Article Text

Download PDFPDF

Proportionate or disproportionate secondary mitral regurgitation: how to untangle the Gordian knot?
  1. Victor Kamoen,
  2. Simon Calle,
  3. Marc De Buyzere,
  4. Frank Timmermans
  1. Cardiology, University Hospital Ghent, Ghent, Belgium
  1. Correspondence to Professor Frank Timmermans, Department of Cardiology, University Hospital Ghent, Ghent 9000, Belgium; frank.timmermans{at}


Recent randomised percutaneous mitral intervention trials in patients with heart failure with secondary mitral regurgitation (SMR) have yielded contrasting results. A ‘relative load’ or ‘proportionality’ conceptual framework for SMR has been proposed to partly explain the disparate results. The rationale behind the framework is that SMR depends on the left ventricular dimension and not vice versa. In this review, we provide an in-depth analysis of the proportionality parameters used in this framework and also discuss the regurgitant fraction. We also consider haemodynamic observations in SMR that may affect the interpretation and comparisons among proportionality parameters. The conclusion is that the proportionality concept remains hypothetical and requires prospective validation before envisaging its use at individual patient level for risk stratification or therapeutic decision-making.

  • mitral regurgitation
  • echocardiography
  • cardiac imaging and diagnostics

Statistics from

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.


Secondary mitral regurgitation (SMR) is a diagnostic and therapeutic challenge.1 Because SMR usually develops as a consequence of left ventricular disease in patients with heart failure (HF),2 guideline-directed medical treatment (GDMT)3 and cardiac resynchronisation therapy are recommended,4 but surgical intervention of SMR in patients with HF has not provided survival benefit.5 Percutaneous mitral intervention with MitraClip in SMR has shown promising results,6 but two recent randomised trials (MITRA-FR and COAPT)7 8 showed discordant results which triggered debate on the differences in trial design and patient characteristics between these studies.9 The concept of ‘relative load’ or ‘(dis)proportionality’ in SMR was proposed as a framework to partly explain the disparate results between MITRA-FR and COAPT.5 10 This concept is based on the improved outcomes of patients with HF with ‘disproportionate’ SMR (COAPT) following MitraClip on top of GDMT, whereas in ‘proportionate’ SMR (MITRA-FR), SMR may be an innocent bystander and targeting these patients with MitraClip may not alter their clinical outcome compared with GDMT alone.10 11

To define the proportionality concept, SMR severity has been numerically modelled by introducing proportionality coefficients.9 11–13 These parameters normalise the traditional metrics of SMR (such as the effective regurgitant orifice area (EROA) and the regurgitant volume (RV)) for left ventricular end-diastolic volume (LVEDV) and LV function, and may therefore better reflect the true haemodynamic burden of SMR.9 In addition to the proportionality coefficients EROA/LVEDV, RV/LVEDV, the regurgitant fraction (RF) as another measure for ‘relative load’ also gained significant interest in the field of SMR, but the theoretical constructs remain hypothetical and await further proof-of-concept. We here provide an in-depth analysis of these parameters and we highlight the strengths and potential issues when considering these parameters for grading SMR and decision-making in clinical practice.

Defining severe SMR

SMR occurs frequently in patients with HF with reduced ejection fraction (EF). SMR is caused by an incompetent mitral valve because the structural and functional alterations of the LV create an imbalance between mitral leaflet tethering and (synchronised) closing forces.1 In addition, evidence suggests that intrinsic structural remodelling (‘primary-like’ component in SMR) or inappropriate growth of the mitral leaflets themselves may contribute to MR in these patients.14 15

Patients with SMR mostly have dilated hearts with advanced LV dysfunction, and it is believed that SMR itself imposes a volume load on the diseased heart that initiates a vicious cycle and negatively affects the natural history of these patients.16 This concept of ‘valvular-driven heart failure’ or ‘SMR-begets-SMR’16–20 is appealing to explain excess morbidity and mortality in patients with HF with SMR.21–24 In addition, the regurgitation may worsen pulmonary pressures and reduce forward stroke volume (FSV), which may contribute to adverse prognosis in SMR. However, because SMR is grossly a ventricular disease and only accounts for a small increase of LV volume5 12 and mitral intervention has yielded variable results on LV reverse remodelling,5 the ‘volume overload’ or ‘MR-begets-MR’ hypothesis is controversial, but there are arguments to support SMR being both a consequence5 12 and a driver of a vicious cycle of more severe SMR and HF.16–18 25 The latter implies a formidable challenge in identifying such patients for timely mitral intervention during the possible window of opportunity,19 because GDMT at later stages of severe SMR17 or late correction of SMR26 may not relieve the associated hazard of mortality even if SMR regresses.

