Clinical Study
W Wiersinga, M Žarković
and others
Graves’ orbitopathy predictive
score
178:6
635–643
Predictive score for the development or
progression of Graves’ orbitopathy in patients
with newly diagnosed Graves’ hyperthyroidism
Wilmar Wiersinga1,*, Miloš Žarković2,*, Luigi Bartalena3, Simone Donati4, Petros Perros5,
Onyebuchi Okosieme6, Daniel Morris7, Nicole Fichter8, Jurg Lareida8, Georg von Arx8, Chantal Daumerie9,
Maria-Christina Burlacu9, George Kahaly10, Susanne Pitz11, Biljana Beleslin2, Jasmina Ćirić2, Goksun Ayvaz12,
Onur Konuk13, Füsun Balos̜ Törüner12, Mario Salvi14, Danila Covelli14, Nicola Curro15, Laszlo Hegedüs16 and
Thomas Brix16 on behalf of EUGOGO (European Group on Graves’ Orbitopathy)
1
Department of Endocrinology and Metabolism, Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands,
Department of Endocrinology, School of Medicine, University of Belgrade, Belgrade, Serbia, 3Endocrine Unit, Ospedale di Circolo,
4
Department of Medical and Surgical Sciences, School of Medicine, University of Insubria, Varese, Italy, 5Department of Endocrinology,
Royal Victoria Infirmary, Newcastle upon Tyne, UK, 6Department of Endocrinology, Institute of Molecular and Experimental Medicine,
Cardiff University School of Medicine, Cardiff, UK, 7Cardiff Eye Unit, University Hospital of Wales, Cardiff, UK, 8Department of
Ophthalmology, Interdisciplinary Centre for Graves’ Orbitopathy, Olten, Switzerland, 9Department of Endocrinology, Université
Catholique de Louvain, Cliniques Universitaires Saint-Luc, Brussels, Belgium, 10Department of Medicine I, Johannes Gutenberg
University Medical Center, Mainz, Germany, 11Orbital Center, Ophthalmic Clinic, Buergerhospital, Frankfurt, Germany, 12Departments
of Endocrinology and Metabolism, 13Ophthalmology, Faculty of Medicine, Gazi University, Ankara, Turkey, 14Graves’ Orbitopathy Unit,
Department of Clinical Science and Community Health, Fondazione Ca’Granda IRCCS, University of Milan, Milan,
Italy, 15Department of Ophthalmology, Fondazione IRCCS Ca’Granda Ospedale Maggiore Policlinico, Milan, Italy, and
16
Department of Endocrinology and Metabolism, Odense University Hospital, University of Southern Denmark,
Correspondence
Odense, Denmark
should be addressed
European Journal of Endocrinology
2
to M Žarković
Email
milos.zarkovic@med.bg.ac.rs
*(W Wiersinga and M Žarković contributed equally to this work)
Abstract
Objective: To construct a predictive score for the development or progression of Graves’ orbitopathy (GO) in Graves’
hyperthyroidism (GH).
Design: Prospective observational study in patients with newly diagnosed GH, treated with antithyroid drugs (ATD) for
18 months at ten participating centers from EUGOGO in 8 European countries.
Methods: 348 patients were included with untreated GH but without obvious GO. Mixed effects logistic regression
was used to determine the best predictors. A predictive score (called PREDIGO) was constructed.
Results: GO occurred in 15% (mild in 13% and moderate to severe in 2%), predominantly at 6–12 months after start
of ATD. Independent baseline determinants for the development of GO were clinical activity score (assigned 5 points
if score > 0), TSH-binding inhibitory immunoglobulins (2 points if TBII 2–10 U/L, 5 points if TBII > 10 U/L), duration of
hyperthyroid symptoms (1 point if 1–4 months, 3 points if >4 months) and smoking (2 points if current smoker). Based
on the odds ratio of each of these four determinants, a quantitative predictive score (called PREDIGO) was constructed
ranging from 0 to 15 with higher scores denoting higher risk; positive and negative predictive values were 0.28
(95% CI 0.20–0.37) and 0.91 (95% CI 0.87–0.94) respectively.
Conclusions: In patients without GO at diagnosis, 15% will develop GO (13% mild, 2% moderate to severe) during
subsequent treatment with ATD for 18 months. A predictive score called PREDIGO composed of four baseline
determinants was better in predicting those patients who will not develop obvious GO than who will.
European Journal of
Endocrinology
(2018) 178, 635–643
www.eje-online.org
https://doi.org/10.1530/EJE-18-0039
© 2018 European Society of Endocrinology
Printed in Great Britain
Published by Bioscientifica Ltd.
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
Clinical Study
W Wiersinga, M Žarković
and others
European Journal of Endocrinology
Introduction
A subset of patients with Graves’ hyperthyroidism (GH)
also have Graves’ orbitopathy (GO) (1). Development of
GO is feared by patients because it substantially impairs
quality of life due to interference with visual functions
and unmistakable changes in appearance (2). Fortunately,
not all patients with Graves’ disease develop GO. At
the time of diagnosing GH, about 75% have no GO (3).
