Full text of DOI:10.1007/BF03343979

J. Endocrinol. Invest. 25: 152-157, 2002
Does smoking increase relapse rates in Graves’ disease?1
L.E. Kimball*, E. Kulinskaya**, B. Brown*, C. Johnston*, and N.R. Farid*,**
*The Department of Endocrinology, Hemel Hempstead General Hospital, Hemel Hempstead;
**University of Hertfordshire, Hatfield, Hertfordshire, United Kingdom
ABSTRACT. We wondered whether the relapse
rate of Graves’ hyperthyroidism was increased
amongst patients who smoked. To this end, a
retrospective analysis of clinical and laboratory
data of consecutive patients with Graves’ dis-
ease was carried out. All patients were treated
with thionamide anti-thyroid drugs (ATD) for at
least one year and remission was defined as
continued and biochemical euthyroidism, at
least 6 months after discontinuing ATD. The
study comprised 221 subjects with Graves’ dis-
ease from a hospital-based population over 9
years. We took the following variables into ac-
count when assessing contribution to disease
relapse: Goiter, Thyroid Associated Ophthalmo-
pathy (TAO), and Time to euthyroidism after
starting ATD and interaction between smoking
and sex. Smoking had a marginally significant
(p=0.081) deleterious effect on the likelihood
of remission after ATD treatment for Graves’
disease. The effect of smoking was, however,
highly significant in males and indeed the dele-
terious effect on remissions may be restricted
to males (odds ratio, 11.1; 95% confidence in-
terval, 1.25 to 98.5). The presence of goiter
(odds ratio, 3.8; 95% confidence interval, 2.05
to 7.1) and TAO (odds ratio, 1.8; 95% confi-
dence interval, 0.993 to 3.18) forecasted low-
er chances of achieving remission. The shorter
the time a patient became euthyroid after start-
ing ATD the more likely his disease was to re-
mit. We conclude that cigarette smoking in-
creases the likelihood of Graves’ disease recur-
rence in males treated with anti-thyroid drugs.
Thus, smoking appears to be an important risk
factor in the pathogenesis and outcome of
Graves’ disease patients at least in subsets of
patients.
(J. Endocrinol. Invest. 25: 152-157, 2002)
©2002, Editrice Kurtis
INTRODUCTION
We have previously identified two subsets of Gra-
ves’ disease patients, which differed in severity and
prevalence of TAO. Patients with severe disease
had high rates of disease recurrence after withdra-
wal of anti-thyroid drugs, large goiters, family his-
tory of the disease, laboratory parameters of auto-
aggression and high prevalence of Major Histoco-
mpatibility Complex susceptibility alleles for Graves’
disease (2, 3). The influence of smoking on disease
relapse rate has not been previously specifically ad-
dressed.
Graves’ disease is characterized by remissions and
relapses and many factors affect the length of re-
mission following the cessation of anti-thyroid ther-
apy (4-10). We wondered whether smoking has an
influence on remission rate in patients treated with
anti-thyroid drugs. To this end, we retrospectively
analyzed the clinical and laboratory data from a
large group of patients with Graves’ disease treat-
ed with anti-thyroid drugs. From this we sought to
answer the question of whether smoking is associ-
ated with higher recurrence rates of Graves’ dis-
ease.
Cigarette smoking is an important risk factor in the
pathogenesis and outcome of an increasing number
of diseases. Smoking is the most visible environmen-
tal risk factor for Thyroid Associated Ophthalmopathy
(TAO). It has been suggested that smoking may in-
fluence ophtalmopathy through direct irritative ef-
fects or by modulating immune reactions that occur
in Graves’ disease. Smoking probably also adversely
influences the outcome of the therapy of eye disease.
Specifically, smoking and the degree of smoking have
been found to affect the outcome of orbital radiation
therapy and high-dose glucocorticoid treatment in
patients with severe ophthalmopathy (1).
1Presented in part at the Annual Meeting of the Endocrine Society, San
Diego, Ca, June 12-15, 1999.
Key-words: Graves’ disease, remission, antithyroid drugs, smoking, lo-
gistic regression.
Correspondence: Prof. Nadir R. Farid, Osancor Biotech Inc, 31 Woodland
Drive, Watford, Herts WD17 3BY, UK.
