Productive long term collaborations can be built on our powerful - - PowerPoint PPT Presentation

▶
productive long term collaborations can be built on our
SMART_READER_LITE
LIVE PREVIEW

Productive long term collaborations can be built on our powerful - - PowerPoint PPT Presentation

Jan 05, 2024 437 likes •691 views

Productive long term collaborations can be built on our powerful statistical support Qin Liu MD, Ph.D. The Wistar Institute Statistical Practice in Cancer Conference March 1st, 2019 Statisticians task Provide critical collaborative

slide-1
SLIDE 1

Productive long‐term collaborations can be built on

  • ur powerful statistical support

Qin Liu MD, Ph.D. The Wistar Institute Statistical Practice in Cancer Conference March 1st, 2019

slide-2
SLIDE 2

Statistician’s task

  • Provide critical collaborative statistical support for

diverse biomedical projects

– Grant applications – Research designs – Data analysis results interpretations – Manuscripts

  • Develop novel statistical approaches for accurate

data analysis

slide-3
SLIDE 3

Example 1 Analysis of Mouse Tumor Growth Data

Determine treatment effects using longitudinal tumor growth data from a mouse model

slide-4
SLIDE 4

Tumor Growth Data from A Mouse Model Experimental Procedure

Randomization 10 mice / treatment Measure tumor volume over time Treatment Arm 1 Treatment Arm 2

slide-5
SLIDE 5

Example: Problems with Longitudinal Tumor Growth Data Analysis

Problem: The ANOVA test outcome at a single time point. It does not detect a statistically significant difference in longitudinal tumor growth between the BRAF inhibitor (PLX4720) alone vs. in combination with the CSF1R inhibitor (GW2580)

Single (PLX) Combo ANOVA compares fold-change 10 mice / treatment Single (GW) Control

slide-6
SLIDE 6

The ANOVA test compares fold‐changes of tumor volume from baseline at 14 days

One-way ANOVA with Bonferroni’s adjusted p-values were reported Control GW2580 PLX4720 GW2580 0.911 PLX4720 0.007 0.277 GW2580 + PLX4720 <0.001 0.005 0.693

slide-7
SLIDE 7

Mouse Tumor Growth Data

Problem: The ANOVA test compares a single time point. It does not detect a statistically significant difference in longitudinal tumor growth between the BRAF inhibitor (PLX4720) alone vs. in combination with the CSF1R inhibitor (GW2580)

Single (PLX) Combo Compare fold-change 10 mice / treatment Single (GW) Control

slide-8
SLIDE 8

Solution: Compare Tumor Growth Trends (Velocities) Using A Fitted Mixed model

p=0.002 p=0.012

Bonferroni’s adjusted

100 200 300 400 500 600 700 800 1 2 3 4 5 6 7 8 9 10 11 12 13 14 days Predicted trend (Control) Observed mean (Control) Predicted trend (GW2580) Observed mean (GW2580) Predicted trend (PLX4720) Observed mean (PLX4720) Predicted trend (Combined) Observed mean (Combined)

Tumor volume

p<0.001

Wang,…, Liu,…Herlyn, Kaufman. BRAF Inhibition Stimulates Melanoma-Associated Macrophages to Drive Tumor Growth. Clinical Cancer Research. 2015.

Control GW PLX Combination

slide-9
SLIDE 9

Example 2 Clinical Data Analysis

Examine the effect of aging on the response of melanoma to therapy

slide-10
SLIDE 10

Results From A Pre‐clinical Study ‐ Mouse Model of Melanoma

  • Tumors in young mice grew faster but they responded robustly to BRAF

inhibitor PLX4720

  • Tumors in aged mice responded poorly to BRAF inhibitor (PLX)
slide-11
SLIDE 11

Clinical Data Analysis

Kaur, …, Liu, …, Weeraratna. sFRP2 in the aged microenvironment drives melanoma metastasis and therapy resistance. Nature 2016; 532: 250-266. Data mining with stratifying patients by treatment and treatment history In patients treated by Vemurafenib without prior therapies,

  • lder individuals show muted tumor response to BRAF inhibitor Vemurafenib.
slide-12
SLIDE 12

Example 3 High Throughput Screening Data Analysis How can we evaluate the potential synergistic effect of two drugs from in vitro data efficiently and accurately?

A new statistical approach development and applications

slide-13
SLIDE 13

Why Do We Study Drug Combinations?

Limitations of single drug treatment

  • A single drug often has limited anti‐tumor effect
  • Resistance is a major issue

Advantages of treatment with drug combination

  • May achieve a desired efficacy at lower dose

with less side effects

  • May reduce the resistance
  • May have Synergistic effect
slide-14
SLIDE 14

Drug Combinations Can Target the Same

  • r Different Biological Pathways

Modified from Chudnovsky, Y. et al. Journal of Clinical Investigation. 2005;115:813-824 A B Loewe additivity Bliss independence C D

slide-15
SLIDE 15

Definition of Synergism

Bliss independence criterion

  • Additive effect – For given doses, the combined response (inhibition rate of

cancer cell growth) from two drug compounds equals the sum of each single drug response subtracted by the multiplication of each single drug response.

