HLA-DQA1, -DQB1, and -DRB1 Alleles Associated with Acute Tubulointerstitial Nephritis in a Chinese Population: A Single-Center Cohort Study

Acute tubulointerstitial nephritis (ATIN) is a common pattern of acute kidney injury (AKI) and accounts for 15–27% of intrinsic renal AKI biopsy specimens (1). The cause varies between cases, with drug hypersensitivity being the most common, followed by systemic autoimmune diseases, tubulointerstitial nephritis and uveitis (TINU) syndrome, infections, and malignancies (2). Although cellular immunity is involved in the disease initiation and progression (3), the pathogenesis of tubulointerstitial nephritis remains unclear.

HLA genes are located on the short arm of chromosome 6 and encode numerous immunologically functional molecules, including HLA class I and II molecules. HLA genes have extremely high levels of polymorphism and heterozygosity and are associated with most autoimmune disorders (4, 5). Over the past decade, HLA alleles conferring genetic susceptibility to TINU syndrome have been revealed in small case series by researchers from Finland (HLA-DQA1*0104, -DRB1*14) (6), Spain (HLA-DQB1*01, -DR*14) (7, 8), Italy (HLA-DQA1*0102, -DQB1*0503) (9, 10), the US (HLA-DRB1*0102,-DQB1*05) (11–13), Australia (HLA-DQA1*01,-DQB1*05) (14), Germany (HLA-DQB1*0503,-DRB1*1401) (15), Japan (HLA-DQB1*050101, -DQB1*050201) (16, 17), the U.K. (HLA-DR*14), (18) and China (HLA-DQB1*0503) (19) (Supplemental Table I). These reports provided evidence for the relevance of HLA-DQA1, -DQB1, and -DRB1 to TINU syndrome in different populations. However, no studies have focused on HLA genetic susceptibility in ATIN because of other causes, and the specificity of the reported TINU-related HLA susceptibility alleles has not been well evaluated.

In the current study, we examined the HLA-DQA1, -DQB1, and -DRB1 allele frequencies in 154 adult patients of a Chinese Han cohort with ATIN of various causes. Associations between specific HLA alleles/haplotypes and drug hypersensitivity–related ATIN (D-ATIN) and TINU syndrome were analyzed. Potential clinicopathological associations of the specific HLA haplotype were further investigated.

Patients were from a prospective cohort of adults (≥18 y) with ATIN that had been clinicopathologically diagnosed in the Renal Division of Peking University First Hospital from Jan 1, 2005, to Dec 30, 2015. The enrollment criteria included the following: 1) consent to the study, 2) availability of DNA samples from peripheral WBCs, and 3) availability for follow-up for at least 12 mo after biopsy. Patients who had glomerular diseases or hereditary renal diseases were excluded from the study. Those who had ATIN secondary to malignancy infiltration or deposition, such as lymphoma, leukemia, L chain disease, and multiple myeloma, were also excluded (Fig. 1). Two hundred healthy individuals, who were ethnically matched and voluntarily recruited from Chinese blood donors, were included in the control group. All participants in this study were self-reported Chinese Han. The study was approved by the Committee on Research Ethics of Peking University First Hospital.

The cause of ATIN was determined after renal biopsy, according to the most likely cause determined by the treating nephrologists and verified through follow-up by a working team that included nephrologists and ophthalmologists. All patients were screened for autoimmune diseases, malignancy, and infectious diseases, and they accepted ophthalmological examination to identify TINU syndrome during hospital stay. Diagnoses of D-ATIN, TINU syndrome, Sjogren’s syndrome (SS), and IgG4-related ATIN (IgG4-RD) were made based on previously described criteria (20–23). Clinical parameters and laboratory data were documented. Renal dysfunction was evaluated by the levels of serum creatinine (Scr). AKI was defined and staged according to the Kidney Disease: Improving Global Outcomes criteria (24). Estimated glomerular filtration rate (eGFR) was calculated by the Chronic Kidney Disease Epidemiology Collaboration equation and is expressed as milliliters per minute per 1.73 m² (25). Patients were scheduled for monthly follow-ups for 6 mo and then every 3 mo until at least 1 y after renal biopsy. Clinic visits included event investigation, physical examination, laboratory testing, and treatment.

