Incidence and prevalence of thyroid dysfunction in type 1 diabetes Christa Nederstigt, Eleonora P.M. Corssmit, Eelco J.P. de Koning, Olaf M. Dekkers PII: DOI: Reference:
S1056-8727(16)00002-7 doi: 10.1016/j.jdiacomp.2015.12.027 JDC 6629
To appear in:
Journal of Diabetes and Its Complications
Received date: Revised date: Accepted date:
18 November 2015 30 December 2015 31 December 2015
Please cite this article as: Nederstigt, C., Corssmit, E.P.M., de Koning, E.J.P. & Dekkers, O.M., Incidence and prevalence of thyroid dysfunction in type 1 diabetes, Journal of Diabetes and Its Complications (2016), doi: 10.1016/j.jdiacomp.2015.12.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Incidence and prevalence of thyroid dysfunction in type 1 diabetes
b
Department of Internal Medicine, Section Endocrinology and Department of Clinical Epidemiology, Leiden
RI P
a
T
Christa Nederstigta,b,c , Eleonora P.M. Corssmita, Eelco J.P. de Koninga, Olaf M. Dekkersa,b,d
c
University Medical Center, Postbus 9600, 2300 RC Leiden, The Netherlands. Medisch Centrum Haaglanden, d
SC
Department of Internal medicine, Lijnbaan 32, 2512 VA Den Haag, The Netherlands. Department of Clinical
MA NU
Epidemiology, Aarhus University Hospital, Olof Palmes Allé 43-45, DK-8200 Aarhus, Denmark.
Corresponding author: C. Nederstigt MSc, MD, Leiden University Medical Center, Postbus 9600, 2300 RC Leiden, The Netherlands. e-mail: [email protected], fax +31 71 526 6994 Other Authors: [email protected], [email protected], [email protected]
ED
Short title: Thyroid dysfunction in type 1 diabetes
AC
CE
PT
Word count: 3399
1
ACCEPTED MANUSCRIPT
RI P
T
ABSTRACT
Aims: To estimate prevalence and incidence of auto-immune thyroid disease and thyroid auto-antibodies in an
SC
unselected cohort of patients with DM1, including stratification by age, gender and duration of diabetes.
MA NU
Methods: Patients with T1D visiting our outpatient clinic between 1995 and 2011 were included. We calculated the prevalence of AITD at first screening and estimated prevalence and incidence rates during follow-up. Results: A total of 1304 patients were included, 48.9% being female. Mean age of diabetes onset was 18.7 years. Of all patients without known thyroid disorder first screened for AITD, 10.3 % (n=104) was diagnosed with
ED
hypo- or hyperthyroidism. The average prevalence of AITD in our population was 112/1000 patients. We found 128 new cases of AITD, 101 cases of hypothyroidism and 27 of hyperthyroidism between 1995 and 2011 with
PT
accompanying incidences of 11.2/1000 person years (95%CI 9.5-13.4), 8.9/1000 person-years (95%CI, 7.3-10.8) and 2.4/1000 person-years (95%CI, 1.6-3.5), respectively. Age-stratified incidence of AITD was comparable at
CE
all ages in both males and females, with an approximately two times higher incidence in females.
AC
Conclusions: The incidence of AITD among T1D patients is high, but stable among all ages and independent of diabetes duration.
Key words: Endocrinology, Diabetes, Epidemiology, Thyroid disease.
2
ACCEPTED MANUSCRIPT 1. INTRODUCTION Type 1 diabetes (T1D) is caused by an auto-immune reaction against insulin producing pancreatic beta cells. It
T
is associated with higher risk of other auto-immune diseases such as auto-immune thyroid diseases (AITD) and
RI P
celiac disease.(Lorini et al., 1996, Radetti et al., 1995, Crone et al., 2003, Warncke et al., 2010, Shun et al., 2014) The prevalence of antibodies against thyroid tissue such as anti-thyroid peroxidase (TPO) antibodies is
SC
markedly higher (up till 25 percent) in patients with T1D (Bilimoria et al., 2003, Frohlich-Reiterer et al., 2008, Kakleas et al., 2009a, Radetti et al., 1995, Araujo et al., 2008, Holl et al., 1999, McCanlies et al., 1998, Betterle
MA NU
et al., 1984) compared to the general population (10 percent) (Hollowell et al., 2002). Of antibody positive T1D patients 3-55 percent will develop auto-immune thyroiditis (Frohlich-Reiterer et al., 2008, Umpierrez et al., 2003, Radetti et al., 1995, Dagdelen et al., 2009, Franzese et al., 2000). The larger part of these patients develop hypothyroidism, but hyperthyroidism (M. Graves), characterized by antibodies against the TSH
ED
receptor, is also associated with T1D.(Franzese et al., 2000, Prazny et al., 2005) Patients with T1D are often screened for auto-immune thyroid disease at regular intervals .(Frohlich-Reiterer et
PT
al., 2008, Prazny et al., 2005) Protocols for screening for AITD are generally based on measurement of TSH levels and anti-thyroid antibodies.
