15/03/2012

1

Biomarker study approachesWillem Hanekom

Willem.Hanekom@uct.ac.za

G Poste. Nature 2011;459:156

15/03/2012

2

Biomarker for what?

1. Correlate of protection against infection2. Correlate of protection against disease3. Correlate of risk of infection4. Correlate of risk of disease5. Correlate of infection6. Correlate of disease7. Correlate of disease severity8. Correlate of prognosis9. Correlate of drug resistance10.Correlate of response to therapy11.Correlate of relapse following therapy

Biomarkers for TB vaccinology

• Correlate of risk of diseaseTargeted populations in efficacy trials

• Correlate of protection against diseaseQuick screens of vaccine candidates

15/03/2012

3

Is this important?

• 1.7 million die from TB each year

• Models: Vaccines that interrupt transmission will have major impact

Is this novel?

• Validated biomarkers have not been identified to date, using our approach

15/03/2012

4

What do you expect to find?

T cellMtb

Lung macrophage IFN-γ, IL-2, TNF

What do you expect to find?

T cell responses“Hypothesis-driven”

Gene expression profiles“Unbiased”

15/03/2012

5

“Unbiased” approaches

1. Transcriptomics2. Proteomics3. Metabolomics

etc.

Transcriptomics: measure mRNA

Gene expression

15/03/2012

6

Transcriptomics: 1. DNA microarrays

Transcriptomics: 2. RNA-Seq

15/03/2012

7

Biomarker

Hypothesis-driven

approaches

“Unbiased” approaches

Candidate approaches

How would you do a CoR study?

L Qin JID 2007;196:1304

15/03/2012

8

How would you do a CoP study?

L Qin JID 2007;196:1304

5,72

4 in

fant

s enr

olle

d

No TB disease: controls*

TB disease: cases* TB disease: cases*

2 yrs

BCG

To determine correlates of risk of TB disease, following BCG vaccination

Birth 10 wks

*Training and validation sets

15/03/2012

9

6,36

3 ad

oles

cent

s en

rolle

d

To determine correlates of risk of TB disease, following infection with Mtb

Tom Scriba, Adam Penn-Nicholson, Hassan Mahomed, Dan Zak, Alan Aderem, many others.

2 yrs

No TB disease: controls*

TB disease: cases*TB disease: cases*

*Training and validation sets

Matching• Cases randomly assigned to training and validation sets• Cases were divided into groups (bins) according to the

following variables (in order of importance):1. Passive or active original study arm2. Previous episode of TB3. Age at enrollment4. Gender5. Ethnicity

• Controls that matched all 5 variables were assigned to each case at a 2:1 ratio (for cross-sectional) or 1:1 ratio (for longitudinal)

To determine correlates of risk of TB disease, following infection with Mtb

Inclusion criteria• QFT and/or TST+• No TB for 6 months• HIV-negative

Slide courtesy of Tom Scriba

15/03/2012

10

15/03/2012

11

Which sample to use for analysis?

15/03/2012

12

PBMC MonocytesPurify CD14+ Monocytes

Antigens Ag85A/B ‐

ESAT‐6/CFP‐10 ‐

H37Rv H37Rv

Unstimulated Unstimulated

Purify T cells

RNA isolation, RNA‐seq RNA isolation, RNA‐seq

Specific gene expression analysis in the adolescent study

Slide courtesy of Adam Penn-Nicholson

15/03/2012

13

Thaw PBMC

Lyse cells, inactivate Mtb.  Freeze.

Monocytes H37Rv(6hr)

AutoMACS CD14+ “Possel”

PTC AutoMACS“Deplete”

Isolate RNA

Multicolourflow

Freeze Negfraction for HLA typing

Collect sup

2) (12hr)PBMC Peptides (BSL‐2) (12hr) (12hr)

PBMC H37Rv (12hr)

PTC OctoMACS

Collect supernatant, filter 

BSL‐3

Specific gene expression analysis in the adolescent study

Slide courtesy of Adam Penn-Nicholson

Bioinformatics approaches….1.

vs.

