Can MRI predict subsequent pseudarthrosis resulting from osteoporotic thoracolumbar vertebral fractures?

The purpose of this study was to assess the ability of short inversion time inversion-recovery (STIR) in magnetic resonance imaging for predicting the prognosis of osteoporotic vertebral fractures.

We analyzed 63 vertebrae of 56 patients who had osteoporotic vertebral fracture (Th10-L2) prospectively. Image finding of a homogeneous high signal change on a fractured vertebra was evaluated and all vertebrae were divided into "homogenous high signal change group" or "non-homogenous high signal change group". On the other hand, image finding of linear black signal area was evaluated and all vertebrae were divided into "linear black signal area group" or "non-linear black signal area group".

Sixteen and 24 vertebrae were included in the homogenous high signal change group or the linear black signal area group, respectively. The 16 homogenous high signal change cases did not result in non-union, and 47 non-homogenous high signal change cases resulted in 14 non-unions, a significant difference. Twenty-four linear black signal area and 39 non-linear black signal area cases resulted in 10, and 4 non-unions, respectively, also a significant difference. The kyphosis progression rate of the linear black signal area group (mean 35%) was significantly higher than that of non-linear black signal area group (mean 23%). The visual analog scale of back pain of the linear black signal area group (mean 35 mm) was significantly higher than that of the non-linear black signal area group (mean 23 mm).

STIR was useful for predicting bone union, kyphosis, and back pain in patients with osteoporotic vertebral fracture.