28 BioPharm NOVEMBER 2001

Process Development

selection of the product (daughter) ion in thethird quadrupole after the collision of thefather ion in a collision cell (the secondquadrupole). That mode of ion selection anddetection is called selected reactionmonitoring (SRM).

Joining MS and LC. MS operates in a high-vacuum environment. In contrast, today’spremium separation technology, liquidchromatography (LC), is performed underatmospheric pressure. Historically, itsaqueous mobile-phase flow rate wasincompatible with MS. However, innovativeand successful research efforts on the designof an effective interface connection betweenLC and MS over the past 25 years havemade LC compatible with MS. Electrosprayionization (ESI) and atmospheric-pressurechemical ionization (APCI), collectivelycalled atmospheric pressure ionization(API), have matured into the reliableinterface necessary for quantitativeLC/MS/MS bioanalysis.

Biological systems are complicated toanalyze, and the development of robustLC/MS/MS methods suitable for routinelyanalyzing thousands of samples remains adifficult, time-consuming task. Thepopularity and growth of LC/MS/MS rose soquickly in the early 1990s that some users ofthe method abandoned sound techniques ofsample preparation and clean-up, discardedthe basic principles of chromatography, andneglected the fundamentals of chemistry.Such oversights ultimately created problemsattributed to the analytical method. ManyLC/MS/MS challenges have been reviewed,discussed, and published (3,4).

LC/MS/MS remains one of the mostuseful tools available for bioanalysis.Successful use requires understanding theextraction process and the underlyingprinciples of both chromatography and MS.A systematic approach for developing robustmethods and automation (wherever possible)is necessary to realize long-term benefits and

Systematic Troubleshooting for LC/MS/MSPart 1: Sample Preparation and Chromatography

LC/MS/MS provides superiorsensitivity and selectivity, rapidanalysis, maximized developmentefficiencies, and improvedturnaround times. Although large-scale LC/MS/MS analysis isfraught with challenges, you canlearn to overcome the obstacleswith careful planning and thesetroubleshooting techniques. Part 1presents troubleshootingtechniques related to samplepreparation and chromatography.

Development and validation ofbioanalytical methods forpharmaceutical product analysis arecommon rate-limiting steps inbiopharmaceutical product

development. A rational, strategic approachto developing robust, automated, validatedbioanalytical methods can reduceslowdowns and bottlenecks in drugdevelopment and contribute to synergistic,consistent, long-term performance.

Use of liquid chromatography withtandem mass spectrometry detection(LC/MS/MS) for bioanalysis has grownexponentially in the pharmaceutical industrysince the 1980s. The popularity of thismethod is attributed to its superiorsensitivity, extraordinary selectivity, andrapid rate of analysis. Adoption ofLC/MS/MS has been driven by the need fortimely, high-quality data at various stages ofthe drug development process: from high-throughput screening of drug candidates andprompt data generation for preclinicalstudies to almost “real-time” analysis ofclinical samples (1). Rapid and rationalLC/MS/MS methods play an important rolein bioanalytical sample analysis andshortening cycle times forbiopharmaceutical product development (2).

LC/MS/MS RationaleThe underlying principle of MS is theproduction of ions from analyzedcompounds that are then separated orfiltered based on their mass-to-charge ratio(m/z) and detected in a spectrometer. Themass spectrum generated plots theabundance of the produced ions as afunction of m/z. The most dominantapplications for quantitative bioanalysisemploy tandem mass spectrometers(MS/MS) that use a triple quadrupoleinstrument. Two mass analyzers are used:one for selection of the precursor (father)ion in the first quadrupole, and the other for

Naidong Weng and Timothy D.J. Halls

Corresponding author Naidong Weng is associatedirector of bioanalytical chemistry, and Timothy D.J. Halls is vice president ofpharmaceutical chemistry at Covance LaboratoriesInc., 3301 Kinsman Boulevard, Madison, WI 53704,608.242.2652, fax 608.242.2735,naidong.weng@covance.com, www.covance.com.

30 BioPharm NOVEMBER 2001

Process Development

efficiencies. This month we highlightrational method development and validationand includes troubleshooting tips for soundsample preparation, for analyte stability inbiological matrices, and for chromatographicconditions and injection solvents. Part 2 willfurther discuss development strategies forLC/MS/MS analysis of biopharmaceuticalproducts including proteins and peptides. Itwill present troubleshooting tips for carry-over, recovery and matrix effects, andselectivity; and for automation wherepossible to improve the overall reliability andefficiency of LC/MS/MS.

Sound Sample Preparation The role of sample clean-up in LC/MS/MSis critical. In addition to assessment ofanalytes in the usual picogram/mL tonanogram/mL range, biological sampleanalysis is complicated because the samplescontain macromolecules and othercompounds such as proteins, endogenousand exogenous compounds, andcoadministered pharmaceuticals(collectively referred to as unwantedcompounds). Those compounds will be athigher concentrations than the analyte.

The first step in sample clean-up is toremove as many of the unwanted compoundsas possible without significant loss of theanalytes of interest. Solid-phase extraction(SPE), liquid–liquid (LL) extraction, andprotein precipitation (PP) are often thetechniques of choice. Unwanted compoundscan be present still in higher concentrations

than the analytes of interest after the firstclean-up. A second stage of clean-up,typically involving LC separation furtherseparates analytes of interest from theunwanted compounds. MS/MS offers a thirdstage of separation through selection ofappropriate precursor and product ion pairsso that unwanted compounds are notregistered (unseen) by the detector.

However, those unwanted and MS/MSunseen compounds present significantchallenges for LC/MS/MS practitioners. In

the LC/MS interface, unwanted compoundscompete with analytes for ionization and cancause inconsistent matrix effects that aredetrimental to quantitative LC/MS/MS. If notseparated from the analytes, some conjugatedmetabolites break down in the interface sothat analyte concentration is overestimated.LC/MS/MS practitioners must use caution intheir analyses and envision that behind everyanalyte peak, “unseen” MS peaks fromcontaminants may cause or contribute topotential problems in the assessment.