Defining severe SMR in patients with HF is a challenging task, because no gold standard method exists for grading MR and because echocardiographic grading of SMR is fraught with difficulties.27 Patient characteristics and comorbidities in patients with HF also complicate the use of a SMR ‘severity’ threshold or decision for mitral intervention. Therefore, an association of SMR severity with risk does not confirm that risk as being treatable, as this may depend on potential confounders in patients with SMR such as coexisting NYHA class IV28 or EF <30 %,28 29 that may prohibit beneficial effects of mitral intervention. In addition, the dynamic nature of SMR1 and the inability to accurately estimate genuine LV contractility30 (and its reserve) in SMR make it difficult to predict the individual impact of SMR on the LV structure and function. Finally, as a continuous linear relationship may exist between MR severity and outcome,22 defining a cut-off for severe SMR remains an arbitrary construct (but this applies for many other parameters in medicine). For all these reasons, it is very challenging to grade SMR based on the attributable risk at individual patient level. Moreover, if we would consider the independent nature of SMR from a statistical perspective, then the ‘independency’ of SMR with regard to outcome may still have to compete with unknown, unmeasured or immeasurable confounders. Finally, if SMR would be a reasonable legitimate for mitral valve intervention, then this should translate into improved prognosis, but mitral intervention has yielded controversial results with no clear survival benefit.5 Yet, as patients with HF have a dismal prognosis, the target for mitral intervention may not be improving survival, but symptom relief and improving quality of life.5

All these considerations may explain why (a) guidelines on SMR strongly depend on expert opinions; (b) various definitions or SMR severity cut-offs have been used and (c) controversy exists on the role of SMR with regard to outcome and benefit from mitral valve intervention.31

The EROA/LVEDV and RV/LVEDV proportionality coefficients

Proportionate SMR refers to the situation where the extent of EROA or RV is expected based on the degree of LV remodelling or LV size (LVEDV), and the LV enlargement is generally larger than might be expected from SMR alone (‘MITRA-FR profile’).11 12 Disproportionate SMR refers to an out-of-proportion SMR relative to LV size, that is, ‘there is more SMR than expected by LV size’ (high ratio of EROA/LVEDV or RV/LVEDV; ‘COAPT patient’) (figure 1 and table 1 for examples). This supports the view that SMR is a surrogate for LV disease in proportionate SMR and thus the myocardium itself, but not SMR, is amenable for successful GDMT.12 This contrasts with disproportionate SMR, where GDMT may be less effective, because additional factors may be involved in the pathogenesis of SMR in these patients10 11 (figure 1).

Figure 1

Two patients with SMR (Patients 1 and 2 in table 1) are shown. Above, an example of SMR proportionate to the degree of LV size is shown. Below, SMR disproportionate to LV size (due to a focal inferolateral myocardial lesion) is shown. For each example, the first and second panels show the LV and the mitral valve features and the third panel shows the MR colour Doppler jet in the LA. The right upper panel shows a significant reduction of SMR following GDMT in proportionate SMR, whereas the right lower panel shows significant reduction of SMR following MitraClip on top of GDMT. See text for explanation. AML, anterior mitral leaflet; AO, aorta; GDMT, guideline-directed medical treatment; LA, left atrium; LV, left ventricle; MRjet, mitral regurgitant jet; PML, posterior mitral leaflet; SMR, secondary mitral regurgitation.

