years, the ULSAM study) the association between plasma (P)-PTH and HF hospitalization was investigated adjusted
for established HF risk factors (myocardial infarction, hypertension, diabetes, electrocardiographic left ventricular
hypertrophy, smoking, and hypercholesterolaemia) and variables reflecting mineral metabolism (S-calcium, S-phosphate,
P-vitamin D, S-albumin, dietary calcium and vitamin D intake, physical activity, glomerular filtration rate, and blood draw
season). During follow-up (median 8 years), 75 individuals were hospitalized due to HF. In multivariable Cox-regression
analyses, higher P-PTH was associated with increased HF hospitalization (hazard ratio for 1-SD increase of PTH, 1.41,
95% CI 1.12–1.77, P ¼ 0.003). Parathyroid hormone also predicted hospitalization in participants without apparent
ischaemic HF and in participants with normal P-PTH.
In a large community-based sample of elderly men, PTH predicted HF hospitalizations, also after accounting for estab-
lished risk factors and mineral metabolism variables.
Our data suggest a role for PTH in the development of HF even in the absence of overt hyperparathyroidism.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Parathyroid hormone † heart failure † community † prognosis
Plasma parathyroid hormone and risk of
congestive heart failure in the community
Emil Hagstro ¨m1,2*, Erik Ingelsson3,4, Johan Sundstro ¨m1,5, Per Hellman2,
Tobias E. Larsson6,7, Lars Berglund1, Ha ˚kan Melhus5, Claes Held1,
Karl Michae ¨lsson1,2, Lars Lind5, and Johan A¨rnlo ¨v3,8
1Uppsala Clinical Research Center, Uppsala University, 751 85 Uppsala, Sweden;2Department of Surgical Sciences, Uppsala University, Uppsala, Sweden;3Department of Public
Health and Caring Sciences/Geriatrics, Uppsala University, Uppsala, Sweden;4Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden;
5Department of Medical Sciences, Uppsala University, Uppsala, Sweden;6Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden;
7Department of Nephrology, Karolinska Institutet, Stockholm, Sweden; and8The School of Health and Social Studies, Dalarna University, Falun, Sweden
Received 8 June 2010; accepted 10 June 2010; online publish-ahead-of-print 26 August 2010
In experimental studies parathyroid hormone (PTH) has been associated with underlying causes of heart failure (HF)
such as atherosclerosis, left ventricular hypertrophy, and myocardial fibrosis. Individuals with increased levels of PTH,
such as primary or secondary hyperparathyroidism patients, have increased risk of ischaemic heart disease and HF.
Moreover, increasing PTH is associated with worse prognosis in patients with overt HF. However, the association
between PTH and the development HF in the community has not been reported.
In a prospective, community-based study of 864 elderly men without HF or valvular disease at baseline (mean age 71
The hyperparathyroid disorders, such as primary hyperparathyr-
oidism (pHPT) and secondary hyperparathyroidism (sHPT), are
characterized by elevated levels of parathyroid hormone (PTH),
osteoporosis and a disturbed mineral metabolism, but also with
an increased risk for cardiovascular morbidity and mortality.1,2
Heart failure (HF) is a major cause of morbidity and mortality
worldwide.3However, the potential role of elevated levels of
PTH in the development of HF has been less well studied.
Recent data support a causal role for PTH in the underlying
pathology leading to HF. Interestingly, previous experimental and
clinical studies suggest that elevated PTH levels may not only
promote vascular pathology leading to atherosclerosis and
ischaemic HF,4,5but also distinct cardiac pathology, such as myo-
cardial calcification,6fibrosis,7and hypertrophy,8that could lead
to non-ischaemic HF. Moreover, higher PTH levels have been
shown to predict HF morbidity in select patient groups such as
patients with overt hyperparathyroidism but without HF,1,9,10and
patients with symptomatic HF but without known mineral
* Corresponding author. Tel: +46 186110000, Fax: +46 18 50 66 38, Email: email@example.com
Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2010. For permissions please email: firstname.lastname@example.org.
European Journal of Heart Failure (2010) 12, 1186–1192
metabolism abnormalities.11,12However, no previous studies have
reported on PTH levels as a predictor for HF hospitalization in the
On the basis of previous data, we hypothesized that higher PTH
levels would be associated with higher risk for HF in the commu-
nity. Accordingly, the aim of the present study was to investigate
the association between PTH and HF hospitalization in a large
community-based sample of elderly men free from HF at baseline,
adjusting for established risk factors for HF and mineral metabolism
variables. Moreover, as a secondary aim, we investigated PTH as a
predictor of non-ischaemic HF.
