The effect of preoperative smoking cessation and smoking dose on postoperative complications following radical gastrectomy for gastric cancer: a retrospective study of 2469 patients

Gastric cancer (GC) is ranked as one of the most prevalent malignancies worldwide with about one half of cases occurring in China. At present, curative resection is the only mainstay of therapy [1, 2]. Despite a downward trend in the incidence of postoperative complications following gastrectomy, due to significant improvements in surgical techniques, about 10.3–23.6% of patients experienced morbidity, resulting in longer hospital stays, increased healthcare costs, and even perioperative death [3‐6]. In order to reduce postoperative complications and improve the patients’ quality of life, it is essential to identify potential risk factors and those patients at high risk. To date, various factors have been identified as adverse predictors for morbidity after gastrectomy, such as advanced age, obesity, and blood transfusion [3, 5, 6]. Unfortunately, most of these factors are not modifiable [7]. There is growing evidence that smoking is a significant risk factor for postoperative complications following a variety of operations, especially pulmonary problems and surgical site infections [8‐11]. As a result, quitting smoking before surgery is likely to improve the short-term outcomes [12]. Other scholars have argued that cessation of smoking shortly before surgery will not decrease postoperative complications [13, 14]. Possible explanations for the conflicting data were the variable duration of smoking cessation before surgery and inconsistency in patient inclusion criteria.

In fact, most of the previous studies were conducted in patients who underwent thoracic surgery or taking varied abdominal surgery procedures together. But in fact, the incidence and risk factors for postoperative pulmonary problems following lower and upper abdominal surgery may differ significantly. For example, patients who underwent hepatobiliary and upper gastrointestinal surgery were twice as likely to develop pulmonary complications than those from any other surgical categories [15]. Thus, to reduce postoperative complications, it is important to understand the incidence and risk factors for complications for each type of surgical procedure. To our knowledge, no study has focused solely on GC patients and the influence of cigarette dose on postoperative complications following gastrectomy has not been investigated [16]. We hypothesize that smoking may serve as an independent risk factor for postoperative complications following radical gastrectomy for GC, especially pulmonary problems, and there may be a dose-dependent effect between cigarette dose and the incidence of complications, and an optimal duration of smoking cessation preoperatively could reduce its incidence. This question was addressed by conducting a retrospective study to investigate the potential influence of smoking, cigarette dose, and the duration of smoking cessation prior to surgery on the incidence of postoperative complications in GC patients who underwent radical gastrectomy.

Smokers were defined as current or former smokers, and non-smokers were considered to be those who had never previously smoked. Smokers were divided into three categories depending on their smoking cessation period before surgery (≤ 2, 2 to 4, and ≥ 4 weeks), as reported for the majority of previous similar studies [12, 15, 20‐22]. The cigarette dose was calculated by the average number of packs of cigarettes smoked everyday multiplied by the duration of smoking years (pack-years). In addition, smokers were subsequently classified into 3 subgroups according to the 25th and 75th cigarette dose, indicating light, moderate, and heavy smokers (20 and 40 pack-years in the present study, respectively), similar to but not exactly the same with a previous study, in which pack-year was divided into quartiles [9].

Postoperative complications were noted for ≤ 30 days following surgery and classified according to the Clavien-Dindo classification [23]. As reported by Aoyama et al. [24], grade I complications were excluded as they are considered to have little clinical significance. Major complications were defined as Clavien-Dindo grade III or greater. In some patients who suffered from more than one complication, each of the complications was recorded and analyzed.

All patients were classified into two groups according to whether they had ever smoked or not. Differences between the groups were compared by univariate analyses, a χ2 or Fisher’s exact test for categorical variables. The test data for normal distribution was checked by the Shapiro-Wilk test, and a Student t test or Mann-Whitney U test was used for continuous data which was normally distributed or not. Continuous and multi-classification categorical variables were converted into binary outcomes according to the standard clinical or widely accepted thresholds, as reported in our previous studies [6, 19]. Univariate analysis by a χ² test was carried out to test the impact of clinicopathological variables on postoperative complications. Multivariate binary logistic regression analysis (forward: LR) was further used to test factors with a P value < 0.05 following univariate analysis. The prognostic performance of the model was evaluated by the area under the curve (AUC) for the receiver operator characteristic (ROC) plot. Subgroup analysis was performed basing on the smoking cessation period before surgery and the pack-years and was compared with the non-smoker group as the reference using logistic regression analysis. All statistical analyses were performed with IBM SPSS Statistics for Windows (Version 24, NY: IBM Corporation, US), and a P value < 0.05 was considered to be significant.

