Subclinical hyperthyroidism impacts left ventricular deformation: 2D and 3D echocardiographic study

Abstract

Background. We sought to investigate left ventricular (LV) function and mechanics assessed by three-dimensional echocardiography (3DE) and speckle tracking in patients with subclinical hyperthyroidism (SCH). Methods. We included 35 untreated women with SCH and 35 healthy control women matched by age. All participants underwent laboratory analyses which included thyroid hormone levels, and complete 2DE and 3DE examination. Results. 2DE LV longitudinal and circumferential strain was significantly decreased in the SCH subjects. 2DE LV systolic and early diastolic strain rates in longitudinal and circumferential directions were reduced, whereas late diastolic strain rates were increased in SCH individuals. 3DE LV end-diastolic volume and cardiac output were significantly elevated in the SCH patients. 3DE LV deformation in all three directions, as well as 3DE area strain, were significantly lower in the SCH group. Serum thyroid-stimulating hormone (TSH) and free thyroxine (FT4) levels correlated with 2DE LV mass index, 2DE longitudinal strain, and 3DE LV area strain in the whole study population. Conclusion. LV deformation evaluated by 3DE and speckle tracking imaging are significantly impaired in SCH subjects. TSH and FT4 levels correlate with 2DE and 3DE LV structure and mechanics.

Key words::

• echocardiography
• function
• left ventricle
• subclinical hyperthyroidism
• three-dimensional deformation

Introduction

Subclinical hyperthyroidism (SCH) is characterized by low or undetectable serum thyroid-stimulating hormone (TSH) levels, with normal ranges of free thyroxine (T4) and total or free triiodothyronine (T3) levels (1,2). However, the cutoff values for thyroid hormone levels have not been established yet. The prevalence of SCH in the USA is estimated at less than 0.7% when SCH is defined with TSH level less than 0.1 mIU/l, and 3.2% when TSH cutoff is 0.4 mIU (l,3).

The effect of SCH on cardiovascular morbidity and mortality has not been clarified because results on this topic are still conflicting. On one side, authors have shown that SCH is associated with increased all-cause and cardiovascular morbidity and mortality (4). On the other side, investigators demonstrated that SCH was associated with atrial fibrillation development, but not with other cardiovascular diseases or mortality (5).

The remodeling of the left ventricle in SCH has not been investigated enough; results are inconsistent and authors mainly included patients with exogenous or long-lasting SCH and used traditional two-dimensional echocardiographic (2DE) parameters (6–10). Only recently the same group of authors reported the results of left ventricular (LV) remodeling obtained by 2DE strain analysis in subjects with exogenous SCH due to differentiated thyroid carcinoma and long-term TSH-suppressive levothyroxine substitution therapy (11,12).

We hypothesized that SCH could significantly impact LV mechanical function, thus the aim of the present study was to investigate LV function and deformation by comprehensive two- and three-dimensional echocardiography and speckle tracking analysis in individuals with endogenous SCH.

Methodology

Our study enrolled 41 female subjects with untreated subclinical hyperthyroidism and 35 healthy female controls of similar age. The SCH and control subjects were included between January 17, 2011 and January 29, 2014. The inclusion criteria were the decreased serum TSH level (< 0.4 mIU/ml) with normal levels of T4 and free thyroxine (FT4) in at least two separate measurements taken 3–6 months apart. However, 2 subjects refused to perform further echocardiographic investigation, whereas 4 persons were excluded due to concomitant disease (arterial hypertension, diabetes mellitus, and mitral stenosis). Thus, investigation included 35 SCH participants. The control group included individuals who were referred to our outpatient clinic due to innocent murmur, palpitations, or within annual screening examination. All subjects from the control group were included simultaneously with the SCH subjects. Exclusion criteria were cardiovascular disease (arterial hypertension, myocardial infarction, atrial fibrillation, heart failure, congenital heart disease, and valvular disease), obesity (body mass index [BMI] ≥ 30 kg/m²), asthma, chronic obstructive lung disease, neoplastic disease, cirrhosis of the liver, kidney failure, sleeping disorders, type 2 diabetes mellitus, and treatment with L-thyroxine due to hypothyroidism.

