Avoiding the Pitfalls When Quantifying Thyroid Hormones and Their Metabolites Using Mass Spectrometric Methods: the Role of Quality Assurance

Molecular and Cellular Endocrinology xxx (2017) 1e13

Contents lists available at ScienceDirect

Molecular and Cellular Endocrinology

journal homepage: www.elsevier.com/locate/mce

Avoiding the pitfalls when quantifying thyroid hormones and their metabolites using mass spectrometric methods: The role of quality assurance

* Keith Richards, Eddy Rijntjes, Daniel Rathmann, Josef Kohrle€

Institut für Experimentelle Endokrinologie, Charite-Universitatsmedizin€ Berlin, Berlin, Germany article info abstract

Article history: This short review aims to assess the application of basic quality assurance (QA) principles in published Received 18 November 2016 thyroid hormone bioanalytical methods using mass spectrometry (MS). The use of tandem MS, in Received in revised form particular linked to liquid chromatography has become an essential bioanalytical tool for the thyroid 20 January 2017 hormone research community. Although basic research laboratories do not usually work within the Accepted 20 January 2017 constraints of a quality management system and regulated environment, all of the reviewed publications, Available online xxx to a lesser or greater extent, document the application of QA principles to the MS methods described. After a brief description of the history of MS in thyroid hormone analysis, the article reviews the Keywords: Thyroid hormone metabolites application of QA to published bioanalytical methods from the perspective of selectivity, accuracy, Tandem mass spectrometry precision, recovery, instrument calibration, matrix effects, sensitivity and sample stability. During the last 3-Iodothyronamine decade the emphasis has shifted from developing methods for the determination of L-thyroxine (T4) and 0 Iodothyronine 3,3 ,5-triiodo-L-thyronine (T3), present in blood serum/plasma in the 1e100 nM concentration range, to metabolites such as 3-iodo-L-thyronamine (3-T1AM), 3,5-diiodo-L-thyronine (3,5-T2) and 3,3’-diiodo-L- 0 thyronine (3,3 -T2). These metabolites seem likely to be present in the low pM concentrations; consequently, QA parameters such as selectivity and sensitivity become more critical. The authors conclude that improvements, particularly in the areas of analyte selectivity, matrix effect measurement/documentation and analyte recovery would be beneﬁcial. © 2017 Elsevier B.V. All rights reserved.

Contents

1. Scope of the review ...... 00 2. Background ...... 00 3. The history of mass spectrometry in thyroid hormone analysis ...... 00 4. Quality assurance and method validation ...... 00 4.1. Selectivity ...... 00 4.2. Accuracy, precision and recovery ...... 00 4.3. Instrument calibration ...... 00 4.4. Matrix effects and the use of ISTD ...... 00 4.5. Sensitivity ...... 00 4.6. Stability ...... 00

0 Abbreviations: CLIA, Chemiluminescence immunoassays; 3,5-T2, 3,5-Diiodo-L-thyronine; 3,3 -T2, 3,3‘-Diiodo-L-thyronine; ELISA, Enzyme-linked immunosorbent assay; FDA, Food and Drug Administration; GC-MS, Gas chromatography-mass spectrometry; IA, Immunoassay; 3-T1AM, 3-Iodothyronamine; ID, Isotope dilution; LC, Liquid chromatography; LC-MS, Liquid chromatography mass spectrometry; LC-MS/MS, Liquid chromatography tandem mass spectrometry; LLOD, Lower limit of detection; LLOQ,

Lower limit of quantiﬁcation; m/z, Mass/charge ratio; MS, Mass spectrometry; QA, Quality assurance; TH, Thyroid hormones; THM, Thyroid hormones and metabolites; T4,L- 0 0 3 Thyroxine, 3,3’,5,5 -Tetraiodo-L-thyronine; T3, 3,3 ,5-Triiodo-L-thyronine; SRM, Selected reaction monitoring; MRM, Multiple reaction monitoring; MS , Triple MS; ISTD, Internal standard; SPE, Solid phase extraction; nM, nanomolar (nmolL 1); pM, picomolar (pmolL 1). * Corresponding author. Charite-Universit atsmedizin€ Berlin, Institut für Experimentelle Endokrinologie, Augustenburger Platz 1, 13353 Berlin, Germany. E-mail address: [email protected] (J. Kohrle).€ http://dx.doi.org/10.1016/j.mce.2017.01.032 0303-7207/© 2017 Elsevier B.V. All rights reserved.

4.7. Equilibration ...... 00 5. Immunoassays ...... 00 6. Summary ...... 00 Funding...... 00 Conflict of interest statement ...... 00 Acknowledgements ...... 00 References ...... 00

1. Scope of the review review are equally applicable to all biological matrices.

The use of mass spectrometry (MS) for the analysis of thyroid 2. Background hormones. (TH) and TH metabolites (THM) in biological samples has THM are a group of low molecular mass iodine-containing become widespread. MS methods for the determination of TH and hormonally active compounds derived from the amino acid L- THM in human (Siekmann, 1987; Thienpont et al., 1994; De tyrosine. The pro-hormone L-thyroxine (T4, Fig. 1A), the more Brabandere et al., 1998; Tai et al., 2002, 2004; Thienpont et al., 0 biologically active 3,3 ,5-Triiodo-L-thyronine (T3, Fig. 1B) and the 1999; Van Uytfanghe et al., 2004; Soukhova et al., 2004; Gu et al., 0 di-iodometabolites 3,5-T2 and 3,3 -T2 (Fig. 1D and E respectively) 2007; Yue et al., 2008; Saba et al., 2010; DeBarber et al., 2008; belong to the thyronine family of compounds. The de-carboxylated, Galli et al., 2012; Jonklaas et al., 2014; Kunisue et al., 2011a; de-iodinated derivatives of T4, include 3-iodothyronamine (3- Wang and Stapleton, 2010; Kahric-Janicic et al., 2007; Ackermans T1AM, Fig. 1C); these are known as thyronamines. et al., 2010) and animal (Saba et al., 2010; DeBarber et al., 2008; Endogenous TH are found in tissues, blood and other body fluids Kunisue et al., 2011a; Wang and Stapleton, 2010; Ackermans and their accurate measurement in these biological matrices is a et al., 2010; Hackenmueller and Scanlan, 2012; Nagao et al., 2011; key requirement for both clinical diagnostic and research purposes. Kunisue et al., 2011b; Kiebooms et al., 2014; Hansen et al., 2016) TH and THM may exist in both free and protein-bound forms in blood serum/plasma, animal tissues (Saba et al., 2010; Ackermans biological matrices and it is desirable to have available bioanalytical et al., 2010, 2012; Kunisue et al., 2011c; Kunisue et al., 2010), cell methods to determine both free and total (i.e. free plus bound) culture lysates (Saba et al., 2010; Piehl et al., 2008) and superna- concentrations in both blood and tissues. Among low molecular tants (Rathmann et al., 2015), saliva (Higashi et al., 2011), and urine mass hormones the avid and extremely high affinity binding of TH (Aqai et al., 2012; Fan et al., 2013) have been published. Over the and THM to several binding proteins in the blood and various last 25 years the availability of affordable, high sensitivity benchtop subcellular compartments is a unique challenge to every diagnostic tandem mass spectrometers, in particular those coupled to liquid approach intending their quantification (Refetoff, 2000; Roy et al., chromatography (LC-MS/MS) and as a consequence of the manu- 2012; Little, 2016). facturer's constant search for new markets and hence new applications, the thyroid research community has access to this additional powerful tool to compliment traditional immunoassay 3. The history of mass spectrometry in thyroid hormone (IA)-based methods. The move of LC-MS/MS into new application analysis areas brings with it not only a new bioanalytical approach and gains in productivity, but also potential pitfalls. These pitfalls are perhaps T4 and T3 are 99.97 and 99.7% respectively bound to carrier more familiar to the early adopters of the technique, who tended to proteins in the circulation (Spencer, 2000). The availability of TH to be active in application areas subject to quality management sys- cells is considered to be dependent on their free rather than tems and accredited/regulated environments. The need for quality protein-bound concentrations (Cheng et al., 2010), making the assurance (QA), including method validation has been recognised concentrations of free T3 and T4 of major diagnostic interest to and taken on board by the majority of TH research groups pub- clinicians. Hence, free T4 and T3 concentrations are clinically lishing bioanalytical MS methods; however, the implementation of important parameters that are routinely measured on a high QA and the interpretation of data is variable. The limited availability throughput daily basis in blood samples from adult patients. The of certified reference materials and isotopically labelled internal accurate, precise determination of total versus free T4 and T3 in standard (ISTD) compounds and the lack of proficiency testing are blood serum present quite different challenges related to their particular limitations. The aim of this review is not to compre- respective concentration ranges: free T4 and T3 concentrations are hensively document the development of MS in the analysis of THM, in the low pM range, in contrast to the low to mid nM concentra- but to bring attention to the potential pitfalls and ambiguities and tions that are found for total T4 and T3. If one surveys the literature suggestions for how to overcome them. This is especially important pertaining to the chromatographic separation and detection of total considering that routine TH analytics in the clinical setting depend iodothyronine concentrations in biological matrices prior to the primarily on the application of highly specific TH antibodies and late 1990s, one finds isotope dilution gas chromatography mass state of the art immunoassay technology. The information pre- spectrometry (ID-GC-MS) as the method of choice, requiring deri- sented here is targeted particularly towards researchers new to, or vatisation of the analyte prior to analysis ((Thienpont et al., 1994), considering entering, the field with perhaps limited knowledge of and references therein). LC methods lacked a satisfactory MS MS. The bulk of LC-MS/MS published work in TH research concerns interface and therefore relied upon UV or fluorescence detectors the quantification of TH and THM in blood serum/plasma, so this (the latter requiring derivatisation) and were characterised by long review naturally focusses on this subject. Within the last 10 years, run times interferences and poor sensitivity (Hearn and Hancock, however, an increasing number of studies on tissues and cell cul- 1979; Burman et al., 1981). The advent of electrospray ionisation ture experiments have been reported. The points raised in this in the early 1990s opened up the possibility of analysing relatively polar and non-volatile small molecules by LC-MS/MS. The

