| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Endocrinology and Metabolism |
1 ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, Utah; 2 Department of Pathology, University of Utah School of Medicine, Salt Lake City, Utah; 3 Department of Medicine, University of Utah, Salt Lake City, Utah.
aAddress correspondence to this author at: ARUP Institute for Clinical and Experimental Pathology, 500 Chipeta Way, Salt Lake City, UT 84108-1221. Fax 801-584-5207; e-mail bingfang.yue{at}aruplab.com.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Methods: ED was selected as the separation step. Serum samples were dialyzed 1:1 against a simple protein-free buffer for 20 h at 37 °C. Thyroid hormones in dialysates were purified by online solid-phase extraction (SPE), then chromatographically separated and quantified in positive ion and multiple reaction monitoring modes.
Results: For FT4 and FT3, the lower and upper limits of quantification were 1 ng/L (pg/mL) and 400 ng/L with total imprecision <10%. The method correlated well with an ED-RIA, 2 direct immunoassay methods for FT4, and 1 direct immunoassay and 1 tracer dialysis method for FT3. The adult reference intervals were 12.8–22.2 ng/L for FT4 and 3.62–6.75 ng/L for FT3. Reference intervals for the second trimester of pregnancy (14–20 weeks of gestation) were also established.
Conclusions: We developed a simple protein-free buffer and ED procedure. The performance characteristics and high throughput of the LC-MS/MS method with online SPE for FT4 and FT3 (also reverse T3) are sufficient for the intended clinical use.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Most clinical laboratories measure free T4 (FT4) and free T3 (FT3) by automated immunoassays (IAs)(7)(8)(9)(10). The analytical performance of IAs has been questioned(11). Significant biases have been reported due to either endogenous factors (e.g., abnormal binding proteins, dialyzable protein binding competitors, heterophile antibodies, autoantibodies) or in vitro factors (free fatty acids, assay antibodies, analogs, intrinsic dilution). Disagreement between results from different methods has been a perplexing problem, limiting their clinical utility(12)(13).
An ideal method to measure free thyroid hormones should employ a separation step to physically separate the free fraction from the protein-bound fraction before quantitative analysis. The separation step should not disturb the endogenous equilibrium(9)(14). Theoretically, both equilibrium dialysis (ED) and ultrafiltration (UF) can be valid methodologies. RIA of thyroid hormones in dialysate(15)(16)(17) or ultrafiltrate(18)(19) has been considered the reference methodology. Variability can occur in both separation and detection steps, however, resulting in significant differences among results by different UF- or ED-RIA methods(14). The major concerns with UF include nonequilibrium, absorption, protein leakage, and temperature control, and with ED, dilution and dialysis buffer composition(14). A procedure based on ED combined with determination of thyroxine in the dialysate with a trueness-based measurement procedure has been proposed for a candidate international conventional reference measurement(20)(21).
Isotope dilution mass spectrometry (ID-MS) could be a superb detection method for thyroid hormones with high specificity in comparison to RIA or IA. Methods based on gas chromatography (GC) and ID-MS with selected ion monitoring (SIM) have been developed to measure total T4 and T3, but they require laborious sample cleanup and derivatization(22)(23)(24)(25)(26). Several methods have been reported using liquid chromatography (LC) coupled to MS in SIM mode, or tandem mass spectrometry (MS/MS) in multiple reaction monitoring (MRM) mode, to measure total T4(27)(28)(29) and total T3(29)(31).
Soldin et al.(32) reported the first method for FT4 measurement using UF and LC-MS/MS. T4 in ultrafiltrates was trapped on a short C-18 column and eluted with a steep methanol gradient (up to 100%) using MRM detection in the negative ion mode with a sample cycle time of approximately 10 min. Major concerns include no pH adjustment, UF at 25 °C rather than 37 °C, high ultrafiltrate yield, 2H2-T4 as internal standard (IS), protein leakage, absorption, and potential interferences. Recently, van Uytfanghe et al.(33) reported the first method for FT4 using ED-LC-MS/MS as a candidate reference measurement procedure(34). The dialysate, with 13C6-T4 as IS and 13C9-T4 as carrier, was purified using solid-phase extraction (SPE). A 15-cm C-18 column provided 16-min LC separation before detection in positive MRM mode.
