Statistical Analysis Plan

STATISTICAL ANALYSIS PLAN (SAP) for ALIGN-PD Protocol: ALIGN-PD Clinical Study Protocol SAP Version: v1.0 SAP Date: 2025-10-13 Sponsor/Site: University Clinic Cologne - Neurology Chief Investigator: Professor Michael Thomas Barbe Registry ID (DRKS): DRKS00037920 1. Version history Version: v1.0 Date: 11th Feburary, 2025 2. Abbreviations AE – Adverse Events LFP-adaptive-DBS – Local-feed-potential guided adaptive Deep Brain Stimulation Anatomy-continuous-DBS – anatomy-informed continuous Deep Brain Stimulation CRF – Case Report Form DBS – Deep Brain Stimulation FAS – Full Analysis Set ICH E9 – International Council for Harmonisation Guideline E9 (Statistical Principles for Clinical Trials) ITT – Intention-to-Treat mITT – modified Intention-to-Treat MI – Multiple Imputation MDS-UPDRS – Movement Disorder Society–Unified Parkinson’s Disease Rating Scale NI – Non-Inferiority PD – Parkinson’s Disease PDQ-39 – Parkinson’s Disease Questionnaire-39 PP – Per-Protocol SAP – Statistical Analysis Plan SOP – Standard Operating Procedure SPIRIT – Standard Protocol Items: Recommendations for Interventional Trials TFL – Tables, Figures, and Listings TEED – Total Electrical Energy Delivered 3. Introduction and purpose 3.1 Overview: Prospective, single-centre, randomised, blinded two-period crossover in Parkinson's disease with DBS; interventions: local-feed-potential-guided adaptive DBS (LFP-adaptive-DBS) vs anatomy-informed continuous DBS (Anatomy-continuous-DBS). 3.2 Purpose: Define all confirmatory and descriptive analyses for the main report. Substantial deviations from this analysis plan will require a versioned amendment. Further changes to the analysis plan will be reported in the clinical study report. 3.3 Source documents: ALIGN protocol; ethics approval; CRFs; data management/monitoring SOPs; ICH E9(R1); CONSORT 2010. 4. Objectives and endpoints 4.1 Primary objective: Superiority of LFP-adaptive-DBS vs Anatomy-continuous-DBS with respect to patient preference after both periods. 4.2 Primary endpoint: Forced preference at visit A4 (1 = LFP-adaptive-DBS preferred; 0 = Anatomy-continuous-DBS preferred). A3 and A4 can have -/+ 5 days for change of settings after completion. 4.3 Secondary endpoints: • Movement Disorder Society–Unified Parkinson’s Disease Rating Scale, Part I/II/III (Motor Examination)/IV 1 – MDS UPDRS I/II/III/IV • Parkinson’s Disease Questionnaire-39 2 (PDQ39) • Freezing of Gait Questionnaire 3 (FOG) • Voice Handicap Index 4 (VHI) • Motor Symptom Diary • Wearable-Derived Motor Metrics • Total Electrical Energy Delivered 5 (TEED) • Treatment Expectation for LFP-adaptive-DBS (TEX) • Treatment Expectation for Anatomy-continuous-DBS (TEX) • Local Field Potential Patterns (LFP) • Adverse Events / Serious Adverse Events • Projected Battery Life 4.4. Measuring • Primary endpoint (forced preference): assessed at A4; window +/- 5 days. • MDS-UPDRS III (video): at S1, A3 and A4. • Motor diary/wearables: ≥3 valid days required. • Questionnaires (PDQ-39, FOGQ, VHI, TEX, MDS-UPDRS I/II/IV): at given timepoint, • Safety (AE/SAE): continuous A1 toS1; period-tagged If measurement is outside the defined timepoint, the following rules apply: • Primary endpoint can be measured even if a phase one or two phases were not completed; deviations from completion of the blinded phases by +/- 2 days must be documented. • A five-item Likert-scale assessment of patient preference will be assessed exploratively. • MDS-UPDRS III video are obtained at S1, A2 or A3 and in the specified state regarding medication (ON/OFF) and stimulation (ON/OFF). Rating of videos recorded at the predefined time can be performed by blinded rater at any given time prior to unblinding. Wrong state at filming of video leads to exclusion of the rating for the allocated analysis timepoint. • Motor diary/wearables should at least collect data for > 3 valid days and per day at least 16h of recording. A valid day is considered > 16h of recording. Motor diaries should be filled out beginning the fourth day after change of programs. • Questionnaires are to be obtained at their respective visit of A1, A3 and A4. If items are filled in within window 48h post-visit are allowed for evaluability, however. • Measurement units according to the respective questionnaire metrics. • LFP patterns are measured according to the Medtronic Percept and saved in the respective JSON files after sessions. 