2020 Volume 84 Issue 7 Pages 1105-1111
Background: Parathyroid hormone (PTH) has been associated with cardiovascular disorders; however, it is unknown whether plasma PTH concentrations are associated with atrial fibrillation (AF) in patients with chronic kidney disease (CKD).
Methods and Results: The present cross-sectional study analyzed baseline data of 3,384 patients registered in the Fukuoka Kidney Disease Registry Study, a Japanese multicenter prospective cohort study of patients with non-dialysis-dependent CKD. The outcome was prevalence of AF, and the main risk factor was plasma intact PTH concentration. Associations between plasma intact PTH concentration quartiles (Q1–Q4, from lowest to highest) and the presence of AF were analyzed using logistic regression. In all, 185 patients had AF; 22, 34, 59, and 70 patients were in Q1, Q2, Q3, and Q4 of PTH concentrations, respectively. The prevalence of AF increased incrementally with increases in plasma intact PTH. In the logistic regression model, patients with higher plasma intact PTH concentrations (Q2–Q4) had higher adjusted odds ratios (95% confidence intervals) for the prevalence of AF relative to the reference group (Q1), namely 1.33 (0.76–2.34), 1.82 ([1.06–3.13), and 1.99 (1.08–3.64), respectively (P=0.016).
Conclusions: Higher plasma intact PTH concentrations were significantly and incrementally associated with an increased prevalence of AF in non-dialysis-dependent CKD patients.
Atrial fibrillation (AF) is frequently seen in clinical practice and is a risk factor for the development of heart failure and brain infarction.1,2 In addition, AF may cause distressing symptoms, such as palpitations and chest discomfort, and eventually lead to a decline in an individual’s quality of life. Therefore, identification of modifiable risk factors for the development of AF is important.
To date, many factors have been found to be associated with the development of AF, such as aging,3 hypertension,4 diabetes,5,6 hyperthyroidism,7 mitral valve disease,5 heart failure,5 and renal dysfunction.8 Of these, renal dysfunction is closely linked to other pathological conditions, including uremia, anemia, and mineral and bone metabolism disorders.9 Chronic kidney disease (CKD)-associated mineral and bone disorder often manifests as hyperphosphatemia, hypocalcemia, and elevated serum concentrations of fibroblast growth factor (FGF) 23 and parathyroid hormone (PTH). FGF23, first identified as a phosphaturic hormone, has been shown to have various effects on the cardiovascular system.10,11 Recently, elevated serum FGF23 concentrations were reported to be significantly associated with incident AF in patients with CKD.12 PTH is another phosphaturic hormone that has frequently been associated with cardiovascular disorders, including hypertension,13 vascular calcification,14 left ventricular hypertrophy,15 and heart failure.16 Given these associations, it is possible that AF is also caused by increased circulating PTH concentrations in response to decreased kidney function and resulting relative phosphate overload; however, few studies have focused on the association between PTH and AF, especially in CKD patients.
The aim of the present study was to investigate the association between plasma intact PTH concentrations and the prevalence of AF in non-dialysis-dependent CKD patients.
In the present cross-sectional study, baseline data from the Fukuoka Kidney Disease Registry (FKR) Study, an ongoing Japanese prospective multicenter cohort study of non-dialysis-dependent CKD patients, were analyzed. As described previously,17 the study population consisted of 4,476 outpatients, aged ≥16 years and under the care of a nephrologist at 1 of 12 clinical facilities in Fukuoka and Saga prefectures in the northern region of Kyushu, Japan. All patients had been diagnosed with CKD due to abnormal kidney structure or function of at least 3 months duration, based on the Kidney Disease: Improving Global Outcomes (KDIGO) criteria for CKD.18 Patient enrollment occurred between January 2013 and March 2017, and all patients were scheduled to be followed-up for at least 5 years. Of the 4,476 patients, 1,092 were excluded because of lack of data for either baseline characteristics or outcome. A detailed breakdown of the reasons for exclusion is shown in Supplementary Figure 1. Consequently, data from 3,384 patients were analyzed in the present study.
The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Kyushu University Hospital Institutional Review Board for Clinical Research (No. 469-04). All patients provided written informed consent prior to participating in the study.
