Integrating an exploratory BM in an early clinical stage ... · In collaboration with TransMed Experts and biomarker groups model-based analysis of efficacy or safety biomarker data

Ulrich Kunz, Translational Medicine and Clinical Pharmacology, Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany

Integrating an exploratory BM in an early clinical stage in Pharma R&D

Content of Workshop

1. Introduction

2. Getting information about BM1: Assay request process (Biology of the BM and intended use)

3. Translation into a bioanalytical strategy: Choice of an analytical method for BM1

4. Validation of the BM assay for the first phase I study

5. Sample measurement: results and data evaluation

6. Outcome and consequences on the further development

U. Kunz, June 2016 2

Clinical development Plan

Overview of the (ideal) Process of identifying BM for early clinical studies at BI

Research Teams Disease Indication Teams

Biomarker Medical Subteam (responsible for BM strategy, Pk, Pharmacometrics, Statistics, Clinical, Pre-clinical, Research, lead: TMCP)

TMCP-Plan • Project strategy early clinical development • Outlook on late stage clinical development • (planned studies, population, which BM

and why)

Translatorical Studies

Literature

Contacts to external researchers

Lead Optimization

Start of clinical development

Bioan

alytical sup

port

(assay feasibility, search

for meth

ods, reagen

ts, assay d

evelopm

ent, valid

ation, m

easurem

ent of sam

ples tran

sfer to oth

er sites)


Analyze samples 1. phase I study

Flow for biomarker method establishment


Bioanalyst + Sponsor (TMBE): Translation into a Bioanalytical

strategy and assay design

Bioanalyst + Sponsor: Agree on future intended

use of BM and assay requirements

Bioanalyst + Sponsor: Qualify BM assumptions +

evaluate assay performance

Fail

Success

Set up/modify and validate the level 2/qualified assay

Analyze samples Phase II study

BM Medical Subteam, Translational Medicine and BM expert:

BM strategy

Feasibility study/ method set up and validation

Level 3/Screening assay

Continuing clinical and analytical interpretation of data from further

studies (monitoring of assay performance)

TM & CP plan

CTP (Expl. BM) + Analyt. Workplan

Validation plan

(Re-) Validation plan + Validation Report

CTP (Expl. BM) + Analyt. Workplan

Assay request process Understanding biology

and intended use of BM

Assay request process

Sponsor (Translational Medicine and Biomarker Experts) has to provide sufficient information prior to start of any lab activities Comprehensive questionaire to fill out:

• General status of method, BM and scope of analytical activities

• Structure and Biology of BM

• Intended use of BM results

• Prerequisites for the analytical method

• Planned clinical studies

Thereafter, joint discussion between sponsor and bioanalyst:

• Practical aspects of requested lab support


BI tiered method validation approach for BM assays


Tiered approach: Validation level may change dependent on: • intended use of data, • clinical qualification status of BM and • knowledge about BM-assay and BM

Advantages: Fast , flexible and cost effective (avoid frontloading of work)

• Explorative biomarker

• To gain experience with the biomarker and the biomarker assay

• Comparability within a single study (narrow time window)

Level 3 (Screening assay)

• Explorative and probable valid biomarker

• Comparability within a central lab over several studies (e.g. whole development)

Level 2 (Qualified assay)

• Confirmatory biomarker (probable valid, valid)

• Comparability between labs • External by diagnostic companies or CRO labs with

extensive cross validation, Global reference system established and proven metrological traceability

Level 1 (Cross) validated assay)

Clinical development Plan

How BM1 was nominated as explorative BM for phase I clinical study - very late, phase I clinical trail has already started -

BI Research Teams Disease Indication Teams

Biomarker Medical Subteam (responsible for BM strategy, lead: TMCP)

TMCP-Plan Version 2

Translatorical Studies

Literature

BM1

Contacts to external researchers

Start of phase I clinical trial

Bioan

alytical su

pp

ort (search

for meth

ods,

validation

, m

easurem

ent of

samp

les)

Amendment to Clinical Trial Protocol 1. Phase 1 study

Interim analysis planned


TMCP Plan

The Translational and Biomarker strategy for this program has two main objectives:

1. Identify disease specific markers that change early after treatment with drug and correlate with clinical endpoints.

2. Identify mechanistic or disease specific markers that can predict clinical response to treatment with drug

Efficacy Markers in the target tissue:

......

