Neutrophil-to-Lymphocyte Ratio as a Predictor of Worsening Renal Function in Diabetic Patients (3-Year Follow-Up Study)

Abstract

Background: Previous studies have demonstrated the role of inflammation in diabetic nephropathy (DN). Neutrophil to lymphocyte ratio (NLR) rather than other white cell parameters was found to be a useful inflammatory marker to predict adverse outcomes in medical and surgical conditions. Nevertheless, the value of NLR in predicting DN has not been elucidated. Method: An observational study included 338 diabetic patients, who were followed at our clinic between 2007 and 2009. We arranged our patients into tertiles according to their 2007 NLR. The primary outcome was continuous decrease of GFR >12 mL/min between 2007 and 2009 with the last GFR <60 mL/min. Result: The lowest NLR tertile had fewer patients (2.7%) with primary outcome (i.e., worsening renal function) compared with middle and highest NLR tertiles, which had more patients with primary outcomes (8.7% and 11.5%, respectively) with a significant p-value 0.0164. When other potential confounders were individually analyzed with NLR tertile, the NLR tertiles remained a significant predictor of poor GFR outcome in the presence of other variables (hemoglobin A1C, systolic blood pressure, diastolic blood pressure, age, and congestive heart failure with p-values 0.018, 0.019, 0.017, 0.033, and 0.022, respectively). Conclusion: NLR predicted the worsening of the renal function in diabetic patients. Further studies are needed to confirm this result.

• neutrophil
• lymphocyte
• NLR
• diabetes mellitus
• chronic kidney disease

INTRODUCTION

Diabetic nephropathy (DN) is the leading cause of renal failure in the United States.^1,2 Over the last 20 years, the prevalence of end-stage renal disease (ESRD) markedly declined among diabetes mellitus (DM) type I population to be 2–7%.^3–5 This decline is believed to be due to more diabetic care, which included the usage of hemoglobin A1C targeting treatment and availability of early markers for DN (e.g., microalbuminuria). Similar decline in the prevalence of ESRD in DM type II was also found over the last few decades.⁶ Microalbuminuria was found to be one of the earliest markers for DN.⁷ The reduction in microalbuminuria was associated with a slower decline in glomerular filtration rate (GFR) and decreased cardiovascular risk.^8,9 Later, the role of inflammation and inflammatory cytokines in DN was studied over the last few years.^10–16 Inflammation secondary to ischemia and endothelial dysfunction was found to be associated with the development of DN.^17–19 Some medications were found to slow down the progression of DN through suggested anti-inflammatory actions.^20–26 Neutrophil to lymphocyte ratio (NLR) rather than other white cell parameters (e.g., total white cell, monocyte count, and absolute neutrophil count) was found to be a useful inflammatory marker that predicts adverse outcomes in many medical and surgical conditions.^27–30 This finding was found not only in acute conditions but also in stable and preoperative medical conditions.^31–33 In a cross-sectional study, Chung et al.³⁴ found a higher neutrophil and lower lymphocyte count in patients with microalbuminuria and overt nephropathy compared with diabetics without albuminuria. Nevertheless, the usefulness of NLR as a predictor of worsening of renal functions among diabetic patients has not been elucidated.

METHODS

In a retrospective longitudinal study, our study population of nonselected 338 patients, who visited our diabetes clinic from January to December 2007, was followed up until December 2009. All included patients underwent complete physical exam and routine biochemical analysis [including complete blood count (CBC) and basic metabolic panel (BMP)] in 2007 and had at least one set of BMP on their 2008 and 2009 follow-up visits. We excluded patients with active cancer, hematological proliferative disorder, active or chronic inflammatory or autoimmune disease, and those who received steroids or any other immunosuppressive therapy or chemotherapy just before or during the study period. Before the commencement of the study, the primary end point was set to be a continuous dropping of the estimated GFR (eGFR) more than 12 mL/min over the 3-year follow-up period, with the last eGFR <60 mL/min. Each included patient’s chart was reviewed by two independent physicians. The hematological variables [total white blood cell (WBC), neutrophil, lymphocyte, and monocyte absolute counts, hemoglobin, hemoglobin A1C] were obtained by using Coulter counter® technique (Coulter Gen.S Hematology Analyzer, Beckman Coulter Corp., Hialeah, FL, USA). The GFR was estimated with the 4-variable Modification of Diet in Renal Disease (MDRD) equation using serum creatinine, age, race, and gender. Positive smoking status included both current and prior smokers. The study was approved by the Institutional Review Board at Staten Island University Hospital, New York.

