Article Text

Download PDFPDF

Original article
Incremental prognostic value of early postoperative right ventricular systolic function in patients undergoing surgery for isolated severe tricuspid regurgitation
  1. Kyungil Park1,
  2. Hyung-Kwan Kim2,
  3. Yong-Jin Kim2,
  4. Goo-Yeong Cho2,
  5. Kyung-Hwan Kim3,
  6. Ki-Bong Kim3,
  7. Dae-Won Sohn2,
  8. Hyuk Ahn3,
  9. Byung-Hee Oh2,
  10. Young-Bae Park2
  1. 1Division of Cardiology, Department of Internal Medicine, Dong-A University College of Medicine, Busan, South Korea
  2. 2Division of Cardiology, Department of Internal Medicine, Seoul National University College of Medicine, Seoul, South Korea
  3. 3Department of Thoracic and Cardiovascular Surgery, Seoul National University College of Medicine, Seoul, South Korea
  1. Correspondence to Hyung-Kwan Kim, Division of Cardiology, Department of Internal Medicine, 28 Yongon-dong, Chongno-gu, Seoul 110-744, South Korea; cardiman73{at}


Background Preoperative right ventricular end-systolic area (RV-ESA) and haemoglobin level have been suggested to be independent predictors of long-term prognosis in patients undergoing corrective surgery for isolated severe tricuspid regurgitation (TR).

Aims To investigate whether early postoperative echocardiography provides useful prognostic information in addition to preoperative clinical and echocardiographic variables.

Methods 69 consecutive patients undergoing corrective surgery for isolated severe TR (60 women, mean 57.6±8.9 years) were studied. Comprehensive preoperative echocardiography was performed in all patients, with early postoperative echocardiography in all patients except one. During follow-up (median 40 months, range 6–86 months), clinical events were investigated, defined as operative mortality, cardiovascular death, repeated open heart surgery and readmission due to cardiovascular problems.

Results 28 patients (41%) were categorised as New York Heart Association class II, 36 (52%) as III and 5 (7%) as IV. 63 patients (91.3%) had undergone prior left-sided valve surgery. Seven (10.1%) patients died before discharge. Of the remaining 62 patients, three died during follow-up, and eight required readmission due to cardiovascular problems. On multivariate analyses using clinical and preoperative variables, RV-ESA (p=0.006) and haemoglobin level (p<0.001) were independent predictors of event-free survival. When early postoperative echocardiography variables were included, preoperative haemoglobin and early postoperative RV-FAC were predictors of long-term event-free survival. On receiver-operating characteristic curve analysis, early postoperative RV-FAC≥31% predicted event-free survival with a sensitivity of 90% and a specificity of 83% (p<0.001). The addition of early postoperative echocardiographic RV-FAC markedly improved the prognostic utility of the model containing preoperative haemoglobin level and echocardiographic RV-ESA (p<0.001).

Conclusion Early postoperative RV-FAC, measured by echocardiography, provided valuable information additional to preoperative RV-ESA and hemoglobin level that was useful for predicting long-term clinical events following corrective TR surgery.

  • Tricuspid valve insufficiency
  • surgery
  • treatment outcome
  • echocardiography
  • surgery-valve
  • tricuspid valve disease

Statistics from

Tricuspid regurgitation (TR) has long been neglected, based on the false belief that it is rare and that it is clinically insignificant. However, recent studies have clearly indicated that it is not a rare condition and that its prevalence is rapidly rising in patients who have undergone prior left-sided valve surgery.1–4 More importantly, the development of TR is closely related to exercise intolerance and augurs a dismal prognosis,5 6 even in the absence of left ventricular (LV) dysfunction or pulmonary hypertension.7 Although morbidity and mortality related to TR surgery remain high,8–10 surgical correction of severe TR results in substantial clinical and haemodynamic improvement, including reverse remodelling of the right ventricle (RV), when performed in a timely manner.11 12 Preoperative right ventricular end-systolic area (RV-ESA) measured by echocardiography and haemoglobin level have recently been suggested as indices to determine appropriate surgical timing for severe TR.11 Nevertheless, not all patients respond in the same way after TR surgery; even after successful procedures, some patients will not show any clinical or haemodynamic amelioration and, not surprisingly, will have poorer prognoses. This finding may be explained by the fact that latent, subtle RV dysfunction cannot be always unmasked pre-surgery due to the limitations of currently available imaging modalities, emphasising that assessment of RV function early after corrective TR surgery (the time when RV dysfunction might be clinically manifested, if present) should be helpful in this population. However, it remains unclear whether this approach provides meaningful information for predicting long-term outcomes in patients undergoing corrective TR surgery. Hence, we conducted this prospective study to investigate whether early postoperative quantitative assessment of RV function by echocardiography provides incremental value for the prediction of long-term prognosis of patients undergoing corrective TR surgery, in addition to preoperative clinical and echocardiographic variables.


