Agilent Technologies at TIAFT 2013

Agilent Technologies at TIAFT 2013 Agilent Technologies GC/MS portfolio for the growing challenge of designer drugs analysis Juan Aybar Gas Chromatography Marketing Jaume Morales MS Product Specialist Funchal, Madeira September 2-6 September 12, 2013

Outline 1. Designer drugs in perspective 2. Agilent GC & GCMS Portfolio 3. New Target Deconvolution Workflow 4. High resolution GCMS possibilities 5. What s 7200 GCQTOF 6. Example of Street Heroin Classification and Characterization 2

Designer Drugs in Perspective Designer Stimulants represent an increasingly important health risk due to several factors: 1. Rapid changes in the market and accesibility. Availability of analogues. Internet access 2. Lack of Quality Control 3. Suggested dosages and reported effects not always accurate 4. Marketing General sense of legal safety (you ll pass the drugs tests!!), safety, cheap and availability 3

Designer Drugs Chemical Problem 1. Complexity of Samples 1. Sample Preparation Techniques 2. Multiple analytes, Similar analytes (isomers, analogues, similar spectral cleavage ) 3. Instrument protection vs contaminants and interferents 2. Quantify the known substances 1. MS/MS Quadrupolar Techniques 3. Identify the unknown 1. Search Multiple Databases Spectral similarity 2. Examine fragmentation 3. Deconvolution / Accurate Mass 4. Reference materials availability 5. Create home-made databases spectra/rt of emerging materials 6. Constant evaluation of unidentified spectra / chromatogram peaks 7. 4

Agilent GC and GCMS Portfolio 1. Sample Preparation Workbench 1. Simplifies and accelerates sample prearation including tasks like dilution, derivatizacion, mixing, alliquoting 2. Agilent 7890B w/ Capillary Flow Technology 1. Helps prevent contamination in the MS source by facilitating backflushing, facilitated column exchange or deans switching 3. Agilent 5977 GC/MS 1. MS Spectral deconvolution to confirm expected compounds and discover unexpected analytes 4. Agilent 7000C GC/TQ 1. Most sensitive MS/MS Quantitation with excellent matrix resistence for trace analyte determination and facilitated SW tools with MassHunter SW 5. Agilent 7200 GC/Q-TOF 1. High Resolution/Accurate Mass possibilities with excellent sensitivity for low level compounds 6. Agilent Pre-Configured Analyzer and Kits for forensics and toxicology application 5

Synthetic Cannabinoids Example Problems: Low levels, Multiple Analytes, Similar Analytes, Similar Fragmentation, Coextractants, Derivatization DECONVOLUTION BENEFITS Underivatized Tailing Worst integration GCMS Synthetic Cannabinoid Sample BSTFA Derivatized No Tailing Better Integration Better identification and confirmation of analytes in high matrix samples Reduces data review time, especially for large screening methods Reduces false positives and false negatives in dirty samples Identification based on matching entire spectrum cleaned of interferences against library. Much more reliable than target/qualifier ratio method 6

How Does Deconvolution Work? Ions with the same abundance vs time profile are grouped together to create spectra cleaned of interferences from overlapped peaks Confidentiality Label September 12, 2013 7

Agilent 5977 GC/MS New Features Target Deconvolution TARGET DECONVOLUTION MassHunter Deconvolution New feature fully integrated into Mass Hunter Quant B.06.00 Provides additional confidence in the identification of target analytes by using deconvoluted full scan mass spectra Legacy MSD Chemstation Methods and data files can be converted for use in MH Interactive data review using Mass Hunter Batch at a Glance (BAG) Rapid data review using Compounds at a Glance (CAG) and Outlier flagging Very fast.pdf reporting of individual samples or entire batch Confidentiality Label September 12, 2013 8

Target Deconvolution Workflow in Mass Hunter Quant SW TIC Targets are identified by comparison to locked RTs and 1 Quant + 3 Qualifier ion ratios, then quantified using Quant ion area versus calibration table MH deconvolutes component spectra and performs spectral matching of deconvoluted spectra vs target MS database using RT Window and Library Match Score as qualifiers All data processing performed within Mass Hunter Quant SW Quant Results Qualitative Results Review results in Mass Hunter Batch-at-a-Glance and Compounds-at-a-Glance Combined quantitative and qualitative Target Deconvolution.pdf Summary report Confidentiality Label September 12, 2013 9

