Clinical Outcomes of Stroke in MalIgnanCy

BACKGROUND AND RATIONALE Cancer and stroke increasingly coexist because both conditions are common and survival after cancer has improved. Contemporary evidence indicates that approximately 10-15% of patients with ischemic stroke have a history of cancer, with active malignancy present in a substantial proportion. The relationship is biologically heterogeneous: conventional mechanisms such as atrial fibrillation and atherosclerosis remain frequent, while cancer may additionally promote arterial thrombosis through systemic hypercoagulability, nonbacterial thrombotic endocarditis (NBTE), disseminated intravascular coagulation (DIC), tumor embolism, paradoxical embolism, treatment-related vascular injury, and other mechanisms.

Patients with cancer-associated ischemic stroke are clinically important because recurrent thromboembolism, disability, and mortality are frequent. Earlier cohort data demonstrated a high burden of recurrent arterial and venous thromboembolic events after ischemic stroke in active cancer, while the 2026 American Heart Association (AHA) scientific statement formalized a pragmatic classification of cancer-related ischemic stroke into probable, possible, and uncertain causal-link categories. This framework is particularly relevant to COSMIC because it will capture the required components, including active cancer status, disease spread/progression, D-dimer, NBTE, overt DIC, infarct distribution, transcranial Doppler microemboli, and the susceptibility vessel sign.

Intracerebral hemorrhage (ICH) in patients with malignancy represents a distinct but less systematically studied problem. Systemic cancer has been associated with worse in-hospital outcomes after ICH in large administrative datasets, while the causal spectrum includes hypertensive arteriopathy and cerebral amyloid angiopathy (CAA), anticoagulant exposure, thrombocytopenia/coagulopathy, DIC, tumor or metastasis-related hemorrhage, cerebral venous thrombosis, vascular malformations, and cancer-treatment-related mechanisms. Cancer therefore introduces competing mechanisms of bleeding and thrombosis that cannot be adequately captured by a single stroke subtype or treatment variable.

Despite growing literature, there is no broadly harmonized multicenter registry that jointly characterizes AIS and spontaneous ICH in both active cancer and cancer survivors using the same stroke, oncology, imaging, laboratory, treatment, and longitudinal outcome framework. COSMIC is designed to fill this gap through standardized prospective data collection at discharge, 90 days, and 1 year.

1. STUDY DESIGN COSMIC is a multicenter, observational, prospective, real-world registry. Eligible patients will be identified consecutively at participating centers during a 2-year enrollment period beginning on the date of central ethics committee approval. Each participant will be followed prospectively for 1 year. The study is non-interventional: all investigations and treatments are determined by the treating clinicians according to local practice.
2. OBJECTIVES 3.1 Primary objective To prospectively estimate the 1-year cumulative incidence of any recurrent stroke after index AIS or ICH in patients with current or previous malignancy.

3.2 Secondary objectives

1. To estimate all-cause and vascular mortality at 90 days and 1 year.
2. To characterize functional outcome using the full ordinal mRS distribution at discharge, 90 days, and 1 year.
3. To quantify recurrent AIS, recurrent spontaneous ICH, myocardial infarction, VTE, ISTH major extracranial
4. To evaluate whether active cancer status, cancer type/stage, progression/recurrence, metastatic disease, recent cancer treatment, and treatment interruption/modification are associated with outcomes.
5. STUDY POPULATION, INCLUSION AND EXCLUSION CRITERIA Participating centers should prospectively screen consecutive adult patients with AIS or spontaneous non-traumatic ICH and a current or previous malignancy. To reduce ascertainment bias toward patients with already known active cancer and to allow capture of malignancies newly identified during the index admission, participation is restricted to centers that routinely apply an occult-malignancy screening protocol in patients with cryptogenic ischemic stroke.

4.1 Participating-center eligibility Participating centers must have a routine clinical protocol for occult-malignancy screening in patients with cryptogenic ischemic stroke. At minimum, this protocol must include blood testing and total-body CT with or without intravenous contrast, according to local clinical indications and contraindications. This requirement is intended to support systematic identification of occult cancer and avoid preferential inclusion of strokes occurring in patients with previously known active malignancy. The registry does not mandate additional diagnostic testing beyond each center's routine clinical protocol.

