Adverse Events Associated With Deep Brain Stimulation for Movement Disorders: Analysis of 510 Consecutive Cases

Abstract

Background

Although numerous studies have focused on the efficacy of deep brain stimulation (DBS) for movement disorders, less is known about surgical adverse events, especially over longer time intervals.

Objective

Here we analyze adverse events in 510 consecutive cases from a tertiary movement disorders center at up to 10 years postoperatively.

Methods

We conducted a retrospective review of adverse events from craniotomies between January 2003 and March 2013. The adverse events were categorized into two broad categories –immediate perioperative and time-dependent post-operative events.

Results

Across all targets, perioperative mental status change occurred in 18 (3.5%) cases, and symptomatic intracranial hemorrhage occurred in 4 (0.78%) cases. The most common hardware-related event was skin erosion in 13 (2.5%) cases. The most frequent stimulation-related event was speech disturbance in 16 (3.1%) cases. There were no significant differences among surgical targets with respect to the incidence of these events. Time-dependent postoperative events leading to revision of a given DBS electrode for any reason occurred in 4.7±1.0, 9.3±1.4, and 12.4±1.5 percent of electrodes at 1, 4, and 7 years post-operatively, respectively. Staged bilateral DBS was associated with approximately twice the risk of repeat surgery for electrode replacement versus unilateral surgery (p=0.020).

Conclusion

These data provide low incidences for adverse events in a large series of DBS surgeries for movement disorders at up to 10 years follow-up. Accurate estimates of adverse events will better inform patients and caregivers about potential risks and benefits of surgery and provide normative data for process improvement.

Introduction

Deep brain stimulation (DBS) has become a routine therapy for patients with disabling symptoms of movement disorders. As DBS is proposed both for earlier stages of Parkinson's disease and for new indications in neurology and psychiatry^1–6, it is increasingly important to understand the risk for adverse events associated with the surgical intervention. Although considerable work has documented DBS efficacy across a variety of movement disorders, relatively less is known about adverse events, because clinical trials are typically conducted over a short duration and serious adverse events are relatively infrequent^1,7–13. Prior studies have shown that serious adverse events are more common with simultaneous bilateral DBS for Parkinson's disease (PD) versus best medical therapy alone at up to 12 months follow-up^14,15. We and others have argued that unilateral DBS surgery followed by staged bilateral surgery (if or when needed) provides significant clinical benefit for most PD patients and may spare risk; however, there are few published data comparing adverse events associated with unilateral versus bilateral surgery^16–18.

Beyond evaluating different surgical approaches, better characterizing adverse events in a large sample of DBS patients over longer time intervals can clarify the pre-operative risk/benefit assessment and provide normative data for process improvement. Here we describe surgical adverse events from DBS in a consecutive series of 510 unilateral and staged bilateral craniotomies for Parkinson's disease, essential tremor, and dystonia over 10 years of routine care at a tertiary movement disorders center.

Methods

We performed a query of our Institutional Review Board approved outcomes database to evaluate every DBS electrode placement at the University of Alabama at Birmingham Hospital between January 1, 2003 and March 1, 2013, yielding 510 craniotomies and 1,020 total procedures. Informed consent was not obtained individually from patients because this was a de-identified, retrospective review of an outcomes database designed for quality improvement. The surgical procedures analyzed in this study included insertion of the DBS electrode followed by placement of the pulse generator beneath the clavicle and its connection to the DBS wire on approximately the seventh post-operative day. We reviewed patient demographics and identified both peri-operative adverse events (including visual inspection of all post-operative brain MRIs) and time-dependent complications (infection and electrode repositioning). Procedure-related adverse events included intracranial hemorrhage and subdural hematoma (ICH and SDH, respectively, either symptomatic or asymptomatic), air embolus, intraoperative or postoperative seizure, cerebrospinal fluid leak, mental status change, and pneumonia. ICH and SDH were identified from review of post-operative imaging by an independent radiologist and a neurosurgery resident. Hardware-related adverse events were subdivided into subgroups including hematoma and seroma, lead fracture, skin erosion, and infection. The stimulation-related adverse events were categorized from follow-up clinic visits and include speech disturbance (dysarthria or hypophonia), ballism (abnormal swerving or jerking movements), eyelid apraxia (difficulty with eye opening and visual problems), and corticospinal effects (problems with voluntary movement of one side of the body or extremities).

