Blood transcriptome and clonal T cell correlates of response and nonresponse to eltrombopag therapy in a cohort of patients with chronic immune thrombocytopenia

We evaluated the longitudinal effect of eltrombopag, a thrombopoietin receptor agonist, on gene expression in heavily pretreated patients with chronic immune thrombocytopenia (ITP) using blood transcriptome analysis on pretreatment, and following 1 week and 1 month of treatment samples. A strong stimulation of 208 eltrombopag-induced genes was identified in responders at 1-week time point. 90% of these genes are present in platelets, with GATA1, THPO, VIPAS39 and TGFB1 as top upstream regulators. Subsequently, despite a continued increase in platelet in responders, the eltrombopag-induced gene expression decreased at 1 month compared with the level at 1-week time point. This novel observation of the longitudinal effect of eltrombopag therapy over time suggests different stages of eltrombopag response. Non-responders did not exhibit eltrombopag-induced gene expression changes. This study provides novel insights into the mechanisms of eltrombopag response and nonresponse. With informed consent, 17 patients with chronic ITP receiving eltrombopag monotherapy (75 mg daily via oral administration) during the period of blood sample collections were enrolled in this study. The criteria for response assessment are modified from the International Working Group guidelines (Rodeghiero F, 2009, PMID: 19005182). Response (R) in this study includes both complete response (platelet count >100 x 109/L and absence of bleeding) and response (platelet count ranges from 30 to 100 x 109/L, and at least a two-fold increase of the baseline count without bleeding) that defined in the International Working Group guideline. Nonresponse (NR) was defined as platelet count <30 x 109/L or failed to double baseline platelet count within 90 days of treatment, or bleeding. Peripheral blood samples were collected into PAXgene blood RNA tubes at pretreatment (pre), and at 1-week (1wk) and 1-month (1mon) on treatment. Total of 46 samples were included in this blood transcriptome analysis. Total RNAs were extracted from these blood samples. Human alpha- and beta-globin mRNAs were depleted prior to RNA sequencing library construction. We then performed 3’-end sequencing for expression quantification (3SEQ) analysis on these samples (Beck AH, 2010, PMID: 20098735). 3SEQ is a type of RNA-seq that focuses on quantitative analysis of transcriptome by generating a directional sequencing library targeting 3’UTRs and flanking regions in near upstream of poly-A tail and ensuring that one read is produced and measured per transcript. The sequencing libraries were constructed, and Index associated 36 bp sequences of each sample were generated using Illumina HiSeq 2000 system. Next, 3SEQ data were filtered and mapped to human transcriptome hg19, and read counts for each transcript were generated as previously described (Guo X, 2013, PMID: 24342436). For identification of differentially expressed genes, we employed significance analysis of microarrays (SAM)-Seq algorithm (SAMseq) for paired analysis between different time points of the same patient group using following cut off criteria: fold change > 2; average transcript per million (TPM) > 0.25; and q-value < 0.05. As we performed globin mRNA reduction procedure before 3SEQ library construction, the expressions of hemoglobin subunit alpha 1 (HBA1), alpha 2 (HBA2), and beta (HBB) were excluded from the SAMseq analysis.