Preoperative serum lactate dehydrogenase level as a predictor of major complications following thoracoscopic lobectomy: a propensity-adjusted analysis

Abstract

Open in new tabDownload slide

OBJECTIVES

Despite the positive effects of a thoracoscopic approach on improving postoperative outcomes, the risk of major complications following thoracoscopic lobectomy is not negligible. We sought to assess the usefulness of the preoperative determination of serum biomarkers to refine risk stratification in this patient population.

METHODS

From 2009 to 2017, 626 patients (285 women, 341 men; median age: 67 years) underwent thoracoscopic lobectomy or anatomical segmentectomy for confirmed or suspected early-stage lung cancer or metastasis at our institution. Preoperative serum biomarkers, including albumin, C-reactive protein, haemoglobin and lactate dehydrogenase (LDH), were examined as predictors of postoperative cardiopulmonary complications using logistic regression analyses followed by causal inference.

RESULTS

The 90-day mortality, cardiopulmonary complication and overall morbidity rates were 1.0%, 13.1% and 18.1%, respectively. Although serum albumin, C-reactive protein and haemoglobin were not associated with cardiopulmonary complications in regression analyses, preoperative serum LDH level emerged as an independent morbidity predictor (odds ratio 1.008, 95% confidence interval 1.002–1.013; P = 0.006). The causal inference using the covariate balancing generalized propensity score methodology demonstrated similar results and an approximately positive linear relationship between the odds of cardiopulmonary complications and preoperative serum LDH level. For every 100 U/l increase in preoperative serum LDH, a 2-fold increase in the odds of cardiopulmonary complications was observed.

CONCLUSIONS

Our results suggest that the preoperative serum LDH level is an independent predictor of 90-day cardiopulmonary complications following thoracoscopic lobectomy or segmentectomy, even in properly selected patients. Therefore, we recommend incorporating early serum LDH measurements as a readily available method into the risk assessment process prior to major lung resection.

INTRODUCTION

Accurate risk stratification is crucial for the identification of patients who are amenable to major lung resection and pertains to the most important responsibilities of surgeons and physicians. In this decade, thoracoscopic lobectomy has been increasingly adopted and is the current preferred surgical approach for the treatment of early-stage non-small-cell lung cancer (NSCLC) [1]. A body of evidence has demonstrated that, relative to thoracotomy, thoracoscopic lobectomy is associated with reduced postoperative mortality and morbidity, particularly in high-risk patients [2, 3]. Despite this positive effect of a thoracoscopic approach, the rate of cardiopulmonary complications as major adverse events following thoracoscopic lobectomy is not negligible [3, 4]. Moreover, cardiopulmonary complications are often the primary causes of postoperative death [4, 5]. Accordingly, interest is growing in refining the current risk assessment algorithms, which are predominantly based on clinical studies published over the last decade and are applicable only to major lung resections via thoracotomy [6].

As a readily available method, the preoperative determination of certain serum biomarkers has been increasingly used as an aid in predicting postoperative adverse outcomes [7]. Regarding lung resection, our thorough literature review demonstrated that albumin, C-reactive protein (CRP), fibrinogen, B-type natriuretic peptide, haemoglobin and lactate dehydrogenase (LDH) were capable of predicting postoperative morbidity after major lung resection (Supplementary Data S1). However, existing data are commonly characterized by various limitations, particularly the small sample sizes, the wide variety of surgical approaches and the considerable number of patients undergoing pneumonectomy included in the study cohort. Notably, previous studies have only explored the associations of certain serum biomarkers with postoperative adverse outcomes, while the causal effects of these biomarkers remain obscure. More importantly, no published study has specifically focused on patients undergoing thoracoscopic lobectomy. Therefore, we reviewed our institutional database and assessed the utility of preoperative serum biomarkers in predicting major complications following thoracoscopic lobectomy or segmentectomy.

