Perioperative cardiovascular pathophysiology in patients undergoing lung resection surgery: a narrative review

Summary

Although thoracic surgery is understood to confer a high risk of postoperative respiratory complications, the substantial haemodynamic challenges posed are less well appreciated. This review highlights the influence of cardiovascular comorbidity on outcome, reviews the complex pathophysiological changes inherent in one-lung ventilation and lung resection, and examines their influence on cardiovascular complications and postoperative functional limitation. There is now good evidence for the presence of right ventricular dysfunction postoperatively, a finding that persists to at least 3 months. This dysfunction results from increased right ventricular afterload occurring both intraoperatively and persisting postoperatively. Although many patients adapt well, those with reduced right ventricular contractile reserve and reduced pulmonary vascular flow reserve might struggle. Postoperative right ventricular dysfunction has been implicated in the aetiology of postoperative atrial fibrillation and perioperative myocardial injury, both common cardiovascular complications which are increasingly being appreciated to have impact long into the postoperative period. In response to the physiological demands of critical illness or exercise, contractile reserve, flow reserve, or both can be overwhelmed resulting in acute decompensation or impaired long-term functional capacity. Aiding adaptation to the unique perioperative physiology seen in patients undergoing thoracic surgery could provide a novel therapeutic avenue to prevent cardiovascular complications and improve long-term functional capacity after surgery.

Editor's key points.

• •Lung resection has a significant impact on cardiovascular physiology, and postoperative cardiovascular complications are common.
• •Bringing together historical understanding of the right ventricular and pulmonary vascular pathophysiology of one-lung ventilation and lung resection, this review highlights the role that right ventricular dysfunction can play in postoperative morbidity and functional limitation.
• •Further work is required to explore the hypothesis that aiding adaptation to the unique perioperative physiology will provide a novel therapeutic target to prevent complications and improve functional capacity after surgery.

Thoracic surgery is recognised to confer a high risk of postoperative respiratory complications. Less well appreciated however, are the substantial haemodynamic challenges posed and the significant potential for cardiovascular complications. Here we examine the hypothesis that the cardiovascular consequences of thoracic surgery significantly influence short- and long-term outcome and therefore merit further consideration. This review will highlight the influence of cardiovascular comorbidity on outcome, review the complex pathophysiological changes inherent to one-lung ventilation and lung resection, and examine their influence on cardiovascular complications and functional limitation.

Haemodynamic challenges of thoracic surgery

Short-term intraoperative physiology during one-lung ventilation and lung resection

During a period of one-lung ventilation (OLV) through a combination of gravitational effects and increased non-dependent (operative) lung pulmonary vascular resistance (PVR) (owing to hypoxic pulmonary vasoconstriction, lung collapse, surgical manipulation, and pulmonary artery clamping), cardiac output is redirected to the dependent, ventilated, non-operative lung which serves to maintain oxygenation and minimises shunt.¹ Using a technetium-labelled single-photon emission computed tomography technique to assess pulmonary blood flow distribution in a porcine model of OLV and thoracotomy, Kozian and colleagues² observed that only a minimal percentage of cardiac output continues to pass through the operative lung during OLV. This near doubling in dependent lung blood flow has been consistently demonstrated to result in a 25–35% increase in pulmonary artery pressure (PAP) and a 20–50% increase in PVR3, 4, 5 during the period of OLV.

Haemodynamic adaptation to these conditions of acutely increased afterload relies both on the ability of the pulmonary circulation to accommodate this increased flow (pulmonary vascular flow reserve) and the ability of the right ventricle (RV) to maintain cardiac output in the face of the ensuing increased afterload (RV contractile reserve). As occurs in health at times of increased cardiac output, for example during exercise, pulmonary vascular flow reserve reflects the ability of the pulmonary circulation to accommodate increased cardiac output without excessive increases in PAP.

The ability of the RV to adapt in the face of acutely increased afterload reflects a complex autoregulatory response involving both heterometric (where ventricular volume changes) and homeometric (where ventricular volume remains constant) responses.⁶ In initial response to acutely increased RV pressure (as occurs at the institution of OLV), the RV dilates. Such dilation induces myocardial stretch, and stroke volume is initially maintained through the Frank–Starling mechanism. Occurring subsequently over a period of minutes, homeometric autoregulation (the Anrep effect) describes a reflex mediated increase in intrinsic RV contractility in response to increased afterload.⁷ Examination of RV pressure–volume loops between baseline and during OLV after lobar pulmonary artery clamping clearly demonstrates this phenomenon in humans undergoing lung resection in experimental conditions (Fig 1).

Fig 1.

Annotated pressure–volume loop illustrating right ventricular haemodynamic changes during one-lung ventilation (OLV) and lobectomy. The same right ventricle (RV) pressure–volume (PV) loop is reproduced in parts a–c. Solid line represents preoperative baseline; dashed line represents intraoperative situation during OLV and after lobar pulmonary artery (PA) ligation. (a) During OLV and after lobar PA ligation, the RV end-systolic pressure (RVP_ES) is increased whereas stroke volume (SV) remains unchanged. (b) In response to this acutely increased afterload, SV is maintained by homeometric autoregulation with a sympathetically mediated increase in RV contractility (the Anrep effect), illustrated by an increase in the gradient of the RV end-systolic pressure–volume relationship (ESPVR). (c) Maintaining SV in the face of increased afterload results in increased RV stroke work (RVSW) illustrated by the changes in the internal area of the PV loop from baseline (red area only) to OLV and PA ligation (red and pink shaded areas). Novel illustration drawn by the author from data by Wink and colleagues.³.

