ANALYTICAL

BIOCHEMISTRY
Analytical Biochemistry 334 (2004) 403–405
[Link]/locate/yabio

Notes & Tips

Quantifying trimethylamine and trimethylamine-N-oxide

in human plasma: interference from endogenous
quaternary ammonium compounds
Marcus A. Baina,*, Randall Faullb, Gianfranco Fornasinic, Robert W. Milnea,
Russell Schumannd, Allan M. Evansa
a
Centre for Pharmaceutical Research, School of Pharmacy and Medical Sciences, University of South Australia, Adelaide, SA 5000, Australia
b
Renal Unit, Royal Adelaide Hospital, Adelaide, SA, Australia
c
Sigma-Tau Pharmaceuticals Inc., Gaithersburg, MD, USA
d
Levay & Co. Environmental Services, Ian Wark Research Institute, University of South Australia, Mawson Lakes, SA, Australia

Received 1 June 2004

Available online 5 August 2004

Trimethylamine (TMA)1 is a dietary-derived tertiary mal, with helium used as the carrier gas at a flow of
amine that has the characteristic odor of rotting fish 60 mL/min. A carboxen-polydimethylsiloxane (75 lm)
[1]. TMA is efficiently absorbed from the gastrointesti- SPME fiber (Supelco), was used for analysis.
nal tract [2] and is subsequently metabolized by the fla- For the analysis of TMA, 400 lL of human plasma
vin containing monooxygenase (EC [Link]) isoform 3 was immediately acidified upon collection with 600 lL
enzyme (FMO3) in the liver to form trimethylamine-N- of hydrochloric acid (0.01 M), vortex mixed for 30 s,
oxide (TMNO) [3]. Previous studies on the quantifica- and transferred to a 4-mL glass vial containing a mag-
tion of TMA and TMNO in biological samples, using netic stirrer. Subsequently, milli-RQ water (800 lL) and
GC [4,5], high-performance liquid chromatography [6], 200 lL deuterated TMA (D9 Æ DCl salt, Cambridge Iso-
nuclear magnetic resonance spectroscopy [7], and fast tope Laboratories, Massachusetts, USA) as internal
atom bombardment mass spectrometry [8], have used standard (14.7 lM, free base equivalent) was added.
urine as the sample matrix and there are relatively few Samples (unknown plasma sample treated as above,
studies of TMA and TMNO measurements in human acidified calibration standards, and acidified quality con-
plasma [9,10]. Based on previous urine methods [4,5], trol samples) were placed on ice and 50 lL potassium
we have developed a GC method using solid-phase mic- hydroxide (5 M) was added. The vial was immediately
roextraction (SPME) to quantify TMA and TMNO in sealed with a Teflon-lined cap, vortex mixed for 30 s,
human plasma. Herein we report on artifacts that can and heated to 50
C for 5 min. The SPME fiber was in-
significantly interfere with sample analysis. serted into the headspace for 5 min and then removed
A Varian Star 3400 CX gas chromatograph (Califor- and inserted into the GC-MS. For the analysis of
nia, USA) fitted with a SPB-1 sulfur (30 m · 0.32 mm · TMNO, 300 lL of hydrochloric acid (0.01 M) was added
4.0 lm film thickness) fused silica capillary column to 50 lL of plasma in a 4-mL glass vial. Titanium(III)
(Supelco, Pennsylvania, USA) and Varian Saturn 2000 sulfate in dilute sulfuric acid (10 lL, 45% w/v) was added,
mass spectrometer were used for analysis. The final the vial was sealed and vortex mixed for 30 s, and reduc-
GC-MS conditions were splitless injection, injector tem- tion of TMNO to TMA was allowed to proceed at room
perature 250
C, and column temperature 120
C isother- temperature for 15 min. Deuterated TMA as internal
*
standard (100 lL, 14.7 lM free base equivalent) and
Corresponding author. Fax: +61-8-8302-2389. milli-RQ water (550 lL) were then added, and while on
E-mail address: [Link]@[Link] (M.A. Bain).
1
Abbreviations used: TMA, trimethylamine; FMO, flavin contain-
ice 100 lL of potassium hydroxide (5 M) was added
ing monooxygenase; TMNO, trimethylamine-N-oxide; SPME, solid- and the vial immediately sealed. The samples were then
phase microextraction. analyzed for TMA using the method above.

