Understanding heterogeneity in JIA
medwireNews: In a session a the Annual European Rheumatology Conference (EULAR) 2017 providing insights into juvenile idiopathic arthritis (JIA) heterogeneity, EPOCA data were presented suggesting that geographic variability in JIA epidemiology, treatment, and outcomes may go some way to explaining it, while research deciphering the biologic basis of the condition could help combat it.
Geographic variations in JIA epidemiology and treatment
Alessandro Consolaro (Istituto Giannina Gaslini, Genoa, Italy) discussed the findings from the EPOCA – Epidemiology treatment, and outcome of Childhood Arthritis – study, which found each of these factors varied across different countries throughout the world.
His premise was that geographic or racial differences in JIA phenotypes and the variability in treatment approaches, not least as a result of biologic availability and affordability, could impact on disease outcome.
A total of 49 countries took part in the study, spanning eight geographic areas, including Northern, Central, Mediterranean, and Eastern Europe; North and Latin America; North Africa plus the Middle East; and South-East (SE) Asia. This provided data on 9137 patients with JIA and 4000 healthy children as controls.
The prevalence of JIA subtypes varied widely across the geographic areas, with oligoarthritis, for instance, being particularly prevalent in Mediterranean Europe, whereas it was seen at almost the same frequency as rheumatoid factor-negative polyarthritis in North and Latin America.
In Latin America and North Africa, systemic JIA was the most prevalent subtype, while in SE Asia, the “picture is pretty different,” said Consolaro. Here, systemic JIA and enthesitis-arthritis patients had “by far the highest percentage of prevalence.”
The age at which patients developed JIA also varied. Patients in Mediterranean Europe and Northern Europe developed the disease early, at 3.5 and 4.7 years, respectively, whereas patients in North America and SE Asia tended to have a later onset of disease. The differences were evident at the country level as well, with the median age ranging from close to 2 years to above 10 years. There was a large difference even within Europe, with a 3.1-year age difference between age at onset in Mediterranean and Eastern Europe, at 3.5 years and 6.6 years, respectively.
Consolaro commented that the variance in JIA subtype distribution is a probable explanation for the differences in age at disease onset, but he added that “it is not the only one;” given that the differences were still evident when the different JIA subtypes were assessed separately.
Variability also existed in comorbidities associated with JIA, with anterior uvetis being the most common. But while it occurred at a frequency of more than 15% in Northern and Mediterranean Europe, the frequency was lower in other areas, at nearer 10%.
This comorbidity was not only prevalent in patients with oligoarthritis, Consolaro noted; there was also a high frequency in patients with rheumatoid factor-negative polyarthritis and psoriatic arthritis, at least in Europe.
Moving on to the treatment of JIA in these different areas, Consolaro reported that half of JIA patients in Northern Europe had received at least one biologic agent, contrasting with the lowest frequency, of 35%, among patients in SE Asia.
Patients with systemic JIA and rheumatoid factor-positive polyarthritis in North America and North/Mediterranean Europe were the most likely to receive biologic treatment, at over 60%, and patients in these areas had the lowest disease activity, according to the 10-point Juvenile Arthritis Disease Activity Score.
Significantly higher disease activity was seen in patients treated in North Africa and the Middle East and in Eastern Europe compared with other areas.
Indeed, there was “modest” negative correlation between the countries’ Gross National Income (GNI) per capita and disease activity, and a positive correlation with the frequency of prescriptions for biologics, Consolaro noted.
Similarly, the percentage of patients with inactive disease was over 30% in Latin America, North America, and Mediterranean and Central Europe, which Consolaro said “goes together with the frequency of patients treated with biologics and the frequency of patients with active disease.”
The frequency of articular and extra-articular damage followed the same pattern, being higher in South East Asia and Eastern Europe, at 39% and 26%, respectively, and lowest in Central Europe and North American, at a corresponding 14% and 8%. And the frequency of patients with damage from countries with below-median GNI and health expenditure per capita was about double that of countries where income and expenditure were above the median.
Consolaro summarized that there are wide differences in JIA characteristics across geographic regions and that patients with JIA living in non-Western countries had higher levels of disease activity and cumulative damage than patients in North American and Western Europe.
