The Change of Laboratory Tests Could Be Predictive Factors for Infection after McKeown Esophagogastrectomy

Aim To analyze whether the change of laboratory tests (postoperative day 1 (POD1) minus pre-operation) could be predictive factors for postoperative infection in patients who have undergone McKeown esophagogastrectomy. Methods We retrospectively investigated the clinical data of 358 patients who have undergone McKeown esophagogastrectomy, and divided them into infection and noninfection groups. SPSS 22.0 software was performed for data analysis. Results In the two groups, smoking status (66.7% vs. 42.3%; P = 0.014), male gender (86.1% vs. 72.0%; P < 0.001), hoarseness (23.6% vs. 8.7%; P < 0.001), poor coughing ability (51.4% vs. 9.1%; P < 0.001), the change of WBC count (5.59 ± 4.75 × 109/L vs. 4.51 ± 4.11 × 109/L; P = 0.05), the change of glucose (6.03 ± 3.97 g/L vs. 3.78 ± 3.18 g/L), the change of ALB (−12.83 ± 3.45 g/L vs. −10.69 ± 3.86 g/L), the change of CRE (0.17 ± 19.94 umol/L vs. −4.02 ± 15.40 umol/L, P = 0.047) were significantly different. These factors were assessed using logistic regression analysis, and factors with P ≤ 0.05 in the univariate analysis were entered into multivariate analysis based on the forward stepwise (conditional) method. Poor coughing ability (odds ratio [OR], 11.034, 95% confidence interval [CI], 5.358–22.724), smoking status (OR, 4.218; 95% CI, 2.110–8.431), the change of WBC count (OR, 1.079; 95% CI, 1.000–1.164), the change of serum ALB level (OR, 0.849; 95% CI, 0.772–0.935), and the change of blood glucose levels (OR, 1.237; 95% CI, 1.117–1.371) were determined as independent risk factors for postoperative infection. We established a scoring system based on these 5 factors, and the area under the curve for this predictive model was 0.843 (range, 0.793–0.894); the sensitivity, specificity, and cut-off score were 70.8%, 85.3%, and 2.500, respectively. Conclusion Poor coughing ability, smoking habit, the high change of WBC and blood glucose levels, and low change of serum ALB levels can be used to predict the occurrence of postoperative infections among patients who have undergone McKeown esophagogastrectomy.


Introduction
Up to now, Esophageal cancer is the sixth-most common cause of cancer-related death all around the world, and in developing countries, it is the fi h most frequent cause of deaths [1]. Furthermore, the incidence and mortality of patients with esophageal cancer in China were the highest globally in 2009 [2]. Surgery remains the standard treatment for resectable esophageal cancer. However, esophagogastrectomy is a complex procedure, with morbidity and mortality rates of 23%-50% and 2%-8%, respectively [3,4].
Patients undergoing McKeown esophagogastrectomy are exposed to a higher risk of infection compared with those receiving other types of surgery. Moreover, patients with esophageal cancer are at a greater risk of antimicrobial exposure due to their impaired immunological functions and are also at an increased risk of infection with multidrug-resistant bacteria.
In this study, we assumed that the change of laboratory tests (laboratory tests within 24 h a er surgery minus pre-operation) will be associated with the infections following McKeown esophagogastrectomy, and developed recommendations for clinicians treating patients with these risk factors. , and lactic acid levels) between the groups. All the postoperative day 1 (POD1) indicators were analyzed within 24 h a er surgery. And we used POD1 indicators minus preoperative ones to calculate the change of laboratory tests.

Inclusion and Exclusion Criteria.
e inclusion criteria were as follows: patients aged >18 years with esophageal cancer who underwent McKeown esophagogastrectomy and developed an infection during hospitalization. And those with infection prior to hospital admission were excluded from the study.

Statistical Analysis.
Categorical variables were expressed as number and percentage, and continuous variables were expressed as mean ± standard deviation. Student's t-test was used to examine continuous variables, and the Chi-squared test or Fisher's exact test was used to assess categorical variables. Multi-variate analysis was performed to determine the predictors of postoperative infection, and the forward stepwise (conditional) method was used to identify factors to enter into the multivariate regression model. Receiver operating characteristic (ROC) curves were constructed to estimate the sensitivity, specificity, and the area under the curve (AUC) for various cutoff points of the relevant indicators. Statistical significance was set at 푃 ≤ 0.05, and all statistical analyses were computed using SPSS Version 22.0. Table 1 describes the characteristics of the 72 patients (20.1%) with postoperative infection, from among the 358 patients who had undergone McKeown esophagogastrectomy in the present study. We compared the patients' baseline characteristics and clinical disease features between groups, and identified significant differences in smoking habits and gender between the two groups. e smoking habit frequency (66.7% vs. 42.3%; 푃 < 0.001) and proportion of males (86.1% vs. 72.0%; 푃 = 0.014) were greater in the infection group than in the noninfection group (Table 1 and Figure 1).

Differences in the Perioperative Clinical Features.
In the present study, the factors of hoarseness (23.6% vs. 8.7%; 푃 = 0.001) and poor coughing ability (51.4% vs. 9.1%; 푃 < 0.001) were significantly different between the groups; both were more frequent in the infection group. However, other perioperative clinical features, including wound pain, increased heart rate and respiratory rate, chest pain/chest distress, and atrial fibrillation, did not exhibit a significant difference (Table 2 and Figure 2).

