Combined Thyroxine/Liothyronine Treatment Does Not Improve Well-Being, Quality of Life, or Cognitive Function Compared to Thyroxine Alone: A Randomized Controlled Trial in Patients with Primary Hypothyroidism

Abstract

T₄ is standard treatment for hypothyroidism. A recent study reported that combined T₄/liothyronine (T₃) treatment improved well-being and cognitive function compared with T₄ alone. We conducted a double-blind, randomized, controlled trial with a crossover design in 110 patients (101 completers) with primary hypothyroidism in which liothyronine 10 μg was substituted for 50 μg of the patients’ usual T₄ dose. No significant (P < 0.05) difference between T₄ and combined T₄/T₃ treatment was demonstrated on cognitive function, quality of life scores, Thyroid Symptom Questionnaire scores, subjective satisfaction with treatment, or eight of 10 visual analog scales assessing symptoms. For the General Health Questionnaire-28 and visual analog scales assessing anxiety and nausea, scores were significantly (P < 0.05) worse for combined treatment than for T₄ alone. Serum TSH was lower during T₄ treatment than during combined T₄/T₃ treatment (mean ± sem, 1.5 ± 0.2 vs. 3.1 ± 0.2 mU/liter; P < 0.001), a potentially confounding factor; however, subgroup analysis of subjects with comparable serum TSH concentrations during each treatment showed no benefit from combined treatment compared with T₄ alone. We conclude that in the doses used in this study, combined T₄/T₃ treatment does not improve well-being, cognitive function, or quality of life compared with T₄ alone.

THYROXINE (T₄) IS THE standard replacement therapy for hypothyroidism (1), but in some patients, symptoms of ill health persist despite T₄ treatment. It is not clear whether this is because of comorbidity or because standard T₄ replacement is in some way inadequate for some individuals (2–4).

T₄ has little intrinsic biological activity, and its metabolic effects are achieved by peripheral conversion to liothyronine (T₃). The thyroid also secretes T₃ directly, and in humans this accounts for about 20% of the body’s total T₃ production (5). A truly physiological thyroid replacement regimen would therefore include both T₄ and T₃. Combined T₄/T₃ therapy was widely used in the past, in the form of desiccated thyroid, but with the availability of synthetically prepared T₄, the latter preparation became preferred on account of its long half-life, stable pharmacokinetics, and more precise standardization (6). A clinical trial published in 1970 compared combined T₄/T₃ treatment with T₄ alone and found that T₄ was better tolerated (7); however, the doses of both T₄ and T₃ that were used would now be regarded as excessive.

Recent studies revived interest in combined T₄/T₃ treatment for hypothyroidism. In rats rendered hypothyroid by thyroidectomy and radioiodine treatment, T₄ alone failed to normalize circulating and tissue concentrations of T₄ and T₃, but this was achieved with combined T₄/T₃ treatment (8, 9). Subsequently, a small clinical trial reported by Bunevicius et al. (10) found that partial substitution of liothyronine for T₄ resulted in improved mood, well-being, and measures of cognitive function compared with T₄ alone and was preferred by most of the patients. If these results are confirmed, combined T₄/T₃ treatment might become standard thyroid replacement therapy.

We conducted a double-blind, randomized, controlled trial comparing the effects of combined T₄/T₃ treatment and T₄ alone on symptoms of hypothyroidism, quality of life, cognitive function, and subjective satisfaction with T₄ therapy.

Patients and Methods

Patients and recruitment

Recruitment to the study commenced in April 2000, and the study was completed in November 2002. We aimed to enroll both subjects with persistent symptoms despite T₄ therapy and also subjects who felt well while taking T₄, but specific recruitment targets were not set for these subgroups. Inclusion criteria were primary hypothyroidism of at least 6-month duration, a stable dose of T₄ of 100 μg/d or more, no change in T₄ dosage in the previous 2 months, and serum TSH concentration between 0.1–4.0 mU/liter (reference range in our laboratory, 0.3–4.0 mU/liter) at a screening visit. The diagnosis of hypothyroidism was confirmed from medical or laboratory records or by contacting primary care physicians; in a few patients with long-standing hypothyroidism, this was not possible.