5. Trial schedule (short) Visits: Screening; A1; S1; A2 (repeatable, start period 1); A3 (switch); A4 (end period 2, preference); N (3-month phone). Randomisation via R (R Core Team (2021) and randomizeR 7 with fixed seed into permuted blocks 4/6/8 and balanced AB/BA. Allocation concealment via sequentially numbered, opaque, sealed envelope, the seed to generate the list is stored in a password protected file on site. Unblinded investigator opens the allocated envelopes after eligibility is confirmed and programs according to sequence in the envelopes. Sequence is not revealed to blinded sta. Rater for motor MDS-UPDRS and questionnaire assessments will remain blinded. Emergency unblinding possible in case of a medical emergency – reasons and unblinding will be documented in CRF. If possible unblinding will be prevented by employing the unblinded investigator for clinical care. STUDY PERIOD Enrolment Allocation Post-allocation Close -out TIMEPOINT Screenin g A1 (Baselin e, outpatie nt visit) S1 (inpatient program ming) A2 (Start crossover) Week 1-2 A3 (Post 1st phas e Visit) Week 3-4 A4 (Post 2nd phas e Visit) N (3 months , phone) ENROLMENT Eligibility x Informed consent x β-peak detection x LFP Timeline x x x x x Stimulation Parameters x x x x LFP-adaptive-DBS / Anatomycontinuous-DBS programming x x x Allocation x INTERVENTION Anatomycontinuous - DBS x (x) LFP-adaptive-DBS (x) x ASSESSMENTS Disease/Demograp hic data x Questionnaires x x x x Motor exam (MDS - UPDRS III) Med OFF/ Stim OFF x Med ON/Stim On x x AE assessment x x x x x x x x Accelerometry & motor diaries x x Blinding assessment x x Medication documentation x x Program preference x 6. Sample size One-sided exact binomial test (α=0.05) for H₀: p ≤ 0.50 vs H₁: p > 0.50. N = 30 would give power at p=0.75 with >0.80. N=26 would give ≈0.82 power. N=30 chosen to allow 10– 15% dropouts. 7. Statistical methods 7.1 General principles One-sided exact binomial test at a significance level of α = 0.05 for the confirmatory primary analysis. In the case of a non-significant result in the superiority analysis, a non-inferiority analysis with δ = 0.10 will be conducted; this is pre-specified as exploratory, intended to support clinical interpretation, and is not used as the basis for sample size determination. A sensitivity analysis with smaller δ values will be conducted exploratorily; however, the non-inferiority criterion is considered met if the original δ = 0.10 threshold is satisfied. Secondary analyses are exploratory unless stated otherwise. Primary analysis as mITT as specified below. 7.2 Analysis sets Modified Intention-to-Treat (mITT): All randomised participants with given preference will be included in summaries and analyses as randomised. Conservative ITT-aligned sensitivity: Include all randomized participants; impute missing preference = “not preferring LFPadaptive-DBS.” Per-Protocol (PP): Subset of mITT with no major protocol deviations expected to materially aect the primary inference; operational criteria and handling are described alongside endpoint definitions and the schedule of assessments. As-Treated (AT): Analyses according to treatment received; presented as sensitivity. General note. Evaluability, time windows, and deviation handling are governed by the endpoint definitions and the SPIRIT schedule. Any additional clarifications will be documented in the data review plan and reflected in listings. 7.3 Hypotheses (primary) H0: p <= 0.50 (proportion preferring LFP-adaptive-DBS is at most 50 percent); H1: p > 0.50 (LFP-adaptive-DBS is prefered by majority of patients one-sided). Forced preference at A4 (categories: LFP-adaptive-DBS / Anatomy-continuous-DBS); primary coding = 1 for LFP-adaptive-DBS, 0 for Anatomy-continuous-DBS. 7.4 Estimation and models (crossover): Supportive: test for sequencing eect: Fisher Exact will be conducted and reported exploratorily. Sensitivity analysis will be reported. 7.5 Secondary analyses. Secondary endpoints will be analysed using appropriate descriptive and inferential methods compatible with their scale (continuous or categorical), accounting for sequence and period as design factors where relevant. For quantitative secondary outcomes measured after each treatment, data will be analyzed as repeated measures within subjects. Depending on data completeness and distribution, either paired analyses (paired t-test or Wilcoxon signed-rank test) or linear mixed-eects models will be applied to compare the LFP-adaptive-DBS and Anatomycontinuous-DBS conditions. Clinically relevant baseline covariates (e.g., sex, age, disease duration) may be included to improve precision and assessed in statistical analyses. Analyses are exploratory without multiplicity adjustment; results will be reported as eect estimates with descriptive p-values. Subgroup analysis for disease subtype, disease duration, age, LEDD, electrophysiological phenotype based on LFP, MDS-UPDRS subscores. 