Study sample

We prospectively recruited consecutive patients who underwent corrective surgery for isolated severe TR (ie, without any other valvular dysfunction) between March 2003 and April 2009. To be included in the present study, the following three criteria for severe TR had to be met based on the preoperative echocardiography: (1) TR jet >30% of right atrial (RA) area; (2) inadequate cusp coaptation; and (3) systolic flow reversal in the hepatic vein. Inadequate coaptation of the tricuspid valve was determined to be present when the gap between the septal and anterior leaflets of the tricuspid valve that was measured using zoomed images of the modified apical four-chamber view was visually identified and estimated to be ≥5 mm. Inadequate coaptation was linked to tricuspid annular dilation and apical tethering of the leaflets in all patients. Patients with significant left-sided valve disease who underwent concomitant left-sided valve surgery were excluded from the study. All patients enrolled in this study were free of significant coronary artery disease, as evidenced by preoperative invasive coronary angiography. Ultimately, a total of 69 patients (mean age 57.6±8.9 years; 60 women (87%)) were included in the current study. Of the 69 patients enrolled, 60 patients were included in our previous report.11 In all patients, clinical and echocardiographic follow-up was performed at least 6 months after corrective TR surgery at our institution. Information on clinical events was obtained by reviewing hospital records and by conducting telephone interviews. The study protocol was approved by the institutional review board of our hospital, and all patients gave written informed consent to participate in the study.

Clinical assessment, management and follow-up

Clinical events were defined as operative mortality (death within 30 days after surgery or before discharge), cardiovascular death, repeated open heart surgery, and readmission due to cardiovascular problems. Each patient's medical history and the results of a clinical examination were obtained at baseline by each attending physician. After TR surgery, all patients were managed and followed by their personal surgeons or physicians at our institution. Collection of the last follow-up data was completed in February 2010. No patients were lost to follow-up.

Echocardiographic examination

Comprehensive echocardiographic evaluation was performed before surgery in all patients using commercially available equipment (Sequoia, Siemens Medical Solutions or Vivid 7, GE Medical Systems). LV end-diastolic and end-systolic diameters and LV ejection fraction (EF) were measured by M-mode in the parasternal short-axis view at the papillary muscle level. Measurements of RV-ESA and RV end-diastolic area (RV-EDA) were made in the apical four-chamber view, and RV fractional area change (RV-FAC) was calculated using the following formula: RV-FAC=(RV-EDA−RV-ESA)/RV-EDA×100%.11 Great care was exercised not to procure foreshortened images of the apical four-chamber view. RA area was measured at end-systole in the apical four-chamber view. The TR jet area and RA area were obtained at the highest possible Nyquist limit in a given patient. Tricuspid annular diameter was defined as the widest diameter that could be measured from an end-diastolic still frame. Systolic tricuspid annular velocity was obtained by placing the sample volume of the pulsed wave tissue Doppler imaging at the lateral side of the tricuspid annulus. Pulmonary artery systolic pressure was measured by the sum of the peak systolic trans-tricuspid pressure gradient and estimated RA pressure. The diameter of the inferior vena cava and its respirophasic variation were measured 1.0–2.0 cm from the RA junction in subcostal view during the cardiac cycle to estimate RA pressure. The percentage decrease in diameter was used to determine mean RA pressure; the RA pressures were categorised as normal (at least 50% decrease), mildly elevated (dilated more than 17 mm with 50% decrease), moderately elevated (dilated more than 17 mm with less than 50% decrease), or severely elevated (dilated more than 17 mm without any collapse), as recommended by current guidelines.12 Early postoperative echocardiography was performed without intravenous fluid loading or tachycardia at a median of 11 days (IQR (7–15.5). All echocardiographic examinations were performed by a skilled sonographer and interpreted by an experienced cardiologist (K-HK or K-YJ). All echocardiographic measurements were averaged from three consecutive beats in patients with sinus rhythm and from five consecutive beats in patients with atrial fibrillation.

