key: cord-0302715-9t5brqma authors: Damsky, W.; Wang, A.; Young, B. D.; Ayasun, R.; Ryu, C.; McGeary, M. K.; Fazzone-Chettiar, R.; Pucar, D.; Gulati, M.; Miller, E. J.; Bosenberg, M.; Flavell, R.; King, B. title: Treatment of sarcoidosis with cutaneous involvement with tofacitinib date: 2021-07-05 journal: nan DOI: 10.1101/2021.07.01.21259700 sha: 95e8e4c0a7f3ea563d33e11471ca8d034d3b955c doc_id: 302715 cord_uid: 9t5brqma Sarcoidosis is an idiopathic inflammatory disorder that is commonly treated with glucocorticoids and there are no approved steroid-sparing medications. There is emerging evidence that Janus kinase (JAK) inhibitors, which inhibit JAK-dependent cytokine activity, may hold promise in sarcoidosis. In this open-label trial, 10 patients with recalcitrant sarcoidosis with cutaneous involvement were treated with tofacitinib 5 mg twice daily. There was no washout period and patients were permitted to continue, taper, or discontinue other treatments. The primary outcome was the change in the Cutaneous Sarcoidosis Activity and Morphology Instrument (CSAMI) activity score after 6 months. Change in internal organ disease activity was also assessed using total lesion glycolysis (TLG) determined by full-body positron emission tomography. A mean reduction in the CSAMI activity score of 82.7% was observed, with 6 patients showing a complete response. Internal organ response data was available in 8 patients; a decrease in TLG of [≥]50% was noted in 5 patients, with complete or near complete resolution in 3 (>98% reduction in TLG). Patients were generally able to significantly taper or discontinue their baseline immunosuppressive regimen, which included prednisone in 5 patients. Single cell RNA-sequencing, bulk RNA-sequencing, and high-throughput proteomic analyses were performed on skin and blood as a function of treatment in order to delineate changes in immunologic signals with therapy. We identified CD4+ T cell derived IFN-gamma as a central cytokine driver of sarcoidosis and inhibition of its activity was achieved with tofacitinib and correlated closely with clinical improvement. Tofacitinib appears to have impressive activity in treatment of sarcoidosis and likely acts by inhibiting IFN-gamma, larger, controlled studies are warranted. Sarcoidosis is an inflammatory disorder that most commonly affects the lungs; however, any organ including the skin can be involved. Glucocorticoid-based regimens (prednisone and corticotropin gel) are the only approved therapies. Prednisone remains the recommended first-line treatment for both pulmonary and extra-thoracic sarcoidosis 1 . However, this is not ideal as chronic therapy is often required and glucocorticoid-associated toxicities are common 2 . Methotrexate is commonly used as a steroid-sparing agent, but it is often inadequate 1 . TNF-α inhibitors have also been evaluated; however, the benefit in controlled trials has been marginal, with several studies finding no benefit [3] [4] [5] [6] [7] [8] . TNF-α inhibitors can also induce or exacerbate sarcoidosis 9 . A hallmark of sarcoidosis is the non-caseating granuloma, which is observed microscopically in affected tissues. Although the pathogenesis of granuloma formation in sarcoidosis is complex, it appears to involve coordinated activity of several cytokines, chemokines, and other signals 10 . Of these, many cytokines including IFN-γ, Type I IFNs (IFN-α/β), IL-2, IL-4, IL-6, IL-12, IL-13, IL-23 and GM-CSF signal via the JAK -signal transducer and activator of transcription (STAT) pathway. Indeed, JAK-STAT pathway activation has been reported in tissues and blood of patients with sarcoidosis [11] [12] [13] [14] [15] . JAK inhibitors are oral small molecules that attenuate the activity of JAK-STAT-dependent cytokines. We and others have recently described effective treatment of individual patients with sarcoidosis using tofacitinib (JAK1/2/3 inhibitor) or ruxolitinib (JAK1/2 inhibitor) 14-20 . patients had a compete response (defined as CSAMI activity score of 0). Postinflammatory hyperpigmentation and/or scarring often persisted (e.g. Figure 1F ). A skin biopsy was performed before and after 6 months of treatment in 6 patients; of which 3 showed a complete response and 3 showed partial improvement. In patients with a complete clinical response, sarcoidal granulomas were no longer evident in skin. In those with a partial clinical response, biopsy of active areas showed persistence of granulomatous inflammation ( Figure 1G ). Of the 10 patients, 8 had pulmonary involvement, and 1 had myocardial