About 25% of patients with essential thrombocythemia (ET) or primary myelofibrosis (PMF) carry a somatic mutation of CALR, the calreticulin gene [N Engl J Med. 2013;369:2379-90]. So far, more than 50 different indels in CALR exon 9 have been found, but a 52-bp deletion (type 1 mutation) and a 5-bp insertion (type 2) are the most common lesions. All indels generate a novel C-terminus of the mutant protein, in which the endoplasmic reticulum retention signal KDEL is lost and the negatively charged amino acids are replaced by neutral and positively charged amino acids, disrupting the Ca-binding site. This suggests that both cellular dislocation and impaired Ca-binding activity may be involved in the abnormal proliferation of cells expressing a mutant calreticulin. It is still unclear, however, why the same mutant gene is associated with 2 different disease phenotypes (ET and PMF). In particular, little in known about the effect of the mutant protein on megakaryocyte biology and bone marrow collagen deposition.

We studied the relationships between CALR mutation type, megakaryocyte biology, and clinical phenotype in patients with myeloproliferative neoplasms. According to the 2008 WHO criteria, 716 out of 892 patients had ET and 176 had PMF. Overall, 578 (65%) patients carried JAK2 (V617F), 230 (26%) had a CALR indel, and 84 (9%) had nonmutated JAK2 and CALR. Patients with MPL mutations were excluded.

Twenty-six different types of CALR lesions were identified: 120 (52%) patients had type 1 mutation, 75 (33%) had type 2, and 35 (15%) carried other indels. The frequency of type 1 mutation was significantly higher in PMF than in ET (71% vs 46%, P=.004). All these variants involved 3 different stretches of negatively charged amino acids, with an increase in the isoelectric points (pI) of the mutant protein. As type 1 and type 2 mutations affected stretch I and III, respectively, the 26 indels were categorized into 3 groups on the basis of the stretch they affected: i) type 1-like (61%), affecting stretch I; ii) type 2-like (36%), stretch III; iii) and other types (3%), stretch II. The pI values were significantly different in the 3 groups (P<.001). The frequency of type-1 like mutations was significantly higher in PMF than in ET (82% vs 55%, P=.001).

In vitro differentiated megakaryocytes from CALR-mutant patients displayed a significant increase in the extent of both intracellular Ca2+ release from the endoplasmic reticulum and extracellular Ca2+ entry inside the cytoplasm, as compared with healthy controls. Megakaryocytes carrying type 1-like CALR mutations exhibited the highest amplitude of Ca2+ flows regardless of the type of disease. In ET, impaired Ca2+ homeostasis was accompanied by atypical proplatelet architecture (ie, more branches and bifurcations).

With respect to clinical phenotype at diagnosis, ET patients with type 2-like CALR mutation showed a trend towards higher PLT count (P=.063) and lower age (P=.053), and significantly lower LDH values (P=.021) than those with type 1-like mutation. In a hierarchical cluster analysis including demographic, clinical and molecular data, CALR mutation type (1 vs 2) identified the 2 clusters with the highest dissimilarity.

Considering all patients, those with type 2-like CALR lesions had a better survival than those with JAK2 (V617F) (96.1% vs 84.4% at 10 years, P=.039), while no difference was found between the 2 CALR mutation types. ET patients with type 2-like CALR mutations showed a lower risk of thrombosis than those with JAK2 (V617F) (P=.010). By contrast, ET patients with type 1-like CALR mutations had a higher risk of myelofibrotic transformation that those with type 2-like CALR mutations (P=.029) and especially those with JAK2 (V617F) (P=.011). Finally, PMF patients with type 1-like CALR variants had a better survival than those with JAK2 (V617F) (80.1% vs 48% at 10 years, P=.008).

In summary, abnormalities in megakaryocyte calcium metabolism and proplatelet architecture are found in patients with CALR-mutant myeloproliferative neoplasms, and their extent is related to mutation type. Type 2-like CALR mutations are more likely to be associated with isolated thrombocytosis without bone marrow fibrosis, ie, with an ET phenotype. By contrast, type 1-like CALR mutations are generally associated with bone marrow fibrosis, ie, with a PMF phenotype. Thus, in CALR-mutant myeloproliferative neoplasms, the mutation type is a major determinant of the clinical phenotype.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution