Predictors of dementia-free survival after bilateral subthalamic deep brain stimulation for Parkinson's disease
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.258056
Source of Support: None, Conflict of Interest: None
Keywords: Deep brain stimulation, dementia, Parkinson's disease, subthalamic nucleus, survival
Deep brain stimulation (DBS) of the subthalamic nuclei (STN) is currently the standard of care for patients with Parkinson's disease (PD) experiencing disabling motor complications of dopaminergic treatment., Though the benefit of surgery on motor complications persists over the long term, it fails to retard the progression of PD, and the emergence of treatment-refractory axial motor and non-motor dysfunction, notably dementia and falls.,,,,,, Though some studies have shown that cognitive side effects attributable to DBS do occur, they are generally clinically insignificant and similar to medically treated patients.,,, Patients with significant cognitive dysfunction are not offered surgery. However, subclinical cognitive dysfunction is commonly detected in the preoperative neuropsychological tests of patients being offered DBS after several years of the disease. Following DBS, with the natural course of disease unhalted, around a quarter of patients eventually become demented within 10 years.,,, This could be an under representation as most of these studies report a significant drop-out during the long-term follow up.,,, The development of dementia offsets many of the benefits of DBS on the quality of life (QOL) and independent functioning. It would, therefore, be very useful to predict the dementia-free survival in patients who are otherwise eligible for STN DBS so that the patients and their families can take informed decisions.
We reviewed the baseline clinical and neuropsychological data and the protocol-based annual follow-up information of patients who underwent bilateral STN DBS for advanced PD at our center during the period 1999–2014, with an aim to assess the dementia-free survival and its predictors in this cohort.
The pre- and post-operative evaluation methods, the criteria for selection for STN DBS, and the surgical procedure followed by our team have been reported earlier., For this study, the case records of all the consecutive PD patients who underwent bilateral STN DBS surgery from August 1999 to June 2014 and who had a minimum of 2 years follow-up data, were reviewed. The dementia-free versus demented status was decided based on the last follow-up assessment. Dementia of PD was diagnosed by retrospectively applying the criteria proposed by Emre et al., when the patient was judged to have a dementia syndrome of insidious onset and slow progression; and, the history, clinical and mental examination revealed impairment in more than one cognitive domain, a significant decline from the baseline level and an impairment of daily life activities. The baseline data collected for the analysis included age of onset of the disease, age at surgery, duration of PD at surgery, history of depression or treatment for depression before DBS, history of psychosis necessitating treatment, presence of freezing of gait in the OFF phase (verified by a review of the pre-operative drug OFF phase videos), and levodopa-equivalent daily dose (LEDD) of drugs. The Unified Parkinson's Disease Rating Scale (UPDRS) part II, III, and IV scores were collected for assessing the activities of daily living (ADL), motor signs, and motor complications, respectively. For the baseline UPDRS III, in addition to the composite scores, the “TRIB” (tremor, rigidity, and bradykinesia) and “axial” (items 18, 19, and 27–30 in UPDRS III) scores and their levodopa response were calculated separately. The baseline neuropsychological data collected included mini-mental status examination (MMSE), Addenbrooke's Cognitive Examination (ACE),, and Wisconsin Card Sorting Test (WCSE)., The exclusion criteria for surgery included freezing of gait in the ON phase, the presence of dementia, and clinically significant and active depression or psychosis. The history of freezing of gait in the OFF phase, and the history of treatment for depression and psychosis were included in the analysis because of their association with cognitive changes in PD and the risk of future cognitive decline.,, Patients with depression or psychotic symptoms at baseline evaluation underwent DBS only after achieving complete and stable remission of symptoms. The protocol was reviewed and approved by the Institutional Ethics Committee.
The association of clinical and historical data with the dementia-free survival during the follow up period was assessed using the Cox univariate regression analysis. P values <0.01 were considered significant for the univariate analysis. The survival effects of categorical and dichotomized continuous variables were assessed using the Kaplan–Meier survival curves and comparisons were done using the Mantel–Cox log rank test. Cox proportional hazards regression model was used to determine the independent predictors of dementia-free survival. To avoid multicollinearity in the setting of a relatively small sample size, only the number of categories passed in the WCST was included for the analysis. The analysis of sub-scores of ACE for individual cognitive domains was not attempted for the same reason.
