Conventional epidurography (CE) is thought to have insufficient usefulness on percutaneous epidural adhesiolysis (PEA). The investigators aimed to evaluate the association between the outcome of PEA and three dimensional-rotational epidurography (3D-RE). The investigators performed 30 PEA in 26 patients, and evaluated their post-PEA image findings. Two independent clinicians categorized and recorded the occurrence of contrast at intra-canal ventral and extra-foraminal regions on CE; and contrast at dorsal canal (DC), ventral canal (VC), dorsal foramen (DF), and ventral foramen (VF) on 3D-RE. Reproducibility was assessed using intra-class correlation coefficients (ICCs). The symptom relief after one month for the patients receiving PEA and the contrast distribution patterns of CE and 3D-RE and were determined.

Patient Selection. This prospective study protocol was approved by the ethics committee, and informed consent was obtained from all subjects. The investigators analyzed the medical and radiographic records included patients who underwent lumbar PEA procedures at a single university hospital. The inclusion criterion was FBSS and SS with unilateral radiculopathy. All patients reported a history of discogenic or radicular symptoms refractory to conservative treatments and epidural steroid injection for a minimum of 6 weeks. Diagnoses were established using magnetic resonance imaging (MRI) and/or computed tomography (CT) performed before PEA. The PEA was standardized to all patients receiving the procedure. Each patient received an epidural injection, and if the symptoms persisted or the relief was insufficient, the patient received PEA >6-week interval between the epidural steroid injection. Positive provocative test during PEA was used to confirm the affected spinal level [11]. All patients underwent CE before and after PEA; and 3D-RE after PEA. Patients with a history of spinal surgery and those with cauda equina syndrome, bleeding diathesis, associated somatic or psychiatric disease, vertebral fractures, pregnancy, and tumors or other underlying systemic diseases that could significantly affect the procedural outcomes were excluded. The patients had bilateral symptoms and did not react to the provocation during PEA were also excluded. All procedures were performed by one of the authors (YCH) using the same procedural protocol.

PEA Procedure. A standard PEA procedure was used to lyse adhesions and achieve nerve blockades in all patients, as previously described [1,2]. A 1-day standardized protocol was followed. The patient was placed on a radiolucent table in the prone position, and the procedure was performed under fluoroscopic guidance. The coccygeal and sacral regions were disinfected with 10% Betadine, and the surgical site was draped in the usual aseptic manner. The sacral hiatus was anesthetized with 1% xylocaine. Then, a 13.6-G Coaxial needle was introduced into the epidural space below the level of S3 with a combined use of US-guidance and fluoroscopy insurance. The US was provided using a scanner (Xario 100; Toshiba, Tokyo, Japan) with a 7~18 MHz linear transducer (PLU-1204BT). A total of 0.5 mL of the contrast agent (Omnipaque, GE Healthcare, Ireland) was instilled to confirm the epidural space. Both anteroposterior, right-oblique, left-oblique, and lateral fluoroscopic views were obtained. The investigators also assessed the development of any adverse reactions. On confirmation of the target for PEA, an angiographic catheter (Cobra 4-Fr; Cordis, USA) and/or coaxial supporting catheter (Chiba 6-Fr; Cook, USA) was gently inserted toward the target site. Once it reached the target site, a second CE was obtained by injecting 3 mL of the contrast agent for the identification of filling defects or cutoff signs surrounding the target area. The investigators choose the target nerve roots for PEA by clinical dermatome involvement and provocation tests during PEA. The catheter tip was positioned at the ventral epidural space of the target site, and in the case of foraminal diseases, was placed at the opening of the foramen [12]. When the tip of the catheter touched the target site or the contrast agent exerted pressure on the lesion, patients were asked to report provoked symptoms [4]. According to the surgical records, they frequently reported pain similar to what they had been suffering. Both mechanical and fluid adhesiolysis were performed. The former was achieved through pushing, pulling, and rotating movements of the catheter, while the latter was achieved by the injection of 0.9% normal saline (10 mL). Following PEA, a third CE and 3D-RE was obtained using 3 mL of the contrast agent. All CE and 3D-RE images were saved in the Digital Imaging and Communications in Medicine format for future analysis, and the CE and 3D-RE after PEA was used for analysis in our study. Finally, a 40 mg triamcinolone acetonide (YungShin, Taichung, Republic of China) was slowly injected.

3D-RE technique. After performing PEA, 3D-RE are obtained on a commercial digital bi-plane angiography system (Allura Xper, Philips). A 240-degree rotation forward and backward of the tube-camera unit around the patient's longitudinal axis within 4 seconds is performed by using an acquisition matrix of 1024 x 1024 pixels. The forth and back rotation results in 120 radiographs. Raw data were automatically sent to a dedicated workstation (Philips Xtra vision workstation), where the rotational projection images are prepared for computed 3D reconstruction. 2D radiographs are used to determine the volume of interest. After correction of gain and distortion, a 3D data set is calculated, resulting in 512 transverse CT-equivalent sections. Depending on the initially chosen volume of interest, the voxel size varies from 0.1 to 0.6 mm. The investigators used a resolution of 0.14-mm voxel size, which resulted in a 3D cuboid of 25 x 19 x 25 cm.

