New-onset atrial fibrillation after anatomic lung resection: predictive factors, treatment and follow-up in a UK thoracic centre

BACKGROUND

Atrial fibrillation remains one of the most common complications occurring after thoracic surgery, with a reported incidence of between 4 and 37% [1–3]. De novo postoperative atrial fibrillation (POAF) may significantly increase morbidity and mortality after thoracic surgery in both the short term and long term [1–4]. Furthermore, through increases in hospital stay and resource utilization, POAF may confer considerable costs to patient care [5, 6].

The pathophysiology involved in the development of POAF is multifactorial, and involves a complex interaction between ‘triggering’ stimuli and ‘sustaining’ processes acting on a myocardial substrate that may be predisposed to developing tachyarrhythmia. Non-modifiable factors, such as increasing age and male gender, may increase predisposition to AF through increasing the arrhythmogenicity of the substrate atrial myocardium [7]. In addition, modifiable factors, such as electrolyte imbalance and systemic infection, may result in triggering stimuli that increase the likelihood that re-entrant circuits may become established within the atrium [8, 9].

The extent of pulmonary resection has also been associated with POAF, with patients undergoing pneumonectomy experiencing greater rates of POAF than those undergoing lobar and sublobar resections [10]. More recently, there has likewise been much focus on reducing postoperative morbidity through increased utilization of video assisted thoracoscopic surgery (VATS) and robotic approaches to lung resection, leading to a suggestion that such techniques may also reduce the incidence of POAF.

Although POAF is usually a self-limiting arrhythmia with the majority of thoracic surgical patients returning to sinus rhythm at the time of discharge, the degree of recurrence of AF after discharge and the reason for the persistence of AF in some patients and not others remain unclear. One recent study utilizing implantable loop recorders to follow up CABG patients developing new onset AF suggests that up to 60% developed episodic recurrence of the arrhythmia in the first year after surgery [11].

The aims of this study were, therefore, to first determine the perioperative factors correlating with POAF in our UK patient population and whether modification of perioperative factors may reduce the incidence and morbidity associated with POAF after anatomical lung resection. Secondly, we aimed to evaluate the course of POAF after its onset, its persistence at follow-up and the relationship of persistent AF to perioperative factors and pharmacotherapy.

METHODS

Patient selection and recruitment

Consecutive patients undergoing anatomical lung resection for any indication at Guy's Hospital between 1 January 2015 and 31 December 2015 were considered for inclusion. Patients undergoing wedge resections, taking long-standing antiarrhythmic medications (excluding beta-blockers), or with a prior history of atrial fibrillation or other cardiac arrhythmia were excluded. The choice of an open vs VATS approach was left to the surgeon. Pulmonary vein division was performed using the Covidien Ultra Universal Endo GIA™ Tri-staple™ (Covidien Ltd.) in all cases. Postoperatively, all thoracotomy patients received paravertebral analgesia via a paravertebral catheter. VATS patients received either intercostal blocks given by the surgeon under direct visualization intraoperatively or a paravertebral catheter depending on surgeon preference. All patients received patient-controlled analgesia with morphine as the first-line agent for the first 1–2 postoperative days.

Demographic data collected included age, sex and any preoperative comorbidities. Surgical factors recorded included surgical approach (thoracotomy or VATS) and type of anatomical lung resection. Postoperative complications, medication use and length of stay outcomes were also recorded. De novo postoperative atrial fibrillation (POAF) was defined by the presence of AF on 12-lead ECG taken when there was a clinical suspicion of arrhythmia. The ongoing presence of AF or return to sinus rhythm was also assessed on 12-lead ECG prior to discharge and at 1-month follow-up.

Statistical analysis

Logistic regression was used to model POAF as a dichotomous outcome variable against pre-, intra- and postoperative predictor variables. Where the predictor variable was categorical, a multinomial logistic regression model was used. For binary predictor variables, the χ² statistic was used to determine any significant association with the occurrence of POAF. On the basis of the results of univariate analysis, a multivariable model was constructed to determine the relative importance of these univariate predictors of POAF. A P-value less than 0.05 was considered statistically significant. Data analysis was performed using StataMP version 12.1 (StataCorp, Texas).

RESULTS

Three hundred and seventy-seven patients (44.3% (167/377) male) with a mean age of 67.1 years underwent anatomical lung resection. A summary of the preoperative demographics of included patients is given in Table 1. An overall summary of results is given in Table 2.

