A Systematic Review Comparing Radiofrequency versus Focused Ultrasound Pallidotomy in the Treatment of Parkinson’s Disease

Parkinson’s disease (PD) is the second most common age-related neurodegenerative disorder [1], affecting 0.5–2% of individuals above 65 years old [2]. Surgical treatments with permanent lesions or deep brain stimulation are often sought when patient symptoms fluctuate despite medical optimization and when medication escalation is limited by induced involuntary movements (dyskinesias) [3]. Although deep brain stimulation is more commonly employed than lesion-based approaches today, lesion-based approaches have shown similar efficacy [4] and are preferred in patients for whom a chronically implanted stimulation device carries considerable risks, such as those with medical comorbidities or challenges with long-term follow-up [5, 6].

A well-established lesion-based therapy for PD is radiofrequency (RF) pallidotomy [7‒9]. This open surgical procedure produces thermolesion to the internal segment of the globus pallidus (GPi) via RF waves delivered through an intracranial probe [10]. Recently, magnetic resonance imaging-guided high-intensity focused ultrasound (FUS) has been approved for pallidotomy in patients with PD [11]. In contrast to RF, FUS is an incisionless method, producing lesions via ultrasound waves that can penetrate the skull [11]. Additionally, FUS may offer improved control of lesion location and size in comparison to RF, as MR thermography enables monitoring of lesion size throughout a procedure whereas RF lesions are dependent on intraoperative clinical assessments [12]. However, FUS lesion size may be limited by patient tolerance and skull density ratio [13]. Given the differences in modalities, it is important to compare both the efficacy and risk profiles of RF and FUS pallidotomy. To address this gap, we performed a systematic review comparing the efficacy and risk profile of RF and FUS pallidotomy in patients with PD.

We performed a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [14] with the goal of addressing two questions. First, what is the relative efficacy of RF versus FUS pallidotomy? Second, how do the risks of each procedure compare? Two databases were queried to identify relevant reports. PubMed was queried with (Parkinson’s disease) OR (PD) AND (pallidotomy), and Embase was queried with (“parkinson disease” OR “pd”) AND “pallidotomy.” Published reports from prior to 4/2/24 were included.

All investigators (J.A.G., S.K., and D.D.W.) independently screened all titles and abstracts. A first round of screening was performed to identify original reports of RF or FUS pallidotomy in patients with PD. To this end, reports were excluded if they met any of the following conditions: (1) written in a non-English language, (2) non-primary literature (e.g., review, meta-analysis, opinion piece, description of meeting, guidelines or society statement, erratum), (3) non-human subjects, (4) fewer than 5 subjects underwent RF or FUS pallidotomy, (5) clinical outcomes not assessed (none of Hoehn and Yahr, UPDRS pre- and post-surgery, failure rate, or adverse effects reported), (6) only outcomes following bilateral operations considered, (7) subjects did not have PD, (8) subjects had prior surgical treatment for PD, (9) results not separated by treatment group, (10) outcomes assessed at less than 3 months or an indeterminate amount of time, (11) ambiguity in reported methods or results, (12) unable to obtain report (shown in Fig. 1).

We defined three outcomes of interest for our review: UPDRS III pre- and post-surgery scores, failure rate, and adverse effects. In a second round of screening, reports were excluded if they met any of the following conditions: (1) critical risk of selection bias per Robins-I [15], (2) no outcomes of interest were reported with a sufficiently long follow-up time (Unified Parkinson’s Disease Rating Scale (UPDRS) III pre- and post-surgery scores or overall outcomes at ≥6 months or adverse effects with a maximum follow-up time across participants ≥12 months), (3) concern for shared patients with an included report (shown in Fig. 1).

