Subclinical thyroid
disorders:
still a matter of controversy
Simon HS Pearce
Plan
• Background
• Subclinical hypothyroidism
-Vascular risk
• Subclinical hyperthyroidism
-Understand the pathophysiology
-Approach to Management
What is normal?
• 16 healthy individuals, having monthly TFTs for 1 year
• People stick to their own “reference” interval
• Extrapolating to Free T4 values -setpoint +/- ~2.5 pmol/l
• “My normal range is different from yours”
Andersen et al. JCEM 2002
TSH in centenarians and offspring
Δ 232 Ashkenazim, age
97
o 366 of offspring, age 69
 177 age-matched
controls
Atzmon et al. JCEM 2009
Lancet 1971; I: 203
Possible mechanisms
• Dyslipidaemia
• Cardiac systolic & diastolic dysfunction
• Hypertension
• Endothelial dysfunction
• Hypercoagulability
Hard outcomes
• Rotterdam Heart Study
– Community-based cross sectional survey
– 1149 women (mean age 69 +/-7 yrs)
– 10.8% had “subclinical hypothyroidism” (TSH>4.0, N FT4)
– Odds ratio for MI= 2.3 (CI; 1.3-4.0)
– OR for aortic atherosclerosis 1.7 (1.1-2.6)
– Population attributable risk of TSH to MI estimated
to be 14%
• N.B. Diabetes 14%, Smoking 15%
Hak et al. Ann Intern Med 2000;132: 270
Meta-summary of meta-analyses
• Relative risks (5-95% confidence intervals)
Author
Number Cardiovascular Cardiovascular All cause
events
mortality
mortality
Singh 2008
13,267
1.53 (1.31–1.79)
1.28 (1.02–1.60)
1.12 (0.99-1.26)
Ochs 2008
14,449
1.20 (0.97-1.49)
1.18 (0.98-1.42)
1.12 (0.99-1.26)
NI
NI
1.22 (0.95-1.57)
Haentjens 2008 14,619
Razvi 2008
29,022
1.23 (1.02– 1.48)
1.09 (0.84 –1.41)
NI
Rodondi 2010
55,287
1.18 (0.99- 1.42)
1.14 (0.99- 1.32)
1.09 (0.96-1.24)
Thvilum 2012
35,740
NI
NI
1.17 (1.00-1.37)
Meta-summary of meta-analyses
• Relative risks (5-95% confidence intervals)
Author
Number Cardiovascular Cardiovascular All cause
events
mortality
mortality
Singh 2008
13,267
1.53 (1.31–1.79)
1.28 (1.02–1.60)
1.12 (0.99-1.26)
Ochs 2008
14,449
1.20 (0.97-1.49)
1.18 (0.98-1.42)
1.12 (0.99-1.26)
NI
NI
1.22 (0.95-1.57)
Haentjens 2008 14,619
Razvi 2008
29,022
1.23 (1.02– 1.48)
1.09 (0.84 –1.41)
NI
Rodondi 2010
55,287
1.18 (0.99- 1.42)
1.14 (0.99- 1.32)
1.09 (0.96-1.24)
Thvilum 2012
35,740
NI
NI
1.17 (1.00-1.37)
All cause mortality in SCH
M Thvilum, F Brandt, TH Brix & L Hegedüs. Nat Rev Endocrinol 2012
Janus response: Age
•Thanks to Stefano Mariotti & David Cooper
Meta-analysis
• Performed by Salman Razvi/ Abdul Shakoor
• Longitudinal or cross sectional studies of
independent community-based subjects
• 14 studies fitted stringent criteria
• 2,531 SCH participants
• 26,491 euthyroid individuals
• Divided studies according to age of inclusion
• <65 yr vs 65 and above: median 60 & 74 yr
IHD prevalence in cross-sectional studies of
SCH & euthyroid controls
Younge
r
Older
IHD incidence in longitudinal studies of
SCH & euthyroid controls
Younger
Older
Cardiovascular mortality in longitudinal
studies of SCH & euthyroid controls
Younger
Older
Summary
• Prevalent and incident IHD, and IHD mortality is
increased in SCH compared to euthyroid
population
• Evidence of increased IHD confined to studies that
have included people aged less than 65 years
Razvi et al. JCEM 2008
Patient-level analysis
• 55,287 participants; 3,450 with SCH (6.2%)
