Diagnostic Accuracy of Transthoracic Echocardiography and Transesophageal Echocardiography for Ruptured Chordae Tendineae: A Meta-Analysis

Echocardiography is signicant for the diagnosis of ruptured chordae tendineae (RCT) for which transesophageal echocardiography (TEE) is always better than transthoracic echocardiography (TTE), but the diagnostic accuracy of TTE still remains confusing. A meta-analysis of included papers was performed to evaluate the diagnostic accuracy of TTE and TEE, with surgical ndings of RCT as the gold standard of patients with suspected RCT.


Abstract Background
Echocardiography is signi cant for the diagnosis of ruptured chordae tendineae (RCT) for which transesophageal echocardiography (TEE) is always better than transthoracic echocardiography (TTE), but the diagnostic accuracy of TTE still remains confusing.

Methods
A meta-analysis of included papers was performed to evaluate the diagnostic accuracy of TTE and TEE, with surgical ndings of RCT as the gold standard of patients with suspected RCT.

Results
The literature search yielded 862 papers, 6 met the inclusion criteria, included 505 patients, for detecting RCT, the sensitivity and speci city of TTE were 48% (95% CI: 35-62%) and 98% (95% CI: 89-100%), and those of TEE were 99% (95% CI: 64-100%) and 94% (95% CI: 87-98%), Conclusions TTE is not highly sensitive for the detection of RCT but is highly speci c, so there is a potential of missing diagnosis, and subsequent TEE is almost required in highly suspected patients.

Background
Ruptured Chordae Tendineae (RCT) is increasingly reported as one of the important causes of severe mitral regurgitation (MR) [1], which is a progressive disease with acute clinical symptoms that eventually requires mitral valve (MV) surgery [2].
Echocardiography is widely used in diagnosing RCT, which has an irreplaceable position [3]. The de nite diagnosis plays a decisive role in the formulation of treatment plans. It is generally known that the methods we used to detect RCT include transesophageal echocardiography (TEE) and transthoracic echocardiography (TTE). However, compared with TTE, TEE has better views and intensive resources, but uncomfortable for patients and associated with complication risks [4]. Under this situation, TTE is particularly important for the diagnosis of RCT.
A series of studies shown the diagnostic performance of TTE and TEE for RCT. Previous studies demonstrated that the sensitivity of TEE was higher than that of TTE for the detection of RCT, which could reach 100% [5,6]. Some of the studies showed the high sensitivity of TTE for the diagnosis of RCT, the highest could reach 90% [7]. However, others showed the low sensitivity of TTE for detection of RCT, the lowest was 0% [6]. According to the large range differences in sensitivity of TTE, we want to evaluate the actual value of TTE and TEE in the identi cation of the accuracy of the RCT causing MR. Therefore, we conducted this meta-analysis to evaluate the diagnostic value of TTE and TEE using RCT ndings on surgical as the gold standard. And discussing whether TEE can be replaced by TTE in diagnosing RCT.

Search strategy
We searched the databases of Embase, Pubmed and Cochrane library for relevant English studies up to July 20, 2019. Terms of "Chordae Tendinae'' or "Chorda Tendinea rupture", "echocardiography", "sensitivity" OR "speci city speci city" OR 'diagnostic accuracy' OR 'surgery' were used. Moreover, the reference lists of the retrieved systematic and narrative reviews were also manually searched to identify additional relevant studies.

Study selection
Studies were included if they met all of the following criteria: 1. clinically suspected MR due to any cause, 2. surgery ndings as the gold standard for RCT, 3. su cient data to allow extraction of a 2 × 2 diagnostic table, 4. papers after the year of 1989. Reviews, letters, case reports and commentaries were excluded.
Two authors (Dong, FJ and Liao, LS) with the same expertise and experience selected the eligible studies independently. Discrepancies between them were resolved by the 3rd author (Chen LX).

Data Extraction and Quality evaluation
All papers were screened by titles and abstracts. Full text and reference lists of relevant papers ful lling the inclusion and exclusion criteria were reviewed in detail. Data for 2 × 2 tables were extracted.
For each included full-text article, quality was evaluated by the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) or potential biases.