We report here a rapid and selective LC-MS/MS method with an analytically and clinically valid ED procedure for quantifying FT4 and FT3 in serum (for reverse T3, see the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue4).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
99.0%), NH4OH (29.7% as NH3), NaCl, Na3PO4, Na2HPO4, KH2PO4, KCl, MgSO4 · 7H2O, HEPES (free acid), urea, CaCl2, NaN3, and NaOH were purchased from Sigma-Aldrich, and were of the highest purity available unless otherwise specified. We purchased T3 (>99% by HPLC) and rT3 (>98% by TLC) from EMD Bioscience and 13C6-T4 (L-thyroxine, 99% tyrosine-ring-13C6 purity, 90%) from Cambridge Isotope Laboratories. The impurities (T4, T3, and rT3) in 13C6-T4 were found to be insignificant. We obtained acetonitrile (Optima), methanol (Optima), and formic acid (99%) from Fisher Scientific. We prepared stock solutions of T4, T3, rT3, and 13C6-T4 (1 g/L) in a mixture of methanol and 30% NH4OH (50/50, vol/vol) using a volumetric flask made of high-purity polypropylene. A combined 300 ng/L (pg/mL) working solution and 100 µg/L intermediate IS working solution were prepared in methanol with 100 mmol/L NH4OH. We prepared the IS working solution (20 ng/L 13C6-T4) in a mixture (20/80, vol/vol) of methanol and phosphate buffer (50 mmol/L Na2HPO4, 150 mmol/L NaCl, pH 10.9). Calibrators were freshly spiked with combined working solution at equivalent concentrations of 0, 5, 10, 20, 30, and 40 ng/L (pg/mL) for each analyte in PBS containing 3 mmol/L KH2PO4, 8 mmol/L Na2HPO4, and 150 mmol/L NaCl, pH 7.2, as blank matrix, in each run.
We obtained the 96-well dialysis plate from Harvard Apparatus. It is designed to use a dialysand to dialysate ratio of 1:1 and up to 200 µL in each compartment. Each well has its own regenerated cellulose membrane (5 kDa molecular weight cutoff). The procedure for preparing the dialysis buffer is described in the online Data Supplement. The Institutional Review Board of the University of Utah approved all studies with samples from human subjects.
sample preparation
Aliquots of 200 µL serum samples or quality controls were dialyzed against 200 µL dialysis buffer at 37 °C for approximately 20 h. Dialysate (150 µL) was mixed with 150 µL IS working solution in a 96-well polypropylene plate (Phenomenex).
online spe-lc-ms/ms
The instrument consisted of an API 5000 triple-quadrupole mass spectrometer from Applied Biosystems/MDS Sciex, 2 1100-series binary pumps and 1 column oven from Agilent, a HTC PAL autosampler from Leap Technologies, and a 6-port switching valve from Valco Instruments Co. The LC pumps were reconfigured to reduce gradient delay volume. The switching valve was connected like an injector with an extraction column in place of the sample loop, without backing flushing.
We used a guard column of wide-pore C5 (8 by 2 mm; Phenomenex) for online SPE at ambient temperature with 10 mmol/L formic acid in water (A) and methanol (B) as mobile phase, and a Synergy Polar-RP column (2 µm, 100 Å, 50 by 2 mm; Phenomenex) with 10 mmol/L formic acid in water (A) and acetonitrile (B) for LC separation at 45 °C. Mobile phase for online SPE was delivered at 2 mL/min with the following program: 15% B for 0.1 min, linear ramp to 95% B in 0.6 min and hold for 0.8 min, linear ramp to 15% B in 0.3 min and hold for 0.7 min. Mobile phase for LC separation was delivered at 0.75 mL/min with the following program: 30% B for 0.6 min, linear ramp to 42.5% B in 1.6 min, linear ramp to 95% B in 0.1 min and hold for 0.2 min. At 0.6 min after injection, the switching valve was switched for 0.3 min to elute the analytes from the extraction column to the LC column. The injection volume was 250 µL.