The study is based on the Uppsala Longitudinal Study of Adult Men
(ULSAM), a health investigation aimed at identifying risk factors for car-
diovascular diseases.13All 50-year-old men, born between 1920 and
1924 living in Uppsala, Sweden, were invited to participate between
1970 and 1974. The present analysis was based on a follow-up investi-
gation, when the participants were ?71 years old (1991–95, n ¼
1221). We excluded 20 participants due to previously diagnosed HF
and 14 due to diagnosed valvular disease. Individuals with missing data
on established risk factors for HF or PTH values were also excluded
(n ¼ 323).Thus,864menwereeligiblefortheprimaryanalyses.Theinci-
dence of HF hospitalization was similar in the participants who were
excluded due to missing data [incidence rate per 100 person-years at
risk for participants not included in the study 1.07 (n ¼ 323), and for
thepresentstudysample(n ¼ 864)1.05].Analyseswerealsoperformed
in the following sub-groups: individuals without myocardial infarction
(MI) prior to the baseline examination (n ¼ 794) and participants
without pHPT or sHPT [plasma (P-) PTH ,6.8 pmol/l and serum (S-)
calcium ,2.6 mmol/l, n ¼ 771]. All participants gave written consent
and the Ethics Committee of Uppsala University approved the study.
Venous blood for biochemical analyses was drawn after an overnight
fast and stored at 2708C until analysis. Intact P-PTH was measured
with solid-phase two-site chemiluminescent immunoassay using an
Immulite 2500 (Diagnostics Product Corporation, Los Angeles, CA,
USA).14S-calcium, albumin, and phosphate were measured with spec-
trophotometry.14The instruments used for the biochemical analyses
were Hitachi 717 or 911 (Hitachi, Japan). P-25-OH vitamin D was
determined with high-performance liquid chromatography together
with atmospheric pressure chemical ionization and mass spectrometric
detection (Vitas A/S, Oslo, Norway).14Circulating NT-proBNP, tropo-
nin I, cystatin C, and C-reactive protein were measured as previously
described.15Glomerular filtration rate was calculated from S-cystatin
C results using the formula y ¼ 77.24x21.2623.16Fasting P-glucose
was measured by routine laboratory analysis. Insulin sensitivity was
determined by euglycaemic insulin clamp,1724-h ambulatory systolic,
and diastolic blood pressures were recorded as previously described.18
Urinary albumin excretion rate (UAER) was determined in the urine
collected during the night and analysed with a radioimmunoassay
method (Albumin RIA 100, Pharmacia, Uppsala, Sweden). Electrocardio-
graphic left ventricular hypertrophy (LVH) was defined as high-
amplitude R-waves according to the revised Minnesota code19together
with a left ventricular strain pattern.20Height, weight, body mass index
[weight (kg)/height2(m), BMI], electrocardiogram, and supine systolic
and diastolic blood pressures were measured under standardized
conditions.13Hypertension was defined as systolic blood pressure
≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or use of antihy-
pertensive medication. Diabetes mellitus was defined as fasting
P-glucose ≥7.0 mmol/l, 2 h post-load glucose levels ≥11.1 mmol/l, or
use of oral hypoglycaemic agents or insulin. Hypercholesterolaemia
was defined as total S-cholesterol ≥5 mmol/l or use of lipid-lowering
treatment. Smoking status and leisure time physical activity was obtained
from questionnaires. Presence of valvular disease (International Classifi-
cation of Disease [ICD]-9 codes 394–397 and 424 or [ICD-10] codes
I05–I08 and I34–I37) and prior MI ([ICD-9] code 410 or [ICD-10] code
I21) were assessed from the hospital discharge register.
Dietary intake of calcium and vitamin D were recorded using a 7-day
The Swedish Hospital Discharge Register was used to define HF
([ICD-8] codes 427.00, 427.10, and 428.99, [ICD-9] code 428, or
[ICD-10] code I50) or hypertensive heart disease with HF ([ICD-10]
code I11.0). Medical records from the relevant hospitalizations were
reviewed by two physicians (E.I. and L.L.), blinded to the baseline
data, who classified the cases as definite, questionable, or miscoded.
The classification relied on the definition proposed by the European
Society of Cardiology,21and the review process has previously been
described.22After validation, 75 definitive HF cases were included.
Sixty-three HF hospitalizations occurred in participants with normal
PTH and 12 HF hospitalizations occurred in those with elevated
PTH. We did not include fatal HF events in our endpoint, as the validity
of the HF diagnosis as a cause of death is uncertain and we wanted to
limit the risk for misclassification.
Logarithmic transformation was performed to achieve normal distri-
bution for skewed variables (PTH, phosphate, UAER, troponin I,
NT-proBNP, cystatin C, and C-reactive protein). The relation of
PTH to incidence of HF hospitalization was investigated in the
primary analysis using Cox proportional hazards regression in the
whole cohort, using the following four models:
(B) adjusted for established risk factors for HF (hypertension, prior
MI, electrocardiographic LVH, diabetes mellitus, smoking, BMI,
(P-25-OH vitamin D ,37.5 nmol/L)], glomerular filtration rate,
dietary intake of calcium and vitamin D, leisure time physical
activity, blood draw season (winter, summer). In the winter
season (November–April) vitamin D is synthesized at a lower
rate at high latitudes;23
(D) adjusted for established risk factors for HF (Model B) and vari-
ables associated with mineral metabolism (Model C). Owing to
the large number of covariates in this model, we also used pro-
pensity scores to support the validity of the findings in this par-
ticular model.24This approach has been suggested to be a good
alternative to control for confounding when there are seven or
fewer events per confounder.25
scores, defined as the conditional probability of having PTH
.5.23 or ≤5.23 pmol/L (quartile 4 vs. 1–3), were estimated by
multivariable logistic regression models.
