Robotic-Assisted Correction of Adult Spinal Deformity

Introduction of technology

Adult Spinal Deformity: Present Practice and Standard of Care

Adult spinal deformity (ASD) is a heterogeneous condition of typically the elderly patient, which involves misalignment in the sagittal and/or coronal plane. ASD can be asymptomatic or, depending on its type and severity, present with varying degrees of, and oftentimes excruciating, pain and disability typically exceeding the health impact of other chronic diseases, including those of the heart and lung. Even though nonoperative treatment is typically recommended first in patients presenting without significant or progressively worsening pain, neurological deficit, or deformity, studies have so far been unable to detect significant improvement in the health status of ASD patients with conservative measures, whereas adequate surgical treatment consistently alleviates pain and leads to clinically meaningful decrease of disability.

Over the last decades, improvements in surgical technology, including both techniques and instrumentation, have advanced ASD surgery from supportive to corrective and expanded the patient population that can now hope to regain functional ability and health-related quality of life (HRQoL) after successful treatment. Surgical treatment is challenging, however, for a number of reasons, including patient selection (risk-benefit assessment of surgery), patient optimization (e.g., increase in bone quality, prehabilitation, smoking cessation, reduction of risk factors for complications in typically elderly and frail patients), surgical planning (e.g., extent, combination of approaches, need for osteotomies), use of operative technology (e.g., intraoperative imaging, navigation, electrophysiological monitoring), technical nuances during the case, operative length, and complication management intra- and postoperatively. Besides surgical excellence, modern anesthesia and critical care management are required to deal with potentially excessive blood loss during and after ASD correction.

With complication and reoperation rates as high as 70% and 28% at the 2-year follow-up, there is a need to improve safety and durability of ASD surgery. The evolving minimally invasive surgery (MIS) techniques for ASD aim to reduce morbidity, lower complication rates, and shorten postoperative recovery time. The options of the MIS surgeon include circumferential MIS techniques with anterior or lateral discectomy (indirect decompression), followed by posterior percutaneous instrumentation or hybrid techniques, again with anterior or lateral discectomy but followed by a conventional open posterior approach for instrumentation and decompression and osteotomies, as needed. Not all ASD patients are suitable for MIS techniques, and algorithms to guide decision-making with regard to the optimal approach have been proposed.

Current Use of Robotic Surgery for Adult Spinal Deformity

Robots have been the logical next step in the toolbox of MIS spine surgeons, building on the platform of navigation but optimizing critical issues in MIS surgery, including radiation exposure and precision in patients with challenging anatomy. Deformity surgery requires meticulous fine motor skills and a steady hand to exploit small working corridors and minimize collateral damage during exposure, but the arduous and long procedures predispose to mental and physical fatigue. Moreover, the individual patient’s segmental vertebral anatomy may not be well visualized in either the AP or lateral plane using C-arm imaging, and both 3D navigation with oblique image reconstruction and robotic guidance allow accomplishing complex screw trajectories with higher confidence. The fundamental advantages of the robot are geometric precision (allowing for exact localization of the starting point, orientation, and trajectory length), infinite reproducibility, perfect “memory,” lack of physiological tremor or fatigue, and insensitivity to radiation. The synergy between those qualities and the spine surgeon’s inherent judgment, experience, and adaptability are key to successful robotic procedures. In general, three different levels of robotic assistance can be differentiated:

tele-surgical systems with remote command stations from where the surgeon controls the machine (e.g., the da Vinci Surgical System [Intuitive Surgical, Inc.]);
supervisory controlled systems with pre-programmed machines that autonomously perform certain actions; and
co-autonomy type shared-control models where robot motion is concurrently controlled (e.g., SpineAssist/Renaissance/Mazor X).

For application in spine and ASD surgery in particular, the available robotic systems belong in the third category.