The MS instrument was operated in positive MRM mode. Two transitions were monitored for each compound: m/z 777.7–731.6 and 604.7 for T4; m/z 651.8–605.8 and 478.8 for T3 and rT3; and m/z 783.8–737.8 and 610.7 for IS. Chromatograms acquired from a patient serum sample are illustrated in Fig. 1
.
|
evaluation of method performance
All experiments described include the ED procedure, except when explicitly stated as "ED not included," and all replicates were separate preparations instead of multiple measurements from 1 preparation. To determine the lower and upper limit of quantification (LLOQ and ULOQ), we analyzed freshly spiked PBS (0.5, 1.0, 2.0, 100, 150, 200, 400 ng/L) in triplicate in each run, in 3 runs and 3 days, without the dialysis step. We evaluated imprecision by analyzing 7 serum pools in triplicate in each run, in 6 runs and 6 days. We also studied imprecision with spiked dialysis buffer (half HEPES concentration) at 3 concentrations, without the dialysis step, analyzed in triplicate in each run, in 5 runs and 5 days.
We evaluated ion suppression using a postcolumn infusion method(35). A standard solution (10 µg/L) was introduced into the flow path after the analytical column, resulting in an increased baseline for every transition. A set of 24 dialyzed patient samples (with low thyroid hormone concentrations) were injected. We examined the chromatograms for evidence of ion suppression. We searched 3 databases (CAS Chemical Registry, METLIN Metabolite Database, and Merck Index) for compounds potentially interfering with T4 and T3/rT3 and further investigated the potential interferences by examining chromatograms and quantitative results of patient samples acquired with 2 different transitions(36) for each compound (n > 1000).
For stability studies, aliquots of 2 serum pools were stored at ambient temperature, 4 °C, and –20 °C for 6 days and analyzed in duplicate in each run, in 6 runs and 6 days. The aliquots stored at –20 °C were frozen and thawed 6 times. Another 4 serum pools were processed in 8 replicates and stored at ambient temperature. For 4 consecutive days, 2 replicates of 8 were analyzed each day in 1 run to study the stability of processed samples. We evaluated carryover by injecting a negative control after spiked samples containing 400 ng/L T3, rT3, or T4. We studied the effect of the serum volume used for dialysis by dialyzing 100, 150, or 200 µL of 4 patient sera with 200 µL dialysis buffer in duplicate in each run, in 3 runs and 3 days.
method comparison
We used 3 different sets of serum samples because sample volumes were not sufficient for all comparison methods. For FT4 comparison, we used a set of 90 serum samples previously analyzed by an ED-RIA method (Nichols Institute Diagnostics) and stored at –70 °C for approximately 1 year. Another set of 96 serum samples was collected for comparison with the Vitros ECi FT4 immunoassay (Ortho Diagnostics) and Modular Analytics E170 FT4 and FT3 immunoassays (Roche). Another set of 30 serum samples were used for comparison with a tracer dialysis method (Quest Diagnostics).
reference intervals
We followed the CLSI guideline (C28-A2). Serum samples were collected from a total of 150 healthy adults and analyzed for thyroid-stimulating hormone (TSH). We excluded 13 samples because of abnormal TSH and used 137 (67 female, 70 male) samples with normal TSH concentrations (0.3–4.0 mU/L) to establish reference intervals for free thyroid hormones. Serum samples from 818 pregnant women were collected and analyzed to establish reference intervals by week of gestation. Ethnicity was 12% Asian, 23% black, 25% Hispanic, and 41% white; ages ranged from 13.3 to 45.4 years; gestation duration was 14 to 20 weeks. All samples were anti-Tg and anti-TPO negative, with TSH between 0.15 and 3.11 mU/L as described by La’ulu et al.(37). Statistical data analysis was performed with EP Evaluator (v. 7.0; David G. Rhoads Associates) and Excel (Microsoft).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
For FT4 and FT3, serum samples were stable for all conditions evaluated for 6 days, and no degradation was detected after 6 freeze-thaw cycles (single-factor ANOVA, P value <0.1 and >0.05 with 
= 0.05). The differences from the initial concentrations were <10% at all storage time points (P value <0.1 and >0.05 with = 0.05) for the processed samples stored at ambient temperature for 4 days. It is worth pointing out that the sera used to make the pools for these 2 studies were not freshly collected, were in –80 °C storage for about 1 year, and may have previously undergone multiple freeze-thaw cycles.