16, 17
These antibodies are considered the best predictor for development of
CE
auto-immune thyroiditis. Treatment however is based on thyroid stimulating hormone (TSH) levels, as the sole presence of antibodies has no clinical consequences yet. In the past decades clinicians tended to screen for
AC
AITD and antibodies more frequently and over time various screening strategies are proposed. (FrohlichReiterer et al., 2008, Kakleas et al., 2009a, Karavanaki et al., 2009, Kordonouri et al., 2002, Prazny et al., 2005, Perros et al., 1995) With regard to cost-effectiveness, and selection of patients with increased risk for AITD the optimal screening strategy is not known yet. (Shun et al., 2014) Although the prevalence of AITD in patients with T1D increases with age, it is not clear if risk to develop AITD depends on age or duration of diabetes. Detailed information on the incidence of AITD is of direct relevance for clinical practice and might guide the improvement of screening-strategies. In this study we aimed to estimate prevalence and incidence rates of auto-immune hypo- and hyperthyroidism in patients with diabetes type 1 in more detail. Secondly, we aimed to assess whether incidence rates depended on age, sex or duration of diabetes.
3
ACCEPTED MANUSCRIPT 2. SUBJECTS, MATERIALS AND METHODS
2.1 Study population:
T
We included all patients with type 1 diabetes who visited the outpatient clinic of internal medicine,
RI P
endocrinology or the pediatrics department of the Leiden University Medical Center in the Netherlands at least once between January 1995 and December 2010. To secure that visits were diabetes related, we excluded
SC
patients without any HbA1c measurement. We assessed clinical records of each patient, containing information on outpatient visits as well as medical correspondence and diagnostics. We extracted available information on
MA NU
the diagnosis of diabetes, the diagnosis of hypo- or hyperthyroidism, presence of other thyroid (related) diseases including use of (iodine containing) medication and diagnosis of multinodular goiter or thyroid carcinoma, diabetes related co-morbidity and death. Patients were excluded if the date of T1D diagnosis could not be retrieved (n=11).
ED
We collected laboratory measurements including thyroid function (FT4, TSH and TPO-ab) for all patients from
PT
January 1995 until December 2010. During that time interval thyroid function tests (TSH, FT4) were performed as part of a two-yearly routine screening protocol. This protocol included antibody measurements of associated
CE
diseases, including TPO-ab and TG-ab every two years, irrespective of results of TSH and FT4, although in clinical practice screening frequency varied. Besides thyroid function parameters, cholesterol levels and kidney
AC
function were measured. The study protocol was approved by the local ethical committee. 2.2 Disease definitions:
In our study T1D was defined as insulin dependence within one year after diabetes diagnosis combined with either diagnosis before the age of 25 years, diabetic ketoacidosis at onset, low fasting c-peptide shortly after diagnosis or GAD positivity. (Alberti and Zimmet, P. Z., 1998, American Diabetes Association, 2012, The expert committee on the diagnosis and classification of diabetes mellitus, 2003) Definition of subclinical hypothyroidism was elevated TSH values and free thyroxine (FT4) values within normal range. Clinical overt hypothyroidism was defined as elevated TSH and FT4 values below 10.0 pmol/l. Hypothyroidism with accompanying TPO-ab positivity was considered Hashimoto’s disease. Definitions of clinical overt and subclinical hyperthyroidism were decreased TSH values and FT4 values exceeding 24 pmol/l or decreased TSH
4
ACCEPTED MANUSCRIPT values with FT4 values in the normal range, respectively. Graves’ disease was defined as hyperthyroidism with positive antibodies against the TSH receptor. Definition of AITD included: 1. Patients with thyroid dysfunction and positive thyroid antibodies (Hashimotho and Graves’ disease) and 2. all patients with thyroid dysfunction
T
not attributable to an alternative (not auto-immune mediated) disease irrespective of antibody positivity,
RI P
because for a substantial part of the patients no antibody measurement was available in general or at
files, including reported lab results and/or clinical opinion.