Demographic Factors and Biochemical MeasurementsDemographic and clinical data were recorded at enrollment. These included age, sex, blood pressure, the use of angiotensin-converting enzyme inhibitors (ACEIs), angiotensin II receptor blockers (ARB), β-blockers, other antihypertensive agents, or diuretics, diabetes, hyperthyroidism, a history of ischemic heart diseases, body mass index (BMI), smoking status, and alcohol intake. The data were obtained from medical records or patient questionnaire (for smoking status and alcohol intake) by trained research coordinators. Blood and serum biochemical parameters (i.e., hemoglobin, serum albumin, total cholesterol, high-density lipoprotein cholesterol [HDL-C], high-sensitivity C-reactive protein [hs-CRP], creatinine, albumin-corrected calcium, phosphate, plasma intact PTH, and urinary protein, creatinine, and phosphate) were determined by a central laboratory from blood and urine samples collected at the time of enrollment and stored at −80℃ until measurement. An albumin-corrected serum calcium concentration was calculated when serum albumin was <4 g/dL according to the formula of Payne et al.19 For patients <18 years of age, estimated glomerular filtration rate (eGFR; mL/min/1.73 m2) was calculated using the formula of Schwartz et al,20 whereas in patients ≥18 years of age the following formula21 was used:
eGFR = 194 × creatinine−1.094 × age−0.287 (× 0.739 in women)
The fractional excretion of phosphate (FEPi), which is the ratio of phosphate clearance to creatine clearance, was calculated using as follows:
FEPi = (urinary phosphate/serum phosphate) / (urinary creatinine/serum creatinine) × 100
Definitions of Outcomes and CovariatesThe primary outcome of the present study was the prevalence of AF. The presence of AF was determined from a review of medical records for clinical diagnosis. There was no distinction made between types of AF (persistent or paroxysmal) in the present study. The main risk factor was plasma intact PTH concentration, which was determined by the central laboratory.
Statistical AnalysisCategorical data are expressed as numbers and percentages, whereas continuous variables (normally and non-normally distributed) are expressed as the median and interquartile range. Patients were categorized according to plasma intact PTH concentrations into quartiles (Q1–Q4, from lowest to highest). The distribution of clinical factors (stratified by plasma intact PTH concentration quartiles) was analyzed for trends: the Cochran–Armitage test was used for categorical variables and the Jonckheere–Terpstra test was used for continuous variables. Multivariable-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the prevalence of AF were estimated using a logistic regression model. The following covariates were included in the multivariable-adjusted logistic regression, selected based on a priori clinical judgment: age, sex, systolic blood pressure (SBP), use of antihypertensive agents and diuretics, diabetes, hyperthyroidism, a history of ischemic heart diseases, BMI, smoking status, alcohol intake, hemoglobin, serum albumin, total cholesterol, HDL-C, log[hs-CRP], albumin-corrected calcium and phosphate, eGFR, urinary protein to creatinine ratio, and FEPi. Subgroup analysis was performed to elucidate interactions in the association between plasma intact PTH concentration and the prevalence of AF. To investigate the cut-off concentration of plasma intact PTH for the prediction of AF, receiver operating characteristic (ROC) curve and area under the curve (AUC) analyses were performed. In all analyses, 2-tailed P<0.05 was considered significant. Statistical analyses were performed using JMP version 13.2 (SAS Institute, Cary, NC, USA) and R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria).
All data were stratified according to plasma intact PTH concentration quartiles (Table 1). Patients with a higher plasma intact PTH level were significantly older, more likely to be male, had higher SBP and BMI, had a higher prevalence of diabetes mellitus, and a history of ischemic heart disease. Hemoglobin, albumin, total cholesterol, HDL-C, albumin-corrected calcium, and eGFR were significantly lower in patients with higher plasma intact PTH concentrations. In contrast, serum hs-CRP and phosphate, the urinary protein to creatinine ratio, and FEPi were significantly higher in patients with higher plasma intact PTH concentration. Patients with a higher intact PTH concentration used diuretics and antihypertensive agents, including ACEIs, ARBs, and β-blockers significantly more frequently.
