Treatment - The Journal of Clinical Endocrinology & Metabolism

J Clin Endocrinol Metab, October 2009, 94(10):3700–3707 jcem.endojournals.org 3703
Laboratory analysis
The laboratory analyses were performed at the different study
centers. The reference intervals as well as the analysis methodology
were consequently different. Also, the methods changed
during the time of the study. The lower reference limit for free T 4
at randomization was 8–11.7 pmol/liter, and the upper limit was
18–28 pmol/liter. Disregarding the fact that different methods
had been used, the levels of serum free T 4 at randomization were
not different in the two groups [in group I, 56.8 (SD 18.0) pmol/
liter; and in group M, 56.4 (SD 19.2) pmol/liter].
Statistical analysis
The primary endpoint hypothesis was tested by means of a � 2
test. The influence of possible prognostic factors on the time to
TAO was investigated with Cox regression analysis. The development
within the various subgroups was illustrated by means of
Kaplan Meier technique. P values were two-sided, and P � 0.05
was considered significant. The MEDLOG software system (release
2008-2; Information Analysis Corp., MEDLOG Systems,
Crystal Bay, NV) was used for storage of data and statistical
analysis.
Results
The two treatment groups were similar for the characteristics
listed in Table 1. Among the patients randomized for
iodine-131, two patients relapsed and had thyroid surgery
and further radioiodine treatment, respectively. Among
the patients randomized for and accounted for medical
treatment, 12 patients experienced adverse events that led
to change of therapy to radioiodine. Furthermore, 33
(22%) of the patients randomized for medical treatment
relapsed. Of those, five were treated with surgery, 25 with
radioiodine, and three with further medical therapy.
Among the patients in group M who received iodine-131
as treatment for relapse, only one subsequently developed
TAO (see Appendix II). Early administration of L-thyroxine
after radioiodine treatment was not associated with
arrhythmia or serious adverse effects.
In patients randomized to treatment with radioiodine,
a significantly higher cumulative incidence of worsening
or development of TAO (63 patients; 38.7%) was observed
compared with the group randomized to antithyroid
drugs (32 patients; 21.3%; � 2 , P � 0.001). The log
rank Mantel-Haenszel test for equality of the curves for
worsening or development of TAO confirmed this difference
(P � 0.001; see Fig. 2A).
In the patients with TAO already at the first visit (13.1%),
the overall course of the ophthalmopathy was not significantly
affected by the choice of treatment, but de novo development
of TAO was significantly more frequent in group
I (Fig. 1; P � 0.001). Of the 41 patients with preexisting
TAO, 10 patients in each group improved; one patient in
group I and two patients in group M had initial improvement
Number of patients
60
50
40
30
20
10
0
10
9
with a subsequent deterioration. Two patients in group I had
no change; nine and seven patients deteriorated in groups I
and M, respectively. If patients with only eyelid retraction
were included in the group with preexisting TAO, a Mantel-
Haenszel comparison between the treatment groups still
yielded a nonsignificant difference for worsening (P �
0.24) and an increased risk for de novo development of
TAO (P � 0.001).
Independent of the choice of treatment, smoking increased
the risk of worsening or development of TAO (P �
0.001; Fig. 2B). However, in smokers the choice of treatment
for Graves’ hyperthyroidism did not have a significant
impact on the outcome (P � 0.38; Fig. 2D), in contrast to
nonsmokers where iodine-131 treatment significantly increased
the risk of worsening or development of TAO (P �
0.001; Fig. 2C).
In a Cox regression analysis, both the choice of therapy
and smoking habits were shown to influence the risk of
worsening or development of TAO (Table 2). The effect of
therapy for hyperthyroidism on the time to worsening or
development of TAO was significantly different in the
group of smokers and nonsmokers (P � 0.01). In the Cox
regression statistics, this was analyzed by using the interaction
term iodine-131 therapy in smokers (Table 2). The
differences are demonstrated graphically in Fig. 2, C and D.
The associations between TAO and pretreatment laboratory
levels of thyroid hormone could merely be estimated,
because the methods and reference intervals differed
between the centers. However, taking these limitations into
consideration, the Cox regression analysis suggested that
higher free T 4 levels were associated with an increased risk
for TAO (Table 2).
In the post hoc analysis, using the set criteria for the
change in TAO, worsening or development of TAO was
also found to occur more often in group I (40 patients;
25%) than in group M (16 patients; 11%; P � 0.002). The
Mantel-Haenszel comparisons between different groups
as shown in Fig. 2, A–D, showed comparable results. The
53
23
worse pre-existing new development
Iodine-131
Medical
FIG. 1. Number of patients with worsening of preexisting TAO [10 of
22 patients (45%) in group I; and nine of 19 patients (47%) in group
M] and de novo development of TAO [53 of 141 patients (38%) in
group I; and 23 of 131 patients (18%) in group M] at any time during
follow-up.