Complications and Learning Curve Associated with an Imageless Burr-Based (CORI) Robotic-Assisted Total Knee Arthroplasty System: Results from First 500 Cases

Abstract

Background

The use of robotic-assisted total knee arthroplasty (RA-TKA) is gaining traction. There is evidence to suggest that RA-TKA can help to optimize the precision and accuracy of implant positioning and that there may be protective effects on surrounding bony and soft tissues. Yet, there are important differences between the various RA-TKA systems currently on the market. One such newly introduced RA-TKA system uses imageless technology and performs bony cuts with the use of a burr-based device. The learning curve and complications unique to this system have yet to be assessed.

Methods

We evaluated 500 consecutive RA-TKA cases using a newly developed burr-based and imageless system which were done by a single surgeon between the months of October 2021 and February 2023. Operative times were recorded and compared to the previous 150 conventional TKA cases allowing for the learning curve to be calculated using the CUSUM method. Intraoperative and postoperative complications were categorically profiled.

Results

The learning curve of this RA-TKA system was found to be 6 cases. Intraoperative complications included unintended bony over resection (n = 3), soft tissue injury (n = 2), and robotic system hardware (n = 2) or software (n = 2) malfunction. Postoperative complications consisted of superficial pin site infection (n = 1) and periprosthetic fracture near the pin sites (n = 1). There were no identified cases of prosthetic joint infection, instability events, or wound complications.

Conclusions

The learning curve and the complication profile of a newly introduced imageless and burr-based RA-TKA system were described. This information serves to guide surgeons in adopting this technology and can counsel them regarding the potential pitfalls and challenges associated with its integration into practice. The work sheds light on the complexity and learning curve of the recently released imageless burr-based RA-TKA system. This important information is intended to help surgeons accept this cutting-edge technology by providing advice on any errors and difficulties that can occur when integrating it into clinical practice. This information can help surgeons navigate the complexities of integrating this new burr-based robotic technology into knee replacement procedures, enabling them to make well-informed decisions and receive guidance.

Introduction

Total knee arthroplasty is one of the most commonly performed procedures in orthopedics and its volume is projected to continue to rise dramatically [1]. Robotic-assisted total knee arthroplasty (RA-TKA) has garnered significant attention as a potential means of improving precision and optimizing the accuracy of implant positioning [2]. As a result, device manufacturers have introduced several robotic systems over the past decade with both semiautonomous and autonomous design platforms. Among these robotic systems, there are disparities in the type and degree of haptic feedback that the surgeon receives; and while some systems rely on advanced imaging such as computed tomography, others employ imageless technologies [3]. Further, some systems accomplish bone resection via cutting burrs, while others use saws or cutting guides. The refinement in technology is based on improvization in both the hardware and software of the system with a focus being on machines being faster, accurate, easy to use, and safe.

Compared to conventional total knee arthroplasty (C-TKA), RA-TKA has been shown to demonstrate improved implant positioning, [4, 5] better soft tissue and bony protection, [6, 7] and decreased local inflammation [8]. However, clear evidence touting its superiority over C-TKA with regards to long-term outcomes, implant survivorship, and revision rates is yet to be proven although isolated studies are now emerging demonstrating varied clinical benefits of use of the robotic system [3]. Robotic-assisted technologies have been associated with several unique complications [9]. For example, most RA-TKA systems utilize computer navigation pins that are typically inserted into the tibial and distal femoral cortices. The use of such pins has been associated with superficial pin site infections and even periprosthetic fractures [10–13]. Additionally, the integration of RA-TKA technologies into surgical workflow has been shown to incorporate a noteworthy but not insurmountable learning curve for the surgeon [14–16]. As new RA-TKA systems continue to be introduced to the market, it is imperative that surgeons be able to recognize and treat the complications unique to those systems as well as understand the input required to adopt these technologies into their practice so as to avoid or predict complications associated with the technology. While most studies so far have focused on complications related to pin tracts and planning accuracy, there is a paucity of data on more comprehensive evaluations of the robotic systems, including both technique- and technology-related failures.

The aim of this study was to (1) calculate the learning curve of a recently introduced burr-based and imageless RA-TKA system, and (2) report on the complications specific to the use of this particular robotic system.

