Musculoskeletal Imaging
Original Research
Sacral Fractures After Lumbosacral Fusion: A Characteristic Fracture Pattern
Audio Available | 
Share
OBJECTIVE. The purpose of this study was to describe the radiologic pattern of sacral fractures after lumbosacral fusion and to identify clinical characteristics relevant to the radiologic diagnosis.
MATERIALS AND METHODS. A search of CT, nuclear medicine, and MRI radiology reports over a 5-year period at our institution revealed a total of 23 patients with sacral fractures after lumbosacral fusion. Two radiologists reviewed all of the images to determine the sacral fracture pattern. The clinical records of these patients were reviewed for interval after surgery, fusion length, hardware, approach, preoperative diagnosis, symptoms, treatment, and risk factors.
RESULTS. All 23 sacral fractures were horizontal through the sacral body, involved the screw holes, and exited through the posterosuperior sacral alae. The fractures occurred within 3 months of fusion in 19 of 23 patients. All 23 patients had symptoms at the time of fracture. Seventeen of 23 fusions were long (more than four vertebrae). Four of 23 patients had osteoporosis. Eleven of 23 fractures healed without surgery, and 12 were managed with transiliac fixation.
CONCLUSION. Sacral fractures after lumbosacral fusion have a characteristic transverse pattern through the sacral screw holes that differs from the configuration of more common sacral insufficiency fractures. Most of these fractures occur within 3 months after surgery, and many of the patients need additional surgical fixation. Because few of the patients had osteoporosis and most underwent long fusion, the fractures might have been caused by hardware-related stress raisers in the sacrum.
Keywords: fusion, osteoporosis, sacral fracture, stress raiser
Sacral insufficiency fractures represent failure of bone under normal physiologic loads. These fractures occur in the sacral alae with a vertical orientation, parallel to the sacroiliac joints. Horizontal fractures through the sacral body often occur in conjunction with the vertical alar fractures.
Sacral fracture after lumbosacral fusion has been reported in the surgical literature [1–10]. The authors of the reports presumed osteoporosis to be an important risk factor based on patient demographics (elderly women). Obesity has also been theorized as a risk factor because of the increased load at adjacent segments after multisegmental fusion [8]. Mechanical factors, such as iliac crest graft harvesting, number of segments fused as a measure of the length of the lever arm, and balance of the spine in the sagittal plane, also have been suggested [9]. However, the reports have focused on treatment and outcome rather than on radiologic diagnosis. The purpose of this study was to describe the radiologic pattern of sacral fractures after lumbosacral fusion and to identify clinical characteristics relevant to the radiologic diagnosis.
| Materials and Methods |   |
|---|
This study was approved by the institutional review board at our hospital and conducted in compliance with HIPAA guidelines. The requirement for informed consent was waived because the study was retrospective. We conducted a search of the PACS and radiology information system for CT, nuclear medicine, and MRI radiology reports from January 1, 2005, to March 30, 2010, containing the keywords “sacral fracture” and “fusion.” To be included, patients must have undergone an index procedure during which instrumented fusion of the lumbar spine to the sacrum was performed, must have had imaging evidence of a sacral fracture after the index procedure, and must have undergone CT at or immediately after fracture. The cases of 23 patients (six men, 17 women; mean age, 67.7 years; range, 49–81 years) fulfilling the inclusion criteria were identified. Two of the 23 patients were previously described in a case series [8]. The mean age of the men was 70 years (range, 59–80 years) and of the women was 67.1 years (range, 49–81 years).
Unenhanced CT of the lumbosacral spine was performed with a 16-MDCT scanner (MX8000, Philips Healthcare). MRI studies were performed with a 1.5-T MR imager (HDx, GE Healthcare) with an eight-channel cervicothoracolumbar coil. Radiographs consisted of postoperative 14 × 36 inch (36 × 91 cm) anteroposterior and lateral standing long digital cassette views of the thoracic and lumbar spine. Bone scans were obtained after IV administration of 99mTc-methylene diphosphonate with a standard dose of 25 mCi and were acquired with a dual-head gamma camera (Skylight, Philips Healthcare).