The principal exclusion criteria were major comorbidity, current or recent T₃ treatment, history of thyroid cancer requiring suppression of TSH secretion, cardiac disease, and use of drugs that affect thyroid hormone secretion, metabolism, or bioavailability or measures of thyroid hormone action. Untreated major depression was an exclusion criterion, but patients receiving antidepressant treatment were eligible, provided treatment had been unchanged for the past 3 months and was likely to continue for the duration of the study.

Recruitment was from endocrinology outpatient clinics and private practices (n = 49), primary care physicians (n = 9), and local advertisement (n = 52). Thirty-one subjects who responded to advertisements were screened by telephone and were ineligible or declined to participate; a further 20 subjects attended a screening visit, but were deemed ineligible or declined to participate.

At the screening visit, subjects were classified either as satisfied (adequate clinical response to T₄ with no persistent symptoms) or dissatisfied (persisting symptoms despite T₄ replacement) by a single clinician (J.P.W.) based on the clinical history. Typical complaints among dissatisfied patients were tiredness, impaired well-being, or weight gain. A clinical history and physical examination were performed, and routine blood tests (blood count, serum creatinine, and liver function tests in all patients; serum calcium and iron studies where indicated) were performed to exclude comorbidities that might account for symptoms of ill-health.

Study design, treatments, and evaluation

The study had a double-blind, crossover design, with the order of treatment randomized in permuted blocks of 10 using sealed envelopes. Fifty-six patients were randomized to T₄, followed by combined T₄/T₃ treatment, and 54 patients were randomized to combined treatment first. At study entry, patients reduced their daily T₄ dose by 50 μg and took the study medication (either 50 μg T₄ or 10 μg liothyronine (Tertroxin, Boots Healthcare Australia, North Ryde, New South Wales, Australia) in capsules of identical appearance) in addition to their reduced T₄ dose. Treatment periods lasted 10 wk, separated by a 4-wk period during which patients resumed their usual T₄ dosage. Liothyronine (10 μg/d) was used because this dose is thought to be bioequivalent to 50 μg T₄ (11) and is similar to the dose of 12.5 μg/d used by Bunevicius et al. (10). In Australia, liothyronine is available only as a 20-μg tablet, and it was therefore impractical to replicate exactly the dose used by Bunevicius et al.

At baseline and at the end of each treatment period, subjects attended after an overnight fast and before taking T₄ or study medication (i.e. 24 h after the previous dose). Venous blood and a random urine sample were collected for measurement of serum TSH, free T₄, and free T₃; serum SHBG and plasma cholesterol (markers of thyroid hormone action on liver); plasma alkaline phosphatase, serum osteocalcin, and urinary deoxypyridinoline/creatinine ratio (markers of thyroid hormone action on bone). Symptoms and signs of hypothyroidism were assessed using the Billewicz scale as modified by Zulewski et al. (12), which gives a tissue hypothyroidism score out of 13. Resting pulse rate and blood pressure were measured in the supine position as cardiovascular markers of thyroid hormone action, and ankle reflex relaxation time was assessed using a photomotogram (13). Treatment compliance was assessed by counting unused capsules.

At each visit, patients self-administered three questionnaires: the Short Form 36 (SF-36), the General Health Questionnaire 28 (GHQ-28), and the Thyroid Symptom Questionnaire (TSQ). The SF-36 (14) was selected as a generic quality of life instrument that has been validated in the Australian population (15). It consists of eight individual scales and two composite scales: the physical component summary and mental component summary scores. The GHQ-28 (16) was selected as a well-validated and widely used measure of psychological function or disturbance, which has previously been used to assess patients with thyroid disease (3, 17). It consists of a global score and four subscales: somatic symptoms, anxiety/insomnia, social dysfunction, and severe depression. The TSQ (3) was selected as a disease-specific instrument and has been shown to be sensitive in detecting impaired well-being in T₄-treated subjects. The SF-36 was scored by standard methods (14). The GHQ-28 and TSQ were scored using a four-point Likert scale, with each response scored 0, 1, 2, or 3 (3, 16).

