|
Objectives
|
Identifying biomarkers for early prediction of therapeutic outcomes |
Identifying biomarkers associated with anti-TNF drug concentrations |
Identifying biomarkers of clinical response to VDZ and USTE |
|
Treatments
|
Infliximab, adalimumab, or VDZ |
Adalimumab and Infliximab |
VDZ and USTE |
|
Study population and sample size
|
Discovery cohort: n=14 Replication cohort: n=23 |
Adalimumab: n= 187 Infliximab: n=198 |
Amsterdam Discovery Cohort: VDZ n= 64, USTE n=62 Oxford validation
cohort: VDZ n= 25, USTE n=33 |
|
CD or UC
|
Discovery cohort: 4 CD and 10 UC Replication cohort: 14 CD and 9 UC |
All participants had CD |
All participants had CD |
|
Sex distribution %female
|
Discovery cohort: 50% Replication cohort: 47.8% |
Adalimumab: 47.1% Infliximab: 56.1% |
Amsterdam Discovery Cohort: VDZ: 50%, USTE: 68% Oxford validation
cohort: VDZ: 36%, USTE: 55% |
|
Average age
|
Discovery cohort: 38.6 Replication cohort: 37.1 |
Adalimumab: 37.2 Infliximab: 35.3 |
Amsterdam Discovery Cohort: VDZ - R: 36, NR: 28 USTE - R: 38, NR: 35
Oxford validation cohort: VDZ - R: 41, NR: 42 USTE - R: 43, NR: 30
|
|
Ethnic background of participants
|
Not explicitly stated. Discovery cohort was likely German and
Replication cohort was likely Hungarian based on where the studies were
conducted |
Adalimumab: 94.1% White, 2.1% South Asian, 3.7% other Infliximab: 95%
White, 2% South Asian, 3% Other |
Amsterdam Discovery Cohort: VDZ - 75% White USTE - 76% White Oxford
validation cohort: VDZ - 88% White USTE - 85% White |
|
Sample type
|
Whole blood DNAm and RNA |
Whole blood DNAm |
PBL DNAm |
|
Sampling period
|
RNA: Baseline, 4h, 24h, 72h, week 2, week 6, week 14 DNAm: Baseline,
week 2, week 6 |
Baseline, week 14 week 30, week 54 |
Baseline and 6-9 months post treatment |
|
Results
|
- 85,728 DMPs, 357 DMRs, and 3043 DEGs in remitters
- 58,347
DMPs, 1163 DMRs and 389 DEGs in non-remitters
- Methylation
changes appeared more stable in remitting patients
- No clear
predictive signature for treatment failure at baseline, however week 2
DNAm and RNA response to treatment appeared to predict remission or
treatment failure |
- 4999 DMPs were identified between baseline and 14 weeks
- 323
DMPs were associated with elevated drug concentrations (sign of positive
response)
- 125 DMPs could be correlated to patient clinical
biomarkers
- Very few (13) DMPs were found when comparing
infliximab and adalimumab |
- Machine learning could predict response with an area under the curve
(AUC) of 0.87 for VDZ and 0.89 for USTE.
- VDZ had a sensitivity
of 0.769 and a specificity of 0.67
- USTE had a sensitivity of
0.73 and a specificity of 0.73
- The models developed in
Amsterdam correctly predicted non-response in 88.9 % of VDZ NR and 92.3%
of USTE NR |
|
Limitations
|
- Small cohort and sample size
- Lack of discussion of patient
demographics like ethnic background
- Combined CD and UC groups
may prevent accurate DNAm patterns
- Whole blood DNAm may
confound results |
- Outcome data may be improved with endoscopic outcomes
-
Whole blood DNAm may confound results
- Patient background was
grand majority European |
-Lower statistical power in the USTE group
- Approximately 70%
of participants had less stringent response assessments as they could
not receive endoscopies
- PBLs are a heterogeneous cell
population and subject to confounding |
