Computational Approaches to Assessing Safety of Host Cell Proteins and ‘Impurities’Or – how sequence-based information is integral toThe assessment of monoclonal antibodies and their ‘impurities‘
• PDA Europe • Workshop on Monoclonal Antibodies• Rome, 20-21 September 2016
• Annie De Groot MD• EpiVax, Inc.
2
Disclaimer – I am CEO/CSO of EpiVaxThus there is an inherent COI in this presentation.I will do my best to present the work in an unbiased manner.
3
Sequence-based Assessment of Monoclonal Development Risk
Pre-ClinicalAssessmentofImmunogenicitybasedonsequenceComparisontoexistingmonoclonalantibodydata(andimmunogenicitydata)Retrospectiveandprospectivevalidation(published)Currentlyinuseby9outof12largepharmaandmany,manysmallerbiotechcompanies
Applicationstobiologics/biosimilars:• Sequence-basedcharacterizationofrisk• Characterizationofimpurities(modifiedsequences– riskofimmunogenicity?)• CHOandotherHCPimmunogenicityriskassessment
4
Sequence-based Assessment of Monoclonal Development Risk
Examples:
• ACHO(HCP)Proteinthatco-purifieswiththebiologic.Isitpotentiallyaproblem?
• Sequencesthatvary– e.g.peptidethathasslightlyvariantaminoacidsas‘impurities’
• Newmonoclonalantibodyderivativeswithunusualcomposition(Fabs,Dabsetc).
• Novelscaffoldproteins(alsobasedonhumanproteinsequences).
• Others?
5
Computational Analysis as part of a Continuum (and part of ‘Quality by Design‘)
Non-HumanInVivo
InVitro&ExVivoInSilico
Modifysequencestoreduce
immunogenicity
Screenmultipletherapeuticcandidates
HumanSCIDmousemodel
HLA-transgenicmousemodel
Either
Examine:• Characteristics
ofTcellandAPCresponse
• EffectofHLAdonordiversity
• Effectofdonorpathologies
• Effectofformulationandpost-translationalmodifications
Rankforimmunogenicity
basedonpredictedepitopecontent
ProceedtoDrug
Development
Peptideandprotein
screening:EffectonT
cellpopulations
MemoryandnaïveTcells
HLAbindingstudies
Cell-basedassays
Wholeprotein
screening:ProcessingbypurifiedDCs
http://bit.ly/The_TCWP
6
In vitro validation
Test articles:whole protein,
domains, peptides
• FluoroSpot (dual cytokines)• Flow cytometry• Peptide elution (in development)
Readout format
Assay(cell culture)
• Cytokine production• Cell proliferation• Cell phenotype – Teff vs. Treg
Endpoints
• Exposed blood• Fresh (active cells)• Re-stimulation (memory)
• Naïve blood• Long term culture (IVIP)
http://bit.ly/EpiLab
7
ComputationalTools- Background
T Cell Activation B Cell ActivationWith T Cell Help – Drive Ab Response
NO T Cell Activation NO B Cell Activation
Without T Cell Help – Lose Ab Response
EpiVax
10
Biologics, like Foreign Proteins, are processed by Immune system
EpiVax
11
Biologics, like Foreign Proteins, are processed by Immune system
• T cell epitopes are drivers of immunogenicity• Sequence analysis enables T cell epitope identification• EpiVax uses EpiMatrix to predict epitopes
–matrix based prediction algorithm
–Full suite of HLA-based predictions are available; Class II usually used for biologics.
Mature APC
MHC II Pocket
Epitope
Protein
12
Epitope Content Drives Immunogenicity
EpiVax - confidential 12
- 80 -
- 70 -
- 60 -
- 50 -
- 40 -
- 30 -
- 20 -
- 10 -
- 00 -
- -10 -
- -20 -
- -30 -
- -40 -
- -50 -
- -60 -
- -70 -
- -80 -
ThrombopoietinErythropoietin
AlbuminIgG Fc Region
GM-CSF
Follitropin-Beta
Beta-2-Microglobulin
Interferon-Beta
Human Growth HormoneTetanus Toxin
Influenza Hemagglutinin
Immunogenic Antibodies*
Non-Immunogenic Antibodies†
Your Candidate A
Your Candidate B
* Average of antibodies known to induce anti-therapeutic responses in more than 5% of patients
† Average of antibodies known to induce anti-therapeutic responses in less than 5% of patients
All scores are adjusted for the presence of Tregitopes.