Table 1

Representative examples of patients with SMR are shown with haemodynamic parameters

‘Disproportionate’ MR is typically seen in patients with primary MR.12 Some have coined disproportionate SMR as ‘tertiary MR’,32 where the mitral valve is more distorted than expected based on the extent of myocardial injury.10 This could for instance occur in a patient with a focal inferolateral myocardial lesion where retraction of the posterior mitral leaflet causes an out-of-proportion ‘SMR’. In figure 1, such a patient with a disproportionate phenotype is shown (Patient 2). The mitral leaflets appear structurally normal but clear malcoaptation is observed due to severe posterior leaflet tethering. In contrast, the same myocardial lesion located at the anterior myocardial wall may not cause such mitral valve distortion and thus less or proportionate SMR.10 Whether maladaptive mitral leaflet growth may contribute to disproportionate SMR, or some structural component may coexist in these patients remains to be investigated.14 15 33

Patient 2 with disproportionate SMR has an EROA/LVEDV value of 0.23, consistent with the COAPT profile,11 whereas Patient 1 with proportionate SMR has an EROA/LVEDV of 0.11, which matches the MITRA-FR profile.11 Patient 1 with proportionate SMR was successfully treated with GDMT that eventually abolished SMR. In contrast, the patient with disproportionate SMR was refractory to GDMT and required mitral clipping following two episodes of lung oedema. Since the procedure, the patient did not experience HF events anymore and only mild SMR persists (figure 1).

Considerations on the EROA/LVEDV and RV/LVEDV proportionality coefficients

The graphs in figure 2 represent the proportionality framework with patient examples shown in table 1. Patients 3 and 4 have a different EROA and RV, but the same EROA/LVEDV and RV/LVEDV coefficients; thus, for higher EROA-RV values, LVEDV ‘increases’ to maintain proportionality (Patient 4). The connecting lines (‘slopes’) between Patients 3 and 4 in figure 2, therefore, indicate similar proportionality coefficients. Importantly, and as illustrated by Patients 5 and 6 compared with Patients 3 and 4, respectively, the same proportionality coefficients do not necessarily bear similar EFs or RFs. Therefore, EROA/LVEDV and RV/LVEDV normalise SMR for ‘passive’ LV size only. A RV/LVEDV of >0.2 and an EROA/LVEDV of >0.14 are proposed as cut-offs to categorise patients as disproportionate SMR, based on calculations and assumptions made from the MitraClip trials.11 12

Figure 2

The EROA and RV versus LVEDV relation is shown. Each line represents a fixed EROA or RV/LVEDV value, for a given EF value and at a fixed RF of 50%. The same plot is shown for RV versus LVEDV. All the marked patients from table 1 (square, triangle and circle) are depicted in the graphs with the specific values. EF, ejection fraction; EROA, effective regurgitant orifice area; LVEDV, left ventricular end-diastolic volume; RF, regurgitant fraction; RV, regurgitant volume.

However, the proportionality framework may seem too reductive to predict outcomes or to select individual patients for mitral intervention. First, although Patients 3 and 4 have different SMR severity based on EROA and RV, the proportionality concept assumes similar SMR severity in these patients, but whether this translates into similar or different outcomes awaits further proof. In our hands, when normalising EROA, RV or the average pixel intensity (API) values for LVEDV or LV end-systolic volume, the prediction of major adverse cardiac events was not better compared with non-normalised values.23 Second, it is important to stress that for MITRA-FR and COAPT, mean EROA/LVEDV and RV/LVEDV values have been calculated.11 12 Obviously, these values cannot be used to derive cut-offs for categorising individual patients, because without knowledge on the standard deviation (ie, the spread of the individual patients around the mean), substantial overlap between patients with ‘proportionate’ and ‘disproportionate’ SMR may exist, which may be an important struggle in applying this concept at patient level. In fact, when only considering strict ‘proportionate’ patients in our SMR cohort,23 the correlation between LVEDV and EROA is weak (R2 ranging between 2% and 32%, ie, 2%–32% of EROA variability is attributed to LVEDV). This may be due to echo inaccuracy and variability when measuring EROA and LVEDV. In addition, biological variables such as interpapillary muscle distance,34 dyssynchrony and intrinsic mitral leaflet changes1 14 15 also determine EROA independent from LVEDV and thus affect the proportionality parameters. Therefore, the concept of proportionality (coefficients) may be difficult to settle in clinical practice.