| Plasma intact PTH quartile | P for trend | ||||
|---|---|---|---|---|---|
| Q1 (n=842) | Q2 (n=850) | Q3 (n=848) | Q4 (n=844) | ||
| PTH concentrations (pg/mL) | 5–46 | 47–66 | 67–108 | 109–1,660 | |
| Demographics | |||||
| Age (years) | 65 [49–73] | 66 [56–74] | 69 [60–77] | 70 [61–78] | <0.001 |
| Male sex | 446 (53.0) | 454 (53.4) | 509 (60.0) | 503 (59.6) | <0.001 |
| SBP (mmHg) | 126 [115–138] | 129 [119–139] | 131 [120–142] | 133 [122–146] | <0.001 |
| DBP (mmHg) | 73 [66–80] | 75 [68–81] | 75 [67–83] | 74 [66–82] | 0.168 |
| Antihypertensive agents used | 604 (71.7) | 661 (77.8) | 715 (84.3) | 790 (93.6) | <0.001 |
| ACEI or ARB | 548 (65.1) | 583 (68.6) | 619 (73.0) | 676 (80.1) | <0.001 |
| β-blockers | 54 (6.4) | 84 (9.9) | 148 (17.5) | 235 (27.8) | <0.001 |
| Diuretics | 82 (10.4) | 117 (12.9) | 182 (21.1) | 315 (38.0) | <0.001 |
| Diabetes | 168 (20.0) | 187 (22.0) | 226 (26.7) | 324 (38.4) | <0.001 |
| Hyperthyroidism | 8 (1.1) | 11 (1.2) | 12 (1.4) | 11 (1.3) | 0.489 |
| History of IHDs | 55 (6.5) | 74 (8.7) | 105 (12.4) | 147 (17.4) | <0.001 |
| Body mass index (kg/m2) | 22.4 [20.2–25.0] | 23.1 [21.0–25.6] | 23.1 [20.8–25.9] | 22.9 [20.6–25.6] | 0.008 |
| Current smoker | 97 (11.5) | 83 (9.8) | 91 (10.7) | 98 (11.6) | 0.794 |
| Current drinker | 523 (62.1) | 526 (61.9) | 531 (62.6) | 501 (59.4) | 0.315 |
| Blood and urine tests | |||||
| Hemoglobin (g/dL) | 13.4 [12.2–14.5] | 13.3 [12.1–14.5] | 12.8 [11.6–14.2] | 11.3 [10.5–12.5] | <0.001 |
| Serum albumin (g/dL) | 4.1 [3.9–4.4] | 4.2 [3.9–4.4] | 4.1 [3.8–4.3] | 3.9 [3.6–4.1] | <0.001 |
| Serum total cholesterol (mg/dL) | 195 [175–219] | 198 [174–224] | 193 [169–218] | 178 [154–206] | <0.001 |
| Serum HDL-C (mg/dL) | 59 [48–73] | 60 [47–74] | 57 [44–70] | 52 [42–64] | <0.001 |
| eGFR (mL/min/1.73 m2) | 56.0 [41.8–74.9] | 48.5 [36.5–63.4] | 37.4 [26.1–52.6] | 18.0 [12.3–26.4] | <0.001 |
| Serum hs-CRP (mg/dL) | 0.0 [0.0–0.1] | 0.0 [0.0–0.1] | 0.1 [0.0–0.1] | 0.1 [0.0–0.2] | <0.001 |
| Corrected serum calcium (mg/dL) | 9.6 [9.4–9.9] | 9.6 [9.3–9.8] | 9.4 [9.2–9.7] | 9.2 [8.9–9.5] | <0.001 |
| Serum phosphate (mg/dL) | 3.4 [3.0–3.8] | 3.3 [3.0–3.7] | 3.3 [2.9–3.7] | 3.7 [3.2–4.3] | <0.001 |
| Urinary protein (g/gCr) | 0.2 [0.1–0.6] | 0.2 [0.1–0.7] | 0.4 [0.1–1.1] | 1.2 [0.4–2.8] | <0.001 |
| FEPi (%) | 12.4 [8.3–18.0] | 14.2 [9.9–19.6] | 18.2 [12.5–25.3] | 33.9 [22.5–44.4] | <0.001 |
Data are expressed as the median [interquartile range] or as n (%). The Cochran–Armitage test was used to determine P-values for trends in categorical variables. The Jonckheere–Terpstra test was used to determine P-values for trends in continuous variables. Two-tailed P<0.05 was considered significant. Conversion factors are as follows: to convert total cholesterol from mg/dL to mmol/L, multiply by 0.0259; to convert high-density lipoprotein cholesterol (HDL-C) from mg/dL to mmol/L, multiply by 0.0259; to convert high-sensitivity C-reactive protein (hs-CRP) from mg/dL to nmol/L, multiply by 9.524; to convert albumin-corrected Ca from mg/dL to mmol/L, multiply by 0.25; to convert phosphate from mg/dL to mmol/L, multiply by 0.323. ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; FEPi, fractional excretion of phosphate; IHDs, ischemic heart diseases; PTH, parathyroid hormone; SBP, systolic blood pressure.
