Tumors of the sacrum present a difficult problem for patients and surgeons. The complex presacral anatomy is frequently distorted because of mass effect, radiation, or scarring. This makes resection of tumors in this region demanding. The functional consequences of resection must be carefully considered when dealing with sacral malignancies. The sacrum not only contains important neurological structures, but also serves as the only mechanical connection between the axial skeleton and the lower extremities through the sacroiliac joints. It is critical to be aware of the consequences of resections in this region on bony stability and neurological function.
Approximately 2/3 of all sacral tumors present at or below the S2 level.5 Transverse partial sacrectomy is a surgical option for treating tumors of the caudal sacrum. This procedure can potentially afford excellent local control while maintaining pelvic ring and spinal column stability. However, the mechanical consequences of high partial transverse sacrectomy are not well understood. In a large series of patients undergoing partial sacral resection for chordoma, three of nine patients undergoing sacrectomies involving a portion of the S1 body experienced postoperative fractures (Fig 1).5 These three fractures resulted in subsequent complex reconstruction (iliolumbar arthrodesis) using spinal instrumentation.
Gunterberg et al2 pioneered the study of sacropelvic mechanics in 1976 by performing a cadaveric study that investigated the effects of transverse partial sacrectomy. Transverse partial sacrectomies were performed through the S1 foramina (between S1 and S2 vertebral bodies) or through the S1 body (1 cm below the sacral promontory).2 The specimens were then loaded vertically and tested to failure.2 The resections weakened the pelvis by 30% and 50%, respectively.2 Based on the published data of the time regarding forces transmitted at the L5/S1 articulation, they reported that even the higher of the two osteotomies did not weaken the pelves to the extent that they could no longer withstand normal standing forces imparted at the L5/S1 articulation.2 They recommended that it was safe to allow full weightbearing after submaximal resection of the sacrum.2 However, their study did not entirely simulate in vivo loading conditions.2 The cadaveric specimens in the study by Gunterberg et al2 were stabilized during testing by potting the base (the rami) in epoxy resin. This configuration created false loading and response conditions by two separate means. First, the reactant forces to the load placed at the L5/S1 disk space were applied through the potted portion of the specimen (the ischium, rather than at the hip, which is the natural loading condition). By moving the reactive forces to a more central position, the moment imparted on the iliac wings decreases substantially, which alters the load distribution across the sacroiliac joint. Second, with the rami potted in epoxy resin, the iliac wings were restricted against any significant angulation or splaying during loading. This is an important point because the sacrum is a wedge and would normally act to splay the pelvis in the physiologic loading situation. The combination of these two loading conditions during testing would work together to falsely increase the apparent ability of the cadaveric pelves to withstand a vertically oriented load at the lumbosacral junction.
We raised several issues. First, we sought to determine the load to failure of the normal pelvis and pelves with transverse partial sacrectomies above and below the level of the S1 neural foramina. Related questions included: How much of the sacroiliac joint is resected when sacrectomy is performed above and below the S1 neural foramina? What are the stiffness and most common “mode” of failure when loading the pelves after transverse sacrectomy at these levels? How do the individual elements of the mechanical system move with respect to one another during loading? Lastly, how do the failure loads that we discovered compare to normal physiologic loads experienced across the lumbosacral junction?
MATERIALS AND METHODS
Ten pelves were randomized into one of three groups. The Control Group (three pelves) was tested with completely intact sacrums. The sacrospinous and sacrotuberous ligaments were left undisturbed in these specimens. In Group I (four pelves), trans- verse partial sacrectomy was performed just caudal to the S1 ventral foramina and parallel to the exiting S1 nerve roots in the plane of the true pelvis. In Group II (three pelves), partial sacrectomy was performed cephalad to the S1 foramina in the plane of the pelvic ring (Fig 2). Specimens that had bone tumors (primary or metastatic) or had undergone radiation therapy to the pelvis were excluded. The age, height, weight, and cause of death were recorded for each specimen. Soft tissues were stripped from the pelves with the exception of the sacroiliac ligaments, sacrotuberous ligaments, and sacrospinous ligaments. The symphysis pubis was left undisturbed and the hips were disarticulated. Several morphometric parameters, including true pelvic dimensions, transischial distance, and transfoveal distance were measured and recorded. The demographic and morphometric data for three groups were similar (Table 1). The lumbar spine was amputated at the L4/L5 disk space, leaving the L5 vertebral body and distal segments attached to the pelvis. We obtained prior institutional review board approval.
A testing apparatus was constructed to simulate the loading conditions in the pelvis (Fig 3). Two 46-mm metallic spheres were placed on vertical posts. These supported the pelves through the acetabulum. The ball and post supports were placed on x-y slides, allowing free translation in the coronal plane without significant friction. This allowed the iliac wings to splay or angulate without restriction during loading. A curved anterior blocking plate was hooked around the symphysis to stabilize the pelves against rotation in the sagittal plane during loading without restricting motion of the innominate bones. The pelves were oriented so the anterior superior iliac spine and pubic tubercle were in the vertical plane during testing to simulate standing. This simple testing arrangement allowed for the free, unrestricted response of each component of the bony pelvis to the loads applied.
