Increased fT4 concentrations in patients using levothyroxine without complete suppression of TSH

Introduction

Thyroid hormone measurements are commonly performed in clinical chemistry laboratories since thyroid disorders are highly prevalent and the prevalence is even rising (1). The first step in the diagnosis of thyroid disorders is the measurement of serum thyroid-stimulating hormone (TSH) concentration, possibly followed up with serum free thyroxine (fT4) concentration when TSH is above or below the reference interval. Once a thyroid disorder is diagnosed and treatment is started, thyroid hormone status is regularly monitored by measuring TSH and/or fT4 concentrations depending on the etiology of the thyroid disorder. To illustrate, Hashimoto’s hypothyroidism can sufficiently be monitored by measuring TSH solely and fT4 may, whether necessary or not, sometimes be used to ascertain adequate thyroid hormone status. On the other hand, TSH cannot be measured reliably in patients with central hypothyroidism and fT4 is the key parameter to monitor in this group. Over the past years, physicians in our hospital noticed a discrepancy between TSH and fT4 concentrations in the laboratory, more specifically high fT4 concentrations (above the upper limit of normal, ULN) without complete suppression of TSH (complete suppression meaning TSH ≤ 0.01 mU/L) in blood samples of patients at the outpatient clinic. These discordant results appeared to occur more frequently following the introduction of a new electrochemiluminescence fT4 immunoassay (ECLIA, Roche Cobas e601) in the laboratory in 2017. TSH measurements were, since 2008, performed using an ECLIA assay (Roche Cobas e601) with a reference interval set at 0.5–5.0 mU/L. The observation of fT4 concentrations above the ULN with discordant, slightly suppressed TSH (0.02–0.5 mU/L) or even TSH concentrations within the reference intervals (0.5–5.0 mU/L) may have several causes. First, the set reference intervals may not fit the population. Second, the newly implemented Cobas fT4 immunoassay suffers from analytical issues. Third, patient-related factors could play a role in the discordant results. Therefore, the aim of our study was to investigate whether these three possible causes could underlie the observed discordant TSH and fT4 results.

Materials and methods

Results

Re-evaluation of the 95% double-sided reference interval of the Delfia and Cobas immunoassay confirmed the reference intervals in use. The sample of one participant was excluded because of a TSH < 0.01 mU/L. Non-parametric analyses found a reference interval of respectively 10.8–22.6 pmol/L (Delfia, in use 10–23 pmol/L) and 12.6–22.3 pmol/L (Cobas, in use 12–22 pmol/L).

Second, we found no evidence for the existence of analytical flaws in the Cobas fT4 immunoassay. The internal quality controls of the fT4 Cobas immunoassay showed a long-term interassay CV of 2.2% at 17.1 pmol/L and 2.0% at 41.8 pmol/L that remained within 2 standard deviation (SD) limits by weekly adjustments. The fT4 Cobas immunoassay also performed well in EQAS with an annual (2018) precision of +4.1% and a trueness of −0.6%, and optimal Sigma SA and TE scores of 6.0, compared to other Cobas users in the Netherlands.

Third, we investigated 3174 fT4 + TSH requests using the former fT4 Delfia immunoassay and 3408 fT4 + TSH requests using the current fT4 Cobas immunoassay. FT4 concentrations ranged from 23.1 to 70 pmol/L in the Delfia cohort and from 22.1 to 47.6 pmol/L in the Cobas cohort. Figure 1 shows the distribution of the requests and frequency of discordant pairings. The percentage of fT4 results > ULN accompanied with a slightly suppressed TSH or TSH within the reference interval differed statistically significantly between the Delfia and Cobas cohorts (P = 0.032; P < 0.001 respectively), although the absolute difference was limited (Fig. 1). We further studied the patient records of the discordant fT4 and TSH pairings to check for notable similarities and found that 81% of the samples with fT4 > ULN and a slightly suppressed TSH and 72% of the samples with fT4 > ULN and a TSH within the reference interval belonged to levothyroxine (L-T4) users, as shown in Fig. 1. Furthermore, we noticed that, respectively, 4.7 and 8.2% of the discordant samples belonged to amiodarone users. Table 1 shows the mean fT4 and median TSH concentrations of the complete Cobas and Delfia cohorts. Table 2 shows the median fT4 and TSH concentration of patients from the Cobas cohort presenting with an increased fT4 and discordant TSH concentration. Results were disaggregated between L-T4 users, amiodarone users, and patients using neither L-T4 nor amiodarone. No notable differences emerged from these results.