Mean carryover was <0.09% and <0.04% for FT4 and FT3, respectively, in 3 runs. For FT4, 150 µL serum produced 1.61% to 11.11% higher results (mean 5.29%) than 200 µL serum (12.60, 15.50, 18.55, 22.35 ng/L [pg/mL]), whereas 100 µL serum produced 13.21% to 18.12% increases (mean 14.71%). For FT3, 150 µL serum produced 0.14% to 7.03% higher results (mean 3.20%) than 200 µL serum (3.65, 4.20, 4.23, 4.52 ng/L [pg/mL]), whereas 100 µL of serum produced 6.79% to 20.04% increases (mean 14.90%).
There was no evidence of ion suppression observed on the chromatograms close to the retention times of T4, T3, and 13C6-T4. In 3 database searches, we produced a list of potential interfering compounds; however, all compounds were eliminated for interference study because they were not commercially available or not likely to be present in human serum. No evidence of interference was observed for all analytes and the IS in the results of >1000 patient samples. We performed evaluations for interferences by examining peak shape, split peak, peak shoulder, and peak area ratio of the 2 transitions(36) acquired for each compound. No interfering compounds have been identified to date.
Results for method comparison are illustrated in Fig. 2
for FT4 and Fig. 3
for FT3. Percentage bias plots for both FT4 and FT3 are presented in Fig. 4
. The observed correlation between FT4 and FT3 was y (FT3) = 0.22x (FT4) + 1.57, Sy x = 5.45, r = 0.87, n = 96 by the described method. The reference interval for normal adults was established as 12.8 (95% CI, <11.7 to 13.6) to 22.2 (20.2 to >28.6) ng/L for FT4 and 3.62 (<3.11 to 4.14) to 6.75 (6.60 to >7.05) ng/L for FT3 using nonparametric method (CLSI C28-A). One sample was observed with exceptionally high FT4 (28.6 ng/L) and marginal FT3 (6.6 ng/L). For 13 samples excluded with abnormal TSH, a range of 12.4–19.4 ng/L was obtained for FT4 and 4.79–7.95 ng/L for FT3. The reference intervals for pregnant women in the second trimester are presented by week of gestation in Table 1
.
|
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
). Completely resolved peaks were obtained for T4, T3, and rT3. Thorough cleanup by online SPE with high-efficiency LC separation and specific MRM detection provided high selectivity, as demonstrated by the absence of ion suppression and interferences in patient samples. Moreover, the LC column equilibration critical for the gradient method was performed during sample loading and washing of the extraction column, and cleaning and equilibration of the extraction column were performed during the chromatographic separation on the analytical column. A throughput of 480 samples per day could be achieved with <3.0 min sample cycle time. Compared to manual SPE, the advantages of this method include simplicity, high recovery, specificity, and throughput.
sample preparation
See the online Data Supplement for studies related to sample preparation procedure and dialysis buffer preparation, as well as studies related to absorption loss in preparing standard solutions and absorption loss onto both the 96-well polypropylene plate and the 96-well dialysis plate.
Protein leakage through the membrane, with spuriously high results, was observed during method development. It was the result of the membrane being pierced by pipette tips during sample handling, not defective devices. The 5-kDa molecular weight cutoff of the regenerated cellulose membrane used in the dialysis plate is low enough to stop proteins from diffusing through, consistent with several studies indicating that the dialysates are free from protein leakage(15)(38)(39). The online SPE-LC method is sensitive to the presence of proteins in the dialysate, which results in clogging the LC column or reduced retention time. These observations may serve as indicators of the protein leakage into the dialysate. In more than 6 months of routine operation, and >3000 samples analyzed in duplicate, protein leakage has not been observed with dialysis plates from different batches.
