Evaluating the Impact of Correcting for Endogenous T_{4} Baseline on the Bioequivalence of Levothyroxine Sodium Formulations in Healthy Volunteers
The objective of this study was to evaluate the impact of various methods for correcting for endogenous T_{4} baseline on the bioequivalence of levothyroxine sodium formulations in healthy volunteers.
This Phase 1, singledose, openlabel, study was conducted according to a threeperiod, randomized crossover design in healthy volunteers. The total dose given was 600 µg levothyroxine sodium for Regimen A, 450 µg levothyroxine sodium for Regimen B and 400 µg levothyroxine sodium for Regimen C. Subjects received one of six sequences of Regimen A (twelve 50 µg Synthroid^{®} tablets), Regimen B (nine 50 µg Synthroid^{®} tablets) or Regimen C (eight 50 µg Synthroid^{®} tablets) under fasting conditions at approximately 0830 on Study Day 1 of each period. A washout interval of at least 44 days separated the doses of the three study periods.
Blood samples (sufficient to provide approximately 2 mL serum) for total levothyroxine (T_{4}), total triiodothyronine (T_{3}) and thyroid stimulating hormone (TSH) assay were collected by venipuncture into 5 mL evacuated siliconized collection tubes as follows:
At approximately 0 hours and at 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 10, 12 and 18 hours after the 0hour collection on Study Day –1 in each study period.
At approximately –30 minutes, –15 minutes and at 0 hours prior to dosing and at 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 10, 12, 18, 24, 36, 48, 72 and 96 hours after dosing on Study Day 1 in each study period.
Serum concentrations of T_{4} and T_{3} were determined using validated radioimmunoassay (RIA) methods. The lower limit of quantification of T_{4} was 1.00 µg/dL. The lower limit of quantification of T_{3} was 0.25 ng/mL. Serum concentrations of TSH were determined using a validated IRMA assay; lower limit of quantification was 0.250 µIU/mL.
Subjects were male
and female volunteers between 19 and 50 years of age, inclusive. Subjects were judged to be euthyroid and in
general good health based on the results of medical history, physical
examination, vital signs, 12lead electrocardiogram and laboratory tests. Females were postmenopausal, sterile, or if
of childbearing potential, were not pregnant or breastfeeding and were
practicing an acceptable method of birth control.
Thirtysix subjects (18 M, 18 F) participated in the study, with mean age of 32.9 years, mean weight of 74.5 kg and mean height of 172 cm. Three subjects received study drug in only one period and thus were not included in any of the pharmacokinetics analyses. Thirtythree subjects (16 M, 17 F) were included in the pharmacokinetic analyses, with mean age of 33.1 years, mean weight of 73.5 kg and mean height of 171 cm.
The pharmacokinetic parameters of total levothyroxine (T_{4}) were estimated using noncompartmental methods. These included: the maximum serum concentration (C_{max}) and time to C_{max} (T_{max}), the area under the serum concentrationtime curve (AUC) from time 0 to 48 hours (AUC_{48}), time 0 to 72 hours (AUC_{72}) and time 0 to 96 hours (AUC_{96}). For T_{4}, values of these parameters (C_{max}, T_{max}, AUC_{48}, AUC_{72} and AUC_{96}) were determined without correction for endogenous T_{4} levels and after correcting all postdose concentrations using each of following three methods:
Correction Method 1: The predose baseline value on the day of dosing was subtracted from each postdose concentration. The predose baseline value was calculated as the average of the three concentrations at –0.5, –0.25 and 0 hours prior to dosing in each period.
Correction Method 2: For each time of postdose sampling, the observed concentration was corrected assuming that the endogenous T_{4} baseline level at 0 hours declines according to a halflife of 7 days.
Correction Method 3: The T_{4} concentration for each time of postdose sampling was corrected by the concentration observed at the same time of day during the 24 hours preceding the dose.
For all three methods of correction, the corrected 0hour concentration was assumed to be 0.
For uncorrected and corrected T_{4} an analysis of variance (ANOVA) with fixed effects for sex, sequence, sexbysequence interaction, period, regimen and the interaction of sex with each of period and regimen, and with random effects for subjects nested within sexbysequence combination was performed for T_{max}, and the natural logarithms of C_{max} AUC_{48}, AUC_{72} and AUC_{96}. A significance level of 0.05 was used for all tests.
The bioavailability of each of Regimen B (450 µg dose) and Regimen C (400 µg dose) relative to that of Regimen A (600 µg dose) for uncorrected and corrected T_{4} was assessed by the two onesided tests procedure^{1} via 90% confidence intervals obtained from the analysis of the natural logarithms of AUC_{48} and C_{max}. Bioequivalence was concluded if the 90% confidence intervals from the analyses of the natural logarithms of AUC_{48} and C_{max} were within the 0.80 to 1.25 range. Likewise, the bioavailability of Regimen B (450 µg dose) relative to that of Regimen C (400 µg dose) was assessed. The same was done using each of AUC_{72} and AUC_{96} in place of AUC_{48}.
A repeated measures analysis was performed on the T_{4} concentration data of Study
Day –1 for each period. To investigate
the possibility of carryover effects, an ANOVA was performed on the logarithms
of the Study Day –1 AUC_{24}.
The mean serum concentrationtime plots for uncorrected T_{4} after administration of levothyroxine sodium on Study Day 1 are presented in Figure 1. The mean T_{4} serum concentrationstime profiles are fairly consistent after administration of the three regimens. Mean T_{4} concentrations prior to dosing are approximately 7.5 µg/dL and increase to about 13 to 14 µg/dL at maximum before declining. The mean T_{4} concentrations remain at approximately 9 µg/dL at 96 hours after administration of these large doses of levothyroxine sodium to the healthy volunteers.
Figure 1. Mean Levothyroxine (T_{4}) ConcentrationTime Profiles on Study Day 1 Following Single Dose Administration of Levothyroxine Sodium
– Uncorrected for Endogenous T_{4} Baseline Concentrations
Mean ± standard deviation (SD) pharmacokinetic parameters of T_{4} after administration of the three regimens without correcting for endogenous T_{4} baseline concentrations are listed in Table 1.
Table 1. Mean ± SD Pharmacokinetic
Parameters of Levothyroxine (T_{4}) Without Correcting for Endogenous T_{4} Baseline Concentrations 