15/03/2012

14

Bioinformatics approaches….2.

Unsupervised

Supervised

Helen Fletcher, Ali Filali, Rafick-Pierre Sekaly, many

others.*DNA microarray analysis

of RNA from PBMC from 10 wks of age, incubated

with BCG for 12 hours.

Gene expression signatures that associatewith risk of TB, after Mtb infection

15/03/2012

15

Tom Scriba, Adam Penn-Nicholson, Dan

Zak, Alan Aderem, many others.

*RNA-Seq using RNA from Paxgene tubes, 6

months prior to TB Dx.

Gene expression signatures that associatewith risk of TB, after Mtb infection

Bioinformatics approaches….3.

Correlates

Biology

15/03/2012

16

Myeloid cell activation and inflammation associate with risk of TB, after BCG

Helen Fletcher, Ali Filali, Rafick-Pierre Sekaly, many others.*GSEA pathways from, DNA microarray analysis of RNA from PBMC from 10 wks

of age, incubated with BCG for 12 hours.

-0.5

0

0.5

1

M2.3M3.1

M1.2

M3.4

M5.12

M4.2

M5.1

M6.13

M3.2

M4.14

M4.6M6.6

M4.7

M6.9

M4.1

M6.15

M6.19

M4.10

M3.6

M4.15

M4.3

M4.5

Inflammation

Interferonresponse

Myeloid lineage

T cells

B cells

DOWNUP

Erythrocytes

Expression pathways: 6 mo prior to TB Dx in adolescents

Lymphoid lineage

Tom Scriba, Adam Penn-Nicholson, Dan

Zak, Alan Aderem, many others.

*RNA-Seq using RNA from Paxgene tubes.

Cytotoxicity

Protein synthesis

15/03/2012

17

Bioinformatics approaches….4.

Validation is critical!

Validated expression signatures that associate with risk of TB, after BCG

Helen Fletcher, Ali Filali, Rafick-Pierre Sekaly, many others.

*qPCR analysis of RNA from PBMC from 10 wks of age, incubated with

BCG for 12 hours.

15/03/2012

18

6363

ado

lesc

ents

enr

olle

d

To determine correlates of risk of TB disease, following infection with Mtb

Tom Scriba, Adam Penn-Nicholson, Hassan Mahomed, Dan

Zak, Alan Aderem, many others.

2 yrs

No TB disease: controls

TB disease: casesTB disease: cases

VALIDATION SET

Cases: n = 14Controls: n = 28

Cases: n = 8Controls: n = 8

Cases: n = 35Controls: n = 70

• Longitudinal Analysis

Cases: n = 28Controls: n = 28

TRAINING SET

• Cross-sectional Analysis

L Qin JID 2007;196:1304

15/03/2012

19

Validated expression signatures that associate with risk of TB, after BCG

Helen Fletcher, Ali Filali, Rafick-Pierre Sekaly, many others.

*qPCR analysis of RNA from PBMC from 10 wks of age, incubated with

BCG for 12 hours.

15/03/2012

20

ID in accessible body fluid

ID in accessible body fluid

Practical assay

Practical assay

MechanismsMechanisms

Modified, after M Disis 2011; Cancer Immunol Immunother 66:433.

Why do the biomarker study? Common steps in biomarker discovery

15/03/2012

21

Hypotheses

G Poste. Nature 2011;459:156

Clinical 

design

Engin

eering

15/03/2012

22

*Funding

*External

A few methodological lessons learnt

1. The clinical phenotypes are the most NB!2. Pay particular attention to sample size3. Make sure validation cohort adequate4. Immediate sample processing needs to be

optimal5. Optimal biobanking indispensible6. Define everything beforehand!7. Regardless, methods WILL change during the

study!

15/03/2012

23

Funders and partners