Abstract
Background Sacroiliac (SI) joint fusion is increasingly used to treat chronic SI joint pain. Multiple surgical approaches are now available.
Methods Data abstraction and random effects meta-analysis of safety and efficacy outcomes from published patient cohorts. Patient-reported outcomes (PROs) and safety measures were stratified by surgical technique: transiliac, including lateral transiliac (LTI) and posterolateral transiliac (PLTI), and posterior interpositional (PI) procedures.
Results Fifty-seven cohorts of 2851 patients were identified, including 43 cohorts (2126 patients) for LTI, 6 cohorts (228 patients) for PLTI, and 8 cohorts (497 patients) for PI procedures. Randomized trials were only available for LTI. PROs were available for pain (numeric rating scale) in 57 cohorts (2851 patients) and disability (Oswestry Disability Index [ODI]) in 37 cohorts (1978 patients).
All studies with PROs showed improvement from baseline after surgery. Meta-analytic improvements in pain scores were highest for LTI (4.8 points [0–10 scale]), slightly lower for PLTI (4.2 points), and lowest for PI procedures (3.8 points, P = 0.1533). Mean improvements in ODI scores were highest for LTI (25.9 points), lowest for PLTI procedures (6.8 points), and intermediate for PI (16.3 points, P = 0.0095).
For safety outcomes, acute symptomatic implant malposition was 0.43% for LTI, 0% for PLTI, and 0.2% for PI procedures. Wound infection was reported in 0.15% of LTI, 0% of PLTI, and 0% of PI procedures. Bleeding requiring surgical intervention was reported in 0.04% of LTI procedures and not reported for PLTI or PI. Breakage and migration were not reported for any device. Radiographic imaging evaluation reporting implant placement accuracy and fusion was only available for LTI.
Discussion Literature support for SI joint fusion is growing. The LTI procedure contains the largest body of available evidence and shows the largest improvements in pain and ODI. Only LTI procedures have independent radiographic evidence of fusion and implant placement. The adverse event rate for all procedures was low.
Level of Evidence 1.
Introduction
The sacroiliac (SI) joint is a common cause of chronic low back pain (15%–30%).1,2 In patients with low back pain following lumbar fusion, SI joint pain may be a more common cause (as high as 40%).3 Nonsurgical treatments for SI joint pain include physical therapy, pain medications, SI joint injections (corticosteroids or other substances), and radiofrequency ablation.
Fusion of the SI joint was first reported in the early 1900s.4 While several technical variations on open SI joint fusion have been reported,5 open fusion techniques solely to address discrete SI joint pain are now rarely performed due to significant operative and postoperative morbidity.6 The American Medical Association (AMA) assigned a Current Procedural Technology (CPT) to the open SI joint fusion procedure in the 1980s and refined the description in 2014.
In the early 2000s, minimally invasive transiliac procedures for SI joint stabilization and fusion were developed. These procedures follow the lateral trajectory described by Routt for the stabilization of traumatic injuries involving the SI joint.7 The earliest (and most common) version of this procedure involves the placement of metallic devices laterally to medially through the ilium, across the SI joint and into the sacrum (Figure 1). The term lateral transiliac (LTI) is used here to describe this procedure. Although standard bone screws were used initially, in the late 2000s, triangular titanium implants (TTI) (iFuse Implant System, SI-BONE, Santa Clara, CA) were designed specifically for SI joint fusion procedures using an LTI approach. Typically, 3 devices are placed such that the implants traverse 3 bony cortices and terminate near or at the sacral body. Termination at or near the sacral body is advantageous as the body has superior bone quality compared with the ala.8,9 These implants stabilize the SI joint acutely through their triangular shape, which resists rotation within bone, as well as forming multiple points of fixation. Long-term (permanent) stabilization is achieved via bone growth onto the implant surface and through the implant fenestrations and across the SI joint. Several biomechanical studies provide additional support for these implants.10–12 Early studies of the use of TTI for SI joint fusion via the LTI procedure showed that this approach provides immediate and sustained improvement in pain and related disability, with improvements that exceeded those of nonsurgical treatment.13–15 With accumulating evidence, AMA CPT published a code (27279) that describes minimally invasive SI joint fusion using this LTI approach. Subsequently, and not surprisingly, additional metallic devices were developed and cleared for SI joint fusion by the US Food and Drug Administration.