The individual propensity
Parathyroid hormone was modelled as a continuous standardized
variable. We also used multi-category models comparing risk in
PTH and heart failure
PTH quartiles 2, 3, and 4 with that in quartile 1 (lowest) and
threshold models [modelled as quartile 4 vs. quartiles 1–3 (PTH .
In secondary analyses, we investigated whether the association
between PTH and HF was confounded by other distinct pathophysio-
logical pathways that are not primarily reflected by the established risk
factors: atrial fibrillation, inflammation (C-reactive protein), ventricular
dysfunction (NT-proBNP), renal dysfunction (cystatin C), myocardial
cell damage (troponin I) and kidney damage (UAER), insulin sensitivity,
and 24 h ambulatory systolic and diastolic blood pressures. We also
investigated the association between PTH and HF in individuals
without previous MI at baseline (n ¼ 794) or during follow-up, being
at risk until censored at the date of the first MI (n ¼ 739). Additionally,
as it is possible that some of the predictive information of the estab-
lished cardiovascular risk factors could be lost due to the categoriz-
hypercholesterolaemia/non-hypercholesterolaemia, we also replaced
the dichotomous hypertension variable with systolic and diastolic
blood pressures and anti-hypertensive treatment and the hypercholes-
lipid-lowering treatment. Moreover, as different antihypertensive treat-
ments may have different effects on the levels of PTH,26,27we also
replaced the antihypertensive treatment variable in the above model
with use of angiotensin-converting enzyme (ACE) inhibitors, alpha-
blockers, beta-blockers, calcium antagonists, or diuretics.
Test of linearity was assessed by adding a quadratic term of PTH to
multivariable model B. To gain additional insights into potential nonli-
nearity of the association between PTH and HF, we examined the Cox
regression model using penalized splines with five degrees of freedom
[knots at 5th (1.79 pmol/l), 35th (3.38 pmol/l), 65th (4.62 pmol/l), and
95th percentile (7.88 pmol/l) of PTH]. The proportional hazards
assumptionsfor the models
Nelson-Aalen plots and confirmed formally by Schoenfeld’s tests.
Additionally, we performed tests for effect modification for hyperten-
sion, prior MI, electrocardiographic LVH, diabetes mellitus, smoking,
and BMI by including multiplicative interaction terms with these
variables and PTH. None of the interaction terms reached statistical
significance [although the multiplicative interaction term for LVH was
borderline significant (P ¼ 0.063)].
In order to control for the potential inclusion of participants with
HF at baseline, additional analyses were performed where all subjects
who developed HF within 2 years from baseline were excluded from
The population-attributable risk (PAR) proportion was calculated as
p*[(HRexposed2 HRunexposed)/HRexposed] where p is the proportion of
cases that were exposed and the hazard ratio (HR) is adjusted for
the established cardiovascular risk factors, model B.24
P-values of ,0.05 from two-sided tests were considered statistically
significant. STATA 10.1 (Stata Corporation, College Station, TX, USA)
and SAS v. 9.1 (SAS Institute, Cary, NC, USA) were used.
The authors had full access to and take full responsibility for the
integrity of the data. All authors have read and agree to the manuscript
Baseline characteristics are shown in Table 1. During follow-up
(median 8.74 years, range 0.09–11.4 years), 75 participants were
hospitalized due to HF (incidence rate 1.04/100 person-years at
Body mass index (kg/m2)
S-high-density lipoprotein (mmol/l)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Intact P-PTH (pmol/l)*
Albumin-corrected S-calcium (mmol/l) 2.33+0.09 2.33+0.10
S-25-OH vitamin D (nmol/l)
Glomerular filtration rate
Dietary intake of calcium (mg/day)
Dietary intake of vitamin D (mg/day)
S-C-reactive protein (mg/l)*
P-troponin I (mg/l)*
P-NT pro-brain natriuretic peptide
S-cystatin C (mg/l)*
Microalbuminuria (20-200 UAER
Electrocardiographic left ventricular
Leisure time physical activity
Lipid lowering treatment
Table 1 Baseline characteristics
58 (6.7)53 (6.7)
62 (7.2) 49 (6.2)
Values are means (+SD) for normally distributed continuous, median (interquartile
ranges) for skewed (denoted with *), and n (%) for categorical variables. Conversion
factor for plasma PTH (from pmol/L to pg/mL): (pmol/L)/0.1053.
