Intracranial hemorrhage risk factors of deep brain stimulation for Parkinson's disease: a 2-year follow-up study.

OBJECTIVE: This study aimed to analyze the risk factors of intracranial hemorrhage (ICH) after deep brain stimulation (DBS) for idiopathic Parkinson's disease (PD).
METHODS: Patients who received DBS from March 2014 to December 2016 were retrospectively analyzed. The hemorrhage index was derived by combining the hemorrhagic volume and clinical manifestations of ICH. All patients with IHC were followed up for 2 years.
RESULTS: Computed tomography showed 13 events of ICH in 11 patients (nine cases in the subthalamic nucleus), including eight cases with symptomatic hemorrhage (seven cases in the subthalamic nucleus). Hemorrhage was characterized by intracranial hematoma in the electrode puncture tract. Male sex and hypertension were significant risk factors for ICH. Hemorrhage in the preferred puncture side was significantly higher than that in the non-preferred puncture side. The mean hemorrhage index was 2.23 ± 0.83 in 11 patients, and no permanent neurological impairment was found during the 2-year follow-up. The effect of DBS on motor symptoms was similar in patients with and without ICH.
CONCLUSION: Male sex and hypertension are risk factors of ICH after DBS in PD. The risk of hemorrhage on the first puncture site is significantly higher than that on the second puncture site.

Introduction

The long-term efficacy of deep brain stimulation (DBS) for treatment of Parkinson’s disease (PD) is widely recognized.[1-8] Approximately 9000 patients with PD have undergone DBS. There have been many surgical technique reports on how to reduce the risk of intracranial hemorrhage (ICH) in patients with PD who received DBS.[9,10] Therefore, many methods have been used to reduce the incidence and volume of ICH. Problems that need to be addressed are reducing the risk of ICH and neurological outcomes that patients will have after ICH. Therefore, this study aimed to determine risk factors and the long-term prognosis of ICH in patients with idiopathic PD who underwent DBS.

Materials and methods

Patients’ enrollment criteria and clinical data

From 1 March 2014 to 30 December 2016, 352 patients with PD were treated with DBS in Shanghai Changhai Hospital. Idiopathic PD in patients was diagnosed according to the United Kingdom Parkinson’s Disease Society Brain Bank diagnostic criteria,[11] in line with the current consensus of Chinese experts on DBS in PD.[12] This study was approved by the Medical Ethics Committee of Changhai Hospital. All of the patients who participated in this study agreed to sign informed consent.

We collected data, including sex, age, course of disease, and the scores of the preoperative Unified Parkinson’s Disease Rating Scale ([UPDRS] assessed by a qualified physician, and drug onset (med-on) and off (med-off) status were recorded), Mini-mental State Examination, Montreal Cognitive Assessment, Non-motor Symptom Questionnaire, and Hamilton depression scale. An electrocardiogram and blood pressure were monitored during the operation and within 3 days after operation. If systolic pressure was >150 mmHg or/and diastolic pressure >100 mmHg, the patient was diagnosed with hypertension. Intraoperative adjustment of the electrode location was recorded. A cranial computed tomographic (CT) scan was performed 6 days after the operation.

Surgical procedures and follow-up

A 3.0 T magnetic resonance imaging (MRI) scan without a frame was performed before the operation, including T1-weighted, T2-weighted, and susceptibility-weighted sequences. The Leksell G headframe and Surgiplan surgical planning system (Elekta AB, Stockholm, Sweden) were used. The Medtronic 3387 or PINS L302 (Medtronic, Inc., Minneapolis, MN, USA) was selected for the internal globus pallidus (GPi) target and the Medtronic 3389 or PINS L301 was selected for the subthalamic nucleus (STN) target. The subarachnoid space closure technique was used to reduce cerebrospinal fluid loss.[13] After the electrode arrived at the target, bio-glue was used to seal the skull bone hole, and then an external temporary stimulator was connected to perform the intraoperative macrostimulation test. We simultaneously recorded the efficacy of electrical stimulation and the adverse reaction threshold. After fixation of the electrode by Stimloc (Medtronic) or Leadloc (PINS), the scalp was sutured. A 1.5 T MRI scan was performed to confirm whether the electrode position was satisfactory and whether there was any ICH. An extension wire and implantable pulse generator (IPG) were then implanted under general anesthesia. The IPG stimulation parameters were set 1 month postoperation. Clinical motor symptom follow-up data included UPDRS-III scores (med-off, IPG-on [IPG was used]) at 3, 6, and 12 months, and 2 years postoperation.

