The prognostic factors in sepsis patients after operation of gastrointestinal tumors in ICU

Objectives: We studied the clinical proles and the prognostic factors in patients with sepsis after thegastrointestinal tumor surgery in ICU. Methods: We retrospectively screened patients who underwent the gastrointestinal tumor surgery at the Peking University Cancer Hospital from January 2015 to December 2019. Among them, 181 patients who were diagnosed with sepsis in ICU were enrolled in our study. Cox regression was performed for multivariate adjusted factor analyses. Results: The 90-day all-cause mortality rate was 11.1% in our study. The univariate analysis showed that BMI, shock within 48h after entering ICU, number of blood leukocytes, the ratio of lymphocytes to neutrophils, INR, creatinine, procalcitonin, lactic acid, oxygenation index, SOFA score within 24h after entering ICU, APACHE II score within 24h after entering ICU were statistically signicant. In multiple analysis, we found that BMI (cid:0) 20kg/m 2 was a protective factor, while lactic acid (cid:0) 3 mmol/L after entering ICU and APACHE II score (cid:0) 20 within 24h after entering ICU were independent risk factors for the prognosis of sepsis after the gastrointestinal tumor surgery in ICU. Conclusions: The 90-day all-cause mortality rate was 11.1% in our study. BMI (cid:0) 20kg/m 2 was a protective factor, while lactic acid (cid:0) 3 mmol/L after entering ICU and APACHE II score (cid:0) 20 within 24h after entering ICU were independent risk factors for the prognosis of sepsis after the gastrointestinal tumor surgery in ICU.


Background
Sepsis was a worldwide problem. It is estimated that there are 31.5 million patients with sepsis every year in the world, which causes about 5.3 million deaths every year [1] . Sepsis is associated with high mortality. About 14000 people died of its complications every day in the world. Sepsis can be caused by any types of infection. Pathogenic microorganisms include bacteria, fungi, viruses and parasites. As the previous de nition of sepsis (infection+SIRS) is too sensitive, the new de nition of sepsis is the lifethreatening organ dysfunction caused by the host's maladjusted response to infection. Organ dysfunction manifests that sepsis-related organ failure score was no less than two points [2] .
The most common postoperative infection of gastrointestinal tumor is abdominal infection. Early identi cation of infection, control of infection source, proper use of antibiotics and rapid resuscitation of critical patients are the cornerstone of abdominal infection management [3][4][5][6] . There are numerous factors affecting the prognosis of sepsis. It has been reported that the prognosis of sepsis is related to lactic acid, interleukin-6, PCT, CRP, HFABP and so on [7][8][9][10][11] . However, as the de nition describes, sepsis is a syndrome with extreme heterogeneity. In the past, there were various reports on mortality of sepsis, though, few studies on the prognosis of patients with sepsis after operation. The purpose of this study is to investigate the prognostic factors of patients with sepsis who were admitted to ICU after gastrointestinal operation.

Patients
From January 2015 to December 2019, a total of 1636 patients were admitted to ICU after surgery from the gastrointestinal tumor center of Peking University Cancer Hospital. According to the new de nition of sepsis, 181 patients diagnosed with sepsis were enrolled in this study. Exclusion criteria: patients were admitted to ICU for other reasons or did not have sepsis during ICU stay, or patients' sepsis occurred out of ICU stay.

Therapeutic method
For sepsis patients, we have active anti-infection treatment, sent etiology examination and tried to nd the source of infection. For patients with septic shock, we took the following measures: First, bestowed broad-spectrum antibiotics within one hour, and took the etiological examination before giving antibiotics. Second, 30 ml/kg of crystalloid uids was used for rapid volume resuscitation within three hours, and the initial target of mean arterial pressure (MAP) was 65 mmHg. If the MAP was not achieved the target within one hour, noradrenaline was used for increasing the blood pressure. Third, we looked for and controlled the source of infection actively. For patients with a de nite source of infection, we controlled the source of infection by minimally invasive drainage, surgical debridement, removal of a central venous catheter and sputum aspiration by bronchoscope, etc. To emphasize; we follow the guidelines for other treatment strategies [5,6] .