7.6 Interim analysis: An interim analysis has been added to the study design and submitted to the local ethics committee as a protocol amendment prior to first patient in. The interim analysis will be conducted only after formal approval has been obtained prior to the planned interim time point. Interim would proceed after 18 evaluable preference assessments. The O’Brien-Fleming alpha-spending approach would be applied so that early termination in favor of superiority is only possible if at least 15 out of 18 patients prefer LFP-adaptive-DBS. A non-binding futility recommendation would be issued if fewer than 11 out of 18 patients favor LFP-adaptive-DBS. In all other cases, recruitment will continue up to the full target of 30 patients. This protocol will be updated timely, once ethics approval has been granted. 7.7 Missing data and intercurrent events (ICE): Primary: mITT; if periods are missing or incomplete but preference was given primary comparison is possible. If no preference was given, no preference is imputed. Sensitivity: conservative imputation counting non-available preference as not LFPadaptive-DBS. ICE (envelope opening, programme change, rescue): treatment-policy as main strategy; hypothetical sensitivity analyses and reporting of protocol breaches regarding envelope opening, programme change, rescue. For multi-item scales (e.g., MDS-UPDRS III), if up to two items are missing, impute with group mode; if whole score missing, exclude from primary analysis. 7.8 Multiplicity: Only the primary objective is confirmatory. Secondary endpoints are exploratory without alpha allocation. 7.9 Estimands (ICH E9 R1): Treatment: LFP-adaptive-DBS vs Anatomy-continuous-DBS; Population: adults with PD and LFP-adaptive-DBS ready DBS who meet inclusion criteria and do not meet exclusion criteria, Variable: preference indicator as forced choice between LFPadaptive-DBS (1) and Anatomy-continuous-DBS (0); ICE strategy: as in 7.7; Summary: proportion preferring LFP-adaptive-DBS contrasted against 0.5 7.10 Baseline and disposition: Descriptive statistics. 7.11 Safety: AE/SAE tables by period and overall, graded by severity/relatedness; descriptive comparisons. 7.12 Wearables and TEED (optional): Exploratory summaries for wearable metrics and TEED calculations. 7.13 Software: R (v4.x)(R Core Team (2021) and MATLAB (R20xxb)9 8. Data management and quality Capture via paper CRF system with controlled corrections. Entry in a digital double. Pseudonymisation, roles and permissions, risk-based monitoring. Trial monitoring by the ZKS Köln – Center for Clinical Trials Cologne. Prior to unblinding (before database lock), a blinded review of the cleaned dataset will be conducted using masked datasets by blinded personnel to verify analysis readiness, including data completeness, adherence to visit windows, classification of protocol deviations, and definition of analysis populations. If input from the operationally unblinded investigator is required, it will be limited to factual/operational clarifications and will not involve disclosure or discussion of treatment allocation; the blinding of other team members will be maintained throughout. Any clarifications arising from this review will be documented and, if relevant, incorporated into a versioned SAP amendment signed before unblinding. 9. Deviations from SAP All deviations will be documented and justified; substantial changes may require ethics/authority notification according to local law. 10. Tables, figures 10.1 Table: baseline characteristics. 10.2 Table: primary preference analysis overall and by sequence. 10.3 Figures and Tables secondary endpoints Layouts are not prespecified; formatting and column order may be adapted to journal style without SAP amendment. Content (estimates, confidence intervals, populations) remains as specified 1. Goetz, C. G. et al. Movement Disorder Society-sponsored revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov. Disord. 23, 2129–2170 (2008). 2. Jenkinson, C., Fitzpatrick, R., Peto, V., Greenhall, R. & Hyman, N. The Parkinson’s Disease Questionnaire (PDQ-39): development and validation of a Parkinson’s disease summary index score. Age Ageing 26, 353–357 (1997). 3. Giladi, N. et al. Construction of freezing of gait questionnaire for patients with Parkinsonism. Parkinsonism Relat. Disord. 6, 165–170 (2000). 4. Jacobson, G. P. et al. The Voice Handicap Index (VHI): Development and Validation. Am. J. Speech Lang. Pathol. 6, 66–70 (1997). 5. Koss, A. M., Alterman, R. L., Tagliati, M. & Shils, J. L. Calculating total electrical energy delivered by deep brain stimulation systems. Ann. Neurol. 58, 168–168 (2005). 6. R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 7. Uschner, D., Schindler, D., Hilgers, R.-D. & Heussen, N. randomizeR: An R Package for the Assessment and Implementation of Randomization in Clinical Trials. J. Stat. Softw. 85, 1–22 (2018). 8. O’Brien, P. C. & Fleming, T. R. A Multiple Testing Procedure for Clinical Trials. Biometrics 35, 549 (1979). 9. The MathWorks, Inc. MATLAB (Version R20xxb). The MathWorks, Inc. (2024).