Article Text

Original article
Heart-type fatty acid-binding protein in the early diagnosis of acute myocardial infarction
  1. Miriam Reiter1,2,
  2. Raphael Twerenbold1,2,
  3. Tobias Reichlin2,3,
  4. Mira Mueller2,
  5. Rebecca Hoeller1,2,
  6. Berit Moehring1,2,
  7. Philip Haaf2,
  8. Karin Wildi1,2,
  9. Salome Merk2,
  10. Denise Bernhard2,
  11. Christa Zellweger Mueller1,
  12. Michael Freese2,
  13. Heike Freidank4,
  14. Isabel Campodarve Botet5,
  15. Christian Mueller1,2
  1. 1Department of Internal Medicine, University Hospital, Basel, Switzerland
  2. 2Department of Cardiology, University Hospital, Basel, Switzerland
  3. 3Cardiovascular Division, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts, USA
  4. 4Department of Laboratory Medicine, University Hospital, Basel, Switzerland
  5. 5Servicio de Urgencias, Hospital del Mar—IMIM, Barcelona, Spain
  1. Correspondence to Prof Christian Mueller, Department of Cardiology, University Hospital Basel, Petersgraben 4, Basel CH-4031, Switzerland; Christian.mueller{at}usb.ch

Abstract

Objective To investigate the diagnostic and prognostic role of heart-type fatty acid-binding protein (hFABP) compared with copeptin and in addition to high-sensitivity cardiac troponin T (hs-cTnT) in patients with chest pain suspected of acute myocardial infarction (AMI).

Design Diagnostic and prognostic performances of hFABP, copeptin and hs-cTnT were evaluated and compared. The final diagnosis was adjudicated by two independent cardiologists.

Setting This prospective observational multicentre study took place in four primary and one secondary hospital from April 2006 to September 2009.

Patients We enrolled 1247 consecutive patients with suspected AMI to the emergency department. For analysis, patients were included, if baseline levels for hs-cTnT and hFABP were available (n=1074), patients with ST-segment elevation myocardial infarction (STEMI) were excluded for the diagnostic analysis (n=43).

Interventions Treatment was left to the discretion of the emergency physician.

Main outcome measures AMI and mortality.

Results 4% of the patients had STEMI and 16% of the patients had non-STEMI. Patients with AMI had significantly higher levels of hFABP at presentation (p<0.001). Neither the combination with hFABP nor with copeptin increased the diagnostic accuracy of hs-cTnT at admission, quantified by the area under the receiver operating characteristic curve (AUC) (p>0.05). The negative predictive value regarding 90-day, 1-year and 2-year mortality was 100% (99–100), 99% (98–100) and 98% (96–99), respectively, for hFABP levels below the median (p<0.001). The accuracy of hFABP to predict 90-day mortality was moderate (AUC 0.83; 95% CI 0.77 to 0.90).

Conclusions hFABP and copeptin do not improve the diagnosis of patients with chest pain without ST-segment elevation, but may be useful for risk stratification beyond hs-TnT.

  • MYOCARDIAL ISCHAEMIA AND INFARCTION (IHD)
  • CORONARY ARTERY DISEASE

Statistics from Altmetric.com

Background

Acute myocardial infarction (AMI) is a major cause of death and disability worldwide. Its rapid and accurate diagnosis and risk stratification are critical for the initiation of effective evidence-based medical management and treatment,1 ,2 but still an unmet clinical need. ECG and cardiac troponin (cTn) form the diagnostic and prognostic cornerstones of clinical assessment.2 ,3 But particularly in patients without ST-segment elevation in the ECG, diagnosis of non-ST-elevation myocardial infarction (non-STEMI) often requires serial blood sampling and causes prolonged time spent in the emergency department (ED), increasing patients’ uncertainty and anxiety, and costs for the healthcare system.4 ,5 Furthermore, delayed rule-in increases morbidity and mortality.6 ,7 Recently, high-sensitive cTn assays have been shown to provide higher accuracy at presentation to the ED as compared with conventional cTn assays.8–10