E. Hagstro ¨m et al.
risk, number at risk at baseline 864), and 47 developed HF without
prior MI. Baseline characteristics for participants with vs. without
HF are shown in Supplementary material online, Table S1.
Parathyroid hormone levels and heart
Cox regression models
In Cox proportional hazards analyses, 1 SD increase of P-PTH was
associated with a 32–43% higher risk for hospitalization in crude
analysis, as well as after adjusting for established risk factors for
HF, for factors associated with mineral metabolism or both
(models A–D, Table 2). Also, men in the highest quartile of
P-PTH had up to a three-fold higher risk of being hospitalized com-
pared with those in the first quartile and up to two-fold higher risk
for hospitalization compared with participants in quartiles 1–3 in
models A–D (Table 2, Figure 1). The associations between
P-PTH and HF in model D were similar when using propensity
score (HR for quartile 4 vs. quartiles 1–3: 2.07, 95% CI 1.23–
3.50, P ¼ 0.006). The relation between P-PTH and HF showed
no deviation from linearity when introducing a quadratic term of
PTH (P ¼ 0.37). Also, the regression spline graph suggests that
there appears to be a fairly linear increase in hazard for HF with
increasing PTH levels (Figure 2).
The results remained essentially the same in the sub-sample of
participants without prior MI (n ¼ 794, Model D: HR for 1 SD
increase 1.35, 95% CI 1.05–1.72, P , 0.05) or in the subsample
without prior MI and being at risk until censored at the time of
MI during follow-up (Model D: HR for 1 SD increase 1.33, 95%
CI 1.04–1.71, P , 0.05) as well as in participants without signs
of pHPT or sHPT (Model D: HR for 1 SD increase 1.84, 95% CI
1.19–2.84, P ¼ 0.006). Also, the results were unaltered when
excluding those who were hospitalized due to HF during the
initial 2 years after baseline (n ¼ 12, Model D: HR for 1 SD
increase 1.46, 95% CI 1.16–1.85, P ¼ 0.002).
The association between P-PTH and HF also remained significant
after the separate addition of other described risk factors for HF to
models B and D; atrial fibrillation, NT-proBNP, cystatin C, troponin
I, C-reactive protein, UAER, insulin sensitivity, and 24 h ambulatory
systolic and diastolic blood pressures (see Supplementary material
online, Table S2). Moreover, the results remained essentially the
same when we replaced the hypertension variable in model B with
systolic and diastolic blood pressures (continuous) and antihyper-
tensive treatment and replaced the hypercholesterolaemia variable
with total S-cholesterol (continuous) and lipid lowering medication
(HR for 1 SD increase 1.28, 95% CI 1.05–1.57, P ¼ 0.016) or when
we replaced antihypertensive treatment in the above model with
use of ACE inhibitors, alpha-blockers, beta-blockers, calcium
antagonists, or diuretics (HR for 1 SD increase 1.28, 95% CI 1.04–
1.57, P , 0.01, see Supplementary material online, Table S3).
Even though the multiplicative interaction term for LVH did not
formally reach statistical significance (P ¼ 0.06), we performed stra-
tified analyses in participants with and without LVH, as previous
experimental and clinical data indicate an important role for LVH
as a mediator of the association between PTH and HF. In this
1 SD increase1.32 (1.11–1.58)†
Q1 (,2.93 pmol/l)Referent Referent
Q2 (2.93–3.95 pmol/l)0.84 (0.39–1.81) 0.88 (0.40–1.94)
Q3 (3.96–5.23 pmol/l) 1.68 (0.86–3.28)1.54 (0.78–3.05)
Q4 (.5.23 pmol/l)2.01 (1.05–3.83)* 1.97 (1.00–3.86)*
Q4 vs. Q1–3 (.5.23 pmol/l) 1.72 (1.08–2.77)*1.70 (1.04–2.77)*
Table 2 Relations of plasma parathyroid hormone to heart failure hospitalization (n 5 864)
Model B, established
Model C, mineral
Model D, established risk
factors 1 mineral metabolism
Q, quartile. Data are Cox proportional hazards ratios (95% CI) for a 1 SD plasma parathyroid hormone increase in; unadjusted; adjusted for established risk factors for heart
failure; adjusted for factors associated with mineral metabolism; combination of multivariable models.
*P , 0.05;†P , 0.01;§P ¼ 0.052.
Figure 1 Nelson-Aalen plot of cumulative incidence rate of
heart failure hospitalization in the whole sample, by two groups
[parathyroid hormone quartiles 1–3 (solid line) vs. quartile 4
PTH and heart failure
stratified analysis, the association between P-PTH and HF appeared
stronger in participants with LVH when compared with those
without [HR, (Model B) with LVH (n ¼ 62) 1.67, 95% CI 1.07–
2.60, P ¼ 0.023; without LVH (n ¼ 802) 1.19, 95% CI 0.93–1.52,
P ¼ 0.17].