Discordance between EROA/LVEDV and RV/LVEDV

Apart from the difficulties in determining EROA, RV and LVEDV,27 another issue that received little or no attention is the non-linear relationship between EROA and RV.35 Based on the orifice equation (RV=EROA × Time Velocity Integral (TVI)12), which is commonly used in clinical practice, an ERO of 40 mm2 is concordant with a RV cut-off of 60 mL if TVI is ‘fixed’ at 150 cm,36 that is, 60 mL=40 mm2×150 cm. In figure 3, the TVI of 150 cm represents the linear line (constant slope) that links the three guideline-recommended respective EROA and RV cut-off pairs: ESC (20 mm2 and 30 mL),37 ASE/ACC (30 mm2 and 45 mL)27 38 and AHA/ACC (40 mm2 and 60 mL).36 However, TVI is not fixed among patients, because its value may decrease with increasing EROA and decreasing EF in SMR,35 which is shown by the curved line in figure 3. Therefore, a non-linear relationship between RV and EROA exists,35 and this may contribute to the recently reported discordance between EROA and RV in reaching their severity thresholds in SMR.39 For example, Patient 2 with disproportionate SMR depicted in figure 3 (and table 1) shows a very low TVI value (93 cm) because of a large EROA (41 mm2), combined with high left atrial pressure. Although the AHA/ACC EROA SMR severity threshold of 40 mm2 is reached in this patient, the RV threshold of 60 mL is not (38 mL=41 mm2×93 cm).36 Consequently, the discordance between EROA and RV in reaching their severity cut-offs eventually leads to discordance between EROA/LVEDV and RV/LVEDV. In fact, for the disproportionate Patient 2 in figure 1, the EROA/LVEDV coefficient is 0.23, which strongly fits with the COAPT profile,11 whereas the RV/LVEDV value of 0.20 approaches a proportionate MITRA-FR profile.12 Discordance between EROA and RV may partly reconcile why the COAPT patients were recently classified as ‘proportionate’ (MITRA-FR profile) based on RV/LVEDV,12 but as ‘disproportionate’ when considering the EROA/LVEDV parameter.11 However, because of the limitations on the assumptions made from the echocardiographic COAPT and MITRA-FR data,9 12 and because RV values are not reported in COAPT,8 40 the use or comparisons between EROA/LVEDV and RV/LVEDV may not be valid at this time, but this issue should be further investigated along with greater transparency of trial data.9 Also, it should be addressed whether the EROA-RV discordance or the apparent non-linear hydraulic interaction between EROA and RV based on TVI is merely attributed or exacerbated by the PISA method. This may be key because the PISA-method underestimates MR severity due to the elliptical regurgitant orifice in SMR.41 42 In true severe or torrential SMR, the significant underestimation of PISA-EROA multiplied by low TVI values may therefore synergize the underestimation of the RV.

Figure 3

The correlation between EROA and RV is shown as a function of TVI (RV=EROA×TVI). The dots represent individual patients from our updated SMR cohort.23 A TVI value at 150 cm connects the paired guideline-recommend EROA and RV thresholds for severe secondary mitral regurgitation, as indicated by ESC, ASE/ACC and AHA/ACC. For example, an EROA severity threshold of 40 mm2 endorsed by AHA/ACC is in accordance with a RV of 60 mL in case TVI is at least 150 cm (60 mL=0.4 cm2×150 cm). However, because TVI deflects, the disproportionate Patient 2 indicated by a square meets the AHA/ACC severity cut- off of 40 mm2, but not the RV threshold of 60 mL. The RV for this patient is 'only' 38 mL (38=0.41 cm2×93 cm). Therefore, the EROA/LVEDV value (0.23) is consonant with a disproportionate profile, whereas the RV/LVEDV value is closer to a proportionate profile (0.20). ACC, American College of Cardiology; AHA, American Heart Association; ASE, American Society of Echocardiography; EROA, effective regurgitant orifice area; RV, regurgitant volume; TVI, time-velocity integral.