Regimens 

Pharmacokinetic Parameters (units) 
A: 600 µg Dose 
B: 450 µg Dose 
C: 400 µg Dose 

T_{max} 
(h) 
3.1 ± 2.4 
3.2 ± 2.1 
3.5 ± 3.3 
C_{max} 
(µg/dL) 
14.3 ± 2.14 
13.2 ±
2.05^{*} 
13.2 ± 2.45^{*} 
AUC_{48} 
(µg•h/dL) 
518 ± 71.8 
493 ± 72.7^{*} 
484 ± 73.6^{*} 
AUC_{72} 
(µg•h/dL) 
741 ± 102 
712 ± 108^{*} 
691 ± 102^{*,+} 
AUC_{96} 
(µg•h/dL) 
951 ± 133 
919 ± 139 
892 ± 133^{*,+} 
* Statistically significantly different from Regimen A (ANOVA, p < 0.05). + Statistically
significantly different from Regimen B (ANOVA, p < 0.05). 
The bioequivalence/bioavailability results for uncorrected T_{4} are listed in Table 2.
Table 2. Bioequivalence and
Relative Bioavailability–Uncorrected Levothyroxine (T_{4}) 

Regimens 



Relative Bioavailability 

Test vs. 
Pharmacokinetic 
Central
Value^{*} 
Point 
90% Confidence 

Reference 
Parameter 
Test 
Reference 
Estimate^{+} 
Interval 
450 µg vs.600 µg 
C_{max} 
13.0 
14.0 
0.928 
0.890 – 0.968 

AUC_{48} 
481.7 
504.8 
0.954 
0.927 – 0.982 

AUC_{72} 
694.9 
721.9 
0.963 
0.936 – 0.990 

AUC_{96} 
896.2 
925.6 
0.968 
0.941 – 0.996 
400 µg vs. 600 µg 
C_{max} 
12.9 
14.0 
0.921 
0.883 – 0.960 

AUC_{48} 
469.6 
504.8 
0.930 
0.904 – 0.958 

AUC_{72} 
670.4 
721.9 
0.929 
0.903 – 0.955 

AUC_{96} 
865.7 
925.6 
0.935 
0.909 – 0.962 
450 µg vs. 400 µg 
C_{max} 
13.0 
12.9 
1.007 
0.967 – 1.050 

