In patients undergoing mitral surgery for ischaemic mitral regurgitation is it preferable to repair or replace the mitral valve?

Summary

A best evidence topic was written according to a structured protocol. The question addressed was whether patients undergoing coronary bypass grafting and mitral intervention for moderate to severe ischaemic mitral regurgitation are best treated with mitral repair or replacement. Five hundred and fifty papers were found using the reported search. Based on the 14 non-randomised studies judged to represent best evidence, we concluded that whilst there is some evidence that the operative mortality may be less following mitral valve repair, long-term data are equivocal. Even with contemporary techniques, recurrent mitral regurgitation is not uncommon following repair. Replacement was more frequently performed for patients with greater co-morbidity. Whilst two studies attempted to control for this using propensity analysis, in the majority of studies this introduced considerable bias. No data was available on long-term functional outcomes and quality of life. As there is currently insufficient evidence to inform clinical practice, a randomised trial is warranted in this important area.

Ischaemic mitral regurgitation, Mitral valve replacement, Mitral valve repair

1. Introduction

A best evidence topic was constructed according to a structured protocol [1].

2. Clinical scenario

A patient is referred with dyspnoea and angina. Angiography shows severe triple vessel disease. Echocardiography shows restricted posterior mitral valve leaflet motion with moderate regurgitation. Ventricular function is mildly impaired. The referring cardiologist suggests concomitant mitral surgery because of the adverse prognosis of residual mitral regurgitation (MR) following coronary artery bypass grafting (CABG) in this patient group [2]. The surgeons present disagree whether it is better to repair or replace the mitral valve.

3. Three-part question

In [patients undergoing mitral surgery for Ischemic Mitral Regurgitation (IMR)] is it preferable to [repair or replace the mitral valve] in terms of [operative mortality, re-operation rate, echocardiographic, functional outcomes and long-term survival]

4. Search strategy

Medline search from January 1947 to May 2010 was performed using the OVID interface: (exp Mitral Valve Insufficiency/or exp Mitral Valve Prolapse/) and (exp Ischemia/or exp Myocardial Ischemia/) and (replacement or repair).mp. LIMIT to human studies. American Heart Association (AHA) and National Institute for Health and Clinical Excellence (NICE) guidelines were also searched.

5. Search outcome

Five hundred and fifty papers were found of which 15 were deemed to be represent best evidence (Table 1 ).

Table 1.

