Incidence of SDH recurrence/ progression at 1 year post-procedure

Rate of successful embolization of the target vessels (MMA trunk and branches) with ONYX base on DSA imaging

Change in hematoma thickness based on CT/MRI imaging at 90 days post-procedure

Changes in hematoma volume at 90 days post-procedure

Change in Midline shift based on CT/MRI imaging at 90 days post-procedure

Change in the Modified Rankin Scale score (mRS) Grade 0 (no symptoms) to 6 (death) at 90 days and 1 year post-procedure

Percentage of patients with favorable functional outcome defined as Modified Rankin Scale of 0 to 3 at 90 days and 1year post-procedure

Percentage of patients with good functional outcome defined as Modified Rankin Scale of 0 to 2 at 90 days and 1 year post-procedure

Quality of life assessed by (EuroQol) EQ-5D scale Grade 0 (worst health) to 100 (best health) at 90 days and 1 year post-procedure

Total patients with SAEs within 90 days post-procedure

Incidence of neurological death within 90 days post-procedure

Incidence of procedural serious complications within 30 days post-procedure:
o    symptomatic procedure-related intracranial hemorrhage
o    any procedure-related intracranial hemorrhage
o    any procedure-related neurological deficit
o    CNS infection caused by procedure
o    procedure-related artery dissection, vessel wall damage and vessel perforation
o    procedure-related ischemic event
o    retroperitoneal hematoma (femoral access)/wrist hematoma (radial access)
o    neuropathy at the puncture site
o    contrast agent allergy or encephalopathy

Symptomatic SDH is defined as an SDH that fulfills one or more of the following criteria AND shows neurological symptoms (headache, short-term cognitive decline, speech difficulty or aphasia, gait impairment, focal weakness, sensory deficits, seizures) that are attributable to the hematoma

CONTENTS

Version History
1. Summary of Protocol
1.1. Statistical Analysis Plan and Protocol Discrepancies
1.2. Study Objectives and Endpoints
1.3. Study Design
2. Statistical Hypothesis
3. Sample Size Considerations
4. Planned Analyses
4.1. Interim Analyses
4.2. Final Analysis
5. Analysis Population
5.1. Protocol Deviations
6. Considerations for Data Analyses and Data Handling Conventions
7. Study Population Analyses
7.1. Overview of Planned Study Population Analyses
7.2. Description Of Study Population Analysis
7.2.1. Demographic and Baseline Characteristics
8. Primary Efficacy Endpoints Analyses
8.1. Overview of Planned Primary Efficacy Endpoints Analyses
8.2. Specifications of Primary Efficacy Endpoints Analyses
8.2.1. Definitions and Derivations
8.2.2. Analysis Methods
8.2.3. Model Specifications
8.3. Subgroup Analyses
9. Secondary Statistical Analyses
9.1. Secondary Efficacy Endpoint Analyses
9.1.1. Overview of Planned Analyses for Secondary Efficacy Endpoints
9.1.2. Specifications of Secondary Efficacy Endpoint Analyses
9.2 Safety analyses
9.2.1 Overview of Planned Safety Analyses
9.2.2 Specifications Of Safety Analyses
9.2.2.1 Adverse Events
10. References
12. APPENDICES
1 2.1 Appendix 1: Treatment Period and Study Period
12.2 Appendix 2: Data Display Standards and Handling Conventions
12.3 Appendix 3: Handling of Early Withdrawals and Missing Data
12.4 Appendix 4: Abbreviations Table

1.1. Statistical Analysis Plan and Protocol Discrepancies
All changes to the content of the analyses originally specified in Protocol 3.0 (2022-09-21) are summarized in Table 1.
Table 1 Changes to the analysis plan defined in the protoco

1.2. Study Objectives and Endpoints

1.3. Study Design

Hypothesis testing will be performed on hematoma recurrence or progression rates.
• Null hypothesis: there is no difference in the incidence of recurrence or progression of hematoma between the MMA embolization group and the control group at 90 days after randomization;
• Alternative hypothesis: there is difference in the incidence of recurrence or progression of hematoma between the MMA embolization group and the control group at 90 days after randomization.

