Left atrial disease and left atrial reverse remodelling across different stages of heart failure development and progression: a new target for prevention and treatment.

The left atrium is a dynamic chamber with peculiar characteristics. Stressors and disease mechanisms may deeply modify its structure and function, leading to left atrial remodelling and disease. Left atrial disease is a predictor of poor outcomes. It may be a consequence of left ventricular systolic and diastolic dysfunction and neurohormonal and inflammatory activation and/or actively contribute to the progression and clinical course of heart failure through multiple mechanisms such as left ventricular filling and development of atrial fibrillation and subsequent embolic events. There is growing evidence that therapy may improve left atrial function and reverse left atrial remodelling. Whether this translates into changes in patient's prognosis is still unknown. In this review we report current data about changes in left atrial size and function across different stages of development and progression of heart failure. At each stage, drug therapies, lifestyle interventions and procedures have been associated with improvement in left atrial structure and function, namely a reduction in left atrial volume and/or an improvement in left atrial strain function, a process that can be defined as left atrial reverse remodelling and, in some cases, this has been associated with improvement in clinical outcomes. Further evidence is still needed mainly with respect of the possible role of left atrial reverse remodelling as an independent mechanism affecting the patient's clinical course and as regards better standardization of clinically meaningful changes in left atrial measurements. Summarizing current evidence, this review may be the basis for further studies.

Introduction

The left atrium (LA) is a thin‐walled dynamic chamber that plays an important role in global cardiac performance, contributing to left ventricle (LV) filling and cardiac output and maintaining a dynamic interaction with the LV. In response to varying stressors (i.e. electrical, mechanical and metabolic), the LA can undergo a remodelling process resulting in its persistent dilatation with functional impairment. , LA enlargement and dysfunction are well‐established markers of diastolic and systolic dysfunction, as well as predictors of cardiovascular (CV) outcomes, including atrial fibrillation (AF), , stroke, , heart failure (HF), and subsequent mortality. This relationship is highly related to LV function and has been shown across all the ejection fraction, , , , even in advanced HF. However, recent data suggest a possible independent contribution of LA dysfunction in HF.

In the past decades, the concept of LA remodelling has evolved, and the contribution of the LA to the pathophysiology of HF has grown in appreciation. , , , , Indeed, changes in LA structure and function may be either a marker of HF severity or a mechanism actively contributing to the progression of cardiac dysfunction. One manifestation of LA disease, AF, contributes substantially to morbidity in patients with HF. Recent data suggest that timely therapeutic intervention may attenuate the process of deleterious LA remodelling and potentially promote reverse LA remodelling with a consequent improvement in clinical symptoms and outcomes. , In contrast, age and endurance exercise are related to LA enlargement even in absence of LA disease. ,

Prevention of LA dysfunction and treatment of LA disease may therefore be of critical importance. The relationship between changes in LV structure and function and subsequent CV outcome is well established. We propose here a similar conceptual approach to the LA reflecting dynamic changes in relation to HF progression. The aim of this review is to summarize the concepts and the pathophysiologic mechanisms behind the potential beneficial effects of directed medical therapy on LA reverse remodelling and its clinical implications across all HF stages.

Definition of left atrial reverse remodelling

Atrial disease, as stated in the recent European Society of Cardiology (ESC) HF guidelines, can be defined as a complex of subclinical structural, electrophysiological, and functional changes that affect the atria with the potential to produce clinical consequences (Graphical Abstract). Signs and symptoms may be those of HF and/or of atrial arrhythmias and be due to both a failing LV and a dysfunctional LA, proving the active role of atrial disease in HF pathophysiology , and in other pathophysiological processes. Impaired atrial contractile reserve and LA enlargement independently predict mortality and CV morbidity, as well as AF onset or recurrence and severity of functional impairment in HF patients. , , ,

The temporal process occurring after the removal of external stressors, leading to a reduction in LA volume and/or a restoration of specific functional parameters is called reverse remodelling.

Only recently the attention shifted to quantifying and evaluating LA reverse remodelling as a potential biomarker that responds to specific therapeutic interventions, potentially providing prognostic information. One critical aspect is that the response to therapeutic interventions is potentially different in the acute or chronic settings or in the early or late stage of atrial failure. In the early stage, the LA is less abnormal and potentially more capable of responding to the changes in myocardial structure and function. When stressors act for a longer time, LA remodelling progresses and reversal might be less likely (Figure  ). Reversibility and, ultimately, the clinical significance of LA reverse remodelling in the late stages of atrial failure, have to be proven, yet. Some studies have indicated that LA dysfunction may precede changes in LA volume in different clinical settings , , , and in normal subjects. , Thus, changes in LA function may be more sensitive than changes in LA volume to detect abnormal myocardial function and the effects of therapeutic interventions. However, this has not been proven, yet.

Figure 1

Progression of left atrial remodelling. The progression of left atrial remodelling goes through several stages, along with progressive worse cardiovascular outcomes. When causative stressors are not removed (by preventive measures or treatments), worsening atrial failure inexorably reaches an irreversible stage. On contrast, acting in the earlier phases may arrest and also reverse the pathophysiologic process. CV, cardiovascular; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle.

