Orthopedic Surgery and Post-Operative Cognitive Decline in Idiopathic Parkinson’s Disease: Considerations from a Pilot Study

Abstract

Background:

Post-operative cognitive dysfunction (POCD) demarks cognitive decline after major surgery but has been studied to date in “healthy” adults. Although individuals with neurodegenerative disorders such as Parkinson’s disease (PD) commonly undergo elective surgery, these individuals have yet to be prospectively followed despite hypotheses of increased POCD risk.

Objective:

To conduct a pilot study examining cognitive change pre-post elective orthopedic surgery for PD relative to surgery and non-surgery peers.

Methods:

A prospective one-year longitudinal design. No-dementia idiopathic PD individuals were actively recruited along with non-PD “healthy” controls (HC) undergoing knee replacement surgery. Non-surgical PD and HC controls were also recruited. Attention/processing speed, inhibitory function, memory recall, animal (semantic) fluency, and motor speed were assessed at baseline (pre-surgery), 3 weeks, 3 months, and 1 year post- orthopedic surgery. Reliable change methods examined individual changes for PD individuals relative to control surgery and control non-surgery peers.

Results:

Over two years we screened 152 older adult surgery or non-surgery candidates with 19 of these individuals having a diagnosis of PD. Final participants included 8 PD (5 surgery, 3 non-surgery), 47 Control Surgery, and 21 Control Non-Surgery. Eighty percent (4 of the 5) PD surgery declined greater than 1.645 standard deviations from their baseline performance on measures assessing processing speed and inhibitory function. This was not observed for the non-surgery PD individuals.

Conclusion:

This prospective pilot study demonstrated rationale and feasibility for examining cognitive decline in at-risk neurodegenerative populations. We discuss recruitment and design challenges for examining post-operative cognitive decline in neurodegenerative samples.

The term “Post-Operative Cognitive Dysfunction” – or POCD - denotes post-operative memory and/or thinking problems that have been corroborated by neuropsychological testing [1]. While not yet recognized as a clinical diagnosis, clinicians are beginning to use the term POCD as a general description of post-operative patients who complain of memory and thinking problems. Typically, these problems include reduced attention and concentration, word finding problems, and difficulty learning information relative to pre-surgery abilities. Researchers show that post-operative cognitive change associates with increased post-operative and long-term care costs [2, 3] as well as mortality [4, 5]. Types of post-operative cognitive change predict functional limitations; older adults with isolated executive changes experience more functional limitations and increased caregiver burden [6].

Despite numerous attempts to identify anesthetic or surgical mechanisms for operative related cognitive decline, there are no definitive explanations for POCD. General anesthesia remains an unconfirmed influence on cognitive decline in numerous randomized trials [7–9], despite animal studies suggesting that inhalational anesthetics enhance oligomerization and cytotoxicity of amyloid β peptides (a protein change associated with Alzheimer’s Disease) [10, 11], tau phosphorylation [12], and associated neuroinflammatory responses in humans [13, 14]. Surgery-related mechanisms on post-operative cognitive dysfunction (POCD), most often studied in cardiac populations, remain equally inconclusive [4, 15–17].

Pre-operative patient characteristics are, by contrast, important predictors for post-operative cognitive decline and complications [5, 18–22]. A large scale non-cardiac study prospectively examining age differences shows significantly higher rates of POCD for older adults (i.e., 3-month post-operative cognitive decline involving two standard deviations below baseline performance: young = 5.7% , middle = 5.6% , older = 12.7% [5]). Lower education is a risk as well [5, 20], with higher education and better outcome attributed to better brain status (being resistant to disease), better test-taking abilities, and the interrelationships between educational advancement, social support, and better post-operative medical care. Other proposed risk factors are lower pre-surgery executive function [23], higher rates of pre-surgical depression [23], greater vascular disease state (e.g., renal disease, congestive heart failure; [24]), pre-surgical evidence of non-symptomatic stroke [5, 25], and pre-surgical evidence of small vessel vascular disease [26].

