Prediction of hospital mortality after colorectal perforation surgery from inflammation-based prognostic scores

Background Inflammation-based prognostic scores have prognostic value in cancer or cardiovascular disease patients. This study evaluated the prognostic value of inflammation-based prognostic scores in colorectal perforation patients. Methods Data of 97 patients who underwent surgery for colorectal perforation were reviewed. We calculated various inflammation-based prognostic scores and analyzed the relationship between inflammation-based prognostic score and hospital mortality due to colorectal perforation. Results Multivariate analyses of hospital mortality revealed neutrophil–lymphocyte ratio (P = .0021), C-reactive protein/albumin ratio (P = .0224), and prognostic nutritional index (P = .0078) as independent predictive factors. The Kaplan–Meier analysis showed that patients who met all of the following parameters avoided hospital death: neutrophil–lymphocyte ratio < 30, prognostic nutritional index ≥ 27.2, age < 75 years, and perforation of the left colon. Conclusion Neutrophil–lymphocyte ratio, C-reactive protein/albumin ratio, and prognostic nutritional index were superior to other inflammation-based prognostic scores in predicting mortality of colorectal perforation. Neutrophil–lymphocyte ratio, prognostic nutritional index, patient's age, and sidedness of the perforation site may be useful parameters to identify subgroups in which a favorable prognosis can be expected.


INTRODUCTION
Colorectal perforation has an extremely poor prognosis due to generalized peritonitis and septic shock. As a result of this condition, disseminated intravascular coagulation (DIC) can arise leading to multiple organ failure, resulting in the high mortality rates [1]. In many cases, surgical approaches, including extensive drainage and perioperative intensive treatment with antibiotics, inotropes, polymyxin-B direct hemoperfusion, and respiratory support, are required [2,3]. Despite such aggressive therapy, few patients survive. To improve the prognosis, it is important to assess the condition of patients, choose appropriate surgical procedures, and discern the necessity of perioperative treatment. Previous studies have evaluated several indicators for predicting the prognosis of colorectal perforation [3][4][5][6][7]. Kayano et al showed that a low psoas muscle index is a poor prognostic factor for lower gastrointestinal perforation [3]. Yamamoto et al showed that a high patient age and low preoperative systolic blood pressure were independent risk factors for mortality in patients with colorectal perforation [4]. In addition, scoring systems such as the Acute Physiology and Chronic Health Evaluation II score, physiological and operative severity score for the quantification of mortality and morbidity score, and the Sepsis-related Organ Failure Assessment score were used to investigate the prognostic value of patients who underwent surgery for colorectal perforation [5][6][7].
On the other hand, inflammation-based prognostic scores (IBPSs), which reflect malnutrition and systemic inflammatory responses, have been reported to predict outcomes in patients with several clinical conditions such as malignant tumors [8][9][10][11][12][13][14][15] and cardiovascular diseases [16,17]. Typical IBPSs include the neutrophil-lymphocyte ratio (NLR), platelet-lymphocyte ratio (PLR), C-reactive protein/albumin ratio (CAR), prognostic nutritional index (PNI), Glasgow Prognostic Score (GPS), and prognostic index (PI). In addition, in patients with gastrointestinal perforation, IBPSs may reflect the degree of peritonitis and systemic inflammation, the time course from the onset, and the nutritional status and may be associated with postoperative prognosis. However, few studies have examined the relationship between the various IBPSs and gastrointestinal perforation. To our knowledge, only one study has demonstrated that NLR and PLR are superior to other IBPSs in predicting the mortality of patients with gastrointestinal perforation [18]. That considered, there are several differences in clinical condition, surgical stress, and survival rates between perforation of the upper and lower gastrointestinal tract. Considering the variation of perforation sites in this study, including the stomach, small intestine, colon, and appendix, we specifically analyzed patients who underwent surgery for colorectal perforation and evaluated the prognostic value of IBPSs such as NLR, PLR, CAR, PNI, GPS, and PI. Moreover, we suggested a new score based on IBPSs and routinely available parameters to identify subgroups in which a favorable prognosis can be expected following surgery.