As several reports did not explicitly specify whether scores reflected measurements from bilateral limbs or only those contralateral to the operated side, we defined UPDRS III scores as putatively contralateral to the operated side or putatively bilateral in cases where laterality was not reported. The distribution of pre-treatment RF and FUS UPDRS III was bimodal, consistent with the report of both kinds of scores across studies (shown in online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000539911). We defined putative contralateral scores as those with an average pretreatment score less than 40, and putative bilateral scores as those with an average pretreatment score greater than or equal to 40 (shown in online suppl. Fig. 1). As the single included FUS study reported putative contralateral UPDRS III scores, we excluded reports with putative or stated bilateral UPDRS III scores for fairer comparison of RF and FUS pallidotomy conditions. If UPDRS III pre- and post-surgery scores were reported only in graphs [16], scores were estimated from graphs using a digital ruler in Adobe Illustrator.

To exclude reports based on concern for shared patients, we first identified pairs of reports with explicit or likely overlap in patients as those for which any of the following held: (1) one or both reports explicitly stated sharing patients with the other report, (2) reports had overlapping surgery dates and shared ≥1 hospital affiliation, (3) surgery dates were not available for at least one report, reports shared ≥1 hospital affiliation and ≥1 authors, and reports had publication years ≤2 apart. If a report belonged to more than one pair, the pairs were joined in a group. A single report was then selected from each group. To do so, each group was first narrowed to reports that reported UPDRS III pre- and post-surgery scores in the text, if these were included in at least one report in the group. If at least one report in the group reported these scores as mean values, the group was further narrowed to reports that included mean values. Among the remaining reports, the report with the most patients who underwent RF or FUS pallidotomy was included.

From each report identified above, we extracted patient demographics including age, sex, and PD duration; post-treatment lesion volume; and outcomes of interest including UPDRS III pre- and post-surgery scores, overall outcomes, and adverse effects. For numerical quantities, we extracted the average, standard deviation (SD), and range where available. For the average, we used the mean where available, and otherwise the median. If SD was unavailable but a 95% confidence interval (CI) was available, we converted the CI to SD assuming normally distributed data: CI = mean ± 1.96 * (SD/sqrt(n)), where ‘sqrt’ is the square root and “n” is the number of subjects. When outcomes of interest were reported at multiple time intervals from surgery, we used those closest to 12 months in the case of UPDRS III pre- and post-surgery scores and failure rate; we reported adverse effects closest to and ≥12 months. All investigators (J.A.G., S.K., D.D.W.) verified the accuracy of study inclusion or exclusion and study outcomes.

For each report, we evaluated risk for bias in a comparison of RF and FUS pallidotomy. We used the ROBINS-I tool to assign a rating for the risk of bias as low, moderate, serious, or critical in several domains: confounding, selection of participants, classification of interventions, deviation from intended interventions, missing data, measurement of outcomes, and selection of reported results [15]. Reports were considered to have at least a moderate risk of bias due to confounding if they were not randomized controlled trials and confounding was expected to exist [15]. Reports that did not characterize important confounders were considered to have at least a serious risk of bias [15]. To obtain a lower bound on the number of reports with a serious risk of bias from confounding, we defined age, disease duration, and lesion volume as important confounders, and reported the fraction of studies that did not characterize all of these quantities. For the selection of participants, reports were considered to have at least a moderate risk of bias if they did not explicitly detail both inclusion and exclusion criteria. Reports were considered to have at least a serious risk of bias if they included a subset of patients with PD in whom outcomes may deviate from the overall population, such as those undergoing staged bilateral pallidotomy. For missing data, reports were considered to have at least a moderate risk of bias if patients were lost to follow-up or otherwise excluded from analysis. For the measurement of outcomes, reports were considered to have at least a moderate risk of bias if patients and providers were not blinded to outcomes used in this review. For selection of reported results, reports were considered to have at least a moderate risk of bias if there was no evidence of pre-registration of outcomes.

We estimated the mean difference and accompanying 95% confidence intervals of UPDRS III post- and pre-pallidotomy scores within RF and FUS conditions and post-pallidotomy scores across RF and FUS conditions using the frequentist approach (MetaInsight v5.1.2 2024, https://crsu.shinyapps.io/MetaInsight/) [17]. Mean difference was considered significant if the accompanying 95% confidence interval did not include zero.

