Accuracy of Patient-Specific Instrument Resections In Vivo in Total Ankle Arthroplasty on Postoperative Weightbearing CT Scan

Abstract

Background:

Total ankle arthroplasty (TAA) using patient-specific instrumentation (PSI) has increased in popularity with evidence for decreased operative duration, fluoroscopy usage, and increased implant placement accuracy. To date, no comparisons have verified the accuracy of PSI in vivo using preoperative and postoperative weightbearing computed tomography (WBCT). Our study aims to supplement the literature by quantifying the accuracy and precision of PSI-guided implant positioning using WBCT. The primary study outcome is to identify positioning deviations in degrees or millimeters in any plane for talus and tibial PSI-guided resections and subsequent implant placement. Secondary outcomes were correlation and regression analyses performed to identify variables that correlate to resection or implant placement deviation, as well as comparison to existing radiography-based PSI techniques.

Methods:

This was a single-surgeon, single-implant consecutive patient retrospective study where preoperative and postoperative WBCTs were obtained. TAA was performed by the senior author with the same low-profile implant for all cases. Talus and tibial resection analysis was performed in computer-automated fashion on postoperative segmented WBCTs and is described as 3 possible deviations from plan (cut height in millimeters, sagittal slope, varus/valgus deviation). Overall deformity in this group was not severe. Implant placement factors independent of PSI guides are described as center-of-mass translations (mm) and axial rotation (degrees). Desired accuracy for the PSI system was within ±2 mm or ±2 degrees of the preoperative plan. Statistical analysis of data collected included Student t test and linear regression analyses.

Results:

Thirty patients were included. Four talus implants were excluded per intraoperative surgeon discretion (deviation from PSI plan, use of conventional instruments). All postoperative tibial measurements were within the desired limits, except for mediolateral center of mass positioning (P = .003) and slope (P = .013). Two of six talar parameters also diverged from preoperative measurements: axial rotation (P = .015) and anteroposterior positioning (P = .002). In addition, no correlations exceeding r = 0.5 were noted between preoperative deformity measurements and postoperative positioning.

Conclusion:

For the 26 ankles that did not require an intraoperative deviation from PSI plan and/or use of conventional instruments, total ankle replacement performed with PSI using WBCT results in component placement with relatively little deviation from the preoperative plan. In addition, the lack of correlation between preoperative deformity and implant placement suggests that the magnitude of preoperative deformity in this group was not associated with the accuracy of PSI-guided component positioning.

Level of Evidence:

Level III, retrospective cohort study.

Introduction

Ankle arthritis treatment options have progressed throughout the years, with more recent developments focusing on furthering the advancement of total ankle arthroplasty (TAA). ⁶ The goals of arthroplasty are to provide the patient with a stable weightbearing construct that alleviates pain while preserving ankle range of motion.^6,15 With improvements in surgical technique and preoperative planning, TAA has become an increasingly popular and viable treatment option for ankle arthritis similar to that of other major joints of the lower extremity. However, implant survivability is influenced by several factors; specifically, component misalignment has been identified as a cause of premature implant failure leading to necessary revision surgery.^8,9,11 This underscores a need for reliable intraoperative alignment methods.

Thus, technological advances such as the use of patient-specific instrumentation (PSI) has soared in popularity. PSI for TAA has potential advantages including minimizing bone loss stock for potential revision surgeries along with decreasing hospital cost, fluoroscopy usage, and operative time. ¹⁸ There is dissension in the literature regarding implant placement accuracy improvements with PSI; although PSI was demonstrated to exhibit increased implant placement accuracy on postoperative radiographs in several studies,^7,12,14,20 another study investigating this subject found no difference between the 2 surgical methods. ¹⁸

