Platelet-lymphocyte ratio and its dynamic changes predict mortality in septic acute kidney injury patients: a retrospective multi-center study using U.S. database and Chinese hospital data

Introduction

Acute kidney injury (AKI) is a heterogeneous and multifactorial syndrome characterized by sudden deterioration of renal function. It occurred in 10–15% of hospitalized patients, and over 50% of critically ill patients in the intensive care unit (ICU), with septic AKI accounts for 45–70% (Mehta et al., 2016; Hoste et al., 2018; Ostermann et al., 2025). Septic AKI often leads to prolonged ICU and hospital stays, and higher mortality, which ranges from 60–70% (Bellomo et al., 2004; Hofhuis et al., 2013). Though microvascular dysfunction, dysregulated inflammation, and metabolic reprogramming might play roles in the development of septic AKI, the precise mechanisms have not been clearly described; hence, effective and specific interventions for septic AKI are still lacking (De Backer et al., 2011; Wang et al., 2012; Maiden et al., 2016; Gómez, Kellum & Ronco, 2017; Peerapornratana et al., 2019; Liu et al., 2024). Given the notable prevalence, obscure pathophysiology, and poor prognosis, early identification of septic AKI trajectory and timely prevention of poor outcomes are of great importance.

Systemic inflammation plays a crucial role in the progression of septic AKI. Complex interactions among platelets, granulocytes, the vascular endothelium, and invading pathogens finely modulate the inflammatory response through intertwined pro-inflammatory and anti-inflammatory pathways (Li, Huang & Sun, 2025, pp. 2015–2020).

Increasing attention has been directed toward identifying reliable inflammatory markers for septic AKI, particularly those derived from complete blood cell counts due to their accessibility and cost-effectiveness (Liu et al., 2020; Atreya et al., 2023; Yin et al., 2024; Wei et al., 2024; Wei et al., 2025). Among these indices, the neutrophil-to-lymphocyte ratio (NLR) has been widely investigated and recognized as a robust predictor of kidney disease, as it reflects both innate (neutrophil-driven) and adaptive (lymphocyte-mediated) immune responses (Buonacera et al., 2022). Elevated NLR was consistently associated with increased risk and severity of chronic kidney disease (CKD) and erythropoietin resistance in hemodialysis (Carollo et al., 2025b; Carollo et al., 2025a; Li, Huang & Sun, 2025, pp. 2015–2020). Notably, both high NLR and ΔNLR were independently associated with all-cause mortality and disease severity in septic AKI patients (Wei et al., 2023; Wei et al., 2024; Shangguan et al., 2025). NLR outperformed other markers, such as white blood cell (WBC), procalcitonin, and monocyte-to-lymphocyte ratio (MLR) in predicting the prognosis of septic AKI, achieving the highest area under the receiver operating characteristic curve (AUROC) for 30-day mortality prediction (0.618 and 0.624 in two large studies) (Wei et al., 2024; Shangguan et al., 2025). The platelet-to-lymphocyte ratio (PLR) is another clinically available and cost-effective index associated with systemic inflammation. Platelets function not only in hemostatic roles but also as immunomodulators by binding pathogens, releasing cytokines, and forming aggregates with leukocytes (Dewitte et al., 2017; Chen et al., 2020b). Lymphopenia is a hallmark of sepsis-induced immune suppression, whereas platelet activation reflects a pro-thrombotic, pro-inflammatory state; hence PLR implicitly quantifies this immune-thrombotic axis (Iba & Levy, 2018; Gasparyan et al., 2019; Nicolai et al., 2020). PLR has shown prognostic value across a wide range of diseases, including malignancy, cardiovascular disease, and AKI (Sunbul et al., 2014; Li et al., 2015; Zheng et al., 2017; Jeon et al., 2023). Both high NLR (adjusted HR = 1.97) and PLR (adjusted HR = 2.62) were regarded as independent risk factors for all-cause mortality in patients with CKD (Umeres-Francia et al., 2022). However, lower PLR was associated with autologous arteriovenous fistula (AVF) dysfunction (odds ratio 9.97; 95% CI [2.53–39.25]) and positively correlated with AVF patency duration (Rho = 0.254, p = 0.002) (Messina et al., 2024). For peritoneal dialysis (PD) patients, multivariate Cox regression analysis showed that the all-cause mortality rate was higher in the high PLR group than in the low PLR group (HR = 1.64) (Liu et al., 2021). Similarly, a study found that the mortality in the high NLR group (18.8 vs. 7.5 %) and high PLR group (18.8 vs. 7.5 %) increased for hemodialysis individuals. Of note, adjusted Cox regression analysis showed the association of mortality and high NLR was lost, while the significance of the association of high PLR and mortality increased (Yaprak et al., 2016).