A musculoskeletal radiology fellow and an attending musculoskeletal radiologist (36 years of experience) reviewed the CT studies, radiographs, MRI studies, and nuclear medicine bone scans. The sacral fracture pattern was determined on the CT scans. The sacral fracture was characterized as to location (ala, body), presence of foraminal involvement, relation of the fracture plane to the sacral instrumentation, direction of the sacral fracture (horizontal, vertical, or oblique), and exit locations of the fracture planes.
The musculoskeletal radiology fellow reviewed the clinical notes and operative records for the following data: age; sex; presence of osteoporosis according to results of a preoperative dual-energy x-ray absorptiometry study performed in the 6 months before surgery or history of previous characteristic osteoporotic fractures; body mass index (BMI); history of diabetes, smoking, or other chronic disease; more than 2 weeks of corticosteroid use in the 6 months before surgery; use of bisphosphonates before surgery; history of radiation therapy to the pelvis; preoperative diagnosis (reason for surgery); length of spinal fusion; type of hardware used; preexisting pelvic instrumentation; approach (anterior or posterior); use of bone morphogenetic protein during surgery; time from surgery to diagnosis of sacral fracture; symptoms at fracture diagnosis; type of treatment of the sacral fracture; and time to and reason for surgery on patients who needed transiliac fixation.
 View larger version (171K) | Fig. 1A72-year-old man with osteoporosis. A, Coronal CT image obtained 45 days after L4–S1 fusion shows nondisplaced transversely oriented fracture (arrows) through sacral screw holes and exiting superior alae. Asterisk indicates donor graft site. |
 View larger version (141K) | Fig. 1B72-year-old man with osteoporosis. B, Coronal CT image obtained 3 months after A shows fracture is more prominent owing to resorption along fracture line as part of normal healing process. |
 View larger version (157K) | Fig. 1C72-year-old man with osteoporosis. C, Coronal CT image obtained 11 months after B, after long-term parathyroid hormone therapy, shows multiple areas of osseous bridging across fracture plane indicative of healing. |
| Results | |
|---|
All 23 patients had symptoms at the time of diagnosis. Symptoms included pelvic pain, lower back pain, buttock pain, pain in one of the lower extremities, and in one patient, a cracking noise. In 17 cases, the diagnosis of sacral fracture was made with CT initially. In the other six cases, CT was performed immediately after the diagnosis was suggested at MRI or bone scintigraphy. The fracture was visible on CT scans in all cases. The initial diagnosis was made with MRI in five patients, but seven patients underwent MRI performed at or immediately after fracture diagnosis, and the fracture was visible on the MR images of six of these seven patients. Dephasing artifact from the adjacent hardware obscured the fracture on MR images of one patient. The initial diagnosis was made with bone scintigraphy of one patient, but two patients underwent bone scintigraphy at or immediately after the fracture diagnosis, and increased activity within the sacral body was visible in both patients. However, evaluation of the alae was limited by postsurgical change. Although all patients underwent radiography at or immediately after the fracture diagnosis, owing to cortical stepoff at the anterior margin of S1 at the fracture site, a fracture was visible in retrospect on the radiograph of only one patient.
All 23 sacral fractures had a transverse orientation through the sacral body and exited the sacrum through both of the posterosuperior sacral alae (Figs. 1A, 1B, 1C, 2A, and 2B). Twenty-two of 23 fractures extended through both of the screw holes. One fracture extended through the right screw hole and above the left screw hole. Five of 23 fractures exhibited anterior displacement of the sacral body, the largest amount of displacement measuring 1 cm in one patient (Figs. 3A and 3B). Two patients had mild angulation of the sacral body at the fracture site. Only 1 of 23 fractures had vertical components within the sacral alae that extended inferiorly from the major transverse component.
The average time to fracture diagnosis after the index procedure was 97.6 days (median, 45 days; range, 6–722 days [23 months 22 days]). The fracture diagnosis was made within 3 months of fusion in 19 of 23 cases. The times to fracture diagnosis in the other four cases were 246, 363, 614, and 722 days. An analysis of these four outliers revealed that three of four patients were women; three had undergone lumbar surgery, two of the three having preexisting long fusion; and three underwent long fusion as the index procedure. No trends with regard to other clinical or operative risk factors were found.
