CONFIDENTIAL

Quantitation of a Therapeutic Insulin Analogue by Immuno-Affinity Extraction LC-MS/MS

(or running a real peptide project with a hybrid assay: questions and some solutions?)

Michael BlackburnCovance Laboratories Ltd

Alnwick, UK

5th EBF Open Symposium14th November 2012

Barcelona, Spain

CONFIDENTIAL

Acknowledgements

• Martin Gerl and his team, Sanofi, Frankfurt : Scientific leadership of this project, supply of IA-gel.

• Dietmar Schmidt, Ronald Schmidt and team, Sanofi, Frankfurt: Dialogue, expertise with insulins and training in Immuno-Affinity extraction.

• Isotope Chemistry Group, Sanofi, Frankfurt: Production of iodated and stable labelled internal standards

• Barbara Bell, Nick Gray and Jennifer Mackie, Alnwick: Method development, sample analysis.

• Stuart McDougall: leadership of the Alnwick team.

CONFIDENTIAL

Insulin Analogues• Glargine, glulisine, lispro, aspart etc, are modified insulins, also known as insulin

analogues (IAs).• IAs have a modified isoelectric point (pI) which reduces solubility at physiological pH.

• Changing pI changes adsorption into body: get fast & slow-acting insulins

Longer Lasting Insulin Development Compound: LoLA

• Molecular Weight 6.2 kDa approximately: modified at A & B chain• Requirement for a reliable LC-MS/MS assay in animal & human plasma • LLOQ requirement of 200 pg/mL in plasma or better

Structure of Human Insulin: 51 amino acids, a large peptide

G I V E Q C C T S I C S L Y Q L E N Y C N

F V N Q H L C G S H L V E A L Y L V C G E R G F F Y T P K T

S S

S

S

S

S

NH2-

NH2-

-COOH

-COOH

CONFIDENTIAL

LoLA Q1 Mass Spectra & Manipulation of Charge StateSAR161271_full_scan #21-33 RT: 0.38-0.61 AV: 13 NL: 2.62E6T: + p ESI Q1MS [ 750.000-1400.000]

800 900 1000 1100 1200 1300 1400m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e A

bund

ance

782.85

1043.56

894.80 1252.16

900.20

905.42883.911002.52871.31 962.29 1202.84 1290.491154.431095.48 1339.36

6+5+7+

8+

20mM ammonium formate/acetonitrile LoLA charge distribution envelope: Methanol/formic acid/ipa 10%

5+

6+

7+

6+

5+

4+•Charge state envelope can be to some extent controlled

using mobile phase composition

•Some difference observed between different source typesWaters Z spray, Thermo, also APi5000 at Frankfurt

Waters Z spray, Quattro Ultima

Thermo Quantum Ultra

CONFIDENTIAL

Direct Assay (physico-chemical assay) Approaches

• A combination of solid phase extraction of plasma (1mL) combined with LC/MS/MS• MS/MS transition chosen depended on cluster chosen for fragmentation• Initially used 1252 > 219• Extraction by Waters Oasis HLB• Limit of quantitation using a direct assay was 500-1000 pg/mL LoLA from 1mL plasma• For all assays used the addition of a surfactant to stop binding of standard solutions

RT: 2.94 - 7.00

3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5Time (min)

0

10

20

30

40

50

60

70

80

90

1000

10

20

30

40

50

60

70

80

90

100

Rel

ativ

e A

bund

ance

500pg/mL Lantus in rat plasma LoLA ISTD

RT: 5.32

6.06

6.305.40 6.346.01 6.383.08

5.56 5.915.103.26 4.493.73 4.434.01 4.823.576.976.61

RT: 4.99

5.224.56 6.375.42 6.276.063.49 4.463.953.30 3.76 5.84 6.976.71

NL: 2.10E2TIC F: + c ESI SRM ms2 1213.400@cid53.00 [174.850-175.150] MS 30March09_013

NL: 2.80E3TIC F: + c ESI SRM ms2 1252.200@cid60.00 [218.950-219.250] MS 30March09_013