PATIENTS AND METHODS

We performed a retrospective study on thoracoscopic major lung resections using our institutional database, which included retrospective data from 2009 to 2014 and prospective data thereafter. From January 2009 through 2017, a total of 665 patients underwent thoracoscopic lobectomy or segmentectomy at our institution. Of those patients, 626 patients with confirmed or suspected early-stage NSCLC or pulmonary metastasis were included in the present study. Thirty-nine patients with infectious diseases (e.g. bronchiectasis) were excluded because they were exposed to a much higher comparative risk of postoperative complications. The preoperative serum biomarkers were recorded after reviewing patient charts. At our institution, the CRP, LDH and haemoglobin levels were routinely obtained prior to surgery, whereas the preoperative albumin level was routinely measured starting in October 2015. The measurements of B-type natriuretic peptide and fibrinogen were, and are not currently, part of routine preoperative laboratory testing. Instead, these are only performed for patients with heart failure and suspected coagulation disturbance. This study was approved by the local Institutional Review Board, and specific patient consent was waived.

All patients with confirmed or suspected NSCLC underwent computed tomography, positron emission tomography and brain magnetic resonance imaging for clinical staging. Patients with suspected mediastinal nodal metastases underwent endobronchial ultrasound-guided fine-needle aspiration or cervical mediastinoscopy. Our multidisciplinary lung oncology team preoperatively reviewed all cases. Lung cancer staging was performed according to the American Joint Committee on Cancer 7th edition manual in patients undergoing lung resection up to 2016 and the American Joint Committee on Cancer 8th edition starting in 2017. The pathological stage was reported based on the final histopathological findings following lung resection and systematic mediastinal lymph node dissection. All patients underwent a physiological evaluation prior to surgery. The predicted postoperative values of forced expiratory volume in the first second expressed as the % predicted (ppoFEV₁%), and of diffusing capacity of the lung expressed as the % predicted (ppoDLCO%) were calculated using the functional segment technique [1]. Patients who demonstrated impaired pulmonary function (ppoFEV₁% or ppoDLCO% <40) during preoperative lung function testing performed a stair climb test or cardiopulmonary exercise test (cycle ergometry) for additional risk stratification prior to lung resection. When performance in the stair climb test was not satisfactory, cycle ergometry was used to determine eligibility. A maximum oxygen consumption >10 ml/kg/min or 35% predicted was considered sufficient to warrant major lung resection.

The performance status was considered marginal or poor when the Eastern Cooperative Oncology Group (ECOG) scale score was ≥2. All patients were assigned a score based on the modern version of the American Society of Anesthesiologists physical status (ASA-PS) scale by the responsible anaesthesiologists prior to surgery [8].

Thoracoscopic lobectomy and anatomical segmentectomy were performed using a 3-port approach, including a 3-cm anterolateral access incision in the 4th intercostal space without rib spreading and with visualization only through the monitor. Lobar or segmental vessels and the bronchus were individually divided. The hilar and mediastinal lymph nodes were dissected.

At our institution, patients undergoing major lung resection were followed up for 6 weeks and 3 months after the discharge in the outpatient department. Postoperative mortality and morbidity were defined as death and adverse events within 90 days of surgery. Postoperative complications were defined according to the Society of Thoracic Surgeons and The European Society of Thoracic Surgeons (STS/ESTS) joint standardization of variable definitions and terminology [9]. Specifically, pulmonary complications included pneumonia, atelectasis requiring bronchoscopy, adult respiratory distress syndrome, initial ventilator support for >48 h, unplanned reintubation or tracheotomy. Cardiovascular complications were defined as acute myocardial infarction, pulmonary embolism and atrial or ventricular arrhythmia requiring intervention. Other complications included bronchopleural fistula, empyema, wound infection, other postoperative infections, cerebrovascular events, delirium, renal failure, postoperative bleeding, chylothorax, recurrent laryngeal nerve injury and other relevant events. Overall morbidity was defined as the occurrence of any of these complications, including mortality, but except for technical complications, such as postoperative bleeding, chylothorax or recurrent laryngeal nerve injury. In addition, the length of hospital stay was not used to assess the postoperative outcomes, as it is dependent on many factors other than postoperative adverse events.