RF to define severe SMR: rationale

Some investigators have numerically modelled SMR severity based on the assumption that a RF cut-off of 50% is considered severe SMR.9 13 The RF is a volumetric index and is defined as the fraction of the total LV stroke volume (TSV) that leaks back (the RV) into the left atrium during systole, that is, RF%=RV/TSV. In clinical practice, RV is mostly obtained by subtracting the FSV from the TSV (see table 1 for illustration).27 Thus, whereas RV and EROA are normalised for ‘passive’ LV size only in the RV/LVEDV and EROA/LVEDV coefficients, the ‘proportionality’ RF index normalises the RV values for TSV, ignoring the absolute EROA or RV. For example, when considering the haemodynamic parameters of Patients 3 and 4 in table 1, according to the 50% RF-based severity cut-off, both patients meet the definition of severe SMR, despite different absolute RVs (26 vs 37 mL, respectively). Conversely, Patients 7 and 8 show how similar RVs can occur with similar EF, but with different LVEDV and RF. Therefore, the RF parameter better represents the relative load concept in SMR compared with crude EROA/LVEDV and EROA/LVEDV measures, because it integrates a functional component. In a recent report, a RF of 50% was shown to predict events in patients with SMR having EROA values between 20 and 30 mm2.22 In (these) relative lower ranges of SMR severity however, the haemodynamic 50% RF cut-off parameter probably also reflects poor reciprocal FSVs (eg, Patient 3). Therefore, potential collinearity between RF and forward stroke distance, volume or flow rate in predicting outcome in these patients is plausible, but this remains to be investigated.

The 50% RF severity cut-off in SMR: does one size fit all?

It is important to note that the proposed 50% RF severity cut-off has been adopted from small pilot studies in primary MR,43 but the use of this cut-off in SMR is not supported by substantial evidence, apart from one recent outcome study.22 Here, a linear relationship between RF and outcome was demonstrated with events occurring along a broad range of RF values.22 In addition, a recent echocardiographic COAPT report showed similar outcomes in patients with RF < or >36%,40 and a cardiac magnetic resonance (CMR) study used a RF severity cut-off of 35%.44 These findings22 40 44 on RF with respect to outcome in SMR mandate more consensus on the optimal RF cut-off to define severe or (dis)proportionate SMR. Finally, it has not been vitally discussed or considered whether a similar RF (thus similar SMR severity according to proportionality) but with different absolute RVs (eg, Patients 3 and 4) has the same or different impact on pulmonary pressures in the diseased hearts of patients with SMR. The volume threshold that may increase pulmonary pressures and affect outcome in the heterogeneous HF population is likely to be variable, which is again a reminder to the complex definition or what constitutes severe SMR, but this is not considered by the proportionality coefficients or the RF. The pulmonary pressure integrates the complex conundrum of pressure-volume relationships in the cardiovascular tree, but pulmonary pressure has been rarely considered in multivariable analysis by previous SMR studies.23 We showed that pulmonary pressure is the strongest independent determinant of outcome in patients with SMR,23 and it also demonstrated to be a major and independent driver of outcome in the conservative treatment arm of the COAPT trial40 and in big data analysis.45

Echocardiographic assessment of RF

The accuracy, reproducibility and uniformity of 2D echocardiographic measures or methods used to assess RF in randomised intervention studies have been recently discussed.40 46 For the RF parameter, each of the components (TSV, FSV or RV) has intrinsic error, because multiple measures are combined into each of these volume components, and the error increases when the volumes are subtracted.27 In figure 4 and table 1 (Patient 9), we show how echocardiographic volumetric measurement variability47 of RF may shift a patient from proportionate (RV/LVEDV=0.14 at RF=50% and EF=30%) towards a disproportionate profile (RV/LVEDV=0.23 and RF=59% and EF=39%), assuming the proposed RV/LVEDV cut-off being 0.2012 and/or RF >50%.13 Indeed, apart from biological variability, the SD for the patients in the conceptual framework (‘their position in the EROA/LVEDV and RV/LVEDV graphs’) when measured with the volumetric method is determined by echocardiographic measurement variability of LVEDV, LVESV (EF), FSV (RF) and TVI, which may be substantial.27 47 Therefore, additional imaging such as 3D echocardiography47 or CMR may be considered44 to allow better volumetric assessments. For 2D echocardiography, consistency of RV and RF measures should be checked by measuring FSV with Doppler.46

Figure 4

An example (Patient 9 from table) of proportionate SMR is shown in the RV/LVEDV graph (indicated with white star). The RV/LVEDV value is 0.14, which is below the 0.2 threshold for disproportionate SMR, which is indicated by the upper bold line. The RV was assessed using the volumetric method (RV=TSV-FSV). Due to volumetric measurement error by echocardiography (indicated by black star), the patient shifts towards the disproportionate zone. EF, ejection fraction; LVEDV, left ventricular end-diastolic volume; RF, regurgitant fraction; RV, regurgitant volume; SMR, secondary mitral regurgitation.

