Imageless robotic-assisted revision arthroplasty from UKA to TKA

Background and objective It is evident from the national joint registries that numbers of revision knee arthroplasty operations are rising. The aim of this article is to introduce a new robotic-assisted approach in UKA to TKA revision arthroplasty and investigate the alignment accuracy, implant component use and surgery time and to compare it to primary robotic-assisted TKA arthroplasty. Methods This retrospective, case-control study included patients undergoing image-less robotic-assisted revision arthroplasty from UKA to TKA (n = 20) and patients undergoing image-less robotic-assisted primary TKA (control group, n = 20) from 11/2018 to 07/2020. The control group was matched based on the BMI and natural alignment. Comparison of groups was based on postoperative alignment, outlier rate, tibial insert size, lateral bone resection depth, incision-to-wound closure time. All surgeries were performed by a single senior surgeon using the same bi-cruciate stabilizing TKA system. Statistical analysis consisted of parametric t‑testing and Fisher’s exact test with a level of significance of p < 0.05. Results The two groups showed no differences in mean BMI, natural alignment (p > 0.05) and mean overall limb alignment. No outlier was found for OLA and slope analysis. The smallest insert size (9 mm) was used in 70% of the cases in the revision group (n = 14) and in 90% of the cases in the primary group (n = 18, p = 0.24), distal femoral and tibial resection depth showed no statistical difference (p > 0.05). The incision to wound closure time was longer in the revision group but showed no significant difference. Conclusion Image-less robotic-assisted revision arthroplasty from UKA to TKA showed a comparable surgery time, and alignment accuracy in comparison to primary robotic-assisted TKA. Comparable bone preservation and subsequent tibial insert size use was observed for both groups.


to TKA is presented and the accuracy of Introduction
Unicompartmental knee arthroplasty (UKA) is an established and successful procedure in contemporary orthopaedic surgery. In addition to the benefits of UKA implantation, such as bone conserving, ligament-sparing procedure, and restoration of normal knee kinematics, the UKA procedure continues to show higher revision rates in national prosthesis registries in comparison to total knee arthroplasty (TKA) procedures [8,14,15,20]. The UKA procedure is technically demanding and depends mainly on the experience of the surgeon [17]. The British National Joint Registry (NJR) showed that the revision rate for UKA is significantly higher for surgeons with < 30 UKA procedures per year (4% vs. 1% annual revision rate) [16]. In recent years, higher numbers of UKA procedures have been observed in the NJR for arthroplasty in Germany (EPRD, Endoprothesenregister Deutschland). The number of UKA procedures of all primary knee arthroplasty procedures in 2019 was 13.5% and about 4% higher than the previous value in 2015 [6]. For this reason, with an increasing number of UKA procedures, an increase in the number of necessary UKA to TKA revisions can be expected in future years. Furthermore, conflicting evidence still exists regarding clinical outcome and revision rates  [17,21]. In contrast, recent registry studies from Sweden and Norway and further meta-analyses showed reduced clinical outcomes and higher rerevision rates for UKA to TKA procedures compared to primary TKA [4,5,11,13,27]. Some authors suspected the reason for this is the mostly necessary use of augmentations, stems and higher inlays due to the increased bone loss compared to primary TKA [27]. In most cases, the increased bone loss during these revision surgeries is due to bone loss during implant removal and resection. As a result, larger inserts and more constrained inserts often have to be used in UKA to TKA revisions [22]. Robotic-assisted primary TKA (RA-TKA) is increasingly gaining popularity in the orthopaedic domain [9]. The main advantages of RA-TKA include improved component alignment and realtime control of soft tissue balance [12,19,23,26,31]. The use of RA-TKA in the context of UKA revision is a promising approach to improve the results of these revision procedures. The use of robotic assistance might reduce alignment outlier, intraoperative bone loss, and improve soft tissue balance. For this reason, the aim of this study was to examine the extent to which alignment goals can be achieved and bone-saving surgery can be performed. In addition, the workflow of imageless robotic-assisted UKA revision is presented.