Assessment of left atrial reverse remodelling

Assessment of LA remodelling is based on various non‐invasive parameters. To date, there are no definite cutoffs identifying a clinically relevant LA reverse remodelling. Hence, monitoring LA remodelling without such standardization makes routine examination challenging. Another major limitation of current evaluations of changes in LA function is that they are almost completely based on LA size. Measurements of LA strain and the different passive and active components might, on the other hand, better define atrial function in different clinical settings (Table  ). , , , Table  shows some of the definitions of LA reverse remodelling reported from the literature. , , , , , , , , Most reports have used arbitrarily a 15% change of LA volume measured with either echocardiography and/or cardiac magnetic resonance (CMR). Functional aspects of LA reverse remodelling have been extensively described using strain analysis. A recent systematic review and meta‐analysis including more than 2500 patients showed reference values for reservoir, conduit and contractile strains. The variability was mostly explained by heart rate, body surface area, and sample size. Nevertheless, little is known about the percentage change of LA strain which may be clinically significant. The difficulties related to the assessment of LA reverse remodelling are due to time‐consuming issues, the reliability of the measurement and the lack of standardization across inter‐vendor packages. However, the reproducibility of both LA volume and LA strain seems acceptable. ,

Table 1

Imaging evaluation of left atrial reverse remodelling. Evaluation of left atrial size and function

Parameter	Meaning	Advantages	Disadvantages
LA volume 2D‐TTE	↓ LA volume = LA reverse remodelling	Large body of evidence Good reproducibility	Underestimation compared to 3D‐TTE, CT and CMR No LA functional assessment
LA volume 3D‐TTE	↓ LA volume = LA reverse remodelling	Less geometric assumptions → correlation with CT and CMR	Time‐consuming Low frame rate for 3D acquisition and consequent poor spatial resolution Few evidences of LA reverse remodelling
Peak LA longitudinal strain	Quantification of functional reverse remodelling ↑ LA strain = ↓ LA fibrosis = LA reverse remodelling	Semi‐automated Angle‐independent Less load‐dependent 37 Evaluation of phasic LA function 38 Acceptable reproducibility	Time‐consuming Inter‐vendor variability Lack of standardization
LA volume CT	↓ LA volume = LA reverse remodelling ↑ LA ejection fraction	Short time for scanning and analysis More accurate than TTE Good correlation with CMR Anatomical characterization for AF ablation	Need for contrast agent Need for breath‐hold
LA volume CMR	↓ LA volume is the surrogate measure of LA reverse remodelling ↓ LA fibrosis = ↓ DE 39	Gold standard for LA volume assessment High spatial and temporal resolution Tissue characterization (DE)	Time‐consuming Challenging with current technology Expensive Renal function Lack of evidence Availability/access

2D, two‐dimensional; 3D, three‐dimensional; AF, atrial fibrillation; CMR, cardiac magnetic resonance; CT, computed tomography; DE, delayed enhancement; LA, left atrium; TTE, transthoracic echocardiography.

Table 2

Definitions of left atrial reverse remodelling

Author	Imaging modality	Definition	Type of choice	Reproducibility and agreement
Westenberg et al. 41	CMR	≥15% LAV reduction	Arbitrary, pre‐specified	Assessed: good
Antonini‐Canterin et al. 42	TTE	≥15% LAV reduction	Arbitrary, pre‐specified Based on previously reported data	Assessed on 10 pts: good
Tops et al. 43	TTE	≥15% LAV reduction	Arbitrary, pre‐specified	Assessed on 15 pts: good
Brenyo et al. 44	TTE	High responders: ≥20% LAV reduction	Arbitrary	Not mentioned
		Low responders: <20% LAV reduction	≥lower‐quartile response in the CRT‐D group vs. <lower‐quartile response in the CRT‐D group
Marsan et al. 45	3D‐TTE	≥15% LAV reduction	Arbitrary Based on previously reported data	Assessed on 20 pts: acceptable
Candan et al. 46	TTE	>15% LAV index reduction	Arbitrary, pre‐specified	Assessed on 20 pts: good
Gelsomino et al. 47	TTE	≥15% LAV reduction	Arbitrary, pre‐specified	Assessed: acceptable
Kloosterman et al. 16	TTE	≥10% reduction in LAV	Arbitrary, pre‐specified	Not mentioned
Mathias et al. 48	TTE	>30% LAV reduction	Arbitrary	Not mentioned
			Median value of LAV percent decrease

3D, three‐dimensional; CMR, cardiac magnetic resonance; CRT‐D, cardiac resynchronization therapy‐defibrillator; LAV, left atrial volume; pts, patients; TTE, transthoracic echocardiography.

Less load‐dependent

Evaluation of phasic LA function

↓ LA fibrosis = ↓ DE

Stages of heart failure

In line with the 2021 HF Society of America, HF Association of the ESC and Japanese HF Society universal definition of HF, a four‐stage classification has been proposed for atrial disease. We discuss below the potential benefits of prevention and treatment on LA reverse remodelling, across the four HF stages.

Patients at risk for heart failure and with pre‐heart failure (stages A and B)

The importance of adopting effective measures to control risk factors and thus to prevent future HF development is an emerging public health need. Even in the absence of overt HF, LA remodelling has a crucial role in this process. LA structural and functional reverse remodelling has been used to assess the therapeutic utility of different treatments, both in animal models and in humans (Table  ). , , , , , , , , , , , , , , , , , , , , , , , Risk of HF and pre‐HF, i.e. structural heart disease and/or abnormal cardiac function in the absence of symptoms (stages A and B) represent two crucial steps of the progression of LA remodelling in which preventive measures and targeted treatments have shown favourable results.

Table 3

Left atrial reverse remodelling in patients at risk of heart failure and pre‐heart failure (stages A and B)

