The Relationship between Cause of Death in Myelofibrosis and the Dynamic International Prognostic Scoring System

Background:

The Dynamic International Prognostic Scoring System (DIPSS) and DIPSS Plus are well established systems for providing prognoses to patients with primary myelofibrosis. Five risk factors are included in the DIPSS: age > 65, hemoglobin < 10 g/dL, leukocyte count > 25 x 10⁹/L, circulating blasts ≥ 1%, and presence of constitutional symptoms. The DIPSS plus score also factors in three additional variables: presence of an unfavorable karyotype, transfusion dependency, and platelet count < 100 x 10⁹/L. The bulk of current research focuses on overall survival. The goal of this study was to examine if the presence of these risk factors were associated with cause of death (COD) in patients with myelofibrosis (MF).

Methods:

Follow-up information on 139 patients referred to Michigan Medicine with an initial diagnosis of myelofibrosis, who subsequently died between April 2007 and February 2017, was collected through retrospective chart review. Nine patients who had an unknown cause of death (COD), and 17 who died from complications due to BMT were excluded, leaving 113 evaluable patients (61 male, 52 female). DIPSS plus risk factors were collected at time of referral, though there was insufficient data available to include an analysis of transfusion-dependency or karyotype. P-values were obtained from a chi-squared test for independence between each variable and COD.

Patients

The median age at diagnosis was 70.7 years (IQR 63.0, 74.9), the median age at death was 73.1 years (IQR 65.8, 77.9), and the median time from diagnosis to death was 2.3 years (IQR 0.9, 4.0). 61.9% of patients had primary MF (n = 86), 23.0% had post-essential thrombocythemia MF (n = 32), and 14.4% had post-polycythemia vera MF (n = 20); one patient's history was unknown. 48.2% had a positive JAK2 mutation (n = 67), 24.5% were tested and negative (n = 34); 12 patient's JAK2 mutation status were unknown.

Results:

Seven causes of death were predefined, namely death from acute myeloid leukemia (n = 26), bone marrow failure (n = 21), congestive heart failure or pulmonary hypertension (n = 20), spleen or liver complications (n = 17), infection not related to cytopenias (n = 7), failure to thrive (n = 7), or any other COD (n = 15). Among the seven characteristics, a significant association was observed among patients with a platelet count < 100 x 10⁹/L; disproportionately more died due to bone marrow failure and less died due to cardiac or pulmonary causes (p = 0.012). This should be interpreted in the context of no statistically significant associations between COD and the other five clinical risk factors. Overall DIPSS score ≥ intermediate-2 also did not lead to significant associations. Of the 113 patients, sixteen had missing DIPSS scores and/or unknown JAK2 status. Among the remaining 97, a larger-than-expected number of patients with a JAK2 mutation had an Intermediate-1 DIPSS risk; conversely, more patients without a JAK2 mutation had an Intermediate-2 risk. A statistical test for independence yielded a p = 0.009, suggesting that patients with JAK2 mutations have lower DIPSS-calculated risk.

Conclusions:

DIPSS Plus variables are often collected at referral to provide prognostic information to patients. It appears that the presence of thrombocytopenia is more likely to predict bone marrow failure as a patient's ultimate COD compared to the presence of anemia. Additionally, patients with JAK2 mutations seem to have lower DIPSS-calculated risk. Knowledge of this clinical information can aid management of myelofibrosis.