Furthermore, there appear to be varying types of cognitive decline, with some individuals primarily experiencing executive based changes, others declarative memory difficulties, and others mixed cognitive profiles [6]. Our research group [6, 27] as well as others [10, 28] hypothesize that pre-operative cognitive profiles and disease risk states could indicate cognitive reserve deficiencies that may be accelerated by operative related events. Retrospective studies suggest that evidence of pre-surgery, mild cognitive impairment may increase the risk for developing post-operative cognitive decline and Alzheimer’s disease (AD) [29]. Prospective post-operative cognitive studies to date, however, have been limited to non-dementia adults without a known neurodegenerative disease. Thus, the magnitude of risk that a pre-existing neurodegenerative disease may pose on post-surgery cognitive outcome remains largely unknown [30].

Individuals with Parkinson’s disease (PD) may be at particular increased risk for acceleration of motor and cognitive dysfunction after surgery. PD impacts at least 1.5 million people in the United States alone, with most patients diagnosed after age 60. The disease is not only associated with motor changes, but also cognitive, mood, pain, and fatigue disturbances. Like their non-PD peers, individuals with PD may seek orthopedic joint replacement for quality of life improvement (e.g., total knee replacement), pain management, or for urgent events such as falls and fractures. We know of no studies examining how individuals with PD change cognitively after major surgery. There are only a few retrospective studies examining changes in range of motion or pain in PD patients after orthopedic surgery [31]. Since PD is associated with a 1.7–5.9 greater risk for dementia relative to non-PD peers [32–34], it appears prudent to examine how operative events may contribute to PD related cognitive decline.

The primary aim of this pilot study was to examine post-operative cognitive decline in patients with PD. Reliable change methods examined individual changes in PD patients relative to control surgery and control non-surgery peers. Non-surgical PD participants were also recruited. Outcome measures included markers of processing speed, inhibitory function, memory, semantic fluency, and motor function. We hypothesized that surgery related cognitive changes in PD would be primarily apparent on measures assessing processing speed and inhibitory function. This hypothesis is based on previous work showing that PD initially alters frontal-subcortical cognitive pathways associated with processing speed, working memory and inhibitory function [35] which may place these patients at a resource disadvantage during operative events and/or stressors.

METHODS

Study design

The study was a prospective longitudinal pilot study with control arms involving a baseline (pre-surgery) neuropsychological assessment, with repeat assessments at 3 weeks, 3 months, and 1 year post-surgery. Patients were assessed for delirium and followed during inpatient surgery related hospitalizations. The study was approved by the University of Florida Institutional Review Board (315–2002). It required a participant consenting process and followed guidelines from the Declaration of Helsinki.

Participants

Potential surgery participants were referred from the UF Center for Movement Disorder and Neurorestoration, the Departments of Orthopedics and the Anesthesiology within University of Florida, as well as from private orthopedic surgery centers throughout Florida. Surgery participants underwent joint replacement surgery (total knee, hip, or spinal fusion) that required general anesthesia. Non-surgery non-PD control participants were referred from the same departments, and also recruited through newspaper advertisements, radio announcements, speaking engagements, and community memory screenings. Individuals with idiopathic PD had to be diagnosed by a movement disorder specialist and were required to meet the United Kingdom PD Society Brain Research Criteria for diagnosis, with “on” medication Hoehn and Yahr scale range of 1–3. Individuals with PD hadto be non-demented and were required to be receiving dopaminergic treatment. Exclusion criteria included the intention to obtain another surgery during the study period (including deep brain stimulation, although DBS acquired 6 months or more prior to surgery was permitted), evidence of a learning disorder, evidence of secondary or atypical parkinsonism or other disorder as suggested by the presence of any of the following: 1) history of major stroke(s); 2) exposure to toxins or neuroleptics; 3) history of encephalitis; or 4) neurological signs of upper motor neuron disease, cerebellar involvement, supranuclear palsy, or significant orthostatic hypertension. A substantial portion of the surgery and non-surgery non-PD participants are reported upon in another publication [26] and are not commented upon here in this manuscript.

