Fast, Automatic, and Accurate Determination and Identification of Targeted Analytes in High-Throughput Analysis by Chromatography – Tandem Mass Spectrometry Steven J. Lehotay and Yelena Sapozhnikova Agricultural Research Service Eastern Regional Research Center Wyndmoor, Pennsylvania; USA
67
Embed
Fast, Automatic, and Accurate Determination and ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Fast, Automatic, and Accurate Determination and Identification of Targeted Analytes in High-Throughput
Analysis by Chromatography – Tandem Mass Spectrometry
Steven J. Lehotay and Yelena Sapozhnikova
Agricultural Research ServiceEastern Regional Research Center
Wyndmoor, Pennsylvania; USA
DisclaimerMention of brand or firm name does not constitute an endorsement
by the USDA above others of a similar nature not mentioned.
• Aminoglycosides can be analyzed at the same time as other veterinary drugs by adding the ion-pairing reagent to combined final extracts from two sample preparation methods.
• Smaller test portions (2- 5 g) in QuEChERS are possible using the Blixer for many (but not all) commodities.
• High quality, rugged results can be achieved for hundreds of ultratrace analytes in diverse foods using automated high-throughput analysis by QuEChERS + ITSP+LPGC-MS/MS and UHPLC-MS/MS in parallel without matrix-matched calibration followed by summation function chromatographic peak integrations and post-run processing to yield accurate determinations and trustworthy identifications with little need for human review.
High Throughput Efficiency Start to Finish
1) Sample Processing (Blixer and/or Cryomill?)
2) Batch extraction by QuEChERS
3A) UHPLC-MS/MS 3B1) Automated clean-up (ITSP)
3B2) Fast, low-pressure (LP) GC-MS/MS
4) Trustworthy automatic peak integrations and identifications without human review
0 5 10 15 20 25 30
DataProcessing
Analysis
SamplePreparation
SampleProcessing
minutes/person/sample
Recent Past
Current
Near Future
Sample Throughput to Analyze Chemical Residues mini-QuEChERS + UHPLC- & ITSP+LPGC- MS(/MS)
UHPLC- and Fast GC-MS/MS
Automated in Parallel
Summation Function Chromatographic Peak Integration
This future is now!
• Lehotay et al. (2016) "Automated mini-column solid-phase extraction cleanup for high-throughput analysis of chemical contaminants in foods by low-pressure gas chromatography – tandem mass spectrometry" Chromatographia, 79, 1113-1130
•Han et al. (2016) “Method validation for 243 pesticides and environmental contaminants in meats and poultry by tandem mass spectrometry coupled to low-pressure gas chromatography and ultrahigh performance liquid chromatography” Food Control 66, 270-282
• Sapozhnikova and Lehotay (2015) “Review of recent developments and applications in low-pressure (vacuum outlet) gas chromatography” Anal. Chim. Acta 899, 13-22
• Lehotay et al. (2015) “Current issues involving screening and identification of chemical contaminants in foods by mass spectrometry” Trends Anal. Chem. 69, 62-75
•Lehotay and Cook (2015) “Sampling and sample processing in pesticide residue analysis” J. Agric. Food Chem. 63, 4395-4404
• Sapozhnikova and Lehotay (2015) “Evaluation of different parameters in the extraction of incurred pesticides and environmental contaminants in fish” J. Agric. Food Chem. 63, 5163-5168
Some Recent Publications of Note
Major Classes of Antibiotics
Aminoglycosides
Sulfonamides Tetracyclines-Lactams
Macrolides Quinolones
Penicillin G Sulfadimethoxine Tetracycline
Gentamicin C1 EnrofloxacinErythromycin
Currently, 219 vet. drugs (including >100 antibiotics) are on our list,but have targeted and evaluated ≈180 so far in (UHP)LC-MS/MS.