 View larger version (165K) | Fig. 2A57-year-old woman with osteopenia. A Sagittal (A) and coronal (B) T2-weighted MR images of lumbosacral junction 64 days after T10–S1 fusion show nondisplaced transversely oriented fracture (arrows) extending through inferior aspect of both sacral screw holes and exiting superior alae. |
 View larger version (93K) | Fig. 2B57-year-old woman with osteopenia. B, Sagittal (A) and coronal (B) T2-weighted MR images of lumbosacral junction 64 days after T10–S1 fusion show nondisplaced transversely oriented fracture (arrows) extending through inferior aspect of both sacral screw holes and exiting superior alae. |
Eleven of 23 patients underwent preoperative dual-energy x-ray absorptiometry. According to the World Health Organization criteria, five of these patients had osteopenia, and two had osteoporosis. The lowest T score among the total hip, femoral neck, and spine was used for patient categorization. Among the 12 patients who did not undergo preoperative dual-energy x-ray absorptiometry, two patients were given the diagnosis of osteoporosis on the basis of a history of insufficiency fracture in a nonsacral location.
The average BMI of the 23 patients was 26.9 (range, 18.7–44.4). Six patients were obese (BMI, > 30), and six were overweight (BMI, 25–30). Six of the 23 patients had a history of smoking, and three had a history of type 2 diabetes. One patient had a history of Parkinson disease. Six of 23 patients were taking bisphosphonates at the time of surgery, and three had used corticosteroids for more than 2 weeks in the most recent 6 months. One of 23 patients had a history of radiation to the pelvis for endometrial cancer 1 year before the lumbosacral fusion.
 View larger version (171K) | Fig. 3A83-year-old woman who underwent vertebral fusion. A Sagittal T2-weighted MR image (A) and reformatted sagittal CT image (B) of lumbosacral junction 42 days after T10–S1 fusion show anteriorly displaced fracture (arrow) through sacral body that was treated with transiliac fixation 1 day after diagnosis. |
 View larger version (179K) | Fig. 3B83-year-old woman who underwent vertebral fusion. B, Sagittal T2-weighted MR image (A) and reformatted sagittal CT image (B) of lumbosacral junction 42 days after T10–S1 fusion show anteriorly displaced fracture (arrow) through sacral body that was treated with transiliac fixation 1 day after diagnosis. |
Thirteen of the 23 patients had undergone thoracic or lumbar surgery. The preoperative diagnoses among these patients were recurrent stenosis, pseudoarthrosis, anterolisthesis, hardware loosening, adult scoliosis, and junctional degeneration. In the 10 of 23 patients who had no history of surgery, the preoperative diagnoses were spinal stenosis, anterolisthesis, and adult scoliosis.
Seventeen of the 23 fusions were long (more than four vertebrae). The longest fusions were from T4 to S1 and from T10 to S1. The shortest fusion was from L4 to S1. All 23 patients underwent posterior fixation with pedicle screws. Twenty-two patients underwent fixation with titanium pedicle screws and rods, and one underwent fixation with stainless steel screws and rods. All 23 patients underwent bilateral posterolateral fusion.
Sixteen patients underwent interbody fusion at one disk space level or more with polyetheretherketone interbody cages from various manufacturers. The L5–S1 level was included in all of these interbody fusions. Seven of these interbody cages were placed by an extreme lateral approach through the retroperitoneum, four were placed by an anterior approach, and the others were placed via a posterior approach, either laminectomy or a transforaminal approach. Bone morphogenetic protein was used in 10 of 23 patients during surgery. In all patients, a larger screw diameter was chosen for the sacrum than for the lumbar vertebrae (7.5 vs 6 mm). All sacral screws were placed parallel to the S1 endplate, in a convergent direction, and bicortically for better pullout strength.
Eleven of 23 fractures healed nonoperatively, and 12 patients underwent transiliac fixation. No trends were found between these two groups with regard to surgeon performing the procedure, interval to fracture diagnosis, sex, BMI, or history of smoking, diabetes, bisphosphonate use, steroid use, or osteoporosis.