Summary table

Author, date and country	Patient group	Outcomes	Key results	Comments
Study type
(level of evidence)
Bonow et al., (2008),	Updates on behalf of the	Not applicable	Despite presenting no data	The studies included
Circulation, USA,	ACC and AHA on		on the comparative	are not informative
[3]	valvular heart disease.		outcomes or repair vs.	and are flawed while
	The authors analyse 14		replacement they	other, important
Systematic review	papers and make		recommend MV repair for	studies have not been
(level 1a)	recommendations based		some patients with IMR.	included. The authors
	on these limited studies		Indeed they fail to mention	fail to make any
			replacement as an option at	recommendation on
			all	replacement as a
				treatment option
Magne et al., (2009),	Study of 370 sequential	30-day mortality	Higher in-hospital	Failed to define the
Circulation, Canada,	patients form 1995 to		mortality rate in the	cohort as at least
[4]	2008 with undefined		replacement group (17.4%)	moderate severity of
	severity of IMR based		compared to repair (9.7%)	IMR and so may
Unmatched retrospective	on echocardiography			include patients with
cohort study (level 2b)	who underwent MV	Six-year survival with a	Better five-year survival of	little indication for
	repair (184 – using rings	mean follow-up of 45	mitral repair group (67%)	mitral intervention
	in all cases the majority	months of unknown	compared to mitral
	of which were 28 mm or	completeness	replacement (73%)	Contemporary
	below) or MV			techniques with all
	replacement with 100%	Early echocardiography	There was a higher rate of	patients leaving the
	preservation of the sub-	follow-up	greater than mild residual	operating room with
	valvular apparatus, 21%		regurgitation in the repair	greater than mild
	bioprosthesis) with		group (18%) than with	residual regurgitation
	unknown completeness		replacement. Neither of
	of revascularisation		these differences were still	This informative
			present when propensity	study shows that
	Patients with acute MR,		matching was used	replacement was used
	other mitral aetiologies or			in the sicker cohort
	previous mitral surgery			and that when this
	were excluded			selection bias was
				corrected for using
	No intraoperative			propensity scoring
	conversions were			there was no
	reported			difference in
				mortality or survival.
	To account for			Despite the
	differences between the			standardised
	cohorts subsequent			techniques of repair
	propensity scoring was			and confirmed
	performed			efficacy of repair –
				no patient left the
				operating room with
				greater than mild
				MR, the early
				follow-up of
				durability showed
				that 18% of repairs
				had recurrent MR
				No functional data or
				long-term echo data
				were presented
Micovic et al., (2008),	Study of 138 sequential	30-day mortality	Higher in-hospital	Failed to define the
Heart Surg Forum, Serbia,	patients from 2000 to		mortality rate in the	cohort as at least
[5]	2006 with undefined		replacement group (9.6%)	moderate severity of
	severity of IMR who		compared to repair (5.8%)	IMR and so may
Unmatched retrospective	underwent MV repair			include patients with
cohort study (level 2b)	(86 patients – using rings	Five-year survival with a	Better five-year survival of	little indication for
	in the majority of cases	mean follow-up of 84	mitral replacement group	mitral intervention
	but of undefined size) or	months that was 83%	(82%) compared to mitral
	MV replacement (52	complete	repair (77%)	In common with
	patients – 100%			many earlier studies,
	mechanical valves	Reoperation rate	No reoperations in either	there was a higher
	without preservation of		group	operative mortality
	the subvalvular			to replacement but
	apparatus) with	Functional outcomes	Improved functional	this was possibly as
	unknown completeness		outcomes in each group	a consequence of the
	of revascularisation.		after mitral intervention but	sicker patients having
	Patients with other mitral		no direct comparison of the	replacement and the
	aetiologies or previous		functional outcomes	inclusion of
	mitral surgery were		between the interventions.	intraoperative
	excluded		The improvements were	conversions as within
			empirically greater in the	the replacement
	Intraoperative conversion		replacement group	group. Again, as with
	from repair to			other authors, this
	replacement, were	Late postoperative	Marginal improvement in	initial higher
	reported as replacement	echo	ejection fraction after mitral	mortality contrasted
			intervention with no	with a higher five-year
			difference between the	survival in the
			groups	replacement group.
				The inclusion of echo
				and functional data is
				useful but does not
				allow comparison
				between the mitral
				interventions
				The operative
				techniques for
				replacement were
				sub-optimal while the