Enrollment required two study phases. After informed consent, there was an initial screening for eligibility and for cognitive function via a telephone-based assessment (Telephone Interview for Cognitive Status, [36]). This was followed by an in-person cognitive assessment to identify patients who met inclusion/exclusion criteria. For the knee and hip surgery patients, a standardized anesthetic protocol was used that involved ropivacaine for nerve blocks and post-operative infusion, pre-medication with intravenous midazolam, induction with fentanyl and thiopental (mechanically ventilated with an air/oxygen mixture to maintain an end-tidal CO₂ at 35 ± 5 mm anesthesia maintained with inhaled isoflurane and a Bispectral Index Monitor (BIS^TM) maintained between 40 and 50) and intravenous fentanyl and rocuronium. Delirium was assessed each post-operative day until discharge [37]. With the exception of one surgery (hip surgery) that involved communication to the anesthesiology and surgery team via telephone correspondence and fax, all surgeries were conducted within the same hospital.

Neuropsychological protocol

A trained psychometrician administered all cognitive measures. For the current study, learning/memory, attention/processing speed, inhibition, semantic fluency, and motor function were analyzed. For each cognitive domain, z-scores were created from standardized scores referenced to published test norms [38, 39]. Two neuropsychologists (CP, DB) reviewed the baseline data. Data were double scored and entered by trained individuals who were blinded to diagnosis or condition.

Measures

The Hopkins Verbal List Learning-Revised (HVLT-R; [40]) is a 12-item word list that assesses learning and memory. The words were orally presented at the rate of one word per 2 seconds. After three learningtrials, participants were instructed to recall the words first immediately, and again after a 20-minute delay. Dependent variable = the total correct after the 20-minute delay.

The Digit Symbol Coding subtest of the WAIS-III [41] was used to assess attention/concentration and processing speed. Participants were instructed to quickly match digits with their corresponding symbols without making errors within a 120 second time limit. Dependent variable = total correct within this time limit.

The Stroop Color Word Test, Part 3 [42, 43] was used to assess inhibitory function. In part 3 of this test, the participant views words that are printed in different color ink (e.g., red is printed in blue ink) and are instructed to name the color of the ink and not read the word. Dependent variable = total correct in 45 seconds.

Animal Fluency (Animals; [44]) required participants to name as many animals as possible within 60 seconds. Dependent variable = total correct exemplars.

The Finger Tapping Test [45] assessed fine motor speed in the dominant and non-dominant hand. Dependent variable = mean speed of both dominant and non-dominant hands.

Statistical analysis

Descriptive statistical analyses for POCD were based on reliable change analyses using a modification of Jacobson and Truax’s Reliable Change Index (RCI; see ISPOCD publications, e.g., [1, 26]): (Δ X - Δ X_c)/ SD _(ΔXc) . Two RCI scores were computed for each PD participant at each time point (2 weeks, 3 months, and 1 year) using mean change (Δ X_c) and standard deviation of the mean change (SD _(ΔXc) ) calculated from the HC surgery group and the HC non-surgery group. HCs change means were stratified by age (i.e., 60–70 and ≥71) to appropriately match age ranges of the participants with PD. Reliable change was defined by setting alpha at 0.10 (two-tailed), which corresponds to a 90% confidence level. Reliable decline was determined as a z-score ≤ –1.65 and improvement as a z-score ≥ +1.645. Changes at these cutoffs occur randomly in only 5% of cases in each direction, and is thus considered a clinically significant change [46]. Chi-square analyses were conducted to explore group frequency by type of cognitive decline for PD surgery relative to PD non-surgery peers.

RESULTS

Enrollment

Over a two-year period, we screened 152 older adult surgery or non-surgery candidates, with 19 of these individuals having a diagnosis of PD. Of the individuals with PD, 8 met inclusion/exclusion criteria and elected to participate in the study. The remaining individuals with PD chose not to participate due to surgery location (surgery at another hospital), travel concerns, or surgery postponement. The final sample included 5 PD orthopedic surgery (PD-S), 3 PD non-surgery (PD-NS), 47 surgery control (HC-S), and 21 non-surgery control (HC-NS) participants. For the PD surgery group, none dropped out at 2 weeks and 3 months. One participant did not return for 1 year follow up. There were no dropouts in the PD-NS group. For the control surgery and non-surgery participants, the rate of attrition was 0% at 3 weeks, 6% (3/47) at 3 months, and 13% (6/47) at 1 year. Of these controls, two were unavailable at the 3 month time point, but were tested at 1 year. Of the PD participants havingsurgery, three had total knee replacement, one had hip replacement, and one had spinal surgery. All of the HC participants had elective total knee replacement surgery. No participant experienceddelirium.

