Platelet RNA chips dip into thrombocytosis

In this issue of Blood, Gnatenko and colleagues have conducted studies to determine whether platelet RNA profiling can predict causes of thrombocytosis.¹  The authors show that expression levels of only 4 platelet transcripts are able to predict JAK2 V617F–negative ET in more than 85% of samples.

Thrombocytosis is a relatively common hematologic abnormality and is associated with iron deficiency, malignancy, and chronic inflammatory processes. Chronic myeloproliferative disorders (polycythemia vera [PV], essential thrombocythemia [ET] and primary myelofibrosis [PMF]) are important albeit less common causes. Hematologists are often consulted to exclude myeloproliferative disease (MPD) as a cause of high platelet counts. Most often, the clinical picture does not present difficulties distinguishing MPD from reactive thrombocytosis (RT). Among the MPDs with thrombocytosis, ET is typically considered after PV and PMF have been excluded. The somatic mutation V617F in the Janus kinase 2 gene (JAK2) distinguishes MPD from RT, but only half of patients with ET express the JAK2 V617F mutation.²  Thus, biomarkers specific for ET patients who are JAK2 V617F–negative would have diagnostic value and possibly provide mechanistic insights.

The authors' earlier work has shown that platelet gene expression can distinguish ET from healthy persons.³  In the current study, 126 subjects were recruited: 48 healthy controls, 40 ET patients, and 38 RT patients. The investigators have fabricated a novel “platelet gene chip,” which contains probes for 432 mRNAs. Most of these transcripts are reportedly expressed exclusively or predominantly in platelets; a small subset is expressed predominantly in leukocytes, permitting assessment of leukocyte contamination. Using an initial cohort of subjects as a “training set,” sophisticated statistical analyses identified 11 transcripts, the expression of which effectively distinguished the 3 study groups. One hundred percent of the RT subjects and 87.5% of the ET patients were classified correctly using the 11 transcripts measured by the microarray analysis. Importantly, these findings were validated by qRT-PCR in 10 randomly selected subjects from each of the ET and RT cohorts. To validate further this set of biomarkers, the 11 genes were used to predict the phenotype in 31 additional subjects with thrombocytosis. Using platelet RNA, qRT-PCR was able to correctly classify 87% of these subjects. The authors provide evidence that expression levels of only 4 transcripts (HIST1H1A, SRP72, C20orf103, and CRYM) were able to predict JAK2 V617F–negative ET patients in more than 85% of samples.

RNA expression profiling has been used in a variety of diseases for the purpose of identifying novel genes involved in the pathophysiology, as well as for diagnostic and prognostic purposes. Only a few studies have considered differential platelet RNA expression in clinical hemorrhagic or thrombotic phenotypes, including complex phenotypes such as sickle cell disease or cardiovascular disease.^4-7  Platelet RNA profiling of ET has particular appeal considering that the transcripts are derived from the tissue of primary clinical and pathophysiologic interest. The success of these genomic approaches is critically dependent on the precision of the patient phenotyping, generally large numbers of subjects, and appropriate bioinformatic and statistical analyses. Regarding the precision of phenotyping, there is no “gold standard” for the diagnosis of ET, such that misclassification (other MPDs may masquerade as ET) would undermine any genetic association. Other variables—such as age, sex, and platelet-lowering therapies—could also impact on megakaryocyte/platelet gene expression. Although larger numbers of patients would be needed to replicate the results of the current work, this study, nevertheless, lays the foundation for the use of RNA expression profiling in identifying genes associated with specific platelet phenotypes. Of particular interest is the “platelet chip” generated by the authors, which could help to move this field forward, although more details on the selection of the included genes are needed. Validation of such a chip would be very helpful in overcoming the known difficulties of preparing leukocyte-free platelet preparations. Extending RNA expression profiling to other platelet phenotypes may also permit identification of genes involved in the interindividual variation in platelet reactivity and other platelet-dependent disorders of bleeding and clotting. Last, there is a great need for accurate predictors of thrombotic and hemorrhagic risk in ET. It is hoped that genomic approaches such as those used by Gnatenko et al will be fruitful in this respect, but this will likely require consortia with large numbers of patients.