  • Synergistic effect – the observed response from combination of two compounds

exceeds their additive effect (excess over Bliss independence)

Bliss CI. The toxicity of poisons applied jointly. Annals of Applied Biology 1939; 26:585–615.

1,2 = 1 + 2 – 1* 2, 0< <1 1,2 > 1 + 2 – 1* 2

slide-16
SLIDE 16

Common Analysis Method Excess over Bliss

Predicted 1,2 = 1 + 2 – 1* 2, 0< <1 Additivity : Observed 1,2 – Predicted 1,2 = 0 Synergism : Observed 1,2 – Predicted 1,2 > 0 Antagonism : Observed 1,2 – Predicted 1,2 < 0

slide-17
SLIDE 17

Heat Map Using "Excess over Bliss”

2 0.666667 0.222222 0.074074 0.024691 0.00823 2 15 1 ‐2 ‐3 ‐11 0.666667 ‐13 8 ‐7 0.222222 ‐11 2 ‐7 ‐14 ‐27 0.074074 ‐9 4 3 25 15 8 0.024691 ‐12 14 21 13 5 2 0.00823 ‐12 9 7 25 8 13

Excess over Bliss scores = 0 indicates that the combination treatment is additive; Excess over Bliss scores < 0 indicates the combination is less than additive (antagonism); Excess over Bliss scores > 0 indicates activity greater than additive (synergism).

Gamitrinib PI3K inhibitor BKM110 µM

slide-18
SLIDE 18

Chou TC and Talalay P. Generalized equations for the analysis of inhibitions of Michaelis‐Menten and higher‐order kinetic systems with two or more mutually exclusive and nonexclusive inhibitors. European Journal of

  • Biochemistry. 1981; 115: 207‐216.

Zhao, W., Sachsenmeier, K., Zhang, L., Sult, E., Hollingsworth. R. E., and Yang,

  • H. (2014), “A new Bliss independence model to analyze drug combination

data,” Journal of Biomolecular Screening, 19(5), 817–821. Whitehead, A., Su, T‐L., Thygesen, H., Sperrin, M., Harbron, C. (2013), “Investigation of the robustness of two models for assessing synergy in pre‐ clinical drug combination studies,” Pharmaceutical Statistics, 12, 300‐308.

Other Analysis Methods Combination Index

slide-19
SLIDE 19

Two‐stage Nonlinear Dose‐Response Surface Model

Liu Q, Yin X, Languino LR, Altieri DC. Evaluation of drug combination effect using a Bliss independence dose-response surface model. Statistics in Biopharmaceutical

  • Research. 2018. 10:2, 112-122, DOI: 10.1080/19466315.2018.1437071
  • ,

,

  • ,,

1

  • and are the doses of Drug A and B used in combination, reach the inhibition rate of .

, and , are the doses of single drugs A and B that reach the same inhibition rate, .

First stage:

  • Estimate the Hill Slope and IC50 for drugs A and B, respectively;
  • Bootstrapping the parameters estimated from stage I (random samples of Hill slope &

IC50 drawn from bivariate-normal distribution for each drug with corresponding means & variance-covariance estimated from stage I); Second Stage:

  • Estimate Interaction Index
slide-20
SLIDE 20

Plate map for different drug concentrations combined with constant Cisplatin

Evaluation of drug combination effect

  • drugs in combination with Cisplatin

µM

20 6.67 2.22 0.74 0.25 0.08 0.03 0.01 0.003 0.001 20 6.67 2.22 0.74 0.25 0.08 0.03 0.01 0.003 0.001

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 A

DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO

B

DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO

C

DM SO DM SO Dox DM SO

D

DM SO DM SO Dox DM SO

E

DM SO DM SO Dox DM SO

F

DM SO DM SO Dox DM SO

G

DM SO DM SO Dox DM SO

H

DM SO DM SO Dox DM SO

I

DM SO Dox DM SO DM SO

J

DM SO Dox DM SO DM SO

K

DM SO Dox DM SO DM SO

L

DM SO Dox DM SO DM SO

M

DM SO Dox DM SO DM SO

N

DM SO Dox DM SO DM SO

O

DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO

P

DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO DM SO

CI‐994 CI‐994# SGC0946 SGC0946 LLY507 LLY507 PFI‐3 PFI‐3 C646# C646 LAQ824 LAQ824 bromosporine bromosporine GSK2801 GSK2801 PFI‐1# PFI‐1 SGC‐CBP30 SGC‐CBP30 I‐CBP112 I‐CBP112 JQ1/SGCBD01 JQ1/SGCBD01

slide-21
SLIDE 21

Evaluation of Drug Combination Effect with Interaction Index

slide-22
SLIDE 22

Validation of the Improved Efficacy of JQ1 and Cisplatin Combination in vivo

Yokoyama, …, Liu, …, Zhang. Inhibition of BET protein BRD4 activity synergizes with cisplatin in

  • varian cancer by targeting ALDH activity through an ALDH1A1 super-enhancer and the associated

enhancer RNA. Cancer Research. 2016

Tumor volume Survival

slide-23
SLIDE 23

Collaborations During The Past 5 Years

Two P01 Three U01 Several R01 and DOD grants > 80 coauthored scientific publications

slide-24
SLIDE 24

Thank you!