Dopa responsiveness in Parkinson’s disease ============================================ * Sacha E Gandhi * Anahita Nodehi * Michael A Lawton * Katherine A Grosset * Vicky Marshall * Yoav Ben-Shlomo * Donald G Grosset ## Abstract **Background** Dopaminergic responsiveness is a defining feature of Parkinson’s disease (PD). However, there is limited information on how this evolves over time. **Objectives** To examine serial dopaminergic responses, if there are distinct patterns, and which factors predict these. **Methods** We analyzed data from the *Parkinson’s Progression Markers Initiative* on repeated dopaminergic challenge tests (≥ 24.5% defined as a definite response). Growth-mixture modeling evaluated for different response patterns and multinomial logistic regression tested for predictors of these clusters. **Results** There were 1,525 dopaminergic challenge tests from 336 patients. At enrolment, mean age was 61.2 years (SD 9.6), 66.4% were male and disease duration was 0.5 years (SD 0.5). 1 to 2 years after diagnosis, 48.0% of tests showed a definite response, but this proportion increased with longer duration (51.1-74.3%). We identified 3 response groups: ‘Striking’ (n = 29, 8.7%); ‘Excellent’ (n = 110; 32.7%) and ‘Modest’ (n = 197, 58.6%). Significant differences were as follows: striking responders commenced treatment earlier (P = 0.02), were less likely to be on dopamine agonist monotherapy (P = 0.01), and had better cognition (P < 0.01) and activities of daily living (P = 0.01). Excellent responders had higher challenge doses (P = 0.03) and were more likely to be on combination therapy (P < 0.01). **Conclusion** Three distinct patterns of the dopaminergic response were observed. The proportion of PD cases with definite dopa responsiveness increased over time, so the initial treatment response may be an unreliable diagnostic aid. A good dopaminergic response typifies Parkinson’s disease and is a key part of the diagnostic process1. Variable responses are reported in clinical trials2, cohort3 and autopsy studies4, 5 and impact upon clinical outcomes2, 3. However, information is limited on the evolution of the dopaminergic response in PD. Some patients fail to respond to an acute L-dopa challenge in early PD but subsequently respond well to chronic dopaminergic therapy6, 7. Methods of assessing the dopaminergic response vary. An overall grading of the L-dopa response is reported in two studies with autopsy diagnostic confirmation, either in 3 (marked, moderate, and nil response)5 or 4 response groups (definite, good, limited and nil-to-poor)4, while other studies reported cross-sectional findings from on-off motor scoring3, 8. Previous observational studies that have used challenge tests to evaluate the response over time involved small cohorts and applied the superseded Modified Webster scale9–13. A previous report showed that the response amplitude increased over the first 5 years but then remained steady, although longer-term data was available in only 24 cases11. The topic of longitudinal dopaminergic responsiveness is important for two main reasons. First, there are clinical implications: patients with a limited response in early disease have greater motor severity, faster motor progression3 and a lower likelihood of motor complications11. Second, L-dopa responsiveness is a key feature that helps to distinguish PD from other degenerative parkinsonian conditions where the response is more limited and wanes over time. The MDS diagnostic criteria define a clear and dramatic beneficial response to dopaminergic therapy as a supportive criterion for PD, and the absence of an observable response to high dose L-dopa despite at least moderate disease severity an absolute exclusion1. The finding, however, that a small subset of autopsy-proven PD cases appear not to respond to L-dopa5, or have a nil-to-poor response4, questions the concept of dopa- responsiveness being a defining feature of PD. The *Parkinson’s Progression Marker Initiative* (*PPMI*) study has performed standardized serial on-off scoring in relation to prevailing dopaminergic therapy, providing unique prospective information about both initial and subsequent treatment responses14. We analyzed *PPMI* data with 3 aims: (1) to examine whether dopaminergic responsiveness changes over time; (2) to examine if there are, or are not, distinct patterns of the dopaminergic response, using a data-driven analytical approach; and (3) assuming different groups, to test for clinical characteristics as predictors of these treatment response groups. ## Methods *PPMI* is an international multicenter observational study, in which 423 PD patients from 24 centers were followed longitudinally to establish clinical, biospecimen, imaging, and genetic biomarkers of progression14, 15. Patients were recruited up to 2 years after diagnosis if they were 30 years or older, untreated, and if they had at least two of bradykinesia, rest tremor or rigidity, or, alternatively, either asymmetric bradykinesia or an asymmetric rest tremor. Patients were required to have a dopaminergic deficit in dopamine transporter (DaT) imaging with 