Among the 3,384 patients, 185 had AF; of 185 patients these, 22 patients were in Q1, 34 were in Q2, 59 were in Q3, and 70 were in Q4. The prevalence of AF increased incrementally as plasma intact PTH concentrations increased (Figure 1). In the un-adjusted, age- and sex-adjusted, and multivariable-adjusted logistic regression models, patients with higher plasma intact PTH concentrations (Q3 and Q4) had higher adjusted ORs for the prevalence of AF compared with the reference group (Q1; Table 2). The adjusted OR for every 50-pg/mL increase in plasma intact PTH concentration was 1.09 (95% CI 1.00–1.20; P=0.076) for the prevalence of AF.
Prevalence of atrial fibrillation in each quartile of plasma intact parathyroid hormone (PTH) concentration. Logistic regression analysis was used to analyze the significance of differences, and 2-tailed P<0.05 was considered significant.
| Unadjusted | Age and sex adjusted | Multivariable adjusted | |||||||
|---|---|---|---|---|---|---|---|---|---|
| OR (95% CI) |
P-value | P for trend |
OR (95% CI) |
P-value | P for trend |
OR (95% CI) |
P-value | P for trend |
|
| Plasma Intact PTH quartile | <0.001 | <0.001 | 0.016 | ||||||
| Q1 (PTH 5–46 pg/mL) | 1.00 (Ref.) |
– | 1.00 (Ref.) |
– | 1.00 (Ref.) |
– | |||
| Q2 (47–66 pg/mL) | 1.55 (0.90–2.68) |
0.113 | 1.38 (0.79–2.40) |
0.252 | 1.33 (0.76–2.34) |
0.321 | |||
| Q3 (67–108 pg/mL) | 2.79 (1.69–4.59) |
<0.001 | 2.10 (1.26–3.48) |
0.004 | 1.82 (1.06–3.13) |
0.029 | |||
| Q4 (109–1,660 pg/mL) | 3.37 (2.07–5.50) |
<0.001 | 2.40 (1.46–3.96) |
<0.001 | 1.99 (1.08–3.64) |
0.026 | |||
| Per 50-pg/mL increment in PTH |
1.08 (1.03–1.14) |
0.007 | 1.09 (1.03–1.16) |
0.009 | 1.09 (1.00–1.20) |
0.076 | |||
Unadjusted, age- and sex-adjusted, and multivariable-adjusted ORs were analyzed with logistic regression modeling. The covariates included age, sex, SBP, the use of antihypertensive agents and diuretics, diabetes, hyperthyroidism, a history of IHDs, body mass index, smoking status, alcohol intake, hemoglobin, serum albumin, total cholesterol, HDL-C, log[hs-CRP], albumin-corrected calcium and phosphate, eGFR, urinary protein, and FEPi. Two-tailed P<0.05 was considered significant. CI, confidence Intervals; ORs, odds ratios. Other abbreviations as in Table 1.
To evaluate heterogeneity in the association between plasma intact PTH concentration and the presence of AF, subgroup analysis was performed. The association between plasma intact PTH concentrations and AF tended to be enhanced in patients with a higher than lower eGFR (P=0.063; Figure 2). Furthermore, the association was significantly enhanced in patients with a lower FEPi (P=0.014).
Multivariable-adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for the presence of atrial fibrillation by 50-pg/mL increases in plasma intact parathyroid hormone (PTH) in subgroups stratified according to different clinical factors. Open and filled symbols denote point estimates of the OR, with error bars representing the 95% CI. Two-tailed P<0.05 was considered significant. eGFR, estimated glomerular filtration rate; FEPi, fractional excretion of phosphate; hs-CRP, high-sensitivity C-reactive protein. Conversion factors are as follows: to convert hs-CRP concentrations from mg/dL to nmol/L, multiply by 9.524; to convert albumin-corrected Ca from mg/dL to mmol/L, multiply by 0.25; to convert phosphate from mg/dL to mmol/L, multiply by 0.323.
The cut-off plasma intact PTH concentration for predicting AF was 69 pg/mL, with sensitivity, specificity, and AUC of 0.697, 0.527, and 0.625, respectively (Supplementary Figure 2).