AUC_{48} 
481.7 
469.6 
1.026 
0.997 – 1.055 

AUC_{72} 
694.9 
670.4 
1.037 
1.009 – 1.065 

AUC_{96} 
896.2 
865.7 
1.035 
1.007 – 1.064 
* Antilogarithm of the least squares means for logarithms. + Antilogarithm
of the difference (test minus reference) of the least squares means for
logarithms. 
The mean serum concentrationtime plots for T_{4}, after correction for endogenous baseline levels of levothyroxine using each of the correction methods, are presented in Figure 2 for Correction Method 1, Figure 3 for Correction Method 2, and Figure 4 for Correction Method 3. The mean T_{4} serum concentrations after correcting for endogenous baseline levels by any of the three methods of correction were higher after administration of Regimen A (600 µg dose) than after administration of Regimens B (450 µg dose) and C (400 µg dose) throughout the 96hour sampling period. The mean baseline corrected T_{4} concentrations for Regimens B (450 µg dose) and C (400 µg dose) were comparable throughout the 96hour sampling period. The baseline corrected T_{4} concentrations prior to dosing were assigned a value of zero for each of the three methods of correction. However, 96 hours after administration of these large doses of levothyroxine sodium to healthy volunteers the mean baseline corrected T_{4} concentrations remain at approximately 1 to 2 µg/dL for Correction Methods 1 and 3 and approximately 3 to 4 µg/dL for Correction Method 2.
Figure 2. Mean Levothyroxine (T_{4}) ConcentrationTime
Profiles after Correction for Endogenous Baseline Levels of T_{4} Using
Correction Method 1
Figure 3. Mean Levothyroxine (T_{4}) ConcentrationTime
Profiles after Correction for Endogenous Baseline Levels of T_{4} Using Correction Method 2
Figure 4. Mean Levothyroxine (T_{4}) ConcentrationTime Profiles
after Correction for Endogenous Baseline Levels of T_{4} Using Correction Method 3
Mean ± SD pharmacokinetic parameters of T_{4} after administration of the three regimens after correcting for endogenous T_{4} baseline concentrations are listed in Table 3.
Table 3. Mean ± SD Pharmacokinetic
Parameters of Levothyroxine (T_{4}) after Correcting for
Endogenous T_{4} Baseline Concentrations 


Regimens 

Pharmacokinetic Parameters (units) 
A: 600 µg Dose 
B: 450 µg Dose 
C: 400 µg Dose 

Correction Method 1 

T_{max} 
(h) 
3.1 ± 2.4 
3.2 ± 2.1 
3.5 ± 3.3 
C_{max} 
(µg/dL) 
7.05 ± 1.66 
5.54 ± 1.53^{*} 
5.72 ± 1.44^{*} 
AUC_{48} 
(µg•h/dL) 
172 ± 40.4 
126 ± 39.0^{*} 
123 ± 45.4^{*} 
AUC_{72} 
(µg•h/dL) 
222 ± 56.0 
161 ± 55.5^{*} 
149 ± 68.6^{*} 
AUC_{96} 
(µg•h/dL) 
259 ± 72.5 
184 ± 69.9^{*} 
169 ± 92.5^{*} 
Correction Method 2 

T_{max} 
(h) 
3.3 ± 2.8 
5.8 ± 9.3 
3.7 ± 3.5 
C_{max} 
(µg/dL) 
7.15 ± 1.64 
5.68 ± 1.50^{*} 
5.83 ± 1.45^{*} 
AUC_{48} 
(µg•h/dL) 
204 ± 40.9 
160 ± 40.1^{*} 
156 ± 43.4^{*} 
AUC_{72} 
(µg•h/dL) 
292 ± 56.9 
235 ± 58.2^{*} 
221 ± 62.7^{*} 
AUC_{96} 
(µg•h/dL) 
379 ± 74.0 
312 ± 74.6^{*} 
295 ± 82.2^{*} 
Correction Method 3 

T_{max} 
(h) 
3.5 ± 3.1 
3.6 ± 2.3 
3.6 ± 4.0 
C_{max} 
(µg/dL) 
7.03 ± 1.64 
5.85 ± 1.78^{*} 
5.56 ± 1.69^{*} 
AUC_{48} 
(µg•h/dL) 
176 ± 36.9 
131 ± 39.2^{*} 
120 ± 28.4^{*} 
AUC_{72} 
(µg•h/dL) 
226 ± 49.4 
166 ± 52.9^{*} 
146 ± 45.4^{*,+} 
AUC_{96} 
(µg•h/dL) 
263 ± 64.8 
189 ± 65.6^{*} 
167 ± 67.2^{*} 
* Statistically significantly different from Regimen A (ANOVA, p < 0.05). + Statistically significantly different from Regimen B (ANOVA, p
< 0.05). 
The bioequivalence/bioavailability results for T_{4} using Correction Method 1, Correction Method 2, and Correction Method 3 are listed in Tables 4, 5, and 6, respectively.
Table 4. Bioequivalence and Relative
Bioavailability for T_{4} (Correction Method 1) 

