Applying prognostic models for myelofibrosis in practice

Genetically-Inspired Prognostic Scoring System

The GIPSS reflects a purely genetic prognostic model and was developed to complement the MIPSS70+ v2.0. The GIPSS categorized patient karyotype as very-high-risk, unfavorable, or favorable (Table 1). The GIPSS was non-inferior to the MIPSS70+ and DIPSS and was validated for OS and LFS in a reference cohort.

Newly emerging models¹

Recent work analyzed 69 cancer genes in 2,000 patients with MPN by whole-genome sequencing. Interestingly, p53 mutations were associated with an inferior outcome. The model also integrated clinical and cytogenetic variables to create a more individualized prediction of OS and risk for transformation to acute myeloid leukemia. This MPN Personalized Risk Calculator provides a patient-specific approach based on cancer genomics.

The Myelofibrosis Transplant Scoring System (MTSS) was established to predict the prognosis for MPN patients potentially undergoing HSCT. The MTSS was evaluated in patients with PMF and post-ET/PV MF who underwent HSCT and found unfavorable outcome variables, such as age ≥ 57 years, Karnofsky performance status < 90%, platelet count < 150 Gpt/L, white blood cell count > 25 Gpt/L, HLA-mismatched donor, ASXL1 mutation, and non-CALR/MPL driver mutation. Other variables in a pre-transplant setting (constitutional symptoms, cytogenetic risk, Hb level, ≥ HMR mutations) were not found to be prognostic. The MTSS was more accurate in predicting outcomes in HSCT patients than DIPSS, MIPSS70, and MYSEC-PM.

Application of prognostic models in clinical settings¹

Currently, guidelines recommend the use of IPSS, DIPSS, and DIPSS+ for risk stratification in PMF. Although these models are easy to apply in daily clinical practice, risk stratification approaches that incorporate molecular and cytogenetic information such as MIPSS70+ v2.0 provide better risk discrimination, especially for patients with intermediate-2 to high-risk classification according to IPSS, DIPSS, and DIPSS+ who are candidates for HSCT. Also, scoring systems that consider genetic and cytogenetic data seem to have increased prognostic value to estimate the risk of leukemia progression (Figure 1).

Figure 1. Risk stratification approach in primary myelofibrosis¹

An unresolved problem is the accurate timing of HSCT. The decision of immediate or delayed HSCT may need to be discussed primarily in younger patients with good clinical status and high genetic risk (as assessed by GIPPS, for example). Here, prognostic models can provide adequate guidance for initiating bone marrow transplantation.

Special situations¹

Post-ET/PV myelofibrosis

MPN management guidelines do not differentiate between therapeutic management for PMF patients versus those with post-ET/PV myelofibrosis. Using the MIPSS70 is not applicable in these cases as relevant HMRs for PMF, such as ASXL1, EZH2, and IDH1, are not predictive for post-ET/PV myelofibrosis outcomes. Therefore, the MYSEC-PM was developed explicitly for patients with post-ET/PV myelofibrosis. It identified six variables (age, Hb < 11g/dL, platelets < 150 Gpt/L, peripheral blasts ≥ 3%, constitutional symptoms, lack of CALR mutation) associated with poor outcome (Table 1) and subgroups patients into four risk levels accordingly. In a validation study in a post-ET/PV cohort, MYSEC-PM was superior to IPSS and DIPSS, allowing reclassification of patients from higher to lower risk categories. Therefore, the authors recommend using MYSEC-PM in patients with post-ET/PV myelofibrosis, although routine use needs to be further evaluated. Of note, age is a continuous variable in the MYSEC-PM and has a significant impact. This may unduly reduce the number of patients found eligible for HSCT, and it is therefore recommended to use other models (DIPSS or DIPSS+) for help with therapeutic decision making.

Prefibrotic myelofibrosis

In 2016, PMF was divided into two stages: prefibrotic and overtly fibrotic. Prefibrotic PMF patients have outcomes intermediate between ET and overtly fibrotic patients. The IPSS could discriminate low-, intermediate-, and high-risk patients. However, prefibrotic myelofibrosis patients with intermediate-risk scores had a significantly longer OS (> 10 years) than overtly fibrotic patients. Hence, prefibrotic PMF patients with intermediate risk (based on IPSS) should not undergo HSCT. However, high-risk patients have inferior OS, suggesting HSCT as a therapeutic option in this setting. MIPSS70, MIPSS70+, and MIPSS70+ v2.0 were validated in a cohort of patients with overtly and prefibrotic myelofibrosis. For patients with no available genetic data, the IPSS, DIPSS, and DIPSS+ may be useful alternatives for determining treatment intensity.

Conclusions and future directions

Research in past decades acquired enormous clinical and molecular data in PMF, leading to prognostic models with multiple variables and high data granularity. Earlier models based mostly on clinical data have evolved, and increasingly genetic information is integrated for risk assessment. However, these new models need further validation, especially when applying them to determine eligibility for HSCT.

A significant limitation of existing models is the lack of treatment options for patients with PMF. While prognostic models are critical to identify patients with high-risk features who benefit from HSCT, they have little impact on treatment decisions in transplant-ineligible patients. Due to a lack of drugs for these patients, management is similar between low- and high-risk groups and consists mainly of ruxolitinib, fedratinib, and hydroxyurea. Moreover, treatment initiation is primarily based on symptoms rather than genetic risk type. Treatment strategies are needed that consider those novel genetic markers, to maximize the utility of prognostic models.