Heat-Map Visualization of Gas Chromatography-Mass ...

Heat-Map Visualization of GasChromatography-Mass Spectrometry BasedQuantitative Signatures on Steroid Metabolism

Ju-Yeon Moon,a,b Hyun-Jin Jung,a,b Myeong Hee Moon,b

Bong Chul Chung,a and Man Ho Choiaa Life Sciences Division, Korea Institute of Science and Technology, Seoul, Koreab Department of Chemistry, Yonsei University, Seoul, Korea

Abnormalities in steroid hormones are responsible for the development and prevention ofendocrine diseases. Due to their biochemical roles in endocrine system, the quantitativeevaluation of steroid hormones is needed to elucidate altered expression of steroids. Gaschromatographic-mass spectrometric (GC-MS) profiling of 70 urinary steroids, containing 22androgens, 18 estrogens, 15 corticoids, 13 progestins, and 2 sterols, were validated and itsquantitative data were visualized using hierarchically clustered heat maps to allow “steroidsignatures”. The devised method provided a good linearity (r2 � 0.994) with the exception ofcholesterol (r2 � 0.983). Precisions (% CV) and accuracies (% bias) ranged from 0.9% to 11.2%and from 92% to 119%, respectively, for most steroids tested. To evaluate metabolic changes,this method was applied to urine samples obtained from 59 patients with benign prostatichyperplasia (BPH) versus 41 healthy male subjects. Altered concentrations of urinary steroidsfound and heat maps produced during this 70-compound study showed also differencesbetween the ratios of steroid precursors and their metabolites (representing enzyme activity).Heat maps showed that oxidoreductases clustered (5�-reductase, 3�-HSD, 3�-HSD, and17�-HSD, except for 20�-HSD). These results support that data transformation is valid, since5�-reductase is a marker of BPH and 17�-HSD is positively expressed in prostate cells.Multitargeted profiling analysis of steroids generated quantitative results that help to explaincorrelations between enzyme activities. The data transformation and visualization describedmay to be found in the integration with the mining biomarkers of hormone-dependentdiseases. (J Am Soc Mass Spectrom 2009, 20, 1626–1637) © 2009 American Society for MassSpectrometry

Mass spectrometry based metabolite profilingreveals the metabolic states of biological sys-tems and provides comprehensive insights by

allowing comparisons between many metabolites si-multaneously present in cells, tissues, or organisms [1,2]. This technique promotes the establishment of rela-tionships between phenotypes and metabolisms byproviding descriptions of the distributions of metabo-lites and their biological functions. In fact, quantitativeanalyses of sets of metabolites in biochemical pathwayshave been used for physiological monitoring, toxicolog-ical evaluations, and clinical diagnosis [3–6].

Many naturally occurring steroids with similarchemical structures could yield biological information[7]. Endogenous steroids are divided into five groups,namely, androgens, estrogens, corticoids, progestins,and sterols, which are generally synthesized from cho-lesterol in the adrenal cortex, ovaries, and testes(Scheme 1). In biosynthetic pathways of steroid hor-

Address reprint requests to Dr. M. H. Choi, Life Sciences Division, Korea

© 2009 American Society for Mass Spectrometry. Published by Elsevie1044-0305/09/$32.00doi:10.1016/j.jasms.2009.04.020

mone, two major types of enzymes are involved, cyto-chrome P450 and steroid oxidoreductase. Abnormalities ofthese enzymes often lead to hormonal imbalances thathave serious consequences, and which are responsiblefor the development of hormone-dependent diseases(see Supplementary Table 1, which can be found in theelectronic version of this article). For example, concen-trations of corticoids and their metabolic ratios providediagnostic evidence of apparent mineralocorticoid ex-cesses caused by 11�-HSD deficiency [8] and congenitaladrenal hyperplasia, which are caused by deficienciesof enzymes like hydroxylase (at C-11, 17, and 21) or3�-HSD [9]. In addition, enhanced androgen activitygenerated by the conversion of testosterone to dihy-drotestosterone (DHT) by 5�-reductase was utilized toallow early therapeutic intervention in young men [10].

Enzyme activity profiles can be used to describe thefunctional diversities of biological systems, which aredriven by genetic diversity. Although enzyme activitieshave been monitored by following reactions betweentargeted enzymes and substrate molecules, over-estimations by radioimmunoassay (RIA) and enzyme

Published online May 5, 2009r Inc. Received February 7, 2009

Revised April 25, 2009Accepted April 28, 2009

1627J Am Soc Mass Spectrom 2009, 20, 1626–1637 QUANTITATIVE STEROID SIGNATURES BY GC-MS

ies, limit the applicability of these assays, and further-more, only single enzymes can be estimated at a time[11–14]. In contrast to conventional enzyme assays, gaschromatography-mass spectrometry (GC-MS)-basedtechniques have better quantitative reproducibility [15,16]. For these reasons, GC-MS profiling has been widelyused for steroid analysis [17–20], and offers the basis fortechniques that can be applied to large-scale clinicalstudies [21].

Clinical significances are generally expressed in ta-bles or bar graphs that show changes in analytes acrossgroups of interest. For studies involving few com-pounds, these visualizations are enough to differentiateclasses showing metabolic differences. However, quan-titative datasets of multiple compounds are much moredifficult to represent visually. Statistical clustering of-fers one such approach, and has been utilized to sup-port genomic and proteomic studies [22–24]. In a simi-lar fashion, quantitative results obtained by metaboliteprofiling can be directly compared between samplesand utilized as metabolic biomarkers. In recent, MS-based quantitative data generated by hierarchical clus-tering analysis (HCA) has been subjected to patternanalysis for metabolite profiling [2, 25]. The concept of“metabolite signature” is a result of this process, andsuch signatures are useful for measuring and visualiz-

Progestins, Sterols

Corticoids

17αααα-OH-PregPreg

Chol

17α-hydroxylase

24S-OH-CholHO

OH

HO

O

HO

O

H

OH

17αααα-OH-ProgProg

3β-HSD 3β-HSD

O

O

O

OH

O

20,22-desmolase

11-DeoxyB 11-DeoxyF

21-hydroxylase 21-hydroxylase

O

OH

O

11β-hydroxylase 11β-hydroxylase

Cortisol (F)O

OHHO OH

O

O

OHOH

O

O

O

HO OH

THF / allo-THF, allo-DHF5β- / 5α-reductase

3β-HSD

11-DehydroB

17α-hydroxylase11β-hydroxylase

O

O

HO

11ββββ-OH-Prog

HO

O

HOH

HO

O

OHOOH

O

11β-hydroxylase

11β-HSD

Cortisone (E)

O

O

OHO

O

OH

HOH

HOOH

O

O OH

HOH

OH

O

HOH

OH

THDOC

THE

5β- / 5α-reductase3β-HSD

H

HO OH

O

O

allo-DHB

20αααα-OH-progH

O

O

5αααα-DHP

HOH

O

P-oneP-diol

21β-hydroxylase

5β-reductase3β-HSD

5α-reductase

17,20-lya

17,20-ly

O

O

HO OH

21-DeoxyF

5α-reductase

HO

20α-HSD 3α-HSD

11β-HSD

20α-HSD

B

Scheme 1. General scheme for steroid metabonames of steroid hormones.

ing the relative analyte concentrations. Although many

groups have used GC-MS based steroid analysis formining biomarkers, technical improvements that moreeffectively allow the visualization of steroid metabolitesare required [26]. However, no studies have reportedMS-based steroid signatures generated by HCA to date.

