Lessons from the TASTE trial
Prospective Registry based
Randomized Clinical Trials (R-RCT) –
a new concept for clinical research
Stefan James, Uppsala University
Sweden
SWEDE
HEART
Which Treatment is Best for Whom?
High-Quality Evidence is Scarce
3
Tricoci P et al. JAMA 2009;301:831-41
Level of Evidence A
Current Guidelines
AF
11.7%
Heart failure
26.4%
PAD
15.3%
STEMI
13.5%
Perioperative
12.0%
Secondary prevention
22.9%
Stable angina
6.4%
SV arrhythmias
6.1%
UA/NSTEMI
Valvular disease
23.6%
0.3%
VA/SCD
9.7%
PCI
11.0%
CABG
19.0%
Pacemaker
4.9%
Radionuclide imaging
4.8%
0%
10%
20%
30%
Thrombus aspiration
in Sweden
SCAAR
SWEDE
HEART
TAPAS / Swedish registry data
TA+PCI (N=3 666)
PCI alone (N=16 417)
HR (95% CI): 1.21 (1.08-1.35)
Vlaar, P.J. et al. The Lancet 2008; 371:1915-20
Fröbert, O. et al. Int J Cardiol. 2010; 145:572-3
Register studies
Obeservational studies (None-inverventional)
Strengths
• Ideal for description of standars
• Unselected patient populations –generalizable
• Large number of events – makes it possible to identify rare events
• Inexpensive
Weaknesses
• Data quality variable and questionable
• Cannot be used for comparative outcomes research
• Confounding factors can not be adjusted for despite advanced
statistical models
Randomized Controlled/Clinical Trials - RCT
Non randomized
Observational studies
Randomized
Studies (RCT)
Randomized Clinical Trials- RCT
Strengths
 Correctly designed studies with adequate power are gold standard
 Extinguishes confounding
Weaknesses
 Highly selected populations due to exclusion criteria
 Often selected specialized study centers
 Often surrogate endpoints
 Long time to plan and complete
 Expensive
 Often sponsored by industry- only studies with economic interest will be
performed
SWEDE
HEART
Register based Randomized Clinical trials- R-RCT
Is a prosective randomized trial but it uses a clinical
registry for one or several major functions for trial
conduct.
Registry based Randomized Clinical trials - R-RCT
Strengths
•
Correctly designed studies with adequate power are gold standard
•
Extinguishes confounding
•
Unselected patient populations –generalizable
•
Large number of events – makes it possible to identify rare events
•
Inexpensive
Weaknesses
• Data quality lower
• Variable definition
Number of cases annually: 80 000
RIKS-HIA
73 CCU hospitals, 100%
SCAAR
30 PCI hospitals, 100%
Percutaneous valves
7 hospitals, 100%
Heart surgery
7 hospitals, 100%
Secondary prevention
65 hospitals, 85%
>200 variables
(Baseline data, procedural and outcome measures)
At monitoring: 95-96% agreement between files and registry.
SWEDE
HEART
Name, personal ID number
Data entry on line by the
operator
Administrative data
Automatic linkage with
population registry
Clinical background and prior CV disease
Automated data checks
Angiographic background data
Two questions need to be
answered:
Did the patient consent orally?
Are inclusion and no exclusion
criteria met?
Did the patient consent?
Are inclusion and exclusion crieteria met?
Information for consent
Did the patient consent?
Are inclusion and exclusion crieteria met?
Randomize and store data
Did the patient consent?
Are inclusion and exclusion crieteria met?
TASTE inclusion rate
Patients
All primary PCI:s
Randomized
7244 patients
Date
TASTE and previous studies
TASTE
TASTE
TAPAS
JETSTENT
AIMI
INFUSE-AMI
VAMPIRE
PREPARE
Chevalier
Kaltoft
MUSTELA
X AMINE ST
PIHRATE
EXPIRA
DEAR-MI
Liistro
0
1000
2000
3000
4000
Number of patients
5000
6000
7000
8000
TASTE trial enrollment flow chart
Enrolled in Denmark
N=247
All patients with STEMI in Sweden and Iceland undergoing
primary or rescue PCI. N=11 709 *)
Enrolled in TASTE
N=7259
Erroneous
enrollments
N=15
Not enrolled
N=4697
Randomized in TASTE
N=7244
N=3621 assigned
to thrombus aspiration
N=3623 assigned
to conventional PCI
N=3399 underwent
thrombus aspiration
N=222 underwent
conventional PCI
N=3445 underwent
conventional PCI
N=178 underwent
thrombus aspiration
N=3621 were
followed up
N=3623 were
followed up
N=1162 underwent
thrombus aspiration
N=1162 were
followed up
N=3535 underwent
conventional PCI
N=3535 were
followed up
All-cause mortality up to 1 year
HR up to 1 year 0.94 (0.78 – 1.15), P=0.57
HR up to 30 days 0.94 (0.72 - 1.22), P=0.63
Stent thrombosis
Reinfarction
2.7
2.7
HR 1 year 0.97 (0.73 – 1.28), P=0.81
HR 30 days 0.61 (0.34 - 1.07), P=0.09
HR 1 year 0.84 (0.50 – 1.40), P=0.51
HR 30 days 0.47 (0.20 - 1.02), P=0.06
Lagerqvist NEJM 2014
R-RCT vs. classical RCT
 New concept for clinical research
 Combines the advantages of a clinical registry and
randomized study
 Complement to classical RCT –No substitute
•RRCT
•Evaluation of therapeutic
options available/used in
routine clinical care
RCT
Approval of new
pharmaceutical agents and
medical devices
What can a registry do?