Regimens 



Relative Bioavailability 

Test vs. 
Pharmacokinetic 
Central Value^{*} 
Point 
90% Confidence 

Reference 
Parameter 
Test 
Reference 
Estimate^{+} 
Interval 

450 µg vs.600 µg 
C_{max} 
5.4 
6.9 
0.783 
0.727 – 0.844 


AUC_{48} 
119.7 
167.3 
0.715 
0.658 – 0.778 


AUC_{72} 
151.4 
215.7 
0.702 
0.636 – 0.774 


AUC_{96} 
170.2 
250.2 
0.680 
0.602 – 0.768 

400 µg vs. 600 µg 
C_{max} 
5.6 
6.9 
0.803 
0.745 – 0.865 


AUC_{48} 
118.9 
167.3 
0.711 
0.653 – 0.773 


AUC_{72} 
144.9 
215.7 
0.672 
0.609 – 0.741 


AUC_{96} 
165.1 
250.2 
0.660 
0.584 – 0.746 

450 µg vs. 400 µg 
C_{max} 
5.4 
5.6 
0.975 
0.906 – 1.049 


AUC_{48} 
119.7 
118.9 
1.007 
0.926 – 1.094 


AUC_{72} 
151.4 
144.9 
1.044 
0.948 – 1.150 


AUC_{96} 
170.2 
165.1 
1.031 
0.914 – 1.163 

* Antilogarithm of the least squares means for logarithms. + Antilogarithm of the difference (test minus reference) of the
least squares means for logarithms. 

Table 5. Bioequivalence and Relative
Bioavailability for T_{4} (Correction Method 2) 

Regimens 



Relative Bioavailability 

Test vs. 
Pharmacokinetic 
Central
Value^{*} 
Point 
90% Confidence 

Reference 
Parameter 
Test 
Reference 
Estimate^{+} 
Interval 

450 µg vs.600 µg 
C_{max} 
5.6 
7.0 
0.793 
0.739 – 0.850 


AUC_{48} 
154.5 
199.1 
0.776 
0.721 – 0.835 


AUC_{72} 
227.5 
284.9 
0.799 
0.729 – 0.875 


AUC_{96} 
301.6 
369.5 
0.816 
0.743 – 0.897 

400 µg vs. 600 µg 
C_{max} 
5.7 
7.0 
0.807 
0.753 – 0.866 


AUC_{48} 
148.4 
199.1 
0.745 
0.693 – 0.802 


AUC_{72} 
207.9 
284.9 
0.730 
0.666 – 0.800 


AUC_{96} 
277.3 
369.5 
0.750 
0.683 – 0.824 

450 µg vs. 400 µg 
C_{max} 
5.6 
5.7 
0.982 
0.916 – 1.051 


AUC_{48} 
154.5 
148.4 
1.041 
0.969 – 1.119 


AUC_{72} 
227.5 
207.9 
1.094 
1.001 – 1.197 


AUC_{96} 
301.6 
277.3 
1.088 
0.992 – 1.192 

* Antilogarithm of the least squares means for logarithms. + Antilogarithm of the difference (test minus
reference) of the least squares means for logarithms. 

Table 6. Bioequivalence and Relative
Bioavailability for T_{4} (Correction Method 3) 


Regimens 



Relative Bioavailability 


Test vs. 
Pharmacokinetic 
Central Value^{*} 
Point 
90% Confidence 


Reference 
Parameter 
Test 
Reference 
Estimate^{+} 
Interval 


450 µg vs.600 µg 
C_{max} 
5.7 
6.9 
0.820 
0.757 – 0.888 



AUC_{48} 
125.1 
172.9 
0.723 
0.672 – 0.779 



AUC_{72} 
158.7 
222.0 
0.715 
0.645 – 0.792 



AUC_{96} 
177.7 
256.6 
0.693 
0.631 – 0.760 


400 µg vs. 600 µg 
C_{max} 
5.3 
6.9 
0.775 
0.715 – 0.839 



AUC_{48} 
115.4 
172.9 
0.667 
0.620 – 0.718 



AUC_{72} 
135.9 
222.0 
0.612 
0.553 – 0.678 



AUC_{96} 
164.0 
256.6 
0.639 
0.582 – 0.702 


450 µg vs. 400 µg 
C_{max} 
5.7 
5.3 
1.058 
0.979 – 1.145 



AUC_{48} 
125.1 
115.4 
1.084 
1.008 – 1.165 



AUC_{72} 
158.9 
135.9 
1.168 
1.057 – 1.291 



AUC_{96} 
177.7 
164.0 
1.084 
0.989 – 1.188 


* Antilogarithm of the least squares means for logarithms. + Antilogarithm of the difference (test minus reference) of the
least squares means for logarithms. 