4.2 Patient inclusion criteria

1. Age ≥18 years at the time of the index stroke.
2. Confirmed acute ischemic stroke or spontaneous non-traumatic intracerebral hemorrhage. AIS is defined as an acute focal neurological deficit attributable to cerebral ischemia with CT and/or MRI showing one or more ischemic lesions consistent with the clinical presentation. ICH is defined as non-traumatic bleeding within the brain parenchyma confirmed by CT and/or MRI.
3. Mandatory neuroimaging confirmation of the qualifying event by at least one brain CT and/or MRI examination.
4. Current or previous documented qualifying malignancy. Eligible patients may have active cancer or previous cancer in complete remission/no evidence of disease (NED). The diagnosis must be supported by pathological, oncological, radiological, or hematological documentation in the medical record.
5. Sufficient baseline information to characterize, at minimum, the index stroke type, qualifying neuroimaging findings, and cancer status/history.

4.3 Patient exclusion criteria

1. Age <18 years.
2. No neuroimaging confirmation of the index cerebrovascular event.
3. Primary subarachnoid hemorrhage.
4. Traumatic intracranial hemorrhage.
5. Hemorrhagic transformation of an ischemic infarction without a primary spontaneous ICH. Hemorrhagic transformation remains classified as a complication of AIS.
6. Cerebrovascular events other than AIS or spontaneous non-traumatic parenchymal ICH, including isolated subdural or epidural hemorrhage.
7. No documented current or previous malignancy.
8. Basal cell carcinoma or cutaneous squamous cell carcinoma as the only malignancy history.
9. STUDY PERIOD AND TIMEPOINTS 5.1 Enrollment period Prospective enrollment will begin on the date of central ethics committee approval and will continue for 2 years. Follow-up assessments are targeted at 90±14 days and 365±30 days. Overall study completion will occur after the last enrolled participant completes the 365±30-day follow-up window, followed by database cleaning and lock.

5.2 Follow-up hierarchy and ascertainment Follow-up may be completed in person, by telephone/video, medical-record review, registry linkage, family/caregiver contact, or another documented source. The source must be recorded. If a patient dies before a scheduled follow-up, the date and primary cause of death will be recorded and mRS will be coded as 6 at that timepoint.

1. COLLECTED VARIABLES AT EACH TIMEPOINT 6.1 Baseline and index hospitalization

• Center and index-stroke dates; stroke type; known versus newly diagnosed cancer during the index admission.
• Demographics, body size, pre-stroke mRS, and pre-stroke ECOG performance status.
• Vascular and thromboembolic history, including hypertension, diabetes, dyslipidemia, smoking, prior stroke/TIA, prior ICH, coronary disease/MI, AF/flutter, heart failure, CKD, end-stage liver disease, COPD, prosthetic heart valve, prior/recent VTE, central venous catheter, and recent surgery/procedure.
• Malignancy characteristics: index primary site and histology, status at stroke, date of diagnosis, solid-tumor or hematologic stage/status, metastatic sites, brain metastasis, leptomeningeal disease, multiple primaries, progression/recurrence, disease spread, and active-cancer components.
• Cancer treatment: prior and recent systemic therapy, immunotherapy, targeted/anti-VEGF treatment, endocrine therapy, radiotherapy, surgery, cellular therapy, treatment timing, and treatment interruption/modification because of stroke.
• Pre-stroke antithrombotic, lipid-lowering, antihypertensive, and antidiabetic therapy, including recent antithrombotic interruptions and reasons.
• Stroke onset, admission date, admission BP, GCS, NIHSS, seizures, ICU/neurocritical care, and admission/follow-up brain imaging.
• Vascular imaging, echocardiography, valvular disease, cardiac vegetation/blood cultures, right-to-left shunt, DVT evaluation, rhythm monitoring, and implantable monitoring devices.
• Admission laboratory variables from the first available sample at presentation or within 24 hours, preferably before major treatment effects: hemoglobin, platelets, white cells, neutrophils, lymphocytes, CRP, D-dimer and unit/ULN status, fibrinogen, PT/INR/aPTT, ISTH overt DIC score, creatinine/eGFR, glucose, and LDL cholesterol.

6.2 AIS module • IV thrombolysis, thrombolytic agent, treatment timing and principal reason for non-treatment; large-vessel occlusion, mechanical thrombectomy, culprit vessel, and final mTICI.