After the transverse partial sacrectomies were completed, each pelvis then was mounted on our mechanical testing apparatus (Fig 4). A vertical load was applied through the L4/L5 disk space at a rate of 1 mm/second using a servohydraulic mechanical tester (MTS, Eden Prairie, MN). Force and displacement data were collected (Fig 5). Reflective markers were placed on the ileum and sacrum bilaterally, adjacent to the sacroiliac joints. The position of these markers was tracked by video dimensional analysis (VDA) to measure translation and rotation of the various bony components as they occurred during loading (CCD cameras, Eva 6.0 HIRES VDA software; Motion Analysis Corporation, Santa Rosa, CA). Testing continued until failure occurred (Fig 6). Failure was defined as a 10% reduction in load capacity from the peak value.
After mechanical testing all pelves were carefully inspected. Fracture patterns were identified and recorded. The fractures were described according to Denis et al1 as Zone I (occurring lateral to the sacral foramina), Zone II (through the sacral foramina), or Zone III (medial to the sacral foramina). The sacroiliac joints were dissected free and disarticulated. We used digital imaging to measure the surface area of the remaining sacroiliac joint after osteotomy. This was recorded as a percentage of the entire area of the sacroiliac joint before osteotomy. In Group I, an average of 16% was resected. In Group II, an average of 25% was resected. The percentage resection was different between the three groups (Control Group versus Group I, p = 0.002; Control Group versus Group II, p = 0.0003; Group I versus Group II, p = 0.0309).
The mean, standard deviation (SD), and 95% confidence interval (CI) were calculated for each endpoint. The load data collected consisted of the force applied and the displacement of the load applicator. Using these values, the yield strength and apparent stiffness were calculated. The VDA provided independent spatial coordinates of the sacral ala and iliac wings bilaterally during loading. Video dimensional analysis was used to measure the diastasis at the upper and lower aspects of the remaining sacroiliac joint and angulation of the iliac wings during loading. These data were scrutinized to help determine the translation and angulation of the individual elements as the pelves failed.
The data from each of the three test groups were compared using Student's t-tests. Probability values of less than 0.05 were considered significant. The Shapiro-Wilk test for normality was applied to each data set and all were considered normal distribution populations within the 0.05 level.
The average load to failure decreased with more cephalad sacrectomies (3014 N for control group, 2166N for Group I, and 1044N for Group II). We observed differences between Group II and the Control Group (p = 0.03), and Group II and Group I (p = 0.04) (Table 2).
The average stiffness also trended down (between Groups I and II, p = 0.04, but not between Control Group and Group I or Group II) with more cephalad sacrectomies (353 N/mm in the Control Group, 248 N/mm in Group I, and 101 N/mm in Group II).
The diastasis at the upper sacroiliac joint was similar between all groups and averaged 1.41 mm in the Control Group, 0.97 mm in Group I, and 0.93 mm in Group II. The mean diastasis at the lower sacroiliac joint was similar in the three groups: 1.58 mm in the Control Group, 1.77 mm in Group I, and 1.55 mm in Group II. The mean angulation of the iliac wings also was similar in the three groups: 3.76° in the Control Group, 4.29° in Group I, and 3.49° in Group II.
Failure patterns were most commonly the result of vertically oriented fractures in the paramedian sagittal plane (Denis Zone 2, see Table 3). Most occurred in Denis Zone II (Fig 7). Most failures in Groups I and II occurred from sagittal plane rotation at the lumbosacral junction, whereas fractures in the Control Group occurred mostly from vertical migration. This rotation caused fractures paramedi- ally first on one side and then the other. None of the failures occurred at the sacroiliac articulation. One specimen in the Control Group had sacral and transverse ace- tabular fractures at failure. Another specimen failed at the sacrum and the symphysis pubis anteriorly. Seven of the eight specimens in the sacrectomy groups failed with rotation of the sacral segment, whereas each of the Control Group specimens failed with vertically oriented displacement only.
Approximately 2/3 of sacral tumors arise at or below the level of S2.5 These lesions of the caudal sacrum can sometimes be effectively resected with transverse partial sacrectomy. Great care is taken to resect only the portion of the sacrum necessary for local disease control, sparing as much of the sacroiliac joint and as many of the lumbosacral nerve roots as possible. The critical clinical question to ask is “What is the lowest nerve root that can be saved after choosing a level cephalad enough to achieve a wide margin?” After this is determined radiographically, a chevron osteotomy just caudal to the lowest salvageable nerve root is planned.