Demographics

Table 1. PD patients (n = 8) and controls (HC; n = 68) were similar regarding general demographic variables and general cognitive status (all p >  0.05). Aside from having PD, all were considered healthy with low comorbidities, and there were no statistically significant group baseline differences in cognitive measures with the exception of memory (PD <  HC, p = 0.03).

Table 2. PD-S and PD-NS patients were statistically different in age (PD-S <  PD-NS, p = 0.011). All other demographic variables were similar. Data regarding the PD surgery characteristics are shown in Table 3.

Frequency of PD cognitive changes based on HC non-surgery change scores (Table 4)

At the 3 week post-surgery time point, four out of five PD surgery individuals (80% ) experienced significant decline (at or greater than the –1.645 reliable change score). Three PD surgery participants experienced significant decline at 3 month and 1 year post-surgery assessments. By contrast, only one (33.33% ) non-surgery PD participant experienced decline at 3 month and 1 year post-baseline assessments.

A review of Table 4 shows that the majority of these PD surgery cognitive changes occurred on measures of processing speed and inhibitory function. Exploratory analyses in this sample comparing group type (PD-S, PD-NS) supports a trend for the 3 week assessment [x²(2) = 4.80, p = 0.09]. Group mean reliable change measures are shown in Table 6.

Frequency of PD cognitive changes based on HC surgery change scores (Table 5)

Four out of five (80% ) PD individuals experienced significant decline (at or more than the –1.645 reliable change score) at 3 weeks post-surgery. Three PD surgery participants experienced significant decline at 3 months and 1 year. Only one non-surgery PD participant (33.3% ) experienced reliable decline at 3 month post-baseline assessment.

Exploratory analyses comparing group type (PD-S, PD-NS) by domain relative to the HC Surgery group supports a trend for greater processing speed changes at 3 weeks post-surgery ([x²(2) = 4.80, p = 0.09]. Group mean reliable change measures can be seen in Table 7.

DISCUSSION

This prospective pilot study demonstrates the importance of studying POCD in neurologically at-risk populations. Our data show that the majority of individuals with PD who elected surgery experienced greater cognitive decline relative to the non-PD “healthy” surgery and non-surgery peers. These findings are significant, given that the healthy surgery peers have been reported, (in a separate paper), to have cognitive changes of their own [26]. For the PD participants, these changes were significant in the domains of processing speed and inhibitory functions. Based on the comparison group of PD non-surgery peers, the severity of cognitive change after surgery may be more pronounced than would be expected by PD progression alone. If these findings are confirmed by a larger investigation, the collective findings will support the concept of a threshold effect; i.e., pre-operative cognitive and brain status heralds a type of post-operative insult. A larger, potentially multicenter study is clearly needed to investigate this topic in detail. With the global presence and increase in PD [47–51], it is important that we understand how to provide medical and surgical care and support to those with PD, without further accelerating neurodegeneration.

It is not known why some individuals experience cognitive changes after surgery. In 1955, Bedford [52] published findings suggesting that anesthesia induced cognitive changes. This publication reported upon a chart review of 1,193 adults over the age of 65 who had received a variety of surgical interventions under general anesthesia. Of these patients, 120 individuals (10% ) were reported to have experienced post-operative changes. A subset of 120 patients (n = 18, 1% ) was described as “human vegetables” in the medical records. Since the publication by Bedford, research on POCD has been steadily increasing, yet we still do not know the mechanism(s) that cause POCD. It has been postulated that anesthesia does not contribute to the syndrome at all [8]. Long-standing changes have not traditionally been blamed on lingering pharmacologic effects of anesthetic drugs because of the relatively short half-lives of these agents [53]. Most patients affected by POCD recover in a matter of months [54], which evokes the picture of a slow-healing injury. The fact that not every patient is affected raises a question about genetic influence. The elderly are at greater risk than younger patients [5], and it is known that with age the central nervous system, as with many other physiological systems, is more susceptible to damage. Within our study, the findings suggest that this may be even more pronounced for individuals with a neurodegenerative disease that has been described by some [55] as accelerated aging(i.e., PD).