• Acute infarct pattern: multiplicity, vascular territory, three-territory sign, and susceptibility vessel sign when assessable.
• Hemorrhagic transformation on follow-up imaging using Heidelberg categories and symptomatic intracranial hemorrhage.
• Conventional TOAST etiology; detailed other-determined mechanisms; ESUS phenotype after standard work-up.
• AHA 2026 cancer-related ischemic stroke framework, including NBTE, overt DIC, tumor embolism/invasion, cancer spread/progression, D-dimer, multiple-territory infarction, bilateral HITS, and susceptibility vessel sign.

6.3 ICH module

• ICH location and side, baseline volume and measurement method, intraventricular extension, acute hydrocephalus, and ICH Score.
• NCCT expansion/prognostic markers (including island, blend, black hole, satellite, swirl, fluid-level/sedimentation, heterogeneous density, irregular shape) and CTA spot sign/active extravasation.
• Follow-up imaging at 6-36 hours when available, follow-up volume, and hematoma expansion.
• Tumor/metastasis-related hemorrhage; principal presumed ICH mechanism; Boston Criteria v2.0 for CAA; vascular malformation type; and cerebral venous thrombosis site(s).
• Hemostatic/reversal therapies and neurosurgical interventions.
• Acute care-bundle components: intensive BP lowering and achievement of SBP target; glycemic management; temperature management; and relevant reversal of anticoagulation.
• MRI markers including remote DWI-positive ischemic lesions, Fazekas grade, cerebral microbleed burden, cortical superficial siderosis, white-matter hyperintensities, enlarged perivascular spaces, and lacunes.

6.4 Discharge

• Discharge date, length of stay, 24-hour NIHSS where available, discharge mRS, and discharge destination.
• Antithrombotic regimen and date of post-stroke therapeutic anticoagulation/antiplatelet initiation.
• Lipid-lowering, antihypertensive, and antidiabetic therapy; palliative-care/hospice plan.

6.5 90-day follow-up • Vital status, date/cause of death, vascular death, and mRS.

• Recurrent ischemic stroke (including first date, number, and TOAST etiology), recurrent spontaneous ICH (date and location), MI, VTE, ISTH major extracranial bleeding, seizure, and AFDAS.
• Current antithrombotic and lipid-lowering therapy; anticoagulant/antiplatelet discontinuation and reason.
• Cancer status, ongoing cancer-directed treatment, interval cancer surgery, and hospice/palliative-only care.
• Derived cumulative outcomes: any recurrent stroke, major arterial thrombotic event, and net adverse clinical event.

6.6 1-year follow-up The 1-year assessment mirrors the 90-day assessment but captures events occurring after the 90-day assessment through 1 year to avoid double counting. Cumulative 1-year endpoints will be derived from the 0-90-day and 90-day-to-1-year intervals. AFDAS at 1 year will be derived cumulatively.

1. STUDY DEFINITIONS Index stroke: The qualifying AIS or spontaneous non-traumatic ICH that establishes study entry.

Active cancer (AHA 2026): For analytic stratification, cancer diagnosed or treated within the previous 6 months (excluding basal/squamous cell skin cancer), known recurrent, regionally advanced, or metastatic cancer, or cancer not in complete remission/NED. Active cancer is not required for COSMIC eligibility because patients with previous cancer in complete remission/NED are also included.

Index malignancy in patients with multiple cancers: The malignancy that best represents oncological status at the index stroke. Active malignancy takes precedence over historical disease in remission; if multiple cancers are active, select the cancer driving current treatment, disease burden, or prognosis; if all are historical/NED, select the most recently diagnosed or treated malignancy.

AIS: Acute focal neurological deficit attributable to cerebral ischemia with CT and/or MRI evidence of an ischemic lesion consistent with the presentation.

Spontaneous ICH: Non-traumatic bleeding within the brain parenchyma confirmed by CT and/or MRI. Primary SAH, isolated subdural/epidural hemorrhage, and hemorrhagic transformation without primary ICH are excluded.

TOAST etiology: Large-artery atherosclerosis, cardioembolism, small-vessel occlusion, other determined etiology, or undetermined etiology, assigned using conventional TOAST principles.

ESUS phenotype: Non-lacunar ischemic stroke with no ≥50% atherosclerotic stenosis in an artery supplying the infarct, no major-risk cardioembolic source, no other specific cause, and a standard diagnostic evaluation completed to a reasonable extent.