However, GO may become evident after starting treatment
for hyperthyroidism. In patients with newly diagnosed
GH presenting without obvious GO, it might be useful if
one could estimate the risk of developing GO. In patients
with sufficiently high risk, one may prefer treatment with
antithyroid drugs or total thyroidectomy rather than
radioactive iodine in view of the adverse effect of 131I on
the course of GO (4, 5, 6). In high-risk patients, one could
also envisage prophylactic treatment with selenium,
which does prevent worsening of GO in mild cases (7).
Therefore, a prospective study was initiated in patients
with newly diagnosed GH without overt GO, aiming at
constructing a predictive score for developing GO during
antithyroid drug treatment. Well-known risk factors for
GO are smoking (8) and high serum concentrations of
TSH receptor antibodies (9, 10). Age, gender and thyroid
function were also evaluated as putative predictors,
because older age, male sex and biochemically severe
hyperthyroidism have all been associated with more
severe GO (5, 11, 12).
Patients with overt GO were excluded, but mild eyelid
swelling and mild conjunctival redness were no reason
to exclude patients. The rationale behind this is that
minimal swelling and redness of the eyes in the absence of
other ophthalmological signs compatible with GO are not
necessarily diagnostic of GO: some swelling and redness
of the eyes are common in many subjects regardless of
Graves’ disease. Thus, minimal swelling and redness in
our study participants at baseline was considered as either
a finding unrelated to the diagnosis of Graves’ disease, or
representing very mild, almost subclinical GO. However,
the presence of even minimal swelling and redness score
2 points on the clinical activity score (CAS), a widely
used tool to assess activity of GO (13). That score by itself
should not be taken as evidence for the existence of GO
as it should be realized that the CAS is not specific for
the diagnosis of GO. Any inflammatory eye or orbital
disease may result in swollen, red or painful eyes, and
thus, in positive CAS scores. Therefore, it was decided
to include the evaluation of baseline CAS as a putative
Graves’ orbitopathy predictive
score
178:6
636
predictor of developing GO. Furthermore, the predictive
value of the Vancouver Orbitopathy Rule was assessed, a
short questionnaire for self-reporting of GO symptoms in
patients with a recent diagnosis of GH (14).
Subjects and methods
Patient recruitment
Patients were recruited from ten participating centers of
the European Group on Graves’ Orbitopathy (EUGOGO)
in the period May 2009–May 2014. Inclusion criteria were
untreated GH, absence of overt GO and planned treatment
with antithyroid drugs for 18 months. Definition of
GH was (a) decreased TSH, elevated FT4 and/or FT3 and
(b) diffusely enlarged thyroid gland (either by palpation
or ultrasonography) and/or homogeneous thyroid uptake
at scintigraphy. Antithyroid drugs could be administered
according to either the titration method or the block-andreplace regimen, in line with standard policies of each
participating center.
The mere presence of mild conjunctival redness
and/or mild eyelid swelling (according to Figs 6 and 7
respectively in the color atlas by Dickinson and Perros)
(15) was not a reason to exclude patients, provided
other signs suggesting GO were absent. Exclusion
criteria were (1) previous or planned treatment with
131I or thyroidectomy, (2) presence of GO, defined
as one or more of the following eye changes: (a) soft
tissue changes (moderate or severe eyelid/conjunctival
redness, moderate or severe eyelid/periorbital swelling)
as depicted in the color atlas (15); (b) proptosis above
the upper normal limit (Asians 18 mm, Caucasians
20 mm, blacks 22 mm); (c) diplopia (intermittent,
inconstant or constant); (d) decreased visual acuity
attributable to GO, (3) drugs interfering with natural
course of GO (e.g. steroids, cytokines, anticytokines,
thiazolidinediones, selenium), (4) drugs interfering
with thyroid function (e.g. amiodarone, lithium,
iodine supplements), (5) drug or alcohol abuse, (6) no
informed consent.
Approval of institutional review boards or local
ethical committees was not deemed necessary because
the study protocol did not require additional procedures
beyond those done in the delivery of usual care.
Nevertheless, consent has been obtained from each
patient after full explanation of the purpose and nature
of all procedures.