Anthropometric measures (height and weight) and laboratory analyses (the level of thyroid hormones, total cholesterol, high-density lipoprotein (HDL) and low-density lipoprotein cholesterol, and triglycerides) were taken from all the subjects included in the study. Fasting venous blood samples were drawn between eight and nine o'clock in the morning. None of the participants used any medications before inclusion in the study. Normal ranges for T3, T4, FT4, and TSH were 1.3–2.6 nmol/L, 58–161 nmol/L, 11.5–22.7 pmol/l, and 0.4–4 mIU/l, respectively. T3, T4, and FT4 levels were assessed by IMMULITE 2000 enzyme-labeled chemiluminescent competitive immunoassay; TSH level was determined using IMMULITE 2000, third-generation TSH, two-site chemiluminescent immunometric assay.

BMI and body surface area (BSA) were calculated for each patient. The study was approved by the Local Ethics Committee, and informed consent was obtained from all the participants.

Echocardiography

Echocardiographic examination was performed using a 2.5-MHz transducer and 3D data set acquisition of the left ventricle obtained using the 3 V matrix probe and a Vivid 7 ultrasound machine (GE Healthcare, Horten, Norway).

Standard two-dimensional echocardiographic examination

Values of all 2DE parameters were obtained as the average value of three consecutive cardiac cycles. The LV end-systolic and end-diastolic diameters (LVEDD), LV posterior wall thickness (PWT), and interventricular septum thickness were determined according to the current recommendations (13). Relative wall thickness was calculated as (2xPWT)/LVEDD. LV ejection fraction was estimated by the biplane method. LV mass was calculated using the Devereux formula (14), and indexed for BSA. Left atrial volume was measured just before the mitral valve opening, according to the biplane method in four- and two-chamber views, and all the values were indexed for BSA (13).

Transmitral Doppler inflow and tissue pulsed Doppler were obtained in the apical four-chamber view. Pulsed Doppler measurements included the transmitral early diastolic peak flow velocity (E), late diastolic flow velocity (A), and their ratio (E/A)(15). Tissue Doppler imaging was used to obtain LV myocardial velocities in the apical four-chamber view, with a sample volume placed at the septal and lateral segment of the mitral annulus during systole, and early and late diastole (s, a´, and e´). The average of the peak early diastolic relaxation velocity (e´) of the septal and lateral mitral annulus obtained by the tissue Doppler was calculated, and the E/e´ ratio was computed.

Two-dimensional left ventricular strain and strain rate

2DE strain analysis was performed using three consecutive cardiac cycles of 2DE LV images in apical (long-axis four-chamber and two-chamber view) and parasternal short-axis views (just below the mitral level) (16). The frame rate ranged between 50 and 70 Hz. A commercially available software 2DE Auto LVQ (EchoPAC 112, GE-Healthcare, Horten, Norway) was used for 2DE strain analysis.

The 2DE longitudinal strain and strain rate were calculated by averaging all the values of the regional peak longitudinal strain and strain rate obtained in two-chamber and four-chamber apical views. 2DE circumferential strain and strain rate, as well as 2DE radial strain and strain rate, were assessed as the average of the LV six regional values measured in the parasternal short-axis view, just below the mitral valve. LV twist values, which represent the difference between average peak degree of rotation at six basal segments and four apical segments, as well as untwisting rate, were provided by the software (16).

Three-dimensional echocardiographic examination

A full-volume acquisition of the LV required for further analyses was obtained by harmonic imaging from an apical approach. Six electrocardiogram-gated consecutive beats were acquired during end-expiratory breath hold to generate full volume. The frame rate was higher than 30 frames/s.

All data sets were stored digitally and analyzed by a commercially available software 4D Auto LVQ software (EchoPAC 112, GE-Healthcare, Horten, Norway), which was used for the off-line analysis. 3DE-derived LV mass was also indexed for BSA. The software automatically identified the LV cavity endocardial border in 3DE and provided the LV volumes, cardiac output, stroke volume, ejection fraction, and LV sphericity index (17). After that, an automatic trace of the epicardial border was displayed to detect the region of interest required for LV mass and 3DE myocardial deformation parameters (speckle tracking). The 3DE deformation parameters, including global longitudinal strain, global circumferential strain, global radial strain, and global area strain, were calculated as weighted averages of the regional values from the 17 myocardial segments at end-systole (17). If ≥ 3 segments were rejected and global strain values were not calculated, that patient was excluded.

