Background and purpose: The clinical features and molecular markers of primary myelofibrosis (PMF) in Asian population have rarely been reported. We examined the clinical relevance of molecular markers in a large cohort of PMF patients in Taiwan.

Methods: Bone marrow or blood samples at initial diagnosis from 145 patients consecutively diagnosed with PMF based on WHO criteria in Chang Gung Memorial Hospital-Linkou, Taiwan, were examined. EEC assay was performed in a serum free culture system. PRV-1 mRNA expression in granulocytes was measured by real-time RQ-PCR TaqMan assay. Pyrosequencing was used to detect JAK2 V617Fand its allele burden as well as 46/1 rs12343867 genotype in granulocytes. Mutational analysis of MPL (exon 10) was performed by PCR assay followed by direct sequencing. CALR (exon 9) mutations were screened by GeneScan analysis followed by sequencing for those with length changes. Ten of 20 patients progressed to secondary AML (sAML) had matched paired diagnosis and sAML samples available for comparative analysis.

Results: Of the 145 patients with PMF, the median age was 64 years, 76 were male, IPSS low risk 25, Int I 23, Int II 41, and high risk 56 patients. In a median follow-up of 35.8 months (range 1.1 to 275.5 months), 20 patients progressed to sAML, 88 patients died with a median overall survival (OS) of 67.4 months. JAK2 V617F was detected in 52% (74/143) patients, CALR mutations in 30% (41/135) (type1 n=29; type 2 n=5; and others n=7), MPL mutations in 4% (5/141) (n=2/2/1 for W515L/K/A), and 11.0% of PMF patients were triple-negative. The incidence of 46/1 haplotype in 112 patients analyzed was TT 32 %, CT 36 %, and CC 32 %; C-allele frequency was significantly higher in PMF compared with 50 normal subjects (50% vs. 24%; P< 0.0001).EEC growth was detected in 48.9% (45/92) of patients examined. PRV-1 over-expression was present in 40% (28/70) of patients. Of the 10 matched paired PMF/sAML samples, 6 patients had CALR mutations with similar allele burden at both phases of disease whereas sAML evolved from a non-JAK2 V617F clone in one of the 3 patients carrying JAK2 V617F at diagnosis. Patients with EEC growth or PRV-1 over-expression were significantly associated with younger age, higher WBC and platelet counts. EEC-positive patients had higher Hb level and lower circulating blasts. JAK2 V617F was closely associated with higher WBC and platelet counts whereas patients with CALR mutations had lower WBC counts. None of these molecular markers had a correlation with constitutional symptom, IPSS, occurrence of thrombosis or risk of sAML transformation. EEC growth conferred a favorable leukemia-free survival (LFS) (P =0.019) and OS (P =0.013) compared with those without EEC. PRV-1 over-expression was associated with better OS (P =0.036). JAK2 V617F and MPL mutations did not influence LFS and OS. Allele burden of JAK2 V617F had no impact on outcomes. CALR mutations were associated with a favorable OS compared with mutation-negative patients (P =0.034). There were no difference in outcomes between type 1 and type 2 mutations of CALR. Patients with triple-negative mutations had a significantly inferior OS (P =0.020). CT genotype (46/1) was associated with shorter LFS (P =0.026). EEC growth was strongly associated with PRV-1 over-expression and JAK2 V617F mutation, whereas EEC formation and CALR mutations were mutually exclusive. In multivariate analysis, EEC growth was the most important predictor for LFS (HR 0.058; 95% CI: 0.005-0.676, P =0.023) and OS (HR 0.21; 95% CI 0.076-0.581, P =0.003) among the molecular markers; CALR mutations also held favorable OS (HR 0.245; 95% CI 0.085-0.709, P =0.009).

Conclusions: Approximately 90% of PMF patients in Taiwan had JAK2 V617F, CALR, or MPL mutations, half were associated with C-allele genotype, 78% had EEC growth and /or PRV-1 over-expression. EEC growth was the most important independent factor for predicting better outcomes and CALR mutations also conferred a favorable OS. (Grant support: NSC96-2314-B-182-003, CMRPG330303, OMRPG3C0021, and MOHW103-TD-B-111-09)


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

Sign in via your Institution