The predictive value of red blood cell distribution width and platelet-to -lymphocyte ratio for acute kidney injury in critically ill patients

Red blood cell distribution width (RDW) and the platelet-to-lymphocyte ratio (PLR) are associated with different types of prognoses in critically ill patients. But, the value of RDW and PLR in predicting the occurrence of acute kidney injury (AKI) in critically ill patients are unknown. The purpose of the study was to explore the associations of RDW and PLR with AKI incidence. we examined the associations using we


Introduction
Acute kidney injury (AKI) is a frequent and serious clinical condition in intensive care unit (ICU) patients, with an incidence of 20-70% [1][2][3] . It is associated with increased mortality (approximately 40%) and a high nancial burden and even affects long-term mortality [3][4][5] . In recent decades, research has suggested that early detection of AKI would be bene cial in predicting the prognosis of and providing more effective care for patients with AKI 6,7 . Unfortunately, the clinical prediction of AKI progression remains a challenge.
Several novel serum and urinary kidney injury biomarkers that can be detected to provide an early diagnosis of AKI have been discovered; however, most of these biomarkers require well-controlled conditions and are applicable in only restricted populations. Therefore, they are not widely used currently 2,5 .
In ammation and the immune response play important roles in the pathogenesis of AKI 8, 9 . Increasing evidence suggests that the platelet-to-lymphocyte ratio (PLR) and neutrophil-to-lymphocyte ratio (NLR), which are based on complete blood cell counts, are in ammation-associated parameters and are predictors of the development of AKI [10][11][12] . Clinically, red blood cell distribution width (RDW) is an indicator of anaemia, but previous studies have revealed that it is correlated with the outcome of AKI [13][14] . Zhu et al 14 and Wang et al 15 reported the predictive value of RDW in mortality: a higher RDW related to a higher risk of morality. However, Elhosseiny et al 16 demonstrated that RDW was not correlated with the occurrence of contrast-induced AKI in patients with acute coronary syndrome. However, little attention was taken to the role of RDW in the prediction of AKI in ICU. Therefore, we considered the correlation with RDW and the occurrence of AKI in critically ill patients to elucidate the predictive value of RDW for early diagnosis in AKI patients. In addition, we explored the relationship of RDW and PLR with in-hospital mortality.

| Patient cohort
The study population comprised all patients who attended to the ICU in the First A liated Hospital of the University of South China between 1st January 2016 and 31st December 2019.AKI diagnosis was based on the classi cation of the Kidney Disease: Improving Global Outcomes (KDIGO) . 17 The lowest value of serum creatinine(Scr) detected in the emergency clinic or general ward before admission to the ICU was considered the baseline creatinine value. When this value could not be obtained, the Modi cation of Diet in Renal Disease (MDRD) calculation was performed, assuming that the normal glomerular ltration rate(GFR) was 75 ml·min -1 ·1.73 m -2 . 18 Patients were excluded if they had received a diagnosis of chronic renal insu ciency; had undergone kidney transplantation; were younger than 18 years of age; were readmitted to the ICU; were admitted to the ICU for less than 24 hours; had undergone dialysis within 1 month before admission or were undergoing dialysis when admitted to the ICU; or had no renal function or routine blood test data within 2 days after admission to the ICU.

| Data Extraction
Patient demographic, complete blood count, blood biochemistry, in ammatory marker, Acute Physiology and Chronic Health Evaluation II (APACHE II) score and primary disease data were recorded for each patient. The baseline characteristic within 24 hours of ICU admissions and complete blood count data were recorded also. The PLR was calculated by dividing the absolute platelet count by the absolute lymphocyte count. The development of AKI was used as the primary endpoint in the study, and all-cause hospital mortality was the secondary endpoint.

| Statistical Analysis
Complete statistical analysis was conducted with SPSS 16 software (Chicago, IL, USA). Two tailed p<0.05 signi ed statistical signi cance for all the analyses.In case of continuous data, variables were given as medians with interquartile range and categorical variables were shown as frequency counts (%).
Comparisons between groups were computed with the test of chi-square. The association of RDW with other variables was assessed with Spearman's correlation analysis. Multivariable logical regression was used to examine the associations of RDW and the PLR with the development of AKI and hospital mortality.To determine the diagnostic ability of RDW and PLR for AKI occurrence and in-hospital mortality,we generated receiver operating characteristic (ROC) curves.Cut-off values with sensitivity and speci city of variables were determined by Youden index.