More recently, a variation of the LTI procedure, in which the devices are placed at a posterior lateral angle across the SI joint (see Figure 1), was developed. Here, this is referenced as a posterolateral transiliac (PLTI) procedure. Examples of US Food and Drug Administration–cleared devices for PLTI procedures include RIALTO (Medtronic), Sacrofuse (Sacrix), SI-LOK (Globus Medical), and Transloc (Foundation Fusions Systems). As these procedures involve placement of devices through the ilium across the SI joint into the sacrum, they are considered a variant on transiliac procedures described by CPT 27279. In contrast to LTI, the PLTI technique has a more posterior starting point and takes an oblique trajectory across the SI joint. It was developed to lower the risk of injury/irritation of the S1 and S2 nerves within the foramina and to the branches of the superior gluteal artery.
Another procedure involves placing implants (either structural bone allografts or metallic devices) directly into the SI joint from a posterior approach without transfixing the joint, that is, a posterior interpositional or intra-articular (PI) approach. Examples of metallic devices include NADIA (Ilion Medical) and DIANA (SIGNUS). The same PI procedure can be performed with structural allografts, which are available from multiple vendors. The PI procedure relies on ligamentotaxis for early/short-term SI joint stabilization and distraction arthrodesis (DA) for long-term SI joint fusion. The AMA CPT panel provided clarification that such PI procedures are fundamentally different than transiliac procedures (LTI and PLTI), and a new tracking code (0775T) was assigned. A new AMA CPT Category I code for minimally invasive SI joint fusion with a PI procedure will become effective on 1 January 2024.
Finally, some recently marketed devices are placed in a posterior procedure with device elements that span or bridge the SI joint and engage in both the medial ilium and the lateral sacrum. Examples of such devices include Catamaran (Tenon) and Transfix (Aurora). Published clinical literature describing the safety and/or effectiveness of this type of device and this procedure variation is currently lacking.
Previous meta-analyses of the safety and effectiveness of various procedures for minimally invasive SI joint fusion have been published. To date, no meta-analysis has characterized minimally invasive SI joint fusion safety and effectiveness on the basis of implant placement procedure using the most recently updated AMA CPT procedural descriptions. In the present article, we identify and analyze published clinical cohorts of patients undergoing minimally invasive SI joint fusion using transiliac (LTI and PLTI) procedures and PI procedures.
Methods
The reporting of this systematic review was guided by the standards of the Preferred Reporting Items for Systematic Review and Meta-Analysis Statement.
Literature Search
Published literature describing cohorts of patients undergoing SI joint fusion was identified through structured Medline searching using the search term “sacroiliac joint fusion” or “sacroiliac joint arthrodesis.” Articles published between January 2010 and June 2023 were included. A literature search was performed on 28 March 2023 and updated in July 2023. Studies were included if they included patients seeking operative care for chronic SI joint pain diagnosed via medical history, physical examination, and diagnostic SI joint block and reported operative and/or long-term (>3 months) outcomes after minimally invasive SI joint fusion using metallic devices or allografts. Studies with incomplete reporting, or those reporting only follow-up scores (e.g., Beck16), were excluded. Where possible, duplicate reports of the same cohort were excluded.
Data Abstraction
Abstracts from Medline searches were reviewed, and relevant full-text articles were obtained. Studies were reviewed by at least 2 authors. Any discrepancies in data abstraction were resolved by discussion. For each article, information shown in Table 1 was reviewed and extracted into an Excel worksheet. Results from studies reporting the use of more than 1 device were, where possible, extracted separately. Hence, reporting is on the basis of cohorts, not studies.
Study design is a potentially important source of bias. Studies that noted prospective capture of data but no systematic reporting of follow-up data (eg, number withdrawn, loss to follow-up, etc) were deemed retrospective.
Statistical Analysis
Excel data were imported into and analyzed in R on the RStudio platform. Random effects meta-analysis and meta-regression were performed using the metafor library.17 Meta-regression focused on change scores only. Graphical analysis was performed using the ggplot2 library.
Poolability was assessed on the basis of the combination of study design, target patient population, interventions, and data reporting. Although study reporting was not all complete, all identified studies appeared to target the same patient population and use similar approaches to data capture.
Many studies did not report change scores or measures of variation (ie, SD). Missing SDs were imputed as the average of available reported SDs. Studies not reporting baseline scores were generally excluded. For those studies using more than 1 device that reported outcomes separately, results were separately abstracted. When duplicate studies were identified, only the latest study was used for abstraction. Funnel plots were examined within procedure type to detect evidence of publication bias. Data are available upon reasonable request.
