Retinoblastoma loss in cancer: casting a wider net

Summary:

Capturing both genomic and non-genomic mechanisms of retinoblastoma gene dysfunction has potential to improve risk stratification and patient selection for biomarker-driven therapy. A 186-gene expression signature is capable of identifying Rb loss across cancer types, providing a new framework for assessing Rb dysfunction based on transcriptome data.

In this issue of Clinical Cancer Research, Chen and colleagues(1) report a novel gene expression signature of retinoblastoma (Rb) loss that may predict Rb pathway dysfunction across cancer types occurring through multiple mechanisms. This study not only paves the way for assessment of Rb functional loss when only transcriptome data is available, but also re-emphasizes the important role of Rb as a tumor suppressor and the unanswered questions that still exist regarding the clinical application of Rb as a biomarker in cancer.

The retinoblastoma gene (RB1) is a tumor suppressor, first described over thirty years ago, that is commonly lost in several cancer types (2). Normally, the Rb protein prevents excessive cell growth by inhibiting G1 to S progression in the cell cycle. Rb binds to and inhibits transcription factors in the E2F family, leading to repression of E2F- regulated cell proliferation genes such as cyclin E and cyclin A. The activity of Rb is reversibly turned on and off through Rb phosphorylation controlled by cyclin-dependent kinases (CDKs). Beyond its role in cell cycle, Rb also interacts with other chromatin regulators or transcription factors and has been shown to regulate differentiation, DNA repair, and other cellular programs (2-4).

Identifying Rb loss has several potential clinical implications. For instance, some tumors such as small cell carcinomas harbor near universal loss of Rb, and indeed clinical grade immunohistochemical assays exist and may be used to confirm Rb protein loss associated with certain diagnoses. Rb loss is also prognostic in several malignancies, and the presence of Rb loss has been associated with treatment resistance, cellular plasticity, and lethal progression. Since it may be a late event, repeat biopsy to capture Rb loss may be required. How to treat Rb deficient tumors remains a significant gap in knowledge and a clinical unmet need. Patients are less likely to benefit from CDK 4/6 inhibitors (which are approved for breast cancer, and in clinical trials for other cancer types) since these drugs act to block phosphorylation of Rb in G1 phase to trigger cell cycle arrest. Patients with Rb-deficient tumors may preferentially benefit from DNA damaging agents (e.g, platinum chemotherapy) or potentially alternative therapeutic approaches based on its noncanonical functions(3,4) or synthetic lethality(5).

With these emerging implications, identifying patients with Rb deficient tumors has increasing clinical relevance. As with any molecular biomarker, understanding the advantages and limitations of the available assays is paramount (Fig. 1). With wider adoption of genomic sequencing, including both tissue-based tumor profiling and circulating tumor DNA, RB1 gene mutations and deletions are commonly identified. Whole genome sequencing or epigenetic profiling, though not routinely used clinically, may capture other means of RB1 disruption such as complex rearrangements or DNA methylation leading to gene silencing. These DNA events may or may not lead to loss of RNA or protein expression, which is assessable by in situ approaches such as ISH or immunohistochemistry, with the advantage of capturing intra-tumoral heterogeneity not often appreciated by genomic sequencing. Rb may also become dysfunctional in the absence of genomic or protein loss through hyper-phosphorylation by CDKs, a mechanism that is not currently assessible using clinical assays. Ultimately, functional assessment of Rb may be the most relevant, as this could potentially capture any of these upstream mechanisms that lead to the same downstream effects of Rb loss in a single test.

Figure 1. Mechanisms of Rb pathway loss and assays for their detection.

