Introduction

Type I interferonopathies are a new class of disorders linked to the upregulation of type I interferon (IFN) [1, 2]. Stimulator of IFN genes (STING)-associated vasculopathy with onset in infancy (SAVI) is caused by gain of function mutations in TMEM173 [3], which lead to a constitutive production of high levels of type I IFNs without infectious triggers [3,4,5]. SAVI is characterized by systemic inflammation, chronic anemia, growth failure, cutaneous necrotic lesions, and interstitial lung disease [3, 6]. Low-titer autoantibodies (i.e., anti-nuclear antibody, anti-cardiolipin antibodies) can also be present. Therapeutic management is challenging: steroids are partially effective; patients respond poorly to disease-modifying therapies, such as methotrexate, mycophenolate mofetil, anti-malarials, infliximab, and rituximab [3, 6, 7]. Prognosis is poor especially in patients with a severe lung involvement, with a high mortality in the first two decades of life [3].

In light of the encouraging results obtained with the use of JAK inhibitors in other interferonopathies [8], and the ability of blocking type I IFN pathway activation in SAVI PBMC in vitro [3], a few SAVI patients have been treated with JAK1/2 inhibitors with good responses [9, 10].

We describe three SAVI patients with severe pulmonary involvement in which treatment with the JAK1/2 inhibitor, ruxolitinib, led to marked amelioration of disease manifestations that was sustained in one patient, but only transient in the other two.

Patients and Methods

The local ethics committees of Giannina Gaslini Institute and Ospedale Pediatrico Bambin Gesù approved the study.

IFN signature was performed as described [11] with minor modifications. Briefly, RNA was extracted from whole blood from peripheral venipuncture drawn in PAXgene tubes using PAXgene Blood RNA Kit (Qiagen, Hilden, Germany). cDNA was retrotranscribed using SuperScript® VILO™ cDNA Synthesis Kit (Invitrogen, Carlsbad, California, USA). Selected IFN-stimulated gene (IFI27, IFI44L, IFIT1, ISG15, RSAD2, SIGLEC1) expression was quantified by real-time PCR using gene-specific primers and probes (Roche) with the ddCt method relatively to a healthy donor calibrator using HPRT and G6PD as reference genes. In case samples were run in different assays, the same healthy donor calibrator was used and a positive control was added to assess inter-assay variability.

Mutations in the TMEM173 gene were assessed by standard Sanger sequencing (P1 and P2) or targeted resequencing using a customized panel and analyzed with the NextSeq sequencing platform (Illumina, San Diego, California, USA) (P3). Sequences are analyzed using a bioinformatics pipeline integrated in the BaseSpace Illumina System, and variants are called by VariantStudio Software. Variants identified by next-generation sequencing have been validated by Sanger sequencing.

The dyspnea score was defined as follows: 0: absence of dyspnea; 1: dyspnea only during intense physical activity, such as gymnastics at school; 2: dyspnea on moderate physical activity, such as climbing steps; 3: dyspnea walking; and 4: dyspnea at rest.

Results

Clinical Presentation

Patient 1 (current age 13 years old), born from healthy unrelated parents, presented at the age of eight months with isolated erythematosus-infiltrated skin lesions with pustular evolution followed by scarring, chilblains, and severe nail dystrophy by three years of age (Fig. 1a). Skin biopsy of an active nummular violaceous lesion and a scar resolution of a preexisting lesion revealed granulomatous nodular dermatitis, with deep granulomatous folliculitis and secondary fibrosis. The child presented recurrent episodes of bronchospasm, recurrent fever episodes, and worsening of skin lesions. At the age of eight years, focal thickening of the interlobular septa with areas of ground glass opacities with predominant sub-pleural distribution were evident at chest CT (Fig. 1a). A lung biopsy revealed lymphocytic aggregates in the peribronchial interstitial areas with aspects of capillaritis, and contiguous focal sub-atelectasis with macrophages infiltrating the alveoli (Table 1). Immunological studies revealed increased IgG and IgE, normal lymphocyte subset distribution, and normal lymphocyte proliferation in response to mitogens. Tetanus toxoid-specific IgG were below the protective level when tested at nine years of age (Table 2). Spirometry revealed mixed restrictive and obstructive features (Fig. 1a, Table 3). Differently from the cases described so far, mild renal involvement (microscopic hematuria, mild proteinuria, no biopsy was performed) was also present, associated with hypertension. Steroids (prednisone 1 mg/kg/day) ameliorated the clinical manifestations with normalization of inflammatory markers. However, disease relapses were observed during steroid tapering. Azathioprine and etanercept were ineffective. Peripheral blood type I IFN signature revealed the activation of type 1 IFN-induced gene expression (Supplementary Fig. 1A). Sanger sequencing of the TMEM173 gene identified a de novo c.463G>A p.Val155Met mutation [14].