0003-2697/$ - see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/[Link].2004.07.004
404 Notes & Tips / Analytical Biochemistry 334 (2004) 403–405

A calibration curve for the TMA assay, prepared

from linear regression analysis of the relationship be-
tween the TMA/internal standard peak area ratio and
the concentration of TMA added to plasma was linear
over the range 0.169–33.8 lM (y = 1.6002x + 0.0109,
R2 = 0.9999). Peak areas for TMA were corrected for
the response from the basal endogenous level of TMA.
A similar linear relationship between the TMA peak
area/internal standard peak area ratio and the standard
TMNO concentration in plasma was determined for
the TMNO assay over the range 6.66–266 lM (y =
0.3296x + 0.0163, R2 = 0.9989). When calculating the
TMNO concentration in the plasma samples, the
calibration standard curve was adjusted for baseline
endogenous TMNO levels.
With previous methods, it has become standard prac-
tice for potassium hydroxide to be added to the urine
sample to favor movement of TMA into the headspace
[4,5]. An important observation from the development
of the current assay was a notable increase in the Fig. 2. Evidence of the Hofmann elimination reaction resulting in the
TMA/internal standard peak area ratio with increasing formation of trimethylamine during assay development. (A) Increasing
amounts of potassium hydroxide added to the plasma amounts of potassium hydroxide (5, 10, and 21 mmol) were added to
human plasma and a disproportionate increase in the trimethylamine/
matrix (Fig. 2A). Upon further investigation, it was re- internal standard peak area ratio (means ± SD, n = 4) was observed,
vealed that this increase was evident only with a plasma reinforcing the formation of trimethylamine from endogenous quater-
matrix and not with an aqueous matrix. This is most nary ammonium compounds. (B) Increasing amounts of potassium
likely to be due to the breakdown of endogenous quater- hydroxide (5, 10, and 21 mmol) were added to an aqueous solution of
L -carnitine (1 mM) with subsequent GC-MS analysis for trimethyla-
nary ammonium compounds such as L -carnitine and
mine. An increase in the trimethylamine/internal standard peak area
choline, with the formation of TMA via Hofmann elim- ratio (means ± SD, n = 4) was found, suggesting the formation of
ination (Fig. 1) [11]. To confirm this, an aqueous solu- trimethylamine through the Hofmann elimination reaction.
tion of L -carnitine was assayed for TMA following the
addition of increasing amounts of potassium hydroxide interference by L -carnitine was further reinforced fol-
and the results (Fig. 2B) confirmed the formation of lowing the injection of a concentrated L -carnitine aque-
TMA in a manner similar to that noted with the plasma ous solution, revealing the presence of TMA [12]. In our
matrix. The formation of ‘‘ghost’’ peaks with previous preliminary studies assessing the applicability of GC-
GC methods for quantifying TMA may in part be due MS for the analysis of TMA in human plasma, a signif-
to the reaction described above. In the method by Mac- icant overestimation of TMA was directly related to the
cari et al. [12], the authors noticed that the TMA peak amount of potassium hydroxide added. For instance,
areas were 49% higher for urine when L -carnitine was the plasma TMA concentrations in a healthy subject
not removed from the matrix by a purification step, indi- were found to be approximately 0.576 and 4.23 lM fol-
cating the potential for significant overestimation. The lowing the addition of 0.25 and 5.35 mmol of potassium

