Preoperative three-dimensional computer planning for reverse total shoulder arthroplasty and bone grafting for severe glenoid deformity

Abstract

Background

Computer assisted planning without patient specific instrumentation may be utilized to guide reverse total shoulder arthroplasty baseplate placement. The purpose of this study was to determine the difference between planned and achieved inclination and retroversion correction with three-dimensional preoperative computer assisted planning in reverse total shoulder arthroplasty without patient specific instrumentation with bone grafting for severe glenoid erosion.

Methods

Preoperative three-dimensional computer assisted planning without patient specific instrumentation was performed on 15 patients undergoing primary reverse total shoulder arthroplasty with glenoid bone grafting for severe glenoid erosion. On preoperative and immediate postoperative computed tomography slices, two-dimensional retroversion and inclination were measured. Preoperative three-dimensional baseline retroversion and inclination and planned postoperative three-dimensional retroversion and inclination were measured. Planned and achieved version and inclination changes were compared.

Results

The planned and achieved retroversion corrections were 18° and 12°, respectively (p < 0.001). The planned and achieved inclination corrections were 11° and 11°, respectively (p = 0.803).

Conclusions

Three-dimensional computer assisted planning without patient specific instrumentation in the setting of reverse total shoulder arthroplasty with severe glenoid erosion requiring bone grafting can accurately guide baseplate placement. All cases in which failure to correct retroversion or inclination within 10° of planning occurred in patients with severe erosion (B3 or E3 glenoids), therefore patient specific guides may be warranted in these cases to improve accuracy of implantation.

Level of evidence

Level IV, retrospective case series.

Introduction

Reverse total shoulder arthroplasty (RTSA) has become the standard treatment for patients with rotator cuff arthropathy with pseudoparalysis. With the evolution of surgical techniques, increasingly complex bone deformity is now being addressed with RTSA and glenoid bone grafting in a single stage that previously was either avoided or treated with multiple stages.^1–3 Baseplate and glenosphere position is critical for optimal range of motion and function as well as avoidance of complications including instability and scapular notching.^3–5 Computer assisted planning and patient specific instrumentation (PSI) are tools that have been developed to improve the accuracy of implant positioning during shoulder arthroplasty. ⁶ With the treatment of increasingly complex deformity, utilization of these tools may provide an opportunity in avoiding malposition that would otherwise potentially not occur with simpler pathology.

Computer assisted planning has been evaluated in improving accuracy of glenoid or baseplate component positioning in anatomic and RTSA. In anatomic arthroplasty using standard instruments, computer assisted 3D planning alone has been shown to improve the accuracy of glenoid component placement with regards to version and inclination both in vitro and in vivo in comparison to 2D planning.^4,7 Computer assisted planning and PSI have been investigated during RTSA, both in vitro and in vivo, and have been shown to be accurate and reliable and to reduce variability in baseplate positioning.^8,9 No clinical studies have evaluated the benefit of computer assisted planning using 3D images without PSI compared to 2D planning. This method potentially gains the benefits of preoperative planning in 3D while avoiding the cost associated with PSI. As bone grafting in the setting of RTSA becomes more commonplace for the treatment of severe deformity, identifying the utility of computer assisted 3D planning alone will have increasing importance.

The purpose of this study was to determine the difference between planned and achieved inclination correction, retroversion correction, and bone graft thickness with 3D preoperative computer assisted planning without PSI in RTSA with bone grafting for severe glenoid erosion. We hypothesize that computer assisted 3D planning alone will allow accurate recreation of planned changes in version, inclination, and bone graft thickness.