In light of the results of these studies, it is reasonable to speculate that the PLR might affect the prognosis of septic AKI. Moreover, to the best of our knowledge, few studies have explored whether dynamic changes in PLR during hospitalization add incremental prognostic information. To address this gap, we conducted a retrospective study from two large cohorts, Medical Information Mart for Intensive Care (MIMIC-IV) and West China Hospital of Sichuan University, to investigate the association between PLR/PLR dynamics with the mortality of septic AKI by internal test and external validation.

Method

Result

Baseline characteristics

The flowchart illustrating patient enrollment is presented in Fig. 1. In total, 1,478 patients diagnosed with septic AKI were included in this study, comprising 1,285 individuals from the MIMIC-IV database and 193 from West China Hospital. Of these, 982 patients survived hospitalization (895 from MIMIC-IV and 87 from West China Hospital), making them eligible for analyses involving ΔPLR. Baseline characteristics of patients from the two cohorts are summarized in Table 1. Only hemoglobin had one missing value, and all other variables were complete. The median age was 65 years old, with females accounting for 38.07%. Compared with the West China Hospital cohort, patients from MIMIC database showed higher WBC, neutrophils, platelets, urea nitrogen, creatinine and elevated prevalence of comorbidities of CAD, ARDS, UTI, and CKD. By contrast, lower hemoglobin, albumin, total bilirubin, triglycerides, and likelihood of pneumonia were observed in the MIMIC database. No significant differences were observed in other laboratory parameters or comorbidities between cohorts.

Table 1:

Baseline characteristics of participants from MIMIC and West China Hospital.

