Longitudinal study of 2 patients with cyclic thrombocytopenia, STAT3 and MPL mutations

The production of megakaryocytes and platelets in the bone marrow (BM) is strictly regulated by the balance of thrombopoietin (TPO) and platelet mass to maintain platelet homeostasis. Patients with cyclic thrombocytopenia (CTP) have substantial periodic platelet oscillations with amplitudes ranging from thrombocytopenic to normal/thrombocytosis levels. Patients with CTP typically have a cycling time of 3 to 5 weeks.^1,2 Various potential etiological associations including cyclical megakaryopoiesis, menstrual cycles, cyclical autoimmune platelet destruction, and T-cell clonality have been observed, yet the pathogenesis remain elusive.^1-3 

The oscillatory nature of hematopoiesis in general has been recognized and mathematically modeled as a nonlinear dynamical system with feedback loops involving time delays.⁴ After perturbations, such systems entail oscillations with a constant period of about twice the time delay, and certain conditions introducing moderate thrombocytopenic baselines may increase the “gain” of the system and result in a stable and exaggerated oscillatory stage. Morley, Mackey, and others postulated that CTP could be a consequence of errors in or perturbation of thrombopoiesis.^5-7 Biological findings correlating with these concepts are needed to improve specific modeling for CTP.⁷ We hypothesized that inherited or acquired genetic mutations affecting thrombopoiesis could create a vulnerability of this system to transition to a CTP phenotype and that patterns of gene expression during cycling could shed light on the regulation of this process.

The study was approved by the Stanford University Institutional Review Board and conducted in accordance with the Declaration of Helsinki. Two patients enrolled with informed consent. Twenty-four consecutive blood samples from patient 1 (CT1) and 15 from patient 2 (CT2) were collected every 3 to 4 days.

MPL was Sanger sequenced. MPL c.1210G>A mutation was constructed. Wild-type (WT) or the mutant MPL cDNA was stably transfected into Ba/F3 cells for functional analyses.

The Stanford Tumor Actionable Mutation Panel for Hematopoietic and Lymphoid Neoplasms, or Heme-STAMP in short, is a targeted next-generation sequencing (NGS)-based panel. Heme-STAMP was performed on whole blood DNA to screen for hematolymphoid neoplasm–associated mutations in 164 genes.

Peripheral blood mononuclear cell DNA were used for NGS-based T- and B-cell receptor clonality profiling (LymphoTrack, Invivoscribe).

Blood RNAs were extracted after red blood cell lysis and sequenced for longitudinal transcriptome analysis using SAMseq algorithm to investigate platelet count–associated gene expressions.

Details of these methods are provided in supplemental Methods.

CT1, a 58-year-old male, exhibited platelet count oscillation (range, 1-409 × 10⁹/L; mean, ∼128 × 10⁹/L) and an inverse plasma TPO oscillation (6-2745 pg/mL) with a prolonged cycling period of 39 days.⁸ CT2, an 86-year-old male, also exhibited reciprocal platelet count (range, 7-142 × 10⁹/L; mean, ∼65 × 10⁹/L) and TPO oscillations (12-405 pg/mL) over a period of 26 days (Figure 1A). Neither patient had known comorbidities or medications known to affect platelet number or function.

Whole blood transcriptome analyses identified 236 genes that were quantitatively correlated with platelet count in both patients. These genes can be parsed into 3 groups based on their expression patterns (Figure 1B; supplemental Table 1). The largest group consisted of 151 platelet-specific genes that oscillated preceding platelet count fluctuations by 3 to 4 days (Figure 1B-D). Thirty-four of these 151 genes (eg, MYL9, TUBB1, ITGA2B) demonstrated over 50-fold and 10-fold expression changes during platelet count cycles in CT1 and CT2, respectively (Figure 1B). As platelet RNAs are inherited from megakaryocytes and are positively correlated with the number of newly formed platelets,⁹ the expression patterns suggest cyclical platelet production underlie platelet count oscillations in both patients (Figure 1E).

