The Role of an Array of Routine Clinical Variables to the Occurrence and Severity of Postoperative Pneumonia in Non–Small Cell Lung Cancer Patients

Lung cancer (LC) has become the leading cause of death in China, regardless of gender or region.¹ Surgery is still the cornerstone of management and treatment for LC patients, and currently is the only therapy offering a potential hope for cure, especially for patients diagnosed in the early stages. Postoperative pneumonia (POP) is one of the most common complications in LC patients after therapeutic surgical resection, as it potentially increases mortality risk and hospitalization expenses, prolongs length of stay in hospital, and significantly influences the sequential recovery and prognosis of patients; also, POP may be a marker of increased long-term mortality after LC surgery.^2–4

To better prevent or predict pneumonia in surgical patients with LC, up-to-date data on POP risk factors are urgently necessary, as recent studies have revealed that the ratios of POP are quite different.^2–4 Furthermore, an effective assessment of risk of POP would allow physicians to set realistic expectations for postoperative therapies and care. Various genetic and molecular biomarkers show promise for being predictors of POP; however, the majority of them may not yet be available in routine practices, not to mention the high cost and need for specialized equipment and expertise. Therefore, analyzing the risk factors of POP in some routine variables that may carry significant importance is considered to be more feasible or practicable. Based on this, we performed a cohort study to examine patient-related risk factors for POP as well as severity of POP following therapeutic LC lobectomy in routine variables.

Materials and Methods

Results

Population characteristics

Data from 1058 patients were included in this retrospective analysis. All of them were diagnosed as LC with the histology of NSCLC (mean age: 61.3 ± 10.0 years; range: 39–86 years); 55.5% (587/1058) of them were males and 21.3% (225/1058) were current-smokers; of them, 150 (14.2%) patients suffered from POP and divided into the POP group, compared with 908 (85.8%) in the N-POP group. Details of baseline characteristics were listed in Table 2. Patients in the POP group were older than in the N-POP group (64.5 ± 12.4 versus 58.8 ± 9.4, P < 0.001). Meanwhile, higher proportions of patients with American Society of Anesthesiologists (ASA) scores ≥3 (17.3%, 26/150 versus 8.0%, 73/908, P < 0.001); Charlson comorbidity index (CCI) scores ≥3 (22.0%, 33/150 versus 7.6%, 69/908; P < 0.001); body mass index (BMI) ≥30 (8.7%, 13/150 versus 2.9%, 26/908; P < 0.001) were also found (Table 2).

Table 1  POP severity grades classified by Clavien-Dindo classification of surgical complications

View Fullscreen

Table 2  Baseline characteristics between the groups

View Fullscreen

Clinical characteristics

Higher WBC count as well as NLR and longer operation time were identified in the POP group compared with the N-POP group (WBC: 7.4 ± 2.9 versus 5.9 ± 1.7 × 10⁹; P < 0.001; NLR: 3.2 ± 1.2 versus 2.4 ± 0.9; P < 0.001; operation time: 124.2 ± 29.4 versus 118.8 ± 32.2 minutes; P < 0.001). Patients in the POP group had longer total hospital and postoperative stay length of stay than those in the N-POP group (total hospital stay: 22.0 ± 9.2 versus 15.9 ± 4.1 days; P < 0.001; postop stay: 13.2 ± 9.0 versus 7.8 ± 3.2 days; P<0.001). Also, the POP group's total in-hospital and postoperative expenses were higher than the N-POP group (in-hospital: ¥57,499.7 ± ¥19,019.4 versus ¥45,962.2 ± ¥8452.5; P < 0.001; postoperative: ¥12,389.8 ± ¥1466.9 versus ¥6012.0 ± ¥2563.7; P < 0.001), with no difference observed in preoperative expense between the groups (¥3108.5 ± ¥1106.2 versus ¥3126.1 ± ¥1235.4; P = 0.276; Table 3).

Table 3  Clinical characteristics between the groups

View Fullscreen

Risk factors of POP and POP severity

We performed multivariate analyses via binary logistic regression analysis and ordinal polytomous logistic regression for analyzing risk factors analysis of POP as well as POP severity.

