Article Text
PDF
Abstract
Aim Coeliac disease(CD) is a highly prevalent, gluten-dependent, autoimmune enteropathy. While the diagnosis is based on serological and histological criteria, genotyping of the human leucocyte antigens (HLA) DQ2 and DQ8 has been shown to have substantial clinical utility, especially in excluding the diagnosis in patients who do not carry either antigen. As a result, HLA genotyping is now being performed by more laboratories and has recently become one of the most frequently requested genetic tests in Australia. To date, there has been little scrutiny on the accuracy and reporting of results by laboratories new to HLA typing. In response to clinician feedback that identified potentially clinically significant discrepancies in HLA typing results, the Royal College of Pathologists of Australasia Quality Assurance Programs (RCPAQAP) undertook a pilot study to assess laboratory performance in the detection of HLA-DQ2/DQ8 and their associated HLA-DQA1 and HLA-DQB1 alleles.
Methods DNA was extracted from 5 patients and sent to 10 laboratories for external quality assurance (EQA) testing. Laboratories were assessed for reporting in genotyping, interpretation and methodology.
Results Our findings showed that at least 80% of laboratories underperform with respect to recommended guidelines for HLA typing and reporting for CD, with 40% of laboratories failing to provide any clinical interpretation or full genotyping data. This suboptimal level of reporting may lead to ambiguities for downstream clinical interpretation that may compromise patient management.
Conclusions These findings highlight the importance of adherence to standardised guidelines for optimal performance and reporting of HLA results and substantiate the need for EQA and proficiency testing for laboratories providing this service.
- coeliac disease
- quality assurance
- HLA
Statistics from Altmetric.com
Introduction
Coeliac disease (CD) is a chronic inflammatory enteropathy caused by an autoimmune response to dietary gluten in genetically predisposed individuals.1 The diagnosis of CD relies on demonstrating the presence of coeliac-specific antibodies and small intestinal villous atrophy with improvement on exclusion of dietary gluten.2 A key feature of CD is its strong dependence on the presence of susceptibility alleles encoding for human leucocyte antigens (HLA) DQ2.5, DQ8 or half the HLA DQ2.5 heterodimer, typically DQ2.2.3 These specific HLA types are seen in more than 98% of Europeans and Australians with CD4 5 and encode immune recognition molecules that facilitate pathogenic CD4 +T cell responses to gluten.1 HLA typing therefore achieves near perfect sensitivity and negative predictive value (NPV) for CD in the general population, making it exceptionally useful as a test to exclude CD when the susceptibility genotypes are absent.3 However, as approximately half of the Australian population express HLA DQ2.5, DQ8 and/or DQ2.2, the presence of these HLA types has poor positive predictive value and low specificity for CD.5
HLA typing in the diagnostic workup of CD is increasing and may in part be driven by the increased awareness of CD and the high community uptake of the gluten-free diet, which renders traditional testing with serology and histology uninformative.3 As HLA testing is often used to exclude a diagnosis of CD, it is imperative that the results are both accurate and clearly reported to general practitioners to ensure appropriate patient care. In recognition of this, Australasian guidelines were developed that outline an optimal approach to testing and reporting of HLA results; however, their uptake appears limited to date.3 We recently reported a series of patients that had discrepant HLA genotyping results that would otherwise have impacted patient care had they not been detected.6 This finding challenged conventional clinical wisdom that HLA results are definitive and beyond doubt.
Here we report a more systematic assessment of this issue as part of the Royal College of Pathologists of Australasia Quality Assurance Programs (RCPAQAP). External quality assurance (EQA) programs exist to monitor the consistency and accuracy of clinical diagnostic interpretation and reporting. This pilot study’s findings highlight issues with the detection of HLA-DQ2/DQ8 and the reporting of the results for CD association. First, they underscore an urgent need to ensure all pathology providers offering HLA typing are involved in an EQA program. It is noted that the laboratories performing HLA typing for transplantation have participated in an EQA program under the umbrella of the Asia Pacific Histocompatibility and Immunogenetics Association since the early 1990s. Second, they highlight the requirement to adhere to an agreed set of standards in HLA typing and result reporting.
Material and methods
Laboratories
A total of 10 laboratories (7 from Australia and 3 from New Zealand) participated in this RCPAQAP CD pilot program. All laboratories perform multiple diagnostic testing in addition to HLA genotyping. The public sector was represented by eight laboratories and the private sector was represented by two laboratories. All participating laboratories returned data.