Fig. 1
figure 1

Clinical abnormalities in SAVI patients and response to JAK inhibitor therapy. Images of skin lesions in the three patients are shown in a, b, and c. CT scan showing interstitial lung disease and ground glass appearance in P1 (a) and dense patchy areas of ground glass opacity (probably correlated to previous pulmonary hemorrhages), which appear centrilobular and lobular mainly in the upper lobes in P3 (c). Pulmonary function test of P1 (a) and P2 (b) showing a mixed restrictive and obstructive alteration. Response to therapy is shown in the lower panels. Curves represent patients’ FVC, with L/s in the y-axis and L in the x-axis. Straight lines represent expected values by age. In P3 (c), diffuse multifocal regions of ground-glass opacity and cystic airspace in the left lower lobe (arrow) were present six months after therapy. PFT were not performed due to age. d H&E staining of punch skin biopsy from P3 showing parakeratosis and basal vacuolar degeneration of the epithelium and voluminous dermal interstitial and perivasal neutrophil inflammatory infiltrates without vasculitic damage of vessel walls at different magnifications. L liters, FVC forced vital capacity

Full size image
Table 1 Patient characteristics
Full size table
Table 2 Immunological parameters
Full size table
Table 3 Response to therapy
Full size table

Patient 2 (current age ten years old) was recently described [15]. She presented at the age of three months with growth failure and respiratory distress (Table 1). Livedo reticularis was present on the lower limbs (Fig. 1b). The patient required several ICU admissions for ventilator support during lower airway infections (bronchiolitis and pneumonia). CT scan showed extensive ground glass abnormalities (Fig. 1b). Chronic Pseudomonas aeruginosa infection was documented. Lung biopsy showed interstitial fibrosis without signs of vasculitis. Echocardiographic imaging showed indirect signs of pulmonary hypertension and mild dilation of the pulmonary artery. Steroids were only partially effective, and association of methotrexate and infliximab were ineffective, resulting in steroid dependency. Daily oxygen support and non-invasive ventilation at night were required. Cheeks and nose telangiectatic skin lesions and unilateral vocal cord paresis were also observed (Fig. 1b, Table 1 and Table 3). Except for the elevation of total IgA and IgG and a mild lymphopenia, immunological studies did not reveal major alterations (Tables 2 and 3). Peripheral blood type I IFN signature was positive (Supplementary Fig. 1A). Targeted sequencing revealed a c.842G>A p.Arg281Gln mutation in TMEM173 [15].

Patient 3 (current age three years old), the only child of healthy, unrelated parents, presented at three days of life, with an erythematosus vesicular rash on the nose and cheeks, later spreading to her hands and feet with evolution to pustules and scars (Fig. 1c and Table 1). She was admitted at the age of three months with recurrent low-grade fever, cough, diarrhea, and dermatitis variably responsive to glucocorticoids and antibiotics in the suspect of a combined immunodeficiency. Skin biopsy was consistent with neutrophilic dermatosis with karyorrhexis throughout the vessel wall (Fig. 1d). Chest CT revealed focal thickening of the interlobular septa with areas of ground glass opacities. Blood investigations revealed microcytic anemia, increased acute-phase reactants, hypergammaglobulinemia, mild reduction of C3 levels, and low-titer autoantibodies (anti-cardiolipin), with a normal distribution of lymphocyte subsets. Lymphocyte proliferation was normal in response to PHA and low but present in response to OKT3 (Tables 1 and 2). Peripheral blood type I IFN signature was positive (Supplementary Fig. 1A). The patient was started on glucocorticoids (1 mg/kg/day prednisone) with improvement of clinical and laboratory findings, but recurrence of symptoms at tapering. Targeted genome sequencing revealed a de novo heterozygous mutation in TMEM173 c.461A>G (p.Asn154Ser).

Ruxolitinib Therapy Efficacy, Limits, and Side Effects

Treatment with ruxolitinib was started initially at 0.25 mg/kg/day in two doses, and subsequently progressively increased until clinical efficacy was obtained, namely, to 0.65 mg/kg/day in P1, 0.7 mg/kg/day in P2, and 1.25 mg/kg/day in P3.

All patients showed an initial clinical response starting in the first weeks of treatment and evident after three months (Figs. 1 a and 2 a, Table 3), with amelioration of lung disease (assessed by pulmonary functional tests and CT scans) and resolution of cutaneous lesions that was stable in P1 and only transient in P2 and P3. Furthermore, P1 showed resolution of the microhematuria and was able to progressively taper steroids and stop them after two years.

Fig. 2
figure 2

Response to JAK inhibitor treatment and side effects. Steroid dose and lung disease evolution in response to ruxolitinib therapy (a). Relative quantification of gene expression of six interferon-stimulated genes at baseline and after ruxolitinib treatment; “m + number” indicates months of therapy (b). Infection occurrence in P2 (c). HD healthy donors, FVC forced vital capacity

Full size image

P2 also showed amelioration of the skin phenotype, reduction of the dyspnea that correlated with an improvement in the 6-min walking test (O2 saturation at the end of the test improved from 84 to 98%), and increase in total forced vital capacity (Figs. 1 b and 2 b, Table 3). Furthermore, echocardiographic improvement of the indirect signs of pulmonary hypertension was documented, with normalization of pulmonary artery size.

P3 initially showed a good clinical response with amelioration of skin lesions and radiological lung findings (Fig. 1c, Table 3).