Author	Type of study – sample	Population	Intervention/drug	Follow‐up	LA outcomes	Clinical outcomes
Arterial hypertension
Dernellis et al. 54	Observational prospective – 48	Essential hypertension	Enalapril ± chlorthalidone vs. untreated	16 weeks	LARV: 35.4 to 29.3 ml (−17%), p < 0.05 LACV: 43.8 to 51.3 ml (+17%), p < 0.05 LA ejection force: 20.9 to 18.1 kdynes (−13%), p < 0.05
Mattioli et al. 55	Observational prospective – 120	Hypertensive patients with mild to moderate LV hypertrophy	Telmisartan	12 months	LAV max: 35 to 32 ml (−9%), p < 0.05
Matsuyama et al. 56	Observational	Animal: hypertension	Olmesartan	8 weeks	↓ LA fibrosis No difference in LA action potentials
Matsuyama et al. 56	Observational prospective – 174	Hypertension	Irbesartan or nebivolol	12 months	LAV max: Irbesartan 45.1 to 39.9 ml (−12%), p < 0.001 Nebivolol 47 to 40 ml (−15%), p < 0.001
					LA‐GLS: Irbesartan 37.71% to 40.45% (+ 7%), p < 0.001 Nebivolol 36.4 to 41.52 (+14%), p < 0.001
Atrial fibrillation
Shi et al. 57	Observational prospective – 20	Animal: CHF due to rapid atrial pacing	Enalapril vs. control	5 weeks	↓ LA fractional area shortening (−42%, p = 0.0001)
					↓ LA fibrosis
Kumagai et al. 58	Observational prospective – 20	Animal: sustained AF by rapid right atrial pacing	Candesartan vs. control	5 weeks	↓ LA fibrosis	↑ AF duration in control group
Cha et al. 59	Observational prospective – 20	Animal: CHF due to rapid ventricular pacing	Omapatrilat vs. control	5 weeks	↑ LA area index
Lee et al. 60	Observational prospective – 15	Animal: CHF due to rapid ventricular pacing	Pirfenidone vs. control	3 weeks	↓ LA fibrosis	↓ Arrhythmogenic atrial remodelling ↓ AF vulnerability
Li et al. 61	Observational prospective – 27	Animal: persistent AF by rapid right atrial pacing	Cilazapril or valsartan vs. control	6 weeks	↓ LAV ↑ LA ejection fraction	↓ AF inducibility and duration
Kunamalla et al. 62	Observational prospective – 21	Animal: CHF due to rapid ventricular pacing	Gene‐based expression of dominant‐negative type II TGF‐β receptor	3/4 weeks	↓ Increase in conduction inhomogeneity ↓ LA fibrosis	↓ AF episodes
Boldt et al. 63	Observational prospective – 261	Permanent AF versus sinus rhythm	ACEi (various) vs. control	6 months	↓ Collagen I expression
Perea et al. 64	Observational prospective – 90	Paroxysmal AF	Catheter ablation	4–6 months	Recurrence of AF: LAV (CMR) 126.2 to 103.5 ml (−17%), p < 0.001 No recurrence of AF: LAV (CMR) 98 to 84.9 ml (−13%), p < 0.001
Tops et al. 43	Observational prospective – 148	Paroxysmal or persistent AF	Catheter ablation	13.2 ± 6.7 months	Responders (63%): LAVI max 31 to 22 ml/m2 (−29%), p < 0.05 Non‐responders (37%): LAVI max 29 to 31 ml/m2 (+ 7%), p < 0.05 p‐value between groups: < 0.001	Baseline LA strain and LAVI max predictor of LARR at multivariate analysis
					Responders: LA strain 19% to 22% (+16%), p < 0.05 Non‐responders: LA strain 14% to 15% (+ 7%), NS p‐value between groups: < 0.001
Gelsomino et al. 47	Observational prospective – 33	Paroxysmal AF	Minimally invasive atrial fibrillation surgery	12 months	↓ LAVI (LARR in 72%) ↑ LA strain
Coronary artery disease
Milliez et al. 65	Randomized – 88	Animal: acute MI	Spironolactone ± lisinopril ± atenolol vs. untreated	3 months	↓ LA hyperexcitability ↓ LA fibrosis (spironolactone > atenolol and lisinopril)
Yoon et al. 66	Randomized – 38	Animal: acute MI	Losartan vs. control	4 weeks	↓ Increase in LA diameter ↓ LA fibrosis (↓ connexin‐43 expression)	AF inducibility and duration not different between groups
Ahn et al. 67	Observational prospective – 105	Acute MI	PCI	6 months	Occurrence of LA reverse remodelling after 6 months according to myocardial perfusion grade
Other risk factors
Karason et al. 68	Observational prospective – 63	Obese patients	Bariatric surgery vs. control	12 months	Surgery: LAV 71 to 63 ml (−11%), p < 0.05 Control: LAV 73 to 69 ml (−5%), NSp‐value between groups: NS
Willens et al. 69	Observational prospective – 17	Obese patients	Bariatric surgery	7.6 ± 3.6 months	LA diameter: 35 to 37 mm (+ 6%), NS
Di Bello et al. 70	Observational prospective – 39	Obese patients	Bariatric surgery vs. control	6–24 months	LA diameter: 37.9 to 33.5 (−12%), p < 0.05
Owan et al. 71	Observational prospective – 882	Obese patients	Bariatric surgery vs. untreated	24 months	LAV: 55.3 to 54.4 ml (−2%), NS
Luaces et al. 72	Observational prospective – 41	Obese patients	Bariatric surgery	12 months	LAVI: 31.31 to 32.83 ml/m2 (+ 5%), NS
Pathak et al. 73	Observational prospective – 149	Human: AF patients with BMI ≥27 kg/m2 and ≥1 CV risk factor	Risk factors management (more intensive vs. treating physician) after AF ablation	42 months	Intensive: LAVI 42.5 to 30.4 ml/m2 (−28%), p < 0.001 Control: LAVI 42.4 to 39.5 ml/m2 (−7%), NSp‐value between groups: < 0.001	Arrhythmia‐free survival greater and less symptoms in intensive management group
Pathak et al. 74	Observational prospective – 308	Paroxysmal/persistent AF and BMI ≥27 kg/m2	Exercise programme and individual risk factors management; AF: medical therapy or ablation	49 ± 19 months	No significant difference in LAVI between patients with cardiorespiratory fitness gain and without	Cardiorespiratory fitness predicts arrhythmia recurrence in obese individuals with symptomatic AF
Pathak et al. 75	Observational prospective – 355	AF and BMI ≥27 kg/m2	Weight loss	48 months	WL ≥10%: 37.6 to 30.9 ml/m2 (−18%), p < 0.001 WL 3%–9%: 39.5 to 34.7 ml/m2 (−12%), p < 0.001 WL <3%: 39 to 40.4 ml/m2 (+4%), p = 0.02p‐value between groups: < 0.001	Long‐term sustained WL is associated with significant reduction in AF burden and maintenance of sinus rhythm

ACEi, angiotensin‐converting enzyme inhibitor; AF, atrial fibrillation; BMI, body mass index; CHF, chronic heart failure; CMR, cardiac magnetic resonance; CV, cardiovascular; LA, left atrium; LACV, left atrial conduit volume; LA‐GLS, left atrial global longitudinal strain; LARR, left atrial reverse remodelling; LARV, left atrial reservoir volume; LAV, left atrial volume; LAVI, left atrial volume index; LV, left ventricle; MI, myocardial infarction; NS, not significant; PCI, percutaneous coronary intervention; TGF‐β, transforming growth factor‐beta; WL, weight loss.