Disease Markers in Peripheral Blood :

BM1 is a marker shown to be highly expressed in both the target tissue and serum of patients and higher levels correlated with higher clinical symptoms. Changes in serum levels of BM1 will also be assessed for PK/PD correlations .

BIOMARKER ANALYSIS STRATEGY

In collaboration with TransMed Experts and biomarker groups model-based analysis of efficacy or safety biomarker data may be conducted. The impact of serum levels of BM1 protein concentrations on the longitudinal, population model-based analysis of clinical outcomes is planned as an exploratory analysis using study data.


Assay request: Biology of BM1

Structure of BM (link to proteome database)

Protein MW: 6 kDa, cationic, cystein rich

relevant literature Some available about use of BM1 as biomarker in clinical studies

Biological Variability within and between populations (e.g. Differences Healthy Volunteers – Patients)

Lit: Patient level > healthy volunteers (< 1ng/mL) Variability between patients high, correlation with disease status

Presence in biological context, e.g. monomer, multimer (mono-, hetero-), splice variants, homologies, chemical isomers

Several similar proteins present in blood, no info about multimers, commercial available anti-BM1 antibodies and kits states selectivity regarding BM1

Interaction with other matrix compounds (binding proteins, receptor, drug....)

Unknown, interaction with anionic peptides possible

Circadian rhythm known? Influence of extrinsic factors (e.g. fasted – non-fasted, coagulation....)?

Lit: no evidence observed in healthy serum BM1 levels

Stability of BM (any hints for instability?) Lit: „no large differences found in serum samples taken from the same individuum after a two year interval“ !!

Preferred matrix Serum and plasma (depending on what aliquots are remaining for BM1)


Assay request: Intended use of BM1

Biomarker category • Known: Physiological Response Pd-Biomarker • Option: Outcome Related Pd-Biomarker?

Planned evaluation of results (for details please see: Trial Statistical Analysis Plan)

1. Relative change from baseline levels during treatment 2. Correlation of baseline levels and clinical symptoms

Intended purpose Outcome related BM: 1. How much reduction of peripheral levels of BM1 at week X do we

need to achieve a 80% reduction in clinical symptoms at week 12? 2. Are baseline levels of BM1 predictive of clinical response? 3. Is BM1 a surrogate marker for clinical symptoms, independent of

mechanism specific intervention? → PK/PD with BM1 data.

What is the expected effect during treatment? • not known • about 50% • 1 - 50% • 150 – 200% • 200 – 1000% • >1000%

Reduction > 50% = 1 -50% of baseline

Which Lower Limit Of Quantification of the assay is needed?

Healthy volunteer levels (Lit: about 1ng/mL)

Information about existing analytical methods and current experience

• Lit about use of commercial kits and own ELISAs • Kits from multiple vendors available


Assay request: Intended use of BM1

1. Expected number of samples in first studies?

2. Timelines of first studies 3. Interim analysis necessary?

1. About 300 in phase I, about 1000 in phase II (per indication)

2. Already running, about 3 month from now on (for interim)!!!!!

3. Yes after third dose cohort

Use in further studies? Useful in other indications/projects?

Yes, most likely phase II, support PK/PD modelling through expansion of kinetic data (together with mRNA expression) Yes, data might support the transition into alternative indication/diseases


Question 1

1. Was the available information about BM1 and its intended use sufficient? What was missing? What assay format would you choose?


Translation into the bioanalytical strategy for (running) phase I and beyond (sponsor + bioanalyst)

Biomarker of interest (incl. details about special isoforms or interesting epitopes or assay conditions)

Total BM1 (no infos about isoforms, special selectivity of assays used in lit.)