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
Clinical Study
W Wiersinga, M Žarković
and others
European Journal of Endocrinology
Putative determinants
Putative determinants for developing GO were assessed at
baseline before starting antithyroid drugs. They included
age and gender, family history of autoimmune thyroid
disease, other autoimmune diseases in the patient, duration
of hyperthyroid symptoms until start of antithyroid drugs,
biochemical severity of hyperthyroidism (TSH, FT4, FT3),
immunological severity (TBII, TPOAb), smoking behavior
(never smoker, ex-smoker, current smoker), CAS and
the Vancouver Orbitopathy Rule (14). The Vancouver
Orbitopathy Rule consists of five questions: (1) do you have
redness in your eyes or eyelids? (2) do you have swelling
or feeling of fullness in one or both of your upper eyelids,
(3) do you have bags under the eyes? (4) do your eyes seem
to be too wide open? (5) is your vision blurry, even with
glasses/contacts? If answers to questions 2 or 3 was ‘yes’
and in addition if any of the answers 1, 4 or 5 was ‘yes’,
the result is labeled positive (sensitivity 0.76 and specificity
0.82 for the diagnosis of GO in patients with GH) (14).
Study protocol
In this prospective cohort study, baseline assessment of
eye changes and putative determinants were performed
mostly by endocrinologists. Soft tissue changes were
recorded according to the color atlas (15), lid aperture
in mm in midline by ruler, proptosis in mm by Hertel
exophthalmometer; CAS scores were at a scale from 0 to
7 (13). Patients were asked about double vision (none,
intermittent, inconstant, or constant), decreased visual
acuity and disturbed color vision (if present, patient was
referred to the ophthalmologist). Antithyroid drugs were
started after blood sampling.
Follow-up visits at 6–12–18 months included blood
sampling and re-assessment of smoking behavior and eye
changes, again by the endocrinologist.
End-points of study were development of GO
as defined under exclusion criterion 2 (GO was also
diagnosed if proptosis values had increased by more
than 2 mm) or otherwise at 18 months when antithyroid
drugs were discontinued. Premature stops could happen
if 131I therapy or thyroidectomy was performed earlier
than 18 months or in case of severe intercurrent illness.
Discontinuation of antithyroid drugs before 18 months or
pregnancy (n = 2) were no reason to exclude patients.
Statistical analysis
Assays of thyroid function and thyroid antibodies were
done at each center in the local laboratory. As different
Graves’ orbitopathy predictive
score
178:6
637
assays with different reference values were used, all
obtained results were standardized by dividing obtained
results by the upper normal limit of that assay, and
then multiplied by the most common upper limit of the
reference range. Normally distributed data are reported
as mean ± s.d., while data not normally distributed are
reported as median with range. Categorical variables are
recorded as numbers and percentage. TBII was measured
by second generation assays.
The objective of the statistical analysis was to
identify determinants of developing GO and to construct
a predictive score (called PREDIGO). Comparisons were
made between patients who did or did not develop GO
during the 18-month follow-up period. Odds ratios for
developing GO were calculated for each of the putative
determinants assessed at baseline, using univariate
logistic regression. For multivariate analysis, mixed
effects logistic regression was performed to find which
factors were independently related to GO. R and Ime4
software were used (16, 17). As fixed effects, clinical,
laboratory and demographic data were entered, and as
a random effect intercept for center. Because of the wide
distribution of the data, logarithmic transformations for
TSH, FT4, TPOAb, TBII and duration of hyperthyroid
symptoms were used. To find the optimal model,
multiple models were created and compared. P values
were obtained by likelihood ratio tests of the model
with the effect in question against the model without
the effect in question. 95% confidence intervals for
effects were calculated using the Wald method. For
the predictive score, significant parameters were used
as obtained by the logistic regression. Number of score
points assigned to each parameter was calculated using
the Framingham method (18). The sum of the score
points constituted the PREDIGO score. Optimal cut-off
for the PREDIGO score was obtained by analysis of the
ROC curve and minimizing distance between ROC plot
and point (0, 1) (19). The characteristic of the predictive
score was assessed by bootstrapping.