We note several important limitations. Although we attempted to emulate the natural loading conditions to the best of our ability, we were not able to exactly duplicate in vivo conditions. In addition, the small numbers of specimens used in this study limit the power of the study statistically and we cannot ensure the trends we report would occur clinically given the small numbers and the limited range of in vivo loading situations. Although we excluded specimens with certain histories (radiation treatment or malignancy), we did not verify the consistency of specimen bone quality (ie, using dual-energy xray absorptiometry scans). Inconsistencies in bone quality between specimens could influence our results.
The mechanical consequences of transverse partial sacrectomies are poorly understood. Destabilization of the spinopelvic segment may require bony stabilization and reconstruction using complex methods that involve lumboiliac arthrodesis (Fig 8). Such a reconstruction effort can add significant complexity and morbidity to sacral resection alone. A transverse partial sacrectomy below the S2 foramina typically would not involve resection of the sacroiliac joints and would likely have little effect on stability. Transverse osteotomies involving the S2 and S1 sacral bodies are less well understood and were the focus of our investigation.
Under normal conditions, the sacroiliac articulation is stabilized by a number of different factors. Its wedge shape is positioned between the two large iliac wings. This spatial arrangement conveys stability against caudal migration of the sacrum. The sacroiliac joint has an irregular lining and forms broad, interlocking surfaces with the ileum, which resist motion effectively. The strong sacroiliac, sacrotuberous, sacrospinous, and lumbosacral ligaments also play an important role in stabilization as these are among the strongest ligaments in the body. The combination of these factors renders the spinopelvic segment especially stable.
Our testing system was designed to mimic actual loading conditions as accurately as possible. The force was applied vertically at the L4/L5 junction with the pelvis in the standing position (the anterior superior iliac spine and the pubic tubercle in the vertical plane). The supports were applied at the acetabula via the spherical ball and post constructs. The spherical ball engaged the acetabulum and acted as a universal joint, supporting against vertical load, but not inhibiting the ability of the iliac wing to rotate in any plane. The ball and post were set on x-y slides that were able to move in the coronal plane without significant friction. This allowed for unrestricted splaying of the pelves when loaded. In addition, the anterior blocking plate controlled rotation in the sagittal plane. The pelvis was otherwise unconstrained and allowed to react naturally to the load applied at the L4/L5 disk space.
Fractures occurred in a similar pattern throughout the three groups. In all cases, the initial failure occurred paramedially in a vertical orientation. In the sacrectomy groups, the fractures often seemed to be the result of spin of the promontory in the sagittal plane when loading occurred. As in the study by Gunterberg et al,2 there were no failures at the sacroiliac joint. It seemed the weak mechanical link in our system was the inability of the base of the sacral ala to resist sagittal plane rotation (Fig 6).
To determine how much stability is enough to allow early weightbearing after partial sacrectomy one must consider what loads are transmitted through the lumbosacral junction during normal activity. Sato et al4 studied intradiscal pressure measurements in normal subjects and patients with back pain at the L4/L5 disk (Table 4). They found great variation in intradiscal pressures depending on the patient's position. Standing with flexion elicited the highest pressures.4 The mean intradiscal pressure averaged 1324 kPa, and the average cross-sectional area at the L4/5 disk space was 15.9 square cm.4 This translated to a vertical load of 2105 N transmitted along the lower lumbar spine in this position.4
Using these numbers as an estimate of loads transmitted across the lumbosacral junction, it would seem that Group I may be able tolerate weightbearing immediately postoperatively. None of the normal load values discovered exceeds the average failure load of Group I (2166 N). It may be advisable for this group to avoid high-stress postures, such as forward flexion while standing, until bone adaptation has sufficiently compensated for the redistribution of forces postoperatively. Group II resections would not likely tolerate the forces seen during routine activities. The normal loads in the standing-flexion (2105 N), sitting- flexion (1801 N), and sitting-extension (1170 N) positions all exceed the average failure load found in Group II (1044 N). The normal loads in the standing-upright (857 N), standing-extension (954 N), and sitting-upright (990 N) positions all approached the average failure load in Group II as well. These values indicate that under normal conditions, depending on the posture assumed by the patient, they would fail or approach failure unless lying in the supine or decubitus position.
We believe our findings contribute to the understanding of the mechanical consequences of high transverse partial sacrectomy. The data may be useful in determining which patients should be considered for reconstruction after partial sacral resection.
1. Denis FS, Davis T, Comfort T. Sacral fractures: an important problem. Retrospective analysis of 236 cases. Clin Orthop Relat Res
2. Gunterberg B, Romanus B, Stener B. Pelvic strength after major amputation of the sacrum. An experimental study. Acta Orthop Scand
3. Nachemson A, Elfstrom G. Intravital Dynamic Pressure Measurements in Lumbar Disks
. Stockholm, Sweden: Almqvist & Wiksell; 1970.
4. Sato K, Kikuchi S, Yonezawa T. In vivo intradiscal pressure measurement in healthy individuals and in patients with ongoing back problems. Spine
5. Todd LT Jr, Yaszemski MJ, Currier BL, Fuchs B, Kim CW, Sim FH. Bowel and bladder function after major sacral resection. Clin Orthop Relat Res