Two- and 3-year outcomes of the childhood arthritis and rheumatology research alliance FiRst Line Options for sJIA Treatment (FROST) trial

Abstract

Background

Systemic juvenile idiopathic arthritis (sJIA) is a severe autoinflammatory disease associated with substantial morbidity, including macrophage activation syndrome (MAS), sJIA-associated lung disease, and glucocorticoid (GC)-related toxicity. CARRA’s FiRst Line Options for sJIA Treatment (FROST) study demonstrated favorable short-term outcomes with early biologic therapy, but long-term outcomes have not yet been described.

Methods

Patients with new-onset sJIA enrolled in FROST (2016–2019) were followed longitudinally through the CARRA Registry. Four consensus treatment plans (CTPs) were evaluated, including IL-1 inhibition (either anakinra or canakinumab), IL-6 inhibition (tocilizumab), and two non-biologic arms (methotrexate or GC monotherapy). Long-term follow-up occurred through 36 months. The primary outcome was achievement of clinical inactive disease (CID) by Wallace/ACR provisional criteria without concurrent GC use at 24 and 36 months. Secondary outcomes included CID irrespective of GC use, cJADAS-10 ≤ 2.5 without fever, medication utilization, and adverse events. Adverse events were expressed as events per 100 person-years.

Results

Seventy-three patients were enrolled; 87.6% initiated biologic therapy making comparison to non-biologic therapy infeasible. At 24 and 36 months, 41 and 32 patients, respectively, had evaluable data for the primary outcome. CID without GC use was achieved by 58.5% of patients at 24 months and 65.5% at 36 months. At both time points, over 80% of patients achieved cJADAS-10 ≤ 2.5 without fever and without GC use. Nearly all patients were off GCs at 24 and 36 months (> 90%). The cumulative proportion of patients achieving CID at any time during follow-up increased to 78.8% by 24 months and 87.6% by 36 months. At last follow-up, 42.5% of patients were off all disease-modifying antirheumatic drugs (DMARDs). Over 228.2 person-years of observation, serious adverse events occurred at a rate of 6.6 per 100 person-years; MAS was the most common SAE. No cases of sJIA-associated lung disease were identified.

Conclusions

Although results were limited by loss to follow-up, patients enrolled in FROST demonstrated increasing rates of CID and high rates of glucocorticoid discontinuation during long-term follow-up. Serious adverse events were uncommon and largely related to disease activity. These findings support early biologic therapy as an effective and durable first-line strategy for sJIA.

Clinical trial number

Not applicable.

Background

Systemic Juvenile Idiopathic Arthritis (sJIA) is an autoinflammatory condition characterized by an abnormal innate immune response and significant systemic inflammation driven by massive overproduction of proinflammatory cytokines, in particular IL-1 and IL-6 [1, 2]. The most common symptoms include fever, arthritis, and a distinguishing evanescent rash. sJIA is associated with life threatening complications including macrophage activation syndrome (MAS) and the more recently described sJIA associated lung disease (sJIA-LD) [3–5]. Historically, there was also significant morbidity related to chronic steroid exposure which was often necessary to control disease.

Considerable advances in the understanding of the inflammation in sJIA over the past two decades have led to more targeted therapies including IL-1 and IL-6 inhibitors. In 2022 the Childhood Arthritis and Rheumatology Research Alliance (CARRA) published the results of a prospective observational study called FiRst Line Options for sJIA Treatment (FROST), which was designed to compare the effectiveness of 4 Consensus Treatment Plans (CTPs) in new-onset systemic JIA (sJIA) [6].

In FROST, 86% of patients were treated according to one of the biologic CTPs (IL-1 inhibitors or IL-6 inhibitors) at enrollment. The most commonly reported reasons for choosing a biologic CTP were likelihood of effectiveness for systemic features, minimization of systemic glucocorticoids, and likelihood of effectiveness for arthritis. Half of the patients who started on a non-biologic CTP at enrollment started a biologic during the first three months of follow up. The small number of patients treated with non-biologic CTPs made it impossible to compare rates of achieving clinical inactive disease (CID) between the biologics and nonbiologic treatment groups [7]. Nevertheless, there were encouraging results from the study, which included that the majority of patients were able to achieve a state of CID without concurrent use of glucocorticoids at 9 months. The most frequently reported adverse events included MAS and hospitalization for flare of underlying disease.