Variables	Total (n = 1478)	MIMIC (n = 1285)	West China Hospital (n = 193)	Statistic	P
Age, year	65.00 (53.00, 74.00)	65.00 (54.00, 73.00)	63.00 (48.00, 75.00)	Z = −0.96	0.336
Gender, n(%)				χ2= 2.77	0.096
Female	562 (38.07)	499 (38.83)	63 (32.64)
Male	916 (61.93)	786 (61.17)	130 (67.36)
Serum laboratory test
Hemoglobin, g/L	101.00 (85.00, 118.00)	10.00 (8.50, 11.70)	106.00 (84.75, 132.00)	Z = −2.57	0.010
WBC, K/uL	11.11 (7.29, 17.08)	11.60 (7.60, 17.50)	9.12 (6.26, 14.04)	Z = −4.78	<.001
Neutrophils, K/uL	8.95 (5.43, 14.07)	9.23 (5.71, 14.41)	7.38 (4.08, 11.37)	Z = −4.41	<.001
Lymphocytes, K/uL	0.88 (0.52, 1.44)	0.88 (0.51, 1.46)	0.87 (0.56, 1.36)	Z = −0.03	0.979
Platelets, K/uL	185.00 (120.00, 273.50)	192.00 (125.00, 282.00)	153.00 (83.00, 223.00)	Z = −5.27	<.001
Urea nitrogen, mg/dL	28.84 (17.00, 46.00)	30.00 (18.00, 49.00)	17.50 (12.04, 26.52)	Z = −9.58	<.001
Albumin, g/dL	2.80 (2.40, 3.20)	2.80 (2.30, 3.20)	3.13 (2.81, 3.65)	Z = −8.83	<.001
Total bilirubin, umol/L	11.00 (6.84, 22.23)	10.26 (5.13, 22.23)	13.40 (8.50, 22.30)	Z = −3.16	0.002
Triglycerides, mmol/L	0.70 (0.00, 1.81)	0.00 (0.00, 1.81)	1.27 (0.88, 1.82)	Z = −8.60	<.001
Creatinine, mg/dL	1.30 (0.90, 2.00)	1.40 (1.00, 2.10)	0.84 (0.68, 1.06)	Z = −12.80	<.001
Comorbidities
Hypertension, n(%)	462 (31.26)	391 (30.43)	71 (36.98)	χ2= 3.36	0.067
Received Transplant, n(%)	25 (1.69)	18 (1.40)	7 (3.63)	χ2= 3.76	0.053
Diabetes Mellitus, n(%)	349 (23.60)	307 (23.89)	41 (21.24)	χ2= 0.68	0.409
Coronary Artery Disease, n(%)	320 (21.64)	296 (23.04)	24 (12.44)	χ2= 11.08	<.001
ARDS, n(%)	351 (23.73)	320 (24.90)	30 (15.54)	χ2= 8.22	0.004
Pneumonia, n(%)	439 (29.68)	291 (22.65)	147 (76.17)	χ2= 229.79	<.001
COPD, n(%)	159 (10.75)	139 (10.82)	20 (10.36)	χ2= 0.03	0.852
Tumor, n(%)	246 (16.63)	197 (15.33)	48 (24.87)	χ2= 10.86	<.001
UTI, n(%)	303 (20.49)	280 (21.79)	23 (11.92)	χ2= 10.01	0.002
SLE, n(%)	12 (0.81)	9 (0.70)	3 (1.55)	χ2= 0.65	0.422
CKD, n(%)	189 (12.78)	181 (14.09)	7 (3.63)	χ2= 16.68	<.001

DOI: 10.7717/peerj.20522/table-1

Notes:

Abbreviations

MIMIC

Medical Information Mart for Intensive Care

WBC

white blood cell

ARDS

acute respiratory disease syndrome

COPD

chronic obstructive pulmonary disease

UTI

urinary tract infection

SLE

systemic lupus erythematosus

CKD

chronic kidney disease

The MIMIC set was then randomly allocated into training and test cohorts with the ratio of 7:3. The comparisons of the two datasets are shown in Table S1. Baseline characteristics between these two subsets were comparable, except for a slightly higher proportion of malignancy in the training cohort. Using ROC analysis in the training set, optimal cut-off values for predicting 28-day mortality were identified as 335.44 for baseline PLR and -46.69 for ΔPLR. The patients were hence divided into the “Low PLR” group (PLR < 335.44, n = 643)/“High PLR” group (PLR ≥ 335.44, n = 256) and “Low ΔPLR” group (ΔPLR < −46.69, n = 108)/“High ΔPLR” group (ΔPLR ≥−46.69, n = 519).

The association of PLR and 28-day and 90-day mortality

Baseline features comparisons between the two groups were displayed in Table S2. Patients in the High PLR group were older and exhibited higher platelet counts, whereas the Low PLR group had significantly higher WBC counts, lymphocyte counts, and total bilirubin (P < 0.05). Kaplan–Meier analyses indicated that patients in the Low PLR group had significantly better survival probabilities at both 28-day (HR = 0.674, 95% CI [0.500–0.907], P = 0.009) and 90-day (HR = 0.706, 95% CI [0.566–0.881], P = 0.002) follow-ups (Figs. 2A and 2B). Correspondingly, the mortality rates were significantly lower in the Low PLR group compared to the High PLR group at both 28-day (18.66% vs. 26.56%, P = 0.009) and 90-day (35.61% vs. 46.48%, P = 0.003) (Table 2). In the univariate model 1, the connection between PLR and 28-day and 90-day mortality was statistically significant (28-day: HR = 0.67, 95% CI [0.50–0.91], P = 0.009; 90-day: HR = 0.71, 95% CI [0.57–0.88], P = 0.002). In multivariate model 2 adjusted by age, WBC, and TBIL, the PLR remained significantly associated with 28-day and 90-day mortality (28-day: HR = 0.70, 95% CI [0.52–0.94], P = 0.019; 90-day: HR = 0.72, 95% CI [0.58–0.90], P = 0.004). This predictive value of PLR was also confirmed in model 3 additionally adjusted by gender, hemoglobin, urea nitrogen, concurred with hypertension, diabetes mellitus, CAD, tumor, and CKD (28-day: HR = 0.72, 95% CI [0.53–0.97], P = 0.030; 90-day: HR = 0.74, 95% CI [0.59–0.93], P = 0.010) (Table 3).

