Introduction

Chronic kidney disease (CKD) is recognized as a major public health burden with substantial morbidity and mortality [1]. CKD was related to an increased risk of all-cause and cardiovascular mortality [2, 4]. Cancer risk was increased in mild to moderate CKD and among transplant patients, and cancer-related mortality was significantly higher among patients with kidney disease [3]. Therefore, identifying the high-risk patients with CKD is needed in clinical practice.

Traditional cardiovascular disease (CVD) and cancer risk factors are prevalent among patients with CKD but do not fully explain the increased mortality rates among these subjects [3, 5]. Models based on traditional clinical risk factors are poor in predicting long-term prognostic among CKD patients, and there is still a lack of good tools for risk stratification [6]. Previous studies have demonstrated that extending traditional guideline-based clinical risk factors by adding inflammatory biomarkers may improve risk stratification in CVD prevention [7, 8]. In addition, CKD subjects are at a higher risk of developing cancer due to heightened inflammation and immune dysfunction [9,10,11]. Persistent low-grade inflammation and immune dysfunction are now considered a hallmark feature of CKD, being involved in the mortality of these patients [12]. Recently, systemic immune inflammation index (SII), which combines neutrophil counts, lymphocyte counts, and platelet counts, is recently proposed as a prognostic indicator comprehensively reflecting patients' inflammatory and immune status [13]. Previous studies have demonstrated that SII is not only associated with adverse outcomes in subjects with CVD [14], but also a potential superior marker in predicting death among different cancer patients [15]. However, the SII for CKD has not been reported to date, and little is known about its prognostic value for CKD.

Therefore, the objective of this multi-center research was to reveal the prognostic impact of immune inflammation levels on all-cause, cardiovascular, and cancer mortality in a large cohort, which may help clinicians early recognize patients at high risk for mortality and initiate the prompt, appropriate treatment for prevention.

Method

Data sources and patient selection

This cohort study analyzed data from the Cardiorenal Improvement II (CIN-II) study, which is a multi-center cohort of patients enrolled at five large tertiary hospitals (Cardiorenal Improvement II, ClinicalTrials.gov NCT05050877) in China between January 2007 and December 2020. A total of 22,898 patients diagnosed with CKD (estimated glomerular filtration rates [eGFR] < 60 mL/min/1.73m2) undergoing coronary angiography (CAG) on initial admission were enrolled in this multi-center cohort. The exclusion criteria were as follows: (a) patients missing laboratory parameters; (b) patients receiving dialysis or those who were unclear to have received it or not; (c) patients missing survival information; (d) patients aged < 18 years; (e) patients have known cancer (Supplementary Fig. 1). The Ethics Committee of the Guangdong Provincial People's Hospital approved the study (No.GDREC2019-555H-2). All participating sites received institutional review board approval from their own ethics committees. It was conducted in accordance with the principles of the Declaration of Helsinki.

Baseline data collection

The patient data were extracted from the electronic clinical management system (ECMS). The baseline data included demographic characteristics, complications, procedures, laboratory examinations, and medications at discharge. Biochemistry data including platelets, neutrophils, lymphocytes, and serum creatinine (Scr) were tested on initial admission. Survival information was obtained by cause-specific surveillance data from the Public Security System and Centers for Disease Control and Prevention, recorded and matched with the ECMS for all participant centers.

Clinical definition and endpoint

The endpoints were all-cause, cardiovascular and cancer mortality (the median follow-up period was 4.5 years [interquartile range: 2.4–7.5]). SII was calculated as total platelet counts × neutrophil-to-lymphocyte ratio. The estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation [16], and CKD was defined as eGFR < 60 mL/min/1.73 m2 [17]. Congestive heart failure (CHF) was defined as New York Heart Association class > 2 or Killip class > 1. Anemia was defined as a hematocrit ≤ 39% for males or ≤ 36% for females. Acute myocardial infarction (AMI), diabetes mellitus (DM), and hypertension (HT) were defined according to the 10th Revision Codes of the International Classification of Diseases (ICD-10).

