Prevalence of extravertebral cement leakage after vertebroplasty: procedural documentation versus CT detection

Background: Reported incidence of extravertebral cement leakage after vertebroplasty varies widely across studies. Purpose: To retrospectively compare the relative detection rates of extravertebral leakage noted under intra-procedural fluoroscopic surveillance, postprocedure plain radiographs, and postprocedure computed tomography (CT) in a cohort of patients undergoing vertebroplasty. Material and Methods: With IRB approval, we retrospectively identified 181 patients with 277 levels treated with percutaneous vertebroplasty among a total of 1255 patients undergoing vertebroplasty between 1999 and 2010 who had subsequently undergone a CT examination that included the treated level(s). Categories of leakage were paravertebral, end plate, epidural, and prevertebral venous leakage. CT-detected leak rates were then compared to those noted on the vertebroplasty procedure reports and the archived fluoroscopic images for this same cohort using Pearson's chi(2) test. Results: One hundred and forty-nine (82%, 95% Cl 76-87%) of 181 patients demonstrated evidence of some type of leakage on CT at one or more treated levels. Sixty-two (34%, 95% Cl 28-42%), and seventy-seven (50%, 95% Cl 43-57%) of 149 CT-detected leaks were reported in the procedural dictation or detected on plain radiography (P = 0.01 and 0.006, respectively). The most common type of leakage noted on CT was end plate (n = 81, 45%, 95% Cl 38-52%), followed by paravertebral (n = 64, 35%, 95% Cl 29-43%), epidural (n = 36, 20%, 95% Cl 15-26%), and prevertebral venous (n = 32, 18%, 95% CI 13-24%). Conclusion: Cement leakage after vertebroplasty is common and is often not reported by operators in procedural dictations. CT detects substantially more leaks than plain radiography.