Population-attributable Risk Proportion
Elevated PTH (quartile 4 vs. quartile 1–3 cut-off, P-PTH
.5.23 pmol/l) accounted for 10% (95% CI 1–16%) of the PAR
for HF. Population-attributable risk for established risk factors
for HF and for recently described risk factors are shown in
Table 3 and Supplementary material online, Table S4.
In the present community-based sample of elderly men, PTH pre-
dicted HF hospitalization independently of both established risk
factors for HF and mineral metabolism factors. The results
remained robust in participants with normal levels of PTH or
without clinically overt pHPT or sHPT. Furthermore, PTH also
remained an independent risk factor for HF in the subgroup of indi-
viduals without previous MI at baseline or during follow-up, indicat-
ing that higher levels of PTH also predict non-ischaemic HF.
Elevated levels of PTH accounted for 10% of the PAR proportion
for HF, comparable to several other established risk factors such as
previous MI, diabetes mellitus, and smoking, as well as to more
recently described risk factors for HF.
Comparison with previous studies
The present findings extend results of previous clinical studies that
have reported associations between higher PTH and increased
incidence of HF in select patient groups, such as patients with
pHPT and sHPT,9,10or patients with clinically overt HF.12
However, our study is the first to report the prospective associ-
ation between PTH and HF incidence in the community. More-
over, we are aware of no study that has reported the association
between PTH and non-ischaemic HF.
Several mechanisms may explain the results in the present study:
First, higher PTH may promote specific vascular pathology such
as endothelial dysfunction and atherosclerosis4,5that leads to
cardiac ischaemia and consequent HF.
Second, both experimental and clinical data suggest that PTH
has direct detrimental myocardial effects that could lead to
increased myocardial susceptibility for cardiac ischaemia or an
increased risk for HF of non-ischaemic origin. For instance, myo-
cytes have been shown to be target cells for PTH and higher
PTH may induce myocyte hypertrophy, fibrosis, and LVH.7,8,28,29
Additionally, one community-based study found increased levels
of PTH to be a predictor for LVH in the elderly30and reversal
of LVH was seen after parathyroidectomy and renal transplantation
in both pHPT- and sHPT-patients.31–34The notion of a direct
detrimental myocardial effect of PTH is further supported by the
association between PTH levels and non-ischaemic HF in the
present study and that the association appeared stronger in partici-
pants with LVH (although this finding should be interpreted with
caution given the small number of participants with LVH).
Arguing against the notion of detrimental myocardial effects of
PTH are a few previous experimental studies reporting an inotro-
pic effect of PTH.35,36
Third, it is possible that individuals with higher PTH have
undiagnosed HF at baseline. In previous studies of patients with
primary9,37and secondary HPT10,38these disorders have been
associated with decreased left ventricular function, and with
Figure 2 Association between plasma parathyroid hormone
and heart failure hospitalization. Solid line show estimated
hazard ratios (with 95% confidence limits) for heart failure hospi-
talization in relation to plasma parathyroid hormone as a function
of penalized regression splines. Q, quartile.
Highest quartile of parathyroid hormone
Previous myocardial infarction
Electrocardiographic left ventricular hypertrophy
Table 3 Population-attributable risk proportions for
heart failure of different established risk factors for
Highest quartile of parathyroid hormone was defined as a value .5.23 pmol/L.
Diabetes mellitus wasdefined as fastingplasma glucose ≥7.0 mmol/l, 2 h post-load
glucose levels ≥11.1 mmol/l, or the use of oral hypoglycaemic agents or insulin.
Obesity was defined as a body mass index ≥30 kg/m2, calculated as [weight (kg)/
height2(m)]. Hypertension was defined as systolic blood pressure ≥140 mmHg
and/or diastolic blood pressure ≥90 mmHg and/or use of antihypertensive
medication. Electrocardiographic left ventricular hypertrophy was defined as
high-amplitude R-waves according to the revised Minnesota code together with a
left ventricular strain pattern. Hypercholesterolaemia was defined as total serum
cholesterol ≥5 (mmol/l) or the use of pharmacological treatment for
dyslipidaemia. Smoking status: current smoking vs. no-smoking.
E. Hagstro ¨m et al.
worsening HF.11It is also possible that the present association
between PTH and HF hospitalization could be a reflection of
altered mineral metabolism secondary to HF. In parallel with
declining levels of vitamin D, a progressive increase in PTH may
occur which could be mediated at different levels of the vitamin
D activating pathway; a lower nutritional intake of vitamin D in ill
patients limits the substrate for further activation of vitamin D, a
decreased sun light exposure in physically disabled HF patients
renders a lower skin conversion of vitamin D to active metab-
olite,39or renal dysfunction that leads to a decreased vitamin D-
activation. Yet, the fact that results remained similar after adjusting
for sun light exposure (estimated by time of year for biochemical
analysis and leisure-time physical activity), vitamin D-deficiency,
oral intake of vitamin D, and for kidney- and cardiac-dysfunction
markers, as well as after exclusion of those who developed HF
during the first 2 years after baseline, argues against reverse causa-
tion as an explanation of our findings.