Necessary grade of quantification Relative quantitative (comparison of pre-dose levels necessary)

Validation level for first study Level 3 (explorative BM, proof of hypothesis from lit, test of assay suitability, no inhouse experience with assay so far = not used in pre-clinic)

Special requirements for assay development or validation, (e.g. sampling)?

No (nothing known about needs for special sampling procedures, instabilities)

Selected technology • First choice: commercial ELISA kit (already available, high sensitivity not necessary)

• in principle LC-MS might be an option

Patient samples available (contact)? Some banked patient samples available

Location of sample analysis (internal lab, external CRO)

• Internal until end of phase I (quick results needed)

Request to bioanalytical lab: • Feasibility and selection out of various commercial assays • Validation (level 3) • Measurement of phase I samples (serum and plasma)


Selection of assay/Assay feasibility

• Various BM 1 ELISA kits from different vendors available

• Several anti-BM1 antibodies and recombinant BM1 available

Feasibility test of 6 different ELISA kits from 6 different vendors

Tested parameter:

• Calibration range, curve fitting, signal to noise ratio lowest C/blank

• Reactivity of two external calibration standards (rec. Protein, synthetic peptide)

• Test of three individual sera from healthy volunteers + corresponding EDTA plasma in three dilutions (1:1 – 1:5)

• Test of a patient serum with expected high endogenous level in several dilutions


Selection of ELISA kit - Assay feasibility

Calibration range, curve fitting, signal to noise ratio lowest C/blank

ok ok ok No curve! ok Only 4 calibrators

Reactivity of two separate calibration standards (rec. Protein, synthetic peptide)

Parallel curves

Parallel curves

Parallel curves

- Conc. Independent response = blank!?

Parallel curves

Test of three individual sera from healthy volunteers + corresponding EDTA plasma in three dilutions (1:1 – 1:5)

Non parallel Serum>plasma



- High signals, cannot be blocked with anti-BM1 antibody

Sensitivity not sufficient

Test of a patient serum with expected high endogenous level in several dilutions

Non parallel Parallel 1:5 – 1:50

Non parallel

- - Non parallel

Within replicate precision worth ok ok - ok ok

comment preferred Kit does not work

What does this kit measure?


Question 2

2. Has the intended use of BM1 data been translated adequately into the bioanalytical strategy? (choose of assay, feasibility experiments, plans for future validation and analyzing strategy)?


Assay validation plan, commercial kit#2, validation level 3

Validation plan experiment Calculation, method acceptance

VCs (validation samples)

Three individual sera (low, mid, high) and EDTA plasma, aliquoted and stored frozen

Precision

At least 3 runs at 3 day, VCs in independent duplicates

Tabulate and calculate overall precision (CV), optimal: <30% CV

Calibration range, LLOQ, ULOQ

Data from at least 3 runs at 3 days, whole curve + blank

Tabulated back calculated results of Cs, LLOQ = lowest valid C that can be distinguished from blank, ULOQ = highest valid C * highest valid dilution in parallelism exp.

Parallelism 4 individuals serum ( Healthy and patients), 4 individuals plasma

linear regression of log transformed values (0.8 < slope < 1.2, r²>0.98)

Baseline range Healthy Volunteers

N=12 serum and EDTA plasma Just report mean, min, max, CV, ratio plasma/serum

Short term stability endog. Analyte

24h RT, +1, +3 freeze/thaw cycles (-20°C) Two individual samples (2 serum + 2 plasma) in duplicates (VCs),

Relative deviation stressed aliquots vs. frozen reference < 30%

Expected BM effect >50% decrease


Assay validation results

Validation plan

experiment Calculation, acceptance Results

Precision

4 runs at 3 days, VCs in independent duplicates

Tabulate and calculate overall precision (CV), target: <30% CV

Serum: 4-11% CV (N=8), Plasma: 5-13%CV (N=8)

Calibration range, LLOQ, ULOQ

4 runs at 3 days, whole curve + blank

Tabulated back calculated results of Cs, LLOQ = lowest valid C that can be distinguished from blank, ULOQ = highest valid C * highest valid dilution in parallelism exp.

