Disease flares with baricitinib dose reductions and development of flare criteria in patients with CANDLE/PRAAS

Objectives Patients with chronic atypical neutrophilic dermatosis with lipodystrophy and elevated temperature/proteasome-associated autoinflammatory syndrome (CANDLE/PRAAS) respond to the janus kinase inhibitor 1/2 inhibition with baricitinib at exposures higher than in rheumatoid arthritis. Baricitinib dose reductions to minimise exposure triggered disease flares which we used to develop ‘flare criteria’. Methods Of 10 patients with CANDLE/PRAAS treated with baricitinib in an open-label expanded-access programme, baricitinib doses were reduced 14 times in 9 patients between April 2014 and December 2019. Retrospective data analysis of daily diary scores and laboratory markers collected before and after the dose reductions were used to develop ‘clinical’ and ‘subclinical’ flare criteria. Disease flare rates were compared among patients with <25% and >25% dose reductions and during study visits when patients received recommended ‘optimized’ baricitinib doses (high-dose visits) versus lower than recommended baricitinib doses (low-dose visits) using two-sided χ2 tests. Results In the 9/10 patients with CANDLE with dose reduction, 7/14 (50%) times the dose was reduced resulted in a disease flare. All four dose reductions of >25% triggered a disease flare (p <0.05). Assessment of clinical and laboratory changes during disease flares allowed the development of disease flare criteria that were assessed during visits when patients received high or low doses of baricitinib. Disease flare criteria were reached during 43.14% of low-dose visits compared with 12.75% of high-dose visits (p <0.0001). Addition of an interferon score as an additional flare criterion increased the sensitivity to detect disease flares. Conclusion We observed disease flares and rebound inflammation with baricitinib dose reductions and proposed flare criteria that can assist in monitoring disease activity and in designing clinical studies in CANDLE/PRAAS.


c. Assessment of the CANDLE/PRAAS Disease Flare Criteria in visits stratified into high dose and low dose visits
Six patients (P1, P3, P4, P5, P6, P10) had several visits on low and high baricitinib doses and were included in assessment of the CANDLE/PRAAS flare criteria during visits on high doses and low doses of baricitinib.Patients 2 and 9 were excluded as they did not have a low dose period.Patient 8 was excluded since patient did not have a dose reduction during the study period.Patient 7 was excluded (he was included to development of flare criteria) since patient developed azotemia secondary to presumed BK nephropathy and baricitinib was subsequently discontinued. 1

DAILY DIARY SCORE (DDS) ASSESSMENT
Disease-specific patient daily diary for CANDLE/PRAAS were collected prospectively during JAGA program. 1 Patients with CANDLE/PRAAS or their parents recorded daily symptoms of fever, rash, musculoskeletal pain, headaches, and fatigue.Each symptom was rated on a scale of 0 to 4, with 0=no symptoms, 1=mild symptoms, 2=moderate symptoms, 3=more severe symptoms, and 4=severe symptoms (possible range 0-20).At each visit, the diary score was calculated as follows: a.Average score of each symptom was calculated using data entered since the previous visit and correcting for any day for which diary scores were not recorded.b.The calculated average score for each symptom was summed up and divided by the number of assessed symptoms to calculate the average score for each patient.
Retrospectively reviewed and analyzed DDS data for assessment of baricitinib dose reduction associated clinical flares.Mean DDS was calculated for the period of seven days including the period of three days before dose reduction, day of dose reduction and three days after dose reduction.
If the mean DDS of the reference visit was zero and a patient developed any symptoms during a subsequent visit, it was considered to be a significant change.Therefore, a mean DDS greater than zero is considered indicative of a clinical flare for these patients.
The percent (%) changes in DDS and the laboratory biomarkers were compared between the last visit before dose reduction and the first visit after dose reduction (supplementary table 6a).