At least 2 factors are responsible for the increased concentrations of free hormones when lower sample volumes are used for dialysis. First, it is equivalent to a larger dilution if dialyzing less serum volume against a constant volume of buffer, and dilution has a complex effect on the obtained results(14)(15)(40). More importantly, the dialysand/dialysate should have a lower pH with the lower serum volume. It was confirmed that lower pH was associated with the higher results for free thyroid hormones (see the online Data Supplement).
performance evaluation
The CVs obtained with spiked dialysis buffers were similar to serum pools, implying that insignificant imprecision was introduced by the dialysis step. This is consistent with the fact that ED is a separation step, not a quantitative step.
Two forms of matrix effect, ion suppression and interferences, could severely impact the performance and usage of a LC-MS/MS method. Different means to reduce matrix effects include high-efficiency separation and thorough sample clean-up as used in the described method. Several protocols have been developed to study the presence of ion suppression(35). For a LC-MS/MS method, a potential interfering compound has to satisfy at least 3 criteria: first, it must at least partially coelute with the analyte; second, its ions, including any isotope series or any kind of adduct ions, must share the same m/z as the analyte/adduct ions under the ionization mode used; third, its ions must share the fragment ions chosen for the analyte. Thus, a list of potential interfering compounds can be generated by searching chemical databases and evaluating the probability of their presence in real samples. This approach, and monitoring 2 MRM transitions(36), showed no interferences in the analysis of >1000 patient samples by the described method.
method comparison
There was no full agreement observed with any evaluated method for FT4 or FT3 (Fig. 2
, 3
, and 4
). However, overall comparison strongly supports the validity of the reported method. A slope of 0.99 was obtained in comparison with the ED-RIA method. In the lower concentration ranges, excellent correlation was obtained for FT4 (E170 and Vitros ECi) and FT3 (E170), with a small intercept, correlation coefficient (r) close to unity, and small Sy x, although a slope different from unity could be explained by method calibration. Saturation of antibody was suspected in both E170 and Vitros ECi immunoassay for FT4; however, it was not observed with FT3, possibly due to much lower endogenous concentrations of FT3 than FT4. The presence of large discrepancies between methods (especially ED-RIA for FT4) could be explained by individual method design and performance. For example, the dilution effect (1:1 vs 1:12 serum-to-buffer ratio) could in part explain the large scattering of data points in comparison with the ED-RIA method (see section for dialysis buffer in the online Data Supplement).
reference intervals
The narrow normal distribution of FT4 and FT3 for the reference interval study is in agreement with tight control by the hypothalamus-pituitary-thyroid axis(1)(2). However, one exceptionally high FT4 was observed with a normal FT3 and TSH. The 13 samples rejected by TSH for reference interval study produced FT4 results within the established reference interval, whereas FT3 shifted toward the upper normal range. Furthermore, FT4 and FT3 correlated poorly. The above observations suggest the differences in clinical relevance of FT4, FT3, and TSH(1)(2). Results of the reference interval study in pregnant women suggest a shift toward lower values of FT4 and FT3 compared with normal adults. There is a small trend toward lower values of FT4 from week 14 to week 20 of gestation.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
|
Acknowledgments |
|---|
Financial Disclosures: None declared.
Acknowledgment: This study was supported by the ARUP Institute for Clinical and Experimental Pathology.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  A A A Ismail Interference from endogenous antibodies in automated immunoassays: what laboratorians need to know J. Clin. Pathol., August 1, 2009; 62(8): 673 - 678. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Jonklaas, N. Kahric-Janicic, O. P. Soldin, and S. J. Soldin Correlations of Free Thyroid Hormones Measured by Tandem Mass Spectrometry and Immunoassay with Thyroid-Stimulating Hormone across 4 Patient Populations Clin. Chem., July 1, 2009; 55(7): 1380 - 1388. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M. Thienpont A Major Step Forward in the Routine Measurement of Serum Free Thyroid Hormones Clin. Chem., April 1, 2008; 54(4): 625 - 626. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
