Using Mass Spectrometry to Measure Proteins in the Clinical Laboratory

Transcription

1 Using Mass Spectrometry to Measure Proteins in the Clinical Laboratory AACC Mass Spec October 10, 2014 Andy Hoofnagle, MD PhD University of Washington

2 Learning Objectives Describe the benefits of using mass spectrometry to measure proteins in clinical samples. Outline the procedure for measuring proteins in complex samples by mass spectrometry. List two methods for to improve the sensitivity of mass spectrometric protein assays.

3 The History of Proteomics: Living with the Black Eye Lancet, 2002

4 Ransohoff's World Left-of-the-red-line Goal: Early Detection of Stage I Ovarian Cancer Stage IV cancer vs. normals Cancer drawn in vs. normals Cancer prepared first vs. normals second Discovery Proteomics Stage I cancer vs. benign ovarian masses Contemporaneous draws Complete randomization of preps & analyses Discovery Proteomics

5 The History of Proteomics: Living with the Black Eye Lancet, 2002

6 Lancet, 2002 The History of Proteomics: Living with the Black Eye To good to be true?

7 The History of Proteomics: Living with the Black Eye Sorace and Zhan, BMC Bioinformatics, 2003 Re-analyzed the data from the Petricoin paper "The ability to discriminate between cancer and control subjects based on the m/z values of and reveals the existence of a significant non-biologic experimental bias between these two groups... [which] might include medication or lifestyle change... variation in sample collection, processing and preservation, as well as bias introduced at the time of analysis [e.g., the order of analyses]."

8 Shotgun Proteomics Database searching of MS/MS data to identify and quantify LC-MS/MS Proteins Peptides

9 HPLC-ESI-MS/MS

10 Proteomics Database Searching LC-MS/MS Time Peptides Chromatogram

11 Proteomics Database Searching Time Chromatogram

12 Proteomics Database Searching

13 Proteomics Database Searching

14 Proteomics Database Searching DLGPLTK

15 Proteomics Database Searching DLGPLTK LDGPITK NIGAIGAK Software makes it possible

16 Shotgun Proteomics Approach is aimed at identification not quantification Inherently less precise Significant false-discovery rate The workhorse of proteomics for decades

17 Targeted Quantitative Proteomics Isotope dilution MS/MS to quantify known targets * * * LC-MS LC-MS/MS * Proteins Peptides

18 Multiple Reaction Monitoring Peptide AETHLISK Internal standard AETHLISK Peptide VIFDANAPR Internal standard VIFDANAPR

19 LC-MS/MS Protein Measurements Many steps, potential variability 1. Denature the proteins 2. Reduce and alkylate 3. Digest with enzyme 4. Solid phase extraction 5. Autosampler 6. Chromatography 7. Mass spectrometric detection 8. Peak integration

20 Immunoaffinity Enrichment Anti-protein antibody Anti-peptide antibody Y Y LC-MRM/MS LC-MRM/MS

21 Immunoaffinity Peptide Enrichment

22 Protein Biomarkers Why quantify proteins in complex mixtures? Basic science Discovery Select patients for clinical trials Toxicity Efficacy Diagnosis Prognosis Therapeutic management

23 Experiments Basic science Cells Animals A few human samples Discovery Expect trends to repeat not necessarily the numbers

24 Experiments Human subjects, retrospective Discovery Verification Qualification May not need IRB approval Freezer diving

25 Experiments Patients, prospective clinical trials Diagnosis Prognosis Management Need IRB approval for these studies These assays need to be fully validated These assays need to be fully validated Using novel biomarkers in patient care

26 Validation Regulated Rigorously define performance characteristics Suggested acceptability best practices Documentation Transparency not really a component

27 Validation Unregulated Publishing guidelines Funding agency guidelines Peer reviewers Training Transparency must be a fundamental tenet

28 Tiers of Assay Quality Carr, MCP, 2014

29 Tiers of MRM Assay Quality Tier 1 Tier 2 Tier 3 Clinical diagnosis Research Discovery Internal standards Internal standards Highly precise Precise Less precise QC procedures Values repeatable Trends repeatable Large magnitude changes One to a few analytes 10s-100s of analytes 10s-100s of analytes Fit-for-purpose approach Carr, MCP, 2014

30 Replicating Tier 2: Peptides Experiment 1: Response Curve Experiment 2: Mini-validation of Repeatability Preliminary validation of peptide quantification in complex mixtures CPTAC Assay Development Working Group

31 NCI Assay Portal CPTAC Assay Development Working Group

32 Translating Tier 2: Proteins Validation Reproducibility Peptide degradation Linearity LLOQ Interferences Stability Experiment 5x5 experiment Spike IS peptides before/after digestion Mix pools together Dilution experiments of one or more pools Add potential interferences to pools Stress pools before and after sample prep Validating protein quantification for publication Grant, ClinChem, 2014

33 Guidance Documents: CLIA & CLSI Performance specification Precision Linearity Analytical sensitivity Specificity and interferences Accuracy and method comparison Multiple instruments Carryover Quality control References EP05, EP29 EP06 EP17 EP07, EP14 EP09 EP31 EP10 EP28 Validating protein quantification for the care of patients

34 Accuracy The accuracy of a measurement is the closeness of agreement between the test result and the true result Precision Within batch Between batch Between laboratory Bias Calibration materials Sample-specific matrix effects How similar will a result be if measured in Seattle on Monday and in London on Wednesday? Is it right?

