Until recently, most clinicians would have treated these patients with levothyroxine at thyroid-stimulating hormone (TSH)-suppressing doses because TSH is the major thyroid stimulator for both function and growth [8, 9]. Thus, TSH suppression would be expected to cause either nodule reduction or growth inhibition [10, 11].

In the last 5 years, however, study findings have challenged this policy: Two randomized controlled studies in which ultrasonography was used to evaluate nodule volume changes failed to prove the efficacy of levothyroxine [12, 13]; and recent reports of bone mineral density decreases in patients treated with levothyroxine have raised concern about the risk–benefit ratio of levothyroxine administration [14-16].

General clinical experience, however, and previous open, noncontrolled studies [10, 11] indicate that at least some cold nodules decrease in size when treated with levothyroxine. This controversy suggests the need for further randomized clinical trials of levothyroxine treatment. Because potassium iodide, either alone or in combination with levothyroxine, is used outside the United States for treating diffuse and nodular goiter caused by iodine deficiency [17-19] and for treating sporadic goiter [20], we tested iodide treatment in patients with solitary cold thyroid nodules.

Methods

Top Methods Results Discussion Author & Article Info References

Patients and Study Design

We enrolled euthyroid patients with benign solitary solid cold nodules of the thyroid who were referred to our thyroid clinic. We first selected patients who had a solitary thyroid nodule and no major concomitant disease at clinical examination. Radioiodine scanning and ultrasound examination of the thyroid (Diasonics DRF 250 ultrasound scanner equipped with a linear 10-MHz probe; Sonotron SA, Les Ulis Cedex, France) and fine-needle aspiration biopsy of the nodule were then done. Because of the probe used, the accuracy of measures by ultrasonography could be obtained only for nodules with a diameter of 3.5 cm or less; we therefore excluded patients with larger nodules. We also measured the following in all patients: levels of serum thyroxine, total triiodothyronine, free triiodothyronine, and free thyroxine sub (commercial radioimmunoassay methods); TSH and thyroglobulin levels (immunoradiometric methods); antithyroglobulin and antimicrosomal antibodies (hemoagglutination method); and urinary iodine levels (colorimetric method). We included only newly diagnosed nodules (recognized less than 1 year before the study began).

Using ultrasonography, we measured nodule size in three planes and recorded it on film. Nodule volume was calculated according to the following formula for a spherical ellipsoid: volume = (/6) x AP x width x length, where AP is the anteroposterior diameter [21]. For each patient, ultrasound measurements were done by the same operator, who had no access to the patient's clinical and laboratory data or group assignment. Intraobserver variation of nodule measurement was assessed before the study began by a pilot investigation of 25 patients with a thyroid nodule. In these patients, nodules were measured twice at baseline and after 20 to 30 days, resulting in a weighted statistic of 99.9% [22].

We selected only solid nodules, including nodules with absent or minimal (<10%) cystic component. In addition, we carefully examined both thyroid lobes for the presence of additional nodules. Patients with a second thyroid nodule not evident at physical examination were included if the maximum diameter of the second nodule did not exceed 50% of the maximum diameter of the main nodule. We also measured contralateral thyroid lobe diameters and recorded the data [23].

A radioiodine scan (25 microcuries) was done in all patients to exclude all "hot" nodules, which are usually autonomous and unresponsive to medical treatment [24]. A fine-needle aspiration biopsy examination was then done as previously described [3]. Patients with a cytologic diagnosis of malignancy, follicular lesion, cyst-hemorrhagic lesion, or thyroiditis were excluded; we selected only colloid-parenchymatous nodules that showed colloid and benign follicular cells in variable proportion. In addition, we excluded patients with abnormal thyroid hormone or TSH serum levels, circulating thyroid antibodies, or concomitant cardiovascular or liver diseases, osteoporosis, or pregnancy. All patients lived in an area with a sufficient iodine supply and a goiter prevalence in schoolchildren of less than 1%. Urinary iodine excretion in this area ranges from 80 to 300 µg/d. However, because iodine intake may affect nodule growth and response to therapy, the urinary iodine concentration of our patients was measured in a urine sample taken in the morning. Urinary iodine excretion less than 8.0 µg/dL or exceeding 27.0 µg/dL was a criterion for exclusion.