The purpose of this study is to describe the long-term clinical outcomes and adverse events for patients enrolled in the FROST trial.

Methods

The methodology used to develop the four CTPs has been previously reported [9, 10], as has the design of the FROST study [6, 11]. FROST enrolled patients from 2016 to 2019 from 32 CARRA sites throughout the US and Canada. The CTPs were grouped into biologic CTPs, in which patients were initially treated with an IL-1 inhibitor (anakinra or canakinumab) or IL-6 inhibitor (tocilizumab), and non-biologic CTPs, which included glucocorticoids (GC) alone or methotrexate with or without glucocorticoids. Selection of treatment plan, glucocorticoid tapering, and DMARD switching were determined through shared decision-making between the provider and the patient or parent. Patients were excluded if they had MAS or if the treating physician felt that they could not safely be treated with one of the treatment arms. All treatment arms included optional GCs including the option for pulses doses at diagnosis. Steroid tapering was at the discretion of the treating physician with the goal of reaching 50% of the starting dose by 3 months and discontinuing GCs at 6 months.

All patients enrolled in FROST were also enrolled in the CARRA Registry [8], which allowed for continued observation of clinical status, medication usage, and adverse events after the conclusion of the FROST study at 24 months. Long-term follow up included study visits occurring every 3 months between 12- and 24-months post enrollment, subsequently continuing to have follow up visits every 6 months (+/- 3-month window) in the CARRA Registry until they aged out of pediatric rheumatology care, transferred care to a non-CARRA Registry site, were discharged from rheumatology care due to long-term remission, or were lost to follow up, defined as no study visit for 15 consecutive months.

The CTPs included recommendations for the timing of clinical reassessment and adjustment of glucocorticoids and DMARDs over the first 9 months of treatment. Beyond 9 months, the timing of clinical follow up, tapering of glucocorticoids, and sequence of medication changes was determined solely by the treating physician and the patient/family. Serious adverse events (SAE) were recorded, along with prespecified adverse events that may not have met SAE criteria, including MAS.

Study Outcomes

The primary study outcome was achievement of CID according to the Wallace/American College of Rheumatology provisional criteria without use of glucocorticoids at 24- and 36-months [7]. Selected secondary outcomes included CID irrespective of current GC use, clinical juvenile idiopathic arthritis disease activity score based on 10 joints (cJADAS-10) ≤ 2.5 without current fever or GC use, cJADAS-10 ≤ 2.5 without current fever irrespective of GC use, GC use independent of CID, any biologic or non-biologic DMARD use, and proportion of the entire cohort achieving CID at any point during follow up. The proportion of the cohort having achieved CID at each study visit and the proportion achieving CID at any point up to and including each study visit were both calculated and plotted over time.

Criteria for CID requires that patients have (1) no active arthritis (2) no systemic features of sJIA (3) no active uveitis (4) physician global assessment of disease activity score of zero (5) duration of morning stiffness ≤ 15 min (6) ESR or CRP within normal limits, if available. The cJADAS-10 ranges from 0 to 30 and is composed of joint count (maximum of 10), physician global assessment (0–10), and patient/parent global assessment (0–10). We used the cutoff for defining inactive disease in polyarticular JIA of a cJADAS-10 of ≤ 2.5 but also required the absence of fever [12]. Because the height of fever is not collected in the CARRA Registry (only the presence or absence of fever attributable to sJIA), the systemic Juvenile Arthritis Disease Activity Score (sJADAS) could not be calculated [13].

Both the 24- and 36-month visits had a 6 month visit window (i.e., 18 to 24 months and 30 to 36 months, respectively). For patients with more than one study visit during the six-month window, they were considered to have achieved CID if they met the definition of CID at any one of the study visits.

Composite outcomes (CID and cJADAS) were assessed using all available component data at each time point. Patients were classified as not achieving the endpoint if any observed component indicated active disease.

Classification as achieving the endpoint required sufficient data to confirm that all components met remission criteria; therefore, patients with incomplete data who could not be definitively classified were excluded from the denominator for that specific analysis. As a result, denominators varied depending on component availability and the specific composite definition used. Recent medication use at a study visit was defined as either being on the medication at the visit or having stopped the medication within the 2 months prior to the visit to ensure that disease was in stable remission off medication. Data at the time of last study follow-up visit is also presented. Adverse events are expressed as events per 100 person-years, defined as the cumulative duration of safety follow-up across all participants. The CARRA Registry and FROST were approved by the Duke IRB.

