1 INTRODUCTION AND AIMS

Chronic kidney disease (CKD) is diagnosed either when a person has glomerular filtration rate (GFR) < 60 ml/min/1.73 m2 or when they have structural or functional abnormalities of the kidney that could lead to renal failure (The Cochrane Collaboration, 2019). In England, over 1.8 m people are affected and it is estimated that over 40,000 of these people die prematurely each year (NHS, 2012). The cost of treating CKD in England in 2019 was approximately £1.5bn (Health Service Journal, 2020), representing 1% of all NHS expenditure. Lower kidney function is associated with decreased quality of life and increased risks of mortality, stroke, myocardial infarction, infection and hospital admission (Go et al., 2004; NHS, 2012; Tonelli et al., 2006; Weiner et al., 2004).

According to the Global Burden of Disease 2015 Study (Mortality & Causes of Death, 2016), 1.2 million people worldwide died due to CKD in 2015 (an increase of 32% on 2005). CKD is now 17th in the list of diseases which cause the most ‘years of lost life’ (having been 21st in 2005 and 25th in 1990). Understanding the causes of kidney dysfunction could have important applications in terms of diagnosis and treatment of one of the most globally significant diseases.

CKD has a high heritability index (30–75%) (Canadas-Garre et al., 2019), with over 100 genomic regions reported to contribute towards it (Xu et al., 2018). Many previous studies have identified human leukocyte antigen (HLA) types that are associated with increased risk of end-stage renal disease (ESRD), which is when renal replacement therapy (such as dialysis or transplantation) becomes necessary. Other HLA associations indicate a protective effect. Most of these studies focus on subjects from specific ethnic cohorts or particular geographical regions, and HLA associations reported may be contradictory due to variation in disease prevalence and population stratification.

This study performed a systematic search of existing literature. The aim was to identify, summarize and appraise all previous studies which have attempted to find associations between HLA type and renal function. The HLA region was selected for analysis because it is closely linked to many disorders of the kidney (Robson et al., 2018). This suggests that it may have an impact on kidney function.

2 METHODS

This literature review investigates HLA genotypes associated with renal function in global populations. It was defined in terms of the PICO framework, a process which advises clear definitions of the participants, intervention, comparison and outcome that are to be studied (Huang et al., 2006). The participants (i.e. the subjects to be included) were global populations aged 18 years and over. The interventions (the independent variable) were HLA class I and II types. The comparisons (controls) were subjects without renal disease. The outcome (the dependent variable) was renal function (either increased or decreased). As of 19 September 2019 there is no review protocol for studies of HLA and renal function (as per a search of PROSPERO), though there is one for studies of HLA mismatches and kidney transplant outcomes (Shi et al., 2017).

A literature search was carried out on 30 November 2018 and updated on 17 February 2021 using Medline, Embase and Cochrane Central Register of Controlled Trials on the Ovid platform. These databases are a comprehensive and relevant source of papers and reviews which date back to 1946, 1974 and 2005 respectively. Only primary studies which had been published as full, peer-reviewed papers (rather than abstracts only) were considered. This helped to ensure that only high-quality, reliable evidence was used. Papers were also required to be written in the English language, to guarantee comprehension by the researchers. Their database thesaurus (or ‘index’) terms were required to include words or phrases associated with the following concepts: HLA, renal failure and genetic association. Medline scope notes were checked and all terms associated with key terms within the MeSH thesaurus tree were examined.

We originally intended to consider only papers investigating subjects of white ethnicity, as this most closely reflected the subjects of a research project that we were planning at the time and have since conducted (Lowe et al., 2021). However, ultimately any ethnicity was included as ethnicity was not reliably indexed in the databases searched, and filtering using this criterion would have lost relevant studies. Medline, for example, had 62,789 results for the search term ‘European Continental Ancestry Group’ (the index term for the word ‘Caucasian’). If the term had been properly indexed, the number of papers identified would have been much higher. The search criterion for white ethnicity was therefore not used. For similar reasons, the review initially intended to consider only subjects of middle-age or older, but age was ultimately not selected as a search term because requiring the index terms ‘aged’ or ‘middle-aged’ reduced the yield of publications by approximately 75%. As a result of excluding the concepts of ethnicity and age from the search, papers were identified which related to participants with a wide range of ethnic origins and ages. The observations contained within these papers, therefore, cannot necessarily be applied directly to the middle-aged white population as had originally been intended. A number of more sensitive searches were attempted, using broader search terms. For example, the search term ‘haplotypes’ was added to the genetic association concept and to the HLA concept. However, the additional papers highlighted through this strategy were not relevant to the research question, so the term was released.