Patients selected by these procedures gave informed consent and entered the study. We then randomly assigned patients to receive one of three treatments [using randomized blocks with a coin slightly biased in favor of treatment groups and with allocation blinded only to the ultrasonography operator]: 1) no treatment; 2) oral levothyroxine at an initial dose of 1.0 µg/kg body weight per day to be taken in one single dose in the morning; and 3) oral iodine supplementation at 1.5 mg every 2 weeks, provided as potassium iodide tablets. For patients in the second group, the dose was increased to 1.8 µg/kg per day after 15 days and then individually adjusted after the first 4 months to allow a serum TSH level less than normal values (<0.3 mU/L). At that point, the average levothyroxine dosage was 1.94 ±0.16 µg/kg per day.

Each treatment was continued for 12 months. Patients were evaluated at 4-month intervals by clinical and ultrasound examination. Levels of free triiodothyronine, free thyroxine, TSH, thyroglobulin, urinary iodine excretion, and antithyroid antibodies were also measured every 4 months.

Compliance with therapy was individually controlled in patients receiving levothyroxine by carefully asking the patient and by measuring TSH serum levels at 4-month intervals. Urinary iodine was also measured in all patients at 4-month intervals to check both treatment compliance (in patients receiving potassium iodide) and the absence of iodine contamination in patients receiving no treatment and in those receiving levothyroxine.

End Points

We considered two end points: 1) the type of nodule volume variation from 0 to 12 months and 2) a nodule reduction of 50% of the initial volume after the 12-month observation period.

Sample Size

According to a previous report [12], a spontaneous decrease of 50% or more of the initial volume could be assumed to occur in 20% of patients. We therefore considered therapy successful when it caused a similar ( 50%) volume reduction at least 2.5 times more frequently than occurred with no treatment. Given a type I error of 0.05 (two-sided) and a power of 0.9 and considering that approximately 10% of patients might be lost to follow-up, we estimated the necessary total sample size to be 160 patients.

Statistical Analysis

We planned an interim analysis of the results after 12 months of observation for the first half of the patients who entered the study. Because nodule volumes were distributed in a skewed manner, appropriate transformation such as logarithms were applied to these data for statistical inference, and proper means (that is, geometric means) were used to describe results. We evaluated the statistical significance of nodule size variation in the three groups by repeated-measures analysis [25] and considered data obtained at months 0, 4, 8, and 12 from patients who completed 1 year of follow-up. We then fitted a linear random-effects model with a structured variance-covariance matrix to each log_e-volume profile at 0, 4, 8, and 12 months of observation and considered such covariates as group assignment (no treatment, levothyroxine, or potassium iodide), patient age (in years), cytologic findings (colloid or parenchymatous nodular hyperplasia), and echographic patterns (hypoechogenic, isoechogenic, hyperechogenic, or mixed nodules). We subsequently fitted a second model to the same data, also taking into account the nodule volume class ( 5 mL, 5.1 to 10 mL, or >10 mL). We used the BMDP statistical package (Statistical Software, Inc., Los Angeles, California).

We used a log-linear model [26] to test differences in the proportion of volume reduction by the GLIM statistical package (Royal Statistical Society, London, United Kingdom). A paired t-test was used for comparisons between values before and after 12 months of therapy. Every test of hypothesis was done at a 0.05 level of significance (two-sided).

Results

Top Methods Results Discussion Author & Article Info References

The study was stopped because at the interim analysis, we obtained clinically important results for the first 80 patients who entered the study; we did not include 18 patients who were still being studied at the time of interim analysis.

