William Check, PhD
“My goodness, child, you’ve grown like a weed.” No area of molecular pathology deserves this grandmotherly accolade more than the exploding field of microRNA, which has moved from discovery to clinical application in one short decade. The family to which microRNAs belong—short (21 to 22 nucleotides), non-coding RNAs—was known for many years to be important for development of the roundworm Caenorhabditis elegans. But it was only in 2001 that microRNAs were found to be ubiquitous in vertebrates as well. Over the next few years alterations in microRNA expression profiles were associated with cancers and other disease conditions. Now the technology is coming to fruition, with half a dozen microRNA-based assays for diagnosis and prognosis either available or expected later this year. Impressive accomplishments indeed for a 10 year old.
With such a strong promise to improve patient care, microRNA might be regarded less as a weed and more like the genetically engineered strain of high-producing rice that fostered the Green Revolution. It’s too early to say whether microRNA will foster a laboratory revolution. But molecular pathologists who work with microRNA expect big things from this precocious youngster.
“In all of the years I have been doing molecular pathology, I don’t think we have ever had a biomarker or a class of biomarkers that is this exciting,” says Gregory J. Tsongalis, PhD, HCLD, professor of pathology at Dartmouth Medical School. Dr. Tsongalis, who is director of molecular pathology and co-director of the translational research and pharmacogenomics programs, attributes the excellence of microRNAs as biomarkers to their high specificity, either for cell type or tissue type. “And now,” he adds, “we are learning that they are specific for disease type as well.”
Specificity is a prominent reason why microRNAs have been adopted so fast, Dr. Tsongalis says. New informatics tools are another. “Not only do we know what microRNAs are, but they are being quickly linked to either up- or down-regulation of gene expression,” Dr. Tsongalis says. “On the oncology side, those genes [that are regulated by microRNAs] are the ones we think about when we talk about molecular genetic cancer. It appears that microRNAs are allowing us to answer clinical questions that we couldn’t address before, or for which existing biomarkers gave us equivocal results.”
Dr. Tsongalis has collaborated with a diagnostics company, Asuragen, to explore microRNA in pancreatic cancer. “We asked whether we can differentiate normal pancreas from pancreatic cancer using a microRNA profile,” he told CAP TODAY. “We found that we can.”
Others are equally optimistic. “MicroRNAs have the potential to add tremendously to our diagnostic ability,” Timothy J. O’Leary, MD, PhD, says. Dr. O’Leary, who is deputy chief research and development officer in the Office of Research and Development, Department of Veterans Affairs, agrees that “Alteration of microRNA expression is common in cancer, and in many cases these profiles differ between benign and malignant tumors along a prognostic spectrum.” One way that microRNAs will be useful is as an add-on to traditional anatomic pathology tools in helping to determine the primary site of a metastatic tumor of unknown origin.
“Because microRNAs are regulated similarly to other elements affecting gene expression,” Dr. O’Leary says, “alterations of microRNAs are directly tied to pathogenesis of disease. This means their method of action is amenable to therapeutic intervention by small molecules. So, in addition to being of immense diagnostic significance, microRNAs are of therapeutic importance, not only in cancer but in other diseases as well.”
Another practical virtue of microRNAs: “They are relatively easy to assess,” he says, “so microRNA assays are likely to be available to more labs than many other molecules in the past.”
Several talks at this year’s meeting of the Association for Molecular Pathology will focus on microRNAs and their applications in oncology. Federico Monzon, MD, says members of the AMP program committee selected the topic because molecular laboratories are likely to be dealing with microRNA assays in the near future. “In the last decade, microRNAs have come to prominence as important regulatory tools in cell differentiation,” says Dr. Monzon, associate professor of pathology at Methodist Hospital System in Houston. “They are important both for normal development and for oncogenesis and appear to be implicated in a variety of tumors, so they are being explored as a new class of biomarkers.”
MicroRNAs are good candidates for biomarkers for two reasons, he says. First, they are fairly specific to cell lineages. “Because microRNAs are involved in differentiation, cells from the same type of tissue express the same microRNA profile,” he explains. Second, because microRNAs are very small molecules, they are well preserved in formalin-fixed paraffin-embedded tissue samples. “This allows us to extract microRNA from archived blocks, so they lend themselves to be easily tested with a variety of methods, such as qPCR and microarrays,” Dr. Monzon says.
Precisely how microRNAs are involved in oncogenesis is not known, he cautions. “However, we do know that microRNAs are dysregulated in almost every tumor that people have looked into, which leads to applications using microRNA expression to differentiate between cancer and noncancer and between aggressive and nonaggressive cancer.” Which application of this type will be first to be adopted widely? “I don’t see a clear frontrunner right now,” he says.