Categorical variables were expressed as percentages and were evaluated using the Fisher’s exact test. Continuous data were reported as the median and interquartile range (IQR 25th–75th percentile) and were compared using the Mann–Whitney U-test. Logistic regression analyses were performed to assess the relationships among the serum biomarkers and cardiopulmonary complications. We used 2-sample comparisons of proportions to calculate the sample size for logistic regression. We assumed that postoperative complications occurred in 10% and 20% of patients with low and high risks, respectively [odds ratio (OR) = 2.25]. Accordingly, we estimated that 550 patients were required to reach a statistical power of 90% at a 5% significance level. To estimate the causal effect of the preoperative serum LDH level, the covariate balancing generalized propensity score (CBPS) methodology of Fong et al. [10] was applied to account for the well-established predictors of major complications as covariates. In this method, inverse probability weighting using propensity scores was used to balance the covariate distribution across the treatment value (LDH level). As the most commonly used approach, a parametric propensity score model was assumed and included age, male gender, ppoFEV₁%, ppoDLCO%, smoking history and ASA-PS (>2 vs ≤2). Missing data were imputed using the random forest method of Ishwaran et al. [11]. Subsequently, the balanced data were used to estimate the effect size of the serum LDH level on postoperative cardiopulmonary complications in a logistic regression model with weights determined by covariate balancing propensity scores. To check the robustness of the above parametric approach, the non-parametric extension of the CBPS methodology was then applied as a more flexible approach. Although propensity score matching with or without replacement is currently a popular method for causal inference, this method has been confined to binary treatment variables [10]. In contrast, CBPS is one of few currently available approaches for causal effect analyses of continuous treatment variables and is believed to be an appropriate method for our study [10]. Statistical significance was declared as P-value <0.05. CBPS and data imputation were conducted using the statistical computing environment R, version 3.3.3, along with the R packages ‘CBPS’ 0.18 and ‘randomForestSRC’ 2.5.1 [12–14]. The other statistical analyses were performed using SPSS, version 22.0 for Windows (SPSS, Chicago, IL, USA).

RESULTS

A total of 285 women and 341 men with a median age of 67 years (IQR 59–74 years) were included. The surgical procedures included 562 lobectomies (89.8%) and 64 anatomical segmentectomies (10.2%). In 25 patients (4.0%), the thoracoscopic procedures had to be converted to thoracotomy. As the present study aimed to assess the risk for major complications in thoracoscopic lobectomy or segmentectomy as an initially intended treatment, conversed cases were included in the analyses. Among the entire cohort, the final pathological analyses demonstrated confirmed NSCLC in 562 (89.8%) patients, benign disease in 34 (5.4%) patients and pulmonary metastasis in 30 (4.8%) patients. Patient demographics and clinical characteristics are shown in Table 1. Impaired pulmonary function (ppoFEV₁% or ppoDLCO% <40) was found in 132 patients (21.1%). Of those patients, 58 underwent cycle ergometry, with measurements of the maximum oxygen consumption ranging from 10.2 to 27.0 ml/kg/min (38.0–162.0%).

Six (1.0%) postoperative deaths occurred in the entire patient cohort. Four patients died from postoperative pneumonia. One patient developed a bronchopleural fistula followed by sepsis and multiple organ failure and died on postoperative day 12. Another patient developed pneumonia and atrial arrhythmia on postoperative day 5; this was followed by a bronchopleural fistula, and the patient died on postoperative day 79. Postoperative cardiopulmonary complications and overall morbidity occurred in 83 (13.1%) and 113 (18.1%) patients, respectively. The incidences of single postoperative complications are listed in Table 2.

Table 3 provides the values of the preoperative serum biomarkers, the number of patients measured and their relationships with cardiopulmonary complications. Based on univariable logistic regression analyses, the cardiopulmonary complication rate was not related to albumin, CRP or haemoglobin. In contrast, the LDH level was significantly associated with the risk of cardiopulmonary complications. In the multivariable regression analysis adjusted for age, male gender, pulmonary function, smoking history and ASA-PS, LDH emerged as the single predictor of cardiopulmonary complications [OR 1.008, 95% confidence interval (CI) 1.002–1.013; P = 0.006, Table 4].