E-mail: farid@osancor96.fsnet.co.uk
Accepted August 27, 2001.
152

Smoking and Graves’ remission
SUBJECTS AND METHODS
16 (30%) were goitrous (goiter 2-4), and 39 (74%)
had serum thyroglobulin Ab titers >1/40. In the sub-
group with relapsing disease (no.=130), 42 (32%)
were smokers, 81 (62%) had TAO, 80 (62%) were
goitrous, and 115 (89%) had thyroglobulin Ab titers
>1/40.
Of the 38 (17.2%) male patients, Graves’ disease
remitted in 13 (34%). Only 1 patient (8%) of the sub-
group with remitting disease was a smoker: 4 (31%)
had TAO, 3 (23%) were goitrous, and 7 (54%) had
thyroglobulin Ab titers >1/40. Of the 25 with re-
lapsing disease, 12 (48%) were smokers, 16 (64%)
had TAO, 14 (56%) were goitrous, and 20 (80%) had
thyroglobulin Ab titers >1/40.
This study identified 263 patients with Graves’ dis-
ease visited at West Hertfordshire, England over 9
yr (1988-1996). Two hundred and twenty-one were
deemed to have complete evaluable records and
met the diagnostic, therapeutic and follow-up crite-
ria of the study. All patients were treated with anti-
thyroid drugs (ATD) for at least one year (range 12 to
96 months). Remission was defined as continued clin-
ical and biochemical euthyroidism at least 6 months
after discontinuing ATD. This cut-off point is not with-
out precedent, was logistically convenient and valid
in that, in retrospect, >90% of a smaller subset of pa-
tients who were in remission at 6 months were so at
the end of a one year of follow-up.
Characteristics examined included sex, age, age at
diagnosis of Graves’ disease, smoking, goiter size,
presence of TAO (11), serum thyroid hormone lev-
els, thyroid auto-antibody levels, time to chemical
euthyroidism after institution of ATD and total du-
ration of therapy. These characteristics were relat-
ed to remission of Graves’ hyperthyroidism. Smo-
king status was determined at the time of diagno-
sis and because of the nature of this retrospective
data analysis the quantity of cigarettes smoked was
difficult to ascertain. Patients were regarded as
smokers, ex-smokers or non-smokers at the time of
diagnosis. In this study the attribute ”smoking” ap-
plies to current smokers.
Statistical analysis
The SPSS package was used for our statistical anal-
ysis. Analyses of the effect of categorical attributes
on the odds against remission were performed us-
ing Fisher’s exact test. Odds ratios with 95% confi-
dence intervals were also calculated. The Mann-
Whitney test was used to analyze differences in con-
tinuous characteristics between the patients whose
disease remitted and those with recurrent disease.
Normal approximation was used to calculate the
standardized normal value (Z), and the p-value.
Logistic regression analysis was used to explore dif-
ferent models of attributes to predict remissions.
Inspection and palpation at the time of diagnosis
estimated thyroid size on a scale of: 0-4 (0=normal
thyroid; 1=small, 2=medium, 3=large and 4=very
large goiter) based on an estimated thyroid weight
of 20 g (grade 0) for a normal thyroid and ꢀ40 g for
a very large thyroid (grade 4). Serum samples were
drawn at the time of diagnosis for analysis of T3, T4,
free T4 and TSH. Total T3, T4, fT4 and TSH were
measured using double-antibody radioimmunoas-
says, the normal reference ranges being 1.2-3.0
nmol/l for T3, 70-140 nmol/l for T4, and 9-22 pmol/l
for free T4. A basal TSH level <0.5 mU/l was re-
garded as pathological. TG and thyroid microso-
mal antibodies were measured with a solid-phase
immunosorbent radioassay. TSH receptor antibod-
ies were not measured.
The total study group consisted of 221 patients.
The average age of the overall group was 41.73
yr 13.7 ( SD). One hundred and twenty-nine
(58.4%) patients were considered to have some de-
gree of TAO. Seventy (31.7%) were current smokers
and 25 (11.3%) had recently stopped smoking.
