Trial Review

Please note the ANZCTR will be unattended from Friday 20 December 2024 for the holidays. The Registry will re-open on Tuesday 7 January 2025. Submissions and updates will not be processed during that time.

Registering a new trial?

To achieve prospective registration, we recommend submitting your trial for registration at the same time as ethics submission.

The safety and scientific validity of this study is the responsibility of the study sponsor and investigators. Listing a study does not mean it has been endorsed by the ANZCTR. Before participating in a study, talk to your health care provider and refer to this information for consumers
Trial registered on ANZCTR


Registration number
ACTRN12624001432505
Ethics application status
Approved
Date submitted
16/09/2024
Date registered
6/12/2024
Date last updated
6/12/2024
Date data sharing statement initially provided
6/12/2024
Type of registration
Prospectively registered

Titles & IDs
Public title
BRAIx RCT: A study on the safety and efficacy of reader replacement with artificial intelligence in population breast cancer screening
Scientific title
BRAIx RCT: A multi-state, single-blinded, randomised controlled trial assessing the impact of reader replacement with artificial intelligence on interval cancer rates in population breast cancer screening
Secondary ID [1] 312852 0
None
Universal Trial Number (UTN)
Trial acronym
BRAIx RCT
Linked study record

Health condition
Health condition(s) or problem(s) studied:
Breast cancer 334999 0
Breast cancer screening 335000 0
Condition category
Condition code
Cancer 331513 331513 0 0
Breast
Public Health 331514 331514 0 0
Health service research

Intervention/exposure
Study type
Interventional
Description of intervention(s) / exposure
Intervention: AI-integrated screening mammogram reading using investigational product; BRAIx AI Reader (Software as a Medical Device).
An AI reader (Deep Learning Classification Model) implemented to automatically analyse and assign a classification score to screening mammograms replacing the second radiologist reader in the independent double-reading of every mammogram, while maintaining the arbitration radiologist reader if needed. The first radiologist reader will not know the AI reader result in accordance with the current clinical routine (BreastScreen Australia National Accreditation Standards). If arbitration is required, the arbitration radiologist reader will be unblinded to both the AI and radiologist reader’s decision and have access to image annotations indicating Region of Interest of suspicion according to current clinical practice. Information about AI classification score and threshold will be available to the arbitration reader. The intervention will be implemented in the context of breast cancer screening, accordingly all women aged 40 and above who are eligible to participate in population-based mammography screening.
The approximate turnaround time for this mammogram double-reading procedure involving the AI reader, is 48 hours, which is well within the standard set by the BreastScreen Australia National Accreditation Standards, where screening reading results are provided to the client within 14 days.
A data log file procedure will be implemented to monitor adherence to the intervention if applicable, i.e. audit of patient medical records.
Intervention code [1] 329402 0
Early detection / Screening
Comparator / control treatment
Control arm: Current screening mammogram reading standard
Screening mammograms are read independently by two radiologists (independent double reading) who recall the client for assessment if there is suspicion of cancer. If there is a discrepancy a third arbitration radiologist, who will have access to image annotations indicating Region of Interest of suspicion according to current clinical practice, will make the final decision.
Control group
Active

Outcomes
Primary outcome [1] 339388 0
To assess the effect of AI-integrated mammogram reading compared with the current mammogram reading standard (independent reading by two radiologists and a third if discrepancy) on interval cancer rate (ICR).
Timepoint [1] 339388 0
24 months post final participant enrolment
Secondary outcome [1] 439731 0
To assess the safety of AI-integrated mammogram reading by monitoring the screen detected cancer rate (SDCR)
Timepoint [1] 439731 0
Every 6-months during trial intervention.
Secondary outcome [2] 439732 0
To assess the effectiveness of AI-integrated mammogram reading in breast cancer population screening in terms of accuracy and efficiency combined. Accordingly, effectiveness will be assessed as a composite secondary outcome evaluating the following measures together:
- screen detected cancer rate (SDCR)
- recall rate (RR)
- false positive rate (FPR)
- arbitration read rate (ARR)
- time to results (TTR)
- screen reading workload (SRW)
Timepoint [2] 439732 0
Every 28-days during trial intervention and 28-days post intervention
Secondary outcome [3] 441948 0
Sensitivity (derived from True Positive rate and False Negative rate) and Specificity (derived from False Positive rate and True Negative rate) will be assessed as a composite secondary outcome in the form of under the curve (AUC) in a receiver operating characteristic (ROC) curve to determine accuracy of the AI reader in the intervention compared to the human readers in the intervention arm. The AUC is an effective measure of accuracy because it takes into account the trade-off between sensitivity and specificity.
Timepoint [3] 441948 0
24 months post final participant enrolment
Secondary outcome [4] 442281 0
Positive Predictive Value will be derived from True Positive Rates and False Positive Rates to derive how accurate the AI reader is in the intervention compared to the human readers in the intervention arm.
Timepoint [4] 442281 0
24 months post final participant enrolment

Eligibility
Key inclusion criteria
To be included in the study, subjects must meet the following criteria:
- Women eligible to participate in population-based mammography screening in Victoria and South Australia.
Minimum age
40 Years
Maximum age
No limit
Sex
Females
Can healthy volunteers participate?
Yes
Key exclusion criteria
No exclusion criteria applied.