Reconstruction time was 30 seconds. Detailed information regarding technical performance and reconstruction procedure has been reported [13]. Postprocessing techniques provided by the software included real-time 3D volume rendering and multiplanar reformatting. Real-time 3D volume rendering creates a 3D model of the examined object. The software allows emphasizing bony structures or soft tissue by changing intensity, brightness, and opacity of different X-ray structures. Additionally, rotation of the 3D object in all directions is possible, as is a virtual stereoscopic view provided by the software in combination with special glasses. The multiplanar reformatting modus generates virtual sections according to the three main axes and free defined axes. The section planes can be freely chosen, and curved sectioning is also possible.

Conventional epidurography and 3D-RE Contrast Distributions. Post-PEA CE contrast patterns were defined and classified into 2 types according to the system proposed by Park et al. and Gupta et al.[4,6]: Limited intracanal ventral spread (ICV) and extended extraforaminal spread (EF) (Figure 1). The investigators defined the contrast distributions into 4 areas by post-PEA 3D-RE : Dorsal Canal (DC), contrast spread to the dorsal zone of the ipsilateral epidural space not extending to the neural foramen; Ventral Canal (VC), contrast spread to the ventral zone of the ipsilateral epidural space not extending to the neural foramen; Dorsal Foramen (DF), contrast spread to the dorsal zone of the ipsilateral epidural space extending to the neural foramen; Ventral Foramen (VF), contrast spread to the ventral zone of the ipsilateral epidural space extending to the neural foramen (Figure 2). The CE and 3D-RE images were analyzed and recorded by a musculoskeletal radiologist (Y.C.H.) and a surgeon (C.T.T.) who blinded to the results of the clinical data. The investigators made a consensus if there was discrepancy between the analyses of these images.

Data collection. All patients were clinically evaluated before and 1 month after PEA by a nurse specialized in pain management and blinded to the treatment details. The intensity of leg and back pain prior to PEA was assessed using a subjective visual analog scale (VAS) calibrated from 0 to 10 (0 = no pain and 10 = the worst pain imaginable). The patients' symptoms relief was rated using a Likert scale: 5 = greatly relief; 4 = some residual symptoms and symptoms relief > 50%; 3 = some residual symptoms and symptoms relief = 50%; 2 = residual symptoms and symptoms relief < 50%; 1 = no symptoms relief. For the comparison of clinical outcomes according to the symptoms relief, patients with grade 3, grade 4, and grade 5 were assigned to a group exhibiting >= 50% symptoms relief, while patients with grade 1 and grade 2 were assigned to a group exhibiting < 50% symptoms relief. Regarding the radiation exposure of PEA, simulations were performed with patient-specific input parameters (weight and length) and the actual 3D-RE system settings for each frame, including the automatic modulation of beam energy and dose level, and collimation. Effective dose (ED), which has been generally accepted single number index reflecting patient radiation risk, was subsequently calculated using the latest ICRP 103 weighing factors, published in 2007 [14].

StatisticalAnalysis. Age, the duration of discomfort prior to PEA, and VAS scores are expressed as means ±standard deviations. Demographic data within the two groups of greater symptoms relief and less were compared by using the chi-square test or the Fisher's exact test or unpaired t-test. Inter-reader agreement was evaluated using the intra-class correlation coefficients (ICCs), calculated according to Landis and Koch[15], for the contrast distribution of CE and 3D-RE. Simple linear regression analysis was used to determine the prediction of the outcome of PEA by the contrast distribution of CE and 3D-RE. All statistical analyses were performed using the Statistical Package for the Social Sciences software (SPSS Inc., Chicago, Illinois, USA). A P value of <0.05 was considered statistically significant.

Inclusion Criteria:

• FBSS or SS with unilateral radiculopathy.
• History of discogenic or radicular symptoms refractory to conservative treatments and epidural steroid injection for a minimum of 6 weeks.
• The PEA was standardized to all patients receiving the procedure.
• Each patient received an epidural injection, and if the symptoms persisted or the relief was insufficient, the patient received PEA >6-week interval between the epidural steroid injection.
• Positive provocative test during PEA was used to confirm the affected spinal level.
• All patients underwent CE before and after PEA; and 3D-RE after PEA.

Exclusion Criteria:

• History of spinal surgery and those with cauda equina syndrome
• Bleeding diathesis
• Associated somatic or psychiatric disease
• Vertebral fractures
• Pregnancy
• Tumors
• Other underlying systemic diseases that could significantly influence the procedural outcomes were excluded.
• Bilateral symptoms
• Did not react to the provocation during PEA

Patients who had FBSS or SS with unilateral radiculopathy underwent PEA.