One hundred and eighty-three patients (82.8%)
were female. Of these hyperthyroidism remitted in
53 (29%). Of the subgroup with remitting Graves’
disease, 15 (28%) were smokers, 28 (53%) had TAO,
RESULTS
Sixty-six patients (53 women and 13 men) achieved
remission of Graves’ disease. This is 29.9% of total
number of patients. The average age of those
whose disease remitted was identical to those with
recurrent disease (41.76 yr 13.7 vs 41.86 years
13.11).
The influences of categorical attributes, e.g. sex
and smoking on likelihood of remission, were ex-
amined and are shown in Table 1. Fisher’s exact
test was used to test for associations. Odds ratios
and 95% confidence intervals were also calculated.
Seventy (31.7%) patients were current smokers, and
25 (11.3%) were ex-smokers. Among all of the pa-
tients, hyperthyroidism was in remission in 29.9%. In
descending order of odds ratios cigarette smoking
in males, the presence of goiter, anti-thyroglobu-
lin antibody titer (as a measure of thyroid-specific
autoimmunity) of 40 or greater and TAO vitiate
against a person with Graves’ disease treated with
ATD for at least one-year achieving remission. It is
noteworthy that only 1 of 13 males who smoked
achieved remission.
We next examined continuous characteristics, as
well as potential interactions between categorical
153

L.E. Kimball, E. Kulinskaya, B. Brown, et al.
Table 1 - Effect of categorical attributes on the odds against remission.
Attribute
Odds ratio* (95% CI)
Significance
(Fisher’s exact test, one sided)
Sex (female)
1.275 (0.607-2.680)
0.322
Smoking
Female
Male
1.671 (0.870-3.209)
1.209 (0.599-2.439)
11.077 (1.245-98.548)
0.081
0.365
0.013
TAO
1.778 (0.993-3.180)
3.817 (2.045-7.092)
2.935 (1.451-5.936)
0.037
0.000015
0.003
Goiter
Thyro-Ab titer**
TAO=thyroid associated ophthalmopathy; thyro-Ab titer=thyroglobulin antibody titer. *Odds ratio against remission. **In this analysis 1/40 titer was used
as a division between positivity or otherwise for thyroglobulin antibodies.
and continuous attributes, which may influence re-
missions of Graves’ hyperthyroidism. Mann-Whitney
(U) was used to analyze this aspect of the data.
Normal approximation was used to calculate the
standardized normal value (Z), and the p-value.
The data in Table 2 shows that the speed with
which euthyroidism is achieved with ATD drugs pre-
dicts remissions. This is also reflected in the mean
length of therapy. Interestingly, in this data set
younger women rather than older men were more
likely to go on to remissions. Median age of females
whose Graves’ disease went into remission was 41
yr than compared to 46 in those who did not remit;
whereas for men the median ages were respectively
57 and 46 yr. Likewise, for those in remission, wom-
en were significantly younger at diagnosis than men
(U=171.5, p=0.0026).
are shown in Table 3 for female and male patients
separately. Fisher’s exact test was used to test for
the association of smoking with remission of Gra-
ves’ disease. There were no significant effects of
smoking. The reason for this negative finding may
well be the small subgroup sizes. This simple anal-
ysis is included to explain the necessity for more
sophisticated multivariate analysis to follow, i.e. the
logistic regression analysis.
We next sought to explore multivariate models in
which we factored in the above attributes and that
would allow us to predict remissions. For this we
used logistic regression analysis. We investigated
the following possible predictors for remission: goi-
ter size (0-3), presence of TAO, presence of thyroid
auto-antibodies, time to euthyroidism after starting
ATD, main effects and interaction between smoking
and gender, age and its interaction with gender.
Of the large number of models explored we favor
the one given in Table 4, which takes into account
goiter size (p=0.0007), presence of thyroid auto
antibodies (p=0.0021), gender (p=0.0276), time to
euthyroidism (p=0.000), and an interaction of age
To further explore the effect of smoking on likeli-
hood of remission controlling for other categorical
attributes influencing remission, we examined eight
different subgroups of patients varying in pres-
ence/absence of ophtalmopathy, thyroglobulin an-
tibodies titer of 40 or greater, and goiter. Results
Table 2 - Continuous variables influencing remission for all patients and for males and females separately. Values are median (in-
terquartile range).