Results
392 patients were recruited. Eleven patients were
subsequently excluded (9 opted for thyroidectomy
and 1 for radioactive iodine, and 1 patient suffered a
stroke). Another 33 patients were lost to follow-up,
leaving 348 patients for analysis. Characteristics of the
study population are listed in Table 1. Two patients
became pregnant during the study period; one of them
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
Clinical Study
W Wiersinga, M Žarković
and others
Graves’ orbitopathy predictive
score
638
178:6
Table 1 Baseline characteristics of 348 patients with Graves’ hyperthyroidism treated with antithyroid drugs for 18 months: 295
Age†
Female/male gender
Caucasian/other ethnicity
Family history of AITD
Other AI disease in patient
Never/ex/current smokers
Pack-years*
Cigarettes/day†
Duration of symptoms months§,*
Antithyroid drugs: titration/block-and-replace
TSH (U/L)*
FT4 (pmol/L)*
FT3 (pmol/L)*
TPO (kU/L)*
TBII (U/L)*
CAS score (0/1/2)
Vancouver Orbitopathy Rule negative/positive
Lid aperture (mm)†
Proptosis (mm)†
348 patients
295 patients without
GO group A (85%)
53 patients with new
GO group B (15%)
42.9 ± 13.2 year
82%/18%
94%/6%
20%
6%
58%/13%/29%
15 (7.5–24)
12.7 ± 7.0
3 (1.5–5.25)
80%/20%
0.01 (0.00–0.02)
36.6 (27.2–51.5)
13.6 (8.2–22.1)
200 (46–1000)
6.8 (3.8–13.4)
89%/10%/<1%
98%/2%
9.5 ± 1.7
16.9 ± 2.2
42.4 ± 13.1 year
84%/16%
93%/7%
19%
6%
60%/14%/26%
15 (7.3–22.9)
11.5 ± 5.9
3 (1–5)
80%/20%
0.01 (0.00–0.02)
36.4 (27.1–50.2)
13.6 (8.1–22.0)
206 (47–1060)
6.4 (3.7–12.8)
92%/7%/<1%
99%/1%
9.5 ± 1.8
16.8 ± 2.2
45.7 ± 13.1 year
75%/25%
96%/4%
26%
4%
43%/8%/49%
20 (15.5–32.5)
16.0 ± 8.6
4 (2–6.25)
88%/12%
0.01 (0.00–0.02)
39.4 (29.4–59.4)
13.3 (8.6–23.1)
164 (17–600)
11.0 (5.1–21.5)
73%/25%/2%
94%/6%
9.6 ± 1.7
17.3 ± 2.2
P value
(A vs B)
0.087
0.169
0.550
0.321
0.750
0.002
0.051
0.022
0.007
0.179
0.463
0.102
0.689
0.141
0.006
<0.001
0.075
0.806
0.028
Hyperthyroid symptoms until start of antithyroid drugs; †mean ± s.d.; *median values with interquartile range.
AI, autoimmune; AITD, autoimmune thyroid disease.
§
CAS items was as follows: mild eyelid swelling 10, plica/
caruncle swelling 0, chemosis 0, spontaneous pain 4,
pain on eye movement 3, mild redness of conjunctiva
7, redness of eyelids 14. One patient with CAS 2 had
spontaneous pain and pain on movement, while the
15
20
Varese
10
Cardiff
5
Brussels
Milan
0
developed GO at 18 months. GO developed in 53 (15%)
of the 348 patients, which was mild in 46 patients
(13%) and moderate to severe in 7 patients (2%). Their
characteristics at baseline and at time of GO diagnosis
are shown in Supplementary Table (see section on
supplementary data given at the end of this article).
GO developed in 26 patients at 6 months (7.4%), in
18 patients at 12 months (5.1%), and in 9 patients at
18 months (2.5%) follow-up respectively. There was a
direct relationship between the number of patients who
developed GO and the number of patients recruited by
each individual center (Fig. 1). Mixed model analysis
also confirmed homogeneity between centers with
respect to patient populations. Baseline characteristics
of the 295 patients who did not develop GO and the
53 patients who developed GO are given in Table 1.
Patients who developed GO tended to be older, had
longer duration of hyperthyroid symptoms, were more
often current smokers, consumed more cigarettes per
day and had a higher number of pack-years. Serum TBII
were also higher in patients who developed GO, but
otherwise the biochemical severity of hyperthyroidism
and choice for titration or block-and-replace regimen
were similar in both groups. GO was further associated
with a positive Vancouver Orbitopathy Rule and higher
CAS scores at baseline. CAS >0 at baseline was observed
in 36 patients: CAS of 1 was present in 34 patients and
CAS of 2 in 2 patients. The distribution of the various
Nr. of patients with de novo GO per center
European Journal of Endocrinology
patients remained without GO and 53 patients developed GO during 18 months of follow-up.
Olten
Mainz
Odense
Belgrade
Ankara
0
20
Newcastle
40
60
80
Nr. of patients recruited per center
Figure 1
Direct relationship between number of patients recruited per
center and number of patients developing Graves’
orbitopathy (GO) per center (r = 0.85, P = 0.002) (AITD,
autoimmune thyroid disease; AI, autoimmune; §, hyperthyroid
symptoms until start of antithyroid drugs; †, mean ± s.d.; *,
median values with interquartile range).
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
other with CAS 2 had mild eyelid swelling and mild
redness of conjunctiva. Patients with scores 1 or 2 were
collapsed in a single category (CAS > 0). GO developed
in 12.5% of 312 patients with baseline CAS of 0, and
in 39% of 36 patients with baseline CAS ≥ 1 (P < 0.001).
Development of GO was thus associated with older
age, current smoking, longer duration of hyperthyroid
symptoms, higher TBII, higher CAS and a positive
Vancouver Orbitopathy Rule.
Using mixed-effect logistic regression analysis, four
variables were identified, which were independent of
each other and predicted development or progression to
obvious GO: CAS > 0, current smoking, TBII and duration
of hyperthyroid symptoms; their odds ratios are listed
in Table 2. In the multivariate analysis, age and the
Vancouver Orbitopathy Rule were no longer independent
determinants of GO. For each of the ten centers,
probability of developing GO for the four variables is
presented in Fig. 2.
