Risk of postoperative complications in chronic obstructive lung diseases patients considered fit for lung cancer surgery: beyond oxygen consumption

Abstract

OBJECTIVES

Patients with poor lung function usually undergo cardiopulmonary exercise testing (CPET) and those with a predicted postoperative maximal oxygen consumption (VO₂ max) of >10 ml/kg/min undergo lung resection surgery and still some complications are observed. We aimed to determine other parameters beyond VO₂ able to predict postoperative complications in patients undergoing lung resection surgery.

METHODS

This is an observational study with longitudinal follow-up. Patients with forced expiratory volume in 1 second (FEV₁) or diffusing capacity for carbon monoxide of <40% underwent CPET and those with VO₂ max of >10 ml/kg/min were considered fit for surgery. Patients were followed up prospectively for 12 months and postoperative complications and survival were recorded. Physiological parameters obtained during CPET and pulmonary function tests were analysed.

RESULTS

Eighty-three chronic obstructive pulmonary disease (COPD) patients were evaluated for surgery between 2010 and 2015. Twenty-four patients were considered unfit for surgery and received an alternative therapy. Fifty-five patients had a VO₂ max of >10 ml/kg/min and underwent lung surgery. Among them, 4% died and 41% developed complications postoperatively. Baseline minute ventilation to carbon dioxide output (VE/VCO₂) slope was significantly higher among those who developed postoperative complications or died (P = 0.047). Furthermore, VE/VCO₂ slope of >35 (at maximal exercise) was the single parameter most strongly associated with the probability of mortality and postoperative complications (hazard ratio 5.14) with a survival probability of 40% after 1 year of follow-up. In a multivariable model, VO₂, VE/VCO₂ slope of >35 and work load were independently associated with the probability of having an event.

CONCLUSIONS

VO₂ is not the unique parameter to consider when CPET is performed to evaluate the postoperative risk of lung cancer surgery in COPD patients. The signs of ventilatory inefficiency such as VE/VCO₂ slope predict complications better than VO₂ does.

INTRODUCTION

Chronic obstructive pulmonary disease (COPD) patients frequently suffer from lung cancer, with resection surgery being the best therapeutic option. However, surgery is a challenge in these patients because they are at risk of postoperative complications and mortality [1] and a precise selection of fit for surgery patients is required. These postoperative complications include pneumonia, acute and chronic respiratory failure, mechanical ventilation of more than 48 h, pulmonary embolism, acute pulmonary oedema, acute myocardial infarction, lobar atelectasis requiring bronchoscopy as well as arrhythmias [2]. The preoperative physiological assessment of patients undergoing lung resection surgery should begin with cardiovascular evaluation and lung function assessment that includes spirometry to measure the FEV₁ and the diffusing capacity for carbon monoxide (DLCO). Predicted postoperative (PPO) lung function should be calculated [3]. According to the current recommendations, if the % PPO FEV₁ and % PPO DLCO values are both >60%, the patient is considered at low risk of anatomical lung resection [1], and no further tests are indicated. In general practice, it has been proposed that the shuttle walking test could be useful since it identifies patients with higher oxygen consumption (VO₂) [4]. A cardiopulmonary exercise testing (CPET) is indicated when the PPO FEV₁ and/or PPO DLCO are <30% [3]. A maximal oxygen consumption (VO₂ max) of <10 ml/kg/min or 35% predicted indicates a high risk of mortality and long-term disability for major anatomical resection.

VO₂ max is considered the most important parameter for the prediction of postoperative cardiopulmonary complications as previously described [5]. Smith et al. [2] showed that patients with VO₂ max of less than 15 ml/kg/min had more frequent postoperative complications compared with patients with VO₂ max of greater than 20 ml/kg/min assessed preoperatively by CPET and similar results were also described by Bechard and Wetstein [6]. Brunelli et al. [7] reported that VO₂ max is a reliable predictor of postoperative pulmonary complications and that VO₂ max of <12 ml/kg/min was related to mortality. Additionally, Win et al. [8] found that maximum VO₂ percent of predicted was significantly related to postoperative cardiopulmonary complications and that complications were more frequently observed among those with VO₂ max of <15 ml/kg/min.