Subjects also completed 10 visual analog scales, each consisting of a pair of phrases, such as “as sad as possible” or “as happy as possible” at either end of a 100-mm line, giving a score in millimeters from the left-hand end. The scales assessed general well-being, happiness/sadness, confusion, anxiety, irritability, tiredness, feeling hot/cold, sickness/nausea, blurred vision, and aches and pains. Visual analog scales have been shown to be useful in assessing hypothyroid symptoms and responses to treatment (10, 18). Subjective satisfaction or dissatisfaction with each treatment was rated on a 4-point scale, ranging from “very satisfied: my thyroid treatment seems very effective” (scoring 0) to “very dissatisfied: my thyroid treatment doesn’t seem to work at all” (scoring 3). At the final visit, patients were asked which treatment they preferred.

Cognitive function was assessed by a clinical psychologist using three standard, well validated tests: the Symbol Digit Modalities Test (19), which assesses cognitive efficiency and ability to undertake a novel task; the Trail Making Test Parts A and B (20), which assesses visual search, attention, mental flexibility, and motor function; and the Digit Span Sub-Test (both Forwards and Backwards) of the Wechsler Adult Intelligence Scale III (21), which assesses immediate auditory memory, attention, and concentration.

Biochemistry methods

TSH, free T₄, and free T₃ were measured by chemiluminescence immunoassay on the Abbott Diagnostics Architect (Abbott Diagnostics, North Ryde, Australia). SHBG was measured by enzyme immunoassay using chemiluminescence substrate on Immulite 2000 (Diagnostic Products, Los Angeles, CA). Deoxypyridinoline was measured by an in-house ion paired reverse phase HPLC with fluorescence detection (22). Osteocalcin was measured by in-house RIA (23). Cholesterol and alkaline phosphatase were analyzed by standard biochemical methods on a Hitachi 917 analyzer (Roche, Indianapolis, IN). Intra- and interassay coefficients of variation were as follows: TSH, 1.2 and 2.9%; free T₄, 3.8 and 3.6%; free T₃, 3.0 and 5.1%; SHBG, 4.1 and 6.0%; and osteocalcin, 12.3 and 14.5%, respectively; the interassay coefficient of variation for deoxypyridinoline was 8.0%.

Statistical analysis

Baseline characteristics for the satisfied and dissatisfied groups were compared by t test (for continuous variables), Wilcoxon rank-sums test (for ordinal variables), or Fisher’s exact test; descriptive data are presented as the mean ± sd. Quality of life scores, cognitive function tests, and clinical and biochemical data were analyzed by repeated measures ANOVA using PROC GLM, a procedure within SAS*, to compare the effects of treatments after adjusting for subject and period effects; data are presented as the adjusted mean ± sem. Treatment preference was analyzed by χ² test. The significance level was set at 0.05.

There were four prespecified subgroup analyses: 1) patients classified at baseline as satisfied or dissatisfied with T₄ treatment, as a differential symptomatic response to treatment was possible; 2) patients with serum T₃ less than 3 pmol/liter (the lower limit of the reference range) at baseline, as such patients might be in some way deficient in T₃, compared with those with a normal serum T₃ concentration; 3) patients with serum TSH below 2 mU/liter at baseline compared with those with TSH of 2 mU/liter or more, as some authorities recommend that serum TSH should be in the lower reference range in T₄-treated patients (24, 25); and 4) patients with autoimmune hypothyroidism compared with those with postsurgical and postradioiodine hypothyroidism.

