1. Regulated Bioanalysis:
Science and Art of Meeting the Requirements
India Technology Seminar
Boston, 29 October – November 1, 2007
Dr. Bhaswat S. Chakraborty
30.10.2007

2. Bioanalytical Guiding Principles
 Used in quantitative determination of drugs and
their metabolites in biological fluids
 BE, PK, TK and Meatbolomics studies
 Often the Study is as good as the bioanalytical
 Refinement of all aspects in last 20 years
 Implemented as though regulations
 More and more demanding
 From 1980s to date
 US FDA setting the standards

3. Key Guiding Documents & Publications
 Shah VP et al. Analytical methods validation…. Eur J
Drug Metab Pharmacokinet. 1991
 Draft Guidance on Bioanalytical: USFDA. 1999
 Bioanalytical Workshop. 2000
 Guidance for Industry: Bioanalytical: USFDA. 2001
 DeSilva B et al. Validation of ligand-binding assays…
macromolecules. Pharm Res. 2003
 Draft Guidance on Safety Testing of Drug Metabolites:
USFDA. 2005
 FDA-AAPS Bioanalytical Workshop. 2006

4. Scientific Questions behind Guidance
 Does the method measure precisely and accurately the
analyte(s)?
 If so, how was it validated? How complete and accurate is
the validation?
 Were adequate CC and QC used during incurred sample
runs? Was each and every batch acceptable?
 Were the incurred samples (& repeats) accurately and
reproducibly measured and were they stable at all times?
 Is the documentation comprehensive and reconstructive?

5. Non-Chromatographic Assays: Issues
 E.g., RIA & other ligand binding assays
 Selectivity
 Specific
 Non-specific or Matrix
 Often analyte is also present endogenously
 Quantification
 Non-linear CC
 More points than Chromatographic CC
 Accuracy is questionable, so replicates required
 Cross validation
 Stability is a complex issue
 Stabilty of receptor binding activity as well?

6. Non-Chromatographic Assays: Validation
 Deplete the matrix of the analyte or employ a
“surrogate” matrix
 5 or more validation concentrations
 Intra- and interbatch variation
 LLOQ, ~3 times LLOQ, mid [geometric mean], 75% of
ULOQ,and ULOQ
 At least 2 independent determinations
 Interbatch imprecision (%CV) and inaccuracy (%RE):
±20% (25% at LLOQ and ULOQ).
 Total target error (imprecision and inaccuracy) ≤30%
(40% at LLOQ &ULOQ)

7. Non-Chromatographic Assays: Batch Runs
 CCs
 At least 75% of the standard points should be within 20% of
nominal concentrations (25% at LLOQ)
 Does not apply to “ anchor calibrators, ” those outside the
anticipated validation range and improve “ sigmoidal ” fitting
 QCs
 LQC, MQC, HQC in duplicate
 4 – 6-20 rule
 4 of the 6 QCs must be within 20% of nominal
 At least one QC per conc. meets this criterion
 If additional QCs are used, then 50% of them be within 20%
of nominal

8. Non-Chromatographic Assays:
Cross Validation
 E.g., RIA vs LC-MS
 The comparisons can be done both ways.
 Cross-validation with spiked matrix and subject
samples be conducted at each site or laboratory
 Interlaboratory reliability
 if analyses within a single study are conducted at more
than one site or lab
 If data generated using different analytical techniques

9. Chromatographic and Hyphenated Assays
 HPLC, UPLC, GC, LC-MS
 More selective and accurate than LBAs
 Used for small molecules
 Drugs, metabolites, toxicants
 Linear response
 LC-MS: highly sensitive
 High capacity
 Highly reproducible response stability

10. LC-MS or LC-MS-MS
 LC coupled with one or mass analyzers
 Most popular, very accurate, high throughput
 Quadrupole mass analyzers: most widely used
 Aligns different m/z according to their retention times
 Time-of-flight mass analyzers
 Analyzes different m/z according to the different times taken
to traverse a fixed distance
 FTICR mass analyzers
 Applies a radiofrequency voltage at the same cyclotron
frequency of a m/z, the later latter is moved at a larger radius
than the RF

11. Validation Batches
 Analyze at least 3 batches for accuracy and
precision
 At least 1 validation batch should be made as
large as the largest anticipated sample analysis
batch

12. Calibration Samples & Acceptance Criteria
 CC concentrations
 A minimum of 6 non-zero standards
 Matrix blank: Matrix sample without internal standard
 Zero standard: Matrix sample with internal standard
 Acceptance criteria
 Intra- and inter-batch imprecision (%CV) and inaccuracy
(%RE) ≤15% except at LLOQ where up to 20% can be
allowed

