however, their detection relies entirely on our ability to detect RNA in cancer biopsy samples. With several such candidates showing diagnostic and prognostic promise (7, 10), and others vital for understanding a tumor’s underlying biology, RNA ISH allows for effective and reproducible in situ detection of lncRNA biomarkers (see Figure 2 and case study “Realizing the Potential of Long Non-Coding RNA as a Cancer Biomarker”).

Realizing the Potential of Long Non-Coding RNA as a Cancer Biomarker (10–12)

Researcher: Rohit Mehra, clinical assistant professor of pathology at Michigan Center for Translational Pathology.

Research topic: Long non-coding RNA (lncRNA) plays an important role in the pathogenesis of genitourinary cancers, especially prostate cancer. Of the several lncRNAs important in prostate cancer, one in particular – SChLAP1 – may have clinical utility as a prognostic or diagnostic biomarker. For this, accessible methods for routine in situ lncRNA detection are vital.

Methods:
Discovery – RNA-seq
Comprehensively profiling the transcriptome of over 100 prostate cancer tissues and cell lines revealed that ~20 percent of RNA transcripts in prostate cancer represent novel, uncharacterized lncRNA genes (12). From this set, 121 candidate lncRNAs were nominated for further investigation.

Validation – RNA ISH
One of these candidates was re-named SChLAP1, and in the cohort studied, RNA ISH for SChLAP1 effectively stratified patient outcomes by predicting more rapid biochemical recurrence, clinical progression to metastatic disease (defined by a positive bone scan), and prostate cancer-specific mortality. Mehra commented, “This technology allows us to directly visualize gene expression in the target tissue of interest – for example, within the same sample we can tell whether gene overexpression occurs in benign prostate glands, high grade prostatic intraepithelial neoplasia (HGPIN – a precancerous state) or prostate cancer.”

From bench to bedside

It’s clear that the future of RNA biomarkers in the clinic is promising with the advent of modern RNA ISH technologies. In particular, companion diagnostics – vital in guiding cancer therapeutics – are central to the personalized medicine revolution. A prime example of RNA ISH application is in determining HER2 status in the management of breast carcinoma – where single-cell quantitative in situ RNA has been shown to be advantageous in resolving equivocal HER2 status and tumor heterogeneity (13). Viruses can also contribute to tumor development, and routine detection of viral genes demands a specific, sensitive and accessible technique. Human papillomavirus (HPV), for instance, is a causal agent in head and neck squamous cell carcinoma. Evidence for transcriptionally active HPV oncogenes E6/E7 is the gold standard for determining the presence of clinically relevant infections, but it can be challenging to detect E6/E7 mRNA using conventional techniques. PCR amplification of HPV DNA is more sensitive, but still less specific than DNA ISH – and RNA ISH has been found to provide both sensitive and specific detection (see Figure 3), facilitating a potential diagnostic standard in the future (14).

Because cancer is a disease of gene expression gone awry, understanding the tumor’s dynamic transcriptomic landscape is invaluable for basic and translational research. From characterizing tumor heterogeneity to studying noncoding transcripts, innovative RNA ISH methods are proving their worth in modern-day cancer research. By providing morphological context as well as detection, these methods join a set of powerful approaches enabling cancer researchers to discover, develop and implement a new generation of tissue- and cell-based techniques integral to the promise of personalized medicine.