However, radiography is notably inferior to weightbearing computed tomography (WBCT) in quantifying the accuracy of total knee arthroplasty (TKA) implant placement. ¹⁹ To date, no comparisons have verified the accuracy of PSI for TAA in vivo using preoperative and postoperative WBCT. We hypothesized that the final component positioning of both tibial and talar implants would be within ±2 mm (for center of mass measurements) or ±2 degrees of angulation in coronal, sagittal, and axial planes. Our hypothesis was based on acceptable component placement accuracy as detailed in the literature; although acceptable accuracy for ankle implants has not yet been well defined, several studies use a boundary of ±5 degrees of deformity and consider it adequate.^1,17 The knee arthroplasty world is more stringent, with ±3 degrees generally regarded as acceptable. ⁵ Because of the customized nature of PSI, we posited that the PSI system would be capable of greater accuracy than that previously detailed in the literature. Our study aims to quantify the accuracy and precision of PSI between the preoperative template and postoperative implant location using WBCT. The primary outcome of this study will identify angular deviation over 2 degrees in any plane for talus and tibial resections and thus quantify the accuracy and precision of PSI using WBCT. Secondary outcomes will identify implant placement deviation or translation over 2 mm in any plan from the preoperative plan and if preoperative deformities correlate to postoperative implant placement outcomes.

Materials and Methods

This was a single-surgeon, single-implant consecutive patient retrospective study where preoperative and postoperative WBCTs were obtained following institutional review board approval. All subjects enrolled were screened preoperatively and indicated for TAA. All patients were diagnosed with end-stage ankle arthritis and had failed conservative therapy. The 2 primary etiologies included primary/idiopathic ankle osteoarthritis and posttraumatic ankle osteoarthritis. Each patient indicated for TAA received a preoperative CT scan, which included the full tibia and the foot/ankle, per standardized protocol. Preoperative planning consisted of individualized meetings with the engineering team and the primary surgeon. PSI was generated from measurements taken from the preoperative CT scans, which concerned the existing tibial deformity in all 3 anatomic planes (Figure 1). All patients were treated with the same fixed-bearing low-profile implant, the Vantage Total Ankle System (Exactech, Gainesville, FL).

Figure 1.

Preoperative imaging. Shown in this figure is an example of the preoperative CT scans (left) that are used to generate a 3D rendering of the ankle joint (middle) that assists the surgeon in the PSI planning process. Osteophytes are visualized in green on the 3D rendering. Preoperative coronal radiograph is shown on the right, demonstrating a collapse of the tibial plafond onto the talar dome. PSI, patient-specific instrumentation.

All TAAs included in this study were performed consecutively between September 2021 and February 2024. For inclusion, a patient must have had a TAA using the Exactech PSI software to preoperatively plan implant placements. Notably, the used guides were “coupled,” indicating the tibial resection guide was fixed to the tibia and the talus resection guide was linked to the initial tibial guide’s positioning. If an implant was secured without usage of PSI guides, that resulted in automatic exclusion from the study; 4 talus implants were excluded in this manner. Follow-up visits were conducted in-person as well as virtually.

Postoperatively, each patient was made nonweightbearing until surgical wounds were adequately healed, which typically occurred in 2-4 weeks. At the time of adequate wound healing, they progressed to full weightbearing in a controlled ankle motion (CAM) walker and were then referred for physical therapy. Patients were unrestricted at 8 weeks postoperation to progress to full activity with physiotherapy. Serial radiography is obtained at 4 weeks, 6 months, and annually thereafter.

At the 3-month postoperative mark, patients received a second weightbearing CT scan with the same manufacturer PSI protocol. Twelve total postoperative measurements were obtained, 6 for the tibial component and 6 for the talar component (Figure 2). Translation in all 3 coordinate directions was measured for both implants along with rotation about each axis. Postoperative measurements are listed and defined in Supplemental Table 1.

Figure 2.

Directionality scheme for postoperative implant measurements. This figure represents the directionality scheme referred to when discussing postoperative measurements of both the tibial and talar implants. These measurements are shown on a tibial implant but are valid for the talar implant as well. Both preoperative (gray) and postoperative (red) implants are shown superimposed.

3D Systems Postoperative Implant Analysis Method

Postoperative implant placement analysis of WBCT images is completed at 3D Systems by using the same “Geomagic FreeForm” design file that was used to design the tibia and talus PSI for the specific patient case. The file contains imports of the planned patient ankle anatomy and the planned implant placements that were established during surgical planning. The postoperative tibia and talus implants are compared to the planned tibia and talus implant placements to identify translational (x, y, and z directions) and rotational (anterior-posterior slope/sagittal angle, varus-valgus roll/coronal angle, and internal-external twist/transverse angle) deviations from plan.

