The decade after subthalamic stimulation in advanced Parkinson's disease: A balancing act
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.173663
Source of Support: None, Conflict of Interest: None
Aim: The duration of improvement in quality of life after subthalamic nucleus deep brain stimulation (STN DBS) for Parkinson's disease (PD) and the presurgical identification of factors predicting sustained clinical benefits have implications in patient selection and timing of surgery. These aspects were assessed in patients who underwent yearly assessment for at least 7 years after surgery.
Keywords: Deep brain stimulation; Parkinson's disease; quality of life; subthalamic nucleus
Several multicenter, randomized studies comparing best medical treatment with deep brain stimulation (DBS) established that DBS is the more effective strategy for the management of motor complications of levodopa treatment and improving the quality of life (QOL) of patients with Parkinson's disease (PD).,,,,,, Until recently, DBS was offered to patients 10–15 years after the onset of motor symptoms of PD. Several groups, including ours, reported sustained clinical benefits and improved QOL for 3–5 years following DBS.,,,,,, A few smaller studies on the long-term outcome beyond 5 years of surgery ,,,,,,, found that the motor and cognitive deficits worsened in a manner similar to the medically managed patients, suggesting that DBS does not alter the natural course of PD. Further proof of the duration of benefits on QOL and the identification of predictors of long-term improvement will be useful to choose the ideal timing of surgery and the ideal candidates for DBS. For this purpose, we examined the long-term QOL, motor and cognitive outcomes of bilateral subthalamic nucleus (STN) DBS and the pre-DBS factors that predict sustained motor benefits at or beyond 7 years from surgery.
All patients who underwent STN DBS for PD at our center and who completed the yearly assessment protocol for a minimum of 7 years after STN DBS were included in the study. The patient selection, presurgical evaluation, and surgical protocols have already been reported. In the yearly follow-up visit, all patients were evaluated using the Unified Parkinson's Disease Rating Scale (UPDRS) parts I–IV. Any programming of the neurostimulator, if required for worsening motor symptoms, was done before the assessments. The UPDRS III score was administered in the drug OFF/stimulator ON condition (after the overnight withdrawal of all PD drugs) and in drug ON/stimulator ON condition (after administering their morning dose of PD drugs). We also derived a composite score from the sum of tremor, rigidity, and bradykinesia (“TRIB”) scores in UPDRS part III (items 20–26, 31). The “axial” scores  were derived by summation of items 18, 19, and 27–30 in UPDRS III. Cognitive evaluation was performed using Mini–Mental Status Examination and Addenbrooke's Cognitive Examination., Mood was assessed using Beck's Depression Inventory (BDI). QOL was measured using the Parkinson's Disease Quality of Life (PDQL) Questionnaire., The levodopa equivalent daily dose (LEDD) was calculated  at baseline and at each annual visit, and the optimal stimulation parameters were recorded at each visit. Postoperative improvement in motor scores (representing the effect of DBS) was calculated in stimulation ON, drug OFF state at each follow-up visit using the formula: (baseline UPDRS III score in drug OFF − follow-up UPDRS III score in stimulation ON, drug OFF)/baseline UPDRS III score in drug OFF × 100. The immediate pre-DBS (baseline) levodopa response of motor signs was calculated using the formula: (baseline UPDRS III score in drug OFF − baseline UPDRS III score in drug ON)/baseline UPDRS III score in drug OFF × 100.
Repeated measures paired t-test was used to see the effect of time on the variables examined. Bonferroni correction was applied, and a P < 0.01 was considered statistically significant. Pearson's correlation coefficient was used to assess the predictors of improvement at 7 years or more. Statistical analysis was performed using statistical package for social science software (version 15.0, SPSS, Chicago, IL, USA) for Windows.
From the original cohort of 45 patients previously reported, only 25 were available for a 7-year assessment; 12 of them also completed the 10-year assessment. The mean age of onset of motor symptoms of PD in these 25 patients was 42.1 ± 9.4 years (range: 17–60 years, median: 44 years), the mean age at the time of DBS was 53.4 ± 10.8 years (range: 22–69 years, median: 56 years), and the mean duration of PD was 11.3 ± 5.3 years (range: 3–24 years, median: 11 years). The mean baseline LEDD was 600.0 mg (±190.4 mg), and the mean Hoehn and Yahr stage in the drug OFF state was 3.7 (±1.0). The mean baseline levodopa response of UPDRS III score was 66.4% (±16.3%). Other baseline clinical characteristics of the cohort are shown in [Table 1],[Table 2],[Table 3].