Patients with VO₂ max of <10 ml/kg/min are at high risk of postoperative mortality and non-surgical modalities of treatment should be offered for them [3]. However, VO₂ max in the range of 10–20 ml/kg/min is still associated with moderate risk for postoperative complications [3] with higher incidence in COPD patients [1]. The aim of this work is to evaluate other parameters able to predict postoperative complications and mortality during CPET beyond VO₂ max.

METHODS

Study design and ethics

This is a observational cross-sectional study with longitudinal follow-up that enrolled patients undergoing preoperative evaluation before surgical lung resection for lung cancer treatment according to the current guidelines [3] between 2010 and 2015 in a tertiary hospital. The study protocol was approved by the Ethics Committee of the participating hospital and the written informed consent was obtained from all subjects.

Study population

The study enrolled 83 patients of whom 79 patients were analysed. COPD was diagnosed according to GOLD criteria [9] based on pulmonary function tests. All patients were discussed in a multidisciplinary lung cancer expert board that includes chest physicians, oncologists, radiologists and thoracic surgeons. All patients with FEV₁ or carbon monoxide diffusion capacity (DLCO) predicted postoperative <40% underwent CPET according to the locally accepted guidelines [5]. Those with no significant comorbidity and VO₂ max of >10 ml/kg/min were considered as fit and underwent surgical treatment. Thoracotomy was the commonly used surgical approach for management. All the patients underwent postoperative respiratory rehabilitation according to the institutional protocol, including early mobilization and incentive spirometry with pain control and inhalation of bronchodilators and/or mucolytics. Postoperative cardiopulmonary complications and mortality were recorded during the 12 months after surgery (Fig. 1). We analysed the postoperative pulmonary complications—including bronchopleural fistula defined as persistent chest tube leak of >7 days, empyema, atelectasis, pneumonia, acute respiratory failure requiring >48 h ventilation, adult respiratory distress syndrome, haemothorax, chylothorax, subcutaneous emphysema, pulmonary oedema, pulmonary embolism, lobar torsion, cardiac hernia, oesophago-pleural fistula and new requirement for chronic oxygen therapy—and cardiac complications namely cardiac arrhythmias, acute myocardial infarction and acute coronary syndrome. Moreover, local tumour recurrence during the 12 months following the surgery was recorded.

Characterization of participants

All patients underwent clinical history (including age, gender, associated cardiac comorbidities, smoking history) and pulmonary function testing [including forced spirometry, DLCO and its correction to alveolar volume (KCO) and static lung volumes] that was performed according to the international guidelines.

Cardiopulmonary exercise testing

All patients underwent CPET using an electromagnetically braked cycle ergometer (Ergocard, Medi-soft S.A., Belgium) according to the international guidelines [10]. Reference values were taken from Jones et al. [11]. Serial blood gases analyses were taken every 2 min throughout the test for lactate measurement. All the parameters of the test were recorded at baseline, anaerobic threshold (AT) and maximal work load.

Statistical analysis

All the data were expressed as mean ± standard deviation unless otherwise stated. Categorized variables were expressed as numbers and percentages. Comparisons between different groups were performed using the χ² or unpaired t-test as appropriate. Univariable and multivariable cox regression hazard analysis was applied to study the survival dependency of different CPET parameters. Stepwise selection was used for multivariable model building. The choice of covariables was verified by hands-on modelling, where terms are added or removed in a logical order based on heavily clinically related covariables in combination with a review of the literature rather than solely according to the statistical significance to identify predictors for multivariable analyses. Multiple multivariable models were tried and the one showing best overall goodness-of-fit was chosen. Both the hazard ratio and 95% confidence interval (CI) were shown. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were used to identify the sensitivity and specificity for selected CPET parameters in relation to associated events. Kaplan–Meier survival analysis was plotted to identify the survival probability. A two-tailed P-value <0.05 was considered statistically significant. MedCalc® (version 9.2.1.0, Acacialaan 22, B-8400 Ostend, Belgium) and SPSS package (version 18; Chicago, USA) were used for all analyses.