Materials and methods

Between 6/2016 and 12/2018, 36 patients underwent RTSA with glenoid bone grafting for severe deformity at the University of Utah by one shoulder surgeon (RZT). The surgeon made the decision to perform an RTSA with a structural graft based on the ability to correct baseplate inclination to at least neutral tilt on a standing true anteroposterior radiograph of the shoulder and to within 10° of neutral version on an axillary radiograph without significantly reaming beyond 5 mm of glenoid bone stock to gain correction. The goal of reaming was to correct to 100% baseplate seating and to restore the joint line back to the native joint line. If these goals could not be achieved with reaming alone, then RTSA with structural bone grafting was selected. Bone grafting was not performed if only lateralization of the joint line was desired and if reaming could achieve correction to the limits previously mentioned.

Of these 36 patients, only patients who underwent primary RTSA with a concomitant structural glenoid bone grafting that also underwent preoperative computer assisted 3D planning were considered for inclusion in the current study. Computer assisted 3D planning was performed if the primary surgeon believed severe correction of alignment (version, inclination) with glenoid bone grafting would be required. We excluded patients who underwent RTSA with a structural glenoid bone graft as a revision of a prior arthroplasty as computer planning was not possible on those cases. We also excluded those patients who underwent primary RTSA with a concomitant structural glenoid bone graft where preoperative 3D planning was not performed due to inadequate computed tomography (CT) images. No patients during the time frame of the study underwent bone grafting during primary RTSA that had adequate preoperative CT scans and did not have preoperative computer planning of the graft. A total of 18 patients met these inclusion criteria. Of the included patients, three patients did not have an immediate postoperative CT. The remaining 15 patients underwent immediate postoperative CT scans of the shoulder to assess baseplate and graft position and is the final group of patients evaluated in the current study. Investigational Review Board approval at the University of Utah was obtained to perform a retrospective analysis of radiographic images from the included patients.

Operative procedure

The operative procedures were performed as previously described by Tashjian et al. ³ and Boileau et al. ¹⁰ The Aequalis Reversed Shoulder Arthroplasty (Wright Medical, Bloomington, MN, USA) system was utilized in all cases. A long post (25 mm) baseplate was utilized in all cases engaging at least 5 mm of native glenoid with the central post. Two methods of glenoid bone grafting were performed—structural grafting and BIO-RSA. Structural grafting utilizing a contoured graft slid into the glenoid defect with preparation of the baseplate with the graft in place was performed on six patients as previously described by Tashjian et al. ³ (Figures 1 and 2). A shaped BIO-RSA technique, as described by Boileau et al., ¹⁰ where a graft is harvested, shaped, and placed onto the central post of the baseplate prior to implantation was utilized in nine cases. Both femoral head allograft and humeral head autograft were utilized dependent on bone quality based upon surgeon discretion. Three femoral head structural allografts, two femoral head BIO-RSA allografts, three humeral head structural autografts, and seven humeral head BIO-RSA autografts were performed.

Figure 1.

Grashey anteroposterior radiograph of the shoulder with rotator cuff arthropathy with severe glenoid erosion (E3).

Figure 2.

Postoperative Grashey anteroposterior radiograph of the shoulder after structural glenoid grafting during reverse shoulder arthroplasty utilizing humeral head autograft.

Radiographic analysis

Preoperative CT scans and radiographs (AP, Grashey, Axillary Lateral) were performed on all patients. All patients underwent preoperative 3D templating using the Glenosys software system (Imascap, Brest, France). CT scans were converted to 3D reconstructions through automatic segmentation. All patients included in the study had severe enough erosion to require bone grafting to restore version and inclination and not just for pure lateralization of the joint line. Planning was performed by the same surgeon performing all the operations (RZT) within a week before surgery. The goals of planning were to restore inclination between 0° and 5° of superior inclination and version between 5° and 10° of neutral version. We attempt to restore the joint line to the native joint line without lateralization or medialization based upon the remaining anteroinferior glenoid rim. Auto-calculated 3D measurements for preoperative inclination and version were recorded from the preoperative plans using the Glenosys software. The 3D measurements for the templated RTSA including planned implant inclination, version, and maximal graft thickness were recorded from the preoperative plans using the Glenosys software (Figure 3). Axial and coronal screen shots were taken of the planned axis from the 3D plan for intraoperative use during axis guide pin placement (Figures 4 and 5). Similarly, an enface sagittal view screen shot of the glenoid was taken identifying the location of planned guide pin placement to be utilized intraoperative (Figure 6). Intraoperatively, the 3D plan was visualized on a computer during the surgery and the central guide pin alignment as well as graft shape and size was attempted to be replicated without any use of PSI. Correction of inclination and version during surgery was based upon the guide pin placement for the central post, not necessarily the cut dimensions of the grafts.