| Subject characteristics
After reviewing the records of 2435 eligible subjects who were attended to the ICU of the First A liated Hospital of the University of South China from 1 January 2016 to 31 December 2019.The nal cohort comprised 1500 critically ill patients in the analysis (Fig 1). The subject characteristics of critically ill patients are presented in Table 1. Most patients were men (63.4%); 615 patients (41%) were classi ed in the AKI group, and 885 patients (59%) were allocated to the non-AKI group. The number of patients with stage 1 AKI was 159 (25.9%), with stage 2 AKI was184 (29.9%) and with stage 3 AKI was 272 (44.2%).
The AKI group with a remarkably higher RDW than non-AKI group. Compared with non -AKI patients, the number of older people was larger and the prevalence rates of hypertension, diabetes mellitus, and coronary artery disease were higher in the AKI patients. The AKI patients had remarkably higher PLRs, cystatin C (CystC), Scr, blood urea nitrogen (BUN),C-reactive protein (CRP), and procalcitonin (PCT) levels and APACHE II scores than the non-AKI group. However, haemoglobin was lower in AKI group. Notably, no statistical differences was observed between the AKI group and non-AKI group in the prevalence of chronic obstructive pulmonary disease (COPD) or albumin level and white blood cells(WBC)(Table1).  (Table 3), and those in the PLR-high group had a 9.109-fold higher risk of developing AKI t (OR= 9.109, 95% CI:7.089-11.705)han those in the PLR-low group. Moreover, CystC, CRP, PCT and triglyceride levels and the APACHE II score were identi ed as potential risk factors for the development of AKI.

RDW and PLR predicted the occurrence of AKI
The ROC curve was used to assess the discriminative ability of RDW in predicting AKI in ICU compared with other in ammation index parameters (including the PLR, CRP, PCT and CystC). The AUC for AKI development based on RDW was 0.728 (95% CI:0.702-0.754), and the optimal cut-off value was 14.045; with the sensitivity was 73.3%, and the speci city was 65.5%, which were higher than those for PCT, CRP and the APACHE II score (Fig 2a, Table 4). The AUC for AKI development based on the PLR was 0.780 (95% CI: 0.755-0.804), with a cut-off value of 172.067 (sensitivity: 77.1%, speci city: 73.4%). However, CystC was associated with the highest AUC, at 0.821 (95% CI: 0.800-0.843) (Fig 2b, Table 4).

Risk of in-hospital mortality predicted by multi-variable logical regression
After adjustment for male, age, diabetes mellitus, hypertension, COPD, coronary artery disease, baseline Scr and BUN, the adjusted ORs for in-hospital mortality based on RDW was 1.202 (1.136-1.271); there was no relationship between the PLR and in-hospital mortality. Therefore, we divided the patients into two subgroups for each parameter: RDW-low and RDW-high and PLR-low and PLR-high. The adjusted ORs for in-hospital mortality in the RDW-high group were 2.907 and 1.534, respectively. The adjusted ORs for inhospital mortality based on WBC count, CystC, and the APACHE II score were 1.016, 1.128 and 1.095, respectively (P<0.05). However, albumin, haemoglobin and triglycerides unrelated to mortality(P>0.05).
(Table5)  :0.673-0.738). PLR, CRP and PCT did not show a clear ability to predict in-hospital mortality considering a cut-off value of 0.5 (Table 6).