Here, we introduce the GC-MS quantitative profilingof 70 urinary steroids, including 22 androgens, 18estrogens, 15 corticoids, 13 progestins, and 2 sterols,generated by HCA to evaluate metabolic changes andenzyme reactions in steroid analysis. The aim of thisstudy was to validate a GC-MS profiling method and amethod that allows the quantitative visualization ofmultiple urinary steroids. To visualize quantitative re-sults, a microarray map (a type of heat map) wasdesigned to present the urine sample results of 59patients with benign prostatic hyperplasia (BPH) and 41healthy male subjects; BPH was chosen because steroidmetabolism is known to play a role in the progress ofprostate diseases [27–29]. This study focused on illus-trating the usefulness of steroid signatures for explain-ing both the concentrations of individual steroids andthe activities of enzymes correlated global steroidogen-esis in BPH, and suggest that enzyme activity profilingmay be a useful diagnostic tool and provide a means ofidentifying mining biomarkers in hormone-dependent

drogens

Estrogens

O

H

OH

HO

O

HOH

HEA

ααααββββββββ-diol

An

5α-reductase 17β-HSD3α-HSD

3β-HSD

17β-HSD

aromatase

5αααα−−−−dione

O

O

17β-HSD

3α-HSD

O

HOH

Epi-An

3β-HSD5α-reductase

pi-T

OH

17α-HSD

H

HOH

O

Etio

OH

ββββααααββββ-diol

HOH

ββββαααααααα-diolHO

H

OH

6αααα-OH-DHEA

O

OH

3β-HSD

-dione

O

HO

OH

A-diol

17β-HSD

O

OH3β-HSD

O

OH

HDHT

HOH

OH

ααααααααββββ-diol

5ββββ−−−−dione

O

OH

5β-reductase

HO

O

HO

OH

E1 17ββββ-E2

2-OH-E1HO

O

HO

aromatase

2-OH-E2

2-MeO-E2

2-MeO-E1O

OH

O

HO

OH

H3CO

HO

O

H3CO

4-OH-E1

4-OH-E2

4-MeO-E2

4-MeO-E1

16αααα-OH-E1 Estriol 16-keto-E2

16-Epi-E317-Epi-E3

HO

O

OH

HO

OH

O

HO

OH

OH

HO

OH

OCH3

HO

O

OCH3

HO

OH

OH

HO

OH

OH

HO

OH

O

HO

OH

OH

17β-HSD

methyltransferase

methyltransferase methyltransferase

methyltransferase

2-hydroxylase

17β-HSD

3α-HSD 17β-HSD

17α-HSD

16α-hydroxylase

4-hydroxylase 16-hydroxylase

T

in man. See Supplementary Table 1 for the full

An

HO

D

O

E

1HO

se

ase

AO

H

H

lism

diseases.

1628 MOON ET AL. J Am Soc Mass Spectrom 2009, 20, 1626–1637

Experimental

Chemicals and Materials

Reference standards of the 70 steroids examined in thisstudy (Table 1) were obtained from Sigma (St. Louis, MO,USA), Steraloids (Newport, RI, USA), or NARL (Pumble,Australia). The internal standards used were, 16,16,17-d3-testosterone and methyltestosterone for 22 androgens,2,4,16,16-d4-estradiol for 18 estrogens, 9,11,12,12-d4-cortisolfor 15 corticoids, 2,2,4,6,6,17�,21,21,21-d9-progesteroneand 2,2,4,6,6,21,21,21-d8-17�-hydroxyprogesterone for 13progestins, and 2,2,3,4,4,6-d6-cholesterol for 2 sterols werepurchased from NARL and C/D/N isotopes (Pointe-Claire, Quebec, Canada). For solid-phase extraction (SPE),Oasis HLB cartridge (3 mL, 60 mg; Waters, Milford, MA,USA) was preconditioned with 3 mL of methanol fol-lowed by 3 mL of deionized water. Sodium acetate (re-agent grade), acetic acid (glacial, 99.99� %) and L-ascorbicacid (reagent grade) were obtained from Sigma (St. Louis,MO, USA). A solution of �-glucuronidase/arylsulfatasefrom Helix pomatia was purchased from Roche DiagnosticsGmbH (Mannheim, Germany). The trimethylsilylating(TMS) agents, N-methyl-N-trifluorotrimethylsilyl acet-amide (MSTFA), ammonium iodide (NH4I), and dithio-erythritol (DTE) were purchased from Sigma. All organicsolvents used were of analytical or HPLC grade and werepurchased from Burdick and Jackson (Muskegan, MI,USA). Deionized water was prepared using the Milli-Qpurification system (Millipore, Billerica, MA, USA).

Standard Solutions and Quality-Control Samples

Stock solutions of all reference standards were preparedat a concentration of 1000 �g/mL in methanol andworking solutions were made up with methanol atconcentrations in the range 0.1 to 10 �g/mL. L-Ascorbicacid (1 mg/mL) was added to prevent the oxidations oflabile steroid metabolites. All standard solutions werestored at �20 °C until required and all were stable for aminimum of 3 mo. The urine samples used for calibra-tion and quality-control (QC) purposes were preparedin house from steroid-free urine [30]. Steroids-free urinesamples were prepared by percolating urine samplesthrough Serdolit PAD-1 (0.1–0.2 mm analytical grade;Serva, Heidelberg, Germany). In this way, urinary ste-roids and other potential interfering compounds areretained in the cartridges and the eluates were collected.After checking for endogenous steroids with negativeresult following the solvent extraction procedure, theywere used as matrix for steroid profiling.

Subjects and Sample Collection

First-morning urine samples were collected from 59BPH patients (age: 65.3 � 8.2 y male) and 41 healthymale subjects (age: 56.7 � 7.1 y) in the Department ofUrology at Severance Hospital (Seoul, Korea). All pa-

tant biopsy and had either a suspicious finding bydigitorectal examination or an elevated level of serumprostate-specific antigen (PSA). The age-matched nor-mal controls showed no evidence of serum PSA levelelevation or of BPH. Steroid levels in urine were nor-malized versus urinary creatinine values using Jaffemethod [31].

Sample Preparation

Quantitative metabolite profiling of urinary steroidswas based on previous reports [4, 8, 32]. Briefly, urinesamples (2 mL) spiked with 20 �L of the 7 internalstandards (d3-testosterone and d4-estradiol; 1 �g/mL,d4-cortisol and d8-17�-hydroxyprogesterone; 5 �g/mL,methyltestosterone, d9-progesterone, and d6-cholesterolto 10 �g/mL) were extracted with Oasis HLB SPEcartridges coupled to a peristaltic pump. After loadinga sample onto a cartridge, the cartridge was washedwith 2 mL water and eluted twice with 2 mL ofmethanol. Combined methanol eluates were evapo-rated under a stream of nitrogen and then 1 mL of 0.2 Macetate buffer (pH 5.2), 100 �L of aqueous 0.2% ascorbicacid, and 50 �L of �-glucuronidase/arylsulfatase werewas added. After incubation at 55 °C for 3 h, thesolution was extracted twice with 2.5 mL of ethylacetate: n-hexane (2:3, vol/vol). Combined organic sol-vents were evaporated using a N2 evaporator at 40 °Cand further dried in a vacuum desiccator over P2O5-KOH for at least 30 min. Finally, the dried residue wasderivatized with MSTFA/NH4I/DTE (40 �L; 500:4:2,vol/wt/wt) at 60 °C for 20 min, and 2 �L of theresulting mixture was subjected to GC-MS in selected-ion monitoring (SIM) mode. To maximize extractionefficiencies, two organic solvents, tetrahydrofuran(THF) and ethyl acetate, were evaluated in the pH range5 to 9. In addition, the potential uses of diethyl ether,methyl tert-butyl ether, and n-hexane were investigatedfor promoting the partition of water-miscible THF andethyl acetate.