Direct physico-chemical assay did not givethe required sensitivity

Instrumentation: TSQ Quantum Ultra

Insulin analogue spiked into rat plasma & extracted

CONFIDENTIAL

Sensitivity Enhancement 1: Use of FAIMS-pSRM

Useful for clusters which fragment poorly

In this example fragmentation is relatively poor; at higher collision energy the molecule ‘unzips’

i.e. produces lots of low intensity fragments

Can use the full signal from the precursor ion for best sensitivity and improve selectivity using an alternativetechnique: Ion mobility spectrometry

Combine a specific, high molecular weight Cluster with ion mobility selection(with some collision energy)

and ‘pseudo’ SRM = high energy Q3 SIM

Or can use a 0 constant neutral loss experiment(0 CNL). Gives an alternative specificity for poorlyfragmenting peptides

CONFIDENTIAL

Enhancement 2: Immuno-affinity Extraction

Cleaner extract than SPE

Columns can be re-used: >20 x

Expensive, a little slower

Throughput = 1 plate/analyst per day

Immunoaffinity image: services.leatherheadfood.com

Procedure adapted to use

multi-channel pipettes to

enhance throughput

CONFIDENTIAL

SPE vs IAC (or small & large)

SPE• Commercially prepared columns• Used once• Can be automated

Procedure• Condition column solvent• Equilibrate column aqueous• Load sample• Apply wash (aqueous solvent)• Elute Compound • Discard column/plate

IAC• Columns prepared in the lab•Columns are re-used•Probably best to run manually

but can use multi-channel pipettesProcedure

•Wash column with elution buffer•Wash column with buffer•Slowly apply sample to column•Wash with buffer•Elute with elution buffer (acid/ethanol)•Wash & prepare for storage

Very Good Tip:

Don’t put IAC gels in the freezerWe used Oasis HLB for initial

method development in a vacuum manifold.

CONFIDENTIAL

Result: Hybrid Immunocapture – LC-MS Approaches

• IAC extracts were cleaner, with absolute recovery approximately 80-90%• Limit of quantitation using immuno-capture assay was 200 pg/mL LoLA from

250uL spiked plasma. Immuno-capture gives a more sensitive assay.• This was true whether using FAIMS-pSRM on a Thermo system or MS/MS

on the API5000 system. • Enzyme digestion (Edman) was investigated as a back-up: did not improve

signal vs ‘top down’ approachRT: 3.03 - 5.99 SM: 5B

3.5 4.0 4.5 5.0 5.5Time (min)

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e A

bund

ance

100pg/mL plasma IAC 60uL Laminar flow

RT: 5.29MA: 2248

5.75

5.66

5.50

3.204.89 5.15

5.073.69 4.71

4.34 4.644.23 4.583.553.40

4.124.06

NL:5.43E2TIC MS 15april09_007

100pg/mL LoLA from spiked plasma

Decision: ‘ Hybrid’ IA- LC-MS approach selected for longer lasting insulin project

CONFIDENTIAL

Internal Standards

• Initial method development used an analogue ISTD (e.g. bovine insulin, insulin glargine)

• For early studies used iodated compoundThis ISTD performed well and we did not see evidence of loss of iodine and conversion to parent compound during the period of use. Recommend regular checking of this.

• GLP Studies used 15N isotopically labelled compound: all nitrogens labelled

• Concentration ISTD 2ng/mL (assay range:0.2 – 5 ng/mL)

CONFIDENTIAL

FAIMS method LLOQ & LC conditions

LoLA

I-LoLA ISTD

LLOQ in animal plasma (0.2 ng/mL)

TSQ Vantage with Turboflow system & FAIMS unit

•Load 100uL of extract onto an Oasis on line extraction column HLB 2.1 x 20mm •Elute onto Hichrom Ace 3 C18, 50 x 2.1 mm, 3 μm•Mob Phase A: 20mM ammonium formate pH 3•Mob Phase B: acetonitrile/0.1% formic acid flow rate: 0.4 mL/min•Sample volume 250 uL; injection volume 100 uL

NOTE: 0.2 ng/mL is still a rather high LLOQ for this work. Some low dose levels were not properly covered.The ideal LLOQ was 100 pg/mL or lower

