Page 104 - Recognizing axial spondyloarthritis - Janneke de Winter

CHAPTER SIX
IL-17 production through KIR ligation (31,32). Whereas further research is needed to assess similar functions for HLA-C*07, it is important to note that we observed that HLA-C*07 contributed to axSpA independent from HLA-B*27 in the GESPIC and SPACE cohorts.
As to potential use of HLA-C*07 as a biomarker, the diagnostic value of HLA-C*07 alone appeared to be limited in the SPACE cohort, with a positive predictive value (PPV) of 40.7% (95% CI 36.0%-45.6%) and a negative predictive value of 81.1% (95% CI 75.5%-85.7%). HLA-B*27 has a PPV of 87.9% (95% CI 82.7-91.6%) and a NPV of 91.5% (95% CI 97.7-94.2%). HLA-C*07 negativity in combination with HLA-B*27 positivity increases the PPV to 97.88% (95% CI 95.8-98.9%), the NPV declines to 58.1% (95% CI 50.1-65.8%). Beyond diagnosis, HLA-C*07 genotyping could help to stratify axSpA sub-populations. Although preliminary, our data suggest that HLA-C*07 negativity is linked to a phenotype with higher CRP levels, more radiographic damage and less psoriasis prevalence. Confirmation within other big cohort studies is essential to test for these associations.
This study has several limitations. First, our study was not powered to fully correct for possible confounding caused by linkage disequilibrium (LD) of HLA-C with HLA-B*27 (16). However, HLA-B*27 prevalence did not differ significantly between HLA-C*07 negative and positive axSpA patients, both in GESPIC and in the SPACE cohort. And, although the gathering of axSpA and CBP patients in one cohort in the SPACE cohort enabled us to perform multivariate regression, showing that HLA-C*07 was linked to axSpA independently from HLA-B*27, larger cohorts including both axSpA and non-axSpA individuals will have to answer that question. Second, participants of the DKMS healthy donor cohort self-reported the absence of diseases, including axSpA, and one can thus not formally exclude that some of the healthy donors might have had axSpA or axSpA symptoms at the time of registration. It is however very unlikely that the prevalence of axSpA in the DKMS population is higher than in the general population and, if any selection bias would occur, the true HLA-C*07 prevalence difference would be larger than we measured. Third, GESPIC could not be corrected in detail for ethnicity with the DKMS donor cohort. Different ethnicity might result in different HLA-C allotype frequencies. However, HLA-C allotype frequency of all DKMS donors (n=141,049) did not differ from HLA-C allotype frequencies selected in our population (n=135,160) (data not shown), showing that matching for ethnicity does not influence HLA-C*07 prevalence in this cohort.
102