Heart-type fatty acid-binding protein (hFABP), a small soluble cytosolic protein involved in the transportation of long-chain fatty acids into the cardiomyocyte is released rapidly into the circulation in response to cardiomyocyte injury. Due to its solubility, hFABP can be released more rapidly than structurally bound molecules like cardiac troponins. Furthermore, it may enter the vascular system directly via endothelium because of its small size (15 kDA). Thus, hFABP is regarded as a early sensitive marker of AMI.11–15 In previous studies, hFABP has also been suggested as a prognostic tool in heart failure16 and marker of myocardial ischaemia, even in the absence of necrosis.13 ,17 ,18 Furthermore, it seems to predict long-term mortality in patients with acute coronary syndrome.13 ,18 ,19

In this prospective multicentre study, we assessed the diagnostic and prognostic utility of hFABP in combination with hs-cTnT, and compared with hs-TnT alone and in combination with copeptin.

Material and methods

Study design and population

The Advantageous Predictors of Acute Coronary Syndrome Evaluation (APACE) Study is an ongoing prospective multicentre study designed and coordinated by the University Hospital Basel. From April 2006 to June 2009, a total of 1247 consecutive patients presenting to the ED with chest pain suggestive of AMI with onset or peak within the last 12 h were recruited. Patients with end-stage renal failure requiring dialysis were excluded. For analysis, patients were included if the baseline levels for hs-cTnT and hFABP were available. Patients with ST-segment elevation in the initial ECG were excluded from the diagnostic analysis, because current guidelines recommend immediate revascularisation on the basis of ECG findings in these patients.

The study was performed according to the principles of the Declaration of Helsinki and approved by the local ethics committee. Written informed consent was obtained from all patients. The authors designed the study, gathered and analysed the data, vouched for the data and analysis, wrote the paper and decided to submit it for publication. The assays were donated by the manufacturers, who had no role in the design of the study, data analysis, manuscript or decision to submit for publication.

Routine clinical assessment

All patients underwent an initial clinical assessment that included history taking, a physical examination, 12-lead ECG, continuous ECG monitoring, pulse oximetry, standard blood tests and chest radiography. CTnI or cTnT, CK-MB and myoglobin were measured at presentation and 6–9 h after, or as long as clinically indicated. The precise timing of clinical post-baseline measurements and the treatment of patients were left to the discretion of the attending physician.

Follow-up and clinical end points

After hospital discharge, patients were followed up after 3 and 12 months by telephone calls or in written form. Any reported clinical event—in particular cardiovascular events—since presentation to the ED, was reviewed by asking the patient or a close relative and traced by establishing contact with the respective family physician or treating institution. Information regarding death and cause of death were obtained from the national registry on mortality, the hospital's diagnosis registry or family physician's records. The primary end point was all-cause mortality.

Adjudicated final diagnosis

Adjudication of final diagnoses was performed centrally in the core lab (University Hospital Basel) for all patients twice: once according to conventional cTn levels used onsite (this method was used in the initial analyses to examine the performance of hs-cTn assays)9 ,20–23 and once including levels of Roche hs-cTnT in order to take the advantage of the higher sensitivity and overall diagnostic accuracy offered by hs-cTn assays24 ,25 (this allows the additional detection of small AMIs that were missed by the adjudication based on conventional cTn assays). Two independent cardiologists reviewed all available medical records—patient history, physical examination, results of laboratory testing (including hs-cTnT levels), radiological testing, ECG, echocardiography, cardiac exercise test, lesion severity and morphology in coronary angiography—pertaining to the patient from the time of ED presentation to 90-day follow-up. In situations of disagreement about the diagnosis, cases were reviewed and adjudicated in conjunction with a third cardiologist.

AMI was defined and cTn levels interpreted as recommended in current guidelines.26 ,27 In brief, AMI was diagnosed when there was evidence of myocardial necrosis in association with a clinical setting consistent with myocardial ischaemia. Myocardial necrosis was diagnosed by at least one cTn value above the 99th percentile (or for the conventional cTn assays above the 10% imprecision value if not fulfilled at the 99th percentile) together with a significant rising and/or falling.27–29 The criteria used to define rise and/or fall in conventional cTn and hs-cTnT and the assumption of linearity are described in detail in the online supplementary data.

All other patients were classified as ‘no AMI’ for this analysis, including in this group the categories of unstable angina (UA), non-cardiac origin, cardiac but not coronary (eg, tachyarrhythmias, perimyocarditis) and symptoms of unknown origin with normal levels of hs-cTnT.