0 Fig. 1. Chemical structures of L-T4 (A), L-T3 (B), 3-T1AM (C), 3,5-T2 (D) and 3,3 -T2 (E). development of tandem LC-MS/MS delivered the combination of 200pM or less (Saba et al., 2010; Galli et al., 2012), or close to high sensitivity and selectivity essential for the quantitation of (DeBarber et al., 2008) or below (Ackermans et al., 2010) the lower analytes in complex biological samples such as blood serum and limit of detection (LLOD) of some methods, whilst immunoassay plasma. The publication in 1998 of an LC-MS/MS method for the (IA) data suggest higher concentrations (see below). Table 2 sum- determination of total T4 in human serum that showed quantifi- marises the significant publications over the last four decades cation data comparable to that obtained by the established GC/MS relating to the mass spectrometric analysis of TH and THM in blood technique (De Brabandere et al., 1998) heralded the dawn of LC-MS/ serum/plasma. MS analysis in the thyroid hormone research community. Applied to virtually all published methods, whether GC-MS or The earliest published LC-MS/MS methods concentrated exclu- LC-MS based, is the concept of isotope dilution mass spectrometry sively on the determination of total T4 and T3 in blood serum or (ID-MS) (Lehmann, 1002). In the ideal scenario, stable isotopically 13 2 plasma (Thienpont et al., 1994; De Brabandere et al., 1998; Tai et al., labelled analogues of each analyte (e.g. C6-T4, H4-3-T1AM etc. 2002, 2004; Thienpont et al., 1999; Van Uytfanghe et al., 2004; often known collectively as “internal standards” (ISTD)) are added Soukhova et al., 2004) and were capable of quantification of (spiked) to samples, calibrators and any QA-related samples in endogenous TH in approximately the 1 nM (T3) to 100 nM (T4) known, constant concentrations. Isotopically labelled ISTD possess concentration ranges. Methods to quantify free T4 and T3 concen- physicochemical properties that are almost identical to those of the trations by LC-MS/MS in blood serum became possible with the native (unlabelled) substance. In particular, chromatographic availability of enhanced sensitivity tandem MS (Gu et al., 2007; Yue retention times and MS ionisation properties (for the latter this et al., 2008; Kahric-Janicic et al., 2007). The quantification of free TH includes any potential ionisation enhancement or suppression ef- in blood serum by tandem MS and its performance compared with fects from co-eluting sample matrix, so called “matrix effects”)are immunoassay has been the subject of a 2011 review (Soldin and assumed to be virtually indistinguishable between ISTD and ana- Soldin, 2011) hence it will not be covered in more detail here. lyte. Crucially however, due to differing molecular masses between More recently, the focus has been extended to include emerging analyte and ISTD, they can be separately detected within a single THM. According to reports thus far, the identification and quanti- MS method. Thus, any errors in accuracy incurred by losses of an- 0 fication of emerging metabolites such as 3,3 -T2 (Jonklaas et al., alyte during sample extraction and clean up, fluctuations in sample 2014), and 3,5-T2 (Hansen et al., 2016) also require quantification volume injected into the LC- or GC-MS system and matrix effect in the low pM range, although due to the absence of information on differences between calibrators and samples within a single batch potential binding proteins it is unclear as to whether free or total analysis or between batches can be corrected by taking into account THM concentrations are being measured. At this point, there is little the instrument response of the ISTD. In practice this is achieved for agreement (Hoefig et al., 2016) as to the likely endogenous con- chromatographic instruments by using analyte peak area/ISTD peak centration of another of the emerging THM - 3-T1AM - in blood area (known as the “peak area ratio”) as the detector response, serum, although LC-MS/MS reports so far suggest it to be circa. rather than simply analyte peak area. Analytical methods that