Statistical analysis

The population was categorized by SII quartiles: quartile (Q)1 > 420.09, Q2 > 420.09 and < 646.91, Q3 > 646.91 and < 1063.39, Q4 > 1063.39. For the baseline characteristics, continuous variables were summarized as median and quartiles or mean and standard deviation as appropriate. Categorical variables were described as frequencies and percentages. Student t tests and non-parametric tests (Kruskal–Wallis) were used to compare normally and non-normally distributed variables, respectively. Differences in categorical variables were compared using the chi-square test. To test the relationship between all-cause, cardiovascular, and cancer mortality and SII levels as a continuous variable, restricted cubic splines (RCS) analyses were conducted. Time-to-event data were presented graphically using Kaplan–Meier curves and compared with a log-rank test. The association between SII levels and all-cause mortality was assessed by multivariate Cox regression models, unadjusted and adjusted for age, gender, anemia, AMI, CHF, DM, HT, stroke, percutaneous interventions (PCI), eGFR, low-density lipoprotein cholesterol (LDL-C), angiotensin-converting enzyme inhibitors and angiotensin receptor blockers (ACEI_ARB), beta-blocker, calcium channel blocker (CCB), statins. A variance inflation factor ≥ 5 can be considered as a criterion for multicollinearity [18], suggesting there was no synteny problem among the variables. To better visually depict the influence of the competing risk, we generated cumulative incidence function (CIF) curves for cardiovascular and cancer mortality. Gray's tests were used to assess differences between cardiovascular and cancer mortality. For cause-specific mortality analyses, we further excluded individuals with unknown causes of mortality. The associations of cardiovascular and cancer mortality with SII levels were examined using competing risk Fine and Gray models, unadjusted and adjusted for the same confounders. Mortalities due to other causes were treated as competing risks. Subgroup analyses were also performed to explore the source of heterogeneity according to age [(≥ 65 versus < 65 years of age], gender, CAD, and anemia. Stratification and interaction analyses showed consistency with the main results. To assess whether the accuracy of predicting long-term prognosis would improve after adding SII to a traditional clinical risk factors model (age, gender, HT, DM, CAD, AMI, CHF, stroke, anemia, and low-density lipoprotein cholesterol (LDLC), the C-index, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated. NRI assesses the ability of a model including a new prognostic marker to more accurately reclassify individuals into higher or lower risk (of death) categories compared to traditional clinical risk factor models. The IDI was used to estimate the discrimination improvement in average sensitivity and any potential increase in average 1-specificity [19]. Missing values in the candidate predictor variables were imputed using multivariate imputations by the chained equation method. Data management and statistical analyses were performed by R version 4.0.3. We considered a two-sided P value < 0.05 statistically significant.

Results

Baseline characteristics

Among the 19,327 patients with CKD (the mean age 68.8 ± 10.03 years, 32.0%were women) were included. All subjects were categorized into four groups according to SII levels at baseline: Q1 group (n = 4831), Q2 group (n = 4832), Q3 group (n = 4832) and Q4 group (n = 4832). As baseline SII levels increase, patients are more likely to combine with AMI, CHF, and anemia. On the contrary, higher SII levels were related to lower levels of HGB and eGFR. More data on the baseline characteristics of the study population are shown in Table 1.

Table 1 Baseline characteristics according to SII levels on admission categories
Full size table

With a median follow-up of 4.5 years, 5174 (26.77%) participants died of all causes, of which, 2861 (14.80%) were cardiovascular-specific deaths, and 375 (1.94%) were cancer-specific deaths. The all-cause mortality of Q1 group vs. Q2 group vs. Q3 group vs. Q4 group was 22.9 vs. 24.9% vs. 26.8 vs. 32.4% (p for trend < 0.001). Correspondingly, the rates for cardiovascular and cancer mortality were 12.0 vs. 13.3% vs. 15.3 vs. 18.6% (p for trend < 0.001) and 1.8 vs. 1.9% vs. 1.9 vs. 2.2% (p for trend < 0.001) (Fig. 1).

Fig. 1
figure 1

The incidence of all-cause, cardiovascular, and cancer mortality among the four groups

Full size image

Systemic immune inflammation index and clinical outcomes

When SII as a continuous variable, we performed multivariable regression analysis and it was significantly associated with all-cause, cardiovascular and cancer mortality (all-cause mortality: adjusted hazard ratio [aHR], 1.20, 95% confidence interval (CI): 1.14–1.26, p < 0.001; cardiovascular mortality: aHR, 1.27, 95% CI 1.19–1.36, p < 0.001; cancer mortality: aHR, 1.23, 95% CI 1.04–1.47, p < 0.001) (Table 2). Subsequently, we completed univariate and multivariate regression analysis as a categorical variable to further verify the association of SII levels with the increased risk of total or cause-specific death. After adjusting for confounders, all-cause mortality (Q2, Q3, Q4: HR [95 CI%] = 1.15 [1.06–1.26], 1.30 [1.19–1.42], 1.48 [1.35–1.62], respectively; p for trend < 0.001) and cardiovascular mortality (Q2, Q3, Q4: HR [95 CI%] = 1.16 [1.03–1.31], 1.40 [1.24–1.58], 1.64 [1.44–1.85], respectively; p for trend < 0.001) increased with higher levels of SII. And SII levels were related to an increased risk of cancer mortality comparing the highest quartile to the lowest quartile of SII (Q2, Q3, Q4: HR [95 CI%] = 1.12 [0.83–1.52], 1.22 [0.90–1.67], 1.50 [1.09–2.08], respectively; p for trend < 0.001) (Table 2). Kaplan–Meier curves for all-cause mortality and cumulative incidence function curves for cardiovascular mortality manifested that the highest quartile group displayed the worst all-cause and cardiovascular death in a significant SII levels-dependent manner between four groups. While the highest quartile group showed a significantly increased cancer mortality (Fig. 2). In addition, RCS analysis demonstrated a nearly linear increase in the risk of all-cause mortality (nonlinear p = 0.160) and cardiovascular mortality (nonlinear p = 0.229) as the SII increased. But there is a nonlinear relationship between cancer mortality and SII levels (nonlinear p = 0.032): SII evidently increased the risk of cancer death at relatively high levels (Fig. 3).