Fourth, PTH per se and the hyperparathyroid diseases have in
multiple lines of evidence been associated with HF risk factors
such as coronary heart disease,9,10,40hypertension,2,9LVH,6,30,32,41
diabetes mellitus,42obesity,43dyslipidaemia,44C-reactive protein,45
and euglycaemic insulin clamp glucose disposal rate.46Thus, it is
possible that the association between PTH and HF can be explained
by some other causal factor. However, as our results were indepen-
dent of these factors, confounding does not appear to be the sole
explanation of our findings. Still, we cannot rule out the possibility
of residual confounding or confounding by unmeasured factors.
Despite the abundance of previous experimental and clinical data
suggesting a causal link between PTH and HF, we are aware of
no evidence suggesting that reducing the levels of PTH will
decrease HF risk. Thus, our data should not be construed as imply-
ing a direct benefit of a reduction of PTH levels. However, if vali-
dated, our data could motivate additional interventional studies
evaluating the potential in PTH-lowering in order to prevent the
development of HF. This may prove to be important given that
HF is a major and escalating public health problem with enormous
suffering for the individual and high costs for society.3
Strengths and Limitations
Strengths of this investigation include the large, homogenous,
community-based population, detailed characterization of potential
confounders, and long follow-up. Furthermore, the HF diagnosis
was validated in all cases decreasing the inclusion of false-positive
Limitations include the unknown generalizability to women or
other age- and ethnic groups as we only examined men of the
same age and ethnic background. Another limitation of the study
is that we used the presence of an interim MI during follow-up
as a proxy for coronary heart disease. Even though this is an estab-
lished method for examining ‘non-ischaemic’ HF47,48it would be
better to assess non-ischaemic HF in a more specific way, e.g. by
examining all subjects with coronary angiography. However, this
is not feasible in a community-based cohort. Also, since the HF
diagnosis was based on a review of medical records, it was not
possible to distinguish between systolic and diastolic HF because
echocardiography was not available at the time of diagnosis for
many of the cases. Thus, in our study it is not possible to
examine whether the impact of PTH is different for systolic vs.
diastolic HF. Additionally, in the present study, an ECG was
used to diagnose LVH which is a less sensitive method than
Milder, non-hospitalized cases of HF were not included in our
endpoint, which may be considered a limitation but would tend
to bias the results toward the null hypothesis.
Moreover, some of the multivariable models included many cov-
ariates that could lead to bias and uncertain risk estimates.25,49
However, the consistency of results in all models and sub-samples
and the fact that the risk estimates were similar when using pro-
pensity scores supports the validity of our results.
Higher PTH predicted HF hospitalizations in the community, even
in individuals without signs of a disturbed mineral metabolism.
Further studies are warranted to validate our findings and to elu-
cidate whether PTH is a modifiable risk factor.
Supplementary material is available at European Journal of Heart
This study was supported by The Swedish Research Council
(2006-6555), Swedish Heart-Lung foundation, Thure ´us foundation,
Lisa och Johan Gro ¨nbergs foundation, and Uppsala University.
Conflict of interest: L.L. reports grant support from AstraZeneca.
1. Block GA, Klassen PS, Lazarus JM, Ofsthun N, Lowrie EG, Chertow GM. Mineral
metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc
2. Garcia de la Torre N, Wass JA, Turner HE. Parathyroid adenomas and cardiovas-
cular risk. Endocr Relat Cancer 2003;10:309–322.
3. Berry C, Murdoch DR, McMurray JJ. Economics of chronic heart failure. Eur J
Heart Fail 2001;3:283–291.
4. Perkovic V, Hewitson TD, Kelynack KJ, Martic M, Tait MG, Becker GJ. Parathyroid
hormone has a prosclerotic effect on vascular smooth muscle cells. Kidney Blood
Press Res 2003;26:27–33.
5. Rashid G, Bernheim J, Green J, Benchetrit S. Parathyroid hormone stimulates
endothelial expression of atherosclerotic parameters through protein kinase
pathways. Am J Physiol Renal Physiol 2007;292:F1215–F1218.
6. Stefenelli T, Abela C, Frank H, Koller-Strametz J, Globits S, Bergler-Klein J,
Niederle B. Cardiac abnormalities in patients with primary hyperparathyroidism:
implications for follow-up. J Clin Endocrinol Metab 1997;82:106–112.
7. Amann K, Ritz E, Wiest G, Klaus G, Mall G. A role of parathyroid hormone for the
activation of cardiac fibroblasts in uremia. J Am Soc Nephrol 1994;4:1814–1819.
8. Liu X, Xie R, Liu S. Rat parathyroid hormone 1–34 signals through the MEK/ERK
pathway to induce cardiac hypertrophy. J Int Med Res 2008;36:942–950.