Power calculations

Sample size calculations were based on published data (3, 15, 18, 26); in all cases, α was set at 0.05. For the SF-36, a sample size of 70 subjects gave 90% power to detect a 2-point difference between treatments in the physical or mental component summary scores; this is clinically meaningful, as the 1995 Australian National Health Survey found a 2.2-point difference in the physical component summary score and a 5.4-point difference in the mental component summary score between subjects with and without thyroid disease (Australian Bureau of Statistics, personal communication). No data were available for the GHQ-28 using 4-point Likert scoring on which to base calculations; however, for the abbreviated GHQ-12, a sample size of 100 subjects would give 80% power to detect a clinically meaningful 2-point difference between treatments. This suggested that a sample size of 100 was reasonable in the present study using the GHQ-28. For the TSQ, a sample size of 82 subjects gave 80% power to detect a 2-point difference between treatments. For the Symbol Digit Modality Test, a sample size of 82 subjects gave 90% power to detect a 5-point difference; for the Digit Span Test (Backwards), 38 patients were needed for 90% power to detect a one-digit difference; and for the Trail Making Test (Part B), 100 subjects were required to give 80% power to detect a 5-sec difference between treatments. The required sample size was, therefore, 100, and the recruitment target was set at 110 to allow for withdrawals.

Ethical approval

The study protocol was approved by the human research ethics committee of Sir Charles Gairdner Hospital. Informed consent was obtained from all participants.

Results

Baseline characteristics and retention

The baseline characteristics of the 110 study subjects, and the satisfied and dissatisfied subgroups are shown in Table 1. Ninety-four patients (85%) had autoimmune or idiopathic hypothyroidism; of 12 patients with postsurgical hypothyroidism, three had a history of Graves’ disease, and one of Hashimoto’s disease, whereas three of four patients with radioiodine-induced hypothyroidism had a history of Graves’ disease. Dissatisfied subjects tended to be younger than satisfied subjects and had significantly worse quality of life (determined by SF-36) and more symptoms and signs of hypothyroidism (as determined by TSQ and Zulewski score). A history of depression was more common in the dissatisfied group, but not significantly so (33% vs. 18%; P = 0.13). More dissatisfied subjects were taking antidepressants at baseline (16% vs. 2%; P = 0.02), but this accounted for only a minority of patients. Psychological well-being was significantly worse in dissatisfied patients, as determined by the total GHQ-28 score, but only a few subjects in each subgroup had a GHQ score greater than 39 (4% of satisfied vs. 11% of dissatisfied subjects; P = 0.29), the accepted threshold score for the detection of psychiatric disorder (27). Dissatisfied subjects had worse scores for somatic symptoms, anxiety/insomnia, and social dysfunction, but there was no difference between subgroups in scores for severe depression. Comorbidities other than depression were not more prevalent among dissatisfied subjects. T₄ dosage and baseline serum TSH were not significantly different between subgroups; serum TSH was less than 2 mU/liter in 67% of satisfied subjects and in 74% of dissatisfied subjects.

Of 110 subjects recruited, 101 (92%) completed the study. Of the nine subjects who withdrew, seven were female. Reasons given for withdrawal were pregnancy (n = 2), surgery for unrelated conditions (n = 2), worsening symptoms (n = 2), inability to attend study visits (n = 1), nasal congestion and palpitations (n = 1), and depression and insomnia (n = 1). Compliance with treatment was good, with a mean of 98% of study capsules taken. No significant adverse effects were reported during treatment.