Standard processing of kidney biopsy specimens included light microscopy, immunofluorescence, and electron microscopy. For light microscopy, all cases were stained with H&E, periodic acid–Schiff, Masson’s trichrome, and Jones methenamine silver. A pathologic diagnosis of ATIN required both the presence of prominent interstitial inflammation in the nonfibrotic cortex and tubulitis. Semiquantitative scores for classifying tubular injuries (tubular brush border loss, necrosis, and atrophy) and interstitial injuries (interstitial edema, inflammation, and fibrosis) referred to the criteria of the Banff working classification. We applied a 0–4+ scale as follows: 0 = no lesion, 1+ = ≤ 25% of parenchyma affected by the lesion, 2+ = > 25–50%, 3+ = > 50–75%, and 4+ = > 75% (26, 27).

Immunofluorescence staining was used to identify interstitial-infiltrating inflammatory cells and expression of HLA-DR and -DQ in renal tubules and interstitial cells. The Abs were obtained commercially as follows: Abs against CD3, CD4, CD8, CD68, and CD20 were from Zsbio (Beijing, China), and Abs against CD38, neutrophil elastase, HLA-DR, and HLA-DQ were from Abcam (Cambridge, U.K.). Immunostaining results were evaluated using the single-blind method. All nonoverlapping microscopic fields (×400) of the cortical interstitial area without glomeruli and vessels were selected. For infiltrating cell staining, the mean counted number of positive cells per tubulointerstitial field was calculated. The number of tubules that were positively stained with HLA-DR/DQ was counted under ×400 magnification and expressed as the number of positive tubular cells/×400.

Peripheral blood samples with EDTA anticoagulant were collected from patients with ATIN and healthy controls. Genomic DNA was extracted using a Gentra Puregene Blood Core Kit C (QIAGEN, Germantown, MD) and stored at −80°C. All HLA genotypings were performed in the same laboratory at Beijing Search Biological Technology Co. For HLA-DQA1, the alleles were typed using electrophoresis. HLA-DQB1 alleles were typed by bidirectional sequencing of exons 2 and 3, and HLA-DRB1 alleles were typed by bidirectional sequencing of exon 2, using SeCore SBT Kits (Invitrogen, Brown Deer, WI). HLA typing results were analyzed using the uTYPE 4.0 SBT software.

Statistical analysis was performed using SPSS 17.0 statistical software (SPSS, Chicago, IL). Normally distributed variables are expressed as the mean ± SD and were compared using the t test. Nonparametric variables are expressed as the median and interquartile range and were compared using the Mann–Whitney U test. The HLA allele polymorphism, in both patients and control groups, and the Hardy–Weinberg equilibrium test at each locus in controls were examined by the Arlequin software version 3.5 (28). HLA haplotype analysis was performed with the R Project package Haplo Stats, version 3.3.1. The χ² or Fisher exact test was used to compute the statistical significance of associations between the risk or protective alleles in ATIN patients. The associations are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Correction for multiple comparisons of the detected alleles between ATIN subgroups and controls was carried out by false discovery rate correction software (specifically, an R language–based software package using p.adjust algorithms), and p < 0.05 was considered statistically significant. For comparisons between the ATIN subgroups, the p value was set as 0.008 (0.05/6).

Altogether, 154 patients who were clinicopathologically diagnosed with ATIN were enrolled in this study (Fig. 1). Their age was 47.6 ± 12.2 y old, with female predominance (66.2%). The causes of ATIN included D-ATIN in 76 cases (49.4%), TINU syndrome in 38 cases (24.7%), SS in 25 cases (16.2%), IgG4-RD in seven cases (4.5%), and other causes in eight cases (5.2%). Representative histologic images of patients with ATIN of different causes are shown in Fig. 2. Forty-two patients (27.3%) had an allergic history, such as allergic asthma, rhinallergosis, conjunctivitis, or urticaria. The peak Scr values reached 254.0 (173.5, 428.5) μmol/l, and 24 patients (15.6%) needed renal replacement therapy. The clinical features of ATIN patients are shown in Supplemental Table II.