Both patients also shared fluctuations of immature neutrophil-specific^10,11 (eg, AZU1, ELANE, LCN2) and erythrocyte-specific (eg, ALAS2, EBP42, HBM) genes, with cycling patterns preceding platelet-specific genes (Figure 1B-D). Gene ontology analysis showed enrichment of neutrophil-specific and erythrocyte-specific biological processes in these platelet count–associated genes (supplemental Table 2). In contrast to gene expression patterns, only neutrophil counts in CT1 showed a cycling trend, all within normal range (supplemental Figure 1). This may be due to dampening (buffering) effects of marrow neutrophil release control and long lifespan of erythrocytes that absorbed the mild production fluctuations.⁴ 

The TPO and trilineage gene expression patterns are consistent with TPO-mediated oscillations⁴ in both patients. For CT2, the 14d interval between TPO and platelet count oscillations resembles the normal time delay (about 12 days) of TPO effect on platelet count in healthy people,¹² whereas CT1 displayed a prolonged time delay of 21d. Moreover, compared with CT2 and prior CTP cases,^3,13-16 CT1 had much higher TPO peak levels, similar to those observed in congenital amegakaryocytic thrombocytopenia.¹⁷ These prompted us to sequence the TPO receptor, MPL, in CT1, which revealed a novel germline c.1210G>A heterozygous mutation, resulting in a p.Gly404Arg (G404R) substitution in a highly conserved site (Figure 2A-B; supplemental Figure 2). No MPL mutation was detected in CT2.

In vitro functional analysis demonstrated that MPL G404R is a loss-of-function mutation with deficiencies in TPO-stimulated growth, TPO uptake, and TPO pathway activation (Figure 2C-E; supplemental Figure 2). Thus, this mutation may explain the unusually high TPO peaks without rebound thrombocytosis, and it may also slow the TPO-mediated thrombopoiesis, explaining the prolonged cycle period in CT1. Family studies demonstrated a paternal origin of the mutant allele. The father exhibited nonperiodic fluctuations in platelet count (166-223 × 10⁹/L) during an 80-day period. Therefore, this MPL mutation is insufficient to cause CTP.

Heme-STAMP testing of peripheral blood revealed somatic STAT3 variants in both patients: CT1 had STAT3 Y640F (2% VAF) and STAT3 D661Y (1%VAF); CT2 had STAT3 D661Y (6% VAF) (Figure 2A). These STAT3 variants are pathogenic gain-of-function mutations first identified in T-cell large granular lymphocyte (T-LGL) leukemia and are associated with immune-mediated cytopenias, including neutropenia or pure red cell aplasia, even when present at subclonal (low VAF) levels.^18-21 Background LGL populations were observed in both CT1 and CT2 (supplemental Figure 3), and T-LGL lymphoproliferation has been reported in 2 patients with CTP.^3,22 

NGS profiling demonstrated clonal T cells in both patients (Figure 2F). This is consistent with a recent report where clonal T-cell receptor rearrangement was found in 6 out of 8 patients with CTP, including 1 who later developed T-LGL leukemia.³ The clonal populations did not show cyclical variation in either patient (Figure 2G).

Additional findings were identified in each patient. CT1 had a monoclonal B-cell population by flow cytometry and B-cell receptor clonality profiling (Figure 2A; supplemental Figure 4). CT2 had pathogenic gain-of-function mutations in the last exon of PPM1D (R458X, 1% VAF; L484X, 2% VAF), which impair the DNA damage response.²³ 

In summary, our study revealed novel genetic and cellular associations in 2 patients with CTP, who displayed exaggerated oscillations with their intrinsic periodic rhythms. The fact that pathogenic somatic gain-of-function STAT3 mutations and stable clonal T-cell populations are identified in both patients and are associated with autoimmunity and cytopenias suggest their involvement in the pathogenic baseline of CTP in these patients. In line with the previous model,⁴ we hypothesize that these factors could impair thrombopoiesis, increasing gain of the system, leading to overcorrection and persistence of cycling in the setting of perturbations eliciting a thrombopoietic response. In support of this concept, a CTP phenotype linked to hydroxyurea therapy has been observed in patients with JAK2 V617F–positive polycythemia vera.^24,25 

The true incidence of underlying somatic and germline genetic alterations in CTP such as seen here in STAT3 and MPL will need to be explored in additional patients. Identifying these mutations in patients who require therapy could represent a potential opportunity for individualized, novel targeted therapeutic approaches.

This work was supported by a grant from the National Institutes of Health (NIH National Cancer Institute) (2P01CA04960529A1) and an award from the Department of Pathology, Stanford University (Pathology in Precision Health Research Award).

Contribution: H.Z., M. Chien, Y.H., W.S., R.S.B., C.H., P.H., D.X., N.C.W., A.G.T., J.J., W.H.W, Z.C., and P.A. performed experiments; H.Z., Y.H., B.M.Z., X.G., L.L.T., D.J., M.S.K., and M. Craig analyzed the data; H.Z., J.R.G., and J.L.Z. designed the study; H.Z., J.B.B., and J.L.Z wrote the paper; and all authors provided critical reviews and edits.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: James L. Zehnder, Department of Pathology, Stanford University School of Medicine, 300 Pasteur Dr, Lane 235, Stanford, CA 94305; e-mail: zehnder@stanford.edu.