Quality of life, cognitive function, and treatment preference

The results of the quality of life measures at the end of each treatment are shown in Table 2. For the SF-36, there was no significant difference between T₄ and combined T₄/T₃ treatments for the physical component summary score, the mental component summary score, or any of the eight individual scales of the SF-36. For the GHQ-28, the overall score was significantly higher (indicating worse psychological well-being) for combined treatment compared with T₄ alone (18.3 ± 1.0 for T₄vs. 21.2 ± 1.0 for T₄/T₃; P = 0.033). In each of the four subscales of the GHQ-28, the mean score was higher (indicating worse symptoms) for combined treatment than for T₄ alone, but the difference was statistically significant only for social dysfunction (6.7 ± 0.3 for T₄vs. 7.7 ± 0.3 for T₄/T₃; P = 0.028). For the TSQ, there was no significant difference between T₄ and combined T₄/T₃ treatment.

There were no significant differences between treatments for eight of 10 visual analog scales. Anxiety scores were significantly worse for combined T₄/T₃ treatment than for T₄ (24.9 ± 1.8 for T₄vs. 30.7 ± 1.8 for T₄/T₃; P = 0.026), as were scores for sickness/nausea (12.8 ± 1.6 for T₄vs. 17.4 ± 1.6 for T₄/T₃; P = 0.049). The treatment satisfaction score did not differ significantly between treatments.

The results of the cognitive function tests are shown in Table 3. There was no significant difference between treatments in any of the tests.

Of 101 patients completing the study, 46 preferred T₄ treatment, 36 preferred combined T₄/T₃ treatment, and 18 had no preference. This is not different from results expected by chance (P = 0.32).

On analysis of the prespecified subgroups based on baseline variables [satisfied (n = 46) or dissatisfied (n = 55); TSH, <2 (n = 74) or ≥2 mU/liter (n = 27); free T₃, <3 (n = 34) or ≥3 pmol/liter (n = 67); autoimmune (n = 85), surgical (n = 12), or radioiodine-induced hypothyroidism (n = 4)], no subgroup could be identified in which combined T₄/T₃ treatment improved quality of life or cognitive function compared with T₄ alone. For the SF-36, there was no treatment difference in the physical or mental component summary scores. For individual SF-36 domains, subjects with baseline TSH below 2 mU/liter had better scores during T₄ treatment than combined T₄/T₃ treatment in physical functioning (T₄, 83.6 ± 1.3; T₄/T₃, 80.0 ± 1.3; P = 0.045) and bodily pain (74.0 ± 1.5 vs. 68.2 ± 1.5; P < 0.01), whereas subjects with baseline TSH of 2 mU/liter or more had apparent improvement in social functioning during combined T₄/T₃ treatment (T₄, 70.7 ± 3.8; T₄/T₃, 84.7 ± 3.8; P = 0.016). For the GHQ-28 total score, there were significant treatment differences between T₄ and T₄/T₃ treatment for satisfied patients (T₄, 17.3 ± 1.1; T₄/T₃, 21.0 ± 1.1; P = 0.026), and for subjects with baseline TSH less than 2 mU/liter (18.0 ± 1.1 vs. 21.7 ± 1.1; P = 0.0252); in each case the results favored T₄ therapy. There were no significant differences in GHQ subscales for any subgroup. For the TSQ score, there was a significant treatment difference for patients with baseline TSH below 2 mU/liter (T₄, 11.5 ± 0.6; T₄/T₃, 13.1 ± 0.6; P = 0.042), indicating fewer hypothyroid symptoms during T₄ therapy. For visual analog scales, scores for anxiety (T₄, 22.7 ± 2.6; T₄/T₃, 30.6 ± 2.6; P = 0.026) and blurred vision (T₄, 22.3 ± 3.4; T₄/T₃, 32.9 ± 3.5; P = 0.034) were significantly worse during combined treatment in patients who were satisfied at baseline, whereas nausea was significantly worse during combined treatment in patients with serum free T₃ of 3 pmol/liter or more at baseline (T₄, 10.9 ± 2.1; T₄/T₃, 18.6 ± 2.1; P = 0.011).