CONFIDENTIAL

Correlation of MS with LBA Assay• IAC-LC-FAIMS-MS vs ELISA values for 16 samples from discovery

study (discovery LC-MS assay)

R2 = 0.9178

0

1000

2000

3000

4000

5000

6000

0 1000 2000 3000 4000 5000 6000

ELISA

"LC-MS"

Both x & y axes are LoLA concentration pg/mL

CONFIDENTIAL

Acceptance Criteria

•As this was a hybrid assay, method validated using 20/25 ‘biologicals’ criteriaDecision based on assay complexity (immuno-capture extraction)Also we are at the limits of instrument performance due to the size of analyte

•Same criteria used at Frankfurt. These were inspected and accepted by FDA•Analytical Range 0.2 – 5ng/mL

LoLA concentrations rarely higher than 2 ng/mL •As part of method development/validation we checked analytical cross-over between

LoLA and other insulin analogues

IA-LC-FAIMS pSRM method was used for discovery and GLP pre-clinical studies.

Need to move to a clinical assay. Some concerns about matrix variability in human plasmawith pSRM method.

Precision and Accuracy Data for first GLP Study (IAE-LC-FAIMS-pSRM).

Nominal concentration (ng/mL)

Number of samples Mean calculated concentration (ng/mL)

Mean % Difference Total%CV

0.2 18 0.198 -1.06 22.2

0.5 18 0.512 2.30 18.8

2 18 2.08 3.83 9.06

5 18 4.81 -3.79 8.37

CONFIDENTIAL

Product Ion Mass Spectra for long lasting insulin

SAR161271_dau_spec #108-191 RT: 3.42-6.02 AV: 84 NL: 8.87E4T: + p ESI Full ms2 1043.600@-30.00 [ 100.070-1050.000]

200 300 400 500 600 700 800 900 1000m/z

0

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

100

Rel

ativ

e A

bund

ance

219.1

120.1

1043.8

328.9

251.3 639.5

999.1573.7201.1 362.0

555.0293.3 476.3 983.4

379.9 656.2624.3757.2426.2 527.6 901.4862.9681.7 949.3185.3

B (a2)

Non-specific immonium ions should be avoided where possible: eg. 120 PheSpecific ions from B chain a & b series are available for insulins (Thevis & Schanzer, 2007).Lower charged clusters eg 1252 5+ did not fragment so well, 6+ have higher internal energy and fragment more readily. Need to compromise specificity vs ease of fragmentation.

B chain fragmentations Reference:. Thevis and Schanzer Analyst 2007, 132, p287-291

[M+6H+]6+

CONFIDENTIAL

Clinical Studies: LC-MS/MS Method

•Changed mobile phase to boost ratio of 6+ charge state & increase signal

•Used diagnostic 219 product ion(possible because of clean IAC extract)

•15N labelled internal standard available

drug

ISTD

LLOQ in plasma (0.2 ng/mL)

Mob. Phase B:Methanol/Iso-propanol/formic acid 0.5%Mobile Phase A: 0.5% Formic acidColumn Thermo Hypersil Gold 1.9um 20 x 2.1mmAgilent 1200, Stainless steel LC tubingSemi-UPLC method: Pressure approx 5000psi/360barFlow rate 0.3 mL/min500 uL sample volume; 50 uL injection volume

NOTE: 0.2 ng/mL is still a rather high LLOQ for this workSome low dose levels were not properly covered,particularly moving into clinical studies.Ideally LLOQ should be 100 pg/mL or lower

CONFIDENTIAL

Anti-Drug Antibody Effects: Repeat Dose Study

• Immunogenicity - Any biological therapeutic agent may have an unwanted immune response. This can range from having no effect to a severe hypersensitivity reaction. Immunogenicity can also result in neutralisation of the drug ie no clinical benefit, or at least a reduction in response and a change in pharmacokinetics.

• Immune response must be monitored when there is more than 7 days of drug exposure. All biotherapeutics need to be assessed for immunogenicity by measuring ADa – anti drug antibodies.