View Full Size

Distribution of fT4 + TSH requests and frequency of discordant pairings.

Citation: Endocrine Connections 12, 4; 10.1530/EC-22-0538

• Download Figure
• Download figure as PowerPoint slide

Since 72–81% of the studied patients with discordant fT4 + TSH results used L-T4, literature was researched to explain this phenomenon with a specific focus on time of blood withdrawal and time of L-T4 intake. Multiple studies have investigated serum fT4 concentrations directly after L-T4 ingestion, all including hypothyroid patients treated with a stable dose of L-T4. These studies all reported an equivalent course of fT4 after morning L-T4 intake (3, 4, 5, 6, 7, 8, 9, 10). Figure 2 depicts a summary of this literature search; it shows that fT4 concentrations rise after 1 h and peak between 2 and 4 h after L-T4 intake (+15–25%) followed by a gradual decline and return to baseline within 24 h in accordance with the known time to maximal concentrations of L-T4 (T-max = 2–3 h).

View Full Size

The course of free T4 (fT4) concentration (% change) in patient sera during 24 h following morning L-T4 ingestion. Data are based upon the studies below. The measurement at 0 h indicates the measurement before L-T4 intake (3, 4, 5, 6, 7, 8, 9, 10).

Citation: Endocrine Connections 12, 4; 10.1530/EC-22-0538

• Download Figure
• Download figure as PowerPoint slide

Discussion

This study aimed to investigate which possible causes could underlie discordant results of fT4 above the ULN together with a TSH within the reference interval or a slightly suppressed TSH (0.02–0.5 mU/L). Our study showed that this discordance was not related to poor analytical performance or decision limits and reported that most patients with this finding were using L-T4.

We showed that higher fT4 concentrations along with a discordant, non-suppressed, TSH could not be explained by incorrect reference intervals nor by abnormal analytical variation. The assay variation was as expected from the kit insert and local verification and the Cobas fT4 immunoassay performed well in the Dutch EQAS. Although the frequency of observed discordance was statistically significantly higher in 2018–2019 (Cobas cohort) vs 2015–2016 (Delfia cohort), the difference was limited. Hence, the introduction of the new Cobas immunoassay for the measurement of fT4 could not explain the notification of increased TSH-fT4 discordancy in samples from patients in our clinic. We did not screen the Delfia cohort for use of L-T4 or other striking characteristics, but we have no evidence for changes in the patient population over this period of 5 years’ time. We do believe that informing the physicians about the new Cobas immunoassay may have caused an additional focus on the fT4 results which led to a subjective increase in discordant values. We did assess the use of amiodarone which was present in 4.7 and 8.2% of the cohorts with discordant fT4 and TSH results. We did, however, not investigate the use of other drugs, such as antithyroid drugs, beta-blockers, or glucocorticoids which could have influenced fT4 results as well and should ideally be looked at in future (prospective) studies (11).