Please cite this article in press as: Richards, K., et al., Avoiding the pitfalls when quantifying thyroid hormones and their metabolites using mass spectrometric methods: The role of quality assurance, Molecular and Cellular Endocrinology (2017), http://dx.doi.org/10.1016/j.mce.2017.01.032 4 K. Richards et al. / Molecular and Cellular Endocrinology xxx (2017) 1e13 incorporate ID principles will deliver significantly superior accu- monitoring (MRM). Generally, the precursor/fragment ion transi- racy compared to non-ID methods for complex (dirty) sample tion producing the largest intensity signal is used for quantification matrices. of the analyte (the “quantifier” ion); the less intense transition (the “confirmation” ion) is used to confirm analyte identification. 4. Quality assurance and method validation Providing MS instrument parameters remain stable, the relative intensities of quantifier/confirmation ions (as measured by the ra- QA is a term associated with analytical chemistry, but less so tio of peak areas from MRM chromatograms) remains constant, with basic research. Whilst it is unreasonable to expect thyroid regardless of analyte concentration. Peak identification can be hormone researchers to apply full quality management procedures confirmed by constant measurement of this ratio, applying an to their laboratory environment, it is appropriate to expect an allowable tolerance (often set at 20%). Any analyte producing an adequate standard of QA to be applied to published bioanalytical MRM peak ratio outside the allowable tolerance must be rejected MS methods. What should such a standard include? The US Food either as falsely identified or as subject to interfering co-eluting and Drug Administration (FDA) guidelines for bioanalytical method compounds, making detection unreliable. Most modern MS data validation (B.C. Committee, 2001) state that analyte selectivity, analysis software provide automated procedures for evaluating accuracy, precision, recovery, instrument calibration, sensitivity MRM ion ratios. The use of quantifier/confirmation ions in tandem (i.e. limits of detection) and analyte stability (e.g. freeze/thaw of MS has become standard practice in environmental monitoring for samples, storage of analyte stock solutions) should be evaluated. trace pollutants (Bondoux & Joumier, 2012). Rigorous adoption of These basic requirements would appear to be quite reasonable to this QA check would decrease uncertainty when confirming the achieve within the confines of research institutes active in TH identification of low concentration endogenous THM in biological research and essential to ensure accurate and precise reporting of samples. Fig. 2A and B and Table 1 show a real example for T2 TH and THM concentrations. This review will therefore discuss the isomers in an extracted blood serum sample; samples have been above-mentioned topics and access the extent to which they are accepted/rejected based on the 20% tolerance for MRM peak ratio being successfully addressed in publications. A further topic, criteria. From the respective MRM chromatograms, it is clear that particular to the determination of endogenously occurring hor- for the rejected sample m/z 526.0 / 352.9 shows a contribution mones, is that of sample equilibration. This is included in this from a co-eluting contaminant (panel A), this interference is not section, although strictly speaking it is not a QA parameter. present in the chromatogram for the accepted sample (panel B). There has been much interest in the detection and quantifica- 4.1. Selectivity tion of endogenous 3-T1AM in blood serum since its discovery (Scanlan et al., 2004). A survey of the available literature on the Selectivity is the most important parameter to be tested and determination of this THM in serum demonstrates the challenges should always precede any quantification e the entire analytical faced when attempting to confirm the presence of trace (defined in determination clearly relies upon this and without it, there is this case as sub 1 nM) concentration levels of endogenous hormone simply no point in subjecting a mis-identified analyte to further metabolites. To the knowledge of the authors there has not been a rigorous QA procedures. In the tandem mass spectrometer, an an- single published confirmation of the identification of endogenous alyte molecular ion of m/z X (the so-called “precursor ion”) is iso- 3-T1AM in blood serum/plasma using the MRM ion ratio procedure lated in the first quadrupole section, and then transported into the outlined above, even though in some cases 3-T1AM concentrations collision cell, where it is fragmented to produce fragment ions. A were reported. One published method appeared to monitor only single fragment ion (generally one with high abundance) of m/z Y is one MRM transition for 3-T1AM (Nguyen et al., 2011), whilst for then selected in the third quadrupole section before being detected. another, two or more MRM transitions were monitored, but the Such single mass transitions measurements per analyte are called result of only one was illustrated (Saba et al., 2010; Galli et al., selected reaction monitoring (SRM). In combination with chro- 2012). For yet another, spiked calibrator samples displayed two matographic retention time, one SRM measurement per analyte usable MRM transitions whose area ratios showed consistency may be sufficient for unambiguous identification at relatively high (n ¼ 31, %RSD <10%, although the actual MRM ratio was not pre- analyte concentrations, since the peak intensities will be corre- sented); however endogenous 3-T1AM concentrations in serum spondingly high. This is the case, for example with endogenous T4, were at the detection limit and one MRM transition suffered from which may be present at ca. 100 nM in human blood serum. interferences making it unreliable (DeBarber et al., 2008). Authors Additionally, since background interferences in the sample and/or very recently publishing a multi-analyte method, adopted the in the LC-MS/MS itself tend to reduce with increasing m/z, T4 has a quantifier/confirmation MRM peak ratio with 20% tolerance molecular mass sufficiently high enough to give a pseudo- criteria, but were only able to obtain a single precursor ion to molecular ion (by LC-MS/MS at m/z 778) that is not subject to fragment ion transition for 3-T1AM (Hansen et al., 2016). These excessive chemical noise, making chromatographic peak detection observations would seem to cast doubt as to whether 3-T1AM has and quantification relatively unproblematic. This is not the case, for really been reliably detected at endogenous concentrations in instance, for 3-T1AM. 3-T1AM has a molecular mass (355 Da) low blood serum by LC-MS/MS. enough to be subject to high chemical noise from multiple in- For cases where co-eluting interferences cannot be removed terferences from other endogenous substances in biological sam- using additional clean-up procedures, or where the chemical ples. In addition, endogenous concentrations of 3-T1AM in serum background is simply so high that analyte MRM signals are are likely to be considerably lower (Hoefig et al., 2016) than that of “swamped”, one alternative approach is to confirm the identity of T4, resulting in weak intensity peaks, that may not be unambigu- analytes by applying an extra MS dimension. Some triple quadru- ously identified and quantified. For such analytes it is necessary to poles incorporate a linear ion trap capability in quadrupole three simultaneously measure a second precursor ion/fragment ion SRM. (Hager, 2002) enabling MS3. Fig. 3. Shows the MS3 spectrum m/z 2 This greatly increases the specificity of the measurement, since the 360.1/m/z 342.8/m/z 50e345 for H4-3T1AM. Clearly, the ion at þ probability of an interfering substance at the correct retention time m/z 216.0 (representing [M þ HeNH3eI] ) is a fragment of m/z producing not only a precursor ion of m/z A leading to a fragment 342.8, rather than a fragment arising directly from the pseudo ion of m/z B, but also an additional fragment ion of m/z C, is molecular ion m/z 360.1 (m/z 212.0 / 338.8/356.1 would be the 3 extremely small. This type of MS scan is termed multiple reaction equivalent ions for unlabelled 3-T1AM). Therefore, the MS scan m/

Fig. 2. MRM chromatograms for “Serum_1” (A) and “Serum_2” (B) samples in Table 1.

Table 1

MRM peak area ratio check for 3,5-T2 and 3-T1AM in blood serum sample.

Sample name Sample type Analyte name MRM cResult peak area ratio test aPeak area ratio bExpected peak area ratio Peak area

SERUM_1 Unknown 3,5-T2 526.0/479.8 Fail 0.925 0.245 2.64Eþ04 SERUM_1 Unknown 3,5-T2 526.0/352.9 2.44Eþ04 SERUM_2 Unknown 3,5-T2 526.0/479.8 Pass 0.261 0.245 7.86Eþ04 SERUM_2 Unknown 3,5-T2 526.0/352.9 2.05Eþ04 STANDARD _1 Standard 3,5-T2 526.0/479.8 Pass 0.245 0.245 2.77Eþ05 STANDARD _1 Standard 3,5-T2 526.0/352.9 6.79Eþ04 STANDARD _1 Standard 3-T1AM 356.1/338.8 Pass 0.836 0.836 1.39Eþ05 STANDARD _1 Standard 3-T1AM 356.1/212.1 1.16Eþ05 SERUM_10pM_1 Unknown 3-T1AM 356.1/338.8 Fail 0.593 0.836 3.57Eþ04 SERUM_10pM_1 Unknown 3-T1AM 356.1/212.1 2.12Eþ04 a Peak Area Analyte Name 2/Peak Area Analyte Name 1. b Peak Area Analyte Name 2/Peak Area Analyte Name 1 for STANDARD_1. c Result ¼ ((Peak Area Ratio e Expected Peak Area Ratio)/Expected Peak Area Ratio) x 100 If Result >±20%: Pass. If Result <±20%: Fail.

z 356.1 / 338.8/212.0 is expected to be highly specific for the 3-T1AM (Fig. 4. Panel A), however high chemical background noise confirmation of the presence of 3-T1AM in biological samples. Fig. 4. contributes to the peak areas, causing the MRM peak area ratio to Shows MRM and MS3 chromatograms for an extracted serum fail (Table 1). In contrast, the same sample analysed in MS3 mode sample, spiked with 10pM 3-T1AM. The MRM traces for m/z (Fig. 4, panel B), shows a clearly discernable peak, free of in- 356.1 / 338.8 and 356.1 / 212.0 certainly indicate the presence of terferences, confirming the presence of 3-T1AM.

Year/1st Model/ Sample Pre-analytics MS system Endogenously Landmark Sample Selectivity LOD Accuracy/Precision/ Linear Cal Matrix effect Stability Ref. author speciation amount determined Equilibration determination Recovery study study concentrations