Instrumental Conditions

GC-MS was performed with an Agilent 6890 Plus gaschromatograph interfaced with a single-quadrupoleAgilent 5975 MSD at an electron energy of 70 eV and anion source temperature of 230 °C. Each sample (2 �L)was injected in split mode (10:1) at 280 °C and separatedthrough an Ultra-1 capillary column (25 m � 0.2 mmi.d., 0.33 �m film thickness; Agilent Technologies; PaloAlto, CA, USA). The GC oven temperature was initiallyset at 215 °C, then ramped to 260 °C at 1 °C/min, andfinally increased to 320 °C at 15 °C/min and held for 1min. The carrier gas was helium at a column headpressure of 210.3 kPa (column flow: 1.0 mL/min at oventemperature of 215 °C). For quantitative analysis, char-acteristic ions of each steroid were determined as their

Table 1. The GC-MS information for quantitative analysis of steroids studied

Compounds (trivial name) AbbreviationIon

selecteda

Retentiontime(min) Compounds (trivial name) Abbreviation

Ionselected

Retentiontime(min)

Androgens Estrogens5�-androstan-3�,17�-diol ���-diol 256 12.12 17�-estradiol 17�-E2 416 18.605�-androstan-3�,17�-diol ���-diol 256 12.87 Estrone E1 414 19.12Androsterone An 434 15.27 17�-estradiol 17�-E2 416 19.98Etiocholanolone Etio 434 15.44 2-methoxy-17�-estradiol-3-

methylether2-MeO-E2-3-methylether 388 23.40

5�-androstan-3�,17�-diol ���-diol 241 16.005�-androstan-3�,17�-diol ���-diol 256 16.07 4-methoxy-17�-estradiol 4-MeO-E2 446 23.735�-androstan-3�,17�-diol ���-diol 241 17.04 2-methoxyestrone 2-MeO-E1 444 24.57Epidihydrotestosterone Epi-DHT 434 17.46 2-hydroxy-17�-estradiol-3-methylether 2-OH-E2-3-methylether 446 25.0611-keto-androsterone 11-keto-An 520 17.5511-keto-etiocholanolone 11-keto-Etio 520 17.60 2-methoxy-17�-estradiol 2-MeO-E2 446 25.53Dehydroepiandrosterone DHEA 432 17.85 2-hydroxyestrone 2-OH-E1 502 25.92Epiandrosterone Epi-An 419 18.10 2-hydroxy-17�-estradiol 2-OH-E2 504 26.75Androstenediol A-diol 434 18.58 4-hydroxyestrone 4-OH-E1 502 27.345�-Androstanedione 5�-dione 432 18.60 4-hydroxy-17�-estradiol 4-OH-E2 504 28.42Epitestosterone Epi-T 432 18.77 17-epiestriol 17-epi-E3 504 29.195�-androstan-3�,17�-diol ���-diol 241 18.83 Estriol E3 504 29.82Dihydrotestosterone DHT 434 19.34 16-keto-17�-estradiol 16-keto-E2 487 30.06Androstenedione A-dione 430 19.80 16a-hydroxyestrone 16�-OH-E1 487 30.06Testosterone T 432 20.52 16-epiestriol 16-epi-E3 504 31.1611�-hydroxyandrosterone 11�-OH-An 522 20.73 2-hydroxyestriol 2-OH-E3 592 37.3011�-hydroxyetiocholanolone

11�-OH-Etio 522 21.05 d4-17 �-estradiolb d4-E2 420 19.88

16�-hydroxy-DHEA 16�-OH-DHEA 505 28.49d3-testosteroneb d3-T 435 20.44Methyltestosteroneb Methyl-T 446 24.21

Corticoids ProgestinsTetrahydrodeoxycortisol THS 548 34.98 5�-dihydroprogesterone 5�-DHP 445 20.01TetrahydrodeoxycorticosteroneTHDOC 550 36.26 Epipregnanolone Epi-P-one 447 23.33Tetrahydrocortisone THE 634 38.66 Pregnanolone P-one 447 23.62Tetrahydrocortisol THF 636 41.30 Allopregnanolone Allo-P-one 447 23.97Dihydrodeoxycorticosterone DHDOC 548 42.03 Pregnanediol P-diol 269 25.03Allotetrahydrocortisol Allo-THF 636 42.39 Pregnanetriol P-triol 435 26.3421-deoxycortisol 21-deoxyF 634 42.55 Pregnenolone Preg 445 27.3611-deoxycortisol 11-deoxyF 544 42.76 5�-dihydroprogesterone 5�-DHP 445 28.5711-deoxycorticosterone 11-deoxyB 546 43.51 Progesterone Prog 458 29.90Cortisone E 615 46.03 20�-hydroprogesterone 20�-DHP 445 30.2011-dehydrocorticosterone 11-dehydroB 617 46.80 17�-hydroxypregnenolone 17�-OH-Preg 548 32.68Allodihydrocorticosterone Allo-DHB 636 46.87 17�-hydroxyprogesterone 17�-OH-Prog 546 35.70Allodihydrocortisol Allo-DHF 634 47.01 11�-hydroxyprogesterone 11�-OH-Prog 531 41.32 1629

JA

mSoc

Mass

Spectrom

2009,20,

1626–1637

QU

AN

TIT

AT

IVE

STE

RO

IDSIG

NA

TU

RE

SB

YG

C-M

S

ble

1.C

onti

nued

om

po

un

ds

(tri

vial

nam

e)A

bb

revi

atio

nIo

nse

lect

eda

Ret

enti

on

tim

e(m

in)

Co

mp

ou

nd

s(t

rivi

aln

ame)

Ab

bre

viat

ion

Ion

sele

cted

Ret

enti

on

tim

e(m

in)

ort

ico

ster

on

eB

634

47.7

9d

9-p

rog

este

ron

ebd

9-P

rog

465

29.7

0o

rtis

ol

F63

247

.90

d8-

17�

-hyd

roxy

pro

ges

tero

neb

d8-

17�

-OH

-Pro

g55

235

.53

4-co

rtis

olb

d4-

F63

647

.86

Ste

rols

Ch

ole

ster

ol

Ch

ol

458

40.8

024

S-h

ydro

xych

ole

ster

ol

24S

-OH

-Ch

ol

413

48.9

9d

6-ch

ole

ster

olb

d6-

Ch

ol

464

40.5

5

anti

tati

veio

ns

asth

eT

MS

der

ivat

ives

of

ster

oid

s.in

tern

alst

and

ard

use

din

ster

oid

pro

filin

g.

1630 MOON ET AL. J Am Soc Mass Spectrom 2009, 20, 1626–1637

comparing retention times and matching the heightratios of characteristic ions (Table 1).

Method Validation

QC samples containing all 70 analytes were quantifiedusing MS peak height ratios versus the IS. Calibrationsamples were made up at 15 different concentrations,which depended on the sensitivity and reference valuesof the urinary steroids. Least-squares regression analy-sis was performed on peak height ratios at increasinganalyte levels to obtain calibration linearity. Limits ofdetection (LOD) and quantification (LOQ) were definedas the lowest concentration with a signal-to-noise (S/N)ratio � 3 for LOD and of S/N ratio � 10 for LOQ.Precisions are expressed as coefficients of variation (%CV) and accuracies as percent relative errors (% bias),and were determined using QC samples at three differ-ent concentrations (low; 2 � 50 ng/mL, medium; 20 �200 ng/mL, and high; 100 � 2000 ng/mL) dependingon the sensitivity and reference values of the urinarysteroids. To determine within-day repeatabilities, fourreplicates were analyzed, whereas day-to-day repro-ducibility was measured by running samples on 5different days.

