Translating MicroRNA Discovery Into Clinical Biomarkers in Cancer

In the United States, cancer is the second leading cause of death, exceeded only by cardiovascular disease, and an estimated 500 000 patients with cancer will die this year.^{1 - 2} After cardiovascular and infectious diseases, cancer is the third leading cause of mortality worldwide.³ However, the field of clinical oncology is poised for unprecedented innovation, reflecting the confluence of breakthroughs in decoding disease pathobiology in the context of high-throughput enabling technologies.⁴ Harnessing the full potential of transformative advances is predicated on defining biomarkers that promote targeted cancer prevention, diagnosis, and treatment of individual patients and populations.^{4 - 5} A new generation of molecular technologies, including genomic, proteomic, and metabolomic mapping, hold the promise of translating into practice the use of biomarker panels for increased diagnostic and therapeutic sensitivity and specificity.^{2 ,4} Yet essential elements have resisted definition in developing mechanism-based molecular markers for individualized management of cancer. In particular, the hierarchically organized integrated epigenetic, genetic, and postgenetic circuitry that dictates developmental restriction of cell destiny and underlies tumorigenesis when dysregulated has so far remained poorly understood.

Emerging science has revealed a layer of genetic programmatic coordination by which cells determine their fate; this layer involves posttranscriptional regulation of gene expression by microRNAs (miRNAs).⁶ These regulatory molecules originate by transcription of distinct genes in the noncoding portions of chromosomes as precursor RNA molecules of hundreds to thousands of nucleotides, which undergo distinct nuclear and cytoplasmic processing.^{6 - 7} The resulting double-strand miRNA molecules of approximately 22 bases form the targeting core of a complex multimeric mechanism hybridizing with messenger RNA molecules through nucleotide complimentarity, resulting in their sequestration or degradation, thereby defining the pool of genes available for lineage specification.^{8 - 9} The combinatorial nature of nucleotide complimentarity permits individual miRNAs to regulate the expression of hundreds of genes by posttranscriptional modification of their cognate messenger RNAs. This mechanism represents the most recent addition to the multidimensional complexity characterizing nuclear-cytoplasmic information processing at the interface between epigenetic and genetic mechanisms, as well as transcriptional, translational, and posttranslational regulation. MicroRNAs represent a purely regulatory, as opposed to structural, process that fine-tunes gene expression.⁶ Importantly, their function has provided insight into the integrated genetic circuitry that defines cell differentiation that, when corrupted, supports neoplastic transformation.^{6 ,8 - 9}

The pervasiveness of this novel mechanism, which controls gene expression⁶ and the accepted paradigm of cancer as a genetic disease,^{10 - 11} suggest linkage between miRNA-dependent processes and susceptibility to neoplastic transformation. Indeed, altered miRNA expression has been recently recognized as a trait of tumorigenesis.^{12 - 13} Even though some miRNAs commonly exhibit altered expression across tumors, different tumor types more often express unique patterns of miRNAs, referable to their tissues of origin.^{12 - 18} Further, processes underlying initiation and progression of tumors, including genomic instability, epigenetic dysregulation, and alterations in expression or function of regulatory proteins, directly alter the complement of miRNAs expressed by transforming cells.⁶ Moreover, specific miRNAs regulate fundamental elements integral to carcinogenesis, including tumor suppressors and oncogenes, with induction of their dysregulation promoting cell transformation.^{6 ,13 ,19}

Patterns of miRNA expression appear to be a richer source of pathognomonic tumor information than messenger RNA expression profiling.¹² The presumptive role of miRNAs in tumorigenesis underscores their value as mechanism-based therapeutic targets in cancer that have yet to be fully exploited. Similarly, unique patterns of altered miRNA expression provide complex fingerprints that may serve as molecular biomarkers for tumor diagnosis, prognosis of disease-specific outcomes, and prediction of therapeutic responses.