Pedicle screw constructs are today’s basis of spinal fixation, providing substantial rigidity to facilitate bony healing (fusion) and in the absence of a navigation system, their safety depends largely on the patient’s anatomical landmarks and the surgeon’s experience. It is well documented that even in experienced hands the rate of implant malposition ranges from 5% to >30% for cases with challenging anatomy; since ASD cases often comprise revision procedures, reliable anatomical landmarks are frequently altered. Early robotic systems have thus focused on guiding the placement of pedicle screws by preparing optimal trajectories based on navigation, a task by which most modern systems continue to be confined. Even in the latest surgical robots for the spine, surgeons are still required to place the pedicle screw manually. The combination of surgical robotics with intraoperative CT or O-arm image acquisition has greatly simplified implementation, obviating the need for patient registration based on C-arm fluoroscopy, while increasing the precision of robotic surgery even further.

Beyond assisting in pedicle screw fixation, today’s applications of robots for ASD cases are few. To restore lordosis and obtain solid arthrodesis across the anterior spinal column, interbody fusions including anterior and lateral trajectories are commonly used. The obstacles to anterior lumbar interbody fusion (ALIF) are the ureters and large retroperitoneal blood vessels (aorta, vena cava, and branches) overlying the lumbar spine anteriorly. The da Vinci Surgical System (Intuitive Surgical) has been used successfully in animal models and several smaller case series to mobilize aforementioned critical structures and help facilitate the approach laparoscopically. The visualization inside and around the disk space was considered superior to conventional open or laparoscopic techniques. However, the da Vinci is not yet FDA approved for spinal instrumentation and outside exploratory clinical research its use—in particular for ASD cases—cannot yet be recommended. The authors are not aware of any other robotic application for anterior or lateral discectomies, osteotomies or deformity correction.

Adult Spinal Deformity—Outcomes With and Without Robotic Assistance

Accuracy of Thoraco–lumbar Screw Implantation

High-quality randomized studies comparing outcomes of ASD surgery with or without use of robotic guidance are lacking. As such, our understanding of its value is taken from critical review of pro- and retrospective cohort studies or extrapolated from meta-analyses of trials in non-ASD populations. There are some reports on the accuracy of robotic thoracolumbar pedicle screw instrumentation (summary provided in Table 8.1 ), where the accuracy of pedicle screw insertion is usually graded according to the classification proposed by Gertzbein and Robbins (G&R; Box 8.1 ).

Table 8.1

Review of the current literature, summarizing articles published until 07/2019 that report on the safety and accuracy of robotic-guided thoracolumbar pedicle screw placement for ASD and related spinal disorders.

Authors, year	Study design	Number of patients/screws/type of robot	Assessment of accuracy	Assessment of safety	Assessment of feasibility
Devito et al., 2010 ²⁸	Retrospective multicenter study (14 hospitals); population included about 14% with AIS and ASD	N = 842 patients; n = 635 (3271 implants) available with intraoperative fluoroscopy; n = 139 (646 implants) available with postoperative CT scan; Mazor SpineAssist ®	Intraoperative x-ray and postoperative CT (when available); clinical acceptance rate 97.9%; G&R grades A in 89.3%, B in 9.0%, C in 1.4%, and D in 0.3%; 98.3% within safety zone (G&R grades A and B)	N/A	Of 682 cases in which 3912 implant insertions were planned, 83.6% (3271 implants) were fully executed under robotic guidance (the remainder initiated under robotics guidance and manually continued by the surgeon)
Macke et al., 2016	Retrospective cohort study; population with AIS	N = 48 patients (662 implants); Mazor Renaissance®	Review of postoperative CT; G&R grades A & B in 92.7% (safety zone), C in 4.5%, D in 1.5%; E in 1.2%	N/A	N/A
Fan et al., 2018	Retrospective cohort study; population with ASD	N = 286 patients, including n = 83 with robotic PLIF, n = 75 with template-guided PLIF, n = 109 with CT-navigated PLIF	Review of postoperative CT. Robotic cases: G&R grades A in 91.3%, B in 4.7%, C in 2.8%, and D in 0.6%; 96.0% within safety zone (G&R grades A and B)	Robotic cases: Six complications in total (7.2%), including dural tear, SSI, wound revision. No neurological complications occurred. Two revisions due to cage misplacement were necessary.	N/A

AIS , Adolescent idiopathic scoliosis; ASD , adult spinal deformity; G&R , Gertzbein and Robbins; N/A , not available; PLIF , Posterior Lumbar Interbody Fusion; SSI , surgical site infection.