				repair techniques
				were reasonably
				contemporary
				The marginal
				improvement in
				ejection fraction,
				despite reasonable
				repair techniques and
				good five-year survival
				data, highlight the
				possible weakness of
				echocardiography and
				ejection fraction as a
				suitable primary
				endpoint for
				assessing these
				patients
Milano et al., (2008),	Study of 512 sequential	90-day mortality	Higher 90-day	Considerable
Ann Thorac Surg, USA,	patients from 1986 to		mortality in replacement	differences between
[6]	2006 with moderate to		group (21.6%) compared to	groups in terms of
	severe IMR based on		repair (10.3%)	co-morbidity
Unmatched retrospective	preoperative
cohort study (level 2b)	transthoracic or trans-	Five-year actuarial	Better relative survival of	No data presented on
	oesophageal	survival with a mean	mitral repair group (93.7%)	how patients were
	echocardiography or	follow-up of 85	compared to mitral	allocated to repair vs.
	ventriculography who	months that was 99.2%	replacement group	replacement
	underwent MV repair	complete	(79.1%), which equates to
	(416 patients – using		78% five-year survival in	No data presented on
	rings of undefined size)		repair group compared to	the MV repair or
	or MV replacement (106		72% in replacement	replacement
	patients – 28%			technique. Inclusion
	bioprosthesis using			of results prior to
	undefined percentage of			description of
	subvalvular			downsizing
	preservation) with			annuloplasty (1996)
	unknown completeness
	of revascularisation			Using relative
				survival can be
	Patients with acute IMR			misleading as to the
	and previous mitral			actual survival in this
	surgery were excluded			cohort, which is
				extracted from the
	Intraoperative conversion			Kaplan–Meier curve
	from repair to
	replacement, were			No postoperative
	reported as replacement			echo assessment of
				LV function or degree
				of MR
				No data on need for
				reoperations
				No functional
				outcomes
Bonacchi et al., (2006),	Study of 54 sequential	30-day mortality	Higher in-hospital	Considerable
Heart Vessels, Italy,	patients from 1995 to		mortality in replacement	differences between
[7]	2003 with moderate to		group (16.6%) compared to	groups in terms of
	severe IMR based on		repair group (5.5%)	co-morbidity
Unmatched retrospective	preoperative echo and
cohort study (level 2b)	scintigraphy who	Five-year actuarial	‘Similar’ but undefined	No data presented on
	underwent MV repair	survival with a mean	five-year survival in repair	how patients were
	(36 patients – 83%	follow-up of 39	group compared to	allocated to repair vs.
	annuloplasty rings to	months of unknown	replacement	replacement
	achieve coaptation	completeness
	depth of 10 mm) or MV			Good description of
	replacement (18 patients	Reoperation rate	One patient in each group	mitral repair and
	– using 100% sub-		required reoperation	replacement
	valvular preservation,			techniques
	percentage of	Functional outcomes	Good improvement in
	bioprosthesis not stated)		functional outcomes with	Postoperative echo
	with a mean number of		patients with improvement	and functional data
	grafts >2.4		in the mean NYHA class	not broken down by
			from 3.0 to 1.6	procedure
	The exclusion criteria
	were not reported			Most of the data
				presented is after the
	Intraoperative conversion			description of
	from repair to			downsizing
	replacement, were			annuloplasty
	reported as replacement
Al-Radi et al., (2005),	Study of 202 sequential	In-hospital mortality	Higher in-hospital	Failed to define the
Ann Thorac Surg, Canada,	patients from 1990 to		mortality in replacement	cohort as at least
[8]	2001 with undefined		group (21.0%) compared to	moderate severity of
	severity of IMR based		repair (1.5%)	IMR and so may
Unmatched retrospective	on preoperative echo			include patients with
cohort study (level 2b)	and scintigraphy who	Five-year actuarial	80% five-year survival in	little indication for
	underwent MV repair	survival with undefined	repair group compared to	mitral intervention
	(65 patients – 100%	duration and	75% in replacement. Higher
	annuloplasty rings with a	completeness of	need for reoperation in	Considerable
	mean size of 30) or MV	follow-up	repair group (14.0%)	differences between
	replacement (137		compared to replacement	groups in terms of
	patients – 77% using		(3%)	co-morbidity
	subvalvular
	preservation, 64%	Reoperation rate	Better initial survival	Unusual subgroup
	bioprosthesis) with a		in the repair but no late	breakdown not
	mean number of grafts		survival advantage of repair	widely reported in
	>2.4		particularly when	the literature with
			propensity scores	papillary dysfunction
	The cohort was		adjustment for	not traditionally
	sub-divided into two		co-morbidity	described as chronic
	subgroups of papillary			IMR. Good
	muscle dysfunction vs.			long-term follow-up