Moller€ et al. Serum/human 1 ml Acidify, PPT, LLE, GC-MS/MS T 4: 82.4e108 nM First ID- Y1 SIM None reported Accuracy: LQAL/HQAL Y: in water None None (Moller (1983) N,O- GCMSMS spikes:98.3% reported reported et al., 1983) bis(trifluoroacetyl) method with Precision:1.7% methyl ester good derivative. correlation vs RIA Siekmann Serum/human 0.5e2 ml Acidify to pH2, DIT GC-MS/MS T4: 65.4 First use of DIT Y1SIM None reported Accuracy: vs RIA or EIA Y: in None None (Siekmann, e (1987) as stabiliser, cation 187.6 nM stabiliser only solvent reported reported 1987) e exchange resin Precision:1.4 1.9% clean-up, N,O- bis(trifluoroacetyl) methyl ester derivative. LH-20 column clean-up. De Brabandere Serum/human N.D. Add NaCl, acidify LC-MS/MS T4: 47.2 First LC-MS/MS Y1SRM Y (on stds only Accuracy: vs GC-MS/MS Y: in None None (De e et al. (1998) (reconstituted (~1 ml) and PPT, then LLE. 226.1 nM method for T 4 per FIA) only/ solvent reported reported Brabandere e from (control sera) Precision:0.9 2.1% et al., 1998) lyophilized) Thienpont Serum/human Upto 3 ml PPT þ LLE. Prep GC-MS 3T: 0.98e3.8 nM First LC-MS/MS Y1 SRM Y (on stds Accuracy: vs GC-MS/MS Y: in buffer None Y(T 4- > T3)(Thienpont et al. (1999) reconstituted HPLC clean-up LC-MS/MS method for T 3 only) Precision’: 0.9e1.2% reported et al., 1999) from (LCMSMS) lyophilized Recovery from buffer solutions (99.7e101.2%) “ & Tai et al. Serum/human 3 ml DIT as stabiliser, LC/MS4: 63.6 T Serum Y (study) SIM None reported Accuracy: vs routine Y: in Y: in BSA None (Tai et al., (2002) acidify pH2, PPT and e64.2 nM equilibration clinical methods” solvent HSA reported 2002) LLE. SPE clean-up. study. SPE College of American solutions clean-up pathologists (2000 only labs) < 10% bias. Precision: 0.2e0.6% Extraction recovery: 74% 125 (with I labelled T 4) Spike recovery: 99.5 e100.6% Tai et al. Serum/human 3 ml Addition of DIT as LC-MS/MS3: 1.6 T nM Measurement of Y (study) 1 MRM None reported “Absolute recovery” (vs Y: in None Y(T 4- > T3)(Tai et al., (2004) stabiliser, ascorbic SRM 1951b a matrix spike): 65% solvent reported 2004) acid, citric acid and CRM to be made Spike recovery: 98.9 dithiothreitol as available to labs. e99.4% anti-oxidants. Add Precision 0.8e1.6% NaCl, PPT and SPE (inter batch) clean-up. 1.9e2.6% (intra batch) m Van Uytfanghe Serum/human 60-200l Dilution to pH 10 LC-MS/MS T 4:66 Serum applied Y1SRM Y Accuracy: vs GC/MS Y (no detail Y: extracts None (Van e þ et al. (2004) and direct SPE 250 nmol/l directly to SPE ( 2.7% to 0.7% given) vs reported Uytfanghe clean-up. (no pre- difference) and vs 4 calibration et al., 2004) extraction) Laboratories in standards: European project no evidence (reported elsewhere). of ion Precision: 1.7e1.8% suppression Recovery: ~80% found. Scanlan et al. Serum/mouse Not Not reported LC-MS/MS T1AM: no First report of None 2 MRM (Scanlan fi (2004) reported quanti cation endogenous 3- reported (one et al., 2004) T1AM in serum reported) Soldin et al. Serum/human 600ml Ultrafiltration at LC-MS/MS fT 4: 0.011 First free-T 4 by None 1 SRM Y Accuracy: Correlation Y: in None None (Soldin et al., (2005) 25 C, online SPE e0.026 nM ultrafiltration/ reported 0.97 vs equilibrium solvent reported reported 2005) with column LC/MS/MS dialysis/IA; 0.33 vs switching direct IA 032 laect hsatcei rs s ihrs . ta. viigteptal hnqatfigtyodhroe n hi eaoie sn mass using metabolites their and hormones http://dx.doi.org/10.1016/j.mce.2017.01. thyroid (2017), quantifying Endocrinology when Cellular pitfalls and Molecular the Avoiding assurance, quality al., of et role K., The Richards, methods: as: spectrometric press in article this cite Please Precision: inter-batch 6.7e7.1%, intra-batch 4.1e6.6% Recovery: not reported Van Uytfanghe Serum/human 1 ml Dialysed at pH 7.4. LC-MS/MS4: fT 0.0177 nM First Y1 SRM Y Accuracy: 98.0e100.0% Y: in Y. No None (Van et al. (2006) Dialysate made Equilibrium for certified reference solvent indication of reported Uytfanghe alkaline, then SPE dialysis-LC-MS/ standards. Comparison ion et al., 2006) clean-up MS method. vs IA: concentrations suppression varied by 3to33%. Precision: 2.8% (intra- batch), 2.3% (intra- batch) Recovery: absolute from before and after extraction spikes: 85% ml Direct SPE LC-MS/MS T AM: no First report of None 2 MRM (Braulke Braulke et al. Plasma/ 200 1 quantification pre-analytics for reported (one et al., 2008) (2008) hamster endogenous 3- reported) T 1AM in plasma DeBarber et al. Serum/rat, 200ml PPT at pH4 and SPE LC-MS/MS No 3-T 1AM: None 2 MRM Y Accuracy: not reported Y: in rat Y: Stocks (DeBarber (2008) mouse and clean-up. quantification Validated reported (one Precision: Inter-batch serum and extracts, et al., 2008) human Detection of method reported) 8.4e20.3%; intra-batch storage and endogenous 2.7e13.1% freeze/thaw T 1AM at limit of Recovery: 58e69% detection in all (process efficiency) species Multi-method YT Saba et al. Serum/rat and 1 ml Add NaCl. PPT, LC-MS/MS Rat: 1AM: 2 Y Accuracy: not reported None (Saba et al., ¼ for (2010) human acidify to pH4, then T1AM 0.30; MRM (one Precision: not reported reported 2010) ¼ thyronamines SPE clean-up T0AM 0.04; reported) Recovery: 90% T3 ¼ 1.4; and thyronines fi T4 ¼ 49.6 nM and rst to Human: quantify T 1AM. T1AM ¼ 0.15 e0.20 nM m ± Multi-method None 2 MRM Y Accuracy: low med high Y: in Y: none None (Wang and Wang and Serum/bovine 500l Addition of ascorbic LC-MS/MS T 3 5.7 0.52 nM, including T reported QQAL, reported as % solvent observed on reported Stapleton, Stapleton acid, citric acid and T4 2 107.6 ± 7.0 nM isomers and rT recovery. post extract 2010) (2010) dithiothreitol as 3 anti-oxidants. PPT, 3,5-T 2 (with HPLC Comparison vs RIA and matrix 0 then SPE clean-up. (<0.74 nM), 3,3 - separation of for T 3 vs Tai et al. (2004) spiked T2 (<0.92 nM), isomers). with SRM1951b samples and rT3 Precision: intra-batch (<1.4 nM) were 1.2e9.6% lower than the Recovery (spike): 81.3 method e111.9% detection limits m N/A 2 MRM (2 Y Accuracy: < 6% bias Y: in DMSO/ None None (Ackermans Ackermans Plasma/rat and 100l Proteinase K LC-MS/MS No detection of Protease reported) across linear range 0.1% reported reported et al., 2010) et al. (2010) human digestion, then on- endogenous digestion of line SPE. T 1AM or T 0AM serum proteins. Precision intra-batch HCOOH On-line SPE T 1AM 1.4%, T 0AM 6.1% method for Recovery (spike): T 1AM T 1AM and T 0AM 96%, T0AM 103% Jonklaas et al. Serum/human Method as in Soldin LC-MS/MS 3,32‘:-T 0.013 Endogenous (Jonklaas 2014 et al. (2005) e0.040 nM 3,3‘-T 2 in et al., 2014) hospital patients e fi Hansen et al. Serum/frog 50ml Urea treatment plus LC-MS/MS T 4 7.68, T 3 1.15, Multi-analyte Y2MRM Y Accuracy: 88 103% Y: in matrix Y: quanti ed Y: TH pH (Hansen e 2016 and tadpole SPE clean-up. rT3 0.36, 3,3’-T 2 method. with 20% (HQAL) 102 159% as 7% and et al., 2016) 0.64, 3,5-T 2 0.66, Endogenous ion ratio (LQAL) to 30% photolytic tolerance stability, (continued on next page) 032 4.2. Accuracy, precision and recovery Ref. Most published MS methods are validated for precision and accuracy and some for recovery. Earlier publications tested accuracy by comparison of control sera measured with existing IA freeze thaw and sample extract storage Stability study (Siekmann, 1987; Soukhova et al., 2004) or GC/MS (De Brabandere et al., 1998; Thienpont et al., 1999; Van Uytfanghe et al., 2004) methods. Most laboratories currently access accuracy by spiking known amounts of hormone into the biological matrix and

study measuring the spiked amount to measured amount ratio. Precision is typically determined by analysing intra- and inter-day replicate samples. There have been few reports in the TH-related scientific literature regarding participation in intra-laboratory (proficiency) Linear Cal Matrix effect testing, with the exception of Soldin's group (Gu et al., 2007) for the quantification free T3 and T4 in serum, so precision is generally 24.7%

e quoted as intra-lab (termed repeatability) rather than inter-lab (termed reproducibility). 17.2%, HQAL e In the context of developing reference measurement procedures to meet regulatory requirements, some laboratories have analysed

80% for majority of TH the NIST (National Institute of Standards and Technology) human Precision: MQAL: inter- batch 1.31 inter-batch 5.35 Recovery (spike): 70 e and THM, and surrogate recovery vs neat standards. Accuracy/Precision/ Recovery serum standard reference material SRM 1951b for total T3 (Tai et al., 2004) and T4 (Wang and Stapleton, 2010). More recently NIST have announced their intention to support clinical and research laboratories by providing Certiﬁcate of Analysis reference values for

determination total and free T4 and T3 in SRM 971; reference values will be determined using two independent analytical techniques (including LC-MS/MS) (Long et al., 2016). The same organisation

AM (1 plans to investigate the potential for a 4-part T /T serum standard

1 3 4 except for T SRM) Selectivity LOD reference material obtained from (1st to 3rd trimester) pregnant and non-pregnant women. The availability of standard patient and reference materials, and their potential application in proﬁciency testing, would represent an important step towards stand-

Equilibration ardisation in the assessment of analytical method accuracy and 2 precision. No such samples are yet available for sera of newborns, children and adolescents, where growth-related (percentile) TH reference parameters are used and mainly total but not free T4 and

AM and 3,5-T T concentrations are determined (Van Uytfanghe et al., 2014).