Development of a Scoring System to Predict Postoperative
Infections.
e AUC and cut-off point were 0.575 (range, 0.498-0.651) and 4.420 × 10 9 /L for the change of WBC count, 0.725 (range, 0.660-0.790) and 4.355 mmol/L for the change of blood glucose level, and 0.658 (range, 0.590-0.727) and −11.900 mmol/L for the change of serum ALB level, respectively.
Patients with were assigned a score of 1 for each of the following factors: poor coughing ability, smoking habit, the change of WBC count and blood glucose levels greater than the cut-off values, and the change of ALB level lower than the According to our analysis, smoking acts as one of the independent risk factors for predicting postoperative infection. Liu et al. [5] reported that smoking history was one of the risk factors of postoperative lung infection. Moreover, the study conducted by Kinugasa et al. [6] showed that smoking habit was risk factors for postoperative pulmonary complications. Furtermore, the similar result shown in the study conducted by Ferguson et al. [7]. Smoking history could increase airway resistance, then cause numerous postoperative sputum thrombi. Consequently, the risk of pulmonary infection increases by impairment of the respiratory epithelium cilia structure, damaging to goblet cells, and weakening cilia movement.
e result of our study showed that the change of WBC count was associated with the postoperative infection. Sugita et al. [8] showed that the preoperative WBC count did not differ between infected and noninfected patients, although the POD1 WBC count was significantly higher in infected patients than in noninfected patients. e study conducted by Gomez et al. [9] also showed that the median WBC count was significantly greater in patients with infection than in those without infection during the first 10 postoperative days.
Furthermore, we found that the change of ALB level was an independent risk factor for postoperative infection in patients underwent McKeown esophagogastrectomy. e study conducted by Yin et al. [10] showed that low serum albumin was independently associated with the surgical site infections (SSI). Moreover, Zhao et al. [11] demonstrated that ALB level <35 g/L was an independent risk factor for postoperative infectious complications in patients with hepatocellular cancer. In addition, Yuwen et al. [12] showed that an ALB level of <35 g/L was associated with an increased risk of SSI in patients a er orthopedic operations.
Lastly, our study showed that the change of blood glucose level was an independent risk factor for predicting infection. Similar to our findings, Vriesendorp et al. [13] indicated that the POD1 blood glucose level in esophageal cancer patients a er esophagectomy was only associated with the length of hospitalization. Moreover, Ng et al. [14] showed that the change in the target glucose control in diabetic patients was independently associated with an increase in SSI. Another study conducted by Ambiru et al. [15] demonstrated that the SSI rates were correlated with the hyperglycemia following surgery.
In the scoring system of our study, the continuous variables changed to categorical variables through the cut-off value (higher than cut-off value is number A, lower than the cut-off value is number B), and added value for each factor. All the categorical variables used assigns 20% for each factor in the total score because it was useful and easy in clinical practice.
I think our study is quite novelty. On the one hand, our study talked about specific disease (esophageal cancer), and undergoing specific kind of operation (McKeown cut-off value; patients who did not meet these requirements were assigned a score of 0 each. e AUC of this predictive model was 0.843 (range, 0.793-0.894); the sensitivity, specificity, and cut-off score were 70.8%, 85.3%, and 2.500, respectively ( Figure 4 and Table 5).

Pulmonary Complication through Clinical Diagnosis.
e differences of changes of laboratory tests in the groups divided by pulmonary complication were showed in the supplementary data.

Discussion
Results of comparing the infection and noninfection groups in the present study indicated that poor coughing ability, pulmonary function tests in our clinical practice; we need prospective study to address this problem for better diagnosing the respiratory complication, which is one of the major complications of esophagectomy [16,17]. Our study did not compare the minimally invasive esophagectomy with open invasive esophagectomy, some studies showed that minimally invasive esophagectomy had lower incidence of postoperative infection than open invasive esophagectomy [18][19][20]. In addition, other factors should be contained in further study, such as operation time, intraoperative bleeding, the application of antacids, and so on. esophagogastrectomy); on the other hand, in our study, we used the changes of these laboratory tests different from other studies. e limit of our study is that it is a single center retrospective research, and the study population comprised only Asian participants. e patient number enrolled in our study was relatively small, so some risk factor, such as alcohol consumption were not included in the independent risk factors. Probably, in the future study, we could get more data to conduct a tendentious matching analysis to better identify the risk factors. Moreover, we do not have postoperative

Conclusion
Patients are exposed to high risks of predicting postoperative infection a er McKeown esophagogastrectomy, although poor coughing ability, smoking habit, the change of WBC count, the change of ALB level, and the change of blood glucose level may be as independent risk factors for postoperative infections in these patients. At last, we used a scoring system comprising these 5 factors, and observed that the AUC of this predictive model was 0.843 (range, 0.793-0.894), whereas the sensitivity, specificity, and cut-off score were 70.8%, 85.3%, and 2.500, respectively.
Data Availability e datasets used and/or analyzed in the current study are available from the corresponding author upon request.

Consent
All authors have agreed to the publication of this manuscript. level within 24 h a er surgery minus pre-operation. Change of glucose: blood glucose level within 24 h a er surgery minus pre-operation. With regard to the scoring system, patients were assigned a score of 1 for each of the following factors: poor coughing ability, smoking habit, change of WBC count and blood glucose levels greater than the cut-off values, and change of ALB level lower than the cut-off value, whereas patients who did not meet these requirements were assigned a score of 0 each.