Clinical Research Analytical Method Validation: How & Why it is Vital?

Transcription

Clinical Research Analytical Method Validation: How & Why it is Vital?
Analytical Method Validation:
How & Why it is Vital?
Marilyn A. Huestis, Ph.D.
Chief, Chemistry & Drug Metabolism
Intramural Research Program,
National Institute on Drug Abuse, NIH
CAT November 6, 2007
Clinical Research
ƒ Evaluate cognitive, subjective & physiological
effects & specific brain activities (fMRI) of licit
& illicit drugs
ƒ Identify important biomarkers of drug exposure
ƒ Fundamental hypothesis that a relationship
exists between pharmacological or toxicological
response to drug & concentration of drug in an
accessible compartment
Clinical Research
ƒ Conduct controlled drug administration
research in human drug users
ƒ IRB approved, written informed consent,
placebo-controlled, randomized, balanced,
double blind studies
ƒ Multiple doses & routes of administration
ƒ Subjects reside on closed research unit to
prevent exogenous drug administration
ƒ Ethical hurdles becoming prohibitive
Clinical Pharmacokinetics
ƒ Enables us to correlate pharmacokinetics with
pharmacodynamic effects
ƒ Enhance our understanding of physiological,
biochemical, subjective & cognitive drug effects
ƒ Controlled drug administration provides
framework for understanding these processes &
for interpreting drug concentrations in biological
fluids
Biological fluids & tissues
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
Blood
Plasma
Urine
Oral Fluid
Hair
Sweat
Breast milk
Brain
Skin
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
Meconium
Placenta
Cord Blood
Umbilical cord
Amniotic fluid
Urine
Hair
Method Validation
ƒ Before you start:
ƒ Define validation experiments
ƒ Performance characteristics & acceptance criteria
Quantification of Drugs &
Metabolites by GCMS & LCMSMS
ƒ Irreplaceable specimens of limited volume
ƒ High concentrations of major & low levels of
minor analytes in same limited specimen
ƒ Multiple drug classes
ƒ Throughput & turn-around time
ƒ Instrument time limited
ƒ Time required to troubleshoot assay or identify
interference
ƒ Most important: producing accurate results
Full Method Validation
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
Calibration model
Sensitivity
Dynamic linear range
Selectivity (Specificity)
Bias/Recovery (Accuracy)
Imprecision
Extraction efficiency
Stability
Carryover
Dilution integrity
Partial Method Validation
ƒ Method transfers between laboratories
ƒ Instrument &/or software platform changes
ƒ Change species matrix (rat to human plasma)
ƒ Change matrix (human urine to plasma)
Cross Validation
ƒ Comparison 2 bioanalytical methods
ƒ One serves as reference other comparator
ƒ Compare fortified samples & true specimens
ƒ Selectivity from metabolites
ƒ Selectivity from other medications
ƒ Change detection systems
ƒ Change specimen processing
ƒ Change anti-coagulant
ƒ Limited volume changes (pediatrics)
Calibration
ƒ Number of required calibrators
ƒ Generally, prefer 5 to 6 calibrators
ƒ Never force the curve through zero
ƒ Y intercept important descriptor of bias
ƒ Linear least squares
ƒ Homoscedastic data (constant variance across the
linear dynamic range)
ƒ Weighted least squares (1/x)
ƒ Heteroscedastic data (significant differences
between variances at the lowest & highest
concentrations) can handle larger dynamic range
Calibration
ƒ Quadratic (1/x2)
ƒ Extends the linear dynamic range
ƒ Requires more calibrators in the non-linear area of
curve
ƒ Historical curves