Measurement of hs-cTnT

Blood samples for determination of hs-cTnT (Roche) were collected at presentation to the ED in serum tubes. Additional samples were collected at 1, 2, 3 and 6 h. When treatment required transferring the patient to the catheter laboratory or coronary care unit, because the diagnosis of AMI was certain, serial sampling was disrupted. After centrifugation, samples were frozen at −80°C until assayed in a blinded fashion in a dedicated core laboratory. The Roche hs-cTnT assay was measured on the Elecsys 2010 (Roche Diagnostics). The limit of blank and limit of detection (LoD) were determined to be 3 and 5 ng/l, respectively. The 99th percentile of a healthy reference population was reported at 14 ng/l with an imprecision corresponding to 10% coefficient of variation (CV) at 13 ng/l. Levels below 5 ng/l were considered undetectable.30 Calculation of the glomerular filtration rate was performed using the abbreviated Modification of Diet in Renal disease formula.31

Investigational assays

For the determination of hFABP, we used the QuickSenshFABP assay (8sens.biognostic GmbH, Berlin, Germany), which is an immunochromatographic lateral-flow test applicable with whole blood, plasma or serum. The assay contains two different specific monoclonal antibodies for hFABP, one of which is gold-labelled and the other one biotinylated; results of the single measurements were evaluated and accurately quantified using the mobile optoelectronic reader system QuickSensΩ100 (8sens.biognostic GmbH), with an LoD of 0.6 ng/ml, a 99th percentile cut-off point of 5.76 ng/ml, measured in a population of healthy volunteers and a CV of less than 20% (20%CV) at 5 ng/ml.

Copeptin was measured using a commercial sandwich immunoluminometric assay (B.R.A.H.M.S LUMItest CT-proAVP, BRAHMS AG, Hennigsdorf/Berlin, Germany).32 Since the initial publication, the capture antibody was replaced by a murine monoclonal antibody directed to amino acids 137–144 (GPAGAL) of pro-Arginine-Vasopressin in order to improve the accuracy of the assay. The lower detection limit was 0.4 pmol/l and the functional assay sensitivity (20%CV) was <1 pmol/l. The 97.5 percentile in a healthy reference population was 16.4 pmol/l.33

Statistical analysis

Continuous variables are presented as means (±SD) or medians (with the IQR), and categorical variables as numbers and percentages. Continuous variables were compared with the use of the Mann-Whitney U test and categorical variables with the use of the Pearson's χ2 test. Receiver operating characteristic (ROC) curves were constructed to assess the sensitivity and specificity of measurements and to compare their ability to diagnose AMI. Logistic regression was used to combine hs-cTnT levels at presentation with early changes in hs-cTnT levels. The comparison of areas under the ROC curves (AUC) was performed as recommended by DeLong et al.34 In case of missing values, we made use of listwise deletion. The optimal cut-off values were determined by the Youden Index, pictured by the point farthest from the bisector of the graph. For the analysis of the prognostic value of the hs-cTnT assays, we did Kaplan–Meier analysis and presented cumulative survival rates at 2 years; 95% CIs were estimated by the SE. Furthermore, we performed a univariate Cox regression analysis for hFABP and multiple cox regression analysis, including elevation of hs-cTnT above the 99th percentile, adjusted for pre-existing coronary artery disease, age, sex and cardiovascular risk factors, which represented independent predictors for death (arterial hypertension, hyperlipidaemia, familiarly disposition of cardiovascular diseases and diabetes) in univariate regression models. All hypothesis testing was two-tailed; for Cox regression models, we included variables with p values of less than 0.1 in univariate regression and for all other testing p values of less than 0.05 were considered to indicate statistical significance. All statistical analyses were performed with the use of SPSS for Windows, V.20.0 (SPSS), and MedCalc software, V.10.3.0 (MedCalc).

Results

Characteristics of patients

Baseline characteristics of the 1074 patients with baseline values of hFABP and hs-cTnT are shown in table 1. Forty-three patients (4%) had ST-segment elevation AMI. Copeptin values were available for 1063 patients of the whole cohort and for 1021 patients without ST-segment elevation. The adjudicated final diagnosis was non-STEMI in 16% of the patients in the whole cohort, UA in 11%, cardiac non-coronary disease in 14%, non-cardiac symptoms in 47% and symptoms of unknown origin in 9% (see online supplementary figure S4).

