Background

More than 5 million patients are admitted to intensive care units (ICUs) each year in America [1, 2], and 6–24% of these patients have acute kidney injury (AKI) [3]. In the presence of AKI, patient mortality increases to as high as 60–70%, especially within 1 year after ICU admission [4,5,6]. Considering the high incidence of AKI in the ICU and its poor prognosis, an increasing number of observational studies over the past 2 decades have been devoted to identifying the clinical predictors of mortality in AKI.

Systemic inflammation is an integral part of disease progression in critical illness and is commonly associated with sepsis, leading to an increased risk of mortality [7, 8]. Inflammation plays an important role in the initiation and progression of AKI [9,10,11], and morphological and/or functional changes in vascular endothelial cells and/or in the tubular epithelium are observed in patients with AKI. Leukocytes, including lymphocytes, infiltrate the injured kidneys and the entire body via the circulatory system and induce the generation of inflammatory mediators such as cytokines and chemokines, which damage the kidney and other organs [12]. The antithrombotic effects of platelets can evolve into atherogenesis via the secretion of proinflammatory cytokines [13], whereas the binding of platelets to endothelial cells can trigger leukocyte transmigration and adhesion, especially in the presence of shear stress [14]. The platelet-to-lymphocyte ratio (PLR) has been introduced as a potential marker of inflammation in cardiovascular disease (CVD) and tumors, which are also inflammation-related diseases [15,16,1. Participants with higher calibrated PLRs (PLR > 311) were more likely to be elderly, female, and white and to report a history of chronic pulmonary disease, metastatic cancer, solid tumors, and iron-deficiency anemia; they also had higher levels of serum potassium, BUN, white blood cells, neutrophils, platelets, urine output, and eGFR and were more likely to use RRT than those with lower PLRs (PLR < 90).

Fig. 1
figure 1

HRs (95% CIs) for all-cause mortality across fitted groups of platelet-to-lymphocyte ratios (fitted groups: model 1 and model 2)

Full size image
Table 1 Baseline characteristics of the study population
Full size table

Association between platelet-to-lymphocyte ratio and 30-day and 90-day outcomes

A total of 2277 thirty-day and 3112 ninety-day deaths occurred during the follow-up period. A U-shaped relationship was observed between the PLR and 90-day mortality, and the patients in the mid-PLR group (90–311) had the lowest 30-day mortality rate when compared with rates in the PLR < 90 and PLR > 311 groups (both P < 0.001). The adjusted HRs (95% CIs) for PLRs < 90 and > 311 were 1.25 (1.12–1.39) and 1.19 (1.08–1.31), respectively. A similar trend was observed for 30-day mortality, and the risk was less evident with higher PLRs (P = 0.047 for PLR > 311). Following the stratification of PLRs into quintiles and using the second quintile (PLR 101.2–155.6) as a reference, both extremely low (<101.2) and extremely high (>330.2) PLRs were associated with an increased risk of 90-day mortality. For 30-day mortality, the lowest PLRs (<101.2, P = 0.001) were associated with an increased risk, whereas marginally increased risk was associated with extremely high PLRs (>330.2, P = 0.088) after adjustment for potential confounders (Figs. 1 and 2 and Table 2).

Fig. 2
figure 2

HRs (95% CIs) for all-cause mortality across quintile groups of platelet-to-lymphocyte ratios (quintile groups: model 1 and model 2)

Full size image
Table 2 HRs (95% CIs) for all-cause mortality across groups of platelet-to-lymphocyte ratios
Full size table

Subgroup analyses

In the subgroup analyses, the association between the PLR and the risk of 90-day mortality was similar for most strata (P = 0.083–0.983) (Table 3). Significant interactions were observed only for age (P < 0.001). Patients younger than 65 years of age had a significantly higher risk of 90-day mortality for a PLR < 90 (HR 1.37, 95% CI 1.16–1.62, P < 0.001), whereas only older patients (≥65 years) showed an increased risk for a PLR > 311 (HR 1.25 95% CI 1.22–1.40, P < 0.001). Similarly, patients with a heart rate < 89.4 beats per minute had a significantly higher risk of 90-day mortality with a PLR > 311 (HR 1.29, 95% CI 1.22–1.48, P < 0.001), whereas patients with a heart rate ≥ 89.4 beats per minute (aged ≥ 65 years) showed an increased risk only with a PLR < 90 (HR 1.39, 95% CI 1.20–1.61, P < 0.001).

Table 3 Subgroup analysis of the associations between 90-day all-cause mortality and the platelet-to-lymphocyte ratio
Full size table

Discussion

In this study, we observed a U-shaped relationship between the PLR and 30-day and 90-day mortality, and both low and high PLRs were associated with increased all-cause mortality. Proctor et al. [21] investigated the correlation between the PLR and overall survival in a large-scale cohort of 8759 patients with cancer. In contrast to our results, their study showed a positive correlation between the PLR and mortality when using a similar PLR cutoff (PLR < 150, HR 1; PLR 150–300, HR 1.19; P < 0.001; PLR > 300, HR 1.71, P < 0.001). Yaprak et al. [22] recently evaluated the correlation between the PLR and mortality in a small cohort of patients with end-stage renal disease (ESRD) and demonstrated that the PLR could independently predict all-cause mortality in this population. A main reason for this difference is the insufficient number of patients with low PLRs. However, our study addresses this limitation. In the Framingham Heart Study, blood cell composition was treated as a prognostic factor for CVD, and the association between hematocrit and CVD mortality showed a U-shaped association in both men and women [23].