Box 8.1

The Gertzbein-Robbin (G&R) classification is illustrated, discriminating between five grades and describing the degree of screw deviation from the “ideal” intra-pedicular trajectory.

G&R type	Degree of cortical breach	Description
A	0 mm	Intra-pedicular screw without breach of the cortical layer of the pedicle
B	<2 mm	Screw that breaches the cortical layer of the pedicle but does not exceed it laterally by 2 mm
C	2–3.9 mm	Penetration of <4 mm
D	4–6 mm	Penetration of <6 mm
E	>6 mm	Screws that do not pass through the pedicle or that, at any given point in their intended intra-pedicular course, breach the cortical layer of the pedicle in any direction by more than 6 mm

In 2010, Devito et al. described the accuracy rates of n = 646 implants deployed with the Mazor SpineAssist robot as G&R grades A in 89.3%, B in 9.0%, C in 1.4%, and D in 0.3%, reviewing their series of robotic screw guidance in a population that included about 14% of adolescents with scoliosis and adults with spinal deformity.

In 2013, Hu et al. assessed screw placement accuracy based on intraoperative biplanar fluoroscopy, postoperative radiographs, and surgeon’s judgment in a retrospective review of n = 102 consecutive patients, of which the robot (Mazor Renaissance) was successfully used in n = 95, including n = 85 patients (89.5%) with ASD (kyphosis and scoliosis). Of n = 1085 planned screws, n = 960 were implanted and n = 949 were considered successful and accurate (98.9%). Eleven mispositioned screws amounted to a rate of 1.1%, of which one patient with a laterally placed L4 pedicle screw and L3 radicular pain required revision surgery. A total of 110 planned screws were aborted and placed manually, mostly due to poor registration or trajectory issues in the thoracic > lumbar > sacral or iliac level. Fifteen screws were not placed, at the surgeon’s discretion. The authors mentioned difficulties in cases with severe obesity (BMI 49.9 kg/m ² ) and deformity, precluding adequate fluoroscopic image registration. Even though it seems from this report that in deformity and revision cases where the normal anatomical landmarks are obscured the robot is helpful to improve screw accuracy, it must be acknowledged that in studies assessing screw accuracy based on postoperative CT scans the malposition rates are about four times higher than are the rates obtained by clinical estimation.

In 2016, Macke et al. published their series of n = 50 pediatric patients (n = 662 screws) with adolescent idiopathic scoliosis (AIS). The overall accuracy (G&R grades A&B) with the Mazor Renaissance was 92.8% (3% medially breached), and in cases with intraoperative prone position CT scanning, the inaccuracy rate could be reduced to 2.4% (0% medially breached).

The largest comparative study analyzing clinical and radiological outcomes after robotic (Mazor Renaissance; n = 83 patients, n = 1012 screws) versus 3D-printed patient-specific template (n = 75 patients; n = 886 screws) or CT-based navigation (n = 109 patients; n = 1276 screws) in a population of n = 267 patients diagnosed with severe adult degenerative scoliosis was published in 2018. The authors found a significantly higher rate of good (G&R grades A and B) screw position in robotic (96.0%), compared with both template- (90.6%; P < .001) or CT-navigation-based procedures (93.0%; P = .019). The G&R grades for robotic-assisted screws in this challenging population were 91.3% (A), 4.7% (B), 2.8% (C), 0.6% (D), and none were labeled E. Six screws in the robotic group versus n = 30 in the template- (P < .001) and n = 20 in the CT-navigation groups (P = .029) were revised manually from the original device guidance. Pedicle encroachment was lateral in all n = 88, highlighting that “tool skiving” was likely the inciting issue with mal-positioned screws. Significant violation of the proximal facet joint was not found in any of the robotic screws.

Accuracy of Spino–pelvic Screw Implantation

Achieving solid fixation and fusion across the lumbosacral junction is challenging. The tri-cortical S2 alar-iliac (S2Ai) trajectory finds increasing popularity as a method of low-profile pelvic fixation to reduce strain on the S1 screws and replace classic iliac bolt fixation. The path of these screws is surrounded by major neural and vascular structures and the use of robots in obtaining proper S2Ai trajectories has been reported in several publications over the last decade, providing evidence that robotic-guided S2Ai screw placement is feasible and accurate (summary provided in Table 8.2 ).