EpiMatrix Predicted Excess/Shortfall in Aggregate Immunogenicity Relative to a Random Peptide Standard
13
Two types of T cell epitopes can be identified
Epitope can be either effector or regulatory but most algorithms cannot differentiate
this in silico
T regT eff
14
Tregitopes = Treg epitopes found in Monoclonal Antibodies
• Discovered & patented by EpiVax
• Highly conserved peptide sequences in Fc and Fab regions of antibodies
• High affinity, promiscuous binders across HLA alleles
• Activate antigen-specific regulatory T cells
• Can be co-formulated or synthesized with therapeutic proteins or carriers
De Groot A.S., et al., Activation of Natural Regulatory T cells by IgG Fc-derived Peptide “Tregitopes”. Blood, 2008,112: 3303. http://tinyurl.com/ASDeGroot-Blood-2008 http://bit.ly/Treg1
16
EpiMatrix Protein Detail ReportProvides detailed map of epitope content
Frame Frame DRB1*0101 DRB1*0301 DRB1*0401 DRB1*0701 DRB1*0801 DRB1*1101 DRB1*1301 DRB1*1501Start Stop Z-Score Z-Score Z-Score Z-Score Z-Score Z-Score Z-Score Z-Score
1 DIQMTQSPS 9 0.91 0.17 0.35 0.06 0.09 0.86 -0.08 -0.01 02 IQMTQSPSS 10 2.15 1.35 2.26 1.58 2.17 1.88 2.24 2.52 63 QMTQSPSSL 11 1.11 0.28 0.5 0.96 0.29 0.39 0.9 0.4 04 MTQSPSSLS 12 1.87 2.26 2.02 1.88 1.06 1.66 1.79 2.05 75 TQSPSSLSA 13 0.51 0.28 1.31 0.69 -0.04 0.54 1.09 1.05 0. . . . . . . . . . . .. . . . . . . . . . . .
46 LLIYAASTL 54 1.41 1.08 0.99 2.16 1.54 0.85 2.28 2.3 347 LIYFASTLQ 55 1.91 1.85 2.46 2.07 1.79 1.08 1.79 2.1 748 IYFASTLQS 56 1.73 2.01 2.68 1.26 1.58 1.82 2.11 2.07 649 YFASTLQSG 57 0.64 1.6 0.59 0.32 1.42 1.28 0.04 1.09 0. . . . . . . . . . . .. . . . . . . . . . . .
98 FGQGKTVEI 106 0.7 1.9 -0.02 0.2 0.93 0.34 1.41 1.12 199 GQGKTVEIK 107 -0.28 -0.84 0.31 0.01 0.03 -0.57 -0.47 -1.24 0
DRB1*0101 DRB1*0301 DRB1*0401 DRB1*0701 DRB1*0801 DRB1*1101 DRB1*1301 DRB1*1501 Total2.4 2.26 2.8 3.11 2.41 2.29 2.28 2.52 --
23.41 12.17 15.96 18.16 10.2 13.2 16.47 16.99 126.5612 6 7 9 5 7 8 8 62
Deviation from Expectation: 38.46 Deviation per 1000 AA: 48.56Adjusted for Regulatory Epitopes Deviation from Expectation: -21.24 Deviation per 1000 AA: -26.82
Total Assessments Performed: 792
EpiMatrix ReportAccession: YOUR_PROTEIN - Sequence: YOUR_PROTEIN
Maximum Single Z scoreSum of Significant Z scores
Count of Significant Z Scores
AA Sequence Hits
Summarized Results
Assessment
HitEpiBar
EpiMatrix Immunogenicity Score EpiMatrix Tregitope-adjusted Score
17
EpiMatrix Predicted ADA Responses Relative to Observed ADA Responses in Known Antibodies
Slide TitleN
eo E
pito
pe C
onte
nt
Tregitope ContentHigh Low
Low Avastin (0%)
Herceptin (0%)
Mylotarg (3%)Simulect (1%)Synagis (1%)
Hig
h
Campath (45%) Remicade (26%)Rituxan (27%)
All scores are adjusted for the presence of Tregitopes.
mAb Trade Name (ADA%)
Your Antibody
18
High Throughput
http://www.epivax.com/immunogenicity-screening/ht_report/
The High-Throughput Antibody Analysis tool allows users to compare a large number antibody candidates by ranking the immunogenicity of heavy and light chains.