35 Harmonization Trying to make results as similar as possible Examples Two different platforms in the same laboratory The same platform in different laboratories Two different platforms in two different laboratories Methods Non-certified reference material Cooperation within a consortium Recalibration to a single central laboratory Improving between-platform concordance Reducing between-laboratory variability Not necessarily the true answer

36 This is Not Harmonization One FDA-approved Immunoassay, Two Laboratories IGF-1 Site 2 r 2 = 0.66 y = 0.6 x Site 1 Cox, ClinChem, 2014

37 This is Not Uncommon Seven FDA-approved Immunoassays, One Laboratory Comparison Methods PTH Clinical Gold Standard Method Cantor, Clin Chem, 2006

38 Hypothyroidism is Diagnosed with TSH Google Maps Three hospitals, 0.5 miles apart, three platforms, two different outcomes Roberts, Clin Chem, 2004

39 Standardization Trying to make results from all platforms in all laboratories give the same true result Methods Reference method procedures Certified reference materials Traceable calibration materials Examples Vitamin D Standardization Program Hormone Standardization Program Hemoglobin A1c Standardization (NGSP) Pre-established acceptance criteria Commutable materials

40 Keeping it real over time Proficiency Testing Methods Spiked synthetic matrix Stabilized human samples Minimally-processed native human samples Examples College of American Pathologists Royal College of Pathologists of Australasia DEQAS

41 Proficiency Testing Commutability Matters Comparator Method Patient samples Proficiency testing material Reference Method

42 Proficiency Testing Accuracy Based CAP is moving toward accuracy based programs a great step forward

43 Does calibration have a role in research proteomics assays?

44 Variability of Peptide Liberation Vary the amount of trypsin (50%-200%) Very different results for each peptide

45 Variability of Peptide Liberation Mechanism of agitation matters Peptide Thermomixer Thermomixer, stopped shaking End-over-end Peptide 2 How can two peptides from the same protein be so different?

46 Variability of Peptide Liberation Are we ever done? Agger, ClinChem, 2010

47 Calibration Improves Concordance Between-laboratory repeatability Calibration curve Single-point calibration Four other laboratories Ave CV=16.0% Ave CV=11.1% Comparison laboratory Cox, Clin Chem, 2014

48 Autoantibody Interference Autoantibodies can sequester antigen away from detectable complexes Falsely negative results with sandwich assays

49 Immunoaffinity Peptide Enrichment

50 Thyroglobulin: A Low Abundance Protein A Troubled Biomarker Autoantibody Negative Positive Hoofnagle & Roth, JCEM, 2013

51 Replication at National Reference Labs ARUP Quest Thyroglobulin, ng/ml (Immunoassay) Thyroglobulin, ng/ml (Immuno-MRM) Kushnir, Clin Chem, 2013 Clarke, J Inv Med, 2012

52 Interlaboratory Concordance ImmunoMRM Assay 50 Assay Assay 2 Kushnir, Clin Chem, 2013

53 Conclusions We can accurately quantify proteins using LC-MS/MS Calibration is the cornerstone of repeatability Validation can be fit-for-purpose Effective translation requires adequate validation Transparency is pivotal for generalization Life after validation harmonization standardization proficiency testing

54 Acknowledgements UW Tom Laha Fred Strathmann Zhican Wang Ian de Boer Ken Thummel John Brunzell Tomas Vaisar Jess Becker Mike MacCoss Mark Wener LabCorp Russ Grant ARUP Alan Rockwood Mark Kushnir IGF-1 Working Group Holly Cox Filipe Lopes David Cowan Mark Parkin Andreas Thomas Mario Thevis Getachew Woldemariam Anthony Butch Larry Bowers Waters Don Cooper Andrew Peck Broad Eric Kuhn Sue Abbatiello Steve Carr Fred Hutchinson CRC Jeff Whiteaker Mandy Paulovich Equipment Waters Thermo Funding NIH (NHLBI, NCI, NIDDK, ODS) Waters Partnership for Clean Competition Alliance for Lupus Research Department of Laboratory Medicine