Extraction recoveries were determined using QCsamples at three concentration levels in triplicate foreach steroid by adding known amounts of mixed work-ing solutions to steroid-free urine samples. Absoluterecoveries were calculated by comparing peak heightratios of extracted samples versus those of their nonex-tracted counterparts, using the IS added just before thederivatization step.

Statistical Analysis and Steroid Signatures

Data manipulation was performed using Excel 2007spreadsheets (Microsoft Corp., Seattle, WA, USA), Sig-maplot (SYSTAT Software Inc., San Jose, CA, USA), andSIMCA software (Umetrics Inc., Umeå, Sweden). Quan-titative results are expressed as means � SD, and groupcomparisons are made using the unpaired two-tailedStudent’s t-test. Statistical significance was consideredat P � 0.01.

To visualize differences between the steroid signa-tures of BPH patients and healthy subjects, a supervisedhierarchical clustering algorithm based on Pearson’scorrelation coefficient was used with a TIBCO SpotfireDecisionSite Browser (TIBCO Spotfire, Inc., Somerville,MA, USA). Concentrations of steroids and ratios ofmetabolites to precursors (an indicator of enzyme ac-tivity) in BPH patients and healthy subjects were com-pared using the Student’s t-test, and statistically sig-nificant variables were selected and processed. Forclustering analysis, all prefiltered quantitative results ofsteroids measured and their ratios were log trans-formed and normalized using z-scores by subtractingpopulation means from individual raw scores and di-

a C C C d

aQ

ubA

n

1631J Am Soc Mass Spectrom 2009, 20, 1626–1637 QUANTITATIVE STEROID SIGNATURES BY GC-MS

Z-score transformation [z � (observed value – baselinemedian)/baseline standard variation] ensures that eachanalyte in the subject population has a median value of0 and a standard deviation of 1. In the heat mapgenerated using HCA, color coding in maps indicatingauto range by difference of concentration in each com-pound was gradated at three point; red indicates z-scores � 0, blue z-scores � 0, and white z-scores 0 ofa row (subjects) across all columns (urinary steroids).

Results and Discussion

Urinary Steroid Profiling

Steroid hormones of significance under different biolog-ical conditions fall into several distinct groups, and arefrequently used to evaluate endocrine functions. Com-prehensive extraction methods for GC-MS based ste-roid analysis mainly focus on specific steroid metabo-lites which have the same biological function andchemical structures [15–17, 19]. Due to the chemical andphysical diversities within androgens, estrogens, corti-coids, progestins, and sterols, the described profilingtechnique was devised to simultaneously quantify ste-roid analytes as diverse as lipophilic cholesterol andhydrophilic corticoids in a single extract.

In SPE experiment, urinary steroids are purifiedusing a unique copolymer sorbent, Oasis HLB, whichallows high and reproducible results to be obtained [8,30, 32, 33]. In general, SPE absorbent retains steroidsbased on hydrophobic interactions with free and con-jugated steroids. The following two extractions meth-ods were therefore investigated: (1) 2 mL of urine wasprocessed with a SPE cartridge and the sample obtainedwas hydrolyzed, and (2) urine was first hydrolyzed andthen processed in SPE. However, difficulties in loadingsamples into SPE cartridges after hydrolyzing themwith �-glucuronidase/arylsulfatases, due to an increasein the viscosity of urinary solution hydrolyzed, wereobserved. This result was not observed if �-glucuronidaseonly was present in the enzyme solution [6]. Therefore,SPE procedure was processed first in sample prepara-tion steps.

Although catechol estrogens, such as 2-OH-E1, 2-OH-E2, 4-OH-E1, 4-OH-E2, and 2-OH-E3, are degraded whenexposed to light or to a pH higher than 9.5 [34], 0.2%aqueous L-ascorbic acid was added to prevent oxidativedecomposition before enzymatic hydrolysis. After enzy-matic hydrolysis, deconjugated steroids were extractedwith an ethyl acetate: n-hexane (2:3, vol/vol) mix. Inextraction, THF and ethyl acetate were primarily investi-gated, and diethyl ether, t-butyl methyl ether, and n-hexane were added to expedite THF and ethyl acetatepartitioning. When relatively polar organic solvents, suchas diethyl ether and t-butyl methyl ether, were used,chromatographic interference was encountered. Althoughboth THF and ethyl acetate had high extraction efficien-cies, ethyl acetate was finally selected because of the

Overall extraction efficiencies were not signifi-cantly different for all steroids studied at pH valuesof 5 to 7. The recoveries of catechol estrogens (10.4%� 48.9%) were poor at pH 9 because they are oxidizedto quinones at this pH [35]. In addition, differentratios of ethyl acetate to n-hexane were examined anda 2:3 (vol/vol) mix was chosen having consideredextraction yield and chromatographic properties. Us-ing optimized conditions, all urinary steroids studiedwere extracted in good yields, except 2-OH-E3 and24S-OH-Chol, which are still detectable in urine afterextraction (Figure 1).

Gas Chromatography-Mass Spectrometry

Using the described GC-SIM/MS conditions, 70 steroidhormones were analyzed as their TMS derivatives andpeak identification was straightforward using retentiontimes, characteristic ions, chromatograms, and massspectra, the latter were compared with those of refer-ence standards. The GC oven program used resulted inwell separated, good peak shapes for most steroids withretention times from 12.12 min for ���-diol to 48.99 minfor 24S-OH-Chol. On the other hand, 11-keto-An and11-keto-Etio, and 16-keto-E2 and 16�-OH-E1 were co-eluted and estimated totally in quantification (Figure 2).However, some partial-overlapped compounds (e.g.,5�-dione and 17�-E2, 2-OH-E2-3-methylether and P-diol, 4-OH-E1 and Preg) were differentiated by SIMusing their different characteristic ions (Table 1).

Most steroids were monitored using their molecularions as base peaks, except: ���-diol, ���-diol, ���-diol,which had an abundant [M-180; M-2OTMS]� ion at m/z256; ���-diol, ���-diol, ���-diol with an [M-195;M-2OTMS-CH3]� ion at m/z 241; 5�-DHP, 5�-DHP,Preg with an [M-15; M-CH3]� ion at m/z 445; Epi-P-one,P-one, and Allo-P-one with an [M-15]� at m/z 447;16-keto-E2, 16�-OH-E1 with an [M-15]� ion at m/z 487;THS with an [M-90; M-OTMS]� ion at m/z 548; THE andAllo-DHF with an [M-90]� ion at m/z 634; THF andAllo-THF with an [M-90]� ion at m/z 636; 11-deoxyFwith an [M-90]� ion at m/z 544; cortisone (E) with an[M-90-15; M-OTMS-CH3]� ion at m/z 615; and 11-dehydroB with an [M-15]� at m/z 617. In the cases ofP-diol, P-triol, and 11�-OH-Prog, less intense fragmentsat m/z 269, 435 and 531, respectively, were chosen toimprove selectivity rather than most intense peaks (m/z117, 255, and 546); this was attributed to urinary inter-ference. Regarding the TMS ethers of P-diol and P-triol,the characteristic ions at m/z 269 and 435 were moni-tored, which are attributed to [M-195; M-2OTMS-CH3]�

and [M-117; M-CH[OSi(CH3)3]CH3 side-chain]� ions.The base peak of P-diol is m/z 117 and other importantfragments are m/z 269 [M-90-90-15]�, m/z 284 [M-90-90]�, m/z 347 [M-117]�, and m/z 449 [M-15]�. Amongthese fragments, m/z 269 instead of m/z 117 was chosento improve selectivity from urinary backgrounds. Alsothe P-triol has m/z 255 [M-117-90-90]�, m/z 345 [M-117-