The first step within the analysis is to center the local origin of the preoperatively planned tibia (or talus) implant, and then manually zero the rotational position values associated with it (Figure 3, Step 1). After this, the planned implant and preoperatively planned anatomy are moved together in the software so that the planned implant’s local origin aligns with FreeForm’s global origin, zeroing the translational position values associated with the planned implant (Figure 3, Step 2). Once the rotational and translational values are zeroed, the file containing the details of the postoperative implant’s location relative to the resected postoperative anatomy are uploaded to FreeForm. Then, an alignment function in the software is used to superimpose the postoperative anatomy and implant location with the preoperative anatomy and implant location (Figure 3, Step 3). Once the preoperative and postoperative anatomies and implants are superimposed, measurements of relative implant locations and resection dimensions can then be made in the software (Figure 3, Step 4).

Figure 3.

3D systems’ methodology for implant positional analysis. This is a pictorial depiction of the process used by 3D Systems to analyze positioning of both the tibial and talar implants. Step 1 depicts the preoperatively planned anatomy with the implant in position. Step 2 shows the result of zeroing the axes in the software so the resection analysis can proceed. Once the axes are zeroed, the postoperative scans (tan) can be imported and superimposed on the preoperative anatomy (gray) using a best-fit alignment function as shown in Step 3. Finally, the anatomy can be turned off, leaving only the relative positions of the implants shown, after which measurements regarding the implant’s postoperative position can be made (Step 4).

Statistical Analysis

Data were analyzed using a Student t test assuming unequal variances to determine significant differences between each postoperative measurement and their maximum desired deviations. Statistical significance for this test was set at α = .05. P values were calculated for deviations of ±2 mm (degrees) for all 12 measurements to assess the primary study hypothesis. The secondary study outcome was analyzed via a correlation calculation relating all variables, both preoperative and postoperative, to each other. Variables having a correlation coefficient greater than |r| = 0.5 were considered to demonstrate statistical effects on one another. A post hoc analysis was used to assess study power.

Results

Preoperative and postoperative measurements were obtained and recorded for 30 patients. All 30 patients met inclusion criteria for the tibial implant; however, 4 talar implants were excluded because of intraoperative deviations from the plan at the discretion of the surgeon (Figure 4). Two of these talar implants were excluded because of usage of a different implant than what was planned preoperatively; that is, a flat top implant was used in lieu of a curved top. The remaining 2 talar implants were excluded because of usage of conventional instrumentation to increase the resections as a result of a tight joint space. Mean, SD, and range were calculated for the 4 preoperative parameters and the 12 total (6 tibial, 6 talar) postoperative parameters. The data are represented briefly in similar fashion in Tables 1 and 2.

Figure 4.

Flow diagram for exclusions. Pictorial representation of the exclusions made in this study; 4 talar implants were excluded because of intraoperative usage of conventional instrumentation at the surgeon’s discretion.

Table 1.

Tibial Deviation Data (n = 30).

	Roll (degrees) (Varus Positive) a	Slope (degrees) (Anterior Positive) a	Axial Rotation (degrees) (Internal Rotation Positive)	x Axis Translation (mm) (Medial Positive)	y Axis Translation (mm) (Anterior Positive)	z Axis Translation (mm) (Superior Positive) a
Mean	0.9	−2.3	0.2	−1.3	0.1	0.7
SD	2.6	2.9	3.7	1.1	1.0	1.3
Minimum	−4.4	−6.7	−7.5	−3.4	−1.9	−1.9
Maximum	6.0	5.5	6.9	0.6	2.5	3.2
Patients outside ± 2 mm (degrees), n/n (%)	15/30 (50)	14/30 (46.7)	20/30 (66.7)	9/30 (30)	1/30 (3.3)	6/30 (20)
P value	.150	.013*	.796	.003*	.737	.140

Abbreviation: PSI, patient-specific instrumentation.

^aPSI controlled variable.

^*Statistically significant.

Table 2.

Talus Deviation Data (n = 26).