123I Ioflupane (or vesicular monoamine transporter (VMAT-2) imaging with 18F AV133 in Australia). Imaging data were analyzed by the *PPMI* Imaging Core and defined against pre- specified criteria: visual assessment combined with uptake ratios (< 65% of age- expected putamen binding ratio). Exclusion criteria included a diagnosis of dementia or drug-induced parkinsonism. Patients were followed for up to 13 years, with assessments 3-monthly during the first year and 6-monthly thereafter. Participants underwent comprehensive clinical assessment, including parts I to IV of the Movement Disorders Society-Unified Parkinson’s Disease Rating Scale (MDS UPDRS), with part III (motor) scores in the ‘on’ and ‘off’’ state16. Challenge tests were performed six-monthly to assess dopaminergic responsiveness over time. Cognition was assessed with the Montreal Cognitive assessment (MoCA), adjusted for years of education. *PPMI* data were accessed in November 2022 at [https://www.ppmi-info.org/](https://www.ppmi-info.org/)15. ### Measured responses The effect of dopaminergic therapy (both levodopa and dopamine agonists) was calculated using the following formula: ![Formula][1] Patients withheld their dopaminergic therapy overnight for at least 6 hours (or 12 hours for long-acting drugs) prior to being scored in the ‘off’ state and were then scored in the ‘on’ state 1.5 hours after taking their usual morning dose (L-dopa, dopamine agonist, or both). For patients with deep brain stimulation (DBS), DBS was switched off 1 hour before ‘off’ scores were recorded and was switched on for assessment in the ‘on’ state. Alternatively, and less commonly, patients were scored in the ‘on’ state and waited until their medication wore off when an ‘off’ score was recorded. Levodopa equivalent doses (LED) were calculated by standard formulas17, updated for opicapone18. Responses were analyzed against the 24.5% threshold in the MDS UPDRS part III score, which equates to 30% change in the UPDRS part III score19 and is defined in diagnostic criteria as representing a marked improvement1. Responses were declared ‘definite’ when the improvement in the MDS UPDRS part III score was ≥ 24.5% and ‘limited’ when < 24.5%, as reported previously3. ### Subjective grading Clinicians were asked at repeated time points to answer yes, no, or not applicable to 2 questions relating to the treatment response: (a) ‘Is there little or no response to L- dopa?’ and, following a protocol update in June 2020, they were additionally asked (b) ‘Is there a clear and dramatic response to L-dopa?’. When there were consistent recordings of ‘little or no response’, a case was classed as having *nil response*. When there was ‘a clear and dramatic response’ at least once, the case was defined as responsive. The overall L-dopa response in individual cases was then categorised as *marked* when at least 50% of the subjective gradings recorded a ‘clear and dramatic response’, and *moderate* when this was less than 50%. Subjective clinician assessments were compared with the measured responses from individual challenge tests when collected at the same visit. If a ‘clear and dramatic response’ was recorded as present, the test was graded *marked.* If ‘little or no response’ was recorded as present, the test was graded as showing *nil response.* If both ‘clear and dramatic response’ and ‘little or no response’ were recorded as ‘no’, the test was graded *moderate*. As the second question ‘Is there a clear and dramatic response to L-dopa?’ was introduced in June 2020, a grade of *moderate* could only be ascribed to cases evaluated after this date. ### Modeling approaches A data-driven approach was used to identify if there were, or were not, multiple response clusters using only percentage response and disease duration in a growth mixture model (GMM)20. The model parameters were estimated using Maximum Likelihood Estimation (MLE) through the Expectation-Maximization (EM) algorithm, and the functional form of time was based on fractional polynomials21. In order to identify the number of classes that best fitted the data, we compared the following goodness-of-fit criteria: sample-size adjusted Bayesian Information Criterion (SABIC), entropy values, and the Akaike Information Criterion (AIC)22, 23. After GMM was applied, univariable multinomial logistic regression was performed, using the largest cluster as the reference group, to evaluate the association between clinical covariates and treatment response clusters. Covariates selected for univariable modeling included age at diagnosis, female gender, body mass index (BMI), time to starting any dopaminergic therapy, drug class (i.e. L-dopa monotherapy, dopamine agonist monotherapy and L-dopa plus dopamine agonists), challenge test and total daily dose, education-adjusted MoCA score and MDS UPDRS part I – IV scores, with part III recorded in the on state. The following covariates with a P-value < 0.1 in at least in one group in univariable modeling were incorporated into the multivariable multinomial logistic regression model: age at diagnosis, female gender, BMI, time to starting dopaminergic