ƒ Not a good idea, certainly not for forensic testing
Calibration
ƒ All methods require calculation of individual
data points against the mean response factor
to assess acceptability
ƒ Frequently ±20% of target
ƒ Also, ±15% of target, except at LOQ ±20% of target
ƒ Split curves
ƒ Enables you to extend the dynamic range &
quantify analyte concentrations with one
extraction & injection
ƒ Required validation criteria met for both curves
Concentrations Evaluated
ƒ Replicates of QC
ƒ Highly preferred to do a one-way ANOVA to
estimate imprecision
ƒ QC independently prepared from different
supplier, vial or weighing than calibrators
ƒ Volume of organic added to fortify <5%
ƒ Number of calibration curves
ƒ Minimum of 4, preferable 5 or more
Concentrations Evaluated
ƒ Minimum of 3 quality control (QC) samples
over dynamic linear range: low, medium & high
ƒ Low QC within 3X LOQ, middle, high end of curve
ƒ Some require LOQ to be tested
ƒ Replicates of QC
ƒ
ƒ
ƒ
ƒ
Varies considerably
Minimum of 5 for intra-assay imprecision
Minimum of 20 for inter-assay imprecision
Or minimum duplicates for 10 days
Sensitivity
ƒ Limit of detection (LOD)
ƒ Lowest analyte concentration identified but not
quantified at exact value
ƒ Lowest analyte concentration reliably differentiated
from background noise
ƒ Signal to noise ratio ≥3 for analyte peak height &
highest & lowest baseline noise around analyte
Sensitivity
ƒ Limit of detection (LOD)
ƒ CDM requirement: empirical determination of lowest
analyte concentration with acceptable
chromatography, relative retention time ±2% of
reference, S/N ≥3 for all ions, & ion ratios within
±20% reference
ƒ Reference can be from 1 or all calibrators
Sensitivity
ƒ Limit of quantification (LOQ)
ƒ Lowest analyte concentration quantified with
suitable precision & accuracy
ƒ Signal to noise ratio of ≥10 for analyte peak height &
highest & lowest baseline noise around analyte
ƒ CDM requirement: empirical determination of lowest
analyte concentration that meets LOD & quantifies
within ±20% of target
ƒ LOD can equal LOQ when limits determined
empirically
Selectivity
ƒ Potential endogenous matrix interference
ƒ Must evaluate each matrix type
ƒ Minimum of 6 - 8 replicates, maybe 10
ƒ Especially important with LCMSMS
ƒ Potential exogenous component interference
ƒ
ƒ
ƒ
ƒ
Hemolysis, icteric, high lipid content
Metabolites
Concomitant medications
Illicit drugs
2D GCMS with Cryofocusing
Imprecision
ƒ Intra-assay variability (repeatability)
ƒ Determine CV%
ƒ Minimum of 5 replicates in same run
ƒ Preferred to perform an ANOVA to evaluate
imprecision within each day of multi-day analyses
ƒ Preferred N=20, 5 X 4, 4 X 5, duplicates in 10 runs
Imprecision
ƒ Inter-assay variability (intermediate precision)
ƒ Differences in operator, equipment or time
ƒ Preferred N=20, 5 X 4, 4 X 5, duplicates in 10 runs
ƒ Acceptance criteria ≤20%
ƒ Reproducibility
ƒ Precision between laboratories (external proficiency
test results)
ƒ Standardizing between laboratories
Imprecision
ƒ Intra-assay variability (repeatability)
ƒ Acceptance criteria ≤20%
ƒ If you fortify QC each day, frequently find significant
differences between days
ƒ Report worst CV for all days & state that intra-assay
imprecision significantly different, but not clinically
significant
Carryover
ƒ Contamination by a preceding sample
ƒ During validation, establish concentration above
which contamination could occur
ƒ Test highest concentration that is expected &
work lower as needed to establish this limit