1 3 reported T Landmark Sample The amount of substance recovered from the sample matrix ﬁ

0 compared with the known amount spiked (or quoted for a certi ed reference material) in percentage terms is calculated for the majority of published methods and often quoted as “recovery”. This

AM 3.11, T value is a measure of the trueness or accuracy of the method, and 1 0.74 nM determined concentrations T may be quoted in terms of percentage bias (Ackermans et al., 2010; Rathmann et al., 2015). It should be pointed out that, in common with most international regulatory organisations, the FDA guidelines define recovery as the instrument response for a known amount of analyte in the sample (for instance spiked into the biological matrix) compared with that for the same amount of analyte in pure solvent. The recovery, thus defined, is a measure of the extraction efficiency of the method combined with any matrix effects arising during sample analysis. This way of defining recovery not only tests the percentage accuracy/bias of a method, but also Pre-analytics MS system Endogenously the efficiency of the sample extraction/clean-up protocol. For an LC- MS/MS method the recovery need not be 100%, but it should be reproducible. The authors found that whilst only one publication in Sample amount the thyroid research field actually used the FDA definition of “recovery” (DeBarber et al., 2008), others have adopted the definition under an alternative descriptor (e.g. “Process Efficiency”, as used by Piehl et al. (2008). and Rathmann et al. (2015).). ) Model/ speciation 4.3. Instrument calibration continued ( The response of the LC-MS/MS is calibrated against known concentrations of target analytes. The linear dynamic range (the Year/1st author

Table 2 range within which the instrument response is proportional to the

3 2 Fig. 3. MS (m/z 360.1/ m/z 342.8/ m/z 50e345) spectrum for H4-3T1AM. analyte concentration in a linear relationship) should at least cover extraction losses, variation in HPLC injectors and effects on analyte the expected range of analyte concentrations to be quantified. ionisation by the sample matrix is essential for good accuracy with Linearity is most commonly qualified in terms of Pearson coeffi- LC-MS/MS methods and was adopted early on as ID-GC/MS cient (r2) using an adequate number of calibrators across range of methods were converted to LC-MS/MS (Thienpont et al., 1999). concentrations. The FDA recommendations specify that calibration Unfortunately, whereas the use of stable isotope ISTDs GC/MS ap- standards should be prepared in the same sample matrix as the real pears to entirely balance out the influence of the matrix, this is not samples to be quantified. The use of pure solvents to prepare cali- always the case with LC/MS (Vogeser and Seger, 2010). Isotopically bration standards has been common practice with both GC- and LC- labelled ISTDs are also not commercially available for all THM; MS (methods (Siekmann, 1987; Thienpont et al., 1994; Tai et al., inevitably, some multi-methods rely on the assignment of a single 2002, 2004; Thienpont et al., 1999; Wang and Stapleton, 2010; ISTD to multiple analytes, only one of which co-elutes with it; Ackermans et al., 2010; Kunisue et al., 2011b)). An equilibrium differential matrix effects may then result. dialysis sample preparation method coupled with on-line SPE It therefore becomes essential to reduce the effect of the matrix clean-up (Yue et al., 2008) used solvent/buffer based calibrators; and to document its effect during validation. The early LC-MS/MS matrix influences were investigated and found to be absent. methods are notable for their scant attention to this issue; publi- However, differential matrix effects between samples and calibra- cations that are more recent describe the assessment of matrix tors are a potential source of error (Vogeser and Seger, 2010), and effects (Van Uytfanghe et al., 2004; Wang and Stapleton, 2010; wherever possible, matrix-matched calibrations are preferable. The Piehl et al., 2008; Rathmann et al., 2015). use of matrix-matched calibrators using either protein-containing One approach to quantifying matrix effects in place in our lab- solutions (De Brabandere et al., 1998; Soukhova et al., 2004), oratory is the comparison of pure solvent spiked versus post- charcoal stripped serum (Nagao et al., 2011), or spiked serum/ extraction spiked samples. Thus, the following formula is used to plasma (Saba et al., 2010; Kiebooms et al., 2014; Hansen et al., 2016) calculate the matrix effect in %. is thus becoming commonplace. post extraction spiked MRM peak area %Matrix effect ¼ *100 spiked solvent MRM peak area 4.4. Matrix effects and the use of ISTD An illustrative example of a spiked serum sample demonstrating The suppression or enhancement of analyte ionisation caused a recovery of circa 34% and a matrix effect of circa 51% is shown in by co-eluting interferences from biological samples and its impact Fig. 5. on measurement accuracy in the clinical laboratory has been In this laboratory matrix effects of 75% are deemed as reviewed (Vogeser and Seger, 2010). For sensitive, accurate and acceptable. precise measurements, it is imperative to minimise and to document matrix effects when validating and implementing LC-MS/MS 4.5. Sensitivity methods for the quantification of THM in biological samples. Measures that can minimise matrix effects include. Analyte sensitivity is quoted in terms of LLOD and lower limit of quantitation (LLOQ). In the TH analysis community these detection The use of isotopically-labelled ISTDs limits are quoted in simple terms of signal:noise of MRM (SRM) The use of added clean-up steps to reduce the amount of bio- chromatograms; LLOD or LLOQ are generally accepted as equivalent logical matrix material present in sample extracts. to the smallest concentrations measurable giving a signal:noise of Preparation of matrix-matched calibration samples 3:1 or 10:1 respectively. Comparing values between publications can be challenging, given the plethora of units in use (ppb, pg/mL, The use of isotopically labelled ISTDs to compensate for sample ng/mL, pM/mL, mg/dL, mg/L, ng/dL nM for serum, ng/g, pmol/g, ng/

Fig. 4. Chromatograms for a serum sample extract spiked with 10pM 3-T1AM.

