Dear Editor,

Calreticulin (CALR) gene mutations (CALR mut) have recently been discovered in about 20–35 % of patients affected by essential thrombocythemia (ET) and primary myelofibrosis (PMF) [1, 2]. Several molecular assays have been developed to detect the most frequent CALR mut (type 1 consisting of a 52-bp deletion, and type 2 of a 5-bp insertion) [3, 4]. All these techniques are useful for identifying CALR mut at the diagnosis, but they are not suitable for minimal residual disease (MRD) monitoring, since the maximum sensitivity is 1 %. The droplet digital PCR (ddPCR) technology is a third-generation PCR method that started to be used in hematological malignancies [5–7]. We describe a ddPCR assay with a sensitivity of 0.01 % developed for the absolute quantification of CALR type 1 and 2 mutations and analyze a cohort of 57 JAK2V617F-negative myeloproliferative neoplasm patients. ddPCR experiments were performed using the QX-200 instrument (BioRad) and specific primers and probes were designed for both type 1 and type 2 mutations (see Supplementary Files). CALR mut load in each sample was expressed as fractional abundance (FA, mutant allele/mutant allele + wild-type allele). The CALR mut allelic burden resulted heterogeneous in both ET (min.13.8 %–max. 51 %) and PMF (min. 34.5 %–max. 51.3 %) patients. We show that the median CALR mut allelic burden at diagnosis was significantly higher in PMF patients as compared to ET case (47.9 vs 43.8 %, p = 0.008) whereas no significant difference was observed between the type 1 and 2 mutations (Fig. 1a). Moreover, no relationship between the gene mutation type and the CALR mut amount was observed within each group of ET and PMF patients. In our ET series, there were 14 (29.7 %) patients with a very low FA, <30 %; this group was not statistically different in terms of hemoglobin, white blood cells and platelet counts, age, sex, thrombosis and/or hemorrhage, and CALR mut type compared to those with FA >30 %. The PMF group included ten patients, too few for any type of statistical considerations.

Fig. 1
figure 1

a Distribution of CALR type 1- and 2-mutated copies determined by ddPCR and expressed in FA in 57 MPN patients. The median of the CALRmut allelic burden at diagnosis resulted to be 43.8 % in ET patients and 47.9 % in PMF, showing a statistically significant difference (p = 0.008). Each dot represents a patient. The lines indicate the median for each group. b Assessment of CALR mutation load by ddPCR in a PMF patient receiving ABMT. CALR mutation was not revealed at the time of AML evolution, whereas a low positivity (FA 0.01 %) was detected 2 months after ABMT and five months before hematological relapse. c 1-D plot showing each droplet corresponding to CALR mutations plotted on the graph of fluorescence intensity versus droplet number. All positive droplets are indicated in blue, whereas negative droplets are indicated in gray

Full size image

Sequential evaluations by ddPCR experiments were performed in three patients to monitor the CALR mut load during treatment. CALR mut load at diagnosis was 15.8 and 48 % in two ET patients. The former patient was treated with interferon-α (IFN-α) and after 5 years from diagnosis the FA was 7.7 %. The latter was also treated with IFN-α and after 2 years from diagnosis, the CALR mut load was 14.7 %. Both patients had stable disease and a well-controlled platelet count. A 44-year-old man at PMF diagnosis showed a FA of 49.7 %; 8 years later, we observed a leukemic transformation. At the time of the AML evolution, the CALR mut load was 0 %; this finding was also confirmed by PCR qualitative analysis. The patient underwent induction chemotherapy, achieving complete remission, then allogeneic bone marrow transplantation (ABMT) from a HLA-matched related donor. Two months later, the FA observed by ddPCR analysis was 0.01 %; 7 months after, the ABMT and AML relapsed and at this time, the CALR mut load was 13.5 % (Fig. 1b, c).

Although the importance of the CALR allelic burden determination has not yet been defined at the disease onset, the utility of and need for a sensitive method, like our ddPCR assay, are unquestionable for the purposes of MRD monitoring [8–10].