AHA 2026 probable cancer-related AIS: Active cancer plus ≥1 of: cardiac vegetation(s) with negative blood cultures consistent with NBTE; ISTH overt DIC score ≥5; or radiographic evidence of tumor invading the left atrium/ventricle, pulmonary veins, or cervicocephalic arteries with infarction in the corresponding territory.

AHA 2026 possible cancer-related AIS: Cryptogenic ischemic stroke after standard diagnostic evaluation in active cancer that has spread beyond the primary site or recently progressed, plus ≥1 of: D-dimer >2500 ng/mL FEU without acute VTE; acute infarction in multiple cerebrovascular territories; bilateral HITS on TCD; or absence of susceptibility vessel sign despite confirmed acute intracranial occlusion.

AHA 2026 uncertain causal link: Active cancer is present but criteria for probable or possible cancer-related AIS are not met.

Symptomatic intracranial hemorrhage after AIS: Intracranial hemorrhage associated with neurological deterioration of ≥4 NIHSS points or death, with hemorrhage considered the likely cause of worsening. Radiological subtype is recorded separately using the Heidelberg Bleeding Classification.

Hematoma expansion: Increase in ICH volume of >6 mL absolute or >33% relative from baseline to follow-up imaging, according to the prespecified CRF definition.

ICH mechanism: A single principal mechanism assigned after work-up: hypertensive arteriopathy; CAA; anticoagulant-associated; tumor/metastasis-related; DIC/coagulopathy/thrombocytopenia; vascular malformation; cerebral venous thrombosis; treatment-related; other determined; or undetermined.

CAA classification: Probable or possible sporadic CAA according to Boston Criteria v2.0 when appropriate MRI and clinical information are available [8].

Remote DWI-positive ischemic lesion after ICH: Focal DWI hyperintensity with corresponding ADC reduction located remote from the hematoma/perihematomal region and consistent with acute ischemic injury. COSMIC uses a pragmatic ≥10-mm separation from the hematoma margin when assessable.

cSVD MRI markers: White-matter hyperintensities, lacunes, cerebral microbleeds, enlarged perivascular spaces, cortical superficial siderosis, and related markers recorded using standardized STRIVE concepts.

INTERACT3-style BP target: For the CRF care-bundle variable: intensive BP lowering aimed at SBP <140 mmHg, preferably within 1 hour of treatment initiation, while avoiding SBP <130 mmHg.

INTERACT3-style glycemic target: For the CRF bundle variable: glucose management to the prespecified study range of 110-180 mg/dL, with avoidance of clinically significant hypoglycemia. This pragmatic range is used for harmonized data capture; analyses may additionally examine INTERACT3 diabetes-specific targets.

INTERACT3-style temperature target: Treat temperature ≥37.5°C and aim for ≤37.5°C as soon as possible, preferably within 1 hour of treatment initiation.

AFDAS: AF or atrial flutter first documented after the index stroke in a patient without known pre-stroke AF/flutter, established by 12-lead ECG, telemetry, Holter/external monitoring, or an implantable cardiac monitor. Device-detected atrial high-rate episodes alone are not automatically coded as AFDAS without ECG confirmation or local adjudication.

ISTH major extracranial bleeding: Clinically overt extracranial bleeding that is fatal, occurs in a critical organ/site, causes a hemoglobin decrease ≥2 g/dL, or requires transfusion of ≥2 units of whole blood/red blood cells. Intracranial bleeding is captured separately in COSMIC.

Vascular death: Death primarily due to index/recurrent ischemic stroke, index/recurrent ICH, myocardial infarction, pulmonary embolism, sudden unexpected vascular death, or another adjudicated vascular cause.

Major arterial thrombotic event: Composite of recurrent ischemic stroke, myocardial infarction, or vascular death. Component events will also be reported separately.

Net adverse clinical event (NACE): Composite of recurrent ischemic stroke, recurrent spontaneous ICH, myocardial infarction, VTE, ISTH major extracranial bleeding, or death from any cause. Components will be reported individually to avoid masking clinically divergent effects.

1. OUTCOME MEASUREMENT AND ENDPOINTS Outcome events will be adjudicated locally from clinical records and follow-up information using the prespecified CRF definitions. When uncertainty materially affects a key endpoint, cases may undergo central adjudication blinded to the comparison of interest where feasible.