ƒ Inject each concentration followed by zero
specimen (can quantify interference)
ƒ If follow by blank or solvent, only eyeball results
ƒ Typically expect result <LOD
Carryover
ƒ If carryover concentration is above the linear
range of method, can’t be sure of concentration
ƒ Typically, determine no carryover at upper LOQ
ƒ Should re-extract following specimen due to
potential carryover into the vial
ƒ Some suggest reinjection & compare 2
quantifications
ƒ Could also reinject high specimen followed by
zero specimen to quantify amount of carryover
Recovery
ƒ Two solution approach
ƒ Prepare 10 tubes with identical blank matrix
ƒ Fortify 5 tubes with native analyte (d0) & deuterated
internal standard (d3) prior to extraction
ƒ Extract 5 tubes & fortify with d0 & d3 after extraction
ƒ % recovery d0 = mean target ion area pre-extraction
mean target ion area post-extraction
ƒ % recovery d3 = mean target ion area pre-extraction
mean target ion area post-extraction
Recovery
ƒ Accuracy is difference between test result &
true result
ƒ Requires comparison against certified reference
standard
ƒ In many toxicological methods, reference standard
not available (exception blood alcohol, some TDM)
ƒ Generally, method bias or analyte recovery
determined
ƒ Measure how mean value of experimental
observations agrees with “accepted value”
ƒ Little agreement on appropriate method for recovery
Recovery
ƒ Acceptance criteria: as close as possible
recovery for d0 & d3
ƒ Advantages
ƒ Accounts for variability between SPE tubes,
injection volume & other factors
ƒ Useful for non-deuterated internal standards
ƒ Only requires 2 tubes
Recovery
Accuracy vs Precision
ƒ Three solution approach
ƒ Prepare 6 tubes with identical blank matrix
ƒ A: fortify 3 tubes with d0 & d3 prior to extraction
ƒ B: fortify 3 tubes with d0 prior to extraction;
add d3 after extraction
ƒ C: Prepare a neat solution of same amount of d0 & d3
ƒ Extraction = mean ratio d0 & d3 areas pre-extraction
efficiency mean ratio d0 & d3 areas post-extraction
ƒ Absolute = mean target ion area pre-extraction
recovery mean target ion area post-extraction
Stability
ƒ Analyte stability in a given matrix under specific
conditions for given time intervals
ƒ During storage prior to analysis
ƒ Room temperature, 4ºC, -20ºC, -80ºC
ƒ
ƒ
ƒ
ƒ
Of derivatized extracts
On autosampler
3 freeze-thaw cycles
All conditions should be determined for realistic
storage times
O
O
X
O
O
O
X
O
O
O
X
O
Not accurate
Not precise
Accurate
Precise
Not Accurate
Precise
CDM stability requirements
ƒ 3 QC levels (3 replicates) quantify ±20%
compared to freshly prepared & analyzed
specimens
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
Room temperature for 8 h
1 - 7 days at 4ºC
1 to 12 wks -20ºC (rarely -80ºC)
Derivatized extracts for 48 to 96 h (temperature?)
On autosampler for 48 to 96 h
3 freeze-thaw cycles
ƒ Our operation not designed for long-term
stability studies- limitation
Dilution integrity
ƒ Document dilution protocol produces accurate
results
ƒ Dilute high concentration QC with blank matrix
to bring concentration within analytical range
ƒ Document that results ±20% of target
ƒ Good practice during routine analysis to always
dilute a QC with patient specimen
ƒ If water or buffer are used to dilute, must
validate that results are equivalent
LCMS, LCMSMS & LCMS-TOF
ƒ Ion suppression (↓ analyte signal) & ion
enhancement (↑ analyte signal) due to matrix
effects
ƒ Matrix effect due to alteration of ionization
efficiency due to presence of co-eluting ions