A large number of patients undergo thoracic surgery with OLV every day however, and clinical experience reveals that it is generally well tolerated; despite this radical redistribution of pulmonary blood flow, systemic arterial blood pressure and cardiac output are maintained in the vast majority. It is likely, however, that in a minority of patients, pulmonary vascular or RV comorbidity results in an inability to adequately adapt. In case reports of thoracic surgery in patients with established pulmonary arterial hypertension (PAH), loss of pulmonary vascular reserve is evidenced by marked increases in PAP^8,9 during surgery. In these limited reports, mortality and complication rates are high^8,10; for these reasons, there has historically been a hesitation to perform thoracic surgery on patients with PAH or RV dysfunction.⁸

Patients with PAH are at increased risk after thoracic surgery, with Ramakrishna and colleagues¹¹ observing an incidence of multisystem composite morbidity in 61.5% of patients undergoing thoracic surgery, compared with 16.7% in patients who underwent gynaecological, urological, plastic, dermatological, or breast surgery. By interrogation of a large US national administrative database, Smilowitz and colleagues¹² reported a prevalence of PAH in patients undergoing thoracic surgery of 2%, although it is likely that this figure, based on International Classification of Diseases, 10th Revision (ICD-10) codes, underestimates the true incidence of PAH.¹³ In this cohort, patients with PAH experienced a two-fold increase in major adverse cardiovascular events.

Historically, invasive assessment of PVR had been explored as a method of identifying patients unable to tolerate the haemodynamic challenge of lung resection.^14,15 More recently, unilateral balloon occlusion of the pulmonary artery (with or without contemporaneous assessment of RV haemodynamics) has been advocated for assessment of patients perceived to be at extreme risk (such as those undergoing pneumonotomy or pleuropneumonectomy), but the technique is invasive, resource-intensive, and associated with significant risks. Examination of the RV response to exercise may offer a less invasive alternative; in 1996 Okada and colleagues¹⁶ demonstrated a clear association between the preoperative RV response to exercise and postoperative complications, where a decline in RV ejection fraction (RVEF) on exercise was associated with increased postoperative cardiopulmonary morbidity and prolonged hospital stay. Although both techniques show promise in identifying patients lacking either pulmonary vascular, or RV contractile reserve, neither has seen sufficiently rigorous validation to be recommended in mainstream practice. None of the current UK,^17,18 European,¹⁹ or American²⁰ guidelines of preoperative assessment for the patient undergoing thoracic surgery advocate any form of cardiovascular assessment beyond risk factor based assessment of revised cardiac risk index.

Longer-term physiological changes after lung resection

The observation of impaired RV function after lung resection has been consistently reported.^5,16,21 Our research group conducted the only sequential cardiovascular magnetic resonance (CMR; the gold standard for assessing RV function) imaging study assessing RV function after lung resection.²² Confirming these earlier reports, this 27-patient study demonstrated a reduction in RVEF from 50.5% [6.9%] preoperatively to 45.6% [4.5%] on postoperative Day 2, with ongoing impairment (44.9% [7.7%]) at 2 months (P=0.003), without any contemporaneous changes in left ventricular (LV) function (Fig 2).²² Impaired RV free wall longitudinal strain was also present 2 months postoperatively, suggesting ongoing RV dysfunction.²³

Fig 2.

Changes in right, but not left, ventricular function after lung resection. Novel illustration drawn from the data of McCall and colleagues,²² in which 27 patients underwent serial cardiovascular magnetic resonance imaging at three perioperative timepoints. Right ventricular (RV) ejection fraction is significantly reduced on postoperative Day 2 (POD2) and at 2 months (∗P<0.01 for both). There were no changes in left ventricular (LV) ejection fraction over the study period (P=0.62, one-way repeated-measures analysis of variance).

The hypothesis that RV dysfunction after lung resection results from increased afterload^21,24, 25, 26, 27 is well established; however, until recently this has not been demonstrated. Although as discussed above, numerous studies demonstrate that both PAP and PVR increase acutely intraoperatively, these changes are transient, returning to baseline within hours of surgery^4,5,16,21 (Fig 3). Importantly, however, although PVR and PAP are commonly used indices in clinical practice, they are relatively incomplete measures of afterload. Both measure opposition to mean flow (static afterload), ignoring the pulsatile components of afterload.²⁸

Fig 3.

Changes in pulmonary vascular resistance and pulmonary artery pressure occurring during and after lung resection. Novel illustration drawn from data reported by Waller and colleagues,⁴ where invasive haemodynamic monitoring is performed in patients undergoing pneumonectomy (n=10) and lobectomy (n=11). Despite transient intraoperative increases, pulmonary vascular resistance (PVR) and pulmonary artery pressure (PAP) return to baseline within 24 h postoperatively. Similar findings have been reproduced several research groups.^{5 15 20} ∗P<0.05 vs baseline value. Insufficient data provided in original publication to allow generation of error bars.

