Profiling type I and II interferon responses reveals distinct subgroups of pediatric patients with autoinflammatory disorders

Abstract

Background: Elevation of type I interferon (IFN-I) is characteristic of a group of diseases known as type I interferonopathies. Several technologies are available to monitor IFN-I, but there is no consensus on their routine use in medical laboratories.

Objective: We aimed to compare the performance of two technologies for this purpose: NanoString, which monitors messenger RNA expression of interferon-stimulated genes (ISGs), and Simoa, which quantifies IFN-α2 protein in an ultrasensitive way. We also designed a NanoString assay to monitor type II ISGs and tested its value to discriminate clinical conditions.

Methods: A total of 196 samples from patients with diseases associated or not with IFN-I pathway activation were analyzed by NanoString and Simoa.

Results: The comparison between NanoString IFN-I score and IFN-α2 Simoa revealed a r ² coefficient of 0.55. We identified IFI27, IFI44L, and SIGLEC1 as the ISGs most closely related to IFN-α2 concentration. Nineteen samples had a positive IFN-I score but undetectable IFN-α2. These samples were also positive according to IFN-II score, pointing to IFN-II as the primary ISG inducer in corresponding patients. By measuring IFN-I and IFN-II scores in a subset of patients with systemic lupus erythematosus and systemic juvenile idiopathic arthritis, we identified two subgroups of patients in whom IFN-I and IFN-II were dominant.

Conclusion: Both IFN-α2 quantification and NanoString reliably distinguish type I interferonopathies from other diseases. Type I and II interferons induce different transcriptomic signatures in vitro and in vivo, and our results highlight the value of monitoring both IFN-I and IFN-II in interferon-related diseases.

Keywords: IFN signature; Interferonopathy; NanoString; Simoa; juvenile idiopathic arthritis; lupus; type I interferon; type II interferon.

Fig 1

IFI27, IFI44L, and SIGLEC1 expressions are most closely related to IFN-α concentration. We analyzed 196 samples from 119 patients by NanoString technology and Simoa. (A) Association between IFN-α2 protein levels and IFN-I score was assessed by Spearman rank correlation coefficient and analyzed by tobit regression model by quadratic polynomials. (B) Association between level of expression of each of 6 interferon-stimulated genes composing IFN-I score and IFN-α2 protein concentration.

Fig 2

Comparison of diagnostic performance of NanoString and Simoa IFN-α2. Samples from 116 patients with either interferonopathies (22 SLE, 7 DM, 4 monogenic type I interferonopathies, 9 other connective tissue diseases) or other diseases (n = 74) and 7 healthy controls were analyzed for Simoa IFN-α2 and IFN-I signature by NanoString. (A) IFN-α2 concentration (fg/mL) and IFN-I score (log) plotted for each sample by disease. Statistical analysis was performed by Kruskal-Wallis test (ns, not significant; ∗P < .05, ∗∗∗P < .001, ∗∗∗∗P < .0001). Black horizontal lines represent mean for each group. (B) Analysis of ROC curve of IFN-I score and plasma IFN-α2. (C) NanoString and IFN-α2 Simoa were performed on 116 samples and interpreted with 3.1 and 4.7 fg/mL used as thresholds for IFN-I score and IFN-α2 Simoa, respectively. Thirty discordant results were identified between these two techniques: 11 IFN-I score^negIFN-α2^pos samples (red dots) and 19 IFN-I score^posIFN-α2^neg samples (blue dots). IFNopathies, Interferonopathies.

Fig 3

IFN-I and IFN-II induce distinct transcriptomic signatures identified with NanoString technology. (A) Fold change increase of expression of two ISGs composing IFN-I score (SIGLEC1, IFI44L) after stimulation of whole blood cells with either IFN-γ or IFN-α2. (B) Fold change increase of expression of 5 previously identified IFN-γ–inducible genes (ANKRD22, FCGR1B, HLA-DMB, HLA-DPB1, HLA-DRB3) after stimulation of whole blood cells with either IFN-γ (blue dots) or IFN-α2 (red dots). (C) Coanalysis of IFN-I score and IFN-II score after IFN-γ (blue dots) or IFN-α2 (red dots) stimulation. IFN-II score was calculated as median of 5 IFN-γ–inducible genes’ relative expression compared to healthy controls. Statistical analysis was performed by Kruskal-Wallis test (ns, not significant; ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, ∗∗∗∗P < .0001).

Fig 4

Concomitant analysis of IFN-I score and IFN-II score reveals different transcriptomic profiles within single disease. (A) IFN-II score calculated for 13 discordant IFN-I score^posIFN-α2^neg samples. Statistical analysis was performed by Wilcoxon-Mann-Whitney U test (∗∗P < .01). Black horizontal lines represent mean for each group. (B) Coanalysis of IFN-I score and IFN-II score of patients with monogenic type I interferonopathies (n = 6, blue dots), patients with MAS (n = 4, red dots), and 13 patients with IFN-I score^posIFN-α2^neg(black dots).(C) Coanalysis of IFN-I score and IFN-II score of patients with SLE (n = 16). (D) Coanalysis of IFN-I score and IFN-II score of patients with sJIA (n = 8).