Study design
This retrospective case control study consisted of two groups including a total of 40 patients. The first 20 consecutive cases of robotic-assisted UKA revision procedures from November 2018 to July 2020 were included in the experimental group. A matched control group consisted of 20 patients undergoing primary robotic TKA surgery in the same time interval matched by body mass index (BMI ± 5 kg/m 2 ) and natural overall limb alignment (nOLA ± 5°). All surgeries were performed by a single senior surgeon (ME) using a bi-cruciate stabilizing TKA system  23]. In the following, the technique of robotic-assisted UKA conversion to TKA is described: Due to the imageless system no preoperative CT scan is necessary. Preoperative analysis included full leg radiographs to determine OLA, mPTA and lateral distal femoral angle (lDFA). Additionally, preoperative full leg radiographs prior to UKA implantation were analyzed to determine the natural alignment and joint line height of the patient. If no radiographs prior to UKA implantation existed, full leg radiographs of the contralateral lesser extremity were used to determine the natural alignment. The intraoperative set-up was comparable to the already published setup [1, 7].
Step 1: Incision, array positioning, registration and soft tissue balance tracking A standard medial parapatellar approach is used, while the proximal incision is 1-2 cm longer to enable pin position on the femur within the same incision. Tibial pins are attached anteromedially 8-10 cm distal to the tibial plateau to avoid conflicts with the saw. All pins are attached within the bone bicortically. Definition of the hip/ knee center and range of motion (ROM) is done by the developer's standard following the workflow of the NAVIO TM software (. Fig. 1). Bone tracking with the handpiece and tracking of the soft tissue balance is done with UKA implants in situ. To achieve adequate varus and valgus stress in soft tissue balance, examination spacer blocks or Hohmann hooks can be used.
Assessment of the soft tissue balance in 0°a nd 90°and within the range of flexion from 0°to 90°should be tracked carefully to enable perfect soft tissue balance in component positioning.
Step 2: Planning of component alignment The aim of any arthroplasty surgery is to reconstruct the joint line and joint line obliquity with respect to the posterior condylar axis. Bone resection depth should not exceed 5.5-6.5 mm lateral (distal + posterior) and, if possible, 7.5-8.5 mm medial (distal + posterior, note the bone loss during implant removal and the thickness of the implant while assessing the medial cut depth) to avoid joint line proximalization. Alignment of the femoral component is done using the 3D planning intraoperative software to verify reconstruction of the posterior condylar axis and the posterior condylar off-set. The femoral component alignment is adjusted to the preoperative detected natural limb alignment (particularly lDFA). Gaps are then balanced especially in 0°and 90°flexion with symmetrical gaps of 1-2 mm mediolaterally. A slight lateral lift-off in flexion is accepted. Adaption of component alignment to gain optimal gap balancing is mainly done virtually prior to execution of cuts. Exceeding limits of 87°of mPTA are avoided An mPTA in excess of 87°should be avoided.
Step 3: Implant removal and bone cutting After virtual alignment of the prothesis components the UKA implants are re-moved. The distal cut of the femur is done with the trephine following the intraoperative plan determined preoperatively with the NAVIO TM software. Later, 4-in-1 cutting blocks are used to perform the last femoral cuts with navigated cutting blocks. The tibial side can be burred alone or preparing by a bone saw with tibial cutting blocks. A probe can be inserted to check gap balancing and postoperative alignment with the NAVIO TM system. If balance differences or limitation of full ROM exist, further bone cuts or soft tissue balancing might be done. Bone defects above 5 mm are compensated with appropriate augmentation.
Step 4: Component implantation After correct execution of bone cuts, component implantation can be done following the clinical routine. After cementing a probe inlay (BCS vs. CCK) and different PE thicknesses can be tested to achieve best soft tissue balance and stability.

Results
Demographic data and alignment analysis results are shown in detail in . Table 1. Both groups showed no difference of mean BMI (p = 0.0149) or preoperative natural coronal limb alignment (p = 0.151). Reasons of UKA revision were aseptic loosening (n = 8, 40%), secondary instability (n = 5, 25%), osteoarthritis progression (n = 4, 20%) and valgus overstuffing (n = 3, 15%). Preoperative mPTA and slope showed no differences between groups (p = 0.627; p = 0.941 respectively), 0 (0%) 0 (0%) -Values in mean ± SD SD standard deviation, UKA unicompartmental knee arthroplasty, TKA total knee arthroplasty, OLA overall limb alignment, mPTA medial proximal tibial angle, n.s. not significant whereas preoperative lDFA showed more valgus alignment in the UKA revision group (85.5°vs. 87.8°, p = 0.03). Incision to wound closure time was higher in the UKA revision group (76.0 min vs. 69.6 min) but did not reach significance level (p = 0.052). Postoperative overall limb alignment showed a slightly more varus limb alignment in the primary TKA group (178.6°vs. 176.0°), whereas all postoperative alignment parameters showed no statistical differences between the groups (. Table 1). Coronal and sagittal alignment accuracy was found to be similar between groups with no outliers for OLA and slope analysis in both groups (. Fig. 2, . Table 2) and the same outlier rate for mPTA analysis (5% each, n = 1). Lateral bone cut depth on the tibial and femoral side was comparable in both groups (. Table 1). Standard BCS inserts were used in 100% of the cases within the primary TKA group and in 90% (n = 18) of the UKA conversion cases (p = 0.49). Hence, CCK inserts were used in two cases of the UKA revision group (10%). No metal augmentation (stems, femoral augments) was used in any of the cases. Mean insert size was slightly higher in the UKA revision group but showed no statistical difference (9.6 mm vs. 9.1 mm, p = 0.11). The minimum onlay size was used in most of the cases in the UKA revision group (70%) but less frequently when compared to the primary TKA group (90%, p = 0.24).