Extracted Compound Chromatograms Toluene 4.95 min. The peak apex of the deconvoluted component must be within the RT Range of the target peak identified by the Quant Engine in order to be used for library matching. If more than one component meets these criteria, consider only the component spectrum with the highest peak area for the deconvoluted quant ion EIC. EIC ECC RT : Seconds 285 290 295 300 305 310 315-10 seconds + 10 seconds Target RT Window econvoluted component spectrum Deconvoluted component spectrum Deconvoluted component Deconvoluted component spectrum Deconvoluted component spectrum spectrum

Deconvolution Report

DRS versus MH TD : Summary Comparison Function DRS Target Deconvolution SW Platform MS Chemstation Mass Hunter Quant B.06.00 Quantitation 1 target ion + up to 3 qualifier ions 1 target ion + up to 3 qualifier ions Full MS Deconvolution Yes Yes Identification Additional identification with NIST Library search Interactive data review Quant results using raw and deconvoluted data RT match + Library search vs target MS library Yes QEDIT Yes RT match + Library match vs target MS library No Batch at a Glance Compounds at a Glance Outlier flagging Yes Reporting HTML.pdf

Agilent 7200 GC/ Q-TOF Possibilities of High Resolution GCMS September 12, 2013 13

Benefits of GCQTOF High resolution full scan spectra > 15K versus < 1K for SQ (15-20X higher) Higher selectivity without MS/MS lower requirement for hi-res GC separations. Avoids the operational cost and complexity of GCxGC. Accurate mass measurements < 5 ppm versus 350-400 for SQ (70-80X better) Better qualitative decisions (molecular formula information) for both MS and MS/MS spectra Fast acquisition of full spectra range 50 Hz max versus typically < 5 Hz max for SQ With consistently greater sensitivity than SQ MS/MS with Product Ion spectra More selective than TQ due to higher resolution With accurate mass information for each product ion Powerful structural elucidation tools Ideal tools for solving complex analytical problems

How much R is enough? Intensity Ratio = 1:1 Flurenol methyl ester m/z = 240.0781 Mass error = 1.7 ppm COELUTING COMPOUNDS & SIMILAR m/z 240.1218 240.0785 Dimetilan m/z = 240.1217 Mass error = 0.4 ppm No IRM corrections applied Dm = 0.043 Da. ~13,500 resolution FWHM

How much R is enough? Intensity Ratio = 1:0.02 COELUTING COMPOUNDS & SIMILAR m/z Flurenol methyl ester m/z = 240.0781 Mass error = -0.4 ppm 240.0780 A resolution of 13,500 is not enough - larger ratios need MS/MS or better GC separation Dimetilan m/z = 240.1217 Mass error = -13.7 ppm 240.1184

GCMS Challenge : complex analytical problems 1 st Approach : The TOF-MS delivers high mass accuracy full scan GC-TOF-MS has unrivalled full spectrum sensitivity, comparable to GC- MS in SIM mode. GC-TOF-MS allows accurate mass measurement for higher selectivity and sensitivity using a very narrow mass window (0.02-0.05Da). BUT, what happens when Resolution and Δ Mass are not enough? 2 nd Approach : GC-QTOF-MS

GCMS Challenge : complex analytical problems 2 nd Approach : The QTOF-MS can : - Reduce noise by Precursor Selection thus delivering more selectivity. - Confirm ID with extra Hi-Res MS/MS spectra. -Allow structural Elucidation.

The Problem Confirm Most Likely Structure Kava Extract - Compound B, C 16 H 14 O 4 (Rings + Double Bonds = 10) EI Full Scan (M H) + 269.0802 Candidate structures

Formula Calculator Determine all possible formulas consistent with measured mass C 5 H 12 O 2 PS 3 m/z = 230.9732

The Problem Confirm Most Likely Structure Kava Extract - Compound B, C 16 H 14 O 4 (Rings + Double Bonds = 10) EI Full Scan (M H) + 269.0802 Candidate structures H MS/MS experimental measurements m/z (experimental) Formula Error (ppm) Score 269.0802 C 16 H 13 O 4 2.2 80.7 C 6 H 5 CH=CH C 6 H 5 CH 2 =CH C 6 H 5 CO CO CH 3 193.0494 C 10 H 9 O 4 0.6 96.7 167.0334 C 8 H 7 O 4 3.0 N/A 166.0259 C 8 H 6 O 4 0.6 N/A 138.0310 C 7 H 6 O 3 1.1 98.1 110.0359 C 6 H 6 O 2 3.0 N/A 95.0127 C 5 H 3 O 2 0.9 99.5