Statistical analysis

Data are expressed as mean±SD or median with IQR for continuous variables, and as numbers (%) for categorical variables, as appropriate. Differences between continuous variables in patients who did and did not experience clinical events were compared using the Student t test or Mann–Whitney test, depending on the results of normality testing by the Shapiro–Wilk test. For comparison of categorical variables, the χ2 test was employed. A Cox proportional hazards model with the use of forward selection based on the likelihood ratio test was implemented for multivariate analysis to determine which prognostic factors identified in the univariate analysis were significantly related to long-term clinical events. To assess cut-off values of the independent predictors of surgical outcomes, receiver-operating characteristic (ROC) curve analysis was used. The cut-off values were selected as those with the maximal sum of sensitivity and specificity. Event-free cumulative survival rates were plotted using the Kaplan–Meier method; comparisons were made between patients with and those without clinical events using the log-rank test. The incremental prognostic value of early postoperative RV-FAC, over preoperative haemoglobin level and RV-FAC, was tested via the likelihood ratio test of Cox models using a global χ2 value. All statistical analyses were performed using SPSS V.17.0 for Windows. A p value of <0.05 was taken as a cut-off value for statistical significance.


Baseline characteristics

Table 1 summarises the baseline characteristics of all 69 patients. Median follow-up duration after corrective TR surgery was 40 months (range 6–86 months). Sixty of the 69 patients (87%) were female, and preoperative ECGs showed atrial fibrillation in 58 (84%) of patients. Prior left-sided valve surgery had been performed in 63 patients (91.3%). At the time of left-sided valve surgery, tricuspid annuloplasty was concomitantly performed in 11 patients (16%). The cause of TR was functional in 55 patients (80%), rheumatic in 13 (19%) and congenital in 1 (1%). In terms of functional capacity, 28 patients (41%) were classified as New York Heart Association (NYHA) functional class II, 36 (52%) as class III and 5 (7%) as class IV. With regard to medications, diuretics were given in 64 patients (92.8%), most of whom took loop diuretics (50 patients, 72.5%) and aldosterone antagonists (55 patients, 79.7%). Digitalis was given in 48 patients (69.6%), ACE inhibitors or angiotensin receptor blockers were given in 10 (14.5%), and β-blockers were given in 10 (14.5%).

Table 1

Baseline clinical, laboratory, and echocardiographic characteristics of study patients*

Preoperative echocardiographic examinations yielded an average of LV end-diastolic diameter of 45.8±8.4 mm, LV end-systolic diameter of 29.5±6.6 mm, LV-EF of 57.6±8.9%, tricuspid annulus diameter of 44.2±7.7 mm, RA area of 53.1±20.1 cm2, TR area of 24.2±11.1 cm2, RV-EDA of 34.6±8.7 mm, RV-ESA of 20.3±6.5 mm and RV-FAC of 41.9±7.9%. Pulmonary artery systolic pressure was 40.8±8.8 mm Hg (table 1).

Measurement variabilities for RV-FAC

Inter- and intra-observer variabilities for RV-FAC were determined from five randomly selected patients by two independent investigators (K-HK and P-K). RV-FAC by these two independent observers were correlated with an r value of 0.91 (SE of the estimate=0.57). In terms of intra-observer variability, the correlation was r=0.93 (SE of the estimate=0.62) without a trend for over- or under-estimation.

Early postoperative echocardiography was successfully performed in all patients except one, who died 2 days after corrective TR surgery.


Eight patients (12%) underwent tricuspid valve repair, and 61 patients (88%) underwent tricuspid valve replacement, with use of tissue valves (n=36) or mechanical valves (n=25). Concomitant Maze operation was performed in 16 patients. The total cardiopulmonary bypass time was 189±52 min, and the aortic cross-clamp time was 104±35 min. No patients died during surgery.

Postoperative outcomes

Of the 69 patients enrolled, 10 died during the follow-up period. Seven of the 10 patients died before discharge, thus operative mortality was estimated to be 10.1%. Of the remaining 62 patients who survived TR surgery, one died of heart failure, one died of hemorrhagic stroke and one patient experienced unexpected sudden death. Eight patients required readmission due to cardiovascular problems, two of whom had an event of embolic infarction, and six had congestive heart failure. Thus, 51 patients (73.9%) remained event-free at the end of follow-up.