Table 8.2

Review of the current literature, summarizing articles published until 07/2019 that report on the safety and accuracy of robotic-guided S2Ai screw placement for ASD and related spinal disorders.

Authors, year	Study design	Number of patients/screws/type of robot	Assessment of accuracy	Assessment of safety	Assessment of feasibility
Bederman et al., 2017	Retrospective case review	14 consecutive patients; 31 S2Ai screws; Mazor Renaissance®	Review of postoperative radiographs and CT; 45.2% accuracy; G&R type A breach in n = 10 (32.2%), type B in n = 1 (3.2%), and type C in n = 6 (19.4%)	No complications related to S2Ai placement occurred.	Difficulties with the screw simulation software, limited to 60-mm screws. Need to probe manually for all screws placed at depth, as robotic drill capacities (28 mm) were exceeded.
Hyun et al., 2017	Retrospective case review	Four patients; 8 S2Ai screws; Mazor Renaissance®	Review of postoperative CT; 100% accuracy	No complications related to S2Ai placement occurred.	No intraoperative difficulties occurred.
Hu and Lieberman, 2017	Retrospective case review	18 non-consecutive patients; 35 S2Ai screws; Mazor Renaissance®	Review of postoperative CT; 100% accuracy	N/A	No intraoperative difficulties occurred.
Laratta et al., 2018	Retrospective case review	23 consecutive patients; 46 S2Ai screws; Mazor Renaissance®	Review of intraoperative O-Arm (CT); 95.7% accuracy (each one breach <3 mm and >6 mm)	No complications related to S2Ai placement occurred.	All planned screws were deployed. No difficulties were reported.
Shillingford et al., 2018	Retrospective matched cohort study	68 consecutive patients (105 S2Ai screws); 23 in robot (46 S2Ai screws) and 28 in free-hand group (59 S2Ai screws); Mazor Renaissance®	Review of intraoperative O-Arm (CT); 95.7% accuracy in robot (2 breaches) and 91.5% in free-hand group (5 breaches; P = 0.463)	No significant intraoperative neurovascular or visceral complications associated with S2AI screw placement occurred in either group.	N/A

ASD , Adult spinal deformity; G&R , Gertzbein and Robbins; N/A , not available; S2Ai , S2 alar-iliac.

In 2017, Laratta et al. used the bone-mounted Mazor Renaissance system to guide S2Ai screw insertion (mean screw length, 80 ± 2.6 mm; mean diameter, 8.5 ± 0 mm). In a retrospective series of 23 consecutive patients treated between January and September 2016, the authors analyzed intraoperative 3D O-arm scans to determine the severity and direction of breach. For the S2 sacral alar corridor, their blinded evaluation revealed n = 2 breaches of the iliac cortex (each one anterior and posterior), of which one was moderate-severe (G&R type E; >6 mm). The overall accuracy was 95.7% for n = 46 S2Ai screws, the robotic-assisted screw placement was safe, as no complications related to the placement occurred in their series.

Also in 2017, Bederman et al., using the same system (Mazor Renaissance) to guide S2Ai screw insertion (mean screw length 80, range 65–90 mm), reported on their retrospective series of 14 consecutive patients and n = 31 screws. According to the G&R scale, they found type A breach in n = 10 (32.2%), type B in n = 1 (3.2%), and type C in n = 6 (19.4%), resulting in a lower overall accuracy rate of 45.2%. Of note, none of the misplacements was intrapelvic, risked visceral or neurovascular structures or needed to be removed or revised. The authors mentioned difficulties with the screw simulation software, which was limited to 60-mm screws at that time. As such, only screws >75 mm were found to breach beyond the bone cortex. Bederson et al. needed to probe manually for all screws placed at depth, as the robotic drill capacities (28 mm) were exceeded.