After randomization, the characteristics of these 80 patients and their nodules were homogeneous in the three groups, and we saw no statistical differences (Table 1). Twelve nodules evenly distributed among the three groups had a cystic component of less than 10%. Of the 80 patients, 70 (87.5%) completed the assigned 12 months of follow-up. Three patients not receiving treatment dropped out (1 patient moved and 2 missed follow-up), as did 4 patients receiving levothyroxine (1 patient decided to have thyroidectomy, 2 refused further treatment, and 1 missed follow-up) and 3 patients receiving iodide (2 patients refused further treatment and 1 missed follow-up). Characteristics of these patients and their nodules were equivalent to those of patients who completed the study (mean age, 37.6 ±12.5 years compared with 39 ±11.0 years; mean nodule volume, 5.2 ±5.8 mL compared with 5.5 ±5.8 mL; mean TSH level, 1.4 ±0.3 mU/L compared with 1.2 ±0.6 mU/L).

Compliance with treatment was good in all studied patients as judged by patient self-report and by laboratory evaluation: In all patients receiving levothyroxine, the serum TSH level was less than 0.3 mU/L, and average thyroid hormone levels were slightly increased but were always within the normal range (Table 2). The mean size of the contralateral thyroid lobe decreased (P < 0.001) (Table 2). Adherence to the protocol in patients who received potassium iodide and those who received no treatment was confirmed because serum TSH and thyroid hormone levels did not change (Tables 3 and 4). In patients treated with iodide, the average urinary iodine excretion increased from 13.8 to 20.2 µg/dL (P = 0.002). No patient in either of the treated groups developed clinical signs or symptoms of thyrotoxicosis or other adverse reactions to treatment.

Nodule Volume Evolution and Response to Treatment

To compare our data with those of previous studies, we first calculated the arithmetic mean of nodule volume. We then used the more appropriate geometric mean of nodule volume for subsequent analysis. At the end of the 12-month period, the geometric mean nodule volume was reduced by 40% of the initial value (P < 0.001) in the patients receiving levothyroxine Table 2 and by 23% (P = 0.053) in the patients receiving potassium iodide (Table 3). In contrast, mean nodule volume increased by 11% (P = 0.085) in the untreated patients (Table 4).

As shown in Table 5, we observed a clinically relevant reduction in nodule volume ( 50%) in 9 of 23 (39%) patients receiving levothyroxine (mean volume reduction in these 9 patients, 62%; range, 50% to 80%), in 5 of 25 (20%) patients receiving potassium iodide (mean volume reduction in these 5 patients, 75%; range, 60% to 90%), and in none of 22 untreated patients. Overall comparison among the three groups within the same log-linear model gave a P value of 0.004. According to the model, there was no difference in odds ratio for nodule volume reduction between the levothyroxine and potassium iodide groups (P = 0.156), whereas receiving no treatment reduced the same odds ratio by a factor of 26 (P = 0.026). Most nodules that were clinically significantly reduced had a volume of 5 mL or less. A clinically relevant volume increase did not occur in the two treated groups but did occur in 3 of 22 patients (13.6%) in the untreated group (increases of 53%, 66%, and 118%, respectively) (Table 5).

A second fine-needle aspiration biopsy was advised for all patients whose nodule had not been substantially reduced; biopsy was actually done in 10 of 14 patients in the levothyroxine group, 14 of 20 patients in the potassium iodide group, and 18 of 22 patients in the untreated group. In all cases, the original diagnosis was confirmed. In addition, the diagnosis was confirmed through histologic examination in 9 of the patients who chose to have surgery.

