Modified Frailty Index Independently Predicts Postoperative Pulmonary Infection in Elderly Patients Undergoing Radical Gastrectomy for Gastric Cancer

Introduction

The incidence of gastric cancer (GC) is gradually decreasing worldwide; however, it is relatively high in Asia.¹ In China, the incidence and mortality of GC both rank third among all the malignancies.² Until now, surgical resection is the mainstay treatment and only curative modality for GC.³ The high incidence of postoperative complications after radical resection is a major contributor to the poor prognosis,⁴ with an estimated incidence ranging from 12.0%⁵ to 34.4%.⁶ Pulmonary infection is one of the most common postoperative complications after radical gastrectomy, and it is associated with a prolonged hospital stay, and increased morbidity and mortality rates.^7,8 Thus, it is essential to determine risk factors associated with pulmonary infection in elderly GC patients.

Frailty is widely recognized as an impaired physical response to stress and it is highly prevalent with increasing age.⁹ Moreover, frailty is associated with increased vulnerability to adverse health outcomes after surgical conditions, including prolonged hospitalization and increased mortality.¹⁰ Until now, no consensus for the definition of frailty has been made. The definitions of frailty varied in different studies, as synonymous with advanced old age,¹¹ comorbidity,¹² or disability.¹³ Increasingly, frailty is widely recognized as a biologic syndrome of cumulative declines of multiple physiologic systems, resulting in decreased energy reserve, cognition, physical ability, resistance to stressors, and increased vulnerability to adverse outcomes.^14,15 As reported by previous studies, frailty is defined as a clinically state of increased vulnerability, which is induced by aging–associated decline in function and reserve across multiple physiologic systems.¹⁶ Fried et al defined frailty as a condition meeting 3 of the 5 phenotypic criteria including unintentional weight loss, exhaustion, low physical activity, slowness, and weakness.¹⁰ However, the modified frailty index (mFI) containing 11 variables based on National Surgical Quality Improvement (NSQIP) is one of the most commonly used measurements.¹⁷ Recently, mFI has attracted a lot of attention due to the gradually increasing aging population. Previous studies have indicated mFI as a predictor for morbidity and mortality among various surgical cohorts.¹⁸ In addition, a recent study by McChesney et al identified that mFI could predict postoperative outcomes in patients undergoing radical pelvic surgery.¹⁹ Moreover, mFI was also an effective predictor of mortality in patients with brain tumor²⁰ and readmission after colorectal surgery for high-risk patients.²¹ However, the correlation between pulmonary infection and mFI remains unknown. This study was designed to investigate the predictive value of mFI for pulmonary infection in elderly GC subjects undergoing radical gastrectomy. This study focused on the association between frailty and pulmonary infection. Our results highlighted mFI as an independent risk factor for pulmonary infection in elderly GC patients. A recent single-center prospective cohort study by Aceto et al²² identified the predictive power of mFI for postoperative pulmonary complications in elderly subjects after major abdominal surgery, which was quite in accordance with our results. Collectively, our results provide a potential useful predictor for pulmonary infection in elderly.

Materials and Methods

Patients

This was a retrospective cohort study carried out at the Department of anesthesiology, Taizhou People’s Hospital with the approval of the ethics committee of our hospital. Elderly GC patients who were scheduled for elective radical gastrectomy were consecutively enrolled between January 2017 and December 2020. The inclusion criteria were as follows: 1) age between 65 and 85 years; 2) American Society of Anesthesiologists (ASA) grade II and III; 3) resectable GC with pathological support; 4) with a follow–up for at least postoperative 30 days; 5) with complete clinicopathologic data. The exclusion criteria were as follows: 1) with other malignancies, or distant metastasis; 2) with preoperative chemoradiotherapy; 3) combined abdominal and thoracic surgery; 4) underwent emergency surgeries due to complications; 5) lost to follow up. All the enrolled patients should provide signed informed consent.

Clinical Data Collection

All the enrolled patients underwent the same perioperative management by the same surgical and anesthetic group. Abdominal computed tomography (CT), gastroscope, X–ray or CT of the chest, and laboratory examination were routinely performed for patients prior to surgery. The gastrectomy and lymphadenectomy were performed based on the Japanese gastric cancer treatment guidelines.²³ Enrolled patients received second-generation cephalosporin for 3–5 days as peri-operative prophylactic antibiotic treatment.

Data were collected as follows: 1) baseline demographic data, including age, gender, body mass index (BMI), American Society of Anaesthesiologists (ASA) physical status, smoking status; 2) clinicopathological data, including the extent of gastrectomy, surgical approach, the extent of lymph node dissection, duration of operation, estimated blood loss, histopathological grade, pathological stage, intra-operative blood transfusion, intensive care unit (ICU) admission, neutrophil–lymphocyte ratio (NLR), controlling nutrition status (CONUT), and mFI; 3) laboratory data, including hemoglobin, white blood cell, albumin, creatinine, urea, C–reactive protein (CRP), and tumor necrosis factor (TNF–α). The histopathological grade and pathological stage were defined according to the 7th edition of the International Union Against Cancer (UICC) TNM classification. Laboratory examinations were performed using the obtained peripheral venous blood samples on the surgery day. As described by previous study,²⁴ CONUT was calculated using serum albumin, lymphocyte, and total cholesterol level. In addition, the cut–off value of CONUT was set at 3 based on the previous report.²⁴