Clinical parameters

There was no difference between treatments in weight, blood pressure, or ankle jerk relaxation time (Table 4). The resting pulse rate was significantly lower during combined treatment than with T₄ alone, but the magnitude of the difference was small (68.8 ± 0.5 for T₄vs. 67.3 ± 0.5 for T₄/T₃; P = 0.048). The tissue hypothyroidism score measured by the methods of Zulewski et al. (12) was significantly higher (indicating a greater number of symptoms and signs of hypothyroidism) for combined treatment (3.5 ± 0.1 for T₄vs. 3.9 ± 0.1 for T₄/T₃; P = 0.041).

Biochemistry results

During combined T₄/T₃ treatment, serum free T₄ was significantly (P < 0.001) lower than during T₄ treatment (Table 5), whereas there was no significant difference in serum free T₃ concentrations. Serum TSH was significantly higher during combined therapy compared with T₄ alone (1.5 ± 0.2 mU/liter for T₄vs. 3.1 ± 0.2 for combined T₄/T₃; P < 0.001). Serum SHBG was significantly (P < 0.01) lower, and plasma cholesterol higher (P = 0.015) during combined treatment compared with T₄ alone.

Subgroup analysis taking into account differences in TSH between treatments

The increase in mean serum TSH concentration during combined T₄/T₃ treatment compared with T₄ alone raised the possibility that beneficial effects of combination therapy were being masked by mild tissue hypothyroidism during this treatment. To explore this possibility, a post hoc subgroup analysis was carried out in subjects (n = 39) whose serum TSH concentrations at the end of the two treatments differed by 0.99 mU/liter or less. The rationale for this was that serum TSH concentrations in individuals fluctuate over time, and if serum TSH differs by 0.99 mU/liter or more in the same (euthyroid) individual on two occasions, the difference is statistically significant at the 1% level (28). The results of this analysis are shown in Table 6. As expected, serum TSH no longer differed significantly between treatments, whereas the difference in serum free T₄ was similar to that in the group as a whole. Serum SHBG remained significantly lower during combined treatment compared with T₄, whereas differences in cholesterol, pulse rate, and Zulewski score were no longer significant. There were no significant treatment differences in SF-36 physical and mental component summary scores, individual scales of the SF-36 (data not shown), TSQ score, GHQ total score and subscales, treatment satisfaction score, or nine of 10 visual analog scales, nor was there any trend toward improved scores during combined treatment compared with T₄. For the visual analog scale assessing anxiety, scores remained significantly (P = 0.035) worse for combined T₄/T₃ treatment compared with T₄. In this subgroup, 16 subjects preferred T₄ treatment, 19 preferred combination therapy, and four had no preference, which was not significantly different from results expected by chance (P = 0.61).

Confounding factors

Potential confounding factors were identified in nine subjects who completed the study: compliance less than 90% (n = 3), intercurrent illness (n = 2), commencement or change in dose of antidepressant (n = 2), undiagnosed pregnancy (n = 1), and stopped sex hormone replacement (n = 1). Excluding these subjects from the analysis did not alter significance of the results, except for pulse rate, which was no longer significantly different between treatments (T₄, 68.5 ± 0.6; T₄/T₃, 67.2 ± 0.6; P = 0.11).

Discussion

In this study, no benefit of combined T₄/T₃ treatment over standard T₄ therapy could be demonstrated on quality of life, hypothyroid symptoms, cognitive function, subjective satisfaction with thyroid replacement therapy, or treatment preference. No subgroup of patients could be identified who benefited symptomatically from combined T₄/T₃ treatment; in particular, there was no evidence of benefit in the clinically important subgroup of patients complaining of persistent symptoms of hypothyroidism despite T₄ replacement.