Three hundred and fifty patients underwent lobectomy (92.8%), 13 underwent bilobectomy (3.4%) and 14 underwent pneumonectomy (3.7%). A VATS approach was used in 188 (49.9%) with the remaining 189 (50.1%) of cases completed through a standard posterolateral thoracotomy.

Postoperative atrial fibrillation occurred in 43 patients (11.4%) with a mean time to development of 3.55 ± 5.22 days. POAF patients had a significantly longer length of stay than those that did not develop AF postoperatively (10.8 ± 5.8 vs 6.78 ± 4.4; P = 0.0014).

Patients developing de novo postoperative AF were significantly older than those remaining in sinus rhythm (72.6 ± 1.19 vs 66.4 ± 0.63 years) (P = 0.0008). There were no significant differences in the prevalence of preoperative comorbidities, including hypertension, ischaemic heart disease, thyroid dysfunction, alcohol intake and beta-blocker use between POAF and non-POAF groups (Table 1).

An open rather than VATS approach (open 26/189 (13.8%); VATS 17/188 (9.0%); P = 0.150) demonstrated a tendency towards an increased incidence of postoperative AF; however, this did not reach statistical significance. When considering operation type, pneumonectomy and bilobectomy did not correlate with significantly higher rates of postoperative AF when compared with lobectomy as a base outcome (P = 0.215 and P = 0.171, respectively). No patients had documented sustained electrolyte abnormality postoperatively. Postoperative chest infection (as evidenced by pyrexia and elevation of inflammatory markers outside of the normal range) correlated significantly with the onset of de novo postoperative AF (22/43 (37.9%) vs. 58/334 (7.6%); P < 0.0001).

A multivariable model was built to include age, gender, surgical approach, degree of resection and postoperative infection. Only age (P = 0.001) and postoperative infection (P < 0.0001) were significant multivariable predictors of POAF in this patient cohort (Table 3).

A summary of the treatment modalities used for POAF patients is given in Table 4. As is our institutional protocol, initial treatment was with potassium and magnesium in the majority of cases (30/43 (69.8%)). Beta-blockers were used in 17 cases (39.5%), digoxin in 23 cases (53.5%) and amiodarone in 2 cases (4.7%). Overall, the most common combined treatment modality included electrolytes and digoxin, used in 14/43 (32.5%) cases. Three patients (16.3%) received a combination of beta-blockers and digoxin. Three (7.0%) patients in the POAF group were started on anticoagulation. Overall, 4/43 (9.3%) patients remained in rate-controlled atrial fibrillation on discharge, and one patient demonstrated paroxysmal AF postoperatively although was in sinus rhythm at the time of discharge. Three of these five patients were formally anticoagulated with warfarin. When considering the two patients that did not receive anticoagulation, there were clear reasons for omitting treatment. The first patient (with paroxysmal AF) refused anticoagulant therapy after counselling on the risks/benefits. The second was discussed with our cardiology team, and the risk of bleeding secondary to recurrent falls was felt to outweigh the benefit of anticoagulant therapy.

Three (3/43; 7%) patients were found to have AF at 4-week follow-up and were referred for further cardiology assessment. When examining these cases in more detail, two were in AF on discharge. The first of these patients was found to have thyroid goitre on clinical examination although thyroid function tests were normal. The second patient experienced a complicated postoperative course with chest sepsis, a prolonged air leak, and was discharged with a chest drain later removed in clinic. The final POAF patient was treated with digoxin and discharged in sinus rhythm, however subsequently presented with AF at 4 weeks postoperatively. This patient had undergone a more extensive resection with chest wall reconstruction for a T3 lesion. She was referred to our cardiology team for further assessment and management of her AF.

DISCUSSION

The results of this study once again demonstrate the importance of increasing age and postoperative infection in the development of POAF after thoracic surgery. However, conversely to previous studies, other risk factors, such as male sex, ischaemic heart disease, extent of resection and surgical approach, were not found to be significantly associated with POAF [10]. Our overall incidence of POAF of 11.4% is in line with previous studies, which estimate the incidence to be between 4 and 37% [1–3]. Only 4 out of 43 POAF patients (9.3%) in this study remained in AF on discharge, and 3 (7%) were found to have sustained AF at 4-week follow-up. Those patients remaining in AF displayed definable characteristics in their preoperative work-up and/or surgical procedure, suggesting a higher risk of AF in the postoperative period.