Effect of STN DBS in the drug OFF state
The UPDRS scores are shown in [Table 1]. The UPDRS I score did not change significantly following DBS at any follow-up. UPDRS II scores showed a significant improvement at 7 and 10 years. Though the scores showed a decline (P = 0.001) between the 1- and 5-year assessment, there was no significant decline after 5 years. The UPDRS III scores showed an improvement of 39.5% (±27.5%) at 7 years and 30.2% (±24.1%) at 10 years, which were lower than the 71.6% (±13.4%) improvement at 1 year and 53.9% (±16.3%) at 5 years. UPDRS III scores declined between 1- and 5-year follow-up visits (P < 0.001), and also between 5 and 10 years (P = 0.007). The composite “TRIB” scores showed a sustained improvement until 10 years. The mean improvement in TRIB scores was 74.6% (±14.9%) at 1 year, 62.7% (±18.7%) at 5 years, 48.7% (±23.5%) at 7 years, and 35.4% (±26.9%) at 10 years. The axial score remained improved only till 5 years. The mean improvements in the axial scores were 62.7% (±16.7%), 30.1% (±21.7%), 10.2% (±51.4%), and 2.4% (±23.4%) after 1, 5, 7, and 10 years, respectively. The subscores for tremor, rigidity, and bradykinesia were in an improved status at 7 years, but at 10 years, only tremor and rigidity scores [Table 2] showed persisting improvement, while the bradykinesia score was not different from the pre-DBS scores. Unlike tremor, which remained uniformly improved till 10 years, a loss of improvement was evident for rigidity between 5 and 7 years (P = 0.003, for rigidity at 5 years vs. 7 years). Among the components of axial scores [Table 2], gait remained improved until 5 years (P = 0.002), but not at 7 years. The loss of benefit for postural stability was evident by 5 years itself. Speech showed no significant improvement by 5 years and was worse than baseline at 7 years.
Effect of STN DBS in drug ON state
UPDRS I scores did not improve at any time. The initial benefit in UPDRS II and III scores in the first year was not sustained. The TRIB scores also reflected the same pattern. The axial scores were back to the baseline levels by 5 years and significantly worse than the baseline levels by 7 years [Table 1]. DBS had no effect on speech (which was worse than the baseline level by 5 years), and on gait and postural stability, which were worse than the baseline level by 7 years [Table 2].
The improvement in dyskinesias and fluctuations (UPDRS IV A and B) persisted until 7 years. In the smaller subgroup, motor fluctuations (UPDRS IV B) remained improved at 10 years but not dyskinesias [Table 1].
Effect of STN DBS on cognition in drug ON state
The baseline scores of BDI, MMSE, and ACE are shown in [Table 3]. None of the patients had clinically significant depression or dementia before DBS. Among the 25 patients, 20 were evaluated with ACE at baseline levels, of which 17 were retested with ACE at 7 years, and 9 patients at 10 years. MMSE scores [Table 3] were available for 24 patients at baseline, 21 patients at 7 years, and 9 patients at 10 years. There was no statistically significant decline in the ACE or MMSE scores in these patients. However, 7 of the 25 patients (28%) had clinical dementia at follow-up; 1 at 5 years, 3 at 7 years, and 3 at 10 years. They had severe dementia and could not complete formal neuropsychological testing to be included in the comparisons.
The mean depression scores (BDI) showed a statistically significant improvement only at 1 year and declined to baseline levels by 5 years; thereafter, the scores remained stable [Table 3]. However, many patients were on higher-dose antidepressant drugs than in the previous years during the assessments (six at 5 years, 2 at 7 years, and 1 at 10 years).
Effect of STN DBS on QOL
The total and subscores of PDQL are shown in [Table 3]. The total score improved after DBS only till 5 years, and the systemic and emotional component scores declined to baseline levels by this time. There was no improvement in PDQL, either in the total or subscores, at 7 years and beyond.
Effect of STN DBS on medication requirement
The LEDD at baseline (600.0 ± 190.4 mg) came down by approximately 50% at 1-year follow-up (312.3 ± 193.4 mg; P > 0.001). The reduction in LEDD was maintained both at 5 years (370.3 ± 186.8 mg; P = <0.001) and 7 years (369.5 ± 194.6 mg; P = <0.001), but not at 10 years (470.0 ± 327.5 mg; P = 0.09).
Long-term changes in the intensity of STN stimulation
The mean values of neurostimulator settings (amplitude: 3.1 ± 0.6 V; pulse width: 66.3 ± 14.2 µs; frequency: 153.4 ± 25.9 Hz) at 1 year did not change significantly at 7 or 10 years (amplitude: 3.1 ± 0.8 V, P = 0.6; pulse width: 67.5 ± 13.4 µsec, P = 0.3; frequency: 150.0 ± 22.8 Hz, P = 0.1). During the 211 patient-years of follow-up, 35 neurostimulators that reached end-of-life were replaced.