All procedures were performed by the same neurosurgeon, using previously published methods¹⁶. The placement of DBS electrode was guided by MRI frame-based stereotaxy using the CRW frame system in conjunction with localization software. The vast majority of patients were awake and responsive throughout the procedure, and microelectrode recordings (MERs) were used to refine targeting in virtually all subthalamic and pallidal cases, and in a minority of ventral intermediate thalamic surgeries for essential tremor. In a few instances, pallidal electrodes were placed under general anesthesia without MERs in children or adolescents with generalized dystonia. We passed one MER electrode per recording trajectory and typically performed between 1 and 3 passes per case (mean number of passes 2.1 ± 1.2; median 2). Postoperative volumetric MRI was obtained within 24 hours of the procedure on a per protocol basis. The pulse generator was placed beneath the clavicle and connected in a separate procedure one week later, using either general anesthesia or conscious sedation. We exclusively performed unilateral placement of the DBS electrode followed by contralateral surgery, if and when it was clinically indicated, as described previously¹⁶. All patients underwent a single channel IPG placement with each electrode, and possible contralateral electrode followed by contralateral single channel IPG. Neurologists who are involved in patient’s care programmed the devices.

The decision and performance of the staged bilateral implantation is based on individual patient and the type of movement disorder. Our surgical approach is a tailored to the individual patient, based upon their personal preference in the context of a risk/benefit assessment provided by the treating clinicians. In our experience, patients with essential tremor generally remain unilateral, especially if the stimulator is placed contralateral to their dominant arm. Some patients with severe essential tremor elect to proceed with staged surgery if their tremor is especially prominent or bothersome in the unstimulated arm or if they have significant midline tremor (head, neck, voice). In patients with generalized and cervical dystonia, bilateral stimulators are usually needed for optimal motor improvement, so we typically advise the staged procedure within the first few months of the initial surgery after optimization of their unilateral stimulator during postoperative programming. Our view is that this simplifies postoperative programming in dystonia patients and may give a more nuanced understanding of the effects of one stimulator on a given patient’s motor symptoms. Patients with Parkinson’s disease vary considerably. Many PD patients with asymmetric motor symptoms have marked motor improvement for greater than 5 years with only unilateral stimulation, but because of the more progressive nature of the disease, many or most patients eventually elect to proceed with the staged contralateral procedure. Despite this, our prior work demonstrated that only 32% of unilateral PD patients underwent the staged bilateral procedure within 2 years of their initial surgery, and importantly, the subgroup of patients who remained unilateral retained significant improvements in the UPDRS parts 2, 3, and 4 versus their preoperative baseline¹⁶. For all patients, the most important motivation for staged placement of a second DBS electrode is to improve residual or progressive motor disability that is severe enough to significantly impair activities of daily living despite optimal medical management.

We performed descriptive statistics evaluating age, gender, and adverse events across the three surgical targets (subthalamic nucleus, globus pallidus interna, ventral intermediate thalamus). The incidence of peri-operative events was evaluated as a mean probability ± standard error, and Fisher’s exact test compared the frequency of adverse events by target, regardless of the underlying diagnosis. For time-dependent post-operative adverse events (infection and electrode repositioning/replacement), we used Kaplan-Meier survival analyses regardless of surgical target or diagnosis with the log rank mean test. The significance threshold for all statistical tests was p=0.05.