ICH and hemorrhage index

Hemorrhagic volume was calculated for patients with ICH as follows: hemorrhagic volume (mL) = maximum cross-sectional length width × layer thickness × layer number/2. The hemorrhage index was derived according to the hemorrhagic volume and the severity of hemorrhagic symptoms. The scores were obtained according to the following grades: a hemorrhagic volume of 0 to 5 mL = 1, 5 to 10 mL = 2, 10 to 15 mL = 3, 15 to 30 mL = 4, and > 30 mL = 5. Clinical symptoms (e.g., lethargy, limb symptoms, mental symptoms) increased the index by 1 point. Another 1 point was added if surgical intervention was required.

Statistical methods

The paired t test was used for comparison of variables before and after hemorrhage. One-way ANOVA was used for analysis of variables between hemorrhagic and non-hemorrhagic patients. The significance of risk factors was analyzed by Fisher’s exact probability test. Statistical analysis was performed using SPSS 21.0 (IBM, Armonk, NY, USA). The value of α was set to 0.05 and thus P < 0.05 (double-tailed) was considered statistically different.

Results

Demographic information of the patients

A total of 352 patients with PD were implanted with 686 electrodes. The baseline information of the patients is shown in Table 1. ICH was confirmed by CT or MRI at 3 to 6 days after the operation in 11 patients (13 events). The mean hemorrhage index was 2.23 ± 0.83 in 11 patients. No anticoagulant and antiplatelet drugs were used before the operation in all patients with ICH, and no abnormal coagulation function was found before the operation. Eight patients had symptomatic hemorrhage and three had asymptomatic hemorrhage. The initial symptoms included headache in two patients, drowsiness in three patients, and hypodynamia or dysesthesia in two patients. Postoperative delirium and manic symptoms were found in one patient.

Table 1.

Demographic information of the patients.

	All patients	Hemorrhagic patients	Non-hemorrhagic patients	P value
Males	195	10	185	0.026
Females	157	1	156
Hypertension	103	8	95	0.003
Disease duration (years)	10.12 ± 3.38	12.91 ± 6.24	10.03 ± 4.24	>0.05
Age (years)	62.22 ± 6.08	64.91 ± 4.30	62.17 ± 7.93	>0.05
Preoperative improvement rate	52.51% ± 13.43%	54.32% ± 15.38%	51.52% ± 14.24%	>0.05
Electrode number	686	13	673
Single side	18	1	17
Bilateral side	334	10	324
STN	325 (633 electrodes)	9	316	>0.05
GPi	27 (53 electrodes)	2	25
Motor function
UPDRS-III score postoperatively	55.62 ± 12.32	60.73 ± 16.95	55.45 ± 15.81	>0.05
UPDRS-III score during on time	26.30 ± 9.10	27.27 ± 8.09	26.26 ± 11.80	>0.05
Non-motor function
MMSE	27.02 ± 11.34
MoCA	24.06 ± 3.60
NMSQ	17.3 ± 4.8
HAMD	20.26 ± 5.8

STN: subthalamic nucleus; GPi: internal globus pallidus; UPDRS: Unified Parkinson’s Disease Rating Scale; MMSE: Mini-mental State Examination; MoCA: Montreal Cognitive Assessment; NMSQ: Non-motor Symptom Questionnaire; HAMD: Hamilton depression scale.