In the test and external validation cohort, patients were also clarified into “High PLR” group and “Low PLR” group by the optimal PLR cutoff of 335.44, according to the training cohort. The PLR grade was also a strong predictor for 28-day and 90-day mortality for septic AKI patients, which was verified by 386 individuals in the test cohort from MIMIC (28-day: HR = 0.683, 95% CI [0.466–1.001], P = 0.049; 90-day: HR = 0.690, 95% CI [0.502–0.948], P = 0.021) (Figs. 3A and 3B) and 193 patients in the external validation cohort from West China Hospital (28-day: HR = 0.544, 95% CI [0.317–0.934], P = 0.024; 90-day: HR = 0.652, 95% CI [0.414–1.027], P = 0.061) (Figs. 3C and 3D).

To further explore the predictive value, two multivariate models were also performed for test cohort, which were adjusted by crude factors in the training cohort. The result remained robust in Model 2 (28-day: HR = 0.69, 95% CI [0.47–1.00], P = 0.058; 90-day: HR = 0.70, 95% CI [0.51–0.97], P = 0.032). In Model 3, lower PLR was a potentially protective indicator, but with only borderline statistical significance (28-day: HR = 0.71, 95% CI [0.48–1.06], P = 0.097; 90-day: HR = 0.71, 95% CI [0.51–0.99], P = 0.047). For external validation with patients from West China Hospital, low PLR also showed potential protective role for survival in Model 2 (28-day: HR = 0.56, 95% CI [0.32–0.98], P = 0.042; 90-day: HR = 0.66, 95% CI [0.41–1.05], P = 0.081) and Model 3 (28-day: HR = 0.54, 95% CI [0.31–0.93], P = 0.025; 90-day: HR = 0.64, 95% CI [0.41–1.02], P = 0.060) (Table 3).

The association of ΔPLR and 28-day and 90-day mortality

With regard to the dynamic changes of PLR, referred to as ΔPLR, patients with Low ΔPLR displayed reduced hemoglobin, TP, ALB, and TBIL, but elevated WBC, Scr, neutrophil counts, platelet, urea nitrogen, and proportion of male (P < 0.05) (Table S2). Remarkably, ΔPLR demonstrated strong mortality prediction, with Low ΔPLR associated with 2.3-fold higher 28-day mortality risk and 1.7-fold increased 90-day mortality (28-day: HR = 2.261, 95% CI [1.255–4.074], P = 0.005; 90-day: HR = 1.702, 95% CI [1.093–2.649], P = 0.017) (Figs. 2C and 2D). The 28-day and 90-day mortality was significantly elevated in the Low ΔPLR group (28-day: 14.81% vs. 6.94%, P = 0.007; 90-day: 24.07% vs. 15.41%, P = 0.029) (Table 2). In the univariate model 1, the connection between ΔPLR and 28-day and 90-day mortality was statistically significant (28-day: HR = 2.26, 95% CI [1.25–4.07], P = 0.005; 90-day: HR = 1.70, 95% CI [1.09–2.65], P = 0.017). In the multivariate model 4 adjusted by gender, hemoglobin, WBC, ALB, TBIL, and Scr, the ΔPLR remained significantly associated with 28-day and 90-day mortality (28-day: HR = 2.47, 95% CI [1.32–4.61], P = 0.005; 90-day: HR = 1.61, 95% CI [1.01–2.57], P = 0.045). This predictive value of ΔPLR was also confirmed in model 5 additionally adjusted by age, concurred with hypertension, diabetes mellitus, CAD, tumor, and CKD (28-day: HR = 2.57, 95% CI [1.36–4.88], P = 0.004; 90-day: HR = 1.66, 95% CI [1.04–2.67], P = 0.035) (Table 3).