Notably, the clinical responses did not correlate with decreased type I IFN signatures, which improved only transiently in P1 during concomitant treatment with high dose steroids and ruxolitinib (Fig. 2b).

During three years of ruxolitinib treatment, P1 presented three flares of the skin lesions requiring brief cycles of steroids. No severe infections occurred. BK virus in the urines was intermittently detected (viral copies from thousands to 2 million/ml), without BK viremia, nor alteration of kidney function.

P2 (follow-up of 18 months) starting from seven months of therapy, experienced an increase in hospitalization because of respiratory infections, caused by Pseudomonas aeruginosa exacerbation, several episodes of rhinovirus infection, influenza A, varicella, and coronavirus (the last requiring mechanical ventilation and extracorporeal membrane oxygenation for 11 days), prompting the decision of reducing ruxolitinib dosage (Fig. 2c). At 18 months of therapy, CT scan revealed a worsening of the interstitial disease with a ground glass appearance and interlobular thickening (Fig. 1b).

P3 (follow-up of 12 months) after ten months on treatment on ruxolitinib presented clinical and radiological relapse of lung disease requiring glucocorticoid therapy (2 mg/kg/day of prednisone) with a prompt response (Fig. 2a). She never presented viral or bacterial infections since the beginning of therapy. Cell blood counts and ruxolitinib serum levels were regularly monitored (Supplementary Fig. 1B, Supplementary Table 1). No cytopenia was observed during the follow-up (Table 3). All patients are continuing ruxolitinib treatment at the time of writing.

Discussion

Our report confirms that JAK inhibition is a therapeutic resource worth considering for patients with SAVI, including very young children, as in the case of P3, also taking into account that the disease poorly responds to any other available immune-modifying agents. However, contrary to published reports, we observed several episodes of severe viral infections in one patient, suggesting the possibility that JAK inhibitors might significantly increase the risk of infections possibly in cases of more severe lung involvement or in the presence of genetic modifiers, not investigated in our patient, with consequent deterioration of the lung disease. Interestingly, the most frequent virus isolated was rhinovirus, which may be a direct consequence of effective type I IFN inhibition by ruxolitinib at least in respiratory epithelial cells, where IFN-β is required to control rhinovirus [16, 17]. On the other end, these infections might result from a cumulative effect of the drug with the reported developmental and in vitro proliferative defects of STING mutant T lymphocytes [18, 19]. Considering the severity of lung disease and the lymphopenia (Table 2) present before ruxolitinib treatment, P2 was started on antibiotic prophylaxis with Bactrim and Azithromycin; that, however, did not seem to prevent the febrile episodes, mostly if not always caused by viral pathogens. Of note, the patient is now on IVIG prophylaxis and no severe infections have been reported so far. Whether this is due to ruxolitinib reduction and/or IVIG treatment, it is difficult to understand.

We observed a variable medium- and long-term clinical response to the therapy. Only P1 seemed to reach a stable control of the disease, while P2 and P3 required at some point the introduction (P3) or the increase (P2) in steroids to control flares. Several factors might have played a role, including variability of disease severity and inadequate plasma drug levels. Indeed, while all patients were steroid dependent and required several hospital admissions to treat flares, P2 and P3 appear to have a more severe phenotype, with P2 presenting several life-threatening exacerbations requiring admission to ICU and ventilator support during infection episodes and P3 presenting a very early disease onset (three days of life). Additionally, despite dose adjustment and plasma level monitoring, drug dosage appeared to be stably adequate only in P1 (Supplementary Fig. 1 and Supplementary Table 1). Notably, we reduced ruxolitinib dosage in P2 after one year of therapy, in the attempt to prevent infections (Fig. 1c).

Overexpression of a set of interferon-stimulated genes (ISGs) in peripheral blood has consistently been found in all reported patients with SAVI syndrome [6, 14, 15]. Modulation of IFN signature following treatment may represent a surrogate biomarker. However, we did not observe a consistent decrease in IFN signature during the treatment. In previous reports of ruxolitinib treatment in patients with SAVI [9, 10, 20], only a partial or no decrease in IFN signature was observed. The lack of downregulation of IFN signature might be explained by the fast kinetic of JAK inhibition by ruxolitinib as shown by the transient decrease in pSTAT1 following drug intake [9] which seems not to be coupled with a clear impact on ISG expression (Supplementary Fig. 2). Otherwise, it might suggest that the six gene–based test is not appropriate to assess type I interferon activation at least in patients with SAVI treated with a JAK inhibitor.

Finally, it is important to consider that ruxolitinib inhibits not only the signaling of type I interferon receptor but also that of other receptors involved in important inflammatory pathways such as IL6, IL12/23, and interferon γ. It is possible that the inhibition of some or all of these pathways is responsible for the therapeutic efficacy (and side effects) we see in SAVI.

In conclusion, SAVI disease still appears to be orphan of ideal therapeutic targets. Despite the encouraging results obtained in this and previous reports [9, 10], prospective studies with a greater number of patients are warranted to address all the crucial issues linked to JAK inhibitor use in patients with SAVI.