therapy, drug class, challenge test and total daily dose, education-adjusted MoCA and MDS UPDRS II scores. MDS UPDRS III and IV were excluded from multivariable modeling as they closely correlate with percentage response. The multivariable model mutually adjusted for all included covariates. Results are reported in terms of multinomial odds ratio (95% CI, p-values). In a sensitivity analysis, we adjusted GMM for dosage to assess its impact on cluster identification. An additional analysis excluded observations beyond 9 years to evaluate the influence of late-stage dropout on clustering. All analyses were undertaken using R 4.3.0 software. ## Results Of 423 cases enrolled, 6 (1.4%) had a change in diagnosis and were excluded, leaving 417 cases, of whom 336 (80.6%) had at least one challenge test. 287 of these (85.4%) underwent more than one challenge test. Of the 336 cases, 223 were male (66.4%) and 113 female (33.6%). At enrolment, the mean age was 61.2 years (SD 9.6) and mean disease duration 0.5 years (SD 0.5). Tables 1 and 2 present the demographic and clinical features of the cohort. A total of 1,527 challenge tests were performed, of which 2 were excluded as outliers, leaving 1,525 tests for analysis. View this table: [TABLE 1:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/T1) TABLE 1: Demographic and clinical characteristics in 336 patients with Parkinson’s disease, according to data-derived treatment response group. View this table: [TABLE 2:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/T2) TABLE 2: Challenge test results and dopaminergic therapy in 336 patients with Parkinson’s disease, according to data-derived treatment response group. The 2 excluded challenge tests were from 2 separate cases; one test had a mismatch in the time of the challenge dose and the on and off scores, and the other had clearly inconsistent observations compared with their other scorings. The remaining challenge tests from these 2 cases were included in the analysis. At 1-2 years after diagnosis, 48.0% of tests reached the 24.5% threshold for a definite response, but this proportion was higher for all subsequent 2-year time periods (between 51.1 and 74.3%) (Figure 1). In the 287 cases with more than one challenge test, an initially limited response became definite in 116 cases (40.4%) during a mean follow-up of 28.8 months (SD 19.7), whereas an initially definite response became limited in 72 cases (25.1%) over a mean follow-up of 22.9 months (SD 17.4). The remaining cases did not change category. In the 49 cases with only one challenge test, 25 (7.4% of 336 cases) had a limited response and 24 (7.1%) had a definite response. A limited response below the 24.5% threshold was more likely when motor scores were low, but was seen for all disease durations (Supplementary Figures 1 and 2). Duration of follow-up was similar between cases with an initially limited (55.4 months, SD 35.7) and initially definite response (50.0 months, SD 34.9). 85.6% of initially limited responders and 85.2% of initially definite responders went on to have 1 or more challenge tests. ![Supplementary Figure 1:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F6.medium.gif) [Supplementary Figure 1:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F6) Supplementary Figure 1: Scatterplot of percentage improvement versus ‘off’ motor scores. ![Supplementary Figure 2:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F7.medium.gif) [Supplementary Figure 2:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F7) Supplementary Figure 2: Scatterplot of percentage improvement versus ‘off’ motor scores, according to disease duration. ![FIG. 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F1.medium.gif) [FIG. 1.](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F1) FIG. 1. Percentage change in MDS-UPDRS part III scores in 1525 challenge tests in 336 cases of Parkinson’s disease. Both the challenge test and total daily dose of dopaminergic medication increased over time (Table 3). Dopaminergic drug class affected percentage responsiveness: a significantly greater percentage response was seen with L-dopa compared to a dopamine agonist alone (P < 0.0001). A similar proportion had a definite response when a combination of drug classes was used compared to L-dopa alone (P = 0.37). View this table: [TABLE 3:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/T3) TABLE 3: Challenge test responses according to drug class and disease duration ### Cluster identification from modeling In GMM, we observed identical SABIC values for 2 or 3 clusters (13235) and similar entropy (0.67 and 0.61, respectively) and AIC values (13226.8 and 13223.5, respectively). After truncating the data at 9 years and repeating GMM, SABIC and AIC were lower for three (SABIC=11210.6 and AIC=11199.9) compared to two clusters (SABIC=11212.1, AIC=11203.7) with similar entropy values (3 clusters= 0.63, 2 clusters=0.68). As lower SABIC and AIC values are preferred, a 3-cluster model was favored (Figure 2, Supplementary Figure 3). Entropy provides insight into the "overlap" between clusters, reflecting uncertainty about which cluster