Table 2:

The associations of PLR/ΔPLR and clinical outcomes for septic AKI patients.

Variables	PLR						ΔPLR
	Total (n = 899)	High PLR (n = 256)	Low PLR (n = 643)	Statistic	P		Total (n = 627)	High ΔPLR (n = 519)	Low ΔPLR (n = 108)	Statistic	P
Hospital duration, M (Q1, Q3)	22.43 (13.52, 37.02)	23.57 (13.97, 36.17)	21.89 (12.90, 37.41)	Z = −0.57	0.572		22.69 (13.92, 37.85)	21.64 (13.45, 36.20)	28.64 (17.56, 44.64)	Z = −2.88	0.004
ICU duration, M (Q1, Q3)	7.06 (2.57, 15.39)	7.00 (2.14, 15.82)	7.12 (2.80, 15.07)	Z = −0.24	0.811		6.54 (2.42, 15.39)	6.54 (2.35, 15.75)	6.52 (2.57, 13.23)	Z = −0.01	0.996
28-day death, n(%)	188 (20.91)	68 (26.56)	120 (18.66)	χ2= 6.91	0.009		52 (8.29)	36 (6.94)	16 (14.81)	χ2= 7.30	0.007
90-day death, n(%)	348 (38.71)	119 (46.48)	229 (35.61)	χ2= 9.12	0.003		106 (16.91)	80 (15.41)	26 (24.07)	χ2= 4.77	0.029
Invasive mechanical ventilation, n(%)	320 (35.60)	80 (31.25)	240 (37.33)	χ2= 2.95	0.086		222 (35.41)	184 (35.45)	38 (35.19)	χ2= 0.00	0.958
RRT need, n(%)	71 (7.90)	18 (7.03)	53 (8.24)	χ2= 0.37	0.543		45 (7.18)	39 (7.51)	6 (5.56)	χ2= 0.51	0.473
Septic Shock, n(%)	635 (70.63)	189 (73.83)	446 (69.36)	χ2= 1.76	0.185		394 (62.84)	325 (62.62)	69 (63.89)	χ2= 0.06	0.804
Readmitted, n(%)	594 (66.07)	174 (67.97)	420 (65.32)	χ2= 0.57	0.449		460 (73.37)	378 (72.83)	82 (75.93)	χ2= 0.44	0.508

DOI: 10.7717/peerj.20522/table-2

Notes:

Abbreviations

AKI

acute kidney injury

PLR

platelet-to-lymphocyte ratio

ICU

intensive care unit

RRT

renal replacement therapy

Table 3:

The Cox proportional hazards models of 28-day and 90-day mortality for the training, test, and validation cohorts.

Cohorts	PLR						ΔPLR
	Model1	P	Model2	P	Model3	P	Model1	P	Model4	P	Model5	P
	OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)		OR (95% CI)
Training cohort
28-day mortality	0.67 (0.50∼0.91)	0.009	0.70 (0.52∼0.94)	0.019	0.72 (0.53∼0.97)	0.030	2.26 (1.25∼4.07)	0.005	2.47 (1.32∼4.61)	0.005	2.57 (1.36∼4.88)	0.004
90-day mortality	0.71 (0.57∼0.88)	0.002	0.72 (0.58∼0.90)	0.004	0.74 (0.59∼0.93)	0.010	1.70 (1.09∼2.65)	0.017	1.61 (1.01∼2.57)	0.045	1.66 (1.04∼2.67)	0.035
Test cohort
28-day mortality	0.68 (0.47∼1.00)	0.049	0.69 (0.47∼1.01)	0.058	0.71 (0.48∼1.06)	0.097	2.69 (1.26∼5.72)	0.007	2.56 (1.18∼5.52)	0.017	2.92 (1.29∼6.60)	0.01
90-day mortality	0.69 (0.50∼0.95)	0.021	0.70 (0.51∼0.97)	0.032	0.71 (0.51∼0.99)	0.047	2.85 (1.48∼5.49)	0.001	2.76 (1.41∼5.39)	0.003	3.06 (1.51∼6.23)	0.002
External validation cohort
28-day mortality	0.54 (0.32∼0.93)	0.024	0.56 (0.32∼0.98)	0.042	0.54 (0.31∼0.93)	0.025	2.59 (0.58∼11.57)	0.196	4.01 (0.75∼21.40)	0.104	4.95 (0.65∼37.53)	0.122
90-day mortality	0.65 (0.41∼1.03)	0.061	0.66 (0.41∼1.05)	0.081	0.64 (0.41∼1.02)	0.060	2.32 (0.71∼7.61)	0.152	2.39 (0.66∼8.66)	0.186	2.20 (0.52∼9.28)	0.281

