Blood perfusion in osteomyelitis studied with [15O]water PET in a juvenile porcine model

Abstract

Background: Osteomyelitis is a serious disease which can be difficult to treat despite properly instituted antibiotic therapy. This appears to be related at least partly to degraded vascularisation in the osteomyelitic (OM) lesions. Studies of perfusion in OM bones are, however, few and not quantitative. Quantitative assessment of perfusion could aid in the selection of therapy. A non-invasive, quantitative way to study perfusion is dynamic [¹⁵O]water positron emission tomography (PET). We aim to demonstrate that the method can be used for measuring perfusion in OM lesions and hypothesize that perfusion will be less elevated in OM lesions than in soft tissue (ST) infection. The study comprised 11 juvenile pigs with haematogenous osteomyelitis induced by injection of Staphylococcus aureus into the right femoral artery 1 week before scanning (in one pig, 2 weeks). The pigs were dynamically PET scanned with [¹⁵O]water to quantify blood perfusion. OM lesions (N = 17) in long bones were studied, using the left limb as reference. ST lesions (N = 8) were studied similarly.

Results: Perfusion was quantitatively determined. Perfusion was elevated by a factor 1.5 in OM lesions and by a factor 6 in ST lesions.

Conclusions: Blood perfusion was successfully determined in pathological subacute OM lesions; average perfusion was increased compared to that in a healthy bone, but as hypothesized, the increase was less than in ST lesions, indicating that the infected bone has less perfusion reserve than the infected soft tissue.

Keywords: Bone infection; Osteomyelitis; Perfusion; Positron emission tomography; Swine; [15O]water.

Fig. 1

Example of scanning (pig no. 1). Top: CT image. Middle: Fusion of CT and [¹⁵O]water PET images (averaged over the 5 min scanning time). Bottom: Fusion of CT and parametric image of the K ₁ parameter, i.e. the perfusion, with colour scale from 0 to 80 mL/min/100 cm³. In the images, a section is shown of the 3D VOIs drawn in the femoral medullar canal and in the thigh muscles. In this example, the positioning of the pig was very symmetric; in cases where pigs were slightly tilted to one side or had variation in how much the limbs were flexed, VOIs were drawn to correspond anatomically, not necessarily centred within the same slice

Fig. 2

Example of lesion (arrow) and spherical VOIs drawn at the lesion and at the corresponding position in the non-infected limb. The upper image is CT, and the lower image shows CT fused with parametric image of perfusion. This is the same pig and same colour scale as in Fig. 1

Fig. 3

Abscess at metatarsus II in pig no. 4. Top: CT image with chosen VOI position reported in Table 4, determined from CT with guidance from [¹⁵O]water PET image. Middle: Fused PET/CT image with a profile of VOIs at vertical spacing of 2 voxels. Bottom: Perfusion profile from these VOIs. The chosen VOI corresponds to distance 0 mm and turns out not to be maximum value, but judged from other side of the profile, the value appears to be typical for the non-necrotic part of the abscess

Fig. 4

Bland-Altman plots for perfusion in long bone and soft tissue lesions. Upper: simple differences. Lower: logarithmic differences (a difference ∆ corresponds to a factor of exp(∆), e.g. exp(1.1) ≈ 3 and exp(2.3) ≈ 10). Underlying data are presented in Table 3. Note the more homogeneous spread of OM and ST in the logarithmic plot