The above hypothesis tests will be two-sided and statistical significance will be determined as a p-value of＜0.05, and two-sided 95% confidence intervals will be reported.

The study hypothesizes that patients undergoing middle meningeal artery embolization may have lower symptomatic recurrence (in the trepanation/drainage group) and lower rates of symptomatic hematoma progression (in the non-surgical group) of subdural hematoma compared to patients who do not go MMA embolization. Two interim analyses will be conducted. With an expected symptomatic recurrence rate (trepanation/drainage group) / symptomatic progression rate (non-surgical group) after MMA embolization of 5% compared with 12% in the control group, and considering a loss to follow-up rate of 8%, a sample size of 722 cases with a twosided alpha of 0.05 would yield 90% power.

4.1. Interim Analyses
Two formal interim analyses was planned to review data relating to treatment efficacy, participant safety and quality of trial conduct using an Haybittle-Peto stopping rule for efficacy. Given the conservative rule used and the negligible amount of type-I error rate spent at the interim analysis, the significance threshold will remain at 5% for the final analysis.
4.2. Final Analysis
The planned final analysis will be performed after the following steps are completed sequentially:
1. All subjects have completed the study as defined in the protocol.
2. The Statistical Analysis Plan (SAP) has been finalized and approved.
3. All required database cleaning is completed, and the head of Data Management department announces the final database release and database freeze.
4. Randomization codes are issued according to the unblinding process at the end of the trial. Only blinded data will be reviewed prior to database lock. In addition, no database lock will occur prior to finalization of this SAP.

Population	Definition/Criteria	Analyses Evaluated
All included Subjects Set	• All subjects who signed informed consent.	• Study Population
Intent-to-treat Analysis Set (ITT)	• Including all randomized subjects based on the intention-to-treat (ITT) principle, i.e. analysis of subjects based on initial treatment assigned by randomization	• Study Population • Efficacy
Per-Protocol Analysis Set (PPS)	• The PPS will include all subjects in the FAS who did not have any major protocol deviations. • Protocol deviations excluding subjects from the PPS are defined in Section 5.1 (Protocol Violations). • Analysis in PPS will be based on the treatment the participant assigned by randomization. • Additional efficacy analyses based on the PPS set will not be performed if the PPS set includes 90% or more of subjects in the FAS and the difference in analysis results between the two analysis sets is expected to be negligible.	• Study Population • Efficacy
Safety Set (SS)	• Includes all randomized subjects. • As treated • ITT analysis set	• Safety

5.1. Protocol Deviations
• Important protocol deviations, including those related to study inclusion/exclusion criteria, trial conduct, subject management, or subject assessment, will be summarized.
• Protocol deviations will be tracked by the study team during the course of the study according to relevant regulations in routine monitoring.
− All data will be reviewed prior to database lock to ensure that all important protocol deviations and deviations that could lead to exclusion of subjects from the analysis can be captured and classified in the protocol deviation dataset.
− This dataset will serve as the basis for summaries of protocol deviations.

Analyses will be performed using SAS ® System Version 9.4 or higher, and all tables, figures, and listings will be generated as RTF files.
For descriptive statistics, the number of observations (n), standard deviation, median, minimum, maximum，Q1 and Q3 will be presented for continuous variables and number and percentage of subjects will be presented for categorical variables, unless otherwise specified.
Table 2 provides an overview of the appendices in the SAP for outlining general rules for data analyses and data handling conventions.