Table 2 shows the clinical characteristics between patients with and without clinical events. Compared with patients who did not develop clinical events, patients with clinical events showed a lower prevalence of NYHA functional class II (24 patients (47.1%) without clinical events versus 4 patients (22.2%) with clinical events, p=0.048), a higher serum creatinine level (p=0.07), a lower albumin level (p=0.02), a lower cholesterol level (p=0.047), a lower haemoglobin level (p=0.001) and a lower platelet count (p<0.001). Among preoperative echocardiographic parameters, patients with clinical events were characterised by having a higher RV-ESA (p=0.03) and a lower RV-FAC (p=0.01). With regard to early postoperative echocardiographic parameters, LV end-systolic diameter (p=0.03) and RV-FAC (p<0.001) were the two variables that best discriminated between patients with and without clinical events.

Table 2

Comparison between patients with and without clinical events

Predictors of clinical outcome in the long term

On univariate survival analyses using a Cox proportional hazards model, NYHA functional class (as a categorical variable; NYHA II vs III or IV), blood urea nitrogen, serum creatinine, serum albumin, total cholesterol, haemoglobin, platelet count, preoperative RV-EDA and RV-ESA, preoperative RV-FAC, early postoperative LV end-systolic diameter, early postoperative RV-ESA and early postoperative RV-FAC were significantly associated with subsequent clinical events (table 3). When clinical and preoperative echocardiographic variables were taken into consideration, haemoglobin level (p<0.001) and RV-ESA (p=0.006) emerged as independent predictors of clinical events in a multivariate Cox proportional hazards model. A Cox proportional hazards model encompassing clinical and preoperative as well as early postoperative echocardiographic variables showed that preoperative haemoglobin (p=0.003) and early postoperative RV-FAC (p<0.001) were the two most important predictors of clinical events (table 3). In ROC curve analyses, we found that early postoperative RV-FAC ≥31% effectively predicted event-free survival, with a sensitivity of 90% and a specificity of 83% (p<0.001, area under the curve=0.93) (figure 1). The mean event-free survival rate estimated by Kaplan–Meier analysis was longer in patients with early postoperative RV-FAC ≥31% (65.7±2.5 months) than in those with early postoperative RV-FAC <31% (29.3±7.5 months) (p<0.001) (figure 2). The 2-year event-free survival rate was 92.2% in patients with early postoperative RV-FAC ≥31% (figure 2).

Table 3

Univariate and multivariate analysis of clinical and preoperative as well as early postoperative echocardiographic predictors of clinical events

Figure 1

Receiver-operating characteristics (ROC) curve for long-term event-free survival in relation to early postoperative right ventricular fractional area change (RV-FAC), preoperative haemoglobin level and preoperative RV end-systolic area (RV-ESA).

Figure 2

Kaplan–Meier survival curve comparing event-free survival rate according to the early postoperative right ventricular factional area change (RV-FAC).

On the other hand, preoperative haemoglobin level ≥11.5 g/dl showed sensitivity and specificity of 73% and 68% (p<0.001, area under the curve=0.78), respectively, for predicting event-free survival. With regard to preoperative RV-ESA, sensitivity and specificity were 63% and 69%, respectively, with an RV-ESA of 20 cm2 as the cut-off value (p=0.02, area under the curve=0.68). Preoperative NYHA functional class failed to demonstrate its value as an independent predictor of long-term outcome on multivariate Cox regression analysis (p=0.39) (figure 3).

Figure 3

Long-term survival according to preoperative NYHA functional class. Despite no statistical difference between the two groups, there was a clear separation of the two survival curves throughout the follow-up period.

Incremental value of early postoperative RV-FAC in the prediction of clinical events after corrective TR surgery

We attempted to evaluate the benefit of assessing early postoperative RV-FAC for the prediction of future clinical events following corrective TR surgery in addition to laboratory and preoperative echocardiographic parameters (haemoglobin level and preoperative echocardiographic RV-ESA). As figure 4 shows, the addition of early postoperative echocardiographic RV-FAC markedly improved the prognostic utility of a model containing preoperative haemoglobin level and echocardiographic RV-ESA.