Predictors of sustained improvement at 7 and 10 years after STN DBS
We determined the baseline factors (age at onset of PD, age at the time of surgery, duration of PD at the time of surgery, baseline UPDRS III, TRIB, and axial scores in drug ON and drug OFF state, and levodopa response of motor signs) that were associated with a good motor outcome (improvement in UPDRS III scores compared with baseline scores in the drug OFF state) at 7 and 10 years. At 7 years, the improvement in UPDRS III scores from baseline had a positive correlation (R = 0.60; P = 0.002) with the baseline levodopa response of UPDRS III scores, a negative correlation with the baseline UPDRS III scores in the drug ON state (R = −0.45; P = 0.02) [Figure 1] and [Figure 2], and no correlation with the other factors. The analysis for 10-year follow-up was limited by the small number of subjects (n = 12), but the improvement in UPDRS III scores after DBS showed a negative correlation (R = −0.63; P = 0.03) with the pre-DBS axial scores in the drug ON state [Figure 3].
Patients without follow-up
Among the first 45 patients included in our original report  and who completed at least 7 years after DBS, 20 were not available for the 7-year assessment. Eight of them (from other states of the country) did not wish to travel and were under local care. Seven were confirmed to be dead (the cause of death could not be ascertained by telephonic contact in three; two patients died of coronary artery disease; another patient died of pneumonia two years after DBS at the age of 75 years; and one patient committed suicide 3 years after DBS). Four patients reported significant gait and cognitive dysfunction hindering their travel. One patient was injured by a psychotic relative, resulting in lead fracture and explantation and was not included in the analysis. There was no statistically significant difference in the age at onset (46.4 ± 12.9 years; P = 0.2) of motor symptoms, age at surgery (57.2 ± 10.9 years; P = 0.3), duration of PD at the time of surgery (10.8 ± 6.4 years; P = 0.7), baseline Hoehn and Yahr stage in drug OFF state (3.8 ± 1.0; P = 0.6), or baseline levodopa response (63.2 ± 20.9%; P = 0.6) between the patients who were followed up and those who were missed. The patients who missed follow-up had a better baseline UPDRS III scores in the OFF state (42.2 ± 12.0; P = 0.002) compared with those who had not; however, there was no significant difference in the baseline UPDRS III scores in the ON state (16.3 ± 11.7; P = 0.7).
At the time of the analysis, 20 of the 25 patients who were assessed at 7 years had completed 10 years after DBS. However, the 10-year follow-up evaluation could not be performed in eight of them. Three were confirmed to be dead (one died of coronary artery disease; and, another patient died 9 years after surgery due to cirrhosis caused by chronic hepatitis B virus infection. He was seropositive before surgery. The cause of death could not be ascertained by telephonic interview in the third patient). Four patients did not come for follow-up owing to personal inconvenience. One patient had a fall resulting in lead fracture, and reimplantation was refused by the family.
There are only few reports evaluating the long-term effects of STN DBS in PD beyond 7 years,,,,,,, and the number of subjects in each study is limited [Table 4]. The changes in QOL, an important end point in assessing the outcome of DBS, has been addressed by only one of these studies. Besides aiding in patient selection and counseling, whether the benefits last for at least a decade is an important consideration, particularly in regions of the world where DBS treatment is an out-of-the pocket expense for the family.
From the systematic, yearly assessment data of the first 45 PD patients who underwent bilateral STN DBS at our center and included in our previous report, we were able to obtain data of a cohort of 25 patients who completed the 7-year assessment and 12 patients who completed the 10-year assessment. The most striking observations were that patients continued to perform better in their activities of daily living when compared with the pre-DBS state, suffered less severe motor fluctuations and motor deficits of PD even at 7 and 10 years, and required much less dopaminergic medications until 7 years. Those who had greater improvement in motor deficits with DBS were those who had greater levodopa response to their motor deficits prior to surgery. Patients with more motor deficits after levodopa prior to their DBS procedure tended to have lesser motor benefits in the long term. The initial improvement in axial scores was lost by 5 years and in the QOL, by 7 years. Nearly one-third of the patients were demented in the long term. At 10 years, in the smaller subgroup, many of the patients had increased their drug intake to improve motor dysfunction and had recurrence of dyskinesias. In this subgroup, worse axial scores in drug ON pre-DBS state predicted a worse motor outcome at 10 years.