Materials and Methods

Study Group

This prospective cohort was comprised of 500 consecutive patients undergoing total knee arthroplasty by a single surgeon at an urban tertiary referral center and research hospital between the months of October 2021 and October 2023 with approval from the Institutional Review Board and Ethical Committee (HNH/IEC/2022/OCS/ORTH/96). Inclusion criteria included patients above 18 years of age with radiographic and clinical evidence of knee arthritis undergoing primary TKA utilizing a novel imageless robotic-assisted TKA system known as the CORI Surgical System (Smith & Nephew plc, Watford, England). There was a minimum follow-up of 1 year postoperatively. For the purposes of our analysis, we excluded the following: revision TKA (n = 7), conversion of unicompartmental knee arthroplasty (UKA) to TKA (n = 3), and history of prior knee joint infection (n = 1). Operative time of this robotic-assisted TKA surgeries was compared with that of conventionally TKA operated by the same surgeon previously, for analyzing learning curve. Intraoperative and postoperative complications in general and those specific to this novel burr-based system were determined to assess the safety of this system. For each case, routine preoperative radiographs comprising weight bearing anteroposterior, lateral, and full length standing films were obtained. The nature of the procedure was explained in depth to all patients undergoing surgery and written informed consent for the surgery was attained preoperatively.

Surgical Technique

A standard medial parapatellar approach was used in all cases. For the purpose of registration, distal femoral and proximal tibial pins were inserted bi-cortically. Two 4mm distal femoral pins were placed through the arthrotomy site. An additional two 4mm tracking pins were placed at the proximal tibial diaphysis through two small incisions. Intraoperative mapping to create a geometric 3D model for the robotic system was done. Joint balance and kinematics were recorded using both unstressed and varus/valgus stress throughout range of motion. The surgeon was able to use the planning screen to obtain the desired alignment and balance. The LEGION Total Knee Implant System (Smith & Nephew) was used in all cases. The choice of cruciate retaining versus substituting implants was made at the discretion of the surgeon guided by the gaps displayed by the robotic screen. Execution of bone cuts was performed by a hand held burr-based sculpting device. After trialing, joint alignment and balance was recorded. Cemented implants were used in all cases. The patella was not resurfaced in any case.

Learning Curve

Operative time was recorded for every case. Operative time was defined as the time from surgical incision to wound closure. The learning curve for the new robotic system was assessed by a method previously described by Kayani and colleagues while evaluating a separate robotic platform [16]. Cumulative summation (CUSUM) is a statistical technique developed by E.S. Page used for monitoring change detection. CUSUM involves calculation of a cumulative sum. Samples form a process x_n are assigned weights w_n and summed as follows:

$${\text{S}}_0=0$$

$${\text{S}}_{\text{n}+1}=\text{max}\left(0,{\text{S}}_{\text{n}}+{\text{x}}_{\text{n}+1}-{\text{w}}_{\text{n}}\right)$$

Change in value is found when the value of S exceeds a certain threshold value. CUSUM sequential analyses were used to assess the learning curve associated with operative time. The operative time from the previous 150 C-TKA cases was used as a reference value for the CUSUM analyses. Based on the description of Kayani et al, a minimal clinical difference was set at 5 min and the standard deviation was set at 10 min [16]. A minimum of 60 patients were determined to be required to detect a minimal difference in operative time using two-tailed, two sample t-tests with a power of 80% and achieving a 5% significance level. Sequential CUSUM analyses were used with overall mean values of case duration from the RA-TKA group. CUSUM values were then representative of the running total of the differences between each data point and the standardized target value of the respective outcome. All statistical analyses were performed using standard statistical software programs.

Intraoperative Complications

Intraoperative safety of the robotic arm was defined in terms of unintended bony or soft tissue damage occurring during the procedure. We were able to categorize these intraoperative complications into three groups recorded as:

1. Bony over resection by the handheld burr (determined by pink and red patches on bony surfaces displayed over the robotic screen),

2. Ligamentous injury,

3. Soft tissue injury other than ligamentous injury,

4. Intraoperative hardware complications and

5. Software malfunctions (Fig. 4).

Fig. 4.

Periprosthetic fracture at pin tract on the femoral side reported 3 weeks post operation. Patient later underwent periprosthetic fracture fixation using a bridging locking plate. Patient recovered and fracture healed 3-month post fixation

Burr over-resection in particular was defined as burr penetration over 2x2 mm beyond the desired resection depth, which is indicated by more than three dark red areas as demonstrated by the software screen.