Four of the patients with fractures that healed without surgery had undergone previous thoracic or lumbar spinal surgery. Two had displacement at the fracture site, and two had angulation at the fracture site. Seven of these 11 patients had undergone long fusion, and four had undergone short fusion. All patients recovered with bedrest followed by light physical therapy and gradual return to activity. One patient was also treated with a spinal bone stimulator, and five patients received parathyroid hormone therapy (teriparatide). In one patient, the fracture had healed, but because of the fracture, screw loosening occurred in the sacral segment, allowing anterolisthesis with angular kyphosis to develop at L5–S1 and requiring pelvic fixation approximately 1 year after diagnosis (Figs. 4A, 4B, 4C, and 4D).
 View larger version (135K) | Fig. 4A49-year-old woman 46 days after T11–S1 fusion. A Coronal (A) and sagittal (B) reformatted CT images show transversely oriented fracture (arrows) through sacrum involving screw holes and exiting superior sacral body and alae. Asterisk (A) indicates graft harvesting site. |
 View larger version (132K) | Fig. 4B49-year-old woman 46 days after T11–S1 fusion. B, Coronal (A) and sagittal (B) reformatted CT images show transversely oriented fracture (arrows) through sacrum involving screw holes and exiting superior sacral body and alae. Asterisk (A) indicates graft harvesting site. |
 View larger version (128K) | Fig. 4C49-year-old woman 46 days after T11–S1 fusion. C Follow-up coronal (C) and sagittal (D) reformatted CT images 1 year after A and B show healing of fracture (arrowhead, D), but lucency is evident around sacral screws (arrows, C), reflecting screw loosening likely due to initial fracture, leading to L5 anterolisthesis and L5–S1 angular kyphosis. Screws were found loose during surgical revision. |
 View larger version (125K) | Fig. 4D49-year-old woman 46 days after T11–S1 fusion. D, Follow-up coronal (C) and sagittal (D) reformatted CT images 1 year after A and B show healing of fracture (arrowhead, D), but lucency is evident around sacral screws (arrows, C), reflecting screw loosening likely due to initial fracture, leading to L5 anterolisthesis and L5–S1 angular kyphosis. Screws were found loose during surgical revision. |
Four of the 12 patients treated surgically underwent transiliac fixation within 3 days of the fracture diagnosis. Two of the patients were treated immediately to prevent displacement, and the other two patients were operated on because of displacement. An initial conservative treatment regimen was attempted for the other eight patients, who eventually underwent transiliac fixation. The median time to pelvic fixation after the fracture diagnosis for these eight patients was 23 days. Six of the eight patients were treated within 30 days of the fracture diagnosis because of persistent severe intolerable pain with simple sitting or movement. Two of eight patients were treated conservatively for 3 months and 6 months before needing pelvic fixation. In both cases, surgery was necessary because there was no evidence of radiographic healing, and the patients had persistent severe symptoms.
Nine of the 12 patients treated surgically had undergone previous thoracic or lumbar surgery. Ten of these patients were in the long fusion group, and two were in the short fusion group. Two patients were treated with a unilateral transiliac screw, and 10 patients were treated with bilateral transiliac screws. No specific reason was cited in the operative notes or chart notes for the use of a unilateral transiliac screw versus bilateral transiliac screws. In two cases a transsacral cage was placed from L5 through S2 to add stability.
| Discussion | |
|---|
Twenty-seven cases of sacral fracture after lumbosacral fusion have been previously reported [1–10]. The prevalence is estimated to be approximately 1% for short fusion and 3% for long fusion [6]. Compared with the findings in these previously reported cases, age, sex, time to fracture diagnosis after the index procedure, clinical presentation, and underlying preoperative diagnosis were similar in our study.
The common sacral insufficiency fracture is characterized by fractures in one or both sacral alae parallel to the sacroiliac joints [11]. A horizontal component crossing the sacral body often occurs, accounting for the H-shaped Honda sign appearance typical of an insufficiency fracture on bone scans [12]. The pattern of sacral fractures after lumbosacral fusion was different. All 23 of our patients had a characteristic pattern with a horizontal orientation through the sacral body, fracture through one or both of the screw holes, and exit planes through both posterosuperior sacral alae. This fracture pattern was best appreciated on coronal reformatted images, the sagittal reformatted images being useful for assessing displacement and angulation.