The carriers’ frequencies of HLA-DQA1, -DQB1, and -DRB1 in patients with D-ATIN, TINU syndrome, SS, and IgG4-RD were summarized and compared between groups (Supplemental Table IV, Table I). The HLA allele distribution in the patients with ATIN of other causes is shown in Supplemental Table III. Interestingly, there was no difference between patients with D-ATIN and TINU syndrome in these three HLA loci (Table I). The drugs that had been potentially involved in the development of D-ATIN in our patients included various antibiotics in 26 cases (such as β-lactams, quinolones, macrolides etc.; 26/76, 34.21%), nonsteroidal anti-inflammatory drugs in 22 cases (22/76, 28.95%), Chinese herbs in 16 cases (16/76, 21.05%), proton pump inhibitors in five cases (5/76, 6.58%), and other drugs in eight cases. There were no HLA allele frequency differences detected among D-ATIN patients that had different culprit drugs. Nor was there a difference between TINU patients with or without late-onset uveitis (data not shown). Patients with SS-ATIN and IgG4-RD presented a different pattern of the tested HLA alleles than patients with D-ATIN and TINU syndrome (Table I).

We next investigated the associations between the allele frequencies and the presence of ATIN with different causes. Unlike the healthy controls, patients with D-ATIN or TINU syndrome carried the same risk alleles in HLA-DQA1, -DQB1, and -DRB1, including DQA1*0104, DQB1*0503, and DRB1*1405 and *1454. Alleles DQA1*0102 and *0302, DQB1*0303 and *0602, and DRB1*0901 and *1501 were negatively associated with the presence of D-ATIN. These alleles were also protective against TINU syndrome, but the p value did not reach statistical significance after false discovery rate correction, which could have been because of the limited number of patients with TINU syndrome. There were no significant associations between the HLA alleles and the presence of SS, IgG4-RD, or other rare causes of ATIN.

As patients with D-ATIN and TINU syndrome had the same risk alleles, we combined the two groups of patients and observed more pronounced associations for the above detected risk alleles: DQA1*0104 (p value corrected by false discovery rate method [Pc] = 4.72 × 10⁻²², OR = 13.81, 95% CI 7.70–24.79), DQB1*0503 (Pc = 1.95 × 10⁻¹⁴, OR = 9.51, 95% CI 5.08–17.80), DRB1*1405 (Pc = 8.06 × 10⁻¹⁹, OR = 12.80, 95% CI 6.81–24.07), and DRB1*1454 (Pc = 1.13 × 10⁻⁷, OR = 1.16, 95% CI 1.07–1.24). Based on these results, in the following evaluation of the potential clinicopathological relevance of HLA loci, we considered patients with D-ATIN and patients with TINU syndrome to be one group to get a larger sample size (n = 114).

The haplotype analysis in patients with D-ATIN/TINU syndrome was performed via the R project package haplo stats version 3.3.1. Altogether, 87 haplotypes were detected in our studied population, and three haplotypes showed significantly different distributions between D-ATIN/TINU patients and the healthy controls. HLA-DQA1*0302/DQB1*0303/DRB1*0901 was the most common haplotype in healthy controls and was therefore set as the reference haplotype. The threshold of statistical significance was set at p < 0.0006 (0.05/87). HLA-DQA1*0104/DQB1*0503/DRB1*1405 was strongly associated with the presence of D-ATIN/TINU (p = 7.01 × 10⁻¹⁶, OR = 55.60, 95% CI 14.51–223.04).

Altogether, 72 D-ATIN/TINU patients had consecutive renal function monitoring for at least 12 mo after renal biopsy. We evaluated the clinicopathological relevance of HLA haplotype in these patients. HLA-DQA1*0104/DQB1*0503/DRB1*1405 was the most common detected haplotype in these patients, with 56.0% (28/50) of patients with D-ATIN and 40.9% (9/22) of patients with TINU syndrome. As shown in Table II, patients carrying HLA-DQA1*0104/DQB1*0503/DRB1*1405 had higher peak Scr (320.0 [227.0, 506.0]) versus 222.0 ([177.0, 323.0], p < 0.05) and more severe renal tubulointerstitial inflammatory impairment, as evaluated by the degree of interstitial inflammation and tubulitis on renal histopathology examination, than those who did not carry this specific HLA haplotype. There was no significant difference in the age, gender, allergic history, or systemic inflammatory response, as represented by the values of erythrocyte sedimentation rate, C-reactive protein, and IgG, between the groups of patients (Table II).