RESULTS

Study population

Twenty-four COPD patients were considered unfit for surgery and received an alternative therapy, whereas 55 patients had a predictive postoperative VO₂ max of >10 ml/kg/min and underwent a lung surgery (Fig. 1). Seventy-two percent of the patients met the GOLD criteria of moderate to very severe airflow obstruction. There was no statistical difference between those fit and unfit for surgery, regarding the basal characteristics and pulmonary function test (P > 0.05) except for KCO% (P = 0.026; Supplementary Table S1). In 4 patients who underwent a surgery, the tumour was considered irresectable intraoperatively and no resection was performed. Fifty-one patients had a resection surgery. Eight patients (14.5%) underwent a pneumonectomy, 31 patients (56.4%) a lobectomy and 12 patients (21.8%) either a segmentectomy or atypical resection of pulmonary nodule. Among them, 4% died and 41% developed complications (Fig. 1). The postoperative complications were lung atelectasis (4 patients; 7.8%), atrial fibrillation (2 patients; 4%), pulmonary oedema (1 patient; 1.96%), postoperative pneumonia (2 patients; 4%), haemothorax (1 patient; 1.96%), subcutaneous emphysema (3 patients; 5.9%), acute respiratory failure (2 patients; 4%) and 17.6% (9 patients) required oxygen therapy postoperatively. During the follow-up of the studied population, we reported no recurrence of the local tumour. The clinical characteristics and functional parameters of the patients considered fit for surgery according to the development of postoperative complications or mortality are summarized in Table 1. Moreover, the clinical characteristics and functional parameters of those developed postoperative complications versus postoperative mortality are given in Supplementary Table S2.

Cardiopulmonary exercise testing and pulmonary function test

The different CPET parameters of the studied cohort, both at baseline and at maximum exercise, are provided in Supplementary Table S3. Patients considered fit for surgery had higher heart rate and systolic blood pressure at maximal exercise and achieved higher work load, higher minute ventilation, higher oxygen uptake and carbon dioxide (CO₂) output at maximum exercise as well as higher lactic acid than unfit patients. At the point of AT, there was no significant difference in the studied parameters, other than VO₂/kg (10.79 ± 2.7 vs 8.96 ± 1.7 ml/kg/min, P = 0.003, Supplementary Table S3).

Moreover, there was no statistical significant difference when comparing the different CPET parameters at maximal exercise or AT among those who underwent surgical lung resection, i.e. 51 patients (Fig. 1) regarding the presence or absence of overall events including postoperative complications and mortality (Table 2). However, patients who developed complications or died postoperatively had higher VE/VCO₂ at baseline (51.3 ± 8.4 vs 46.4 ± 8.6, P = 0.047; Table 2) and lower diffusing capacity (P = 0.031 and 0.026 for DLCO% and KCO%, respectively; Table 1). Furthermore, patients who died postoperatively had statistically significant higher systolic blood pressure at maximal exercise (P < 0.001; Supplementary Table S4), but no significant differences regarding the other CPET parameters or pulmonary function (Supplementary Tables S3 and S4).