Up to half of the hydraulic power in the main PA is contained in the pulsatile components of flow comprising, resistance, capacitance, inertia, and pulse wave reflection. As such, true RV afterload, the pulmonary input impedance, is a composite of both static and pulsatile components.^28,29 Assessment of pulmonary input impedance is technically challenging, requiring simultaneous high-fidelity assessment of pulmonary arterial pressure and flow, and is therefore rarely performed clinically. The main distinction (and complexity) of impedance analysis, from other methods of afterload analysis, is that it is performed and interpreted in the frequency domain. Flow and pressure profiles are measured simultaneously at the same point in the pulmonary artery and then decomposed by Fourier analysis into a collection of sinusoidal waves of increasing frequencies. Impedance spectra are formed by the modulus of pressure and flow for each harmonic/frequency with the impedance modulus across the range of frequencies being representative of different components of afterload. This allows the relative contribution of afterload generated at a small vessel level (low frequency domains – PVR), resulting from pulse wave reflection within the pulmonary vascular tree and resulting large vessel stiffness (high frequency domains – characteristic impedance) to be distinguished from one another.^29,30

In the CMR study described above, pulmonary artery acceleration time and pulmonary artery distensibility, surrogate measures of pulsatile afterload, were both reduced.²² Furthermore, there was a unilateral increase in wave reflection in the branch of the pulmonary artery supplying the operative lung, which led to redistribution of blood through the non-operative lung.²³ Both the wave reflection in the operative lung and the resultant redistribution of blood flow were associated with impaired RVEF and chronic impairment in RV free wall strain, finally demonstrating the widely hypothesised association between increased RV afterload and deterioration in RV function after lung resection.²³

Previous work in animal models of pneumonectomy supports the hypothesis that afterload is increased postoperatively, but is relatively invisible to conventional measures. In a porcine model of pneumonectomy, Heerdt³¹ demonstrated that 3 days after pneumonectomy whereas there is a modest increase in PVR of 16%, characteristic impedance is increased by 56%.³¹ In a longer-term canine pneumonectomy model, Lucas and colleagues³² observed persistent changes in characteristic impedance (and to a lesser extent, increased PVR), 5 yr after resection.³²

RVEF is a notoriously load-dependent marker of RV function²⁸; although increased afterload as a cause of reduced RVEF after lung resection is intuitive and an association between indices of afterload and decline in RVEF has now been demonstrated, the potential for changes in intrinsic contractility to contribute to this decline has seen little attention. Animal models simulating pulmonary thromboembolism (PTE) by performing anaesthesia, thoracotomy, and transient pulmonary artery clamping (an analogous situation to lung resection) demonstrate a persistent decline in RV function hours after removal of clamps and normalisation of afterload – a persistence that would not be expected if mechanical afterload were the sole cause of RV dysfunction.³³ In microsphere embolisation models of PTE, inflammatory gene expression and neutrophil accumulation is evident after 18 h, providing evidence of acute RV inflammatory injury triggered by increased afterload.^34,35 It is plausible that acute increases in afterload occurring intraoperatively and immediately postoperatively in patients undergoing thoracic surgery could lead to induction of myocardial inflammation and impairment of RV function by a similar mechanism. Data from Heerdt and colleagues³⁶ demonstrating oxidative and nitrosative stress associated alterations in myocardial calcium signalling in a porcine model of thoracotomy and lung resection supports such a hypothesis. Furthermore, there is evidence from animal models that systemic oxidative and nitrosative stress related to lung ischaemia (during the period of OLV) and reperfusion is associated with wide ranging remote organ injury, including cardiac dysfunction.^37,38

Cardiovascular comorbidity influences outcome after thoracic surgery

Cardiovascular comorbidities are common in patients undergoing thoracic surgery; ischaemic heart disease (IHD) is reported to affect 12–75% of patients undergoing lung resection.^39,40 Patients with IHD undergoing lobectomy have been shown to have a five-fold increase in risk of 30-day mortality,³⁹ and those undergoing pneumonectomy have a five-fold increase in risk of in-hospital mortality.⁴¹ Furthermore, Sandri and colleagues³⁹ demonstrated that although patients with IHD undergoing video-assisted thoracic surgery (VATS) lobectomy have similar rates of major adverse cardiac events (MACE), when cardiovascular or pulmonary complications occurred, patients with IHD were less likely to recover (‘failure to rescue’), an effect they hypothesise is being mediated by reduced aerobic reserve. Congestive cardiac failure (CCF) is also a significant risk factor for mortality after lung resection; the risk of in-hospital mortality is five times higher in patients with CCF undergoing lobectomy, and 23 times higher in those undergoing pneumonectomy.^41,42