DATA EXTRACTION a. Development of CANDLE/PRAAS Clinical and Subclinical Disease Flare Criteria
For all visits with dose reductions, clinical and laboratory data were extracted from the last visit prior to baricitinib dose reduction (reference visit), and the first follow-up visit after the dose was reduced (flare visit).Data included DDS, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), white blood cell count (WBC), hemoglobin (HGB), absolute lymphocyte count (ALC), platelets (PLT), and 25-gene IFN score 2 (when available).Physician notes were reviewed for documentation of clinical symptoms and physical exam findings.Absolute and percent changes were calculated (supplementary table 6a).
b. Assessment of the CANDLE/PRAAS Disease Flare Criteria in visits stratified into high dose and low dose visits The visit prior to dose reduction was used as a reference visit to calculate post dose reduction changes.For the assessment of the established disease criteria in high-dose and low-dose visits, we determined a reference visit for each patient separately when the patient was on optimal/clinically effective baricitinib dose and clinically stable considering his/her own disease course, based on expert rheumatologist judgement.For the reference visit, all patients were clinically stable or fulfilled remission criteria that was published by Sanchez et al. 1 Three patients (P4, P5 and P10) were in remission and off glucocorticoid (GC)s.Three patients (P1, P3 and P6) had more than 50% reduction in their daily GC dose from baseline.Patient 7 had reduction in his GC dose from baseline (1.2 mg/kg/day at first JAGA visit) however he was still on 0.82 mg/kg/day GCs at reference visit since he had azotemia and his baricitinib dose was reduced.Laboratory changes after dose reductions that resulted in a clinical flare and laboratory changes after dose reductions that did not result in a clinical flare were systematically evaluated.
For the criteria, DDS increases were used to define clinical symptoms.In a sub-analysis, we assessed the value of adding the IFN score as a laboratory biomarker to increase the sensitivity of disease flare detection and computed the number of additional clinical and subclinical flares that were identified.

Definition and clarification of data extraction:
The number of baricitinib dose reductions, reasons for the dose reductions, the amount of dose reduction, and the first date when the patient took the lower dose were documented.Data were extracted in an excel spreadsheet.
Assessment of daily diary score changes: Clinical notes and daily diaries were retrospectively reviewed for the periods baricitinib dose reductions occurred.Laboratory biomarkers of inflammation and clinical symptoms recorded on a daily diary and/or clinical notes that can constitute a disease flare including fever, rashes, headaches, fatigue, and joint and musculoskeletal pain were extracted.In some instances image of rashes were sent by e-mail.Clinically relevant changes in laboratory biomarkers were included as flare criteria and confirmed previous clinical observations reported in patients with interferonopathies during active disease. 3 4finition of abnormal biomarker cut off values: CRP was considered a clinically abnormal value if 5mg/L or greater, and ESR was considered a clinically abnormal value if 20 mm/hr or greater.CRP either remained within normal limits or increased to above 5mg/L.Increasing CRP values were a component of the flare criteria.When CRP increased to higher than 5mg/L, the % change was computed.In addition to the change in CRP score resulting in a clinically abnormal value, a cutoff of a 40% increase in CRP was considered a change in cases when the CRP was elevated in the reference visit and a cutoff of a 20% increase in ESR was considered a change in cases when the ESR was elevated in the reference visit.We used the same cut off of 20% and requirement of the change resulting in a clinically abnormal value for IFN score.

Determination of % change in biomarker cut off values:
We assessed the laboratory changes at the reference visits and the visits when the patients developed a disease flare post dose reduction (supplementary table 6a) We calculated the % change (increase or decrease) in each laboratory biomarker for each patient comparing the reference visit with the flare visit.Then we determined the lowest meaningful % change for each laboratory biomarker that was associated with clinical symptoms and was consistent with a clinically meaningful change in biomarker.We used these percentages as cut off when developing the flare criteria.We subsequently applied these criteria to all visits to identify possible flare visits during both the low dose and the high dose periods and assessed the % changes to those visits identified as flare visits by using the criteria.The % changes were similar in the high dose and low dose flare visits in the confirmation phase (supplementary table 6b) to those in the baricitinib withdrawal flare visits, which strengthened the notion that the cut off values selected were meaningful in long-term monitoring settings.

