Many roads lead to MPN

In this issue of Blood, Lundberg et al correlate the presence of known mutations in patients with myeloproliferative neoplasms (MPNs) with clinical outcome, thereby proposing a molecular risk stratification.¹ 

The clinical presentation of patients with MPNs is heterogeneous, and the individual disease course is difficult to predict at diagnosis. Although in some patients the disorder remains indolent for many years, others experience multiple complications and rapid disease progression. It is therefore gratifying to read that Lundberg et al can corroborate this clinical heterogeneity at the molecular level.¹  The authors investigated the “clonal architecture” of MPNs, that is the nature of different mutations detected in individual patients and the order in which they appear.

Because the authors selected known cancer genes for analysis, many of which have been previously shown to be affected in MPNs, the message of this study is less in the nature of the mutations found but rather in the variable pattern of their acquisition, which this study demonstrates. However, it is noteworthy that in this cohort, mutations in some novel genes, such as p53 and NF-E2, appear to be more frequent than others, such as c-Cbl or c-Mpl, which have been known for several years.^2-4  A model presented in Figure 5 of the Lundberg et al paper depicts the many different constellations observed and uses them to stratify patients by risk of leukemic transformation. As is the case in other myeloid neoplasias, such as acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS),^5,6  a higher number of mutations is associated with poorer outcome. In MDS, the number of mutations is likewise correlated with the time to leukemic transformation.⁶  The question remains, however, whether the acquisition of additional mutations is inherent in the disease process. If so, it would occur stochastically such that individual risk could not be assessed a priori. Alternatively, patients could present with high risk mutations at diagnosis—these then simply need to be identified and considered when assessing therapeutic options.

One hypothesis for explaining the acquisition of mutations and the observed variability in disease progression is the presence of a “hypermutable state” in MPNs.^7,8  In this model, individual patients are thought to acquire different additional mutations during disease progression, and the nature of these mutations directs the clinical phenotype. The data presented by Skoda’s group clearly argue against a hypermutable state in MPNs. The vast majority (95%) of all mutations detected were already present in the first sample analyzed. In addition, using 2 different methods, the authors calculate a mutation rate of 1 mutation in the genes analyzed in 45 to 66 patient-years. This calculation, however, raises 1 question. Of the 197 patients analyzed, 33% already carried ≥2 mutations. The average age at diagnosis in this patient cohort was 51 years for essential thrombocythemia (ET) and 58 or 61 years, respectively, for polycythemia vera (PV) and primary myelofibrosis (PMF) patients. Nonetheless, 27% of the ET patients had already acquired ≥2 mutations, in a time frame calculated to suffice only for 1 mutation. Prior to disease manifestation, therefore, some patients must have incurred a higher mutation rate.

Because transformation to acute leukemia, which is often highly refractory to treatment, is clinically the most challenging complication experienced by these patients, early predictors of leukemic risk are of utmost importance. Lundberg et al show that mutations in the tumor suppressor p53 are present at very low levels (so-called “subclonal levels,” where a very small percentage of the patient’s cells carry the aberration) in a small number of MPN patients. With 1 exception, 4 of the 5 patients carrying p53 mutations transformed to AML, with a latency of between 5 and 10 years. Because the p53 mutations were observed at such low levels, modern sequencing technologies (next-generation sequencing [NGS]) are required for their detection.

Given their clinical importance, detection of p53 mutations by NGS should be considered in MPN patients, especially in light of 1 unexpected observation in this data set: of the 5 patients in which p53 mutations were identified, 3 were diagnosed with ET, generally considered to carry a significantly lower risk of leukemic transformation than PV or especially PMF. Moreover, ET is frequently diagnosed at a younger age, as also seen in the current cohort. ET patients with p53 mutations therefore present with an unanticipated high risk, one that is inapparent by clinical means, but may frequently be good candidates for bone marrow transplantation (BMT). BMT is the only curative approach to MPNs and one that may preempt leukemic transformation. A similar argument may apply to select patients with TET2 mutations, which are also shown to be associated with poor outcome in this study. However, the rate of leukemic transformation in TET2-mutated patients was only 30% compared with the 80% in p53-mutated patients (with the caveat of small number errors in both cohorts); hence, the decision for BMT must consider this.

The rapidly decreasing costs of NGS analysis may soon allow an economical use of this technology for the detection of clinical risk in MPN patients, a cohort that today appears undiscernibly heterogeneous in outcome. In this way, MPNs may follow the successful path forged by >10 years of molecularly guided therapeutic trials in AML, which have led to both improved molecular risk stratification and the development of targeted therapies for select molecularly defined groups of patients.⁹