Statistical Analysis

Patients were divided into three equal groups (tertile model) according to their NLR in 2007 (cutoff points were the 33rd and 66th NLR values). The distributions of continuous and categorical variables were presented as means, standard deviations, frequencies, and percentages. Group comparisons used χ² analysis for categorical data. Continuous variables were compared with analysis of variance or Kruskal–Wallis analysis of ranks, depending on the probability distribution of the variable. All probabilities were two sided and p-values <0.05 were considered statistically significant. The exact two-tailed Cochran–Armitage test for trend was used to test for an association between NLR tertile and the incidence of a poor GFR outcome. A univariate screen of all potential predictors of poor GFR outcome was performed. Due to the small number of poor outcomes (n = 26), a logistic model would not support more than one covariate in addition to NLR tertile. Accordingly, a modification to this multivariable approach was taken whereby several logistic models were constructed, corresponding to the NLR tertile and each of the above-mentioned potential confounders individually. The objective was to see whether the effect of tertile was changed when adjusting for a given covariate. All data were analyzed using the Statistical Analysis System, version 9.2 (SAS^® Institute, Cary, NC, USA).

RESULTS

The Baseline Characteristics

The patients were arranged into three equal groups according to their baseline 2007 NLR values [first tertile (NLR < 1.6) = 110 patients, second tertile (NLR 1.6–2.36) = 115 patients, and third tertile (NLR > 2.36) = 113 patients]. There were significantly higher frequencies of male and Caucasian patients in the third (highest NLR) tertile compared with the lower NLR tertile(Table 1). Additionally, the patients in the higher NLR tertile were significantly older with higher average body mass index and higher frequencies of coronary artery disease and chronic lung disease compared with the lowest NLR tertile. Likewise, there was a trend of more patients with congestive heart failure in the higher NLR tertile. The rates of treatment with aspirin, nonsteroidal analgesics, clopidogrel, statin, angiotensin-converting enzymes inhibitor, and angiotensin receptor blockers were nonstatistically different among the three NLR tertile groups. There was no difference in baseline systolic blood pressure, diastolic blood pressure, and hemoglobin A1C among the three NLR tertiles.

Table 1. Baseline characteristics of our diabetic population by initial (2007) neutrophil to lymphocyte ratios (NLRs).

	NLR < 1.6		1.6 ≤ NLR ≤ 2.36		NLR > 2.36		p-Value
Demographic characteristics	N = 110	%	N = 115	%	N = 113	%
Gender (males)	33	30.0	40	34.8	49	43.4	0.0385
Race (Caucasians)	26/92	28.2	48/96	50.0	56/90	62.2	0.0001
Age (years, mean and SE)	56.05	1.15	60.03	1.07	58.18	0.96	0.0225
Weight (kg, mean, SE)	84.79	1.94	87.28	2.06	92.79	1.83	0.0061
BMI (mean, SE)	31.53	0.74	31.45	0.73	33.91	0.77	0.0333
Medical history and risk factors
Hypertension	87	(79.1)	91	(79.1)	83	73.5	0.5041
Coronary artery disease	11	10.0	36	31.3	30	26.6	0.0004
Congestive heart failure	2	1.8	10	8.7	10	8.9	0.0524
Cerebrovascular event	3	2.7	8	7.0	4	3.5	0.2600
Chronic lung disease	5	4.6	18	15.7	19	16.8	0.0092
Smoking	27	24.6	40	34.8	31	27.4	0.2162
Neuropathy	26	23.6	23	20.0	34	30.1	0.2013
Retinopathy	12	10.9	12	10.4	11	9.7	0.9589
Diabetic foot ulcer	15	13.6	12	10.4	22	19.5	0.1459
Medications
Statins	77	70.0	90	78.3	84	74.3	0.3665
NSAIDs	13	11.8	13	11.4	12	10.6	0.9594
Aspirin	64	58.2	71	61.7	64	56.6	0.7243
Clopidogrel	12	10.9	16	13.9	20	17.7	0.3463
Angiotensin receptor blocker	25	22.7	30	26.1	20	17.7	0.3088
Angiotensin convertase inhibitor	65	59.1	57	49.6	62	54.9	0.3553

Note: NSAIDs, nonsteroidal anti-inflammatory drugs.