Results

Baseline characteristics of the 73 patients enrolled in FROST are shown in Table 1 including those who were included in the analysis of the primary outcome at 24 and 36 months (N = 41 and N = 32, respectively). Due to missed visits, not all patients included in the 36-month outcome were included in the 24-month outcome assessment. As expected, the majority of patients in long-term follow-up had started on a biologic CTP (87.6%). During the follow-up period, the most frequently administered medication was anakinra, utilized in 68.5% of patients at some point. Canakinumab was the second most common treatment, prescribed to 56.2% of patients, followed by tocilizumab, which was used in 26% of cases. Notably, 19.1% of patients received both an IL-1 inhibitor and an IL-6 inhibitor at some time during the follow-up period. By 36 months there was only 1 patient in follow up who had started on a non-biologic CTP. Disease activity at baseline, measured by median cJADAS-10, physician global assessment, patient global assessment, joint count and all lab measures, for patients included in long-term follow up was similar as for the entire cohort.

Table 1.

Baseline patient characteristics for the entire cohort (n = 73) as well as baseline characteristics for patients with follow up available at 24 (n = 41) and 36 months (n = 32). CTP, Consensus Treatment Plan, CHAQ, Childhood Health Assessment Questionnaire, cJADAS, Clinical Juvenile Arthritis Disease Activity Score

n	All Patients	Month 24 Patients	Month 36 Patients
	73	41	32
Follow Up (years)	3.3 [1.6, 4.4]	3.5 [2.4, 4.5]	4.4 [3.4, 5.5]
CTP, n (%)
Glucocorticoids	6 ( 8.2)	3 ( 7.3)	0 ( 0.0)
Methotrexate	4 ( 5.5)	2 ( 4.9)	1 ( 3.1)
IL-1 inhibitor	59 ( 80.8)	35 ( 85.4)	30 ( 93.8)
IL-6 inhibitor	4 ( 5.5)	1 ( 2.4)	1 ( 3.1)
Age	6.8 [4.1, 11.0]	7.1 [4.0, 11.5]	6.1 [4.6, 11.3]
Male sex, n (%)	44 ( 60.3)	25 ( 61.0)	23 ( 71.9)
Physician global assessment	7.0 [5.0, 7.0]	7.0 [5.0, 7.5]	7.0 [5.0, 7.0]
Patient global assessment	6.0 [3.2, 8.0]	5.5 [3.0, 8.0]	6.0 [3.0, 8.0]
Number of active joints	4.0 [2.0, 8.0]	5.0 [2.0, 9.5]	5.0 [2.0, 8.8]
Fever, n (%)	73 (100.0)	41 (100.0)	32 (100.0)
Arthritis, n (%)	73 (100.0)	41 (100.0)	32 (100.0)
Rash, n (%)	70 ( 95.9)	38 ( 92.7)	30 ( 93.8)
Lymphadenopathy, n (%)	24 ( 32.9)	14 ( 34.1)	11 ( 34.4)
Hepatomegaly or Splenomegaly, n (%)	15 ( 20.5)	7 ( 17.1)	5 ( 15.6)
Serositis, n (%)	7 ( 9.6)	5 ( 12.2)	5 ( 15.6)
ESR (mm/hr)	73.0 [57.0, 97.0]	75.0 [58.0, 97.0]	67.5 [53.0, 97.0]
CRP (mg/dL)	9.0 [4.7, 16.0]	12.2 [5.4, 16.2]	8.9 [5.2, 15.5]
Ferritin (ng/mL)	829.0 [249.0, 2603.0]	604.5 [234.4, 1543.7]	859.0 [346.0, 2367.2]
Hemoglobin (g/dL)	10.1 [8.9, 11.2]	10.0 [8.9, 11.4]	10.1 [9.0, 11.2]
WBC (x10^3)	12.2 [8.5, 19.1]	11.5 [8.8, 15.8]	11.3 [8.9, 19.0]
Platelets (x10^3)	458.0 [353.0, 571.0]	399.0 [331.5, 570.5]	397.0 [360.5, 500.2]
CHAQ	1.4 [0.5, 2.0]	1.2 [0.2, 1.9]	1.1 [0.1, 2.0]
cJADAS-10	17.0 [10.5, 21.5]	17.0 [10.8, 21.2]	17.0 [11.5, 21.0]
Race (%)
Asian	6 ( 8.2)	4 ( 9.8)	5 ( 15.6)
Black	7 ( 9.6)	5 ( 12.2)	2 ( 6.2)
Hispanic	13 ( 17.8)	4 ( 9.8)	4 ( 12.5)
White	42 ( 57.5)	24 ( 58.5)	18 ( 56.2)
Other	1 ( 1.4)	0 ( 0.0)	0 ( 0.0)
Missing	4 ( 5.5)	4 ( 9.8)	3 ( 9.4)