CONFIRMATION OF THE CANDLE/PRAAS DISEASE FLARE CRITERIA
Definition of a visit: Patients had study visits with clinical and laboratory evaluations, every 3-6 months on average at NIH.However, they were also seen by their local providers and had disease monitoring labs performed in between NIH visits.A patient in a disease flare who requires close monitoring by their local provider may require more frequent clinical and laboratory evaluations.To prevent overinterpretation of data obtained from a period where patients were in states of prolonged disease flare, we determined that each calendar month represents a visit if the patient had clinical and/or laboratory evaluation.If a patient had blood work at multiple occasions within the same calendar month, those multiple visits were considered as one study visit and labeled as flare vs no flare visit based on the worst clinical and laboratory findings.e.g., if a patient had blood work 3 times in a given calendar month and 1 out 3 was consistent with disease flare, we considered this visit as a flare visit.

Definition of high-dose and low-dose visits:
To assess the performance of the flare criteria, we assessed the flare criteria during visits when patients received currently recommended dose/clinically effective dose (supplementary table 4), 1 5 and lower than effective dose.Patient visits on lower doses were categorized as "low-dose" visits and those on equal or higher than effective/recommended doses were categorized as "high-dose" visits.In this analysis we only included patients who had both, "low dose" and "high dose" visits (n=6; P1, P3, P4, P5, P6 and P10).We excluded a total of four patients: three patients who had no "low dose" visits (P2, P8 ad P9) and one patient who had azotemia (P7).Patient 7 was excluded since determining the clinically effective baricitinib dose in the setting of azotemia and renal insufficiency may not be accurate.
We identified 51/153 visits when patients received low dose baricitinib and 102/153 visits when patients received high dose baricitinib, over 9.5 and 25.2 patient years respectively.The same reference visit for each patient was used for comparison with each visit in the low dose and each visit in the high dose period.The clinical and subclinical CANDLE/PRAAS flare criteria were used to calculate the flare rate during low-dose and high-dose visits.
Treatment decisions (steroid adjustments/baricitinib adjustments/no action) during "flare visits": To determine whether clinical and subclinical "flare visits" that were identified by systematic application of the flare criteria had treatment actions implemented in at the time of the visit, we extracted the drug changes and the impact on disease activity on subsequent visits.

BK viral load extraction before and after dose reduction:
To assess the impact of the baricitinib dose reduction on viral load in blood and urine, we extracted the viral load at the visit before and after dose reduction and compared whether the load was lower, the same or higher.