Beyond the diagnoses of cardiovascular comorbidities, objective assessment of cardiovascular health has also been shown to predict postoperative outcomes after thoracic surgery. Natriuretic peptides (both B-type [BNP] and its prohormone precursor, NTpro-BNP), secreted by ventricular myocytes during periods of volume overload, pressure overload, or both, have been shown to stratify risk. Nojiri and colleagues found that patients undergoing lung resection with preoperative BNP levels >100 pg ml⁻¹ had an 87% rate of cardiovascular and respiratory complications, compared with 47% when BNP was 30–100 pg ml⁻¹, and 11% with when BNP <30 pg ml⁻¹ (P<0.0001).⁴³ In a post hoc analysis of a small cohort of patients undergoing thoracotomy and lobectomy, Young and colleagues⁴⁴ demonstrated that preoperative BNP predicted postoperative functional decline 3 months after resection.⁴⁴ Current combined guidelines from the European Society of Anaesthesia and European Society of Cardiology recommend ‘considering the use of preoperative BNP to obtain prognostic information in high-risk patients prior to non-cardiac surgery’,⁴⁵ but at present the role of natriuretic peptides in aiding perioperative decision-making in patients undergoing lung resection is unclear, and clear recommendations for their use cannot be made.

Cardiopulmonary exercise testing (CPET) has been widely used to assess cardiovascular reserve and has been demonstrated to predict perioperative outcomes in a wide range of surgical specialities.⁴⁶ Peak oxygen uptake (VO₂) and the ventilatory equivalent for carbon dioxide (expired minute volume divided by pulmonary carbon dioxide output – VE/VCO₂) have been most consistently identified as parameters that predict outcomes after lung resection47, 48, 49, 50; however, both measures will be influenced by both cardiovascular and respiratory comorbidities. The oxygen pulse (VO₂/heart rate) is considered to be a surrogate measure of stroke volume during CPET and is therefore thought of as a more independent measure of cardiac performance.⁵¹ A single retrospective study in 2010⁵² has shown that patients undergoing lung resection who have a lower oxygen pulse during preoperative CPET (and therefore decreased ability to increase stroke volume in the face of exercise) are at a higher risk of cardiopulmonary complications; however, these findings have not been replicated in other thoracic surgery studies.^47,50

Elevated preoperative resting heart rate and elevated heart rate during peak exercise (during preoperative CPET) have both been identified as risk factors for cardiopulmonary complications after lung resection.^52,53 Impairment of heart rate recovery after preoperative exercise testing has similarly been demonstrated to predict cardiopulmonary complications after lung resection.⁵⁴

Cardiovascular complications are common after thoracic surgery

Although cardiorespiratory complication rates in excess of 25% are commonly reported after thoracic surgery,55, 56, 57 the widespread practice of reporting cardiovascular and respiratory complications grouped together makes the true incidence of postoperative cardiovascular complications difficult to define. It is clear, however, that some cardiovascular complications, particularly perioperative myocardial injury (PMI) and postoperative atrial fibrillation (POAF), are common after thoracic surgery with POAF often cited as one of the more common of the composite complications.55, 56, 57

Perioperative myocardial injury

Definition

The definition of PMI varies throughout the literature. The two most common definitions are myocardial injury after noncardiac surgery (MINS)^58,59 and PMI.^60,61 Both definitions rely on identification of postoperative troponin elevation.

MINS is defined as myocardial injury caused by ischaemia which occurs within 3 days postoperatively identified by a high-sensitivity (hs) troponin-T assay of more than 20 ng L⁻¹.⁵⁹ To diagnose MINS, myocardial ischaemia does not need to be demonstrated, but troponin increase must be perceived to have occurred in the absence of ‘evidence of a non-ischaemic cause’. More recently, the Standardized Endpoints in Perioperative Medicine initiative (StEP) initiative⁶⁰ have advocated the use of the term PMI to describe a troponin increase in excess of the 99th percentile upper reference limit (in the absence of overt ischaemia) regardless of mechanism, highlighting their belief that it is not clinically possible to distinguish which increases of troponin are attributable to which mechanisms. If one or more ischaemic features are identified alongside a perioperative troponin increase (e.g. chest pain or ischaemic ECG changes), then the fourth universal definition for myocardial infarction is fulfilled and myocardial infarction (rather than injury) has occurred.⁶²

Incidence

Although myocardial infarction after thoracic surgery is rare (estimated incidence, 0.2–1.1%⁶³), there is growing evidence that a postoperative troponin increase is common. Within thoracic surgery, the incidence of PMI ranges from 14%⁶⁴ to 28%⁶⁵ (Table 1). In the largest of these studies, the Vascular Events in Noncardiac Surgery Patients Cohort Evaluation (VISION) study (with more than 40 000 patients), demonstrated that the incidence of MINS after thoracic surgery was 19.8% (17.6–22.2%).⁶⁶ Smaller observational studies suggest that smoking history and the extent of surgery are independent risk factors for PMI (Table 1).

Table 1.

Summary of studies reporting perioperative myocardial injury after thoracic surgery. ∗Adult patients with cardiac risk factors, defined as patients ≥65 yr old or patients <65 yr old with known cardiovascular pathology (history of cardiac, cerebral, or peripheral vascular pathology) who underwent elective thoracic surgery. Bilobe, bilobectomy; HR, hazard ratio; lobe, lobectomy; MINS, myocardial injury after noncardiac surgery; NTproBNP, N-terminal pro-brain natriuretic peptide; PMI, perioperative myocardial injury; pneumo, pneumonectomy; POD, postoperative day.