some individuals belong to. Despite slightly higher entropy values for two clusters (0.67 for 2 clusters, 0.61 for 3 clusters), the posterior probability of belonging to the assigned class in the 3-cluster model remains high at a 50% threshold (Supplementary Figure 4, Supplementary Table 1). The 3 cluster model was additionally considered more clinically informative and a better reflection of clinical experience (Figure 2, Supplementary Figure 3). ![Supplementary Figure 3:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F8.medium.gif) [Supplementary Figure 3:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F8) Supplementary Figure 3: GMM using 2 clusters. ![Supplementary Figure 4:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F9.medium.gif) [Supplementary Figure 4:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F9) Supplementary Figure 4: GMM with raw data points, shape-coded according to cluster. ![FIG. 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F2.medium.gif) [FIG. 2.](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F2) FIG. 2. Trajectory of dopaminergic responsiveness over time according to data- driven cluster categorization. The following labels were used to describe these 3 clusters: (i) Striking response in 29 cases (8.7%), (ii) Excellent response in 110 cases (32.7%), and (iii) Modest response in 197 cases (58.6%). Sensitivity analysis excluding data beyond 9 years identified the same 3 clusters (Supplementary Figure 5), indicating that attrition did not influence cluster identification, allocation, or trajectory. Loss to follow up rates were similar in striking (20.7%) and modest (23.4%) responders, and most patients were followed for the study duration (76.6% to 84.5% across the 3 clusters) (Supplementary Table 2). ![Supplementary Figure 5:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F10.medium.gif) [Supplementary Figure 5:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F10) Supplementary Figure 5: Sensitivity analysis excluding data beyond 9 years. GMM identified the same three treatment response groups, with no indication of selective dropout of cases from the modest treatment response group. GMM trajectories were additionally adjusted for challenge test dose and to visualize this modelling approach we used a reference dose of 200mg (Supplementary Figure 6). Dose varied by group and time (Supplementary Table 3), most notably in striking and excellent responders. When comparing GMM with and without dose adjustment, modest and excellent responders had very similar trajectories. For the striking group there was a greater reduction in response in later disease, suggesting that higher doses contributed to the better response in these cases. ![Supplementary Figure 6:](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F11.medium.gif) [Supplementary Figure 6:](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F11) Supplementary Figure 6: GMM adjusted for dose using a 200mg reference dose. When comparing GMM with and without dose adjustment, modest and excellent responders have very similar trajectories. Whilst the trajectories for striking responders differ in later disease, this partly reflects the fixed 200mg dose used in the adjusted model. In univariable modeling, striking and excellent responders were significantly more likely to be female, younger, have lower MDS UPDRS II and III ‘on’ scores, higher MDS UPDRS IV and MoCA scores, and have a higher challenge dose compared with modest responders (Figure 3). A striking response was associated with significantly lower BMI. In the multivariable model (Figure 4), striking responders started dopaminergic therapy significantly earlier (P = 0.02), were less likely to be on dopamine agonist monotherapy (P = 0.01) and had significantly better cognition (P < 0.01) and MDS UPDRS II scores (P = 0.01). Excellent responders were significantly more likely to be on a combination of drug classes (P < 0.01) and had a higher challenge dose (P = 0.03). ![FIG. 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F3.medium.gif) [FIG. 3.](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F3) FIG. 3. Odds ratios from univariable modeling of factors associated with dopa responsiveness. ![FIG. 4.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F4.medium.gif) [FIG. 4.](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F4) FIG. 4. Odds ratios from multivariable modeling of factors associated with dopa responsiveness. We then examined how challenge test responses and motor complications differed according to cluster and disease duration (Supplementary Table 4). Most differences between clusters persisted over time, though in the case of daily dose and MDS UPDRS IV the differences between clusters increased over time and then weakened (or were consistent with chance) at 10-13 years of disease, which had the least power. Striking and excellent responders were significantly more likely to have a definite treatment response than modest responders, and had lower motor ‘on’ and ‘off’ scores (Figure 5) and higher MDS UPDRS IV scores (Table 2). ![FIG. 5.