Results
Dataset
The literature search resulted in 464 hits. After review of abstracts and addition of 4 previously known cohorts not identified via Medline hits, a total of 122 full-text articles were obtained and reviewed, of which 56 met eligibility criteria. An additional literature search in July 2023 revealed 5 additional relevant cohorts. One study of placement of plates (not minimally invasive) or implants placed in a sacroalar iliac trajectory (not commonly performed in the United States and Europe) was excluded.18
Efficacy
Efficacy results were reported in 57 cohorts (2851 patients, Table 2) for pain scores and 37 cohorts (1978 patients) for Oswestry Disability Index (ODI). The largest number of studies involved use of iFuse Implant System (SI-BONE, Inc., Santa Clara, CA, USA), placed through the LTI procedure. There were fewer PI studies reporting pain or ODI scores, but these studies were more likely to be prospective (Table 3).
Baseline (green) and last follow-up (blue) mean scores for each cohort are shown in Figure 2 for pain scores and Figure 3 for ODI.
ODI was reported in 31 LTI cohorts (1584 patients), 2 PLTI cohorts (135 patients), and 3 PI cohorts (259 patients). Prospective study designs were used in 25% of LTI cohorts, 0% of PLTI cohorts, and 67% of PI cohorts. Meta-analytic mean scores at baseline and last follow-up were as follows: for LTI procedure, 56.2 (52.5–60) and 30.9 (26.8–34.9); for PLTI procedure, 51.5 (47.4–55.6) and 44.8 (39.3–50.3), and for the PI procedure, 54.6 (48.3–60.9) and 40.6 (28.2–53.1).
Change scores for pain ratings and ODI are shown in Figure 4 and Figure 5. These figures similarly show that meta-analytic mean improvements for pain scores were larger for LTI procedures (4.8 points, 95% CI 4.5–5.2), slightly lower with PLTI (4.2 points, 2.6–5.8), and lowest with PI procedures (3.8 points, 2.9–4.7, analysis of variance [ANOVA] P = 0.1533). Similarly, ODI scores were larger for LTI procedures (25.8, 22.8–28.9) vs PLTI (6.8, 4.0–9.6) and PI (16.3, 12.0–20.6, analysis of variance P = 0.0095) procedures. This analysis was limited as only 2 PLTI studies and 3 PI studies included ODI. Funnel plots (not shown) did not indicate publication bias.
Meta-regression showed no impact of study design (ie, prospective vs retrospective studies) on change scores for either SI joint pain or ODI.
Safety
Safety outcomes by procedure and study are shown in Table 4. In total, 63 cohorts reported outcomes on 3162 patients (Table 5). Prospective designs were used in 9/47 (19%) of LTI cohorts, 0/8 (0%) of PLTI cohorts, and 3/8 (38%) of PI cohorts. Most published studies involved LTI placement of metal implants (47 cohorts and 2348 patients). In the LTI category, iFuse Implant System was the most commonly reported device used (32 cohorts and 1637 patients). The PLTI procedure was reported in 8 cohorts (317 patients). The PI procedure (with placement of structural allograft[s] or metallic implants) was reported in 8 cohorts (497 patients).
The rate of selected safety outcomes is shown in Table 6. Acute implant malposition was reported in 0.43% of patients undergoing placement of devices in the LTI procedure, 0% in the PLTI procedure, and 0.2% in the PI procedure. In the PI procedure, malposition was reported for a metallic device but not for structural allografts (though radiographic assessment was not performed in any PI allograft study).65 Several outcomes were not reported in any study (fracture, dislodgment of bone into the foramen, implant breakage, or bowel perforation). Bleeding requiring surgery was reported in 6 LTI studies (1 patient each, meta-analytic rate of 0.039% [95% CI 0%–0.163%]) but not in any PLTI or PI report. Device removal for pain was reported for all procedures (rates of 0.06% for LTI, 1.1% for PLTI, and 0.48% for PI). Wound infections were reported in LTI studies only (rate of 0.15%).
Including all studies, meta-regression showed that study design (prospective vs retrospective) did not statistically significantly impact safety outcome estimates. Similarly, study design did not have any effect on safety estimates within LTI studies. Meta-regression was not done for PLTI and PI studies due to the small number of reports.
Within LTI studies, meta-regression showed that device type was significantly associated with 2 of 9 safety outcomes. Acute malposition occurred more commonly with the Samba screw (2/9 cases in one study,50 χ2 P < 0.0001). Late removal due to pain occurred more commonly with SI-LOK (4/157 cases, 2.5%, χ2 P = 0.0165); this result was driven by a single study53 reporting 4 removals in 40 patients. Analysis across device types was not performed for PI and PLTI studies due to the low number of cohorts.
