The Dark Matter of Levodopa Pharmacodynamics

While clinicians have used levodopa preparations with great effect for decades, we have a surprisingly limited understanding of the basic features of and the mechanisms underlying levodopa effects in Parkinson disease (PD). A fundamental distinction in levodopa effects is between the short-duration response (SDR), which is locked to plasma and presumably striatal levels of levodopa, and the long-duration response (LDR), an effect that follows chronic treatment, building up (and decaying after treatment cessation) over days to weeks, independent of plasma levels. As measured by standard clinical rating scales, the SDR and LDR each account for approximately half of the total therapeutic effect of levodopa.1-5 The combined SDR-LDR effects explain the consistent treatment response seen in mild-to-moderate PD patients despite the short half-life of levodopa. The complexity of levodopa pharmacodynamics (LDPD) was recognized by Cotzias in his seminal publications on levodopa treatment of PD.6, 7 Muenter characterized, and coined the terms SDR and LDR, in the early 1970s.8 Despite this early recognition and the clinical importance of these phenomena, there has been relatively little systematic study of LDPD. A Pubmed search for ‘levodopa’ returns >25,000 entries. A search using the terms ‘levodopa’, ‘short-duration response’, and ‘long-duration response’ results in <100 entries, many tangential to LDPD. The thin prior literature on LDPD is doubly surprising in view of the fact that study of striatal dopaminergic signaling was and is a major focus of basic neuroscience research. What accounts for the neglect of these clinically and scientifically salient phenomena? Part of the answer is the difficulty in studying LDPD, particularly the LDR, in vivo. Peter Kempster described the LDR as the “dark matter” of LDPD, as its effects are inferred.9 Muenter's pioneering approach was to study previously untreated PD subjects at pretreatment baseline, after a single acute treatment to measure the SDR, and after chronic treatment in the practical “OFF” and “ON” states to assess the LDR and SDR + LDR states. This approach is limited to de novo subjects, which in high-income nations means individuals with early PD. The alternative is to study motor performance after treatment withdrawal, but for how long? The prolonged withdrawals likely required (see later) are neither feasible nor ethical. Preclinical model experiments should be instructive, but while SDR-like effects can be readily demonstrated, there is no validated LDR model. The absence of validated LDR models is attributable to several factors. Following the seminal work of Schultz on striatal dopaminergic mediation of reinforcement learning, research on striatal dopaminergic signaling over the last generation focused on phasic signaling, which may play a limited role in LDPD.10, 11 Most neuroscientists are not aware of the complexities of LDPD, generally assuming that the SDR is all there is to levodopa action in PD. Finally, inadequate characterization of potentially differentiating behavioral features of the SDR and LDR is an obstacle to specifying strong criteria for model validation. Work in the 1990s substantiated and extended Muenter's findings, also providing some insight into potential mechanisms. A credible initial model was that both the SDR and LDR reflected a relatively rapid paracrine effect of extracellular levodopa metabolized to dopamine. In this model, the SDR was driven by plasma levels and the LDR resulted from a residual intrastriatal reservoir, likely surviving nigrostriatal terminals, releasing dopamine between plasma level peaks. Nutt and a group of Italian investigators established that the LDR could be induced/maintained by dopamine agonists.12, 13 These results suggested an alternative explanation, namely that the LDR is a neuroplastic consequence of chronic receptor stimulation. This inference is consistent with more recent post-mortem analysis indicating that the dopamine synthetic capacity of surviving posterior putaminal nigrostriatal terminals is negligible within 3–4 years after diagnosis, undermining the reservoir concept.14 Prior work left open major questions about critical features and the natural history of the LDR. What is the LDR half-life? Hauser and Holford used clinical trial washout data to derive an estimate of approximately 8 days.15 This analysis assumed that the time constants of LDR induction and decay are identical. Larger time constants are possible. Nutt and Holford used pharmacodynamic modeling of some trial and other data to suggest that levodopa effects plateau after months of treatment.16 The trajectory of LDPD is also uncertain. An attractive model was that LDR magnitude declined with disease progression (Fig. 1A).17, 18 The fading LDR was thought to contribute to the emergence of motor fluctuations and indirectly to peak dose dyskinesias as the relative magnitude of the SDR increased. This conceptualization was challenged by Cilia et al. in a study of PD patients in sub-Saharan Africa who gained access to levodopa only after a mean of 7 years (±4 years) after disease onset.3 They documented a substantial LDR in these more advanced PD subjects at approximately 60% of the total response and constant across the three different disease duration intervals (<5, 5–10, and >10 years) studied (Fig. 1B). In this issue of Movement Disorders, Schnalke and Feige et al. utilize longitudinal data from the Parkinson's Progression Markers Initiative (PPMI) to explore the natural history of the LDR.5 They included 148 PD patients with available International