• Information derived from 11 studies
• 9664 deaths; 2168 from CHD
• SCH defined as TSH 4.5-19.99 mU/l (N FT4)
Rodondi et al. JAMA 2010
Patient-level analysis: TSH
Patient-level analysis: TSH
Patient-level analysis: Age
Interim Summary
• Meta-analysis with many thousands of patient
events shows vascular death is associated
with SCH
• Effect is greater at higher TSH levels,
reaching significance at TSH of 7.0 mU and
above
• Effect is attenuated at older ages
UK General Practice Research
Database
• Primary care resource linking ~10 million patient
records, labs, prescriptions & death certificates
• During 2001 there were 322,291 TSH measurements
• Identified 4,735 people >40 yrs with TSH 5.0- 10.0
mU/l, normal FT4
• Excluded individuals on L-T4, ATDs, previous thyroid
disease, previous IHD, stroke, other vascular disease
Razvi S et al. Arch Intern Med 2012
UK General Practice Research
Database
• Participants followed until March 2008 (median 7.6 yrs)
• People aged 40- 70 yrs (n=3093) and >70 yrs (n=1642)
• 52.9% and 49.9% were treated with L-thyroxine during follow
up (Primary Care decision)
• Analysed outcomes for incident IHD, vascular and all cause
mortality over follow up period (Cox regression MVA)
L-Thyroxine treated group
• 94% of people continued to take L-T4
• Median dose 75μg (12.5-175 μg) daily
Untreated group
• 1.3% developed overt hypothyroidism
-(TSH >10, or  FT4)
• 58% remained with elevated TSH
• 38% reverted to euthyroidism
• 2.5% developed low TSH
Baseline characteristics
40-70 yrs
>70 yrs
Untreated
L-T4 Rx
Untreated
L-T4 Rx
Number
1459
1634
823
819
Age
55.9 ± 8.3
55.9 ± 8.4
79.9 ± 6.5
79.4 ± 6.2
Females
82.5%
87.4%
75.6%
84.6%
Serum TSH (mU/l)
6.3 ± 1.3
6.7 ± 1.4
6.3 ± 1.2
6.8 ± 1.4
Serum FT4 (pM)
13.4 ± 4.4
12.9 ± 3.0
14.6 ± 4.4
13.9 ± 3.4
BMI (Kg/m2)
27.8 ± 5.9
28.1 ± 6.2
25.4 ± 4.6
26.3 ± 5.1
Systolic BP (mmHg)
136.5 ± 20.0 135.2 ± 19.3 149.4 ± 23.5
149.4 ±
22.0
T Cholesterol (mM)
5.86 ± 1.34
5.82 ± 1.21
5.93 ± 1.36
5.95 ± 1.25
Diabetes
18.0%
18.1%
26.9%
26.6%
Smokers (current)
18.3%
17.9%
10.9%
10.1%
Deprivation index
17.5
16.75
15.86
16.58
GP contacts/yr
1.2
1.3
2.3
2.4
Fatal & non-fatal vascular events
40-70 yrs
HR 0.61 (0.39- 0.95); p=0.02
All cause mortality
40-70 yrs
HR= 0.36 (0.19 – 0.66) ; p<0.001
Fatal & non-fatal IHD events
>70 yrs
HR 0.99 (0.59- 1.33); p=0.56
All cause mortality
>70 yrs
HR= 0.71 (0.56 – 1.08) ; p=0.11
Event rate stratified by age
• LT4 vs untreated; Fatal + non fatal CV events
Degree of serum TSH elevation
• Median serum TSH 6.6 mU/l
• Reference group (HR=1) is untreated patients
Hazard Ratio for vascular events P value for trend
TSH 6.6 or less
TSH > 6.6
40-70 yrs
0.62 (0.39-0.96)
0.41(0.26-0.81)
0.007
>70 yrs
1.02 (0.66-1.82)
1.19 (0.74-1.80)
NS
Razvi et al. Arch Intern Med; 2012
Atrial fibrillation
Hazard Ratio for AF/ month 5-95% CI
L-T4 exposure
40-70 yrs
0.998
0.995- 1.001
>70 yrs
1.000
0.999- 1.001
Summary
• L-T4 treatment of SCH was associated with a
lower CV mortality and CV event rate in
patients <70 yrs
• Importantly, L-T4 treatment was not
associated with AF
• Not an RCT study, but represents outcome of
real-life practice
Razvi et al. Arch Intern Med 2012
Who should we treat?