As of April 2024, there were 27 FROST patients still actively following up in the CARRA Registry. Thirty-five patients had been lost to follow up. The remaining 11 patients either withdrew consent, transitioned to adult rheumatology or were deceased (Table 2).

Table 2.

Status within the CARRA Registry at the time of the analysis for the 73 patients enrolled in FROST

Status	n	%
Still active	27	37.0%
De facto loss-to-follow-up	35	47.9%
Disease remission	6	8.2%
Subject withdrew consent	3	4.1%
Transitioned to adult rheumatology	1	1.4%
Deceased	1	1.4%

The primary outcome of CID without concurrent use of glucocorticoid was available for 41 patients at 24 months and 32 patients at 36 months. More patients had data available for calculating cJADAS-10 at both visits. Table 3 summarizes the clinical outcomes at 24 and 36 months. At 24 months 58.5% of patients were in CID without concurrent use of glucocorticoids. This percentage increased to 65.5% at 36 months. There was 1 additional patient in CID but still on glucocorticoids at 24 months. At 24 months and 36 months the percentage of patients that had cJADAS-10 ≤ 2.5 without fever and without current glucocorticoid use was 81.6% and 80.0%, respectively. Almost all patients were off glucocorticoids at 24 months (90.5%) and at 36 months (90.6%).

Table 3.

Clinical outcomes at 24 and 36 months (Both timepoints had a 6 month window. The 24 month visit was from 18 months to 24 months. The 36 month window was from 30–36 months). CID, Clinical Inactive Disease, cJADAS, Clinical Juvenile Arthritis Disease Activity Score, GC, glucocorticoid

CID without current GC use	24 months	36 months
	24/41 (58.5%)	21/32 (65.6%)
CID irrespective of current GC use	25/41 (61.0%)	21/31 (67.7%)
CID without GC or recent biologic use	10/49 (20.4%)	8/33 (24.2%)
cJADAS ≤ 2.5 and no fever without current GC use	40/49 (81.6%)	32/40 (80.0%)
cJADAS ≤ 2.5 and no fever irrespective of current GC use	42/47 (89.4%)	32/36 (88.9%)
cJADAS ≤ 2.5 and no fever without current GC use or recent biologic use	19/53 (35.8%)	17/40 (42.5%)
On GC	4/42 (9.5%)	3/32 (9.4%)
Recent Biologic use	30/41 (73.2%)	25/32 (78.1%)

Figure 1 shows the cumulative proportion of patients reaching CID at any time during follow up. This percentage continued to increase throughout the duration of the long-term follow-up reaching 78.8% of patients at 24 months and 87.6% by 36 months. Figure 2 shows how the proportion of patients in CID at the specified study visit increases up to 15 months after enrollment then fluctuates for the duration of the study period.

Fig. 1.

Cumulative proportion of patients achieving CID (Clinical Inactive Disease) for the first time during long term follow up

Fig. 2.

Proportion of patients in CID (Clinical Inactive Disease ) at each study visit with 95% confidence interval

Adverse events are summarized in Table 4. There was a total of 228.2 person-years of safety observation with a median of 3.3 years (IQR 1.6, 4.4) per patient. There were 15 SAEs observed for a frequency of 6.6 per 100-patient years. The median time between enrollment and SAE was 218 days. There was 1 death due to acute liver failure occurring 2.6 years after enrollment which was previously reported [6]. The most commonly reported SAE was MAS which occurred in 6 subjects. Two of those patients were not receiving biologics or methotrexate at the time of MAS and the others were all receiving IL-1i at the time of MAS. No patients enrolled in FROST were subsequently diagnosed with sJIA-LD during long-term follow-up; however, screening for sJIA-LD was not systematic or protocolized.

Table 4.