Statistical analysis

Continuous variables were presented as mean ± standard deviation (SD) and were compared using the two-tailed Student's t-test in case of normal distribution of data or by Mann–Whitney U test in case where the data were not distributed normally. Comparisons between the controls and the patients were performed by an independent-samples t-test. The differences in proportions were compared using the χ² test. The correlations were determined by the Pearson's correlation test, after TSH values were transformed by natural logarithm logarithmic in order to obtain normal distribution. Inter- and intraobserver variability was examined using Pearson's bivariate two-tailed correlations for randomly selected 10 SCH subjects and 10 controls. The p value < 0.05 was considered statistically significant.

Results

We did not find any difference in age, BSA, BMI, blood pressure, heart rate, and lipid levels, except for HDL, between the controls and the SCH group (Table I). HDL level was higher among the SCH subjects. Regarding the thyroid hormone level, T3 level was similar between the observed groups, but T4 and FT4 levels were higher in the SCH group, although in normal ranges (Table I). TSH was significantly lower among the SCH individuals, by definition (Table I).

Table I. Demographic and clinical parameters of study population.

	Controls (n = 35)	SCH (n = 35)	p
Age (years)	43 ± 12	43 ± 11	1.000
BMI (kg/m^2)	23.0 ± 3.7	23.4 ± 3.8	0.657
BSA (m^2)	1.71 ± 0.13	1.70 ± 0.12	0.739
Systolic blood pressure (mmHg)	119 ± 7	121 ± 8	0.271
Diastolic blood pressure (mmHg)	71 ± 7	73 ± 10	0.335
Heart rate (beat/min)	70 ± 8	74 ± 10	0.069
T3 (nmol/l)	1.80 ± 0.35	1.91 ± 0.38	0.256
T4 (nmol/l)	106 ± 22	130 ± 24	< 0.001
fT4 (pmol/l)	13.3 ± 1.5	16.5 ± 2.3	< 0.001
TSH (mIU/ml)	2.14 ± 1.07	0.15 ± 0.13	< 0.001
Triglycerides (mmol/l)	1.15 ± 0.53	1.11 ± 0.65	0.779
Total cholesterol (mmol/l)	5.05 ± 0.98	4.93 ± 1.22	0.652
LDL cholesterol (mmol/l)	3.06 ± 0.93	2.91 ± 1.12	0.544
HDL cholesterol (mmol/l)	1.44 ± 0.27	1.85 ± 0.64	< 0.001

BMI, body mass index; BSA, body surface area; BP, blood pressure; HDL, high-density lipoproteins; FT4, free thyroxine; LDL, low-density lipoproteins; NS, non-significant; T3, triiodothyronine; T4, thyroxine; TSH, thyroid-stimulated hormone, SCH, subclinical hyperthyroidism.

Conventional two-dimensional echocardiographic parameters

The observed groups did not differ in diameters of aorta, left atrium, and left ventricle. However, the thicknesses of interventricular septum and posterior wall of the LV, as well as LV mass index, were increased in the SCH group (Table II). Left atrial volume and volume index were also increased in the SCH patients (Table II).

Table II. Echocardiographic parameters of LV structure and function in study population.

	Controls (n = 35)	SCH (n = 35)	P
Aorta diameter (mm)	27.1 ± 1.7	26.8 ± 1.8	0.473
Left atrium diameter (mm)	35.0 ± 2.2	35.2 ± 1.9	0.685
LA volume (ml)	40.5 ± 6.4	44.1 ± 7.6	0.035
LA volume index (ml/m^2)	23.6 ± 3.9	25.9 ± 4.3	0.021
LV end-diastolic diameter (mm)	46.3 ± 3.0	47.0 ± 2.9	0.323
LV end-systolic diameter (mm)	29.1 ± 3.4	28.6 ± 3.7	0.558
Interventricular septum thickness (mm)	8.7 ± 1.0	9.3 ± 0.9	0.011
Posterior LV wall thickness (mm)	8.5 ± 0.8	9.2 ± 1.1	0.003
Relative wall thickness	0.37 ± 0.03	0.39 ± 0.04	0.020
LV mass/BSA (g/m^2)	88.1 ± 16.4	100.8 ± 19.7	0.005
Ejection fraction (%)	64 ± 3	64 ± 4	1.000
E (cm/s)	73 ± 23	66 ± 19	0.170
A (cm/s)	51 ± 14	56 ± 15	0.155
E/A ratio	1.43 ± 0.17	1.18 ± 0.24	< 0.001
e´ (cm/s)	14.4 ± 3.3	11.1 ± 3.1	< 0.001
E/e´	5.1 ± 1.0	6.0 ± 1.7	0.006
s (cm/s)	10.1 ± 2.8	8.5 ± 2.2	0.01