Hypertension

In hypertensive patients, evidence of LA reverse remodelling has been provided by the efficacy of specific therapies and a better blood pressure (BP) control. In fact, some studies have demonstrated a reversal of LA dimension with associated functional improvement , , , , (Table  ). Telmisartan showed a reduction in LA maximum and minimum volumes along with an increase in atrial ejection force and a decrease in BP, after 12 months of treatment. Along with structural reverse remodelling, the use of antihypertensive therapy has also been shown to be associated with functional improvement in measures of LA strain. In a group of 160 patients treated with irbesartan or nebivolol, speckle tracking analysis showed a significant improvement in LA global peak reservoir strain after 12 months (from 37% to 40% and from 36% to 41% for irbesartan and nebivolol, respectively). Whether these results may be translated to a clinically relevant LA reverse remodelling has yet to be demonstrated. Also, given the observational nature of these studies, results should be considered as exploratory.

The use of angiotensin‐converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARBs) has been hypothesized to be associated with LA reverse remodelling by a dual mechanism: an improved haemodynamic status, mediated by BP reduction, with a secondary improvement in LV diastolic function and a direct anti‐fibrotic effect.

Accordingly, careful monitoring of LA structure and function over time may serve as a surrogate marker to assess the efficacy of antihypertensive therapies. However, no study to date has assessed the prognostic significance of LA reverse remodelling among hypertensive patients.

Atrial fibrillation

The relationship between LA remodelling and AF is reciprocal. The progression of LA disease may predispose to the occurrence of AF and, on the other hand, AF perpetuates and might worsen adverse LA remodelling in a vicious circle. The basis of this process seems to be related, at least in part, to an inflammatory state and the development of fibrosis. In addition, the structural alterations associated with an enlarged LA lead to electrical remodelling.

Reverse remodelling identified by a shrinking LA volume has been associated with fewer AF recurrences in HF patients. , , Several studies have demonstrated the benefits of both medical therapy and catheter ablation on LA reverse remodelling (Table  ). Even in animal models of HF due to rapid atrial pacing, the use of ACEi or ARB resulted in a significant LA size reduction and decreased amount of fibrosis. , The reduction of LA fibrosis has also been reported in one human study. The amount of collagen I was higher in AF versus sinus rhythm patients undergoing cardiac surgery. Interestingly, when divided into groups with or without ACEi treatment, there was a significant lower expression of collagen in AF with ACEi versus AF without ACEi.

The extent of pre‐existing fibrosis is a crucial determinant of interventional success in patients undergoing AF treatment. , Pre‐procedural LA strain is associated with rhythm outcome after LA catheter ablation. Despite ablation itself may result in increased LA fibrosis, , performing the procedure has shown to favour LA reverse remodelling. Two studies reported LA response to catheter ablation or surgical treatment of AF. , In one study, LA reverse remodelling (Table  ) occurred in 63% of patients undergoing radiofrequency catheter ablation. While LA maximum volume and LA minimum volume significantly decreased in the overall population, only the responder group reported a significant increase in LA strain. In this study, baseline LA strain and LA maximum volume were independent predictors of LA reverse remodelling. Similarly, in a cohort of patients who underwent surgical treatment of AF, 73% had LA reverse remodelling after a 12‐month follow‐up.

Coronary artery disease

Left atrial volume increases in patients with recent myocardial infarction (MI), especially in the early phase. LA remodelling after 1 month and baseline LA size are both independent predictors of morbidity and mortality. LA enlargement often occurs in parallel with LV remodelling, even in the absence of LV ejection fraction (LVEF) deterioration. ,

Very few studies have assessed the direct effect of coronary interventions on LA reverse remodelling (Table  ). Ahn et al. investigated the impact of myocardial perfusion on LA remodelling in 105 patients with acute MI treated with successful primary percutaneous coronary intervention. Despite no overall change of LA volume after 6 months, evidence of significant LA reverse remodelling according to myocardial perfusion grade was reported. In addition, perfusion grade and anterior MI location were independent determinants of LA remodelling at multivariate analysis (both p < 0.001).

In post‐MI animal models, renin–angiotensin–aldosterone system (RAAS) inhibition produced some effects on LA remodelling, preventing LA enlargement (Table  ). The anti‐fibrotic effect of ARBs was highlighted by a reduction in the expression of connexin 43. Similarly, in ischaemic HF rats treated with ACEi, beta‐blocker or mineralocorticoid receptor antagonists (MRAs), atrial hyperexcitability was reduced by all drugs, but only spironolactone reduced atrial fibrosis. The hypothesis generating nature of these findings should be replicated in humans to validate consistency in clinical practice.

Other risk factors

Recent evidence showed that dietary modification, physical activity and weight loss may reverse atrial abnormalities and may impact on arrhythmias incidence and recurrence in obese patients (Table ).

A recent meta‐analysis reported conflicting data about cardiac structural and functional changes after bariatric surgery. Karason et al. demonstrated LA reverse remodelling in obese patients following bariatric surgery when compared to the group with only dietary restrictions. Nevertheless, the difference in LA volume between the interventional and the dietary group was not significant either at baseline and after 1‐year follow‐up. In general, LA reverse remodelling occurred in parallel with LV mass reduction and LV filling pressure improvement.