IFN score addition as biomarker in a sub analysis:
The 25-gene IFN score 2 was collected as a secondary endpoint in the compassionate use study 1 but was initiated later and was therefore not available for all study visits.Based on control data, a normal IFN score was defined as below 44.2 (cut off is 95%ile in healthy controls). 2 the baricitinib reduction visits, the 25-gene IFN score was only available before and after dose reduction for one patient (P1) out of the seven patients who developed a clinical flare post baricitinib dose reduction.The IFN score rose highly when the baricitinib dose was reduced in patient (P1).The original flare criteria were therefore established without the inclusion of the IFN score due to paucity of data.However, for the validation comparing high-dose and lowdose visits, the 25-gene IFN score was available for 108 out of 153 visits, which allowed us to assess the performance of the IFN score at high-dose and at low-dose visits.We then assessed the flare criteria by adding the IFN score to determine whether the flare criteria could capture more "flare visits".The addition of the IFN score to the flare criteria was assessed by determining how many additional flares were identified.
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) Cetin Gedik K D, dose; NA, not available; P, patient * P1: Baricitinib dose was reduced from 9 mg/day to 8 mg/day and then to 7 mg/day four days later, P3: Baricitinib dose was reduced from 10 mg/day to 8 mg/day and then to 6 mg/day the following day.P3: Baricitinib dose was reduced from 8 mg/day to 7 mg/day and then to 6 mg/day 11 days later.P5: Baricitinib dose was reduced from 11 mg/day to 10 mg/day and then to 9 mg about four weeks later.** Patient 7 had one additional dose reduction from 8 mg/day to 6mg/day by patient's local provider at another time point, due to anemia during a hospitalization for a presumed CANDLE/PRAAS disease flare.Baricitinib dose was increased to 8 mg/day 3 weeks later and no laboratory data is available for this dose reduction.This dose reduction was excluded since a. the patient had been admitted with a presumed CANDLE/PRAAS disease flare prior to baricitinib dose reduction, b. missing data for this time frame.& Patient 7's dose reduction started with dose reduction from 8 mg/day to 6 mg/day.Then rapid dose reduction and discontinuation occurred over 3 months in the context of renal failure.Flare occurred with holding baricitinib for one day although it was restarted at a dose of 2.7 mg/day the following day.It was discontinued permanently one week later.After discontinuation of baricitinib, he continued to have active disease that was controlled with high doses of glucocorticoids.He had a major flare presenting as macrophage activation syndrome (MAS).One week after discontinuation of baricitinib, the patient was admitted for persistent fevers, tachycardia, abdominal distention and fluctuating increased work of breathing with increased oxygen requirement.# D13 occurred on study day 521.P10's baricitinib dose was increased from 10 mg/day back to 12 mg/day eventually.On study day 899, P10 had another dose reduction (D14) from 12 mg/day to 11 mg/day.^In addition, patient 10 had mistakenly taken extra dosing (24 mg/day instead of 12 mg/day) for 45 days since he mixed up 4mg vs 1 mg tablets Baricitinib dose reductions by 1-2 mg/day resulted in lower BK viral load in the blood and urine.BK viremia became negative in P1 and decreased in P3 and P10.With the development of BK nephropathy in one patient (P7), we recommend monitoring BK viremia and suggest keeping BK viral load in blood as low as possible, but all times below log 4 copy/ml (10,000 copies/ml), consistent with recommendations made for kidney transplant recipients. 6 ** Three dose reductions (P3/D#4, P3/D#5, P9/D#12) resulted in changes in DDS and they had labs drawn 2 weeks, 3 months, and 2.5 weeks after recording of the symptoms in DDS respectively.At that time there were no laboratory changes observed.We did not consider these DDS as disease flares.P3: We observed mild and intermittent rash and MSK symptoms at pre dose reduction visits as well; these symptoms were considered baseline fluctuations.P9: Patient developed fatigue only and DDS change was secondary to fatigue.In the absence of additional CANDLE/PRAAS findings such as fevers, rashes, MSK symptoms, headaches, it was considered to be insufficient to call this as a disease flare.Patient was asymptomatic otherwise.^P4 received lower dose of baricitinib for three days.Patient had clinically significant laboratory changes post dose reduction with no clinical symptoms.P7's baricitinib was discontinued secondary to azotemia and was on high dose steroids.P7 had clinically significant laboratory changes post dose reduction with no clinical symptoms (both patients fulfilled subclinical flare criteria, please see supplementary table 6a).D, dose reduction; DDS, daily diary score; NA, not available; P, patient.

Supplementary Table 3
Summary of laboratory biomarkers, mean DDS, glucocorticoid data and clinical status at reference visits for patients included in the confirmation of the flare criteria * P7 was not included in the confirmation phase as he was considered to have active disease and remained on high doses of GC. ** Retrospectively reviewed and analyzed DDS data for assessment of baricitinib dose reduction associated clinical flares.Mean DDS was calculated for the period of seven days including the period of three days before dose reduction, day of dose reduction and three days after dose reduction.If the mean DDS of the reference visit was zero and a patient developed any symptoms during a subsequent visit, it was considered to be a significant change.Therefore, a mean DDS greater than zero is considered indicative of a clinical flare for these patients.^Low hemoglobin was drug related (baricitinib induced anemia) ALC, absolute lymphocyte count; CRP, C-reactive protein; DDS, daily diary score; ESR, erythrocyte sedimentation rate; GC, glucocorticoid; HGB, hemoglobin; IFN, interferon; P, patient; PLT, platelets; WBC, white blood cell.