Our findings of increased risk of POAF with increasing age complement the outcomes of previous studies. In their 10-year analysis of patients undergoing thoracic surgery, Amar et al. similarly showed a correlation between increasing age and a risk of POAF that increased from 4% in patients <50 years to 25% in patients aged above 70 years [12]. The association between increasing age and atrial fibrillation in the general population is well recognized, with up to 20% of over 80-year-olds in Europe and the USA thought to be living with this arrhythmia [13, 14]. With increasing age, a number of structural changes are observed within the atrial myocardium, subsequently predisposing to the initiation and maintenance of AF re-entrant circuits. On a cellular level, apoptosis, impairment of gap junction function, increased interstitial fibrosis, altered connexin expression and localization, extracellular matrix, fibroblast and collagen deposition may all promote local areas of conduction block, re-entry and fibrillatory contraction [15–18]. On a macroscopic level, these structural changes result in altered atrial stretch and atrial dilatation, which may subsequently produce activation of the renin–angiotensin system and the release of natriuretic peptides [16, 17]. In de novo postoperative AF, ‘triggering’ factors, such as increased oxidative stress and inflammation, are likely to subsequently provide sufficient stimuli to initiate AF in the more susceptible myocardium of patients with advanced age [7].

The role of postoperative infection in POAF development is also not surprising. Patients treated for clinical infection (as evidenced by pyrexia and elevation of inflammatory markers outside of the normal range) were significantly more likely to develop POAF than those without infectious complications (37.9 vs. 7.6%; P < 0.0001). Sepsis creates a proinflammatory response, which is characterized by the release of inflammatory cytokines, such as interleukins-1, 2, 6, and 8, tumour necrosis factor-α (TNF-α), C-reactive protein (CRP), heat shock proteins, and matrix metalloproteinases. This cytokine surge subsequently promotes increased automaticity, autonomic dysfunction, structural and electrical remodelling leading to an increased risk of atrial fibrillation [19]. Furthermore, systemic sepsis precipitates an increase in oxidative stress and free radical production as a result of increased mitochondrial respiration, neutrophil activation, and the metabolism of arachidonic acid [20]. Activation of neutrophils increases nitric oxide free radical and subsequent peroxinitrite production, which may in turn alter cardiomyocyte calcium signalling and excitation–contraction coupling. Furthermore, through post-translational modifications in myofibrillar proteins, peroxynitrite may also potentiate arrhythmogenesis by impairment of myofibrillar contractile function [21].

Despite a higher rate of POAF associated with thoracotomy when compared with a VATS approach (9.0 vs 13.8%), this did not reach statistical significance. It may be expected that through avoidance of thoracotomy, a reduction in postoperative pain, improved early mobilization and earlier discharge, VATS approach would reduce the overall surgical stress associated with lung resection, thus alleviating a proportion of the inflammatory and oxidative stress associated with open surgery [22]. Indeed, VATS approach was associated with earlier hospital discharge (median length of stay 5 (2–25) days) when compared with open surgery (7 (3–41) days) in this patient cohort. However, our findings of comparable AF rates between VATS and open approaches echo work by Park and colleagues who also failed to show a positive effect of VATS in reducing POAF in a matched cohort of 244 patients undergoing pulmonary lobectomy [23]. In our experience, we believe this may, in part, relate to improvements in postoperative pain management, early mobilization and the implementation of enhanced recovery protocols in both open and VATS cases. We promote the return of all patients to the ward after surgery unless there is a specific indication for high dependency or intensive care unit admission. Early drain removal and the availability of portable digital suction systems, such as the Thopaz (Medela, IL, USA) device, have further encouraged early mobilization in all patients with strong support from intensive physiotherapy and nursing care.

Similarly, we may also have expected that the extent of resection would significantly impact on the rate of POAF, with pneumonectomy patients at greater risk than those undergoing bilobar, lobar or segmental resections due to the changes in cardiac output and right heart pressures in the early postoperative period [10]. However, this was not the case, and we observed similar rates of AF across all resection subgroups (Table 2). While we acknowledge that the relatively small number of patients undergoing pneumonectomy may limit the interpretation of this outcome, we also believe that enhanced recovery and standardized multidisciplinary management of these patients plays a vital role in averting the sequence of postoperative pain, basal collapse and infection which, in turn, may precipitate AF arrhythmogenesis.

