Background
A good response to levodopa is a key feature of Parkinson's disease (PD), and a poor response suggests an alternative diagnosis, but the extent of variation in the levodopa response in definite PD is not well defined.

Literature Review
A systematic review of articles reporting pathologically confirmed PD and levodopa responsiveness from 1971 to 2018 was performed using the medical subheadings “postmortem,” “Parkinson's disease,” "levodopa," and “l ‐dopa” in PubMed, Embase, and Latin American and Caribbean Health Sciences Literature (LILACS) databases.

Cases
A total of 12 articles described 445 PD cases: 61.7% male, age at disease onset 64.0 years (SD 9.6), age at death 77.1 years (SD 7.2). Levodopa responsiveness was reported in 399 cases (89.7%) either as a graded or a binary response. In the 280 cases (70.2%) describing a graded response, it was excellent in 37.5%, good in 45.7%, moderate in 12.1%, and poor in 4.6%. In the 119 cases describing a binary response (29.8%), 73.1% were levodopa responsive, and 26.9% were nonresponsive. Comorbid brain pathology was present in 137 of 235 cases assessed, being cerebrovascular in 46.0% and Alzheimer's disease in 37.2% of these, but its contribution to levodopa responsiveness was unclear.

Conclusions
The levodopa motor response varies in definite PD. Explanations other than diagnostic inaccuracy should be explored.