Author, year	N	Patient population	PMI/MINS definition	Incidence of PMI	Risk factors for PMI	Association with outcome
Lucreziotti and colleagues,67 2007	n=50: 8 (16%) pneumo 30 (60%) lobe/bilobe 8 (16%) sub-lobar 4 (8%) pleural decortication	Surgery by thoracotomy with postoperative stay more than 5 days	Troponin I increase ≥0.06 μ L−1 on Days 1, 3, and 5	20% within 5 days	- ≥2 coronary risk factors - Chronic antiplatelet therapy - Pneumonectomy/pericardectomy - Transient ST-segment change	Postoperative outcomes not reported.
Muley and colleagues,64 2011	n=64 20 (31%) pneumo 24 (38%) lobe/bilobe 20 (31%) sub-lobar	Thoracic surgery by thoracotomy	Troponin I ≥0.32 ng ml−1 on first postoperative day	14%	- Intrapericardial procedures	Association between PMI and NTproBNP elevation. No association between PMI and early postoperative outcomes or 90-day survival.
Hua and colleagues,68 2016	n=491 No procedural detail	Thoracic surgery under single surgeon over 5 yr	Troponin I ≥0.04 ng ml−1 on first postoperative day	16%	- Increasing age - Hyperlipidaemia - Smoking history - Peripheral vascular disease	Majority referred for cardiac assessment. No difference in mortality between those with and without PMI.
González-Tallada and colleagues,65 2020	7 (4%) pneumo 104 (59%) lobe 39 (22%) sub-lobar 27 (15%) pleural/chest wall	High cardiovascular risk patients undergoing elective thoracic surgery∗	Troponin I ≥0.04 ng ml−1 on POD 1 and 2, after excluding non-ischaemic causes	27.3%	- Smoking history - Type of surgery (pneumonectomy > lobectomy)	No difference in 30-day mortality or postoperative cardiovascular complications between those with and without PMI (stroke/angina/myocardial infarction/atrial fibrillation).
Uchoa and Caramelli,69 2020	n=151 13 (9%) pneumo 80 (53%) lobe/bilobe 58 (38) sub-lobar	Elective lung resection procedures	Troponin I ≥0.16 ng ml−1 within 48 h postoperatively	49.7%	- Not reported	Troponin I of 0.16–0.31 on POD2 had a HR of 12.0 (1.8, 80.0) for 1-yr mortality. Troponin I of ≥0.32 had a HR of 21.5 (1.5, 311.6) for 1-yr mortality.

Mechanisms of perioperative myocardial injury after thoracic surgery

Unlike a type-1 myocardial infarction, PMI is not thought to be attributable to coronary artery obstruction and disruption of blood supply to the myocardium. Rather, PMI or MINS is widely described as having the same mechanism as a type-2 myocardial infarction,⁶² in which oxygen supply/demand imbalance leads to transient subclinical myocardial injury.⁷⁰

As described above, thoracic surgery is associated with marked, transient increases in PAP and PVR with evidence of impaired RV function postoperatively (but preservation of LV function).²² Contemporaneous troponin measurement revealed that elevation in postoperative troponin was associated with RVEF but not left ventricular ejection fraction (LVEF),⁷¹ offering the hypothesis that after thoracic surgery, the right-sided haemodynamic stresses may be enough to cause an elevated troponin in some patients. This is supported by the observation that the level of troponin elevation is also associated with the volume of lung resected, with pneumonectomy being associated with a greater troponin increase than lesser resections.^65,67

Outcomes

It is increasingly being appreciated that PMI is an important marker of a covert pathophysiological process during the perioperative period and is associated with increased mortality at both 1 month and 1 yr postoperatively^61,66,72,73 in patients undergoing major noncardiac surgery. In thoracic patients, MINS has been associated with increased length of hospital stay⁶⁵ but its effect on mortality is unclear. Two observational studies found no increase in 30-day mortality,^65,68 whereas one demonstrated that a postoperative elevated troponin I was associated with increased 1 yr mortality⁶⁹ (Table 1). In two of these observational studies, none of the patients with MINS experienced clinical signs of ischaemia^65,67; it has been suggested that patients undergoing thoracic surgery are less likely to report ischaemic cardiac chest pain owing to the high analgesic demands of the surgery itself.

Preventative strategies

The Perioperartive Ischemic Evaluation-2 (POISE-2) trial demonstrated that neither aspirin nor clonidine reduced the incidence of postoperative MI, and that the administration of aspirin was associated with increased bleeding risk.⁷⁴ Intraoperative hypotension has been associated with postoperative myocardial injury,⁷⁵ and so intuitively avoiding hypotension may reduce the risk of PMI; however, such a strategy remain to be proven. A recent review by Ackland and Abbott,⁷⁶ however, highlights that the interaction between perioperative hypotension and postoperative organ dysfunction is potentially more complicated than commonly perceived, offering the hypothesis that hypotension may be consequence rather than cause of underlying organ injury.