0.0312 – 1.5 ng eq/mL ok, CV <4% (N=4), %dev: -3% - +5% LLOQ = 0.31 ng eq/mL (1:10) ULOQ = 750 ng eq/mL (1:500)

Parallelism 4 individuals serum and 4 plasma

linear regression of log transformed values (0.8 < slope < 1.2, r²>0.98)

2 out of 4 non-parallel serum 3 out of 4 non-parallel plasma


N=12 serum and EDTA plasma

Just report mean, min, max, CV Serum: Mean 2.6 ng eq/mL (N=11), Min 0.62, Max 6.1, inter-indiv. CV = 71% Plasma: Mean 2.2 ng eq/mL (N=11), Min 0.63, Max 5.1, inter-indiv. CV = 69% Plasma/Serum = 0.85


24h RT, +1, +3 freeze/thaw cycles (-20°C) Two individual samples in duplicates (VCs),

Relative deviation stressed aliquots vs. frozen one < 30%

Serum: stable 24h RT, +1, +3 freeze/thaw cycles (-20°C) Plasma: stable 24h RT, +1, +3 freeze/thaw cycles (-20°C)


Validation results - Precision


Run

Date Kit Lot#

Curve

Number

QC Low

(ng eq/mL)

QC Mid

(ng eq/mL)

QC High

(ng eq/mL)

ratio

Low/Mid

ratio

High/Mid

24-Aug-13 #2 V_0001 2.10 3.43 18.7 0.61 5.5

2.30 3.45 16.8 0.67 4.9

27-Aug-13 #2 V_0002 1.97 3.35 21.4 0.59 6.4

2.10 3.31 21.1 0.64 6.4

27-Aug-13 #2 V_0003 1.67 3.15 18.1 0.53 5.7

1.67 3.14 17.6 0.53 5.6

28-Aug-13 #2 V_0004 1.97 3.45 18.5 0.57 5.3

1.98 3.37 18.2 0.59 5.4

mean 1.97 3.33 18.8 0.59 5.6

sd 0.22 0.13 1.62 0.05 0.51

CV 11.0% 3.8% 8.6% 8.1% 9.1%

Validation results - Parallelism

Acceptance criteria:

0.8 < slope < 1.2

AND r² > 0.98

sample: concentration[ng eq/mL]

dilution 1: #P1 #P2 #P3 #P4 #S1 #S2 #S3 #S4

10 5.28 3.17

20 5.39 3.84

30 26.1 5.32 8.18 26 21.2 4.36 38.8

40 24.7 4.87 7.77 24.9 20.4 4.53 33.9

50 24.3 4.53 6.95 23.6 18.1 4.11 28.6

100 21.7 198 5.13 20.4 15.9 24.3

200 20.3 170 17.2

300 158

500 138

CV 10% 15% 7% 19% 16% 13% 13% 20%

rel. Range 24% 37% 16% 41% 37% 28% 33% 46%

linear regression

log (1/dilution) log(conc.)

-1.00 -0.28 -0.50

-1.30 -0.57 -0.72

-1.48 -0.06 -0.75 -0.56 -0.06 -0.15 -0.84 0.11

-1.60 -0.21 -0.91 -0.71 -0.21 -0.29 -0.95 -0.07

-1.70 -0.31 -1.04 -0.86 -0.33 -0.44 -1.09 -0.24

-2.00 -0.66 0.30 -1.29 -0.69 -0.80 -0.61

-2.30 -0.99 -0.07 -1.07

-2.48 -0.28

-2.70 -0.56

slope 1.133 1.221 1.086 1.403 1.220 1.248 0.804 1.384

r² 0.99983 0.999863 0.994746 0.998475 0.999889 0.998258 0.983918 0.994972

Parallel? 1 0 1 0 0 0 1 0 Failed ! U. Kunz, June 2016 20

Questions 3

3. Has the assay been sufficiently proven/validated prior to start of sample measurements? What experiments are missing, not adequate?


