Case Studies Using the Singulex Erenna to Develop Sensitive Custom Biomarker Assays. Alison Joyce AAPS NBC 2015, San Francisco, CA

Transcription

1 Case Studies Using the Singulex Erenna to Develop Sensitive Custom Biomarker Assays Alison Joyce AAPS NBC 2015, San Francisco, CA

2 Outline Technology background Case Studies: -Assay 1: Bead-based homebrew assay in urine matrix -Assay 2: Plate-based homebrew assay in human serum -Assay 3: Bead-based homebrew assay in monkey serum Insights and tips along the way! 2

3 Singulex Erenna Immunoassay Platform 2-step process to achieve maximum sensitivity and minimum non-specific binding: Step 1: Execute a microparticle (MP)-based sandwich immunoassay Step 2: Elute the sample and utilize a unique single molecule counting technology capture Ab coated MP fluorescent-labeled detector antibody Elute fluor-labeled detector/ transfer to 384-well plate Plasma sample MP Analyte Count single molecules with Erenna System MP Incubate and wash MP Incubate and wash MP DE s EP s TP s STEP 1: Sandwich Immunoassay STEP 2: Single Molecule Counting 3

4 Attaining Ultrasensitive Detection Sensitivity range: fg/ml to low pg/ml Fun Fact: Erenna are jellyfish Ultrasensitivity is achieved by: 1) Bead-based binding -provides efficient analyte capture 2) Concentration of sample -small volume elution 3) Digital readout -single positive fluorescent events 4) Low background -low protein binding components (beads, plates) -Multiple wash and plate transfer steps -Luminous siphonophores -Both luminescent and autofluorescent 4

5 Case Study 1: Homebrew Biomarker Assay Format in Urine Matrix (Analyte A) Expected Concentrations: low sub pg/ml Urine can be very dilute; easy workflow desired Compare bead vs plate assay formats Assay Parameter Bead Plate Sensitivity Low-sub pg/ml 1-20 pg/ml Sample volume 200 ul 50 ul Assay time 6 hrs 4.5 hrs Workflow Difficult Easy Streptav bead Analyte A Anti-Analyte A*Alexa 647 Ms anti-analyte A-biotin ROQ = pg/ml In neat urine matrix Plate: 2 pg/ml LLOQ Bead: 0.36 pg/ml LLOQ ~10 fold sensitivity improvement DoE critical factors for sensitivity: 1) Ab:Bead ratio 2) Bead concentration in well 3) Detector concentration 5

6 Challenges with Urine Matrix Normalization Concentration of analyte varies since volume varies Creatinine frequently used; not perfect Composition Urine varies sample to sample ph, specific gravity, solutes May contain excreted dyes that fluoresce Clarification May have particulates or sedimentation Could require centrifugation or filtration Vendor recommends clarification Can result in loss of analyte 6

7 pg/ml Analyte A pg/ml Analyte A Clarification Effects: Filtration of Analyte A (Recombinant vs Endogenous) 50 Effect of Filtration on Spiked QC's Filtration Effects on Analtye A Measurement in urine QC1 QC2 QC3 QC4 QC5 unfiltered filtered unfiltered plate filtered 0.2 um filtered Sample prep Filtering spiked QC s results in ~ 25% loss of recovery of recombinant analyte Filtering unspiked urine matrix results in significant loss of signal Are we filtering out non-specific signal, or true endogenous Analyte A? 7

8 Immunodepletion of Analyte A from Urine Matrix pg/ml Analyte A Immunodepletion of Analyte A from Urine Samples Undepleted Depleted Pool Individual control spike* Sample LLOQ Immunodepletion of signal to LLOQ or near LLOQ levels using specific antibody pull-down demonstrates specificity of signal Urine cannot be clarified or we suffer analyte loss This will impact precision, so triplicates were run ~7% of data points needed to be deactivated due to %CV > 20% 8

9 % R e c o v e r y Selectivity in Urine Matrix and Reagent Stability S e le c tiv ity T e s tin g U L O Q (4 1 p g /m l) L L O Q (0.3 6 p g /m l) +/- 30% of nominal acceptance criteria In d iv id u a l U rin e D o n o r Acceptable recovery of ULOQ and LLOQ spikes in neat matrix (1:2 MRD) 5 Reagent Prep Storage time Lot 1 Beads Lot 2 Beads Lot 1 Detector Storage Temp Singulex kit 2 years 4 o C Commercial 6 months 4 o C Singulex kit 2 years 4 o C Lot 2 Detector In-house 2 months -80 o C 9

10 Patient Samples vs Controls Median = 0.9 Median = 0.5 Study samples: 47/100 (47%) were BLOQ (<0.36 pg/ml) Controls: 10/20 (50%) were BLOQ (<0.36 pg/ml) Urine concentration is variable and will affect the measured concentration of specific analyte in the matrix -Used creatinine to normalize analyte concentrations Censored data analysis shows medians are not statistically different (0.36 and 0.32 pg Analtye A/mg Cr for study samples and controls, respectively) 10

11 Conclusions for Case Study 1 Bead-based approach was chosen over plate-based due to sensitivity requirements in sub pg/ml range Urine is a challenging matrix; varies in concentration and may contain lots of particulate matter Filtration of samples was not feasible due to loss of specific analyte Triplicates were run to control for any precision issues Selectivity showed sample differences did not affect recovery of spiked analyte Good reagent stability was demonstrated for Analyte A assay Up to 2 years at 4oC for both capture and detector antibodies Study support showed study analyte A levels were not statistically elevated compared to control levels Used standard censored data analysis techniques since ~50% of the samples scored BLOQ 11