Discussion
Evidence for the safety and efficacy of minimally invasive SI joint fusion has grown substantially over the past decade. Multiple procedural approaches are now available, including LTI, PLTI, and PI (less common). However, previous meta-analyses have not distinguished outcomes by procedure type. This distinction is key as the procedural approaches differ in their approach to acute and long-term stabilization and fusion of the complex SI joint. Thus, the clinical outcomes and safety data for the different procedures should not be generalized; they are fundamentally different. These fundamental differences have been recognized by the AMA, resulting in separate CPT codes to describe the procedures.
The evidence base for minimally invasive SI joint fusion is fairly large. However, the great majority of published evidence involves the LTI procedure. The LTI procedure was based, in part, on previous reports of percutaneous screw fixation for sacral fractures as well as early results of an identical procedure for the treatment of chronic SI joint pain. Within the LTI procedure, the most commonly reported device (constituting approximately 3 quarters of all LTI patients reported) was triangular titanium implants (iFuse Implant System, SI-BONE). The evidence base for PLTI and PI procedures (7.6% and 15.2% of all patients with follow-up pain scores) was far less.
Our results generally suggest that minimally invasive SI joint fusion provides high levels of pain relief and disability improvement (as reflected by improvements in ODI) along with a reasonable safety profile. Of note, no cohort showed mean worsening of pain and/or disability. Meta-regression provided evidence to suggest that improvements in pain and disability were highest in LTI procedures and lower in the other procedures. In the PLTI procedure, implants are placed obliquely across the SI joint axis of rotation. A biomechanical study showed that this approach provided less stability compared with placement of implants parallel to the axis of rotation (as done with the LTI procedure).71 This decreased stability may be due in part to less bony engagement in the sacrum and implant termination point in the sacral ala vs nearer the sacral body. Moreover, because implants placed in the PLTI trajectory traverse the ligamentous portion of the joint, bone surfaces may be separated by several millimeters of ligamentous tissue, which is not conducive to new bone formation.72 In a comparative study, PTLI procedures had a higher incidence of persistent SI joint pain with radiographic evidence of lucency and nonunion compared with LTI.30
LTI procedures also appeared to show superior efficacy (pain and ODI response) compared with PI procedures. The PI procedure relies on DA, namely stabilization of the SI joint through ligamentotaxis achieved by distracting the joint with an implant. DA was previously used in the lumbar spine, but the technique is now rarely used due to substantial rates of implant subsidence, low rates of permanent fusion, and poor overall responses.73–75 Rather, current practices in the lumbar spine involve full preparation of the disc space with complete removal of disc material, preparation of the endplate, and stabilization of the motion segment with rigid anterior and/or posterior instrumentation. In the SI joint, DA with metallic devices was successful at achieving fusion only when the joint was thoroughly prepared and augmented with off-label use of rh-BMP.64,65 In early PI studies using metallic implants, the rate of poor implant position was high (5/1964 and “large percentage”).65 Also note that DA of the SI joint relies on accurate intra-articular placement of implants; a recent case series suggests a high rate of structural allograft placements outside the joint with consequent lack of joint fusion.76
The rates of some events (major bleeding and symptomatic implant malposition) have been called into question, especially for the LTI procedure.77 Our results suggest that event rates are low across all procedures, including the LTI. Implant malposition causing symptoms resulting from inadvertent nerve root irritation is more likely in the LTI procedure compared with other procedures as implants are directed toward the sacral foramina. However, the rate of this outcome was low in the LTI procedure (0.43%). This rate is consistent with postmarket surveillance reported by the manufacturer,78,79 including a decrease in rate over time.79 This rate is also consistent with the reported symptomatic malposition/revision rate for pedicle screws during lumbar fusion placed freehand (0.8%–3.25%).80,81 The rate of implant malposition varied across LTI device types, but this finding was driven by only 2 studies. Symptomatic malposition was reported in some PI procedure studies but no PLTI procedure studies. Direct comparisons of adverse event rates across procedures are generally not available. A single-center retrospective comparison of LTI and PLTI procedures published in an abstract only suggests a lower rate of adverse events with the PLTI approach.82 Detailed analyses of revision rates (mostly for implant malposition) have been reported for triangular implants only.78,79 Information on implant location in SI joint fusion procedures has not been fully evaluated. Symptom recurrence with implant malposition after PI procedures (allografts) has been reported.76
The rates of other outcomes (implant removal, dislodgment of bone fragments into the sacral foramina, fracture, wound infection, and bleeding requiring surgery) were low in all groups. The rate of LTI implant removal varied across device types, but these results were driven by a small number of studies. No study reported breakage or migration of implants. In studies of the PI procedure using structural allograft, follow-up radiographic assessment was not performed. This prevents assessment of migration, subsidence, fracture of the structural allograft, or fusion status.
