Definitions of white matter hyperintensity change: impact on estimates of progression and regression

Background White matter hyperintensity (WMH) progression is well documented; WMH regression is more contentious, which might reflect differences in defining WMH change. We compared four existing WMH change definitions in one population to determine the effect of definition on WMH regression. Methods We recruited patients with minor non-disabling ischaemic stroke who underwent MRI 1–3 months after stroke and 1 year later. We assessed WMH volume (in absolute mL and % intracranial volume) and applied four different definitions, including two thresholds (based on SD or mL), percentile and quintile approaches. Results In 198 participants, mean age 65.5 (SD=11.13), baseline WMH volume was 15.46 mL (SD=19.2), the mean net WMH volume change was 0.98 mL (SD=2.84), range −7.98 to +12.84 mL. Proportion regressing/stable/progressing WMH were threshold 1 (SD), 29.8%/55.6%/14.6%; threshold 2(mL), 29.8%/16.7%/53.5%; percentile approach, 28.3%/21.2%/50.5%. The quintile approach includes five groups with quintile 3 reflecting no change (N=40), quintiles 1 and 2 any WMH decrease (N=80) and quintiles 4 and 5 any WMH increase (N=78). Conclusions Different WMH change definitions cause big differences in how participants are categorised; additionally, non-normal WMH distribution precludes use of some definitions. Consistent use of an appropriate definition would facilitate data comparisons, particularly in clinical trials of potential WMH treatments.