“This disparity in disease outcome may be partially due to differences in the availability or affordability of biologics, as usage of these medications was more common in Western pediatric rheumatology centers,” he concluded.
Clues from biology for deciphering heterogeneity in JIA
In another presentation, Rae Yeung, from the University of Toronto in Canada, outlined how looking at the biology of JIA as well as the clinical signs and symptoms could help further our understanding of the heterogeneity surrounding the disease.
Conditions comprising JIA are “incredibly” different, with the seven International League of Associations for Rheumatology (ILAR) subtypes showing very different clinical phenotypes, not only in their disease trajectory, but also in their response to treatment and their clinical course, Yeung commented.
“It is very important that we understand and take the successes that we’ve learnt […] from biology to better understand and dissect the heterogeneity between these groups,” she proposed.
They gathered multilayers of information, including clinical phenotype, protein expression, cellular phenotype, gene expression, single-nucleotide polymorphisms, and metabolome, on children with newly diagnosed arthritis from the ReACCh-Out study and another independent validation cohort. Then, using dimensionality reduction, they compressed the hundreds of variables in a principle components analysis and identified four that best described the large sources of variation among the nearly 200 patients.
Using these four components they were able to describe 45% of all of the differences between the patients. Yeung pointed out that, although these composite descriptors were obtained in a very data driven way, “they make sense.”
She reported the “incredibly surprising” finding that the variables making up the four key components are all “very familiar variables.”
The first component (PC 1) was made up of the pro-inflammatory cytokines, the second component (PC 2) comprised common measures of disease activity in patients with JIA, such as joint counts and inflammasome activation, the third component (PC 3) divided the cytokine profile in a more robust way, into macrophages as well as T cells, and also identified age at onset of disease as an important contributing factor. The fourth component (PC 4) further divided the biologic variables into T-helper cell (Th)1, Th17, and Th2.
They are “very logical, meaningful, clinical, and biologic measures that divide very nicely into these composite descriptors of these patients,” Yeung remarked.
Distribution of the patients across these four components revealed five unique and distinct clusters. For instance, cluster V patients scored highly for PC 1 and so had increased circulating levels of proinflammatory cytokines, while cluster III patients scored highly for PC 2 and had the highest measures of disease activity.
Yeung and team then considered whether their four components perform better than the current ILAR criteria. They had a much higher significance than the ILAR criteria, with both the clinical and biologic variables also outperforming the ILAR criteria when considered separately.
Yeung pointed out that the new criteria are a lot more homogenous in the way they group patients, reducing some of the heterogeneity seen when using the ILAR criteria, and are very stable. Kruskal Wallis analysis showed that about 30 variables could be removed without disrupting the cluster in which the patients were grouped.
The variants that were identified as being key were erythrocyte sedimentation rate (ESR), number of active joints, number of effused joints, levels of interleukin (IL)-1α, IL-6, enthesitis, IL-4, and C-reactive protein, Child Health Assessment Questionnaire score, and visual analog scale for pain score.
In terms of clinical relevance, Yeung pointed out that the disease trajectories in each of the five clusters are much more uniform than those seen using ILAR criteria.
She used cluster V patients as a particular example. These patients have very low levels of disease activity (PC 2) and so look well at the bedside, yet have very high levels of pro-inflammatory cytokines (PC 1). This group actually does poorly and so subcritical disease activity is not recognized when biologic measures of the cytokine profile are not included, she stressed.
Yeung added that their current research into how JIA-related genes are associated with these five biologically distinct groups suggests that “only one” gene, in fact, distinguishes the group enriched for pro-inflammatory cytokines and she hopes to be able to present these data very soon.
She concluded by saying that regardless of the analysis method used, they were able to come up with these five groups of patients that are biologically distinct. And whether looked at according to gene expression, cytokine expression, or other variables, they remain biologically unique, separating “beautifully” into five different groups.
By Lucy Piper
medwireNews is an independent medical news service provided by Springer Healthcare. © 2017 Springer Healthcare part of the Springer Nature group