Results

We evaluated 510 consecutive unilateral and staged bilateral DBS electrode placements in 313 male (61.4%) and 197 female (38.6%) patients aged 59.1 ± 14.3 years over an average follow-up duration of 49.3±1.4 months. The most common DBS target was the subthalamic nucleus (STN) (n=270, 52.9%), followed by the ventral intermediate thalamus (VIM) (n=140, 27.5%) and the globus pallidus interna (GPI) (n=100, 29.6%) (Table 1).

Table 1.

Demographics

		Target
	Total	GPI	STN	VIM	P-value
Number of cases n (%)	510 (100.0)	100 (19.6)	270 (52.9)	140 (27.5)
Male	313 (61.4)	42 (42.0)	179 (66.3)	92 (65.7)	<0.0001
Female	197 (38.6)	58 (58.0)	91 (33.7)	48 (34.3)
Age at Surgery mean ± std	59.1 ± 14.3	48.2 ± 19.4	60.7 ± 9.8	63.8 ± 13.5	<0.0001

STN= subthalamic nucleus, GPI= globus pallidus interna, VIM= ventralis intermediate nucleus

p-value for gender was determined using chi-square statistic.

p-value for age was determined using F-statistic from Proc GLM (General Linear Model)

Adverse events in the perisurgical period were relatively uncommon. These events are itemized specifically in Table 2. Intraoperative adverse events related to electrode placement including intracranial hemorrhage, subdural hemorrhage, air embolus, and seizure occurred in 5.1 ± 1.0 % of cases. Intracranial hemorrhage was associated with permanent neurological symptoms in 4/510 electrode placements (0.78±1.83 %). Complications in the immediate postoperative period occurred in 6.1±1.0% of electrode placements, including seizure, spinal fluid leak, mental status change, and pneumonia. Acute or subacute mental status change was the most frequent immediate postoperative complication. The mental status change associated with surgery was present in 18 cases (3.5±0.8%). Hardware-related complications occurred in 4.9±1.0% of cases including hematoma or seroma, lead fracture, skin erosion, and infection. Only one subject was concerned with repeated infectious complications. The stimulation-related adverse events occurred in 5.7±1.03% of cases including speech disturbance, ballism, eyelid apraxia, and corticospinal effects.

Table 2.

Adverse events

	Total (n=510)		GPI (n=100)		STN (n=270)		VIM (n=140)
	No. of Electrodes	%	No. of Electrodes	%	No. of Electrodes	%	No. of Electrodes	%	P-value
Procedure
ICH	15	2.94	4	4.00	6	2.22	5	3.57	0.57
Symptomatic ICH	4	0.78	1	1.00	3	1.11	0	0	0.53
Asymptomatic ICH	11	2.16	3	3.00	3	1.11	5	3.57	0.19
SDH	9	1.76	1	1.00	5	1.85	3	2.14	0.91
Air embolus	1	0.20	1	1.00	0	0	0	0	0.20
Intraop_seizure	1	0.20	0	0	1	0.37	0	0	1.00
Post-op seizure	7	1.37	0	0	5	1.85	2	1.43	0.52
CSF leak	4	0.78	2	2.00	2	0.74	0	0	0.17
Mental status change	18	3.53	2	2.00	13	4.81	3	2.14	0.36
Pneumonia	2	0.39	0	0.00	1	0.37	1	0.71	1.00
Total	72	11.2±2.03
Hardware
Hematoma/seroma	4	0.78	0	0	4	1.48	0	0	0.32
Lead fracture	2	0.39	1	1.00	0	0	1	0.71	0.22
Skin erosion	13	2.54	2	2.00	3	1.11	8	5.71	0.02 *
Infection	6	1.18	3	3.00	0	0	3	2.14	0.008 *
Total	25	4.9±0.96
Stimulation
Speech disturbance	16	3.14	2	2.00	8	2.96	6	4.29	0.62
Ballism	6	1.18	0	0	6	2.22	0	0	0.08
Eyelid apraxia	2	0.39	1	1.00	1	0.37	0	0	0.43
Corticospinal effects	5	0.98	1	1.00	3	1.11	1	0.71	1.00
Total	29	5.7±1.03

^*Statistically significant at p < 0.05 indicated in bold font.