8.1 Primary endpoint The primary endpoint is any recurrent stroke within 90-day and 1 year after the index event, defined as the first occurrence of recurrent ischemic stroke or new spontaneous non-traumatic ICH. The endpoint will be analyzed as time-to-first event and as a cumulative risk. Death without recurrent stroke will be treated as a competing event for cumulative-incidence analyses.

8.2 Key secondary endpoints

• Recurrent ischemic stroke and recurrent spontaneous ICH separately at 90 days and 1 year.
• All-cause mortality and vascular death at 90 days and 1 year.
• Ordinal mRS distribution at discharge, 90 days, and 1 year; functional dependence (mRS 3-6) as a secondary dichotomous measure, particularly among patients with pre-stroke mRS ≤2.
• Myocardial infarction at 90 days and 1 year.
• Venous thromboembolism (DVT, pulmonary embolism, catheter-related thrombosis, or other adjudicated VTE) at 90 days and 1 year.
• ISTH major extracranial bleeding at 90 days and 1 year.
• Major arterial thrombotic event and NACE at 90 days and, where components permit, cumulatively at 1 year.
• Post-stroke seizure/status epilepticus at 90 days and 1 year.
• AFDAS at 90 days and 1 year.
• Cancer progression/recurrence, ongoing cancer-directed therapy, and transition to hospice/palliative-only care at follow-up.

8.3 AIS-specific secondary endpoints

• In-hospital hemorrhagic transformation and symptomatic intracranial hemorrhage.
• Distribution of TOAST etiologies, ESUS phenotype, and AHA 2026 cancer-related AIS categories.
• Reperfusion treatment use, final mTICI after thrombectomy, and reasons for non-treatment.
• Recurrent AIS etiology and concordance with the index mechanism. 8.4 ICH-specific secondary endpoints
• Hematoma expansion on follow-up imaging.
• Achievement of BP, glycemic, temperature, and anticoagulation-reversal care-bundle components.
• Distribution of presumed ICH mechanisms, including tumor/metastasis-related hemorrhage, anticoagulant-associated hemorrhage, coagulopathy/DIC, CAA, and cerebral venous thrombosis.
• Frequency and distribution of remote MRI DWI-positive ischemic lesions. 9. SAMPLE SIZE COSMIC is designed as a registry rather than a single hypothesis-driven comparative trial; therefore, no fixed a priori sample size is mandated. Each center will be asked to include all consecutive eligible patients during the 2-year enrollment period. The final sample size will determine the precision of incidence estimates and the complexity of multivariable models. For multivariable time-to-event analyses, the number of model parameters will be constrained by the number of observed outcome events, and penalized or parsimonious models will be used when event counts are limited. No center-level minimum will be required for inclusion in the main descriptive analyses, although center-volume sensitivity analyses may be performed.

1. STATISTICAL ANALYSIS PLAN 10.1 General statistical principles Statistical analyses will be prespecified in a detailed statistical analysis plan before database lock for the main prospective analyses. Categorical variables will be summarized as counts and percentages, while continuous variables will be reported as mean±SD or median (IQR), according to their distribution. Baseline characteristics will be described overall and across clinically relevant subgroups, including index stroke type and cancer status. Time-to-event outcomes, including recurrent stroke and mortality, will be analyzed using survival methods and, when appropriate, competing-risk approaches accounting for death as a competing event. Functional outcome measured with the modified Rankin Scale will primarily be assessed across its full ordinal distribution using proportional-odds regression, with alternative models considered if model assumptions are not met. Other vascular, thrombotic, hemorrhagic, neurological and cardiac outcomes will be summarized using cumulative risks and incidence rates, with recurrent-event analyses considered for selected secondary outcomes. Multivariable analyses will rely on clinically prespecified covariates, including relevant demographic, clinical, cancer-related, treatment, imaging and laboratory characteristics, and will account for clustering by participating center where appropriate. Prespecified subgroup and exploratory analyses will investigate potential differences according to stroke subtype and mechanism, cancer activity, stage and type, metastatic disease, recent cancer-directed treatment, antithrombotic exposure, and other clinically relevant features. Nonlinear associations may be modeled using flexible approaches when supported by sample size. Effect estimates will be presented with 95% confidence intervals; two-sided P values <0.05 will be considered statistically significant for primary analyses, with adjustment for multiple comparisons applied to families of secondary or exploratory analyses when appropriate.