NLR as a Predictor of the Primary Outcome

Patients in the higher NLR tertile had significantly higher baseline serum creatinine and lower eGFR compared with lower NLR tertile (Table 2). The lowest NLR tertile had fewer patients (2.7%) with primary outcome (i.e., dropping GFR >12 mL/min/3 years) compared with middle and highest NLR tertiles, which had more patients with primary outcomes (8.7% and 11.5%, respectively) with a significant p-value = 0.0164 according to the exact two-tailed Cochran–Armitage test (Figure 1). When each of the potential confounders was individually analyzed with the NLR tertile, the NLR tertile remained a significant predictor of poor GFR outcome in the presence of other variables (hemoglobin A1C, systolic blood pressure, diastolic blood pressure, age, and congestive heart failure with p-values 0.018, 0.019, 0.017, 0.033, and 0.022, respectively). When the urine albumin over creatinine ratio (UACR) was considered, the NLR showed a trend of poor outcome in the highest versus the lowest NLR tertiles (p = 0.055) and the UACR was a significant predictor of poor renal outcome (Table 3). Additionally, there was a significantly higher prevalence of chronic kidney disease (CKD) both in 2007 and 2009 in the highest NLR tertile compared with the lower tertiles (Figure 2). Further subanalysis demonstrated a higher primary outcome in the lowest lymphocyte tertile compared with the highest tertile (p = 0.01). However, the highest neutrophil tertile had a higher nonsignificant primary outcome compared with the lowest neutrophil tertile (p = 0.09).

Figure 1. Primary outcome (i.e., dropping of GFR >12 mL/min/3 years) according to initial (2007) neutrophil to lymphocyte ratio tertiles; p = 0.01 according to the two-tailed Cochran–Armitage test.

Figure 2. Prevalence of chronic kidney disease (CKD) according to initial (2007) neutrophil to lymphocyte ratio tertiles.

Table 2. Baseline characteristics of our diabetic population by initial (2007) neutrophil to lymphocyte ratios (NLRs).

	NLR < 1.6		1.6 ≤ NLR ≤ 2.36		NLR > 2.36
Clinical and laboratory variables	Mean	SE	Mean	SE	Mean	SE	p-Value
Systolic blood pressure	127.95	1.72	130.48	1.81	127.29	1.64	0.5628
Diastolic blood pressure	77.10	0.99	77.16	0.96	75.28	1.04	0.2293
White blood cell count (k/cc)	6.63	0.19	7.1	0.16	8.04	0.20	<0.00001
Neutrophil (k/cc)	3.21	0.11	4.21	0.11	5.51	0.15	<0.00001
Lymphocyte (k/cc)	2.67	0.08	2.15	0.05	1.71	0.05	<0.00001
NLR	1.22	0.03	1.97	0.02	3.44	0.12	<0.00001
Monocyte (k/cc)	0.47	0.02	0.51	0.01	0.54	0.02	0.0064
Hemoglobin (gm/dL)	13.05	0.13	12.85	0.15	12.84	0.16	0.4876
Hemoglobin A1C (%)	8.06	0.20	7.71	0.16	7.96	0.18	0.4608
Serum creatinine (mg/dL)	0.92	0.03	1.03	0.04	1	0.03	0.028
Blood urea nitrogen (mg/dL)	15.22	0.65	17.04	0.77	16.73	0.84	0.294
GFR (mL/min)	88.44	2.48	76.64	2.57	77.71	2.13	0.0014
Low density lipoprotein (mg/dL)	106.37	3.75	100.92	3.60	95.44	3.12	0.1152
High density lipoprotein (mg/dL)	43.30	1.45	42.61	1.46	40.43	1.30	0.2041
Urine albumin/creatinine (mg/gm)	92.34	27.28	172.21	48.61	211.7	47.45	0.2482

Table 3. The predictive value of NLR adjusted to other confounding variables.

Covariate included with NLR tertile	p-Value for covariate	p-Value for NLR tertile
UACR	<0.0001	0.0558
Hemoglobin A1C	0.3864	<0.0184
Systolic BP	<0.0188	<0.0198
Diastolic BP	0.0712	<0.0178
Age	0.1495	<0.0330
Gender	0.2152	<0.0256
Race	0.9300	<0.0124
Hypertension	0.9102	<0.0178
Congestive heart failure	0.4431	<0.0220

Note: UACR, urine albumin over creatinine ratio.