Sample Analyte Extraction Species Matrix LLOQ–ULOQa Volumeb

Albuterol SPE human serum 0.05–10 0.4

Clonidine LL human serum 0.01–1 1

Fentanyl SPE human plasma 0.050–5 0.25

Fexofenadine SPE human plasma 5–2,500 0.05

Fluconazole LL human plasma 0.5–100 0.5

Fluoxetine LL human plasma 0.5–250 0.1Norfluoxetine 0.5–250

Hydrocodone SPE human plasma 0.1–100 0.3Hydromorphone 0.1–100

Ketoconazole LL human plasma 20–10,000 0.25

Loratadine LL human plasma 0.01–1 1Descarboethoxy-loratadine 0.025–2.5

Midazolam LL human plasma 0.1–100 0.41-OH midazolam 0.1–1004-OH midazolam 0.1–100

Morphine SPE human plasma 0.5–50Morphine-3-glucuronide 1–100Morphine-6-glucuronide 10–1,000

Nicotine LL human plasma 1–200 0.25Cotinine 10–2,000

Omeprazole LL human plasma 5–2,000 0.2

Oxycodone SPE human plasma 0.1–50 0.4Oxymorphone 0.1–50Noroxycodone 0.1–50

Paroxetine LL human plasma 0.05–50 0.4

Protease inhibitors PP human plasma 0.1Amprenavir 10–10,000Indinavir 10–10,000Nelfinavir 10–10,000Ritonavir 10–10,000Saquinavir 10–10,000

Pseudoephedrine SPE human plasma 10–2,500 0.1human urine 10–2,500 0.1

Ribavirin PP human plasma 10–10,000 0.1

Rosiglitazone LL human plasma 1–1,000 0.05

Sertraline SPE human plasma 0.25–100 0.25Desmethyl-sertraline 0.5–100

Sildenafil SPE human plasma 1–500 0.35Desmethyl-sildenafil 1–500

Triazolam LL human plasma 0.1–50 0.2

aUpper and lower limits of quantitation in ng/mL bin mL

Table 1. LC/MS/MS methods using silica columns and aqueous–organic mobile phases

Figure 1. Unexpected results from reversed-phase chromatography: retention (capacityfactor) of ketoconazole and its internalstandard R51012 on Hypersil BDS C18column (50 � 3 �m) using a mobile phaseof acetonitrile–water–formic acid (x:(100-x):1, v/v/v) where x is the percentage ofacetonitrile in the mobile phase) at a flowrate of 0.5 mL/min

6

5

4

3

2

1

040 50 60 70 80 90

KetoconazoleR51012

% Acetonitrile

Cap

acity

fact

or k

32 BioPharm NOVEMBER 2001

Process Development

the analyte, the column, and the mobilephase can interact with unexpected results.For example, Figure 1 shows that whenorganic content of the mobile phasecomposition was increased (a desirablecharacteristic for improving sensitivity), thecapacity factor (k�) decreased initially butincreased after the deflection point. Thatsystem demonstrated an initial reversed-phase mode with primary retention becauseof hydrophobic interaction between theanalytes and the alkyl chains. It reverted tonormal-phase where the primary retentionwas because of hydrophilic interactionbetween the analyte and the residual silanolgroups. That kind of dual retention behaviorwas observed for various analytes ondifferent brands of reversed-phase columns.Even the extensively end-capped reversed-phase column had at least 30%nonendcapped residual silanol groups.Analyte on-column retention, therefore,depends on the characteristics andinteractions of the analyte, the column, andthe mobile phase.

Polar ionic retention. For analysis of polarionic compounds, reversed-phaseLC/MS/MS can be problematic. Anionformation occurs when the pH is higher thanthe pKa in acidic compounds, and cationformation occurs when the pH is lower thanthe pKa of basic compounds. Columnretention decreases matrix effects. Ionizationof polar analytes decreases column-retention, and promotes matrix effects thatcan impact analysis bioanalysis. To retainpolar ionic compounds, highly aqueousmobile phases or ion-pair chromatography isrequired. However, neither of those methodsis conducive to the spray conditions required

Analyte StabilityCertain drug product analytes are subject todegradation. Endogenous enzymes foundwithin biological matrices can accelerateanalyte degradation. Information aboutdegraded analytes should be obtained duringmethod development. A generaltroubleshooting approach to address thatproblem would be stabilizing analytes bychoosing an appropriate anticoagulant, pH,or enzyme inhibitor. If analytes cannot bestabilized, then in situ derivatization of theanalytes to more stable forms in biologicalmatrices should be attempted.

Anticoagulants in the plasma can havesignificant stabilizing effects on the analyte.Some ester-containing analytes are unstablein plasma containing sodium heparin as theanticoagulant, but they may be relativelystable in plasma where sodium fluoride isthe anticoagulant (5).

Establishing analyte stability in thebiological matrix at an early stage of methoddevelopment is crucial. For example, when anew investigational drug containing esterfunctional groups was recently tested, usingsodium heparin as the anticoagulant, 98% ofthe analyte was degraded at roomtemperature. Decreasing the temperature to�20 °C resulted in 25% degradation by

plasma esterase ex vivo in 20 hours. Thesame analyte showed no degradation at allwhen the same experiment was conductedusing sodium fluoride as the anticoagulant.That information provided criticalinformation for the appropriate design of theclinical study protocols.

Some amine-containing nucleosides aremetabolized ex vivo by plasma deaminase.Addition of small amounts oftetrahydrouridine (THU) inhibit deaminaseactivity and stabilize the analytes (6). Thio-(sulfhydryl-containing) compounds areusually unstable in plasma. In situderivatization of thio- compounds with N-ethylmaleimide (NEM) (7) or with methylacrylate (MA) (8) lead to stable derivatives,which subsequently can be reliably analyzed.

Chromatographic ConditionsReversed-phase chromatography isuniversally applied in LC/MS/MS. Theadvantages of reversed-phase columns withmobile phases of various compositionsinclude excellent stability, columnefficiency, and versatility for a wide varietyof compounds.

