Effect of steroids and relevant cytokine analysis in acute tubulointerstitial nephritis

The study design was approved by the institutional review boards of all involved centers (nos. H-1801-043-914, B-1801/447–409, and 20,180,124/30–2018-3/023) and complied with the Declaration of Helsinki. Informed consents were waived under this approval. A total of 7196 patients had a kidney biopsy performed at three tertiary referral hospitals (Seoul National University Hospital, Boramae Medical Center, and Seoul National University Bundang Hospital) over 18 years (from January 2000 to December 2017). Among them, 270 patients were diagnosed with ATIN. Patients were excluded from the analysis if they also had glomerulonephritis (n = 122), pyelonephritis (n = 3), or other pathologies, including thrombotic microangiopathy (n = 6) and atheroembolism (n = 1), they were younger than 18 years (n = 5), and they had an insufficient follow-up period (< 3 months) (n = 11). We also excluded patients without azotemia at the time of biopsy (n = 9: 5 were renal tubular acidosis, 2 were drug-induced ATIN, 1 was renal tuberculosis, and 1 was related with Sjögren syndrome) because the effect of steroids on outcomes of ATIN could not be examined in these patients. Finally, the analyzed cohort comprised 113 patients.

All of the patients’ clinical data such as age, sex, body mass index, comorbidities, including diabetes mellitus, hypertension, and chronic kidney disease, renal outcomes, and steroid treatment for ATIN, were collected. Baseline blood laboratory data, such as serum creatinine, hemoglobin, cholesterol, uric acid, and albumin levels, were measured. The dipstick test was used to evaluate proteinuria, which was semi-quantitatively scored. Microscopic evaluation of urine was performed to define hematuria and pyuria. The random urine protein to creatinine ratio was measured at the time of biopsy, but this variable was not available in five patients. The etiology of ATIN was classified into idiopathic if there were no specific factors inducing kidney dysfunction. Two nephropathologists reviewed all of the slides and semi-quantitative graded scores for tubular atrophy and interstitial fibrosis, and leukocyte infiltration, according to the Banff working classification [18].

Renal recovery, which was primary outcome, was defined as a decrease in serum creatinine levels to ≤1 .3mg/dL or ≥ 50% from its peak value. Renal recovery was evaluated at multiple time points, such as at 6 months after kidney biopsy and at the last follow-up. The outcome of ESRD was defined as initiation of renal replacement therapy or kidney transplantation due to failed kidney function. The events of ESRD and all-cause mortality were obtained from the Kidney Renal Registry and the National Database of Statistics Korea, respectively.

Thirty three patients’ plasma and urine were acquired at the time of kidney biopsy and before any treatment was provided. Forty healthy individuals’ plasma and urine were also recruited at the time of a health checkup for comparison with the patients. We performed a bead-based multiplex assay according to the manufacturer’s instruction (LEGENDplex™; BioLegend, San Diego, CA, USA). This assay included 13 cytokines and chemokines which were known to play a crucial role in renal ischemia-reperfusion injury [19‐22] and were possibly related to tubular inflammation pathways [23], such as interleukin (IL)-1β, interferon (IFN)-α2, IFN-γ, tumor necrosis factor-α, monocyte chemo-attractant protein-1, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-18, IL-23, and IL-33. Cytokine- and chemokine-bound beads were read on a flow cytometer (BD LSRFortessa™; BD Biosciences, San Jose, CA, USA). Data were analyzed using the analysis program recommended by the manufacturer’s instructions (LEGENDplex™, Data Analysis v8.0; VigeneTech, Carlisle, MA, USA).

All statistical analyses were performed using SPSS version 23.0 (IBM Corp., Armonk, NY, USA) and R software (version 3.4.1; The Comprehensive R Archive Network: http://​cran.​r-project.​org). Continuous variables are expressed as the mean value and standard deviation if they had a normal distribution and as the median value and interquartile range if they were not normally distributed. A normality test was performed using the Kolomogrov–Sirnov test. Categorical variables are expressed as proportions. The chi-square test was used for comparison of categorical variables (Fisher’s exact test if not applicable) and the Student’s t-test was used for continuous variables. The Mann–Whitney test was used if variables were not normally distributed. Analysis of variance analysis was used when normality was satisfied for comparison between group, and the Kruskal–Wallis test was used if normality was not satisfied. Logistic and Cox proportional hazard ratio models were conducted to calculate odds ratios (ORs) and hazard ratios (HRs) of the outcome risk, respectively. All of the covariates shown in Table 1, except for the random urine protein to creatinine ratio (missing in five cases), were adjusted in the multivariate models. To assess the effect of steroids in ATIN, we pooled the data of three previous retrospective cohort studies, and compared the ESRD progression rate between the steroid-treated and non-treated groups. Pooled estimates of the relative risks and 95% confidence intervals (95% CIs) were obtained using the DerSimonian and Laird random effects model. Heterogeneity was assessed using the Cochran Q statistic and I². All P values were set to two-sided, and values ≤0.05 were defined as significant.

Baseline characteristics of the patients are shown in Table 1. The etiology of ATIN was idiopathic in 77 (68.1%) patients, drug-induced in 25 (22.1%), autoimmune in five (4.4%), malignancy in five (4.4%), and infection in one (0.8%). In the case of drug-induced etiology, herbal medication was the most common cause (n = 8, 32%), followed by nonsteroidal anti-inflammatory drugs or salicylate (n = 6, 24%), antibiotics (n = 6, 24%), proton pump inhibitors (n = 2, 8%), and others (2 cases of oxaliplatin, 1 case of tranexamic acid). In the case of autoimmune ATIN, there were three cases of anti-neutrophil cytoplasmic antibody-associated vasculitis, one case of tubulointerstitial nephritis and uveitis, and one case of Sjögren syndrome. All cases of malignancy (n = 5) were multiple myeloma, and infectious-ATIN was associated with Hantaan virus.