Table 1

Baseline characteristics of the patients

Levels of hFABP according to diagnosis and hs-cTnT at presentation

Patients with AMI had significantly higher levels of hFABP than those with other diagnoses (median level 8.8 (IQR 4.2–25.5) vs 2.3 (IQR 1.5–3.6), respectively, p<0.001). Notably, patients with STEMI had significantly higher levels of hFABP than those with non-STEMI (median level 19.2 (IQR 8.4–61.8) vs 8.0 (IQR 3.4–21.4), respectively, p<0.001; see online supplementary figure S4). There was a moderate but significant correlation of hFABP with hs-cTnT, (r=0.61, p<0.001). About 82% of the patients with hs-cTnT above the 99th percentile had hFABP levels above the median in this cohort, whereas 67% of the patients with hs-cTnT below the 99th percentile had hFABP levels below the median of 2.7 ng/ml.

Diagnostic value of hFABP, copeptin and hs-cTnT as single markers

In the overall cohort, the diagnostic performance of hFABP was moderate to high, its AUC amounted to 0.84 (95% CI)  0.81 to 0.86), which was significantly lower than hs-cTnT (AUC 0.94, 95% CI 0.92 to 0.95, p<0.001 for comparison with AUC of hFABP alone). Copeptin as a single marker showed poor performance in the distinction of AMI (AUC 0.70, 95% CI 0.67 to 0.73, p<0.001 for comparison with hs-cTnT and hFABP, see figure 1).

Figure 1

Diagnostic performance of high-sensitive troponin T and the combination of cardiac troponin with heart-type fatty acid-binding protein (hFABP) in the early diagnosis of acute myocardial infarction (AMI). (A) Receiver operating characteristic (ROC) curves describing the prognostic performance of hFABP (red line), high sensitive troponin T (hs-cTnT, orange line) and a combination of hFABP with hs-cTnT (blue line) in the diagnosis of AMI at presentation. (B) ROC curves describing the prognostic performance of hFABP (red line), high sensitive troponin T (hs-cTnT, orange line) and the combination of hFABP with hs-cTnT (blue line) in the subgroup of patients with symptom onset within 3 h from presentation. This figure is only reproduced in colour in the online version.

In early presenters (within 3 h from symptom onset), hFABP alone was superior to copeptin (AUC 0.85 (95% CI 0.82 to 0.88) vs AUC 0.70 (95% CI 0.65 to 0.74), respectively, p<0.001) and inferior to hs-cTnT (AUC 0.92 (95% CI 0.89 to 0.94), p<0.001 for comparison with hFABP, see figure 1).

The diagnostic accuracy of hFABP alone was independent of the time from onset of the symptoms (all p>0.05; see figure 2). The explorative ROC-derived optimal cut-off for hFABP at presentation to discriminate AMI from other causes of chest pain was 4.2 ng/ml (for diagnostic parameters, see table 2).

Table 2

Diagnostic performance of hFABP and cardiac troponin (95% CI)

Figure 2

Diagnostic accuracy of heart-type fatty acid-binding protein (hFABP), high-sensitive cardiac troponin T (hs-cTnT) and copeptin at presentation according to time since the onset of chest pain. The area under the receiver operating characteristic curve is shown, according to the time since the onset of chest pain, for Roche hs-cTnT (orange line), hFABP (red line) and copeptin (green line), as well as the combinations of hs-cTnT with hFABP and copeptin (dashed blue and grey lines). This figure is only reproduced in colour in the online version.

Diagnostic value of hFABP and copeptin with hs-cTnT

In the overall cohort, the combination with hFABP decreased the diagnostic accuracy as quantified by the AUC curve for hs-cTnT (0.88 (95% CI 0.86 to 0.90) vs 0.94 (95% CI 0.92 to 0.95), p<0.001). In early presenters, the combination of hs-TnT with hFABP did not provide incremental diagnostic value compared with hs-TnT alone either (0.88 (95% CI 0.84 to 0.91) vs 0.92 (95% CI 0.89 to 0.94), respectively, p=0.035; see figures 1 and 2).