On-column retention. Reversed-phasecolumns do not always behave in a“reversed-phase” way. The characteristics of

Figure 2. Silica column stability for a LC/MS/MS method of analyzing ribavirin in humanserum; column: Betasil silica 50 � 3 �m; mobile phase: acetonitrile–water–trifluoroaceticacid (TFA) (95:5:0.05, v/v/v); sample preparation: protein precipitation

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

RibavirinInjection #14

1.53(a)

0.410.55

0.77 1.09 1.39 1.78 2.052.17

2.34 2.622.51

25,000

20,000

15,000

10,000

5,000

Time, min

Inte

nsity

, cps

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6

RibavirinInjection #357

1.54(b)

0.380.27 0.58 0.86

0.74 1.30

1.09

1.76 2.082.20

2.40 2.6

13,000

11,000

9,000

7,000

5,000

3,000

1,000

Time, min

Inte

nsity

, cps

Retention Time (minutes)

AMP RIT NEL IS SAQ IND0.67 1.54 2.43 2.43 2.50 3.780.69 1.54 2.42 2.44 2.50 3.800.69 1.54 2.42 2.42 2.50 3.780.69 1.52 2.42 2.44 2.50 3.78

acolumn: Betasil silica 50 � 3 �m; mobilephase: linear gradient elution of acetonitrile–water–formic acid from 95:5:0.2 to 30:70:0.2in two minutes; flow rate: 0.5 mL/min;sample preparation: protein precipitation

bAMP: amprenavir; RIT: ritonavir; NEL: nelfinavir; IS: internal standard:SAQ: saquinavir; IND: indinavir

Table 2. Reproducibility of gradient elutionon silica column for four injections

Process Development

to achieve good LC/MS/MS sensitivity.Good spray conditions are necessary forstrong, stable analyte signals.

A number of drugs have basic functionalgroups, and MS in the positive ion modeoften detects such components as protonatedions. Protonation is the most important meansof ionization in positive-ion electrospraymass spectrometry. Acidic mobile phases

often are used to ensure that analytes are intheir protonated forms. However, the chargedanalytes will have poorer retention onreversed-phase columns. That is undesirablebecause detrimental matrix effects canobscure the analysis.

That problem can be overcome bycapturing polar ionic compounds on a silicacolumn (9). Depending on the conditions,sensitivity on silica columns can increasefive- to eightfold. Figure 1 illustrates dualretention on a reversed-phase columnthrough the use of silanol to retain polaranalytes. Various compound analytes (Table 1) have been successfully elutedusing a silica column (9–20). Basiccompounds are eluted with an acidic mobilephase to create cations for electrospray andion detection.

Silica columns demonstrate excellentstability. Typically, one column can be usedfor at least 500 injections of extractedsamples. Figure 2 shows column stability ofthe LC/MS/MS analysis of ribavirin inhuman plasma after 350 injections. Evenwith the protein precipitation extractionmethod, no retention-time changes wereobserved. The combination of an aqueous-organic mobile phase and a silica columnalso demonstrated an excellent, reproduciblegradient elution. Reequilibration time wascomparable with that of the reversed-phasecolumn. Table 2 lists retention times for fourconsecutive injections of five differentprotease inhibitors. The resultingchromatograms demonstrate excellent peakshape and resolution, shown in Figure 3.

Injection SolventsA trend toward faster, more efficientLC/MS/MS has prompted the use ofcolumns with smaller dimensions (50 �m � 2 �m). In contrast toconventional columns (250 �m � 4.6 �m),a one column offers advantages that includefaster analysis time, better concentrationsensitivity, and lower solvent consumption.

Figure 3. Chromatograms of gradient elution of five protease inhibitors on silica column withaqueous–organic mobile phase; column: Betasil silica 50 � 3 �m; mobile phase: lineargradient elution of acetonitrile–water–formic acid from 95:5:0.2 to 30:70:0.2 in two minutes;flow rate: 0.5 mL/min; sample preparation: protein precipitation; AMP: amprenavir; RIT:ritonavir; NEL: nelfinavir; IS: internal standard; SAQ: saquinavir; IND: indinavir

2 4

AMP SAQ

(a) 0.67

3.53

4e5

3e5

2e5

1e5

Time, min

Inte

nsity

, cps

2 4

(b)

0.64

2.50

80,000

60,000

40,000

20,000

Time, minIn

tens

ity, c

ps

2 4

NEF RES (IS)

(e) 2.434e5

3e5

2e5

1e5

Time, min

Inte

nsity

, cps

2 4

(f)

0.62

2.43

8,000

6,000

4,000

2,000

Time, min

Inte

nsity

, cps

2 4

RIT IND

(c) 1.541.0e6

8.0e5

6.0e5

4.0e5

2.0e5

Time, min

Inte

nsity

, cps

2 4

(d)

0.65

3.7810,000

8,000

6,000

4,000

2,000

Time, min

Inte

nsity

, cps

Figure 4. Theory of analyte enrichment onanalytical column

30

20

10

00% 10% 20% 30% 40%

Small moleculePeptide

% stronger solvent in mobile phase % stronger solvent in injection solution

Enr

ichm

ent

These criteria should be consideredwhen selecting an injection solution:

analyte adsorption, enrichment,retention, solubility, and stability

retention mechanism(s)

solvent evaporation

Injection Solution Selection Criteria

34 BioPharm NOVEMBER 2001

36 BioPharm NOVEMBER 2001

Process Development

composition, and the stationary phase. Ifretention on the C18 column were blindlyassumed to be reversed-phase, water wouldbe used as the weaker elution solvent and,therefore, the injection solution. Very poorpeak shapes would result. The influence ofinjection solvents on chromatography isdemonstrated in Figure 5.

If an analyte is unstable in the mobilephase but stable in a weaker component ofit, the mobile phase can still be used, as longas the weaker solvent is used forreconstitution. For example, omeprazole isunstable in acidic solution. To obtain goodpeak shape and sensitivity, an acidic mobilephase can be used. After extraction,omeprazole can be reconstituted inacetonitrile and injected onto a silica columnwith an acidic aqueous–organic mobilephase (13). Because of the short run time(three minutes), on-column degradation ofomeprazole is not observed. In that situation,both analyte stability and on-columnstacking are achieved because acetonitrile isa weaker solvent for the silica column (seethe “Injection Solution Selection Criteria”box).