12 D E 's O D n m Case Study 2: Plate-based Erenna vs ELISA Homebrew Assay (Analyte B) Expected normal donor levels for Analyte B: ~100 pg/ml Diseased individuals expected to have reduced concentrations Collaborator lab only has ELISA capabilities S ta n d a rd C u rv e C o m p a ris o n E L IS A S in g u le x Diluted normal individuals fall here 1 0 Mouse anti-analyte B -Alexa 647 Rat-anti-Analyte B Lots of dynamic range available [A n a ly te B p g /m L ] Diluted normal individuals fall here 0.1 ~10-fold sensitivity improvement ELISA LLOQ = 90 pg/ml Singulex LLOQ = 10 pg/ml 12

13 ELISA vs Singulex Erenna Assay Workflow ELISA: LLOQ = ~90 pg/ml O/N 4oC 100 ul/well 1 hr 3 hr 200 ul/well 1 hr 100 ul/well 30 mins 100 ul/well 30 sec Singulex Erenna: LLOQ = 10 pg/ml 384-well plate O/N 4oC 50 ul/well 1 hr 2 hr 50 ul/well 1 hr 50 ul/well 30 mins 50 ul/well 5 mins 40uL/10uL 1.5 hrs Erenna plate-based has similar workflow to ELISA Erenna has reduced sample volume requirements Flexibility to freeze and store eluted/neutralized plate Freeze plate option 13

14 [A n a ly te B p g /m l] Endogenous Levels of Analyte B in Normal Serum Donors: ELISA vs Singulex A n a ly te B S e r u m L e v e ls E L IS A v s S in g u le x Mean = 130 pg/ml Mean = 189 pg/ml S in g u le x (MRD 1:8) E L IS A (MRD 1:2) Serum spike Recovery % 60-70% (MRD = minimum required dilution) E L IS A S in g u le x Singulex measured levels in ALL samples tested (64 total), whereas 65% of samples were BLOQ in the ELISA Measured concentration of endogenous Analyte B are statistically different in ELISA vs Singulex (130 vs 189 pg/ml, respectively). -Difference most likely due to low recovery in ELISA at 1:2 MRD 14

15 Case Study 2 Conclusions 10-fold improvement in sensitivity using Erenna vs ELISA Allows accurate baseline measurement (well above LLOQ) Allows meaningful measurement of decreased Analyte B levels in diseased individuals Potential to differentiate between normal and diseased samples, and severity of disease Singulex assay had measureable endogenous Analyte B in 100% of serum samples tested (64/64); ELISA had 65% samples score BLOQ Decreased sample volume required 12.5 ul vs 200 ul (Singulex vs ELISA, respectively) Conserves valuable sample Flexibility to freeze/store eluted/neutralized plate at -80oC and read at a later time, ship to different site, etc. Workflow and cost are essentially the same as ELISA for homebrew assay 15

16 Case Study 3: Bead-Based Homebrew Assay Format for Analyte C Anti-Drug Ab*Alexa 647 Forced complex off-line with excess drug Drug Analyte C Streptavidin Bead Anti-Analyte C biotin (commercial) Assay measures total (free and drug-bound) Analyte C Initial Range of Quantitation: pg/ml in neat monkey serum (1:4 MRD) 16

17 p g /m l A n a ly te C Parallelism for Analyte C Assay P a r a lle lis m A n a ly s is o f S e r u m S a m p le s S a m p le 1 S a m p le 2 S a m p le 3 S a m p le 4 S a m p le D ilu tio n F a c to r 1:4 initial dilution, some samples exhibited non-parallelism as they were diluted out Increase MRD to 1:20 to overcome matrix effects in most samples Impact: LLOQ increased to 7 pg/ml in neat serum -Since this is a total assay, dosed samples will show significant increase in Analyte C levels (target engagement) -fg/ml sensitivity not required, however baseline levels may not be measureable 17

18 D E 's D E 's Capture Reagent Issue with Analyte C Assay B e a d L o t C o m p a r is o n (C o m m e r c ia l) L o t 1 B e a d s L o t 2 B e a d s B e a d L o t C o m p a r is o n (C o m m e r c ia l v s in -h o u s e ) L o t 1 : v e n d o r, u n k n o w n M C R L o t 2 : v e n d o r, u n k n o w n M C R L o t 3 : in -h o u s e, 5 :1 M C R L o t 4 : in -h o u s e, 2 5 :1 M C R * * p g /m L A n a ly te C p g /m L A n a ly te C In-house labeled 25:1 MCR performed similarly to vendor Lot 1 Lot to lot differences in commercial biotinylated antibody -This impacted sensitivity Successfully biotinylate in-house -Different MCR s mattered -25:1 condition worked best 18

19 Case Study 3 Conclusions Parallelism should be tested on samples with measureable endogenous analyte levels Matrix interference may dictate a higher MRD, thus affecting sensitivity Commercially available biotinylated capture antibodies may be used May vary lot-to-lot Can affect assay performance, particularly sensitivity Different MCR s for in-house biotinylations may impact assay performance 19

20 Acknowledgements Thank You! Yiqun Zhang Sheldon Leung 20