Dihydrostreptomycin
Streptomycin
Hygromycin
Spectinomycin hydrate
Spectinomycin
Neomycin
Gentamicin (C2+C2a)
Apramycin
Amikacin
Kanamycin
UHPLC w/ agent added
50 mM sodium 1-heptanesulfate in final extract
Aminoglycosides Multiclass, Multiresidues
2 g tissue + 20 mL of 10 mM NH4OAc, 0.4 mM EDTA,2% trichloroacetic acid, and 0.5% NaCl in water + IS
2 g tissue + 10 mL 4/1 (v/v) acetonitrile/water + IS
Shake 5 min on pulsed vortex platform shaker (80% setting, max pulsation)
Centrifuge 3 min at 3700 rcf
Centrifuge 3 min at 3700 rcfTransfer 10.75 mL (1 g equiv. sample) to 15 mL tube
Condition 50 mg WCX* DPX† tips with 3 mL eachof methanol and water
Tissue equivalence 0.174 g/mL
(no cleanup)
407 µL extract(71 mg sample equiv.)
71 µL extract(71 mg sample equiv.)
+ 272 µL 138 mM sodium 1-heptanesulfate ion-pairing (IP)reagent in water/acetonitrile
Yields 95 mg/mL final extract for each method in 34/66 (v/v) acetonitrile/watercontaining 50 mM IP reagent and 0.85% HO2CH 4 µL injection = 0.38 mg equiv. sample on column
Table 1: Results for the veterinary drugs spiked at 0.5X, 1X, and 2X levels, n=10 each, in the bovine tissues; (tR = retention time); aminoglycosides in blue text.
Goal: Develop a multi-class, multi-residue method for analysis of LC- and GC- amenable pesticides as well as legacy and emerging environmental contaminants in food:
Pesticides
Polychlorinated biphenyls (PCBs), including dioxin-like PCB congeners
Polycyclic aromatic hydrocarbons (PAHs)
Polybrominated diphenyl ethers (PBDEs)
Novel alternate flame retardants (FRs)
150 pesticides + 69 environmental &
other contaminants
99 pesticides & internal/ QC standards
LPGC-MS/MS
UHPLC-MS/MS
55 overlapping pesticides
>240 Analytes in Parallel using 10 min Analyses
See Han et al. (2016) Food Control 66, 270-282
Fast Low-Pressure (LP)GC-MS/MS
MS(/MS)
GC Oven
Injector
Restriction Capillary
Mega-BoreColumn
5 m
0.18mm
15 m 0.53 mm 1 m xx-5ms
No special adaptations needed; can be implemented in any GC-MS(/MS).
Review of dozens of publications using LPGC-MS(/MS):Sapozhnikova and Lehotay, Anal. Chim. Acta 899, (2015) 13-22
LPGC-MS is Much Faster
600 800 1000 1200 1400 1600 1800
2e+006
4e+006
6e+006
8e+006
1e+007
Time (s)
100 150 200 250 300 350 400 450 500
2.5e+006
5e+006
7.5e+006
1e+007
1.25e+007
1.5e+007
1.75e+007
2e+007
2.25e+007
Time (s)
25 min
LPGC-MS
Traditional GC-MS
and more sensitiveIs fast micro-bore with back-flush better? Watch out for carry-over.
1. New GC-QQQ provided high-speed, sensitive, selective, and
precise detection of nearly all targeted analytes at <10 ng/g
LOQ/LOI using 9:1 SPLIT INJECTION. Merely 100 µg equivalent
sample introduced! BEST PERFORMANCE, LONG-TERM!
2. LPGC-MS(/MS) is a faster and more rugged technique shown
to give taller peaks with less tailing (and lower LOQs) than
conventional GC-MS(/MS). MEGA-BORE = MEGA-RUGGED!
3. The micro-bore restrictor is inserted ≈1 cm into the mega-bore
analytical column, achieving zero dead volume, and metal
ferrules are used for a leak-free system. COMMERCIALLY
AVAILABLE AS CUSTOM ORDER AT NORMAL COLUMN COST.
4. Sandwich injection (air gap) using 5 µL syringe to ensure full
sample introduction and avoid analyte discrimination.
Aspects in the LPGC-MS/MS Analysis
5. A 1 m integrated guard column goes into the hot transfer line,
which avoids damage to the stationary phase from H2O/O2 in
every extract, even when the oven is cool. This improves long-
term chromatographic performance and reduces column
bleed.