A predictive score (Prediction of Graves’ Orbitopathy
(PREDIGO)) was constructed based on the four
independent variables. Points were assigned to the
presence or absence of each variable, and the sum of the
points provides the numerical predictive score that ranges
from 0 to 15 (Table 3). Scores >6 have some predictive
value for developing GO, whereas scores of ≤6 are not
predictive of developing GO. Sensitivity of the predictive
score is 0.56 (95% CI: 0.42–0.70), specificity 0.75
(0.70–0.79), positive predictive value 0.28 (0.20–0.37) and
negative predictive value 0.91 (0.87–0.94). The area under
the ROC curve of the predictive score is 0.71 (0.63–0.789)
(Fig. 3). The proportion of patients who developed GO
Table 2 Baseline variables predicting development of GO in
patients with Graves’ hyperthyroidism treated for 18 months
with antithyroid drugs.
Graves’ orbitopathy predictive
score
178:6
639
among all patients with a particular predictive score is
depicted in Fig. 4.
Discussion
Rate of GO development after the onset of
Graves’ hyperthyroidism
In this multicenter study, 15% of patients with newly
diagnosed GH but without overt GO, developed GO
Ankara
Belgrade
Brussels
Cardiff
Mainz
Milan
Newcastle
Odense
Olten
Varese
0.8
0.6
0.4
Probability of GO
W Wiersinga, M Žarković
and others
0.2
0.0
0.8
0.6
0.4
0.2
0.0
0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
TBII (U/L)
Smoking
no
yes
Ankara
Belgrade
Brussels
Cardiff
Mainz
Milan
Newcastle
Odense
Olten
Varese
0.6
0.4
Probability of GO
European Journal of Endocrinology
Clinical Study
0.2
0.0
0.6
0.4
0.2
Baseline variable
predicting GO
CAS 0
CAS ≥ 1
TBII < 2 (U/L)
TBII 2–10 (U/L)
TBII > 10 (U/L)
Duration hyperthyroid
symptoms
<1 month
1–4 months
>4 months
Current smoker no
Current smoker yes
Univariate odds ratio
(95% CI)
1.00
4.45 (2.11–9.42)
1.00
1.82 (0.93–3.54)
3.75 (1.57–9.00)
P value
0.0
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
<0.01
Symptom duration (months)
CAS Present
0.08
<0.01
0
>0
Figure 2
Probability of developing Graves’ orbitopathy in patients with
1.00
1.62 (0.82–3.19)
3.35 (1.47–7.64)
1.00
2.77 (1.53–5.05)
Graves’ hyperthyroidism during 18-month treatment with
0.16
<0.01
<0.01
antithyroid drugs per participating center, as a function of TBII
and current smoking (top panels) and as a function of
duration of hyperthyroid symptoms and clinical activity score
(bottom panels).
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
W Wiersinga, M Žarković
and others
Table 3 Predictive score for development of GO (PREDIGO)
Graves’ orbitopathy predictive
score
178:6
640
1
Clinical Study
in patients with Graves’ hyperthyroidism treated with
TBII
Duration of hyperthyroid
symptoms
Current smoking
Cut-off
Score
0
≥1
<2 (U/L)
2–10 (U/L)
>10 (U/L)
<1 months
1–4 months
>4 months
No
Yes
0
5
0
2
5
0
1
3
0
2
0.2
CAS
Sensitivity
0.4
0.6
Baseline variable
0.8
antithyroid drugs for 18 months.
European Journal of Endocrinology
during a follow-up of 18 months. The orbitopathy
was mild in 13%, moderate to severe in 2% and
occurred predominantly at 6–12 months after starting
antithyroid drugs. Differences between centers did not
seem to play a role for the frequency of GO, as the
number of patients who developed GO was directly
related to the number of patients recruited from that
center (Fig. 1).
The data are remarkably similar to those of a recent
single-center Italian study from Varese published in 2013
(3). In that study, 194 patients with GH but without overt
GO were followed for 18 months during/after treatment
with antithyroid drugs. GO developed in 12.9% (mild
GO in 10.3%, moderate-to-severe GO in 2.6%), and most
appeared 6–12 months after the start of treatment. At the
time of diagnosis of hyperthyroidism, GO was absent in
74% and present in 26%. New GO following the diagnosis
of GH occurred in 9.5% (0.129 × 74%), rendering a total
GO frequency of 35.5% (26% + 9.5%) in a population
of GH. GO was diagnosed at the same time as GH in
73% (26 out of 35.5), whereas GO onset followed that of
hyperthyroidism in 27% (9.5 out of 35.5). These figures
are at variance with reports in the older literature, stating
that GO develops after, simultaneously with or before
GH in about 47, 38 and 15% respectively (20, 21, 22).