The mean serum concentrationtime plots for baseline T_{4} on Study Day –1 prior to dosing with levothyroxine sodium in each Period are presented in Figure 5. Analysis of the T_{4} concentration data obtained during the 24 hours of Study Day –1 of each period confirmed that T_{4} has a diurnal cycle with statistically significant differences across time. The diurnal variation in baseline T_{4} concentrations prior to dosing are consistent with the observed diurnal variation in the serum concentrations of TSH (Figure 6).
Analysis of the 24hour AUC for Study Day –1 revealed that the regimens (dose levels) had statistically significantly different carryover effects from one period to the next (firstorder carryover) and from Period 1 to Period 3 (secondorder carryover).

Figure 5. Mean Levothyroxine (T_{4}) ConcentrationTime
Profiles on Study Day –1 Prior to Dosing with Levothyroxine Sodium by Period
The mean serum concentrationtime plots for TSH for the 24 hours prior to and 96 hours after administration of levothyroxine sodium on Study Day 1 are presented in Figure 6. The serum concentrations of TSH appear to clearly show diurnal variation, prior to dosing. During the 24hour period prior to dosing, the concentrations of TSH decline during the morning hours until reaching the lowest levels at approximately 1200 before starting to increase to maximum values at 0200 the next morning, i.e., the morning of Study Day 1 (18 hour sample on Study Day –1).
Administration of any of the three large doses of levothyroxine sodium substantially, but not completely, suppressed the TSH serum concentrations throughout the 24hour period after dosing on Study Day 1. TSH serum concentrations continued to be suppressed throughout the 96hour sampling period after dosing; the concentrations did not return to baseline values even after 96 hours. The rank order of suppression of the TSH serum concentrations was consistent with the rank order of the size of levothyroxine sodium dose administered in each of the three regimens with the greatest suppression of TSH serum concentrations associated with administration of the largest dose (Regimen A, 600 µg).
Figure 6. Mean TSH ConcentrationTime Profiles for
the 24 Hours Prior to (Study Day –1) and for the 96 Hours after Administration
of Levothyroxine Sodium on Study Day 1
The mean T_{3} concentration for the 24hour period prior to dosing and throughout the 96‑hour period after dosing were in the very narrow range of 1.1 to 1.3 ng/mL after administration of the large doses of levothyroxine sodium to healthy volunteers.
Determination of the bioavailability of levothyroxine sodium products in healthy volunteers presents significant challenging issues. Levothyroxine is naturally present in the blood, with total endogenous baseline T_{4} levels ranging from 4 to 14 µg/dL. Thus, to compare the bioavailabilities of levothyroxine sodium formulations after a single dose in healthy volunteers, FDA Guidance^{2} recommends administration of 600 µg, several times the normal clinical dose, to raise the levels of the drug significantly above baseline and to hopefully reduce the influence of endogenous levels. However, results from several bioavailability studies and a stochastic simulation study with levothyroxine products suggested that, given very reasonable assumptions about endogenous levothyroxine behavior in healthy subjects, the use of baseline uncorrected C_{max} and AUC_{48} values would result in a high probability of declaring two products bioequivalent when they actually differ by as much as 35%.^{3}
The current study was designed to evaluate how much two
formulations could differ and still pass the bioequivalence criteria specified
in the current guidance when not correcting for endogenous T_{4} baseline levels. The results from this study clearly indicate
that the use of baseline uncorrected C_{max}, AUC_{48}, AUC_{72} and AUC_{96} values would result in declaring two products
bioequivalent when they actually differ by as much as 25% to 33% (450 µg and
400 µg versus 600 µg). Utilizing
the criteria specified in FDA Guidance,^{2} both the 450 µg dose (Regimen B) and the
400 µg dose (Regimen C) would be declared bioequivalent to the 600 µg dose
(Regimen A) because the 90% confidence intervals for evaluating bioequivalence
obtained without correcting for endogenous T_{4} baseline levels were contained within the 0.80
to 1.25 range. Furthermore, the 450 µg
dose would be declared bioequivalent to the 400 µg dose because the 90%
confidence intervals for evaluating bioequivalence without correcting for
endogenous T_{4}
baseline levels were contained within the 0.80 to 1.25 range. Considering the margin by which the
conditions for declaring bioequivalence were passed in this study, products
that differ by more than 33% would have a good chance of being declared
bioequivalent on the basis of uncorrected data.
The results of this study clearly demonstrate the significant
limitations and problems with the current methodology and criteria for
assessing the bioequivalence of levothyroxine sodium products in healthy
volunteers without correcting for endogenous T_{4} baseline levels.