involvement (Table 1) . Whole body PET-CT was performed in these 9 patients and total lesion glycolysis (TLG) was determined to assess sarcoidosis activity. Baseline studies were performed while taking a stable preceding immunosuppressive regimen that included prednisone plus methotrexate in 3 patients, and prednisone or methotrexate monotherapy in 2 and 1 patients, respectively ( Figure 1A) . A subsequent PET-CT after 6 months of tofacitinib was obtained. One follow-up study could not be interpreted due to nonadherence to the dietary preparation. A decrease in TLG of ≥ 50% was noted in 5 patients, with complete or near complete resolution in 3 (>98% reduction in TLG) (Figure 2A) . One patient had essentially stable TLG, but did appear to have improvement in mild lymph node avidity below the detection threshold ( Figure 2B) . Two patients had an increase in TLG, however; both were able to discontinue methotrexate and prednisone during the study and the change was not clearly clinically significant. Further, both experienced improvement in their skin (e.g. Figure 2C ). Examples of complete or near complete resolution in pulmonary ( Figure 2D -E) and myocardial disease activity ( Figure 2F ) are illustrated. Of 5 patients taking prednisone at baseline, three were able to discontinue it entirely and one was able substantially decrease the dose. One patient with preexisting, Achilles tendinopathy responsive to prednisone increased his prednisone dose during the study for this. All patients taking methotrexate were able to discontinue it. Several patients reported subjective improvement in confirmed or suspected nasal sinus involvement. A patient with a hoarse voice due to laryngeal involvement noted normalization of her voice. In general, the skin tended to improve to a greater degree than internal organ involvement. There was weak, non-significant correlation between the degree of improvement in cutaneous and internal organ involvement ( Figure 2G ). Definitive interpretation of this trend was limited by varying baseline treatment regimens. In all 10 patients, the treatment regimen including tofacitinib led to disease improvement not achieved with the prior regimen. All 10 patients reported improved skin-related quality of life ( Figure S1 ). Tofacitinib was well tolerated, there were no significant or dose-limiting adverse events (Table S1) . In order to better understand the potential molecular targets of tofacitinib in sarcoidosis, single cell RNA sequencing (scRNA-seq) was performed on lesional skin All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint from 3 of the patients prior to treatment and compared with normal skin from 3 healthy controls (Table S2) . 24034 cells were analyzed with a median of 4029 unique molecular identifiers. The data were visualized using uniform manifold approximation and projection (UMAP), revealing 37 clusters, corresponding to 11 major cell types ( Figure S2 ). Although transcriptional variation was noted in several cell types, when comparing sarcoidosis to normal, we focused our analysis on T cells and myeloid cells given the important role that these cell types play in granuloma formation 24 Differential gene expression and pathway (IPA) analysis were used to compare predominant T h populations in sarcoidosis (clusters 2, 7, 12) to controls (clusters 0, 5). revealing a unique activated T h 1 cluster in each sarcoidosis patient (T h 1-SAR) that was not present in controls ( Figures 3B, S3) . IFNG (IFN-γ) was by far the most differentially upregulated cytokine in T h 1-SAR, which expressed high levels of canonical T h 1 transcription factors (TBX21, STAT1). Increased expression of CSF2 (GM-CSF) was also present. TNF (TNF-α) was expressed at modest levels and there was minimal IL-17 expression ( Figure 3C, S3) . Chemokines CCL3, CCL4, and CCL5 were also highly upregulated ( Figure S3 ). Predicted upstream regulators of T h 1-SAR included IL-6, IL-12, and IL-15 ( Figure S3 ). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint We next analyzed myeloid clusters. A total of 1748 myeloid cells were analyzed, revealing 14 clusters which included both macrophages and dendritic cells (DCs) ( Figure 3D-E, S4 ). Macrophages were abundant and somewhat diverse in sarcoidosis ( Figure 3E ). Expression patterns among predominant macrophage populations in sarcoidosis (Mac-SAR, clusters 0, 1, 4, 6, 9, 10) were compared to control clusters phenotype that produced IL-12 were present almost exclusively in the sarcoidosis samples ( Figure S4 ). IFN-γ was the most highly significant inferred upstream cytokine regulator of Mac-SAR based on their transcriptional profile ( Figure S4 ). Together, these data strongly implicated IFN-γ derived from an aberrant T h 1 response as a key driver of sarcoidosis pathology, consistent with the conserved role of IFN-γ in classical macrophage activation and granuloma formation 25, 26 . Through these and other analyses (below) we identified other cytokine signals including GM-CSF, IL-15, IL-6, TNF-α, which we hypothesize play a secondary/reinforcing role ( Figure 3G ). Proposed cytokine and chemokine interactions among T cells and macrophages were further supported by a bioinformatic analysis of receptor-ligand interactions (Cellphone DB) 27 ( Figure 3H) . Interestingly, IL-15 and IL-6 were produced predominantly by fibroblasts and endothelial cells ( Figure S5 ). Of these signals IFN-γ (JAK1/2), GM-CSF (JAK2), IL-6 (JAK1/2), and IL-15 (JAK1/3) are expected to be inhibited by tofacitinib. In 6 patients for whom matched skin biopsy specimens were available (Table S3) , bulk RNA sequencing was performed on lesional skin before and after tofacitinib. This data was compared to skin from 6 healthy controls and further contextualized using an independent gene expression data set consisting of 15 cutaneous sarcoidosis biopsies and 5 controls (Judson et al) 28 . IPA was performed to compare sarcoidosis to controls in our data and the Judson data and showed that an IFN-γ transcriptional signature was the most significant change in both data sets. IL-15, IL-6, GM-CSF, and TNF-α were also among the most significant predicted regulators ( Figure 3I ). Analysis of pre-and during-treatment skin biopsies from trial patients showed that in those achieving complete response, there was near-complete suppression of T h 1-SAR and Mac-SAR genes (from scRNA-seq experiments) and an IFN-γ response signature ( Figure 3J) . Whereas, in patients with partial improvement, T h 1-SAR, Mac-SAR genes and the IFN-γ response signature were still detected. GM-CSF, IL-15, IL-6, and TNF-α response signatures also correlated somewhat with response patterns, albeit not as closely as IFN-γ, as assessed using principal component analysis ( Figures 3J, S6) . Overall, the bulk expression data was highly consistent with the scRNAseq data and further implicated IFN-γ as a key driver of sarcoidosis. Tofacitinib reduces cytokine, chemokine, and macrophage activation marker levels in plasma All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint We used high throughput proteomics to study plasma from 9 patients in whom matched samples were available before and after 6 months of tofacitinib and compared these with 11 healthy controls (Table S4) . Of the 1536 proteins analyzed with this approach, we found IFN-γ, IL-6, and TNF-α to be among the most upregulated in sarcoidosis at baseline relative to controls ( Figure 4A ). IL-15 was more modestly increased (GM-CSF was not part of this panel). Further, we found a significant correlation between levels of proteins in plasma and mRNA expression in cutaneous sarcoidosis, suggesting upregulation of several of these markers in plasma was a direct reflection of disease activity ( Figure 4B ). Next, we overlaid 1) proteins upregulated in sarcoidosis plasma at baseline versus controls with 2) proteins modulated by tofacitinib therapy in sarcoidosis plasma, and in doing so found a 14-protein signature or sarcoidosis that we termed, sarcoidosis plasma signature (SPS). The SPS included IFN-γ, IFN-γ targets (CXCL9, CXCL10, CXCL11), macrophage activation proteins (CHIT1, CHI3L1, and FBP1) and TNF-α ( Figure 4C ). We then evaluated whether differences in SPS profiles correlated with response to tofacitinib. To do so, we compared SPS levels in the 4 best responders (complete/near complete response) to the other 5 patients (partial improvement) ( Figure 4D ). We found that the best responders tended to have lower baseline SPS activity and tofacitinib resulted in relative normalization of SPS to levels comparable to healthy controls. In contrast, the other patients tended to have higher baseline SPS activity (or were heavily immunosuppressed when the baseline samples were collected, e.g. Pts 4,5) with SPS levels tending to remain elevated above those seen in healthy controls even with tofacitinib ( Figure 4D ). Given that relatively higher baseline SPS levels in plasma and persistent IFN-γ activity in skin were associated with partial improvement on tofacitinib, we hypothesized that in some patients, a higher dose of tofacitinib might be required. In