MA NU
2.3 Laboratory assays:
SC
presentation of thyroid dysfunction. Diagnosis of AITD, before 1995 was based on information from clinical
Until 10-’97 TSH was measured by IFMA (Delfia, Perkin Elmer-Wallac Oy, Turku, Finland); reference values 0.4 to 4.0 mU/l. From 10-’97 a Modular Analytics E-170 system (Roche Diagnostic Systems, Nederland BV, Almere, The Netherlands) was used; reference values 0.3 to 4.8 mU/l. FT4 levels were measured with IMx until 10-’97
ED
(Abbott Laboratories, Abbott Park) and from 10-’97 with a Modular Analytics E-170 system (Roche Diagnostic Systems, Nederland BV, Almere, The Netherlands). Reference range: 10.0 to 24.0 pmol/l. Until 12-’05 TPO-ab
PT
were measured with an anti-TPO radioimmunoassay (RIA) kit (Brahms, Germany) and from 12-’05 with Immulite 2500 (Siemens Medical Solutions Diagnostics). Values >30.0 kU/L and >35.0 kU/L respectively were
CE
considered positive. Antibodies against the TSH receptor were initially measured with a radioimmunoassay
AC
(RIA) kit (Brahms, Germany) >10.0 IU/L and later with a competitive radioreceptor assay (TRAK assay Brahms, Germany) > 1.0 IU/L).
2.4. Statistical analysis: 2.4.1. Patient characteristics Patient characteristics were reported as proportions (binary data) or means with accompanying 95% confidence intervals (95% CI) or interquartile range (IQR). Follow-up ended at death, last HbA1C measurement or the occurrence of a thyroid disease other than AITD that potentially could affect thyroid function (for example operation for goiter).
5
ACCEPTED MANUSCRIPT
2.4.2. Prevalence
RI P
point divided by the total number of patients at risk at yearly intervals.
T
Prevalence of an outcome was calculated as number of patients with the outcome (AITD) at a specific time-
2.4.3. Yield of first screening for AITD
SC
We calculated how many patients screened for the first time for thyroid diseases had (sub)clinical thyroid disease. Patients diagnosed before 1995, or patients that had not been screened for thyroid disease during
MA NU
follow-up were excluded to estimate the yield of first time screening. 2.4.4. Incidence of AITD between 1995 and 2011
We estimated incidence rates for biochemically confirmed AITD in the period 1995-2011. Incidence rates were calculated as number of patients with newly diagnosed AITD per 1000 person years (py) at risk. Incidence
ED
calculations were restricted to patients without 1) a previous diagnosis of AITD before 1995 2) subclinical thyroid dysfunction at first TSH screening in order to obtain a cohort of patients at risk for the outcome at
PT
baseline. The second group was excluded because with diagnosis of subclinical disease at first screening (with usually absence of clinical symptoms) in strict sense it is unknown how long thyroid dysfunction was present
CE
and what the specific time point should be for calculating the incidence. We stratified incidence rates by age (ten year intervals), duration of diabetes (ten year intervals) and sex. We tested for trends of the stratified
AC
rates using a Mantel-Haenszel method. Hereby we calculated an approximate estimate of the rate ratio for increase of one decade of age or duration of diabetes. 2.4.5. Prevalence of auto-antibodies In the majority of the patients at least one TPO-ab and TG-ab measurement was done. We calculated prevalence of TPO-ab and TG-ab positivity in all patients with and without AITD. Data were analyzed using Stata Statistical Software: Release 12. College Station, TX: StataCorp LP. Figures were made using GraphPad Prism 5.00 (GraphPad Software, Inc.)