For each report that provided a description of overall outcomes of RF or FUS pallidotomy or scores on a per-patient basis, we defined a “failure rate.” To do so, we used definitions of failure where available. If no definition of failure was provided, we used definitions of success where available. In these cases, we defined failure as the absence of success, and failure rate as one minus the success rate. Finally, if only per-patient scores were available, we defined failure as no improvement in scores following pallidotomy. Across reports, we compared failure definitions and computed the mean failure rate for reports with the same treatment type.

For each report that described adverse effects, we assigned each reported adverse effect to one of the following categories: hemiparesis, speech difficulty, imbalance/gait disturbance, visual disturbance, facial weakness, cognitive deficit, salivation, death, and others. If the same adverse effect was reported at more than one time-point, we counted it only once. If hemorrhage was described as asymptomatic or subclinical and did not require evacuation, we did not count it as an adverse effect. If a report explicitly stated or implied that there were no adverse effects in a category, we assigned a count of zero. On the other hand, if a count could not be determined (e.g., if adverse effects in a category were not assessed), no value was assigned.

Across reports, we summarized counts and fraction of subjects with adverse effects in each category within each treatment type. To assess for a significant difference in counts in each category across treatment types, we used the Fisher exact test as implemented in the SciPy Python package. The Fisher exact test is preferred in the setting of small counts (e.g., fewer than five adverse effects in a treatment group), as it does not require approximating the binomial distribution with a normal distribution which requires large counts to be valid [18].

Our selection process identified 37 RF [7, 19‒54] and 3 FUS [16, 54, 55] reports with at least one outcome of interest (shown in Fig. 1). Patient demographics for these reports are shown in Table 1. The average age of participants was 59.3 years for RF reports and 62.1 years for FUS reports. The average PD duration was 16.0 years for RF reports and 10.0 years for FUS reports. The average post-treatment lesion size was 129.6 mm³ for RF reports and 20.5 mm³ for FUS reports.

Four RF reports [23, 32, 49, 50] and 1 FUS [16] report met the inclusion criterion for analysis of UPDRS III scores (shown in Table 2). In each included report, average UPDRS III scores significantly decreased post-pallidotomy relative to pre-pallidotomy (shown in Fig. 2a). Pre-pallidotomy scores were compared to post-RF or post-FUS pallidotomy scores pooled across reports (shown in Fig. 2b). Across RF reports, there was a significant reduction in UPDRS III scores from baseline following pallidotomy (−13.0, 95% CI: [−17.1, −8.9], n = 96 patients at follow-up) (shown in Fig. 2c). This was also the case for the FUS condition (−11.9, 95% CI: [−19.4, −4.5], n = 6 patients at follow-up) (shown in Table 2). There was not a significant difference in the post-pallidotomy UPDRS III scores between RF and FUS conditions (−1.1, 95% CI: [−9.6, 7.4]).

Eighteen RF reports [7, 20, 22, 23, 25‒28, 30, 33, 34, 37, 39, 41, 44, 46, 49, 52] and 2 FUS reports [54, 55] described overall outcomes with a follow-up of at least 6 months and were included in the failure rate analysis (shown in Table 3). The mean failure rate was 14% across RF reports, and 24% across FUS reports (shown in Table 3). Failure definitions were relatively stringent in both FUS studies, considering outcomes to be a failure if scores did not increase by a fixed, positive amount [54, 55], whereas this was not the case for all RF reports [7, 20, 22, 25‒28, 33, 39, 41, 44, 46, 49] (shown in Table 3).