The rates of GO cases developing after diagnosis of GH
are apparently much higher in the past (47%) than
at present (27%). Earlier diagnosis and treatment of
hyperthyroidism, identification of the risks conferred by
131I therapy and post-radioiodine hypothyroidism and
focus on the detrimental effects of smoking have likely
contributed to this secular decline. The lower prevalence
and incidence of late-onset GO seems to be the result of
effective primary and secondary prevention of GO (23).
0
Maximum score 15.
0
0.2
0.4
0.6
0.8
1 − Specificity
1
Figure 3
Receiver-operator curve with confidence intervals of the
PREDIGO score for predicting development of Graves’
orbitopathy in patients with Graves’ hyperthyroidism treated
for 18 months with antithyroid drugs.
Nowadays it is rare for a patient with GH who presents
without GO to contract moderate-to-severe GO later in
the course of the disease. Indeed, the clinical phenotype
of Graves’ disease is presently milder than that in the
past (24, 25).
Determinants of GO development
We identified four independent determinants as predictors
of GO development in patients with newly diagnosed
GH: CAS, TBII, current smoking and longer duration of
hyperthyroid symptoms. Associations between these items
and GO are known from previous studies (8, 9, 10, 11,
12). However, this is the first time the association of these
items with development of GO has been demonstrated
in a prospective manner in hyperthyroid patients who
had no overt GO at study entrance. It is not immediately
clear why a longer duration of hyperthyroid symptoms
is a predictor of GO. In a previous study among 251
consecutive patients with newly diagnosed GH, duration
of symptoms >4 months was associated with younger
age, higher TPOAb (thyroid peroxidase antibodies) levels,
larger goiter size, but not with higher TBII (TSH-binding
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
W Wiersinga, M Žarković
and others
Proportion of patients
at each score who developed GO
Clinical Study
100
75
50
25
0
0
5
10
15
PREDIGO Score
Figure 4
Predictive score for development of Graves’ orbitopathy in
patients with Graves’ hyperthyroidism treated with
European Journal of Endocrinology
antithyroid drugs for 18 months, called the PREDIGO
(PREDIctor of Graves’ Orbitopathy). Scores ≤6 are predictive of
not developing GO, whereas scores >6 have some predictive
value for developing GO.
inhibitor immunoglobulins) or higher prevalence of GO
(although the prevalence of GO was 19% at symptoms
duration ≤4 months and increased to 28% at a duration
≥12 months, the trend was not significant) (26). One may
hypothesize that longer duration of symptoms means
longer exposure to TBII, which is associated with larger
goiter size and by analogy may also lead to larger size of
extraocular muscles and orbital fat.
Biochemically more severe GH has been associated
with a higher risk of GO in a Swedish prospective
randomized clinical trial (5). The slightly higher baseline
serum FT4 in patients who developed GO in our study,
although not statistically significant, points in the same
direction.
Assessment of the extent of swelling and redness
of the eyes remains subjective, and minor degrees may
remain unnoticed and/or not necessarily indicate GO.
We tried to increase objectivity in deciding whether
eyelid swelling or conjunctival redness was absent or
mild (allowing inclusion), or moderate to severe (patient
excluded) with the help of a color atlas of eye changes
in GO (15). Retrobulbar pain, mild eyelid swelling and/
or mild conjunctival redness count in the CAS, which
proved to be one of the main independent predictors
of GO development. One may argue that these minor
degrees of swelling, redness and pain already indicated
the presence of GO at baseline. Even so, the PREDIGO
Graves’ orbitopathy predictive
score
178:6
641
score retains its value in predicting progression of very
mild (almost subclinical) GO to more overt GO. It is
interesting that the Vancouver Orbitopathy Rule (asking
among others also for redness in the eyes and redness
or fullness of eyelids) had some predictive value in
the univariate analysis, providing confirmation in an
independent cohort of the association between a positive
Vancouver Orbitopathy Rule and GO development (14).
Its question ‘Do your eyes seem to be wide open?’ may
reflect the slightly higher baseline values of lid aperture
and proptosis in our patients who developed GO.
Other putative determinants of developing GO in our
study could have been differences in the rate of control of
hyperthyroidism and the extent to which euthyroidism
was restored. Thyroid function tests were, however, not
different between groups at 6, 12 and 18 months during
treatment with antithyroid drugs (data not shown). There
were no differences with respect to development of GO
between patients who were treated with antithyroid drugs
according the block-and-replace regimen or the titration
regimen.
Quantitative predictive score for GO development
Based on the four independent determinants, we were
able to construct a quantitative predictive score for the
development of GO, called PREDIGO. The score ranges
from 0 to a maximum of 15. Low scores are associated
with a low probability of developing GO, and vice versa.
PREDIGO had a high negative predictive value of 0.91
but a rather low positive predictive value of 0.28. In
other words, it seems better at identifying those patients
who escape GO than those who will develop GO.