Looking AheadLC/MS/MS provides superior sensitivity,selectivity, and rapid analysis. Methoddevelopment for large-scale LC/MS/MSanalysis is fraught with challenges, however.Obstacles can be prevented through carefulplanning and the application of logicalproblem-solving techniques. Automationand integration of information systems intobioanalytical laboratory processes andplatforms have been shown to providesynergistic improvements in consistency,performance, and error reduction. Part 2 willpresent troubleshooting tips for carry-over,recovery and matrix effects, and selectivity;and for automation where possible toimprove the overall reliability and efficiencyof LC/MS/MS.

AcknowledgmentsWe wish to thank the many scientists and associatesfrom our bioanalytical facilities located at Madison,WI, and Indianapolis, IN, for their contributions.

References(1) M.S. Lee and E.D. Kerns, “LC/MS

Applications in Drug Development,” MassSpectrom. Rev. 18, 187–279 (1999).

(2) N. Weng et al., “Simultaneous Development ofSix LC-MS-MS Methods for theDetermination of Multiple Analytes in Human

be obtained by diluting the sample with theweaker component in the mobile phase andincreasing the injection volume.

C18 columns. Before choosing a dilutingsolvent, you should determine the retentionmechanism. Retention mechanisms dictatewhether water or organic solvent will bestronger for elution. For reversed-phase LC,water is weaker than organic solvents, suchas acetonitrile, are stronger. For normal-phase LC, the solvent strengths are reversed:Water is the stronger elution solvent, andorganics the weaker. Even the name“reversed-phase C18 column” can bemisleading. The retention mechanism on aC18 column can be complicated by the dual-retention mechanism, as shown in Figure 1.The actual retention mechanism depends onthe analyte itself, the mobile phase

However, when smaller quantities ofpacking material are used, thechromatography can be more easilydisturbed. Mismatches between the injectionsolution and the mobile phase are a commonproblem in compounds that demonstrateearly elution peaks (21).

Choosing a reconstitution solventcompatible with the mobile phase canincrease sensitivity. When the injectionsolution has a weaker eluting strength thanthe mobile phase, the effects of the injectionsolution on peak shape and chromatographyefficiency can be enriched. Figure 4 shows atypical enrichment profile. Maximumchromatography efficiency can be achievedusing injection solution at the weakestelution strength. From a practical point ofview, better chromatography efficiency can

Figure 5. A common mistake: assuming that C18 columns are always reversed-phase;column: Hypersil BDS C18 50 � 2 �m, mobile phase: water–acetonitrile–formic acid(20:80:0.2, v/v/v); injection solution: A, water; B, mobile phase; C, acetonitrile

0.5 1.0 1.5

ALB

A

B

C(a)

0.22 0.54

0.82

1.74

Albuterol

HO

HO

OH H

N C(CH3)3

1.5e5

1.0e5

5.0e4

Time, min

Inte

nsity

, cps

0.5 1.0 1.5

BAM

(b)

0.16 0.48

0.88

1.36

40,000

20,000

Time, min

Inte

nsity

, cps

0.5 1.0 1.5

NIC

(c)

0.22 0.71

1.09

50,000

Time, min

Inte

nsity

, cps

0.5 1.0 1.5

COT

(d)

0.39

0.69

5e5

Time, min

Inte

nsity

, cps

BamethanHO

OH H

N

N

N

CH3

CH3

Nicotine

N

N O

CH3

Cotinine

www.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1900.html.

(16) Y-L. Chen et al., “A Quantitative Method forAssay of Nicotine and Cotinine in HumanPlasma Using LC/MS/MS on a Silica Columnand a Robotic Liquid Handling System,” paperpresented at the AAPS Annual Meeting andExposition, 29 October–2 November 2000,Indianapolis. Available at www.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1014.html.

(17) Y-L. Chen, G.D. Hanson, and N.Weng, “ARapid LC/MS/MS Method for the SimultaneousDetermination of Hydrocodone, and ItsMetabolite Hydromorphonein Human PlasmaUsing Automatic Sample Preparation,” paperpresented at the 49th ASMS Conference onMass Spectrometry, 27–31 May 2001, Chicago.

(18) T. Addison, G. Hanson, and N. Weng,“Validation of an Ultrasensitive Method for theDetermination of Oxycodone, Oxymorphone,and Noroxycodone in Human Plasma UsingLiquid Chromatography with Tandem MassSpectrometric Detection,” paper presented atthe 49th ASMS Conference on MassSpectrometry, 27–31 May 2001, Chicago.

(19) W.Z. Shou et al., “An Automatic 96-Well SolidPhase Extraction and Liquid Chromatography–Tandem Mass Spectrometry Method for theAnalysis of Morphine, Morphine-3-Glucuronide and Morphine-6-glucuronide inHuman Plasma,” J. Pharm. Biomed. Anal.,accepted for publication (2001).

(20) W.Z. Shou et al., “A Highly Automated 96-Well Solid Phase Extraction and LiquidChromatography/Tandem Mass SpectrometryMethod for the Determination of Fentanyl inHuman Plasma,” Rapid Commun. MassSpectrom. 15, 466–476 (2001).

(21) N. Weng et al., “Importance of InjectionSolution Composition for LC/MS/MSMethods,” J. Pharm. Biomed. Anal., in press(2001). BP

38 BioPharm NOVEMBER 2001

Process Development

Plasma,” J. Pharm. Biomed. Anal.,unpublished (2001).

(3) D.L. Buhrman, P.I. Price, and P.J. Rudewicz,“Quantitation of SR 27417 in Human PlasmaUsing Electrospray Liquid Chromatography-Tandem Mass Spectrometry: A Study of IonSuppression,” J. Am. Soc. Mass Spectrom. 7,1099–1105 (1996).

