Neutrophil to lymphocyte ratio and platelet to lymphocyte ratio are superior to other inflammation-based prognostic scores in predicting the mortality of patients with gastrointestinal perforation

Background The neutrophil to lymphocyte ratio (NLR) is gaining interest as an independent predictor of survival in patients with various clinical conditions. No study to date has reported an association between inflammation-based prognostic scores, including the Glasgow Prognostic Score (GPS), NLR, platelet to lymphocyte ratio (PLR), Prognostic Nutritional Index (PNI), and Prognostic Index (PI), and mortality in patients with gastrointestinal perforation (GIP). We compared the prognostic value of these measures. Findings A total of 32 patients with GIP were retrospectively enrolled. Patients were assessed according to the GPS, NLR, PLR, PI, and PNI. Multivariate analyses were performed to identify variables associated with mortality. Receiver operating characteristic (ROC) analyses were also performed. Overall survival rates (in-hospital mortality) were calculated using the Kaplan–Meier method, and differences in survival rates between groups were compared by the log-rank test. Multivariate analysis of significant variables revealed NLR (HR 1.257, 95% CI 1.035–1.527, P = 0.021) and PLR (HR 1.004, 95% CI 1.001–1.007, P = 0.016) at the time of admission to the intensive care unit to be independently associated with in-hospital mortality. AUC analysis revealed Sequential Organ Failure Assessment-Glasgow Coma Scale (SOFA-GCS) (0.73) to be superior to NLR (0.57) and PLR (0.58) for predicting mortality, and a high SOFA-GCS score was associated with reduced overall survival (P < 0.05). Conclusions NLR and PLR were superior to other inflammation-based prognostic scores in predicting the mortality of patients with GIP.


Introduction
The neutrophil to lymphocyte ratio (NLR) has gained interest as an independent predictor of survival in patients with various clinical conditions, ranging from oncological to cardiovascular diseases. NLR has also been reported to predict bacteremia better than other infection markers [1], and an NLR > 7 was reportedly an independent marker of mortality in patients with bacteremia [2].
Gastrointestinal perforation (GIP) is a life-threatening disease with a high mortality rate; GIP often leads to shock and usually requires active rescue in the intensive care unit (ICU) and emergency laparotomy [3]. No previous study has reported an association between inflammation-based prognostic scores and outcomes in patients with GIP.
We hypothesized that NLR measured at the time of admission to the ICU may better predict in-hospital mortality in patients with GIP, as compared with other inflammationbased prognostic scores. To test this hypothesis, we compared the prognostic value of various inflammation-based prognostic scores in patients with GIP.

Methods
We conducted a single-center retrospective study in a 16-bed ICU. The study protocol was approved by the Ethics Committee of Osaka Medical College (Osaka, Japan). A total of 40 patients diagnosed with GIP, who underwent surgery and were treated in the ICU of Osaka Medical College Hospital between January 2014 and June 2016, were retrospectively enrolled. Of these, 32 patients were evaluated, excluding those who were aged 18 years or younger; who were pregnant; who had immunosuppressive disease (e.g., HIV), or were undergoing immunosuppressive therapy (e.g., chemotherapy, chronic use of steroids, autoimmune disease treatment) within 1 month of the study; and who had cardiac arrest at the time of ICU admission. Individual patient consent was not obtained since all data used in this study were acquired retrospectively from the laboratory information system without any additional blood sampling or laboratory analysis. The main outcome measure was inhospital mortality. The following demographic and clinical data were collected: age, sex, comorbidities (cancer, coronary artery disease, diabetes mellitus, hypertension, and renal disease), Sequential Organ Failure Assessment-Glasgow Coma Scale (SOFA-GCS) score at ICU day 1, length of hospital stay, and in-hospital mortality. Since our study population included intubated patients under sedation with propofol and/or dexmedetomidine at ICU admission, we excluded the GCS item from the SOFA score. Blood samples were obtained upon ICU admission for measurements of CRP, albumin, white blood cell count, neutrophil count, lymphocyte count, and platelet count. The GPS, NLR, PLR, PI, and PNI were obtained as shown in Table 1.
Descriptive analysis was performed for all variables. Continuous variables were expressed as median (interquartile range), and categorical variables as counts (percentage). Patient characteristics were compared between survivors and non-survivors using Fischer's exact test. Univariate analysis and multivariate analysis (Cox proportional hazards model) were used to examine associations between patient characteristics and prognostic factors. Analyses using Cox proportional hazards models were performed by forward selection of variables which were found to be significant by univariate analysis and inflammation-based prognostic scores. Receiver operating characteristics (ROC) curves were generated for variables which were significant in the multivariate analysis, and areas under the curve (AUCs), cutoff values, sensitivities, specificities, and predictive values were calculated. Using these cutoff values, overall survival rates (in-hospital mortality) were calculated with the Kaplan-Meier method, and differences in survival rates between groups were compared by the log-rank test. A P value < Table 1 Inflammation-based prognostic scores CRP (≤ 10 mg l −1 ) and white blood cell count (≤ 11 × 10 9 l −1 ) 0 CRP (≤ 10 mg l −1 ) and white blood cell count (> 11 × 10 9 l −1 ) 1 CRP (> 10 mg l −1 ) and white blood cell count (≤ 11 × 10 9 l −1 ) 1 CRP (>10 mg l −1 ) and white blood cell count (>11 × 10 9 l −1 ) 2

Discussion
NLR and PLR were found to be superior to other inflammation-based prognostic scores in predicting the mortality of patients with GIP. NLR and PLR are based primarily on the physiological link between neutrophilia and lymphopenia with systemic inflammation. Jilma et al. [4] studied changes in white blood cell types after inflammation and reported a 300% increase in circulating neutrophils, 96% decrease in monocytes, and 85% decrease in lymphocytes 4 to 6 h after inflammation. Below, we discuss the literature surrounding NLR and the prognostic capabilities of NLR for GIP. Growing evidence suggests the usefulness of NLR in the prediction of survival in various contexts, such as lung cancer, colorectal cancer, orthotopic liver transplantation for primary hepatocellular carcinoma, postoperative coronary artery bypass grafting, chronic heart failure, pulmonary emboli, and acute pancreatitis [1,5,6]. Moreover, NLR was a more sensitive parameter than increased white blood cell count in patients with suspected appendicitis [7]. These data suggest the importance of NLR in multiple patient populations.
In the present study, NLR and PLR had a positive predictive value of 38.5% and a negative predictive value of 84.2%, suggesting that NLR and PLR may be more useful for ruling out mortality, rather than predicting it. NLR and PLR can be obtained easily, cheaply, and rapidly and can provide relevant information for necessary interventions within the first few hours of hospital admission. As discussed earlier, studies have shown that NLR predicts bacteremia better than other infection markers [1] and an NLR > 7 was reportedly an independent predictor of mortality in patients with bacteremia [2]. In another study, the initial NLR   Fig. 2 Kaplan-Meier survival curves for inflammation-based prognostic scores. a NLR. b PLR. c SOFA-GCS measured at ED admission was independently associated with 28-day mortality in patients with severe sepsis and septic shock [8]. However, these previous studies did not assess associations between inflammation-based prognostic scores, including the GPS, NLR, PLR, PNI, and PI, and mortality in patients with GIP. Our results are informative in this respect.
This study has a potential limitation. Given the retrospective, single-center design of the study and small cohort, multivariate analysis may be difficult to apply. A large-scale prospective validation study will be needed to confirm our results.

Conclusion
NLR and PLR were superior to other inflammationbased prognostic scores in predicting the mortality of patients with GIP.