13. CC: Intra-Batch
N 4 4 4 4 4 3 4 4
MEAN 68.858 56.445 51.053 36.433 15.498 7.957 0.983 0.490
SD(±) 4.5328 4.0874 4.2432 2.8389 1.1250 0.4143 0.0785 0.0183
%CV 6.58 7.24 8.31 7.79 7.26 5.21 7.99 3.73
%NOMINAL 93.08 95.38 103.51 98.49 104.71 107.52 99.24 100.00

14. QC Samples & Acceptance Criteria
 QC concentrations
 LLQC: About 3 times the LOQ; MQC: ~geometric
mean of LQC & HQC; HQC: ~70% to 85% of
ULOQ; Dilution QC: sufficient to cover highest
anticipated dilution
 Acceptance criteria
 Intra- and inter-batch imprecision (%CV) and
inaccuracy (%RE) ≤15% except at LLOQ where up
to 20% can be allowed

15. QC: Inter-Batch
N 22 24 21
Mean 0.979 35.482 55.913
SD (+) 0.0670 2.6945 3.6018
CV (%) 6.85 7.59 6.44
Min. 0.88 31.75 50.57
Max. 1.10 40.86 63.63
%Nominal Conc. 98.90 95.90 94.58

16. Selectivity (Non-interference from Matrix)
 Review noninterference in at least 6 sources of
matrix for non-MS assays
 For MS assays determine MFs in 6 sources if
the nonisotopically labeled IS is used
 If isotopically labeled IS is used, demonstrate
that IS normalized MF is close to unity
 Interference in blank matrix ≤20% of LOQ

17. Example
Matrix effect accuracy QC %nominal value 100.70
Hemolysis effect accuracy QC % nominal value
For LQC is 101.41 and
For HQC is is 100.74
Hemolysis effect precision QC Coefficient of variation
For LQC is 9.57 and for
HQC is 5.65

18. Reproducibility of the Method
 Precision and accuracy:
 Inter-run precision and accuracy of the QCs
 Second column or instrument verification:
 Reproducibility of the method on an alternate column or
instrument
 A batch of precision and accuracy samples is analyzed on a
different column or instrument on one of the days of
validation. Good practice but not mandatory
 Reproducibility using incurred samples:
 Sample availability can be an issue

19. Example: Column to Column Ruggedness
Conc.ng/mL QC LOW QC MED QC HIGH
0.99 ng/mL 36.99 ng/mL 59.18 ng/mL
0.84 – 1.14 31.44 – 42.54 50.03 – 68.06
Conc.found %Nom Conc.found %Nom Conc.found %Nom
0.94 94.75 32.90 88.96 58.82 99.39
0.89 89.54 33.03 89.28 58.28 98.47
1.02 103.18 31.75 85.83 56.18 94.92
1.01 101.94 33.32 90.08 53.02 89.59
1.14* 114.98 33.93 91.72 53.33 90.11
0.95 95.93 35.26 95.32 49.62* 83.84
N 5 6 5
MEAN 0.962 33.365 55.926
SD 0.0536 1.1704 2.7002
Min 0.89 31.75 53.02
Max 1.02 35.26 58.82
%CV 5.57 3.51 4.83
%NOM 97.17 90.20 94.50

20. Analyte Recovery
 Detector response of extracted vs direct input
 Extent of recovery of analyte and IS
 Demonstrate that it is
 Consistent
 Precise
 Reproducible
 Determine recovery at 3 concentration levels
 Does not have to be close to100%

21. Example: Drug Recovery
Low QC
Extracted response Unextracted response
325712 605788
330450 611672
332840 621078
309710 607364
345282 606242
375431 602309
N 6 6
Mean 336570.83 609075.50
SD (±) 22257 6608
CV (%) 6.61 1.08
Mean recovery (%) 55.26

22. Stability of Samples
 Stock solution:
 Minimum of 6 hours at room temp. & Fridge temp. for 24 hr
 Postpreparative (extracted samples/autosampler tray):
 Longest time from preparation through analysis. Vs. fresh
standards
 Benchtop:
 At ambient temp. (or processing temp.) – for extraction duration
(typically ~4-24 hr)
 Freeze-thaw:
 QC samples at least 2 conc., 3 cycles, completely thawed,
refrozen for 24/12 hr, at anticipated storage temp.
 Long-term: Can be postvalidation
 For longest time – collection to for any sample (3 aliquots; low
and high conc. with fresh standard curves); assess vs. nominal

23. Example
Auto sampler stability QC Mean % Change
(~49 hr) LQC is –4.28 and for
HQC is –2.35
Bench Top Stability in matrix at room temperature QC Mean Change for
(~ 25.5 Hrs.) LQC is 4.56 for
HQC is 3.73
Freeze and thaw stability QC Mean % Change for
At –30±5°C after 3 cycles LQC is 5.60 and for
HQC is 3.47
Dry extract stability QC Mean % Change for
At –30±5°C after (Approx 51 Hrs.) LQC is -6.75 and for
HQC is 3.16
Short-term stability in solution at room temperature Mean % change is -1.10
(~16.5 Hrs.)
Short-term stability in solution at refrigerator Mean % change is –0.48
(~ 96 Hrs.)