Unless otherwise noted, statistical significance was determined using Fisher's exact test.

Among surgical targets, skin erosion and infection were statistically significant at p < 0.05 (Table 2), with VIM target most likely to result in skin erosion (n=8, 5.71%, p=0.02) and GPI site most likely to result in infection (n=3, 3%, p=0.008). At our level of statistical power, there were no significant differences among surgical targets with respect to the incidence of any other adverse events.

Overall, electrode revision (infection, repositioning, device mechanical failure) occurred in 4.7±1.0, 9.3±1.4, and 12.4±1.5 percent of cases at 1, 4, and 7 years post-operatively (Figure 1). Electrode revision for both infection and repositioning (complete replacement) occurred at the highest rate during the first post-operative year (1.6±0.6% and 1.7±0.7%, respectively) but continued to accumulate at a slower rate thereafter (4.5±0.6% and 4.7±1.5% at 7 years). Adjustment of the DBS electrode position within the same trajectory (without complete electrode replacement) occurred less frequently and usually slightly later versus complete replacements of migrated or frankly malpositioned electrodes (0.7±0.3% at one year). The electrode replacement was typically performed to raise it dorsally to avoid capsular side effects, providing a larger number of DBS electrode contacts with a wider therapeutic window.

Figure 1.

Probability of DBS electrode revision over time. We separately characterize overall revision, device infection, complete reposition, and repositioning the same DBS electrode without complete electrode replacement. Revision for both infection and repositioning occurred at the highest rate during the first post-operative year but continued to accumulate at a slower rate over the subsequent years.

At our level of statistical power, there was no significant difference among surgical targets (p=0.19) or between brain hemispheres (p=0.14) with respect to the probability of revision for location, nor were there differences in the probability of post-operative infection (p=0.09 and p=0.42, respectively). We were typically unable to salvage the intracranial lead when the IPG or the connector wire became infected. The infection of either the connector wire or IPG in most cases would track up and cause problems of the main intracranial lead and increase the risk of intracranial infection. Therefore, intracranial lead was difficult to recover.

Furthermore, staged bilateral DBS is associated with electrode revision (infection, repositioning, and device mechanical failure) in 7.5±2.7, 16.3±3.8, and 25.5±5.7 percent of cases at 1, 4, and 7 years post-operatively (Figure 2). At our level of statistical power, there was a statistically significant difference in the probability of surgery for DBS electrode revision between staged bilateral and unilateral DBS surgery (p=0.020).

Figure 2.

Probability of DBS electrode revision over time for staged bilateral and unilateral DBS surgery. We characterize overall revision between these two techniques and show that bilateral DBS is associated with more long-term surgical adverse events than unilateral DBS and the difference between the unilateral and bilateral DBS electrode repositioning is statistically significant.

Discussion

Serious adverse events from DBS surgery are uncommon, but their occurrence underscores specific risks associated with this elective procedure. In this large, comprehensive sample from a single tertiary center, we provide several new findings regarding adverse events associated with DBS surgery. First, we provide evidence that staged bilateral DBS is associated with more surgical adverse events than unilateral DBS alone. Additionally, we characterize adverse events across stimulation targets and diagnoses over a longer follow-up interval than prior studies, allowing estimates for additional adverse events that occur years after the initial electrode placement. Clinical trials focus on DBS efficacy over relatively short time intervals (6 to 12 months postoperatively), potentially leading to underestimates of the frequency of surgical adverse events over time^3,19–21 (Table 3). Furthermore, we characterize the three main categories of complications, including the procedure-, hardware-, and stimulation-related adverse events. The prior studies on adverse events often comment on selected subgroups of adverse events or focus more narrowly on one or two DBS targets in a single disease state (Table 3). More recently, Hamani et al evaluated hardware-related problems through a systematic review, but did not mention other surgical adverse events like many previous authors who focused on hardware adverse events^22–26. Also, Kleiner-Fishman et al, Videnovic et al, and Pollack et al performed a meta-analysis and commented on the prevalence of adverse events following DBS for PD, but they lacked sufficient investigation into several of the complications and the three main movement disorders^12,27,28 (Table 3).