Data collection and follow-up
The clinical data and laboratory examination of the patients were collected as follows: age, body mass index (BMI), underlying diseases, the length of the rst operation, pathogen susceptibility test, antibiotics used, whether shock occurred within 48 hours after entering ICU, the number of blood leukocytes, lymphocyte percentage, neutrophil percentage, international standardized ratio (INR), activated partial thromboplastin time (APTT), albumin, creatinine, cardiac troponin I (TNI), procalcitonin (PCT), lactic acid (Lac), base excess (BE), oxygenation index (PaO2 / FiO2) after entering ICU, sequential organ failure (SOFA) score and acute physiological function and chronic health evaluation ( APACHE) II score within 24h after entering ICU. Unless otherwise stated, the rst test after entering ICU was used for analysis.
They were followed up to 90 days in the clinic or by phone call.

Statistical analyses
The data of continuous variables were statistically described by mean ± standard deviation. The noncontinuous variables were described by median (quartile 1 [Q1], quartile 3 [Q3]). The counting variables were described by numerical value (percentage). The Kaplan-Meier method was used to calculate the survival rate of patients and log-rank test was used for the univariate analysis. Cox regression was used for the multivariate adjusted analyses with forward LR method. Statistical analyses were carried out using SPSS version 24.0 and P values less than 0.05 (two-tailed) were considered signi cant.

Patient characteristics
According to the new de nition of sepsis, a total of 181 patients were diagnosed with sepsis and 86 of them were diagnosed with septic shock within 48 hours after entering ICU. There were 13 patients complicated by abdominal bleeding or gastrointestinal bleeding, 16 cases by deep-vein thrombosis, one case by cerebral infarction and one case by myocardial infarction. See Table 1 for baseline characteristics of patients.

The Univariate and multiple survival analyses
The univariate analysis is shown in Table 2. All the sepsis patients were followed up for 90 days; 20 patients died, and the 90-day all-cause mortality rate was 11.1%. Univariate analysis showed that there were statistically signi cant differences in BMI, shock within 48h after entering ICU, the number of blood leukocytes, the ratio of lymphocyte to neutrophil, INR, creatinine, PCT, Lac, BE, oxygenation index after entering ICU, SOFA score and APACHE II score within 24h after entering ICU. Among them, BMI more than 20 kg/m 2 was a protective factor; the others were risk factors. Especially, shock within 48h after entering ICU, INR, creatinine, Lac, oxygenation index after entering ICU, SOFA score and APACHE II score within 24h after entering ICU had P values less than 0.01. See Table 3 for multiple analyses. Those factors with P values less than 0.05 were enrolled in the Cox regression analysis. The results showed that BMI, lactic acid after entering ICU and APACHE II score within 24h after entering ICU were independent prognostic factors. The BMI 20kg/m 2 was a protective factor, while lactic acid 3 mmol/L after entering ICU and APACHE II score 20 within 24 hours after entering ICU were risk factors. The survival curves of these three factors are shown in Figure 1, gure 2 and gure 3.

Discussion
Sepsis is one of the most common causes of death in critically ill patients. At present, there are only a few studies focused on postoperative sepsis. This study focuses on sepsis after operation of gastrointestinal tumor. In this study, the mortality rate is lower than that of sepsis reported in the literature [12] , which may be related to the fact that most of the infection sources of the patients we selected are abdominal infection, and we can actively control the infection sources by multidisciplinary cooperation. In this study, 181 patients with sepsis who were admitted to ICU after operation of gastrointestinal tumor were analyzed retrospectively and we found that BMI 20kg/m 2 , lactic acid 3 mmol/L after entering ICU and APACHE II score 20 within 24h after entering ICU were independent pro gnostic factors.