Our results differ from those of Bunevicius et al. (10), who reported improved well-being and cognitive function with combined T₄/T₃ treatment. There are several possible explanations for this discrepancy. Firstly, the previous study was smaller (33 patients), and sample size calculations were not reported in the paper. Secondly, the majority of patients in the previous study had a history of thyroid cancer and were presumably receiving T₄ suppressive therapy (which sometimes causes adverse effects) rather than standard replacement therapy. They were therefore not representative of hypothyroid subjects in general. Thirdly, treatment periods in the previous study were only 5 wk, with no intervening washout period, which is barely long enough for steady state to be reached after changes in T₄ dosage. This may well have confounded the results. The treatments used in the two studies differed slightly, in that we substituted 10 μg liothyronine for 50 μg of the patients’ usual T₄ dose, whereas Bunevicius et al. used 12.5 μg liothyronine, but it is unlikely that this accounts for the markedly different results of the studies. Our results are, however, consistent with those of another study (reported in abstract form) of 48 subjects with a parallel design, in which combined T₄/T₃ treatment did not improve symptoms compared with T₄ alone (29).

During combined T₄/T₃ treatment, the mean serum free T₄ concentration decreased, free T₃ was unchanged, and serum TSH increased compared with values during treatment with T₄ alone. The lack of difference in free T₃ concentrations was expected, because blood samples were taken 24 h after the previous dose of T₄ and T₃, and the plasma half-life of T₃ is less than 1 d (30). The increase in serum TSH concentrations was somewhat unexpected, as 10 μg liothyronine and 50 μg T₄ are thought to have similar biological potencies (11, 31, 32), and in the study by Bunevicius et al. (10), serum TSH concentrations did not differ significantly during T₄ and combined T₄/T₃ treatments. A key difference is that Bunevicius et al. took blood samples 2 h after ingestion of T₄ and liothyronine, when serum T₃ concentrations are likely to be at their peak (30), whereas we took samples 24 h after the previous dose. If circulating T₃ concentrations regulate TSH secretion (33), then because of the short half-life of T₃, it is possible that during combined T₄/T₃ treatment serum TSH concentrations increase slightly as serum T₃ concentrations decline and would be highest 24 h after the previous dose of thyroid hormone. This would explain the discrepancy between the two studies, but is speculative, as there have been no studies examining diurnal variation in TSH secretion during combined T₄/T₃ treatment. On the other hand, there is evidence that TSH secretion is effectively regulated by circulating T₄, rather than T₃ (34). The basis for this is the observation (in rats) that brain and pituitary derive a larger proportion of intracellular T₃ from local deiodination of circulating T₄ than do other tissues, in which intracellular T₃ is mainly derived from circulating T₃ (35). If this is true of humans, then it is possible that partial substitution of liothyronine for T₄ maintains euthyroidism in peripheral tissues despite increased TSH secretion. Finally, it is possible that the widely quoted liothyronine to T₄ potency ratio of 5:1 (on a microgram to microgram basis) is simply incorrect, because it is based on early studies using bioassays (30, 31) that predate sensitive TSH assays. Some authorities suggest that the potency ratio is lower, about 3:1 or 4:1 (34). The small, but statistically significant, changes in Zulewski score, pulse rate, cholesterol, and SHBG during combination treatment compared with T₄ alone support this possibility.

Whatever the correct explanation, the increased serum TSH concentrations during combined T₄/T₃ treatment raised the possibility that subjects were relatively underreplaced during this arm of the study, thus masking the beneficial effects of combined treatment on well-being and symptoms of hypothyroidism. The results of the post hoc subgroup analysis make this unlikely, as no benefit of combined T₄/T₃ treatment over T₄ alone was observed in subjects whose serum TSH concentrations were similar during each treatment. Although such an analysis results in a loss of statistical power, one would expect to see a trend toward improved quality of life or cognitive function scores if combined treatment really were preferable to T₄. No such trend was evident. Furthermore, the number of subjects in this subgroup analysis (n = 39) is still larger than that in the study by Bunevicius et al. (10), which reported beneficial effects of combined treatment.