A, late diastolic mitral flow (pulse Doppler); E, early diastolic mitral flow (pulse Doppler); e´, average of the peak early diastolic relaxation velocity of the septal and lateral mitral annulus (tissue Doppler); LA, left atrium; LV, left ventricle; s, systolic velocity across the septal mitral annulus.

LV systolic function, assessed by 2DE ejection fraction, was normal in both groups, without any difference (Table II). On the other hand, tissue Doppler assessment indicated reduced LV systolic function in the SCH participants. Parameters of LV diastolic function (E/A and E/e’) were worse in the SCH subjects comparing with healthy controls, but still in the normal ranges (Table II). There was no difference in pulsed Doppler-derived transmitral velocities during early and late diastole (E and A) between the observed groups, whereas tissue Doppler-derived e’ was significantly lower in the SCH subjects (Table II).

Two-dimensional strain analysis

2DE LV mechanics was decreased in longitudinal and circumferential direction in the subjects with SCH, whereas there was no difference in radial strain and strain rates between the two observed groups (Table III). Systolic deformation of the LV, evaluated by systolic strain rates, was decreased in the SCH comparing with the healthy individuals. Additionally, LV diastolic mechanical function, estimated by early and late diastolic strain rate, was impaired in patients with SCH (Table III). LV twist and untwisting rate were increased in the SCH subjects in comparison with the healthy controls (Table III).

Table III. Two- and three-dimensional speckle tracking assessment of the LV function in the study population.

	Controls (n = 35)	SCH (n = 35)	p
Two-dimensional parameters
Peak systolic strain (%)
Longitudinal	− 21.3 ± 2.0	− 18.5 ± 1.7	< 0.001
Circumferential	− 22.4 ± 2.7	− 19.1 ± 3.0	< 0.001
Radial	49.5 ± 9.3	45.2 ± 11.2	0.085
Systolic strain rate (s^− 1)
Longitudinal	− 1.5 ± 0.2	− 1.2 ± 0.2	< 0.001
Circumferential	− 1.7 ± 0.3	− 1.3 ± 0.3	< 0.001
Radial	2.1 ± 0.4	1.9 ± 0.5	0.071
Early diastolic strain rate (s^− 1)
Longitudinal	1.7 ± 0.3	1.5 ± 0.2	0.002
Circumferential	2.0 ± 0.4	1.7 ± 0.3	< 0.001
Radial	− 2.3 ± 0.5	− 2.2 ± 0.4	0.359
Late diastolic strain rate (s^− 1)
Longitudinal	1.3 ± 0.3	1.6 ± 0.4	< 0.001
Circumferential	1.5 ± 0.4	1.9 ± 0.5	< 0.001
Radial	− 1.9 ± 0.5	− 2.1 ± 0.4	0.068
Twist (°)	16.3 ± 4.5	19.2 ± 5.6	0.022
Untwist rate (°/s)	− 95 ± 23	− 109 ± 29	0.031
Three-dimensional parameters
LV end-diastolic volume (ml)	119 ± 20	131 ± 22	0.020
LV end-systolic volume (ml)	44 ± 14	50 ± 16	0.102
Stroke volume (ml)	75 ± 17	81 ± 18	0.155
Cardiac output (l/min)	5.3 ± 0.8	6.0 ± 0.9	0.001
LV ejection fraction (%)	62 ± 3	62 ± 4	1.000
LV mass index (g/m^2)	80.1 ± 13.2	89.6 ± 15.1	0.006
Sphericity index	0.41 ± 0.05	0.39 ± 0.05	0.407
Global longitudinal strain (%)	− 18.6 ± 2.5	− 16.8 ± 2.0	0.001
Global circumferential strain (%)	− 19.0 ± 2.7	− 17.1 ± 2.4	0.003
Global radial strain (%)	42.3 ± 6.1	37.0 ± 5.1	< 0.001
Global area strain (%)	− 30.2 ± 3.8	− 28.0 ± 3.0	0.009

LV, left ventricular.