Although several studies have investigated the influence of smoking and short-term preoperative smoking cessation on postoperative outcomes after various types of surgery, the overall quality of evidence is moderate and limited by the small number of studies contributing to the key analyses, leading to contradictory and perhaps even confusing conclusions [8‐16, 20, 25]. The majority of previous studies concluded that smoking adversely affects the short-term outcomes after surgery, and smoking cessation several weeks before surgery was effective in reducing the incidence of complications [8‐12, 16]. In contrast, other scholars have argued that short-term smoking cessation before surgery was not associated with a decrease in wound healing and pulmonary or overall postoperative complications [13, 14, 20]. Only one study investigated the influence of preoperative smoking cessation on postoperative complications for GC patients in particular, in which Jung et al. [16] reported that postoperative complications, especially impaired wound healing, pulmonary problems, and leakage, occurred more commonly in smokers than in non-smokers. Further subgroup analysis by this research group found that smoking cessation for at least 2 weeks preoperatively was required to reduce the incidence of complications. However, their conclusions have not been verified by other studies, and the potential impact of cigarette dose has never been investigated in GC patients undergoing curative gastrectomy.

To the best of our knowledge, this is the largest study to date that has evaluated the impact of smoking on postoperative complications and the first study to investigate the attributable risk of cigarette dose on postoperative complications, basing on GC patients undergoing radical gastrectomy. According to previous studies, besides pulmonary problems, impaired wound healing and leakage were also confirmed to be significantly associated with smoking [9, 16]. Tissue hypoxia caused by acute exposure to smoking and lacking proper fibroblast migration to form the healing tissue were considered to be the most important cause of this link [26, 27]. Moreover, heavy smokers had an obviously increased incidence of bleeding requiring transfusion, but the exact reasons for this bleeding remain to be established [10]. Thus, complications were divided into surgical site infection (including wound healing and leakage), pulmonary complication, bleeding, and others in this study, similar with that of the previous study [16]. In this study, we revealed that smoking was an independent risk factor for postoperative complications, especially pulmonary problems. All of the 65 patients developing pneumonia received antibiotic treatments, and 12 among who developed respiratory failure needing ventilation were transferred into the intensive care unit. And finally, 3 patients died due to pneumonia within 30 days after surgery. But they did not differ significantly between the smoker and non-smokers. In addition, severe smoking (≥ 40 pack-years) was identified to be significantly associated with surgical site infections.

The incidences of overall complications were comparable among patients who stopped smoking ≤ 2 weeks, 2 to 4 weeks, and ≥ 4 weeks preoperatively. Compared with the incidence of postoperative pulmonary problems in non-smokers (3.3%), it was comparable with that in those who stopped smoking ≥ 4 weeks before operation (3.1%, P = 0.914), but significantly higher in those who stopped smoking < 4 weeks (5.7%, P = 0.007). Postoperative pulmonary problems have been identified to be the most costly of all major postoperative complications and resulted in the longest duration of hospital stay, as identified by the National Surgical Quality Improvement Program (NSQIP) [28]. In addition, previous studies have found that patients who underwent upper abdominal surgery were twice as likely to develop pulmonary problems than those undergoing any other surgical category [15]. Given the high incidence of and the significant adverse impact of postoperative pulmonary complications, it is very important for patients who underwent gastrectomy to quit smoking before their operation, which has been identified as a modifiable risk factor. Based on the present study, at least 4 weeks cessation may be needed to prepare for surgery, a finding consistent with the conclusion reported by Lindström et al. [12], who found that smoking cessation 4 weeks before surgery decreased the risk of complications. A meta-analysis of 6 randomized controlled trials and 15 observational studies revealed that each week of preoperative abstinence increased the relative risk reduction of postoperative complications by 19% [22]. And at least 4 weeks of smoking cessation had a significant larger treatment effect than shorter abstinence from smoking. Jung et al. [14] even suggested that as short as 2 weeks of smoking cessation before surgery was effective in reducing its adverse influence on surgical outcomes. One possible explanation is that the acute toxic effect of recent smoke inhalation is a major mechanism for the delayed wound healing and infection risk [8, 16]. Thus, even short-term smoking cessation can reduce the damage caused by acute exposure to smoking and lead to a significant reduction in postoperative complications.