Definitions and Outcomes

According to previous report,²⁵ mFI was evaluated using 11 preoperative items, including diabetes mellitus, congestive heart failure, hypertension requiring medication, transient ischemic attack or cerebrovascular accident, Functional status 2 (not independent), myocardial infarction, peripheral vascular disease or rest pain, cerebrovascular accident with neurological deficit, chronic obstructive pulmonary disease (COPD) or pneumonia, percutaneous coronary intervention (PCI), prior cardiac surgery, or angina, and impaired sensorium. One positive point was assigned for each of the 11 items (with the presence or history), and mFI was calculated by dividing positive points by 11. The primary observational endpoint was set as the development of pulmonary infection within postoperative 30 days. As described by previous reports, postoperative pulmonary infection was defined based on new or progressive radiographic infiltrate, bacterial sputum cultures, plus two or more of the following: antibiotic treatment, temperature >38°C, leukocytosis (WBC > 12 x10⁹/L) or leucopenia (WBC count <4 x10⁹/L), and/or purulent secretions.^26,27 Enrolled patients were categorized into pulmonary infection and non–pulmonary infection groups, based on the presence of pulmonary infection within postoperative 30 days.

Statistical Analysis

The statistical analysis was conducted using SPSS 22.0 (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 8.0 (GraphPad Inc., CA, USA). Continuous variables are expressed as mean with standard deviations (SD), while numerous variables as number (n) with percentage. Chi–square test, Fisher exact, Student’s t–test, and Mann Whitney U–test were used for data analyses as appropriate. ROC curves were plotted to evaluate the cut–off and predictive values of factors. Binary univariate and multivariate logistic regression analyses were employed to determine risk factors for postoperative pulmonary infection using the “Enter” method. Multicollinearity analysis was performed among potential risk factors in the logistic regression model. A two–sided P value <0.05 was considered statistically significant.

Results

As showcased in Figures 1, 405 elderly GC patients were initially enrolled according to our inclusion criteria. Thereafter, 59 were excluded following exclusion criteria and a total of 346 patients were finally included in data analysis. The mean age of all the patients was calculated to be 73.7 years and the majority of patients were males (67.9%, 235/346). Of the enrolled 346 patients, pulmonary infection was observed in 51 patients within postoperative 30 days, with an incidence of 14.7% (51/346). The demographics and clinical characteristics associated with pulmonary infection are summarized in Table 1. No significant differences between patients with or without postoperative pulmonary infection were observed in terms of gender, BMI, type of operation, reconstruction method, oesophageal reflux, extent of lymph node dissection, estimated blood loss, histopathological grade, pathological stage, and ICU admission (P > 0.05). Patients with an older age (P = 0.002), higher ASA physical status (P = 0.017), longer duration of operation (P = 0.014), and higher CONUT score (P = 0.023) were more likely to develop pulmonary infection after surgery. Moreover, patients with the presence of current smoking habits (P = 0.039), intra-operative blood transfusion (P = 0.020), and operation via laparotomy (P = 0.016) were significantly more susceptible to pulmonary infection. Besides, patients who developed pulmonary infection showed a significantly higher percentage of high mFI level (≥0.225) when comparing with those without pulmonary infection (P < 0.001). In addition, the most common positive items in mFI were diabetes mellitus and hypertension (see Figure 2).

Table 1 Demographics and Clinical Characteristics Associated with Pulmonary Infection in Elderly GC Patients After Radical Gastrectomy

Figure 1 The flow chart. Abbreviation: GC, gastric cancer.

Figure 2 Percentage of positive items of mFI in total patients. Abbreviations: mFI, modified frailty index; COPD, chronic obstructive pulmonary disease; PCI, percutaneous coronary intervention.

Laboratory data associated with pulmonary infection are displayed in Table 2. Patients with higher preoperative CRP (P = 0.012) and lower albumin levels (P = 0.016) were also associated with increased risks of pulmonary infection. There was no significant difference between the two groups concerning hemoglobin, white blood cell, creatinine, urea, and TNF–α (P > 0.05).

Table 2 Preoperative Laboratory Variables Associated with Pulmonary Infection in Elderly GC Patients After Radical Gastrectomy

To assess the predictive values of continuous variables for pulmonary infection, we generated ROC curves. As layout in Figure 3, age (AUC: 0.611, P = 0.0012), mFI (AUC: 0.770, P < 0.001), and CRP (AUC: 0.594, P = 0.032) were three predictors for pulmonary infection. Based on the cut–off values calculated by ROC curves, the five continuous variables were categorized into high and low expression groups. Subsequently, potential risk factors (P < 0.05 in Tables 1 and 2) were enrolled into the univariate logistic regression models. As shown by the forest plot in Figure 4, age (≥73.5 years), ASA grade III, current smoking habit, mFI (≥0.225), and CRP (≥8.0 mg/L) were six potential risk factors for pulmonary infection. These five factors with P < 0.05 in univariate analysis were further included in the multivariate analysis. The results indicated that preoperative mFI (≥0.225) was the only independent risk factor for pulmonary infection (OR: 2.72, 95% CI: 2.02–3.31, P = 0.011, see Figure 5).