To analyze the temporal pattern of nodule variation, we applied repeated-measures analysis to the geometric means of nodule volumes at months 0, 4, 8, and 12. The results indicated a progressive reduction in nodule volume in both the levothyroxine (P < 0.001) and potassium iodide groups (P = 0.015) but not in the untreated group (P > 0.2). Thus, both levothyroxine and potassium iodide effectively reduced nodule volume compared with no treatment (P < 0.001 and P = 0.028, respectively). We observed no difference (P = 0.097) in nodule volume reduction between the levothyroxine and potassium iodide groups (Figure 1). Covariates such as cytologic diagnosis, echographic patterns, and patient age did not significantly affect nodule volume variations. We applied the same analysis to data stratified according to the class of initial nodule volume ( 5 mL, 5.1 to 10 mL, or >10 mL) (Figure 2). We observed a progressive volume reduction only in nodules with volumes of 5 mL or less (P < 0.001) and 5.1 to 10 mL (P = 0.002) in patients receiving levothyroxine and in nodules with a volume of 5 mL or less in patients receiving potassium iodide (P = 0.003). In the untreated patients, nodule volume did not change in any of the initial volume classes (P > 0.25). In nodules with a volume less than 5.0 mL, volume reduction was almost twofold greater in the levothyroxine group than in the potassium iodide group (P = 0.046).

View larger version (14K): [in this window] [in a new window]   Figure 1. Geometric means (loge scale) of observed nodule volumes (symbols) at months 0, 4, 8, and 12 in patients receiving levothyroxine (l-thyroxine), potassium iodide, or no treatment. Expected values (obtained by fitting a linear random-effects model to the data) are indicated by lines. Estimated slopes of the linear regression equations with 95% confidence intervals were as follows: patients receiving levothyroxine, –0.04(CI, –0.0567 to –0.0233;P < 0.001); patients receiving potassium iodide, –0.02(CI, –0.0361 to –0.0039;P = 0.015); patients receiving no treatment, 0.0061 (CI –0.0110 to 0.0232; P > 0.2). Comparisons among slopes were as follows: levothyroxine group compared with untreated group, P < 0.001; potassium iodide group compared with untreated group, P = 0.028; levothyroxine group compared with potassium iodide group, P = 0.097. Slopes reported refer to loge nodule volume variation for every unit-time (4 months) through the 12-month follow-up.

View larger version (21K): [in this window] [in a new window]   Figure 2. Geometric means (loge scale) of observed (symbols) and expected (lines) nodule volumes in patients receiving levothyroxine [l-thyroxine], potassium iodide, or no treatment according to the size of nodules. Estimated slopes of the linear regression equations and 95% confidence intervals were as follows: 1) levothyroxine group: nodule volume of 5 mL or less, –0.049(CI, –0.067 to –0.031;P < 0.001); nodule volume of 5 to 10 mL, –0.035(CI, –0.057 to –0.013;P < 0.002); nodule volume of 10 mL or greater, –0.023[CI, –0.050 to 0.004; P = 0.091]; 2) potassium iodide group: nodule volume of 5 mL or less, –0.026(CI, –0.044 to –0.008;P < 0.003); nodule volume of 5 to 10 mL, –0.012(CI, –0.034 to 0.010; P = 0.276); nodule volume of 10 mL or greater, –0.0002[CI, –0.028 to 0.027; P > 0.2]; 3) untreated group: nodule volume of 5 mL or greater, –0.002(CI –0.020 to 0.016; P > 0.2); nodule volume of 5 to 10 mL, 0.013 (CI, –0.012 to 0.038; P > 0.2); nodule volume of 10 mL or greater, 0.024 (CI –0.017 to 0.065; P > 0.2).

Seven patients (two receiving no treatment, three receiving levothyroxine, and two receiving potassium iodide) had a second small nonpalpable thyroid nodule detected by ultrasonography at first observation. The mean volume of these nodules was 0.18 ±0.09 mL (range, 0.06 to 0.28 mL; median, 0.23 mL). After 1 year, one of these small, nonpalpable nodules disappeared in the levothyroxine group, two disappeared in the potassium iodide group, and one disappeared in the untreated group. During the study period, two nonpalpable new nodules appeared in two patients receiving iodide and one patient receiving no treatment.