6. An insert reduced oven volume, and 220 V rapid heating
option installed. Oven program of 50°C/min ensured that
the temperature was within consistent control.
7. Ion source and final oven temperature were 320°C (4.6 min)
to reduce ghost peaks and keep system clean. Post-sequence
oven at 250°C help, too.
8. MS/MS dwell times set to >7 points across analyte peaks
yielded greater analytical accuracy than using fewer points.
Aspects in the LPGC-MS/MS Analysis
9. Final extracts contained 0.1% HCO2H to improve stability of
base-sensitive pesticides, such as chlorothalonil.
10. Sample preparation used automated mini-cartridge SPE to
provide better cleanup than dispersive-SPE.
11. Analyte protectants were added to all extracts to coat active
sites in the liner, columns, transfer line, and ion source, even
if matrix material has built up from previous injections.
Aspects in the LPGC-MS/MS Analysis
HO O
OH
O O
OH
ethylglycerol
1 mg/ mL
ethylglycerol
1 mg/ mL
O
HO
HO
O
OHHO
O
HO
HO
O
OHHO
gulonolactone
0.1 mg/ mL
gulonolactone
0.1 mg/ mL
HO OH
OH OH
OH
HO OH
OH OH
OH
sorbitol
0.1 mg/ mL
sorbitol
0.1 mg/ mL
shikimic acid
0.05 mg/ mL
shikimic acid
0.05 mg/ mL
HO
HO
HO
OH
O
2.5 0.25 0.25 0.125
Effect of Analyte Protectants
w/ analyte protectants
w/o analyte protectants
NOTE: BE SURE TO WASH SYRINGE WELL WITH H2O IN SOLVENT MIX
See: Mastovska et al., Anal. Chem. 77 (2005) 8129-8137
Slope = 5.3% ME/tR
Slope = -0.5% ME/tR
-75%
-60%
-45%
-30%
-15%
0%
15%
30%
45%
60%
75%
90%
105%
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Avg
Mat
rix
Effe
ct (
ME)
Retention Time (min)
Avg MEs among the 4 Matrices for the Analytes
ME vs. QC Std
ME vs. Int. Std
Oh, and we may have eliminated matrix effects in GC-MS …
… via use of appropriate int. stds + analyte protectants in split inj’n
-0.04
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00
Dif
f. in
Ret
. Tim
e f
rom
Rgt
Day
1 (
min
)
Retention Time (min)
Cattle Day 1 Chicken Day 2 Pork Day 3
Rgt Day 2 Rgt Day 3
All of which leads to very consistent peaks and tR!3-Day Validation Experiment of 202 Pesticides analyzed by LPGC-MS/MS
40 matrix (muscle) spks and blks + QC = 70 injections per dayAvg tR (min) of reagent stds and matrices throughout the run (SD <0.040)
Analyte Protectants Added to all Final Extracts
phenanthrene
azinphos
dibenz(ah)-anthracene
Don’t Trust the “Advanced” Software
And don’t trust the analyst, either. This mistake was caught after preparing the previous slide for this presentation.
× ×
Int. Std.
Quant.Ion
Qual.Ion
RESOLVED: Human review takes too long!
High-throughput (or even low-throughput) multi-analyte monitoring applications:
G.F. Pang et al. (Beijing, China) include 1,138 pesticides in their GC- and LC- MS/MS monitoring approach. Large team of chemists conduct analyses and review results.
USDA: 240 analytes × 2-4 ion transitions × 50 samples/batch = 36,000 peaks!Analyst review and re-integration at 1 s per peak = 10 hours w/o breaks
on each instrument!
Summation Integration in Chromatography
SIMPLIFY, don’t COMPLIFY!
• Draw a straight line at the baseline just before the start of the expected peak to just after its expected end EASY PEASY!
• e.g. Elkin et al. “Computer-controlled mass fragmentographywith digital signal processing” J. Chromatogr. 81 (1973) 47-55
Rules in Automatic Post-Run Identification(e.g. in Excel or Instrument Software)
1) Ret. time (tR) for each ion (Quant. and Qual.) must be ≤|0.1| min from the contemporaneous tR(ref.)
