Abnormal [18F]FDG PET/MRI Findings In Paraspinal Muscles of Patients with Suspected Cerebrospinal Fluid Leak.

1533 Introduction: The most characteristic symptom of a CSF leak is persistent headache that worsens when upright and improves with recumbency (orthostatic headaches).1 Unfortunately, no one diagnostic test can consistently exclude the diagnosis of CSF leak or identify the site(s) of leakage.2 Misdiagnosis is common, preventing proper treatment as the symptoms are often mistaken for those of other diseases.3,4 In this report, we hypothesized that [18F]FDG PET/MRI could identify inflammation around the sites of CSF leak due to its pro-inflammatory properties.5 Methods: IRB approval was obtained. Seven patients were included who had orthostatic headache and suspected CSF leak. They had received standard-of-care imaging (including CT myelogram), without identifying a leak. Six controls were also recruited. Subjects were imaged using a GE SIGNA PET/MRI system (time-of-flight PET; 3.0T bore; 4-8 min/bed position) from head to toe. One hour after a 10-mCi injection of [18F]FDG, we performed a whole-body PET/MRI scan of subjects (8-10 consecutive imaging stations). PET and MRI raw data were reconstructed using algorithms equipped in the scanner.6,7 Abnormal [18F]FDG hotspots were identified, and maximum standardized uptake values (SUVmax) were measured using image analysis software (Horos v.3.3.5). Hotspots were categorized into 5 types: paraspinal muscle, interspinous ligament, osseous tissue, neuroforamina, and fluid. SUVmax measurements of each type were compared with SUVmax of corresponding areas in controls using a two-sided Mann-Whitney U-test (P-value<0.05 considered significant). Results: In all seven patients, increased uptake of [18F]FDG was found in paraspinal structures at various levels (Figure 1). However, only 3 patients showed MRI abnormalities, which were in the same region as PET abnormalities (Figure 1). The [18F]FDG SUVmax in lesions from the patients showed a higher mean value than the corresponding areas from controls in all tissue types. SUVmax of the abnormal lesions in the patient group ranged from 1.148 to 6.565, while SUVmax of the corresponding tissues in controls ranged from 0.416 to 2.893. A Mann-Whitney U-test comparing these two groups indicated a significant difference. Figs. 2 and 3 present different cases of abnormally increased uptake in patients: in cervical paraspinal muscles (Fig. 2), at the level of the interspinous ligament in the lumbar spine (Fig. 3A). Asymptomatic controls did not show areas of focally increased [18F]FDG uptake in these regions (Fig. 3B). Note that the cases in Figs. 2 and 3 show no abnormal MRI signal change. Epidural blood-patching was performed for four patients based on information from standard-of-care diagnostic methods. Two patients underwent epidural blood-patching at the sites covering abnormalities identified by our study and reported temporary improvement in symptoms. Conclusions: While other nuclear medicine approaches such as radioisotope cisternography have been used to identify CSF leaks, to our knowledge neither [18F]FDG PET nor simultaneous PET/MRI have been used for this purpose until now.8Extradural CSF may irritate and inflame tissues outside the nervous system5 thus providing a mechanism for PET detection using [18F]FDG. This irritation is reflected clinically in studies of patients with chronic CSF leaks following accidental dural puncture, who report higher rates of spinal pain at the site of CSF leak than control patients who underwent epidural injection without dural puncture.9,10 Such inflammatory lesions have increased metabolism and become more glucose-avid than normal tissues, which can be imaged by PET using a glucose analog radiotracer such as [18F]FDG. In our results, more abnormalities were found with [18F]FDG PET than 3T MRI, which suggests [18F]FDG PET might be better suited than MRI for the detection of early inflammatory changes due to CSF leak.