Discussion with sponsor

1. Assay feasibility tests went fine (kit lot#1) But validation failed (due to different kit lot #2?) -> Consultation with the kit supplier (old lot #1 no longer available, third lot failed too)

2. Unfortunately, not sufficient time for an alternative assay, clinical study has already started -> Consultation with sponsor (Translational Medicine and Biomarker Expert)

3. Decision to continue with this assay as a quasi-quantitative assay.


Consequence for bioanalysis

• Keep the validation as it is

• Avoid reporting of relative concentration values (misleading)

• Pre-screening of pre-dose samples in various dilutions, titer determination (LLOQ = cut-off), reporting of titers for pre-dose samples only

• Selection of optimal individual dilution for each subject (upper range of calibration curve)

• Measurement of all samples of a subject in the fixed optimal dilution

• Calculation and reporting of post-/pre-dose ratio for post-dose samples

-> Discussed in the validation summary and as part of the Analytical Workplan for BM1 measurement in the first study


Question 4

4. Assay acceptance after validation. What about the consequences of the failed parallelism experiment? quasi-quantitative evaluation of data ok?


In-study acceptance criteria (technical)

General, all samples

• Is the CV of the response values of the dependent replicates (one preparation of sample in multiple wells) above 25%? IF yes THEN deactivate the whole sample (automatically done by Watson).

Dynamic range = Check of calibration curve

• Are there calibration standards with a relative deviation from nominal/target of more than 20%? IF yes THEN deactivate the whole standard (all replicates).

• Recalculation of calibration curve Calibration range is truncated if lowest or highest standard failed.

• IF more than 25% of all replicates were excluded OR less than 5 valid calibration standards are remaining OR r² < 0.98 THEN the plate/run failed; → Reject the run; STOP. ELSE continue

No target values for QCs! QC results will be monitored and evaluated retrospectively (concentrations + ratios between QC results within 30% of mean).

-> focus on comparability of runs not on relative accuracy


Run

Date Kit Lot#

Curve

Number

QC Low

(ng eq/mL)

QC Mid

(ng eq/mL)

QC High

(ng eq/mL)

ratio

Low/Mid

ratio

High/Mid

30-Sep-13 #2 S_0003 2.22 3.43 19.9 0.65 5.8

2.15 3.62 19.5 0.59 5.4

30-Sep-13 #2 S_0004 2.27 3.3 16.5 0.69 5.0

2.21 3.37 18.2 0.66 5.4

30-Sep-13 #2 S_0005 1.91 3.38 19.7 0.57 5.8

1.72 3.53 18.2 0.49 5.2

01-Oct-13 #2 S_0006 2.16 3.67 18.6 0.59 5.1

2.3 3.9 20.3 0.59 5.2

01-Oct-13 #2 S_0007 2.27 3.92 20.6 0.58 5.3

2.18 4.12 18.9 0.53 4.6

04-Jun-14 #4 S_0008 2.2 4.11 20 0.54 4.9

2.19 3.73 21.3 0.59 5.7

04-Jun-14 #4 S_0009 2.66 4.26 24.7 0.62 5.8

2.18 4 22.4 0.55 5.6

mean 2.19 3.74 19.9 0.59 5.3

sd 0.21 0.32 2.00 0.054 0.38

CV 9.4% 8.4% 10.1% 9.2% 7.2%

mean -30% 1.53 2.62 13.9 0.41 3.7

mean +30% 2.84 4.86 25.9 0.76 6.9

Measurement of study samples Retrospective QC evaluation and run acceptance

1. Test of normal distribution

2. All batches fulfills +-30% acceptance criterion vs. mean of all study runs. All batches are considered comparable and valid. Red values = outliers > 3 * inter-quartile range