A total of 242 papers were identified through the database search, but an earlier scoping exercise had revealed three additional publications which were not included in the database search (Davood et al., 2008; Mosaad et al., 2014; Nassar et al., 2015), possibly due to their selected keywords. The keywords of two papers did not include the concept of genetic association (Mosaad et al., 2014; Nassar et al., 2015). The keywords of the third paper suggested that this publication should have been captured by the search (Davood et al., 2008); possible explanations for this omission are that either the keywords were not indexed correctly or the journal (Research Journal of Biological Sciences) was not included in the databases searched. All three papers were included, giving a total of 245 publications selected. Of these 245 papers, 32 were duplicates of the same publication found in different databases, leaving 213 unique papers for screening. The titles and abstracts of these 213 papers were read to assess their relevance to the research question. A total of 155 were deemed irrelevant to the research question and were removed. The primary reason for each exclusion is documented in Table 1. After this initial screening exercise, 58 papers remained for consideration.

TABLE 1. Primary reason for exclusion after screening of titles and abstracts Reason for exclusion Number of results excluded Not related to kidneys 42 Investigating transplantation but not kidney failure 38 Not related to HLA class I and II 33 Only tangentially related to kidney failure 16 Case study 14 Study of children 6 New HLA allele discovery 4 Not an article 2

In 17 cases, no full text version of the paper was available, either because the reference related to a conference poster or abstract which had not been published as a paper or because the full text was inaccessible. The University of Manchester library was consulted for help in obtaining these papers, but was unsuccessful. This left 41 full text papers for assessment. A further six publications were removed after the full text articles were assessed, either because the paper was not related to HLA class I or II (n = 3), the paper did not measure renal failure subjects against healthy controls (n = 2), or because the paper was not a primary study (n = 1). Thirty-five articles, therefore, were deemed relevant to the research question. Figure 1 summarizes the entire process of exclusion following a strategy adapted from Moher et al. (2009), which outlines the process for completing a systematic review.

Process of including and excluding papers

The 35 papers were analysed and the results extracted from tables in the Section 3 and prose in the Section 4. No contact with original researchers was made to verify their data or request further information such as funding sources. Each paper was also subject to an assessment using the Critical Appraisal Skills Programme to ensure it was of sufficient quality to be included (Critical Appraisal Skills Programme, 2018). The search was carried out by a single researcher with consultation and advice from two librarians. All screening and assessments of eligibility were performed independently by two researchers who discussed any discrepancies until consensus was reached. Extraction of data from the papers was performed by a single researcher. Different studies used different methods to determine whether HLA types were associated with renal function. The principal summary measures were odds ratio, relative risk and hazard ratio.

3 RESULTS

Associations between HLA and renal function were identified in 30 of the 35 papers. In all, there were a total of 181 associations reported, relating to 58 different HLA class I types, 37 class II types and 14 three-locus haplotypes. None of the 14 findings which related to a haplotype was independently replicated, but 31 of the findings relating to a single HLA type were replicated. There were 20 types which were found to be associated with increased risk of kidney disease by at least one study, but protective against kidney disease by at least one other.