Implants for SI joint fusion are likely effective only if placed accurately in the target location. Placement accuracy as assessed by implant engagement length into the sacrum was reported in one LTI study.14 Placement accuracy was indirectly assessed in other LTI studies of the iFuse Implant System through independent radiographic analysis of fusion outcomes.27 No other studies reported implant placement accuracy. Inaccurate placement of structural allografts leading to revision surgeries has been reported.76 To our knowledge, prospective studies of PI allograft placement have not included radiographic assessment. Only studies of iFuse implants have used independent radiologists for assessment.
Other than for some LTI devices, published studies of SI joint fusion have not extensively evaluated radiographic joint fusion. Whang et al reported high joint fusion rates (88%) for iFuse implants based on independent radiographic analysis.27 Some studies reported high joint fusion rates but without independent assessment.55,62 Other studies reported “probable fusion” on the basis of clinical findings (absence of loosening on x-ray plus symptom improvement).51 As noted above, some studies excluded all radiographic assessment, including fusion.66 There appears to be no consensus on the definition of fusion of the SI joint. Successful fusion of the lumbar spine appears to predict increased improvement in pain and disability scores.83 The rate of SI joint fusion across procedures, as well as the correlation between SI joint fusion and symptom relief, remains underexplored.
Overall literature quality was moderate. A minority of included cohorts were prospective, that is, a higher level of study design. Most studies were retrospective case series. No prospective studies were available for the PLTI procedure. Only LTI included randomized trials. Except for one peer-reviewed cohort showing no differences between an LTI and a PLTI procedure,30 prospective comparative data are not available.
Of interest is the difference between metallic devices and structural allografts. Whether head-to-head studies need to be executed is a matter of discussion.
Detailed per-study information was not sufficient to determine with confidence whether pooling results across studies was reasonable. However, it appears that the same diagnostic algorithm was used for diagnosis in most studies (history, physical examination with the use of 3 or more physical examination tests that provoke typical pain near the PSIS, and diagnostic intra-articular SI joint block using a small amount of local anesthetic). Moreover, all studies showed improvement in one or both efficacy parameters (pain and ODI), and all studies showed low overall adverse event rates, suggesting the observation of similar phenomena across studies. Funnel plots showed no systematic bias.
Other aspects of SI joint fusion procedures remain underexamined. For example, cost-effectiveness has been estimated only for triangular titanium implants.84 Opioid consumption is reported in a minority of studies (not reviewed further here) with variable results (some showing reduced opioid use25 and others showing no reduction49).
Conclusions
Substantial literature supports the safety and effectiveness of minimally invasive SI joint fusion. The evidence base for the LTI procedure is largest and of the highest quality. However, only one implant (iFuse) has level 1 evidence and independent radiographic assessment of joint fusion. Studies reporting outcomes with the LTI procedure show larger improvements in pain and disability compared with those on PLTI or PI procedures. All procedures appeared to have low rates of safety outcomes; direct comparisons are not available; such studies are potentially limited by event rarity. Outstanding issues for PLTI and PI procedures include implant placement accuracy, fusion rates, and impact on opioid use. No information is available for devices placed posteriorly that span the joint.
Footnotes
Funding The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests Three authors are employees of a device manufacturer whose implant is included in the systematic review, 5 authors are consultants for various implant manufacturers included in this paper.
Disclosures P.W. reports grants/contracts from Bioventus and OssDsign; consulting fees from SI-BONE, Bioventus, Boston Scientific, Medtronic, Orthofix, and Stryker; speakers bureau for SI-BONE, Medtronic, Orthofix, and Stryker; Advisory Boards for Centinel Spine, Cerapedics, Empirical Spine, NuVasive, Providence Medical Technologies, and OssDsign. V.P. reports consulting fees from DePuy Synthes, Mainstay Medical, Zimmer, Stryker, Medtronic, Aesculap, Spine Welding AG, Orthobond, J&J, Baxter, and SI-BONE, Inc. B.D. reports consulting from SI-BONE, Inc. and Intrinsic Therapeutics. B.S. reports consulting from SI-BONE and SRL. DC, CR, and RC are employees of SI-BONE, Inc. D.P. reports consulting from Globus, SI-BONE, and Alexion and royalties from SI-BONE.
- This manuscript is generously published free of charge by ISASS, the International Society for the Advancement of Spine Surgery. Copyright © 2023 ISASS. To see more or order reprints or permissions, see http://ijssurgery.com.