Remission
(no.=66)
No remission
(no.=155)
U
Z
p-value
(one-sided)
Time to Eu* (months)
Time of Rx (months)
5 (3-7)
7 (4-10)
18 (15-25)
46 (37-57)
46 (40-54)
39.5 (30-50)
42 (33-48)
3698.5
3276.5
2888.0
107.0
-3.2694
-4.240
-1.714
-1.709
-0.905
-2.111
0.0006
0.00001
0.043
14 (13-17)
41 (34-50)
57 (49-64)
37 (30-47)
54 (44-62)
Age of women (years)
Age of men (years)
0.044
Diagnosis age, women (yr)
Diagnosis age, men (yr)
3151.0
94.0
0.183
0.0175
*Eu=euthyroidism. Time to euthyroidism refers to the period between onset of therapy with anti-thyroid medications and the normalization of thyroid
function.
154

Smoking and Graves’ remission
Table 3 - Effect of smoking on remission, controlling for other categorical attributes influencing remission for men and women sep-
arately. Values are numbers (percentages) in each subgroup.
Sex
Smoking
TAO
No
Thyroglobulin
Ab titer*
Goiter**
No
Relapse
Remission
Total
p-value, Fisher’s
exact test
Female
No
Yes
No
No
Yes
No
Yes
No
Yes
Yes
1 (50%)
1 (33%)
1 (50%)
2 (67%)
2
3
1.0
-
No
Yes
Yes
No
No
4 (67%)
1 (25%)
2 (33%)
3 (75%)
6
4
0.524
-
No
Yes
No
13 (56%)
3 (60%)
10 (44%)
2 (40%)
23
5
1.0
-
No
Yes
No
Yes
3 (43%)
1 (100.0%)
4 (57%)
-
7
1
1.0
-
No
Yes
Yes
Yes
No
No
18 (60%)
9 (64%)
12 (40%)
5 (36%)
30
14
1.0
-
No
Yes
Yes
3 (75%)
1 (50%)
1 (25%)
1 (50%)
4
2
1.0
-
No
Yes
Yes
16 (80%)
11 (85%)
4 (20%)
2 (15%)
20
13
1.0
-
No
Yes
Yes
Yes
30 (88%)
15 (100.0%)
4 (12%)
-
34
15
0.298
-
Total
130 (71%)
53 (29%)
183
-
Male
No
No
Yes
No
No
Yes
Yes
No
Yes
No
No
Yes
No
Yes
No
Yes
Yes
No
No
No
Yes
No
Yes
Yes
Yes
1 (33%)
2 (67%)
-
3
1
1.0
-
Yes
1 (100%)
No
Yes
-
2 (100%)
-
2
1
0.333
-
1 (100%)
No
Yes
2 (33%)
1 (100%)
4 (67%)
-
6
1
0.429
No
Yes
-
-
2 (100%)
-
2
-
N/A
No
Yes
3 (60%)
2 (100%)
2 (40%)
-
5
2
1.0
-
No
Yes
2 (100%)
-
-
-
2
-
N/A
-
No
Yes
3 (100%)
1 (50%)
-
3
2
0.400
-
1 (50%)
No
Yes
2 (100%)
6 (100%)
-
-
2
6
-
-
Total
25 (66%)
13 (34%)
38
*In this analysis thyroglobulin Ab titers >1/40 were considered positive. **Goiter scores 2-4 taken as positive. TAO: thyroid associated ophthalmopa-
thy; Ab titer: antibody titer.
with gender (p=0.0184). From Table 4 we can see
that goiter size of 2 or above carries a significant
risk of non-remission. The comparatively short time
to euthyroidism after starting ATD, age (being
younger for females, and older for males), and ab-
sence of thyroid antibodies are all predictive of re-
mission. The quality of this model is explored in
Table 5. In this model prediction of remission for
patients with remitting disease (sensitivity) is cor-
rect in 45.5% of cases and prediction of relapse for
patients whose hyperthyroidism relapsed (speci-
ficity) is correct in 89.7% of cases. Overall correct
prediction (remission/no remission) is 76.5%. The
predictive value of this model is quite high.
Prediction of relapse (negative predictive value) is
correct in 81.41% and prediction of remission (pos-
itive predictive value) is 64.8% (Table 5).
Inclusion of TAO (p=0.1507) improved sensitivity
of the above model to 50%, at the expense of a re-
duction of specificity to 87.7%. Further inclusion of
155

L.E. Kimball, E. Kulinskaya, B. Brown, et al.