Samples
DNA was extracted from five patients and a total DNA concentration of 1 µg (50 ng/µL) was sent to each participating laboratory. Laboratories were informed that each DNA sample represented a suspected CD case. Fictional patient details were supplied. Table 1 represents actual patient details and associated HLA typing status.
Patient details and HLA typing status
Procedure
Participants were instructed to measure the concentration of DNA and perform their routine testing for HLA DQ2/DQ8 typing. A clinical patient report was requested for RCPAQAP evaluation of all generated data.
Assessment criteria
Assessment criteria were based on key recommendations previously endorsed by the RCPA.3 This guideline reported the following as best practice: (1) Genotyping: the report should denote all at risk alleles present, ideally defining both HLA-DQA1 and HLA-DQB1 alleles to determine the complete genotype and, when possible, zygosity status. Estimates of relative-risk for CD based on HLA genotype and zygosity may improve clinical risk stratification. However, the major value lies in excluding CD based on the absence of an at-risk genotype. A comment should emphasise the limited positive predictive value of HLA typing in isolation. (2) Interpretation and reporting: should be clear and concise to aid clinical interpretation. Results should include a simple summary statement on whether genetic susceptibility to CD is present or absent and provide a comment on the non-diagnostic nature of HLA typing in isolation. (3) Methodology: the PCR coding sequence-based methodology provides more accurate and reliable HLA typing results than single nucleotide polymorphism (SNP)-based approaches and is the preferred methodology for individual patient management.
For this analysis, key components of these recommendations were condensed and developed into a deductive scoring system (table 2) to provide greater objectivity in assessment as used by a range of QA providers in assessing genetic testing approaches and modelled on the approach reported by Patton et al (2014).7 Components were weighted by importance in terms of contributing to an accurate report as judged by the authors. For example, a critical genotyping error is clearly important and results in a heavier point deduction compared with not providing references to support a clinical comment. Laboratories were deducted points for errors from a maximum score of two in each category. A maximum score of 6 (2 points for each category) is therefore achievable for each individual case. A final average percentage score taking into account all three categories was derived to provide an indication for each laboratory’s overall performance.
Assessment criteria and deductive scoring used for quality monitoring CD reporting
Points were deducted according to the following
Genotyping
Points were deducted for making a critical genotyping error (ie, getting the genotype incorrect, 2 points), not determining the complete HLA allele sequence (1 point), not reporting zygosity (0.2 points) and not correctly using current HLA nomenclature (0.2 points).
Interpretation
Points were deducted for making a critical interpretation error (2 points), for not providing any clinical interpretation (1.5 points), for not providing a comment on a negative result to exclude CD (1 point), for not providing a comment on using HLA typing in isolation (0.5 points), no statement on observed genotype and susceptibility to CD provided (0.5 points) and references in support of clinical interpretation not provided (0.2 points).
Methodology
Points were deducted for not providing a statement regarding the specific testing assay used (1.5 points) and for not providing information regarding the limitations/sensitivity of the testing assay (0.5 points).
Results
Diagnostic platforms
The reported assays used for HLA typing mainly consisted of primer/probe-based allele-specific detection for single or multiple SNP (table 3). Specific measuring platforms used to determine CD genotyping data were not provided by 60% (6/10) of laboratories. Two laboratories (20%) did not provide any information regarding their specific testing methodology used for CD HLA testing.
Specific methods and platforms used for data analysis in the 2016 Coeliac Disease HLA Genotyping pilot module
Data reporting
All participants submitted clinical reports on the five RCPAQAP patient DNA samples. Reports were largely similar in size with 80% (8/10) of reports being a single document consisting of five pages (one page for each case). One laboratory submitted a single page report covering all five cases and one laboratory submitted five documents relating to each individual case.