Serious Adverse Events. CTCAE, Common Terminology Criteria for Adverse Events, IV abx, intravenous antibiotics, MAS, Macrophage Activation Syndrome

Event	CTCAE grade	Current Biologic and Non-Biologic use
Acute Liver Failure	5	Anakinra
Hepatitis	4	Anakinra, Tocilizumab
Protein Losing Enteropathy	3	Anakinra
Severe injection site reactions	3	Anakinra
Infection requiring IV abx	3	Canakinumab
Appendicitis	3	Golimumab
MAS	4	Anakinra
MAS	3	Canakinumab
MAS	3	Canakinumab
MAS	3	None
MAS	3	Anakinra, Canakinumab
MAS	3	None
sJIA Flare	3	Canakinumab
sJIA Flare	3	Anakinra, cyclosporine
sJIA Flare	2	Anakinra, Cyclosporine, methotrexate

Disease status at the last study visit was analyzed for the entire cohort. Importantly, for 35.6% of patients, CID could not be calculated at the most recent study visit due to missing data. Therefore, analysis was limited to patients whose last study visit included all the variables needed to calculate CID status (n = 47). Overall, 50.6% of these patients were in CID at the last study visit. This includes 9 patients whose last visit occurred within 3 months after enrollment. When limited to subjects whose last study visit occurred at 24 months or later (n = 38), 50% s were in CID. At the last study visit 42.5% of all 73 patients were off all DMARDs.

Discussion

The long-term outcomes of the FROST trial provide valuable insights into the evolving therapeutic landscape of sJIA, underscoring the effectiveness and safety of early biologic use in achieving a state of inactive disease. As previously reported, the majority of patients in FROST had achieved CID (57%) or had a cJADAS-10 of ≤ 2.5 without fever or glucocorticoids (75%) at 9 months. This current study demonstrated that for patients who did not have complete response in the first 9–12 months there is still opportunity for them to achieve CID. This was best demonstrated by the increase in the proportion of patients meeting the primary endpoint of CID at 24 months from 58.5% to 65.5% at 36 months a trend best demonstrated in Fig. 1. It was also highly encouraging to see that nearly all patients were off glucocorticoids at 24 months after enrollment while nearly half of patients were off all DMARDs at the last study visit.

Our observations regarding sJIA outcomes are consistent with other cohorts with longer observation times that showed patients with sJIA have the highest chance of remission off medications among all JIA categories. In a prospective single center cohort of children with systemic juvenile idiopathic arthritis (sJIA) managed with a treat to target approach and followed for five years, 96% of patients achieved remission at the five-year time point, with the majority (75%) discontinuing all medications by that time [14]. This study, in conjunction with a similar follow up study, suggests that early initiation of anakinra, followed by biomarker-guided discontinuation, may be associated with high remission rates both within the first year and at long-term follow-up [15]. Similar long term outcomes were reported from a large retrospective cohort in which 98% of patients were in remission at a median of 5 years after starting treatment with 48.8% in remission off medication [16]. In a small cohort of children with JIA from Nordic countries, more than 80% of patients with sJIA had inactive disease and 85% were off medications at 18 years of follow up [17]. In Canadian children with sJIA, 70% of children were in remission off medications after an average of 6 years of follow up [18]. Of note, sJIA phenotypes, response to therapy, and disease course may possibly vary in different regions of the world.

When interpreting the long-term data, it is important to consider the potential clinical courses that patients with sJIA might follow. Many patients will have a monophasic course while others will follow a polyphasic course characterized by periods of disease remission and disease flares which may vary in severity, or a chronic course which does not remit [14, 19–21]. It is possible that a significant proportion of patients lost to follow up were patients with monophasic disease who self-elected to no longer follow up due to remission off medications, which is a common scenario in pediatric rheumatology. Only patients who were discharged from care by the treating physician were included in the “disease remission” category. This may have resulted in a falsely low proportion of patients achieving CID at 2 and 3 years in our study, as these patients were not included in the numerator or the denominator. Patients with polyphasic or relapsing chronic disease may be a reason that we saw that the cumulative incidence of achieving CID was higher than the proportion of patients in CID in the specified time points for the cross-sectional analysis. It is possible that many of the subjects who were not in a state of CID at 24 and 36 months had been in CID, but experienced a flare of their disease at that visit. This highlights the critical need to better understand the factors that lead to disease flares once patients achieve CID. This study was not designed to evaluate flares, much less the etiology of flares; thus, we cannot identify if flares were related to potentially modifiable factors such as intentional medication withdrawal or non-adherence or from non-modifiable factors such as incidental infection.