Several mathematical and statistical methods can be used to correct for the contribution of T_{4} baseline levels, based on different biologic assumptions about the behavior of endogenous T_{4} following administration of exogenous levothyroxine. When a single dose of exogenous levothyroxine sodium is given to healthy subjects, one could assume that endogenous levothyroxine levels remain constant if there is no suppression of endogenous production (Correction Method 1). If production were completely suppressed, via feedback through the hypothalamicpituitary axis, the endogenous levothyroxine would decline at an average rate defined by its halflife, which is approximately 7 days (Correction Method 2). Thus, a constant baseline of endogenous levothyroxine (Correction Method 1) versus a baseline that decays exponentially with a 7day halflife (Correction Method 2) defines the limits for endogenous levothyroxine following a dose of exogenous levothyroxine sodium. This assumes that no other components of the thyroid system would impact the turnover of T_{4} and T_{3}. The third method of baseline correction (Correction Method 3) employed in this study corrected the T_{4} concentration for each time of postdose sampling by the baseline T_{4} concentration observed at the same time of day during the 24 hours preceding the dose, i.e., on Study Day –1.
One of the objectives of the current study was to better understand the impact of three different methods of correction for endogenous T_{4} baseline on the bioequivalence evaluation of levothyroxine sodium formulations in healthy volunteers. In contrast to the results with uncorrected data, for all three correction methods for endogenous T_{4} baseline, neither the 450 µg dose nor the 400 µg dose would be declared bioequivalent to the 600 µg dose. However, as with the uncorrected data, the 450 µg dose would continue to be declared bioequivalent to the 400 µg dose after correcting for endogenous T_{4} baseline levels using any of the three correction methods because the 90% confidence intervals for evaluating bioequivalence after correcting for endogenous T_{4} baseline continue to be contained within the 0.80 to 1.25 range. The 50µg difference between the 450 µg dose and the 400 µg dose represents a 12.5% difference.
Correction Method 1 relies on the assumption that there is no suppression of endogenous production when a single large dose of exogenous levothyroxine sodium is given to healthy subjects, thus assuming a constant baseline of endogenous levothyroxine. This assumption is clearly not true since TSH levels after dosing with levothyroxine sodium in the study were definitely suppressed, though not completely. Thus, it is very unlikely that endogenous T_{4} production would be constant after administration of large doses of levothyroxine sodium to healthy volunteers. This method of correction has also several undesirable characteristics. The method will sometimes produce a negative value for AUC as was observed with one of the subjects in this study. Furthermore, the method relies completely upon the results from only three samples obtained during an interval of only 30 minutes just prior to dosing. Just from a consideration of randomness alone, the influence of the average of these three concentrations could be significant. More troubling than the small number of observations is the brief time span from which they are taken. It is known that there is a circadian effect on hormone levels, and the Day –1 data from this study clearly confirmed the presence of the circadian effect. Therefore, unless a subject's expected T_{4} levels during the 30 minute time frame just prior to dosing happens also to be the expected average for a 24hour cycle, the corrected AUC by this method is in error.
Correction Method 2 depends upon the assumption that endogenous production of levothyroxine is completely suppressed when a single large dose of exogenous levothyroxine sodium is given to healthy subjects. Therefore, already available endogenous levothyroxine will decline at rate defined by its halflife, which is assumed to be 7 days. This method also has several undesirable characteristics. Method 2 gives a reasonable correction only if production of endogenous T_{4} abruptly and completely stops when study drug is administered and does not resume during the sampling period. Even if this unlikely assumption is true, the correction will be in error for a given subject, with the size of the error depending on how much the given subject's elimination halflife differs from 7 days. The halflife of levothyroxine is not very well documented in healthy volunteers and the 7day halflife is an approximation based on data from isotope studies with levothyroxine. As previously noted, TSH levels after dosing with levothyroxine sodium were definitely suppressed, but not completely. Thus, it seems very unlikely that endogenous T_{4} production would be reduced to zero, with an accompanying 7‑day halflife. The use of a single value for levothyroxine halflife for all healthy subjects (regardless of gender, race, and age) at all times is clearly a significant oversimplification. However, estimation of a levothyroxine halflife for each subject in each period is not possible using the currently recommended design in healthy volunteers. Moreover, as with Method 1, Method 2 relies heavily on the average of three concentrations taken immediately before dosing. In particular, for the case in which a subject randomly has a predose average considerably higher than typical for that subject, the corrected AUC is more likely to be negative.