	ventricular dysfunction			demonstrating the
				poor prognosis of
	Patients with acute IMR			this patient group.
	and previous mitral			No data presented on
	surgery were excluded			how patients were
				allocated to repair vs.
	Intraoperative conversion			replacement but
	from repair to			showed that annular
	replacement, were			dilation and leaflet
	reported as replacement			restriction usually got
				a ring only as repair
				technique and
				papillary dysfunction
				patients were more
				likely to have
				replacement. Large
				mean size of
				annuloplasty ring
				points to an
				underutilisation of
				downsizing
				annuloplasty with
				half the timescale of
				the study prior to the
				introduction of this
				procedure
				No postoperative
				echo assessment of
				LV function or degree
				of MR
				No functional
				outcomes
Calafiore et al., (2004),	Study of 102 sequential	In-hospital mortality	Higher perioperative	Considerable
Ann Thorac Surg, Italy,	patients from 1988 to		mortality among	differences between
[9]	2002 with moderate to		replacement group (10%)	groups in terms of
	severe IMR based on		compared to repair (3.2%)	co-morbidity
Unmatched retrospective	preoperative
cohort study (level 2b)	echocardiography or	Five-year actuarial	75.6% five-year survival in	The groups compared
	ventriculography who	survival with a mean	repair group compared to	within the study
	underwent MV repair	follow-up of 39	66% in replacement	were mismatched as
	(82 patients – using	months of unknown		the indication for
	non-ring based	completeness		repair vs.
	annuloplasty) or MV			replacement were
	replacement (20 patients	Functional outcomes	Improvement in NYHA	pre-specified and so
	– 55% bioprosthesis		class from 3.2 to 2.1 in	not comparable
	using 100% subvalvular		repair and from 3.5 to 2.5
	preservation) with		in replacement	Poor application of
	unknown completeness			downsized
	of revascularisation	Late postoperative	86.1% completeness of	annuloplasty without
		echo	follow-up echo. 50% of	use of rings
	The exclusion criteria		repairs had moderate MR
	were not reported		on late follow-up with only	Post echo confirmed
			small improvements in	poor repair
	No intraoperative		ejection fraction	technique
	conversion from repair to
	replacement reported			No data on need for
				reoperations
				Substantial
				percentage of cohort
				operated on prior to
				introduction of downsizing
				annuloplasty as procedure
				of choice for IMR
				No data on need for
				reoperations
Reece et al., (2004),	Study of 110 sequential	In-hospital mortality	Higher perioperative	Failed to define the
Ann Surg, USA,	patients from 1995 to		mortality among	cohort as at least
[10]	2002 with undefined		replacement group (10.7%)	moderate severity of
	severity of IMR based		compared to repair (1.9%)	IMR and so may
Unmatched retrospective	on preoperative			include patients with
cohort study (level 2b)	echocardiography who	Perioperative	No difference in the	little indication for
	underwent MV repair	complications	incidence of postoperative	mitral intervention
	(54 patients – using an		complications
	undersizing
	annuloplasty size 26 or			There is no indication
	28) or MV replacement			of how patients were
	(106 patients – undefined			assigned to each group
	percentage of			or the severity of the MR
	bioprostheses using
	100% subvalvular			Considerable
	preservation) with a			differences between
	mean number >2.2 grafts			groups in terms of
				co-morbidity
	Patients with acute IMR
	were excluded			Good surgical
				techniques with high
	No intraoperative			mean number of
	conversion from repair to			grafts per case
	replacement reported
				No medium or
				long-term survival
				reported
				No data on need for
				reoperations
				No postoperative
				echo assessment of
				LV function or degree
				of MR
				No functional
				outcomes
Mantovani et al., (2004),	Study of 102 sequential	In-hospital mortality	Higher perioperative	Comparable patient
J Heart Valve Dis,	patients from 1993 to		mortality among repair	groups undergoing
Italy, [11]	2003 with moderate to		(8.2%) compared to	procedures using
	severe IMR based on		replacement group (7.3%)	contemporary
Unmatched retrospective	preoperative			techniques
cohort study (level 2b)	echocardiography who	Five-year actuarial	66.6% five-year survival in
	underwent MV repair	survival with a mean	repair group compared to	Good description of
	(61 patients – using ‘a	follow-up of 36.8 that	73.4% in replacement	valve pathology with
	moderately undersized’	was 100% complete		97% either type I
	annuloplasty ring) or			dysfunction or IIIb
	MV replacement (41	Functional outcomes	NYHA and CCS class
	patients – 24%		improved greatly after	One of the few
	bioprostheses using		surgery with slight attrition	papers to provide
	100% subvalvular		of this improvement on late	functional follow-up