(4) B.K. Matuszewski, M.L. Constanzer, and C.M.Chavez-Eng, “Matrix Effect in QuantitativeLC/MS/MS Analysis of Biological Fluids: AMethod for Determination of Finasteride inHuman Plasma at Picogram per MilliliterConcentrations,” Anal. Chem. 70, 882–889(1998).

(5) N. Weng, J. Lee, and J.D. Hulse, “ChiralHPLC Quantitation of MethocarbamolEnantiomers in Human Plasma,” J. Liq.Chromatogr. 17, 3747–3758 (1994).

(6) K. B. Freeman et al., “Validated Assays for theDetermination of Gemcitabine in Human Plasmaand Urine Using High-Performance LiquidChromatography with Ultraviolet Detection,” J.Chromatogr. B. 665, 171–181 (1995).

(7) M. Jemal and D. Hawthorne, “HighPerformance Liquid Chromatography/IonSpray Mass Spectrometry of N-Ethylmaleimide and Acrylic Acid EsterDerivatives for Bioanalysis of ThiolCompounds,” Rapid Commun. Mass Spectrom.8, 854–857 (1994).

(8) M. Jemal and D. Hawthorne, “QuantitativeDetermination of BMS-186716, a ThioCompound, in Dog Plasma by High-Performance Liquid Chromatography–PositiveIon Electrospray Mass Spectrometry afterFormation of the Methyl Acrylate Adduct,” J. Chromatogr. B. 693, 109–116 (1997).

(9) N. Weng et al., “Novel LiquidChromatographic–Tandem MassSpectrometric Methods Using Silica Columnsand Aqueous-Organic Mobile Phases forQuantitative Analysis of Polar Ionic Analytesin Biological Fluids,” J. Chromatogr. B. 754,387–399 (2001).

(10) M. Pelzer et al., “Validation of a Method toDetermine Clonidine in Human Serum byLC/MS/MS on a Silica Column,” paperpresented at the AAPS Annual Meeting andExposition, 29 October–2 November 2000,Indianapolis. Available at www.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1898.html.

(11) W.Z. Shou et al., “A Normal Phase LC/MS/MSMethod for the Determination of Albuterol inHuman Serum,” paper presented at the AAPSAnnual Meeting and Exposition, 29 October–2November 2000, Indianapolis. Available atwww.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1899.html.

(12) T. Addison et al., “Method Development andValidation of a Highly Sensitive Normal PhaseLC/MS/MS Method for Analysis of Loratadineand Descarboethoxyloratadine in HumanPlasma,” paper presented at the AAPS AnnualMeeting and Exposition, 29 October–2November 2000, Indianapolis. Available atwww.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1029.html.

(13) M. Pelzer, S. Harsy, and N.Weng, “Validationof a Method to Determine Omeprazole inHuman Plasma by LC/MS/MS on a SilicaColumn,” paper presented at the AAPS AnnualMeeting and Exposition, 29 October–2November 2000, Indianapolis. Available atwww.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1023.html.

(14) A. Eerkes et al., “Development and Validationof a Method to Determine Triazolam in HumanPlasma by LC/MS/MS on a Silica Column,”paper presented at the AAPS Annual Meetingand Exposition, 29 October–2 November 2000,Indianapolis. Available at www.aapspharmaceutica.com/scientificjournals/pharmsci/am_abstracts/2000/1026.html.

(15) W.Z. Shou et al., “A Normal Phase LC/MS/MSMethod for the Determination of Ribavirin inHuman Plasma,” paper presented at the AAPSAnnual Meeting and Exposition, 29 October–2November 2000, Indianapolis. Available at

Info #24

22 BioPharm JANUARY 2002

Process Development

The gradient elution was actually determinedto be the cause of the carry-over effect.

To address the issue, an isocratic elutionwith a mobile phase that contained 45%acetonitrile was used to eliminate the carry-over. A new problem surfaced when theoriginal internal standard was not retainedon the column in the mobile phase, so a newone had to be used. The new standard hadretention properties similar to those of theanalyte. Figure 2 depicts the resolution ofthe carry-over problem. The method wasvalidated and successfully used for sampleanalysis.

The ritonavir example is one of the morecomplicated troubleshooting scenarios. Bothchromatographic condition and internalstandard were changed as a result of theproblem and its solution. Occasionally,carry-over problems have simpler solutions.In another example, simply raising theneedle height solved the carry-over effect.Carry-over is often avoided by merelyminimizing the contact surface betweenanalyte and needle.

Recovery and Matrix EffectsLC coupled with MS/MS detection is aspecific and sensitive method for druganalysis in biological matrices. Because ofthe highly sensitive nature of MS/MSdetection, extensive chromatographicresolution may not be required, and shortrun times can be used to obtain very highthroughput for sample analysis. Materials inthe extracted biological matrix, however,can exist in much higher concentrations thanthe analyte. Some materials may have thesame m/z for both father and daughter ionsand will be observed on the chromatogramas interference peaks. Though the peaks maybe unseen on the LC/MS/MS chromatogram,what happens more often is that materialfrom the extracted biological matrix elutesclosely to the analyte. Ionization is affectedand results in high imprecision and loss of

Systematic Troubleshooting for LC/MS/MSPart 2: Large-Scale LC/MS/MS and Automation

Meeting the challenges of large-scale LC/MS/MS can improve theanalytical processes that supportbiopharmaceutical drugdevelopment. The conclusion ofthis article presentstroubleshooting techniques forLC/MS/MS and illustrates anautomation strategy.

LC/MS/MS can provide superiorsensitivity and selectivity, rapidanalysis, maximized developmentefficiencies, and improved turnaroundtimes — its challenges are in large-

scale application. In Part 1 (BioPharm,November 2001), we offeredtroubleshooting techniques for samplepreparation and chromatography (1).LC/MS/MS remains one of the most usefultools available for bioanalysis. A rational,strategic approach of developing robust,large-scale, and automated LC/MS/MSmethods can reduce slowdowns andbottlenecks in drug development andcontribute to synergistic, consistent, long-term performance. Part 2 discusses furthertroubleshooting techniques and presents anautomation strategy that improves methodrobustness and performance.