Left (DNA): Wild-type RB1 may be lost through genomic alterations (e.g., deletions, missense/nonsense mutations, indels or rearrangements) or epigenomic modifications (e.g., methylation of the RB1 gene promoter). Fluorescence in situ hybridization (FISH) is a method for detecting gene copy losses or rearrangements but not point mutations or indels; however, FISH can be performed from very sparse material and allows for in situ visualization of intratumoral heterogeneity. Next generation sequencing that includes exonic regions of the RB1 gene locus may be performed from clinical biopsy specimens or circulating tumor DNA to detect point mutations and indels; these assays also detect rearrangements when located within exonic regions and can usually identify mono- or bi-allelic deletions, though copy number losses are more challenging to detect in samples with low tumor DNA content. Whole genome sequencing would be required to detect complex genomic alterations outside exonic regions of RB1 (not depicted), and specific assays such as bisulfite sequencing would be required to detect epigenetic modifications of the RB1 gene. Middle (Protein): Biallelic loss of the RB1 gene locus leads to loss of RB1 protein expression by immunohistochemistry (IHC), but point mutations and frame shifts may result in expression of a hypofunctional or truncated polypeptide product that could potentially be recognized by RB1 antibodies for IHC. Non-genomic alterations such as epigenetic modifications, decrease in gene transcription/translation, or increased degradation could lead to decreased Rb protein levels by IHC, but the degree of decrease in levels may be indeterminate with regards to functional outcomes. Right (mRNA): Transcriptome-based assays may be performed from clinical biopsy specimens. RNA-based assays may not be able to distinguish upstream mechanisms of Rb pathway loss.

Towards this end, Chen et al analyzed data from 951 cell lines in the Cancer Cell Line Encyclopedia to generate a 186-gene expression signature of Rb loss. The signature was developed based on the presence of bi-allelic genomic alterations (i.e., deep deletion, shallow deletion with additional DNA mutation, or 2+ DNA mutations). They then applied this signature to patient tumor data from the Cancer Genome Atlas (TCGA) pan-cancer datasets which included over 11,000 tumors across 33 tumor types. The Rb loss signature had high accuracy in predicting RB1 bi-allelic loss and was also able to identify Rb functional loss due to promoter hypermethylation. As opposed to most signatures, the gene list encompassed but was not limited to canonical cell proliferation genes. High Rb signature score was associated with worse progression free survival, disease free survival, and overall survival in the TCGA cohorts as well as an advanced prostate cancer cohort.

A clinical assay for Rb pathway loss would be relevant for prognostic and predictive applications, and its utility would depend on feasibility in relation to biospecimens available, cost, and impact on clinical decision-making. Two copy loss of the RB1 gene by fluorescence in situ hybridization or DNA sequencing, as well as complete absence of protein expression by immunohistochemistry (IHC) are more or less unequivocal as far as conferring functional outcomes associated with Rb pathway loss. However, in the case of mutations including variants of uncertain significance, indeterminate results by IHC, or in biospecimens with no evidence of Rb loss by conventional assays, a functional assay based on gene expression could play an important role. It is important to note that Rb pathway loss by gene expression may not be interchangeable with complete Rb protein loss with regards to prognostication or prediction of therapeutic vulnerabilities in all cases. For example, gene expression conferred by loss of CDKN2A loss or amplification of D-type cyclins or CDK4/CDK6 would be expected to have some overlap with Rb protein loss; however, these upstream alterations would not be expected to lead to the same resistance to CDK4/6 inhibition as Rb loss. As such, ideally a clinical assay would be prospectively validated for predicting therapeutic vulnerability or resistance related to a specific agent or class of agent. The described Rb signature provides a new framework to assess the biologic and clinical implications of functional Rb loss. Teasing out the differences between mechanisms of loss, cancer specific changes, and timing will be important. Indeed, Chen and colleagues identified some transcriptomic differences between RB1-loss due to deletion and due to bi-allelic RB1 mutations. Understanding the impact of common co-occurring alterations (e.g. TP53 loss) and how these influence downstream pathway changes and therapy response will also be critical.

While additional studies are needed, this study represents a step towards casting a wider net to capture tumors with Rb deficiency. Although the RB1 gene was the first tumor suppressor gene to be described in cancer, much is still to be learned about its complex function and clinical utility. This paradigm may also be applied to other genomic alterations to better understand their frequency of functional loss and context within and across cancer types. Ultimately, transcriptome approaches may inform future clinical decision making and choice of therapy by complementing existing tools for precision oncology(6).