Some or all parts of trial
•
•
•
•
•
•
Identify patients
Randomize
Collect baseline and procedure characteristics (CRF)
Assist with and collect consent forms
Identify clinical endpoints (endpoint detection)
Control clinical outcome events (adjudication, CEC)
Study design
RCT
R-RCT
Strategy
+
Device – CE mark, used
+
Device, first in man
+
Approved drugs used in clinical
practise
Drugs for new indication
+
+
New drugs
+
+
Study design
• Simple hypotesis, one question- one answer
•
•
•
•
•
Sub-studies limited and simple
Treatment alternatives available
Well defined randomized options
Open lable with blinded evaluation of events (PROBE)
Blind, placebo controlled?
Endpoints
-
Well defined, death optimal
Clinical
Complete
Available (Delay for Swedish hospital admission
registry)
- Central clinical event committee (CEC) is needed
if not well defined events- particularly for open
label trials
Data base
Other national
registries (PAR, LM, )
Clincial register
(variabler ex.
personual ID)
Extra study specific
variables
Informed consent
Randomisation code
Incl-/exclusion critera
Study database
Alla variabler
Personal ID replaced to
study coode
Cannot be changed
Analyse databas
Personal ID replced
with study code
Relevant registry
variables
Available
Available for registry staff/ PI
Possibility to remove patients
from registry
Available for registry
staff/ for registry
staff/trialists
Not possible to remove
patients from a trial
Data checks
All patients kept untial
behålls tills ev återtaget
samtycke
Study design
Eligible patient*:
After informed consent
1:1 online randomisation
in ambulance or ED
Oxygen
*Inclusion criteria:
• symptoms suggestive of AMI
within 6h
• SpO2 ≥ 90%
• ≥ 30y
• ECG changes indicating ischemia
and/or elevated troponin levels
Air
6l/min for (6-)12h via
Oxymask
Primary Endpoint: 1-year total mortality
Additional secondary endpoint and sub studies
Data analysis through SWEDEHEART registry and national mortality registry
Ongoing R-RCT
VALIDATE (n=6000)
Bivalirudin versus Heparin in NST and ST- Elevation myocardial infarction in patients on modern
antiplatelet therapy in SWEDEHEART,
DETOX-AMI (n=7000)
DETermination of the role of OXygen in Acute Myocardial Infarction,
SWEDEPAD (n=2480)
SWEdish Drug Elution trial in Peripheral Arterial Disease. DES vs BMS and DEB vs POBA.
IFR SWEDEHEART (n=2000)
Instantaneous Wave-Free Ratio versus Fractional Flow Reserve in ACS
PROSPECT-2 (n=1200, hybrid trial)
Providing Regional Observations to Study Predictors of Events in the Coronary Tree. Evaluate
future events from cholesterol plaques detected by near infrared spectroscopi
DISCO (n=2480)
Evaluate if patients with out of hospital cardiac arrest should undergo routine coronary angiography
U-CARE (n=500)
Evaluation of internet based cognitive behavioural therapy (iCBT) versus usual care in patients with
depression/anxiety post MI.
Conclusions
• Large need for randomized trials (RCT) particularly for the
evaluation of strategies, devices, pharmacological
therapies
• Classical RCTs are often not performed in broad
representative patient populations
• The national clinical registries are strong networks for
collaboration and enroll complete patient populations
• Prospective Registry based Randomized Clinical Trials
(RRCT) is a new opportunity for clinical research
• RRCT is ideal for one clinically important hypothesis with
reliable hard endpoints