Figure 3.

The 3D measurements of preoperative glenoid version and inclination as well as templated 3D position of baseplate version, inclination, and maximal graft thickness utilizing Glenosys.

Figure 4.

Coronal 3D planned view showing trajectory of central guide pin.

Figure 5.

Axial 3D planned view showing trajectory of central guide pin.

Figure 6.

Enface sagittal view showing 3D location of guide pin placement on glenoid face.

Immediate postoperative CT scans were performed on postoperative day 1 to assess graft position, thickness, and implant alignment.

All patients underwent retrospective radiographic review by a separate shoulder and elbow fellowship trained surgeon (PNC) who was not the primary surgeon who did the planning or the surgical procedures. All preoperative CT scans were exported to the digital imaging and communication (DICOM) format and uploaded to a third-party DICOM viewer for analysis (OsiriX, Pixmeo, Geneva, Switzerland). CT scans were first reoriented into the plane of the scapula as previously described. ¹¹ Measurements of glenoid version and inclination were performed on the reformatted 2D axial and coronal images as previously described in a method validated for accuracy and reliability (Figures 7 and 8). ¹² Inclination was measured as the angle between the glenoid and the scapular spine on the coronal CT slice at the deepest point of the supraspinatus fossa as measured on the most lateral sagittal slice to contain the scapula spine. Retroversion was measured as the angle between a line from the anterior to posterior glenoid rim and a line extending from the medial aspect of the scapula to the middle of the glenoid on the axial slice midway between the most proximal slice containing the glenoid and the most distal slice containing the glenoid (scapula body centerline). Glenoid deficiency was classified on preoperative radiographs and CT scans using the Walch classification for patients with glenohumeral osteoarthritis and the Favard classification for patients with rotator cuff arthropathy.^13,14 Postoperative CT scans were exported to the DICOM format, uploaded into Osirix viewer, reformatted into the plane of the scapula, and analyzed as described above. The 2D retroversion and inclination of the postoperative baseplate was recorded from these reformatted images as well as the maximal thickness of the glenoid bone graft (Figures 9 and 10). The 3D measurements were not performed postoperatively because the Glenosys software is not able to process scans with implants in place.

Figure 7.

Retroversion measured on reformatted 2D axial CT scan image in the plane of the scapula.

Figure 8.

Inclination measures on reformatted 2D coronal CT scan image in the plane of the scapula.

Figure 9.

The 2D postoperative retroversion of the baseplate measured on axial 2D image reformatted in the plane of the scapula.

Figure 10.

The 2D postoperative inclination of the baseplate measured on coronal 2D image reformatted in the plane of the scapula.

Statistical analysis

Descriptive statistics were calculated for preoperative and postoperative variables, and the Kolmogorov–Smirnov test was used to assess data normality. Planned changes in retroversion and inclination were calculated comparing preoperative 3D measured and 3D planned variables. Achieved changes in retroversion and inclination were calculated comparing preoperative 2D measured and postoperative 2D measured variables. Preoperative planned graft thickness was measured on 3D planning and postoperative achieved graft thickness was measured on postoperative 2D imaging. Planned versus achieved change in inclination and version and planned versus achieved maximal graft thickness were compared with paired Student’s t-tests as all data were normally distributed. P-values of < .05 were considered statistically significant. Because this is an uncommon procedure examined with a retrospective design, no power analysis was conducted, and all available subjects were included.