This study also highlights the variability of treatment modalities available for POAF. Our protocol is to initially treat with electrolyte replacement, including intravenous magnesium and potassium where necessary, and we do not routinely use prophylactic anti-arrhythmic medications in the pre- or postoperative period. As a second line, digoxin (53.5%) and beta-blockers (39.5%) were sufficient to permit rate control and/or cardioversion without the need for amiodarone in all but two cases [24]. Overall 4/43 (9.3%) patients remained in rate-controlled atrial fibrillation on discharge, and one patient had paroxysmal AF although was in sinus rhythm at the time of discharge. Three were formally anticoagulated with warfarin, considered on a case-by-case basis [25]. Our current practice is to anticoagulate where consent is obtained and there is no absolute contraindication.

At present, there is little evidence to suggest which POAF patients will fail to cardiovert back to sinus rhythm despite pharmacotherapy, and how frequently AF recurs after discharge. In our cohort, persistence of AF at 4-week follow-up was found to be present in three patients. Each of these cases had potential risk factors for non-resolution of AF, including one case of thyroid goitre, one prolonged chest sepsis and one extensive chest wall resection with reconstruction. These results suggest that post-discharge persistence of POAF after anatomical lung resection is uncommon, and is much less prevalent than documented in reports after cardiac surgery, where AF recurrence rates may be as high as 60% [11].

A number of perioperative factors may contribute to the persistence of POAF, and defining patients at high risk remains difficult in the pre- and early perioperative period. Of note, one patient in this cohort was discharged in SR, but developed recurrence of AF at 4-week follow up. This case highlights the importance of rhythm assessment in the clinic with follow-up ECG, rigorous clinical review and consideration of anticoagulation where paroxysms of AF persist. Larger studies, ideally employing continuous postoperative rhythm monitoring, are now required to examine in more detail both the incidence and significant risk factors associated with POAF persistence at follow-up. In this way, we may identify these patients during their inpatient stay, prompting early anticoagulation and cardiology review.

Strengths and limitations

Although this study represents a relatively large and homogeneous patient population, our results should be considered in the context of their limitations. First, as our patients return to the ward on the day of surgery, it is not our routine practice to use telemetry to record cardiac rhythm in postoperative thoracic patients in our unit. As such, we may underestimate the true incidence of POAF when defining AF according to Heart Rhythm Society (HRS) guidelines [26]. Similarly, postoperative AF assessment at follow-up was by means of clinical interview and examination. ECG was performed where clinical suspicion existed to confirm or refute the presence of AF; however, it is possible that cases of paroxysmal or subclinical AF may be under-reported. Second, given the retrospective nature of this study, other potential confounders that may impact on the rate of POAF, such as surgical time, blood transfusion and extent of mediastinal lymph node dissection, the use of perioperative beta-blockers and/or postoperative inotropic support, were not included in this analysis. Finally, when considering the effect of volume of resection on POAF, small numbers of bilobectomy and pneumonectomy cases relative to the overall sample size may result in insufficient power for this study to demonstrate any effect of these factors on the development of POAF.

CONCLUSIONS

Postoperative AF remains a common complication after anatomical lung resection resulting in increased postoperative morbidity and hospital stay. Despite advances in minimally invasive surgery, increasing age and postoperative infection remain the key risk factors for the development of POAF. Although age is non-modifiable, stringent monitoring and early treatment of postoperative infection may help to reduce POAF and its associated morbidity, particularly in older patients. In our experience, enhanced recovery, early involvement of physiotherapy and pain management teams are fundamental in averting the sequence of postoperative pain, basal collapse and ultimately infection that increases the risk of POAF.

AF persistence or recurrence after discharge remained infrequent in this patient cohort, suggesting that the majority of cases are indeed self-limiting. However, follow-up rhythm assessment is crucial to identify persistent AF and clinicians should be vigilant for recurrence of AF at routine follow-up. Future prospective studies employing continuous home rhythm monitoring may help better identify the recurrence rate of subclinical AF in thoracic surgical patients, leading to improved selection of those likely to require long-term antiarrhythmic therapy, anticoagulation and ultimately a reduction in AF associated morbidity.

Conflict of interest: none declared.

REFERENCES