Table 2 Overview of Appendix

Component
Appendix 1: Treatment Period and Study Period
Appendix 2: Data Display Standards and Handling Conventions
Appendix 3: Handling of Premature Withdrawals and Missing Data
Appendix 4: Abbreviations Table

7.1. Overview of Planned Study Population Analyses
Unless otherwise stated, the analyses will be based on the ITT set
Table 3 provides an overview of the planned study population analyses.
Table 3 Overview of Planned Study Population Analyses

	Data Presentation Generated
	Table	Figure	Listing
Subject Disposition 1, 3
Subjects who entered screening and failed screening	Y		Y
Randomized Subjects
Subjects receiving any study treatment - Received burr hole drainage - Received MMA embolization - Received optimal medical therapy
Completed Subjects
Subjects who premature withdrew and reasons
Entered Intent-to-Treat Set (ITT)
Entered Per-Protocol Analysis Set (PPS)
Entered Safety Set (SS)
Protocol Deviations
Important protocol deviations	Y		Y
Any protocol deviations	Y		Y
Demographic and Baseline Characteristics
Sex 1	Y		Y
Age (years) 2
Ethnic 1
Weight (kg) 2
Height (m) 2
BMI (kg/m 2) 2
Smoking status 1
Alcohol status 1
Baseline mRS score2
Prior medical history 1
Prior medical history	Y		Y
Prior medication history 1
Prior antiplatelet, anticoagulant use	Y		Y
Prior statin use
Prior hormonal use
Baseline cranial imaging diagnosis (CT/MRI)
Massive cerebral infarction 1	Y		Y
Intracranial space-occupying lesion 1
Midline shift 1
If yes, shift distance 2
Subdural hematoma 1
If yes, hematoma Side Positions 1
Maximum thickness of hematoma on the enrolled side 2
Volume of hematoma on the enrolled side 2

7.2. Description Of Study Population Analysis
7.2.1. Demographic and Baseline Characteristics
• Age (years) will be calculated as (date of informed consent – date of birth)/365.25 rounded down.
• BMI (kg/m²) will be calculated as [weight(kg) / (height(cm)/100) ^{^2}].

8.1. Overview of Planned Primary Efficacy Endpoints Analyses
Unless otherwise stated, the primary efficacy endpoints analysis will be based on the Intentto-Treat (ITT) and Per-Protocol (PPS) sets. Intent-to-Treat (ITT) set will be the primary analysis set.
Table 4 provides an overview of the planned primary statistical analyses.
Table 4 Overview of planned primary efficacy endpoints analyses

8.2. Specifications of Primary Efficacy Endpoints Analyses
8.2.1. Definitions and Derivations
• Symptomatic SDH recurrence (in the trepanation/drainage group) refers to a situation in which the maximum thickness of hematoma of patients exceeds 10 mm, combined with neurological symptoms, or the need for repeat surgery.
• Symptomatic SDH progression (in the non-trepanation/drainage group) refers to a situation in which the maximum thickness of increases by more than 3 mm compared with baseline, combined with neurological symptoms, or the need for surgery.
• The results of central imaging evaluation and analysis of CT/MRI examinations required for the determination of the subject 's primary endpoint in this study will be presented in tables and listings, respectively. Central image evaluation will be performed by the independent Core Lab. Evaluation of neurological symptoms will be derived from the Outcome Assessment Committee (OAC) review data.
• It should be noted that if there is a record of reoperation in participants who have undergone trepanation/drainage, and there is a record of surgical treatment in subjects who have received medical therapy without trepanation/drainage, the occurrence of the primary endpoint can be directly determined, and hematoma thickness or neurological symptoms are no longer required.
• In case of non-negligible amounts of missing data (>5%), we will use the MCMC method based on the Missing at random (MAR) assumption to impute missing data. The regression model will include group, sex, age, BMI, baseline mRS score and baseline maximum hematoma thickness. The number of imputations is 10, seed=1234, nbiter=200. SAS code examples for multiple imputation will be provided in Appendix 3.
• Hematoma recurrence (trepanation/drainage group) or symptomatic hematoma progression (non-surgical group) at day 90 will be analyzed as follows. If there is any discrepancy between EDC and central imaging assessment scan date in the determination of visit window, the date determined by central imaging assessment shall prevail.