In the present study serum SHBG was lower, and plasma cholesterol higher during combined T₄/T₃ treatment compared with T₄ alone, suggesting a lesser effect of thyroid hormone on the liver. The difference in SHBG remained significant in the subgroup analysis correcting for differences in serum TSH. In contrast, Bunevicius et al. (10) found that SHBG increased during combined T₄/T₃ treatment compared with T₄, whereas cholesterol was unchanged. These discrepancies between the studies are unexplained.

Our results appear inconsistent with studies in rats, in which combined T₄/T₃ treatment was required to achieve tissue euthyroidism. This is readily explained by interspecies differences in thyroid hormone secretion and metabolism. In rats, direct thyroidal secretion accounts for a far higher proportion of total T₃ production than in humans (40% vs. 20%) (35). Accordingly, it may well be necessary to give both T₄ and T₃ to achieve physiological thyroid replacement in rats, but this appears to be unnecessary in humans.

The strengths of our study include its large sample size and crossover design, giving a high degree of statistical power, and the inclusion and categorization of subjects who were satisfied and those who were dissatisfied with T₄ treatment. It could be argued that the inclusion of dissatisfied subjects made the study group less representative of hypothyroid subjects in general, because it is thought that most hypothyroid subjects have a satisfactory symptomatic response to T₄ treatment. It is, however, the suboptimal response to T₄ in some patients that makes combined T₄/T₃ treatment a subject of interest, and so it would have been unreasonable to exclude dissatisfied subjects from the study. The drop-out rate in the study was not excessive, and compliance with treatment was high. A weakness of our study was the use of a fixed quantity of liothyronine as partial substitution for T₄ regardless of baseline T₄ dosage. This meant that during combination treatment, the ratio of T₄/T₃ administered differed between subjects, and differed from the theoretically optimal ratio of approximately 10:1 (36). In addition, liothyronine was administered once daily, whereas for optimal replacement, divided doses or a slow-release preparation would be preferable. However, these limitations also apply to the study by Bunevicius et al. (10), and as our primary intention was to confirm or refute that study, they do not detract from our conclusions. It remains possible that different dosing regimens or routes of administration (such as transdermal preparations) of thyroid hormones would achieve more physiological thyroid replacement and conceivably improve well-being in patients, but this remains to be demonstrated.

The reason why some patients with hypothyroidism experience persistent symptoms of ill health despite apparently adequate T₄ replacement is not known. Possible explanations include incorrect diagnosis, comorbidities, and suboptimal prescription or monitoring of T₄ therapy (2, 4). These are unlikely to account for symptoms in the dissatisfied subjects in our study, because the diagnosis was verified in most cases, patients with major comorbidities (other than treated depression) were excluded, and most patients had serum TSH concentrations in the lower reference range. Undiagnosed depression in dissatisfied patients is a possible explanation; although depression scores on the GHQ-28 subscale for depression were not different between dissatisfied and satisfied subjects, this subscale is designed to detect severe, rather than mild or moderate, depression (16, 27). More detailed clinical studies of unselected patients with hypothyroidism are required to explore this possibility further.

In conclusion, we found no evidence that combined T₄/T₃ replacement (in the dosage regimen used in this study) resulted in improved well-being, cognitive function, quality of life, or increased thyroid hormone action on peripheral tissues compared with T₄ alone. We were unable to confirm the results reported by Bunevicius et al. (10). Unless beneficial effects of combined T₄/T₃ treatment over T₄ alone can be convincingly demonstrated by others, T₄ should remain the standard treatment for hypothyroidism.

Acknowledgments

We thank our colleagues for referring potential participants, Boots Australia for donating liothyronine tablets, Graylands Hospital Pharmacy for preparing study medications, Thyroid Australia and the Australian Thyroid Foundation for publicizing the study, and the participants.

This work was supported by Sir Charles Gairdner Hospital Research Foundation.

Abbreviations:

 
• GHQ-28,

  General Health Questionnaire 28;

 
• SF-36,

  Short Form 36;

 
• TSQ,

  Thyroid Symptom Questionnaire.

References