DOI: 10.7717/peerj.20522/table-3

Notes:

Abbreviations

PLR

platelet-to-lymphocyte ratio

OR

odd ratio

In the test and external validation cohort, patients were also clarified into “ΔHigh PLR” group and “ΔLow PLR” group by the optimal PLR cutoff of −46.69, according to the training cohort. The connection between low ΔPLR grade and 28-day and 90-day mortality for septic AKI patients was also significant from 268 individuals in the test cohort from MIMIC (28-day: HR = 2.689, 95% CI [1.264–5.720], P = 0.007; 90-day: HR = 2.847, 95% CI [1.478–5.487], P = 0.001) (Figs. 4A and 4B), while in the external validation cohort from West China Hospital, despite showing a similar trend, no statistically significant associations were observed (28-day: HR = 2.590, 95% CI [0.579–11.575], P = 0.196; 90-day: HR = 2.323, 95% CI [0.709–7.613], P = 0.152) (Figs. 4C and 4D).

Furthermore, two multivariate models were also performed for the test cohort, which were adjusted by crude factors in the training cohort. The results were similar in Model 4 (28-day: HR = 2.56, 95% CI [1.18–5.52], P = 0.017; 90-day: HR = 2.76, 95% CI [1.41–5.39], P = 0.003) and Model 5 (28-day: HR = 2.92, 95% CI [1.29–6.60], P = 0.010; 90-day: HR = 3.06, 95% CI [1.51–6.23], P = 0.002). For external validation with patients from West China Hospital, given the limited sample size, low ΔPLR also did not showed significant predictive role for mortality in Model 4 (28-day: HR = 4.01, 95% CI [0.75–21.40], P = 0.104; 90-day: HR = 2.39, 95% CI [0.66–8.66], P = 0.186) and Model 5 (28-day: HR = 4.95, 95% CI [0.65–37.53], P = 0.122; 90-day: HR = 2.20, 95% CI [0.52–9.28], P = 0.281) (Table 3).

Mediation analysis of PLR/ΔPLR on 28-day and 90-day mortality

This study further assessed the regulating role of inflammation-associated biomarkers:

CRP/ΔCRP, WBC/ΔWBC, and neutrophils/Δneutrophils in the association between PLR/ /ΔPLR and 28-day mortality in the septic AKI patients. This study found that ΔWBC mediated 27.99% of the association between ΔPLR and 28-day mortality (Fig. 5). However, these inflammation factors didn’t serve as the mediator for the impact of ΔPLR on 90-day mortality (Fig. 5). The regulating role of CRP, WBC, and neutrophils was not observed between PLR level and both 28-day and 90-day mortality (Fig. S1).

Discussion

In this study, we found that a higher baseline PLR was a strong predictor of 28-day and 90-day mortality in patients with septic AKI, as demonstrated in both the internal and external validation cohorts. Subgroup analysis revealed that the prognostic value of PLR was more pronounced in non-UTI patients for 90-day mortality. Interestingly, male patients appeared to benefit more from a lower baseline PLR. In contrast, a lower ΔPLR during hospitalization was associated with poorer prognosis, although this finding was not confirmed in the external validation cohort. The external validation result should be interpreted with caution, as it was inconclusive and likely underpowered, possibly due to the limited sample size and population heterogeneity between the US and Chinese cohorts. Future multicenter, prospective studies are warranted to confirm this association. The potential prognostic role of ΔPLR may be partially mediated by ΔWBC. Furthermore, a lower ΔPLR was significantly associated with prolonged length of hospital stay.