9. Vestergaard P, Mollerup CL, Frokjaer VG, Christiansen P, Blichert-Toft M,
Mosekilde L. Cardiovascular events before and after surgery for primary hyper-
parathyroidism. World J Surg 2003;27:216–222.
10. De Boer IH, Gorodetskaya I, Young B, Hsu CY, Chertow GM. The severity of sec-
ondary hyperparathyroidism in chronic renal insufficiency is GFR-dependent,
race-dependent, and associated with cardiovascular disease. J Am Soc Nephrol
11. Arakelyan KP, Sahakyan YA, Hayrapetyan LR, Khudaverdyan DN, Ingelman-
Sundberg M, Mkrtchian S, Ter-Markosyan AS. Calcium-regulating peptide
PTH and heart failure
hormones and blood electrolytic balance in chronic heart failure. Regul Pept 2007;
12. Sugimoto T, Tanigawa T, Onishi K, Fujimoto N, Matsuda A, Nakamori S,
Matsuoka K, Nakamura T, Koji T, Ito M. Serum intact parathyroid hormone
levels predict hospitalisation for heart failure. Heart 2009;95:395–398.
13. Byberg L, Siegbahn A, Berglund L, McKeigue P, Reneland R, Lithell H. Plasminogen
activator inhibitor-1 activity is independently related to both insulin sensitivity and
serum triglycerides in 70-year-old men. Arterioscler Thromb Vasc Biol 1998;18:
14. Hagstrom E, Hellman P, Larsson TE, Ingelsson E, Berglund L, Sundstrom J,
Melhus H, Held C, Lind L, Michaelsson K, Arnlov J. Plasma parathyroid
hormone and the risk of cardiovascular mortality in the community. Circulation
15. Zethelius B, Berglund L, Sundstrom J, Ingelsson E, Basu S, Larsson A, Venge P,
Arnlov J. Use of multiple biomarkers to improve the prediction of death from car-
diovascular causes. N Engl J Med 2008;358:2107–2116.
16. Larsson A, Malm J, Grubb A, Hansson LO. Calculation of glomerular filtration rate
expressed in mL/min from plasma cystatin C values in mg/L. Scand J Clin Lab Invest
17. Hagstrom E, Hellman P, Lundgren E, Lind L, Arnlov J. Serum calcium is
independently associated with insulin sensitivity measured with euglycaemic-
hyperinsulinaemic clamp in a community-based cohort. Diabetologia 2007;50:
18. Ingelsson E, Bjorklund-Bodegard K, Lind L, Arnlov J, Sundstrom J. Diurnal blood
pressure pattern and risk of congestive heart failure. J Am Med Assoc 2006;295:
19. Prineas R, Crow R, Blackburn H. The Minnesota Code Manual of Electrocardiographic
Findings: Standards and Procedures for Measurement and Classification. Bristol: John
20. Levy D, Larson MG, Vasan RS, Kannel WB, Ho KK. The progression from hyper-
tension to congestive heart failure. J Am Med Assoc 1996;275:1557–1562.
21. Guidelines for the diagnosis of heart failure. The Task Force on Heart Failure of
the European Society of Cardiology. Eur Heart J 1995;16:741–751.
22. Ingelsson E, Arnlov J, Sundstrom J, Lind L. The validity of a diagnosis of heart
failure in a hospital discharge register. Eur J Heart Fail 2005;7:787–791.
23. Macdonald HM, Mavroeidi A, Barr RJ, Black AJ, Fraser WD, Reid DM. Vitamin D
status in postmenopausal women living at higher latitudes in the UK in relation to
bone health, overweight, sunlight exposure and dietary vitamin D. Bone 2008;42:
24. Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable
fractions. Am J Public Health 1998;88:15–19.
25. Cepeda MS, Boston R, Farrar JT, Strom BL. Comparison of logistic regression
versus propensity score when the number of events is low and there are multiple
confounders. Am J Epidemiol 2003;158:280–287.
26. Perry HM 3rd, Jensen J, Kaiser FE, Horowitz M, Perry HM Jr, Morley JE. The
effects of thiazide diuretics on calcium metabolism in the aged. J Am Geriatr Soc
27. Seely EW, LeBoff MS, Brown EM, Chen C, Posillico JT, Hollenberg NK,
Williams GH. The calcium channel blocker diltiazem lowers serum parathyroid
hormone levels in vivo and in vitro. J Clin Endocrinol Metab 1989;68:1007–1012.
28. Nilsson IL, A˚berg J, Rastad J, Lind L. Left ventricular systolic and diastolic function
and exercise testing in primary hyperparathyroidism-effects of parathyroidect-
omy. Surgery 2000;128:895–902.
29. Bogin E, Massry SG, Harary I. Effect of parathyroid hormone on rat heart cells.
J Clin Invest 1981;67:1215–1227.