However, other studies have revealed that smoking cessation about half a month, 8 weeks, or even 16 weeks before surgery did not reduce the level of overall complications [13, 14, 20]. In the opinion of these researchers, the chronic cumulative effects of smoking on pulmonary functions, such as increased mucus production and a reduction in immune functions, played a more important role than acute toxic exposure, which could result in postoperative pulmonary infection or even respiratory failure [8]. It has been reported that pulmonary function generally improved after about 8 weeks of smoking cessation [25]. Thus, the optimal duration of smoking cessation prior to an operation to reduce postoperative complications remains poorly defined. Because almost all of the patients underwent surgery within 2 weeks from the diagnosis of GC in our department, the patients with smoking cessation > 2 weeks before operation had stopped smoking when they were admitted, whereas patients with cessation time ≤ 2 weeks were generally current smokers until they were advised to stop smoking during admission. Compared with patients with shorter preoperative smoking cessation period (≤ 2 weeks), those with longer interval (> 2 weeks) seemed to have a higher rate of developing symptoms because of the tumor, such as bleeding and/or vomiting due to pyloric obstruction (26.7% vs 21.1%), although the difference was barely outside the range of significance (P = 0.077). Moreover, patients with longer smoking cessation period were significantly older than those with shorter cessation interval (60.17 vs 56.13 years, P < 0.001), and with significantly lower preoperative hemoglobin levels (118.54 vs 125.05 g/L, P = 0.001), whereas advanced age, lower hemoglobin level, and pyloric obstruction were all well-known adverse predictors for morbidity following gastrectomy [3, 5, 29, 30]. Thus, we hypothesize that although the incidences of postoperative complications were comparable among patients with different smoking cessation period in the present study, the benefit of longer cessation time might be counteracted by the poorer general condition of the patients. Therefore, a prospective randomized large sample size study is needed to confirm this conjecture in the future.

In contrast to smoking status and duration of smoking cessation, the influence of smoking severity on postoperative complications has rarely been investigated. The incidence of postoperative complications was identified to be positively associated with cigarette dose in the present study, as was the surgical site infection. Pulmonary complications were also significantly more common in moderate to heavy smokers (with > 20 pack-years). In a previous study including 20,830 cancer patients to investigate the potential effect of smoking on postoperative outcomes, Gajdos and colleagues [9] divided pack-years into quartiles and split by smoking status. Given the relatively smaller sample size of our study, we classified the smokers into 3 subgroups according to the 25th and 75th cigarette dose. As expected, major adverse events were also significantly more common in heavy smokers, compared with light to moderate smokers or non-smokers. The dose-response relationship was highly suggestive of a causal rather than a mere correlational association. The results were echoed by Hawn et al. [8], who argued that a dose-dependent relationship existed between pulmonary problems and pack-year exposure, and revealed that smoking-associated surgical complications were significantly increased in patients smoking for > 20 pack-years. Similarly, Livingston et al. [31] found that current smoking of > 20 pack-years was related to an increased risk of failure to wean in a series of 575 bariatric patients. In a propensity score-matched analysis of 36,454 patients who underwent plastic surgery, Toyoda et al. [10] concluded that smokers with 11 or more pack-years had significantly higher rates of deep surgical site infections and re-operation following plastic surgery. A dose-dependent association between urine nicotine levels and wound healing impairment in breast reduction patients was also found by Bartsch et al [32] The possible explanation was the chronic cumulative effects of the adverse impact of smoking, such as tissue hypoxia on pulmonary, immune, and wound healing functions [8]. This explanation was indirectly supported by the finding that a high concentration of oxygen following operation decreased the risk of wound healing problems [33]. Therefore, it was reasonable to hypothesize that a high concentration of oxygen during or after gastrectomy for heavy smokers could reduce the incidence of postoperative complications. But further study is needed to unequivocally clarify this suggestion.

The present study has a number of limitations first and foremost being that it was a retrospective and single-institution study. Thus, although smoking was identified to be an independent predictor for postoperative complications, especially pulmonary problems, it does not indicate a causal relationship and further prospective studies are needed. Second, the smoking habits and cigarette dose were only orally obtained by the responsible doctor, instead of through biological monitoring (such as nicotine metabolites), which would have been more objective. Although we recorded the average number of smoked cigarettes and the duration of smoking to calculate the cigarette dose, it should be borne in mind that the content of harmful substances, such as nicotine, in different types of cigarettes is likely to vary significantly, which was not investigated in detail in the present study. Therefore, there is a possibility of misclassification and miscalculation of the risk of complications in the subgroup analysis. Last but by no means the least, the basic clinicopathological characteristics were significantly different in smokers and non-smokers, such as age and gender, which may present as confounders when investigating the relationship between smoking and postoperative complications. A prospective study or retrospective study with large samples using propensity score matching analyses to adjust the potential selecting bias may eliminate this limitation [10].

The present study has identified that smoking was an independent adverse predictor for postoperative complications, especially for pulmonary problems, in GC patients following radical gastrectomy. It is suggested that at least 4 weeks of smoking cessation prior to surgery is required to decrease postoperative pulmonary problems. There was a positive relationship between the incidence of complications and cigarette dose, and more than 20 pack-years were shown to increase significantly the risk of complications. Therefore, to reduce postoperative complications, both surgeons and patients should pay much attention to preoperative smoking cessation, especially for heavy smokers.