Results

Of the 18 patients who met inclusion criteria, 15 underwent postoperative CT scanning (83% follow-up). The mean age at the time of surgery was 75 years (range, 61–92 years). There were eight female and seven male shoulders. The preoperative diagnosis was primary glenohumeral osteoarthritis in 11 shoulders and rotator cuff arthropathy in 4 shoulders. Glenoid deformities were classified as 1—E1, 1—E2, 2—E3, 2—A2, 1—B2, and 8—B3.

Preoperative 2D retroversion and inclination averaged 22° ± 13° (range, 0°–41°) and 8° ± 13° (range, −12° to 34°), respectively. Preoperative 3D retroversion and inclination averaged 26° ± 12° (range, 4°–49°) and 11° ± 14° (range, −9° to 44°), respectively (Table 1). The 3D planned baseplate orientation was 8° ± 2° of retroversion and 2° ± 2° of inclination on average. Average preoperative planned graft thickness was 11 ± 2 mm (Table 2).

Table 1.

Demographics and preoperative superior inclination and retroversion measures.

Patient	Age	Gender	Side	Diagnosis	Glenoid deformity type	Preop 2D sup inclin (°)	Preop 2D retro (°)	Preop 3D sup inclin (°)	Preop 3D retro (°)
1	82	F	R	RCA	E2	34	38	44	49
2	83	M	R	RCA	E3	27	0	23	4
3	61	M	R	OA	B3	−4	28	−4	34
4	70	F	R	OA	B3	10	28	6	31
5	78	F	L	RCA	E3	20	10	19	13
6	79	F	R	OA	B3	15	20	13	23
7	70	F	L	OA	B2	0	21	1	25
8	73	M	R	OA	B3	−12	41	−9	40
9	73	M	L	OA	B3	−6	40	−2	40
10	55	F	R	OA	B3	14	24	19	23
11	70	M	R	RCA	E1	18	14	15	21
12	86	F	R	OA	A2	−5	0	−6	7
13	74	M	L	OA	B3	1.7	28	17	30
14	92	F	L	OA	A2	5	14	4	18
15	86	M	L	OA	B3	−1	29	17	27

OA: osteoarthritis; RCA: rotator cuff arthropathy.

Table 2.

Graft type and preoperative 3D planned and postoperative 2D achieved superior inclination, retroversion and graft thickness.

Patient	Planned 3D sup inclin (°)	Planned 3D retro (°)	Planned 3D graft thickness (mm)	Graft type	Postop 2D sup inclin (°)	Postop 2D retro (°)	Postop 2D graft thickness (mm)
1	5	10	13.1	Structural	0	0	14
2	5	4	9	Structural	−10	0	14
3	0	9	15.4	BIORSA	6.5	13	11
4	3	9	11.5	BIORSA	1	18	10
5	0	9	11.6	Structural	17	6	10
6	0	5	12.1	Structural	−7	6	12
7	0	7	11.3	BIORSA	14	3	6
8	2	10	11.2	Structural	−5	17	12
9	0	7	14.9	Structural	−13	22	11
10	5	6	7.9	BIORSA	11	20	7
11	5	10	9.2	BIORSA	15	18	8
12	0	9	10.2	BIORSA	8	0	7
13	3	7	11.3	BIORSA	3	12	6
14	2	9	9.4	BIORSA	4	16	12
15	4	10	11.3	BIORSA	13	23	7

Postoperative average implant retroversion and inclination averaged 12° ± 8° (range, 0°–23°) and 4° ± 9° (range, − 10° to 17°), respectively. The average postoperative graft thickness was 10 ± 3 mm (Table 2).

The planned and achieved retroversion corrections were 18° ± 11° and 12° ± 11°, respectively (mean ± standard deviation, p < 0.001). The planned and achieved inclination corrections were 11° ± 10° and 11° ± 12°, respectively (p = 0.803). The difference between the planned maximal bone graft width and final graft width was 2.7 ± 2 mm (p = 0.075) (Table 3).