Even in the absence of surgery, weight loss influenced LA reverse remodelling. Dietary modifications among obese patients have been shown to decrease LA volume along with a decrease in body mass index.

Other potentially modifiable risk factors such as sleep apnoea syndrome and arterial stiffness are associated with AF recurrence and LA remodelling. , Future studies will clarify whether targeted treatment of these conditions can reverse atrial remodelling and whether such changes are associated with a reduction in symptoms.

As opposed to LA reverse remodelling, exercise training may also be associated with LA enlargement. Young athletes tend to have larger LA volume when compared to active but not trained patients. LA volume of athletes without AF showed considerable overlap with non‐athletes with AF. However, LA reservoir strain in AF patients was significantly impaired compared with patients in sinus rhythm, regardless of training status, reflecting that the dysfunction acts as the basis of the arrhythmia.

Patients with heart failure (stage C)

Heart failure with reduced ejection fraction

Myocardial injury, haemodynamic changes, and neuro‐hormonal activation cause heart remodelling and this is a key factor for progression of HF with reduced ejection fraction (HFrEF). LA remodelling is a well‐studied entity, as it is associated with adverse CV events. LV reverse remodelling is a critical finding as a surrogate of the beneficial effects of most HF treatments and is associated with an improved prognosis. Although some evidence on LA reverse remodelling has been reported from several HFrEF trials, the real impact on clinical outcomes remains less explored (Table  ) , , , , , , , , , , , , , , , , , , , , , , , , , , .

Table 4

Left atrial reverse remodelling in heart failure (stage C)

