Open versus minimally invasive sacroiliac joint fusion: a multi-center comparison of perioperative measures and clinical outcomes

All patients were diagnosed with SI joint disorders using a combination of detailed history, clinical exam, imaging and diagnostic injections. A positive result on 3 or more pain provocation tests such as Gaenslen’s, flexion abduction external rotation (FABER), compression, distraction and thigh thrust, was used as criteria for further testing to confirm the SI joint as the primary pain generator [11, 31]. Diagnostic imaging studies such as x-ray, CT and MRI were performed on all patients to assess pathology in the lumbopelvic hip complex for differential diagnosis. Image-guided intraarticular anesthetic injections were performed as a final step to confirm the diagnosis. All patients had failed a 6-month course of non-surgical treatment consisting of a combination of medication optimization, activity modification, physical therapy and SI joint injections before they were offered surgery.

Several techniques for open fusion of the SI joint have been reported [7, 20, 21, 24]. However, open SI joint surgery technique varied minimally between sites participating in this study. These variations are described where applicable. All sites in this study performed an open posterior approach to the SI joint. After the administration of general endotracheal anesthesia, the patient was placed prone on a radiolucent table, wired for electromyography, and prepped in the normal sterile fashion. A longitudinal incision was made centered over the posterior-superior iliac spine and deepened to expose the bone. Retractors were used to pull back the soft tissue and expose the posterior portion of the inferior SI joint. An osteotome was used to remove the portion of the posterior iliac crest that overhangs the SI joint. This bone was morselized and was later used as graft. Curettes and rongeurs were used to remove the cartilage from the articular portion of the joint and the interosseous ligament from the fibrous portion of the SI joint. One or two holes to accommodate cages were drilled into the SI joint and enlarged with a reamer. The cage(s) were packed with morselized bone and/or rhBMP (off label for this indication) and placed into position under fluoroscopic guidance. Additional bone material was then packed into the remaining open parts of the SI joint. Two 6.5 × 40 mm cancellous iliosacral lag screws were then placed in standard fashion. Under fluoroscopic guidance, pins were placed from lateral to medial across the ilium, across the SI joint and into the sacrum. The pins were then over drilled with a cannulated drill. The holes were tapped with a 6.5 mm tap and the screws with large washers were placed. At one site (AGS), the fixation consisted of a 6.5 mm pedicle screw placed into the S1 pedicle and a second 6.5 mm pedicle screw placed between the inner and outer tables of the ilium. A 3 cm spinal rod was then used to connect the two screws. As for the fixation, one surgeon (MGM) used recon plating across the SI joint with placed 2 cancellous screws placed in both the sacral ala and the ilium, as well as one long percutaneous cannulated screws across the SI joint. For all surgeons, EMG stimulation was used throughout the procedure to ensure safe placement of instrumentation. The wound was then irrigated, a hemovac drain was placed between the deep and superficial fascia, and the tissue layers were closed sequentially.

Minimally invasive SI joint surgery using a series of triangular, titanium, TPS coated implants (iFuse Implant System) (Figure 1) was performed with the patient on a radiolucent table to facilitate the use of intraoperative fluoroscopy. After general endotracheal anesthesia was administered, the patient was turned prone and prepped in the normal sterile fashion. A lateral incision (3 cm) was made into the gluteal region, positioned over the sacral body as viewed on a lateral fluoroscopic image. The fascia was then bluntly dissected to reach the outer table of the ilium. A Steinmann pin was passed through the ilium across the SI joint to the center of the sacrum (lateral to the neural foramen). After a soft tissue protector was passed over the pin, a hand drill was used to create a pathway across the ilium, across the SI joint and into the sacrum. Finally, a triangular broach was used to further decorticate the bone and prepare a triangular channel to receive the first implant. Using a pin guidance system, a total of three implants were placed. The most cephalad implant was seated within the sacral ala above the first neural foramen. The second implant was located above or adjacent to the S1 foramen and the third between the S1 and S2 foramen (Figure 2). The incision was irrigated and the tissue layers were closed. A variable program of gradual return to full weight bearing was employed based on local practices and patient needs. In general, patients were instructed to ambulate partial weight bearing with the assistance of a walker for the first 3 weeks after which time toe touch ambulation was recommended for another 4 weeks. After a regimen of gradual return to full weight bearing, patients began 4 weeks of physical therapy.

Patient reported clinical outcomes were collected prospectively prior to surgery to establish baseline values and at follow up intervals per each surgeon’s routine practice. For consistency in reporting, pain scores using visual analog scale (VAS) were collected at visits close to 12 and 24 months only.

Clinical improvement was defined using minimum clinically important difference (MCID) and substantial clinical benefit (SCB) values available in the literature. As there are currently no reported MCID or SCB values for SI joint fusion, values were chosen using lumbar spine criteria published by Copay et al. for MCID and Glassman et al. for SCB [32, 33]. MCID for pain is defined as a change of >2.0 pts and SCB is defined as a 2.5-point decrease or raw score of < 3.5.

Baseline demographic variables were summarized with mean, standard deviation and frequency tables. Demographic characteristics were compared across cohorts using t tests for continuous variables and chi-squared tests for nominal variables. Surgical parameters (e.g., operating time, estimated blood loss, length of stay) were compared using t tests.

A key outcome variable was the change in pain scores on VAS. Mean baseline VAS scores, as well as mean change from baseline at 12 and 24 months, were tabulated. To account for potential differences in patient characteristics across cohorts, VAS pain scores were compared after matching, as follows. Patients in the two treatment groups were matched based on gender and age in 5-year intervals (20–25 years, 25–30 years, etc.). This process led to a variable m:n matching where m ≠ n. Variable matching was due in part to missing values of the pain score as well as differences in overall age distribution between the two cohorts.

A linear mixed effects model (SAS PROC MIXED) was used to estimate treatment differences using a cell means formulation with additional covariates of prior lumbar fusion and a random effect for patient. Prior lumbar fusion was included as a covariate since it may increase the risk of SI joint degeneration, and also reflects a history of serious back pain requiring surgery. The cell means formulation has a parameter for each treatment, each block (i.e., gender and 5-year age category), visit, and a random effect for patient. Pairwise differences between treatments for blocks of matched age-category/gender were estimated. Sparse combinations of age/gender blocks that resulted in inability of the model to provide estimates were removed from the model. The final model was estimable and converged. A final treatment effect estimate was calculated by averaging over matched age-category/gender/treatment combinations. Analysis was performed using SAS 9.0 (Cary, NC).

No intraoperative complications occurred. Postoperative complications were slightly more common in the open surgery group (21% of patients) compared to the MIS group (18%) (Table 6). The most common complications reported in both groups were postoperative neuropathy and transient trochanteric bursitis (4 OS and 2 MIS). In the OS group, leg pain (3), neuropathy (4) and wound related issues (6) were more common. In the MIS group, falls (4) and facet pain (4) were more frequent.

Forty-four percent (66/149) of patients in the open surgery cohort subsequently underwent removal of spinal implants. In most cases, removal was for pain at the ilial or sacral screw. In contrast, only 3.5% of patients (4/114) undergoing MIS surgery underwent postoperative repositioning of the implants. In 3 patients, reoperation was due to nerve root impingement discovered on the postoperative CT scan. In one case, reoperation was performed at the surgeon’s discretion based on radiographic findings only.