Objective: To evaluate the structural validity of new cognitive composites in Parkinson’s disease (PCCs).

Background: PCCs created from constituent tasks are promising in their single-index parsimony that is sensitive to the earliest cognitive change with augmented power to reduce trial sample size needs. However, several types of validation are needed, including structural verification of whether the constituent tasks load onto the PCC and whether tasks should be weighted or not in calculating the PCC. Two unweighted PCCs have been published using the Parkinson’s Progression Markers Initiative (PPMI) without structural validation.

Method: We reanalyzed PPMI (March 2025 freeze) selecting individuals at baseline who had completed: Montreal Cognitive Assessment, Letter-Number Sequencing, Symbol Digit Modalities, Category Fluency, Judgement of Line Orientation, and Hopkins Verbal Learning Test. We used confirmatory factor analysis to test two PCCs—Weintraub’s (6 tasks, standardized & raw scores) and Simitsi’s (5 tasks, raw scores)—using robust Satorra-Bentler maximum likelihood estimation. We also compared the validity of weighted (i.e., freely estimated) versus unweighted (i.e., tau-equivalent) PCCs. Validity was evaluated with five fit indices: chi-square, standard root mean residual, root mean square error of approximation, comparative fit index, Tucker-Lewis index.

Results: 4,075 individuals (Mage=64.9, Meducation=16.3, 54.8% men, 95% white) met inclusion criteria. The Weintraub PCC was invalid on all five fit indices. Using raw scores marginally improved the fit on four indices, but was still invalid. Using a weighted approach, the fit of the PCC significantly increased, with borderline validity on one index when using raw scores, but still invalid on four fit indices. The Simitsi PCC was also invalid on all five fit indices. However, when calculated with a weighted approach, it was valid on four indices.

Conclusion: Although mathematically plausible, two proposed PCCs are structurally invalid. When the PCCs were weighted, the Simitsi index was valid, but the Weintraub PCC was not. Caution in use and interpretation of these PCCs—as published—is warranted, and new PCCs should be created with multipronged validation.

Table 1

References: 1. Weintraub D, Brumm MC, Kurth R, York MK, Parkinson’s Progression Markers Initiative. Use of Robust Norming to Create a Sensitive Cognitive Summary Score in De Novo Parkinson’s Disease: An Illustrative Example. Mov Disord. Published online January 6, 2025. doi:10.1002/mds.30111

2. Simitsi AM, Sfikas E, Koros C, et al. Motor and nonmotor features of p.A53T alpha-synuclein PD vs idiopathic PD: longitudinal data from the PPMI study. J Neurol. 2025;272(3):203. doi:10.1007/s00415-024-12836-w

3. Stricker NH, Twohy EL, Albertson SM, et al. Mayo-PACC: A parsimonious preclinical Alzheimer’s disease cognitive composite comprised of public-domain measures to facilitate clinical translation. Alzheimer’s & Dementia. 2023;19(6):2575-2584. doi:10.1002/alz.12895

4. Goetz CG, Choi D, Guo Y, Stebbins GT, Mestre TA, Luo S. It Is as It Was: MDS-UPDRS Part III Scores Cannot Be Combined with Other Parts to Give a Valid Sum. Movement Disorders. 2023;38(2):342-347. doi:10.1002/mds.29279

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/caution-in-the-creation-of-parkinsons-disease-cognitive-composites/