The PLR, proposed as a surrogate marker of systemic inflammation, has emerged as a prognostic indicator across a broad spectrum of diseases. In oncology, a large cohort study involving 8,759 cancer patients demonstrated a positive correlation between elevated PLR and increased mortality (Proctor et al., 2011). Similarly, in cardiovascular disease, PLR was identified as an independent predictor of long-term mortality following non-ST segment elevation myocardial infarction (Azab et al., 2012). In the context of sepsis, a retrospective cohort study using the MIMIC database, which included 5,537 patients, found that higher PLR levels were significantly associated with increased mortality (OR 1.29; 95% CI [1.09–1.53]) (Shen, Huang & Zhang, 2019). Beyond sepsis, PLR has also shown prognostic relevance in kidney-related conditions. Specifically, in both hemodialysis and peritoneal dialysis populations, patients with higher PLR exhibited greater all-cause mortality compared to those with lower PLR (Yaprak et al., 2016). In patients undergoing primary percutaneous coronary intervention, Velibey et al. (2017) reported that elevated PLR independently predicted the risk of contrast-induced AKI. Moreover, PLR also predicted AKI following cardiac surgery with cardiopulmonary bypass (He & Zhou, 2021). Focusing specifically on septic AKI, Chen et al. demonstrated a significant correlation between PLR and 28-day mortality using Spearman correlation analysis (Chen et al., 2020a). Collectively, these findings underscore the prognostic potential of PLR in both inflammatory and renal pathologies. Given its accessibility and clinical relevance, we stratified patients by PLR levels in our study and further explored the association between PLR and mortality in septic AKI patients. However, most of the aforementioned studies were limited by single-center designs, relatively small sample sizes, and a predominant focus on baseline PLR values, without consideration of its dynamic changes over the course of illness.

To address these limitations and enhance the generalizability of findings, we leveraged data from two independent cohorts, including both internal and external validation sets. The study found that baseline PLR strongly predicted 28-day and 90-day mortality in the training cohort from the MIMIC database, which was also been confirmed by internal test and external validation from West China Hospital of Sichuan University. This multicohort approach allowed us to confirm the prognostic value of baseline PLR in a broader septic AKI population. Although it is increasingly recognized, the underlying biological mechanisms remain to be fully elucidated. One proposed explanation involves the stage-dependent immunological shift that occurs during sepsis, in which the dynamic balance between pro-inflammatory and anti-inflammatory responses is critical for both disease progression and organ recovery (Liu et al., 2024). In the acute phase of sepsis, platelets are actively involved in immune responses by releasing inflammatory cytokines and directly interacting with pathogens and immune cells. These actions amplify the systemic inflammatory response, contributing to endothelial injury, microcirculatory dysfunction, and ultimately organ damage (Azab et al., 2012; Cho et al., 2014; Kim et al., 2015; Nording, Seizer & Langer, 2015). Conversely, a reduction in circulating lymphocytes, a hallmark of sepsis-induced immunosuppression, is associated with impaired host defense and a higher risk of secondary infections. Therefore, an elevated PLR may reflect a state of immune dysregulation, characterized by excessive inflammation (platelet activation) and inadequate immune surveillance (lymphopenia). This imbalance may hinder the timely transition from the hyperinflammatory phase to a reparative, anti-inflammatory phase, thereby delaying renal recovery and worsening prognosis (Chen et al., 2020a). In line with this mechanistic hypothesis, our study demonstrated that higher baseline PLR was significantly associated with increased 28-day and 90-day mortality, although it was not significantly linked to other clinical outcomes such as ICU length of stay. These findings suggest that PLR may serve as a marker of immune imbalance rather than a general severity index, and may be particularly useful in identifying patients at risk for poor long-term outcomes.