Discussion
In this respective study, we observed that a high RDW (RDW≥ 14.045) and a high PLR (PLR ≥ 172.067) were independent predictor for the occurrence of AKI in ICU patients. Furthermore, we found that patients with high-RDW denoted 2.9 fold risk of hospital mortality than those in the RDW-low group, but the PLR did not show a clear predictive value of in-hospital mortality (AUC:0.552). The results suggested that RDW and PLR, two parameters that are easy and inexpensive to detect, may be valuable in predicting the development of AKI.
The high incidence of AKI has caused substantial social and economic burdens worldwide, and early diagnosis and intervention are important. In recent decades, many studies on potential biomarkers of AKI, including CystC and kidney injury molecule-1 (KIM-1), have been conducted by clinicians and researchers, with the goal of improving early diagnosis of AKI [19][20] ; however, few studies have been conducted in the clinic because these biomarkers are expensive and di cult to examine. Some studies have explored the value of some in ammatory factors, such as NLR and PLR, in prediction of the development and prognosis of AKI [10][11][12] . RDW is a parameter detected in routine blood tests; however, the association between RDW and the occurrence of AKI in ICU patients is still unclear.
The study indicated that RDW shown clear predictive value of the occurrence of AKI, RDW and the PLR outperformed CystC in the prediction of AKI. This nding supports the signi cance of RDW and PLR as indictors of AKI [21][22] . We compared the predictive value of RDW with those of other in ammatory factors and found that a high RDW and high PLR were better predictors of AKI than CRP, PCT, and WBC count.
This result is similar to those of studies conducted by Elhosseiny et al 16 and Fatih Akin et al 23

. Yanfei
Shen et al 24 suggested that the PLR related to increased risk of mortality in sepsis patients. However, contrary to their study result, in the current study, the PLR did not show a clear ability to predict in-hospital mortality in the study. This may be because the population comprised critically ill patients, the elevated PLR was due to in ammation, and in ammation was not the only primary reason for in-hospital mortality.
Although the predictive value of RDW for AKI in ICU patients has been revealed, the underlying mechanisms are still unknown. In ammation and renal ischaemia-reperfusion are major primary causes of AKI. RDW is usually a parameter associated with anaemia, but some studies have demonstrated that RDW can also be a surrogate biomarker of in ammation [25][26] . It has been reported that an elevated RDW value increases the risk of progression of cardiovascular diseases, potentially leading to a decrease in renal blood ow and reduced renal artery pressure, which could cause prerenal AKI [15][16] .An increasing amount of evidence has demonstrated that poor nutritional statusand in ammatory cytokines directly in uence iron status, decrease erythropoietin production and enhancing erythropoietin resistance 15-16 .
Lin Pei Jia et al demonstrated that systemic in ammatory response syndrome (SIRS) values were increased in the high-RDW group in a large-sample study in critically ill patients 29 . Our study also suggested that RDW was positively associated with CRP and PCT and negatively associated with low haemoglobin. Therefore, RDW is related to prognosis, partly because of the role of in ammation and poor nutritional status.
Although RDW was remarkbly negatively related to haemoglobin and albumin levels, RDW showed a good ability to predict the development of AKI and in-hospital mortality according to the ROC analysis. In the current study, both RDW and the PLR were correlated with other in ammatory markers, and RDW was also associated with the PLR (r = 0.182, P=0.000) which contracted to the results of the study conducted by Jiefu Zhu 25 ; therefore, we did not combined the two parameters.
In addition to the intrinsic limitations of all retrospective designs, other aspects of this study were subject to certain limitations, which should be acknowledged. First, our study was a retrospective observational study which conducted in a single centre, and control confounders and selective bias were di cult to control. However, patients with a hospital stay less than 48 hours and patients with less than two blood tests were excluded to avoid selection bias caused by other factors. Therefore, the results of this study can still represent the epidemiological characteristics of AKI in the ICU. Second, we de ned AKI based solely on the change in the Scr level because diuretics could affect urine output, so the actual incidence of AKI may be higher than the result. Finally, the study investigated only short-term prognosis and did not analyse long-term prognosis; the latter will require the long-term follow-up of patients with AKI.

Conclusions
In summary, our study supports the independent predictive and prognostic power of high RDW in critically ill AKI patients. A high PLR also provides valuable information for the early diagnosis of AKI but is not predictive of in-hospital mortality. Availability of data and materials The datasets generated and/or analyzed during the current study available from the corresponding author on reasonable request.

Con ict of interest
None Ethics Statement The study was in accordance with the ethical principles of the Declaration of Helsinki and approved by the ethics committee of the rst a liated hospital of South China (certi cation number 20201211LL012). And the study has been granted an exemption from requiring informed consent by the ethics committee of the rst a liated hospital of South China .