For copeptin, added to hs-cTnT, there was neither diagnostic improvement in the whole cohort (AUC for hs-cTnT and copeptin 0.88 (95% CI 0.86 to 0.90), p<0.001 for the comparison with hs-cTnT alone), nor in the subgroup of early presenters (AUC for hs-cTnT and copeptin 0.86 (95% CI 0.83 to 0.89), p<0.001 for the comparison with hs-TnT alone, results not shown). The diagnostic parameters of hs-cTnT and hFABP alone and their combination in the overall cohort are shown in table 2.

Diagnostic value of hFABP for unstable angina

The ability of heart-type FABP to discriminate UA from other non-cardiac causes of acute chest pain was very poor (AUC 0.57 (95% CI 0.52 to 0.63)).

HFABP and hs-cTnT during serial sampling

In serial measurements at presentation, 1, 2 and 3 h, the absolute values of hs-cTnT increased among patients with AMI, including STEMI and non-STEMI, whereas hFABP levels decreased during serial sampling. In patients with chest pain of non-cardiac origin, troponin values as well as hFABP values were stable (see online supplementary figure S5). Furthermore, during serial sampling, diagnostic accuracy of hFABP and copeptin in the diagnosis of non-ST-segment elevation AMI, as quantified by the AUC, decreased slightly (all p>0.05), whereas the diagnostic accuracy of cTnT assays increased during the first 2 h (all p<0.05; see online supplementary table S3A). The diagnostic performance of hFABP increased when hFABP levels at presentation were combined with absolute changes at 1 h (p<0.05; see online supplementary table S3B).

Prognostic value of hFABP

In Cox regression analysis of the whole cohort, adjusted for age, sex and cardiovascular risk factors, hFABP represented an independent predictor for death (HR 1.017 (95% CI 1.007 to 1.029), p=0.002). Also, when adjusted to hs-cTnT levels, age, sex and cardiovascular risk factors, hFABP had additional predictive value regarding mortality (HR 1.023 (95% CI 1.011 to 1.036), p<0.001) beyond hs-cTnT (p>0.05). This was also the case for copeptin (HR 1.004 (95% CI 1.002 to 1.006), p<0.001). In Kaplan–Meier analysis, the group of patients with elevated levels above the median of hFABP for the whole cohort (2.7 ng/ml) showed significantly higher mortality rates, especially within the first 3 months after hospital admission (p<0.001, compared with log-rank test; see online supplementary figure S6). The negative predictive value regarding 90-day, 1-year and 2-year mortality was 100% (99–100), 99% (98–100) and 98% (96–99), respectively, for an hFABP level below the median (χ2-test, p<0.001). The prognostic ability to predict mortality within 90, 360 and 720 days, quantified by the AUC, was 0.83 (95% CI 0.77 to 0.90), 0.80 (95% CI 0.78 to 0.83) and 0.79 (95% 0.76 to 0.81), respectively.

Prognostic value of hFABP in combination with hs-cTnT

We created four different subgroups according to the baseline levels of hFABP and hs-cTnT: (1) patients, who had hFABP below the determined median and hs-cTnT values below the 99th percentile (n=467), (2) patients with elevated hFABP levels above the median and hs-cTnT levels below the 99th percentile (n=213), (3) patients with hFABP below the median but elevated hs-cTnT levels above the 99th percentile (n=71) and (4) patients with elevated levels of hFABP and hs-cTnT at presentation (n=323).

In Kaplan–Meier analysis, the group of patients with elevated levels above the 99th percentile of hs-cTnT and above the median of hFABP showed significantly higher mortality than all other groups (survival probability within 2 y: 0.80 (95% CI 0.75 to 0.85) vs 0.97 (95% CI 0.96 to 0.98) vs 0.97 (95% CI 0.95 to 0.98) and vs 0.96 (95% CI 0.91 to 1), respectively, log-rank test: p<0.001; figure 3). The negative predictive value regarding 90-day, 1-year and 2-year mortality was 100, 98 and 98%, respectively, for hs-cTnT below the 99th percentile or hFABP level below the median (χ2-test, p<0.001; see figure 3).