Table 4 - Logistic regression to predict remission. Regression coefficients, adjusted odds ratios (OR) for remission and 95% confidence
intervals for ORs. Goiter categories 0-3 are compared to the highest goiter category (4); OR of females to males; OR of thyroid an-
tibodies being negative; OR of each extra month to euthyroidism*; OR of extra year of age for females**.
Predictor
Regression
coefficient (B)
p-value,
Wald test
Adjusted odds
ratio [Exp(B)]
95% Confidence interval
for adjusted odds ratio
Lower
Upper
Gender (F)
2.8626
0.0276
17.5067
1.371
223.541
Goiter
-
0.0007
0.0045
0.0008
0.0992
0.0564
-
-
-
Goiter 0
Goiter 1
Goiter 2
Goiter 3
2.5046
2.8622
1.5115
1.7266
12.2383
17.5001
4.5335
5.6217
2.1743
3.2912
0.7519
0.9541
68.8853
93.0536
27.3362
33.1249
Thyro-Ab titer
Time to Eu
1.2562
-0.1425
-0.0598
0.0021
0.000
3.512
0.8672
0.9419
1.579
0.8117
0.8962
7.8113
0.9263
0.9900
Age*sex (F)
0.0184
*Adjusted odds ratio below 1 means the longer the time a patient became euthyroid after starting thionamide anti-thyroid drugs (ATD) the less likely
his disease was to remit. **Adjusted odds ratio below 1 means the older a female patient was the less likely her disease was to remit. EU: euthyroidism;
Ab titer: antibody titer.
Table 5 - Logistic regression to predict remission. Quality of
indeed the effect of smoking was restricted to
prediction for the model given in Table 4.
males. This finding raises the possibility that males
Remission
Observed
Predicted
Total Percent correct
are intrinsically susceptible to the immunologic and
other effects of smoking, resulting in severe disease
characterized by high recurrence rates after ATD
treatment and high prevalence of TAO.
This study upholds the reasonable expectation that
mild Graves’ disease is likely to respond more
rapidly to standard doses of thionamides as com-
pared to more severe disease (2, 3, 9, 10). It follows
that, on average, patients with milder disease re-
ceive treatment for a shorter period than those with
severe disease. The latter statistic also reflects the
reluctance (in the past) of physicians to refer pa-
tients with apparent non-remitting disease for ra-
dio-iodine therapy.
Some of the determinants of remission such as goi-
ter size, surrogate markers of auto-aggression and
rapidity with which euthyroid status is achieved af-
ter ATD affirm previous observations (2, 3, 9, 10).
Although not measured in this study, TSH receptor
antibodies have been noted to be less valuable in
predicting remission than more simple clinical and
immunologic measures (2, 6, 10).
No
Yes
16
30
46
No
139
36
155
66
89.68%
45.45%
Yes
Total
175
221
Predictive value
81.47% 64.8%
139 30
Overall correct
prediction
221
76.47%
smoking (p=0.4476) and its interaction with gender
(p=0.2029) somewhat increased sensitivity to 53%
without changing specificity of 87.7%. To summa-
rize, neither TAO, nor smoking and its interaction
with gender were significant factors, the latter be-
ing perhaps the result of comparatively small num-
ber of males (38) in this study.
DISCUSSION
TAO contribution in forecasting disease severity is
worthy of special consideration. Thus, while it was
significant in the examination of categorical at-
tributes (Table 1), it did not contribute significantly
to the logistic regression model because its effect
was confounded with that of some of the other vari-
ables already in regression. As such it did not add
any further information to the model.
The remission rate of 30% achieved in this study is
the lower end of the spectrum quoted in the liter-
ature but is in line with many recent studies (4-10).
The principal question we sought to answer was
whether smoking was associated with higher re-
currence rates of Graves’ disease. The deleterious
effect of smoking on remission rates was of mar-
ginal significance in the study group as a whole. In
male patients the relationship of smoking to recur-
rent disease was highly significant (p=0.013) and
The lack of predictive power of some of the factors
e.g. familiarity (2, 3) may have to do with the fact
156

Smoking and Graves’ remission
3. Frecker M., Preus M., Kozma L., et al. Heterogeneity by
cluster analysis techniques of Graves’ patients typed for
HLA DR and IgG heavy chain markers. Mol. Biol. Med.