ƒ Exact mechanism unknown
ƒ Competition between non-volatile matrix ions &
analyte ions for transfer to gas phase
ƒ Greater problem with electrospray ionization
than atmospheric pressure chemical ionization
Robustness
ƒ Measures method susceptibility to small
changes occurring during routine analysis
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
pH
Mobile phase composition
Flow rate
Column temperature
Lots of extraction materials, columns, solvents
Derivatization temperature
Different analytes, instruments
LCMS, LCMSMS & LCMS-TOF
ƒ More suppression with polar ions elute early
with many matrix components
ƒ Post-extraction addition of analytes
ƒ Compare analyte areas in matrix & mobile phase
ƒ Static method: data about matrix effects at time of
analyte elution
ƒ Post-column infusion:
ƒ Infuse analyte continuously into MS, inject matrix
after different specimen preparation techniques
ƒ See areas of matrix effect across run time
LCMS, LCMSMS & LCMS-TOF
ƒ Additives in mobile phase
ƒ Acetate vs trifluoroacetate (reduces ionization)
ƒ Phospholipids produce ion suppression
ƒ Glycerophophocholines major type in plasma
ƒ Monitor 184 to 184 ion to see phospholipids
ƒ In routine practice can help determine when
necessary to change analytical column
LCMS, LCMSMS & LCMS-TOF
ƒ Minimize matrix effect
ƒ Improve specimen preparation
ƒ Change chromatography to move analytes of
interest to area of chromatogram without effects
ƒ Choose different ion source
ƒ Deuterated internal standard minimizes matrix effect
ƒ Need internal standard that mimics extraction efficiency
from matrix & ionization efficiency
ƒ Specimen containers
ƒ Preservatives
LCMS, LCMSMS & LCMS-TOF
ƒ Matuszewski et al Anal Chem 2003;75:3019-30
ƒ Suggested assay calibration curve in at least 6
blank matrices & compare %CVs
ƒ Taylor LCMS Method Validation IATDMCT 07
ƒ Suggested assay 5 replicates of 3 QCs in 5 different
specimen matrices
ƒ Add d0 & d3 pre & post-extraction & in mobile phase
ƒ 45 injections for matrix effect, absolute recovery,
process efficiency & inter-subject variability
LCMS, LCMSMS & LCMS-TOF
ƒ Taylor IATDMCT 2007
ƒ
ƒ
ƒ
ƒ
Matrix effects: Post - Pure/Pure *100
Absolute recovery = Pre/Post *100
Process efficiency: Pre/Pure *100
Inter-subject variability: CV of 5 Pre samples
ƒ Huestis:
ƒ Absolute recovery = Pre/Pure *100
ƒ Process efficiency: Pre/Post *100
ƒ CDM uses ratios rather than area abundance to
remove variability by including internal standard
Useful References
ƒ Guidance for Industry: Bioanalytical Method
Validation
http://www.fda.gov/cder/guidance/4252fnl.htm
ƒ Clinical & Laboratory Standards Institute
(formerly NCCLS) http://www.nccls.org/
ƒ International Conference on Harmonization:
Chemical & Pharmaceutical Quality Assurance
ƒ International Organization for Standardization
ƒ ISO 9000 Quality Management Principles
ƒ http://www.iso.org/iso/home.htm
Analytical Method
ƒ Develop & validate a sensitive & specific
analytical method for detection & quantification
of MDMA & metabolites MDA, HMMA & HMA
ƒ 2D-GC/MS for sufficient sensitivity &
specificity
ƒ LOQ ≤ 10 ng/mL necessary for characterizing
pharmacokinetics of MDMA & metabolites
ƒ To analyze specimens from controlled MDMA
administration study
ƒ Erin Kolbrich’s doctoral dissertation
Useful References
ƒ Shah et al 2000 Pharmaceutical Research 17:15511557
ƒ Hartmann et al 1998 J Pharmaceutical & Biomedical
Analysis 17:193-218
ƒ Goldberger et al 1997 Forensic Science Review 9;5980
ƒ Taylor PJ in Polettini A ed Application of Liquid