Results, Data Transfer and Reporting

Visit 2 = pre-dose NOS = no sample

1/

PTNO CPEVENT DILUTION RESULT TITER RESULT RATIO PKQ RUN

209 VISIT 2 500 10000 1 P735_0006

209 VISIT 6 500 0.98 P735_0006

209 VISIT 8 500 0.85 P735_0006

209 VISIT 10 500 NOS P735_0006

210 VISIT 2 50 1400 1 P735_0006

210 VISIT 6 50 0.72 P735_0006

210 VISIT 8 50 0.12 P735_0006

210 VISIT 10 50 NOS P735_0006

212 VISIT 2 30 600 1 P735_0006

212 VISIT 6 30 0.51 P735_0006

212 VISIT 8 30 0.1 P735_0006

212 VISIT 10 30 NOS P735_0006

213 VISIT 2 100 1400 1 P735_0006

213 VISIT 6 100 NOS P735_0006

213 VISIT 8 100 0.06 P735_0006

213 VISIT 10 100 0.05 P735_0006

214 VISIT 2 500 9900 1 P735_0006

214 VISIT 6 500 0.5 P735_0006

214 VISIT 8 500 0.05 P735_0006

214 VISIT 10 500 0.04 P735_0006


Questions 5

5. Would you agree with the level3 in-study validation concept in which runs are preliminary accepted based on calibration curve data only and QC results are compared retrospectively to identify outlier runs?


Evaluation of BM1 results

• Significant dose dependent effect

• BM1 correlates with physiological effect

• Hints for an outcome related biomarker (early prediction of efficacy)


Planning for phase II

• BM 1 confirmed as very promising physiological response biomarker and maybe outcome related biomarker

• Team decision to measure BM1 in all phase II studies also in other indications

Consequences for bioanalysis:

• Phase I approach is too much effort for the expected number of samples (Pre-screening/titer determination of pre-dose samples from each subject in order to define optimal dilution)

• Decision to invest in a complete new inhouse assay development using commercial antibodies

• MSD for a broader dynamic range

• Selection of antibodies and calibration standard to increase parallelism and enable a relative quantitative assay

• Serum only


Validation of the new MSD BM1 assay, level 2

Validation plan level 2 experiment

VCs (validation samples) Three individual sera (low, mid, high) aliquoted and stored frozen, additional LLOQ sample = higher diluted serum

Precision

At least 6 runs at 4 days, VCs in independent triplicates Mean values are target for in-study batch acceptance

Calibration range, LLOQ, ULOQ Tabulation of back-calculated results

Parallelism 6 individuals sera ( Healthy and patients)


N=24 sera


24h RT, +1, +3, +5 freeze/thaw cycles (-20°C) Three individual samples in triplicates (VCs)

Calibration sample stability Ready-to-use diluted, 2h RT

Stock solution stability, banked calibration standard

Thawed aliquot in the refrigerator, long-term monitoring at -70°C and -150°C

Long term stability serum Isochronic sample sets (stressed vs. -150°C), 3 individuals

Sampling robustness Fresh-blood stability, serum coagulation time, interference of hemolysed blood cells

Banking of sample control (monitoring samples)

Use for long term trends and lot bridging experiments


Questions 6-7

6. After first data available. Decision to develop a complete new assay: overkill or ok?

7. Has this been a typical case study or something special?


Questions for discussion

1. Was the available information about BM1 and its intended use sufficient? What was missing?

2. Was the intended use of BM1 data translated adequately into the bioanalytical strategy? (choose of assay, feasibility experiments, plans for future validation and analyzing strategy)

3. Has the assay been sufficiently proven/validated prior to start of sample measurements? What experiments are missing, not adequate.

4. Assay acceptance after validation. What about the consequences of the failed parallelism experiment? quasi-quantitative evaluation of data ok?

5. Would you agree with the level3 in-study validation concept in which runs are preliminary accepted based on calibration curve data only and QC results are compared retrospectively to identify outlier runs?

6. After first data available. Decision to develop a complete new assay: overkill or ok?

7. Has this been a typical case study or something special?


Documents

Integrating an exploratory BM in an early clinical stage ... · In collaboration with TransMed Experts and biomarker groups model-based analysis of efficacy or safety biomarker data