Elimination of Carry-OverCarry-over is probably one of the mostcommonly encountered problems ofLC/MS/MS method development. Sourcesof problems range from instrument hardwareand the selection of appropriate rinsesolvents to challenges with chromatography.Frequently, resolving the problem requires acombination of individual experimentalsolutions and systematic and logicalinvestigation.

Carry-over is a phenomenon that wasdiscovered during method development inLC/MS/MS of ritonavir in human plasma. Acyano column was used with 0.1% aceticacid and an acetonitrile gradient elution.Blanks injected after high-level standardsshowed increasing carry-over (Table 1).Chromatograms (Figure 1) show both thehigh-level standard and the blank injectedimmediately after the high-level standard.Prior injection of the same blank before thehigh-level standard did not show a ritonavirpeak, which confirmed a carry-over effect.

Naidong Weng and Timothy D.J. Halls

Corresponding author Naidong Weng is associatedirector of bioanalytical chemistry, and Timothy D.J. Halls is vice president ofpharmaceutical chemistry at Covance LaboratoriesInc., 3301 Kinsman Boulevard, Madison, WI 53704,608.242.2652, fax 608.242.2735,naidong.weng@covance.com, www.covance.com.

BioPharm JANUARY 2002 23

sensitivity within a run. For an LC/MS/MSmethod, it is important to identify whetherthe loss of sensitivity is due to poor recoveryor to matrix suppression, because bothcauses give the same result. Carefullydesigned experiments will establish thesource of the problem.

Recovery is determined by comparing thepeak areas of extracted samples with thoseof neat solutions spiked (postextraction) intoa blank matrix. Because both samples havethe matrix ingredients present, the matrixeffects can be considered the same forextracted samples and postextraction spikedsamples. Any differences in response cannow be considered to be due to extractionrecovery. The matrix effect is determined bycomparing peak areas of neat solutionsspiked (postextraction) into blank matrixwith those of other neat solutions. Becausethe analytes are not extracted, the analyteshould have the same response inpostextraction spiked samples and in neatsolutions. The matrix ingredients, therefore,cause whatever differences are noted in theresponses.

A useful method to assess matrixsuppression is postcolumn infusion of ananalyte into the MS detector. The extractedblank matrix is injected by an autosampleronto the analytical column. The setup isshown in Figure 3. The purpose ofpostcolumn infusion with analyte is to raisethe background level so that the suppressionmatrix will show as negative peaks. Thissetup has been successfully used to identifyand troubleshoot matrix suppression peaks.

During LC/MS/MS method developmentfor analysis of a nucleoside compound andits metabolite, lower and inconsistent signalsof the metabolite peak were observed. Anaqueous–organic mobile phase and a silicacolumn were used. The extraction methodwas a simple protein precipitation. Thelower signal was due to matrix suppression,which was confirmed by postcolumninfusion portrayed in Figure 4. A broadsuppression band was observed around theanalyte peak. The problem was overcome bydiluting the extracted sample five-fold withthe weaker elution solvent, in this case,acetonitrile. The suppression was no longerobserved, and the method has since beenvalidated and successfully used for routinesample analysis.

SelectivityAlthough MS/MS is highly selective fordiscriminating analytes from interferencepeaks, the blank screen test routinelyperformed for HPLC-UV is inadequate forLC/MS/MS methods. Because of simplifiedextraction procedures (a benefit of the high-selectivity of MS/MS detection), manyendogenous compounds, metabolites, orcoadministered medicines can coelute withthe analytes of interest and not show up as

Figure 1. Identification of carry-over problem by injecting a standard at high concentration(left panels) followed by a blank (right panels)

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

InternalStandard

0.63 1.36 2.63

2.160.86 1.76

1.032.90 3.80

3.08

1.97

3.60

20,000

16,000

12,000

8,000

4,000

Time, min

Inte

nsity

, cps

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

(d)(c)

0.1 0.7

1.12.7

2.0

4.03.83.3

2.32.5

3.7

8070605040302010

Time, min

Inte

nsity

, cps

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

HighStandard

0.23 1.530.57 2.130.79 2.86

3.14

2.37

3.68

1.4e51.2e51.0e58.0e46.0e44.0e42.0e4

Time, min

Inte

nsity

, cps

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Blank

Analyte

(b)(a)

0.5 1.6 2.62.12.8

3.93.7

3.23.3

2.4

3.5

500

400

300

200

100

Time, min

Inte

nsity

, cps

Figure 2. Carry-over problem was resolved as evidenced by no analyte peak in the blank(right panels) injected immediately after the standard at high concentration (left panels)

New Internal

Standard

0.10.5 1.4

1.00.2

0.60.3 1.5

1.51.91.7

1.7

0.7

1.8

Time, min

Inte

nsity

, cps

(d)(c)

0.10.2

0.3 0.5 1.81.61.20.7

0.90.9 1.4

4,0003,5002,5002,0001,5001,000

500

Time, min

Inte

nsity

, cps

0.2 0.6 1.0 1.4 1.80.2 0.6 1.0 1.4 1.8

0.2 0.6 1.0 1.4 1.80.2 0.6 1.0 1.4 1.8

HighStandard

0.10.60.2

0.70.4 1.2 1.6

0.9

1.71.8

2.0

1.2e5

1.0e5

8.0e4

6.0e4

4.0e4

2.0e4

1.6e51.4e51.2e51.0e58.0e46.0e44.0e42.0e4

Time, min

Inte

nsity

, cps

Blank

Analyte

(b)(a)

0.1

0.3

0.40.5

0.9

0.6

0.7

1.1

1.81.9

1.71.2

1.4

0.81.5

45403530252015105

Time, min

Inte

nsity

, cps

Test Analyte Area

Standard 500 606,022Blank 954

Standard 500 595,334Blank 1,486

Standard 500 608,066Blank 1,935

Table 1. Carry-over test of the analyte byinjecting standards at high concentrationand at blanks

24 BioPharm JANUARY 2002

Process Development

metabolite, and therefore the in-interfacebreakdown products, can elute with thesame retention as the analyte. Those

interference peaks at the m/z channel of theanalytes. Those unseen peaks cause matrixsuppression and inconsistent signals. Theselectivity of LC/MS/MS can becompromised even more by a breakdown ofmetabolites (especially conjugatedmetabolites) at the LC/MS interface. Thosebreakdown products are usually the analytesof interest. Without chromatographicretention and resolution, the conjugated

breakdown products have the sameprecursor–product m/z as the analyte and canbe falsely quantified as analytes of interest,

Figure 4. Postcolumn infusion to determine the matrix suppression profiles of originalextraction and modified extraction: matrix suppression was eliminated by diluting theextracted samples five fold

0.5 1.0 1.5 2.0

Extracted blank diluted five foldExtracted blank

Metabolite R.T.