30. Saleh FN, Schirmer H, Sundsfjord J, Jorde R. Parathyroid hormone and left ven-
tricular hypertrophy. Eur Heart J 2003;24:2054–2060.
31. Stefenelli T, Mayr H, Bergler-Klein J, Globits S, Woloszczuk W, Niederle B.
Primary hyperparathyroidism: incidence of cardiac abnormalities and partial
reversibility after successful parathyroidectomy. Am J Med 1993;95:197–202.
32. Piovesan A, Molineri N, Casasso F, Emmolo I, Ugliengo G, Cesario F, Borretta G.
Left ventricular hypertrophy in primary hyperparathyroidism. Effects of successful
parathyroidectomy. Clin Endocrinol (Oxf) 1999;50:321–328.
33. Hara S, Ubara Y, Arizono K, Ikeguchi H, Katori H, Yamada A, Ogura Y, Murata H,
Mimura N. Relation between parathyroid hormone and cardiac function in long-
term hemodialysis patients. Miner Electrolyte Metab 1995;21:72–76.
34. Ferreira SR, Moises VA, Tavares A, Pacheco-Silva A. Cardiovascular effects of suc-
cessful renal transplantation: a 1-year sequential study of left ventricular mor-
phology and function, and 24-hour blood pressure profile. Transplantation 2002;
35. Ogino K, Burkhoff D, Bilezikian JP. The hemodynamic basis for the cardiac effects
of parathyroid hormone (PTH) and PTH-related protein. Endocrinology 1995;136:
36. Tastan I, Schreckenberg R, Mufti S, Abdallah Y, Piper HM, Schluter KD. Parathyr-
oid hormone improves contractile performance of adult rat ventricular cardio-
myocytes at low concentrations in a non-acute way. Cardiovasc Res 2009;82:
37. Nappi S, Saha H, Virtanen V, Limnell V, Sand J, Salmi J, Pasternack A. Left ventri-
cular structure and function in primary hyperparathyroidism before and after
parathyroidectomy. Cardiology 2000;93:229–233.
38. Nasri H, Baradaran A, Naderi AS. Close association between parathyroid
hormone and left ventricular function and structure in end-stage renal failure
patients under maintenance hemodialysis. Acta Med Austriaca 2004;31:67–72.
39. Zittermann A, Schleithoff SS, Koerfer R. Vitamin D insufficiency in congestive
heart failure: why and what to do about it? Heart Fail Rev 2006;11:25–33.
40. Kamycheva E, Sundsfjord J, Jorde R. Serum parathyroid hormone levels predict
coronary heart disease: the Tromsø Study. Eur J Cardiovasc Prev Rehabil 2004;
41. Randon RB, Rohde LE, Comerlato L, Ribeiro JP, Manfro RC. The role of second-
ary hyperparathyroidism in left ventricular hypertrophy of patients under chronic
hemodialysis. Braz J Med Biol Res 2005;38:1409–1416.
42. Procopio M, Magro G, Cesario F, Piovesan A, Pia A, Molineri N, Borretta G. The
oral glucose tolerance test reveals a high frequency of both impaired glucose tol-
erance and undiagnosed Type 2 diabetes mellitus in primary hyperparathyroidism.
Diabet Med 2002;19:958–961.
43. Kamycheva E, Sundsfjord J, Jorde R. Serum parathyroid hormone level is associ-
ated with body mass index. The 5th Tromsø study. Eur J Endocrinol 2004;151:
44. Hagstro ¨m E, Lundgren E, Lithell H, Berglund L, Ljunghall S, Hellman P, Rastad J.
Normalized dyslipidaemia after parathyroidectomy in mild primary hyperpar-
athyroidism: population-based study over five years. Clin Endocrinol (Oxf) 2002;
45. Øgard CG, Engelmann MD, Kistorp C, Nielsen SL, Vestergaard H. Increased
plasma N-terminal pro-B-type natriuretic peptide and markers of inflammation
related to atherosclerosis in patients with primary hyperparathyroidism. Clin
Endocrinol (Oxf) 2005;63:493–498.
46. Chiu KC, Chuang LM, Lee NP, Ryu JM, McGullam JL, Tsai GP, Saad MF. Insulin
sensitivity is inversely correlated with plasma intact parathyroid hormone level.
47. Chen YT, Vaccarino V, Williams CS, Butler J, Berkman LF, Krumholz HM. Risk
factors for heart failure in the elderly: a prospective community-based study.
Am J Med 1999;106:605–612.
48. Vasan RS, Larson MG, Benjamin EJ, Evans JC, Levy D. Left ventricular dilatation
and the risk of congestive heart failure in people without myocardial infarction.
N Engl J Med 1997;336:1350–1355.
49. Harrell FE Jr., Lee KL, Mark DB. Multivariable prognostic models: issues in devel-
oping models, evaluating assumptions and adequacy, and measuring and reducing
errors. Stat Med 1996;15:361–387.
E. Hagstro ¨m et al.