To our knowledge, previous reports have not described the appearance of the fracture, but in those in which coronal or axial CT images are shown [1, 3, 8], the fracture pattern is the same as in our patients. All of the fractures in our series were visible on CT scans, but some were subtle because of beam-hardening artifact, and careful assessment of the sacrum in all three planes was necessary. One of the fractures was not visible on the initial MR images because artifact from the metal S1 screws obscured the fracture in all imaging planes. Radiography did not show the fractures in 22 of 23 patients, even in retrospect, possibly because these radiographs were long cassette images of the entire spine rather than focused examinations of the lumbar spine and because the sacrum was obscured on both views by overlying pelvic structures or bowel gas.
In a report of a series of nine patients with sacral fracture after lumbosacral fusion, Klineberg et al. [6] suggested that the screws might act as stress raisers. Clinical reports have shown that sacral fusion has a higher failure rate than fusion at other levels in the lumbosacral spine with respect to instrumentation failure, loss of sacral fixation, loss of lumbar lordosis, and pseudoarthrosis, even in patients without osteoporosis [13]. A biomechanical study of pedicle screw fixation in the lumbosacral spine showed that bicortical sacral screw purchase was significantly stronger than unicortical purchase when measurements of mean pullout force were made at different depths of insertion [14]. Vavken and Krepler [9], however, suggested that bicortical screws can increase stress concentration. Another biomechanical study [15] showed that load distributions are altered in long fusion constructions, substantial increases in stress occurring at adjacent spinal levels. That most of the patients in our study underwent long fusion and all fractures involved the sacral screws supports the theory that sacral screws act as stress raisers.
It has been suggested that osteoporosis is an important risk factor, but only four of our 23 patients had osteoporosis. Similarly, obesity has been suggested as a risk factor for sacral fracture after fusion [6–9], and 12 of our 23 patients were overweight, but no definite conclusion can be made that increased weight is a risk factor. In addition, no conclusion can be made about the effects of smoking, diabetes, steroid use, Parkinson disease, and previous radiation therapy given the small number of patients in our study with these conditions. To our knowledge, previous studies have not examined long-term bisphosphonate use as a risk factor for these fractures, although there is mounting evidence of fracture risk in the subtrochanteric femur with bisphosphonate use [16]. Although six of our patients were taking bisphosphonates at the time of surgery, because of the small number of patients, we did not find any causal relation.
Twelve of 23 patients in our study were treated with operative fixation. In comparison, in the case series reported by Klineberg et al. [6], six of nine patients needed surgical fixation, as did all five patients described by Papadopoulos et al. [8]. In contrast, Vavken and Krepler [9] successfully treated three of four patients conservatively, and those investigators' systematic literature review of the cases of 16 additional patients revealed that all of these patients responded well to conservative therapy. In our series, displacement was not a criterion for intervention, although two patients in our study did undergo immediate surgery because of displacement. In most of the cases in our study, the need for operative fixation was based primarily on pain severity and fracture healing after attempts at nonoperative management. The subjective opinion of the surgeon also played a role in the treatment algorithm. For example, one of the surgeons at our institution observes his patients for 3 months, allowing a chance for healing to occur and for pain to decrease. Patients who cannot tolerate the pain undergo transiliac fixation sooner. Evidence of radiographic healing is helpful, but the most important indicator is the degree of resolution of symptoms.
Our study had several limitations, including the retrospective design, which limited our ability to determine bone mineral density for all of the included patients. Clinical notes and operative notes might have been incomplete, resulting in omissions of important risk factors and lack of explanation of treatment parameters and decisions. For example, higher activity level after lumbosacral fixation can predispose patients to sacral fracture, but detailed information on patient activity level was not available in the clinical notes. In addition, the small number of patients precluded statistical comparisons. Although long fusions were present in 17 of 23 patients and there was a suggestion that obesity and osteoporosis may not play as large of a role in these fractures as previously thought, we cannot determine whether these conditions are statistically significant risk factors.
| Conclusion | |
|---|
Sacral fractures after lumbosacral fusion have a characteristic transverse pattern through the sacral screw holes, differing from the configuration of common sacral insufficiency fractures. This complication is difficult to diagnose on radiographs but is readily diagnosed with CT. The entity should be considered when a patient has new symptoms after lumbosacral fusion. Radiologists should carefully assess postoperative CT and MR images for sacral fractures because these fractures can often be partially obscured by beam-hardening artifact on CT scans and dephasing artifact on MR images.
| References | |
|---|