To estimate the causal effect of the preoperative serum LDH level on postoperative cardiopulmonary complications, the distribution of covariates, including age, male gender, ppoFEV₁%, ppoDLCO%, smoking history and ASA-PS, were balanced using the CBPS methodology. As depicted in Table 5, this approach achieved an approximately equal covariate distribution across the preoperative serum LDH level as the treatment variable. Our analysis of balanced data using a generalized linear model revealed a similar effect size for the preoperative serum LDH level (OR 1.007, 95% CI 1.002–1.012; P = 0.011). As illustrated in Fig. 1, an approximately linear increase occurred in the probability of cardiopulmonary complications relative to the elevation in the preoperative serum LDH level. For every 100 U/l increase in the preoperative serum LDH level, a 2-fold increase in the odds of cardiopulmonary complications following thoracoscopic lobectomy occurred. To assess the robustness of the above results, we additionally applied the non-parametric approach of the CBPS methodology, which revealed very similar results (OR 1.008, 95% CI 1.004–1.012; P < 0.001).

DISCUSSION

Despite the recent advances in perioperative care and the widespread adoption of minimally invasive approach, the rate of major complications following anatomic lung resection is still not negligible, even in a properly selected patient cohort, as shown in the present study. The postoperative outcomes in our patient cohort were in line with those of previously published data derived from a large clinical thoracic surgery database and highlight again the need to refine the current risk assessment algorithms, specifically for patients considered for thoracoscopic lobectomy or segmentectomy [3]. In this context, the preoperative determination of serum biomarkers predictive of postoperative adverse events may allow more elaborate risk stratification and better informed decision-making. To date, limited data are available regarding serum biomarkers as predictors of postoperative complications following major lung resection (see the Supplementary File). Notably, no recent study has specifically focused on patients undergoing thoracoscopic lobectomy, which is currently the preferred surgical approach for the treatment of early-stage NSCLC [1].

The major finding of the present study was that the preoperative serum LDH level was an independent predictor of the cardiopulmonary complications following thoracoscopic lobectomy or segmentectomy after accounting for well-established risk factors. To assess the robustness of this finding, we performed causal inference using the CBPS methodology for both the parametric and non-parametric approaches, which revealed a similar effect size for serum LDH. For every 100 U/l increase in the preoperative serum LDH level, a 2-fold increase in the odds of cardiopulmonary complications occurred. To date, few studies have investigated the predictive value of the preoperative serum LDH level for adverse events following lung resection [15–17]. While a preoperative serum LDH level greater than 230 U/l was found to be predictive of pulmonary complications in a series of 89 lobectomies via thoracotomy, there has been report on a more than 4-fold increase in the odds of pulmonary complications following open lung surgeries with various extents of resection, when the preoperative serum LDH level was greater than 320 U/l [15, 16]. In addition, Mitsudomi et al. [17] studied 62 patients who underwent pneumonectomy and found significant associations between a higher preoperative serum LDH level (>178 U/l) and postoperative mortality and complications necessitating rethoractomy. It is important to note that, in addition to the inconsistent findings and various limitations, these previously published studies are characterized by dichotomizing the serum LDH level as a continuous treatment variable. This approach may result in a misleading impression that the odds of cardiopulmonary complications following lung resection are sharply increased when the preoperative serum LDH level exceeds a certain cut-off value. However, our analyses, including causal inference, clearly demonstrated an approximately positive linear relationship between the odds of cardiopulmonary complications and the preoperative serum LDH level. In addition, the present study is the first to embrace the predictive ability of the preoperative LDH level for postoperative adverse outcomes in patients undergoing major lung resection via thoracoscopic approach. According to our results, we believe that the determination of the serum LDH level can aid in more accurate risk stratification prior to major lung resection and better assignation of resources for postoperative care. Patients with higher preoperative serum LDH levels may require closer postoperative monitoring and more intensive pulmonary care. Therefore, we recommend incorporating an early measurement of the serum LDH level into the preoperative risk assessment process.