Those difficulties were overcome in the case series of Hyun et al., who implanted eight S2Ai screws (8.5-mm diameter; 80- or 90-mm length) in four adult patients undergoing corrective surgery for spinal deformity. Those authors used the Mazor Renaissance as guidance, after a software update providing a more expansive view of the pelvis for up to 80-mm screw trajectories. Applying a guiding technique with Jamshidi needles and K-wires beyond the 28-mm pilot hole, they found all eight screws (including two of 90 mm in length) to be placed accurately without cortical violation in postoperative CT studies. No intra- or postoperative complications occurred. The net mean time of the use of the robot was 13 min for screw placement utilizing 5.3 s of fluoro per screw on average.

A further article from 2017 comes from Hu and Lieberman, who summarize their experience with n = 35 Mazor Renaissance-guided S2Ai placed screws (length and diameter not provided). It should be noted that the authors’ series originally comprised 34 patients, but 16 of those were excluded as no postoperative CT data was available. Reviewing the postoperative CT scans in 18 patients with degenerative scoliosis (n = 12), flat back syndrome (n = 2) or other diagnoses (n = 4), the authors found no breach in any of the placed screws. The average deviations from the actual screw to the treatment plan were in the ranges of 3 mm (axial plane) and 1.8 mm (lateral plane) at the entry point and 2.1 mm (axial plane) and 1.2 mm (lateral plane) at a depth of 30 mm. The authors mentioned the possibility of slight lateral skiving of the drill at the irregular and oblique sacral surface at the typical S2Ai entrance point, a deviation they considered minimal and acceptable. No difficulties with the robotic-guided S2Ai screw placement were encountered intraoperatively.

Lastly, Shillingford et al. performed a retrospective matched cohort analysis to compare the accuracy of robotic guidance versus free-hand technique for S2Ai screw placement.

Rates of Intraoperative Complications

Corrective surgeries for ASD are per se associated with extraordinarily high complications rates. Any technology with the potential to lower the commonly encountered intra- and postoperative complications is of potential interest to spine surgeons.

Kantelhardt et al. found a lower rate of intraoperative complications including major hemorrhage and dural tears in robotic- compared to fluoroscopy-guided surgical procedures (4.7% vs. 9.1%, P = .04) in their retrospective series of n = 112 consecutive patients undergoing spinal fusion for a wide range of indications. The differences in postoperative complications, including wound healing disturbances (13.5% vs. 21.4%), surgical site infections (SSIs; 2.7% vs. 10.7%), and cerebrospinal fluid fistulas (0.0 vs. 6.1%) were in favor of robotic-assisted procedures but insignificant.

Even more impressive interim findings with a five-fold decrease in surgical complications ( P < .001) and a seven-fold decrease in revision surgery ( P = .012) were announced for the robotic arm of the MIS ReFRESH, a prospective controlled multicenter study assessing the clinical impact of robotic guidance compared to fluoroscopic guidance. For now, the final results of this study have not been announced, but it should be considered that the trial enrolled patients undergoing short (<4 segments) spinal fusion procedures and the results might therefore not be applicable 1:1 to the typical ASD population.

In a large, comparative analysis of n = 276 ASD patients, no marked differences between robotic- and both template- and CT-navigation-guided procedures were found with regard to estimated blood loss; length of postoperative stay; revision surgeries; and intra- and postoperative complications including dural tears, SSIs, wound revisions, and neurological complications.

Proximal Facet Violation

The exact planning and execution of the pedicle screw trajectory on the uppermost instrumented vertebrae (UIV) included in long fusion constructs theoretically permits the surgeon to spare the proximal facet joint from mechanical damage. This may result in biomechanically superior results, with the possibility to translate into lower rates of adjacent segment disease and failure in the long term after ASD correction. So far, no major differences between robotic versus other sophisticated guiding tools were found in a comparative study on ASD patients. More patient data with long-term follow-up is required to shed light on this potentially valuable benefit of robotic surgery.