	preservation) with >2.5		follow-up	but disappointingly
	grafts			this is shown only
		Late postoperative		graphically without
	Patients with acute MI	echo		the actual data
	or other cardiac
	procedures were			15% residual MR
	excluded			after mitral repair
				with higher mean PA
	No intraoperative			pressures in repair
	conversion from repair to			group
	replacement reported
Tavakoli et al., (2002),	Study of 93 sequential	Immediate	Poor results form repair	Considerable
Eur J Cardiothorac Surg,	patients from 1988 to	postoperative echo	with only 71.4% or repairs	differences between
Switzerland, [12]	1998 with moderate to		attempted resulting in <2+	groups in terms of
	severe IMR based on		MR and a 28% conversion	co-morbidity
Unmatched retrospective	preoperative		to replacement
cohort study (level 2b)	echocardiography and			No data presented on
	ventriculography who	In-hospital mortality	Higher perioperative	how patients were
	underwent MV repair		mortality among repair	allocated to repair vs.
	(30 patients – using		(20.0%) compared to	replacement
	undefined percentage or		replacement group
	size of annuloplasty		(12.7%)	Not a good reflection
	rings) or MV			of the chronic IMR
	replacement (63 patients	Five-year actuarial	Overall 65% five-year	with emergent IMR
	– 6.2% bioprostheses	survival with a mean	survival with no significant	patients making up
	using 100% subvalvular	follow-up of 36.8 that	difference between repair	21% of the cohort
	preservation) with 92%	was 100% complete	and replacement	and a large
	complete			proportion of the
	revascularisation	Functional outcomes	99.4% of survivors in	patients had severe
			NYHA I-II at long-term	LV dysfunction
	Patients with acute		follow-up compared to
	MI/acute IMR were not		90% in NYHA class III-IV	Poor repair
	excluded. Patients having		preoperatively	techniques with high
	additional cardiac			mortality, high
	procedures at the time of	Reoperation rate	Three patients required	intraoperative
	MV repair/replacement		reoperation but it is not	conversion rate and
	and CABG were		specified what index	poor long-term
	included		operation they had	durability
	Intraoperative conversion			Good description of
	from repair to			pathology of valves
	replacement was			with 66% either type
	reported as replacement			I or IIIb dysfunction
				No breakdown of
				reoperations by
				primary operation
				No breakdown of
				functional outcomes
				by index operation
				No postoperative
				echo assessment of
				LV function or degree
				of MR
Grossi et al., (2001),	Study of 223 sequential	Immediate	Higher perioperative	Considerable
J Thorac Cardiovasc Surg,	patients from 1976 to	postoperative echo	mortality among	differences between
USA, [13]	1996 with severe IMR		replacement (20.0%)	groups in terms of
	based on preoperative		compared to repair group	co-morbidity
Unmatched retrospective	echocardiography and		(10.0%)
cohort study (level 2b)	ventriculography who			No data presented on
	underwent MV repair	30-day mortality	64% five-year survival in	how patients were
	(152 patients – 77%		repair group compared to	allocated to repair vs.
	usage of rings) or MV		47% in replacement	replacement but
	replacement (71 patients			showed that annular
	– 82% bioprostheses	Five-year actuarial	82% freedom form	dilation and leaflet
	using subvalvular	survival with a mean	moderate-severe MR in	restriction usually
	preservation ‘whenever	follow-up of 39.6	repair group at last	had repair while acute
	possible’) with 86%	months of undefined	postoperative echo	cases were more
	having concomitant	completeness		likely to receive
	CABG			replacement
		Late postoperative
	Patients with MR of	echo		Good repair
	non-ischaemic aetiology			techniques with good
	were excluded. Acute			durability of repairs
	IMR patients were			on late follow-up
	included and made up the			echo
	majority of cases who
	did not have additional			No data on need for
	CABG			reoperations
	No reoperative data was			No functional
	reported			outcomes
	Intraoperative conversion
	from repair to
	replacement was
	reported as replacement
Gillinov et al., (2001),	Study of 482 sequential	In-hospital mortality	All data is propensity	Considerable
J Thorac Cardiovasc Surg,	patients from 1985 to		matched in a hypothesis	differences between
USA, [14]	1997 with moderate to		driven manner. Therefore,	groups in terms of
	severe IMR based on		it is not possible to get raw	co-morbidity
Matched retrospective	preoperative		mortality, survival or
cohort study (level 2b)	echocardiography and		reoperative rates from the	No data presented on
	ventriculography who		paper. Lower risk patients	how patients were
	underwent MV repair		probably benefit from	allocated to repair vs.
	(397 patients – 69%		repair but that benefit is	replacement
	usage of rings) and MV		less for high-risk patients
	replacement (71 patients			Good description of