EpiVax - confidential
19
Tregitopes – in FDA guidelines
• FDA guidelines mention the importance of differentiating effector and regulatory epitope sequences
• Tregitope-conscious analysis results in more accurate analysis and the opportunity to introduce immunosuppressive sequences in the modified protein
20
R² = 0.17
-5%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
-40 -20 0 20 40 60 80
Obs
erve
d Im
mun
ogen
icity
(% A
DA)
Combined Variable Heavy and Light Chain EpiMatrix Scores
WITHOUT Tregitope-Adjusted Scores to Predict Immunogenicity:AVASTIN
BIVATUZUMABCAMPATH
HERCEPTIN
HUMICADE
HUJ591
HUMIRA
LEUKARRESTLUCENTIS
MYLOTARG
RAPTIVA
REMICADE
REOPRO
RITUXAN
SIMULECT
SOLIRIS
SYNAGIS
Antibody Immunogenicity PredictionsWithout Tregitope Adjustment
http://bit.ly/Treg1
21
Slide Title
EpiVax - confidential 19
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
-70 -60 -50 -40 -30 -20 -10 0 10 20 30
Obs
erve
d Im
mun
ogen
icity
(% A
DA)
Combined Variable Heavy and Light Chain EpiMatrix Scores
Using Tregitope-Adjusted Scores to Predict Immunogenicity: AVASTINBIVATUZUMABCAMPATHHERCEPTINHUMICADEHUJ591HUMIRALEUKARRESTLUCENTISMYLOTARGRAPTIVAREMICADEREOPRORITUXANSIMULECTSOLIRISSYNAGISTYSABRIVECTIBIXVISILIZUMABXOLAIRZENAPAX
Failed in trials?
High ADA RiskFirst-Gen Chimerics
Low ADA Risk
MediumADA Risk
y = 0.0086x2 + 0.6727x + 13.32R2 = .76
http://bit.ly/Treg1
22
New – Patent-derived ‘Failed in Trials‘ and pre-clinical Data
23
Treg
sFu
nctio
ning
Non
-func
tioni
ng T
reg
DangerZone– ifTregs notfunctioning– ADA
HLA-specific Tregitope Mapping and Autoimmunity Twist
24
ToolsforAnalyzingtheConservationbetweenBiologicandSelf
Dr. Pini: “Differences between sequences can drive immunogenicity” . . . and conservation with self can diminish immunogenicity
25MHC/HLA
TCR
• Identical T cell-facing residues• Same HLA allele and minimally different
MHC-facing residues
JanusMatrix
Moise, et al. The Two-Faced T cell Epitope: Examining the Host-Microbe Interface with JanusMatrix. Human Vaccines and Immunother. 2013 Apr 12;9(7). http://bit.ly/JanusMatrix.
EpiVax
26
Peptide from Biologic
9-mers that bind HLAHuman Protein where
epitope is found
Visualizing T cell Epitope Relatedness to Human
9-mers with same TCR faceThat bind to same HLA
from Human Genome
Moise L, Beseme S, Tassone R, Liu R, Kibria F, Terry F, Martin W, De Groot AS. T cell epitope redundancy: cross-conservation of the TCR face between pathogens and self and its implications for vaccines and autoimmunity. Expert Rev Vaccines. 2016 May;15(5):607-17.
27
Pathogens have figured this out
Source 9-mer epitope
Human protein with cross-conserved epitopes
Cross-conservedhuman 9-mer epitope
Source (pathogen) protein
27EpiVax - confidential
HCVVaccineepitopesexampleAllInduceTeffectorresponse
HCVepitopeInducedTregresponse
LosikoffPT,MishraS,TerryF,GutierrezA,Ardito MT,FastL,Nevola M,MartinWD,Bailey-KelloggC,DeGrootAS,GregorySH.HCVEpitope,HomologoustoMultipleHumanProteinSequences,InducesaRegulatoryTCellResponseinInfectedPatients. JHepatol.2014Aug22.pii:S0168-8278(14)00613-8.doi:10.1016/j.jhep.2014.08.026.
28
Tregitope JanusMatrix using Cytoscape
HTREG_IGGC-289
HTREG_IGGC-167
hTregitope-167 hTregitope-289
29
Other Examples of Treg epitopes – in Pathogens
HCV peptide HIV peptide
Source 9-mer epitope
Source human protein
TCR-conserved human 9-mer epitope
Source peptide
329/20/16
Moise L, Terry F, Gutierrez AH, Tassone R, Losikoff P, Gregory SH, Martin WD, De Groot AS. Smartervaccine design will circumvent regulatory T cell-mediated evasion in chronic HIV and HCV infection. In:Why vaccines to HIV, HCV and Malaria have so far failed - challenges to developing vaccines againstimmunoregulating pathogens. Frontiers in Microbiology. 2014. Editor (Gowans).http://bit.ly/Smarter_Vaccines_2014. (Open Access)
JanusMatrix Publications
309/20/16
31
ToolsforAnalyzingtheConservationbetweenHCPandSelf:CHOPPI
32
EpitopeContentMatters:ConservationwithHumanEpitopes
What is the impact of the cross-conservation between CHO and Human?
Human CHO
41EpiVax - confidential
33
CHOPPI –Original Publication
Biotechnology and Bioengineering (2014)
Submit an amino acid sequence, name, or accession number into the search box
Review epitope information:• EpiMatrix Score• Total epitopes• Human cross-conserved• CHO-unique
Review results for homology with: • CHO transcriptome• CHO proteome• mouse secretome• validated CHOPs• predicted signal peptide
34
Peptide from HCP
9-mers that bind HLAHuman Protein where
epitope is found
Cross-conservation analysisBetween HCP and Human Genome
9-mers with same TCR faceThat bind to same HLA
from Human Genome
36
Ranking HCP by “HCP-specific epitope content“
HypotheticalHCP
The input HCP sequence has an above-averageCHO-unique epitope content compared to theremainder of the CHO K1 genome.