In the present cross-sectional study of patients with non-dialysis-dependent CKD, we demonstrated that the plasma intact PTH concentration was significantly and positively associated with the prevalence of AF, even after rigorous adjustment for potential confounding factors. This study suggests that PTH, a crucial hormone in mineral and bone homeostasis, plays a role in the development of AF in the CKD population.
AF is defined as the irregular contraction of atrial muscle resulting from abnormal electrical signals in the atrium. Several factors, such as mechanical stress to the left atrium, autonomic nervous system hyperactivity, and disruption of atrial muscle ion channels, are thought to be involved in the development of AF.22–24 Underlying conditions such as mitral valve stenosis, heart failure, and hypertension are typically associated with mechanical stress to the left atrium. Because accumulating evidence shows that hyperparathyroidism causes such underlying conditions like mitral valve diseases,25–27 elevated PTH may indirectly contribute to increased mechanical stress to the left atrium and ultimately to the elevated risk of AF in CKD.
In the association between elevated PTH and AF, a direct effect of PTH on cardiomyocytes is arguable. Although PTH is known to regulate extracellular calcium, it has also been shown to promote overload of free calcium in the cytosol and mitochondria of cardiomyocytes.28,29 This inappropriate increase in intracellular calcium (mediated, in part, by increased oxidative stress) causes adverse structural remodeling, including cardiac hypertrophy and fibrosis.30,31 Because cardiac structural remodeling is important in the pathogenesis of AF,32 PTH-induced intracellular calcium overload may also play a role in the development of AF. In addition, a positive chronotropic action of PTH, another potential cause of AF, has been reported.33 Hara et al demonstrated that PTH and PTH-related protein had direct electrophysiological effects on cardiomyocytes, increasing automaticity.33 Enhanced automaticity may, in turn, cause abnormal electrical signals in the atrium, leading to the development of AF.
Another potential explanation for the positive association between PTH and AF is grounded in the bidirectional interplay of PTH and the renin-angiotensin-aldosterone system (RAAS). Previous research has demonstrated a positive correlation between serum PTH and aldosterone,34 and that spironolactone treatment decreases plasma intact PTH concentrations.35 Furthermore, Isales et al demonstrated that PTH directly stimulates aldosterone secretion by increasing intracellular calcium in the adrenal zona glomerulosa.36 Furthermore, there is growing evidence that activation of the RAAS promotes electrical and structural remodeling in the heart,37 whereas RAAS blockade reduces incident AF.38 Taking all these observations into consideration, increases in PTH may lead to the development of AF via enhanced RAAS activation.
The effect of PTH on the development of AF may be affected by concomitant clinical factors. In the subgroup analysis, preserved renal function tended to strengthen the association between plasma intact PTH concentrations and the prevalence of AF. Alternatively, a community-based observational study showed that renal dysfunction was significantly associated with the development of AF.8 This finding may be explained, in part, by conditions linked to renal dysfunction, including uremia, vascular calcification, and heart failure. Indeed, Aoki et al demonstrated in a 5/6 nephrectomy rat model and in vitro that indoxyl sulfate, a major uremic toxin, contributes to the development of AF.39 Conversely, in patients with preserved renal function, renal dysfunction-related factors are less likely to be present and to exert additional effects on the association between PTH and AF. Finally, previous studies have shown an association between PTH and AF in the general population, mainly composed of non-CKD patients,40,41 and these data are consistent with those of the present study.
In the present study, the association between plasma intact PTH and prevalence of AF was enhanced in patients with a lower FEPi. This may be explained by the link between FGF23 and AF. Both FGF23 and PTH are phosphaturic hormones, and the serum concentrations of both increase with declining renal function to maintain phosphate homeostasis. As described previously, elevated serum FGF23 concentrations are significantly associated with the development of AF.12 Generally, FEPi is higher in patients with higher serum FGF23 concentrations than that in patients with lower serum FGF23 concentrations. Therefore, the effect of PTH on the development of AF may be weakened by the effect of FGF23 on AF in patients with higher FEPi, who are likely to have remarkably higher serum FGF23 concentrations than patients with lower FEPi.
The cut-off concentration of plasma intact PTH for predicting AF (69 pg/mL) determined by ROC curve analysis was very close to the upper limit of the normal range of PTH concentrations. Although the present study was cross-sectional, the data indicate the importance of controlling secondary hyperparathyroidism in CKD patients to prevent AF. It is also important to avoid hyperphosphatemia and hypocalcemia to maintain PTH homeostasis in CKD patients, and the combination of phosphate restriction and the use of phosphate binders or vitamin D receptor activators is a strategy to control these parameters. Further studies are needed to confirm that controlling hyperparathyroidism decreases the onset of AF.