Table 2 Univariable and multivariable cox regression analysis of the association between SII levels on admission and death
Full size table
Fig. 2
figure 2

SII levels on admission and risk of all-cause, cardiovascular, and cancer mortality

Full size image
Fig. 3
figure 3

Restricted spline curve for hazard ratio of all-cause, cardiovascular, and cancer mortality

Full size image

Discrimination and calibration analysis of SII added to traditional risk factors model

The addition of the continuous variable SII to the clinical risk factors model did contribute to a certain increase in C-statistics, which ranged from 0.613 to 0.626 (p = 0.0034) for all-cause mortality and 0.617 to 0.628 (p = 0.0194) for cardiovascular mortality. Reclassification adding SII also displayed an IDI of 0.0014 (p < 0.0001) with 8.3% improvement in NRI (p < 0.0001) in all-cause mortality and cardiovascular mortality (NRI; 0.0743, p = 0.0005; IDI: 0.0003, p = 0.0417), but no significant improvement in cancer mortality (C-statistics:0.618 to 0.620, p = 0.3017, NRI; 0.1162, p = 0.0291; IDI: 0.0001, p = 0.3715) (Table 3).

Table 3 Improvement in discrimination performance and calibration for risk prediction of all-cause, cardiovascular and cancer mortality in the multivariate-adjusted model after including SII
Full size table

Subgroup analysis

When results were stratified by age, gender, anemia, and CAD, a similar prognostic impact on all-cause mortality of different SII levels was found for all the subgroup analyses except for female patients (Supplementary Fig. 2). As concerns the association of SII levels with cardiovascular and cancer mortality stratified by the aforementioned variables, the results demonstrated that there is a consistent effect in subjects (males, anemia, CAD, non-CAD, and regardless of age) for cardiovascular mortality (Supplementary Fig. 3), while there is a consistent effect in patients (youngers, males, anemia, and non-CAD) for cancer mortality. Whereas stratification of inflammation levels was not significantly associated with cancer death in CAD patients (Supplementary Fig. 4). Stratified and interaction analyses demonstrated the association of SII levels with cardiovascular and cancer mortality was evident in gender (p for interaction = 0.004), and age (p for interaction < 0.001), respectively.

Discussion

To our knowledge, this is the first largest multicenter, longitudinal cohort study to report the associations of immune inflammation levels on admission with the risk of all-cause, cardiovascular, and cancer mortality among CKD patients. The results indicated that the highest quartile group showed the worst all-cause and cardiovascular mortality in a significant SII levels-dependent manner among the four groups, while there was a significant impact on cancer mortality when SII was at a high level. Moreover, adding SII to a traditional risk factors model improved the clinical prediction of long-term prognosis. Our findings suggested it is important to account for SII levels in the assessment of mortality risk. Monitoring SII levels may be considered as one focus of reduced mortality efforts in subjects with CKD.

The Global Burden of Disease studies have shown that CKD has emerged as a leading cause of worldwide mortality [20, 21]. The mechanisms of increased mortality among CKD are complicated and involve multiple pathophysiologic alterations, of which, heightened inflammation and immune dysfunction may play an important role in this process [22, 23]. It has been reported that increased levels of inflammatory markers are associated with higher mortality. Elevated platelet levels are related to higher mortality and risk of cardiovascular events, and platelet count may be a useful marker for further stratification of cardiovascular risk, and especially of mortality [24]. In addition, previous research demonstrated that blood neutrophil count may play an important role in accelerating atherosclerosis, which is related to all-cause and cardiovascular mortality in apparently healthy men [25]. Meanwhile, Christelle et al. found that elevated lymphopenia levels are an independent prognostic factor for overall survival (OS), which is conducive to risk stratification when assessing OS among pancreatic cancer subjects [26]. SII, a new and widely available hematologic marker of inflammation, had a superior ability to predict the risk of poor outcomes among subjects with cancer and cardiovascular disease [14, 15]. Our previous studies also have displayed that elevated SII levels were an independent prognostic factor for all-cause mortality following CAG [27]. However, the relationship between SII levels and prognosis among the CKD population is not yet clear. This research indicated that higher SII levels were closely associated with the increased all-cause, cardiovascular, and cancer mortality among CKD subjects, and adding SII to the traditional risk factors model could improve predictive performance. Therefore, it is necessary to monitor SII levels among CKD patients, which could significantly improve long-term prognosis.