Postoperative atrial fibrillation

Postoperative atrial fibrillation (POAF) is common after thoracic surgery, occurs most typically on postoperative Days 2–3,^76–81 and has a reported incidence varying from 6.4%⁷⁷ to 46%.⁷⁸ Roselli and colleagues⁷⁹ found that in a cohort of patients undergoing lung resection, of the 19% of patients who developed POAF, two-thirds of cases occurred in isolation but the remainder were associated with other complications.⁷⁹ Importantly, recent data from Ishibashi and colleagues⁸⁰ suggest that sustained POAF, which carries a higher mortality risk than self-limiting POAF, has a later median onset of nearly 5 days postoperatively, meaning POAF may be undiagnosed in many cases, with patients increasingly being discharged from hospital before Day 5.⁸⁰

Mechanisms

The mechanisms for POAF are complex and not fully understood. As summarised by Dobrev and colleagues,⁸¹ chronic fibrosis of the atrial myocardium (accelerated by comorbidities and toxin abuse) creates a ‘vulnerable substrate’ where atrial ectopic beats can trigger atrial fibrillation (AF). Re-entry circuits are then sustained by a process of perioperative remodelling of connexin ion channels and acute alterations to calcium handling.⁸¹ Identified risk factors for POAF point to the underlying mechanisms for this (Fig 4). In patients undergoing thoracic surgery, hilar, pericardial, and mediastinal lymph node dissection, pulmonary vein ligation, and vagal nerve irritation add thoracic surgery-specific increased risk,^77,82,83,84, 85, 86 explaining the increased incidence of POAF after thoracic surgery than in many other types, and the increased risk observed with increasing extent of lung resection.^83,81

Fig 4.

Schematic representation of the risk factors for and mechanisms of postoperative atrial fibrillation (POAF) after thoracic surgery. Patient risk factors, predominantly promoting atrial fibrosis result in a vulnerable atrial substrate where atrial ectopics can induce fibrillation. The threshold for fibrillation is reduced by the presence mediastinal, pericardial and myocardial inflammation increasing risk of POAF which is further exacerbated by endogenous and exogenous catecholamines and atrial stretch resulting from fluid administration. COPD, chronic obstructive pulmonary disease; IHD, ischaemic heart disease; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle; VATS, video-assisted thoracic surgery.

Perioperative RV dysfunction may also play a role. Matyal and colleagues²⁵ used trans-oesophageal echocardiography to measure the RV myocardial performance index (RV-MPI, an echocardiographic index RV function) in patients undergoing lung resection. RV-MPI was measured before and 10 min after being established on OLV. In those patients with a normal MPI at baseline, if MPI deteriorated on institution of OLV, SVT was significantly more likely (42% vs 10%, P=0.01). Similarly, Amar and colleagues⁸⁷ observed supraventricular tachycardia to be more common in patients with increased RV systolic pressure (as evidenced by increased tricuspid regurgitant jet velocity) postoperatively.

Outcomes

Once thought to be a benign complication, it is increasingly appreciated that POAF results in increased length of hospital stay,^77,88,82,80,89, 90, 91, 92 costs of admission,⁸⁵ and both early and late postoperative mortality.^93,83,80,84 Even when occurring in isolation (i.e. in the absence of other complications), POAF is associated with increased length of stay compared with propensity matched patient pairs, highlighting the clinical burden.⁷⁹ Ishibashi and colleagues⁸⁰ found in their cohort of post-lobectomy patients that 30-day postoperative mortality increased from 0.2% in patients without POAF to 6.1% with POAF; rates of ischaemic stroke were 4.1% in patients with POAF and only 0.2% in those without.⁸⁰ POAF has also been shown to be a strong independent risk factor for death 5 yr after lobectomy (hazard ratio [HR]=3.8; 95% confidence interval [CI], 1.4–9.1).⁹³

Prophylaxis against postoperative atrial fibrillation after thoracic surgery

Clinical guidelines for prevention of POAF are consistent in their message that pharmacological prophylaxis should be reserved for patients with increased risk of POAF, but recommendations for specific drug therapies vary significantly. Indeed, the only strong recommendation consistent across both the (2014) American Association for Thoracic Surgery (AATS) and the (2011) Society of Thoracic Surgeons guidelines on the prevention of POAF in patients undergoing thoracic surgery is that patients taking beta blockers should continue taking them in the perioperative period.^94,95

A more recent trial sequential analysis and meta-analysis of 45 studies by Zhao and colleagues⁹⁶ purports to demonstrate there is now enough cumulative evidence to support the routine use of pharmacological prophylaxis against POAF. Beta blockers were found to be the most effective prophylaxis, followed by angiotensin-converting enzyme (ACE) inhibitors, amiodarone, magnesium, and calcium channel blockers. However, the authors' recommendations that beta blockers should be used as ‘first line prophylaxis against POAF’⁹⁶ are challenged by the well-publicised results of the Perioperative Ischemic Evaluation (POISE) study, which revealed that the acute use of beta blockers in the perioperative period was associated with increased rates of debilitating postoperative stroke and postoperative death.⁹⁷ Without further examination in large-scale, prospective clinical trials, it is difficult therefore to advocate the use of beta-blockade as a method of POAF prophylaxis. In addition, clinical guidelines continue to highlight concern with the widespread use of amiodarone prophylaxis given its association with acute respiratory distress syndrome ARDS in a single trial.^94,95 Van Mieghem and colleagues⁹⁸ stopped their trial comparing verapamil with amiodarone as potential prophylactic agents early, after three patients in the amiodarone group, all of whom had undergone right pneumonectomy, developed ARDS. Although it should be noted that high doses of amiodarone were received by affected patients, the findings of this small trial continue to heavily influence clinical guidelines and practice.⁹⁸