1632 MOON ET AL. J Am Soc Mass Spectrom 2009, 20, 1626–1637

by the loss of 117 mass units, but m/z 435 was monitoredto improve selectivity. In addition, 11�-OH-Prog elutesTHF at the similar retention time (41.30 min), but atypical loss of 90 Da (-OTMS) from the THF (m/z 636)would interfere the detection of 11�-OH-Prog in case ofusing m/z 546, the molecular ion. Therefore, m/z 531[M-90-15]�, which rarely did not exist in THF, waschosen instead of m/z 546 (see Supplementary Figure 1).

Method Validation

Method validation requires the evaluations of linearity,LOD, LOQ, precision, and accuracy using calibrationsamples prepared from in-house steroid-free urine [30].Calibration curves were generated for all analytes usingQC samples fortified with all reference standards at

Figure 1. Extraction recoveries of 70 urinary sterecoveries were calculated by comparing peanon-extracted counterparts.

15.00 20.00 25.00

Abu

ndan

ce

Androsterone Etiocholanolone

P-diol

P-tr

5 -androstan-3 ,17 -diol

11-keto-androsterone11-keto-etiocholanolon

Figure 2. A representative total ion chroma

different concentrations. The devised method wasfound to be linear (correlation coefficient, r2 � 0.994) forall analytes, except for cholesterol (r2 � 0.983). Mostcompounds had an LOQ in the 0.05 � 2.0 ng/mL range,while those of DHEA, Epi-An, 5�-dione, ���-diol,11-dehydroB, 17�-OH-Preg, and 11�-OH-Prog rangedbetween 5.0 and 50 ng/mL (Table 2).

Assay precisions and accuracies were determined byanalyzing three QC samples at different concentrationsof the individual steroids (5, 20, 100, 200, 500, and 2000ng/mL). Intra-day (n � 4) precisions (expressed as %CV) ranged from 0.9% to 11.2%, whereas accuracies(expressed as % bias) ranged from 95.8 to 119.1%, andinter-day (n � 5) precisions (% CV) and accuracies (%bias) ranged from 1.6% to 10.5% and from 91.6% to115.5%, respectively (Table 2). Especially, the accuracy

analyzed using the described method. Absoluteight ratios of extracted samples versus their

0.00 35.00 40.00 45.00(min)

THF

Allo-THF

THS 17α-OH-preg

THE

Cortisone

Cortisol6-keto-17 -estradiol6 -hydroxyestrone

24S-hydroxycholesterol

m obtained from a healthy male subject for

roidsk he

3Time

iol

e11

togra

1633J Am Soc Mass Spectrom 2009, 20, 1626–1637 QUANTITATIVE STEROID SIGNATURES BY GC-MS

Table 2. The validation results of overall method with intra- and inter-day assays

Compounds (trivial name)LODa

(ng/mL)LOQb

(ng/mL)Calibration

range (ng/mL)Linearity

(r2)

intra-day (n � 4) inter-day (n � 5)

CVc

(%)Accuracyc

(%)CV(%)

Accuracy(%)

Androgens���-diol 0.5 1.0 1–200 0.998 5.9 104.0 3.8 97.9���-diol 0.2 0.5 0.5–200 0.998 5.7 107.8 3.1 105.0An 0.1 0.5 0.5–3000 0.996 3.2 119.1 3.5 108.8Etio 0.1 0.5 0.5–3000 0.998 2.5 106.1 2.2 109.7���-diol 0.5 1.0 1–200 0.998 6.8 111.4 3.3 109.6���-diol 0.2 0.5 0.5–200 0.998 4.4 106.2 3.0 109.1���-diol 1.0 2.0 2–200 0.995 5.2 107.5 3.4 103.4Epi-DHT 0.1 0.2 0.2–200 0.994 8.0 106.0 3.4 108.911-keto-An & Etio 0.1 0.5 0.5–1000 0.996 3.7 108.4 3.4 112.3DHEA 5.0 50.0 50–3000 0.997 3.5 106.6 3.3 108.2Epi-An 2.0 5.0 5–1000 0.994 2.5 98.0 2.1 109.5A-diol 0.5 2.0 2–200 0.996 7.4 116.3 10.5 97.25�-dione 0.5 5.0 10–200 0.995 4.8 106.2 2.2 111.8Epi-T 0.05 0.1 0.1–200 0.997 3.8 100.7 1.6 100.8���-diol 10.0 20.0 20–500 0.998 6.9 104.0 3.0 104.0DHT 0.1 0.2 0.2–200 0.998 4.7 109.6 2.4 102.7A-dione 0.05 0.1 0.1–200 0.999 5.7 103.4 2.4 101.7T 0.05 0.1 0.1–200 0.998 4.7 106.1 1.9 102.511�-OH-An 0.2 0.5 0.5–3000 0.997 3.4 104.6 5.8 112.911�–OH-Etio 0.2 0.5 0.5–3000 0.998 2.7 101.3 4.1 113.416�-OH-DHEA 0.1 0.2 0.2–1000 0.996 4.0 104.3 1.7 105.8

Estrogens17�-E2 0.05 0.1 0.1–200 0.998 3.9 98.6 2.6 96.3E1 0.1 0.2 0.2–200 0.998 2.9 99.0 2.0 97.317�-E2 0.05 0.1 0.1–200 0.997 3.5 99.8 2.3 97.22-MeO-E2-3-methylether 0.1 0.2 0.2–200 0.998 2.9 97.3 2.1 96.44-MeO-E2 0.2 0.5 0.5–200 0.998 2.1 98.7 1.8 95.62-MeO-E1 0.05 0.1 0.1–200 0.998 1.9 99.0 1.8 95.82-OH-E2-3-methylether 0.1 0.2 0.2–200 0.998 4.2 101.3 2.4 97.62-MeO-E2 0.05 0.1 0.1–200 0.998 4.3 100.2 2.2 97.12-OH-E1 0.1 0.2 0.2–200 0.998 4.9 104.7 3.7 99.52-OH-E2 0.1 0.2 0.2–200 0.998 6.1 106.4 3.4 99.34-OH-E1 0.1 0.2 0.2–200 0.998 2.9 103.2 2.5 98.84-OH-E2 0.1 0.2 0.2–200 0.998 4.6 100.1 4.2 97.317-epi-E3 0.5 1.0 1–200 0.997 2.4 106.7 2.4 100.3E3 0.2 0.5 0.5–200 0.997 2.7 101.6 2.8 99.716-keto-E2 and 16�-OH-E1 0.1 0.2 0.2–500 0.998 2.7 98.7 8.8 92.516-epi-E3 0.5 1.0 1–200 0.997 3.5 103.2 2.3 100.52-OH-E3 0.5 1.0 1–100 0.997 5.0 107.4 5.3 101.0