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2024/04/27/2024.04.26.24306435/F5.medium.gif) [FIG. 5.](http://medrxiv.org/content/early/2024/04/27/2024.04.26.24306435/F5) FIG. 5. Motor scores in the off and on state in 336 cases of Parkinson’s disease, categorized by test result. ### Subjective grading - individual patients Subjective assessments of the L-dopa response were available in 372 of 417 cases (89.2%), of which 359 (96.5%) had a positive L-dopa response and 13 (3.5%) had ‘little or no response’. In the 359 with a positive response, grading was available for 209 cases and was marked in 196 (93.8%), and moderate in 13 (6.2%). ### Subjective grading - individual challenge tests Clinicians subjectively graded the L-dopa response in 1,366 of 1,525 challenge tests (89.6%); a response was graded as subjectively present in 1,335 (97.7%) and ‘little or no response’ was recorded on 31 occasions (2.3%). The measured response on the challenge test was 32.2% (SD 21.2) when subjectively considered present versus 24.4% (SD 21.5) when tests were subjectively considered to show ‘little or no response’. For cases when the L-dopa response was graded, based on the updated protocol, a marked response corresponded to a measured response of 36.1% (SD 19.0) for 225 tests, while a moderate response corresponded to a measured response of 28.9% (SD 21.2) for 9 tests. ## Discussion The current report is, to our knowledge, the most detailed analysis of the treatment response in PD, covering cases from diagnosis to 13 years later, and assessing the response to both dopamine agonists and L-dopa. Our novel finding is that there are three distinct patterns of dopaminergic responsiveness in PD. A striking response, exceeding 50% improvement, was seen in just under 10% of cases. This degree of improvement was previously reported in a subset of PD cases selected for deep brain stimulation surgery8. An excellent response was seen in around one third, and a modest response in just over half of cases. More modest L-dopa responses have also been reported previously during the first year of treatment in the *ELLDOPA* study, including in patients with a daily dose of 600mg L-dopa2. Our finding was obtained from a primarily data-driven analysis of measured motor scores, without the influence of subjective assessments of the treatment response. Previous observational cohort studies found that on and off motor scores increased9–11, 13, 24, but the response amplitude was either conserved13, 24 or increased over time9–11. Several of these studies were performed in specific PD subgroups with motor fluctuations13, 24, involved the superseded Modified Webster scale9–11, 13, and cohorts that were too small to undertake a detailed data-driven cluster analysis9–13, 24. Our cluster analysis confirms that there is a progressive increase in the difference between ‘on’ and ‘off’ scores over time, both for patients with an excellent and modest response, which represent over 90% of PD cases. In early disease, the lower doses used make it difficult to determine if there is a definite treatment response. However, over time, progressively larger proportions of PD cases fulfil the defined threshold for definite dopaminergic responsiveness. This has significant clinical implications, as many PD cases who initially do not fulfil supportive diagnostic criteria for a positive response to dopaminergic therapy1 will do so later in their disease course. In the 10% of cases with an initial striking response, a further increment was less obvious, perhaps because the ceiling of response had been reached. These differences are not simply explained by higher doses being associated with greater improvement (Supplementary Table 4). From challenge test doses of around 180-220mg in early disease, striking cases had over 50% improvement, whereas excellent responders had 30 to 40% improvement. Modest responders, conversely, had less than 30% improvement from the higher doses (230-260mg) reached in later disease, thereby challenging the concept that greater improvement would be observed at higher doses. Compared with prior reports, a greater proportion of cases were subjectively graded as having a marked response (93.8%), but a similar proportion were identified as poor responders (3.5%). In one study of 69 cases, the response was excellent in 29%, good in 54%, moderate in 13%, and nil-to-poor in 4%4. In another study of 182 cases, 69% had a marked response, 22% were graded as moderate, and 9% had no response5. Both of these reports extracted data from case record review, which may explain the differences from the *PPMI* study, where grading was assessed prospectively. On-off motor scoring using challenge tests in another cohort found that 61% of 1007 cases had a definite response while 39% had a limited response, at 3.4 years3 of disease, which is somewhat higher than the current findings where 51% had a definite response and 49% had a limited response 3-4 years after diagnosis. The lower response in the current analysis may be explained in part by including challenge tests with dopamine agonists, rather than only L-dopa3. There were mismatches between