Page 5/17
The World Health Organization classi ed BMI as follows: BMI 18.5kg/m 2 was underweight, 18.5≤BMI 25 kg/m 2 was normal weight, 25≤BMI 30kg/m 2 was overweight and BMI≥30kg/m 2 was obesity [13] . In this study, we found that patients with BMI 20kg/m 2 had a better prognosis than those with BMI≤20kg/m 2 , so we guessed that BMI 20kg/m 2 might be a protective factor. However, the number of patients in this study was limited, and we did not conduct a more detailed strati ed study. There were numerous reports about the relationship between BMI and the prognosis of sepsis, though the results were still controversial [14,15] . Matthaios PO et al. [16] found that the mortality of obese patients with sepsis increased signi cantly. But one recent meta-analysis divided sepsis patients into three groups: overweight (25 BMI≤30kg/m 2 ), obesity (30 BMI≤40kg/m 2 ) and morbid obesity (BMI 40kg/m 2 ). The results showed that the death risk of overweight patients with sepsis was reduced, while obesity and morbid obesity patients with sepsis did not increase the death risk. The reason for this controversy might be related to the distribution of adipose tissue. It was reported that the visceral fat (VAT) accumulation detected by CT scan was a risk factor for poor prognosis of sepsis. Sepsis patients with a high ratio of VAT area to the subcutaneous fat (SAT) area had an increased risk of death and organ damage [17] . In future, more detailed and rigorous studies should be designed to clarify the relationship between sepsis and BMI.
Lactic acid was constantly produced in metabolism and exercise, but its concentration generally did not rise. Only when the production of lactic acid was accelerated and lactate could not be removed in time, its concentration would increase. Generally speaking, when the energy of the tissue could not be satis ed by aerobic respiration, the tissue could not get enough oxygen or could not deal with oxygen fast enough, the concentration of lactic acid would rise. Hence, sepsis and septic shock guidelines used lactic acid as an indicator of tissue hypoperfusion and as a target for uid resuscitation [5,6] . Many studies had shown that lactate was an independent risk factor for sepsis prognosis [18][19][20] . In our study, it was also con rmed that the lactic acid 3 mmol/L after entering ICU was an independent risk factor for sepsis patients after the gastrointestinal tumor surgery.
There were many scoring systems for evaluating the severity of critical patients, such as SOFA score and APACHE II score [21][22][23] . APACHE II score was considered as the gold standard for risk assessment of critical patients in the past. Several studies con rmed that APACHE II score is an independent risk factor for the prognosis of sepsis patients [24,25] . In our study, we found that the SOFA score and APACHE II score within 24 hours after entering ICU were statistically signi cant in the univariate analysis, while the multivariate analysis showed that only APACHE II score 20 was an independent risk factor in this group.
However, there was evidence that APACHE II score might provide inaccurate information in some patients, for example, in patients with unconsciousness, the score might be too high [26] . Therefore, we need to increase the sample size to con rm this result in the future.
The limitations of this study should be referred. First, this study was a retrospective study and the subjects of this study were patients with sepsis who were admitted to ICU after operation of gastrointestinal tumor, so, whether the results could be extended to all sepsis populations remains to be con rmed. Second, patients with sepsis in the general ward were not included in this study, and most of these patients improved in our hospital. Therefore, the mortality of patients with sepsis after gastrointestinal surgery might be overestimated in our study. In the future, we will design prospective research to verify it. Third, there were several missing data, especially BNP, echocardiography, etc. So we could not accurately evaluate their impact on the prognosis of sepsis patients. Finally, the small sample size of this study increased the risk of type two error which made the study power limited. We hope that there will be more large-scale researches to con rm these results in the future.

Conclusions
The 90-day all-cause mortality rate was 11.1% in our study. BMI 20kg/m 2 was a protective factor, while lactic acid 3 mmol/L after entering ICU and APACHE II score 20 within 24 hours after entering ICU were independent risk factors for the prognosis of sepsis after the gastrointestinal tumor surgery in ICU.

Declarations Ethics approval and consent to participate
We received the approval of the Medical Ethical Committee of Peking University Cancer Hospital. All patients or their near kin in our study provided informed consent.

Consent for publication
Not applicable.

Availability of data and materials
Data will be shared after this study is published.

Funding
The authors received no speci c funding for this work.     Figure 1 Kaplan-Meier survival curves of BMI on overall survival at 90 days Kaplan-Meier survival curves of Lac on overall survival at 90 days