Improvement of the positive predictive value may have to
await the development of more sensitive biomarkers for
GO, which will likely depend on better insight into the
pathogenesis of the disease. It remains to be demonstrated
under which circumstances the score will be most useful
in the management of Graves’ disease patients. One may
consider studies to evaluate whether a high predictive
score would alter management, e.g. in refraining from
131I therapy or starting with selenium. In case of a high
predictive score, this could be used as a disease-specific
argument to convince the patient to stop smoking.
Strengths and limitations
Strengths of our study relate to its prospective nature,
its large sample size and the structured investigation
according to a detailed protocol. Eye changes were
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
Clinical Study
W Wiersinga, M Žarković
and others
recorded predominantly by endocrinologists and not by
ophthalmologists. Slit lamp investigations, which are
performed exclusively by ophthalmologists, are thus not
assessed in the present study. We view this as an advantage
rather than a disadvantage. For newly diagnosed patients
with GH without overt GO are nearly always managed by
internists/endocrinologists. Thereby applicability of the
results of this study for routine endocrine practice may
be inferred.
Supplementary data
This is linked to the online version of the paper at https://doi.org/10.1530/
EJE-18-0039.
European Journal of Endocrinology
Declaration of interest
The authors declare that there is no conflict of interest that could be
perceived as prejudicing the impartiality of this study.
Funding
This research did not receive any specific grant from any funding agency in
the public, commercial or not-for-profit sector.
References
1 Smith TJ & Hegedüs L. Graves’ disease. New England Journal
of Medicine 2016 375 1552–1565. (https://doi.org/10.1056/
NEJMra1510030)
2 Wiersinga WM. Quality of life in Graves’ ophthalmopathy. Best
Practice and Research Clinical Endocrinology and Metabolism 2012 26
359–370. (https://doi.org/10.1016/j.beem.2011.11.001)
3 Tanda ML, Piantanida E, Liparulo L, Veronesi G, Lai A, Sassi L,
Pariani N, Gallo D, Azzolini C, Ferrario M et al. Prevalence and
natural history of Graves’ orbitopathy in a large series of patients
with newly diagnosed Graves’ hyperthyroidism seen at a single
center. Journal of Clinical Endocrinology and Metabolism 2013 98
1443–1449. (https://doi.org/10.1210/jc.2012-3873)
4 Bartalena L, Marcocci C, Bogazzi F, Manetti L, Tanda ML,
Dell’Unto E, Bruno-Bossio G, Nardi M, Bartolomei MP, Lepri A et al.
Relation between therapy for hyperthyroidism and the course of
Graves’ ophthalmopathy. New England Journal of Medicine 1998 338
73–78. (https://doi.org/10.1056/NEJM199801083380201)
5 Tallstedt L, Lundell G, Tørring O, Wallin G, Ljunggren JG,
Blomgren H & Taube A. Occurrence of ophthalmopathy after
treatment for Graves’ hyperthyroidism. The Thyroid Study Group.
New England Journal of Medicine 1992 326 1733–1738. (https://doi.
org/10.1056/NEJM199206253262603)
6 Acharya SH, Avenell A, Philip S, Burr J, Bevan JS & Abraham P.
Radioiodine therapy (RAI) for Graves’ disease (GD) and the effect on
ophthalmopathy: a systematic review. Clinical Endocrinology 2008 69
943–950. (https://doi.org/10.1111/j.1365-2265.2008.03279.x)
7 Marcocci C, Kahaly GJ, Krassas GE, Bartalena L, Prummel M, Stahl M,
Altea MA, Nardi M, Pitz S, Boboridis K et al. Selenium and the course
of mild Graves’ orbitopathy. New England Journal of Medicine 2011
364 1920–1931. (https://doi.org/10.1056/NEJMoa1012985)
Graves’ orbitopathy predictive
score
178:6
642
8 Wiersinga WM. Smoking and thyroid. Clinical Endocrinology 2013 79
145–151. (https://doi.org/10.1111/cen.12222)
9 Eckstein AK, Plicht M, Lax H, Neuhäuser M, Mann K, Lederbogen S,
Heckmann C, Esser J & Morgenthaler NG. Thyrotropin
receptor autoantibodies are independent risk factors for Graves’
ophthalmopathy and help to predict severity and outcome of the
disease. Journal of Clinical Endocrinology and Metabolism 2006 91
3464–3470. (https://doi.org/10.1210/jc.2005-2813)
10 Lytton SD, Ponto KA, Kanitz M, Matheis N, Kohn LD & Kahaly GJ. A
novel thyroid stimulating immunoglobulin bioassay is a functional
indicator of activity and severity of Graves’ orbitopathy. Journal of
Clinical Endocrinology and Metabolism 2010 95 2123–2131. (https://
doi.org/10.1210/jc.2009-2470)
11 Perros P, Crombie AL, Matthews JNS & Kendall-Taylor P. Age and
gender influence the severity of thyroid-associated ophthalmopathy:
a study of 101 patients attending a combined thyroid-eye
clinic. Clinical Endocrinology 1993 38 367–372. (https://doi.
org/10.1111/j.1365-2265.1993.tb00516.x)
12 Kendler DL. The initial clinical characteristics of Graves’ orbitopathy
vary with age and sex. Archives of Ophthalmology 1993 111 197.