6
ACCEPTED MANUSCRIPT 3. RESULTS
A total of 1304 patients (638 female, 48.9%) diagnosed with type 1 diabetes was included. All patients were on
T
insulin treatment. Mean age at first diabetes related outpatient visit was 34 years (IQR 23.2 to 43.8), with an
RI P
average duration of diabetes of 15.3 years (IQR 3.7 to 23.9). Mean age at time of diabetes diagnosis was 18.7 year (IQR 10.3 to 25.3). Based on either results from biochemical screening during follow-up or information
SC
from medical records before 1995, 278 (21.3%) of patients had AITD (128 of which were new cases discovered during follow-up). Of these cases 218 had hypothyroidism (78.4 %) and 60 patients had hyperthyroidism (21.6
MA NU
%). Average duration of diabetes at time of AITD diagnosis was 17.9 years (IQR 7.5-27.8) and average age at AITD diagnosis was 38.4 years (IQR 26.6-49.7). Of the patients with thyroid dysfunction before the diabetes diagnosis (N=15), 6 had hypothyroidism and 9 hyperthyroidism. More female (n=180) than male patients (n=98) developed AITD (OR 1.9, 95%CI 1.5-2.5).
ED
The mean age at diabetes diagnosis was 22.6 years (CI95% 19.3-25.9) of patients who develop hyperthyroidism,
PT
which is slightly higher than patients without thyroid disease (18.2, 95%CI 17.6-18.9) and patients with
3.1. Prevalence
CE
hypothyroidism (19.8, 95%CI, 18.1-21.6).
Mean follow up time from January 1995 until December 2010 of included patients was 10.5 years (range 0 to
AC
16 years). Average population size was 851 patients with 98 cases of AITD on average per calendar year, resulting in an average prevalence of AITD of 115.2/1000 patients (11.5 %) in our population. 3.2. Yield of first screening for AITD A total of 1012 patients without known thyroid disorder were screened for AITD for the first time between 1995 and 2011. At first screening, 10.3 % (n=104) of T1D patients was diagnosed with AITD (fig. 1). Of these patients, 90 were diagnosed with hypothyroidism (16 clinically overt and 60 subclinical, in 14 cases FT4 was missing); in 14 patients hyperthyroidism was discovered (2 clinically overt and 8 subclinical, in 4 cases FT4 level was missing). Treatment with levothyroxine was started in case of clinical overt hypothyroidism (n=16) or complaints suspicious for impaired thyroid function and subclinical hypothyroidism or respecting the applicable
7
ACCEPTED MANUSCRIPT guidelines at the time (n=28). Mean TSH of treated patients with subclinical hypothyroidism was higher compared with patients that remained untreated (11.7 and 5.6 respectively, p=0.395).
T
3.3. Yield of sequential screening: hypothyroidism
RI P
We calculated incidences rates based on all new cases of AITD found by sequential screening between 1995 and 2011. We found 101 new cases of hypothyroidism (23 clinical overt) between 1995 and 2011 with an
SC
incidence rate of 8.9/1000 person-years (95%CI, 7.3-10.8) (table 1). TPO-ab were measured in 94 of these patients with hypothyroidism and positive in 65 patients (69%). In 18 patients (37.0%) TSH was elevated on
MA NU
only one occasion (with a normal FT4) and no therapy was started. Number of patients needed to screen yearly to diagnose one case of hypothyroidism was 113.
3.4. Yield of sequential screening: hyperthyroidism
We found 27 patients with newly diagnosed hyperthyroidism (15 clinical and 12 subclinical) between 1995 and
ED
2011 resulting in an incidence of 2.4 per 1000 person-years (95%CI, 1.6-3.5) (table 1). In 9 patients TSH was decreased with normal FT4 values on only one occasion, which improved spontaneously. In 14 patients with
PT
clinical overt disease TSH receptor antibodies (TSHra) were measured and positive in all cases. Number of
CE
patients needed to screen yearly to diagnose one case of hyperthyroidism a year is 421.
3.5. Incidence rates stratified by age and disease duration
AC
Incidence of AITD was similar across age categories and across categories of diabetes duration (fig. 2a-b). Increase of one decade of age or disease duration did not increase the incidence on thyroid disease with rate ratios of 1.00 (95%CI 0.99-1.02, p=0.52) and 1.00 (95%CI 0.99-1.02, p=0.66), respectively (tested for overall trend). We found no evidence of effect modification by gender.