Studies are beginning to suggest that there are subtypes of POCD among older adults, and that by studying types of cognitive decline we can improve our understanding of mechanisms. Price, Garvan, and Monk [6] showed that within older adults who have POCD after orthopedic surgery, 34% experienced primary executive impairment involving difficulties with mental flexibility and processing speed, 54% experienced primary learning/memory impairment, and 12% showed combined both executive and memory impairment. These findings suggest that adults over 60 years of age can exhibit different variants of cognitive impairment during the three months following surgery. Although these findings require replication, they appear to suggest that there are likely different regions of the brain that can be influenced or associated with POCD. Follow-up investigations e.g., [26] are continuing to show differences in types of post-operative cognitive decline, and that executive changes can be predicted, at least partially, by the integrity of frontal-subcortical systems prior to surgery.

In patients with PD, selective impairment in post-operative processing speed symptoms may be primarily linked to a disturbance in the direct and indirect cortical-subcortical network systems. The current pilot study findings of primary processing speed and inhibitory impairment for the PD participants speak to PD-based, frontal-striatal neuronal vulnerabilities [35]. It has been theorized that individuals can remain above a critical threshold due to cognitive [56] or brain reserve [57] until some combination of factors (e.g., brain damage, neuronal stress) summates to accelerate symptom manifestation (threshold theory’ [57]. Operative events or “stressors” may serve to accelerate frontal-subcortical deficits already developing due to the neurodegenerative disease course. Stressful events include at minimum: (a) emboli (b) damage to the subcortical nuclei due to alteration of the small vessel vascular supply, and (c) neurochemical changes associated with anesthesia.

Certain surgical procedures are associated with venous emboli (e.g., air or clot). Emboli can occur during total knee replacement surgery; e.g., immediately after femoral tourniquet release [58–60]. The comforting image that venous emboli are filtered by the lung – and thus do not gain access to the cerebral circulation – needs to be abandoned. On one hand, up to 30% patients can have a patent foramen ovale that allows material from the right atrium to traverse to the left atrium when atrial pressure on the right exceeds that on the left [61]. This increase in right heart pressure is observed with pulmonary emboli (either air or debris). On the other hand, Sulek and colleagues have observed emboli to pass through the lungs and into the brain, even in patients with closed atrial septums [60]. Emboli during surgery present a risk for stroke. Moreover,arterioles supplying the subcortical nuclei (thalami, basal ganglia, caudate) are particularly vulnerable to blood flow alterations [62], and may be especially prone to operative hemodynamic changes [63]. For all of these reasons, we speculate that embolic insults to individuals with PD may be particularly disruptive due to their already compromised striatal integrity [35]. We encourage future researchers studying POCD in PD to monitor intraoperative emboli and venous pressure changes throughout surgery.

It is also unknown how anesthesia may contribute to post-operative outcome for older adults - and particularly those with Parkinson’s disease. General anesthesia alters activation and blood flow to the brain [64]. Communication between the frontal and posterior brain regions (frontoparietal network) may be disrupted [65] with cerebral blood flow reduction to the frontal cortex, parietal cortex, cingulate, and thalamus [64]. Anesthesia induction is also uniquely associated with decreased spontaneous neuronal firing within the gray matter of the cortex [66], with later firing changes occurring in the subcortical gray regions involved in sensory gating (i.e., thalamic network) [67]. Within these regions, high-frequency rhythms reduce, thereby resulting in the sleep of anesthesia [68]. Interestingly, severely depressed patients under propofol sedation and anesthesia have shown greater reductions in global cerebral blood flow, with particular reduction in the inferior prefrontal region relative to studies of healthy adults [64, 69]. There is some suggestion that depression at time of surgery associates with an abnormal response to anesthesia during the induction phase [27]. These studies, combined with growing appreciation for the rate of depression, apathy, and frontal-subcortical dysfunction in PD [70, 71], supports the argument for investigations on PD and anesthesia responsiveness. Understanding how to predict risk, control the interaction of pre-surgery factors and operative events, and treat POCD will depend on advancing our understanding of the brain, anesthesia and other operative events in at risk and healthy adults.

