A genetically inspired prognostic scoring system (GIPSS) that stratifies primary myelofibrosis (PMF) patients by genetic variants alone was recently proposed. While non-inferior to the dynamic international prognostic scoring system (DIPSS), the lack of overlapping prognostic variables between the models leads to increased risk for disagreement between two valid prognostic models and presents a challenging clinical situation. In an external cohort of 266 molecularly annotated myelofibrosis (MF) patients, we demonstrated that the GIPSS model significantly differentiated between four risk groups (low, int-1, int-2, high) with median OS that was not reached, not reached, 60.5 and 28.9 months, respectively. High-risk patients had significantly inferior leukemia-free survival (LFS) (P < 0.0001). We identified a cohort of prognostically ambiguous patients (n = 39) in which GIPSS and DIPSS models differed by ≥2 risk groups. Among these patients, a similar proportion were up-staged by DIPSS (n = 19) and GIPSS (n = 20). Patients upstaged by GIPSS (genetically high-risk) had a trend toward inferior OS compared with patients upstaged by DIPSS (clinically high-risk) (P = .08) and significantly worse LFS (P = .04). Patients deemed intermediate-2 and high-risk by GIPSS who underwent allogeneic transplant had improved OS compared with those that did not (P = .04). GIPSS is a valid disease-specific prognostic system and outperforms DIPSS in patients where the two models disagree. Additionally, while GIPSS was developed for PMF; the current study shows, however, that the contemporary genetic model performs equally well for both primary and secondary myelofibrosis.
© 2018 Wiley Periodicals, Inc.