More importantly, we introduced and analyzed the ΔPLR during hospitalization—an aspect not previously explored in this clinical context. By stratifying patients among those who survived hospitalization according to ΔPLR levels, we aimed to capture the prognostic significance of PLR dynamics, which may reflect underlying immune recovery or persistent dysregulation, thereby providing novel insights into disease progression and long-term outcomes. In this study, we observed that patients with lower ΔPLR had a significantly worse prognosis compared to those with higher ΔPLR, suggesting that the trajectory of PLR during hospitalization may reflect distinct immune recovery profiles in septic AKI. The low ΔPLR (ΔPLR < −46.69) means the PLR decreases relatively largely compared to the baseline level, suggesting continued thrombocytopenia or excessive, dysregulated lymphocyte expansion (Venet & Monneret, 2018). The non-functional immune reconstitution might lead to failure renal recovery or host defense (Delano & Ward, 2016). On the contrary, a relatively stable or rising PLR (ΔPLR > −46.69) may represent an appropriate resolution phase in which platelet counts normalize and lymphocyte recovery is controlled. The rebalance of platelets and lymphocytes signifies a shift toward immune homeostasis (Tang, Huang & McLean, 2010; Van der Poll et al., 2017). This immune trajectory likely contributes to the repair of damaged organs and a better prognosis. Therefore, ΔPLR may serve not only as a dynamic biomarker of inflammation but also as an indirect indicator of immune system reconstitution during septic AKI, providing valuable insight into patient outcomes beyond static measurements. Notably, we also found that ΔWBC mediated 27.99% of the association between ΔPLR and 28-day mortality. This mediating role of ΔWBC reinforces the notion that ΔPLR captures not only platelet and lymphocyte dynamics but also broader immune-inflammatory interactions.

The limitations of this study should be acknowledged. First, despite the inclusion of both internal and external cohorts, the data were derived from retrospective observational databases, which may introduce residual confounding and selection bias. Notably, the cohorts from two centers differed in some baseline characteristics. The MIMIC-IV cohort exhibited higher levels of inflammatory and renal function markers and a higher prevalence of comorbidities, suggesting a population with greater illness severity and metabolic burden. The Chinese cohort had higher hemoglobin and albumin levels, and a greater incidence of pneumonia reflecting potential variations in disease phenotype and nutritional status. The heterogeneity in population demographics, clinical practice, and disease profiles across regions may limit the generalizability of our findings, the consistent direction of associations between PLR and mortality across both cohorts supports the robustness of our results. Future multicenter prospective studies involving more diverse populations are warranted to further validate and extend these observations. Second, although PLR and ΔPLR were associated with mortality, their mechanistic links to immune and inflammatory pathways were inferred rather than directly measured, and lack validation by functional immunologic markers. Third, certain clinically relevant variables, such as specific infection sources, pathogen types, and immunosuppressive medication use, were not uniformly available, which may affect the generalizability of the findings. Moreover, body-mass index (BMI) and total protein were excluded for analysis for a missing proportion of more than 10%. This exclusion might have led to residual confounding, as both nutritional and metabolic status could influence the prognosis of septic AKI. Although missing data were minimal in the selected cohort, median substitution was applied for hemoglobin. This simple imputation method may also introduce minor bias. Lastly, while ΔPLR showed potential prognostic value, it should be considered with caution, since the prognostic value of ΔPLR has not been validated by external data, which was possibly due to the limited sample size in the external validation cohort. Future large-scale cohorts with more complete datasets are required to confirm the findings.

Conclusion

In conclusion, this study demonstrated that baseline PLR was significantly associated with mortality in patients with septic AKI across both cohorts. By contrast, while ΔPLR showed a potential association with outcomes, this finding was not statistically confirmed in the external validation. Furthermore, ΔWBC partially mediated the association between ΔPLR and mortality, suggesting an interplay between immune recovery and systemic inflammation. These findings underscore the prognostic potential of PLR as a simple, accessible, and integrative biomarker for risk stratification in septic AKI.

Supplemental Information