3.6. Antibody measurements TPO-antibodies were measured in 80.8 % (n=1054) of the population. Of these patients 293 (27.8 %) tested positive on at least one occasion (193/531 female and 100/523 male, RR 1.9, 95%CI 1.5-2.3). Of TPO positive patients a significant larger proportion developed hypothyroidism (46.1 versus 8.9 %, P
S1056-8727(16)00002-7 doi: 10.1016/j.jdiacomp.2015.12.027 JDC 6629
To appear in:
Journal of Diabetes and Its Complications
Received date: Revised date: Accepted date:
18 November 2015 30 December 2015 31 December 2015
Please cite this article as: Nederstigt, C., Corssmit, E.P.M., de Koning, E.J.P. & Dekkers, O.M., Incidence and prevalence of thyroid dysfunction in type 1 diabetes, Journal of Diabetes and Its Complications (2016), doi: 10.1016/j.jdiacomp.2015.12.027
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Incidence and prevalence of thyroid dysfunction in type 1 diabetes
b
Department of Internal Medicine, Section Endocrinology and Department of Clinical Epidemiology, Leiden
RI P
a
T
Christa Nederstigta,b,c , Eleonora P.M. Corssmita, Eelco J.P. de Koninga, Olaf M. Dekkersa,b,d
c
University Medical Center, Postbus 9600, 2300 RC Leiden, The Netherlands. Medisch Centrum Haaglanden, d
SC
Department of Internal medicine, Lijnbaan 32, 2512 VA Den Haag, The Netherlands. Department of Clinical
MA NU
Epidemiology, Aarhus University Hospital, Olof Palmes Allé 43-45, DK-8200 Aarhus, Denmark.
Corresponding author: C. Nederstigt MSc, MD, Leiden University Medical Center, Postbus 9600, 2300 RC Leiden, The Netherlands. e-mail: [email protected], fax +31 71 526 6994 Other Authors: [email protected], [email protected], [email protected]
ED
Short title: Thyroid dysfunction in type 1 diabetes
AC
CE
PT
Word count: 3399
1
ACCEPTED MANUSCRIPT
RI P
T
ABSTRACT
Aims: To estimate prevalence and incidence of auto-immune thyroid disease and thyroid auto-antibodies in an
SC
unselected cohort of patients with DM1, including stratification by age, gender and duration of diabetes.
MA NU
Methods: Patients with T1D visiting our outpatient clinic between 1995 and 2011 were included. We calculated the prevalence of AITD at first screening and estimated prevalence and incidence rates during follow-up. Results: A total of 1304 patients were included, 48.9% being female. Mean age of diabetes onset was 18.7 years. Of all patients without known thyroid disorder first screened for AITD, 10.3 % (n=104) was diagnosed with
ED
hypo- or hyperthyroidism. The average prevalence of AITD in our population was 112/1000 patients. We found 128 new cases of AITD, 101 cases of hypothyroidism and 27 of hyperthyroidism between 1995 and 2011 with
PT
accompanying incidences of 11.2/1000 person years (95%CI 9.5-13.4), 8.9/1000 person-years (95%CI, 7.3-10.8) and 2.4/1000 person-years (95%CI, 1.6-3.5), respectively. Age-stratified incidence of AITD was comparable at
CE
all ages in both males and females, with an approximately two times higher incidence in females.
AC
Conclusions: The incidence of AITD among T1D patients is high, but stable among all ages and independent of diabetes duration.
Key words: Endocrinology, Diabetes, Epidemiology, Thyroid disease.
2
ACCEPTED MANUSCRIPT 1. INTRODUCTION Type 1 diabetes (T1D) is caused by an auto-immune reaction against insulin producing pancreatic beta cells. It
T
is associated with higher risk of other auto-immune diseases such as auto-immune thyroid diseases (AITD) and
RI P
celiac disease.(Lorini et al., 1996, Radetti et al., 1995, Crone et al., 2003, Warncke et al., 2010, Shun et al., 2014) The prevalence of antibodies against thyroid tissue such as anti-thyroid peroxidase (TPO) antibodies is
SC
markedly higher (up till 25 percent) in patients with T1D (Bilimoria et al., 2003, Frohlich-Reiterer et al., 2008, Kakleas et al., 2009a, Radetti et al., 1995, Araujo et al., 2008, Holl et al., 1999, McCanlies et al., 1998, Betterle
MA NU
et al., 1984) compared to the general population (10 percent) (Hollowell et al., 2002). Of antibody positive T1D patients 3-55 percent will develop auto-immune thyroiditis (Frohlich-Reiterer et al., 2008, Umpierrez et al., 2003, Radetti et al., 1995, Dagdelen et al., 2009, Franzese et al., 2000). The larger part of these patients develop hypothyroidism, but hyperthyroidism (M. Graves), characterized by antibodies against the TSH
ED
receptor, is also associated with T1D.(Franzese et al., 2000, Prazny et al., 2005) Patients with T1D are often screened for auto-immune thyroid disease at regular intervals .(Frohlich-Reiterer et
PT
al., 2008, Prazny et al., 2005) Protocols for screening for AITD are generally based on measurement of TSH levels and anti-thyroid antibodies.