Regression models

Cox proportional hazards regression analysis showed that baseline VE/VCO₂ slope was the only CPET parameter associated with postoperative complications, with a hazard ratio of 2.6 (P = 0.008; Table 3). In addition, percentage of predicted minute ventilation (VE%), VCO₂ (l/min), percentage of VO₂ max (VO₂/kg) and respiratory quotient (RQ) were the best single predictors of overall events including both postoperative complications and mortality (P < 0.05; Table 3). Thirty-seven patients (72.2%) of the patients considered fit for surgery had VE/VCO₂ slope at maximal exercise of >35. VE/VCO₂ slope at maximal exercise of >35 had the best sensitivity and negative predictive value of (82.6 and 78%, respectively) in predicting postoperative complications analysed by ROC (AUC = 0.634, CI 95% = 0.487–0.764, P = 0.092). VE/VCO₂ slope of >35 was the parameter most strongly associated with the probability of mortality or postoperative complications in patients with VO₂/kg higher than 10 ml/kg/min (hazard ratio 5.14), and was also associated with a survival probability of 30% after 1-year follow-up (Fig. 2). Furthermore, VE/VCO₂ slope of >35 was still significant in the multivariable analysis (P = 0.017; Table 4); similarly, work load (watts) and VO₂ (l/min) at maximal exercise have a statistical significance in the multivariable analysis. Moreover, VO₂ (l/min) was a significant protector factor (P = 0.036; Table 4) that was not present in the univariable analysis (P = 0.132; Table 3). Kaplan–Meier survival analysis for the presence of complications in the multivariable model was 40% for all covariates after 1-year follow-up (Fig. 3). However, 1-year survival of the high-risk surgical patients (i.e. those with VE/VCO₂ slope of >35) did not significantly differ from those who underwent non-surgical treatment for their lung cancer (P = 0.148; Fig. 4). Among the patients who underwent lung resection, 37% (19 patients) had poor 1-year survival secondary to their cancer stage, whereas 45% (23 patients) had poor 1-year survival due to poor pulmonary function and high VE/VCO₂ slope of >35.

DISCUSSION

We have shown that COPD patients undergoing lung resection surgery for lung cancer are still at risk of complications and death despite preoperatively selecting the fitter patients. By analysing the functional parameters obtained during preoperative CPET, we have found that, even in those with VO₂ max higher than 10 ml/kg/min, parameters of ventilator inefficiency, especially a VE/VCO₂ slope of more than 35, are the strongest predictors of complications and death 1 year after surgery.

Previous studies and interpretation of the findings

We found that 41% developed postoperative cardiopulmonary complications, whereas 4% died despite being at an acceptable risk according to CPET. In accordance with our finding, Stanzani et al. [12] reported 31.6% morbidity and 4.3% mortality in their population subjected to CPET that correlated significantly with COPD. We observed that decreased diffusion capacity was related to postoperative complications and mortality. Similarly, Ferguson et al. [13] found that diminished diffusion capacity was associated with both mortality and pulmonary morbidity. We also observed that VO₂/kg (ml/kg/min) at maximal exercise was not independently related to postoperative complications. Similarly, Campione et al. [14] did not find that measured VO₂ max correlated with postoperative complications. In our population of COPD patients with VO₂ max of >10 ml/kg/min, other factors may play an important role in the preoperative evaluation that are important to predict postoperative complications.

In this study, we found that the baseline VE/VCO₂ slope was significantly higher among the patients who developed complications or died postoperatively with a hazard ratio of 2.6. Ventilatory inefficiency is one of the most important parameters that gained high acceptance in the evaluation of different groups of patients with problems such as COPD, interstitial lung diseases, pulmonary hypertension and congestive heart failure (CHF) [15]. Moreover, we found that VE% predicted at submaximal exercise (AT) and VE% predicted at maximal exercise were good predictors of all postoperative events as well as maximum VO₂/kg (%) in the univariable analysis (Table 3) and VO₂ (l/min) in the multivariable analysis (Table 4). This could be explained by the associated dynamic hyperinflation, an important pathophysiological characteristic of COPD, that increases during exercise and greatly compromises maximal exercise capacity reflected by lower VO₂ max, VCO₂ max and VE max [16]. Additionally, we found that higher RQ (reflecting combined VO₂ and VCO₂ changes) was another predictor of all postoperative events. It is accepted that VE is closely linked to both VO₂ and VCO₂ and more affected with the changes in VCO₂ in a linear relationship and expressed as VE/VCO₂.