Adverse events reported during the long term follow up of the FROST study occurred at a rate similar to or less than what has been reported in prior clinical trials of biologics in systemic JIA [22–25]. No patients were diagnosed with sJIA-LD during follow up. However, lung disease screening was not systematic or protocolized. Additionally, there may have been selection bias during enrollment in FROST, as patients who could not be safely treated with a CTP were not eligible. This may have excluded patients with significant MAS at diagnosis, a factor that has been reported to be associated with increased risk of developing lung disease [4, 5].

The FROST study had several limitations that have been previously discussed, including lack of randomization, missing data, and challenges with enrollment and follow up [6]. The exclusion of patients with MAS at diagnosis may limit the generalizability of the results. In the long-term follow-up cohort a significant percentage of patients were lost to follow prior to the 2 and 3 year outcome assessment or had incomplete data at those visits (43.8% at 24 months and 56.2% at 36 months). This not only limits our ability to draw conclusions on clinical outcomes but may lead to underestimation of the rate of adverse events. In a conservative analysis assuming that all patients with incomplete data or missing visits had active disease, the confirmed remission rate was 33% (24/73) at 24 months and 29% (21/73) at 36 months. Although we cannot say for certain, we suspect some patients were lost to follow-up due to inactive disease and self-discontinuation of care. If these patients had been included in the analysis, the proportion achieving the primary endpoint would have been higher.

Recent European Alliance of Associations for Rheumatology/Paediatric Rheumatology European Society recommendations propose unifying systemic juvenile idiopathic arthritis (sJIA) and adult-onset Still’s disease (AOSD) under the single term “Still’s disease,” reflecting increasing recognition that these conditions exist along a shared disease continuum [26]. The newly proposed classification criteria for Still’s disease do not require the presence of arthritis, in contrast to the International League of Associations for Rheumatology criteria. Because the FROST cohort was defined using ILAR criteria, all enrolled patients were required to have arthritis in at least one joint. As a result, the generalizability of our findings to patients with sJIA who do not have arthritis may be limited. These patients may represent a distinct phenotypic subset with a different clinical course and treatment response. Alternatively, they may reflect an earlier stage of disease before the development of overt arthritis, which could also influence therapeutic response.

In the time since the FROST study was completed, new guidelines put forth by the American College of Rheumatology included recommendations for the initial treatment of patients with treatment-naïve sJIA with and without MAS. For sJIA without MAS the guidelines included a conditional recommendation against oral glucocorticoids as initial monotherapy and strong recommendation against conventional synthetic DMARDS (methotrexate or calcineurin inhibitors) as initial monotherapy. Conversely, biologic DMARDs including anakinra, canakinumab, and tocilizumab were conditionally recommended as initial monotherapy [27]. Similar recommendations for treatment of Still’s disease were published by the European League Against Rheumatism/Pediatric Rheumatology European Society (EULAR/PReS) in 2024 [28]. These recommendations included strong agreement regarding initiation of an IL-1 or an IL-6 inhibitor as early as possible when the diagnosis is established. This approach is supported by studies that showed increased remission rates for patients starting biologic treatment early in the course of the disease. The results of FROST support the recommendations that biologics as first line treatment leads to good outcomes for the majority of patients with sJIA.

Conclusions

Patients with sJIA enrolled in FROST continued to improve during the 2nd and 3rd year of follow-up, with the majority of observable patients achieving CID and discontinuing GCs. SAEs were not common and were predominantly due to disease flare or MAS. The FROST cohort will continue to be followed in the CARRA Registry, providing further longer-term information about the contemporary course of sJIA patients.

Author contributions

TH, GT, YK, VD, and TB were involved in the conception and design of the study, analysis, and manuscript preparation. All listed authors reviewed and approved the manuscript.

Funding

Funding for this project was provided to CARRA, Inc. in part by Genentech, a member of the Roche Group.

Data availability

The data that support the findings of this study are available from CARRA but restrictions apply to the availability of these data, which were used under a data use agreement for the current study, and so are not publicly available. Data are, however, available from CARRA upon reasonable request (carragroup.org).

Declarations

Ethics approval and consent to participate

The CARRA Registry and the FROST study were approved by the Duke IRB. Participating sites had IRB approval from their local IRB. All subjects consented to participate in FROST and in The Registry.

Consent for publication

All subjects enrolled in FROST consented to allowing their data to be published.

Competing interests

TB receives consulting fees from UCB. All other authors have no competing interests.