Validation of a Modified Version of the Myeloproliferative Neoplasm Symptom Assessment Form Total Symptom Score (MPN-SAF TSS)

Background

Patients with myeloproliferative neoplasms (MPNs) are faced with severe disease-related fatigue among a range of other constitutional and spleen-related symptoms. The MPN-Symptom Assessment Form (SAF) is recommended for use to characterize symptom burden (Scherber R, et al. Blood 2011). Within the SAF, a profile of 18 symptom items are evaluated ranging in severity from 0 (absent) to 10 (worst imaginable). The SAF is often summarized to the MPN-SAF Total Symptom Score (TSS) for analysis purposes – a single computed sum of the 10 most clinically meaningful symptom scores, including fatigue (Emanuel R, et al. J Clin Oncol 2012). Though the SAF includes a fatigue item, initial deployments of the MPN-SAF TSS incorporated a 0-10 scaled fatigue item taken from the Brief Fatigue Inventory (BFI; Mendoza T, et al. Cancer 1999). A subsequent version of the MPN-SAF TSS for use within myelofibrosis clinical trials (called the MFSAF v4; Gwaltney C, et al. Leuk Res 2017) employed a harmonized fatigue item. This analysis employing data from two studies was carried out to assess the use of the SAF fatigue item within the MPN-SAF TSS for consistency with the MFSAF v4.

Methods

Both BFI and SAF fatigue items were included in an initial online survey evaluating disease burden among patients with MPNs. Participants were assigned to survey variants as a function of their age. Survey variants included those to receive 1 instance of either the BFI or SAF fatigue item, instances of both BFI and SAF, or 2 instances of the same fatigue item. Surveying was aimed to assess the worst symptom experience in the patient’s last 24 hours. Additionally, an independent survey assessing the impact of COVID-19 among MPN patients was deployed using the SAF fatigue item for the MPN-SAF TSS. This modified version was then used to test internal validity. Pearson correlation (r) and t-tests were used to assess association, Bland-Altman methods were used to evaluate systematic agreement between BFI and SAF fatigue scores, and Cronbach’s alpha was used to measure internal consistency.

Results

There were 229 participants assigned both BFI and SAF fatigue items within the same survey. Among them, 51% (n=117) received the BFI item first and 49% (n=112) the SAF item first. No difference was seen between first and second fatigue scores (mean difference [first-second] = 0.00; 95%CI -0.18, 0.17). BFI and SAF fatigue scores were highly correlated (r=0.88, p<0.001) and showed 88.7% agreement in categorizing severe versus non-severe fatigue (score ≥ 7 versus < 7). Overall concordance in severity category was 73.4% (category [score range]: absent [0]; mild [1-3]; moderate [4-6]; severe [7-10]). Constructing the MPN-SAF TSS using the BFI and SAF fatigue components separately, the original and modified MPN-SAF TSS were nearly identical (r=0.997, p<0.001), and had equivalent internal consistency (both Cronbach’s alpha=0.88). The Bland-Altman plot further indicates the 2 fatigue measures have high agreement with no evidence of directional bias and negligible overall bias (Figure 1: regression slope = -0.04, p=0.25; mean difference=0.22; 95%CI 0.05, 0.39).

Within the COVID-19 survey (n=1217), the modified version of the MPN-SAF TSS was consistent with known correlates among disease characteristics (Emanuel R, et al. J Clin Oncol 2012). For example, more severe MPN-SAF TSS scores were highly correlated with low quality-of-life (n=1156, r = -0.50, p<0.001), and associated with those reporting spleen enlargement (n=301) versus not (n=617) (p<0.001; mean difference=7.7; 95%CI 5.4, 10.1).

Conclusion

The BFI and SAF fatigue items are highly consistent in raw score, severity category, and in contribution to the MPN-SAF TSS. There was no order effect seen in which fatigue item was asked first. The independent COVID-19 survey using the modified MPN-SAF TSS was validated and shows high internal consistent. In the ongoing development to capture the symptom experience, this analysis shows disease-related fatigue is equivalently measured using the SAF fatigue survey item in harmonization with the MFSAF v4.