Figure 4

Incremental prognostic value by Cox proportional hazards model presented as a global χ2 value. The addition of early postoperative right ventricular fractional area change (RV-FAC) resulted in significant incremental improvement in the predictive value on preoperative haemoglobin and preoperative right ventricular end-systolic area (RV-ESA).


We investigated the prognostic utility of early postoperative echocardiography for patients with severe TR undergoing corrective surgery in this prospective study. Preoperative RV-ESA and haemoglobin level were both found to be independent predictors of late event-free survival over a median follow-up of 40 months. Of note, preservation of RV systolic function as measured by echocardiography early after corrective surgery was identified as the most effective and powerful prognosticator of event-free survival and provided incremental prognostic value in addition to preoperative haemoglobin level and RV-ESA.

Based on the mistaken belief that TR is not a significant disease and does not alter exercise capacity or long-term prognosis, corrective surgery remains low on the list of treatment priorities.13 Most patients suffering from severe TR probably receive only diuretics for alleviating RV preload, without consideration of corrective surgery. However, this ‘carefree’ view is no longer justified, because there is considerable evidence that severe TR is becoming a significant clinical burden, and is associated with considerable complication and mortality rates when treated only by medical management.2 7 12 14 15 Although timely correction of severe TR is expected to maintain RV systolic function and thus carry a favourable prognosis,11 determining optimal surgical timing is still a great challenge, primarily due to the difficulty in early detection of subclinical RV dysfunction, even using currently available state-of-the-art technologies such as echocardiography or cardiac magnetic resonance (CMR) imaging.12 Recently, preoperative RV-ESA has gained credit as a useful index for determining optimal timing for TR surgery,11 on the basis of relative independence of preload as demonstrated in mitral regurgitation.16 17 We confirmed in the present study, enrolling a larger number of patients than a previous one with an extended follow-up period, that preoperative RV-ESA is an independent predictor of long-term prognosis. However, the prognostic power of preoperative RV-ESA decreases as follow-up period increases (63% vs 73% for sensitivity and 69% vs 77% for specificity). A similar trend was also observed for preoperative haemoglobin level (73% vs 73% for sensitivity and 68% vs 83% for specificity). More importantly, although the proportion of patients with adverse outcome increases with each increment in preoperative RV-ESA and/or decrement in preoperative haemoglobin level, small preoperative RV-ESA values and/or high haemoglobin levels do not always preclude poor outcomes. Taken together, preoperative RV-ESA and haemoglobin level are insufficient for forecasting long-term event-free survival, although they are valuable for determining appropriate surgical timing.

Determinants of long-term outcomes in patients undergoing corrective TR surgery are multifactorial, incorporating at least two important components: (1) pre-surgical factors; and (2) post-surgical factors reflective of the quality of TR surgery. Surgical success, based on long-term clinical outcome, might be as critical as preoperative factors for determining the adequacy or quality of TR surgery. This is a rational argument, given the high perioperative mortality.8 9 11 The geometric changes in RV and tricuspid valve apparatus observed after TR surgery, both of which are significantly linked to unexpected surgical outcomes, also support this concept.18 The current study validated this concept by showing that an early postoperative RV-FAC of 31% most effectively discriminated patients with clinical events from those without, with high sensitivity and specificity. The prognostic implications of early postoperative RV-FAC were further advocated by a marked escalation of the global χ2 value that occurred after adding early postoperative RV-FAC to a model involving preoperative RV-ESA and haemoglobin level (figure 3). Hence, assessment of early postoperative RV systolic performance may be a practical strategy for segregation of patients with and without clinical events in the long term by providing incremental prognostic information additional to preoperative RV-ESA and haemoglobin level. Early postoperative RV-FAC of 31% constitutes a potentially useful threshold for determining whether TR surgery is successful or not.

Markers suggestive of systemic venous hypertension were reported to be associated with increased risk for early and late phase events in TR patients.19 Given that preoperative haemoglobin level is a mirror indirectly indicating the duration or severity of systemic venous congestion secondary to RV dysfunction,11 it is conceivable that early detection of RV dysfunction should be the main focus for determining optimal surgical timing. Thus, it is not surprising that measurement of preoperative haemoglobin level and RV-ESA by echocardiography (another marker of RV systolic function) is of critical value for clinical decision-making regarding optimal surgical timing, although by itself it is not sufficient. After completing corrective TR surgery, quantitative assessment of RV-FAC during the early postoperative stage (approximately 2 weeks after surgery) is recommended, not merely for a simple measurable index of the surgical success but, more importantly, for assessing long-term prognosis and risk stratification. Therefore, we believe that these three factors, two preoperative and one early postoperative, should be measured in every TR patient; the two preoperative factors can aid in the decision as to whether to proceed with TR surgery, while the early postoperative factor can be adopted as a strong prognosticator of long-term outcomes.

