MSBase Registry Helmut Butzkueven Director, MS Service, Royal - - PowerPoint PPT Presentation

“Real-world” disease outcomes : Experience with the use of the MSBase Registry

Helmut Butzkueven Director, MS Service, Royal Melbourne and Box Hill Hospitals, Australia Managing Director, MS Base Foundation

Data collection

• Data collection has always been a core activity
• f doctors
• “Medical notes”

Clinical practice data collection was very separate from research

Computerisation

General research challenges

• Using these data for research blurs traditional

boundaries of research and service

• When is consent required?
• How specific does it have to be?

Electronic medical record (EMR) for clinical practice and research

• Generic EMR’s are not very useful for disease-

specific data collection, because they are built for generic needs

• Disease-specific modification of EMRs can be

time-consuming and very expensive

• Most centres wishing to collect disease -

specific data have a specific disease interest

– Build their own database – Use an available disease-specific EMR

A registry, like a trial, uses an agreed minimum dataset

The basic language of Multiple Sclerosis (a typical chronic disease)

• Demographics
• Diagnostic test (LP, VEP, MRI Brain and spinal

cord classifiers)

• EDSS/ Kurtzke Functional System Scores
• Relapse date, site, treatment
• Disease modifying drug start and stop dates

Information

• The complexity grows over time….

35 yo woman: 12 years of MS… iMed

The history of MSBase

• The iMed electronic medical record launched by

Serono as a service to doctors in 2001

• Rapidly became very popular in Europe, Australia

and Canada

• Thought-leaders associated with Serono (Nazih

Ammoury, JP Malkowski, Samir Mechati) believed that these iMed records could create codified extract files and that

• These extracts could be combined into a global
• utcomes database: MSBase

MSBase Principles: Investigator Autonomy

• All investigators (centres) agree to a prospective

minimum dataset collection

• Use either iMed or an online data collection tool
• All can propose and conduct studies utilsiing the

minimum dataset

• All can propose and request access to the dataset for

analyses

• Investigators must follow all local rules (consent, ethics

approvals)

• Investigators remain as custodians of their own

datasets

Ethics/Consent

• For research on the minimum dataset
• Not specific

– Demographic trends, global comparisons – Treatment effects – Serious adverse event rates

• Allows

– Collaboration with the pharmaceutical industry – Investigator reimbursements

Governance

• Independence (Not for profit company) provides

great flexibility

• Clear separation of roles

– Administration/Operations – Research teams – Business and scientific leadership

• Formality : Following the mutually agreed rules

– Documentation of procedures, committee terms of reference, delegations of authority

• The MSBase Registry

– 28 Countries – Over 150 participating Centres – Over 34,400 patient datasets – 165,000 patient years of follow up – 338,000 EDSS (neurological score) evaluations – Median visit density is at 5.5 months

Enrolment since 2004

Can create clinically meaningful feedback to clinicians

• Benchmarking
• Severity calculators

Severity calculator

Severity calculator

A few recent analyses from MSBase

• Therapy persistence: DMD discontinuation in

clinical practice

• Head-to-Head treatment comparisons

Therapy persistence

Background

• Interferon-beta and glatiramer acetate are

the most common initial therapies in relapsing

• MS. Their route of administration and

tolerability profiles can limit persistence

• Persistence in clinical practice and major

factors determining persistence remain incompletely characterised.

• We prospectively characterised treatment

persistence in International MS populations using MSBase.