16, 17
These antibodies are considered the best predictor for development of
CE
auto-immune thyroiditis. Treatment however is based on thyroid stimulating hormone (TSH) levels, as the sole presence of antibodies has no clinical consequences yet. In the past decades clinicians tended to screen for
AC
AITD and antibodies more frequently and over time various screening strategies are proposed. (FrohlichReiterer et al., 2008, Kakleas et al., 2009a, Karavanaki et al., 2009, Kordonouri et al., 2002, Prazny et al., 2005, Perros et al., 1995) With regard to cost-effectiveness, and selection of patients with increased risk for AITD the optimal screening strategy is not known yet. (Shun et al., 2014) Although the prevalence of AITD in patients with T1D increases with age, it is not clear if risk to develop AITD depends on age or duration of diabetes. Detailed information on the incidence of AITD is of direct relevance for clinical practice and might guide the improvement of screening-strategies. In this study we aimed to estimate prevalence and incidence rates of auto-immune hypo- and hyperthyroidism in patients with diabetes type 1 in more detail. Secondly, we aimed to assess whether incidence rates depended on age, sex or duration of diabetes.
3
ACCEPTED MANUSCRIPT 2. SUBJECTS, MATERIALS AND METHODS
2.1 Study population:
T
We included all patients with type 1 diabetes who visited the outpatient clinic of internal medicine,
RI P
endocrinology or the pediatrics department of the Leiden University Medical Center in the Netherlands at least once between January 1995 and December 2010. To secure that visits were diabetes related, we excluded
SC
patients without any HbA1c measurement. We assessed clinical records of each patient, containing information on outpatient visits as well as medical correspondence and diagnostics. We extracted available information on
MA NU
the diagnosis of diabetes, the diagnosis of hypo- or hyperthyroidism, presence of other thyroid (related) diseases including use of (iodine containing) medication and diagnosis of multinodular goiter or thyroid carcinoma, diabetes related co-morbidity and death. Patients were excluded if the date of T1D diagnosis could not be retrieved (n=11).
ED
We collected laboratory measurements including thyroid function (FT4, TSH and TPO-ab) for all patients from
PT
January 1995 until December 2010. During that time interval thyroid function tests (TSH, FT4) were performed as part of a two-yearly routine screening protocol. This protocol included antibody measurements of associated
CE
diseases, including TPO-ab and TG-ab every two years, irrespective of results of TSH and FT4, although in clinical practice screening frequency varied. Besides thyroid function parameters, cholesterol levels and kidney
AC
function were measured. The study protocol was approved by the local ethical committee. 2.2 Disease definitions:
In our study T1D was defined as insulin dependence within one year after diabetes diagnosis combined with either diagnosis before the age of 25 years, diabetic ketoacidosis at onset, low fasting c-peptide shortly after diagnosis or GAD positivity. (Alberti and Zimmet, P. Z., 1998, American Diabetes Association, 2012, The expert committee on the diagnosis and classification of diabetes mellitus, 2003) Definition of subclinical hypothyroidism was elevated TSH values and free thyroxine (FT4) values within normal range. Clinical overt hypothyroidism was defined as elevated TSH and FT4 values below 10.0 pmol/l. Hypothyroidism with accompanying TPO-ab positivity was considered Hashimoto’s disease. Definitions of clinical overt and subclinical hyperthyroidism were decreased TSH values and FT4 values exceeding 24 pmol/l or decreased TSH
4
ACCEPTED MANUSCRIPT values with FT4 values in the normal range, respectively. Graves’ disease was defined as hyperthyroidism with positive antibodies against the TSH receptor. Definition of AITD included: 1. Patients with thyroid dysfunction and positive thyroid antibodies (Hashimotho and Graves’ disease) and 2. all patients with thyroid dysfunction
T
not attributable to an alternative (not auto-immune mediated) disease irrespective of antibody positivity,
RI P
because for a substantial part of the patients no antibody measurement was available in general or at
files, including reported lab results and/or clinical opinion.