Table 3.

Difference in planned 3D and final 2D graft thickness, achieved 2D superior inclination and retroversion changes from preoperative to postoperative and planned 3D superior inclination and retroversion changes.

Patient	Difference from planned 3D to final 2D graft thickness (mm)	Achieved 2D superior inclination change from preop to postop (°)	Achieved 2D retroversion change from preop to postop (°)	Planned 3D superior inclination change (°)	Planned 3D retroversion change (°)
1	0.9	34	38	39	39
2	5	37	0	18	0
3	4.4	10.5	15	4	25
4	1.5	9	10	3	22
5	1.6	3	4	19	4
6	0.1	22	14	13	18
7	5.3	14	18	1	18
8	0.8	7	24	11	30
9	3.9	7	18	2	33
10	0.9	3	4	14	17
11	1.2	3	4	10	11
12	3.2	1	0	6	2
13	5.3	1.3	16	14	23
14	2.6	1	2	2	9
15	4.3	14	6	13	17

Discussion

RTSA baseplate placement in the setting of glenoid erosion can be challenging. Restoration of glenoid retroversion and inclination is critical to optimize initial baseplate stability and range of motion and limit complications such as scapular notching and instability. The current study demonstrates that 3D computer assisted templating without PSI can result in inclination corrections within 1° of planned and retroversion corrections within 6° of planned, with a significant under correction in retroversion. Although this difference in retroversion is statistically significant, in the authors’ opinion, a 6° deviation is clinically insignificant. Thus, in the setting of up to 44° of superior inclination and 49° of retroversion, PSI may not be required to appropriately place the baseplate when bone grafting of the glenoid is required. More severe deformity may require a combination of PSI and computer assisted planning to accurately place implants.

The 3D computer assisted planning alone has been evaluated for anatomic shoulder arthroplasty both in vivo and in vitro.^4,7 Iannotti et al. ⁷ performed a study evaluating anatomic glenoid guide pin placement in bone models using standard instrumentation alone, standard instrumentation and preoperative planning, and preoperative planning plus PSI. They determined that the use of standard instruments and 3D planning improved guide pin positioning compared with standard instruments with 2D planning. ⁷ Accuracy of pin positioning increased by 4.5° in version and 3.3° in inclination. They also determined the use of PSI further increased accuracy of pin positioning compared to planning alone. The same authors performed a clinical study evaluating the same three groups. ⁷ They determined that 3D planning alone with standard surgical instrumentation improved the ability to achieve the desired implant position to within 5° of inclination and 10° of version compared to 2D planning with standard instruments. PSI did not significantly improve the ability to achieve the planned goals in this series beyond that obtained with 3D planning and standard instruments alone. These data support the importance of 3D planning alone and parallel our results in patients undergoing RTSA.

In the setting of RTSA, 3D computer planning using standard instrumentation without PSI has not been clinically evaluated. Verborgt et al. ¹⁵ determined that the mean deviation of baseplate version and inclination from the planned version and inclination was on average 4°–5° in a series of patients who underwent RTSA with 3D planning and intraoperative PSI. Dallalana et al. ¹⁶ also reported deviations of baseplate inclination and version from the planned inclination and version using patient specific guides on average of 1° and 2°. These deviations are very similar to our own without the use of PSI. Most of the cases were E0 or E1 glenoids and none had concomitant glenoid bone grafting. Heylen et al. ⁸ performed a clinical study utilizing PSI in the setting of RTSA in 24 patients (12 with PSI and 12 without PSI) and found that PSI prevented extreme values of glenoid component inclination. No study has evaluated the accuracy of 3D planning alone using standard instruments without PSI in the setting of RTSA. Our results are comparable to those of Verborgt et al. ¹⁵ who utilized PSI and planning where they were able to get within 5° of planning and we were able to get within 6° of planning on average. Although larger sample sizes and confirmation of our results by other centers will be necessary to make firm conclusion, based upon our results 3D planning alone is reasonable to accurately place the baseplate even in the setting of severe erosions in the coronal and axial planes requiring bone grafting.