Parkinson and Movement Disorder Society sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) motor scores recorded in their pre-levodopa or dopamine agonist treatment history as baseline as well as in the practically-defined overnight OFF state (to determine the LDR) and subsequent ON state after levodopa intake (to determine the SDR) within 1.5 years after initiation of levodopa. The first “measured” LDR was calculated as MDS-UPDRS-Part III difference from the actual drug-naïve OFF state to the practically defined OFF and was approximately 2 points. In 58% of participants, the “measured” LDR was positive and accounted for more than 53% of the total response to levodopa (ie, LDR + SDR), while in 42% of participants the value for the LDR was negative (meaning that there was probably no LDR or that disease progression over ~1 year overwhelmed the LDR). For further longitudinal analyses, levodopa-naïve OFF motor scores had to be inferred, similar to previous studies, which is an important limitation. In this case, the authors used a mixed linear model on MDS-UPDRS-III scores from a different larger set of 245 untreated PPMI PD patients with at least three consecutive motor assessments without any antiparkinsonian medication. The derived predicted drug-naïve OFF motor scores progressed at 2.65 points/year and allowed calculation of LDRs at follow-up assessments, divided into 2-year intervals, up to 10 years after baseline. Among the 148 participants analyzed, predicted LDR was positive in about 70%, while the remainder had a negative LDR at all available visits. In subjects with positive LDRs, LDR magnitude accounted for 40%–60% of the total levodopa effect and this proportion remained unchanged over the 10-year follow-up period (Fig. 1B). These latter results are consistent with the results of Cilia et al.3 In a meta-analytic approach based on placebo arms of accumulated trial data and analogous modeling of natural history, Pauwels et al. also reported similar results.19 Schnalke and Feige et al. report that relative strengths of the SDR and LDR are stable in progressing disease, while Pauwels et al. suggest that the magnitude of the SDR increases somewhat during disease progression. Schnalke and Feige et al. make other interesting observations and suggestions. Restricting analyses to the group with positive LDRs, there was also no association of LDR magnitudes with motor complications (or non-motor symptom burden and cognitive performance). Though they are careful to specify that they have limited observations in more prolonged PD, the authors also point out that the existence of a stable LDR mandates that the previously hypothesized LDR erosion with disease progression cannot be an explanation for motor fluctuations or peak-dose dyskinesias. Pauwels et al. reached similar conclusions and both suggest that motor fluctuations result from overall disease progression in the face of stable, but ultimately limited, SDR and LDR magnitudes.5, 19 While motor fluctuations are an inevitable aspect of disease progression and occur as soon as the SDR becomes larger than the minimal clinically important difference (MCID; Fig. 1B), peak-dose dyskinesias occur in a large fraction of, but not all, PD patients. Consequently, disease progression, by itself, is not an adequate explanation for peak-dose dyskinesias. They are generally conceptualized as a manifestation of pathologic striatal neuroplasticity secondary to non-physiologic, perhaps pulsatile, receptor stimulation. This framing raises the intriguing prospect that much of what we see in the evolution of LDPD with chronic treatment and disease progression is the consequence of different forms of neuroplasticity – a beneficial form that results in the LDR and a deleterious form resulting in peak-dose dyskinesias. Lastly, as has been seen in other studies, Schnalke and Feige et al. could not document LDR emergence in a substantial fraction, 30%–40%, of the studied PPMI subjects. This may be partly attributable to variation in treatment intensity. In longitudinal analyses, they note that these LDR-negative subjects appear to have faster disease progression and are more likely to exhibit features identified in other studies as associated with more rapid progression of PD. This intriguing suggestion points to an unexpected aspect of disease heterogeneity and may be useful in selecting/stratifying subjects in disease-modification trials. As is typical with interesting results, the work of Schnalke and Feige et al. raises a number of important questions. They present an attractive model of LDPD during disease progression, but at a basic descriptive level we cannot be certain about LDPD features across the spectrum of PD. The “dark matter” nature of the LDR, the relative paucity of longitudinal data, and the need to extrapolate the natural history of untreated PD are obstacles to accurate characterization. We likely need to identify features of the SDR and LDR (and SDR + LDR) states that would permit characterization across the spectrum of PD. It is plausible, for example, that these different treatment states have distinctive behavioral features.4 More speculatively, resting-state functional connectivity or task-based magnetic resonance imaging methods might reveal distinctive signatures. This type of work would also be useful for validating preclinical models. Mechanistic understanding of LDPD would be scientifically fascinating and might offer new targets to enhance treatment of PD. (1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Editing.R.L.A.: 3A, 3B. D.K.L.: 3A, 3B. P.M.: 3A, 3B. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.