MA NU
2.3 Laboratory assays:
SC
presentation of thyroid dysfunction. Diagnosis of AITD, before 1995 was based on information from clinical
Until 10-’97 TSH was measured by IFMA (Delfia, Perkin Elmer-Wallac Oy, Turku, Finland); reference values 0.4 to 4.0 mU/l. From 10-’97 a Modular Analytics E-170 system (Roche Diagnostic Systems, Nederland BV, Almere, The Netherlands) was used; reference values 0.3 to 4.8 mU/l. FT4 levels were measured with IMx until 10-’97
ED
(Abbott Laboratories, Abbott Park) and from 10-’97 with a Modular Analytics E-170 system (Roche Diagnostic Systems, Nederland BV, Almere, The Netherlands). Reference range: 10.0 to 24.0 pmol/l. Until 12-’05 TPO-ab
PT
were measured with an anti-TPO radioimmunoassay (RIA) kit (Brahms, Germany) and from 12-’05 with Immulite 2500 (Siemens Medical Solutions Diagnostics). Values >30.0 kU/L and >35.0 kU/L respectively were
CE
considered positive. Antibodies against the TSH receptor were initially measured with a radioimmunoassay
AC
(RIA) kit (Brahms, Germany) >10.0 IU/L and later with a competitive radioreceptor assay (TRAK assay Brahms, Germany) > 1.0 IU/L).
2.4. Statistical analysis: 2.4.1. Patient characteristics Patient characteristics were reported as proportions (binary data) or means with accompanying 95% confidence intervals (95% CI) or interquartile range (IQR). Follow-up ended at death, last HbA1C measurement or the occurrence of a thyroid disease other than AITD that potentially could affect thyroid function (for example operation for goiter).
5
ACCEPTED MANUSCRIPT
2.4.2. Prevalence
RI P
point divided by the total number of patients at risk at yearly intervals.
T
Prevalence of an outcome was calculated as number of patients with the outcome (AITD) at a specific time-
2.4.3. Yield of first screening for AITD
SC
We calculated how many patients screened for the first time for thyroid diseases had (sub)clinical thyroid disease. Patients diagnosed before 1995, or patients that had not been screened for thyroid disease during
MA NU
follow-up were excluded to estimate the yield of first time screening. 2.4.4. Incidence of AITD between 1995 and 2011
We estimated incidence rates for biochemically confirmed AITD in the period 1995-2011. Incidence rates were calculated as number of patients with newly diagnosed AITD per 1000 person years (py) at risk. Incidence
ED
calculations were restricted to patients without 1) a previous diagnosis of AITD before 1995 2) subclinical thyroid dysfunction at first TSH screening in order to obtain a cohort of patients at risk for the outcome at
PT
baseline. The second group was excluded because with diagnosis of subclinical disease at first screening (with usually absence of clinical symptoms) in strict sense it is unknown how long thyroid dysfunction was present
CE
and what the specific time point should be for calculating the incidence. We stratified incidence rates by age (ten year intervals), duration of diabetes (ten year intervals) and sex. We tested for trends of the stratified
AC
rates using a Mantel-Haenszel method. Hereby we calculated an approximate estimate of the rate ratio for increase of one decade of age or duration of diabetes. 2.4.5. Prevalence of auto-antibodies In the majority of the patients at least one TPO-ab and TG-ab measurement was done. We calculated prevalence of TPO-ab and TG-ab positivity in all patients with and without AITD. Data were analyzed using Stata Statistical Software: Release 12. College Station, TX: StataCorp LP. Figures were made using GraphPad Prism 5.00 (GraphPad Software, Inc.)