Table 3.

Literature review of complications of DBS in series with more than 30 patients

Study’s information						Mor talit y	Procedure related adverse events									Hardware related events					Stimulation related adverse events
							All reported as percentages of total number of electrodes
Authors, year	Surgical approac h (unilater al, sim. bilateral, staged bilateral)	Patients, electrod es (n, n)	Design	Target/D iagnoses	Follow- up (months)	Mort ality %; caus e	ICH total	Sympto matic ICH	SDH	Air embolu s	Seizure	CSF leak	Mental status change	PNA	DVT/ PE	Hemato ma/sero ma	Lead fracture	Skin erosion	Infecti on	Total*	Speech disturb ance	Ballism	Eyelid apraxia	CS effects	Total**
Weaver et al., 2009 15	0, 121. 0	121, 121	Prospect ive RCT	STN, GPi; PD									10.7								9.9
Herzog et al., 2003 30	0, 48, 0	48, 48	Retrospe ctive	STN; PD	24	2.1, MI	2.1		2.1				14.6
Follett et al., 2010 31	0, 121	NR, 121	Prospect ive RCT	STN, GPi; PD	24	8.3, ICH, PNA , sepsi s, MI	3.1						8.0	8.0					15.4		63
Kondziol ka et al., 2002 23	0, 66	66, 66	Retrospe ctive multicen ter	VIM; Tremor	29												15.2	4.5	10.6	30.3
Carlson et al., 2014 21	3, 56	62, 62	Retrospe ctive	STN, GPi; PD	1								22
Binder et al., 2005 35	89,92,99	280, 481	Retrospe ctive	All	24		3.3	0.6
Present study	282, 0, 110	392, 510	Retrospe ctive	All	49.3	0.2, ICH	2.94	0.78	1.76	0.20	1.57	0.78	3.53	0.39	0.00	0.78	0.39	2.54	1.18	4.89	3.14	1.18	0.39	0.98	5.69
Voges et al., 2007 19		993, 1183	Retrospe ctive multicen ter	All	1	0.4, PNA , PE, MS	2.2	0.8						0.6	2				0.4
Videnovi c et al., 2008 12		928, 1154	Systema tic review	STN, GPi; PD			6.3	3.8			1.6	0.2	27.7			3.1	0.7	1.2	4.6	9.6	5.4	0.2	0.7		6.3
Kleiner- Fishman et al., 2006 28		921, NR	Systema tic review & meta- analysis	STN; PD	14.8		3.9				1.5		15.6		0.3				3.6		9.3	2.6	3.6	4.0	19.5
Hamani and Lozano et al., 2006 24		856, 1491	Systema tic review	All													5.0	1.3	6.1	7.4
Sillay et al., 2008 42		420, 1374	Retrospe ctive	All	6														4.5
Pollack et al., 2002 27		300, 515	Literatur e review	All; PD		0.4, PE, ICH	1.7	1.0	0.6		0.19		8.3					1.4	1.9				4.08
Pepper et al., 2013 26		273, 519	Retrospe ctive	All															3
Burdick et al., 2010 20		198, 270	Retrospe ctive	All	6	0.7, NR		5.3	3.7	2	5.3		17.7			0.3	0.7	0.7	4.0	5.4	8.0		1.3	6.7	16.0
Constant oyannis et al., 2005 25		144, 203	Prospect ive RCT														1.4	1.4	7.6	10.4
Limousin et al., 1999 32		110, 135	Retrospe ctive	STN, VIM; PD	12	0.0	2.2												2.2
Umemura et al., 2003 34		109, 179	Retrospe ctive	All	20	1.1, PE, PNA	1.1		1.1		0.5	0.6		0.5	0.5			0.6	2.2
Beric et al., 2001 29		86, 149	Retrospe ctive	All			2.1				1.34		4.7						1.1		1.34		1.34		2.68
Lyons et al., 2004 36		81, 160	Retrospe ctive	All; PD	17			0.6			0.6						1.3	0.6	1.9	3.8
Oh et al., 2002 22		79, 124	Retrospe ctive	All	12												5.1	2.5	15.2	22.8
Hariz et al., 2008 10		69, 69	Retrospe ctive multicen ter	STN, GPi; PD	48																26.6		15.6	15	57.2
Starr et al., 2002 33		44, 76	Retrospe ctive	All; PD	20	0.0	2.6				2.6						1.3		2.6
Kupsch et al., 2006 3		40, 40	Prospect ive RCT	GPi; dystonia	3						2.5		2.5				2.5	2.5	10.0	15.0	12.5
Joint et al., 2002 11		39, 79	Retrospe ctive	All	12.4												5	2.5
Schuurm an et al., 2000 6		34, 34	Prospect ive RCT	All	6	2.9, ICH										2.9			2.9		20.6		2.9		23.5