	Roll (degrees) (Varus Positive) a	Slope (degrees) (Anterior Positive)	Axial Rotation (degrees) (Internal Rotation Positive)	x Axis Translation (mm) (Medial Positive)	y Axis Translation (mm) (Anterior Positive)	z Axis Translation (mm) (Superior Positive) a
Mean	−0.9	−3.9	2.5	1.0	−2.1	0.9
SD	3.0	5.8	4.4	1.6	2.5	2.0
Minimum	−6.5	−14.3	−7.4	−2.0	−5.8	−2.2
Maximum	5.6	9.3	11.9	3.4	3.2	5.9
Patients outside ± 2 mm (degrees), n/n (%)	14/26 (53.8)	19/26 (73.1)	17/26 (65.4)	10/26 (38.5)	18/26 (69.2)	5/26 (19.2)
P value	.205	.950	.015*	.062	.002*	.101

Abbreviation: PSI, patient-specific instrumentation.

^aPSI controlled variable.

^*Statistically significant.

The average age of those included in the study was 64.6 years (range, 42-82) with an average BMI of 31.1 (range, 20.3-41.6) (Table 3). In addition, of the 30 patients included, 14 were male and 16 were female. Mean time to postoperative CT in this study was 4.2 months (range, 0.92-16.23).

Table 3.

Demographics and Preoperative Deformity Data.

Metric	Full PSI Guidance a , Mean (Range) (n = 26)	No Talar PSI Guidance b , Mean (Range) (n = 4)
Age, y	64.7 (42 to 82)	63.5 (58 to 68)
BMI	30.8 (20.3 to 41.6)	33.0 (30.3 to 38.3)
Preoperative coronal deformity (degrees, varus positive)	−0.7 (−18.7 to 15)	−8.5 (−15.1 to 2.8)

Abbreviations: BMI, body mass index; PSI, patient-specific instrumentation.

^an = 9 patients had preoperative deformities greater than 10 degrees varus or valgus in this group.

^bn = 2 patients had preoperative deformities greater than 10 degrees varus or valgus in this group.

Results from the t test indicated that PSI for the tibial implant showed 4 of 6 parameters within the ±2 criteria; the 2 dimensions with significant deviation, compared to the template, were mediolateral implant translation and tibial implant slope. The tibial implant was observed 1.3 ± 1.1 mm (P = .033) lateral compared to the template (range, 3.4 mm lateral to 0.6 mm medial); likewise, the tibial slope was 2.3 ± 2.9 degrees (P = .034) dorsiflexed (range, 6.7 degrees dorsiflexed to 5.5 degrees plantarflexed). All other tibial implant measurements did not significantly deviate from preoperative plan (P > .05) (Tables 1 and 2).

Regarding the talus implant, similar accuracy was observed, with 4 of 6 measurements demonstrating no significant difference between the preoperative template and the postoperative implant placement. The 2 dimensions with significant deviation were talar implant anteroposterior position (2.1 ± 2.5 mm posteriorly, P = .01, range, 5.8 mm posterior to 3.2 mm anterior) as well as axial rotation (2.5 ± 4.4 degrees internally rotated, P = .025, range, 7.4 degrees external rotation to 11.9 degrees internal rotation).

The strongest correlation observed for any of the measured postoperative variables was between talar slope and the posterior translation of the talar implant center of mass, with a positive relation noted (r = 0.64). A negative correlation was seen between 2 preoperative measurements: the coronal plane axes differential and sagittal plane axes differential (r = –0.74). Furthermore, the regression analysis demonstrated that preoperative markers did not correlate with any postoperative variables.

Discussion

In this study, we observed that both tibial and talar components were placed within ±2 mm (degrees) with respect to 4 of 6 total postoperative measurements, for a total of 8 of 12 measurements falling within the desired range. The only criteria not meeting that precision target were tibial slope, tibial mediolateral translation, talar axial rotation, and talar anteroposterior translation, only one of which (tibial slope) is dictated by the PSI cutting block. This initial result suggests that PSI cut-through blocks are capable of precise implant positioning in 3-dimensional space.