Chromatography-Mass Spectrometry in Toxicology
2006
ƒ Matuszewski et al Anal Chem 2003;75:3019-30
ƒ Matuszewski et al J Chrom B 2006;830:293-300
MDMA GCMS Analytical Method
Acid hydrolysis - 0.5 M HCl (time, temp & amt)
↓
pH adjust - 0.1 M phosphate buffer & 10 M NaOH
↓
TM
SPE- UCT Styre Screen DBX
↓
Derivatization- 10 µL HFBA
↓
Back extract - 200 µL TRIS buffer
↓
3 µL splitless GC injection
Analytical Method
ƒ 2D-GC/MS
Analytical Method
FID
TIC
ƒ Instrument: Agilent 6890N GC with 5973 MSD
ƒ GC:
ƒ 1 ° Column: Agilent DB1DB1-MS (non(non-polar)
ƒ 2° Column: Phenomenex ZBZB-50 (↑
(↑ polarity)
ƒ Three “cut”
cut” windows
ƒ Cryotrap: 50°
50°C, then 800°
800°C/min to 275°
275°C
ƒ Oven: 7070-195°
195°C, 195195-100°
100°C (fast), 100100-190°
190°C
ƒ MS
ƒ Ionization: electron impact
ƒ Mass transfer line temperature: 280°
280°C
Analytical Method Validation
ƒ 3 quality control concentrations/analyte
Analytical Method Validation
ƒ Sensitivity & Linear dynamic range
ƒ 2.5, 15, 75 ng/mL MDA & HMA
ƒ 7.5, 40, 200 ng/mL MDMA & HMMA
ƒ
ƒ
ƒ
ƒ
ƒ
ƒ
LOD & LOQ
Bias/Analytical recovery “Accuracy”
Imprecision
Extraction efficiency
Specificity
Dilution integrity, stability & carryover
ƒ 6 concentrations per curve, except HMA = 5
ƒ Pholedrine as IS
ƒ R2 ≥ 0.997
Analytical Method Validation
Analytical Method
Analytical Method
Analytical Method
ƒ Dilution integrity
ƒ 75% dilution with buffer
ƒ All within ± 20% of ¼ mean high QC
ƒ Stability
ƒ Short-term: 12 h room temp, 48 h refrigeration
ƒ Long-term: 3 freeze-thaw cycles
ƒ All within ± 20% of fresh QC, except low HMA
ƒ Carryover
ƒ Low QC injected after 1000 ng/mL
ƒ Quantification & ion ratios within ± 20% target
Participant specimen: 7 h after 106 mg MDMA dose
4.1 ng/mL
110.3
ng/mL
6.7 ng/mL
80.9 ng/mL
Advantages of LC/MS/MS
ƒ Simple specimen preparation
ƒ Some matrices are dilute & shoot (urine)
ƒ Protein precipitation with ACN (blood, oral fluid)
ƒ No derivatization
ƒ Wide range of analytes in a single assay
ƒ Thermolabile & nonvolatile analytes
ƒ Potential for eliminating hydrolysis step &
directly analyzing phase 2 metabolites
ƒ Large dynamic linear range
ƒ Improved fragmentation of opioids
Opiates, Cocaine, & Metabolites
in Oral Fluid
ƒ Why LC/MS?
ƒ Simple specimen preparation
ƒ Protein precipitation with acetonitrile
ƒ No derivatization
ƒ
ƒ
ƒ
ƒ
ƒ
26 cocaine & opiate analytes in a single run
Atmospheric pressure chemical ionization
200 μL oral fluid specimen
Matrix matched calibrators 1 - 500 μg/L
Intra-assay imprecision <15.8%; inter-assay
imprecision <22.8%
Disadvantages
ƒ Cost
ƒ Knowledge & expertise
ƒ Less than for GC/MS
ƒ More than for GC/MS/MS
ƒ Matrix effect
ƒ Suppression & enhancement
ƒ Necessary to quantify effect
ƒ Lack of deuterated internal standards for minor
metabolites & glucuronides problematic
Opiates, Cocaine, & Metabolites
in Oral Fluid
ƒ Demonstrates typical concentrations in
oral fluid after illicit use
ƒ Applicability of SAMHSA cutoff concentrations
for detecting exposure
ƒ Evaluates effectiveness of drug treatment
ƒ Determined drug recovery from oral fluid
collection device & oral fluid volume recovery
ƒ Monitored wider range of analytes
S u b je c t A
Concentration (µg/L)
25
6 -A M
20
H e r o in
15
10
5
0
0
5
10
15
20
25
30
35
Subject B
Morphine
500
25
6-AM
Heroin
400
20
Acetylcodeine
Codeine
300
15
Noscapine
200
10
100
5
0
0
1
5
9
13
Study days
17
21
Concentration acetylcodeine,
codeine, and noscapine (µg/L)
Concentration 6-AM, heroin,
and morphine (µg/L)
S tu d y d a y s
Thank You for Your Attention!