Patients carrying HLA-DQA1*0104/DQB1*0503/DRB1*1405 tended to be treated with more aggressive anti-immune therapy (Table II) compared with the treatments of patients with other haplotypes because of the higher Scr and more severe renal inflammation. Both groups of patients showed good response to steroid therapy and achieved renal function recovery, with eGFR reaching 61.8 ± 22.1 and 60.8 ± 19.9 ml/min per 1.73 m², respectively (p = 0.86), at 12 mo after renal biopsy (Fig. 3, Table II).

We next explored the relationship between HLA haplotype and local kidney inflammatory response in patients with D-ATIN/TINU syndrome. Fourteen patients carrying HLA-DQA1*0104/DQB1*0503/DRB1*1405 and 12 patients carrying other haplotypes were randomly selected. Infiltrated inflammatory cells in the renal interstitial area were detected through an immunofluorescence assay. Patients with HLA-DQA1*0104/DQB1*0503/DRB1*1405 had more renal monocytes/macrophages infiltrating into the interstitial area and the renal tubular wall than patients carrying other haplotypes. There was no difference in the number of other infiltrating inflammatory cells, including T cells, B cells, plasma cells, and neutrophils, in the renal tubulointerstitial area between the groups of patients (Figs. 4A, 5A).

We further investigated HLA-DR and HLA-DQ protein expression in the renal tubulointerstitial compartment. Patients carrying HLA-DQA1*0104/DQB1*0503/DRB1*1405 had significantly higher tubular HLA-DR and HLA-DQ expression than those carrying other HLA haplotypes (Figs. 4B, 5B). The expression of HLA-DQ protein in the interstitial cells was also higher in the patients with the risk haplotype. We then evaluated the association between HLA protein expression and the infiltrating inflammatory cells. Interestingly, there were positive correlations between the number of interstitial CD4⁺ T lymphocytes and the tubular expression of HLA-DR (r = 0.975, p < 0.001) and HLA-DQ (r = 0.832, p = 0.005) (Table III). In the same way, the infiltration of monocytes/macrophages into the interstitial area and into the renal tubular wall was closely correlated with the tubular expression of HLA-DR (r = 0.721, p = 0.004; r = 0.663, p = 0.01) and HLA-DQ (r = 0.615, p = 0.02; r = 0.533, p = 0.05) (Fig. 6, Table III).

HLA-DQA1, -DQB1, and -DRB1 have been associated with the development of TINU syndrome in case reports and small case series of different populations, but information about HLA genetic susceptibility to D-ATIN is very limited. In this study, in a relatively large, single-center prospective cohort of ATIN with various causes, we found that patients with D-ATIN and TINU syndrome shared the same HLA-DQA1, -DQB1, and -DRB1 risk alleles, which were different from those of patients with SS or IgG4-RD. Haplotype HLA-DQA1*0104/DQB1*0503/DRB1*1405 was associated with the development of D-ATIN/TINU and was relevant to the severity of renal dysfunction, the degree of tubulointerstitial inflammation, and the expression of HLA-DR and HLA-DQ molecules in the renal tubular cells.

One important finding of this study is that patients with D-ATIN or TINU syndrome had HLA genetic susceptibility in the HLA-DQA1, -DQB1, and -DRB1 alleles. These patients had been followed-up for at least 1 y, and their diagnosis of the disease (i.e., the cause of ATIN) had been verified through the follow-up course by a working team that included nephrologists and ophthalmologists. Further analysis showed that this HLA allele similarity existed in D-ATIN patients induced by different drugs and in TINU patients with either concurrent or late-onset uveitis. In SS, IgG4-RD, and ATIN because of other causes, this susceptibility in HLA-DQA1, -DQB1, and -DRB1 alleles was not detected nor has it been reported elsewhere. Our recently published study found that 60% of patients with TINU syndrome had a history of taking causative drugs, presented allergic symptoms, and had late-onset uveitis (29). Other researchers have also reported cases of TINU syndrome induced by drugs (30) as well as positive drug-specific lymphocyte stimulating tests (31). These clinical observations indicate that D-ATIN and TINU syndrome resemble each other and might share common immunologic pathogenesis in triggering tubulointerstitial nephritis. The similarity in the HLA-DQA1, -DQB1, and -DRB1 allele susceptibility to these two diseases that we found in this study further supports this hypothesis. However, the internal mechanism awaits in-depth illustration.