Statistical analysis
For detections of RCT on TTE and TEE, speci city, sensitivity, and likelihood ratios were calculated from the abstracted data for each study.
All the analysis was performed by Stata 14.0 (Stata Corp, College Station, TX, USA) with Midas module was used to pool statistics indexes and draw statistical graphs for TTE and TEE separately, such as sensitivity, speci city, the pooled forest graph of the negative likelihood ratio (NLR), the positive likelihood ratio (PLR), and the diagnostic odds ratio(DOR) with corresponding 95% CI (con dence intervals), the area under the summary receiver operating characteristic curve (SROC). 95% CI was calculated for speci city, sensitivity and likelihood ratios using standard methods. SROC was used to examine the diagnostic value of TTE and TEE. Threshold effect was tested by SROC and Spearman correlation coe cient.
RevMan 5.3 (The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark) was used to evaluate the included papers. The inconsistency index (I 2 ) and Cochrane Q statistic (p) were used to estimate the heterogeneity across the included studies. If Cochran Q statistic value p > 0.1 or inconsistency index I 2 < 50%, a xed-effect model will be used; otherwise, a random effects model was selected. Deek's funnel plot asymmetry test will be used to test the potential publication bias, with p < 0.10 for the slope coe cient indicating signi cant asymmetry.

Literature searches
We yielded 862 papers, after excluded the duplications, then screened by title and abstract, nally 15 papers were assessed by full-text for eligibility. 5 papers were further excluded, due to analyzed the TTE or TEE only. 3 papers were further excluded because the authors compared TTE with 3DTEE. 1 articles did not contain enough information to extract 2 × 2 table for analysis .Therefore, 6 studies were ultimately met the inclusion criteria. The screening ow was shown in Fig. 1.

Methodology quality assessment
The QUADAS-2 checklist was used. The included studies' quality was judged to be high (Fig. 2). All the quality assessment items of the included studies had a low risk of bias.
Publication bias Publication bias was tested by Deeks' funnel plot, showed that studies were distributed symmetrically and no clear evidence of publication bias (p = 0.10 in TTE, and 0.43 in TEE, Fig. 7). Regression test of lnDOR against 1/Effective Sample Size 1/2 indicated no obvious small-study bias in our meta-analysis.