The Problem Confirm Most Likely Structure Kava Extract - Compound B, C 16 H 14 O 4 (Rings + Double Bonds = 10) EI Full Scan (M H) + 269.0802 Candidate structures X H C 6 H 5 CH=CH C 6 H 5 MS/MS experimental measurements m/z (experimental) Formula Error (ppm) Score 269.0802 C 16 H 13 O 4 2.2 80.7 193.0494 C 10 H 9 O 4 0.6 96.7 167.0334 C 8 H 7 O 4 3.0 N/A X CH 2 =CH C 6 H 5 CO CO CH 3 166.0259 C 8 H 6 O 4 0.6 N/A 138.0310 C 7 H 6 O 3 1.1 98.1 110.0359 C 6 H 6 O 2 3.0 N/A 95.0127 C 5 H 3 O 2 0.9 99.5

Problem confirm most likely structure EI Full Scan (M H) + 269.0802 Candidate structures H MS/MS experimental measurements m/z (experimental) Formula Error (ppm) Score 269.0802 C 16 H 13 O 4 2.2 80.7 C 6 H 5 193.0494 C 10 H 9 O 4 0.6 96.7 CH=CH C 6 H 5 167.0334 C 8 H 7 O 4 3.0 N/A CH 2 =CH C 6 H 5 CO 166.0259 C 8 H 6 O 4 0.6 N/A 138.0310 C 7 H 6 O 3 1.1 98.1 CO 110.0359 C 6 H 6 O 2 3.0 N/A CH 3 95.0127 C 5 H 3 O 2 0.9 99.5 For the 5 candidate structures, only one fit the losses identified by CID experiments on multiple precursor ions

What is GCQTOF? 7890 + 7000 + 6500 = 7200 GC/Q-TOF Triple Quadrupole GC/MS + = Time of Flight MS Quadrupole Time of Flight MS

7200 Q-TOF Comp NEW Removable Ion Source 7000B TQ NEW Optics 6500 Q-TOF Two 300 L/s Turbos

New Q-TOF System... Yet Totally Proven 4GHz ADC electronics enable a high sampling rate (32 Gbit/s) which improves the resolution, mass accuracy, and sensitivity for low-abundance samples. Dual gain amplifiers simultaneously process detector signals through both lowgain and high gain channels, extending the dynamic range to 105. Hot, quartz monolithic quadrupole analyzer and collision cell identical to the 7000 Quadrupole MS/MS Dual-stage ion mirror improves second-order time focusing for high mass resolution. Proprietary INVAR flight tube sealed in a vacuum-insulated shell eliminates thermal mass drift due to temperature changes to maintain excellent mass accuracy, 24/7. Added length improves mass resolution. Analog-to-digital (ADC) Detector: Unlike time-to-digital (TDC) detectors which record single ion events, ADC detection records multiple ion events, allowing very accurate mass assignments over a wide mass range and dynamic range of concentrations. New Removable Ion Source includes repeller, ion volume, extraction lens and dual filaments Two 300L/s t urbos pump the focusing optics and flight tube Split-flow turbo differentially pumps the ion source and quadrupole analyzer compartments Hexapole collision cell accelerates ion through the cell to enable faster generation of high-quality MS/MS spectra without cross-talk

Removable Ion Source (RIS) Automated Gate Valve RIS Automated Retractable Transfer Line

Removable Ion Source (RIS) RIS Standard on Q-TOF Allows fast EI/CI source swapping without venting Allows swap of complete ion source, including filaments, in ~30 minutes without venting

Different Sample Introduction systems 7693A Injector Thermal Separation Probe CTC with Liquid/Headspace/SPME Headspace Sampler 7697A Thermodesorber

Different Ionization Sources EI Hard ionization technique, extensive fragmentation, limited Molec.ion. Very reproducible, library search allows Compound ID by fragments. PCI Compounds must be able to accept a proton. Amines, alcohols, Very strong molecular ion allows Compound ID by exact mass. NCI Compounds must be able to accept an electron. Halogens, Conj.Systems, phenyl rings Very strong molecular ion allows Compound ID by exact mass.

Capillary Flow Technology- Possibilities Quick Swap Solvent Venting Heart Cutting (Deans Switch) Split to several Detectors Backflushing

Characterization and Classification of Heroin from Illicit Heroin Seizures by GC/Q-TOF 1 Koluntaev Dmitry, 2 Sofia Aronova, 3 Sergei Syromyatnikov and 3 Igor Sarychev 1 InterLab Inc., Moscow, Russia 2 Agilent Technologies, Santa Clara, CA 3 Federal Drug Control Service of Russian Federation, Moscow, Russia 32