Figure 3

Kaplan–Meier curves for all chest pain patients subdivided according to heart-type fatty acid binding protein (hFABP) and high-sensitive troponin T (hs-cTnT) levels at presentation. Kaplan–Meier analysis displaying cumulative 2-year survival during follow-up in patients with chest pain was subdivided into four groups: patients with hFABP levels below the median and hs-cTnT levels below the 99th percentile (blue line), patients with hFABP levels above the median and hs-cTnT levels below the 99th percentile (red line), patients with hFABP levels below the median and hs-cTnT levels above the 99th percentile (yellow line) and patients with hFABP levels above the median and hs-cTnT levels above the 99th percentile (green line) at presentation to the emergency department. This figure is only reproduced in colour in the online version.

Discussion

In this prospective, international multicentre study of 1074 consecutive patients presenting with acute chest pain to the ED, we scrutinised the diagnostic and prognostic value of hFABP. We report five major findings with the potential to improve the early diagnosis and risk stratification.

First, hFABP levels measured at presentation were significantly higher in patients with AMI than in patients with other final diagnoses. Notably, patients with STEMI had significantly higher hFABP levels than those with non-STEMI.

Second, the diagnostic accuracy of hFABP for AMI was moderate to high, but the additional use of hFABP did not significantly increase the diagnostic accuracy provided hs-cTnT. Interestingly, hFABP values in patients with AMI were highest at presentation and decreased in serial sampling, whereas hs-cTnT levels rose in patients with AMI. Furthermore, the diagnostic performance of hFABP to discriminate non-STEMI from other causes of chest pain than AMI seemed to be almost independent from symptom onset, which might indicate a more rapid release due to the small size and solubility of hFABP, in contrast to structurally bound cTn.8 ,9

Third, the ability of hFABP to discriminate UA from other, non-cardiac causes of acute chest pain was poor, which is contrary to other studies, suggesting that hFABP might be a marker of myocardial ischaemia even in the absence of necrosis.13 ,18 ,19

Fourth, hFABP at presentation was a powerful predictor of mortality, irrespective of cardiovascular risk factors, age, gender and hs-cTnT, comparable to copeptin in recent investigations.35 We hereby extend and corroborate previous studies.18 ,19

Fifth, in combination with hs-cTnT, risk stratification into four groups showed that patients with elevated levels of hs-cTnT and hFABP were particularly at risk regarding short-term and long-term mortality. Or in other words, in patients with elevated levels of only one biomarker, hs-cTnT or hFABP, mortality rates were similar to those with no elevation at all. Considering the increased incidence of patients with elevated cTn levels, originating from other diseases than AMI with the use of hs-cTnT assays, the combination of cTn with hFABP may address some of the challenges seen with the clinical application of hs-cTnT.36

The following limitations of the current study merit consideration: First, we evaluated the use of hFABP in combination with hs-cTnT. We hypothesise that our findings also apply to cTnI assays with hFABP. However, additional studies need to confirm this hypothesis. Second, we cannot comment on the patients with end-stage renal failure, as these patients were excluded from the study. Third, this observational study cannot quantify the clinical benefit associated with the increase in prognostic accuracy. To add this important information, interventional studies seem warranted.

In conclusion, the use of hFABP seems to improve early risk stratification, especially regarding short-term mortality, of unselected chest pain patients beyond hs-cTnT. However, it does not provide incremental diagnostic value in patients with symptoms suggestive of AMI but without indicative ST-segment elevations in the ECG.

Acknowledgments

The study was supported by research grants from the Swiss National Science Foundation (PP00B-102853), the Swiss Heart Foundation, Abbott, Roche, Siemens and the Department of Internal Medicine, University Hospital Basel.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • MR and RT contributed equally.

  • Contributors All co-authors have contributed significantly to the manuscript with regard to interpretation of the data and revising it for important intellectual content.

  • Funding The Swiss 21 National Science Foundation (PP00B-102853), the Swiss Heart Foundation, Abbott, 22 Roche, Siemens and the Department of Internal Medicine, University Hospital Basel.

  • Competing interests Dr Mueller has received research support from the Swiss National Science Foundation (PP00B-102853), the Swiss Heart Foundation, the Novartis Foundation, the Krokus Foundation, Abbott, Brahms, Roche and the Department of Internal Medicine, University Hospital Basel, as well as speaker honoraria from Abbott, Brahms and Roche. All other authors declare that they have no conflict of interest.

  • Ethics approval This study was conducted with the approval of the local ethics committees and performed in accordance with the principles of the Declaration of Helsinki.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.