Table 5. Data Handling Rules for Primary Endpoint Analysis

*If percentage of missing primary endpoint is less than 5%, multiple imputation will not be performed and missing data will be regarded as no event.

8.2.2. Analysis Methods
• The primary analysis will compare groups using the CMH chi-square test (the stratification factor is whether or not to perform trepanation/drainage was performed), summarize the 90- day recurrence (trepanation/drainage group) / progression rate (non-surgical group) for the MMA embolization group and the control group using the two-sided 95% CI (ClopperPearson method) and summarize the two-sided 95% CI for the difference in recurrence or progression rates between the two groups using the Miettinen-Nurminen method.
• Sensitivity analyses will use generalized linear mixed-effects model to calculate ORs and their 95% CIs for the comparison of recurrence or progression rates between the two arms, with group as fixed effect, site as random effect. Results of multivariable analysis will also be presented, including treatment group, sex, age, surgical treatment strata (trepanation/drainage vs. no trepanation/drainage), any usage of antiplatelet or anticoagulant therapy, hematoma volume of baseline CT as covariates.

8.2.3. Model Specifications
8.2.3.1. Model Specifications of Primary Analysis

8.2.3.2. Model Specifications of Sensitivity Analysis

8.3. Subgroup Analyses
The primary efficacy endpoint point will be analyzed in the following subgroups:
• Age group(＜70 vs ≥70)
• Sex (male vs female)
• Surgical treatment strata (trepanation/drainage vs. no trepanation/drainage)
• Smoking status(yes vs no)
• Medical history of brain trauma (yes vs no)
• Any usage of antiplatelet or anticoagulant therapy before randomization(yes vs no)
• Any usage of statins within one month before randomization(yes vs no)
• Midline shift of baseline CT (＜10 mm vs ≥10 mm)
• Hematoma thickness of baseline CT (＜10 mm vs ≥10 mm)
• Hematoma volume of baseline CT (＜Q3 of combined hematoma volume data, ≥Q3 of combined hematoma volume data)
• Bilateral SDH (yes vs no)
• Embolized branches of MMA (main trunk only, distal only, main trunk & distal) Subgroup analysis will use the same statistical analysis methods as described in Section 8.2.2.

9.1. Secondary Efficacy Endpoint Analyses
9.1.1. Overview of Planned Analyses for Secondary Efficacy Endpoints
Unless otherwise stated, analyses of secondary efficacy endpoints will be based on the intent-to-treat analysis set (ITT) and per-protocol (PPS) set.
Table 6 Planned Overview of Secondary Efficacy Endpoint Analyses

9.1.2. Specifications of Secondary Efficacy Endpoint Analyses
9.1.2.1. Definitions and Derivations
• Change from 90 days/360 days post study treatment will be calculated as 90 days/360 days measurement – baseline measurement
• Total days in hospital = Date of discharge - Date of hospitalization + 1 (Hospital information page)
• Other definitions and derivations are the same as in Section 8 .2.1

9.1.2.2. Statistical Analysis Methods
• 360 days recurrence or progression of SDH after randomization will be analyzed in the same way as the primary endpoint (Section8.2.2)
• Ordinal logistic regression will be used for intergroup comparison of mRS scale scores at 90 and 360 days after randomization, with treatment group, sex, age, surgical treatment strata (trepanation/drainage vs. no trepanation/drainage), any usage of antiplatelet or anticoagulant therapy within one month before randomization(yes vs no), hematoma volume of baseline CT as covariates.
• Other categorical secondary endpoints will be analyzed in the same way as the primary endpoint (Error! Reference source not found.Error! Reference source not found.Section)
• The linear mixed effects model approach will be used to estimate the mean estimate (95% CI) and the mean estimate (95% CI) of the differences between groups as well as the P values of the differences between groups for the change in maximum hematoma thickness at 90 days after randomization compared to baseline and the change in hematoma volume at 90 days after randomization compared to baseline; Treatment group variables, sex, age, surgical treatment strata (trepanation/drainage vs. no trepanation/drainage), any usage of antiplatelet or anticoagulant therapy within one month before randomization(yes vs no), hematoma volume of baseline CT , visit variables, and visit by treatment group variable interactions will be included in the model and site as the random effect.
• For other continuous secondary endpoints, T tests will be used to compare groups, and 95% confidence intervals for the difference in means between groups will be calculated using normal approximation.