0.040.38 0.73

0.88

1.05 1.411.92 2.20

2.322.45

2.2e52.1e52.0e51.9e51.8e51.7e51.6e51.5e51.4e51.3e51.2e51.1e51.0e59.0e48.0e47.0e46.0e45.0e44.0e43.0e42.0e41.0e4

Time, min

Inte

nsity

, cps

Figure 3. General set-up for identifyingmatrix effect: postcolumn infusion ofcompound while injecting extracted blank

Turbolonspray interface

API 3000

Mass spectrometer

Syringepump

Compound solution

ZDV "T"Union

HPLCcolumn

Injector

HPLC

Info #13

BioPharm JANUARY 2002 25

which can result in severe overestimation ofthe analyte (2).

Long retention times do not necessarilytranslate into large capacity factors (k�),which are the true measurement of on-column retention. A long narrow-borecolumn at a low flow rate will have longerretention times even though the analytes mayactually elute at void volume. Approximatelythe same k� values (k��1) can be obtainedfor an analyte with one-minute retention on a5 cm � 3 mm column at a flow rate of 0.5 mL/min, as for a five-minute retention ona 25 cm � 4.6 mm column at a flow rate of1.0 mL/min for the same analyte.

One approach to selectivity is to ensurethat blanks are truly blank and also to ensurethat good accuracy is obtained for differentlots of matrix spiked with analytes of interest.Coadministered medicines and knownmetabolites should be spiked to QC samplesto demonstrate that their presence does notcause quantitation bias from eitherinterference or matrix suppression.

AutomationEfficiency and speed bottlenecks exist inevery organization, whether a commercialenterprise or an academic laboratory.Laboratory layouts and design issuescontribute to many of the inefficienciesbecause most labs have not been purposefullybuilt for efficient work flow.

One strategic approach to overcomingbottlenecks for high-throughput bioanalysisis to use well-established instrumentation;rigorous, standardized techniques; andautomation, wherever possible, to replacemanual tasks. Automation results in greaterperformance consistency over time and inmore reliable methods transfer from site tosite. Constant assessments of processes,technologies, and procedures are requiredfor continued and incremental processimprovements.

For example, automated 96-well platetechnology is well-established and acceptedand has been shown to effectively replacemanual tasks. The 96-well instruments canexecute automated offline extractions andsample cleanups. They take advantage ofparallel processes, replace manualtechniques, and offer consistent,standardized methods.

Incorporating a disciplined informationtechnology (IT) strategy into the laboratoryis essential for productivity improvements

and consistent performance. The IT strategyshould be integrated into automationprocesses whenever and wherever possible.State-of-the-art IT systems (such ascustomized Laboratory InformationManagement Systems or LIMS) allowanalytical labs to maximize efficiencies andimprove the turnaround time of quality-control data for clients. LIMS-type systemstrack samples and provide validated datastreams, reducing turnaround times and thesubstantial time once allotted to qualitycontrol and data checking.

Figure 5 provides a general approach forautomated 96-well sample analysis and itsbenefits. Automated solid-phase extraction(SPE) and liquid–liquid (LL) and proteinprecipitation (PP) extraction can all be

performed in the 96-well format. Thechromatograms of extracted low limit ofquantitation (LLOQ) and blank plasma orserum samples of three examples are shownin Figure 6. In that example, the Multiprobeliquid handling station (Packard, www.packardbioscience.com) was used to aliquotsamples and add internal standards. TheQuadra 96-320, a 96-well workstation fromTomtec (www. tomtec.com) was used forSPE (fentanyl), LL (fluconazole), or PP(ribavirin) extractions. Aqueous–organicmobile phases on silica columns were usedto achieve the excellent peak shapes andsensitivity.

Table 2 summarizes the blank screen andmatrix effect tests. No interference peaks wereobserved in any of the lots tested. No matrix

Fluconazole (ng/mL) Fentanyl (ng/mL) Ribavirin (ng/mL)

Plasma Std Std Plasma Std Std Plasma Std StdLot # 0.000 0.500 Lot # 0.0000 0.0500 Lot # 0.0 30.0

7496 0.000 0.505 6307 0.0000 0.0503 6065 0.0 31.8

7497 0.000 0.506 6308 0.0000 0.0479 6066 0.0 30.3

7498 0.000 0.518 6309 0.0000 0.0535 5978 0.0 29.1

7499 0.000 0.487 6310 0.0000 0.0486 5979 0.0 31.0

7500 0.000 0.459 6311 0.0000 0.0514 5980 0.0 31.5

7501 0.000 0.522 6313 0.0000 0.0505 5981 0.0 25.9

Mean 0.000 0.499 Mean 0.0000 0.0504 Mean 0.0 29.9

RSD% NA 4.7 RSD% NA 4.0 RSD% NA 7.3

RE% NA �0.1 RE% NA �0.7 RE% NA �0.2

Table 2. Blank screen and matrix effects test for fluconazole (liquid–liquid extraction) inhuman plasma; fentanyl (SPE) in human plasma; and ribavirin (protein precipitationextraction) in human serum

Figure 5. Automated 96-well sample analysis

PE-Sciex API3000

Packard Multiprobe II

Tomtec Quadra 96

Beckman Centrifuge

Zymark Turbovap 9696-well plate format

Shimadzu VP HPLC

Automated aliquotingaddition of IS

Automated SPE/LL/PP

96-wellcentrifuge

Samples

Dry-downconcentration

Results

26 BioPharm JANUARY 2002

Process Development

and peptides. Reversed-phase LC/MS/MS ofproteins and peptides have been extensivelyused in biotechnology analytical laboratories(3). Typically, a gradient elution on areversed-phase column is used with anorganic solvent that ranges in concentrationfrom 0% to 40%–60%. Small amounts oftrifluoroacetic acid (TFA), and to a lesserextent formic acid, are included in the mobilephase to improve the peak shape. Because ofthe gradient elution and characteristics ofproteins and peptides, the carry-over problemcan be significant and should be carefully

lot-to-lot differences were observed becauseall the spiked samples were back-calculatedwithin 15% of the theoretical values.