Postoperative Complications

Postoperative complications in general and those specific to the robotic system were analyzed and recorded. Complications included acute prosthetic joint infection, postoperative instability events, or wound complications. Complications more specific to the use of this robotic TKA system included periprosthetic fracture near pin sites, pin tract infection, pain at the pin tract site, and prolonged pin tract discharge.

Results

Demographics

500 knees that underwent RA-TKA by a novel burr-based and imageless robotic system were included in the study. The mean age of the study group was 67 ± 8.16 years. Male patients comprised 30.52% of the study group, while female patients made up 69.48%. Unilateral RA-TKA was performed in 63.76 % patients and bilateral RA-TKA was performed in 36.24% patients (Table 1).

Table 1.

Patient demographics

Patient demographics
Total knees	500
Number of patients	367
Unilateral TKA	234 (63.76%)
Bilateral TKA	133 (36.24%)
Male gender	112 (30.52%)
Female gender	255 (69.48%)
Mean age	67 ± 8.16 (years)
Varus deformity	467 Knees
Valgus deformity	33 Knees

Learning Curve

Using the cumulative sum analysis method, the learning curve associated with the use of this robotic system was established. The learning curve for the novel robotic system was calculated to be six cases (Fig. 1a). Prior to reaching the six-case learning curve threshold, operative times for the RA-TKA cases were found to be of longer duration compared to the previously recorded 150 C-TKA operations. Overall, there was a decreasing trend noticed in surgical time as the number of cases increased with the average operative time for the first and last 20 cases being 79 and 52 min, respectively (Fig. 1b).

Fig. 1.

a Learning curve. b Illustration of surgical time in entire 500 cases and average time in first and last 20 cases

Complications (Table 2)

Table 2.

Complications

Intraoperative and postoperative complications
	Complication	N (%)
Intraoperative	Over resection of bone by the robotic arm (more than three red areas exceeding 2x2mm)	3 (0.6%)
	Ligamentous injury	0 (0)
	Soft tissue injury (other than ligamentous injury)	2 (0.4%)
	Robotic system hardware complications	2 (0.4%)
	Robotic system software malfunction	2 (0.4%)
Postoperative	Periprosthetic fracture	1 (0.2%)
	Pin tract infection	1 (0.2%)
	Pain or prolonged drainage from pin site	0 (0)
	Acute prosthetic joint infection	0 (0)
	Instability events	0 (0)
	Wound Complications	0 (0)

Intraoperative Complications

Nine patients (1.8%) experienced intraoperative events specific to this system. Three patients demonstrated greater than three dark red areas on the robotic software indicative of over resection by the burr to an area 2x2 mm deeper than planned, which can be either due to malfunction of the robotic arm or because of severe osteoporosis (Fig. 2). In one case, the burr caused a partial tear in the patellar tendon, which required primary repair. A long knee brace was provided for use during ambulation for a two-week period but there was otherwise no change in postoperative restrictions or rehabilitation and no long-term sequelae were identified. In a separate case, a patient suffered a contact burn over the skin due to heat produced by the high speed burr, which healed without complication. Robotic system hardware complications (Fig. 3) included two cases of malfunction in the form of burr sheath breakage. A robotic software malfunction led to locking of the burr in two cases (Fig. 3). In three cases, a replacement robotic system was readily available and the operations were able to be completed with robotic assistance as originally planned. In one case, the operation had to be shifted to a conventional jig-based system with an intramedullary femoral guide.

Fig. 2.

Over-resection by robotic arm. a Over-resection possibly due to osteoporotic bone wherein the pressure of the operator on weak bone may cause resection beyond the intended depth. b Here, the over-resection is likely secondary to malfunctioning in the software or a calibration error causing the drill over-resect bone. Note more uniformly red areas denoting device mis-calibration. c and d Mis-calibration on the tibial side

Fig. 3.

a Software malfunctioning mid procedure being reported by the system b A key performance characteristic (KPC) analysis done later evaluates various elements of possible hardware and software malfunction

Postoperative Complications

One patient (0.2%) sustained a periprosthetic fracture (Fig. 4) of the distal femur one month after surgery that was conceivably related to the use of navigation pins. The fracture was subsequently successfully treated with open reduction and internal fixation. There was one incidence of superficial pin tract infection which was cultured and successfully treated with sensitivity guided antibiotics without requiring a revision surgery. There were zero instances of acute prosthetic joint infections, wound complications, or instability events.

