Objective: To investigate the reliability of the relationship between levodopa response and subthalamic nucleus (STN) DBS outcomes in Parkinson disease (PD).

Background: Levodopa response (LR) is often cited as the gold-standard predictor of potential DBS response in PD, with “ON-OFF” testing regularly used to determine DBS candidacy at major DBS centers¹. However, predictive value of LR for DBS outcome has varied wildly between studies, with one study demonstrating that use of absolute LR as a predictor artificially inflates correlations even when predictors and outcomes are shuffled randomly².

Method: We calculated the correlation between pre-operative LR in “ON-OFF” testing and DBS response in ON-DBS and OFF-medication condition (averaged over the first 12 months postoperatively) for 99 subjects who underwent STN DBS for PD at Washington University between 2007 and 2017. To evaluate reliability of relationships, we calculated raw correlation coefficients as well as shuffled correlation coefficients, randomly shuffling post-operative scores with pre-operative scores using 1000 Monte Carlo repetitions, with zero-centered correlation distributions expected for unbiased correlates. We used absolute and percent change LR and absolute and percent change DBS response in both shuffled and unshuffled conditions.

Results: Absolute LR correlated significantly with DBS response in both unshuffled (r = 0.47) and shuffled (r = 0.43) conditions. Monte Carlo testing confirmed a consistent positive bias in correlation coefficients for shuffled data. Correlation of percent change LR with DBS outcome was also significant but weaker (r = 0.33) in unshuffled data and absent (r = 0.06) with shuffling, with Monte Carlo testing confirming a minimally biased distribution for shuffled data.

Conclusion: Correlating absolute change LR with DBS response is mathematically flawed, with inclusion of baseline scores artificially increasing correlation coefficients even when preoperative scores are dissociated with outcomes. Percent change LR is less biased, but the relationship is not strong enough to be relied on as a sole predictor of DBS response. Development of additional predictors of DBS outcomes is needed to improve patient selection and ensure optimum outcomes and accurate risk-benefit assessments.

References: 1. Welter ML, Houeto JL, Tezenas du Montcel S, et al. Clinical predictive factors of subthalamic stimulation in Parkinson’s disease. [Internet]. Brain 2002;125(Pt 3):575–83.Available from: http://www.ncbi.nlm.nih.gov/pubmed/11872614
2. Zaidel A, Bergman H, Ritov Y, Israel Z. Levodopa and subthalamic deep brain stimulation responses are not congruent. [Internet]. Mov Disord 2010;25(14):2379–86.Available from: http://www.ncbi.nlm.nih.gov/pubmed/20824733

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/levodopa-response-is-an-inadequate-and-often-biased-correlate-of-stn-dbs-outcomes/