Study design
Purpose of the study
Diagnosis
Allocation to intervention
Randomised controlled trial
Procedure for enrolling a subject and allocating the treatment (allocation concealment procedures)
central randomisation by computer
Methods used to generate the sequence in which subjects will be randomised (sequence generation)
Computerised sequence generation.
Masking / blinding
Blinded (masking used)
Who is / are masked / blinded?
The people receiving the treatment/s


Intervention assignment
Parallel
Other design features
Phase
Not Applicable
Type of endpoint/s
Safety/efficacy
Statistical methods / analysis
SAMPLE SIZE CALCULATION
The RCT is designed to assess with 80% power and at a 5% significance level that the interval cancer rate (ICR) in the AI reading model intervention arm is non-inferior to the standard reading model control arm at a non-inferiority margin of 20% of the current rate. The sample size calculation assumes an ICR of 0.00188 (0.188%) in the control arm, and a 5% improvement in the intervention arm, i.e. an ICR of 0.00188 × 0.95 = 0.001786 (0.1786%). The non-inferiority margin is assumed to be 20%, i.e. 0.00188 × 0.2 = 0.000376 (0.0376%). To establish non-inferiority with a power of 80% and an alpha of 5% (i.e. a one-sided 95% confidence interval) requires 102,460 participants in each arm, resulting in a total sample size of 204,920. Sample size was calculated using the standard formulae for comparison of proportions in superiority or non-inferiority trials and calculations were performed in the online software package “Sealed Envelope”.

Non-Inferiority of ICR
ICR in the intervention arm compared to the control arm will be assessed by calculating 95% confidence intervals (one-sided and two-sided) for the difference in proportions. Non-inferiority of the intervention will be established if the one-sided interval does not cross the 20% margin. Superiority of the intervention will be established if the two-sided interval does not contain zero.

Recruitment
Recruitment status
Not yet recruiting
Date of first participant enrolment
Anticipated
6/01/2025
Actual
Date of last participant enrolment
Anticipated
3/08/2026
Actual
Date of last data collection
Anticipated
24/07/2028
Actual
Sample size
Target
205000
Accrual to date
Final
Recruitment in Australia
Recruitment state(s)
SA,VIC

Funding & Sponsors
Funding source category [1] 317297 0
Government body
Name [1] 317297 0
Department of Health and Aged Care - Medical Research Future Fund (MRFF)
Country [1] 317297 0
Australia
Primary sponsor type
Other
Name
ST VINCENT'S INSTITUTE OF MEDICAL RESEARCH
Address
Country
Australia
Secondary sponsor category [1] 319709 0
None
Name [1] 319709 0
Address [1] 319709 0
Country [1] 319709 0

Ethics approval
Ethics application status
Approved
Ethics committee name [1] 316030 0
The Royal Melbourne Hospital Human Research Ethics Committee
Ethics committee address [1] 316030 0
Ethics committee country [1] 316030 0
Australia
Date submitted for ethics approval [1] 316030 0
31/01/2024
Approval date [1] 316030 0
26/02/2024
Ethics approval number [1] 316030 0

Summary
Brief summary
Trial website
Trial related presentations / publications
Public notes

Contacts
Principal investigator
Name 136558 0
A/Prof Helen ML Frazer
Address 136558 0
BreastScreen Victoria; PO Box 542, Carlton VIC 3053
Country 136558 0
Australia
Phone 136558 0
+61 13 20 50
Fax 136558 0
Email 136558 0
hfrazer@breastscreen.org.au
Contact person for public queries
Name 136559 0
Katrina Kunicki
Address 136559 0
St Vincent's Institute of Medical Research; 9 Princes St Fitzroy VIC 3065
Country 136559 0
Australia
Phone 136559 0
+613 13 20 50
Fax 136559 0
Email 136559 0
kkunicki@svi.edu.au
Contact person for scientific queries
Name 136560 0
Katrina Kunicki
Address 136560 0
St Vincent's Institute of Medical Research; 9 Princes St Fitzroy VIC 3065
Country 136560 0
Australia
Phone 136560 0
+613 9231 2480
Fax 136560 0
Email 136560 0
kkunicki@svi.edu.au

Data sharing statement
Will individual participant data (IPD) for this trial be available (including data dictionaries)?
No
No/undecided IPD sharing reason/comment
All participant data will be de-identified and study findings reported at a cohort or aggregate level.


What supporting documents are/will be available?

No Supporting Document Provided



Results publications and other study-related documents

Documents added manually
No documents have been uploaded by study researchers.

Documents added automatically
No additional documents have been identified.