1986, 3: 63-71.
that the study was retrospective, extended over
several years involving several observers scoring
manifestations without explicitly agreed upon cri-
teria. This probably also applies to the stratification
of eye disease but not its presence.
At the other end of the scale some of the influences
such as greater likelihood of remission with de-
creasing age in females and increasing age in males
contradict earlier findings (2, 3, 10). The contribu-
tion of this aspect of the study to our remission
model was of marginal significance.
The question we asked in initiating this study was
whether or not smoking influences the remission rate
of Graves’ hyperthyroidism. Broadly, the answer is
in doubt. There was, however, marked heterogene-
ity of risk between males and females in this respect
– with smoking having a major negative impact on
remission in males but not in females. This finding,
along with the reported higher prevalence of TAO
in males in many studies (1), and perhaps other dif-
ference in genetic susceptibly (12), suggest an im-
portant interaction between smoking and maleness
in determining disease severity. Apparently life
stresses (13, 14) and smoking independently (13) im-
pact on the risk of developing Graves’ hyperthy-
roidism in females but not males (13). It is thus pos-
sible that smoking may have an influence instead,
on disease severity in males. Interestingly, the risk of
smoking for Graves’ disease in females may show
marked age stratification (15).
4. McKenzie J.M. Does LATS cause hyperthyroidism in
Graves’ disease? (A review biased towards the affirmative).
Metabolism 1972, 21: 883-894.
5. Wartofsky L. Low remission after therapy for Graves’ dis-
ease. Possible relation of dietary iodine with antithyroid
therapy results. JAMA 1973, 226: 1083-1088.
6. Schleusener H., Schwander J., Fischer C., et al. Prospective
multicentre study on the prediction of relapse after an-
tithyroid drug treatment in patients with Graves’ disease.
Acta Endocrinol. (Copenh.) 1989, 120: 689-701.
7. Solomon B.L., Evaul J.E., Burman K.D., Wartofsky L.
Remission rates with antithyroid drug therapy: continuing
influence of iodine intake? Ann. Intern. Med. 1987, 107:
510-512.
8. Allannic H., Lorcy Y., Leguerrier A.M., et al. Synthetic an-
tithyroid drugs and Basedow’s disease or the choice of a
therapeutic strategy. Presse Med. 1991, 20: 645-646.
9. Cooper D.S. Antithyroid drugs for the treatment of hyper-
thyroidism caused by Graves’ disease. Endocrinol. Metab.
Clin. North Am. 1998, 27: 225-247.
10. Vitti P., Rago T., Chiovato L., et al. Clinical features of pa-
tients with Graves’ disease undergoing remission after an-
tithyroid drug treatment. Thyroid 1997, 7: 369-375.
11. Classification of eye changes of Graves’ disease. Thyroid
1992, 2: 235-236.
12. Tellez M., Cooper J., Edmonds C. Graves’ ophthalmopa-
thy in relation to cigarette smoking and ethnic origin. Clin.
Endocrinol. (Oxf.) 1992, 36: 291-294.
The findings of the current study and the caveats
raised by it suggest that a prospective study of
smoking habits in patients with Graves’ disease and
its impact on disease severity might be worthwhile.
13. Yoshiuchi K., Kumano H., Nomura S., et al. Stressful life
events and smoking were associated with Graves’ disease
in women, but not in men. Psychosom. Med. 1998, 60:
182-185.
REFERENCES
14. Winsa B., Adami H.O., Bergstrom R., et al. Stressful life
events and Graves’ disease. Lancet 1991, 338: 1475-
1479.
1. Bartalena L., Pinchera A., Marcocci C. Management of
Graves’ ophthalmopathy: Reality and perspectives. Endocr.
Rev. 2000 21: 168-199.
15. Ericsson U.B., Lindgarde F. Effects of cigarette smoking
on thyroid function and the prevalence of goiter, thyro-
toxicosis and autoimmune thyroiditis. J. Intern. Med. 1991,
229: 67-71.
2. Stenszky V., Balazs C., Kozma L., et al. Identification of sub-
sets of patients with Graves’ disease by cluster analysis.
Clin. Endocrinol. 1983, 18: 335-345.
157