The findings in the current study demonstrated that more than half of deaths in patients with CKD are due to cardiovascular disease, and elevated SII levels were a significant risk factor for cardiovascular-specific deaths. CKD continually leads to chronic impairment of cardiac function, and the communication between the heart and kidney occurs through different mechanisms [28]. Of which, inflammatory response is involved in the progression of arterial stiffening among CKD subjects, which could lead to increasing cardiac workload, reducing coronary artery perfusion pressure, and microvascular cardiac ischemia [29]. Moreover, inflammation plays a key role in the progression of coronary artery atherosclerosis [30]. Consistent with previous research, anemia and stroke could significantly increase the risk of all-cause and cardiovascular death. Anemia contributes to increased cardiac output, the development of left ventricular hypertrophy, and congestive heart failure, and the presence of anemia is associated with worse outcomes among patients with CKD [31]. Moreover, CKD is strongly related to ischemic and hemorrhagic stroke, which may have severe neurological deficits and poor vital and functional outcomes because of the limitations of pharmacotherapies [32].

The prevalence of 5-year cancer mortality was 2.0% among patients with CKD in our study. Previous research has highlighted conflicting results about the association of CKD with overall cancer risk. A meta-analysis including 32,057 participants reported a non-significant relationship between reduced renal function and cancer risk, but did with urological cancers [33], Similarly, Guo et al. also found CKD is significantly associated only with genitourinary cancer among different types of cancer [4]. On the other hand, Kitchlu et al. reported that cancer risk was increased in mild to moderate CKD and among transplant recipients, but not in advanced kidney disease. And patients with kidney disease are related to higher cancer death in a large US cohort research [3]. On the other hand, previous studies showed that elevated levels of systemic inflammation markers are related to increased cancer risk and mortality, SII displayed the strong association with cancer risk in many sites evaluated including colorectal, kidney, liver, lung, lymphoma, and myeloma cancers [34]. Moreover, a body of studies showed that SII could be used as a potential prognostic marker among subjects with cancers, elevated SII levels were significantly related to adverse outcomes [14, 35]. Consistently, current study also found that elevated SII levels increased the risk of cancer deaths in patients with CKD.

SII could improve the predictive performance of the traditional risk factors model. Therefore, SII may be a useful tool to assess the risk of death among patients with CKD and to help clinicians identify high-risk population. Besides, based on the association between SII and the risk of mortality, it is important to monitor SII levels and control inflammation with certain anti-inflammatory therapies to reduce the risk of mortality among high-risk population. In addition, ACEI/ARB treatment could protect the heart and kidney function given its anti-inflammatory effects [36]. Moreover, managing comorbidities like stroke, CHF as well as anemia is essential to reduce the risk of mortality. Finally, further investigation is needed to prospectively validate the prediction of SII for all-cause, cardiovascular, and cancer mortality among CKD patients.

Limitation

First, this is a retrospective observational study with the subsequent disadvantages secondary to its nature, so our inferences did not reflect direct causality. Second, our study enrolled CKD patients undergoing CAG, but our research still included a number of patients without CAD, and subgroup analysis showed that our results were still stable in the non-CAD group. Third, despite adjusting for potential confounders, we cannot entirely exclude the possibility that residual, uncontrolled confounding may explain the associations. Fourth, we only evaluated SII levels on admission and did not record changes during the follow-up period. The post-discharge changes in SII level fail to be acquired, and the impacts on long-term mortality remained uncertain. Finally, well-designed and well-executed future research should be conducted to examine the predictive validity of the SII levels on admission for long-term prognosis.

Conclusions

Elevated immune inflammation level is an independent risk factor for all-cause, cardiovascular, and cancer mortality among CKD patients, and SII improved the risk prediction of long-term prognosis than traditional risk factors. Our study demonstrates that the SII levels may provide a more accurate risk stratification for the high-risk CKD population and may prompt an important reference for the subsequent improvement of long-term prognosis among CKD patients through monitoring SII levels.