Assessment criteria scores
Analysis of laboratory clinical reports identified inconsistencies with current CD reporting guidelines for the reporting of genotype, for clinical interpretation of the genotype data and in the reporting of CD diagnostic testing methodology:
Assessment of genotyping
Analysis of reported genotyping data identified that 90% of laboratories did not determine the full genotype for HLA-DQA1 and HLA-DQB1 and did not use the current recommended HLA reporting nomenclature (http://hla.alleles.org/nomenclature/naming.html). Only one laboratory (Lab 10) scored maximum points for correctly genotyping each case in accordance with the correct usage of HLA reporting nomenclature. Notably, one laboratory (Lab 7) only provided DQA1*05 data for case 1 without any interpretation of the data or identifying the associated HLA-DQ antigen (ie, HLA-DQ7). This is representative of a genotyping error given that DQA1*05 is associated with HLA-DQ7 and the at-risk HLA-DQ2.5 antigen. A comparison between the recommended reported genotypes (from the reference testing laboratory) and the genotyping data reported from each participating laboratory are provided in table 4.
Laboratory reporting styles for genotyping and antigen determination for each of the five coeliac disease cases
Assessment of interpretation
Analysis of each clinical interpretation in relation to the identified CD genotype found that for all cases 40% of laboratories did not provide any clinical interpretation, 40% of laboratories did not comment on the limitation of using HLA typing in isolation and 30% of laboratories did not report on the relative CD risk in relation to the specific genotype detected. In addition, 90% of laboratories did not provide any references to support their clinical interpretations for the specific genotypes identified for each case (online supplementary table 1).
Assessment of methodology
Analysis of each laboratory’s CD testing methodology identified that 80% did not report on the sensitivity or limitations on their specific CD testing assay. The remaining 20% of laboratories failed to report any information regarding their testing methodology used for CD analysis (online supplementary table 1).
Assessment of overall performance
An average percentage score was calculated to provide an indication for each laboratory’s overall performance. One laboratory (Lab 10) scored 88% and one laboratory (Lab 1) scored between 65% and 68% for all cases tested. The remaining eight laboratories all scored less than 61% (figure 1).
Laboratory performance for the combined categories of genotyping, interpretation and methodology. Clinical reports from each testing laboratory were individually assessed for each of the five suspected CD cases. CD, coeliac disease.
Discussion
As the development of CD is strongly dependent on the presence of specific HLA susceptibility alleles, HLA genotyping has emerged as an important tool for the diagnostic workup of CD around the globe.2 8 9 As a result, the test is now firmly embedded in clinical practice, with the number of requests rising 14-fold in Australia since 20036 (Medicare Benefits Schedule data, item 71151). The significant rise in requests has been a strong driver for laboratories to develop and offer HLA genotyping for CD and multiple laboratories that otherwise did not perform HLA typing now routinely offer and perform this test. Although our data reflect the current Australasian situation, the findings are likely to be relevant worldwide as they relate to challenges inherent in HLA testing, interpretation and reporting.
The major clinical utility of HLA typing in the workup of CD is its exceptionally high negative predictive value. Information to inform risk stratification from knowledge of the HLA genotype and zygosity is useful but of lesser importance. For example, HLA genotyping in a person at-risk of CD (ie, a 1st degree relative) would identify those who may benefit from CD screening and further follow-up. As the HLA typing result in the clinical workup of CD often leads to excluding a CD diagnosis, an incorrect result can have a substantial adverse impact on patient care. In this study, we have found that commercially available allele-specific SNP detection-based assays were the most frequently offered test by the participating laboratories for CD genotyping. This approach is simpler and cheaper for the detection of a small number of SNPs in comparison to performing full exon typing by PCR and sequencing. However, the complete HLA sequence genotype cannot be determined using these approaches; therefore, the complete HLA genotyping data are inferred rather than confirmed. This can lead to difficulties in distinguishing between the different HLA-DQ antigens (ie, DQ2.2, DQ2.5) associated with the HLA-DQA1 and HLA-DQB1 alleles. This is important given that different DQ antigens are associated with differing levels of CD risk,10 with the highest risk occurring when there are two copies of the DQB1*02 allele as occurs in the homozygous version of HLA-DQ2.5 and HLA-DQ2.5/DQ2.2 (figure 2).
Risk HLA antigens associated with coeliac disease (adapted from Romanos and Wijmenga 2010).10 HLA, human leucocyte antigens.
In this study, assessment of reporting in the three categories of HLA genotyping, data interpretation and CD testing methodology identified shortcomings in all three categories in all 10 laboratories compared with published best practice standards. For example, for genotyping analysis, the largest discrepancies were a failure to determine and report the full HLA DNA allelic sequence according to the recommended HLA reporting nomenclature (90% of laboratories). For clinical interpretation, a significant number of laboratories did not provide any clinical interpretation for all cases (40%) or did not make any comment regarding the usefulness or limitation of using HLA typing in isolation (40%). In addition, clinical statements should be supported by recent evidence and only one laboratory (Lab 6) provided references to support their clinical interpretations for the genotypes identified in each case. For reporting on the diagnostic methodology, the largest discrepancy was a failure to report on the levels of sensitivity or limitations on the specific testing assay (all laboratories).
