The first part of this series alluded to the need for more complete mapping of tumors in order to provide better prognosis and to develop better, more specific interventions. In the second section, we examined immune profiling as it relates to immune-oncology. Here, we will go deeper into gene expression profiling in cancer and immune-oncology.
To reiterate one of the key issues, cancers are complex genetic diseases. They are by definition difficult to attack because they are built to be evasive. As Sharma et al put it in the introduction to their 2017 review, “the intrinsic genomic instability common to all cancers facilitates the escape from cytotoxic or targeted therapies.”
Current Biomarkers are Ineffective
Cancer is complex because two patients with what appear to be similar tumors respond differently to a given treatment. We noted previously that “CD8+, CD4+, PD-1+, and […] PD-L1+ cell densities in pretreatment biopsies can predict response to therapy” (Chen et al, 2016). But according to the same paper, less than half (8%-44%) of metastatic melanoma patients treated with anti-CTLA-4 and anti-PD-1 show a response. What about the other 56%-92%? They display some type of non-responsive or resistant immune signature, so inhibiting immune checkpoint signaling – the pathway targeted by anti-CTLA-4 and anti-PD-1 antibodies – is ineffective.
PD-1 (encoded by PDCD1 in humans) is consistently among the most highly referenced when discussing gene expression in immune oncology; it is rare to find a paper that doesn’t mention this protein. However, it cannot be the whole story or we would see consistent responses across the board.
Profiling the Tumor Microenvironment
There is, therefore, an even deeper level of profiling required to obtain a full picture of a tumor and to identify relevant biomarkers. Numerous reasons exist to explain why gene profiling is a valuable – and rapidly becoming necessary – part of the diagnostic process. However, all those reasons more or less fall under one umbrella: the genetic profile, the molecular subgroup of the tumor, defines its responsiveness to any given therapy. More and more studies are uncovering gene expression profiles that will eventually be used to predict response to therapy.
Prat et al published a 2017 paper in Cancer Research that looked at seven different factors to link “DNA-level genomic alterations [with] immune cell infiltration.” Additionally, they looked at whether those results varied depending on cancer (in this case breast) subtype. This particular study heavily involved RNA-seq. Molecular technologies such as RNA-seq are critical to understanding the immune profile of any given tumor and its microenvironment. While gene sequencing can be useful for discovering cancer-related mutations or neoantigens, RNA-seq offers high-resolution and dynamic insight into the actual expression of those genes, and any others. Immune profiling in oncology is very much concerned with expression levels of normal genes, so RNA-seq is a much stronger tool to assess those. Additionally, RNA-seq can be used to track gene expression over time, for example before and after chemotherapy.
Gene expression analysis is now being used for even more detailed investigation of tumors, going beyond aggregate analysis of a tumor and/or its microenvironment. Chung et al recently noted that “genomic and gene expression profiling are usually used to characterize a bulk tumour in individual cancer patients, whereas cancers display intratumoral heterogeneity that might affect the therapeutic outcome of a targeted treatment.”
The Chung study used high-resolution tools to investigate primary breast cancer, thereby offering a look into potential future applications for gene expression analysis in clinical settings. Their research led to a number of important findings: First, the expression profiles of “carcinoma and tumor-infiltrating immune cells” could be delineated. Second, while the researchers noted the heterogeneity of the sample, they also developed “core gene expression signatures for subtype-specific breast cancer cells.” Lastly, the immune cells were characterized as either activating or suppressing, “suggesting dynamic immune cell interactions and a distinct immune system status in each tumour.” Understanding which immune cells, and which relevant genes, are involved in suppressing tumorigenesis, and which are hijacked to support the tumor. Work such as this suggests that, in the future detailed maps may be created from each patient sample to determine the exact balance of all of these factors.
Coming to the state of the field today, there is a significant and growing database of gene expression information useful in clinical and research contexts. This means that RNA-seq technology can be used to report on gene panels, the same way that microarray does but with far cleaner results. More importantly, though, is the ability to provide comparative analysis of the data returned from one of these panels. Thanks to the accumulated knowledge regarding gene expression profiles of various cancers combined with the volume and accuracy of data provided by RNA-seq, data from one sample can easily be put into context relative to other tumors of the same type. Reports from Cofactor’s Paragon Assay compare results from the tumor sample being tested against a database of gene expression information for that same type of tumor. Therefore, a clinician will know whether the expression level of a particular gene is typical for that cancer, or if it is unique to that specific patient.
In addition, RNA-seq can play a useful role in uncovering novel neoantigens, one of the keys to immune-oncology. In part 1 we noted that neoantigens are a more useful target for profiling tumors than are antigens that are over expressed in tumors. The latter are generally expressed at some level by non-transformed cells, so specifically targeting them is difficult. Neoantigens, on the other hand, are produced by mutated loci, and are therefore unique to the tumor. Additionally, immune cells may be more likely to target (or more effective in targeting) a cell presenting neoantigens as they appear more “non-self” than a cell simply expressing higher-than-normal levels of a wild-type gene.
Identifying neoantigens is, like everything else, more complicated than we might wish it to be. The presence of a mutation doesn’t necessarily relate to expression of that gene as a neoantigen. And, again, the heterogeneity of a tumor means that some cells may have a mutation while others may not. cDNA screening has been the historic norm for identifying neoantigens. In a cumbersome but reasonably effective approach, “cDNA library and MHC molecules were over-expressed in cell lines, and then co-cultured with T cells to identify antigens that could induce the T cell activation” (Lu et al). Now, researchers are developing new protocols using a combination of techniques to both identify and validate the presence of neoantigens. Karasaki et al described one such approach in February, 2017. They found that “integration of both expression data and sequence data from RNA‐Seq with WES” was the best of four approaches to accurate prediction of neoantigens.
In short, new genomics technologies provide powerful tools for researchers and clinicians alike to build a granular picture of any given tumor. Combined with newer methods of pathological examination (e.g. flow cytometry and protein expression analysis), we are starting to reach a point where targeted therapeutics and precise diagnosis, prognosis and treatment plans can be produced.
At Cofactor, we are focused on applying our database of Health Expression Models to immune profiles to better answer these questions. Our ImmunoPrism™ assay offers researchers and clinicians a powerful method to quantify expression of genes, such as PD-1, that are key players in immune avoidance. Additionally, our assay provides context for these expression values, making interpretation and patient comparisons even more powerful. Contact us at [email protected] for more information on gene expression in immune-oncology, or to discuss how ImmunoPrism can be used in your translational or clinical studies.