CorticoidsTHS 0.1 0.2 0.2–200 0.997 0.9 99.2 3.0 98.4THDOC 0.2 0.5 0.5–200 0.996 7.1 100.4 4.1 100.0THE 1.0 2.0 2–500 0.997 7.6 151.4 9.9 151.3THF 0.2 0.5 0.5–200 0.998 6.6 106.1 4.9 104.0DHDOC 0.5 1.0 1–200 0.997 4.2 101.2 4.1 104.9Allo-THF 0.2 0.5 0.5–500 0.996 4.8 103.9 2.6 102.621-deoxyF 0.5 1.0 1–200 0.997 7.3 108.5 4.8 109.311-deoxyF 0.2 0.5 0.5–200 0.998 2.9 98.5 3.5 91.611-deoxyB 0.5 1.0 1–200 0.997 6.9 100.3 2.9 105.1Cortisone (E) 0.2 0.5 0.5–500 0.998 3.2 101.2 2.7 99.811-dehydroB 0.2 5.0 5–200 0.994 2.1 101.8 2.7 100.4Allo-DHB 0.5 1.0 1–200 0.997 6.9 113.1 5.3 108.0Allo-DHF 1.0 2.0 2–200 0.997 2.3 97.5 2.9 107.3Corticosterone (B) 0.2 0.5 0.5–200 0.997 3.3 99.0 3.3 100.3Cortisol (F) 0.1 0.2 0.2–200 0.998 3.0 100.5 3.5 98.0

Progestins5�-DHP 0.1 0.2 0.2–100 0.997 9.9 109.2 3.5 110.5Epi-P-one 1.0 2.0 2–200 0.995 4.1 106.0 3.0 105.8P-one 1.0 2.0 2–200 0.997 7.4 110.1 3.3 115.5

taine

1634 MOON ET AL. J Am Soc Mass Spectrom 2009, 20, 1626–1637

of THE resulted in abnormal values, and it might beproduced by low calibration slope because of poorGC-MS properties.

Quantitative Results of Benign ProstaticHyperplasia

The validated quantitative profiling method was ap-plied to measure 70 endogenous steroids in urinesamples obtained from 59 BPH patients and 41 age-matched healthy male subjects. All quantitative results,corrected by urinary creatinine levels, were compared,and differences between patients and healthy subjectswere evaluated using the Student’s t-test. Although itwas not the purpose of this study to elucidate clinicalimplications, some features were notable, and signifi-cant differences were found between patients andhealthy subjects (see Supplementary Figure 2).

In patients, urinary levels of T, Epi-T, ���-diol,���-diol, ���-diol, 17�-E2, 11�-OH-Prog, and 16�-OH-DHEA were significantly decreased (P � 4.52 � 10–4 �2.77 � 10–13), and those of 11�-OH-Etio, ���-diol,2-OH-E3, 17�-OH-Prog, and 11-deoxyF were signifi-cantly increased (P � 4.56 � 10–7 � 9.42 � 10–18). Someother urinary steroids quantified were also significantlydifferent in patients and controls (P � 0.01). Changes inthe urinary levels of ���-diol and ���-diol were inaccord with previous findings [36, 37], but other steroidchanges observed in this study have not been reportedto date. In particular, urinary 17�-E2 and 2-OH-E3showed remarkable changes with P values of 2.77 �10–13 and 4.25 � 10–9, which might be associated withthe different biochemical roles of estrogen receptor �(ER�) and � (ER�) in various types of prostate cells [28, 38,39]. The involvements of other urinary steroids needs to

Table 2. Continued

Compounds (trivial name)LODa

(ng/mL)LOQb

(ng/mL)Cali

range

P-diol 0.5 1.0 1P-triol 0.2 0.5 0.5Preg 1.0 2.0 205�-DHP 0.2 0.5 0.5Prog 0.5 1.0 120�-DHP 0.2 0.5 0.517�-OH-Preg 5.0 10.0 2017�-OH-Prog 0.2 0.5 0.511�-OH-Prog 2.0 5.0 5

SterolsChol 1.0 2.0 5024S-OH-Chol 0.5 2.0 2

aThe limit of detection was measured S/N ratio � 3.bThe limit of quantification was measured S/N ratio � 10.cPrecision and accuracy were expressed as the mean values of data ob

future study.

Steroid Signatures by Hierarchical ClusteringAnalysis

The present MS-based steroid profiling has become anessential clinical tool and has produced many results inBPH, but it is difficult to identify critical differencesbetween patients and controls because of the largenumber of results. Hierarchical clustering, which in-volves calculating the distance matrices of data objectsand then merging objects that are close to each other toform sub-clusters, has been utilized in biomedical stud-ies to interpret the expressions of biomolecules [2,22–25]. In the present study, after determining therelative levels of each steroid, all data were z-scoretransformed to produce a heat map using a supervisedHCA based on Pearson’s correlation coefficient. Theraw data were the relative concentrations of the 70urinary steroids, and these were filtered by statisticalsignificance (P � 0.01) using the Student’s t-test. Theclustering method and similarity measure were setunweighted average and Euclidean distance, respec-tively. Urinary concentrations were represented by col-ors in the heat map, and each steroid was representedby a single row of colored boxes, whereas columnsrepresented different subjects. Heat maps were com-pared with respect to individual quantitative results,and indicated the effectiveness of steroid signatures.

Figure 3 illustrates in a heat map showing alteredconcentrations of urinary steroids in BPH patients andhealthy subjects. The heat map readily shows changes inthe concentrations of urinary steroids. Increased urinaryexcretion patterns of several steroids, from Epi-DHTdown to 11-deoxyF, were detected in patients, while othersteroid levels were significantly reduced. In addition,visual inspections of steroid signatures complementedquantitative statistics. Although 2-OH-E3 was present at asignificantly higher level in patients (P � 4.25 � 10–9; see

onmL)

Linearity(r2)

intra-day (n � 4) inter-day (n � 5)

CVc

(%)Accuracy

c (%)CV(%)

Accuracy(%)

0.997 6.2 105.6 3.6 107.50.999 6.0 102.8 2.3 105.50.995 8.8 104.1 3.5 98.50.995 7.0 99.2 3.8 104.00.998 4.8 96.0 2.5 98.60.997 9.1 101.7 4.9 103.80.998 3.6 98.7 3.0 102.90.996 3.1 95.8 2.3 96.10.996 11.2 103.9 7.3 94.3

0 0.983 7.3 113.0 7.8 115.00.994 6.5 96.3 4.1 100.9

d from three different concentrations for each analyte.

brati(ng/

–500–500–200–200–100–100–500–200–500

–100–200

Supplementary Figure 2), these increases were only ob-

1635J Am Soc Mass Spectrom 2009, 20, 1626–1637 QUANTITATIVE STEROID SIGNATURES BY GC-MS

served in some patients in the heat map. The levels ofother steroids, 11�-OH-Etio and 11-deoxyF, were alsoinconsistency different in patients (Figure 3).

To illustrate further the usefulness of a steroid sig-nature, the ratio of steroid metabolite to precursor,which reflects enzyme activity (Scheme 1), was exam-ined. The cluster and heat maps of all analyzed sam-ples in this 70 compound study showed differencesbetween groups in these ratios (Figure 4). A dendro-gram of ratios revealed that the oxidoreductasesformed a cluster (5�-reductase, 3�-HSD, 3�-HSD, and17�-HSD, except for 20�-HSD; see SupplementaryTable 1 for full enzyme names), which are differenti-ated with a cytochrome P450 enzyme, 11�-hydroxy-lase. These results support the validity of the datatransformation, since 5�-reductase is a marker of

Healthy

Figure 3. Hierarchically clustered heat map ofhealthy male subjects. Differences between steusing Student’s t-test, and only those that werselected. Row represents subjects and column ucolored based on concentrations in urinary samblue showed a concentration in urine.