Discussion
The main result of this retrospective case control study is that robotic-assisted conversion from UKA to TKA is a precise technique in revision arthroplasty and shows similar alignment outcome parameters when compared to primary RA-TKA.
Alignment accuracy of imageless robotic-assisted UKA conversion to TKA was similar to primary RA-TKA. No radiological outliers for OLA and slope analysis werefound ineither of thegroups. Furthermore, postoperative coronal alignment parameters were all comparable between the groups. In addition to the numerous published studies on improved alignment accuracy in primary robotic-assisted knee Postoperative alignment (b) after robotic-assisted conversion from UKA to TKA with neutral overall limb alignment arthroplasty in comparison to conventional TKA [12,19,23,26,31], this case series was able to demonstrate a comparable accuracy and outcome to roboticassisted primary knee arthroplasty. Intraoperative control of component alignment and joint line restoration in revision arthroplasty is often a challenge due to further bone loss or loss of bony reference points after implant removal. Therefore, malalignment after manual UKA conversion to TKA might be reduced with this robotic-assisted approach. Several studies reported a decreased clinical outcome after UKA to TKA revision when compared to primary TKA. These studies showed an increased use of metal augmentation, higher constraint level and increase of polyethylene (PE) thickness in UKA to TKA revision groups [11,22,29]. Contrary to the published data, a comparable PE thickness was observed in the current case series for UKA revision in comparison to the primary group and the minimum PE size could be used in the majority of the UKA revision cases (70%). Additionally, the robotic-assisted approach enables the intraoperative control of soft tissue balance, which might support the use of standard primary implants and minimum size PE onlays instead of constrained implants and onlays. In this context, imagebased robotic-assisted UKA revision also showed comparable and a tendency towards lower PE thickness compared to manual UKA revision in a recently published study [30]. Alignment parameters were not analyzed in that study. Thus, no comparison between imageless and image-based robotic-assisted UKA revision is possible at this point.
Furthermore, in the UKA revision cases in this series, the use of augments or stems was not required. Individual studies report the need for augments or stems in 30-54% of UKA to TKA revisions [2,29]. Additionally, considering the lateral bone resection height in our study, which was comparable between both groups on the tibial and femoral side, it can be assumed that a bone-conserving revision of UKA to TKA is possible by robotic assistance. Nevertheless, two cases of the UKA revision group (10%) needed a semiconstrained insert, while none of the primary TKA cases needed a CCK insert. Comparable studies by Sarraf et al. [22] and Lunebourg et al. [18] reported the use of constrained implants in 4.2-10.4% of UKA to TKA revision cases. Nevertheless, the exact level of constraint remains unclear.
Despite the technically more difficult procedure of UKA revision, the time from incision to closure was only slightly higher in the UKA revision group compared to the primary TKA group (76.0 min vs. 69.6 min), this difference also did not reach significance level. The incision to wound closure times (ICT) determined in this study are to be compared to frequently published data of single center studies or registry studies [3, 10, 24,25,28]. Published mean ICT range from 67 min to 104 min, regard-less of whether manual or robotic-assisted TKA was used [3, 10, 24,25,28]. Therefore, robotic-assisted UKA conversion is comparable to primary arthroplasty in terms of average time required and does not lead to a significant increase in time from incision to closure.
This study has several limitations. First of all, this was a retrospective study of the first 20 consecutive cases of roboticassisted revision from UKA to TKA analyzing objective operative parameters and alignment outcome only. Thus, the statistical power is reduced. Moreover, the control group was matched by BMI and natural alignment. In some UKA revision cases, preoperative full leg radiographs prior to UKA implantation were not available. Thus, a full leg radiograph of the contralateral side, as performed in the clinical routine, was used to identify the natural alignment. Nevertheless, natural alignment of both lower extremities might differ significantly and therefore might have influenced the classification. Furthermore, no clinical outcome data were obtained because this was not the subject to this study; however, this would give further importantinformationaboutrobotic-assisted UKA revision and should be investigated in further studies. Inaddition, this studycompared UKA revision with primary robotic arthroplasty. No comparison was made between manual and robotic-assisted UKA revision. The aim of this study was to verify whether robotic-assisted UKA revision is comparable to primary arthroplasty in the parameters investigated. The aim was to improve the surgical technique to bring the results of UKA revision in line with primary arthroplasty. Nevertheless, a manual UKA revision comparison group would help to determine whether the roboticassisted technique could achieve an improvement in the accuracy of this operation in a direct comparison.
Conclusion 4 Robotic-assisted revision from UKA to TKA is a reliable approach with accurate component alignment. 4 Robotic-assisted revision from UKA to TKA might help to preserve bone stock and might avoid using revision augmentation material and higher constraint implants.