• Pregnant patients, or planning pregnancy
• Patients with serum TSH > 10.0 mU/l
Who should we consider treating?
•
•
•
•
•
Symptoms or signs of hypothyroidism
Age less than 70 yrs
TSH >7.0 mU/l
Goitre
High vascular risk including
– Ischaemic heart disease
– Diabetes
– Dyslipidaemia
• 380 attendees at ITC 2010
• Electronic voting system
• Female, serum TSH 6.8
Pearce, Wemeau, Vaisman. Eur Thyroid J 2012
Subclinical
hyperthyroidism
What is normal in extreme
old age?
• Age-related decline in median TSH levels (ill people excluded)
Mariotti et al. JCEM 1993
What is normal in extreme
old age?
• Age related decline in T3 levels (ill people excluded)
• FT4 (and TT4) levels remain constant
Mariotti et al. JCEM 1993
Magri et al. 2002
Change to function of HPT axis
• Reduced hepatic thyroid hormone clearance
-glucuronidation, sulfation
• Reduced T4 to T3 conversion
• Reduced type 1 deiodinase activity
• Blunted diurnal TSH secretion
• Flattened TSH response to TRH
Subclinical Hyperthyroidism
Degrees of hyperthyroidism
Degrees of hyperthyroidism
12 months follow up
76% returned to normal
87% remained <0.1
Parle JV et al. 1991 Clin Endo
Prevalence
Both grades
• 1-3% of elderly subjects
in NHANESIII
• 2.1 % in Colorado Health
Fair study
Suppressed TSH
NHANES III
• ~0.7% of TFTs from
people not on T4 at RVI
Evidence
Should we be concerned about
subclinical hyperthyroidism?
Small risk of progression
to overt disease
• Parle et al. 1991
TSH <0.1
2%/ year
• Wiersinga et al. 1995
5%/ year
• Pirich et al. 2000
7%/ year
TSH <0.1
• Schouten et al. 2011
5-8%/ year
• Rosario et al. 2010 TSH 0.1-0.4
1% /year
AF in Framingham survey
TSH
(mU/l)
<0.1
0.1- 0.4
>5.0
0.4- 5.0
Sawin et al. NEJM; 1994
Cardiovascular Health Study
• 3233 US community dwelling individuals over 65, mean age 73
• AF rate 2.0 (CI 1.3-3.0) in Sub Hyper
Cappola et al. JAMA 2006
Overall survival
“Circulatory” survival
• Community-living >60 year olds; overt thyroid disease excluded
Parle et al. Lancet 2001
TSH <0.3mU/l
Normal TSH
TSH>4.8mU/l
• n=558 • Birth cohort design • All 85 yrs at baseline • Leiden, NL
• Hazard Ratio per 2.71 mU/l increase in TSH is 0.77 (0.63-0.94)
Gussekloo et al. JAMA 2004; 292:2591
CHD (fatal & non-fatal)
• 20 yr follow up of population survey, Western Australia;
n=2108
• Mean age 51 (17-89); subclin hypER 1.8%
• No effect of subclinical hyperthyroidism
Busselton Health Study; Arch Intern Med 2005
Meta-analysis of 10 cohort
studies
• 52,000 participants (2188 with SH)
Collet et al. Arch Intern Med 2012
Summary of observational studies
• Increased incidence of AF in SH
• Increased vascular mortality in SH groups in
most, but not all studies
• SH sounds like bad news for your heart
Functional cardiac effects of
subclinical hyperthyroidism
•
•
•
•
•
•
•
Resting tachycardia
LV hypertrophy
Increase LV mass index
Increase cardiac workload
Diastolic dysfunction (impaired relaxation)
Increased systolic function at rest
Impaired systolic response to excercise
Biondi, Kahaly, Klein and others
Non-vascular effects of SH
Effect
Bone mineral density Decreased
Reference
Mudde 1994, Faber
1998, Tauchmanovà
et al. 2004
Fracture
Increased x3 & 4.5;
hip & vertebral
Bauer et al. 2001
Muscle strength
Knee extension
Decreased by 30%
Increased x 3.5
Brennan et al. 2006
Dementia
Kalmijn et al. 2000
OK:
• Subclinical hyperthyroidism is
bad news
•Let’s treat everybody and make
them better
Problem?