Furthermore, we observed that VE/VCO₂ slope of >35 was associated with all postoperative events in COPD patients. It has been previously demonstrated that VE/VCO₂ slope of ≥34 was associated with less survival in CHF [17] and pulmonary hypertension [18] and with higher sensitivity rather than VO₂ max. Torchio et al. [19] found that high VE/VCO₂ slope of ≥34 is an independent predictor of postoperative mortality after lung resection for lung cancer. More recently, Brunelli et al. [20] found that VE/VCO₂ slope of >35 was associated with higher rate of early (30 days) postoperative respiratory complications among COPD patients rather than VO₂ max and they recommended its incorporation in the clinical practice for preoperative evaluation for lung resection. Our results are in keeping with those observations and extend the risk of cardiopulmonary complications during the year after surgery. Furthermore, we found that high VE/VCO₂ slope of >35 denoting ventilatory inefficiency is a more relevant cause of poor survival in COPD patients eligible for lung resection than cancer-related mortality. We also observed that systolic blood pressure at maximal exercise was significantly higher among those who died postoperatively. This could be due to their higher resting hypertension despite not being statistically significant (Supplementary Table S4) or secondary to abnormal control of blood pressure [10].

Associated cardiovascular comorbidity was not found to be a risk factor of postoperative complications in our population, probably reflecting a selection of fitter patients; however, it is well known as a risk factor for perioperative complications and preoperative cardiac evaluation have been incorporated recently [21] in the assessment of the patients with lung cancer planned for surgical resection. Similarly, neither age nor the extent of pulmonary resection was associated with postoperative complications in our population, as has been previously suggested by some authors [1].

Clinical implications

The present study showed first that ventilator efficiency as expressed by VE% predicted, VE/VCO₂ slope and VO₂ (l/min) is an important predictor of the overall postoperative event including complications and mortality. We suggest that these parameters are more reliable in the patient stratification especially among those with COPD and should not be ignored in the preoperative evaluation. Secondly, we assumed that COPD patients with lung cancer and VE/VCO₂ slope of >35 could benefit from optimal pharmacological treatment and pulmonary rehabilitation programme preoperatively with risk reduction of postoperative complications. Recent studies showed that COPD with lung cancer undergoing surgical treatment got benefit from preoperative pulmonary rehabilitation and VO₂ max improved significantly [22]. Furthermore, therapeutic hyperoxia in resting normoxic COPD could change perioperative risk stratification as it was associated with significant improvement in VO₂ max and decreased VE/VCO₂ slope [23].

Study limitations

The current study has some limitations. First, we could not analyse the different CPET parameters in relation to mortality separately due to the limited number of postoperative mortality (only 2 patients) and we used the overall postoperative events for the identification of dependent risk factors. However, we provided a comparison between postoperative complications and mortality including both baseline characteristics and different CPET parameters. Secondly, we did not take into consideration the estimated PPO (VO₂ max) in the comparison between the development of postoperative complications and its absence and we used only the measured VO₂ max. Bolliger et al. [24] found that calculated PPO (VO₂ max) was a reliable factor for predicting postoperative pulmonary complications. However, a meta-analysis of 14 studies confirmed the strong relationship between measured VO₂ max during CPET and the occurrence of postoperative complications and it was accepted in the last updated guidelines for the physiological evaluation of lung cancer patient [3]. Lastly, the number of events analysed in the current study was low. However, this low number (23 events, 45%) did not affect the power of a survival analysis as at least 10 events are needed to be observed for each covariate considered that is not related to the number of participants [25].

CONCLUSIONS

Forty-five percent of COPD patients undergoing lung resection surgery had some form of complications during the following year despite an appropriate risk stratification using CPET. VO₂ max failed to be a good predictor of events by itself, and markers of ventilator inefficiency, especially a VE/VCO₂ slope higher than 35, and work load as well as baseline VE/VCO₂ slope are independently associated with an increased risk of postoperative complications or mortality.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

Funding

This study was funded by Direcció General d’Investigació i desenvolupament tecnològic de la Conselleria d’Innovació, Interior i Justícia de la Comunitat Autònoma de les Illes Balears and Fondos FEDER [Grupos competitivos (PRE-R-22528-2011)].

Conflict of interest: Hanaa Shafiek was granted by the University of Alexandria and Ministry of Higher Education of Egypt as member of ParOwn (the Partnership and Ownership initiative).

ACKNOWLEDGEMENTS

Authors are grateful to all the Lung Function Unit team for their contribution to the logistics of the study.

REFERENCES