Healthy

Figure 4. A heat map for measuring enzymeprecursor to metabolite for each steroid was csignificant at the P � 0.01 level were selected. R

BPH [27–29], and 17�-HSD is positively expressed inprostate cells [40, 41]. Other up-regulated enzymes inthe map might be biomarkers of BPH. Furthermore,the method can be used to visualize quantitativesteroid data in a single graphic using a hierarchicallyclustered heat map. The need for this type of visual-ization arises from the complexity of large experi-mental datasets. One important advantage of using aheat map is that it can be used to visualize clinicaldata across multiple subjects for many analytes,which allows patterns to be easily identified.

Conclusions

Due to the clinical requirement to quantify trace levelsof urinary endogenous steroids, we devised an analyt-

BPH

rinary steroids levels of 59 BPH patients and 41levels in controls and patients were examinednificantly different at the P � 0.01 level werey concentration of individual steroids. Cells areRed represents highly excreted into urine and

BPH

ties related to steroid metabolism. The ratio ofated, and only steroids with level differencesrepresent subjects and column metabolic ratio.

the uroide sigrinarples.

activialculows

1636 MOON ET AL. J Am Soc Mass Spectrom 2009, 20, 1626–1637

ical method to measure the individual concentrations of70 urinary steroids and several enzyme activities basedon steroid signatures using a single assay, which sub-stantially reduces analytical time and sample size. Incontrast to classical biochemical approaches using RIAand EIA that mainly focus on single enzymes, singlemetabolic reactions, and kinetic properties, profilinganalysis involves the collation of quantitative results fora broad series of metabolites to gain an overall under-standing of metabolism.

Steroid signatures as represented by hierarchicallyclustered heat maps use color to communicate under-lying quantitative results, which are much harder toappreciate when presented numerically. This method issimple to implement and maintains the biological in-tegrity of data. Using this technique, inter-relationshipsbetween absolute concentrations and enzyme activitiesin samples from different sources can be recognized.The combined GC-MS profiling technique describedhas adequate sensitivity that could be used for highthroughput analysis, and could be adopted to enablethe quantifications of urinary steroids and indicatemetabolic alterations at biologically relevant levels.

AcknowledgmentsThe authors acknowledge support for this work by an intramuralgrant from the Korean Institute of Science and Technology (KIST),and by grants from the National R and D Program of the KoreanMinistry of Education, Science, and Technology (MEST), and theKorean Science and Engineering Foundation (KOSEF).

Appendix ASupplementary Material

Supplementary material associated with this articlemay be found in the online version at doi:10.1016/j.jasms.2009.04.020.

References1. Nicholson, J.; Lindon, J.; Holmes, E. ‘Metabonomics’: Understanding the

metabolic responses of living systems to pathophysiological stimuli viamultivariate statistical analysis of biological NMR spectroscopic data.Xenobiotica 1999, 29, 1181–1189.

2. Kaddurah-Daouk, R.; Kristal, B. S.; Weinshilboum, R. M. Metabolomics:a global biochemical approach to drug response and disease. Annu. Rev.Pharmacol. Toxicol. 2008, 48, 653–683.

3. Dalle-Donne, I.; Rossi, R.; Colombo, R.; Giustarini, D.; Milzani, A.Biomarkers of oxidative damage in human disease. Clin. Chem. 2006, 52,601–623.

4. Lee, S. H.; Woo, H. M.; Jung, B. H.; Lee, J. A.; Kwon O. S.; Pyo, H. S.;Choi, M. H.; Chung, B. C. Metabolomic approach to evaluate thetoxicological effects of nonylphenol with rat urine. Anal. Chem. 2007, 79,6102–6110.

5. Falk, R. T.; Xu, X.; Keefer, L.; Veenstra, T. D.; Ziegler, R. G. A liquidchromatography-mass spectrometry method for the simultaneous mea-surement of 15 urinary estrogens and estrogen metabolites: assayreproducibility and interindividual variability. Cancer Epidemiol. Biomar-kers Prev. 2008, 17, 3411–3418.

6. Cho, H.-J.; Kim, J. D.; Lee, W.-Y.; Chung, B. C.; Choi, M. H. Quantitativemetabolic profiling of 21 endogenous corticosteroids in urine by liquidchromatography-triple quadrupole-mass spectrometry. Anal. Chim.Acta 2009, 632, 101–108.

7. Falkenstein, E.; Tillmann, H. C.; Christ, M.; Feuring, M.; Wehling, M.

8. Choi, M. H.; Hahm, J. R.; Jung, B. H.; Chung, B. C. Measurement ofcorticoids in the patients with clinical features indicative of mineralo-corticoid excess. Clin. Chim. Acta 2002, 320, 95–99.

9. Caulfield, M. P.; Lynn, T.; Gottschalk, M. E.; Jones, K. L.; Taylor, N. F.;Malunowicz, E. M.; Shackleton, C. H. L.; Reitz, R. E.; Fisher, D. A. Thediagnosis of congenital adrenal hyperplasia in the newborn by gaschromatography/mass spectrometry analysis of random urine speci-mens. J. Clin. Endocrinol. Metabol. 2002, 87, 3682–3690.

10. Choi, M. H.; Yoo, Y. S.; Chung, B. C. Biochemical roles of testosteroneand epitestosterone to 5�-reductase as indicators of male-pattern bald-ness. J. Invest. Dermatol. 2001, 116, 57–61.

11. Wong, T.; Shackleton, C. H.; Covey, T. R.; Ellis, G. Identification of the steroidsin neonatal plasma that interfere with 17�-hydroxyprogesterone radioimmu-noassays. Clin. Chem. 1992, 38, 1830–1837.

12. Minutti, C. Z.; Lacey, J. M.; Magera, M. J.; Hahn, S. H.; McCann, M.;Schulze, A.; Cheillan, D.; Dorche, C.; Chace, D. H.; Lymp, J. F.;Zimmerman, D.; Rinaldo, P.; Matern, D. Steroid profiling by tandemmass spectrometry improves the positive predictive value of newbornscreening for congenital adrenal hyperplasia. J. Clin. Endocrinol. Metabol.2004, 89, 3687–3693.

13. Nelson, R. E.; Grebe, S. K.; O’Kane, D. J.; Singh, R. J. Liquid chromatography-tandem mass spectrometry assay for simultaneous measurement of estradioland estrone in human plasma. Clin. Chem. 2004, 50, 373–384.

14. Santen, R. J.; Demers, L.; Ohorodnik, S.; Settlage, J.; Langecker, P.;Blanchett, D.; Goss, P. E.; Wang, S. Superiority of gas chromatography/tandem mass spectrometry assay (GC/MS/MS) for estradiol for mon-itoring of aromatase inhibitor therapy. Steroids 2007, 72, 666–671.

15. Hsing, A. W.; Stanczyk, F. Z.; Bélanger, A.; Schroeder, P.; Chang, L.;Falk, R. T.; Feras, T. R. Reproducibility of serum sex steroid assays inmen by RIA and mass spectrometry. Cancer Epidemiol. Biomarkers. Prev.2007, 16, 1004–1008.

16. Wood, L.; Ducroq, D. H.; Fraser, H. L.; Gillingwater, S.; Evans, C.;Pickett, A. J.; Rees, D. W.; John, R.; Turkes, A. Measurement of urinaryfree cortisol by tandem mass spectrometry and comparison with resultsobtained by gas chromatography-mass spectrometry and two commer-cial immunoassays. Ann. Clin. Biochem. 2008, 45, 380–388.