Discussion
Our studies ndings demonstrated that for detecting RCT in all patients, TTE had a sensitivity of 48% and speci city of 98%, so it has a likelihood to miss almost 50% potential RCTs. For TEE, the sensitivity and speci city of it can reach a high level, which is 99% and 94% respectively. That means TEE is pretty matched with surgical ndings of previous studies [8]. Monin.et al. [7]reported the diagnosis accuracy of RCT con rmed by TTE could reach 90%. Conversely, Alam.et al.
[6] reported that RCT was seen in none of 23 patients (sensitivity = 0%). In brief, the reported sensitivity of TTE for detecting RCT has varied considerably. Herein, our study of 6 observational studies, which showed differences in sensitivity of TTE for the diagnosis of RCT that ranged from 0-65% [3,5,6,[9][10][11]. The signi cant heterogeneity based on I2 for TTE analyses from the Alam in 1991, probably because the sample size is small.
Rupture of mitral chordae tendineae is one of the causes of acute mitral valve regurgitation, which can lead to acute pulmonary edema and cardiogenic shock [12]. It is one of the most common valvular heart diseases, which is secondary to Marfan syndrome, connective tissue disease, coronary heart disease, congenital heart disease, infective endocarditis, rheumatic heart disease and degenerative valvular disease [13]. It often occurs because of the weakness of mitral chordae tendineae, papillary muscles, and abnormal valve position or shape. Primary person refers to the mitral valve lea et and annular disease caused by degeneration. It is also one of the valvular diseases treated surgically [14]. Early diagnosis of RCT has great signi cance for treatment, which can save the patient's life. Long-term follow-up shows that patients with RCT and MR are more likely to use valve repair [15]. Valve repair has more advantages than valve replacement, such as lower mortality, higher long-term survival rate, lower risk of thrombus embolism and bleeding, better left ventricular systolic function after the operation. The etiology, lesion location and characteristics of mitral valve prolapse are important factors determining the surgical method, among which, mitral valve posterior lobe prolapse and chordae tendineae ruptures are more suitable for the application of valve repair [16]. Therefore echocardiography it is very signi cant, in addition to diagnosis, it also needs to understand the location of the diseased lobe and the pathological characteristics of its accessory structure before the operation.
TEE and TTE are important for the diagnosis of RCT [17]. TEE has a better view because the esophagus is closer to the heart, TEE can avoid the interference of thoracic morphology and lung, and the TEE probe has a higher discrimination rate, so it is easier to view the morphology, texture and motor characteristics of the diseased lobe and chordae tendineae [18]. Therefore, the accuracy of TEE in the diagnosis of valvular heart disease is higher than that of TTE [19]. Besides, TEE can have side effects that make patients feel unpleasant as mentioned before, so inspectors should not overuse TEE method to increase the burden on patients. Yu HT et al. [20] did a study in 2013 showed a high prevalence of unrecognized RCT in mitral valve prolapse patients undergoing valve surgery. RCTs were not found in 124 patients, 73 have no RCT observed at surgery, but 51 surgically proved RCT, which means TTE still has a large chance to miss many subtle RCTs even with rapidly developing imaging facilities.
As we have known the typical echocardiography manifestations of mitral valve prolapse and chordae tendineae rupture are the tip of the diseased valve lobe or the chordae tendineae partially ruptured moving back and forth between the atrioventricular in the following cycles [21].The ruptured end of the systolic chordae tendineae was under the lobe of the healthy side and was impacted by the regurgitation tract. Its residual end pointed to the atrial wall or atrial septum of the healthy side, while it returned to the ventricular chamber during diastole [22]. And, when the course of disease gets longer, the calci cation of disease change place is heavier; TTE is not easy to detect RCT. Under this situation, TTE it is easier to miss diagnoses.
Combined with our research, we had a nding. Although with the developments in the resolution of TTE over the past 30 years, the sensitivity of TTE to detect RCT has no signi cant improvement. Complex co-combination of valve conditions occurs with calci cation or infectious endocarditis. Maybe TTE has a high sensitivity for the typical signs but not for complicated or untypical signs.
Our study also has several clinical implications. For patients with mitral regurgitation, TTE should be the recommended examination for all patients because of convenient and its high speci city. In these patients, if TTE is highly indicative of RCT, the diagnostic accuracy is high because TTE is highly speci c to RCT [23]. Conversely; the absence of valve abnormalities in TTE mostly reduced the likelihood of RCT, so there is no need for TEE. This can eliminate needless hurts that are associated with cost and risk for complication. However, the decision on whether to continue with TEE or not should be determined by the patient's clinical symptoms and doctor. When the valve condition is complex, TTE is di cult to distinguish from the abnormal valve as if due to degenerative mitral valve diseases, vegetation or RCT, TEE should be performed.
This study had several limitations due to potential biases. We included patients that constructed both TTE and TEE could miss many studies that underwent TTE or TEE only.

Conclusions
TTE is not highly sensitive for the detection of RCT, but is highly speci c,so there is a potential of missing diagnosis, and subsequent TEE is almost required in highly suspected patients.

Abbreviations
RCT: Ruptured Chordae Tendineae; TEE:transesophageal echocardiography; TTE:transthoracic echocardiography; MR:mitral regurgitation; Figure 1 Flow chart of study selection process Forest plots for sensitivity and speci city for TTE(A) and TEE(B) of diagnosis accuracy of RCT.

Figure 4
Forest plots for positive likelihood ratio and negative likelihood ratio for TTE(A) and TEE (B).

Figure 5
Hierarchical summary receiver operating characteristic curve (HSROC) of diagnosis accuracy in TTE (A) and TEE (B).

Figure 6
Fagan's Nomogram shows that virtual touch tissue quanti cation as a screening tool for diagnostic accuracy in TTE A pretest probability at 20%, and TEE B pretest probability at 20%.