Pre-surgical baseline features
Among the 127 patients of the surgical cohort, 116 (78 men and 38 women) were eligible for inclusion in the analysis. The pre-surgical demographic and clinical features are presented in [Table 1] and the neuropsychological examination data (baseline as well as at last follow-up) in [Table 2]. Freezing of gait in the OFF phase was present in 65 (56.0%) of the patients. A history of clinically significant depression and psychosis necessitating treatment prior to DBS was present in 73 (62.9%) and 40 (34.5%) patients, respectively.
Dementia status on follow-up
The mean duration of follow-up was 4.7 years (range 2–12 years). During the 542 person–years of follow-up, 30 (25.9%) of the 116 patients developed dementia. The mean dementia-free survival of the cohort was 8.7 years [95%, confidence interval (CI): 7.8–9.6 years]. Six among the demented patients were dead and three were not contactable at the time of collecting of the data for this study. The remaining demented patients were alive. Among the 86 patients who were non-demented at their last follow-up, 29 had dropped out at the time of data collection for the analysis. Thus, 13 were dead, and 16 were either not contactable or under the care of physicians elsewhere. The mean age at death of the 19 patients (6 demented and 13 non-demented at the last follow-up) was 63.5 years (range 36–83 years) and the mean duration after DBS was 5.6 years (range 3–11 years). Five patients died of coronary artery disease and two of pneumonia. One patient each died of cirrhosis of liver, oral cancer, and bowel perforation. Two patients committed suicide and the cause of death could not be ascertained in the remaining seven.
Baseline factors associated with dementia-free survival
[Table 3] and [Table 4] show the results of univariate Cox regression analysis and the Mantel–Cox log-rank test, respectively. Among the variables examined, the history of depression or psychosis, the freezing of gait in the medication OFF phase, the age at surgery, the duration of motor symptoms at the time of surgery, the ADL scores in the ON phase, the UPDRS III axial scores in the ON phase, the lower levodopa response of the UPDRS III axial scores, and a poorer performance in the ACE and WCST testing were associated with a shorter dementia-free survival. The MMSE scores (P = 0.01) and UPDRS III axial scores in the OFF phase (P = 0.02) tended to be significant, while the other variables were not. The Kaplan–Meier survival curves for freezing of gait are presented in [Figure 1], and those for history of depression and psychosis are presented in [Figure 2]. The levodopa response of the UPDRS III and the axial motor sub-scores were dichotomized into two groups (≤50% and >50%) and were in line with the observations in univariate Cox regression analysis. The dementia-free survival was shorter (P = 0.001) in those patients with ≤50% levodopa response for the axial motor scores, while no such difference was noted for the levodopa response of UPDRS III scores (P = 0.215) [Figure 3].
Independent predictors of dementia-free survival
The factors which were significant in the univariate analysis were entered into a Cox proportional hazards regression model, to explore the independent predictors of dementia-free survival. The potential predictors examined in the model were age at the time of surgery, duration of motor symptoms, history of depression, history of psychosis, presence of freezing of gait in the medication OFF pahse, UPDRS II in the ON state, the levodopa response related to the axial signs, and the MMSE, WCST, and ACE scores. The presence of freezing of gait in the OFF phase (hazard ratio: 7.9; 95% CI: 1.95–32.17; P = 0.004) and the WCST scores (hazard ratio: 0.491; 95% CI: 0.311–0.777; P = 0.002) emerged as independent predictors of dementia-free survival.
