This study demonstrates that patients who continue to smoke until the date of curative-intent lung cancer surgery have a higher postoperative morbidity, including higher frequency of PPC, longer hospital LOS, and a higher frequency of ITU admission. In ex-smokers there was trend for reduced frequency of PPC and ITU admission compared to current smokers, but there was no significant difference in observed outcome between patients who quit smoking less <6 weeks versus ≥6 weeks prior to surgery. We found no significant differences in early mortality or long-term survival between any of the groups within our follow up period.
Our study demonstrated that the incidence of PPC in current smokers is 22%, which is greater than 10 times that of never smokers and twice that of ex-smokers. Comparing this observed incidence with results from other studies is challenging because of the differences in the definition of PPC, which may extend to include potentially more severe yet less frequent complications such as pulmonary embolism, prolonged air leaks and bronchopleural fistulas . Despite this, the frequency of PPC in smokers using the MGS is within the wide range of pulmonary events reported in other studies (6.9–43.2%) [5, 9]. Our finding of PPC to be higher in current versus never smokers is also supported by other studies [5, 10].
Despite the observed trend, our study showed no significant difference in PPC frequency between ex-smokers and current/never smokers. Other studies have found no significant difference in ex-smokers and current/never smokers which is most likely due to much smaller study sample sizes , and/or the exclusion of a never smoking control group [11, 12]. In a general thoracic surgery cohort, we recently demonstrated that compared to never smokers, ex-smokers were more than twice as likely to develop a PPC . However, the timing of smoking cessation prior to surgery, and the subsequent effect of postoperative outcome was not investigated.
In our prospective cohort, using 6 weeks as a cut off, we were unable to find an optimum duration of preoperative smoking cessation to reduce PPC incidence. Despite this, our study has demonstrated that stopping smoking 6 weeks prior to surgery seems to reduce, and not increase PPCs compared to current smokers. Interestingly, the study by Nakagawa et al (n = 288)  found the incidence of PPCs among ex-smokers (2–4 weeks prior to surgery) was 53.8%, whilst in current smokers was 43.6%, which were both significantly higher when compared to the never-smokers (23.9%; p < 0.05). However, this study is limited by the retrospective nature of data collection. As defined by Russell Standard , the study found that at least 4 weeks of smoking abstinence was needed for a marked reduction in PPC (53.8% to 34.7%), whilst the incidence of PPC in patients who had a smoke-free period of 9–12 weeks or longer approached the same incidence as those in the never-smoked group. A large prospective review of the General Thoracic Surgery Database from the United States (n = 7990) found that the risk of PPCs following primary resection for lung cancer steadily declined with longer intervals of smoking cessation (4 days - 1 month, 1–12 months, >12 months) . However the differences were not significant enough between the groups to recommend optimal interval of smoking cessation. One critique of this study is that the incidence of PPC was low (6.9% in smokers) as only major pulmonary events were reported, suggesting minor yet more frequent PPCs may have been missed.
Smoking had no direct effect on postoperative mortality or long-term survival during in our study. Nevertheless, the adjusted survival curves for never smokers were clearly above those of the other smoking categories especially after the first 12 months of follow up, suggesting a likely more favorable outcome even though this did not reach statistical significance. This may be in part due to the short follow up period or the smoking habits of patients after surgery; some may stop whilst others may restart. Other studies have shown increased mortality in smokers compared to never smokers, but no significant difference amongst the current and ex-smoking groups [5, 13]. The deleterious effect of continued smoking on long-term survival after being diagnosed with lung cancer is increasingly evident , but few papers have looked into the effect of preoperative smoking cessation on long-term survival in lung cancer patients undergoing curative-intent surgery. A multicenter review of 169 patients undergoing NSCLC resection found that current smokers had a reduced 5-year survival in comparison to never smokers (72% vs. 91%, p = 0.02), and that stopping smoking a year prior to cancer diagnosis had no significant difference in long-term survival . Interestingly, although smoking was not an independent risk factor for late-deaths in our study, PPC development and LOS were independent risk factors for late-deaths, both of which were significantly increased in the current smokers.
We found that lower FEV1% predicted, BMI and COPD diagnosis were associated with current smoking on univariate analysis. Other investigators have found that the prevalence of low and normal BMI was greater among patients who are active smokers; those underweight patients had a greater risk-adjusted LOS compared to normal weight patients . We have previously investigated the risk factors for developing a PPC, and although lower preoperative FEV1% predicted and COPD diagnosis are associated with PPC on univariate analysis, only COPD diagnosis is regarded as an independent risk factor for PPC development or multivariate analysis [1, 2]. Even in these models, current smoking remains the strongest independent risk factor for PPC development. Authors have found carbon monoxide diffusion capacity (DLCO) to be associated with PPC development ; however, DLCO was not included in this study, as it was only performed in patients with limited exercise tolerance so data are limited.
The observed effects of smoking and increasing frequency of PPC could be explained by the suppressive effect of cigarette smoking on the innate immune system. Smoking impairs the mucociliary escalator and the ability to remove foreign bodies/pathogens in the larger airways. This normalises after 15 days following smoking cessation implying that it is not the only etiological factor in PPC development . Smoking seriously impairs antimicrobial [17, 18] and proinflammatory  functions of alveolar macrophages. Furthermore, there is a compounding effect on anaesthesia and smoking on suppression of intraoperative antimicrobial/phagocytic activity in alveolar macrophages. Smoking cessation of 6 months was needed to normalise proinflammatory function whilst only 3 months was required to normalise phagocytic function .
The question of whether a smoking cessation programme in surgery affects postoperative outcomes has been more recently reviewed . Implementation 6–8 weeks prior to elective orthopedic surgery reduced postoperative complications (18% vs. 52%, p = 0.0003) , whilst implementation 2 weeks prior to colorectal surgery made no difference . Despite increasing numbers of small RCTs suggesting intensive smoking cessation treatments work in people with pulmonary diseases such as lung cancer, many patients continue to smoke . In thoracic surgery, pulmonary rehabilitation programmes have shown promise in increasing smoking cessation, with a trend for fewer PPCs in the intervention group ; however, RCTs are needed to confirm efficacy.
In this prospective study, the main limitation is the reliance on self-reporting smoking status in preoperative patients to ascertain smoking cessation, which could result in recall bias and underestimate the real smoking habits of patients. Use of biochemical confirmation by use of exhaled carbon monoxide testing and measuring cotinine levels from urine, saliva or blood could be used in future studies . Furthermore, smoking cessation data was collected as a categorical variable, and therefore additional statistical analysis using ROC curve analysis could not be performed to further inform on a cut-off for the duration of smoking cessation needed to reduce PPC incidence. In our study, there were a low number of VATS cases in our prospective cohort (8.4%), and only 5.6% of patients with stage 1 NSCLC underwent VATS resection. The data of this study preceded the growth of VATS lobectomies in NSCLC patients; the vast majority of lung resection over this time frame was performed via open thoracotomy. This low level of VATS resections could confound the results and make them less generalisable considering that many studies have demonstrated a higher postoperative pulmonary complication (PPC) rate with open thoracotomy vs. VATS resections [27, 28]. Therefore, further studies are required to investigate the effects of smoking cessation on PPC frequency in VATS verses open thoracotomy cases.