Introduction "Street heroin is usually a multi-component mixture comprising, apart from the heroin, of various additives such as pharmacologically active compounds as well as neutral substances. The study benefited from the accurate mass high resolution capability of the Agilent GC/Q-TOF system particularly useful in the identification of the composition of the heroin samples. The characteristic profiles of heroin samples were the basis for further statistical analysis performed in Mass Profiler Professional (MPP). Comparative study of heroin samples and their characterization and classification would help establishing common sources of origin. 33

Experimental: Data Acquisition Two different acquisition methods were utilized to separately acquire data for major and minor components of the samples. The method for major compounds utilized reduced emission current on a filament in order to avoid saturation. The method used to acquire data for minor compounds, had higher emission current but decreased ionization energy during elution of major components. 34

Experimental: Data Acquisition & Data Analysis Two different acquisition methods were utilized to separately acquire data for major and minor components of the samples. The chromatographic deconvolution was performed using MassHunter Unknown Analysis software. Impurities of interest were identified by comparison with the NIST11 mass spectral library. Multivariate statistical software was performed with Mass Profiler Professional (MPP) to find compounds present at distinct levels in different groups of samples. The data were subsequently used to build a classification model. 35

Identification of the Common Components by MPP Most common components of the heroin samples are morphine alkaloids and morphine derivatives. 54 out of 55 samples contained a monoacetyl derivative of morphine: 6- monoacetylmorphine and acetylcodeine. Many other common alkaloids as well as pharmacologically active substances were also detected in the majority of the samples. Alkaloidal Adulterants Noscapine 50 Caffeine 55 Papaverine 50 Dextromethorphan 27 Meconine 43 Tolycaine 18 Morphine 40 Paracetamol 5 Hydrocotarnine 13 Codeine 10 Alkaloidal impurities and non-opiate pharmacologically active cutting agents identified in heroin samples. 36

Library Hits Confirmation Using Accurate Mass For few compounds that did not give a high library match score we performed additional confirmation steps using accurate mass information as well as MassHunter Qualitative Analysis structure elucidation tools. One compound present in 36 out of 55 samples was tentatively identified as 6-acetyl-crotonosine acetate. The identity was evaluated using Molecular Formula Generator (MFG) with library search and Fragment Formula Annotation (FFA) tools. 37

Compound Confirmation Using MFG with FFA and Library Search Annotated mass spectrum of a compound tentatively identified as 6-acetyl-crotonosine acetate. 38

Molecular Structure Correlator (MSC): Another Tool to Evaluate and Confirm Possible Structures of the Compound empirical formula Selected fragment possible substructures corresponding to the fragments evaluated structure Fragment formulas Molecular Structure Correlator (MSC) works with accurate mass MS/MS data to predict and evaluate possible structures of the compound of interest. It can mine a database, e.g. ChemSpider to extract structures corresponding the empirical formula of the compound and rank them according to the compatibility score. 39

PCA Analysis: Major Components No significant separation between the analyzed sample groups was observed when the GC/Q-TOF method for major components. 40

PCA Analysis: Minor Components The data acquired using the method for minor components displayed significant separation into at least two major groups. 41

Hierarchical Clustering Analysis The compounds that contributed to the major difference between isolated sample groups were further visualized using Hierarchical Clustering. Specifically, we were able to observe multiple alkaloids of heroin as well as pharmacologically active agents that contributed into the sample clustering. Hierarchical clustering also confirmed the presence of two major sample groups. 42

Hierarchical Clustering Analysis Tolycaine Dextromethorphane p-isopropoxyaniline Morphinone, 7,8-dihydro-3-desoxy- Meconine Hydrocotarnine Papaverine Codeine Hierarchical Cluster Analysis (HCA) demonstrated separation of the samples into few distinct groups. Shown are few characteristic compounds that are likely to contribute into separation of the samples. 43

Building Class Prediction Model in MPP Two major clusters were further used to construct a Class Prediction Model in MPP 44

Building Class Prediction Model in MPP Volcano plot showing compounds that present at significantly different levels between two groups of samples 45

Validation of the Class Prediction Model To build class prediction model using Partial Least Square Discrimination (PLSD) algorithm all samples that belong to the two major groups excluding all H and Q replicates were used. 46

Prediction Results The samples that were left out during validation and training of the model (outlined in rectangle) were all predicted correctly by applying the class prediction model. 47

Summary Multiple statistical approaches using MPP were utilized to isolate the heroin samples, analyzed by GC/Q-TOF, into distinct clusters Class Prediction model was built in MPP based on two largest groups of heroin samples Accurate mass information in combination with multiple MassHunter software tools helped to confirm the identity of alkaloids and other pharmacologically active compounds found in the heroine samples 48

Thank you for your attention! Any Questions? Page 49