To the best of our knowledge the present large cohort study is the first to show the association between plasma intact PTH concentrations and the prevalence of AF in non-dialysis-dependent CKD patients; however, the study has some limitations. First, the cross-sectional study design precluded investigation of a causal relationship. Indeed, repeated studies have suggested that AF itself causes increases in PTH or PTH-related protein concentrations;42,43 therefore, longitudinal studies are needed to prove causality between increases in plasma PTH concentrations and the development of AF. Second, we did not measure serum FGF23, which is known to be associated with the development of AF;12 however, we attempted to reduce the possible confounding effect of FGF23 on the association between plasma intact PTH and AF by including FEPi as a covariate in the analysis. Third, although heart failure and valvular disease are among the most important risk factors for AF,5 we did not have any data on the history of heart failure or cardiac conditions (apart from ischemic heart disease) or on serum B-type natriuretic peptide concentrations or echocardiography test results. Although we rigorously adjusted for potentially confounding concomitant clinical factors, it is possible that the omission of these or other potential confounders from the analysis affected the results. Fourth, a large number of patients was excluded from the main analyses because of missing data (mostly questionnaire results), and this may have introduced a selection bias. In addition, although all patients had CKD, we did not exclude patients in the acute phase of the illness, and it is possible that this selection decision also affected the laboratory data, including plasma intact PTH concentrations. Finally, we diagnosed AF based on a review of the medical records and did not perform confirmatory electrocardiograms or Holter electrocardiograms; therefore, some patients may have been incorrectly classified as having AF; we consider this to be an important limitation of the study. However, despite these limitations, we believe that the present study furthers our knowledge of the association between PTH and the development of AF in patients with CKD.
In conclusion, the data of the present study show the plasma intact PTH concentrations are significantly and positively associated with the prevalence of AF in CKD patients. In future studies, longitudinal data from the FKR study will provide further robust evidence regarding the relationship between plasma intact PTH concentrations and the development of AF in the CKD population.
The authors thank all the doctors and medical staff who participated in the FKR study. Members of the Steering Committee and Principal Collaborators of the FKR Study Group are listed in Appendix. The authors also thank Eleanor Scharf, MSc(A), from Edanz Group (www.edanzediting.com/ac), for editing a draft of this manuscript.
This study did not receive any specific funding.
T.K. is a member of the Editorial Team of Circulation Journal. The remaining authors report no conflicts of interest.
This study was approved by the Kyushu University Hospital Institutional Review Board for Clinical Research (No. 469-04).
Satoru Fujimi (Fukuoka Renal Clinic), Hideki Hirakata (Fukuoka Renal Clinic), Tadashi Hirano (Hakujyuji Hospital), Tetsuhiko Yoshida (Hamanomachi Hospital), Takashi Deguchi (Hamanomachi Hospital), Hideki Yotsueda (Harasanshin Hospital), Kiichiro Fujisaki (Iizuka Hospital), Keita Takae (Japanese Red Cross Fukuoka Hospital), Koji Mitsuiki (Japanese Red Cross Fukuoka Hospital), Akinori Nagashima (Japanese Red Cross Karatsu Hospital), Ritsuko Katafuchi (Kano Hospital), Hidetoshi Kanai (Kokura Memorial Hospital), Kenji Harada (Kokura Memorial Hospital), Tohru Mizumasa (Kyushu Central Hospital), Takanari Kitazono (Kyushu University), Toshiaki Nakano (Kyushu University), Toshiharu Ninomiya (Kyushu University), Kumiko Torisu (Kyushu University), Akihiro Tsuchimoto (Kyushu University), Shunsuke Yamada (Kyushu University), Hiroto Hiyamuta (Kyushu University), Shigeru Tanaka (Kyushu University), Dai Matsuo (Munakata Medical Association Hospital), Yusuke Kuroki (National Fukuoka-Higashi Medical Center), Hiroshi Nagae (National Fukuoka-Higashi Medical Center), Masaru Nakayama (National Kyushu Medical Center), Kazuhiko Tsuruya (Nara Medical University), Masaharu Nagata (Shin-eikai Hospital), Taihei Yanagida (Steel Memorial Yawata Hospital), and Shotaro Onaka (Tagawa Municipal Hospital).
Please find supplementary file(s);
http://dx.doi.org/10.1253/circj.CJ-19-1201