Discussion

We described the learning curve associated with adopting an imageless and burr-based RA-TKA system. Additionally, we recorded the intraoperative and postoperative complications unique to this system. Robotic systems are the next step in evolution from the ‘conventional’ navigation system as the robotic arm plays a significant role in the execution of surgery. While this represents significant progress in the automation of surgery, these technologies may bring a unique set of problems not usually seen with systems that rely on the dexterity of the operating surgeon. This information will be useful for surgeons hoping to integrate this technology into their practice and can be used to direct surgeons to be able to recognize and treat its common complications.

The reliance on navigation pins by the various RA-TKA systems is associated with several well-recognized complications. Lonner and Kerr found a 0.6% rate of pin site complications in their series of 1064 RA-UKAs, including delayed pin site healing/superficial infection (n = 4), arterial injury (n = 1), and fracture of the tibial metaphysis (n = 1) [11]. In their meta-analysis assessing tracking pin complications in computer-navigated and robotic-assisted knee arthroplasties (including total, unicompartmental, and patellofemoral arthroplasties), Thomas et al found a 1.4% rate of all pin-related complications [10]. Specifically, they reported a 0.6% incidence of superficial pin tract infections and 0.2% incidence of postoperative pin site fracture. Further, they found that most pin-related complications were concentrated at the tibial pin sites (69%). More importantly, and similar to our study, all pin-related complications were properly addressed and had resolved by final follow up [10]. In a recent meta-analysis, Raj and colleagues sought to assess the difference in infection rates among RA- and C-TKAs [17]. They found a summary rate of 0.347% for superficial and pin site infections with a 0.154% rate of deep infections for patients who underwent RA-TKA. Our results are concordant with theirs as we demonstrate a 0.2% rate of pin site infection with zero cases of deep infection.

The use of navigation pins does pose a risk for periprosthetic fracture around the pin sites. Our study recorded one case of periprosthetic pin site fracture; it was located in the distal femur and was treated successfully with open reduction and internal fixation. It is unclear how this case differed from those who did not experience fractures. With biomechanical testing, researchers exhibited that eccentric drilling of the femoral navigation pins increases the risk of fracture and causes the bone to be particularly vulnerable to torsional loads [18]. Literature has noted that the majority of pin site-related fractures occur in the femur (59%) and in women (83%), and while reported fracture cases differ in fracture severity and displacement, all were able to heal with proper treatment [13].

The clinical significance of unplanned burr over resection is unknown. While there are data indicating that planned over resection at the medial or lateral side of the tibial plateau can affect coronal balance, [19] to our knowledge, there are no studies evaluating the consequences of unintentional over resection such as that which was measured in the present study. In the revision TKA literature, small cancellous defects would not be expected to impact implant stability though it has been recommended that bone grafting be used for cystic lesions greater than 5 mm in diameter [20]. We reported four instances of intraoperative hardware or software malfunction. There remains a paucity of data recording intraoperative software or hardware failures in the realm of RA-TKA surgery. While the overall incidence of these complications appears to be low, it is still unclear what impact they may have on final outcomes after surgery. However, it is clear that surgeons must be prepared to resume cases manually should there be robotic system failure mid-operation. It also stresses the need for constant surveillance and periodic checks of robotic equipment. As for software-based errors, more robust algorithms can be optimized to ensure that they remain glitch free. Although not seen in this study and not a major concern for most systems at this point, one must also be cognizant of the potential future dangers of these machines being vulnerable to cyber-attacks in addition to concerns surrounding data privacy if these machines become connected to wider networks.

We reported two instances of iatrogenic intraoperative soft tissue damage—both relating to the cutting burr. These could be attributed to latency times and torque that the machines have while executing a command. Robots are theoretically controlled in “real time,” meaning that any command gets executed almost instantly. However, this near instantaneous response means that there is still a lag of microseconds in command execution and at high speed it can damage surrounding structures unintentionally. Time delay mitigation thus remains an additional frontier for more sophisticated robotic teleoperations. While both soft tissue injuries were managed successfully, surgeons should be keenly aware of some of the unique pitfalls and challenges associated with the use of the burr in this system. Researchers compared iatrogenic intraoperative soft tissue injuries between RA-TKA and C-TKA. Using blinded, fellowship-trained surgeon observers, they revealed that RA-TKA exhibited less macroscopic soft tissue damage compared to conventional jig-based TKA techniques [7]. It has been posited that the robotic systems may reduce the incidence of iatrogenic soft tissue injury by nature of the stereotactic boundaries established by the robotic software at the operative site. Studies have argued that the robots have better control and less soft tissue injury potential and this has been corroborated by measuring inflammatory markers in postoperative patients. By measuring intraarticular interleukin levels, researchers revealed lower levels of local inflammation at the surgical site after RA-TKA compared to manual cases [9]. It is hypothesized that this could be used as a surrogate measure of soft tissue injury caused during any intervention.