9.1.2.3. Model Specifications

9.2 Safety analyses
9.2.1 Overview of Planned Safety Analyses
Unless otherwise stated, safety analyses will be based on the Safety Analysis Set.
Table 7 provides an overview of the planned safety analyses.

Notes:
1. Categorical data.
2. Postoperative serious complications include: symptomatic intracranial hemorrhage, surgery-related intracranial hemorrhage, surgery-related neurological dysfunction , surgery-related central nervous system infection , surgery-related arterial dissection, vessel wall injury, and vessel rupture and perforation , surgeryrelated ischemic events , retroperitoneal/wrist hematoma , puncture site neuropathy , contrast agent allergy, or contrast agent encephalopathy
Y = Plan Generation.

9.2.2 Specifications Of Safety Analyses
9.2.2.1 Adverse Events
• An untoward medical occurrence during a clinical investigation, whether or not related to the investigational medical device, is an adverse event (AE).
• AEs will be coded using MedDRA version 24.0 or higher. System Organ Class (SOC) and Preferred Term (PT) will be reported.
• The investigator will grade the severity of the AE (mild, moderate, severe). AEs with missing severity will not be imputed.
• Adverse events will be summarized using frequency tables in descending order of incidence by SOC and PT unless otherwise specified. If a subject has the same AE reported more than once, the highest severity of this AE for that subject will be reported.
• For adverse events of special interest, the CEC assessment and investigator assessment will be performed, and the results of the two assessments will be presented separately.

9.2.2.3 Concomitant medication
• Concomitant medications are defined as medications that were taken by the patient until or until after randomization.
• Concomitant medications will be summarized by treatment group and the number and percentage of subjects will be listed by medication name.

[1]. Zuo, Q., Ni, W., Yang, P., Gu, Y., Yu, Y., Yang, H., ... & MAGIC-MT investigators. (2023). Managing non-acute subdural hematoma using liquid materials: a Chinese randomized trial of middle meningeal artery treatment (MAGIC-MT)—protocol. Trials, 24(1), 586.
[2]. Newcombe, R. G. (1998). Two‐sided confidence intervals for the single proportion: comparison of seven methods. Statistics in medicine, 17(8), 857-872.
[3]. Newcombe, R. G. (1998). Interval estimation for the difference between independent proportions: comparison of eleven methods. Statistics in medicine, 17(8), 873-890.
[4]. Rayner, J. C. W., & Rippon, P. (2018). Recent extensions to the cochran–mantel–haenszel tests. Stats, 1(1), 98-111.
[5]. Jiang, J., & Nguyen, T. (2007). Linear and generalized linear mixed models and their applications (Vol. 1). New York: Springer.
[6]. Bender, R., & Grouven, U. (1997). Ordinal logistic regression in medical research. Journal of the Royal College of physicians of London, 31(5), 546.
[7]. Lewis, J. A. (1999). Statistical principles for clinical trials (ICH E9): an introductory note on an international guideline. Statistics in medicine, 18(15), 1903-1942.
[8]. Jennison, C., & Turnbull, B. W. (1999). Group sequential methods with applications to clinical trials. CRC Press.

12.1 Appendix 1: Treatment Period and Study Period

12.2 Appendix 2: Data Display Standards and Handling Conventions

12.3 Appendix 3: Handling of Early Withdrawals and Missing Data

*If percentage of missing primary endpoint is less than 5%, multiple imputation will not be performed and missing data will be regarded as no event.