Author	Type of study – sample	Stage of HF – NYHA class	Intervention	Follow‐up	LA outcomes	Clinical outcomes
Reduced ejection fraction
Desai et al. 99	RCT – 413	Stage C – NYHA I–III	ARNI vs. enalapril	12 weeks	ARNI: LAVI 30.4 to 28.2 ml/m2 (−7%) Enalapril: LAVI 29.8 to 30.5 ml/m2 (+2%)p‐value between groups: ≤ 0.001
Januzzi et al. 100	Observational prospective – 794	Stage C – NYHA II–IV	ARNI	6 months 12 months	ARNI: LAVI 37.76 to 32.80 (−13%) to 29.32 (−22%) ml/m2 (both p < 0.001)	↓ NT‐proBNP (p < 0.001)
Brenyo et al. 44	RCT (sub‐study) – 1378	Stage C – NYHA II–III	CRT vs. ICD‐only	12 months	Median LAV reduction: 29% (CRT; IQR 20%–36%) vs. 10% (ICD‐only; IQR 5%–14%)	↓ Risk of supraventricular tachyarrhythmias, HF and deaths
Kloosterman et al. 16	Observational retrospective – 365	Stage C – NYHA II–IV	CRT	6 months	LA + LVRR: LAVI 44 to 34 ml/m2 (−3%) LARR: LAVI 45 to 36 ml/m2 (−20%) LVRR: LAVI 40 to 42 ml/m2 (+5%) noRR: LAVI 43 to 47 ml/m2 (+9%)p‐value between groups: <0.001	HFH and deaths: LARR not different from LARR + LVRR
Mathias et al. 48	RCT (sub‐study) – 533	Stage C – NYHA I–II	CRT vs. ICD‐only	12 months	>30% LAV reduction in 60% of patients	Lower rate of HFH and deaths with LA + LVRR vs. LARR or noRR Lower rate of HFH and deaths with LARR vs. noRR
Valzania et al. 101	Observational prospective – 30	Stage C – NYHA III	CRT	12 months	LA area: 26.6 to 23.1 cm2 (−13%) p < 0.05	↓ AF incidence (p = 0.05) ↓ NYHA class (p < 0.001) No difference for HFH
					LA strain: 11.4% to 16.5% (+45%) p < 0.05
St John Sutton et al. 102	RCT – 419	Stage C – NYHA I–II	CRT‐ON vs. CRT‐OFF (excluded from analysis)	5 years (assessment every 6 months)	Overall LA area change not significant (p = 0.95)	No effect of LA area on mortality/morbidity
Singh et al. 103	RCT – 56	Stage C	Dapagliflozin vs. placebo	12 months	LAVI: −2.6 ml/m2 (p = 0.464)
Lee et al. 104	RCT – 92	Stage C – NYHA II–IV	Empagliflozin vs. placebo	36 weeks	Empagliflozin: LAV (on CMR) 79 to 75.5 ml (−4%) Placebo: LAV (on CMR) 87.9 to 86.3 ml (−2%)p‐value between groups: 0.22
Preserved ejection fraction
Tsang et al. 105	RCT – 21	Diastolic dysfunction + LA enlargement	Quinapril vs. placebo	6 and 12 months	Quinapril: LAVI 43 to 40 to 39 ml/m2 (−9%) Placebo: LAVI 38 to 40 to 44 ml/m2 (+16%)p‐value between groups: <0.05
Mak et al. 106	RCT – 44	Stage C – NYHA II–IV	Eplerenone vs. placebo	6 and 12 months	Eplerenone: LAVI 50 to 49 to 52 ml/m2 (+4%) Placebo: LAVI 45 to 44 to 53 ml/m2 (+18%) Non‐significant difference between groups
Kayrak et al. 107	RCT – 110	Post‐MI	Spironolactone vs. standard therapy	6 months	Spironolactone: LAVI 52.3 to 52.2 ml/m2 (−0.2%) Control: LAVI 50.2 to 49.2 ml/m2 (+2%) Non‐significant difference between groups
					Spironolactone: LAEF 53% to 57% (+8%) Control: LAEF 50% to 47% (−6%)p‐value between groups: 0.013
Edelmann et al. 108	RCT – 64	Stage C – NYHA II–III	Supervised exercise training vs. usual care	3 months	Training: LAVI 27.9 to 24.3 ml/m2 (−13%) Control: LAVI 28.2 to 28.6 ml/m2 (+1%)p‐value between groups: 0.001	Improved maximal exercise capacity
Solomon et al. 109	RCT – 149	Stage C – NYHA I–III	ARNI vs. valsartan	36 weeks	ARNI: LAVI 35 to 32.4 ml/m2 (−7%) Valsartan: LAVI 36.8 to 37.1 ml/m2 (+1%)p‐value between groups: 0.007	↓ NYHA class (p = 0.05)
Deswal et al. 110	RCT – 44	Stage C – NYHA II–III	Eplerenone vs. placebo	26 weeks	Eplerenone: LAV 73 to 64 ml (−12%) p = 0.02 Placebo: LAV 80 to 73 ml (−9%) p = 0.09 Non‐significant difference between groups
Edelmann et al. 111	RCT – 422	Stage C – NYHA II–III	Spironolactone vs. placebo	12 months	Spironolactone: LAVI 28.2 to 27.5 ml/m2 (−2%) Placebo: LAVI 27.8 to 27.6 ml/m2 (−1%) Non‐significant difference between groups
Kurrelmeyer et al. 112	RCT – 48 (only women)	Stage C – NYHA II–III	Spironolactone vs. placebo	6 months	Spironolactone: LAVI 32.5 to 33.3 ml/m2 (+3%) Placebo: LAVI 35.7 to 36.7 ml/m2 (+3%) Non‐significant difference between groups
					↓ Type III collagen levels
Shah et al. 113	RCT – 239	Stage C – NYHA I–IV	Spironolactone vs. placebo	12–18 months	Spironolactone: LAV 59.3 to 60.3 ml (+2%) Placebo: LAVI 58.1 to 60.3 ml (+4%) Non‐significant difference between groups	Reduced composite endpoint of cardiovascular death, HFH, or aborted cardiac arrest due to overall LAV reduction
Soga et al. 114	Observational prospective – 58	Stage C (69% HFpEF, 13% HFrEF)	Dapagliflozin	6 months	LAVI 31 to 26 ml/m2 (−16%) p = 0.001
Valvular heart disease
Antonini‐Canterin et al. 42	Observational retrospective – 79	Stage C	Mitral valve surgery (repair or replacement) for severe degenerative MR	6 months	LAVI 68 to 47 ml/m2 (−29%) p < 0.001
					80% LARR Lower LAVI reduction in >45 years old and hypertensive patients
Kim et al. 115	Observational prospective – 303	Stage C – NYHA II–III	Percutaneous mitral valvuloplasty	After procedure, 1 year and 8 years	Total: LAVI 60.2 to 44.8 (−25%) to 69.1 (+15%) ml/m2, p < 0.001 SR: LAVI 48.8 to 35.7 (−27%) to 55.1 (+13%) ml/m2, p < 0.001 AF: LAVI 85.5 to 65 (−24%) to 100.9 (+18%) ml/m2, p < 0.001	Pre‐procedural LAVI and percentage change of LAV immediately after procedure independent predictor of event‐free survival
Marsan et al. 45	Observational prospective – 65	Stage C	Early repair for severe MR (prolapse)	6 and 12 months	LAVI 43 to 25 to 23 (−47%) ml/m2, both p < 0.05
					85% LARR (1 year)
D'Ascenzi et al. 116	Observational prospective – 32	Stage C – NYHA II–III–IV	TAVR	40 days and 3 months	LAVI 47.3 to 42.8 to 43.5 (−8%) ml/m2, both p < 0.05
					PALS 14.4% to 19% to 19.1% (+33%), p < 0.05 and p < 0.001
Candan et al. 46	Observational prospective – 53	Stage C – NYHA I–II–III	Mitral valve surgery (repair or replacement) for severe MR	6 months	LAVI 58.2 to 43.9 (−25%) ml/m2, p = 0.001 No difference between type of surgery
Hatani et al. 117	Observational retrospective – 83	Stage C	Surgical AVR	1 month, 1 year and 3 years	LAVI 51.3 to 43.7 (−15%) to 44.7 to 46.1 ml/m2, p < 0.001	Residual LA dilatation at 1 year after AVR was associated with reduced event‐free survival
Toprak et al. 118	Observational prospective – 25	Stage C – NYHA II–IV (60%)	Transcatheter mitral valve repair (MitraClip implantation)	12 months	2D‐LAVI max: 51.5 to 50 ml/m2 (−3%) p = 0.512 3D‐LAVI max: 55.2 to 51.5 ml/m2 (−7%) p = 0.018	Pre‐procedural 3D‐LAV and LA reservoir strain associated with MACE (univariate analysis)
					LA reservoir strain: 7.66% to 11.15% (+46%), p < 0.001	LA reservoir strain and 3D‐LAV min correlated with ↓ MR, ↓ NYHA class, ↑ 6MWD
Avenatti et al. 119	Observational retrospective – 35	Stage C – NYHA III	Transcatheter mitral valve repair (MitraClip implantation)	30 days	LAVI max: 74 to 53 ml/m2 (−28%), p = 0.008 LAVI max functional MR: 70 to 55 ml/m2 (−21%), p < 0.05 LAVI max primary MR: 65 to 49 ml/m2 (−25%), p < 0.05	Significant inverse correlation between LA stiffness change and 6MWD in 14 patients
					↓ LA stiffness

3D, three‐dimensional; 6MWD, 6‐min walking distance; 6MWT, 6‐min walking test; AF, atrial fibrillation; ARNI, angiotensin receptor–neprilysin inhibitor; AVR, aortic valve replacement; CMR, cardiac magnetic resonance; CRT, cardiac resynchronization therapy; HF, heart failure; HFH, hospitalization for heart failure; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; ICD, implantable cardioverter‐defibrillator; IQR, interquartile range; LA, left atrium; LAEF, left atrial emptying fraction; LARR, left atrial reverse remodelling; LAV, left atrial volume; LAVI, left atrial volume index; LV, left ventricle; LVRR, left ventricular reverse remodelling; MACE, major adverse cardiovascular event; MI, myocardial infarction; MR, mitral regurgitation; NYHA, New York Heart Association; PALS, peak atrial longitudinal strain; RCT, randomized clinical trial; RR, reverse remodelling; SR, sinus rhythm; TAVR, transcatheter aortic valve replacement.