Patient Follow-up after Termination of the First-year Study

Patients who received 1 year of therapy with either levothyroxine or potassium iodide were asked to discontinue therapy for 4 months so they could be re-evaluated after that period. Twenty of the 23 patients initially treated with levothyroxine and 24 of the 25 initially treated with potassium iodide completed this period of treatment withdrawal. In the 20 patients initially treated with levothyroxine, the geometric mean nodule volume had decreased from 4.2 to 2.3 mL; after levothyroxine was withdrawn, mean volume increased to 2.9 mL. We observed a clinically relevant volume increase in five nodules. In the 24 patients initially treated with potassium iodide, mean nodule volume had decreased from 2.2 to 1.6 mL; it increased to 1.8 mL after iodide was withdrawn. We observed a clinically relevant volume increase in six nodules.

To evaluate whether increases in nodule volume after discontinuation of therapy were related to the effectiveness of the previous medical treatment, we subdivided patients according to whether their nodule volume had or had not decreased during the 1 year of medical treatment. Analysis of data (not shown) indicated that nodule responsiveness to treatment is not predictive of nodule increase after treatment is discontinued.

After termination of the 1-year study, the first 13 patients who received no treatment were switched to the levothyroxine treatment group and entered a second year of follow-up. Levothyroxine dosage and patient follow-up were the same as those previously described for the patients receiving levothyroxine.

Data became available for 11 patients. While patients were not receiving treatment, the geometric mean nodule volume in these patients had increased from 3.6 to 5.4 mL; after 1 year of levothyroxine treatment, it decreased from 5.4 to 3.8 mL (30% reduction [95% CI, 40.3% to 18.6%; P = 0.004]). In 2 of these 11 patients, the reduction in nodule volume was clinically relevant ( 50%). The effect of levothyroxine in these 11 patients did not substantially differ from that observed in patients initially treated with levothyroxine.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study shows that both levothyroxine and potassium iodide were effective in arresting the growth or in reducing the volume of cold thyroid nodules. Mean nodule volume decreased in patients treated for 1 year with either levothyroxine or (to a lesser extent) with potassium iodide, whereas it slightly increased in the untreated patients. We observed a clinically relevant reduction in nodule volume ( 50%) in 39% and 20% of the patients receiving levothyroxine or potassium, respectively, but in none of the untreated patients. In addition, nodule volume did not increase in treated patients but did increase in 14% of untreated patients. When untreated patients whose nodules showed no reduction after 1 year received levothyroxine, the mean nodule volume decreased. Finally, 4 months after discontinuing the therapy that had reduced nodule volume, 25% (11 of 44) of patients previously treated with either levothyroxine or potassium iodide had a clinically significant increase in nodule volume.

Our findings are remarkably similar to those reported by Morita and colleagues [27], who measured nodule volume by ultrasonography after 3 months of treatment with levothyroxine at a dose of 100 µg/d. In addition, a statistically significant nodule reduction during levothyroxine treatment has also been recently reported by Papini and colleagues [28] ( 50% volume reduction in 19% of treated patients and in 6% of controls).

What accounts for the differences between our findings and those of two other controlled studies [12, 13]? In the study by Gharib and colleagues [12], 19% of selected nodules were predominantly cystic (we excluded such nodules from our study), and 22% were functional rather than "cold". Thus, as others have pointed out, they may have been autonomous nodules and thus unresponsive to levothyroxine [29]. The fact that we selected only newly diagnosed nodules and the shorter duration (6 months) of the study by Gharib and colleagues [12] may also have contributed to the difference in results. It is also possible that in geographic areas with different iodine intakes, thyroid nodules differ in responsiveness to levothyroxine therapy. Gharib and colleagues [12] did not measure their patients' urinary iodine excretion, but iodine intake in the Rochester, Minnesota area, is approximately threefold higher than in our area [30].

Reverter and associates [13] also used ultrasound measurement and reported that levothyroxine therapy was ineffective. Their results may have been biased by a comparison of nonhomogeneous groups: Six patients in the treatment group and none in the control group dropped out. A recalculation of their results suggests that 29% (4 of 14) of patients who completed the levothyroxine treatment period and only 15% (3 of 20) of untreated patients had a 50% or greater decrease in nodule volume.

