Network correlates of the cognitive response to levodopa in Parkinson disease

Abstract

Objective: Cognitive dysfunction is common in Parkinson disease (PD), even early in its clinical course. This disease manifestation has been associated with impaired verbal learning performance as well as abnormal expression of a specific PD-related cognitive spatial covariance pattern (PDCP). It is not known, however, how this metabolic network relates to the cognitive response to dopaminergic therapy on the individual patient level.

Methods: We assessed treatment-mediated changes in verbal learning and PDCP expression in 17 patients with PD without dementia who underwent cognitive testing and metabolic imaging in the unmedicated and levodopa-treated conditions. We also determined whether analogous changes were present in 12 other patients with PD without dementia who were evaluated before and during the treatment of cognitive symptoms with placebo.

Results: Levodopa-mediated changes in verbal learning correlated with concurrent changes in PDCP expression (r = -0.60, p < 0.01). The subset of patients with meaningful cognitive improvement on levodopa (n = 8) exhibited concurrent reductions in PDCP expression (p < 0.01) with treatment; network modulation was not evident in the remaining subjects. Notably, the levodopa cognitive response correlated with baseline PDCP levels (r = 0.70, p = 0.002). By contrast, placebo did not affect PDCP expression, even in the subjects (n = 7) with improved verbal learning during treatment.

Conclusions: These findings suggest that cognitive dysfunction in PD may respond to treatment depending upon the degree of baseline PDCP expression. Quantification of treatment-mediated network changes can provide objective information concerning the efficacy of new agents directed at the cognitive manifestations of this disease.

Figures

Figure 1. Correlation between baseline measures of cognitive performance and Parkinson disease–related cognitive pattern (PDCP) activity

(A) Relationship between baseline verbal learning performance and PDCP expression in the levodopa (left) and placebo (right) treatment groups. At baseline, higher PDCP expression was associated with more impaired verbal learning performance in both treatment groups. Squares and triangles refer respectively to cognitive responders (R) and nonresponders (NR) to treatment; see Methods. (B) Relationship between baseline PDCP expression and levodopa-mediated changes in verbal learning performance. Higher baseline PDCP scores correlated with greater improvement in cognitive functioning during levodopa treatment. The horizontal dashed line represents the cutoff (+0.44) for meaningful treatment-mediated change in verbal learning based on the reliable change index (see text). The vertical dashed line represents the baseline PDCP value (+1.01) that corresponded to this behavioral response cutoff. Baseline measures of network activity above this value were found to be associated with improved cognitive functioning during levodopa treatment.

Figure 2. Cognition-related responses to levodopa treatment: Network effects

(A) Changes in verbal learning performance with levodopa treatment differed (p < 0.001; 2 × 2 repeated-measures analysis of variance [RMANOVA], see Results) in cognitive responders (LDR; n = 8) and nonresponders (LDNR; n = 9). (B) Levodopa-mediated changes in Parkinson disease–related cognitive pattern (PDCP) expression differed for cognitive responders and nonresponders (p = 0.03; 2 × 2 RMANOVA, interaction effect), with treatment-mediated declines in the former (p = 0.008; post hoc Bonferroni test) but not the latter (p = 0.78). (C) Levodopa-mediated changes in Parkinson disease–related motor pattern (PDRP) expression also differed for the 2 subgroups (p = 0.001; 2 × 2 RMANOVA, interaction effect). For this motor-related metabolic network, levodopa-mediated reductions were present in both cognitive responders (p < 0.001; post hoc Bonferroni test) and nonresponders (p = 0.03). *p < 0.05, **p < 0.01, ***p < 0.001, Student t tests for comparisons of network activity in each subgroup with corresponding healthy control values; see Methods.

Figure 3. Cognition-related responses to placebo

(A) Changes in verbal learning performance differed for cognitive responders (PLR; n = 7) and nonresponders (PLNR; n = 5) to placebo (p = 0.003; 2 × 2 repeated-measures analysis of variance [RMANOVA], interaction effect). (B) Treatment-mediated changes in Parkinson disease–related cognitive pattern (PDCP) expression did not differ for cognitive responders and nonresponders to placebo (p = 0.11; 2 × 2 RMANOVA, interaction effect). Network activity did not change with placebo in either response subgroup (p ≥ 0.19). (C) There was also no change in Parkinson disease–related motor pattern (PDRP) expression with placebo treatment (p ≥ 0.14) in either response subgroup. *p < 0.05, **p < 0.01, ***p < 0.001, Student t tests for comparisons of network expression in each subgroup to healthy control values.