(Tai et al., 2004) implemented the use of antioxidants (acorbic acid, report an equilibration step (De Brabandere et al., 1998; Saba et al., dithiothreitol, and citric acid) to suppress potential free radical 2010; Hansen et al., 2016) allow 1hr for it to complete. oxidation by lipid hydroperoxides and/or metal induced oxidation in processed sera. This T4/T3 oxidative de-iodination was quan- 5. Immunoassays 2 2 tified by monitoring the conversion of H5-T4 to H5-T3 in spiked serum and was calculated as being up to 3% for certain serum In the decades before the introduction of the LC-MS/MS meth- samples. The use of acid conditions did not seem to be problematic, odology, the thyroid field depended on the use of IA for the since one SPE step used 0.1 M HCl as a wash solution. A recent quantification of TH (Wiersinga and Chopra, 1982; Chopra et al., publication (Hansen et al., 2016) undertook a degradation study of 1972; Brown et al., 1970; Ekins, 1998). In contrast to LC-MS/MS, T4 at 50 C under neutral, basic and acidic conditions, in the pres- early IA did not have multi-analyte capability. The ELISA-based ence of the aforementioned antioxidants; significant TH lability at (enzyme-linked immunosorbent assay) methods in principle use pH4, measurable within 1 h for T4,T3 and other THM was reported, one antibody to immobilize the antigen and a further antibody as a whereas good stability was observed at pH11. This information detection antibody. The latter antibody is labelled with an enzyme would seem to directly contradict that from Thienpont (Thienpont or a fluorophore generating a signal in presence of the correct co- et al., 1999). The authors conclude that more studies are needed to factors and substrates. For many radio-immunoassays a single determine the pH stability of THM. antibody, fulfills the antigen immobilizing role. Quantification is Stability at both 4 C and 20 C has been reported for free T3 possible since added radiolabelled tracer competes with the and T4 in serum (Yue et al., 2008) and freeze/thaw stability for 3- endogenous compound of interest for binding at the antigen T1AM in serum at 80 C storage (DeBarber et al., 2008). recognition site. Pivotal for these IA is the use of a high-affinity antibody with high specificity as primary antigen detection 4.7. Equilibration component, minimizing cross-reactivity to THM (Wiersinga and Chopra, 1982). After initial optimisation for serum, some methods When developing sample preparation procedures for the were then adapted for detection of TH in other (biological) determination of TH/TAM in biological material, then the known or matrices. Up to the introduction of multiplex IA, assays would focus postulated binding of hormones to carrier proteins should be taken on only one analyte of choice. Consequently, a wider interest in the into account. With ID methods, the sample is spiked with an array of TH/THM would have required multiple assays. Nowadays, isotopically labelled analogue of the hormone, differing only in various commercially available antibody based assays are available molecular mass from the native compound. A ligand exchange will for specific hardware and liquid handling platforms, thus allowing take place between protein bound endogenous hormone and ISTD. automated high throughput analysis. It is therefore important to allow this exchange to reach equili- Although free T3 and free T4 concentrations are routinely bration before attempting to extract the hormone; otherwise, measured, they may be less reliable especially in the diseased pa- quantification will likely suffer from poor reproducibility. Tai et al. tient or during pregnancy (Thienpont et al., 2013; Welsh and Soldin, studied the time required for the equilibration of isotopically 2016). Over the past decade, the IFCC Committee for Stand- labelled T4 in human serum, which was found to be complete ardisation of Thyroid Function Tests has done valuable work in the within 30 min (Tai et al., 2002). Based on this finding, those that standardisation and harmonisation and free and total TH assays (Van Houcke et al., 2011; Thienpont et al., 2010, 2014). Since some of the commonly employed IA methods still deviate in measured T3 and T4 concentrations (Yue et al., 2008; Welsh and Soldin, 2016; Steele et al., 2005; Baloch et al., 2003; Faix and Miller, 2016; van Deventer and Soldin, 2013), there is a call for the introduction of a “gold standard” reference method, such that LC-MS/MS, with its limitations, can provide (Thienpont et al., 2013). For many years the scientific and medical focus, however, has been on only a limited number of TH/THM. Over time, the stocks for 0 specific antibodies for other THM such as 3,3 -T2 and 3,5-T2 were either lost or not replenished (Maciel et al., 1979; Wu et al., 1976; Meinhold and Schurnbrand, 1978). Recently, two monoclonal antibody-based chemiluminescence immunoassays (CLIA) were developed for the detection of 3,5-T2 (Lehmphul et al., 2014) and the low molecular weight THM 3-T1AM (Hoefig et al., 2011) respectively. As in the case of LC-MS/MS, quality assurance of these new IA methods is key. The authors clearly described the potential cross-reactivity of a wide set of THM, the accuracy and precision of the respective CLIA and the analyte stability. When adopting the guidelines for the determination of classical TH (B.C. Committee, 2001; Van Houcke et al., 2011; Baloch et al., 2003; Faix and Miller, 2016), there are some aspects that these assays would benefit from:

a) Certified reference material, patient and reference subject material, to aid the method validation and harmonisation of the findings. There are no reference materials available for either 3- Fig. 5. Matrix effect and recovery for 2H -3T AM in a spiked blood serum sample. 4 1 T AM or 3,5-T . Points in duplicate denote the percentage difference in MRM chromatogram peak 1 2 areas from serum sample extracts spiked before (sample numbers 1e2) and after b) A non-interfering calibrator matrix for the assay. The calibrator (sample 3e4) extraction, relative to a spiked solvent sample (samples 5e6). matrix used for the 3-T1AM assay is T3/T4 deficient serum or