Risk factors

There were 10 cases of ICH in male patients and one case in female patients (Table 1). The incidence of hemorrhage in men was significantly higher than that in women (P = 0.026). There was no significant difference in the mean age, course of disease, preoperative improvement rate, preoperative UPDRS-III score, and postoperative UPDRS-III score between patients with hemorrhage and patients without hemorrhage. Mean systolic blood pressure was 145 ± 15.52 mmHg (125–170 mmHg) and mean diastolic blood pressure was 89.64 ± 10.65 mmHg (75–108 mmHg). Eight patients had a systolic blood pressure >150 mmHg and/or diastolic blood pressure >100 mmHg, and three of them had a systolic blood pressure > 160 mmHg. The risk of ICH in hypertensive patients was significantly higher than that in non-hypertensive patients (P = 0.003). Among the 352 patients, 325 (633 electrodes) were treated with STN-DBS, nine of these patients had ICH, and the rate of hemorrhage was 2.7% (Table 2). Twenty-five patients (53 electrodes) were treated with GPi-DBS, two of these patients had ICH, and the rate of hemorrhage was 7.4%. Although the hemorrhage rate of GPi-DBS was higher than that of STN-DBS, there was no significant difference between these two targets.

Table 2.

Clinical data of patients who were treated in the STN.

No.	Sex	Age (years)	Course	Hypertension	Target	Immediatehemorrhage	Detection method	Bleedingposition	Pattern of hemorrhage	Hemorrhagic volume	Neurologicalsymptoms	Preferentialside	ElectrodeAdjustment	Hemorrhage index
1	M	64	25	Y	STN	No ICH, MRI	CT, 6th day	Left frontallobe + tip	2.19 × 2.62-cm high-density shadow	6.5 + 0.5 mL	Headache, loss of consciousness	L	N	3
2	M	64	12	Y	STN	No ICH, MRI	CT, 5th day	Left frontallobe + tip	Small high-density shadow	4.8 + 0.32 mL	None	R	Y (z axis)	2
3	M	68	7	Y	STN	No ICH, MRI	MRI, 6th d	Left frontal lobe + tip	2.7 × 2.2-cm high-density shadow	10.3 + 3.2 mL	None	L	N	3
4	M	73	9	N	STN	No ICH, MRI	CT, 5th d	Right basal ganglia	Small hematoma in right basal ganglia	1.02 mL	Delirium	R	N	1
5	M	62	19	Y	STN	No ICH, MRI	CT, 6th d	Left frontal lobe+right tip	Irregular peri-electrode High-density shadow	(R) 1 + (L)10.8 mL	R limb weakness	L	N	2 & 4
6	F	68	12	Y	STN	No ICH, MRI	CT, 5th d	Right frontal lobe + tip	Right cerebral horns Hematoma	2.42 + 0.54 mL	Headache	R	N	2
7	M	68	20	Y	STN	No ICH, MRI	CT, 3rd d	Bilateral puncture tract	Small hemorrhage in bilateral puncture tracts	(R) 2 + (L)0.4 mL	Drowsy	R	N	2 & 2
8	M	66	7	N	STN	No ICH, MRI	CT, 3rd d	Left frontal lobe + tip	Left frontal hematoma	2.94 + 0.1 mL	None	R	N	1
9	M	62	7	N	STN	No ICH, MRI	CT, 6th d	Left frontal lobe	Small peri-electrode hemorrhage and edema	2.7 mL	Drowsy	L	Y (y axis)	2

M: male; F: female; Y: yes; N: no; STN: subthalamic nucleus; ICH: intracranial hemorrhage; MRI: magnetic resonance imaging; CT: computed tomography; d: day; R: right; L: left.

Surgery-associated factors

Of the 11 patients with ICH, seven had ICH on the preferred puncture side. Two patients had ICH at the non-preferred puncture and two patients had bilateral hemorrhage. The volume of hemorrhage was as follows: 10 to 15 mL in two patients, 5 to 10 mL in four patients, and 0 to 5 mL in five patients. Neurological symptoms occurred in eight patients after ICH (see Table 3 for details). The hemorrhage index in the preferred side was calculated as 0 + 0 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 2 + 3+ 3 + 3 + 4 = 23, and that in the non-preferred side was calculated as 0 + 0 + 0 +0 + 0 + 0 + 0 + 0 + 0 + 0 + 1 + 2 + 2 +2 = 7. Hemorrhage in the preferred puncture side was significantly more harmful than that in the non-preferred side (P = 0.0174) (Figure 1). Thirty electrodes were adjusted in 352 patients, and cerebral hemorrhage occurred in two patients. The electrode in Case 2 was pulled out by 2 mm because the electrode tip was too deep and the right electrode in Case 9 was pulled back and adjusted by 1 mm forward. There was no significant correlation between adjustment of electrodes and ICH.