The symptomatic status of each individual patient is a key component for determining whether or not to proceed with surgery in cases of chronic severe mitral regurgitation.20 If symptoms have already developed in patients with chronic severe mitral regurgitation, surgery frequently leaves residual postoperative LV systolic dysfunction, which is, in many cases, associated with poor prognosis.21 To avoid this devastating event, early surgical correction of severe mitral regurgitation was suggested.22 23 Similar concepts might be applicable in the setting of severe TR. However, despite clear separation of the survival curve between NYHA functional class II versus III or IV (figure 4), statistically significant differences were not achieved after multivariate analyses. In contrast to mitral regurgitation, which induces volume overload to the LV, TR takes place in the RV, which effectively tolerates volume overload.24 Effective handling of volume overload by the RV may contribute to the late onset of symptoms, even in patients with severe TR and incomplete coaptation, and thus subclinical RV dysfunction may frequently be present when surgery is planned, resulting in ‘too late’ surgery in many patients. However, although preoperative NYHA class failed to demonstrate an independent contribution to predicting event-free survival, we still believe that functional capacity should be evaluated at every visit, given the clear separation of the survival curve according to NYHA functional class. This issue should be re-evaluated in another study including a larger number of patients than the present study.

Several limitations of the study should be addressed. First, the aetiologies of TR were heterogeneous, although it was largely functional in origin. However, we prospectively and consecutively recruited patients undergoing surgery for treatment of severe TR and thus believe that our patients are representative of the real TR population seen in clinical practice. Second, the absolute number of patients enrolled is small, and thus definite conclusions cannot be drawn. However, in comparison with prior studies, the sample size of the current study is relatively large and was followed-up for the longest period. Third, because our study had no control group, results of the present study cannot provide advantages of isolated TR surgery over maximal medical therapy in improving prognosis of patients. However, our study was initially designed not to prove the effectiveness of isolated TR surgery, but to evaluate the clinical implications of early postoperative echocardiographic findings in the prediction of long-term outcome additional to preoperative clinical and echocardiographic variables in patients undergoing isolated TR surgery. As a result, the present study highlights the clinical importance of preoperative as well as early postoperative RV systolic function as determinants of long-term outcome in isolated TR surgery. Based on the present study, early postoperative RV-FAC can be used as a surrogate indicator for long-term outcome in future studies. Fourth, we exclusively enrolled patients undergoing corrective surgery for severe TR. Multivalve disease complicated by significant TR is, however, more common in our daily clinical practice rather than isolated TR,25–27 which may limit the application of the present findings to real world clinical practice. Nevertheless, given the increasing recognition of late development of severe functional TR long after left-sided valve surgery as well as its impact on clinical outcome,1 2 6 8 10 14 we believe that our study is of clinical relevance. A variety of possible combinations of other valve lesions in the setting of severe TR is another issue that should be evaluated in the near future. Finally, we used echocardiography in lieu of CMR for assessing preoperative and postoperative RV haemodynamics. However, the relatively high cost of CMR is a pragmatic obstacle to its routine performance in patients before and after corrective TR surgery. Most patients with claustrophobia or implanted pacemaker cannot undergo CMR, also limiting its application.

In conclusion, early postoperative RV-FAC measured by echocardiography provided useful information in addition to preoperative RV-ESA and haemoglobin level for predicting long-term clinical events in patients undergoing corrective surgery for severe TR. An early postoperative RV-FAC of 31% constitutes a potentially useful cut-off value for determining the success of TR surgery. Our findings suggest that this index should be used as an integral component in the routine postoperative care of TR patients, given its high sensitivity and specificity for the prediction of long-term event-free survival.



  • Funding This study was partly supported by grants from the Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of Korea (A090064) and Handok Pharmaceutical 2010 Research Fund.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Seoul National University Hospital.

  • 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.