An important finding of our study was that levothyroxine was effective at a dosage lower than that reported by others [12, 13, 28, 31]. Thyroid-stimulating hormone suppression at values less than 0.3 mU/L but not in the hyperthyroid range was sufficient to reduce nodule volume in approximately 40% of patients. This is important given the increasing concern about the possible adverse effects, especially osteoporosis promotion, of long-term, high-dose levothyroxine therapy. We did not investigate, however, whether higher levothyroxine dosages might produce a greater or more frequent decrease in nodule volume.

Low-dose, discontinuous potassium iodide administration, although less effective than levothyroxine, also reduced nodule volume in 20% of cases. Iodine affects various metabolic variables in thyroid cells, including reduction of the intracellular adenosine 3'5'-cyclic monophosphate accumulation and glucose uptake after exposure to TSH [32, 33]. In addition, iodine antagonizes the mitogenic response of thyroid cells to insulin and insulin-like growth factor type I [34]. Synergistic and alternative mechanisms may also cause the action of levothyroxine and iodide. Consequently, different nodules may respond differently to either levothyroxine or potassium iodide, or both, a possibility we did not explore. However, because long-term therapy with low-dose iodine supplementation is known to be safe and free of the side effects of suppressive thyroxine therapy [35], it represents a potential alternative treatment for patients in whom levothyroxine is contraindicated.

The volume of small nodules was reduced most frequently with either levothyroxine or potassium iodide treatments (Table 5 and Figure 2). None of the nodules whose volume was greater than 10 mL showed a volume decrease of 50% or more. However, if treatment prevented further growth in some nodules, this does not imply that these nodules were nonresponsive. In fact, none of the eight such nodules treated with either levothyroxine or potassium iodide significantly increased in volume, whereas one of the four nodules in the untreated patients increased by more than 50% (Table 5). Apart from initial nodule volume, no other patient or nodule characteristic predicted nodule shrinkage during treatment.

After therapy with either levothyroxine or potassium iodide was discontinued for 4 months, we observed a significant nodule enlargement ( 50%) in approximately 25% of patients. The natural history of cold thyroid nodules is not well defined, and different subsets of solid cold nodules may be recognized because follicular cells are heterogeneous [36], and their growth is influenced by various local and systemic growth factors. This may explain why medical responsiveness to treatment differs and why some but not all nodules resume growing after treatment is discontinued.

On the basis of our study and others [1-7, 28-31], we propose the following guidelines for the medical management of solid solitary cold thyroid nodules that do not exceed 3.5 cm in diameter. Fine-needle aspiration biopsy should be done first, and, in the case of colloid-parenchymatous benign nodules, levothyroxine administration should be regarded as the first-choice therapy provided that no contraindication to levothyroxine is present. Treatment should keep serum TSH levels to less than 0.3 mU/L but not in the frank hyperthyroid range (<0.03 to <0.006 mU/L depending on the assay method) [37]. Nodules with volumes greater than 10 mL (>2.7 cm in diameter) will probably not shrink and could be followed expectantly and treated if they grow further.

The following responses to levothyroxine treatment may occur.

1. Nodule volume decreases: Nodule shrinkage is usually evident after 4 to 8 months. Levothyroxine treatment may be continued until volume no longer decreases. Therapy may then be discontinued and reinitiated if nodules resume growing.

2. Nodule volume remains stationary: These nodules should not necessarily be considered nonresponders because levothyroxine may have prevented further nodule growth. However, treatment should be discontinued and resumed if nodules grow.

3. Nodule volume increases: Because non-neoplastic nodules treated with suppressive levothyroxine are unlikely to grow, an increase in volume strongly indicates surgery.

Administration of discontinuous doses of potassium iodide may be useful in patients with small nodules and contraindications to levothyroxine. Different iodide dosages or association with levothyroxine at varying doses may also be effective, and their efficacy should be tested in future studies.