Table 3.

Improvement rate at the 2-year follow-up in 352 patients.

	Preoperative baseline(med-off)	Postoperative baseline (med-off, IPG-off)	3 months(med-off, IPG-on)	6 months (IPG-on, med-off)	1 year(IPG-on, med-off)	2 years(IPG-on, med-off)
ICH group (n = 10)	60.73 ± 16.95	50.18 ± 13.41	26.91 ± 9.33	26.36 ± 7.30	27.73 ± 9.55	29.64 ± 8.99
Non-ICH group (n = 337)	55.45 ± 15.81	54.64 ± 14.28	26.23 ± 9.01	25.40 ± 7.91	25.13 ± 7.85	27.18 ± 7.33
P value	0.518	0.367	0.588	0.528	0.243	0.162

med-off: no drug use; IPG-off/on: implantable pulse generator off/on; ICH: intracranial hemorrhage.

Figure 1.

Comparison of the hemorrhage index between the preferred side and non-preferred side. The hemorrhage index in the preferred side was significantly higher than that in the non-preferred side (paired t test).

Morphological characteristics of hematomas

In all 11 cases of ICH, the focus of hemorrhage was confined to the periphery of the electrode puncture tract, and there was no hemorrhage caused by subdural hematoma, subarachnoid hemorrhage, epidural hematoma, or venous cerebral infarction. The hematoma did not cause midline deviation, and the brain tissue around the hematoma showed different degrees of brain edema. Six patients presented with ICH in the cerebral cortex and in the tip of the electrode, characterized by “dumbbell-shaped” hematomas (Figure 2) (Cases 2 and 3). The hematoma in the electrode tip of GPi targets showed a “shuttle type” high-density shadow (Figure 2) (Cases 10 and 11).

Figure 2.

Computed tomographic and magnetic resonance imaging of intracranial hemorrhage in 11 patients.

Treatment and follow-up

All hemorrhagic patients were treated conservatively, including absolute bed rest, strict control of blood pressure, oral antiepileptic drugs for 2 weeks, and intravenous application of hemostatic drugs. On the 3rd day after hemorrhage, if cranial CT showed no increase in hemorrhagic volume, the patients were discharged. Case 5 was provided symptomatic antiepileptic treatment after discharge from hospital for 7 months. Patients without ICH received stimulation adaption at the 4th week after the operation and the improvement rate was 51.52% ± 14.24% with significant efficacy (Table 1). Patients with ICH obtained stimulation adaption at the 6th week after the operation and the improvement rate was 54.32% ± 15.38%. There was no significant difference in the baseline UPDRS-III score between patients with ICH and those without ICH. There was also no significant difference in the postoperative UPDRS-III score between patients with ICH and those without ICH. After 2 years’ follow-up, four patients without ICH were lost. There was no significant difference in the improvement rate at 1 month, 6 months, 1 year, and 2 years after the operation. This was a heteroscedastic double-tailed result (P = 0.817, P = 0.675, P = 0.392, P = 0.390, respectively). Therefore, bleeding did not affect the postoperative outcome (Table 3).

Discussion

ICH rate and risk factors after DBS

ICH is the most common and dangerous complication after DBS. The rate of hemorrhage after DBS has been reported as 0.2% to 5.6%,[14-16] and the rate of hemorrhage at each electrode is 0.6% to 3.5%.[17,18] Meta-analysis has shown that the rate of hemorrhage after DBS ranges from 3% to 4%.[19,20] Old age, male sex, PD, and hypertension may be risk factors for ICH after DBS.[14,21-23] The mean age of patients with PD in this study was older than 60 years. The incidence of ICH was 3.1% and 1.9% on each side. The incidence of symptomatic hemorrhage was 2.3% and 1.4% on each side. We found no significant correlations between the age at surgery, duration of disease, preoperative UPDRS-III score, and ICH. However, male sex and patients with perioperative hypertension showed a higher probability of postoperative ICH.

This study showed that the preferred puncture side was more harmful in surgery than the non-preferred side. Vascular injury and infarction caused by puncture are causes of delayed ICH after DBS.[24,25] With regard to the cause of vascular injury, we speculate that it may be due to damage of the vasculature caused by brain tissue displacement owing to cerebrospinal fluid loss and recovery during the operation.