10.2 Missing data Missingness will be described for every key variable. "Not assessed" will be distinguished from true negative findings. Complete-case analyses will be accompanied by multiple imputation under a missing-at-random assumption when missingness is material and the variable structure supports imputation. Variables that are structurally unavailable because an investigation was not performed will not be indiscriminately imputed. Sensitivity analyses will assess the influence of follow-up loss and incomplete diagnostic work-up.

10.3 Software Analyses will be performed using validated R (R Foundation for Statistical Computing; contemporary version at the time of analysis). Reproducible scripts will be archived by the coordinating analytic team.

1. STUDY PERIOD Phase Dates Operational meaning Protocol version 1.0 01 Oct 2026 Protocol/CRF harmonization and ethics/regulatory preparation.

Prospective enrollment From central ethics committee approval through 2 years Prospective consecutive enrollment of eligible index strokes.

Final prospective follow-up Through 365±30 days after the last enrolled participant Completion of 1-year follow-up, followed by database cleaning and lock.

1. ETHICS, CONSENT, DATA COLLECTION AND MANAGEMENT Central ethics review will proceed after expressions of interest from participating centers have been collected. The 2-year study enrollment period will begin on the date of approval by the central ethics committee. Enrollment at each participating site may start only after any additional approval, acknowledgment, or institutional authorization required locally has been obtained. Written informed consent from the patient or legally authorized representative will be obtained when required by the approving authority and applicable local regulations. These central processes do not replace any local ethics or institutional requirements that may apply at participating centers.

Data will be entered into an electronic database using the COSMIC case-report form. No direct identifiers will be entered in the central study database. Each patient will be assigned a coded study record ID; the linkage between study code and identifiable information will remain at the enrolling center and will not be transferred to the coordinating database.

Access to the electronic database will be password protected. Data transmission will use secure encrypted connections. Investigators will enter only data necessary for the scientific objectives. Data retention, controller/processor responsibilities, international data transfer, and deletion schedules will follow the ethics-approved data-management plan and applicable data-protection law, including GDPR where relevant.

1. DATA QUALITY AND MONITORING COSMIC is an observational registry and does not require an independent data and safety monitoring board because no treatment is assigned by the protocol. Data quality will be monitored centrally. Planned procedures include investigator training, standardized field definitions, periodic missingness reports, cross-field consistency checks, and data queries to participating centers. Key outcome dates and internally inconsistent events will be queried before database lock.
2. SAFETY CONSIDERATIONS The registry itself introduces no study-assigned treatment and therefore does not create treatment-related adverse events. Clinical adverse events occurring during routine care are captured as study outcomes where prespecified. Participating clinicians remain responsible for all diagnostic and therapeutic decisions. The study should not delay acute reperfusion therapy, reversal of anticoagulation, neurosurgical care, cancer treatment, or other clinically indicated management.
3. RESULTS DISSEMINATION Study findings will be submitted to national and international scientific meetings and peer-reviewed journals.
4. AUTHORSHIP POLICY Authorship will follow International Committee of Medical Journal Editors (ICMJE) criteria. Participating centers and collaborators who contribute eligible patients, high-quality follow-up, study design, analysis, interpretation, or manuscript preparation may be recognized as named authors or members of the COSMIC Study Group according to a publication-specific authorship plan agreed before manuscript submission. Enrollment numbers alone do not supersede ICMJE requirements. Center-level recruitment and data For publications primarily based on registry recruitment, participating centers will be ranked according to the number of eligible patients contributed to the relevant analysis, provided that submitted data meet prespecified standards for completeness and quality. Centers will then be divided into quartiles according to recruitment performance. As a general rule, centers in the first quartile (highest recruitment) may nominate up to 4 authors, centers in the second quartile up to 3 authors, centers in the third quartile up to 2 authors, and centers in the fourth quartile up to 1 author.
5. PROTOCOL AMENDMENTS AND DEVIATIONS Substantive protocol amendments affecting eligibility, primary endpoints, study duration, or data handling will be version controlled and submitted to ethics/regulatory bodies when required. Minor clarifications that do not alter participant risk or primary scientific objectives may be documented in a protocol history. Deviations affecting eligibility or primary outcome ascertainment will be logged and considered in sensitivity analyses.