LDH is a cytoplasmic enzyme that catalyses the final step in anaerobic glycolysis through the reversible oxidation of pyruvate to lactate [18]. The appearance of LDH in the serum usually indicates cell damage or cell death [18]. As a non-specific biomarker, the total serum LDH level is elevated not only in acute diseases, such as myocardial infarction and severe infection, but also in a host of chronic disorders, including renal and liver diseases, disseminated malignancy and certain haematological disorders [18, 19]. Because we were not able to examine the activity pattern of the serum LDH isoenzymes preoperatively, determining the cause of the elevated preoperative serum LDH level in our patients was fairly difficult. First, moderate or severe renal and liver diseases were rarely present among our patients, who underwent a thorough preoperative work-up and careful selection for major lung resection. Second, although an elevated serum LDH level indicates interstitial fibrosis of the lung following alveolar damage, only 2 patients in our study presented with this disease [15, 20]. Finally, since serum LDH elevation was observed primarily in patients with metastatic lung cancer, early-stage lung cancer was very unlikely to be the source of the increased preoperative LDH level in our study [21, 22]. Of interest, serum LDH has been proposed as an indicator of early organ damage or inflammation, which is not detectable in a routine preoperative work-up [15, 16]. Based on these findings, the determination of serum LDH isoenzymes or the LDH level in bronchoalveolar lavage fluid may improve our understanding of the pathophysiological link between an elevated serum LDH level and an increased risk of morbidity following major lung resection, which is crucial for the arrangement of appropriate risk-reducing interventions prior to surgery, e.g. increased cell viability using anti-inflammatory agents [18, 23, 24].

Interestingly, our analyses failed to associate the preoperative serum albumin, CRP and haemoglobin levels with cardiopulmonary complications following thoracoscopic lobectomy or segmentectomy. We believe that these results are largely attributed to the thorough risk assessment and careful patient selection performed before surgery. Indeed, neither a very low serum albumin level nor mild or severe anaemia, indicating a poor physiological status, was found in our patient cohort [25]. To date, a significant association between the preoperative serum albumin level and pulmonary complications following lung resection was found in only one study involving 103 patients with serum albumin levels ranging from 2.4 to 5.1 g/dl [26]. Furthermore, in the study by Bernard et al. [25], who reported the predictive capability of preoperative haemoglobin for mortality and major complications after pneumonectomy, the haemoglobin level ranged widely from 7.4 to 17.5 g/dl. The same applies to CRP as an acute-phase response protein, which is elevated in response to acute and chronic inflammation. In a study involving over 1400 patients undergoing anatomic lung resections, Lopez-Pastorini et al. [5] found that a preoperative CRP level above 40 mg/l was associated with higher postoperative mortality and morbidity, while patients with a CRP level between 3 and 40 mg/l had no significant increase in morbidity and mortality compared to those with values lower than 3 mg/l. Notably, a CRP level greater than 40 mg/l was present in 16.5% of the patients in their study, which was greater than the number of patients with this level in our study (i.e. 3.7%).

Limitations

We acknowledge potential limitations of the present study. First, inherent bias was present due to the use of retrospective data. Only thoracoscopic major lung resections performed starting in 2015 were prospectively evaluated at our institution. Additionally, our results are derived from data from a single institution, which may not fully reflect the clinical scenarios elsewhere. However, our postoperative outcomes are in line with those of previously published large-scale studies on thoracoscopic lobectomy [3]. Another cause for concern is the selection bias in the present study. Although causal inference using the CBPS methodology mitigated the selection bias to some extent by balancing known confounders, some unknown confounding variables may have influenced our results.

CONCLUSION

In conclusion, the major complications following thoracoscopic lobectomy and segmentectomy are not negligible despite the thorough preoperative physiological evaluation and the positive effect of a thoracoscopic approach on improving postoperative outcomes. The preoperative serum LDH level can predict major complications following thoracoscopic lobectomy after accounting for well-established risk factors. Therefore, we suggest incorporating an early determination of the preoperative serum LDH level as a readily available risk assessment method prior to major lung resection. Closer monitoring and more intensive postoperative care in patients with an elevated preoperative serum LDH level may further reduce adverse events following thoracoscopic lobectomy or segmentectomy. Future research is warranted to generate greater clarity regarding the pathophysiological link between an elevated serum LDH level and increased morbidity following major lung resection.

ACKNOWLEDGEMENTS

The authors acknowledge the substantial contributions from all surgical and anaesthesia team members. The authors also thank Sebastian Ciupa for his kind support during data collection.

Conflict of interest: none declared.

REFERENCES