2) Ion Ratio (IR) = (peak area ion 2)/(peak area ion 1), 3/1, 4/1, etc. (in %); IR(ref.) and tR(ref.) = avg. of contemporaneous high conc. calibration stds in solvent [note: IR(ref.) ≤ 110%]
IR must be |±10| for ≥1 ion or |±20| for ≥2 ions vs. IR(ref.)
3) Conc. must be > reporting level
Conc.(ng/g)
Peak Heights Peak AreasMeCN Matrix MeCN Matrix
2 53% 53% 58% 57%8 16% 18% 20% 24%
32 2% 4% 4% 6%128 1% 10% 10% 17%512 0% 0% 0% 1%
False Negatives Among the ≈700 Calibrations
The true positive identification rates for the calibration stdsat each level are 100% minus the false negatives shown.
Instrument Top Sample Preparation (ITSP)Morris and Schriner (2015) “Development of an automated column solid-phase extraction cleanup of QuEChERS extracts, using a zirconia-based sorbent, for pesticide residue analyses by LC-MS/MS” J. Agric. Food Chem. 63, 5107-5119
www.nacrw.org/2014/presentations/O21-Morris.pdf
Determined performance results in the use of automated mini-SPE cleanup in the LPGC-MS/MS analysis of pesticides and other contaminants in QuEChERS extracts of 10 different matrices.
Used mini-cartridgesshowing removal of chlorophyll and othermatrix components
Robotic liquid handler:3 min cleanup step at 2 µL/s+ 5 min for addition of APs and switching/washing syringes
Final extract volumes = 278 ± 5 µL (n = 255) after 50 µL addition of APs (and/MeCN) solution
See Lehotay et al. (2016) Chromatographia, 79, 1113-1130
ITSP+LPGC-MS/MS takes 13 min per injection cycle
FDA Sampling for Pesticides
• < 25 g units (berries) 1 kg (2.2 lbs)
• 25 – 250 g (apples) 1 kg (≥ 10 units)
• > 250 g (cabbage) 2 kg (≥ 5 units)
• Grains, Tree Nuts 1 kg
• Herbs 0.5 kg
• Spices 0.1 kg
Slide adapted from Jo Marie Cook
CODEX: 1 kg (2.2 lbs)
Pesticide Data Program: 3–5 lbs fresh, 2 lbs processed
USDA-FSIS: 1 lb meat, poultry, fish
Cryogenic Sample Processing
fried bacon
Spex FreezerMill(Cryomill)
Comminuted BroccoliBlixer Blixer + Cryomill
250X magnification
Robot Coupe Blixer has a spatula in the lid to ease and improve comminution
Conclusion: Cryomill = Overkill
250X magnification
Incurred Residues in Peach
0.0
2.0
4.0
6.0
8.0
10.0
1 g 2 g 5 g 10 g 15 g
Co
nce
ntr
atio
n (
ng
/g)
tetrahydrophthalimideBlixer
Cryomill
0
20
40
60
80
100
1 g 2 g 5 g 10 g 15 g
Co
nce
ntr
atio
n (
ng/
g)
fludioxynil Blixer
Cryomill
0
2
4
6
8
10
12
14
16
18
20
1 g 2 g 5 g 10 g 15 g
Co
nce
ntr
ati
on
(n
g/g
)
propiconazoleBlixer
Cryomill
0
2
4
6
8
10
12
14
16
18
20
1 g 2 g 5 g 10 g 15 g
Co
nce
ntr
atio
n (
ng/
g)
azinphos-methylBlixer
Cryomill
Error Contribution due to Sample Processingvs. Test Portion (Subsample Size)
0%
5%
10%
15%
20%
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
RSD
fro
m S
amp
le P
roce
ssin
g
Test Portion Size (g)
Calculated from all residues in 10 commodities processed with Blixer and Cryomill
Conclusions
• Aminoglycosides can be analyzed at the same time as other veterinary drugs by adding the ion-pairing reagent to combined final extracts from two sample preparation methods.
• Smaller test portions (2- 5 g) in QuEChERS are possible using the Blixer for many (but not all) commodities.