Cetin Gedik K
Pre-dose reduction visit: Last visit occurred prior to dose reduction.This visit for each patient who developed post dose reduction disease flares was used as a reference visit for evaluation of the flare criteria.Lowest value used to estimate a cutoff value is highlighted in blue.Values that are above the cut off are highlighted in orange.Flare visit: First visit occurred after baricitinib dose reduction.* see supplementary table 1 for dose reductions ** within normal limits before and during flare ^Patient fulfilled CANDLE/PRAAS disease flare criteria when including IFN score criterion to the flare criteria.Otherwise, patient was unable to meet required laboratory abnormalities to fulfill the flare criteria although patient developed clinical symptoms in association with baricitinib dose reduction.^^IFN score not measured for this visit.Used the mean of IFN scores from the visits before and after the reference visit.% Patient had clinically significant laboratory abnormalities post dose reduction, with no clinical symptoms.Therefore, patient fulfilled subclinical flare criteria.ALC, absolute lymphocyte count; CANDLE/PRAAS, Chronic atypical neutrophilic dermatosis with lipodystrophy and elevated temperature/proteasome-associated autoinflammatory syndrome; CRP, C-reactive protein; D, dose reduction; DDS, daily diary score; ESR, erythrocyte sedimentation rate; HGB, hemoglobin; IFN, interferon; NA, not available; P, patient; PLT, platelets; WBC, white blood cell; WNL, within normal limits.

Actual clinical scenarios observed:
Scenario 1: In patient P4 who had subclinical flares only and in patients P5 and P10 who had clinical flares after fulfilling remission criteria off GC, the baricitinib dose was adjusted after a baricitinib dose reduction during "low-dose visits".The flare resolved in P4 and P5 by adjusting the baricitinib dose back to baseline.In P10, two short courses of GCs were required in addition to a baricitinib dose adjustment to achieve remission again.
However, flares identified in the validation period during the "high dose visits" did not result in GC or baricitinib changes.On subsequent visits P4, P5 and P10 fulfilled flare criteria on 9%, 18% and 18% of visits respectively.Scenario 2: Post flare P1 (S, stable disease) remained on baricitinib 7 mg/day and GC doses between 0.23-0.30mg/kg/day.Over the next ~2 years he had 3 clinical flare visits out of 12 lowdose visits (25% flare visits) and was unable to wean GC.The baricitinib dose was adjusted to 8 mg/day and GCs could be tapered to 0.14 mg/kg/day, however he had 2 clinical and 1 subclinical flare visits out of 8 high-dose visits (38% flare visits) over 2 years.Eventually the baricitinib dose was increased to 9 mg/day, and he had 1 clinical flare visit out of 4 high-dose visits (25% flare visits) and remained on GCs 0.14 mg/kg/day.
In patients P3 (MDA) and P6 (S, stable disease) the baricitinib dose reduction without subsequent dose adjustment resulted in 38% and 57% of subsequent visits fulfilling "flare criteria" respectively during a period of ~2.5 years.In P3 the baricitinib dose was eventually increased to 8 mg/day which allowed a GC taper to doses below 0.15mg/kg/day (criteria for MDA) with no subsequent clinical subclinical flares (flare rate 0%).In P6 the family elected to stay on a lower baricitinib dose and not adjust baricitinib or GC dose.

Suggestion:
The proposed flare criteria can be used to manage patients with CANDLE/PRAAS in remission or with MDA.The scenarios above illustrate their use in finetuning treatment and adjusting steroid doses to the lowest dose possible in patients with MDA.In patients who achieved remission, the "flare rates" may help in quantifying disease control long-term and to better characterize the "level of disease control" that can be achieved on treatment with janus kinase inhibitors.

