Objective: To assess the performance of the new EULAR/ACR criteria, particularly for early detection of cSLE, in comparison to the SLICC criteria among the pediatric population in multiple centers in Saudi Arabia.
Methods: We conducted a retrospective study that enrolled pediatric patients up to the age of 14 years who've been diagnosed with SLE and followed in pediatric rheumatology clinics at 9 multi-tertiary hospitals in Saudi Arabia from 2010 to 2021 as a case group and were compared to a similar group of pediatric patients who've had defined rheumatological diseases other than SLE with a positive ANA titer (≥1:80) as controls. In total, 245 patients were included and distributed as 129 cases (diagnosed by expert pediatric rheumatologists) versus 116 patients in the control group. All relevant clinical information, including history, physical examination findings, and laboratory tests, was documented at the initial presentations. Then, the two sets of SLE classification criteria were applied to both groups to define who's going to meet both or either one of them. The exclusion criteria included those who had insufficient data or had overlapping or undifferentiated diseases. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), receiver operating curve (ROC), and accuracy were calculated for SLICC 2012 and EULAR/ACR 2019 criteria (total scores≥ 10 and ≥ 13). We performed a Chi-squared test to compare sensitivity and specificity of SLICC 2012 and EULAR/ACR 2019.
Results: For SLICC (cut-off ≥4 criteria), the sensitivity was found to be 96.9% (95% CI 92.6%-99.4%) and the specificity was 94.8% (95% CI 89.6%-98.32%), with PPV and NPV of 95.4% and 96.5%, respectively. The ROC for it was 0.96 (95% CI 0.93-0.99), and this criterion had an accuracy of 95%. Regarding EULAR/ACR (total score ≥ 10), the performance measure showed a sensitivity of 99.2% and a specificity of 86.2%. Similarly, PPV was 88.9%; while NPV was a little higher (99.0%) than SLICC. The ROC for EULAR/ACR (total score ≥ 10) was 0.93 (95% CI 0.89-0.96), and this criterion had an accuracy of 93%. However, there was no statistically significant difference between the sensitivity and specificity of either using SLICC or EULAR/ACR (total score ≥ 10), as reflected by a p-value of 0.86 using the Chi-squared test. Although applying the EULAR/ACR with a total score of ≥ 13 revealed lower sensitivity (93.8%) than both the SLICC and the EULAR/ACR (total score ≥ 10), the specificity for it was found to increase up to 91.4% (85.7-96.2%) compared to the (86.2%) specificity of the EULAR/ACR (total score ≥ 10).
Conclusion: In this cohort among the Saudi population with childhood-onset SLE, the new EULAR/ACR 2019 criteria efficiently enable early detection of SLE, although a more frequent rate of false positives was observed with them. Escalating the total score from ≥ 10 to ≥ 13 in the cSLE population improved the specificity close to that of SLICC 2012. Further prospective studies in pediatrics need to be done for the validation of a cut- off score of ≥ 13 in cSLE rather than the traditional score of ≥ 10 in aSLE.
Keywords: Childhood onset systemic lupus erythematosus; SLICC 2012 criteria; new EULAR/ACR 2019 criteria; systemic lupus erythematosus; systemic lupus erythematosus classification criteria.