METHODS
Patients. In this retrospective, single-center, cohort study, we reviewed the data of 105 patients who underwent emergency surgery for colorectal perforation at our institute between August 2010 and December 2019. Among the 105 patients, 8 patients whose clinical data were incomplete were excluded. Finally, data of 97 patients who underwent surgery for colorectal perforation were eligible for analysis. The patients were divided into two groups according to hospital mortality: survivors and nonsurvivors.
Permission to perform this study was provided by the Institutional Review Board of the National Kyushu Medical Center (20C033). The study conforms to the provisions of the Declaration of Helsinki (as revised in Fortaleza, Brazil, October 2013), available at https://www. wma.net/what-we-do/medical-ethics/declaration-of-helsinki/.
IBPSs. The NLR, PLR, CAR, PNI, GPS, and PI were calculated from the patient records. Baseline blood data were obtained by collecting blood from the peripheral vein of each patient before surgery. NLR and PLR were defined as absolute neutrophil count and platelet count, respectively, divided by the absolute lymphocyte count [10,11]. CAR was defined as the serum CRP level divided by the serum albumin level [12]. The PNI was calculated using the following formula: 10 × serum albumin (g/dL) + 0.005 × total lymphocyte count (per mm 3 ) [13]. The GPS was calculated by CRP and albumin using standard thresholds (>1.0 mg/dL for CRP and <3.5 g/dL for albumin) [14]. Calculation of the PI was based on the CRP level and white blood cell count. The upper limits of reference ranges for the CRP level (0.1 mg/dL) and the white blood cell count (11,000/mm 3 ) were used as cutoff values [15]. The PI was considered as 0 if both values were less than the cutoff values, and the PI was 1 if one of the two markers was elevated (Table 1). Receiver operating characteristic (ROC) curve analysis was used to identify the optimal cutoff values of these IBPSs (Fig. 1).
Hospital Mortality and IBPSs. The cause of colorectal perforation, especially the presence of cancer, affects long-term outcomes. Therefore, we discerned that hospital mortality was the most suitable parameter for evaluating the postoperative prognosis of colorectal perforation and applied it for prognosis analysis.
To evaluate the relationship between hospital mortality of colorectal perforation and each biomarker, including IBPSs, univariate and multivariate analyses with a Cox proportional hazard model were performed.
Statistical Analysis. Differences in characteristics between the groups were evaluated using Fisher's Exact Test or unpaired t test. Survival curves were plotted according to the Kaplan-Meier method, and differences were analyzed using the log-rank test. Univariate and multivariate analyses were performed using a Cox proportional hazard model to identify independent prognostic factors. All P values were 2-sided. ROC curve analysis was used to determine the predictive value of the combined index. All analyses were performed using JMP PRO 11 software (SAS Institute Inc, https://www.jmp.com/ja_jp/home.html). Data are presented as number (%), unless otherwise stated. C, cecum; A, ascending colon; T, transverse colon; D, descending colon; S, sigmoid colon; R, rectum; ESD, endoscopic submucosal dissection; EMR, endoscopic mucosal resection.  Operative Outcomes. Operative outcomes were compared between survivors and nonsurvivors ( Table 2). There were no significant differences in mean operative times (150.0 vs 143.1 minutes, P = .6673) and intraoperative blood loss (122.9 vs 99.9 mL, P = .6996) between the 2 groups. The degree of postoperative complications was categorized according to the Clavien-Dindo (C-D) classification [19,20]. The occurrence of all C-D grades of postoperative complications was higher in nonsurvivors than in survivors (survivors, 48.2% vs nonsurvivors, 78.6%; P = .0445). The ratio of severe complications (C-D grade ≥ IIIa) was more remarkable in nonsurvivors (survivors, 13.3% vs nonsurvivors, 71.4%; P < .0001). The median survival time of nonsurvivors was 18.5 days.
New Scoring. When patients were categorized into survivors and nonsurvivors, there were significant differences in age, sidedness of the perforation site, and mean values of NLR and PNI (Table 1). Based on these results, we regarded NLR ≥ 30, PNI < 27.2, age ≥ 75 years, and perforation of the right side of the colon as risk factors for predicting prognosis and established a new score by combining these 4 parameters ( Table 4). The Kaplan-Meier curves according to the new score showed that the survival rates of patients with higher scores were significantly poor, whereas 100% of patients who met all of the following 4 parameters avoided hospital death: NLR < 30, PNI ≥ 27.2, age < 75 years, and perforation of the left colon (Fig 3, A). However, when any one of PNI, age, or sidedness was lacking from the new score, survival rates were not 100% even in patients with a score of 0 (Fig 3, B-E).