Regarding the binary logistic regression analysis for risk factors of POP, variables analyzed included operation time; bleeding amount; CCI score >3; ASA score >3; postoperative predicted diffusion capacity for carbon monoxide of the lung (ppoDlco%); postoperative predicted forced expiratory volume in 1 second (ppoFEV1%); WBC count; NLR; albumin/globulin (A/G) ratio; platelet (PLT); surgical approaches [open/video-assisted thoracic surgery (VATS)]; chronic obstructive pulmonary disease (COPD); cardiovascular diseases; diabetes mellitus; early stage; gender; and age. The following were found to be the independent risk factors of the occurrence of POP: PpoFEV1% (OR: 0.996, 95% CI: 0.993–0.999; P = 0.021); CCI score >3 (OR: 2.694, 95% CI: 1.462–4.965; P = 0.001); ASA score >3 (OR: 2.066, 95% CI: 1.060–4.029; P = 0.033); ppoDlco% (OR: 0.458, 95% CI: 0.090–0.809; P = 0.014); and NLR (OR: 2.171, 95% CI: 1.721–2.737; P < 0.001; Table 4).

Table 4  Multivariate analysis of POP risk factors

View Fullscreen

With regard to POP severity, 19 (12.7%, 19/150) patients were classified as grade I; 89 (59.3%, 89/150) as grade II; 33 (22.0%, 33/150) as grade III; 7 (4.7%, 7/150) as grade IV; and 2 (1.3%, 2/150) as grade V (Table 5). Ordinal polytomous logistic regression was used for risk factor analysis of pneumonia severity. We observed the following to be independent risk factors of POP severity: early stage (OR: 0.626, 95% CI: 0.422–0.929, P = 0.020); CCI score >3 (OR: 1.914, 95% CI: 1.058–3.459, P = 0.032); ppoDlco% (OR: 0.638, 95% CI: 0.445–0.914, P = 0.014) and [insert risk factor here] (OR: 1.218, 95% CI: 1.031–1.436, P = 0.020; Table 6).

Table 5  Proportions of POP grades

View Fullscreen

Table 6  Ordinal polytomous logistic regression for risk factor analysis of POP severity

View Fullscreen

Discussion

The aim of this study is to provide evidence for several important risk factors for POP and POP severity in LC patients undergoing LC surgery, and subsequently, better evaluate patients' physical status after surgery in order that appropriate and prophylactic measures—especially specific prophylactic measures including individualized antibiotic therapy—can be properly carried out. Unlike other studies in which the identified risk factors for POP following LC surgery are similar to those for community-acquired pneumonia in general populations,^6,7 this study includes some routine clinical variables such as pulmonary function tests, preoperative WBC count, and NLR, which are easy to collect. Meanwhile, we used a Clavien-Dindo classification system to assess the POP severity, and investigate the risk factors of POP severity, which may potentially provide an innovative aspect of POP analysis.

The occurrence of POP prolongs the length of hospital stay, increases costs for health care systems, and may indicate the increased long-term mortality in patients after LC surgery. Finding out risk factors in POP after LC lobectomy, and formulating a feasible and scientific model to assess the risk of POP could help physicians develop suitable therapies or preventions. According to our study, patients in the POP group had higher in-hospital costs and longer in-hospital stays compared with patients in the N-POP group. This might validate the negative effects of POP in surgical LC patients.

