Sepsis manifests as a dysregulated immune response to infection (documented or suspected), leading to organ dysfunction. Timely assessment for patients with potentially poor prognosis could warn clinicians about the ongoing pathological process early and to give appropriate care and treatment in time, such as providing intensive care (treatment and monitoring) or referral to another hospital, which could optimize the allocation of limited health care resources. The host response biomarkers play a key role in the early diagnosis, early risk stratification, and early intervention [32]. A systematic review in 2020 summarized 258 identified markers of sepsis, 100 of which were prognostic [2]. NLR is a cheap, simple, rapid and easily available inflammation-related parameter. It has been confirmed that NLR had a high sensitivity for the diagnosis and stratification of sepsis with a good predictive and prognostic value in sepsis [33–35].
Whether NLR is a predictor of poor prognosis in sepsis remains controversy [36]. A retrospective study in 2019 showed no significant association of NLR at admission of emergency department with 28-day mortality in sepsis [19]. The cut-off value of NLR for prediction also various in different studies (Supplementary Table 1). However, several reviews have mentioned the association of a NLR with mortality in sepsis [6, 32, 37], including sepsis due to viral infection [38]. Roman Zahorec proposed that dynamic monitoring of NLR should be done daily to observe its dynamic changes in critical illness [6]. Moreover, he suggested that dramatic increasing of NLR values above 11 indicates continued deterioration of the condition, whereas improving the clinical course of sepsis, lower risk of mortality are associated with decline of NLR values below 7 [6]. Therefore, the dynamic changes of NLR in early sepsis reflecting the intensity of immune-inflammatory reaction are very important. As far as we know, our present study is the first to analyze the predictive value of NLR in in-hospital mortality of septic patients on first three consecutive days and the seventh day after admission. In our analyses at four time points, we found that NLR could be used as a risk factor for in-hospital mortality in septic patients with high sensitivity and specificity, and its cut-off values were 10.769 on day 1, 17.5435 on day 2, 14.3954 on day 3, and 9.1047 on day 7 after admission.
In addition to NLR, PLR and MLR have been reported to be associated with sepsis mortality [39, 40]. However, in our analysis, PLR or MLR at admission is less reliable than NLR or NE% in predicting in-hospital mortality, which is consistent with the findings of Djordjevic et al. in 2018 [41]. Additionally, we found that NE% was also highly correlated with the in-hospital mortality of sepsis. The AUC values were 0.852, 0.790, 0.777 and 0.741, and the cut off values were 87.70, 90.69, 85.00 and 83.93 on day 1, 2, 3 and 7 after admission, respectively. In our univariate Cox proportional hazards model, we did not find a correlation between NE% on day 7 and in-hospital mortality. We speculated that it might be related to the massive missing data on day 7, or it might be related to the fact that the late stage of sepsis was mainly characterized by severe immunosuppression rather than excessive inflammatory response [42].
In the early stage of sepsis, a large number of immune cells are activated, including neutrophils, monocytes, natural killer cells, dendritic cells and lymphocytes [43]. Neutrophils are fundamental components of innate immunity and are essential for microbial eradication, they could be activated in a few minutes after the pathogen invades [44]. In the early stage of sepsis, the number of neutrophils increases rapidly owing to delayed neutrophil apoptosis [43]. Increased circulating neutrophils lead to a dysregulated immune response by releasing inflammatory cytokines and reactive oxygen species at sites distal to the infectious site, contributing to multiple organ failure. As for lymphocyte, it is also actively involved in the immune response in the early stage of sepsis, but the number of lymphocytes is far lower than that of neutrophils, accounting for 20–40% of the WBC in the blood. Uncontrolled apoptosis of T lymphocytes-induced decrease in the absolute counts is critical in the pathogenesis of sepsis [45, 46]. B-cell lymphoma 2 and programmed cell death protein 1, interleukin (IL)-15 and IL-7 are shown to be involved in the apoptosis of T lymphocytes [47]. Excessive apoptosis of B cells in sepsis has also been observed [48]. NLR has been shown to be capable of reflecting the intensity of critical immune-inflammatory reaction [6]. In addition, evidence has shown that dynamic changes of NLR precede the clinical status for several hours and may warn early recognition and timely intervention targeting the ongoing pathological process [6]. Therefore, the increase of NLR in the early stage of sepsis could play an important role in the early diagnosis and prognosis of sepsis [49]. With the continuous clinical improvement of sepsis, the decreased neutrophils and the increased lymphocytes in the peripheral blood could cause NLR decrease continuously [6, 50].
Our study has some limitations. First, this is a retrospective, single-center study, which cannot predict future events and may have some inherent biases. Second, NLR data were missing to a certain extent on the second, third and seventh day after admission, especially on the seventh day after admission, this might have led to some decrease in the accuracy of our analysis. Therefore, large scale, prospective, multicenter and high-quality clinical studies that continuously monitor the dynamic changes of NLR at an early stage of sepsis are needed to be conducted to further clarify the role of NLR in early prediction of the clinical prognosis of sepsis.