Three-dimensional echocardiography

End-diastolic LV volume obtained by 3DE was higher in the SCH than in the controls (Table III). On the other hand, there was no statistically significant difference in LV end-systolic and stroke volumes, although both were higher among the SCH subjects (Table III). This resulted in significantly higher cardiac output in the SCH group. There was no difference in 3DE LV ejection fraction or sphericity index between the groups, suggesting preserved LV systolic function (Table III).

3DE LV strains in all three directions, as well as area strain that represents a combination of longitudinal and circumferential strain, were significantly decreased in the SCH subjects, indicating the impairment of LV deformation in this group (Table III).

Correlation and regression analyses

We found a correlation between the TSH level and 2DE LV mass index (r = − 0.276, p = 0.046), 3DE LV mass index (r = − 0.3, p = 0.032), transmitral E/A ratio (r = 0.38, p < 0.01), 2DE longitudinal strain (r = 0.43, p < 0.01), 2DE circumferential strain (r = 0.35, p = 0.017), LV twist (r = − 0.32, p = 0.024), 3DE LV longitudinal (r = 0.41, p < 0.01), and 3DE LV area strain (r = 0.4, p < 0.01) in the whole study population.

The correlation was also revealed between FT4 level and 2DE LV mass index (r = 0.29, p = 0.045), 3DE LV mass index (r = 0.33, p = 0.01), 2DE LV longitudinal strain (r = − 0.38, p < 0.01), and 3DE LV area strain (r = − 0.34, p < 0.01) (Figure 1).

Figure 1. The relationship between FT4 and left ventricular mechanics.

Interobserver variability

Pearson's correlations—3DE LV longitudinal strain: r = 0.9, p < 0.001; 3DE LV circumferential strain: 0.83, p < 0.001; 3DE LV radial strain: r = 0.57, p < 0.001; 3DE LV area strain: r = 0.8, p < 0.001.

Intraobserver variability

Pearson's correlations—3DE LV longitudinal strain: r = 0.93, p < 0.001; 3DE LV circumferential strain: 0.88, p < 0.001; 3DE LV radial strain: r = 0.64, p < 0.001; 3DE LV area strain: r = 0.86, p < 0.001.

Discussion

Our investigation revealed several new findings: (i) LV cardiac deformation in all directions, assessed by 2DE and 3DE strain, was impaired in the SCH subjects; (ii) systolic and diastolic mechanical LV functions were affected by SCH; (iii) LV twist and untwisting rate were elevated in the SCH patients; (iv) LV volumes and cardiac output, evaluated by 3DE, were increased in the SCH individuals; (v) TSH and FT4 levels correlated with 2DE and 3DE structural and mechanical parameters in the whole study population.

To our knowledge, there are only two studies by the same group of authors that used 2DE strain (11,12). They included 25 patients with a history of differentiated thyroid carcinoma on long-term TSH-suppressive levothyroxine substitution and found that SCH affected LV circumferential and longitudinal myocardial function assessed by strain and strain rate (11). Additionally, they observed a U-shaped relationship between a range of thyroid hormone levels (from hyper- to hypothyroid concentrations) and myocardial longitudinal and circumferential strain (12). Our findings in the present study and the previous study about subclinical hypothyroidism (20) are completely in line with these results. However, we went a step further and determined LV twist, untwisting rate, and 3DE LV deformation. Considering the fact that LV twist mostly represents a marker of LV systolic function (18), while LV untwisting rate is commonly used as a parameter of LV diastolic dysfunction (19), we could conclude that both systolic and diastolic LV mechanical functions are influenced by SCH. Namely, our findings show that LV systolic function, estimated by 2DE and 3DE LV ejection fraction, in SCH is normal and does not differ from the healthy controls; on the other hand, LV systolic mechanics, estimated by 2DE and 3DE strain, is impaired in the SCH subjects. The reason for preserved LV systolic function, despite decreased LV systolic deformation, could actually lie in the paradoxically elevated LV twist that we detected in the SCH patients. LV twist represents a complex movement that involves circumferential motion of the apex with respect to the base of the heart and enables the most efficient contraction of the LV myocardial fibers; it could be increased in overload condition (18). Our results indicate that 3DE LV volumes and cardiac output are increased in the SCH individuals, which suggests hemodynamic overload condition and possibly explains elevated LV twist in this group. Considering our findings that 2DE and 3DE LV ejection fractions were similar between the observed groups, it seems that this conventional echocardiographic parameter is not quite suitable for the estimation of LV systolic function in the SCH subjects.

