Statistics

Patient characteristics were summarized using mean and standard deviation (SD) for continuous and percent for categorical variables. In patients where both pre- and post-L-DOPA and placebo rsFC date were available and analyzable (n = 31; see Results and Supplement), response (change in VS-vmPFC rsFC; post minus pre) to L-DOPA with respect to placebo was first examined by repeated-measures general linear model with plasma CRP concentration as a continuous variable, both with and without clinical and demographic covariates that may confound relationships between inflammation and the brain and behavior (age, sex, race and body mass index [BMI]), as well as study-related variables such as plasma L-DOPA concentrations, order of treatment, etc. (see Supplement for details) in separate analyses. To further interpret a significant treatment by CRP interaction, response to L-DOPA and placebo was plotted across a range of CRP levels to identify a potentially sensitive cut-point. To assess whether rsFC responses to L-DOPA with respect to placebo were significantly different in patients above versus below the identified CRP cut-point (>2 mg/L, see Results and Supplement), and consistent with analyses for variables where only post-challenge data were available, generalized estimating equations (GEE) with identity link function and interchangeable working correlation matrix were implemented both with and without covariates. Linear regression models probed relationships between rsFC response to L-DOPA and CRP (as a continuous variable) and the inflammatory composite score. Backward and forward linear regression using the same criteria for entry and removal were employed to identify which marker (from the composite score) was the most significant predictor. Logistic regression and receiver operating characteristic (ROC) analyses also examined whether rsFC response to L-DOPA (change in rsFC Z-scores, post minus pre) was able to classify patients as having CRP > 2 mg/L versus other potential cut-points (CRP 1 or 3 mg/L). GEE was further utilized to determine whether patients with higher CRP levels had higher VS-vmPFC rsFC after L-DOPA with respect to placebo in the full sample with analyzable post-challenge data (n = 40; see below and Supplement). Similar analyses examined potential effects of CRP level after L-DOPA with respect to placebo on tbFC during reward anticipation (win>neutral, MID), motivation (proportion of high effort choices, EEfRT), and anhedonia response (SHAPS scores, post minus pre), with and without covariates. Relationships between rsFC and behavior were examined by linear regression in models including treatment, CRP level and the effects of their interaction. See detailed Power Calculation in Supplement. All tests of significance were two-tailed with α < 0.05, conducted in IBM SPSS Statistics 28.