Figure 3 Predictive values of factors for pulmonary infection in elderly GC patients by ROC curve analysis. (A) age; (B) duration of operation; (C) mFI; (D) albumin; and (E) CRP. Abbreviations: GC, gastric cancer; ROC, receiver operating characteristic; mFI, modified frailty index; AUC, area under the curve; CRP, C-reactive protein.

Figure 4 Forest plot of the univariate logistic regression analysis for pulmonary infection. Abbreviations: ASA, American Society of Anesthesiologists; NLR, neutrophil–lymphocyte ratio; CONUT, controlling nutrition status; mFI, modified frailty index; CRP, C-reactive protein; OR, odds ratio; CI, confidence interval.

Figure 5 Forest plot of the multivariate logistic regression analysis for pulmonary infection. Abbreviations: ASA, American Society of Anesthesiologists; mFI, modified frailty index; CRP, C-reactive protein; OR, odds ratio; CI, confidence interval.

Discussion

In this retrospective study including 346 elderly GC patients who underwent radical gastrectomy, the incidence of postoperative pulmonary infection was 14.7%, which was quite similar to the 13.4% by previously published report.²⁸ Patients who suffered from pulmonary infection were associated with a higher rate of intensive care and mortality, and longer hospital stay than those without pulmonary infection. Considering the adverse impact of pulmonary infection on post-operative recovery, perioperative management to decrease the incidence was strongly recommended. Although some previous studies have determined the incidence and risk factors for pulmonary infection, these findings were usually with limited patient sample sizes and variables, which may significantly affect the adequate accuracy. In our multivariate logistic analysis, mFI was firstly identified as the independent risk factor for postoperative pulmonary infection in elderly GC patients undergoing radical gastrectomy.

Recently, it has been noted that mFI containing 11 items is one of the most common tools to evaluate frailty and mFI can predict postoperative morbidity and mortality.^17,25 Studies have indicated that those patients with higher mFI are accompanied by significantly increased postoperative complications and mortality.²⁹ Additionally, mFI containing 11 items can be used to predict postoperative complications and mortality in ICU patients after colectomy.³⁰

Recently, a study by Osaki et al reveals that mFI can predict non-home discharge in elderly GC patients after gastrectomy.³¹ Until now, many studies have supported mFI as a predictor for postoperative short-term outcomes in patients who underwent vascular,³² abdominal,³³ head and neck surgery.³⁴ Nevertheless, mFI could be at least a useful adjunct to predict pulmonary infection in elderly GC patients undergoing radical gastrectomy. Without a doubt, major abdominal surgery is a great challenge for the physiological reserve of elderly patients²¹ and the result that higher mFI correlate with increased risk of pulmonary infection is not surprising. In recent years, the predictive role of mFI for postoperative complications has attracted a lot of attention.³⁵ A recent study by Chen et al has revealed mFI as a promising predictor for both postoperative delirium and delayed neurocognitive recovery in elderly subjects after elective total joint arthroplasty.³⁶ This present study also provided insight into the association between mFI and postoperative complications in elderly GC patients, which was in agreement with other prior studies.³⁷ The frailty is associated with impaired organ functions of the body, decreased ability of the immune system to protect organism against infections.³⁸ In addition, diabetes mellitus is an important contributor to the 11–item mFI and it is clearly associated with the increased susceptibility to infection due to various physiopathologic mechanisms.^39,40 These might be possible explanations for the predictive role of mFI for postoperative pulmonary infection. The most common positive item of mFI in this study was diabetes mellitus. Diabetes is recognized as a well–established risk factor for infection in the general population,⁴¹ due to impaired immunity, antioxidant systems, and increased susceptibility to bacterial infections.³⁹ In addition, another common item, preexisting COPD, was significantly associated with a greater risk of incident pneumonia in the elderly.⁴² These results also provided possible explanations for the role of mFI in postoperative pulmonary infection.

This study has some limitations to declare. First, this was a single-center study, and its retrospective nature was associated with some bias. For example, the inclusion of patients with other age ranges or immunosuppressive status after preoperative chemoradiotherapy may possibly affect the results. Moreover, dividing the total patients into development and validation groups may minimize the adverse impact of retrospective study and improve the generalizability of the conclusions. Second, the sample size was relatively small and a larger multi–center trial is required to verify our conclusion. It is also needed to use data from other centers to validate the conclusion. Third, the underlying involved mechanisms remain unclear. Moreover, whether other tools for measuring frailty will lead to the same conclusions remain unclear.

In conclusion, our study indicated that mFI was an independent risk factor for pulmonary infection in elderly GC patients. Moreover, preoperative mFI assessment has the potential to stratify risks and prevent complications. Those patients with higher mFI levels deserved more attention including strict monitoring and timely intervention strategies. In addition, preoperative aggressive controls of some comorbidities (positive items of mFI) may possibly be helpful for pulmonary infection prevention.