In this study of 116 patients who underwent bilateral STN DBS after a mean disease duration of 11 years, yearly clinical and neuropsychiatric assessments revealed that 74.1% of patients remained free of dementia at their last follow-up (range 2–12 years). The mean dementia-free survival was 8.7 years. An older age at surgery, a longer disease duration, a history of clinically significant depression or psychosis prior to selection for surgery, the presence of freezing of gait in the OFF phase, poor ADL scores in the ON phase, poor levodopa response of axial motor scores in UPDRS III, and poor performance in ACE or WCST at the pre-surgical (baseline) evaluation were associated with a shorter dementia-free survival. The severity of UPDRS III motor scores, the degree of levodopa response and the severity of motor fluctuations or dyskinesias did not predict the shorter time to dementia. A poor performance in WCST and the presence of freezing of gait in the OFF phase were independent predictors of a shorter dementia-free survival in the multivariate analysis while the baseline MMSE or ACE scores were not.
Until recently, STN DBS was offered only to advanced PD patients with severe motor fluctuations and dyskinesias that were not controlled by optimal medical treatment. The mean duration of disease at the time of surgery in studies reporting long-term outcome (beyond 5 years) of STN DBS ranged from 11–16 years and these studies reported dementia in 17–29% of the patients.,,,, The emergence of axial motor, non-motor, and cognitive dysfunction consequent to the progression of PD was unhalted by DBS and offset the initial improvement in ADL and independent functioning,, in these patients. The appearance of dementia has a negative impact on the QOL, extent of caregiver burden, and health economics. However, there is scanty information on the predictors of dementia-free survival in the surgical cohorts of PD. Aybek et al., found that around 24.5% of their cohort of 57 PD patients (age at surgery: 63.8 ± 8.0 years, duration: 15.7 ± 5.0 years) who underwent STN DBS developed dementia over a 3-year follow-up period; the patients who developed dementia were older at baseline, had hallucinations, and worse scores in tests of executive functions. Smeding et al., assessed the cognitive functions at baseline and 12 months after STN DBS in 105 patients and found that an advanced age, impaired attention, and low levodopa response at baseline predicted a decline in the cognitive functions in them. A recent report from Korea, with a longer follow-up also reported that impaired attention and executive dysfunction at baseline predicted the global cognitive decline and dementia on follow-up visits. This study evaluated 103 patients (age at surgery: 56.9 ± 7.8 years, duration: 10.6 ± 4.5 years) for a mean follow-up of 42 months (range 12 months–7 years).
In our cohort of 116 patients, whose mean age was 56.5 years (range: 22–78 years) and who had a longer follow-up period of 2–12 years, the univariate analysis showed that an older age at surgery and a longer disease duration were associated with shorter dementia-free survivals. Aging is the strongest risk factor for PD and its progression.,, This relationship is attributed to age-related changes in cellular mechanisms like proteosomal function, autophagic pathways, mitochondrial function, apoptosis, and inflammation, leading to progressive cell death in PD. Aging is also a risk factor for PD dementia., There is probably no increased risk of dementia in post-DBS patients and its development after surgery is likely to be due to the same mechanisms as in medically treated patients. An age >70 years is considered to be a relative contraindication for DBS by many centers and this fact is justified if other cognitive risk factors also co-exist. The prevalence of PD dementia progressively increases with increasing disease duration,, and hence a shorter dementia-free survival is expected with a longer duration of motor symptoms at baseline. This is also supported by a recent report on a cohort of 79 patients with PD who underwent DBS and were evaluated 10 years after the procedure; it was found that 53% of the 55 survivors were demented. The average duration from surgery to the diagnosis of dementia was 5.6 years. This is shorter than the mean dementia-free survival reported in our cohort (8.7 years) and can be explained by the shorter disease duration at surgery in our cohort (15.7 years vs. 11.2 years).
Patients with a history of clinically significant depression or psychosis but symptom-free at the time of surgery, had shorter dementia-free survivals in the univariate analysis. Depression,,, and psychosis,, are the risk factors for PD dementia; recent observations from the Parkinson's Progression Markers Initiative indicated that non-demented PD patients with early psychosis had more cortical atrophy and biomarker profiles compatible with more of amyloid pathology. Similarly, PD patients with depression have been shown to have more severe and more widespread pathology and more frontal atrophy., Active depression and psychosis are considered to be contraindications for DBS.
