This document discusses key concepts in clinical trial methodology, including evidence-based medicine, levels of evidence, randomization, control groups, endpoints, and alternative trial designs such as basket, umbrella, and platform trials. Randomized, controlled trials are considered the gold standard but alternative designs may be needed for rare diseases. Basket and umbrella trials can study multiple treatments targeting specific biomarkers across different tumor types. Platform trials evaluate multiple treatments simultaneously using an adaptive design.

1. Markku Toivonen, MD, PhD
Scientific Director
NDA Regulatory Science Ltd.
REVIEW OF KEY CONCEPTS IN CLINICALTRIAL
METHODOLOGY

2. 2
EVIDENCE-BASED MEDICINE
Care of individual patients is based on evidence
From medical research
From medical expertise and experience
Medical research “gold standard”: randomised,
controlled trial (RCT)

3. LEVELS OF EVIDENCE
Confidence
Risk of bias
http://dx.doi.org/10.1136/ebmed-2016-110401
Murad et al.

4. 4
RANDOMISATION
Study participants are identified based on clear
criteria defined in advance (screening)
Participants fulfilling study criteria are assigned to
treatment groups randomly
 Computerised systems
Neither participants nor study investigators can
influence treatment allocation

5. 5 Methodological principles
ADVANTAGES OF RANDOMISATION
Avoids selection-based bias (systematic error) –
possibility that treatment outcome is influenced by
selection
Objective: treatment groups within a clinical trial are
balanced (similar) and only the treatment is different

6. 6
SHORTCOMINGS?
Randomisation does not guarantee that the study
setting is free from bias, particularly if
The participant, the investigator or both know what
treatment the individual will receive (open-label
study)
Studies should be conducted ”double-blind”
whenever possible

7. 7
CONTROLLED
Control group is a “yardstick” against which efficacy
and safety of experimental therapy can be evaluated
Control treatment (comparator) may be
Placebo
Standard care + placebo
Best supportive care
Established therapy – authorised or not authorised,
but evidence-based

8. 8 Methodological principles
OUTCOME MEASURES/ENDPOINTS
Often called variables
Objective or subjective
Objective: impartial – based on observation/physico-
chemical measure, low degree of interpretation
Subjective – important measure of perceived
benefit/harm
Surrogate endpoints
Biomarkers

9. 9
HIERARCHY OF ENDPOINTS
Defined in advance in protocol
Primary
 Most important measure
 Result defines study success/failure
Secondary
 Measure other important aspects of disease
Tertiary/exploratory
 Other measures of interest

10. 10
NON-STANDARD SITUATIONS
In priciple, development of orphan medicinal products is a
standard situation
Development may require non-standard solutions
Historical control group – natural history studies
• ”RealWorld Evidence”
No control group: single arm study
Adaptive studies
Sequential studies

11. 11
CONVENTIONAL PARALLEL GROUPS RCT -
EXAMPLE
Patients withType-2 diabetes mellitus screened according to protocol
inclusion/eclusion criteria
Randomised to blinded treatment: metformin or investigational drug
Followed for 12 weeks under treatment
Efficacy evaluated based on pre-set criteria
Safety

12. 12
OTHER STUDY APPROACHES – RANDOMISED
OR NON-RANDOMISED
Basket (Bucket) designs
Umbrella designs
Platform designs
Increasingly used in cancer research, but find use in other therapeutic areas
Not one disease – one treatment studies, but
 One disease – several treatments
 Several diseases, one treatment
 Several diseases, several treatments

13. 13
BASKET AND UMBRELLA DESIGNS IN CANCER
Background
Many genomic aberrations are seen across multiple tumor types,
e.g. breast cancer, lung cancer, colorectal cancer
Traditionally studies have selected one tumor type, e.g. lung
cancer, to test the new treatment
Basket and umbrella studies ask if the genomic aberration is more
predictive of treatment effect than the type of tumor

14. 14
BASKET STUDY EXAMPLE
Study population identified by presence of one biomarker (gene
mutation) in tumor
Positive: eligible for study entry
Patients with different tumor types placed in corresponding baskets
Same treatment given to all
Results analysed by combining all baskets and in individual baskets

15. 15
BASKET STUDY EXAMPLE
TREATMENTATARGETS BIOMARKER
TREATMENT
A
TREATMENT
A
TREATMENT
A
TREATMENT
A
ALL
PATIENTS
BIOMARKER
POSITIVE
SMALL CELL
LUNG CANCER
PANCREATIC
CANCER
COLORECTAL
CANCER
MELANOMA

16. 16
UMBRELLA STUDY EXAMPLE
Study targets single tumor type
Several biomarkers (genomic aberrations) present in the tumor type
Several treatments matched to biomarkers

17. 17
UMBRELLA STUDY EXAMPLE
TREATMENTSA, B, CTARGET DIFFERENT BIOMARKERS
SCREEN
PATIENTS
WITH BREAST
CANCER
BIOMARKER A
BIOMARKER B
BIOMARKER C
TREATMENT A
TREATMENT B
TREATMENT C

18. 18
PLATFORMTRIALS
Clinical trial with a single master protocol
Multiple treatments are evaluated simultaneously
In adaptive platform designs some treatments may
be dropped for futility (not effective), declared
superior over others, or new treatments can be
added
Platform trials may find beneficial treatments with
fewer patients, fewer patient failures, less time, and
with greater probability of success than a traditional
two-arm randomised, controlled study

19. 19
ADAPTIVE PLATFORMTRIAL EXAMPLE
EPILEPSY, SEVERAL INVESTIGATIONALTREATMENTS
SCREENAND
RANDOMISE
TREATMENT
A
TREATMENT
B
A + B
TREATMENT
C
TREATMENT
D
1.
2.
3.
4.
BEST
EFFECT
MODERATE
EFFECT
SMALL
EFFECT
SMALL
EFFECT
NO EFFECT,
DISCONT.

20. 20
TO CONCLUDE
Randomised, controlled, blinded studies provide best
scientific evidence of benefit and risk
Not always possible
Not always “efficient”
Alternative designs to be agreed with authorities in
advance