Even though the benefit of DBS on activities of daily living during the medication OFF state were sustained even at a 10-year follow-up, the improvement by 5 years was less than at 1 year, indicating the progression of disease and the appearance of deficits that were not reduced by STN stimulation. This was also reflected in the increasing UPDRS III motor scores with the passage of time. Among the cardinal motor deficits of PD, tremor showed the most stable and sustained response to DBS until 10 years, by which time, rigidity reappeared and bradykinesia was no longer better than that seen at the baseline level. The benefit of DBS on postural stability wore off as early as 5 years, followed by appearance of lack of improvement in gait by 7 years. This differential stability of response to DBS was also reflected in the composite score for tremor, rigidity, and bradykinesia (TRIB), which remained improved at 10 years, while the composite axial score deteriorated to baseline levels by 7 years. This behavior of tremor when compared with rigidity and bradykinesia is congruent with the previous reports.,, It could be due to the fact that the progression of tremor is slower in PD relative to rigidity and bradykinesia , and may have a different pathophysiology  from the other cardinal motor signs. The severity of tremor, therefore, correlates weakly with the degree of nigrostriatal denervation.,,
That speech and axial scores were refractory to both dopaminergic drugs and STN stimulation was evident from the fact that the subscores for both UPDRS III (7 years) as well as the composite axial scores were worse than those seen in the pre-DBS drug ON state at 7 years. The worsening of axial scores beyond 5 years has been observed previously,,,,, and is likely to be due to the natural progression of PD.,, Though programming and modifications in stimulation parameters were attempted in all patients who reported worsening of motor functions during follow-up, the mean intensity of stimulation did not change significantly after 1 year.
Two other major long-term beneficial outcomes of DBS were the reductions in motor fluctuations, dyskinesias, and drug requirement, which were sustained till 7 years. However, in the smaller subgroup, the dose requirement crept back to pre-DBS levels at 10 years, accompanied by a recurrence of dyskinesias. Sustained improvement in dyskinesias, a decade after DBS has been previously reported., However, in these reports, the UPDRS III total score in drug ON state at the last assessment were significantly worse than at baseline levels. Such a worsening was seen only in axial scores in our study and not in the total scores. This may suggest that reinstating dopaminergic therapy may have led to some motor benefits in nonaxial signs, but at the price of dyskinesias.
In our cohort, patients with higher levodopa-responsive motor symptoms before DBS had greater response to STN stimulation, while those with more deficits, especially axial deficits, at baseline levels even after levodopa intake had lesser motor benefits from stimulation on a long-term basis [Figure 1],[Figure 2],[Figure 3]. A higher degree of axial dysfunction at baseline levels could be a predictor of a more rapid progression of the disease pathology to nondopaminergic circuits and hence, resulting in an earlier loss of initial benefits of DBS in axial scores and QOL. Worse axial dysfunction in drug OFF state at baseline levels was previously shown to be associated with an unfavorable DBS response in the long term. The effect of baseline levodopa responsiveness on the long-term motor outcome has not been so far demonstrated. Though there was a high dropout of subjects from long-term assessment, the baseline characteristics of the dropouts, including baseline levodopa response of motor signs, were not different from those available for follow up.
Nearly one-third (28%) of the patients included in the analysis developed dementia by 10 years; the actual figure could be higher, as there were many dropouts after the 7th year of follow-up. The results are compatible with the frequency of dementia encountered in medically treated PD patients.
Systematic QOL assessment is a unique feature of this study. We found that improvement in QOL was maintained only until 5 years and declined to baseline levels at 7 years. This could be due to worsening axial symptoms, and the recurrence of bradykinesia, dyskinesias, and cognitive dysfunction as PD advances. The importance of gait in the QOL of patients with PD has recently been identified. In our cohort, patients underwent DBS at a mean disease duration of 11 years. The fact that in this group, the QOL improvement could be sustained only for an average of 5 years, justifies studies to explore the merits of early DBS on the long-term QOL of PD patients.
The complications of DBS in our original cohort of 45 patients in the initial years have been previously reported. None of the 25 patients in this study had any implant-related complications after 5 years, except in a patient who developed lead fracture following blunt trauma. There were neither further cases of suicides besides a single case at 3 years, nor any deaths directly or indirectly attributable to surgery/stimulation, suggesting that the risk for suicide, if any, due to DBS, is in the early years.
As in other studies reporting long-term outcomes,, patient dropout is a limitation of our study too. Many factors such as worsening physical and cognitive disabilities that hinder travel for follow-up, increased mortality in PD, and disability and death from comorbidities that are common in the elderly, may account for this. We tried to contact all patients/their relatives to account for the dropouts and compared the baseline characteristics of the patients who could be assessed with those who dropped out, and found that there was no significant difference. Even so, pooled analysis of data from larger number of patients from different centers may be required to establish the long-term predictive value of baseline disease characteristics.
Financial support and sponsorship
In-house research funds from Sree Chitra Tirunal Institute for Medical Sciences and Technology, Kerala, India.
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4]