No evidence that treatment is
effective
RCTs of radioiodine in SH
• Dutch trial: poor recruitment rate, other factors:
Trial terminated 2005
• UK Trial: poor recruitment rate (< 10% of SH
patients agreed to randomisation). Trial terminated
2009
• French trial: Prof. B Goichot,- ongoing- AF as
primary endpoint (target 300 patients)
Problem?
No evidence that treatment is
effective
Problems with study designs
• Participant cohort defined by a single TSH
measurement
• AF and mortality followed subsequently
• Many subjects (>50%) will have subsequently
normalised TSH levels during follow up
• How to interpret adverse outcome of low
TSH/ Subclinical hyperthyroidism group?
Formulation 1
• Low TSH represents true endogenous
hyperthyroidism
• Adverse cardiac events & AF are an expected
consequence of excess thyroid hormones
• Need to treat for hyperthyroidism
Formulation 2
• Low TSH represents an effect of ageing and reduced
turnover of pituitary-thyroid axis (biomarker for age)
• Decreased hepatic thyroid hormone clearance,
‘blunted’ diurnal TSH secretion & flattened TSH
response to TRH with ageing
• Adverse cardiac events & AF (+ OP & dementia)
simply are consequences of “biological age”
• No need to treat for hyperthyroidism
Formulation 1.5
• Both the previous suggestions are true
• Adverse cardiac effects are due to excess thyroid
hormones in some
• In others, low TSH is a biomarker for aging & hence
associated with poor outcome
• Need to distinguish between these two groups to
treat some for hyperthyroidism
Cardiovascular Health Study
TSH
N
AF rate/1000
pt yrs
mU/l
Euthyroid 2502
31
Hazard ratio (5-95%
CI)
<0.44
47
67
1.98 (1.29-3.03)
0.1-0.44
40
59
1.85 (1.14-3.00)
1.00 (ref)
• AF rate little different between
grade I and grade II SH
Cappola et al. 2006
TEARS study
• 2004 patients with SH vs 10,111 controls
• 2 TSH measurements 4/12 apart
• Median follow up 5.6 yrs
Vadiveloo et al. JCEM 2011
What to do?
• Observe 3 to 6 months
• Look out for drug/ contrast effects
-Thyroxine
-Opiates
-Glucocorticoid
-Metformin
-L-DOPA
-Amiodarone
-CT contrast (Iopaque)
What to do?
Intrinsic thyroid disease
Ageing process
• TSH <0.1 mU/l
TSH 0.1-0.4 mU/l •
• Age <75
Age >75 •
• Symptoms, goitre
No symptoms •
• Above median FT3, FT4
Below median FT3, FT4 •
• +ve Antibodies, nuclide scan -ve Antibodies, scan •
• Complications, AF
No complications •
Algorithm
* Rare causes
• Pituitary disease
• Tumour hCG
† Repeat testing
• May need investigations
• Maybe yearly re-testing
• Maybe never re-test
Mitchell & Pearce. Clin Endo 2010
Peer opinion: BTA Survey
“There is no reason to treat a patient with
subclinical hyperthyroidism for thyrotoxicosis,
provided they are in sinus rhythm”
Response s
Number of
responses
(%)
Strongly agree
7
(2.7)
Agree
64
(24.3)
Neutral
39
(14.9)
Disagree
114 (43.5)
Strongly disagree
32
(12.2)
Don’t know
6
(2.3)
Total
262 (100)
27% agree
55% disagree
Vaidya B et al. Clin Endo 2008
Summary
• Low serum TSH is common in advanced age
• In many, it is transient & in others may be a
feature of ageing
• The minority will require treatment
• Large evidence vacuum remains
Dr. Salman Razvi
Performed much of the hard work
Acknowledgements
TRISH Investigators and Steering committee
• Amit Allahabadia & Alison Mortimer
• Diana Elbourne, Jayne Franklyn, Malcolm Prentice, Graham
Williams, Janis Hickey, Murray Stewart, Wilmar Wiersinga
Newcastle Co-investigators
• Salman Razvi, Anna Mitchell, Joanna Collerton, Andrew
Kingston & 85+ Core study team
• BTA survey: Bijay Vaidya
& Newcastle Healthcare Charity
The End