Intraoperative electrode adjustment, similar to microelectrode recording, can cause secondary puncture damage to brain tissue. Previous studies have suggested that microelectrode recording may increase the risk of hemorrhage,[19,26] but there is also controversy on this issue.[21,27] In our study, adjustment of the intraoperative electrode position did not increase the risk of postoperative ICH.

Previous studies have shown that GPi-DBS may have a higher probability of ICH than STN-DBS.[27] In our study, the rate of hemorrhage in STN-DBS was 2.7% and that in GPi-DBS was 7.4%, but there was no significant difference in this rate. A reason for this non-significance may be due to the insufficient sample size of patients who received GPi-DBS.

The morphological features of STN-DBS-related hematomas in this study were as follows. ICH was located in the puncture tract where the electrode was located, and the hematoma along the tract could be divided into two parts: subcortical hematoma and deep brain hematoma (Figure 2; Cases 2 and 3). The total hemorrhagic volume was < 30 mL, and there was no subdural hemorrhage, subarachnoid hemorrhage, or extensive brain edema. These findings indicated that the source of hemorrhage was small perforating vessels in the cortex, but not injury or occlusion of the middle cerebral artery branches in the cerebral sulcus or cortical and bridge veins.

ICH in two patients with GPi-DBS showed similar morphological features. The hemorrhage was located at the tip of the electrode, and the shape of the hemorrhage was the same as that of the optic nerve. This may be related to the structure of brain tissue around the GPi nucleus. The source of hemorrhage was probably the perforator artery that branched from the posterior communicating artery to the globus pallidus and a deep cerebral drainage vein. Because the electrode is close to the cistern, the brain tissue is weak. Therefore, a small amount of hemorrhage easily forms in this area, forming a long axis along the optic tract with a “spindle” hematoma shape. The high number of perforating vessels around the GPi is also one of the reasons the GPi has a higher hemorrhage rate than the STN in DBS.[27]

Prognosis of hemorrhage

Eleven patients with ICH in our center were discharged from hospital after conservative treatment. The hemorrhagic volume in 11 patients was mild to moderate, nine had a small amount of hemorrhage (1–10 mL), and two had hemorrhage > 10 mL with a maximum volume of 15 mL. Symptoms of the nervous system caused by hemorrhage were relatively mild in our patients. Eight patients had only mild signs of the nervous system. Three patients were asymptomatic. After conservative treatment, no permanent neurological deficit remained and no death occurred. Case 5 had seizures after discharge and oral antiepileptic drugs were administrated to control the seizures. Our findings suggest that patients with perforating vessel hemorrhage and a hemorrhage index < 4 have a good long-term prognosis.

Overall improvement in patients with ICH was similar to that in patients without hemorrhage. The UPDRS-III (IPG-on, med-off) score at 6 months, 1 year, and 2 years after the operation showed no significant difference between patients with ICH and those without ICH. This lack of finding may have been due to a small hemorrhagic volume in the 11 patients with ICH, relatively mild symptoms of the nervous system, and “micro damage” effects caused by infarction and edema, and non-absorptive hematomas in local tissue. The treatment efficacy of the 11 patients with hemorrhage was not affected after a positive therapeutic effect.

Conclusion

Currently, DBS is a useful technique for treating PD. Postoperative Parkinson’s symptoms have greatly improved and motor complications have been effectively controlled. Although the complications of DBS in minimally invasive surgery are lower than those of other operations, ICH is still unavoidable and is one of the most dangerous complications. Other centers have reported cases of large ICH requiring craniotomy to clear the hematoma and cases of vegetative state or even death resulting from ICH after DBS.[28] Our center has a low rate of hemorrhage, less hemorrhagic volume, and a relatively good prognosis compared with other centers. Our study shows that male sex and hypertension are high-risk factors for ICH after DBS. The preferred puncture side is more likely to have hemorrhage than the non-preferred site. Our short-term follow-up results show that a small amount of hemorrhage does not affect the efficacy of DBS after beginning stimulation. DBS surgery is relatively simple and it is well established. However, minor intraoperative errors may lead to irreparable results. Therefore, control of perioperative risk factors, improvement of surgical methodology, and surgical skills may be effective for reducing the probability of ICH.