Postoperative Pain, Disability, and Health-Related Quality of Life

One report indicates lower postoperative pain levels and a decrease in postoperative opioid use for robotic-assisted procedures, potentially translating into reduced opioid-associated morbidity. However, the added benefit of the robotic technique could be biased by a higher proportion of percutaneous MIS versus open cases in the non-robotic group. So far, the authors are unaware of comparative studies comparing subjective or objective disability or HRQoL outcomes after robotic versus non-robotic ASD surgery; however, the VAS pain, Oswestry Disability Index (ODI), and EuroQol (EQ-5D) questionnaires were included as secondary outcome measures of the MIS ReFRESH study.

Economical Outcomes and Cost-Effectiveness

Currently, to the best of the authors’ knowledge, there is no data on the economical outcomes or cost-effectiveness of robotic-assisted spine surgery, and in particular not for ASD. The high acquisition costs of robotic OR technology might be offset by a reduction in OR time, fewer revision procedures for misplaced screws, and lower rates of adjacent segment disease, but this remains to be proven by large studies with sufficiently long follow-up intervals.

Radiation Exposure

Spinal deformity procedures require thorough knowledge and information about the anatomy, the orientation of the vertebrae in 3D space and their relationship to the underlying and adjacent neurovascular structures. In particular in cases with rotational deformity, many surgeons rely on either intraoperative 2D C-arm fluoroscopy or 3D CT imaging guidance to achieve optimal screw position, which can result in significant radiation exposure to both patients and surgeons alike. With robotics systems, CT and/or fluoroscopy can be acquired with the surgeon and OR staff outside the OR. The procedure may then proceed without additional intraoperative scans. Accordingly, savings in fluoroscopy time and radiation dose have repeatedly been demonstrated for robotic-assisted spinal fusion procedures.

Even more than in adults, minimizing radiation exposure is paramount in the pediatric population, as a potential correlation between exposure and cancer incidence was pointed out. With most robotic systems relying on a limited number of initial 2D or 3D images for registration, the reduction in radiation exposure during AIS and other forms of pediatric deformity surgery is a major advantage of these systems. Even though data is currently lacking for ASD correction, it can be assumed that those savings for the surgeon and OR staff would be similarly impressive for ASD procedures that involve a higher number of spinal implants. As the thin-sliced high-resolution CT scans that most robotic systems require are generally requested as part of most deformity surgeons’ preoperative assessment, the overall radiation exposure for the patient in robotic-assisted spinal ASD correction is likely also less, compared to the additional fluoroscopic radiation exposure.

Revision Rates

A limited number of studies for non-ASD populations exist and uniformly report lower revision rates for robotic comparted to conventional (free-hand) fusion procedures.

In their series of n = 72 robotic MIS fusion cases for spondylolisthesis, Schröder and Staartjes described no need for intraoperative screw reposition, conversion to open procedure using the Mazor SpineAssist system. Their 1-year revision rate was n = 3 (4.2%); none of those revisions were related to screw malposition. In a meta-analysis of the existing literature, the authors found that patients undergoing free-hand fusion procedures were eight times as likely as were patients undergoing robotic fusion procedures to require revision surgery for screw malposition (OR, 8.1; 95% CI, 2–33.3; P < .001). Navigated fusion cases had a comparative revision profile to robotic fusion cases.

In a preliminary report from the MIS ReFRESH study group after the enrollment of 250 patients in the robotic and 79 patients in the fluoro-guided study arms and a minimum of 1-year follow-up, the odds ratio for a revision surgery were 6.4 times lower in the robotic group (2 [0.8%] vs. 4 [5.1%]; P = 0.031).

In their analysis of adult scoliosis cases, Fan et al. found similar revision rates after robotic versus CT-navigated procedures (2/83 [2.4%] vs. 4/109 [3.7%)] P = .512), including zero revisions due to screw malposition in the robotic group and two in the navigated group. Unfortunately the follow-up time remains unclear from their publication.

OR Time

Given its relatively short time of clinical use, studies assessing the impact of robotics systems on surgical efficiency are rare for spine surgery in general and non-existent for ASD. Depending on the number of instrumented spinal segments, multiple registrations will first add to the OR time. However, it can be inferred that, based on the existing literature for image guidance and navigation and considering the increased accuracy of robotics systems, there may be a positive impact on workflow and total OR time. As with the implementation of any new technology, a learning curve exists that, once overcome, may lead to greater surgical efficiency.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here