A history of previous pneumonia is a clinically important predictor, which may be a cost-effective and easy screening tool to identify patients at high risk of POP. Meanwhile, many large-sample studies speculate that a history of community-acquired pneumonia increases the risk of subsequent new episodes of pneumonia, indicating that history of previous pneumonia may be significantly associated with the occurrence of POP after lobectomy.^6,8 Our results revealed that preoperative WBC count was higher in the POP group, which provided evidence for the possibility that history of previous pneumonia resulted in higher preoperative WBC counts, and sequentially the higher risk of incidence of POP. Moreover, numerous studies have been performed to investigate postoperative-related factors including a large number of inflammation-based prognostic markers in the tumor microenvironment and peripheral blood of patients with NSCLC, though the basis of this is not altogether clear.^9,10 White cells subtypes, in particular neutrophils and lymphocytes, play an active role in the inflammation process. NLR, correlating neutrophil and lymphocyte, have been repeatedly reported and confirmed to be useful biomarkers to measure the inflammatory status of the immune system, as the active inflammatory status of the immune system is deemed to expedite the process of tumor invasion and metastasis.¹¹ Plenty of studies are focusing on the association between NLR and postoperative survival of cancer patients, and many of their results have shown a NLR as a predictor of postoperative survival of cancer patients.¹¹ Moreover, NLR may also potentially be a factor to indicate the incidence of postoperative pneumonia. Therefore, we conducted a retrospective study involving LC patients in our department, to assess the association of NLR with the incidence of postoperative pneumonia. The result shows that NLR is an independent factor for POP as well as POP severity, which provides evidence to us to predict the occurrence of POP, though further study is needed to confirm the conclusion.

FEV1 and DLCO are the most commonly used pulmonary function variables, considering the promising correlation with postoperative mortality, complications, and respiratory failure after LC surgery.^12–16 Moreover, FEV1% and DLCO% have been further refined to ppoFEV1% and ppoDLCO%, as ppo values are suggested to more accurately stratify postoperative risk by normalizing to the extent of resection. Several investigators have demonstrated stronger relationships between ppoFEV1% and ppoDLCO%, and postoperative risk after pulmonary resection compared with their parent FEV1% and DLCO% values.^17–20 According to our results, the ppoFEV1% and ppoDlco% can each be valuable for predicting the occurrence of POP as well as POP severity, confirming their correlation with POP, one of the most frequently occurring PPCs (postoperative pulmonary complications).

With the prolonged life expectance, the epidemiology shows an increasing rate of LC patients suffering from multicomorbidities such as pulmonary, cardiac, and vascular diseases, which make the affected patients more susceptible to PPCs compared with youngsters. The risk factor analysis shows that age is not a risk factor for POP, though some studies have reported the impact of age on postoperative complications.²¹ Also, chronic pulmonary diseases including COPD, which is strongly associated with a history of smoking and decreased lung function, is shown to not be the independent risk factor for POP, though the relationship between comorbidities and postoperative complications has been confirmed by numerous researches.^22,23 Moreover, the CCI, which is used to measure comorbidities, is observed as one independent risk factor for both POP and POP severity. Although our study does not confirm smoking as a risk factor for postoperative complications, the impact of smoking on pulmonary function is significant, considering that pulmonary function is closely related to the occurrence and severity of postoperative complications. Many studies have repeatedly confirmed that the relatively low risk of POP in thoracoscopic surgery, compared with the approach of open surgery—though lacking the statistical significance in the result of surgical approaches between the two groups—leads the analysis to be opposite.^24,25

A few limitations to this study are worth noting. First, it is only a retrospective research rather than a prospective randomized trial. Although we endeavored to develop proper inclusive/exclusive criteria to remove other factors that may confound the study, excluding some patients (e.g., patients whose missing data accounted for 30%, who once received antibiotic therapy or had pneumonia before the surgery) would reduce the generalizability of the study, and inevitably affect the results

Moreover, we evaluated the patients undergoing a surgical treatment by several surgeons in a single center, and the study only aims to find out the risk factors of POP or POP severity needing further study to confirm or promote the relevant conclusions. We should focus on and formulate a new, economical, and feasible model using routine variables to predict or assess the risk of POP (e.g., the next step for future studies is the creation of a clinical scoring system). Furthermore, the mechanism of the persistent elevation of the postoperative inflammatory response remains unclear. We only illustrate our clinical findings and present the conclusion. More studies and prospective researches should be performed to refine or confirm the relevant conclusions.

In conclusion, among an array of clinical variables in hospital, major risk factors for POP following LC surgery were ppoFEV1%, ppoDlco%, NLR, ASA score >3, and CCI score >3; meanwhile, ppoDlco, CCI score >3, NLR, and early tumor stage were the positive predictors of POP severity.