	– 59% bioprostheses	Five-year actuarial	Freedom from reoperation	pathology of valves
	using subvalvular	survival with a mean	was 91% at five-year	with 76% either type
	preservation ‘whenever	follow-up of 36.8 that		I or IIIb dysfunction
	possible’) with 95%	was 100% complete
	having concomitant			Questionable
	CABG	Late postoperative		appropriateness of
		echo		repair techniques
	Acute IMR patients
	were included			Confusing
				presentation of
	Cases that involved			outcomes with no
	other cardiac procedures			raw data presented
	were excluded
				No functional
	Intraoperative conversion			outcomes
	from repair to
	replacement was			Poor postoperative
	reported as repairs			echo assessment of
				LV function or degree
				of MR
Prifti et al., (2001),	Study of 49 sequential	30-day mortality	Higher 30-day mortality in	Considerable
J Heart Valve Dis,	patients from 1996 to		replacement group (33%)	differences between
Italy, [15]	2000 with moderate IMR		compared to repair group	groups in terms of
	based on preoperative		(7.0%)	co-morbidity
Unmatched retrospective	echocardiography or
cohort study (level 2b)	scintigraphy who	Three-year actuarial	79.0% three-year survival for	No data presented on
	underwent MV repair	survival with mean	all mitral intervention not	how patients were
	(43 patients – using rings	duration of follow-up	broken down by procedure	allocated to repair vs.
	of undefined size in 86%	of 34 months but		replacement
	of cases) or MV	unknown completeness
	replacement (six patients –			Good description of
	undefined percentage of	Reoperation rate	Three patients were reoperated	pathology of valves
	bioprostheses using		on but the index operation	with 100% of cases
	100% subvalvular		was not reported	either type I
	preservation) with >2.6			dysfunction or IIIb
	grafts
				No data on need for
	Intraoperative			reoperations
	conversion from repair to
	replacement was			No functional
	reported as replacement			outcomes
Hausmann et al., (1999),	Study of 337 sequential	30-day mortality	Higher 30-day mortality in	Very sick cohort of
J Heart Valve Dis,	patients from 1986 to		replacement group (14.2%)	patients with severe
Germany, [16]	1998 with severe IMR		compared to repair group	MR, most of who
	based on preoperative		(12.1%)	had moderate or
Unmatched retrospective	echocardiography or			severe LV
cohort study (level 2b)	ventriculography who	Five-year actuarial	66.8% five-year survival in	impairment, high
	underwent MV repair	survival with unknown	repair group compared to	NYHA class and
	(140 patients – using	mean duration or	73.4% in replacement	high-levels of
	suture or pericardial	completeness of		co-morbidity
	annuloplasty but no use	follow-up
	of ring) or MV			Considerable
	replacement (197	Reoperation rate	4.2% need for reoperation	differences between
	patients – 46.7%		in the repair group	groups in terms of
	bioprostheses using		compared to 0% for	co-morbidity
	100% subvalvular		replacement
	preservation) with >2.2			No data presented on
	grafts			how patients were
				allocated to repair vs.
	Patients with acute IMR			replacement.
	were included
				Poor repair techniques
	Intraoperative conversion			with low rate of ring
	from repair to			annuloplasty
	replacement was			utilisation
	reported as replacement
				Correlates survival
				with decreased
				degrees of MR
				postoperatively
				No functional
				outcomes
Cohn et al., (1995),	Study of 150 sequential	In-hospital mortality	Higher in-hospital	Considerable
Eur J Cardiothorac Surg,	patients from 1984 to		mortality in repair group	differences between
USA, [17]	1994 with severe IMR		(9.5%) compared to	groups in terms of
	based on preoperative		replacement group (8.9%)	co-morbidity
Unmatched retrospective	echocardiography,
cohort study (level 2b)	ventriculography and	Five-year actuarial	56.0% five-year survival in	The groups were not
	TOE who underwent	survival with a mean	repair group compared to	well-matched as the
	MV repair (94 patients –	follow-up of 31.2	91.5% in replacement	authors used
	using rings of undefined	months that was 98%		replacement for
	size in 85%) or MV	complete		tethering of the valve
	replacement (106			and repair for annular
	patients – 71%	Reoperation rate	13% need for reoperation	dilation
	bioprosthesis using		in the repair group
	‘good’ but undefined		compared to 7% for	Good description of
	percentage of		replacement	pathology involved
	subvalvular			with 70.7% having
	preservation) with 93%			type I dysfunction or
	having simultaneous			IIIb
	CABG
				A small number of
	Patients with acute IMR			acute cases were
	were included			included, most of
				who had replacement
	Intraoperative conversion
	from repair to			No assessment of
	replacement was			degree of residual or
	reported as replacement			recurrent MR and LV
				function on
				postoperative echo
				No functional
				outcomes