the subjective interpretation of the L-dopa response and the measured scores, for both the absence of an L-dopa response where there was a definite measured response, and vice versa. Accordingly a PD diagnosis should not be excluded based on a subjective assessment of ‘little or no response’ to L-dopa, particularly in early disease, which is in agreement with the conclusions from the *ELLDOPA* study2. Our univariable model results are partially consistent with previous analyses of the demographic and clinical factors associated with L-dopa responsiveness. Our cases with greater dopaminergic responsiveness were significantly younger and had lower motor scores, as reported in a study of 1007 cases undergoing challenge tests2, 3, but there was no age association in another study assessing longer-term treatment responses2. In the current analysis, female patients had a better response than male patients, which compares to no significant gender difference in one study3 and some evidence of a better response in male patients in another2. However, in the multivariable model, both age and gender were consistent with chance. Several drug factors (higher doses, drug class, and starting treatment earlier), and better cognition and activity of daily living scores remained predictors, which were not included in prior studies2, 3. While the methodology is clearly different for challenge tests versus outpatient derived treatment response, an overall interpretation is that age and gender are not key factors related to the treatment response in PD. GMM was chosen as one of the most sophisticated modeling approaches to identify subgroups with different trajectories25–27. While we considered group-based trajectory modeling (GBTM), GMM provides a better fit for the data, allows for within-cluster variability and most closely approximates the true group trajectories underlying the data28. The similar entropy values of the different models, however, does indicate a degree of uncertainty as to which cluster some cases may belong to, which is a limitation of this study. A further limitation is the lack of pathological confirmation of PD. While benign disorders were excluded by confirming presynaptic dopamine depletion on functional imaging, and an expert panel excluded cases with other degenerative parkinsonian disorders, it is possible that some included cases do not have PD. Motor scores were lower at baseline in striking and excellent responders than in modest responders, likely reflecting greater dopaminergic responsiveness. As percentage response was calculated by off score – on score / off score, this could have had a potential floor effect in modest responders. Despite this, percentage change was still considered preferable to absolute change, as it is defined in the MDS diagnostic criteria1 and is consistent with the methodological approach of previous studies2, 3. A small subset of our cohort (n = 35, 10.5%) underwent DBS. Although striking and excellent responders were more likely to have DBS (striking 17.2% and excellent 13.6% vs. modest responders 7.6%), this is likely to be explained by case selection for DBS based on the degree of dopa responsiveness. It is unlikely that DBS confounded the assessment of dopaminergic responsiveness, as it was generally performed in late disease, was switched off for off scores, and a response to challenge tests would still be expected after DBS implantation. Finally, although some patients were lost to follow-up during the later stages of disease, rates of drop out were similar between striking and modest responders. Moreover, the sensitivity analysis identified the same three treatment response groups with no reduction in the proportion with modest responses. In conclusion, we have identified 3 distinct types of dopaminergic medication response in Parkinson’s using a data-driven analysis. Drug factors appear to contribute more than demographic factors to these groupings. As the proportion of PD cases fulfilling criteria for dopa responsiveness progressively increases over time, the initial response to treatment may be unreliable, which is relevant to the diagnostic process. ## Supporting information Supplementary Table 1 [[supplements/306435_file03.pdf]](pending:yes) Supplementary Table 2 [[supplements/306435_file04.pdf]](pending:yes) Supplementary Table 3 [[supplements/306435_file05.pdf]](pending:yes) Supplementary Table 4 [[supplements/306435_file06.pdf]](pending:yes) ## Data Availability All data produced in the present work are contained in the manuscript. ## Author Roles 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 Critique. SEG: 1B, 1C, 2A, 2B, 2C, 3A, 3B. AN: 1B, 1C, 2A, 2B, 2C, 3A, 3B MAL: 2A, 2C, 3B VM: 2C, 3B YBS: 2A, 2C, 3B KAG, DGG: 1A, 1B, 1C, 2A, 2C, 3A, 3B ## Disclosures ### Funding *PPMI* – a public-private partnership – is funded by the Michael J. Fox Foundation for Parkinson’s Research and funding partners, including 4D Pharma, Abbvie, AcureX, Allergan, Amathus Therapeutics, Aligning Science Across Parkinson’s, AskBio, Avid Radiopharmaceuticals, BIAL, Biogen, Biohaven, BioLegend, BlueRock Therapeutics, Bristol-Myers