Long-term follow-up of a randomized AAV2- GAD gene therapy trial for Parkinson's disease

Abstract

BACKGROUND. We report the 12-month clinical and imaging data on the effects of bilateral delivery of the glutamic acid decarboxylase gene into the subthalamic nuclei (STN) of advanced Parkinson's disease (PD) patients. METHODS. 45 PD patients were enrolled in a 6-month double-blind randomized trial of bilateral AAV2-GAD delivery into the STN compared with sham surgery and were followed for 12 months in open-label fashion. Subjects were assessed with clinical outcome measures and ¹⁸F-fluorodeoxyglucose (FDG) PET imaging. RESULTS. Improvements under the blind in Unified Parkinson's Disease Rating Scale (UPDRS) motor scores in the AAV2-GAD group compared with the sham group continued at 12 months [time effect: F(4,138) = 11.55, P < 0.001; group effect: F(1,35) = 5.45, P < 0.03; repeated-measures ANOVA (RMANOVA)]. Daily duration of levodopa-induced dyskinesias significantly declined at 12 months in the AAV2-GAD group (P = 0.03; post-hoc Bonferroni test), while the sham group was unchanged. Analysis of all FDG PET images over 12 months revealed significant metabolic declines (P < 0.001; statistical parametric mapping RMANOVA) in the thalamus, striatum, and prefrontal, anterior cingulate, and orbitofrontal cortices in the AAV2-GAD group compared with the sham group. Across all time points, changes in regional metabolism differed for the two groups in all areas, with significant declines only in the AAV2-GAD group (P < 0.005; post-hoc Bonferroni tests). Furthermore, baseline metabolism in the prefrontal cortex (PFC) correlated with changes in motor UPDRS scores; the higher the baseline PFC metabolism, the better the clinical outcome. CONCLUSION. These findings show that clinical benefits after gene therapy with STN AAV2-GAD in PD patients persist at 12 months. TRIAL REGISTRATION. ClinicalTrials.gov NCT00643890. FUNDING. Neurologix Inc.

Figure 1. Clinical improvements after gene therapy.

(A) Changes in mean OFF-state Unified Parkinson’s Disease Rating Scale (UPDRS) motor scores at baseline and 1, 3, 6, and 12 months in the AAV2-GAD (red, n = 16) and sham (black, n = 21) groups. After surgery, the patients in both groups showed decreased scores over time (time effect: P < 0.001 by 2-way RMANOVA), with greater improvements in the AAV2-GAD treatment group over all follow-up time points (group effect: P < 0.03; 2 × 5 RMANOVA; *P < 0.05, **P < 0.01, ***P < 0.001, post-hoc Bonferroni tests relative to baseline). (B) Changes in the duration of levodopa-induced dyskinesias (LID) (item 32 of the UPDRS) in the AAV2-GAD (red, n = 16) and sham (black, n = 21) groups over the course of the study. There was a significant difference in duration of LID over time between the two groups (interaction effect: P < 0.02, 2 × 5 RMANOVA). (C) Rate of responders and nonresponders at 12 months. A cutpoint of improvement of 9.0 points identified a significantly greater (χ² = 5.64, P < 0.02) responder rate in the AAV2-GAD group (10 of 16, 62.5%) than in the sham group (5 of 21, 23.8%) at both time points. A cutpoint improvement of 1 hour for diary-based ON time revealed a significant difference between the two groups (χ² = 3.86, P < 0.05), with a larger percentage in the AAV2-GAD group (10 of 14, 71.4%) categorized as responders than in the sham group (7 of 19, 36.8%). Two subjects each in the AAV2-GAD and sham groups were missing 12-month diary data and were omitted from this analysis.

Figure 2. Changes in regional metabolism after gene therapy.

(A) Representative slices displaying significant changes in glucose metabolism in the thalamus, caudate/putamen, prefrontal cortex (BA 8/9/10), and anterior cingulate cortex (BA 23/24) in the AAV2-GAD group (n = 16) compared with the sham group (n = 21) after gene therapy (P < 0.001, uncorrected; SPM RMANOVA). (Increased/decreased metabolism is indicated by red/blue.) (B) Mean metabolic activity in these regions was plotted for the AAV2-GAD (red, n = 16) and sham (black, n = 21) groups. In all of these areas, changes in regional metabolism were different for the two groups (interaction effect: P < 0.005; 2 × 3 RMANOVA), with significant declines over time only in the AAV2-GAD group (P < 0.005; **P < 0.01; ***P < 0.001, post-hoc Bonferroni tests relative to baseline).

Figure 3. Correlation between baseline metabolism in the prefrontal cortex (PFC) and clinical response.

In the AAV2-GAD group (n = 16), an inverse correlation was evident between the baseline metabolism in the PFC and the changes in Unified Parkinson’s Disease Rating Scale (UPDRS) motor scores at 6 months (A: r = –0.51, P < 0.05; Pearson’s correlation) and 12 months (B: r = –0.71, P < 0.003; Pearson’s correlation) from baseline. No correlation was seen in the sham group (n = 21) at either 6 months (C: r = -0.35, P = 0.12; Pearson’s correlation) or 12 months (D: r = -0.04, P = 0.86; Pearson’s correlation).

Figure 4. Trial profile.

Figure 5. Display of abnormal glucose metabolism (P < 0.001, uncorrected, by SPM t test) in two representative patients using the statistical parametric mapping single-case analysis (26).

(Increased/decreased metabolism is indicated by red/blue.) Based on the automated image-based diagnosis for individual subjects (22), one patient was classified as having idiopathic Parkinson’s disease (IPD) with a likelihood of 99%, consistent with this subject’s clinical diagnosis, and was enrolled into the study. The other subject was classified as having multiple system atrophy (MSA) with a likelihood of 99%, despite a clinical diagnosis of IPD, and thus was excluded from the study.