Interestingly, not all 12 of the measured postoperative variables are directly dependent on PSI templating. Three of the tibial measurements (varus/valgus roll, slope, and superior/inferior translation) as well as 2 of the talus implant parameters (varus/valgus roll and superior/inferior translation) are directly controlled by the resection from the PSI block; the rest are influenced to some degree but can be overridden by the surgeon in this ankle replacement system (positioning of the implant itself including translation and rotation, and talus slope).

Separating the variables by direct or indirect PSI control, of the 5 that are directly linked to the PSI template, only 1 parameter (tibial slope) falls outside the stipulated ±2 mm (degrees) tolerance in this single-surgeon series (2.3 degrees dorsiflexed ±2.9). We found open tibial slope to be just outside of the tolerance goal, although still notably quite accurate and within the range of what many surgeons consider acceptable for ankle replacement. Notably, the PSI plan is created from a CT scan and does not account for intraoperative factors such as soft tissue balancing, gutter debridement, and previous surgeries, and surgeon discretion is required to place the implants in an ideal position for the patient. Therefore, there are extraneous variables that are unaccounted for preoperatively by the PSI templating system, and as a result the surgeon may have to alter the positioning of the cutting blocks in order to achieve a desirable outcome.

Specifically, with regard to the sagittal plane tibial slope, we identified several potential sources of error. First, the accuracy could have been negatively affected by the placement of the block itself; variables such as the presence of soft tissue could negatively influence this. Second, the tolerance in the saw slot itself, coupled with the weight of the saw and gravity, may have been sufficient to induce open slope. Finally, the tibial slope may also be resultant from impaction of the press-fit implants, which would apply a force directed anteriorly and cause the tibial tray to fall into relative open slope compared with the preoperative plan. It should be noted that the surgeon in this study prefers 2 to 3 degrees of dorsiflexion; however, this was built into the preoperative plan and thus did not affect the accuracy. Some authors prefer 3 to 5 degrees of open slope as it allows for additional dorsiflexion motion.^4,21 In summary, it can be inferred that the PSI templates allow for precise positioning and placement of tibial and talar implants, and that preoperative templating is useful to the surgeon, although surgeon discretion may be necessitated intraoperatively to achieve an ideal postoperative outcome and not necessarily match the template.

Of the 7 remaining variables indirectly linked to the PSI plan, 3 were calculated to fall outside of the accuracy tolerance and the other 4 were noted to be accurate to within ±2 mm (degrees). The parameter showing the most deviation from plan was anteroposterior translation of the talus implant, with an average posterior center of mass translation of 2.1 ± 2.5 mm (P = .010). Notably, improper positioning of the talar component relative to the tibial component in the sagittal plane has been implicated in early implant failure and poorer outcomes. ⁹ The PSI guides determine the position of the resection, but do not guide either the placement of trials or the instrumentation steps necessary to place the final implant. In this series, the surgeon placed the talus implant freehand based upon their own clinical decision-making. These decisions were based upon bony coverage for the talar implant, rotation matching to that of the second metatarsal and ankle joint, and other decisions such as ligament balancing which is not specifically addressed with PSI templating. The discrepancy between the PSI template and implant positioning is notable, and it is currently unclear if this is due to intraoperative surgeon decision-making, intraoperative surgeon error, PSI templating error, or other unidentified causes. The results of this study confirm this: we found an average posterior center of mass translation of 2.093 ± 2.463 mm with P = .010, indicating strong statistical differentials between the preoperative plan and actual postoperative positioning of the implant.

Axial rotation positioning of the ankle replacement implants is disputed, and previous authors have identified rotation goals as being a medial malleolar gutter, an angle matching bimalleolar gutter bisector, or based on other factors such as matching the second metatarsal to the tibial tubercle. ¹⁶ The template plan was to match rotation to a gutter bisector that was selected by the senior author as a compromise to remove minimal excess bone from the posteromedial tibia. In this system, the resections are completed and flat surfaces are created wherein implants can be set at the surgeon’s discretion. The senior author’s technique is to align rotation of the talus such that the second metatarsal and tibial tubercle are aligned as closely as possible intraoperatively following the resections. Often this aligns the talus along the second metatarsal, but variability of foot posture and deformity may demonstrate some mismatch. Once the talus rotation is set, the talus is instrumented for the implant and a trial implant is placed. The tibial rotation is then matched to the talus trial implant rotation to minimize any rotational mismatch in this fixed-bearing system. Using these techniques, the mean discrepancy between the gutter bisector and what was found to be the “true” position of the talus was 2.5 degrees, indicating that the gutter bisector angle is closer to the selected rotation of the implant than is the medial malleolar gutter angle. Further interpretation of this finding implies that, although talar implant rotation is closely matched to the bimalleolar gutter bisector, the bisector fails to predict precisely the magnitude of rotation necessary for adequate deformity correction, and further research into this area is warranted.