Squibb, Calico Labs, Celgene, Cerevel Therapeutics, Coave Therapeutics, DaCapo Brainscience, Denali, Edmond J. Safra Foundation, Eli Lilly, Gain Therapeutics, GE HealthCare, Genentech, GSK, Golub Capital, Handl Therapeutics, Insitro, Janssen Neuroscience, Lundbeck, Merck, Meso Scale Discovery, Mission Therapeutics, Neurocrine Biosciences, Pfizer, Piramal, Prevail Therapeutics, Roche, Sanofi, Servier, Sun Pharma Advanced Research Company, Takeda, Teva, UCB, Vanqua Bio, Verily, Voyager Therapeutics, the Weston Family Foundation and Yumanity Therapeutics. ### Funding Sources and Conflict of Interest MAL received consultancy fees for advising on a secondary analysis of an RCT sponsored by North Bristol NHS trust. VM has received honoraria from BIAL Pharma, AbbVie and Britannia Pharmaceuticals. KAG has received consultancy fees from Parkinson’s UK. YBS has received consultancy payments from Human Centric DD Ltd and Parkinson’s UK. DGG has received honoraria from BIAL Pharma, Britannia Pharmaceuticals, and UCB Pharma, consultancy fees from the Glasgow Memory Clinic, and grant support from Parkinson’s UK. SEG: none; AN: none. ### Financial disclosures for the previous 12 months MAL received consultancy fees for advising on a secondary analysis of an RCT sponsored by North Bristol NHS trust. VM has received funding for travel to a conference for BIAL and lecture fees from BIAL. KAG has received consultancy fees from Parkinson’s UK. YBS has received consultancy fees from Human Centric DD Ltd and Parkinson’s UK. DGG has received consultancy fees from Parkinson’s UK and NeuroClin Glasgow. SEG: none; AN: none. ### Ethical Compliance Statement *PPMI* was conducted following approval of the local ethics committees of the participating sites, in accordance with national legislation and the Declaration of Helsinki. Written informed consent was obtained from each participant prior to enrollment. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this work is consistent with those guidelines. ## Acknowledgements We would like to thank Dr Tanja Zerenner at the Population Health Sciences Institute at the University of Bristol for her assistance with coding issues. This analysis uses data openly available from *PPMI* (Tier 1), obtained in November 2022 from the *Parkinson’s Progression Markers Initiative* (*PPMI*) database ([www.ppmi-info.org/access-data-specimens/download-data](http://www.ppmi-info.org/access-data-specimens/download-data)), RRID:SCR_006431. For up-to-date information on the study, visit [www.ppmi-info.org](http://www.ppmi-info.org). ## Footnotes * * Dr Sacha E Gandhi and Dr Anahita Nodehi are joint first authors * Received April 26, 2024. * Revision received April 26, 2024. * Accepted April 27, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. ## References 1. 1.Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2015;30(12):1591–1601. 2. 2.Hauser RA, Auinger P, Oakes D. Levodopa response in early Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2009;24(16):2328–2336. 3. 3.Malek N, Kanavou S, Lawton MA, et al. L-dopa responsiveness in early Parkinson’s disease is associated with the rate of motor progression. Parkinsonism & related disorders 2019;65:55–61. 4. 4.Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinical diagnosis of idiopathic Parkinson’s disease: a clinico-pathological study of 100 cases. Journal of neurology, neurosurgery, and psychiatry 1992;55(3):181–184. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiam5ucCI7czo1OiJyZXNpZCI7czo4OiI1NS8zLzE4MSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA0LzI3LzIwMjQuMDQuMjYuMjQzMDY0MzUuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 5. 5.Martin WRW, Miles M, Zhong Q, et al. Is Levodopa Response a Valid Indicator of Parkinson’s Disease? Movement disorders : official journal of the Movement Disorder Society 2021;36(4):948–954. 6. 6.Clarke CE, Davies P. Systematic review of acute levodopa and apomorphine challenge tests in the diagnosis of idiopathic Parkinson’s disease. Journal of neurology, neurosurgery, and psychiatry 2000;69(5):590–594. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiam5ucCI7czo1OiJyZXNpZCI7czo4OiI2OS81LzU5MCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA0LzI3LzIwMjQuMDQuMjYuMjQzMDY0MzUuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 7. 7.Merello M, Nouzeilles MI, Arce GP, Leiguarda R. Accuracy of acute levodopa challenge for clinical prediction of sustained long-term levodopa response as a major criterion for idiopathic Parkinson’s disease diagnosis. Movement disorders : official journal of the Movement Disorder Society 2002;17(4):795–798. 8. 8.Deuschl G, Follett KA, Luo P, et al. Comparing two randomized deep brain stimulation trials for Parkinson’s disease. Journal of neurosurgery 2019;132(5):1376–1384. 9. 9.McColl CD, Reardon KA, Shiff M, Kempster PA. Motor response to levodopa and the evolution of motor fluctuations in the first decade of treatment of Parkinson’s disease. Mov Disord 2002;17(6):1227–1234. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/mds.10244&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12465061&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000179408800012&link_type=ISI) 10. 