The Prognostic Advantage of Calreticulin Mutations in Myelofibrosis Might be Confined to Type 1 or “type 1-like” Calreticulin Variants

Background: Approximately 25% of patients with primary myelofibrosis (PMF) harbor calreticulin (CALR) mutations, which have been associated with longer survival (Klampf et al. NEJM 2013). More than 80% of CALR mutated patients harbor one of two mutation variants: type 1, a 52-bp deletion (p.L367fs*46) or type 2, a 5-bp TTGTC insertion (p.K385fs*47). Recent studies have suggested phenotypic and prognostic differences between these two variants (Tefferi et al. Blood 2014, Leukemia 2014 and AJH 2014). Furthermore, data are emerging that suggest functionally-relevant structural differences between type 1 and type 2 CALR variants, including a higher alpha-helix content of the mutant C-terminus in type 2, compared to type 1 (Eder-Azanza et al. Leukemia 2014).

Objectives: We used statistical models to calculate helix propensity for thirty-one unique amino acid sequences that were altered by CALR mutations and used the results to subclassify non-type 1/2 CALR mutations into “type 1-like” and “type 2-like” variants. Subsequently, we examined the prognostic relevance of these subgroups.

Methods: Calculation of helix propensity, which is the percentage of residues that are predicted to be involved in the formation of an alpha-helix, was performed using AGADIR, which is a statistical approximation algorithm (Munoz et al. Biopolymers 1997). The helix tendency calculations were performed using conditions of pH 7.0, 5 and 25 °C, an ionic strength of 0.1 M and no N- or C-terminal protection.

Results: 532 PMF patients were screened for JAK2, CALR and MPL mutations; the respective mutational frequencies were 58%, 24.6% and 7.3%. Among the 131 CALR-mutated cases, 98 (74.8%) harbored type 1, 15 (11.5%) type 2 and 18 (13.7%) other variants. Based on predicted helix propensity scale, the “other” CALR mutations were subclassified as type 1-like (n=12) or type 2-like (n=6) and respectively grouped with type 1 and type 2 variants, for purposes of phenotypic and prognostic comparisons. The AGADIR-derived predicted helix propensity scale was 29.69 for wild-type CALR and 8.6 or 34.17 for type 1 and type 2 mutant CALR, respectively; accordingly, CALR variants with values that are close to or above the value for wild-type CALR were classified as “type 2-like” (range 26.47-36.12) and those with values close to or below the value for type 1 as “type 1-like” (range 2.11-17.3).

Comparison of “type 1/type 1-like” (n=110) and “type 2/type 2-like” (n=21) CALR mutations showed the latter to be associated with higher DIPSS-plus score (p=0.01), EZH2 mutations (p<0.01), leukocyte count >25 x 10(9)/L (p<0.01), higher circulating blast percentage (p=0.02) and palpable spleen size >10 cm (p<0.01). Comparison of “type 1/type 1-like” CALR and JAK2 mutations (n=309) showed the former to be associated with younger age, higher platelet count, lower transfusion need, higher hemoglobin level, lower leukocyte count and lower DIPSS-plus score (p<0.01 for all comparisons). None of these associations was evident during comparison of “type 2/type 2-like” CALR with JAK2 mutations.

Survival was similar between patients with type 1 and “type 1-like” (p=0.8) and between type 2 and “type 2-like” (p=0.63) CALR mutations. In contrast, survival was significantly shorter in patients with type 2 (HR 2.4, 95% CI 1.2-4.8) and “type 2-like” (HR 3.2, 95% CI 1.0-10.6), when compared to those with type 1 CALR mutations. Survival was also significantly shorter with “type 2/type 2-like” vs “type 1/type 1-like” CALR mutations (p=0.003; HR 2.5, 95% CI 1.4-4.5) and the difference remained significant when analysis was adjusted for age (p=0.047), ASXL1 (p=0.003) or EZH2 (p=0.001) mutations. Similarly, compared to JAK2-mutated cases (n=309), survival was longer in patients with “type 1/type 1-like” (HR 0.4, 95% CI 0.3-0.5) but not in those with “type 2/type 2-like” (HR 0.9, 95% CI 0.5-1.6) CALR mutations; the difference in survival between JAK2 and “type 1/type 1-like” CALR mutated cases remained significant (P<0.01) when analysis was adjusted for age, ASXL1 or EZH2 mutations or DIPSS-plus score.

Conclusions: CALR mutations in PMF might be subclassified into type 1-like and type 2-like variants, based on predicted helical propensity of their mutant C-terminus. The favorable impact of CALR mutations in PMF might be restricted to type 1 or “type 1-like” variants.