Of the 77 9-mer peptides which scored positivelyagainst at least one HLA allele, we found:
• 18 to be exactly preserved within the human genome.• 56 sharing TCR contacts with putative epitopes derived
from the human genome.• 21 sharing some homology with the human genome,
however the sequences with which they can bematched are not predicted to be epitopes, thus reducingthe potential for pre-exposed human T cell populations.These are the “CHO-Unique” epitopes.
37
JanusMatrix for CHO Protein Prediction
Lysosomal protective protein:potentially immunogenic due to
dispersed CHO-unique epitopes
C-X-C motif chemokine 3:potentially tolerated due tofew CHO-unique epitopes
Lysosomal alpha-mannosidase:cluster of CHO-unique epitopes àpotentially immunogenic region
37
Cross-conserved 9-mer epitope
Human protein with cross-conserved epitopes
Cross-conserved human 9-mer epitope
Source (pathogen) protein
CHO-unique 9-mer epitope
38
CHOPPI ResultsInteractive Website Output
≥20:potentiallyimmunogenic
EpiVax - confidential
39
Recent CHOPPI Application
39EpiVax - confidential
BioProcess International(Vol.13Issue4)
Report that the same HCP — phospholipase B-like 2 (PLBL2) — copurifies withmultiple Chinese hamster ovary (CHO)-produced antibody therapeutics.
40
Recent CHOPPI Application:Results
40EpiVax - confidential
Recent CHOPPI Application:Epitopes unique to CHO
42EpiVax - confidential
Cross-conserved 9-mer epitope
Human protein with cross-conserved epitopes
Cross-conserved human 9-mer epitope
Cross-conserved cluster
CHO-unique 9-mer epitope
Source protein
PLBL2:putative T cell epitopes
unique to CHO
43
Recent CHOPPI Application:Conclusions
39EpiVax - confidential
• PLBL2 has an elevated immunogenicity score of 32.89, with 30epitopes that are unique to CHO– 5 of them are EpiBars– 20 are within Clusters
• PLBL2 is potentially immunogenic – but activity may be due to otherfeatures (potential to cause aggregation of IgG)
• Author comments:– “We have no reason to think that our observations are unique to Genentech CHO
cells or the Genentech production process. It is possible that PLBL2 is afrequent impurity in CHO-derived MAb biotherapeutics across the industry. It isalso possible that widely used commercial CHOP ELISAs do not detect thisimpurity, or they could significantly underreport its presence.”
– “We have instituted process changes to reduce PLBL2 levels in all our MAbproducts. As illustrated herein for one product, it is possible to reduce PLBL2 tolevels near or <1 ng/mg. We reduced patient risk with a combination of asensitive assay for PLBL2 and implementation of PLBL2-reducing processchanges.”
44
Recent CHOPPI Application:Jawa V. et al. AAPS J 2016
44
EvaluatingImmunogenicityRiskDuetoHostCellProteinImpuritiesinAntibody-BasedBiotherapeutics.Jawa Vetal.AAPSJ2016.SupplementalFigure1
PLBL2corroboratedinsecondlaboratory,amongotherHCPs
45
Sequence-based Assessment of Monoclonal Development Risk
Pre-ClinicalAssessmentofImmunogenicitybasedonsequenceComparisontoexistingmonoclonalantibodydata(andimmunogenicitydata)Retrospectiveandprospectivevalidation(published)
Applicationstopreparationofdossiersforregulatoryreview:• Sequencebasedcharacterizationofrisk• Characterizationofimpurities(modifiedsequences?)• CHOandotherHCPimmunogenicityriskassessment
46
ISPRI and CHOPPI Approach
A step-wise process for analyzing and reporting on immunogenicity risk
EpitopeMapping
ClusterAnalysis
ARB
NN-align
SMMalign
ReengineeringRankingMapping
EpiMatrix
ClustiMer /EpiBar
OptiMatrix
iTEM
ImmunogenicityScale
DeFT (OptiMatrix)
InVitro/InvivoAssays
JanusMatrix
JMX-AdjustedScore
TargetSelection
Tregitope Tolerization
Do-it-yourself approach
• Andres Gutierrez• Bill Martin• Frances Terry• Matt Ardito• Guilhem Richard
• Lenny Moise• Eduardo Guillen• Sandra Lelias• Christine Boyle
Acknowledgements
Thank you !
Science without fear.
www.epivax.com • [email protected] • 401.272.2123