In this issue of JAMA, Bloomston and colleagues²⁰ compared miRNA profiles from patients with normal pancreas, chronic pancreatitis, and pancreatic cancer, processes that exemplify the paradigm of neoplastic transformation in a specific tissue. Specific patterns of miRNA expression distinguished pancreatic cancer from normal pancreas in 90% of cases and, separately, from chronic pancreatitis in 93% of cases. A subset of miRNAs was of prognostic value, identifying patients with pancreatic cancer who survived longer than 24 months, compared with those who survived less than 24 months. In addition, the expression of 1 specific species of miRNA predicted overall poor survival (median, 14.3 months) compared with patients whose tumors did not express this species (median, 26.5 months).

Beyond diagnosis and prognosis, miRNAs associated with neoplastic transformation in pancreatic tissue were demonstrated previously in other tumors to mediate pathophysiological mechanisms underlying tumorigenesis. While these analyses are the first to define miRNA profiles in adenocarcinoma of the pancreas, they reinforce the usefulness of these biomarkers in defining the molecular taxonomy of tumors.^{6 ,12 - 19} Similarly, these results highlight the potential of miRNA profiling for defining prognosis and risk stratification, identifying low- and high-risk populations of patients with pancreatic cancer. Moreover, the findings underscore the potential value of miRNAs as mechanism-based therapeutic targets in cancer, certainly a critical unmet clinical need in the management of pancreatic cancer.

In the context of these exciting observations in a disease characterized by a dismal prognosis, should clinical oncologists and cancer geneticists begin to apply miRNA profiling to establish stratification of risk or define therapeutic targets in patients with pancreatic cancer? Although the analyses of Bloomston et al provide an initial glimpse into the future of clinical oncology, they reflect the beginning of the continuum integrating discovery, development, regulatory review, and the evidence basis of medicine required to translate advanced technology into clinical practice, a framework that has largely been ignored in the field of biomarkers.^{4 - 5 ,21} Indeed, while biomarkers represent the envisioned future for individualized management of patients with cancer, their potential has yet to be realized.^{2 ,22}

Technological advances generating biomarkers have driven biomedical discovery but have not been systematically validated to define performance metrics, including reproducibility, sensitivity, and precision, required for broad application to clinical specimens.^{2 ,23} Moreover, molecular analytes may be evaluated using different technical platforms for which performances have not been cross-verified. Absence of assay performance standards that reflect rigorous standardization across laboratories and platforms underlies issues of reproducibility, thereby undermining their application to patients.² In the specific case of miRNAs, exploratory analyses have been performed on microarray or bead-based platforms.^{12 ,24} While results from different analyses on these distinct, but complimentary, platforms generally concur,⁶ the challenge remains of rigorously defining the performance metrics of these platforms and their cross-validation for clinical application. Similarly, quantitative and qualitative relationships between biomarkers, patient management, and disease outcomes have not undergone rigorous clinical study, and the proof linking a biomarker with clinical end points may not be readily available.^{4 - 5 ,25} Relationships defining the clinical usefulness of a biomarker should be assessed in randomized clinical trials and subsequently validated in follow-up trials.²²

In the absence of this rigorous approach, the clinical value of biomarkers may be overestimated, reflecting bias and chance resulting from inadequacy of study cohorts.²² For the specific example of miRNAs, unique patterns of expression defining the molecular taxonomy of tumors, and the usefulness of these panels for discriminating differences in patient outcomes, have been defined in exploratory studies in relatively limited populations.^{6 ,12 - 19} In that context, miRNA biomarker panels are poised to advance along the continuum from discovery into development to define their diagnostic, prognostic, and predictive value in hypothesis-driven, appropriately designed, and sufficiently powered clinical trials, using rigorously validated assays and analytical platforms. These studies, and follow-on validation trials, eventually will provide the evidence base for adoption into clinical practice guidelines.

Posttranscriptional regulation by miRNAs is one fundamental component of the hierarchical integrated regulation of gene expression. The corruption of these regulatory elements by processes underlying tumor initiation and promotion contributes to the genetic dysfunction defining and characterizing neoplasia. Beyond molecular mechanisms underlying transformation, the discovery of unique patterns of miRNA expression characterizing tumor taxonomy in pancreatic²⁰ and many other tumors offer the unique opportunity to develop biomarkers for the diagnostic, prognostic, and predictive management of cancer. This is just the beginning in the continuum connecting discovery to clinical application. To advance these seminal observations²⁰ into patient management requires proceeding through development and regulatory approval and establishing the evidence basis for clinical practice.