The identification of patients at high risk for POAF has been identified as a priority^99,100 to allow more focused use of prophylactic agents, and efforts have been made to make models to predict POAF^{84,91,101,102} although these have yet to be adopted widely. Research in this area is further challenged by the lack of an accepted consensus definition, meaning that study endpoints vary considerably (e.g. from self-limiting AF lasting 30 s⁸⁹ to persistent AF requiring acute management¹⁰²). Further work in defining, predicting risk of, and evaluating pharmacological regimens for prevention of POAF is required before firm recommendations for prophylaxis can be made.

Cardiac function determines survival during unplanned ICU admission after thoracic surgery

The UK Association of Cardiothoracic Anaesthesia and Critical Care (ACTACC) recently published the results of its third national collaborative audit examining ‘critical care after lung resection’.^103,104 In this study, reporting on a cohort of more than 110 000 patients of whom 253 (2.3%) required unplanned admission to critical care after lung resection, the ICU diagnosis of RV dysfunction was one of only two factors independently associated with ICU mortality, conferring a 4.8 (2.1–11.4) times increased risk of ICU death (the other independent predictor was the need for both renal replacement therapy and mechanical ventilation).¹⁰⁴ In the face of respiratory failure and the need for invasive mechanical ventilation, ‘normal’ postoperative changes in RV afterload may be magnified as airway pressures increase in combination with the pulmonary vasoconstrictive effects of hypoxia and hypercapnia. In cases of ARDS, this is further compounded by extrinsic vascular compression resulting from interstitial oedema, vasoconstrictor mediator release, endothelial dysfunction, and mechanical obstruction by thromboemboli, neutrophils, and platelets.¹⁰⁵ It is intuitive therefore that after lung resection, patients with respiratory failure receiving mechanical ventilation would be at high risk of RV dysfunction.

Cardiovascular limitation may influence long-term functional outcome after lung resection

Estimation of predicted postoperative (ppo) lung function by pulmonary function testing (PFT) adjusted by a segment counting approach is firmly embedded in clinical practice and widely advocated to predict ‘surgical risk’,^19,20 postoperative complications,18, 19, 20 and the risk of postoperative dyspnoea.^17,18 Although intuitive however, the evidence that calculation of ppo-pulmonary function predicts dyspnoea or functional outcome after resection is limited. Frequently, studies cited as supporting ppoPFTs in prediction of postoperative dyspnoea are small,^106,107 failed to include appropriate ‘predictive’ statistical analysis,106, 107, 108, 109 or in fact made no assessment of functional capacity¹⁰⁶ (commonly examining the association between ppoPFTs and postoperative morbidity or mortality). Pelletier and colleagues¹¹⁰ examined the association between postoperative changes in forced expiratory volume in 1 s (FEV₁) and change in maximum work rate; although significant association was demonstrable, R² was just 30%. Similarly, Larsen and colleagues¹¹¹ observed that for the association between perioperative change in forced vital capacity and peak oxygen consumption, the R² value was just 18%.

An insight into the potential existence of cardiovascular limitation of exercise function after lung resection is provided by Nezu and colleagues,¹¹² who performed pre- and postoperative CPET in patients undergoing pneumonectomy and lobectomy. All patients exhibited a reduction in VO₂ max and maximal workload postoperatively. Both oxygen pulse and maximal heart rate were reduced postoperatively (CPET indices reflective of cardiac function), but breathing reserve was unchanged. Okada and colleagues²¹ performed invasive assessment of pre- and postoperative pulmonary haemodynamics in a cohort of patients undergoing lobectomy or pneumonectomy. While at rest postoperative PVRI and PAP were unchanged compared with the preoperative baseline, on exercise there was a marked increase in PVR and PAP, suggesting a reduction in pulmonary vascular flow reserve which is exceeded when facing the demands of exercise postoperatively. Such an observation is in keeping with findings that in many patients with chronic obstructive pulmonary disease (COPD; a common coexisting condition in the thoracic surgical population), RVEF is already significantly reduced compared with controls,¹¹³ and that many patients with COPD experience similar increases in PVR and PAP and cardiovascular limitation of exercise function even without surgery.¹¹⁴

Cardiovascular considerations in specific populations

Minimally invasive and robotic thoracic surgery

Minimally invasive thoracic surgery (MITS), incorporating VATS and robotic-assisted thoracic surgery (RATS), is seen as an evolution in surgical practice with the majority of lung cancer surgeries now performed using these techniques.¹¹⁵ Proponents suggest benefits in terms of less pain, reduced inflammation, less impairment in pulmonary function, reduced postoperative morbidity, and shorter hospital stays.¹¹⁶