• High quality, rugged results can be achieved for hundreds of ultratrace analytes in diverse foods using automated high-throughput analysis by QuEChERS + ITSP+LPGC-MS/MS and UHPLC-MS/MS in parallel without matrix-matched calibration followed by summation function chromatographic peak integrations and post-run processing to yield accurate determinations and trustworthy identifications with little need for human review.
• Compositional Heterogeneity and Fundamental Error
• Distributional Heterogeneity
• Sample Correctness and Tools
Slide adapted from Jo Marie Cook
Sample Processing w/ Dry Ice
Experimental
Sample preparation (final method): 5-8
10 g homogenized fish + internal standards
Add 10 mL MeCN and shake 10 min on vortex shaker
at 80% setting with max. pulsing
Add 5 g HCO2NH4, shake 1 min,centrifuge 2 min at 3700 rcf
Filter-vial dispersive-SPE:•Add 0.5 mL extract to the PVDF (0.2 µm) filter-vial shell containing 75 mg each anh. MgSO4 + 1/1/1 PSA/C18/Z-Sep• Partially depress the filter-vial plunger and shake for 30 s in an autosampler vial tray• Fully depress the plunger into the filter-vial shell
Sample Preparation
We have validated the approach for analysis of beef, pork, and chicken muscle for implementation by USDA-FSIS
Extraction of Incurred Fish Samples
Extraction of Incurred Fish Samples
Extraction of Incurred Fish Samples
RESOLVED: Good analysts are able to conduct peak
integrations better than current advanced software tools (but good analysts are hard to find, earn wages, get bored reviewing data, and still make misteaks).
RESOLVED: Despite technology and software advancements,
no set of peak integration parameters works consistently for all analytes, concentrations, and matrices in the real-world (at least not yet in my experience).
RESOLVED: Garbage In = Garbage Out• Correct and consistent chromatographic peak integration is
essential to achieving high quality final results.
What the Heck?
× ?Quant.
IonQual.Ion
100 ng/gAzinphos
Analytical chemists are hard-working lazy people who look for short-cuts rather than make manual re-integrations.
Real samples can’t be dropped as outliers.
Integrations of replicate injections were correct 2 out of 3 times
In analytical chemistry, 2 out of 3 ain’t good
In chromatography, the primary parameters are ret. time (tR) and peak shape (width, height/area)
If tR and peak widths are so important and consistent in good methods, why do most (all?) sophisticated (and expensive)
chromatographic peak integration software programs so often choose peaks at the wrong tR with quite variable peak shapes?
Retention Times and Peak Widthsare Rock Solid in UHPLC-MS/MS
# AnalyteDay 1 = 7/17/15 Day 2 = 7/22/15 Day 3 = 7/28/15
3-Day Validation Experiment of 101 Pesticides analyzed by UHPLC-MS/MS40 matrix (muscle) spks and blks + QC = 65 injections per dayAvg tR (min) of reagent stds and matrices throughout the run (SD <0.020)
New mobile phase added for each sequence
Summation Integration Function• ≈1 min to integrate a batch of >60 samples of
≈660 MRMs per sample WITHOUT REVIEW!
• This is a >40 year-old integration function, but LACKING IN SOME DATA PROCESSING SYSTEMS!
% of times results in range
Peak Height Results Peak Area Results
LOQ (ng/g) LOI (ng/g) LOQ (ng/g) LOI (ng/g)
MeCN Matrix MeCN Matrix MeCN Matrix MeCN Matrix
< 1 ng/g 38% 20% 30% 8% 27% 11% 21% 3%
1 - 10 ng/g 54% 70% 56% 59% 62% 78% 60% 55%
10 - 25 ng/g 5% 6% 9% 17% 7% 7% 12% 21%
25 - 100 ng/g 3% 4% 3% 15% 4% 4% 6% 17%
> 100 ng/g 0% 0% 2% 1% 0% 0% 1% 3%
Table III. Average limits of quantification and identification (LOQs and LOIs) ± standard deviations for the pesticide analytes in the 10 commodities analyzed by LPGC-MS/MS on 10 different days using peak heights or areas in summation integration. Bold text indicates when LOI < LOQ and italics signify the lower average between the paired result based on peak height or area.