Myeloproliferative Neoplasm Patient Symptom Burden and Quality of Life: Evidence of Significant Impairment Compared to Controls Using Multivariate Analysis

Introduction: The myeloproliferative neoplasms (MPNs) including polycythaemia vera (PV), essential thrombocythaemia (ET) and primary myelofibrosis (PMF) are rare diseases which contribute to significant morbidity. Symptom management is a prime objective in MPN treatment but current symptom assessment tools have not been validated compared to the general population. Materials and Methods: The MPN Symptom Assessment Form (MPN-SAF) is a reliable and validated clinical tool used to assess MPN symptom burden. The MPN-SAF was administered to MPN patients (n=106) and, for the first time, General Practice and non-blood relative/family controls (n=124) as part of a UK pilot case-control study. Mean scores for individual MPN-SAF items and for the Total Symptom Score were previously compared between cases and controls adjusting for potential confounding variables (Anderson LA, et al. Am J Hematol. 2015). The current analysis employed linear discriminant analysis (a multivariate technique) to compare MPN- 10 symptom profiles between controls and cases, and between controls and patient groups by MPN diagnosis (ET n=55, PV n=37, PMF n=14). Results: MPN patients as a single group had significantly different MPN-10 symptom profiles compared to controls (multivariate analysis of variance [MANOVA] p<0.001). Discriminant loadings which represent correlations between each individual MPN-10 item and the single significant canonical variable were >0.30 for all MPN-10 items, and >0.50 for weight loss, fatigue, night sweats, bone pain, abdominal discomfort, inactivity, concentration problems, and itching. This suggests that all MPN-10 items contribute to distinguishing between patients and controls. The linear discriminate function correctly categorized 72% of all subjects. When considering patients by diagnosis, MPN-10 symptom profiles significantly differed across patients with ET, PV, and PMF and controls (MANOVA p<0.001). Three canonical variables were significant, with discriminant loadings of the first canonical variable being >0.50 for the same MPN-10 items as the previous analysis; of the second canonical variable for fever; and of the third canonical variable for bone pain. This suggests that MPN- 10 items are broadly useful in differentiating between patients and controls, with particular symptoms (e.g., fever and bone pain) contributing to differentiating between patients with different diagnoses. The linear discriminate function correctly categorized 58% of all subjects. Conclusions: MPN patients experience significant morbidity when compared to the general population highlighting the need to manage symptoms effectively. Symptom profiles can not only distinguish MPN patients