17. Palermo, M.; Gomez-Sanchez, C.; Roitman, E.; Shackleton, C. H. L.Quantitation of cortisol and related 3-oxo-4-ene steroids in urine usinggas chromatography/mass spectrometry with stable isotope-labeledinternal standards. Steroids 1996, 61, 583–589.

18. Homma, K.; Hasegawa, T.; Masumoto, M.; Takeshita, E.; Watanabe, K.;Chiba, H.; Kurosawa, T.; Takahashi, T.; Matsuo, N. Reference values forurinary steroids in Japanese newborn infants: gas chromatography/mass spectrometry in selected ion monitoring. Endocrinol. J. 2003, 50,783–792.

19. Meffre, D.; Pianos, A.; Liere, P.; Eychenne, B.; Cambourg, A.; Schuma-cher, M.; Stein, D. G.; Guennoun, R. Steroid profiling in brain andplasma of male and pseudopregnant female rats after traumatic braininjury: analysis by gas chromatography/mass spectrometry. Endocrinol-ogy 2007, 148, 2505–2517.

20. Moon, J.-Y.; Jung, H.-J.; Moon, M. H.; Chung, B. C.; Choi, M. H.Inclusion complex-based solid-phase extraction of steroidal compoundswith entrapped beta-cyclodextrin polymer. Steroids 2008, 73, 1090–1097.

21. Semmes, O. J. Defining the role of mass spectrometry in cancerdiagnostics. Cancer Epidemiol. Biomarkers Prev. 2004, 13, 1555–1557.

22. Rhodes, D. R.; Chinnaiyan, A. M. Integrative analysis of the cancertranscriptome. Nat. Genet. 2005, 37, S31–S37.

23. Ivshina, A. V.; Geroge, J.; Senko, O.; Mow, B.; Putti, T. C.; Smeds, J.;Lindahl, T.; Pawitan, Y.; Hall, P.; Nordgren, H.; Wong, J. E. L.; Liu, E. T.;Bergh, J.; Kuznetxov, V. A.; Miller, L. D. Genetic reclassification ofhistologic grade delineates new clinical subtypes of breast cancer.Cancer Res. 2006, 66, 10292–10301.

24. Ray, S.; Britschgi, M.; Herbert, C.; Takeda-Uchimura, Y.; Boxer, A.;Blennow, K.; Friedman, L. F.; Galasko, D. R.; Jutel, M.; Karydas, A.;Kaye, J. A.; Leszek, J.; Miller, B. L.; Minthon, L.; Quinn, J. F.; Rabinovici,G. D.; Robinson, W. H.; Sabbagh, M. N.; So, Y. T.; Sparks, D. L.; Tabaton,M.; Tinklenberg, J.; Yesavaga, J. A.; Tibschirani, R.; Wyss-Coray, T.Classification and prediction of clinical Alzheimer’s diagnosis based onplasma signaling proteins. Nat. Med. 2007, 13, 1359–1362.

25. Denkert, C.; Budczies, J.; Kind, T.; Weichert, W.; Tablack, P.; Sehouli, J.;Niesporek, S.; Konsgen, D.; Dietel, M.; Fiehn, O. Mass spectrometry-based metabolic profiling reveals different metabolite patterns in inva-sive ovarian carcinomas and ovarian borderline tumors. Cancer Res.2006, 66, 10795–10804.

26. Dettmer, K.; Aronov, P. A.; Hammock, B. D. Mass spectrometry-basedmetabolomics. Mass Spectrom. Rev. 2007, 26, 51–78.

27. Choi, M. H.; Kim, J. N.; Chung, B. C. Rapid HPLC-electrospray tandemmass spectrometric assay for urinary testosterone and dihydrotestos-terone glucuronides from patients with benign prostate hyperplasia.Clin. Chem. 2003, 49, 322–325.

28. McPherson, S. J.; Ellem, S. J.; Simpsn, E. R.; Patchev, V.; Fritzemeier,K.-H.; Risbridger, G. P. Essential role for estrogen receptor � instromal-epithelial regulation of prostatic hyperplasia. Endocrinology2007, 148, 566–574.

29. Luu-The, V.; Bélanger, A.; Labrie, F. Androgen biosynthetic pathwaysin the human prostate. Best Pract. Res. Clin. Endocrinol. Metabol. 2008, 22,207–221.

30. Cho, Y. -D.; Choi, M. H. Alternative sample preparation techniques in

1637J Am Soc Mass Spectrom 2009, 20, 1626–1637 QUANTITATIVE STEROID SIGNATURES BY GC-MS

31. Husdan, H.; Rapoport, A. Estimation of creatinine by the Jaffe reaction.A comparison of three methods. Clin. Chem. 1968, 14, 222–238.

32. Choi, M. H.; Kim, K.-R.; Hong, J. K.; Park, S. J.; Chung, B. C.Determination of nonsteroidal estrogens in breast milk, plasma, urine,and hair by gas chromatography/mass spectrometry. Rapid Commun.Mass. Spectrom. 2002, 16, 2221–2228.

33. Reddy, S.; Iden, C. R.; Brownawell, B. J. Analysis of steroid conjugatesin sewage influent and effluent by liquid chromatography-tandem massspectrometry. Anal. Chem. 2005, 77, 7032–7038.

34. Xu, X.; Veenstra, T. D.; Fox, S. D.; Roman, J. M.; Issaq, H. J.; Falk, R.;Saavedra, J. E.; Keefer, L. K.; Ziegler, R. G. Measuring fifteen endogenousestrogens simultaneously in human urine by high-performance liquidchromatography-mass spectrometry. Anal. Chem. 2005, 77, 6646–6654.

35. Bolton, J. L.; Thatcher, G. R. J. Potential mechanisms of estrogenquinone carcinogenesis. Chem. Res. Toxicol. 2008, 21, 93–101.

36. Wright, F.; Poizat, R.; Bongini, M.; Bozzolan, F.; Doukani, A.; Mauvais-Jarvis, P. Decreased urinary 5�-androstane-3�, 17 �-diol glucuronide

37. Bauman, D. R.; Steckelbroeck, S.; Peehl, D. M.; Penning, T. M. Transcriptprofiling of the androgen signal in normal prostate, benign prostatichyperplasia, and prostate cancer. J. Clin. Endocriol. Metabol. 2006, 147,5806–5816.

38. Tsurusaki, T.; Aoki, D.; Kanetake, H.; Inoue, S.; Muramatsu, M.;Hishikawa, Y.; Koji, T. Zone-dependent expression of estrogen recep-tors &;� and � in human benign prostatic hyperplasia. J. Clin. Endocriol.Metabol. 2003, 88, 1333–1340.

39. Prins, G. S.; Korach, K. S. The role of estrogens and estrogen receptorsin normal prostate growth and disease. Steroids 2008, 73, 233–244.

40. Martel, C.; Rhéaume, E.; Takahashi, M.; Trudel, C.; Couét, J.; Luu-The,V.; Simard, J.; Labrie, F. Distribution of 17 �-hydroxysteroid dehydro-genase gene expression and activity in rat and human tissues. J. SteroidBiochem. Mol. Biol. 1992, 41, 597–603.

41. Délos, S.; Carsol, J. -L.; Fina, F.; Raynaud, J.-P.; Martin, P.-M. 5�-Reductase and 17 �-hydroxysteroid dehydrogenase expression in