In MITS, in contrast to thoracotomy, the operating table is often positioned to promote extreme lateral flexion, opening intercostal spaces and improving surgical access. This extreme position has been shown to result in significant haemodynamic compromise, with reduced mean arterial pressure and reduced cardiac index. Reduction in right atrial and pulmonary artery wedge pressure suggest these changes are from a decrease in venous return.¹¹⁷

In MITS (in particularly robotically assisted surgery), carbon dioxide (CO₂) can be insufflated through sealed ports to create a positive intrathoracic pressure (capnothorax); this encourages lung deflation and can improve operating conditions. This can, however, lead to haemodynamic instability, creating conditions analogous to tension pneumothorax. In both animal models and human studies, CO₂ insufflation has been shown to result in reduced cardiac index,118, 119, 120, 121 lower mean systemic arterial pressure,^119,120,122 higher central venous pressure,^{119,120,123,124} higher PAP,^120,123,124 and higher pulmonary artery wedge pressure.^120,123,124

It has been suggested that the reduced tissue damage, inflammatory injury, and postoperative sympathetic drive resulting from MITS might confer a reduced risk of cardiovascular complications. Although observational studies have reported that the incidence of POAF may be less in patients undergoing MITS lung resection,^77,125 observational reports are challenged by confounding factors influencing the choice of MITS or open surgical technique.⁷⁷ The recent large UK multicentre Video-Assisted Thoracoscopic or Open Lobectomy in Early-Stage Lung Cancer (VIOLET) RCT found no difference in the incidence of in-hospital cardiovascular adverse events between patient randomised to lobectomy via a VATS or thoracotomy approach in patients undergoing thoracotomy (odds ratio [OR]=1.03; 95% CI, 0.54–1.97).¹²⁶

Older patients

It is well recognised that the thoracic surgical population is growing older.¹²⁷ Ageing results in time-dependent post-maturity changes that take place at a cellular level in all organs and result in physiological changes in both the cardiovascular and respiratory systems. Age-related vascular stiffening, which commonly contributes to systemic hypertension in older people, has also been observed in the pulmonary vascular bed leading to pulmonary artery stiffening,¹²⁸ a phenomenon that might be expected to reduce pulmonary vascular flow reserve. In parallel, age-related cardiac remodelling and a downregulation of myocardial catecholamine receptors results in a reduction in contractility.^129,130 Put together, one might hypothesise the older population may cope less well in response to the haemodynamic challenges of OLV and lung resection although this has not been studied. Ageing is also associated with fat infiltration and fibrosis of the cardiac conducting pathways, increasing the risk of arrhythmia.^81,130 Age has consistently been identified as a risk factor for POAF after lung resection alongside many age-related comorbidities^93,79,80,84 Increasing age similarly confers increased risk of PMI.^59,61

Conclusions

The combination of pre-existing cardiovascular comorbidities and the unique pathophysiology resulting from thoracic surgery with OLV, provides a collation of haemodynamic insults that can impact both short- and long-term patient outcomes. The presence of postoperative RV dysfunction has been well demonstrated, although its clinical significance remains uncertain. Although there is evidence to suggest RV dysfunction may be implicated in the aetiology of cardiovascular complications (e.g. POAF and PMI) and may drive impaired functional capacity in some patients, further work is required to improve understanding, identify risk factors, and explore mechanisms. The common cardiovascular complications, POAF and PMI, are increasingly recognised to impact both short- and long-term outcomes but at present preventative strategies are limited.

Aiding adaptation to the unique perioperative physiology seen in patients undergoing thoracic surgery may be novel therapeutic avenues to prevent cardiovascular complications and improve long-term functional capacity after surgery. Plausibly, prehabilitation, augmentation of inotropy, inhibition of inflammatory injury, and manipulation of RV afterload are all potential therapeutic targets which might merit investigation in this context.

Authors' contributions

Conception of the article: BS, PMcC

Editing the final version of the manuscript: BS, PMcC

Literature searches for specific areas of the review: AG, TK, JMcE, CH, BL

Writing of the first drafts: AG, TK, JMcE, CH, BL

Specialist expertise in lung transplantation: NM

Contributed to the manuscript: NM

Declarations of interest

BS is an associate editorial board member of the British Journal of Anaesthesia. PMcC, AG, TK, JMcE, CH, and BL report no conflicts of interest. NM has received honoraria from Cytosorbents, Air Liquide and Edwards Lifesciences, none of which is relevant to the subject of this review.

Funding

The (UK) Royal College of Anaesthetists/National Institute of Academic Anaesthesia British Oxygen Company Chair of Anaesthesia research grant (to BS). The authors' original research on the cardiovascular effects of thoracic surgery has been funded by the Chief Scientist Office (Scotland), National Institute of Academic Anaesthesia, Association of Cardiothoracic Anaesthesia and Critical Care, Scottish Society of Anaesthetists and Association of Anaesthetists of Great Britain and Ireland.

Handling editor: Jonathan Hardman