DISCUSSION

The Main Finding

The association between an increased neutrophil count and type II diabetes, insulin resistance, and DN was already described in the literature.³⁴ The novel finding in our study was that NLR as an inflammatory marker is an independent significant predictor of the worsening of the renal function in diabetic patients, even after adjusting for possible confounders such as age, blood pressure, glucose control, and congestive heart failure.

Superiority of NLR to Other Leukocyte Parameters

The NLR as a marker is superior to other leukocyte parameters (e.g., neutrophil, lymphocyte, and total leukocyte counts) due to the stability of NLR compared with the absolute counts that could be altered by various physiological, pathological, and physical factors. Moreover, NLR represents a combination of two markers of uncontrolled chronic inflammatory condition (i.e., high neutrophil and low lymphocyte). Prior studies demonstrated this superiority of NLR to the total leukocyte count^31,35 and to individual neutrophil and lymphocyte counts^31,36 in predicting adverse outcomes in cardiovascular diseases. In our study, the neutrophil and lymphocyte tertile-adjusted models were not significantly associated with worsening renal functions compared with the NLR model.

Neutrophils, Lymphocytes, Inflammation, and CKD

We postulate that a high neutrophil count is a marker of ongoing destructive nonspecific inflammatory process and a low lymphocyte count is a marker of inadequate regulatory and quiescent immunity pathway. Rapid progression of DN in particular has been shown to be linked to inflammation and episodes of acute renal failure.³⁷ Prominent clusters of inflammatory cells were detected in the renal biopsy specimens.³⁷ Post-ischemic inflammatory syndrome that accelerates diabetic CKD is a critical determinant of injury and can be successfully treated.³⁸ In cancer, neutrophil-derived reactive oxygen species release and subsequent events lead to enhanced angiogenesis.^39,40 A similar sequence of events may be responsible for the development of the abnormal angiogenesis that was described in DN as a microvascular complication of DM.^41,42 The increased spontaneous adherence of neutrophil to endothelial cells was also described as a possible mechanism of DN and proteinuria.⁴³ Moreover, studies demonstrated the usefulness of neutrophil gelatinase-associated lipocalin (NGAL) as a significant marker of DN and CKD progression.^44,45 Interestingly, myeloperoxidase (MPO), which is secreted during activation of neutrophils, may serve as one mechanistic link among persistent inflammation, oxidative stress, and cardiovascular disease. Statin use (p < 0.01) was significantly associated with a decreased MPO activity.⁴⁶ On the other hand, Chung et al.³⁴ demonstrated the association between low lymphocyte count and the severity of nephropathy. Likewise, lymphopenia was associated with adverse outcomes among coronary artery disease and congestive heart failure patients.^47,48

NLR (Marker, Factor, or a Predictor)

While Chung et al.³⁴ reported the association between leukocytosis, low lymphocyte count, and DN in their cross-sectional study, we are reporting not only the higher prevalence of CKD with high NLR but also the worsening of renal function over the follow-up period. We suggest that NLR is not only a marker of presence of nephropathy but a factor in the pathogenesis of the disease. However, we cannot validate the causality of high NLR to this worsening of renal function, which warrants further studies.

Limitations

Although we prospectively followed our study population, the patients were retrospectively recruited from our database. The lack of ethnic diversity in this single-center study limits the generalizability of our findings to other populations. We used the 4-variable MDRD study equation to estimate GFR, which may be imprecise, particularly in the normal renal function range; nonetheless, we used it to assess eGFR changes in the same participants and not as a screening tool for CKD. Finally, the study lacks data on other inflammatory markers (ESR, C-reactive protein, others).

CONCLUSIONS

Higher NLR is a significant predictor of higher short- and long-term all-cause mortality in multiple conditions, and it seems to have the same importance in the prediction of worsening renal function in diabetics. Further prospective, multicenter studies are needed to verify this result, and then, if validated, the reasons for this association of the NLR with CKD in the diabetic population need to be studied at the molecular level. Is the NLR a marker of an etiologic agent in renal disease progression? Can altering the NLR change the prognosis? A proteomic discovery of DN biomarkers that may correlate with higher NLR may be another application to our finding.

Declaration of interest:The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.