Twenty-five RF reports [7, 19, 20, 22‒25, 27‒29, 32, 34‒37, 40‒43, 45, 47‒49, 51, 53] (n = 767 patients) and 3 FUS reports [16, 54, 55] (n = 96 patients) described adverse effects of pallidotomy with a maximum follow-up of at least 12 months (shown in Table 4). For RF reports, adverse effects in order of decreasing prevalence were “other” (15.5%), cognitive deficit (8.8%), speech difficulty (6.4%), facial weakness (5.2%), salivation (3.5%), visual disturbance (3.2%), hemiparesis (2.4%), imbalance/gait disturbance (1.2%), and death (0.7%) (shown in Table 4). For FUS reports, adverse effects in order of decreasing prevalence were “other” (60.4%), speech difficulty (7.3%), imbalance/gait disturbance (3.1%), visual disturbance (2.1%), facial weakness (2.1%), hemiparesis (1.0%), cognitive deficit (1.0%), salivation (0%), and death (0%) (shown in Table 4). Between RF and FUS pallidotomy, there was no significant difference in the prevalence of hemiparesis (p = 0.71), speech difficulty (p = 0.66), imbalance/gait disturbance (p = 0.15), visual disturbance (p = 0.76), facial weakness (p = 0.30), salivation (p = 0.06), or death (p = 1.00) (shown in Table 4). By contrast, cognitive deficit was significantly more prevalent in RF reports (p = 0.004), whereas “other” adverse effects were significantly more prevalent in FUS reports (p = 5.58E-20) (shown in Table 4).

Our full assessment of the risk of bias in the comparison of RF and FUS pallidotomy for individual reports is shown in Table 1. All reports [7, 16, 19‒56] had at least a moderate risk of bias due to confounding, as no study directly compared RF and FUS pallidotomy outcomes in a randomized controlled trial and confounders likely exist [15]. Most reports had a serious risk of bias due to not characterizing possible confounders; however, the risk of bias in other categories was generally moderate or low.

FUS pallidotomy was recently approved for the treatment of PD, but how this new treatment compares to traditional methods of pallidotomy has not been established. To address this gap, we performed the first systematic review comparing FUS pallidotomy with the previous standard of care for pallidotomy in the treatment of PD, RF ablation [6, 57].

UPDRS III scores are commonly used to quantify the motor signs of PD and provide one measure of treatment efficacy. Across reports identified by our systematic review, UPDRS III scores significantly decreased following RF pallidotomy, consistent with a prior meta-analysis [58] and the well-established efficacy of RF pallidotomy in the treatment of PD [9]. The same was true following FUS pallidotomy. We found that the improvement in scores was not significantly different across RF and FUS modalities; however, this could be due to low statistical power given the small sample size in the FUS group (6 patients from one study).

Failure rate is another measure of efficacy and captures whether a treatment was effective overall. The failure rate was lower for RF versus FUS reports (14 vs. 24%). This could be due to the smaller lesion size of FUS pallidotomy compared to RF pallidotomy. However, there was significant heterogeneity in the definition of failure rates that makes this difference difficult to interpret. In particular, for several RF reports, the definition of failure was relatively stricter by requiring clinical worsening. This would be expected to produce lower failure rates in the RF group. Finally, in the majority of RF reports, the definition of failure was not quantitative and instead required subjective assessment. This may introduce variability that also challenges the interpretation of failure rate differences across RF and FUS treatments.

Patients undergoing surgery are subject to both the benefits as well as the risks of the procedure. In addition to comparing the efficacy of RF and FUS reports, we compared the prevalence of adverse effects with both treatments. For most categories, there was no significant difference in adverse effect prevalence across RF and FUS treatments, including hemiparesis, speech difficulty, imbalance/gait disturbance, visual disturbance, facial weakness, salivation, or death. A caveat is that some of these adverse effects, for example, death, were rare with both treatments, such that a very large number of subjects would be required to detect a significant difference if it exists.