Graft thickness was also estimated from the plan and, in general, the actual graft was several millimeters thinner than the planned graft. Planning graft thickness becomes important when determining the method utilized for bone grafting as well as making a decision if grafting is feasible. The shaped BIO-RSA technique, as described by Boileau et al., ¹⁰ requires the baseplate to rest completely on graft where the structural grafting technique we have previously published on allows the baseplate to rest upon some native glenoid bone stock. ³ Both techniques are reasonable for grafting but in general my preference is for grafts larger than 10 mm in thickness to utilize a structural technique and for graft sizes 10 mm or less in thickness to use either the shaped BIO-RSA or the structural graft technique. The concern is that baseplates completely supported by large grafts alone may be at higher risk for failure if there is some graft resorption or failure of incorporation as opposed to grafts supported by some native glenoid. In general, defects requiring a graft over 20 mm in thickness are concerning for grafting using any technique and we have typically utilized a custom metal augmented baseplate in these cases. The current study supports that planning can estimate the correct thickness of final graft size required within 2–3 mm allowing appropriate planning for the method of glenoid defect reconstruction.

On average overall, actual inclination correction was very close to planned inclination correction. If we look at this comparison, the range between actual and planned correction is from 1° to 19°. In general, overcorrection or more inferior tilt was implemented in comparison to the plan understanding the concern for leaving the baseplate with superior tilt. As a result, no implants were placed in greater than 5° of superior tilt on final postoperative CT scans which is correlated with low risk for instability and is associated with a neutrally aligned baseplate on a standing true anteroposterior radiograph as we have previously published.^17,18 The cases in which failure correct to superior tilt within 10° of planning occurred were in two B3 and two E3 deformities, so cases with severe erosion. These cases may warrant the use of a patient specific guide as they had largest differences in planned versus achieved correction of inclination. In terms of retroversion, we almost always under-corrected the amount of version we were attempting to correct based upon our preoperative planned correction. The reason for this is because inclination correction is much easier to visualize intraoperatively as opposed to version. All cases of failure to correct retroversion within 10° of planning occurred in B3 glenoids. Corrections in B2, E2, and A2 glenoids were very close to the planned corrections for both inclination and version. The data suggest patient specific guides might be useful for E3 and B3 deformities whereas their utility is more limited in A2, B2, and E2 deformities.

Limitations of the current study include small sample size and the requirement for 2D postoperative measurement of baseplate position. Because we evaluated only cases of severe erosion requiring significant correction requiring bone grafting, our sample size was limited. Nevertheless, these cases would be considered the “worst case” scenarios and therefore give surgeons information about the utility in cases where a guidance might be of most utility. The small sample size limits the ability to power insignificant findings although the differences in planned and achieved inclination and graft thickness are so small that the differences are not clinically important. Another limitation of the study was that we compared 3D planning versus 2D actual measurements. Glenosys does not have the ability to calculate 3D measurements in the postoperative setting with an implant in placed limiting the ability to determine 3D postoperative implant positioning. We used 3D planning measures because this reflects what we currently do in our practice to plan for a surgical case and the derived values are easier to calculate. In addition, our study does not include control groups of patients without planning or with planning and PSI.

Conclusions

The 3D computer assisted planning alone without PSI in the setting of RTSA with severe glenoid erosion (up to 44° of superior inclination and 49° of retroversion) requiring bone grafting can guide implant placement on average to within 1° of planned inclination changes and 6° of planned version changes. All cases in which failure to correct retroversion or inclination within 10° of planning occurred in patients with severe erosion (B3 or E3 glenoids), therefore patient specific guides may be warranted in these cases to improve accuracy of implantation.