The overall severity of motor deficits in the OFF phase, the resulting limitation of ADL in the OFF phase, LEDD, or the severity of motor fluctuations or dyskinesia had no predictive value for dementia-free survival after DBS in our patients. However, persisting axial motor impairment and the resulting higher limitation of daily activities during the ON period were associated with a shorter dementia-free survival. Persisting axial motor dysfunction in the best ON state could signify a higher burden of extranigral, non-dopaminergic system pathology involving the brainstem postural and gait control mechanisms as well as more diffuse cortical pathology.,, Similarly, we found that the extent of levodopa response of the UPRDS III total scores or the TRIB scores (sum of the tremor, rigidity, and bradykinesia scores) had no relationship with dementia-free survival, while that of the axial motor scores had a significant relation; we had previously reported a similar relationship between the long-term DBS outcome and levodopa response of the axial scores. Our finding justifies the need to consider DBS for PD before the poorly levodopa-responsive axial motor dysfunction emerges in the presentation.,
We found that cognitive screening with MMSE was inferior to a more comprehensive test battery (ACE) and tests sensitive to frontal functions in predicting the future occurrence of dementia. MMSE is poorly sensitive to executive dysfunction in PD patients compared to other screening tests like the Montreal Cognitive Assessment (MoCA). MMSE is also known to have a ceiling effect, which is probably responsible for its comparative inefficiency in predicting dementia-free survival. A poorer performance in the ACE and WCST scores was associated with the earlier development of dementia. Our observation is concordant with those from the earlier reports. Merola et al., showed that in non-demented patients with PD undergoing STN stimulation, mild cognitive impairment (MCI) predicts an earlier dementia though it has no remarkable influence on the motor outcome. The estimated time to develop dementia for their whole cohort was 9.5 years; the median interval to dementia was 6 years for those with PD MCI, while it was 11 years for those with normal cognition at baseline. Similar to our results, they found that an older age at surgery and persisting deficits in the ON phase tended to have a trend for association with an earlier onset of dementia, while the age at onset or motor complications did not have a predictive value.
Among the multiple factors identified, only one motor (history of freezing of gait in the OFF phase) and one cognitive (performance in WCST) factor emerged as independent predictors of dementia-free survival after DBS. This has important pathophysiological and clinical implications. Freezing of gait has anatomically widespread pathophysiological correlates involving the brainstem as well as the cortical regions.,,,, Freezing of gait in the ON stage (levodopa-unresponsive freezing of gait) is a feature of very advanced stages of PD and is generally considered a contraindication for STN DBS; therefore, none of our patients included in this study had the ON stage freezing at baseline. Though the OFF stage freezing is not an exclusion criterion, its response to DBS of the STN is generally less satisfactory compared to other cardinal motor features of PD, encouraging search for alternative/additional targets for neuro-modulation.,,, Freezing of gait has been shown to be associated with regional atrophy involving frontal and parietal cortices,, altered functional connectivity involving executive-attention networks and cognitive, particularly executive, dysfunction. An earlier occurrence of dementia in patients with freezing of gait who undergo STN DBS for PD is accounted for the more widespread cortical and subcortical pathology. Freezing of gait, particularly if unresponsive to levodopa, is considered as an exclusion criterion for STN DBS and its association with earlier dementia is another matter to be considered while selecting such patients. WCST tests set shifting, an important component of executive function. The presence of significant executive dysfunction at baseline, poorly picked up by screening investigations like the MMSE, could be indicative of a less favorable cognitive outcome after DBS. The demonstration of an independent association of freezing of gait with an earlier onset of dementia in subthalamic stimulated patients is novel, to the best of our knowledge. The relevance of freezing of gait in patients considered for STN DBS is two-fold, it is a symptom which may not show a sustained and satisfactory response to DBS, and also, it acts as a surrogate marker for future cognitive decline and dementia at an earlier interval.