Namely, untwisting is the recoil of the systolic torsional deformation; it happens mostly during isovolumic relaxation and contributes to early diastolic filling (19). The evaluation of untwisting may provide an accurate evaluation of LV relaxation. In the present study, untwisting rate is significantly increased in the SCH subjects, which indicates the prolongation of the isovolumic relaxation time and deterioration of LV diastolic function. This is in line with our results regarding LV early and late diastolic strain rates that also show the impairment of LV diastolic mechanical function in the SCH patients.

The assessment of LV myocardial deformation by 3DE has not been done so far. Our findings showed that 3DE LV strain has been impaired in all directions (longitudinal, circumferential, and radial), along with 3DE area strain, in the SCH subjects comparing with the controls. Using this sophisticated method we showed that LV radial strain is also affected by SCH, which was not shown by 2DE strain analysis in our investigation or previously by Abdulrahman et al. (11,12). The fact that 3DE area strain, the parameter of both longitudinal and circumferential function, and accurate index for quantitative assessment of global and regional LV function (20), is reduced among the SCH patients in the present study is more significant.

Our results also indicate the correlation between the TSH and FT4 levels and LV structure, function, and mechanics, assessed by 2DE and 3DE. The progressive reduction of TSH has been associated with increase of LV mass and LV twist, and with decrease of mitral E/A ratio, and LV longitudinal, circumferential, and area strain. Abdulrahman et al. showed a borderline association of 2DE longitudinal and circumferential strain with the TSH levels (11). FT4 inversely correlates with 2DE mass index and longitudinal, as well as 3DE area, strain.

Although the mechanism of LV remodeling in SCH is unclear, there are several possibilities that could explain cardiovascular changes: (i) reduced peripheral resistance and peripheral vasodilatation; (ii) increased activity of renin–angiotensin–aldosterone system; (iii) decreased renal perfusion; and (iv) increased LV volumes and cardiac output. All these changes could induce impairment of LV systolic and diastolic mechanical function, as well as LV hypertrophy (21).

There are several important clinical implications of our study. First, our findings about decreased LV strain could potentially explain cardiovascular morbidity and mortality in SCH patients, especially in long-lasting SCH, because it has been already shown that LV deformation is a good predictor of the outcome (22). Second, LV strain analysis could help us to determine the necessity of therapy introduction. Previous investigations indicate that LV remodeling is reversible (7,8); however, strain analysis could demonstrate improvement much faster and therapy could be stopped earlier. Third, evaluation of LV deformation, due to high sensitivity, may facilitate the follow-up of these patients.

Limitations

The present investigation has several limitations: (i) 3DE estimation of LV structure and function could be significantly influenced by the quality of ultrasound images, especially during the full-volume acquisition; (ii) a small number of patients; (iii) our investigation included only women, which restricts our results to this population; and (iv) coronary artery disease (CAD) was not excluded by coronary angiography, but we included young females without cardiovascular risk factors and free of any symptom of CAD, thus it is unlikely that CAD could explain the LV mechanical myocardial changes that we found.

Conclusion

LV myocardial deformation, evaluated by 2DE and 3DE strain, is impaired in the SCH subjects. There is no difference in LV systolic function, assessed by 2DE and 3DE ejection fraction, between the SCH and the controls. However, other parameters of LV systolic function obtained by tissue Doppler and strain analysis indicate impaired LV systolic function in the SCH subjects, which refers to the inefficacy of ejection fraction in the estimation of LV systolic function in the SCH subjects. It has been shown that the TSH and FT4 levels correlated with LV mass index and LV deformation. Future longitudinal studies with a larger number of patients are necessary to confirm the unfavorable influence of endogenous SCH on LV mechanics and investigate the influence of these changes on long-term outcome. Further research is essential for the assessment of time when therapy should be initiated in the SCH subjects, as well as the possible effect of therapy on LV deformation improvement.

Declaration of interest: The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.