The mean age of the patients was 58 (±15) years and 60% of them were men. At the time of biopsy, the median value of serum creatinine was 3.77 mg/dL and the peak level was 4.50 mg/dL. Dialysis at the time of biopsy was performed in 30.1% of patients, and steroids were used for 81.4% of patients. Patients with autoimmune disease or malignancy were treated with steroids, but the patient with infectious ATIN was not treated. The patients were followed up for a median of 33 months (interquartile range, 19–55 months; maximum of 12 years).

Table 2 shows the overall renal outcomes according to the etiology of ATIN. The rate of renal recovery was 54.9% at 6 months after biopsy and 73.5% during the entire follow-up period. The rate of ESRD progression was 25.7% and the all-cause mortality rate was 17.7% in all patients. Of five patients with malignancy-associated ATIN who were included in the others category, four progressed to ESRD and died. This was the reason for differences in the rates of renal recovery and mortality among the groups.

Steroid use in cases of autoimmune diseases and multiple myeloma was essential for treating the diseases, and steroid use in infectious disease was usually not recommended. Therefore, patients with autoimmune- and malignancy-associated, and infectious ATIN were excluded from the analyses to determine the effect of steroid use. To evaluate the effect of steroid use on outcomes, 102 patients with idiopathic and drug-induced ATIN were analyzed after stratification by the use of steroids. Steroids were used in 80.3% of patients, and the median value of initial steroid dose was 30 mg/day (30–60 mg/day as an equivalent dosage for prednisolone) and the median duration of use was 14 weeks (9–30 weeks). The steroid-treated group was having lower body mass index, less chronic kidney disease, lower uric acid level, and lower score of tubular atrophy and interstitial fibrosis than did the non-treated group (Table 3).

Figure 1 shows the Kaplan–Meier curves of the rate of renal recovery and risks of ESRD and mortality depending on the use of steroids. None of the renal outcomes, including renal recovery and the risks of ESRD and all-cause mortality, were different between the steroid-treated and non-treated groups. Despite adjustment of multiple variables, the steroid-treated group did not have favorable outcomes compared with the non-treated group (Table 4). To balance all of the observed baseline variables between groups, we further used 1:1 propensity score matching. Propensity scores were estimated by logistic regression, with several variables, including age, sex, body mass index, diabetes mellitus, hypertension, chronic kidney disease, serum creatinine levels at biopsy, uric acid levels, albumin levels, proteinuria, hematuria, tubular atrophy and interstitial fibrosis, leukocyte infiltration, dialysis at biopsy, and serum creatinine levels at the time of kidney biopsy. The steroid-treated group had similar rates of renal recovery at 6 months (58.5% vs. 50.0%; OR, 1.05; 95% CIs 0.258–4.269; P = 0.947), renal recovery at the last follow-up (78.0% vs. 65.0%; HR 0.82; 95% CIs 0.401–1.674; P = 0.583), ESRD (23.2% vs. 25.0%; HR 0.76; 95% CIs 0.188–3.032; P = 0.693), and all-cause mortality (14.6% vs. 20.0%; HR 0.33; 95% CIs 0.049–2.174; P = 0.248) compared with the non-treated group (Table 4). In the regression models, diabetes mellitus, but not the other variables, was associated with the renal recovery rate (Additional file 1: Table S1).

Subgroup analysis was performed in the steroid-treated group to determine whether factors related to steroid use, such as pulse therapy of methylprednisolone, initial steroid dose, and delay of starting steroids affected renal outcomes. After adjustment of multiple variables, none of variables were associated with renal recovery at 6 months, including methylprednisolone pulse therapy (OR 1.02 vs. none; 95% CIs 0.987–2.210; P = 0.276), initial steroid dose (OR 0.97 per 1 mg of prednisolone dose; 95% CIs 0.894–1.044, P = 0.381), and a delayed start of steroids (OR 1.00 per 1 day of delay; 95% CIs 0.964–1.041; P = 0.924).

We performed pooled analysis on the effects of steroids on the outcome of ESRD in drug-induced ATIN from three studies [12, 14, 15], as well as the present study (n = 347). We used the DerSimonian and Laird random effects model because of heterogeneity between studies. There was no significant difference in progression of ESRD between the steroid-treated and non-treated groups (relative risk, 1.81; 95% CIs 0.599–5.474; P = 0.292) (Fig. 2).

The above results showed that steroid therapy did not affect the overall outcomes of ATIN. Accordingly, we need to adopt other treatment options to improve patient’s outcomes. Nevertheless, understanding of the pathophysiology of ATIN is not clear. Therefore, we attempted to determine which inflammatory cytokines or chemokines are expressed in patients with ATIN. Plasma levels of IL-1β, IFN-α2, tumor necrosis factor-α, monocyte chemo-attractant protein-1, IL-6, IL-8, IL-17A, IL-18, and IL-23 were higher in patients with ATIN than in healthy individuals (Fig. 3a). Urine levels of IFN-α2, monocyte chemo-attractant protein-1, IL-6, IL-8, IL-12p70, and IL-17A were higher in patients with ATIN than in healthy individuals (Fig. 3b). The other cytokines including IL-10, IFN-γ, IL-23, IL-33 were not different between patients and healthy individuals.