The third method of baseline correction (Method 3) employed in this study corrected the T_{4} concentration at each time of postdose sampling by the corresponding baseline T_{4} concentration observed at the same time of day during the 24hour period preceding the dose, i.e., on Study Day –1. This method provides some advantages in comparison to Methods 1 and 2. The obvious advantages for this method are a) it does not rely on just three samples collected over a very short time period prior to dosing for the correction, and b) the postdose T_{4} concentration is adjusted based on the actual baseline T_{4} concentration at the same clock time of the day before dosing in the same subject in the same period, and thus, this method takes into account the diurnal variation in the baseline T_{4} concentration throughout the day in each subject, which is ignored by Methods 1 and 2.
In contrast to Method 2, for Method 3, endogenous T_{4} production is not assumed to abruptly stop following study drug administration and a constant value for the elimination halflife across subjects is not assumed. However, similar to Method 1, Method 3 relies on the assumption that there is no suppression of endogenous production when a single dose of exogenous levothyroxine sodium is given to healthy volunteers. Furthermore, Method 3 requires the assumption that the circadian pattern in the endogenous T_{4} production does not change when a single large dose of exogenous levothyroxine is administered to healthy subjects.
The impact of administration of large doses of levothyroxine sodium (e.g., 600 µg) on the endogenous production of T_{4} is not known. However, the TSH levels are clearly, but not completely, suppressed after administration of the large doses of levothyroxine sodium to the healthy volunteers in this study. The large exogenous dose may also affect the clearance of total T_{4} via numerous feedback mechanisms. The TSH serum concentrationtime data provide clear evidence of the limitations for each of the three methods of correction utilized in this study. Method 2 assumes that endogenous T_{4} production is abruptly and completely stopped after study drug administration while Methods 1 and 3 assume that there is no suppression of endogenous production when a single dose of exogenous levothyroxine sodium is given to healthy volunteers.
The FDA Guidance^{2} recommended a minimum 35day washout period between the doses of levothyroxine sodium to minimize carryover. The 24hour profiles of the baseline T_{4} serum concentrations on the day before dosing were clearly not the same for the three study periods even though the washout periods between the doses of levothyroxine sodium in this study were 44 days between Periods 1 and 2 and 53 days between Periods 2 and 3. The Day –1 baseline T_{4} data from this study provide convincing evidence that there are carryover effects from the successive study doses, even from the Period 1 dose to the Period 3 dose, and that the carryover effects of the dose levels differ. Carryover effect from the 600 µg dose resulted in higher T_{4} levels than carryover effects of the two lower doses. Exploratory analyses of postdose uncorrected C_{max} and AUC give additional strong evidence of these carryover effects. Also, such unequal carryover effects are present for C_{ma}_{x} with all three methods of correction. Another component of the period effect may be the presence of seasonal and annual variations in hypothalamicpituitarythyroid hormone concentrations in humans. Significant seasonal and annual rhythms in serum TSH and T_{3} levels have been reported in the literature.^{4} However, the amplitude of the circannual rhythm is probably not as large as that of the daily circadian variation.^{4} Therefore, the results from our studies suggest that a much longer washout period between dosing would be required to truly reduce the impact of carryover between dosing periods.
The results of this study strongly suggest that obtaining additional blood samples on Study Day –1 provided data that improved the method of correction for endogenous levels of T_{4}, accounting for the possibility of a circadian pattern. Additional samples during the afternoon and night hours on the day before dosing and on the days after dosing may provide further benefits to this method of correcting for the endogenous baseline.
It is widely recognized that dose initiation and titration need to be done in susceptible groups with the 12.5 µg dosage strength. In the package insert of levothyroxine sodium products,^{5} it states under 'Dosage and Administration – Specific Patient Populations' "the recommended starting dose of levothyroxine sodium in elderly patients with cardiac disease is 12.5 – 25 µg/day, with gradual dose increments at 4 to 6 week intervals. The levothyroxine sodium dose is generally adjusted in 12.5 to 25 µg increments until the patient with primary hypothyroidism is clinically euthyroid and the serum TSH has normalized." NDA approved levothyroxine sodium tablets are available in strengths that differ from their nearest doses by 12 to 13 µg/tablet: that is 75, 88, 100, 112, 125, 137 and 150 µg tablet strengths. The 88 and 112 µg strengths are 12% less or greater, respectively, than the 100 µg strength.