SAS Code Sample:
proc mi data=im1 seed=1234 nimpute=20 out=im2
round=. . . . . 1
min=. . . . . 0
max=. . . . . 1;
;
by trtan;
var sex age bmi basemrs basemth aval;
mcmc chain=multiple nbiter=200 niter=100;
run;
proc sort data=im2;
by _imputation_ trtan usubjid siteid;
run;
proc transpose data=im2 out=im2_t(rename=(_name_=avisit col1=aval));
by _imputation_ trtan usubjid siteid;
var v1 v2;
run;
proc sql noprint;
create table im2_count as
select distinct
_imputation_,
trtan,
avisit,
aval,
count(*) as count
from im2_t
group by _imputation_,trtan,avisit,aval
order by _imputation_,trtan,avisit,aval
;
quit;
/*clopper-pearson 95%ci*/
ods select none;
proc freq data= im2_count;
by _imputation_ trtan avisit;
table aval / binomial (cl=exact) alpha=0.05;
weight count;
ods output binomial=binomial;
run;
ods select all;
proc transpose data=binomial out=binomial_t;
by _imputation_ trtan avisit;
var nvalue1;
id label1;
quit;
proc sort data=binomial_t;
by trtan avisit _imputation_;
run;
ods select none;
proc mianalyze data=binomial_t;
by trtan avisit;
modeleffects proportion;
stderr ase;
ods output parameterestimates=binomial_t_pool;
run;
ods select all;
/*mn*/
proc freq data=im2_t;
by _imputation_ ;
tables trtan *aval / commonriskdiff (cl=score test=score) riskdiff (cl=mn);
ods output commonpdifftests=commonpdifftests1 commonpdiff=commonpdiff1
pdiffcls=pdiffcls1 riskdiffcol1=riskdiffcol11;
run;
data riskdiffcol11;
set riskdiffcol11(where=(row="difference"));
run;
proc sort data=riskdiffcol11;
by _imputation_;
run;
proc mianalyze data=riskdiffcol11;
modeleffects risk;
stderr ase;
ods output parameterestimates=pest1_pool;
run;
/*cmh*/
*** perform cmh test;
proc freq data=im2_t;
tables strata*trtan*aval/cmh;
ods output cmh=cmh;
by _imputation_;
run;
*** apply wilson-hilferty transformation to the cmh statistic and
standardize the resulting normal variable;
data cmh_wh;
set cmh(where=(althypothesis="general association"));
cmh_value_wh=((value/df)**(1/3) - (1-2/(9*df)))/sqrt(2/(9*df));
cmh_sterr_wh = 1.0;
run;
*** combine results;
proc mianalyze data=cmh_wh;
ods output parameterestimates=mian_cmh_wh;
modeleffects cmh_value_wh;
stderr cmh_sterr_wh;
run;
proc sql noprint;
create table im2_count2 as
select distinct
_imputation_,avisit,siteid,trtan,aval,
count(*) as count
from im2_t
group by _imputation_,avisit,siteid,trtan,aval
order by _imputation_,avisit,siteid,trtan,aval
;
quit;
ods select none;
proc glimmix data=im2_count2;
by _imputation_ avisit;
class siteid trtan aval;
model aval/count=trtan / solution;
random intercept / subject=siteid;
ods output parameterestimates=parameterestimates;
run;
ods select all;
proc sort data=parameterestimates(where=(trtan=1)) out=parameterestimates1;
by trtan avisit _imputation_ ;
run;
ods select none;
proc mianalyze data=parameterestimates1;
by trtan avisit;
modeleffects estimate;
stderr stderr;
ods output parameterestimates=parameterestimates1_pool;
run;
ods select all;

Prior/concomitant medications/treatments
Prior medications/therapies are those that ended prior to randomization. Concomitant medications/therapies are medications/therapies other than study treatment or premedication with study treatment taken at any time during the study (on or after randomization), including medications/therapies started and continued prior to randomization.
Incomplete dates for prior/concomitant/post-study medications/treatments recorded in the CRF will not be imputed and will be classified according to the following rules:

Note: "<" refers to "earlier than" and "≥" refers to "same day or later".

12.4 Appendix 4: Abbreviations Table