Patient studied in seen from onset cohort with first treatment initiation

Discontinuation rates 2015

Warrender-Sparkes et al, under revision MSJ

Treatment identity predicting discontinuation: Multivariable Survival model (3)

Predictor Annualised rate HR P-value

IFNβ-1a SC 0.20 1.0 (Ref) IFNβ-1a IM 0.19 0.98 NS IFNβ-1b 0.21 1.10 NS GA 0.23 1.13 NS NAT 0.21 0.93 NS FTY 0.11 0.44 < 0.001

Factors predicting discontinuation: Multivariable Survival model (1)

Predictor Annualised rate HR P-value Female Sex 0.23 1 Male Sex 0.16 0.73 <0.001 Australia 0.33 1 Netherlands 0.28 0.86 NS Canada 0.21 0.86 NS Italy 0.18 0.67 <0.001 Spain 0.15 0.49 <0.001

Head-to-head efficacy studies in MSBase

Introduction to Propensity Score Matching

Covariates Sex Age Treating centre / country Disease duration Any prior immunosuppresive treatment Number of treatment starts Number of treatment starts / disease duration EDSS Total relapse onsets last 12 months Total steroid-treated relapses last 12 months Total relapse onsets last 24 months Total steroid-treated relapses last 24 months

Multivariate logistic regression model

0.41

1 Propensity score

0.41

Treatment A Treatment B

Kalincik et al., PLoS One 8:e63480

• Propensity score: probability of receiving a

treatment based on a series of covariates

• Propensity score estimated using

multivariate logistic regression

Propensity Score

1 0.5

A - Randomized Trial (a posteriori) B – Observational Study C – Unusable Observational Study

Propensity Score 1 0.5 Propensity Score 1 0.5 Propensity Score

Propensity score matching:

• Method that mimics

randomization in observational studies

• Compares individuals who had

a similar probability (propensity score) of receiving the same treatment but actually received different treatments.

Propensity Score Overlapping

0.76 0.33 0.87 0.69 0.79 0.74 0.81 0.82 0.71 0.65 0.90 0.39 0.15 0.42 0.91

Propensity score determined for each patient; patients received Treatment A are matched with patients with a similar propensity score who received Treatment B

Propensity Score Matching

0.74 0.48 0.35 0.95 0.70 0.32 0.82 0.15 0.07 0.35 0.10 0.42 0.15 0.22

Fingolimod or Natalizumab

injectable Relapse Fingolimod Natalizumab

Fingolimod versus Natalizumab

• After relapse on Injectable

– 560 natalizumab switchers – 232 fingolimod switchers – Could match 407 NAT to 171 FNG

Persistence: highly similar in the treatment failure population (in the first two years

• nly)

Kalincik et al, Ann Neurol. 2015;77:425-35.

Time to first relapse

Kalincik et al, Ann Neurol. 2015;77:425-35.

Annualised relapse rate

Kalincik et al, Ann Neurol. 2015;77:425-35.

Disability Progression events- no difference

Kalincik et al, Ann Neurol. 2015;77:425-35.

Disability regression events

Kalincik et al, Ann Neurol. 2015;77:425-35.

Conclusions

• Natalizumab was equal to fingolimod in

– Persistence (over two years- it is likely that natalizumab will persistence will drop after that) – Disability progression

• Natalizumab was superior to fingolimod in

– Relapse rate reduction – Disability regression (20 versus 10 %)

Kalincik et al, Ann Neurol. 2015;77:425-35.

Summary

MSBase registry

• Is user-friendly
• Value-add to clinical practice

– Graphical representations of patient course – Benchmarking – Helps to create better decision tools

MS registries:

• Are a great way to capture large populations

for

– Comparative DMD effectiveness – Long-term disease trends ALSO…. – Pregnancy exposure outcomes data – Safety registries (Cancer, infection, mortality) – National and regional registries

• With special thanks to
• 150 investigating centres, 34400 patients
• Analysts

– Discontinuation (Claire Meyniel, Vilija Jokubaitis, Tim Spelman, Matthew Warrender-Sparkes ) – Head to head (Anna He, Tomas Kalincik, Tim Spelman)

• MSBase Administration (Jill Byron, Eloise Hinson, Lisa

Morgan, James Milesi )

• MSBase Platform and IT (Samir Mechati, Eric Bianchi,

Alex Bulla, Matthieu Corageoud )