Even though the three methods of correction for endogenous T_{4} baseline improve the ability to distinguish between products that are truly different in dose by 25% to 33%, none of the three correction methods were able to distinguish between two products that differ by 12.5%. As stated earlier and similar to the findings with the uncorrected data, the 450 µg dose would continue to be declared bioequivalent to the 400 µg dose after correcting for endogenous T_{4} baseline using any of the three correction methods. Narrowing the 90% confidence intervals for evaluating bioequivalence after correcting for endogenous T_{4} baseline from the standard range of 0.80 to 1.25 would reduce the chance that two products that differ by 12.5% would be declared bioequivalent.
The potential for conducting bioequivalence trials in
athyreotic subjects, a model that minimizes confounding effects from endogenous
T_{4} due to the
absence of residual endogenous hormone, must also be considered. A study in athyreotic subjects would
presumably be a multipledose study and long enough to properly address the
issue of carryover effect. Such a study
in athyreotic subjects would utilize therapeutic doses of levothyroxine sodium
and remove the need for a method of baseline correction.
This study illustrates some important flaws in the design and analysis of singledose crossover studies in healthy volunteers to assess bioequivalence of levothyroxine sodium products, stemming from the significant and complex contribution of endogenous T_{4}. First, the results indicate that the use of baseline uncorrected T_{4} C_{max}, AUC_{48}, AUC_{72} and AUC_{96} values would result in declaring two products bioequivalent when they actually differ by as much as 25% to 33% (450 µg and 400 µg versus 600 µg). The 450 µg dose and the 400 µg dose would both be declared bioequivalent to the 600 µg dose because the 90% confidence intervals for evaluating bioequivalence without correction for endogenous T_{4} baseline were contained within the 0.80 to 1.25 range. Considering the margin by which the conditions for declaring bioequivalence were passed in this study, products that differ by even more than 33% would also have a high likelihood of being declared bioequivalent.
Second, the results from this study indicate that the use of baseline corrected C_{max}, AUC_{48}, AUC_{72} and AUC_{96} values would reduce the likelihood that two products would be declared bioequivalent when they actually differ by 25% to 33%. After correcting for endogenous T_{4} levels using each of the three correction methods employed in this study, neither the 450 µg dose nor the 400 µg dose would be declared bioequivalent to the 600 µg dose because the 90% confidence intervals for evaluating bioequivalence were not contained within the 0.80 to 1.25 range for C_{max}, AUC_{48}, AUC_{72} and AUC_{96}.
Third, the 450 µg dose would continue to be declared bioequivalent to the 400 µg dose utilizing the C_{max}, AUC_{48}, and AUC_{96} values for the baseline corrected T_{4} data by any of the three methods of correction. A 12.5% difference (400 µg versus 450 µg) in levothyroxine sodium products may have a clinically relevant adverse impact on patients. Thus, it is apparent that simple methods of correction for endogenous T_{4} concentrations in healthy volunteers are inadequate since these concentrations not only fluctuate on a diurnal cycle but may also be differentially affected by products with different rates and extents of absorption. Additionally, there is evidence of significant carryover from one dosing period to subsequent periods even with washout periods up to 53 days.
The potential for conducting multipledose bioequivalence trials in athyreotic subjects, a model that minimizes confounding effects from endogenous T_{4} due to the absence of residual endogenous hormone, must also be considered. Such a study in athyreotic subjects would utilize therapeutic doses of levothyroxine sodium and remove the need for a method of baseline correction.
Reference List for AppendixA
1. Schuirman DJ. A comparison of the two onesided tests procedure and the power approach for assessing the equivalence of average bioavailability. J Pharmacokinetics Biopharm. 1987;15:65780.
2. Guidance for Industry: Levothyroxine sodium tablets – in vivo pharmacokinetic and bioavailability studies and in vitro dissolution testing. US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research, December 2000.
3. Riley S, Ludden TM, Simulation study to assess alternative bioavailability calculations, study designs and acceptance criteria for determining the bioequivalence of levothyroxine sodium tablets. GloboMax Technical Report, Project #KNP00500, April 2002.
4. Maes M, Mommen K, Hendrickx D, Peeters D, D'Hondt P, Ranjan R, et. al. Components of biological variation, including seasonality, in blood concentrations of TSH, TT3, FT4, PRL, cortisol and testosterone in healthy volunteers. Clin Endocrinol. 1997;46:587598.
5. Synthroid (levothyroxine sodium tablets, USP). Physician Package Insert. Abbott Laboratories, Inc., North Chicago, IL. 035195R1Rev. July, 2002.