Fig. 1. The Hofmann elimination reaction [11]. When quantifying trimethylamine in biological fluids using GC, it is important to use low
concentrations of potassium hydroxide to ensure negligible conversion of quaternary ammonium compounds found in plasma, such as choline and L -
carnitine, to trimethylamine.
 Notes & Tips / Analytical Biochemistry 334 (2004) 403–405 405

hydroxide, respectively. Using the validated method, the head-space gas chromatography, J. Chromatogr. 584 (1992) 141–
mean plasma TMA and TMNO concentrations (n = 10 145.
[5] G.A. Mills, V. Walker, H. Mughal, Quantitative determination of
healthy adults) were 0.418 lM (range 0.271–0.651 lM) trimethylamine in urine by solid-phase microextraction and gas
and 37.8 lM (range 22.5–79.4 lM), respectively. The chromatography-mass spectrometry, J. Chromatogr. B 723 (1999)
plasma TMA concentration determined using this cur- 281–285.
rent assay method is lower than that reported previously [6] A. Marzo, N. Monti, M. Ripamonti, S. Muck, E. Arrigoni
where overestimation may have occurred [13]. Clearly, Martelli, Determination of aliphatic amines by gas and high-
performance liquid chromatography, J. Chromatogr. 507 (1990)
the conditions for generating TMA in the headspace 241–245.
need to be considered carefully depending on the sample [7] N.G. Abeling, A.H. Van Gennip, H.D. Bakker, A. Heerschap,
matrix under investigation. Importantly, previous GC U. Engelke, R.A. Wevers, Diagnosis of a new case of trimethyl-
methods that have used higher concentrations of potas- aminuria using direct proton NMR spectroscopy of urine, J.
sium hydroxide may have overestimated the TMA Inherit. Metab. Dis. 18 (1995) 182–184.
[8] O.A. Mamer, L. Choiniere, E. Treacy, Measurement of trimeth-
concentration. ylamine and trimethylamine N-oxide independently in urine by
fast atom bombardment mass spectrometry, Anal. Biochem. 276
(1999) 144–149.
References [9] S.R. Dunn, M.L. Simenhoff, L.G. Wesson Jr., Gas chromato-
graphic determination of free mono-, di-, and trimethylamines in
[1] M. Al-Waiz, R. Ayesh, S.C. Mitchell, J.R. Idle, R.L. Smith, A biological fluids, Anal. Chem. 48 (1976) 41–44.
genetic polymorphism of the N-oxidation of trimethylamine in [10] J.R. Cashman, J. Zhang, J. Leushner, A. Braun, Population
humans, Int. J. Clin. Pharmacol. Ther. 42 (1987) 588–594. distribution of human flavin-containing monooxygenase form 3:
[2] M. Al-Waiz, S.C. Mitchell, J.R. Idle, R.L. Smith, The metabolism gene polymorphisms, Drug Metab. Dispos. 29 (2001) 1629–1637.
of 14C-labelled trimethylamine and its N-oxide in man, Xenobio- [11] T.W.G. Solomons, Organic Chemistry, Wiley, New York, 1992.
tica 17 (1987) 551–558. [12] F. Maccari, L. Ferrari, P. Pessotto, M. Ramacci, L -Carnitine
[3] D.H. Lang, C.K. Yeung, R.M. Peter, C. Ibarra, R. Gasser, K. interference in gas chromatographic determination of total
Itagaki, R.M. Philpot, A.E. Rettie, Isoform specificity of trimeth- trimethylamine in urine: a novel assay method, J. Chromatogr.
ylamine N-oxygenation by human flavin-containing monooxy- 532 (1990) 124–129.
genase (FMO) and P450 enzymes: selective catalysis by FMO3, [13] S.H. Zeisel, K.A. Dacosta, J.T. Lamont, Mono-, di- and
Biochem. Pharmacol. 56 (1998) 1005–1012. trimethylamine in human gastric fluid: potential substrates for
[4] A.Q. Zhang, S.C. Mitchell, R. Ayesh, R.L. Smith, Determination nitrosodimethylamine formation, Carcinogenesis 9 (1988) 179–
of trimethylamine and related aliphatic amines in human urine by 181.