Both AKI and chronic kidney disease (CKD) are associated with local and systemic inflammation [24]. Researchers in many observational studies have described high circulating levels of inflammatory mediators and adverse outcomes for these conditions. These inflammatory mediators include blood cells, components of endothelial cells, platelets, lymphocytes, macrophages, mast cells, and fibroblasts. The PLR has been investigated as a new inflammatory marker for predicting major adverse events associated with CVD [16]. In a study of 2563 patients, Velibey et al. [25] demonstrated that increased PLRs are independently associated with a greater risk of contrast-induced AKI in patients undergoing primary percutaneous coronary intervention. A recent study showed that a high PLR is related to the presence of coronary artery disease and is correlated with C-reactive protein and fibrinogen levels [26]. High PLRs in patients with ESRD were also associated with high levels of inflammation. Balta et al. [27] showed that inflammation is better predicted by the PLR than by the neutrophil-to-lymphocyte ratio in ESRD. On the basis of the association between PLR-related inflammation and disease severity, we speculated that excessively high PLRs could predict the same poor outcomes as other inflammation biomarkers in AKI populations.

In a population-based cohort study comparing the mortality rates of 605 ICU patients with those of patients with acute renal failure (ARF), Mehta et al. [28] found that a platelet count < 20,000/mm3 is a criterion for hematologic failure and that the risk of mortality was more than threefold higher among patients who had hematologic failure than among control patients with normal platelet levels (OR 3.39, 95% CI 2.08–5.52). Using data from the Program to Improve Care in Acute Renal Disease, a multicenter observational study of ARF, Chertow et al. [29] examined the correlates of mortality in 618 ICU patients with ARF. Thrombocytopenia was associated with mortality at the time of consultation. In addition, among 512 ICU patients requiring acute dialysis, a platelet count < 50,000/mm3 was a potential risk factor for mortality in multivariate analysis [30]. Thrombocytopenia is common among critically ill patients and is often associated with poor outcomes [31,32,33,34]. The mechanisms underlying thrombocytopenia include either reduced platelet production or excessive platelet destruction related to an underlying illness and therapeutic interventions [35]. Taken together, these findings show that low platelet counts could result from a low PLR in patients with AKI and could lead to high mortality, thus hel** to explain our observation of a U-shaped relationship between the PLR and mortality.

Although AKI in the ICU is associated with high mortality, other factors appear to contribute to poor outcomes. Potential factors that may affect the outcome of AKI include blood pressure [36], renal function [37], urine output [36], and other clinical parameters (i.e., SCr, BUN, and pH [38]), as well as comorbidities (e.g., cardiac disease [38]). In the present study, when patients were stratified according to potential confounders, no significant interactions were observed for sex, ethnicity, PO2, GCS, SBP, potassium, SCr, urine output, BUN, vasopressin use, or ventilator use. Although possible interactions between all-cause mortality and both age and heart rate were observed, there was no heterogeneity of clinical factors among those effects and the PLR. However, little is known about the mechanism underlying the interaction between age and the PLR. Recently, in a multicenter cross-sectional study of a healthy Indian population, Sairam et al. [39] found lower levels of platelets in elderly people. Kweon et al. [40] investigated median PLRs in a healthy Korean population and suggested that the PLR cutoff values for disease evaluation should be established separately according to age. Therefore, we should consider age when evaluating the relationship between the PLR and mortality among critically ill patients with AKI.

The limitations of this study should be acknowledged. First, it was a single-center retrospective analysis, and different conclusions could be reached when using patient records from other centers. Therefore, subject selection bias cannot be ignored, suggesting that a prospective multicenter study is needed. However, the strength of the present study is its representative and ethnically diverse population. Second, because of a lack of data on kidney function prior to 3 months before patient admission, we could not assess CKD status among patients with AKI or determine the role of CKD in the association between the PLR and mortality. Third, the PLR can be measured in patients only upon admission to the ICU. A single measure of the PLR does not fully reflect inflammation, which would be better assessed by simultaneously measuring other inflammatory mediators. Fourth, these preliminary data suggest that the PLR could be a risk adjustment tool with prognostic implications for AKI. Fifth, to establish the PLR as a prognostic marker, researchers must further validate its clinical significance. The cutoff value must be established in one cohort of patients and tested in another, and the number of patients in each group needs to be considered in statistical analyses. Finally, we did not assess the modification of sepsis and shock, which both might increase patient morbidity and predict higher mortality among patients with AKI [41], on association between PLR and outcomes, owing to lack of related data.

Conclusions

We found a U-shaped relationship between the PLR and mortality in which both low and high PLRs were associated with increased overall mortality in critically ill patients with AKI. The PLR is therefore potentially useful in the clinical setting as a cost-effective and readily available biomarker. Our findings need to be confirmed by other studies, especially large prospective studies with longer follow-up.