In two categories, the prevalence of adverse effects differed significantly across RF and FUS reports. First, cognitive deficit was significantly more prevalent in RF than in FUS reports (8 vs. 1%, p = 0.004). A possible explanation for this difference is that as an incision-less procedure, FUS may be less likely to induce damage and inflammation that could impair cognition. On the other hand, patient characteristics may also confound this difference. Patients who underwent RF had a longer PD duration on average and may therefore be more vulnerable in the setting of surgery. Lesions also tended to be larger with RF as compared to FUS treatment in the limited number of studies that reported lesion volumes, although lesion size was not found to correlate with cognitive changes following pallidotomy in a previous report [59].

On the other hand, “other” adverse effects were more common in FUS than in RF reports. This is most likely driven by more extensive reporting of minor adverse effects in FUS as compared to RF reports. As an example of this extensive reporting, the 24 “other” adverse effects in a FUS study with 8 patients consisted of mild headache, pin-site pain, and back pain from lying in a fixed position, all minor effects that are expected to resolve within hours or days. The reason for this difference in reporting may relate to study types. All FUS studies were prospective, whereas some RF studies were retrospective. In comparison to prospective studies, retrospective studies tend to report fewer adverse effects [60, 61].

Our comparison of FUS and RF pallidotomy is limited by several factors. The first is data availability. FUS pallidotomy is a relatively new intervention and very few reports have described its effects. Our screening criteria identified three such reports [16, 54, 55]. Additionally, heterogeneity in the types of outcomes reported by individual studies further reduced the number of participants available for comparisons of FUS and RF pallidotomy outcomes. We anticipate that within the next few years, additional reports of FUS pallidotomy outcomes will enable improved comparisons of RF versus FUS pallidotomy efficacy and adverse effects. In future studies, the use of standardized outcomes would also greatly facilitate these comparisons.

Our comparison of overall outcomes following FUS versus RF pallidotomy was limited by the heterogenous and often subjective nature of outcome definitions across reports, which likely confound differences in failure rates across treatment modalities. This precluded a direct comparison of failure rates across modalities. Although FUS reports provided quantitative overall outcome definitions, the majority of RF reports did not, making it difficult to assess the extent to which definitions of failure were similar or different across RF versus FUS reports.

Our comparison of adverse effect prevalence following RF versus FUS was limited by two factors. First, adverse effects appeared to be less extensively documented in RF than in FUS reports on average. This could skew the perception of safety between RF versus FUS pallidotomy. Second, differences in population characteristics may have confounded differences in adverse effect prevalence. In particular, the average duration of PD was shorter in the FUS population, which could affect the likelihood of reported adverse effects based on differences in baseline symptom severity.

In all comparisons, there is a high risk of bias due to the fact that most studies were not randomized controlled trials, and controlling for potential confounders was limited by incomplete reporting of potential confounding variables such as lesion volume. This limits the reliability of conclusions that can be drawn from comparisons of outcomes at present. In future studies, greater standardization of reported outcomes and potential confounders would greatly facilitate the comparison of FUS and RF efficacy and safety.

FUS pallidotomy is a new treatment modality for PD. Limited data and heterogeneity in outcome reporting challenge comparisons of FUS and RF pallidotomy efficacy and safety. Our systematic comparison of 3 reports of FUS pallidotomy and 37 reports of RF pallidotomy suggests that FUS pallidotomy may have similar efficacy and lower risk of cognitive impairment relative to RF pallidotomy, the previous standard of care for pallidotomy in the treatment of PD, although additional reports of FUS pallidotomy are needed to ensure adequate power and rule out potential confounders of these results.

As only publicly available research was used, study approval and informed consent from participants was not required.

D.D.W. is a consultant for Boston Scientific Inc., Iota Biosciences Inc., and Insightec Inc. The other authors declare no competing interests.

This study was not supported by any sponsor or funder.

Conceptualization: D.D.W. Design and execution of systematic review and writing: J.A.G., S.K., and D.D.W.

Data derived from the cited studies and all quantities analyzed here are presented in the article. Further inquiries can be directed to the corresponding author.