Sixteen HLA types (HLA-A*24, A*26, A*29, A*30, A*32, B*07, B*40, B*44, C*02, DRB1*03, DRB1*04, DRB1*08, DRB1*11, DRB1*13, DQA1*03 and DQB1*06) were found to be protective against ESRD by multiple studies (though nine of these were also found to be associated with increased risk of ESRD by at least one other study: A*24, B*07, B*40, C*02, DRB1*03, DRB1*04, DRB1*11, DRB1*13 and DQA1*03). In total, 38 class I types, 24 class II types and 8 haplotypes were found to be protective against ESRD, though 11 of the class I types and 9 of the class II types were in conflict and also associated with ESRD. Similarly, 15 HLA types (HLA-A*11, B*08, B*15, B*18, B*49, B*50, B*51, B*53, B*55, C*01, DRB1*03, DRB1*04, DRB1*11, DRB1*12 and DQB1*02) were associated with ESRD in at least two studies. However, seven of these were protective against ESRD according to a different study (HLA-A*11, B*08, B*50, DRB1*03, DRB1*04, DRB1*11 and DRB1*12). In total, 31 HLA class I types, 22 class II types and 6 haplotypes were found to be associated with renal dysfunction (though, as mentioned above, 11 of the class I types and 9 of the class II types were also found to be protective against ESRD by different studies).

Many of the findings were directly refuted by other studies. There does not appear to be a consensus around which HLA types have a protective effect and which incur additional risk of ESRD; of the 95 HLA types with a reported association, 20 had the finding refuted by at least one other independent study (21%). Only 10 HLA associations were reported in three or more studies (HLA-A*11, B*07, B*08, B*53, DRB1*03, DRB1*04, DRB1*08, DRB1*11, DQB1*02 and DQB1*06), and 6 of these were refuted by another study (the exceptions being HLA-B*53, DRB1*08, DQB1*02 and DQB1*06). Table 2 shows all reported associations between HLA and renal function (some associations with increased incidence of renal dysfunction, and some with decreased incidence). The table illustrates that there are a large number of HLA types which may have an effect on renal failure. The subjects’ primary diseases are noted in the table (where possible) to allow for comparison of associations based on the underlying cause of renal dysfunction. There are a number of abbreviations in Table 2; these are expanded below the table. Five of the papers did not report any significant associations. These are listed in Table 3.