The clinical impact of these identified discrepancies may have downstream implications on patient management. For example, one laboratory (Lab 7) provided potentially confusing data by just identifying case 1 as being DQA1*05 positive. DQA1*05 is associated with both the low-risk HLA-DQ7 antigen and the at-risk HLA-DQ2.5 antigen. Such inconclusive reporting may lead to inappropriate patient care and unnecessary investigations, for example, invasive endoscopy. Pleasingly, however, there was no patient sample where CD susceptibility alleles failed to be detected when they were present. This is important, as a false negative result could have more severe consequences for the individual who has undiagnosed CD as the diagnosis of CD is likely to be excluded on the basis of a negative HLA result. In addition, the 40% of laboratories that did not provide any clinical interpretation also failed to provide full HLA allele sequence identity and did not use the recommended HLA genotyping nomenclature. This level of reporting may lead to ambiguities for correct clinical classification and risk stratification that could impact on patient treatment. Furthermore, best practise guidelines for the use and reporting of HLA typing results for the workup of CD3 would recommend that each clinical report provide information on genotyping results, clinical interpretation and methodology. The combined performance score of less than 61% in 80% of the laboratories reflects suboptimal performance across these three categories (figure 1).
Conclusion
Shortcomings in the performance and reporting of HLA results can have adverse clinical consequences. This pilot study indicates that evidence-based guidelines developed for HLA testing for the workup of CD3 are not being consistently adhered to. The findings confirm our previous report that HLA typing and reporting errors are occurring in Australian clinical practice.6 Addressing this issue will require greater awareness of the recommended testing and reporting guidelines by both public and private laboratories performing HLA typing, and as a result of this study, approaches to achieve this are being explored. This study also highlights the importance of EQA proficiency testing as a means to allow laboratories to demonstrate appropriate levels of competence, monitor their specific testing over time and to implement corrective strategies where necessary. Importantly, EQA programs assess both the analytical test and the key clinical interpretation, which leads to an improvement in the overall reporting of the diagnostic test. This is particularly important in newer areas of testing where there are not yet agreed guidelines and where the referring clinicians are not as familiar with the likely problems associated with a particular test. The issues identified from this study highlight the important role of EQA providers in helping to raise standards for laboratory diagnostic analyses of genetic disorders. Similar quality assessment evaluations to explore the accuracy of HLA typing in the workup of CD in other countries are indicated.
Take home messages
Coeliac disease (CD) is a chronic inflammatory enteropathy caused by an autoimmune response to dietary gluten in genetically predisposed individuals.
Human leucocyte antigens (HLA) genotyping is now being performed by more laboratories and has recently become one of the most frequently requested genetic tests in Australia.
The majority of laboratories underperformed with respect to recommended guidelines for HLA typing and reporting for CD. This suboptimal level of reporting may lead to ambiguities for downstream clinical interpretation that may compromise patient management.
External quality assurance (EQA) programs exist to monitor the consistency and accuracy of clinical diagnostic interpretation and reporting. Thus, the data from this study underscore an urgent need to ensure all pathology providers offering HLA typing are involved in an EQA program.
Footnotes
Handling editor Runjan Chetty.
Contributors MPH, SYC, NM, LW, JAT-D, JD, MV and TB designed the study, analysed the data, drafted and revised the manuscript. KHT performed the experiments, analysed the data and revised the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests JAT-D is a coinventor of patents pertaining to the use gluten peptides in therapeutics, diagnostics and non-toxic gluten in coeliac disease. He is a shareholder of Nexpep Pty Ltd and a consultant to ImmusanT Inc. Nexpep Pty Ltd and ImmusanT Inc. were formed to develop novel diagnostics and treatments for coeliac disease. This work was designed, developed, prepared and funded by the RCPAQAP. Other authors have no competing interests to declare.
Patient consent Not required.
Ethics approval RCPA Quality Assurance Programs Pty Limited.
Provenance and peer review Not commissioned; externally peer reviewed.