Proposed use of criteria for monitoring CANDLE/PRAAS patients who are in clinical remission (P4, P5, P10) or have minimal disease activity (P3), clinical scenarios:
1. Baricitinib dose reductions to determine the lowest dose tolerated with the goal to: a. Keep BK viral load in blood as low as possible but below log 4 copy/ml (10,000 copies/ml) at all times, consistent with recommendations for BK viral load monitoring for kidney transplant recipients. 6

II. Supplementary Tables Supplementary Table 1 Baricitinib dose reductions and effects on BK viremia/viruria Patient # Dose Reduction Study day Pre-and post- reduction baricitinib dose Dose Reduction by mg (%) Reason for Dose Reduction Effect of the baricitinib dose reduction on BK viremia/viruria
BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s)Ann Rheum Dis doi: 10.1136/ard-2023-225463 -1188.

Table 2
Summary of post-baricitinib dose reduction clinical symptoms and outcomes of clinical flare vs no clinical flare Pre-dose reduction visit (=reference visit): Last visit occurred prior to dose reduction.Flare visit: First visit occurred after baricitinib dose reduction.
* see supplementary table 1 for dose reductions

Table 4
Summary of baricitinib dosing regimen at the time of baricitinib dose reductions

Table 5
Baseline demographics and clinical characteristicsSupplementary Table 6aAbsolute values and percent changes of DDS and laboratory biomarker levels comparing the reference visit with the flare visit

Table 7
Supplementary table 6bRanges of cut-off values of clinical and subclinical flare visits identified during the high and low dose visit phases are in range with the changes seen in the acute baricitinib withdrawals.The ranges of percent changes of components of the flare criteria were extracted for each patient are summarized in the table.The lowest value that is making the cut off is highlighted in blue.All other values fulfilling the flare criterion are highlighted in orange.The lowest value that is making the cut off is highlighted in blue.All other values fulfilling the flare criterion are highlighted in orange.* DDS change not applicable as patients had <15% during subclinical flare visits by definition ALC, absolute lymphocyte count; CRP, C-reactive protein; D, dose reduction; DDS, daily diary score; ESR, erythrocyte sedimentation rate; HGB, hemoglobin; IFN, interferon; Max, maximum; Min, minimum; P, patient; PLT, platelets; WBC, white blood cell; WNL, within normal limits Supplementary Treatment decisions (steroid adjustments/baricitinib adjustments/no action) during "flare visits" * increased *

Flare criteria fulfilled → no baricitinib and no GC change →ongoing disease activity
* in column presumed cause of flare, we added the reason for the presumed flare i.e glucocorticoid taper, baricitinib dose reduction, other.We also determined the level of disease control the patient had prior to the flare at the reference visit (RV): remission (REM) meaning DDS<0.15,normalCRP and off steroids (P4, P5, P10) , minimal disease activity (MDA), DDS<0.15,normalCRP, prednisone equivalent <0.15mg/kg/day (P3) and stable for patient who normalized CRP but had still elevated DDS and were on higher doses of steroids (P1, P6, P7).See supplementary table 7. P7 was on high doses of steroids 0.8 mg/kg/day and could not be tapered.P7 was not included in the validation of the criteria.***all baricitinib dose increases were within the dose range of the provided dosing table (Kim at al. 2018) GC, glucocorticoid; Ind, indication; MDA: minimal disease activity; NA, not available; P, patient; REM, remission; RV, reference visit.BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) * all visits that were identified by fulfilling the proposed flare criteria (clinical and subclinical flares) in the validation phase are listed here.* 7 P3 (MDA)) b.Manage side effects such as anemia (P3) (MDA).c.Find lowest tolerated dose in patients in remission or with minimal disease activity BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) (P3 (MDA) P4 (REM), P5 (REM), P10 (REM))