• Differences in insulins across species may well produce ADaeffects

• LoLA: Repeat dose 28 day preclinical studyDay One Samples: variability of internal standard around 10-30%, recovery highDay 28 samples internal standard recovery was low (< LLOQ of LoLA) and highly variable Evidence of an ADa effect, antibodies were also measured directly

CONFIDENTIAL

Effect of ADa in multi-dose studies

• Antibodies to LoLA formed from 2-3 weeks exposure depending on dose

• These bind to drug & internal standard (iodated LoLA in pre-clinical studies)

• Equilibrium between free (active) and total insulins• Request for a ‘total’ method for bound & unbound insulin• Ab complex is dissociated by acidification with HCl• Leave 1 hour followed by neutralisation and then IAC• Total method has an LLOQ of 1 ng/mL, HLOQ 100 ng/mL• Pre-clinical multi-dose studies used ‘Total’ method

‘free’ LoLA could be assessed by filtration or addition of excess insulin to bind AD-antibodies

CONFIDENTIAL

Metabolite Assays

• Much of insulin activity can be attributed to metabolites of the compound

• Peptides have ‘catabolism’ not metabolismStep-wise loss of amino acids from B chain preserves insulin type activity

example:

• IAC antibody extracts based on insulin epitope on the B chain

will also extract metabolites therefore & we can quantify these and gain extra information regarding compound activity.

M1M2M3

CONFIDENTIAL

Exploratory Metabolite Assay

M1 M2

M3 M4

ISTDLo-La

•These were synthesised in Frankfurt

•Used same LC conditions and extraction procedure as clinical method

•Added MS/MS transitions for major metabolites

•Note for this sample there is no cross-over ofMetabolites > LoLA

•Ran selected samples at higher concentrations to give extra information regarding activity & safety

M1 – M4 5ng/mL, spiked into human plasma, LoLA blank

CONFIDENTIAL

Considerations for LC-MS Assays for Insulins• No waiting required for development of specific antibodies for

immunoassays – get information faster• Greater assay specificity relative to LBA gives confidence and can be used

to generate new informationLevels of different metabolites give insight into action of the analogueNo cross reactivity with other insulin analogues (tested glulisine and 4 other insulin analogues for reactivity) or endogenous insulin (tested at 50 ng/mL)Improved instrumentation (sensitivity, increasing mass range) allows us to reach required LLOQsUsed halogenated & stable labelled internal standard

• IAC gives clean extracts to allow lower LLOQs for insulins and is surprisingly quick and robust

Must consider possible ADa effect multi-dose, esp. across species. So look at ‘total’ or free insulins?

• Acceptance criteria need careful consideration for these assaysSmall molecule limits may not be appropriate or achievable

• Peptide and small protein analysis (<8KDa?) by LC-MS/MS is becoming the standard

CONFIDENTIAL

References• POSTER: TT26B: Experiences on LC-MS Analysis of Large Molecules. D.

Schmidt et al., EBF, Barcelona 2012• POSTER: Quantitation of a Therapeutic Insulin Analogue by Immuno-

Affinity Extraction LC-MS/MS. M Blackburn et al., EBF, Barcelona 2012• Mass spectrometric identification of peptide hormones in doping control

analysis. Thevis and Schanzer., Analyst, 132, 287-291 (2007).• Quantitative Analysis of Biomarkers, drugs and toxins in biological samples

by immunoaffinity chromatography coupled to mass spectrometry or tandem mass spectrometry: A focused review of recent applications. Tsikas, D.I., J. Chromatography B., (2010).

• FAIMS: GLP-Validated quantitation method for a peptide in biological matrix. Klaasen, T., and Romer, A. ASMS Poster Presentation, Denver 2008.

• A Mass Spectrometric Method for quantitation of intact insulin in Blood Samples. Darby, S.M, Miller, M.L., Allen, R.O and LeBeau, M. Journal of Analytical Toxicology, Vol. 25, 2001.

• Clinical significance of insulin antibodies in insulin-treated diabetic patients. Van Haeften, TW, Diabetes Care 12:641-648, 1989.

THANK YOU FOR LISTENING!