TABLE 2. All HLA types and three-locus haplotypes found to be associated with renal function HLA Locus HLA type or haplotype Study Effect Population Number of subjects HLA-A A*01 Karahan et al. (2009) Protective against CGN, HTN nephrosclerosis Turkish 3230 (587 ESRD patients, 2643 controls). Primary diseases: unknown 27%; CGN 23%; HTN nephrosclerosis 17%; PKD 9%; pyelonephritis 9%; VUR nephropathy 4%; diabetic nephropathy 4%; amyloidosis 3%; urologic abnormalities 2%; other 3% Lowe et al. (2021) Reduced eGFR, increased ESRD British 401,307 (with a range of kidney function, eGFR calculated for each) A*02 Karahan et al. (2009) Protective against CGN, HTN nephrosclerosis Turkish See above A*03 Karahan et al. (2009) Protective against CGN, HTN nephrosclerosis Turkish See above Lowe et al. (2021) Reduced eGFR British See above A*09 Rivera et al. (2012) Protective against ESRD Venezuelan 390 (188 ESRD patients, 202 controls). Majority post-streptococcal or other glomerulonephritis origin A*11 Davood et al. (2008) Associated with ESRD Azerbaijani 77 (26 ESRD patients awaiting transplantation, primary disease not specified, 51 controls) Hernandez-Rivera et al. (2019) Associated with ESRD Mexican 3326 (1965 ESRD patients awaiting transplantation, primary disease not specified, 1361 potential kidney donors) Pan et al. (2019) Associated with ESRD Chinese 2083 (499 ESRD patients awaiting transplantation, 1584 controls). Primary diseases: unknown 63%; glomerulonephritis 13%; HTN nephropathy 8%; diabetic nephropathy 6%; interstitial nephritis 5%; PKD 2%; HSPN 2%; obstructive nephropathy 1%; autoimmune diseases 1% Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above A*23 Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above A*24 Cao et al. (2014) Associated with ESRD Cantonese 8285 (4541 ESRD patients awaiting transplantation, 3744 controls). Primary disease: 57% not specified; 43% glomerulonephritis Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above Uygun et al. (2015) Protective against ESRD Turkish 677 (144 highly PRA positive ESRD patients, primary disease not specified, 533 controls) A*25:01 Lowe et al. (2021) Increased eGFR British See above A*26 Hamdi et al. (2014) Protective against ESRD Saudi Arabian 455 (350 ESRD patients, primary disease not specified, 105 controls) Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above A*28 Mosaad et al. (2014) Protective against ESRD Kuwaiti 525 (334 ESRD patients, primary disease not specified, 191 controls) A*29 Karahan et al. (2009) Protective against ESRD Turkish See above Lowe et al. (2021) Increased eGFR British See above A*30 Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above Nassar et al. (2017) Protective against ESRD Yemeni 381 (187 CRF patients on haemodialysis, primary disease not specified, 194 controls) A*31:01 Pan et al. (2019) Associated with ESRD Chinese See above A*32 Lowe et al. (2021) Increased eGFR British See above Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above A*33 Davood et al. (2008) Associated with ESRD Azerbaijani See above A*66 Karahan et al. (2009) Protective against ESRD Turkish See above A*68 Karahan et al. (2009) Protective against ESRD, CGN, HTN nephrosclerosis Turkish See above A*69 Karahan et al. (2009) Protective against ESRD Turkish See above A*78 Crispim et al. (2008) Associated with ESRD Sao Paolo (Brazil) 265 (105 ESRD patients awaiting transplantation, 160 controls). Primary diseases: undetermined 33%; HTN 25%; diabetes 10%; renal cystic disease 8%; Berger disease 5%; glomerulosclerosis 5%; lupus 4%; other 11% HLA-B Bw*4 Prakash et al. (2013) Associated with ESRD North Indian 1024 (512 ESRD patients, primary disease not specified, 512 controls) B*07 Doxiadis et al. (2001) Protective against IgA nephropathy European 2831 (1620 IgA nephropathy patients awaiting kidney transplant, 1211 controls) Lowe et al. (2021) Reduced eGFR British See above Hieu et al. (2019) Protective against ESRD Vietnamese 383 (196 ESRD patients, 187 controls). Primary diseases: CGN 76%; HTN 10%; diabetes 6%; PKD 4%; other 4% Karahan et al. (2009) Protective against ESRD Turkish See above B*08 Doxiadis et al. (2001) Protective against IgA nephropathy European See above Lowe et al. (2021) Reduced eGFR, increased ESRD, increased CKD British See above Hernandez-Rivera et al. (2019) Associated with ESRD Mexican See above Mosaad et al. (2014) Associated with ESRD Kuwaiti See above B*12 Rivera et al. (2012) Protective against ESRD Venezuelan See above B*14 Crispim et al. (2008) Protective against ESRD Sao Paolo (Brazil) See above B*14:01 Lowe et al. (2021) Increased eGFR British See above B*14:02 Lowe et al. (2021) Increased eGFR British See above B*15 Hamdi et al. (2014) Associated with ESRD Saudi Arabian See above Pan et al. (2019) Associated with ESRD Chinese See above B*17 Rivera et al. (2012) Protective against ESRD Venezuelan See above B*18 Hamdi et al. (2014) Associated with ESRD Saudi Arabian See above Hernandez-Rivera et al. (2019) Associated with ESRD Mexican See above B*35 Doxiadis et al. (2001) Associated with IgA nephropathy European See above B*38 Rivera et al. (2012) Associated with ESRD Venezuelan See above B*39 Hamdi et al. (2014) Protective against ESRD Saudi Arabian See above Pan et al. (2019) Associated with ESRD Chinese See above B*40 Cao et al. (2014) Associated with ESRD Cantonese See above Noureen et al. (2020) Protective against ESRD Pakistani 1169 (497 ESRD patients, primary disease not specified, 672 controls) Rivera et al. (2012) Protective against ESRD Venezuelan See above B*42 Yamakawa et al. (2014) Associated with ESRD Brazilian 183 ESRD patients on haemodialysis, primary disease not specified, multiple control groups used B*44 Yamakawa et al. (2014) Protective against ESRD Brazilian See above Lowe et al. (2021) Increased eGFR British See above B*45 Yamakawa et al. (2014) Associated with ESRD Brazilian See above B*48 Rivera et al. (2012) Protective against ESRD Venezuelan See above B*49 Hamdi et al. (2014) Associated with ESRD Saudi Arabian See above Davood et al. (2008) Associated with ESRD Azerbaijani See above B*50 Hamdi et al. (2014) Protective against ESRD Saudi Arabian