Reconciling ‘disproportionate’ and ‘proportionate’ SMR using RF

Although the EROA/LVEDV and RV/LVEDV coefficients represent the proportionality concept, it remains unclear how these coefficients relate or fit to individual RF values in clinical practice. Interestingly, a recent observation integrated all these parameters in selected patients within the EROA/LVEDV and RV/LVEDV scatter plot, with emphasis on the 50% RF cut-off.13 Here, higher EROA/LVEDV and RV/LVEDV values clearly cosegregate with a RF cut-off >50%, indicating a significant link between the proportionality coefficients and the 50% RF cut-off.13 In addition, EROA values >20 mm2 and RV values >30 mL also cosegregate with RF values>50% and with higher EROA/LVEDV and RV/LVEDV values. The same suggestion on EROA versus EROA/LVEDV can be derived from a recent subanalysis of COAPT, where absolute EROA values appear dominant over the EROA/LVEDV coefficient in predicting therapeutic benefit of MitraClip.11 These observations suggest that EROA and RV metrics remain major drivers of outcome in SMR, and both proportionality coefficients and the RF may assist as complementary measures in grading SMR. Thus, a prospective comparative analysis is necessary in order to better position the individual or hierarchical role of all parameters with respect to outcome prediction and therapeutic decision-making.


In conclusion, it is a challenging task to integrate complex biology and hemodynamics into simple mathematical indices, and this is obstructed by practical limitations in measuring these indices with echocardiography (see key messages in box 1). The recently introduced proportionality indices all have potential to assist in grading SMR. However, in order to adopt a new classification system for SMR for clinical translation at individual patient level, head-to-head comparative analyses between EROA, RV, their LV volume-indexed values and RF with respect to outcome in both observational and intervention studies are required using state-of-the-art measurements. As echocardiographic inaccuracy and variability may spoil the proportionality concept in real world practice, more robust or new promising approaches for grading SMR may be considered.48 49

Box 1

Considerations on the proportionality concept and regurgitant fraction

Considerations on the EROA/LVEDV and RV/LVEDV proportionality concept:

  • The proportionality coefficients aim to correct and redefine SMR severity, but it remains hypothetical and proof-of-concept is still lacking.

  • Proportionate SMR assumes a reasonable correlation between SMR severity and LVEDV, but this remains uncertain because:

    1. Echocardiographic inaccuracy and variability of EROA, RV and LVEDV measures introduce bias.

    2. Other variables independent from LVEDV affect severity of SMR and thus the proportionality coefficients.

  • Because TVI of the MR jet is not a fixed value, EROA and RV values are not collinear and therefore, EROA/LVEDV and RV/LVEDV values may not be interchangeable.

Considerations on the RF parameter:

  • The RF is a different approach to define proportionality and it better reflects the true relative load in SMR compared with EROA/LVEDV or RV/LVEDV, as it considers a functional component rather than passive LV size.

  • Issues on the RF parameter:

    1. A RF severity cut-off in SMR remains to be determined.

    2. Superiority of RF to traditional EROA and RV metrics in predicting outcome or selection of potential MitraClip responders is unknown.

    3. RF requires multiple echocardiographic volume measures-assumptions for calculation and is therefore error-prone. This limits its use to select the rare population of MitraClip-eligible patients.50

  • EROA, effective regurgitant orifice area; LVEDV, left ventricular end-diastolic volume; RF, regurgitant fraction; RV, regurgitant volume; SMR, secondary mitral regurgitation; TVI, time-velocity integral.


We thank Christian De Boever for artwork.



  • Contributors VK: manuscript writing. SC: manuscript writing. MDB: manuscript writing. FT: manuscript writing. FT is responsible for the overall content as guarantor. Each of the authors has contributed equally to the review paper.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Patient consent for publication Not required.

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