To our knowledge, the current study is one of the first to estimate the learning curve associated with integrating the novel CORI imageless RA-TKA system into the surgical workflow. A learning curve is a correlation between a learner's performance on a task and the number of attempts or time required to complete the task; this can be represented as a direct proportion on a graph. The learning curve theory proposes that a learner's efficiency in a task improves over time the more the learner performs the task. In a similar study but of a different RA-TKA system, Sodhi et al compared C-TKA operative times to the first 20 robotic-assisted cases and the last 20 robotic-assisted cases out of 200 total. While the initial robotic-assisted cases lasted significantly longer, they found that operative times between C-TKA and the last 20 RA-TKA cases began to equalize but did not assign a specific number of cases required to reach proficiency [21]. Kayani et al reported a learning curve of 7 cases using an alternate robotic system [16] while Grau et al found a 6-case learning curve using the same platform [22]. We calculated a learning curve of 6 cases for adopting this novel RA-TKA system, which is comparable to that found by other researchers using similar robotic-assisted platforms. Analyses of the device company’s antecedent rendition of the RA-TKA system we utilized found learning curves of 7, 11, and 25 cases [23–25].

As technology evolves, it can be expected that newer robotic systems will be offered to surgeons and that each of these will have their advantages and disadvantages. Further, there are different alignment philosophies and surgeon preferences to consider [26]. Not only surgeons, but patients too have their own preferences based on their cultural and functional requirements [27, 28]. To this end, surgeons must know what is the most appropriate system given any scenario. The major players today apart from CORI by Smith and Nephew are Velys by Depuy, Mako by Stryker, CUVIS by Curexo and ROSA by Zimmer. While the core philosophies overlap, the method of execution and planning differs in each, offering a unique opportunity for surgeons to exploit their strengths and at the same time be cognizant of their downsides. This study is a major step toward strengthening surgeons’ operative acumen with a focused description about the safety and learning aspects of one of the key robotic systems.

This study does have several potential limitations. As this is the experience of a single surgeon, its generalizability to all users and all robotic systems is limited. However, this can also be considered a potential strength as the machine was handled by a single surgeon and that there was no inter-surgeon variability as far as operative skills are concerned. This study was conducted within the first two years of the purchase of the robotic system and it may be argued that the other delayed complications related to software and hardware malfunction may come into play after a certain number of cycles have been run and a certain number of surgeries performed. While a valid argument and indeed one that requires monitoring over time, this study fills in a significant gap and offers insight into what complications can be expected and how surgeons can adapt to this new system. One may also argue that the study failed to demonstrate any benefits or report on patient reported outcome measures. While there are several studies and a push to assess the benefits of RA-TKA systems in clinical settings, this study was conducted with the singular aim of determining the learning curve for a particular RA-TKA system and the incidence of robotic specific complications associated with its use.

In conclusion, this study fills a significant gap in the field of RA-TKA by reporting on complications specific to the use of this novel robotic system. This study is the first to report on the learning curve for the CORI RA-TKA system. The findings of this study can help formulate a template for the evaluation and comparison of future generations of RA-TKA systems.

Funding

Ravi Bashyal is a consultant and on the speaker’s bureau for Smith & Nephew and Stryker, and has received honorariums related to this work. No research funding was used for this work.

Declarations

Conflict of interest

Vaibhav Bagaria, Douglas Weaver, and Shobit Deshmukh declare that they have no conflicts of interest.

Ethical approval

The study was conducted between October 2021 and October 2023, with approval from the Institutional Review Board and Ethical committee (HNH/IEC/2022/OCS/ORTH/96). Letter of approval is attached in submission.

Informed consent

For this type of study informed consent is not required.