There have been studies reporting an association between HLA alleles and drug allergy, which were primarily related to the alleles of HLA-A and HLA-B (32). HLA-B*1502 were reported to be associated with carbamazepine-induced Stevens-Johnson syndrome in Asian populations (33–35) and HLA-A*3101 in European, North American, and Mexican mestizo populations (36–38). HLA-B*5801 were reported to be associated with allopurinol-induced Stevens-Johnson syndrome or drug-induced hypersensitivity syndrome (39, 40), and HLA-B*5701 were reported to be associated with abacavir-induced hypersensitivity reaction (41, 42). There were also a few studies showing an association between HLA II alleles and drug allergy; for instance, HLA-DRB1*0701 was related to asparaginase allergy in people of European descent (43), and HLA-DRB9 was related to penicillin-induced immediate hypersensitive reaction in Chinese Han populations (44). In the current study, haplotype HLA-QA1*0104/DQB1*0503/DRB1*1405 presented a much higher frequency in both D-ATIN (58%) and TINU syndrome (43.5%) patients than it had in healthy controls (7.5%). Its frequencies in SS (0.1%), IgG4-RD (0%), and ATIN of other causes (0%) were also low, which strongly support the association of this haplotype with the development of D-ATIN and TINU syndrome in our cohort. Further analysis revealed that the presence of this haplotype was not related to the type of causative drug or the degree of systemic inflammation but was associated with the severity of kidney dysfunction and the degree of renal tubulointerstitial inflammatory impairment. We assume that this certain HLA haplotype could be relevant to the development of kidney-local immune abnormalities and the subsequent inflammatory responses. Because HLA-DQA1*0104, DQB1*0503, DRB1*14 had been associated to TINU syndrome in both Asian and European populations (6, 8, 10, 15, 19), we assume HLA-DQA1*0104/DQB1*0503/DRB1*1405, the risk haplotype that we have detected in our cohort, might be also relevant to other populations, although further validation needs to be developed.

It is generally accepted that cell-mediated immunity has a major pathogenic role in D-ATIN and TINU (45), with CD4⁺ T cells being the most abundant type of interstitial infiltrates (2). Renal tubular cells have been known to express the class II MHC molecules and present Ags to CD4⁺ T cells (46, 47), resulting in the activation of CD4⁺ T cell–mediated immune response. De novo expression of HLA-DR in renal tubular cells can recruit inflammatory cells (48). In the current study, we found a significant upregulation of renal tubular HLA-DR and HLA-DQ protein in D-ATIN/TINU patients that was correlated with the degree of renal inflammatory infiltrates. Interestingly, patients carrying HLA-DQA1*0104/DQB1*0503/DRB1*1405 had more significant MHC-II protein expression, with more CD4⁺ T cells infiltrating and more severe inflammatory injury. We therefore assume that HLA-DQA1*0104/DQB1*0503/DRB1*1405 enhances the Ag-presenting ability of renal tubular cells and thus facilitates the development of tubulointerstitial nephritis under some specific pathogenic stimulus. Further studies are needed to verify the influence of HLA genetic susceptibility on the capacity of renal tubular cells to promote immunologic responses.

The main strength of our study is the prospective, regularly followed-up cohort design in a relatively large number of patients with D-ATIN or TINU syndrome, which enabled the detection of HLA risk alleles and haplotypes and the comparison of HLA-related susceptibility between these diseases. Selection bias originated from the single-center design and the inclusion of patients with biopsy specimen–proved ATIN. The molecular mechanisms implicated in the HLA associations need to be further explored. Additionally, we only focused on HLA-DQA1, -DQB1, and -DRB1 alleles based on previously published studies, which might have caused us to miss other significant loci.

In summary, our data indicate that patients with D-ATIN and TINU have HLA genetic susceptibility in the HLA-DQA1, -DQB1, and -DRB1 alleles. HLA-DQA1*0104/DQB1*0503/DRB1*1405 serves as a significant risk haplotype for the development of D-ATIN and TINU syndrome, which might facilitate renal tubulointerstitial inflammation by enhancing the Ag-presenting capacity of renal tubular cells.

We thank all of the clinicians who were involved in the care of the patients with ATIN for collecting data and Prof. Wanzhong Zou for helping with pathological diagnosis.

The authors have no financial conflicts of interest.