NR= not reported, Sim.= simultaneous

PD= Parkinson’s disease, STN= subthalamic nucleus, GPi= globus pallidus interna, VIM= ventralis intermediate nucleus, IPH=intraparenchymal hemorrhage, MI=myocardial infarction, PE=pulmonary embolism, DVT= deep venous thrombosis, PNA= pneumonia, CS= corticospinal; All= all three targets (STN, VIM, GPi) & multiple indications including movement disorders (essential tremor, dystonia, Parkinson’s disease)

^*Total = total number of hardware events summed if more than three specific sub-categories present

^**Total = total number of stimulation related events summed if more than two specific sub-categories present

Color coding is based on surgical approaches. Light grey is simultaneous bilateral. Medium grey is simultaneous bilateral + staged bilateral. Dark grey includes staged bilateral.

Procedure-related adverse events

Among procedure-related adverse events, changes in mental status or behavior occurred in 18 cases (3.5±0.8%). Although this tended to occur most commonly in STN cases, the STN target exclusively was used in patients with advanced PD, whereas a sizeable number of GPi and VIM cases involved patients with essential tremor and dystonia. These results are much lower than several studies in the previous literature that quote the prevalence of altered mental status from 4.7 – 27.7% in unilateral DBS^{12,20,27–29}(Table 3). Furthermore, our study’s findings demonstrate a lower rate of mental status changes compared to the 10.7-14.6% and 8.0 – 22.0% published rate of encephalopathy after surgery with the simultaneous bilateral procedure^15,30 and staged or simultaneous bilateral DBS^21,31, respectively. One possibility for this difference might be the difficulty in capturing all symptoms of encephalopathy such as anxiety, apathy, mood disturbances, or memory or cognition decline from the chart review. Another likely explanation is that our analysis includes unilateral or staged bilateral DBS procedures, instead of simultaneous bilateral procedure as discussed in the prior study.

Comparing intra-operative events, intracranial hemorrhage occurred in 15 (2.9%) cases; however, in only 4 (0.8%) of these cases, patients experienced long-term neurological symptoms. These findings are within the ICH incidence range of 1.1– 6.3% published in retrospective reviews consisting of sample sizes between approximately 30 and 200 patients^{19,29,30,32–35} and similar to 3.1% ICH rate of a prospective randomized controlled trial³¹. The symptomatic ICH rate of 0.8% is also similar to 0.6% results of Binder et al³⁵ and 0.8% from a multicenter surveillance by Vogue et al¹⁹ (Table 3). DBS target and underlying diagnosis were not associated with differential risk for intracranial hemorrhage at our level of statistical power; however, even with our relatively large sample of more than 500 electrode placements, there were very few hemorrhages.

Hardware-related adverse events

Systematic studies on hardware-related adverse events for all three DBS targets are rare. Although our rate of overall hardware-related problems of 4.9% is less than those in previous studies (5.4 to 22.8%)^{3,6,12,22–25,36}, this may be because we did not include IPG malfunction, lack of benefit premature loss of battery power, or abnormal healing among our hardware-related problems.

