reduce & repeat

23
Reduce & Repeat Non-Clinical Statistics Conference 2014, Brugge October 2014 More Precise XC50s Using Fewer Wells (in vitro) and Fewer Animals (in vivo) Click icon to add classification from picture folder ‘AZ Graphics’

Upload: levi-campbell

Post on 31-Dec-2015

42 views

Category:

Documents


0 download

DESCRIPTION

Reduce & Repeat. More Precise XC50s Using Fewer Wells (in vitro) and Fewer Animals (in vivo). Non-Clinical Statistics Conference 2014, Brugge October 2014. Raw Conc -Response Data. Only Half of the Concs. Only Half of the Replicates. Only Half of the Concs and Half of the Replicates. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Reduce & Repeat

Reduce & Repeat

Non-Clinical Statistics Conference 2014, BruggeOctober 2014

More Precise XC50s Using Fewer Wells (in vitro) and Fewer Animals (in vivo)

Click icon to add classification from picture folder ‘AZ Graphics’

Page 2: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Raw Conc-Response Data

2 Jonathan Bright | September 2014

Page 3: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Only Half of the Concs

3 Jonathan Bright | September 2014

Page 4: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Only Half of the Replicates

4 Jonathan Bright | September 2014

Page 5: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Only Half of the Concs and Half of the Replicates

5 Jonathan Bright | September 2014

Page 6: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Only Half of the Concs and Half of the Replicates and Half of the Controls

6 Jonathan Bright | September 2014

Page 7: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 and 95% Confidence Interval

7 Jonathan Bright | September 2014

Page 8: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Findings #1

• For a “well-behaved” assay, the resource (wells) may be dramatically reduced with little impact on either the estimate of the XC50 or its confidence interval

• “Well-behaved”- Max and min controls that safely position the curve top and

bottom- Conc-response data that have the right sort of sigmoid

shape- Acceptable to overlook details such as biphasic and partial

inhibition• May be exploited

- Throughput- Cost- Compound use

8 Jonathan Bright | September 2014

Page 9: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Second Set of Raw Conc-Response Data

9 Jonathan Bright | September 2014

Page 10: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 and 95% Confidence Interval

10 Jonathan Bright | September 2014

Page 11: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 and 95% Confidence Interval

11 Jonathan Bright | September 2014

Page 12: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Findings #2

• Run-to-run differences in XC50 are massive compared to the small changes in XC50 that occur as a result of reducing the resource (wells) on any given run

• Put in terms of components of variation- Between run variation dominates within-run variation- Within-run variation changes hardly at all as the number of

concs and number of replicates changes• May be exploited

- Reduce the resource per run- Repeat- Average

12 Jonathan Bright | September 2014

Page 13: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

13 Jonathan Bright | September 2014

Page 14: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

14 Jonathan Bright | September 2014

Page 15: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

15 Jonathan Bright | September 2014

Page 16: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

16 Jonathan Bright | September 2014

Page 17: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

17 Jonathan Bright | September 2014

Page 18: Reduce & Repeat

Innovative Medicines | Discovery Sciences

IC50 vs Run

18 Jonathan Bright | September 2014

Page 19: Reduce & Repeat

Innovative Medicines | Discovery Sciences

In Vivo

• A situation similar to the in vitro case has been observed, whereby study-to-study differences are the main component of variation

- Was it a “good day” or a “bad day” for compound X• 2 Start Strategy (Brian Middleton)

- Start half the planned animals (reduce)- Independently run the second half (repeat)- Average

• Gives a superior estimate of the e.g. XC50 or XD50• Provides in some cases a chance to change doses for the

second start

19 Jonathan Bright | September 2014

Page 20: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Summary

• In both in vitro and in vivo settings there are large run-to-run or study-to-study differences when a compound is retested

- Root cause analysis• Exploit by

- reducing the resource (wells or animals) on a given occasion- repeating the experiment- averaging across the experiments

• Reduce- Throughput, cost and compound benefits

• Reduce and Repeat- Precision benefit +

20 Jonathan Bright | September 2014

Page 21: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Acknowledgement and Reference

• Siller H, Taylor JD, Middleton B. Two-start design within a Sephadex inflammatory model – A means to generate reliable ED50 data whilst significantly reducing the number of animals used. Pulm Pharmacol Ther 2012; 25:223-227.

21 Jonathan Bright | September 2014

Page 22: Reduce & Repeat

Innovative Medicines | Discovery Sciences

Extra Slide

22 Jonathan Bright | September 2014

Page 23: Reduce & Repeat

Innovative Medicines | Discovery Sciences23 Jonathan Bright | September 2014

Confidentiality Notice This file is private and may contain confidential and proprietary information. If you have received this file in error, please notify us and remove it from your system and note that you must not copy, distribute or take any action in reliance on it. Any unauthorized use or disclosure of the contents of this file is not permitted and may be unlawful. AstraZeneca PLC, 2 Kingdom Street, London, W2 6BD, UK, T: +44(0)20 7604 8000, F: +44 (0)20 7604 8151, www.astrazeneca.com