Mahesh M. Mansukhani, MD, associate professor of clinical pathology and director of the molecular pathology laboratory, Columbia University Medical Center, collaborated with a second company that is commercializing microRNA technology, Rosetta Genomics, on an assay to distinguish squamous from nonsquamous non-small-cell lung cancer, or NSCLC (Lebanony D, et al. J Clin Oncol. 2009;27:2030–2037). In that work, he and colleagues demonstrated the utility of microRNA-205 expression for this purpose. The initial value of microRNA in diagnostics will be focused rather than broad, Dr. Mansukhani says. “In certain areas right now the biggest value of microRNAs is in identifying cell type,” he says. “This is because the expression levels of many microRNAs differ by orders of magnitude between cell types.” And this is why, he adds, they would be useful in identifying cellular types of tumors of unknown origin, as opposed to differentiating between tumor and nontumor where differences are more subtle and require assays that are more precise.
Rosetta Genomics offers two versions of a tumor of unknown origin test (miRview mets, miRview mets2), as well as an assay for mesothelioma (miRview meso) and the assay for differentiating NSCLC (miRview squamous). An assay for differentiating four histologic types of lung cancer, miRview lung, was introduced in mid-July, says Ayelet Chajut, PhD, executive VP of R&D and head of molecular biology at Rosetta Genomics Ltd. Rosetta expects to launch miRview kidney before the end of the year. (Rosetta’s assays were originally marketed by Prometheus under the “ProOnc” name.)
Asuragen offers a test for pancreatic cancer and has assays for colon and lung cancer in the late stages of validation, says Rollie Carlson, PhD, company president. While the current pancreatic cancer test is done on FFPE tissue, the assay is now being validated for use with fine-needle aspirates. Uptake of the current pancreatic cancer test is “modest,” Dr. Carlson acknowledges: “We wanted to prove that the markers could distinguish chronic pancreatitis from cancer tissue. We expect much greater uptake for the FNA version.”
Both companies perform their tests in their own CLIA laboratories. Asuragen intends to seek FDA approval. “We can provide clinical benefit early on by doing the test ourselves,” Dr. Carlson says. “We intend quickly to make our technologies available to the broader community.”
Whether to make the Rosetta assays available as kits is “a question we often ask ourselves,” Dr. Chajut says. “For now we have decided to retain the assays in our corporate CLIA lab. In the future, we may make assay kits for sale to laboratories.” She says samples are processed in the Rosetta lab with a 10-day turnaround time at a cost of about $3,200.
Laying the groundwork for further microRNA discoveries and applications, Todd A. Fehniger, MD, PhD, assistant professor of medicine at Washington University School of Medicine, and his colleagues used next-generation sequencing to do the first complete characterization of the total “microRNAome”—all 472 known human microRNAs—in a tumor from a patient with acute myeloid leukemia (Ramsingh G, et al. Blood. 2010;116:5316–5326). The work was a proof of principle, Dr. Fehniger says: “We wanted to show that it is possible to measure expression of mature microRNAs and to evaluate microRNA genes for mutations or copy number alterations.” They also analyzed 3’ untranslated regions (3’-UTRs) of messenger RNAs (mRNAs), which are frequent targets of microRNAs. (MicroRNAs suppress translation by binding to 3’-UTRs.) The investigators concluded in their article: “Analysis of the sequence of the 3’-UTRs of all coding genes identified a single somatic mutation in the 3’-UTR of TNFAIP2, a known target of the PML-RAR oncogene.”
Now the group is studying an additional 20 to 30 acute myeloid leukemia patients, looking for aberrant patterns of microRNA expression and mutations in microRNA genes or 3’-UTRs.
Within a year of the finding that microRNA is evolutionarily widespread (Ambros V. Cell. 2001;107:823–826), it was shown to be medically important. Carlo Croce, MD, and colleagues were the first to demonstrate the role of microRNA in a disease condition, chronic lymphocytic leukemia (Calin GA, et al. Proc Natl Acad Sci USA. 2002;99: 15524–15529). “Now it seems to be involved in many diseases,” says Dr. Croce, professor and chair of the Department of Molecular Virology, Immunology, and Clinical Genetics and director of the cancer genetics program at Ohio State University. “It regulates a large fraction of genes and all genes involved in cancer pathogenesis.”