Our current study suggests the potential for prospective investigations. This is despite our difficulty recruiting individuals with PD who were planning orthopedic surgery. Retrospectively, we consider the recruitment issue largely due to communication difficulties (i.e., often individuals with PD did not consider it necessary to inform their neurologist about the likelihood of surgery and we would learn about the surgery post-hoc). Hence, we did not acquire many referrals from our movement disorder center. Addressing communication issues between neurologists and other providers is an important issue for a) reducing complications post hospitalization [72] and b) building prospective team-based studies that target complex medical issues related to medical care. We encourage a large multicenter effort – where each center concentrates on recruiting a small number of participants – and that includes an organized communication system between neurology, surgery departments, and the preoperative anesthesia screening clinic. The downside of a multicenter study would be the variation in surgical technique, but the upside of a properly powered study would outweigh this negative.

We recognize study limitations. First, our sample size is a clear limitation for generalizing rates of POCD in PD. We statistically address this issue by examining change at the PD individual level relative to a group of surgery and non-surgery peer groups. Second, we have an unequal sample size for individuals diagnosed with PD versus PD non-surgery controls. As shown with other diseases [73], an essential study design characteristic for post-operative cognitive research is a comparison group of disease-matched peers. These peers help clarify if the cognitive change over time can be attributed partially to operative events after controlling for age and disease progression. We encourage future researchers to match surgery and non-surgery participants both on demographics and disease severity. In a much larger sample, it may also be feasible to examine PD clinical type (i.e., tremor dominant, primary gait disturbance) and whether the presence of a deep brain stimulator contributed negatively or positively to post-operative outcome. Third, we did not acquire the Unified Parkinson’s Disease Rating Scale [74] (UPDRS) scores for the PD participants, but encourage this for future investigations as well. Fourth, we recognize that our PD orthopedic surgery group included a mixture of knee, hip, and back procedures, while our healthy surgery control group included those having total knee replacement. Orthopedic surgery is not uniform across these surgery types; e.g., knee surgery is associated with emboli from the tourniquet, hip surgery with more potential for blood loss, andback surgery with longer anesthesia time. Examining orthopedic surgery in general, however, is advantageous to studying post-operative cognitive decline. Because orthopedic surgeries are often elective, it allows for prospective enrollment and completion of baseline cognitive assessments. Orthopedic surgeries generally use a routine procedure and rehabilitation approach, thereby introducing less bias when compared to non-cardiac surgeries, such as abdominal procedures, that may vary widely based on patients’ illness. These surgeries are associated with concerning rates of POCD in older adults [5] as well as with negative perioperative events such as emboli and anemia. While we recognize that it would have been more controlled to have the PD participants receive one type of orthopedic surgery (i.e., knee) this was not possible due to recruitment difficulties in our current investigation. We encourage an improved study design to include a similar proportion of PD and non-PD peers within each orthopedic surgery type. Future studies should also examine severity of PD cognitive change after type of orthopedic surgery, if possible. Fifth, we did not assess non-cognitive features that play a significant role in PD, such as quality of life or symptoms of depression, apathy, anxiety, and pain. There is a possibility that for some individuals, a subtle change in cognitive function after elective surgery may be outweighed by decreased pain, improved mobility, and improved quality of life. Given the complexity of PD motor and non-motor related symptoms, we acknowledge that post-operative cognitive decline in PD may be best studied holistically, rather than focused on cognitive function alone.

Despite these weaknesses, we consider our study well organized from a group design perspective and an appropriate pilot study for identifying challenges and stimulating future research. Our findings that the majority (80% ) of the PD surgery sample experienced cognitive decline greater than that of the healthy surgery and non-surgery peers is noteworthy, particularly given our previous findings that the healthy surgery adults in our sample had decline relative to the non-surgery healthy peers [26]. Given the pilot nature of the study, however, we strongly caution clinicians to avoid using these data as a foundation for clinical decisions regarding surgery for individualswith PD.

In summary, this pilot study supports the feasibility and rationale for investigating surgical risk for individuals with PD. We encourage future larger scale multicenter prospective studies to investigate cognitive change associated with operative events in older adults, but in particular those with neurodegenerative diseases such as PD.

DISCLOSURE OF FUNDING

Supported by the National Parkinson Foundation Center of Excellence Grant (CP), Alzheimer Association NIRG-05-14502 (CP), I. Herrmann Anesthesia Foundation, Gainesville, Florida (CP), NIH General Clinical Research Center Grant # MO1-RR00082, NIA F32AG021363, NIH K23NS60660, NIH R01NR014181 and R01 NS082386.