We did not address the question of whether or not DBS has any deleterious effects on the natural progression of cognitive dysfunction in PD in the long run. Long-term prospective studies comparing patients with PD undergoing DBS and medically managed patients are needed to answer this question, and are practically difficult for logistical and ethical reasons. Based on the observations from short-term comparative studies and meta-analyses of cognitive outcomes, DBS is currently considered a relatively safe procedure from the cognitive angle.,, However, cognitive changes attributable to DBS do occur, and include decrease in fluency and a subtle decline in memory and executive functions. These are attributable to the mechanical disruption of fronto-striatal connections as well as stimulation of STN. These structures are involved in not only motor but cognitive functions, as well as affective neural circuits. Our aim was to identify the pre-surgical predictors of dementia-free survival after STN DBS which would help the physician discuss the outcome of this procedure on QOL, with the patient and family members before taking the surgical decision. Nevertheless, the mean dementia-free survival of around 8.7 years in our patient cohort with the mean duration of motor symptoms of around 11.2 years at the time of DBS suggests that the progression of cognitive dysfunction in patients undergoing DBS does not deviate remarkably from that reported earlier in medically managed PD patients who had a longitudinal assessment., Our estimates of the interval to dementia after STN DBS are similar to those from earlier studies which addressed this aspect., The overall conversion to dementia in PD patients undergoing subthalamic DBS is consistent with the natural history of the disease, and thus, DBS does not hasten the onset of cognitive decline in the long run.
We acknowledge that our study has several limitations. The first among them was that the diagnosis of PD dementia on follow-up after DBS was assigned by a review of medical records. The core diagnostic features in the published criteria (historical and clinical information on the impairment of multiple domains, the decline from baseline, and the impairment of daily life activities) were met in all those patients who had the diagnosis of dementia. For mental status examination for the diagnosis of dementia during the follow-up visit, we could only rely upon those cognitive tests which were uniformly done (MMSE, ACE, and WCST) as part of our DBS neuropsychological evaluation protocol; a significant number of patients included in this study were selected and operated many years before the publication of the current diagnostic criteria and recommendations for cognitive testing in PD., Thus, the inherent limitation of the retrospective design of the study precluded the making of the diagnosis of PD dementia strictly based upon the elaborate procedures recommended currently, for cognitive evaluation of PD., However, ACE, which we used, tests the different cognitive domains, and has been validated in large samples of our population. It has been shown to have a fairly good sensitivity and specificity for making a diagnosis of PD dementia. Secondly, the range of age of onset and the duration of follow-up was relatively wide, as we included patients operated between 1999 to 2014, to have sufficient numbers for statistically meaningful results. This was inevitable as DBS is a specialized and costly procedure and the number of drop outs during the long-term follow up is generally high, as is seen in most studies conducted on the long-term outcome of DBS in PD.,,, The high drop-out rates result from worsening disability from axial and non-motor dysfunctions consequent to disease progression (that are not prevented by a symptomatic measure like DBS) and the disability and mortality from co-morbidities in the elderly patients. We used appropriate statistical methods (survival analysis) to circumvent this limitation. Thirdly, the dementia-free survival of the individual patient was estimated as the interval from surgery when they were last examined and found to be not demented; this might have resulted in an underestimation of the dementia-free survival. The fourth limitation was the relatively small sample size which allowed only a limited number of factors to be included in the Cox proportional hazards regression analysis. The sample size limitations and concerns about multicollinearity also precluded exploration of the predictive values of sub-scores of ACE belonging to different cognitive domains, and inclusion of the other scoring parameters of WCST like the number of errors and perseverations. In spite of having these limitations, we believe that the relatively long protocol-based follow-up and the appropriate use of statistical methods are the strengths of our study. These enabled us to estimate the dementia-free survival in patients undergoing STN DBS for standard indications and identify risk factors for impending dementia in them. Our results also underscore the prognostic significance of freezing of gait in relation to the impending dementia. These predictors may prove useful to patients and their families in making an informed decision before they undergo an expensive treatment with surgical risks, albeit small.
The study protocol was reviewed and approved by the Institutional Ethics Committee (IEC) and the study has been conducted in accordance with the ethical standards in vogue. Requirement for written informed consent for the study was waived off by the IEC as the study was of a retrospective nature and all the data were collected by a review of the case files, without including any patient-identifiable data.
Financial support and sponsorship
In-house project (project 5040) of the Comprehensive Care Centre for Movement Disorders, Sree Chitra Tirunal Institute for Medical Sciences and Technology, Thiruvananthapuram, 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]