DISCUSSION
In this study, we evaluated the prognostic significance of various IBPSs in patients who underwent surgery for colorectal perforation and demonstrated that NLR, CAR, and PNI were superior to the other IBPSs in predicting mortality. Moreover, patients with NLR < 30, PNI ≥ 27.2, age < 75 years, and perforation of the left colon were identified as the subgroup in which a better prognosis could be expected.
IBPSs have been reported as predictive factors for the prognosis of various malignant tumors in previous studies. These biomarkers can be evaluated easily by collecting blood from a peripheral vein, and the methods of calculation are simple. IBPSs reflect systemic inflammatory response, immunosuppression, malnutrition, or cachexia due to tumor progression [21]. In addition, a small number of studies have reported that several IBPSs were correlated with prognosis in cases of nonmalignant diseases such as acute pancreatitis, pulmonary embolism, and infective endocarditis [22][23][24]. In these studies, NLR and PNI were mainly applied as predictive factors, which were consistent with our results. NLR is elevated by reflecting neutrophilia and lymphopenia with systemic inflammation. The neutrophil count increases rapidly during acute systemic inflammation, which may occur due to peritonitis and sepsis. In contrast, lymphopenia reflects immunosuppression. Colorectal perforation can cause peritonitis, intra-abdominal abscesses, and septic shock. These complications affect the failure to rescue patients. The condition of immunosuppression reflecting lymphopenia may  be a lethal factor when these complications occur. Therefore, NLR may be an important prognostic factor not only in the condition of malignant diseases but also in acute systemic inflammation or infectious diseases. The PNI value is determined by the serum albumin and total lymphocyte count. Serum albumin level is an indicator of nutritional status. Hypoalbuminemia may occur due to a systemic inflammatory response. Several studies have compared the validity of multiple IBPSs to predict the prognosis of gastrointestinal malignancies [9,21,25,26], and most reports indicated that PNI was a useful prognostic factor that was superior to other IBPSs. Many studies, including ours, suggest that PNI is a versatile prognostic factor, and a low PNI value indicates an unfavorable condition in various diseases.
In the present study, several routinely available parameters other than IBPS were also compared between survivors and nonsurvivors. We focused on patients' age and laterality of the perforation site because the mean age and the incidence of perforation in the right-side colon were significantly higher in nonsurvivors (P = .0312 and .0370, respectively; Table 1). It is presumed that patients of higher age are likely to have more comorbidities that affect the postoperative recovery and they have fewer reserve capacities to withstand peritonitis and surgical stress than younger patients. Regarding laterality, it is known that the gut microbiota between the right and left colon is different. Prevotella, Selenomonas, and Peptostreptococcus were identified in relatively higher abundances in right-side colon, while Fusobacterium, Escherichia/Shigella, and Leptotrichia were relatively abundant in left-side colon among colorectal cancer patients [27]. However, the relationship between these differences in the gut microbiota and the prognosis of colorectal perforation remains unresolved. One conceivable reason why right-sidedness is associated with a poorer prognosis is that perforation of the right colon leads to a higher risk of widespread peritonitis due to unconsolidated stool than perforation of the left colon and rectum. Although only 1 or some IBPSs cannot perfectly predict survival in patients with colorectal perforation, our analysis summarized in Figure 3 indicates that the addition of these 2 parameters, namely, patient's age and sidedness of the colonic perforation, may be able to identify subgroups with better prognosis. In the clinical setting, this new score may be useful for accurate prediction of fatal disease. This would help guide the decision on the need for earlier hospice care after surgery and may aid in determining the specific guidance and counseling offered to patients and their families.
However, this study has a potential limitation. This was a retrospective, single-institution study. However, colorectal perforation occurs suddenly and usually requires emergency surgery. Therefore, designing a prospective study on colorectal perforation is difficult. Few reports have demonstrated the relationship between IBPS and outcomes after surgery for colorectal perforation. Thus, the accumulation of findings from retrospective studies is valuable.
In conclusion, in various inflammation-based prognostic scores, NLR, CAR, and PNI were superior to other IBPSs in predicting mortality in patients who underwent surgery for colorectal perforation. Patients with NLR < 30, PNI ≥ 27.2, age < 75 years, and perforation of the left colon  Table 4. New score consisting of (A) all 4 parameters, (B) 3 parameters except sidedness, (C) 3 parameters except age, (D) 3 parameters except PNI, and (E) 3 parameters except NLR, respectively.
were identified as subgroups in which a favorable prognosis could be expected.

Acknowledgments
We thank the Kaibara Morikazu Medical Science Promotion Foundation, the Uehara Memorial Foundation, the Clinical Research Promotion Foundation, and Fukuoka Public Health Promotion Organization Cancer Research Fund for financial support. We thank Editage Group (https://www.editage.jp/) for editing the draft of this manuscript.

Human Rights Statement and Informed Consent
All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions. Informed consent to be included in the study or equivalent was obtained from all patients.

Author Contribution
KK designed the study. KK, EK, SN, YT, and RY acquired the data. Analysis and interpretation of data: KK analyzed the data. KK performed statistical analyses. KK and TK drafted the article. TK and YS revised the article critically for important intellectual content. TK, YS, and KI approved the final version of manuscript. All authors have read and approved the manuscript.