BSA-PBST (Hoefig et al., 2011), whereas SeraCon II CD serum is Biomed. Chromatogr. 26, 485e490. used for the 3,5-T assay (Lehmphul et al., 2014). It is pivotal that Aqai, P., Fryganas, C., Mizuguchi, M., Haasnoot, W., Nielen, M.W., 2012. Triple bio- 2 affinity mass spectrometry concept for thyroid transporter ligands. Anal. Chem. the calibrator matrix respond equally to the human serum 84, 6488e6493. regarding the antigen accessibility by the antibody or the Baloch, Z., Carayon, P., Conte-Devolx, B., Demers, L.M., Feldt-Rasmussen, U., interference of the calibrator matrix with the detector response. Henry, J.F., LiVosli, V.A., Niccoli-Sire, P., John, R., Ruf, J., Smyth, P.P., Spencer, C.A., fi Stockigt, J.R., N.A.o.C.B. Guidelines Committee, 2003. Laboratory medicine Thus far, no major binding protein has been identi ed for 3,5-T2. practice guidelines. Laboratory support for the diagnosis and monitoring of Apolipoprotein B100 is identified as the major carrier of 3-T1AM thyroid disease. Thyroid 13, 3e126. in human serum, non-covalently binding over 90% of the THM B.C. Committee, 2001. Guidance for industry: bioanalytical method validation. In: CVM, C. (Ed.), FDA, pp. 1e25. http://www.fda.gov/downloads/Drugs/Guidances/ (Roy et al., 2012). ucm070107.pdf. c) Transferability of the method. The transferability of both Bondoux, G., Joumier, J.-M., 2012. Liquid chromatography/mass spectrometry: the methods has not been tested, since neither are commercially method of choice for the qualitative and quantitative analysis of environmental available. pollutants. In: Lebedev, A. (Ed.), Comprehensive Environmental Mass Spec- trometry. ILM Publications, pp. 53e54. Braulke, L.J., Klingenspor, M., DeBarber, A., Tobias, S.C., Grandy, D.K., Scanlan, T.S., Heldmaier, G., 2008. 3-Iodothyronamine: a novel hormone controlling the balance between glucose and lipid utilisation. J. Comp. Physiol. B 178, 167e177. 6. Summary Brown, B.L., Ekins, R.P., Ellis, S.M., Reith, W.S., 1970. A radioimmunoassay for serum tri-iodothyronine. J. Endocrinol. 46 (i). Tandem MS has proved to be an extremely valuable bio- Burman, K.D., Bongiovanni, R., Garis, R.K., Wartofsky, L., Boehm, T.M., 1981. Mea- surement of serum T4 concentration by high performance liquid chromatog- analytical tool for THM, and its combination with liquid chroma- raphy. J. Clin. Endocrinol. Metab. 53, 909e912. tography has become the method of choice for many laboratories. Cheng, S.Y., Leonard, J.L., Davis, P.J., 2010. Molecular aspects of thyroid hormone Although the validation of published LC-MS/MS methods has actions. Endocr. Rev. 31, 139e170. become a routine requirement, there is no generally accepted Chopra, I.J., Ho, R.S., Lam, R., 1972. An improved radioimmunoassay of triiodothyronine in serum: its application to clinical and physiological studies. J. Lab. Clin. “golden rule” as to how a validation procedure should be defined, Med. 80, 729e739. and there are clearly areas of QA that can be improved. In particular, De Brabandere, V.I., Hou, P., Stockl, D., Thienpont, L.M., De Leenheer, A.P., 1998. the use of MRM quantitation and confirmation ions and their Isotope dilution-liquid chromatography/electrospray ionization-tandem mass spectrometry for the determination of serum thyroxine as a potential reference relative ion ratio is not stringently applied; this constitutes a real method. Rapid Commun. mass Spectrom. RCM 12, 1099e1103. risk of mis-identification for THM that occur in the low pM con- DeBarber, A.E., Geraci, T., Colasurdo, V.P., Hackenmueller, S.A., Scanlan, T.S., 2008. centration range in biological samples. Furthermore, the quantifi- Validation of a liquid chromatography-tandem mass spectrometry method to enable quantification of 3-iodothyronamine from serum. J. Chromatogr. A 1210, cation and documentation of matrix effects is not universally 55e59. evident; this is essential, especially considering the fact that stable Ekins, R.P., 1998. Ligand assays: from electrophoresis to miniaturized microarrays. isotopically labelled ISTD are not commercially available for all Clin. Chem. 44, 2015e2030. Faix, J.D., Miller, W.G., 2016. Progress in standardizing and harmonizing thyroid THM. Intra-laboratory accuracy and precision are generally well- function tests. Am. J. Clin. Nutr. 104 (Suppl. 3), 913se917s. addressed, inter-laboratory reproducibility however not so. The Fan, W., Mao, X., He, M., Chen, B., Hu, B., 2013. Stir bar sorptive extraction combined move towards matrix-matched calibration is a welcome improve- with high performance liquid chromatography-ultraviolet/inductively coupled plasma mass spectrometry for analysis of thyroxine in urine samples. ment, as are attempts to assess the stability of THM during the J. Chromatogr. A 1318, 49e57. extraction process (although more work is clearly required here). Galli, E., Marchini, M., Saba, A., Berti, S., Tonacchera, M., Vitti, P., Scanlan, T.S., Sensitivity determination is standard procedure, although the Iervasi, G., Zucchi, R., 2012. Detection of 3-iodothyronamine in human patients: e plethora of quoted units makes direct comparisons between a preliminary study. J. Clin. Endocrinol. Metab. 97, E69 E74. Gu, J., Soldin, O.P., Soldin, S.J., 2007. Simultaneous quantification of free triiodo- methods an arduous task. Finally, most laboratories measure “re- thyronine and free thyroxine by isotope dilution tandem mass spectrometry. covery” by quantifying known concentrations of analyte spiked into Clin. Biochem. 40, 1386e1391. fi fi sample matrix; adopting the FDA definition of recovery, which Hackenmueller, S.A., Scanlan, T.S., 2012. Identi cation and quanti cation of 3- iodothyronamine metabolites in mouse serum using liquid chromatography- measures the real efficiency of the extraction process, would be a tandem mass spectrometry. J. Chromatogr. A 1256, 89e97. welcome advance. Hager, J.W., 2002. A new linear ion trap mass spectrometer. Rapid Commun. Mass Sp. 16, 512e526. Hansen, M., Luong, X., Sedlak, D.L., Helbing, C.C., Hayes, T., 2016. Quantification of 11 Funding thyroid hormones and associated metabolites in blood using isotope-dilution liquid chromatography tandem mass spectrometry. Anal. Bioanal. Chem. 408, 5429e5442. This work is supported by grants from the Deutsche For- Hearn, M.T.W., Hancock, W.S., 1979. High-pressure liquid-chromatography of schungsgemeinschaft within the DFG-SPP 1629 ThyroidTransAct amino-acids, peptides and proteins .14. High-pressure liquid-chromatography KO 922/17-1/2 to JK. of thyromimetic iodoamino acids. J. Liq. Chromatogr. 2, 217e237. Higashi, T., Ichikawa, T., Shimizu, C., Nagai, S., Inagaki, S., Min, J.Z., Chiba, H., Ikegawa, S., Toyo'oka, T., 2011. Stable isotope-dilution liquid chromatography/ Conflict of interest statement tandem mass spectrometry method for determination of thyroxine in saliva. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 879, 1013e1017. Hoefig, C.S., Zucchi, R., Kohrle,€ J., 2016. Thyronamines and derivatives: physiological The authors declare that there are no competing interests. relevance, pharmacological actions, and future research directions. Thyroid 26, 1656e1673. fi € Acknowledgements Hoe g, C.S., Kohrle, J., Brabant, G., Dixit, K., Yap, B., Strasburger, C.J., Wu, Z., 2011. Evidence for extrathyroidal formation of 3-iodothyronamine in humans as provided by a novel monoclonal antibody-based chemiluminescent serum We thank Mr. Ray Monk for expert technical assistance. immunoassay. J. Clin. Endocrinol. Metab. 96, 1864e1872. Jonklaas, J., Sathasivam, A., Wang, H., Finigan, D., Soldin, O.P., Burman, K.D., Soldin, S.J., 2014. 3,3'-Diiodothyronine concentrations in hospitalized or References thyroidectomized patients: results from a pilot study. Endocr. Pract. 20, 797e807. Ackermans, M.T., Klieverik, L.P., Ringeling, P., Endert, E., Kalsbeek, A., Fliers, E., 2010. Kahric-Janicic, N., Soldin, S.J., Soldin, O.P., West, T., Gu, J., Jonklaas, J., 2007. Tandem An online solid-phase extraction-liquid chromatography-tandem mass spec- mass spectrometry improves the accuracy of free thyroxine measurements trometry method to study the presence of thyronamines in plasma and tissue during pregnancy. Thyroid 17, 303e311. and their putative conversion from 13C6-thyroxine. J. Endocrinol. 206, Kiebooms, J.A., Wauters, J., Vanden Bussche, J., Vanhaecke, L., 2014. Validated ultra 327e334. high performance liquid chromatography-tandem mass spectrometry method Ackermans, M.T., Kettelarij-Haas, Y., Boelen, A., Endert, E., 2012. Determination of for quantitative analysis of total and free thyroid hormones in bovine serum. thyroid hormones and their metabolites in tissue using SPE UPLC-tandem MS. J. Chromatogr. A 1345, 164e173.