Automation benefits are shown in Figure 7. The arrows in the Figure indicatethe stages where the samples are tracked bythe LIMS. Manual involvement is kept to aminimum.

Proteins and PeptidesMethod development strategies are wellsuited for LC/MS/MS analysis of proteins

controlled. In some cases, it may even benecessary to inject a blank after each sampleto circumvent the problem.

Proteins and peptides also presentadditional challenges for LC/MS/MSpractitioners. Proteins and peptides easilyform adducts with various types of salts. Theadducts formed can significantly change thecharge of the molecules, which may result inpoor quantitation reproducibility because ofthe shifted m/z values. Salts can alsosuppress the LC/MS/MS signals of analytes(4). Therefore, before LC/MS/MS, adesalting process is important using eitheroffline dialysis or online desaltingprocedures. In the online desaltingtechnique, the analytical column is typicallywashed by the aqueous mobile phase forseveral minutes before gradient elution.

Meeting the ChallengesLC/MS/MS provides superior sensitivity,selectivity, and rapid analysis. Automated 96-well technology has significantly improvedturnaround times and has opened up newopportunities to maximize efficiencies and toenhance drug development.

Method development for large-scaleLC/MS/MS analysis is fraught withchallenges, however. Obstacles can be

Figure 7. Benefits of automation (arrows indicate where LIMS is applied)

Manualprocesses

Retrieve, thaw,centrifuge

AddIS

Transfer todeep-well

plate

Dry-downTurbovap 96

Addsolvent Vortex

PE Sciex API3000

Centrifuge

Reconstitute

Freezerstorage

Unattendedoperations

IT enablingLIMS

LIMSreporting

communication

Multiprobe(automatedprocessing)

Apply to96-wellplate

Tomtec(automated

SPE)

Continued on page 49

Figure 6. Chromatograms of blank (upper panels) and low limit of quantitation (lower panels) for fluconazole (96-well liquid–liquidextraction), fentanyl (96-well SPE extraction), and ribavirin (96-well protein precipitation extraction)

Blank

(e) (f)

(g) (h)

(i) (j)

(k) (l)

(a) (b)

(c) (d)

ISBlank IS

0.240.47

1.49

1.38 1.31

1.68 0.13 1.212.02

2.371.530.68

1.931.181.28

0.641.38

1.71 0.610.72

0.65 2.291.02

200

150

100

50

1.0 2.0

Time, min

FentanylFluconazole Ribavirin

Inte

nsity

, cps

20

16

12

8

4

00.2 1.0 1.8 0.2 1.0 1.8

0.2 1.0 1.8 0.2 1.0 1.8

Inte

nsity

, cps

Inte

nsity

, cps

7,0006,0005,0004,0003,0002,0001,000

1.0 2.0

Time, minTime, min Time, min

Inte

nsity

, cps

50pg/ml

IS0.5ng/ml

IS

0.280.47

1.34 0.68

1.65

1.350.32

1.66

1.960.83

200

150

100

50

1.0 2.0

Time, min

Inte

nsity

, cps

7,0006,0005,0004,0003,0002,0001,000

1.0 2.0

Time, minTime, min Time, minIn

tens

ity, c

ps

Blank IS

0.362.03

400

300

200

100

0

Time, min

Inte

nsity

, cps

2,000

1,600

1,200

800

400

Time, min

Inte

nsity

, cps

10ng/ml IS

0.15

1.73

0.51

2.24

0.821.35

1.672.94

0.26

2.290.66

400

300

200

100

0

Time, min

Inte

nsity

, cps

1.6e41.4e41.2e41.0e4

8,000.06,000.04,000.02,000.0

0.0

Inte

nsity

, cps

1.6e41.4e41.2e41.0e4

8,000.06,000.04,000.02,000.0

0.0

Inte

nsity

, cps

20

16

12

8

4

0

2,000

1,600

1,200

800

400

1.0 2.0 3.01.0 2.0 3.0

1.0 2.0 3.01.0 2.0 3.0

Time, min

Inte

nsity

, cps

2.220.78 1.55

2.41

BioPharm JANUARY 2002 49

overcome through careful planning and through the application oflogical problem-solving techniques. Automation and integrationof information systems into bioanalytical lab processes andplatforms have been shown to provide synergistic improvementsin consistency, performance, and error reduction.

AcknowledgmentsWe wish to thank the many scientists and associates from our bioanalyticalfacilities located at Madison, WI, and Indianapolis, IN, for their contributions.

References(1) N. Weng and T. Halls, “Systematic Troubleshooting for LC/MS/MS:

Part 1, Sample Preparation and Chromatography,” BioPharm 14(11)28–38 (2001).

(2) N. Weng et al., “Development and Validation of a Sensitive Method forthe Hydromorphone in Human Plasma by Normal Phase LiquidChromatography–Tandem Mass Spectrometry,” J. Pharm. Biomed.Anal. 23, 697–704 (2000).

(3) P. Lu et al., “Process Purification of Polypeptides and Proteins byReversed-Phase Column Chromatography: Misconceptions andReality,” BioPharm, 14(9), 28–35 (2001).

(4) R. King et al., “Mechanistic Investigation of Ionization Suppression inElectrospray Ionization,” J. Am. Soc. Mass Spectrom. 11, 942–950(2000). BP

LC/MS/MS Troubleshooting continued from page 26

Info #17