Skin erosion or breakdown of the skin over the lead or the battery was most commonly seen with the VIM surgical target and skin infection was mostly present in GPI site in terms of percentage. In analyzing the actual number of cases, however, it is important to note that the absolute number of infections were equal between VIM and GPI. The cause of the relationship between skin erosion and each target is difficult to pinpoint. It is most likely secondary to multiple factors, including the underlying medical condition, age, and the type of the movement disorder. One potential explanation for this is that the VIM patients primarily suffer from essential tremor and are significantly older in age compared to the patients in STN target. The older patients are more likely to have numerous comorbidities, compromised skin integrity³⁷, increased skin fragility, and rheological abnormalities,³⁸ possibly making them more prone to skin erosion and infection, especially in the background of tremor activity. Furthermore, GPI patients suffer from dystonia and are in poorer condition. Nevertheless, our results lack confidence due to the presence of too few events.

Stimulation-related adverse events

The most common stimulation-associated effect is speech disturbance. It occurred in 16 cases (3.14%). Eyelid apraxia was present in 2 cases (0.39%). Due to the wide variations in the report of these events, it is difficult to compare these prevalences to the previous studies. Nevertheless, our findings are much lower than majority of the previously published rates of 10–45% ocular and speech disturbance^6,10,20,28. This could be attributed to the fact that stimulation-related adverse events are confounded by natural disease progression and medical therapy. One other limitation of our stimulation-related adverse events is the lack of report on weight gain. This adverse effect is well established in the literature^39,40. However, it is difficult to obtain data on weight gain from retrospective chart review because it is not measured or reported systematically.

Time-dependent post-operative adverse events

Overall, electrode replacement or adjustment occurred in 4.7±1.0, 9.3±1.4, and 12.4±1.5 percent of cases at 1, 4, and 7 years post-operatively. Although lead revisions for device infection and repositioning occurred at the highest rate during the first postoperative year, these events continued to accumulate at a slower rate at up to 10 years after the original surgery. This suggests that the prior literature underestimates the probability for these adverse events in real-world clinical care. Although symptomatic intracranial hemorrhage occurred less frequently than infection or electrode repositioning, it can be associated with permanent neurological disability, underscoring perhaps the most important risk associated with DBS.

Our infection rate of 1.2% at 6 months and 1.6% at one year is largely consistent with the published literature^24,41,42. The risk for device infection diminished over time, in contrast to a prior report suggesting that infection is more common following subsequent implanted pulse generator (IPG) replacement²⁶. Early infections occurred most often around the IPG, while later infections were typically erosions of the wire through the surface of the skin. We were typically unable to salvage the intracranial lead when infection of the IPG or connector wire occurred, unlike previous report by Sillay et al⁴². A potential advantage of implanting single channel IPGs is that revision of an infected system involves unilateral rather than bilateral craniotomy for replacement of the system, in contrast to infection of dual channel devices⁴³. However, placement of single channel devices likely results in more frequent IPG replacements for battery expiration in patients with bilateral devices, although there is evidence that Soletra devices typically have slightly longer IPG longevity than the Kinetra dual channel devices⁴⁴.

DBS repositioning surgeries within the first 6 months post-op were typically complete replacement of malpositioned or migrated DBS electrodes, driven by lack of efficacy or unacceptable side effects at low stimulation thresholds. The criteria were always clinical need first, followed by radiographic evaluation. If the postoperative brain MRI suggested that the DBS was in the wrong position and was consistent with side effect thresholds or clinical effects in a particular patient, that would provide further confidence that electrode repositioning was warranted. In the rare instances where the DBS was well-tolerated and postoperative MRI demonstrated satisfactory electrode position, we typically considered the patients non-responders and did recommend electrode repositioning.

