Genetics

Benefiting from genomics in health care depends on data sharing and digital health integration

David P Hansen · Marcel E Dinger · Oliver Hofmann · Natalie Thorne · Tiffany F Boughtwood

To the Editor: We note with interest the recent review of challenges in the management of maturity onset diabetes of the young associated with glucokinase gene mutations (GCK-MODY).1 In Box 2, Bishay and Greenfield reported a prevalence of gestational diabetes mellitus (GDM) of 5–10%. Indeed, in 2010 we reported an estimated prevalence of 10–11% of GDM in south-western Sydney;2 however, the prevalence is now almost double that figure (18.5% of births at Bankstown-Lidcombe hospital in 2015). Bishay and Greenfield also summarised the findings of Chakera and colleagues3 for a population of predominantly European descent: A lower body mass index (BMI, < 25 kg/m2) and a fasting glucose level greater than or equal to 5.5 mmol/L have sensitivity and specificity of 68% and 96%, respectively. It is estimated that among lean women with mild fasting hyperglycaemia, the number of women needed to test is 2.7 to detect a single case of GCK-MODY.1 In contrast, in our recent study of women with GDM, we found that at least 8.1 women would need to be tested to identify one case of GCK-MODY.4 Given our interest in the management of women with GDM, we sought to determine whether these criteria were applicable in a large multi-ethnic cohort of women with GDM. Analysing de-identified, prospectively collected data from all women with GDM in our ethnically diverse population, diagnosed using the Australasian Diabetes in Pregnancy Society (1998) criteria at our institution between 1993 and 2013, we categorised the women into two groups: those with body mass index ≤ 21 kg/m2 (group A1) and those with body mass index > 21 kg/m2 and < 25 kg/m2 (group A2). We collected complete data, including post-partum oral glucose tolerance test results, for 171 women (54, group A1; 117, group A2). The oral glucose tolerance test and post-partum glycated haemoglobin results identified few women (< 14%) in either group who still had possible GCK-MODY. Testing all 171 of these women in pregnancy would have been a costly exercise with a low yield. In testing data in different ethnic groups, we therefore recommend caution regarding the number suggested by Chakera and colleagues.

Jeff R Flack · Glynis P Ross · N Wah Cheung

In reply

Ramy H Bishay · Jerry R Greenfield

To the Editor: The demand for human leukocyte antigen (HLA) typing in the diagnostic work-up of coeliac disease (CD) in Australia has driven a 14-fold rise in testing since 2003 (Medicare Benefits Schedule data, item 71151). Although HLA typing offers limited specificity for CD, its clinical utility results from its exceptional negative predictive value (> 99%) when the specific HLA susceptibility genotypes are not detected.1 Unlike traditional tests for CD, HLA typing results are informative even when the patient is following a gluten free diet. While the accuracy of HLA typing in CD has not been reported, HLA test results are assumed by clinicians to be definitive. Our findings challenge this view. Discrepancies between several patients’ clinical diagnosis of CD and their negative HLA-DQ2 and -DQ8 typing results in AJD’s practice led to repeat HLA testing with another laboratory. The subsequent reporting of a genotype consistent with CD prompted a clinical audit (2013–2016). Of 211 patients with HLA typing results, nine had been coperformed by two separate laboratories (laboratories 1 and 2), either deliberately or inadvertently. Of these nine patients, six returned conflicting results. An additional DNA sample from all six patients was sent for HLA genotyping by a reference laboratory, where genetic susceptibility for CD was confirmed in five patients (Box). Laboratories 1 and 2 differed in the detection of risk alleles and in the interpretation of or reporting of the results in all six cases. Laboratory 2 identified an at-risk allele in only two of the six patients, and of the four patients with a reported negative genotype, two were subsequently confirmed to have definite CD. These preliminary findings raise serious concerns about CD HLA testing errors that adversely affect patient care. Although identified in Queensland, these laboratories routinely outsource their HLA typing to laboratories in New South Wales and Victoria, indicating that several Australian states are involved. We are particularly concerned about laboratories new to HLA testing or laboratories that are not participating in stringent quality assessment programs as the sourced reference laboratory does. Therefore, we suggest that an assessment of the performance and quality control measures of all laboratories offering HLA typing is urgently needed. Consistent adoption of evidence-based guidelines that describe optimal HLA testing and reporting1 should form part of the solution. Box – Human leukocyte antigen (HLA) typing results from three laboratories† Patient Laboratory 1 Laboratory 2 Reference laboratory Confirmed CD “Consistent with DQ2 phenotype; susceptible for CD” Genotype not supplied; “Not susceptible for CD” HLA-DQ2.2/2.5; susceptible to CD Incorrect Confirmed CD “Consistent with DQ2 phenotype; susceptible for CD” Genotype not supplied; “Not susceptible for CD” HLA-DQ2.2; susceptible to CD Incorrect CD excluded “Consistent with DQ2 phenotype; susceptible for CD” Genotype not supplied; “Not susceptible for CD” HLA-DQ2.2; susceptible to CD Incorrect CD not excluded; on GFD “Consistent with DQ8 phenotype; susceptible for CD” Genotype not supplied; “Not susceptible for CD” No susceptibility to CD detected Incorrect Normal CD serology “DQ2 and DQ8 not identified; no genotype susceptibility for CD” “DQA1*0505 has been detected; small percentage susceptible to CD” DQA1*05 (HLA-DQ7); low risk susceptibility to CD Incorrect CD excluded “DQ2 and DQ8 not identified; no genotype susceptibility for CD” “DQA1*0505 has been detected; small percentage susceptible to CD” DQA1*05 (HLA-DQ7); low risk susceptibility to CD Incorrect CD = coeliac disease. GFD = gluten free diet. ND = not detected. † The reference laboratory was in the Victorian Transplantation and Immunogenetics Service in Melbourne. In addition to the incorrect typing results, laboratory 2 failed to report the specific alleles detected and laboratory 1 failed to distinguish between HLA-DQ2.5 and DQ2.2.

A James M Daveson · Michael Varney · Kate E Jackson · Jason A Tye-Din

n/a

Gareth S Baynam · Glenn Pearson · Jenefer Blackwell