6
ACCEPTED MANUSCRIPT 3. RESULTS
A total of 1304 patients (638 female, 48.9%) diagnosed with type 1 diabetes was included. All patients were on
T
insulin treatment. Mean age at first diabetes related outpatient visit was 34 years (IQR 23.2 to 43.8), with an
RI P
average duration of diabetes of 15.3 years (IQR 3.7 to 23.9). Mean age at time of diabetes diagnosis was 18.7 year (IQR 10.3 to 25.3). Based on either results from biochemical screening during follow-up or information
SC
from medical records before 1995, 278 (21.3%) of patients had AITD (128 of which were new cases discovered during follow-up). Of these cases 218 had hypothyroidism (78.4 %) and 60 patients had hyperthyroidism (21.6
MA NU
%). Average duration of diabetes at time of AITD diagnosis was 17.9 years (IQR 7.5-27.8) and average age at AITD diagnosis was 38.4 years (IQR 26.6-49.7). Of the patients with thyroid dysfunction before the diabetes diagnosis (N=15), 6 had hypothyroidism and 9 hyperthyroidism. More female (n=180) than male patients (n=98) developed AITD (OR 1.9, 95%CI 1.5-2.5).
ED
The mean age at diabetes diagnosis was 22.6 years (CI95% 19.3-25.9) of patients who develop hyperthyroidism,
PT
which is slightly higher than patients without thyroid disease (18.2, 95%CI 17.6-18.9) and patients with
3.1. Prevalence
CE
hypothyroidism (19.8, 95%CI, 18.1-21.6).
Mean follow up time from January 1995 until December 2010 of included patients was 10.5 years (range 0 to
AC
16 years). Average population size was 851 patients with 98 cases of AITD on average per calendar year, resulting in an average prevalence of AITD of 115.2/1000 patients (11.5 %) in our population. 3.2. Yield of first screening for AITD A total of 1012 patients without known thyroid disorder were screened for AITD for the first time between 1995 and 2011. At first screening, 10.3 % (n=104) of T1D patients was diagnosed with AITD (fig. 1). Of these patients, 90 were diagnosed with hypothyroidism (16 clinically overt and 60 subclinical, in 14 cases FT4 was missing); in 14 patients hyperthyroidism was discovered (2 clinically overt and 8 subclinical, in 4 cases FT4 level was missing). Treatment with levothyroxine was started in case of clinical overt hypothyroidism (n=16) or complaints suspicious for impaired thyroid function and subclinical hypothyroidism or respecting the applicable
7
ACCEPTED MANUSCRIPT guidelines at the time (n=28). Mean TSH of treated patients with subclinical hypothyroidism was higher compared with patients that remained untreated (11.7 and 5.6 respectively, p=0.395).
T
3.3. Yield of sequential screening: hypothyroidism
RI P
We calculated incidences rates based on all new cases of AITD found by sequential screening between 1995 and 2011. We found 101 new cases of hypothyroidism (23 clinical overt) between 1995 and 2011 with an
SC
incidence rate of 8.9/1000 person-years (95%CI, 7.3-10.8) (table 1). TPO-ab were measured in 94 of these patients with hypothyroidism and positive in 65 patients (69%). In 18 patients (37.0%) TSH was elevated on
MA NU
only one occasion (with a normal FT4) and no therapy was started. Number of patients needed to screen yearly to diagnose one case of hypothyroidism was 113.
3.4. Yield of sequential screening: hyperthyroidism
We found 27 patients with newly diagnosed hyperthyroidism (15 clinical and 12 subclinical) between 1995 and
ED
2011 resulting in an incidence of 2.4 per 1000 person-years (95%CI, 1.6-3.5) (table 1). In 9 patients TSH was decreased with normal FT4 values on only one occasion, which improved spontaneously. In 14 patients with
PT
clinical overt disease TSH receptor antibodies (TSHra) were measured and positive in all cases. Number of
CE
patients needed to screen yearly to diagnose one case of hyperthyroidism a year is 421.
3.5. Incidence rates stratified by age and disease duration
AC
Incidence of AITD was similar across age categories and across categories of diabetes duration (fig. 2a-b). Increase of one decade of age or disease duration did not increase the incidence on thyroid disease with rate ratios of 1.00 (95%CI 0.99-1.02, p=0.52) and 1.00 (95%CI 0.99-1.02, p=0.66), respectively (tested for overall trend). We found no evidence of effect modification by gender.
3.6. Antibody measurements TPO-antibodies were measured in 80.8 % (n=1054) of the population. Of these patients 293 (27.8 %) tested positive on at least one occasion (193/531 female and 100/523 male, RR 1.9, 95%CI 1.5-2.3). Of TPO positive patients a significant larger proportion developed hypothyroidism (46.1 versus 8.9 %, P