Reasons for later repositioning surgeries were more heterogeneous, tending to include minor adjustment of the original DBS electrode within the same trajectory to further improve efficacy or tolerability in the contexts of superimposed disease progression, tolerance/habituation, or stimulation side effects upon attempts at DBS programming. The authors readjust the original electrode within the same trajectory varying the depths of insertion with a micrometric precision. The loss of partial therapeutical effect was based on clinical exam, and variations in depths of the electrode were correlated to ensure there was no symptom rebound.

Unilateral versus staged bilateral DBS

Our standard surgical practice is to place a unilateral DBS electrode on the most affected side of the brain, followed by staged placement of a contralateral electrode if and when it is needed for symptomatic management. The results from this study demonstrate that staged bilateral DBS is associated with more long-term surgical adverse events than unilateral DBS, and our estimates for the risk of adverse events with a single electrode placement compare favorably to the risk associated with bilateral simultaneous electrode placement for PD in clinical trials (Table 3)^15,18,45. These findings are not surprising considering that an additional DBS electrode, connector wire, and battery are implanted in patients with bilateral systems. In patients whose motor symptoms and daily activities improve significantly with unilateral stimulation, simultaneous or immediate staged placement of a contralateral stimulator adds an incremental risk for potentially avoidable adverse events. Although the present data do not address the relative efficacy of unilateral versus bilateral surgery, our group and others have previously demonstrated that unilateral subthalamic DBS provides significant improvement in the UPDRS parts 2, 3, and 4 in patients with advanced PD at up to two years follow-up^10,31,45. In patients who undergo unilateral DBS and retain sufficient clinical improvement with a single DBS electrode, it could be argued that not implanting the second contralateral electrode spares patients from the incremental surgical risk associated with having a bilateral system. Additionally, bilateral simultaneous DBS are known to significantly worsen speech and gait in some patients, as well^46,47. One prior retrospective review of twenty-two patients demonstrates that the rate of adverse events in the staged group (20%) was less than that of the simultaneous group (42%), although the difference was not statistically significant⁴⁸. The relative efficacy and tolerability of unilateral, staged bilateral, and simultaneous bilateral surgical approaches would best be addressed by a prospective, randomized study design.

Strengths

This study has a number of strengths. First, it is a relatively large series from a single center with extensive long-term follow-up, potentially allowing more accurate estimates of the frequency of relatively uncommon serious adverse events. Second, Kaplan-Meier analyses provide better time resolution for adverse events versus other methods that choose arbitrary endpoints for statistical comparisons. Third, rather than focusing on a single type of surgical adverse event (infection, hemorrhage, reposition), this study provides a more comprehensive survey of different types of events, providing a more global view of the risk/benefit assessment associated with the intervention. Also, our data provide the first evidence comparing adverse events between unilateral DBS with staged bilateral DBS. Finally, the data for this study are collected from a single institution where all procedures are performed by one neurosurgeon, ensuring consistency of data capture.

Limitations

These analyses have several potential limitations, as well. First, the probability estimate of DBS revision/replacement events at later follow-up dates may be less precise, because the average duration of follow-up for a given patient was approximately 4 years. This is mitigated in part by the Kaplan-Meier analyses, because patients without shorter follow-up are censored from the probability estimates at later time points. Second, some patients may have died for any reason during the follow-up period, potentially yielding falsely low probability estimates at longer follow-up intervals. Third, although the sample size is relatively large, our study reflects the practice patterns and surgical techniques of a single center that does not perform simultaneous bilateral electrode implants, potentially limiting the generalizability of the results. Despite being a single surgeon study, no trend or decline in complication rate was identified over time. Fourth, this is a retrospective review, potentially introducing selection bias on different surgical targets based upon our local practice patterns or on the population of DBS patients in our region.

Conclusion

Our findings provide a low estimate for the incidence of serious adverse events over time associated with unilateral and staged bilateral DBS for movement disorders. As DBS is proposed in earlier disease stages and for new indications, these findings provide normative data to clarify the risk-benefit assessment for this therapy that can markedly alter health-related quality of life.