We found that an fT4 > ULN together with a discordant TSH was most frequently present in patients using L-T4 which was confirmed by literature (12, 13). There may be various causes contributing to this phenomenon in L-T4 users. Timing of blood withdrawal following L-T4 intake can lead to high fT4 concentrations without (complete) TSH suppression. Hypothyroid patients are mainly treated with L-T4 and the effect of treatment is monitored by measuring serum TSH, sometimes accompanied by fT4. Hypothyroid patients are advised to take a single daily dose of L-T4 orally in a fasting state. L-T4 administration in the morning or at bedtime is considered equally effective as long as L-T4 is taken on an empty stomach to ensure optimal uptake (14, 15). In contrast to fT4, no direct alterations of TSH and T3 have been reported directly after L-T4 ingestion (4, 5, 6, 7, 8). An fT4 course as Fig. 2 presents was found in patients taking L-T4 in the morning before breakfast, and one would also expect an increase in fT4 levels during the night when L-T4 is taken at bedtime (16). However, literature on extensive follow-up of fT4 and TSH levels following L-T4 intake in the morning compared to bedtime is lacking. Ain et al. (7) as well as Hoermann et al. (17) specifically emphasized that fT4 concentrations in L-T4 users were influenced by the time of day, meaning the time interval between L-T4 intake and blood sampling should be considered in the interpretation of fT4 values. In line with this advice, the European Thyroid Association guideline on treating central hypothyroidism advises blood withdrawal for monitoring treatment to be performed before L-T4 intake or at least 4 h after L-T4 intake (18), but other international guidelines do not yet (19) (https://richtlijnendatabase.nl/richtlijn/schildklierfunctiestoornissen/schildklierfunctiestoornissen_-_korte_beschrijving.html; https://richtlijnen.nhg.org/standaarden/schildklieraandoeningen#volledige-tekst-richtlijnen-diagnostiek). To the best of our knowledge, only Conte et al. (20) investigated if timing of blood sampling affected the evaluation of (morning) L-T4 treatment in patients with differentiated thyroid carcinoma and showed an increase in fT4 concentration in the afternoon compared to the morning without TSH deviation in this group. The observed increase (2.8 pmol/L) might not always be determined clinically relevant because treatment decisions are mainly based on changes in TSH. However, such fluctuations in fT4 concentrations may be relevant for evaluating treatment effects if TSH cannot be used as a marker to monitor thyroidal status (e.g. in central hypothyroidism).

Furthermore, a higher fT4/fT3 ratio was seen in patients using L-T4 compared to healthy controls (12, 21, 22). Hypothyroid patients receive L-T4 supplementation which replaces T4 only and thus these patients partially lack the active thyroid hormone triiodothyronine (T3) derived from the thyroid gland. This suggests a need for increased conversion of T4 into T3 in the peripheral tissues to reach adequate tissue T3 concentrations (21), resulting in the need for a higher L-T4 dosage to obtain a normal, or even somewhat lower, concentration of T3 (22, 23). Moreover, TSH is more sensitive to changes in T3 than in T4, clarifying the lack of negative pituitary feedback to increased fT4 concentrations leading to not (completely) suppressed TSH. This phenomenon may be more prominent in patient groups that are characterized by an even greater deficit of endogenous T3 production (e.g. athyroid patients). Even though in athyroid patients some T3 is still produced by deiodinases in several tissues, previous literature showed an outspoken dissociation between fT4 and fT3 concentrations in this group where even a higher L-T4 dosage did not lead to adequate fT3 concentrations (24). Thus, the adapted peripheral conversion of T4 into T3 in L-T4 users leads to an increased fT4 concentration combined with a not (completely) suppressed TSH, which is reflected in a higher fT4/fT3 ratio.

In conclusion, the observed fT4 concentrations above the ULN together with not (completely) suppressed TSH were not caused by incorrect reference intervals or analytical problems but mainly related to L-T4 use. Most likely, a combination of timing of blood withdrawal and the timing of L-T4 intake causes this phenomenon. The retrospective design of our study did not allow us to investigate this further in our cohorts. Physicians and laboratory specialists should be aware of the importance of timing of blood withdrawal and the timing of L-T4 intake to avoid questioning the assay’s performance or, worse, unnecessarily adapting L-T4 dose in patients. The clinical implications of discrepant TSH and fT4 concentrations may differ among specific L-T4 treated groups, as treatment and follow-up approaches are different. One could argue that TSH measurements suffice during treatment follow-up, but measuring TSH alone in L-T4 users has its limitations as well (25). These findings may argue for the application of an integrative approach by adding the measurement of fT3 or the fT4/fT3 ratio as a more reliable reflection of thyroid hormone status in L-T4 users considering the adapted peripheral conversion of T4 into T3 in L-T4 users. Furthermore, specific reference intervals or altered target values of fT4 for L-T4 users could be determined and used, ideally specified for different specific L-T4 treated groups (e.g. Hashimoto’s hypothyroidism, central hypothyroidism, athyroid patients). Last and potentially easiest, it can be advised to draw blood at fixed time points to ensure proper result comparisons over time.

Declaration of interest

There is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This study did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.