24. Dilution Integrity for Concentrations >ULQ
 One or more additional QC >>ULQ prepared
and diluted with blank matrix to bring the
concentration to within the calibration range
and then analyzed
 The acceptance criteria for the diluted QC are
the same as for other QCs
 Intra- and inter-batch imprecision (%CV) and
inaccuracy (%RE) ≤15%

25. Example
 Dilution integrity accuracy (as QC %nominal value)
 For 1/5th dilution factor is 96.13
 For 1/10th dilution factor is 111.49
 Dilution integrity precision (as QC %CV)
 For 1/5th dilution factor is 6.82 and
 For 1/10th dilution factor is 4.76

26. MS Techniques: Matrix Factor
Ion Enhancement
Syringe
Pump MS
Drug ISTD
Autosampler
Ion Suppression

27. MS Techniques: Matrix Factor (MF)
 A quantitative measure of the matrix effects due to
suppression or enhancement of ionization in an MS
detector
 MFs can be determined for the analyte and the IS
 Ratio is called IS-normalized MF for the analyte
 IS-normalized MFs using stable isotope labeled IS
 Usually close to unity for bioanalytical samples
 MF or IS-normalized MF be determined in 6
independent lots of matrices with desirable CV <15%
(not for stable isotope labeled IS using methods)

28. PK Repeats
 SOP
 All PK repeats are
chosen before interim
analysis of data
 Based on sound PK and
Bioanalytical principles
without any bias Ct
 Original value is retained
if the repeat value is
within 15%
 Repeats with one or
more “correct” estimates
t

29. CC Algorithm &
Estimation of the Unknown

30. Data
Conc. (ng/ml) Peak Area Ratio
1501 1.54
1002.5 0.91
703 0.77
249.5 0.25
99.5 0.13
30 0.04
10 0.01
5 0.01

31. Comparison: LS vs. MLE
Least Squares MLE
Intercept 0.009557334 0.009426942
Slope 0.00099529 0.001014858
Std. Error 0.023385281 0.029742873
df 6 6
…very similar in this case but can be very different in other cases

32. Least Squares vs. MLE
1.8
MLE
1.6
1.4
1.2
Peak Area Ratio
Least Squares
1
0.8
0.6
0.4
0.2
0
0 200 400 600 800 1000 1200 1400 1600
Conc. (ng/ml)

33. Comparison: LS vs. MLE
An Example where differences can be substantial
X Y Least Squares MLE
5.66 59
4.63 43 Intercept 3.192698019 10.36237254
5.21 41 Slope 7.424965403 6.621244357
6.01 47 Std. Error 0.157779577 3.300103342
6.17 53 df 10 10
4.82 43
5.08 46
5.51 45
4.95 41
5.13 42
4.66 39
4.74 45

34. Linear Weighting (1/X2 and 1/Y2) & MLE
Least Squares MLE Weighted (1/X^2) Weighted 1/Y^2
Intercept 0.009557334 0.009426942 0.00393 0.00262
Slope 0.00099529 0.001014858 0.00104 0.00102
Std. Error 0.023385281 0.029742873 -- --
df 6 6 6 6

35. Conclusions

36. Current Guidelines are based on
 Expertise
 Correct understanding of the isolation of the analyte from
matrix, metabolites & noise
 Relevant approach to method development & validation, e.g.,
target bias, establishment of method characteristics
 Ethics
 Non-deviation from protocol, QA, QC, audit trail
 Focus
 Specificity, accuracy, precision, reproducibility & stability

37. You are Successful when Your Data
 Enable the regulators know exactly what you
know & come to the same conclusions
 Provide evidence of ‘correct’ determination of
unknown concentrations
 Present validation and analytical report
 Are traceable, accounted for and of proven
integrity
 Are complete

38. In addition ..
 Be aware of
 CC algorithms, their strengths & pitfalls
 Peak smoothing techniques & expertise
 Peak instability, non-reproducibility of a specific assay
 Having correct & enough SOPs
 Long term sample storage
 Assays where replication of clinical samples assay may be
necessary
 Regulatory queries & how to respond to them

39. Acknowledgments:
Dhriti Chakraborty
Chinmoy Ghosh
Adinarayana Andy
THANK YOU