AHA, American Heart Association; IMR, ischaemic mitral regurgitation; MR, mitral regurgitation; CABG, coronary artery bypass grafting.

6. Results

The Joint American Heart Association/American College of Cardiology guidelines [3] omit important references and do not discuss the option of replacement.

The 14 comparative case series identified also have significant limitations, but offer an insight into variations in surgical practice and outcomes [4–17]. No study routinely used preoperative transoesophageal echocardiography, exercise echocardiography or cardiac magnetic resonance imaging to assess the pathology or severity.

The studies often include patients operated on prior to the popularisation of downsizing annuloplasty as the repair technique of choice [18]. Many studies do not adequately describe the techniques used, or report sub-optimal techniques such as incomplete rings, inadequate downsizing. Many include additional leaflet or chordal procedures and obsolete procedures such as pericardial annuloplasty. Similarly, when the mitral valve is replaced, the degree of preservation of the subvalvular apparatus is often not reported or sub-optimal [19].

Thirty-day mortality is generally less for repair than replacement; however this was not consistent in all studies and may be explained by the sickest patients having replacements. The two studies that used propensity matching found no difference, after correcting for co-morbidity [4, 14].

Both procedures were associated with poor survival with little difference between groups. Some studies report no difference in survival [7, 15], some report better survival with repair [4, 6, 8, 9, 13], while others favour replacement [5, 11, 16, 17]. Four studies fail to report the severity of IMR and so risk, including patients with mild IMR who have little to benefit from intervention [4, 5, 8, 10]. Reoperation rate is variably reported, but generally higher in the repair group [8, 16, 17].

Many studies observed that co-morbidity rather than procedure is also the strongest predictor of late outcomes [4, 6, 11, 13]. Significantly the replacement group had a larger proportion of acute cases with greater levels of co-morbidity in many studies [5, 9, 12, 14].

There is significant selection bias towards repair as it is frequently used in annular dilation, while replacement is used in restricted leaflet motion and papillary dysfunction [8, 11], the latter not being classically regarded as chronic IMR. The reporting of failed intraoperative repairs converted to replacement as part of the replacement group is also a cause of bias [5–8, 12, 15–17].

Operative mortality, reoperation rate and survival are unreliable indicators of procedural success in this patient group. More pertinent outcomes measures would include late echocardiography [20], to demonstrate durable repair, and functional outcomes, to demonstrate patient benefit. Unfortunately, unlike for non-ischaemic mitral regurgitation this data is not reported [20]. Functional outcomes, assessed predominantly using dyspnoea and angina scores, were either not reported separately for repair and replacement [5, 7], or showed no difference [9, 12]. Late echocardiographic freedom from greater than mild MR varied from 50 to 85% [4, 9, 11, 13], highlighting the variability in the efficacy and durability of repair.

7. Clinical bottom line

Current guidelines are inadequate to inform clinical practice. Overall, mortality for CABG plus mitral valve repair in the 14 published comparative studies is lower than CABG with mitral valve replacement. However, this finding was not consistent throughout the studies, with several showing higher mortality with repair. This discrepancy is likely to be due to heterogeneous study populations. Further problems with these studies are the widespread application of sub-optimal repair and replacement techniques, often in historical cohorts; the inclusion of acute cases and variations in data reporting (in terms of intention to treat analysis). Few papers present functional or late echocardiographical outcomes.

A randomised trial is clearly warranted [4, 5, 14]. Power calculation depends on the primary end point (mortality or functional outcome) suggesting that patients should be recruited with moderate to severe IMR on preoperative transoesophageal echo, confirmed by exercise testing and randomised to revascularisation combined with either systematically applied downsizing, complete ring, annuloplasty, or complete subvalvular sparing mitral valve replacement, preferably with bioprosthesis. The durability of repair should be documented at early and late time points. To reduce cost and number of patients (approx. 100) primary outcomes should be functional (e.g. oxygen consumption-MV02). Secondary outcomes should include survival and freedom from more than mild MR.