10.Alty JE, Clissold BG, McColl CD, Reardon KA, Shiff M, Kempster PA. Longitudinal study of the levodopa motor response in Parkinson’s disease: relationship between cognitive decline and motor function. Mov Disord 2009;24(16):2337–2343. [PubMed](http://medrxiv.org/lookup/external-ref?access_num=19890972&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) 11. 11.Clissold BG, McColl CD, Reardon KR, Shiff M, Kempster PA. Longitudinal study of the motor response to levodopa in Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2006;21(12):2116–2121. 12. 12.Nutt JG, Carter JH, Lea ES, Sexton GJ. Evolution of the response to levodopa during the first 4 years of therapy. Ann Neurol 2002;51(6):686–693. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/ana.10189&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12112073&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000175863700005&link_type=ISI) 13. 13.Hughes AJ, Frankel JP, Kempster PA, Stern GM, Lees AJ. Motor response to levodopa in patients with parkinsonian motor fluctuations: a follow-up study over three years. J Neurol Neurosurg Psychiatry 1994;57(4):430–434. [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NDoiam5ucCI7czo1OiJyZXNpZCI7czo4OiI1Ny80LzQzMCI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA0LzI3LzIwMjQuMDQuMjYuMjQzMDY0MzUuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 14. 14.Marek K, Chowdhury S, Siderowf A, et al. The Parkinson’s progression markers initiative (PPMI) - establishing a PD biomarker cohort. Annals of clinical and translational neurology 2018;5(12):1460–1477. 15. 15.Parkinson Progression Marker Initiative. The Parkinson Progression Marker Initiative (PPMI). Progress in neurobiology 2011;95(4):629-635. 16. 16.Goetz CG, Tilley BC, Shaftman SR, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Movement disorders : official journal of the Movement Disorder Society 2008;23(15):2129–2170. 17. 17.Tomlinson CL, Stowe R, Patel S, Rick C, Gray R, Clarke CE. Systematic review of levodopa dose equivalency reporting in Parkinson’s disease. Movement disorders : official journal of the Movement Disorder Society 2010;25(15):2649–2653. 18. 18.Schade S, Mollenhauer B, Trenkwalder C. Levodopa Equivalent Dose Conversion Factors: An Updated Proposal Including Opicapone and Safinamide. Movement disorders clinical practice 2020;7(3):343–345. 19. 19.Merello M, Gerschcovich ER, Ballesteros D, Cerquetti D. Correlation between the Movement Disorders Society Unified Parkinson’s Disease rating scale (MDS-UPDRS) and the Unified Parkinson’s Disease rating scale (UPDRS) during L-dopa acute challenge. Parkinsonism & related disorders 2011;17(9):705–707. 20. 20.Muthén B., Shedden K. Finite mixture modeling with mixture outcomes using the EM algorithm. Biometrics 1999;55(2):463–469. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.0006-341X.1999.00463.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=11318201&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000081320800018&link_type=ISI) 21. 21.Tilling K, Sterne JA, Wolfe CD. Multilevel growth curve models with covariate effects: application to recovery after stroke. Statistics in Medicine 2001;20(5):685–704. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/sim.697&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=11241571&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000167199200002&link_type=ISI) 22. 22.Muthén B, Muthén LK. Integrating person-centered and variable-centered analyses: growth mixture modeling with latent trajectory classes. Alcohol: Clinical and Experimental Research 2000;24(6):882–891. [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1530-0277.2000.tb02070.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=10888079&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F04%2F27%2F2024.04.26.24306435.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000087757000020&link_type=ISI) 23. 23.Tofighi D, Enders C. Identifying the correct number of classes in growth mixture models. In: GR H, KM S, eds. Advances in latent variable mixture models. Greenwich: Information Age, 2007:317-341. 24. 24.Santos-García D, de Deus Fonticoba T, Cores Bartolomé C, et al. Response to levodopa in Parkinson’s disease over time. A 4-year follow-up study. Parkinsonism Relat Disord 2023;116:105852. 25. 25.Paul SS, Thackeray A, Duncan RP, et al. Two-Year Trajectory of Fall Risk in People With Parkinson Disease: A Latent Class Analysis. Archives of Physical Medicine and Rehabilitation 2016;97(3):372–379.e371. 26. 26.Gunzler DD, Gunzler SA, Briggs FBS. Heterogeneous pain trajectories in persons with Parkinson’s disease. Parkinsonism & related disorders 2022;102:42–50. 27. 27.Perepezko KM, Gallo JJ, Pontone GM, Hinkle JT, Mills KA. Association of caregiver strain with the trajectory of quality of life in Parkinson’s disease. Parkinsonism & related disorders 2023;108:105292. 28. 28.Den Teuling N, Pauws S, Van den Heuvel E Clustering of longitudinal data: A tutorial on a variety of approaches. arXiv e-prints, arXiv-2111 2021. [1]: /embed/graphic-1.gif