The Prospective Study of Biomarkers, Symptom Improvement, and Ventricular Remodelling During Sacubitril/Valsartan Therapy for Heart Failure (PROVE‐HF) study showed a reduction in LA volume index (from 37 to 29 ml/m2, p < 0.001) after 12 months in patients treated with angiotensin receptor–neprilysin inhibitor (ARNI), along with an increase in LVEF and decrease in LV volumes and E/e′ ratio. Similarly, in the multicentre randomized EVALUATE‐HF trial, the ARNI group showed significant greater reductions in LA volume index, LV volumes and E/e′ ratio as compared to enalapril, while no significant change in LVEF was observed.

The rapid drop in N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) levels during treatment with ARNI , testifies the direct effect of sacubitril/valsartan on ventricular wall stress and a possible acute effect on filling pressures, perhaps related to increased venous capacitance or natriuresis. It has been demonstrated that the significant reduction in circulating NT‐proBNP (and high‐sensitivity cardiac troponin T) is correlated to the LA volume index reduction among HFrEF patients treated with ARNI.

Left atrial reverse remodelling can be considered an active player of the so‐called ‘complete left‐sided reverse remodelling’. , This concept has been elucidated by trials testing the structural response to cardiac resynchronization therapy (CRT) (Table  ). Results from the Multicenter Automatic Defibrillator Implantation Trial with Cardiac Resynchronization Therapy (MADIT‐CRT) study showed that reductions in LA volume with CRT‐defibrillator (CRT‐D) at 1 year were highly associated with reduction in LV volume, suggesting a cause–effect relationship. In this trial, most CRT‐D patients experienced ≥20% reduction in LA volume, while the implantable cardioverter‐defibrillator (ICD)‐only group had a significantly lower LA response. Patients who showed a favourable LA response to CRT‐D experienced a subsequent reduction in the risk of atrial tachycardia, HF events and death. Importantly, atrial tachycardia risk reduction appeared to be related to LA response independently of LV structural changes.

Two studies revealed that 20%–45% of patients with CRT showed LA reverse remodelling in absence of LV reverse remodelling , and contrasting results in terms of outcome have been reported (Table  ). Kloosterman et al. retrospectively assessed LA reverse remodelling in 365 patients who were eligible for CRT. They reported that discordant left side reverse remodelling (LA without LV reverse remodelling) showed comparable mortality and HF hospitalization risk to patients with both LA and LV reverse remodelling. Thus, successful CRT can reduce LA size and improve LA pump function and LA reverse remodelling might improve outcome in CRT patients, even in the absence of LV remodelling.

The LA plays an active role in HFrEF pathophysiology and appears as a promising target to evaluate the efficacy of HF treatments. Although trials showed variable results, additional efforts should be addressed to clarify whether the assessment of LA reverse remodelling, as an independent entity, can provide prognostic information.

Heart failure with preserved ejection fraction

The assessment of LA response to different interventions in HF with preserved ejection fraction (HFpEF) patients has not been fully evaluated and available data have shown contrasting results.

Mineralocorticoid receptor antagonists decrease extracellular matrix turnover and myocardial collagen, one of the mechanisms of atrial disease. Studies investigating the impact of MRAs on cardiac function did not show significant LA reverse remodelling, despite many of them reported improvement in certain parameters of LV diastolic function (Table  ). In the Treatment of Preserved Cardiac Function Heart Failure With an Aldosterone Antagonist (TOPCAT) echocardiographic sub‐study, while spironolactone was not associated with significant change in either LA and LV structure over time, the overall reduction in LA volume was associated with an independent lower risk of subsequent occurrence of CV death, HF hospitalization, or aborted cardiac arrest. The reasons why the effects of antagonizing aldosterone activity did not show a direct effect on LA structure are uncertain. It is possible that the anti‐fibrotic effect of MRAs may require a longer time to result in functional or structural reverse remodelling.

In the phase II Prospective Comparison of ARNI With ARB on Management of Heart Failure With Preserved Ejection Fraction (PARAMOUNT) trial, , while the use of ARNI did not show a significant effect on changes of LV structure or function, diastolic function, or LV mass after 12 and 36 weeks of treatment, LA size and volume significantly decreased after 36 weeks (Table  ). The concomitant inhibition of RAAS activity and the augmentation of the actions of natriuretic peptides have incremental effects on cardiac haemodynamics and structural remodelling. In the aforementioned study, the use of ARNI rapidly reversed LA remodelling without an apparent effect on LV size or diastolic function. This result may be explained by a rapid decrease in LV filling pressure, testified by the reduction of NT‐proBNP levels. LA remodelling reflects increased LV filling pressure in a more robust way than Doppler‐derived measures of diastolic function which are subject to greater variability. Also, the improvement of New York Heart Association (NYHA) functional class after 36 weeks may possibly explain the independent contribution of LA reverse remodelling to improvement of symptoms in HFpEF patients.

The Exercise training in Diastolic Heart Failure (Ex‐DHF) study assessed the effect of structured exercise training compared to usual care in 64 patients with HFpEF. It demonstrated that exercise training positively affected exercise capacity and quality of life in HFpEF, through improvement in LV diastolic function and reduction in LA volume. BP values and body mass index did not change in either group, hence the effect on LA size reduction is not likely to be due to the haemodynamic effect. In fact, a significant reduction of procollagen type I plasma levels induced by exercise training was reported. It seems reasonable that LA reverse remodelling may be related to reduced collagen turnover.

The link between LA reverse remodelling and outcomes in patients with HFpEF needs to be evaluated in larger trials, possibly with longer follow‐up. However, targeting therapies based on LA response in HFpEF seems to be a rational approach.