Dr. Croce spoke at the 2006 AMP meeting on his early CLL work. He will return to the AMP meeting this year to receive the Award for Excellence in Molecular Diagnostics, a sign of how fast this field is maturing. Dr. Croce’s group recently showed that a microRNA that is altered in CLL, miR-29, is also involved in lung cancer (Pekarsky Y, CM Croce. Oncotarget. 2010;1:224–227). “In CLL, miR-29 is for some reason expressed at a high level in indolent disease,” Dr. Croce told CAP TODAY. “We have developed a mouse model with the aggressive form [of CLL], which shows us some very interesting therapeutic targets.” For instance, a gene involved in control of cell death, Mcl-1, is overexpressed. In lung cancer as well, study of miR-29 has revealed potential therapeutic targets. “We can silence certain genes by introducing miR-29 into a lung cancer cell,” Dr. Croce says. “So regulating miR-29 could be used for therapy.”
Before they are used for therapy, however, microRNA will be used for diagnosis and prognosis. Dr. Croce’s group has performed microRNA profiles on individuals enrolled in a study determining whether spiral computed tomography could be an effective screen for those at high risk of developing lung cancer (one-pack-per-day smokers for 20 years). “Over five years, we detected a microRNA signature characteristic of lung cancer before the cancer was detected by spiral CT,” Dr. Croce says. “So essentially we could make a diagnosis of lung cancer before the tumors were clinically or radiologically visible” (Boeri M, et al. Proc Natl Acad Sci. 2011;108:3713–3718). “We also found a microRNA signature that correlates with the behavior of tumors—benign versus progressive,” he says.
Dr. Tsongalis raises practical laboratory considerations. For one, it’s important to be careful with the term “microRNA profile.” “We’re not talking about hundreds of microRNAs,” he says. “It can be as few as two or three or as many as 15 or 20. That gives us enough information to answer some specific clinical questions.” With this small number of analytes, rapid and simple methods can be used. “The quickest and easiest will be real-time PCR,” Dr. Tsongalis says. Because of the short length of microRNA—22 bases—traditional PCR probes would not hybridize with enough stringency. “What has been developed to be able to amplify by PCR is a hairpin loop primer,” Dr. Tsongalis says. (See illustration)
In his own laboratory Dr. Tsongalis has been working with microRNA in both chronic lymphocytic leukemia and pancreatic cancer. “In CLL we now have the ability to potentially use microRNA to identify people with more aggressive, rather than indolent, disease,” he says. One of the things he and colleagues have found is a post-treatment effect on microRNA expression in patients with aggressive disease. “There are some pretty significant differences in responders versus nonresponders. We are still looking at the data, but a handful of microRNAs may identify who may respond to therapy,” he says.
With pancreatic cancer, “What we are trying to do is push the envelope a bit,” Dr. Tsongalis says. “If the assay works for differential diagnosis, what type of specimen can we do that on?” Preferably the assay would be done on a minimally invasive specimen—FNA rather than a resected biopsy. Dr. Tsongalis has been part of Asuragen’s studies that have shown the viability of that approach (Szafranska AE, et al. Clin Chem. 2008;54:1716–1724).
Dr. Monzon and his colleagues have been exploring expression of microRNAs in different types of renal tumors (Powers MP, et al. Diagn Mol Pathol. 2011;20:63–70). They found significant differences in expression in 18 microRNAs among the four tissue types. “We are also exploring the association of microRNAs with response to therapy in renal tumors,” Dr. Monzon says.
Technical considerations should not slow this field, he says. “There are many providers of primers and probes [for qPCR] that can be used to detect microRNAs. Right now the barrier is to have robust clinical associations between a specific microRNA or a microRNA profile and a clinical decision point in which that microRNA result is going to trigger a specific decision—to confirm a diagnosis, determine tumor origin, or label a tumor as more aggressive.”
A technical advantage of microRNAs is that they are small, so they do not get degraded by RNases. “As a result, they are well preserved in most diagnostic specimens, including body fluids,” Dr. Monzon says (Wittmann J, HM Jack. Biochim Biophys Acta. 2010;1806:200–207).
In the total microRNAome analysis that Washington University investigators have done, Dr. Fehniger says some of the expression patterns in the AML patient were “remarkable.” For instance, one microRNA, miR-223, was found to make up about 50 percent of all microRNAs in the patient’s leukemic blasts. “It is hard to know the meaning of that until we do microRNA profiling in more patient samples,” Dr. Fehniger says.
Regarding the mutation in the gene previously implicated in AML, TNFAIP2, Dr. Fehniger says it created a microRNA binding site in the gene’s 3’-UTR that would be predicted to make that gene susceptible to microRNA-mediated repression. The investigators proved that by comparing mutant and normal 3’-UTRs for their binding capacity. “This is one of the first examples where a mutation in a 3’-UTR can lead to alteration in microRNA targeting and potentially to leukemia,” Dr. Fehniger says. “There is a good chance that this microRNA expression in cells themselves or more recently measured in plasma could be used as markers of prognosis.” Many retrospective studies have suggested the prognostic utility of microRNAs, he says, adding, “The next step is to validate and confirm that microRNAs can be used as prognostic biomarkers.”