(https://doi.org/10.1001/archopht.1993.01090020051022)
13 Mourits MP, Prummel MF, Wiersinga WM & Koornneef L. Clinical
activity score as a guide in the management of patients with Graves’
ophthalmopathy. Clinical Endocrinology 1997 47 9–14. (https://doi.
org/10.1046/j.1365-2265.1997.2331047.x)
14 Mohaseb K, Linder M, Rootman J, Wilkins GE, Schechter MT,
Dolman PJ & Singer J. Development and validation of a patient
symptom questionnaire to facilitate early diagnosis of thyroidassociated orbitopathy in Graves’ disease. Orbit 2008 27 419–425.
(https://doi.org/10.1080/01676830802414566)
15 Dickinson AJ & Perros P. Controversies in the clinical evaluation
of active thyroid-associated orbitopathy: use of a detailed protocol
with comparative photographs for objective assessment. Clinical
Endocrinology 2001 55 283–303. (https://doi.org/10.1046/j.13652265.2001.01349.x)
16 R Core Team. R: A Language and Environment for Statistical Computing.
Vienna, Austria: R Foundation for Statistical Computing, 2017.
17 Bates D, Mächler M, Bolker B & Walker S. Fitting linear mixed-effects
models using lme4. Journal of Statistical Software 2015 67 1–48.
(https://doi.org/10.18637/jss.v067.i01)
18 Sullivan LM, Massaro JM & D’Agostino RB. Presentation of
multivariate data for clinical use: the Framingham Study risk score
functions. Statistics in Medicine 2004 23 1631–1660. (https://doi.
org/10.1002/sim.1742)
19 López-Ratón M, Rodríguez-Álvarez MX, Suárez CC & Sampedro FG.
OptimalCutpoints: an R package for selecting optimal cutpoints in
diagnostic tests. Journal of Statistical Software 2014 61 1–36. (https://
doi.org/10.18637/jss.v061.i08)
20 Wiersinga WM, Smit T, van der Gaag R & Koornneef L. Temporal
relationship between onset of Graves’ ophthalmopathy and onset of
thyroidal Graves’ disease. Journal of Endocrinological Investigation 1988
11 615–619. (https://doi.org/10.1007/BF03350193)
21 Marcocci C, Bartalena L, Bogazzi F, Panicucci M & Pinchera A.
Studies on the occurrence of ophthalmopathy in Graves’ disease.
Acta Endocrinologica 1989 120 473–478. (https://doi.org/10.1530/
acta.0.1200473)
22 Burch HB & Wartofsky L. Graves’ ophthalmopathy: current concepts
regarding pathogenesis and management. Endocrine Reviews 1993 14
747–793. (https://doi.org/10.1210/edrv-14-6-747)
23 Bartalena L. Prevention of Graves’ ophthalmopathy. Best Practice
and Research Clinical Endocrinology and Metabolism 2012 26 371–379.
(https://doi.org/10.1016/j.beem.2011.09.004)
24 Bartalena L, Masiello E, Magri F, Veronesi G, Bianconi E, Zerbini F,
Gaiti M, Spreafico E, Gallo D, Premoli P et al. The phenotype of
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access
Clinical Study
W Wiersinga, M Žarković
and others
newly diagnosed Graves’ disease in Italy in recent years is milder
than in the past: results of a large observational longitudinal study.
Journal of Endocrinological Investigation 2016 39 1445–1451. (https://
doi.org/10.1007/s40618-016-0516-7)
25 Perros P, Žarković M, Azzolini C, Ayvaz G, Baldeschi L, Bartalena L,
Boschi A, Bournaud C, Brix TH, Covelli D et al. PREGO (presentation
of Graves’ orbitopathy) study: changes in referral patterns to
Graves’ orbitopathy predictive
score
178:6
643
European Group on Graves’ Orbitopathy (EUGOGO) centres over the
period from 2000 to 2012. British Journal of Ophthalmology 2015 99
1531–1535. (https://doi.org/10.1136/bjophthalmol-2015-306733)
26 Vos XG, Smit N, Endert E, Tijssen JGP & Wiersinga WM. Variation
in phenotypic appearance of Graves’ disease: effect of genetic
anticipation and duration of complaints. European Journal of
Endocrinology 2009 161 113–118. (https://doi.org/10.1530/eje-08-0991)
European Journal of Endocrinology
Received 16 January 2018
Revised version received 21 March 2018
Accepted 11 April 2018
www.eje-online.org
Downloaded from Bioscientifica.com at 06/12/2020 01:22:44PM
via free access