Kunisue, T., Fisher, J.W., Fatuyi, B., Kannan, K., 2010. A method for the analysis of six Spencer, C.A., 2000. In: De Groot, L.J., Chrousos, G., Dungan, K., Feingold, K.R., thyroid hormones in thyroid gland by liquid chromatography-tandem mass Grossman, A., Hershman, J.M., Koch, C., Korbonits, M., McLachlan, R., New, M., spectrometry. J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 878, 1725e1730. Purnell, J., Rebar, R., Singer, F., Vinik, A. (Eds.), Assay of Thyroid Hormones and Kunisue, T., Eguchi, A., Iwata, H., Tanabe, S., Kannan, K., 2011. Analysis of thyroid Related Substances. Endotext, South Dartmouth (MA). hormones in serum of Baikal seals and humans by liquid chromatography- Steele, B.W., Wang, E., Klee, G.G., Thienpont, L.M., Soldin, S.J., Sokoll, L.J., tandem mass spectrometry (LC-MS/MS) and immunoassay methods: applica- Winter, W.E., Fuhrman, S.A., Elin, R.J., 2005. Analytic bias of thyroid function tion of the LC-MS/MS method to wildlife tissues. Environ. Sci. Technol. 45, tests: analysis of a College of American Pathologists fresh frozen serum pool by 10140e10147. 3900 clinical laboratories. Arch. Pathol. Lab. Med. 129, 310e317. Kunisue, T., Fisher, J.W., Kannan, K., 2011. Modulation of thyroid hormone con- Tai, S.S., Sniegoski, L.T., Welch, M.J., 2002. Candidate reference method for total centrations in serum of rats coadministered with perchlorate and iodide- thyroxine in human serum: use of isotope-dilution liquid chromatography- deficient diet. Arch. Environ. Contam. Toxicol. 61, 151e158. mass spectrometry with electrospray ionization. Clin. Chem. 48, 637e642. Kunisue, T., Fisher, J.W., Kannan, K., 2011. Determination of six thyroid hormones in Tai, S.S., Bunk, D.M., White, E.t., Welch, M.J., 2004. Development and evaluation of a the brain and thyroid gland using isotope-dilution liquid chromatography/ reference measurement procedure for the determination of total 3,3',5-triio- tandem mass spectrometry. Anal. Chem. 83, 417e424. dothyronine in human serum using isotope-dilution liquid chromatography- Lehmann, W.D., 2017. A timeline of stable isotopes and mass spectrometry in the tandem mass spectrometry. Anal. Chem. 76, 5092e5096. life sciences. Mass Spectrom. Rev. 36, 58e85. Thienpont, L.M., De Brabandere, V.I., Stockl, D., De Leenheer, A.P., 1994. Develop- Lehmphul, I., Brabant, G., Wallaschofski, H., Ruchala, M., Strasburger, C.J., Kohrle,€ J., ment of a new method for the determination of thyroxine in serum based on Wu, Z., 2014. Detection of 3,5-diiodothyronine in sera of patients with altered isotope dilution gas chromatography mass spectrometry. Biol. Mass Spectrom. thyroid status using a new monoclonal antibody-based chemiluminescence 23, 475e482. immunoassay. Thyroid 24, 1350e1360. Thienpont, L.M., Fierens, C., De Leenheer, A.P., Przywara, L., 1999. Isotope dilution- Little, A.G., 2016. A review of the peripheral levels of regulation by thyroid hor- gas chromatography/mass spectrometry and liquid chromatography/electro- mone. J. Comp. Physiol. B 186, 677e688. spray ionization-tandem mass spectrometry for the determination of triiodo-L- Long, S.E., Catron, B.L., Boggs, A.S., Tai, S.S., Wise, S.A., 2016. Development of Stan- thyronine in serum. Rapid Commun. Mass Spectrom. RCM 13, 1924e1931. dard Reference Materials to support assessment of iodine status for nutritional Thienpont, L.M., Van Uytfanghe, K., Beastall, G., Faix, J.D., Ieiri, T., Miller, W.G., and public health purposes. Am. J. Clin. Nutr. 104 (Suppl. 3), 902Se906S. Nelson, J.C., Ronin, C., Ross, H.A., Thijssen, J.H., Toussaint, B., I.W.G.o.S.o.T.F. Tests, Maciel, R.M., Chopra, I.J., Ozawa, Y., Geola, F., Solomon, D.H., 1979. 2010. Report of the IFCC working group for standardization of thyroid function A radioimmunoassay for measurement of 3,5-diiodothyronine. J. Clin. Endo- tests; part 1: thyroid-stimulating hormone. Clin. Chem. 56, 902e911. crinol. Metab. 49, 399e405. Thienpont, L.M., Van Uytfanghe, K., Poppe, K., Velkeniers, B., 2013. Determination of Meinhold, H., Schurnbrand, P., 1978. A radioimmunoassay for 3,5-diiodothyronine. free thyroid hormones. Best. Pract. Res. Clin. Endocrinol. Metab. 27, 689e700. Clin. Endocrinol. (Oxf) 8, 493e497. Thienpont, L.M., Van Uytfanghe, K., Van Houcke, S., Das, B., Faix, J.D., MacKenzie, F., Moller, B., Falk, O., Bjorkhem, I., 1983. Isotope dilutionemass spectrometry of Quinn, F.A., Rottmann, M., Van den Bruel, A., I.C.f.S.o.T.F. Tests, 2014. A progress thyroxin proposed as a reference method. Clin. Chem. 29, 2106e2110. report of the IFCC committee for standardization of thyroid function tests. Eur. Morreale de Escobar, G., Llorente, P., Jolin, T., Delrey, F.E., 1963. Transient instability Thyroid. J. 3, 109e116. of thyroxine and its biochemical applications. Biochem. J. 88, 526e530. van Deventer, H.E., Soldin, S.J., 2013. The expanding role of tandem mass spec- Nagao, H., Imazu, T., Hayashi, H., Takahashi, K., Minato, K., 2011. Influence of thy- trometry in optimizing diagnosis and treatment of thyroid disease. Adv. Clin. roidectomy on thyroxine metabolism and turnover rate in rats. J. Endocrinol. Chem. 61, 127e152. 210, 117e123. C. International Federation of Clinical, T. Laboratory Medicine Working Group for Nguyen, L.T., Gu, J., Soldin, O.P., Soldin, S.J., 2011. Development and validation of an Standardization of Thyroid Function, Van Houcke, S.K., Van Uytfanghe, K., isotope dilution tandem mass spectrometry method for the simultaneous Shimizu, E., Tani, W., Umemoto, M., Thienpont, L.M., 2011. IFCC international quantification of 3-iodothyronamine, thyroxine, triiodothyronine, reverse T3 conventional reference procedure for the measurement of free thyroxine in and 3,3'-diiodo-L-thyronine in Human serum. Clin. Chem. 57, A82. serum: international federation of clinical chemistry and laboratory medicine Piehl, S., Heberer, T., Balizs, G., Scanlan, T.S., Kohrle,€ J., 2008. Development of a (IFCC) working group for standardization of thyroid function tests (WG- validated liquid chromatography/tandem mass spectrometry method for the STFT)(1). Clin. Chem. Lab. Med. 49, 1275e1281. distinction of thyronine and thyronamine constitutional isomers and for the Van Uytfanghe, K., Stockl, D., Thienpont, L.M., 2004. Development of a simplified identification of new deiodinase substrates. Rapid Commun. mass Spectrom. sample pretreatment procedure as part of an isotope dilution-liquid chroma- RCM 22, 3286e3296. tography/tandem mass spectrometry candidate reference measurement pro- Rathmann, D., Rijntjes, E., Lietzow, J., Kohrle,€ J., 2015. Quantitative analysis of thy- cedure for serum total thyroxine. Rapid Commun. mass Spectrom. RCM 18, roid hormone metabolites in cell culture samples using LC-MS/MS. Eur. Thyroid. 1539e1540. J. 4, 51e58. Van Uytfanghe, K., Stockl, D., Ross, H.A., Thienpont, L.M., 2006. Use of frozen sera for Refetoff, S., 2000. In: De Groot, L.J., Chrousos, G., Dungan, K., Feingold, K.R., FT4 standardization: investigation by equilibrium dialysis combined with Grossman, A., Hershman, J.M., Koch, C., Korbonits, M., McLachlan, R., New, M., isotope dilution-mass spectrometry and immunoassay. Clin. Chem. 52, Purnell, J., Rebar, R., Singer, F., Vinik, A. (Eds.), Abnormal Thyroid Hormone 1817e1821. Transport. Endotext, South Dartmouth (MA). Van Uytfanghe, K., De Grande, L.A., Thienpont, L.M., 2014. A “Step-Up” approach for Roy, G., Placzek, E., Scanlan, T.S., 2012. ApoB-100-containing lipoproteins are major harmonization, Clinica chimica acta. Int. J. Clin. Chem. 432, 62e67. carriers of 3-iodothyronamine in circulation. J. Biol. Chem. 287, 1790e1800. Vogeser, M., Seger, C., 2010. Pitfalls associated with the use of liquid Saba, A., Chiellini, G., Frascarelli, S., Marchini, M., Ghelardoni, S., Raffaelli, A., chromatography-tandem mass spectrometry in the clinical laboratory. Clin. Tonacchera, M., Vitti, P., Scanlan, T.S., Zucchi, R., 2010. Tissue distribution and Chem. 56, 1234e1244. cardiac metabolism of 3-iodothyronamine. Endocrinology 151, 5063e5073. Wang, D., Stapleton, H.M., 2010. Analysis of thyroid hormones in serum by liquid Scanlan, T.S., Suchland, K.L., Hart, M.E., Chiellini, G., Huang, Y., Kruzich, P.J., chromatography-tandem mass spectrometry. Anal. Bioanal. Chem. 397, Frascarelli, S., Crossley, D.A., Bunzow, J.R., Ronca-Testoni, S., Lin, E.T., Hatton, D., 1831e1839. Zucchi, R., Grandy, D.K., 2004. 3-Iodothyronamine is an endogenous and rapid- Welsh, K.J., Soldin, S.J., 2016. Diagnosis of endocrine disease: how reliable are free acting derivative of thyroid hormone. Nat. Med. 10, 638e642. thyroid and total T3 hormone assays? Eur. J. Endocrinol. 175, R255eR263. Siekmann, L., 1987. Measurement of thyroxine in human serum by isotope dilution Wiersinga, W.M., Chopra, I.J., 1982. Radioimmunoassay of thyroxine (T4), 3,5,3'- mass spectrometry. Definitive methods in clinical chemistry, V. Biomed. Envi- triiodothyronine (T3), 3,3',5'-triiodothyronine (reverse T3, rT3), and 3,3'-diio- ron. Mass Spectrom. 14, 683e688. dothyronine (T2). Methods Enzymol. 84, 272e303. Soldin, O.P., Soldin, S.J., 2011. Thyroid hormone testing by tandem mass spec- Wu, S.Y., Chopra, I.J., Nakamura, Y., Solomon, D.H., Bennett, L.R., 1976. trometry. Clin. Biochem. 44, 89e94. A radioimmunoassay for measurement of 3,3'-L-diiodothyronine (T2). J. Clin. Soldin, S.J., Soukhova, N., Janicic, N., Jonklaas, J., Soldin, O.P., 2005. The measurement Endocrinol. Metab. 43, 682e685. of free thyroxine by isotope dilution tandem mass spectrometry, Clinica chimica Yue, B., Rockwood, A.L., Sandrock, T., La'ulu, S.L., Kushnir, M.M., Meikle, A.W., 2008. acta. Int. J. Clin. Chem. 358, 113e118. Free thyroid hormones in serum by direct equilibrium dialysis and online solid- Soukhova, N., Soldin, O.P., Soldin, S.J., 2004. Isotope dilution tandem mass spec- phase extractioneliquid chromatography/tandem mass spectrometry. Clin. trometric method for T4/T3, Clinica chimica acta. Int. J. Clin. Chem. 343, Chem. 54, 642e651. 185e190.