Valvular heart disease

Mitral valve disease is a well‐studied model that represents the consequences of haemodynamic impairment on the LA. In mitral regurgitation (MR), volume overload has a direct effect on the LA causing chamber enlargement and myocyte hypertrophy to compensate the haemodynamic stress and to prevent pulmonary congestion. At a cellular level, atrophy and interstitial fibrosis progressively impair LA function and decrease atrial elasticity.

Candan et al. assessed the relationship between LA peak longitudinal reservoir strain and LA reverse remodelling in 53 patients with severe MR undergoing surgical valve repair or replacement (Table  ). LA volume index significantly decreased after surgery, regardless of the type of intervention. Also, pre‐operative higher LA volume, younger age and higher peak LA longitudinal strain values were independent predictor of LA reverse remodelling.

Left atrial function assessed by speckle tracking echocardiography has been associated with CV outcomes in patients with primary degenerative moderate asymptomatic MR. Data from 87 subjects with degenerative MR enrolled in the Endovascular Valve Edge‐to‐Edge Repair Study II (EVEREST II) trial displayed LA strain changes after both surgical or transcatheter mitral valve repair. LA strain improvement was dependent on baseline LA function as only in patients with normal or high baseline strain values, MR reduction resulted in normalization of LA strain. Moreover, in a cohort of 25 patients undergoing MitraClip, pre‐operative LA volume measured by three‐dimensional echocardiography (but not by two‐dimensional) and LA reservoir strain were associated with 1‐year major adverse CV events.

Recently, transcatheter indirect mitral annuloplasty showed benefits in reducing functional MR in HFrEF patients with a high procedural successful rate; an improvement in NYHA class was reported too. One individual patient data meta‐analysis found that Carillon device implantation resulted in significant left‐sided reverse remodelling, as demonstrated by reduction in LV end‐diastolic and end‐systolic volume and reduction in LA volume.

The mechanisms behind LA remodelling in aortic stenosis (AS) are thought to be related to LV hypertrophy and increased LV filling pressure due to high afterload. The impact of aortic valve replacement (AVR) has yet to be explained. LV and LA reverse remodelling have been assessed in one cohort of patients with severe AS undergoing AVR (Table  ). LA volume index rapidly decreased after 1 month along with a reduction in LV mass index. Patients who did not experience LA and LV reverse remodelling appeared to exhibit worse long‐term outcomes.

Even transcatheter AVR (TAVR) has been demonstrated to be associated with significant LA reverse remodelling. D'Ascenzi et al. reported a significant reduction of LA size at a 40‐day follow‐up after TAVR. Alongside the recovery of LA structure, TAVR was accompanied by a significant increase in global peak atrial longitudinal reservoir strain. Pre‐procedural LA volume and trans‐aortic mean gradient change were reported as predictors of LA volume reduction 3 months following TAVR.

Advanced heart failure (stage D)

When reversal of LA remodelling fails to occur, the disease progresses to stage D of atrial disease. The lack of LA volume reduction despite treatment identifies the irreversible advanced end‐stage, mainly related to a larger amount of fibrosis. Albeit not largely proven, in this stage, no therapeutic intervention may have beneficial consequences on LA function or structure. Only acting at earlier stages may prevent LA remodelling progression (Figure  ).

New treatments and future perspectives

While the effects of sodium–glucose cotransporter 2 inhibitors on atrial function have not been assessed, to date, , , , new drugs acting directly on cardiac function may have an impact also on the LA. Compared with placebo, the myosin activator omecamtiv mecarbil improved LA function, as shown by a reduced LA minimal volume and a larger emptying fraction, in the Chronic Oral Study of Myosin Activation to Increase Contractility in HF (COSMIC‐HF) trial and reduced the risk of new‐onset AF and stroke, likely through an improvement in LA function, in the larger Global Approach to Lowering Adverse Cardiac outcomes Through Improving Contractility in HF (GALACTIC‐HF) trial. , , ,

Rhythm control for AF may improve LA function and this may improve patients' outcomes. These effects, shown also after pulmonary vein isolation, may be particularly effective in patients with HFpEF such as 79% of the patients analysed in the Catheter Ablation versus Antiarrhythmic Drug Therapy for Atrial Fibrillation (CABANA) trial. Improvement in LA function, as well as in outcomes, has been shown also in the Catheter Ablation versus Standard Conventional Therapy in Patients with Left Ventricular Dysfunction and Atrial Fibrillation (CASTLE‐AF) trial, enrolling patients with reduced LVEF.

Our review shows that LA reverse remodelling may occur in clinical practice and be related with better outcomes. Changes in LA size and function may be related with progression of cardiac function and thus be a target of treatment or rather be a surrogate measurement of the severity of cardiac dysfunction. Also in this last case, changes in LA function may allow a more complete assessment of cardiac function and of the risk of cardiac events. LA function depends on both systolic and diastolic LV function and may predict not only HF events, as related with intracardiac pressure and congestion, but also other events including stroke and AF. However, even if measurements of LA function seem more stable than others, more data regarding their reproducibility and their clinically meaningful threshold are warranted. Future studies will also show whether LA function can be a target of treatment. Some interventions, such as pulmonary vein isolation for rhythm control in patients with AF may favourably affect LA function and this may mediate their effects on outcomes. , , , Also percutaneous treatment of MR may have an effect on LA function that may be larger than on LV function.

Conclusions

Left atrial remodelling and LA disease are markers of CV disease and may also have an impact on clinical outcomes. Counteracting the underlying mechanisms and removing the causative factors may restore LA structure and function. Increasing evidence indicates significant LA reverse remodelling after initiation of medical therapy across all the stages of development and progression of HF. Although a clear definition of LA reverse remodelling is needed, assessing LA response to targeted therapies appears to be a valid parameter to re‐define patients' risk stratification. Prospective clinical trials are required to establish the role of LA reverse remodelling as a marker of therapeutic response in clinical practice and to assess its potential independent role as a mechanism of development and progression of HF.