Ultimately, it is possible that microRNAs may give rise to specific therapeutics. In AML pathogenesis, for instance, direct microRNA treatment could come via a microRNA antagonist or mimetic. “There are a number of challenges in translating these concepts into treatment in the clinic,” Dr. Fehniger cautions. The biggest challenge: “How does one deliver a specific microRNA or antagonist to cancer cells without altering normal, nonmalignant cells’ biologic processes?”
Asuragen’s pancreatic test distinguishes cancer from chronic pancreatitis. It was initially developed on FFPE samples. “We just finished a multicenter validation of that test on FNAs and will make it available in the next three months,” Dr. Carlson says. In the FNA version the number of microRNAs will probably go from two to five. No test is available now for this application except for some antibody assays which, Dr. Carlson says, have poor specificity. “We think our FNA assay will be useful on pre-surgery biopsies,” he says. “Doing FFPE at surgery is making the determination late in the game.”
Asuragen is continuing validation studies on a colon cancer assay with large clinical centers, says Elizabeth Mambo, PhD, senior scientist at Asuragen. “There are a number of genetic markers that are proposed to be prognostic for colon cancer,” she says. “We want to identify microRNAs that will outperform or complement those markers.” In work presented at the 2009 AMP meeting, Asuragen’s nine-microRNA assay was more accurate in predicting which stage-two colon cancers would recur than microsatellite instability, which is sometimes used for this purpose.
Validation is also continuing on an eight-microRNA serum-based assay to diagnose lung cancer and to distinguish it from benign lung conditions among those who are at high risk because of age and smoking history.
A company that was spun off from Asuragen, Mirna Therapeutics, is looking at microRNA replacement therapy. It has six microRNAs that it hopes to put into human trials in lung and prostate cancer in the next year, Dr. Carlson says. Therapies include microRNA mimetics and natural microRNAs in controlled-release systems.
Most of the work on Rosetta’s microRNA assays has been presented only in poster form. An exception is the classification assay for NSCLC (Bishop JA, et al. Clin Cancer Res. 2010;16:610–619). Dr. Mansukhani of Columbia University calls this assay “highly accurate.” However, he wonders whether its difference in performance relative to conventional techniques—a combination of morphology and two to three immunostains—is worth the price. No direct comparison has been done. For the assay classifying lung cancer into four histologic types, Rosetta’s Dr. Chajut quotes a preliminary figure of 90 percent accuracy. Again, no comparison was done with conventional methods. “However,” Dr. Chajut says, “we need to remember that the gold standard is the diagnosis from a pathologist, who has to base his or her diagnosis on unreliable immunohistochemical stains.”
Rosetta’s tumor of unknown origin assays, miRview mets and miRview mets2, will compete with two existing assays: Pathwork Diagnostics’ FDA-approved Tissue of Origin test and bioTheranostics’ Cancer Type ID test. Both of the existing assays are based on mRNA. Dr. Mansukhani says, “MicroRNAs are much more robust markers of tissue of origin than mRNA. They are almost like degraded RNA, so they won’t degrade further. MicroRNA even works with decalcified specimens.” No comparison has yet been done, so it is not yet known whether microRNA’s advantages will translate into clinical superiority. However, data have been published on each assay (Rosetta: Rosenwald S, et al. Mod Pathol. 2010; 23:814–823; Pathwork: Pillai R, et al. J Mol Diagn. 2011;13:48–56; Grenert JP, et al. Clin Chim Acta. 2011;412: 1462–1464; bioTheranostics: Erlander MG, et al. J Mol Diagn. epub ahead of print June 27, 2011).
Dr. O’Leary takes a broad perspective on microRNA assays. “Molecular diagnosis is a rapidly evolving field, with continuous introduction of new technologies,” he says. “We very seldom get head-to-head comparisons. That’s why it’s important for lab directors to be able to assess each new approach.” Responding to the claim of high rates of misclassification by conventional diagnosis, he says, “To me this issue is misplaced. The real issue for patients is not the anatomic pathologic diagnosis, but the implications of diagnosis for medical treatment. It’s more important to work on those elements than to strive to match a classic classification system going back to Virchow.”
Look forward, not backward, Dr. O’Leary urges. We would expect such a future orientation to be copacetic to a brash young newcomer like microRNA. Time and data will determine whether microRNA is the new champion or just another contender.
William Check is a medical writer in Wilmette, Ill.