Additionally, current literature on accuracy in both conventional TAA and TAA using radiograph-based PSI was scrutinized to provide a comparison between those existing methods and this WBCT-based method. In general, comparing absolute accuracy with respect to coronal and sagittal component alignment revealed that WBCT-based PSI did not provide more accurate implant placement than radiograph-based PSI; accuracy was comparable to currently published values. ² Conducting an accuracy comparison between conventional TAA and WBCT-based method, PSI-guided TAA is difficult, as there is a scarcity of alignment data in the literature concerning implant placement accuracy of conventional TAA. Few studies exist comparing conventional and PSI-guided TAA; the few that do suggest no difference in alignment exists.^10,13 In addition, several studies show no difference in postoperative clinical outcomes between guided and conventional surgery.^3,13

Furthermore, to investigate potential relations between preoperative deformities and postoperative measurements, a correlation analysis was performed comparing each of the preoperative measurements and the measured postoperative variables. The most meaningful conclusion from this correlation analysis is the lack of correlation between the preoperative deformity and that of the postoperative positioning and angulation of the implants. This suggests that a greater deformity preoperatively does not seem to impact the final positioning of the implants and that patient-specific instruments can be used successfully for deformity correction at the time of surgical treatment with total ankle arthroplasty (Figure 5).

Figure 5.

Postoperative imaging. This figure shows postoperative weightbearing computed tomographs (left) and postoperative radiograph (right), demonstrating the placement of the implants following total ankle arthroplasty. Prophylactic fixation of the medial malleolus was accomplished and is visible in the radiograph. These images can be compared to the patient-specific instrumentation (PSI) plan, represented by the 3D render (middle) which was generated prior to surgery. Osteophytes are again represented in green.

This study is not without limitations. First, this was a single-surgeon, single-center study. The lack of variability in surgical technique along with a limited patient demographic may lend itself to sampling bias. Additionally, the sample size of 30 provides adequate statistical power (80.7%) to confirm statistically significant differences for the 6 tibial postoperative measurements; however, the talar implant exclusions resulted in insufficient power for those measurements (72.2%). As this sample size is relatively small, it may predispose to a lack of generalizability in the results. This study could also be compared to other studies using radiography.

Conclusion

In this group, the 26 of 30 total ankle replacements that were performed with PSI resulted in fairly accurate tibial and talus resections, with little deviation from the preoperative plan. When compared to existing arthroplasty standards, PSI was found to be capable of meeting them. Importantly, the lack of correlation between observed preoperative deformities and postoperative outcomes suggests that implant positioning accuracy with PSI was independent of preoperative deformity magnitude for the range of magnitudes included in this study group.

Supplemental Table 1.

Postoperative Measurement Terms and Definitions for Both Tibial and Talar Implants.

Measurement	Definition
Component coronal plane positioning (roll)	Final implant position relative to the preoperative plan, measured in degrees (varus/valgus, varus positive)
Component sagittal plane positioning (slope)	Final implant position relative to the preoperative plan, measured in degrees (plantarflexion/dorsiflexion, plantarflexion positive)
Component axial plane positioning (axial rotation)	Final implant position relative to the preoperative plan, measured in degrees (internal/external rotation, internal positive)
Component center of mass translation (x axis)	Mediolateral component center of mass translation relative to the preoperative plan, measured in millimeters (medial positive)
Component center of mass translation (y axis)	Anteroposterior component center of mass translation relative to the preoperative plan, measured in millimeters (anterior positive)
Component center of mass translation (z axis)	Proximal/distal component center of mass translation relative to the preoperative plan, measured in millimeters (proximal positive)