order to explore this further, a single patient with partial improvement in cutaneous, pulmonary, and lymph node sarcoidosis after 6 months of tofacitinib 5 mg twice daily during the trial ( Figure 4E -F) was subsequently treated with a higher dose of tofacitinib (10 mg in the morning and 5 mg at night) obtained through insurance after the trial. After an additional 6 months of treatment at the higher dose, there was complete clearance of his skin and after 8 months there was resolution of pulmonary and diffuse lymph node FDG-avidity ( Figure 4E-F) . A relative normalization of the SPS signature was also observed on the higher dose ( Figure 4G ). These data suggest that a higher dose of tofacitinib might be required in patients with higher baseline disease activity. In this open-label trial of tofacitinib, we demonstrate efficacy in 10 sarcoidosis patients with cutaneous involvement. In all 10 patients, disease control with a tofacitinibbased regimen was superior to the patients preceding immunotherapeutic regimen, particularly for skin involvement. Four of five patients entering the study taking prednisone were able to discontinue or significantly reduce the dose. Our mechanistic evaluation suggested that IFN-γ is a key driver of sarcoidosis and is a critical cytokine targeted by tofacitinib with effective treatment. This observation All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint is consistent with prior work showing that IFN-γ is elevated in circulation and in tissues of sarcoidosis patients and correlates with disease activity 29-32 , and makes teleological sense given the fundamental role of IFN-γ in classical macrophage activation, granuloma formation and protection against Mycobacterium tuberculosis 25,26 . Tofacitinib appears to provide an effective means of suppressing IFN-γ, which signals via JAK1/2, in sarcoidosis. We also found that the activity of other cytokines including GM-CSF (JAK2), IL-15 (JAK1/3), IL-6 (JAK1/2), and TNF-α (JAK-independent) are also evident in sarcoidosis. GM-CSF has been shown to promote the differentiation of monocytes into inflammatory macrophages in autoimmunity 33 , IL-15 can re-enforce CD4+ T cell effector responses 34 , and IL-6 is an additional proinflammatory cytokine implicated as a potential treatment in sarcoidosis 35 . Although we do not exclude a role for TNF-α in sarcoidosis, we do implicate these additional cytokines, which may also need to be inhibited for maximal response to therapy. A potential advantage of JAK inhibition (compared to TNF-α inhibition) is the simultaneous, direct inhibition of multiple cytokines. Furthermore, we observe a reduction in TNF-α production and activity with tofacitinib, suggesting that TNF-α production may occur, at least in part, downstream of these JAK-STAT dependent cytokines. While all patients improved, some patients still had disease activity on tofacitinib. This was associated with incomplete suppression of IFN-γ activity, suggesting that higher doses might be required in some patients. Indeed, in a single patient, a higher dose of tofacitinib led to remission of cutaneous and internal organ sarcoidosis. This is consistent with a prior case report, showing a better response of widespread All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; sarcoidosis to tofacitinib 10 mg twice daily than to 5mg twice daily 16 . Moving forward, improved suppression of IFN-γ activity could be achieved through either a higher dose of tofacitinib in some patients, or, potentially evaluation of more targeted JAK inhibitors, such as a JAK1-or JAK1/2-specific inhibitor. Together, these data are promising and support further evaluation of JAK inhibition in the treatment of sarcoidosis. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. stained and CD68 immunohistochemistry on skin biopsies from a representative complete responder (left panels) and partial responder (right panels). **Some potentially identifying patient information has been removed. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint interval (shared area) Cutaneous involvement shown as percent reduction in CSAMI during the study period. Extracutaneous involvement shown as percent reduction in TLG during the study period; for patients with increase in TLG during the study, worsening of 50% was arbitrarily assigned. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint evaluation and specimen acquisition......................................................... Figure S2 . Summary of all cells analyzed using scRNA-seq in sarcoidosis and control skin. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Potentially eligible patients with sarcoidosis were identified in Yale Medicine clinics (dermatology, pulmonology, and cardiology). Inclusion criteria included: 1) a diagnosis of cutaneous sarcoidosis with supportive skin biopsy, 2) Cutaneous Sarcoidosis Activity and Morphology Instrument (CSAMI)(1) activity score of 10 or greater, and 3) taking a stable dose of oral immunomodulatory regimen for at least 3 months, with no plans to change the regimen over the subsequent 6 months. Patients less than 18 years of age, a history of chronic, untreated infections, and malignancy were excluded. Most patients also had internal organ involvement, but it was not used as a criterion for eligibility. Full inclusion/exclusion criteria are listed in the study protocol. Patients provided written informed consent prior to any study related interventions. The study was registered with clincaltrials.gov (NCT03910543). The clinical extent of cutaneous sarcoidosis was quantified using the validated Cutaneous Sarcoidosis Activity and Morphology Instrument (CSAMI)(1), which quantifies disease activity (representing active inflammation) and damage (representing tissue damage/destruction) separately. Only the activity portion of the CSAMI instrument was utilized for this study. Two 4mm punch biopsies of lesional skin were obtained from each patient, prior to initiation of tofacitinib (while taking the preceding immunomodulatory treatment regimen) and again after 6 months of tofacitinib. In patients where disease activity remained on tofacitinib, a lesion that still appeared active was biopsied. In patients with an apparent complete clinical response to tofacitinib, a representative, previously active area of skin was biopsied. In patients with primarily facial involvement, the research biopsies were not mandatory. A portion of the biopsy tissue from each patient was placed in RNAlater reagent (Qiagen) and snap frozen for downstream analyses (see below) and another portion was fixed in 10% buffered formalin and embedded with paraffin for histologic analysis (see below). In 3 patients, a third biopsy was obtained from lesional skin prior to initiation of tofacitinib and was used for single cell RNA sequencing (see below). Skindex-16 is a validated, skin-related quality of life tool (2). Skindex-16 was administered at baseline and after 6 months of tofacitinib therapy. FDG-PET imaging was performed in the morning following a one-day high fat/low carbohydrate diet followed by an overnight fast of greater than 12 hours as previously described (3) . Dietary preparation instructions were provided in writing to the patient, along with telephone contact by clinical staff 48-72 hours prior to the study. Both the cardiac Rubidium-82 (Rb-82) and whole-body FDG-PET imaging were performed on a GE Discovery D690 PET/CT scanner. Low-dose CT images (120 kV, 50 mA for BMI <39, modulated at 50-150 mA for BMI >39) for the purposes of attenuation correction were performed before Rb82 and FDG imaging sequences. Resting ECG-gated dynamic Rb-82 PET imaging were performed using 20-35 millicurie (mCi) of Rb82 consistent with established clinical guidelines and our previous publications(4). Following Rb-82 imaging, the patient was injected with FDG (8-10mCi), followed by a 60-minute waiting period during which the patient relaxed or read in a quiet room. Whole-body FDG images were acquired for 2.5 minutes per bed position from cranial apex to knee, and 2 minutes per bed position from knees to toes. A separate, 8-minute single bed position cardiac acquisition was then performed. Resting Rb-82 perfusion imaging was processed and interpreted visually with 4DM (Invia). Quantitative and visual analysis of cardiac FDG uptake was performed on the AW software platform (GE) on the dedicated 3D cardiac acquisition window (single 47 slice bed position) and reported as previously described (5, 6) . (Quantitative and visual analysis of extra-cardiac FDG update is described below.) When present, the relationship between the location of inflammation and perfusion defects was characterized. The maximum standardized uptake value (SUV) in the heart was used to represent the peak level of inflammation. The volume of inflamed myocardium, cardiac metabolic volume (CMV), was identified as myocardium exceeding a SUV threshold that was derived for each patient by multiplying the left ventricular (LV) blood pool (background) activity by 1.5 as previously described (3) . Cardiac metabolic activity (CMA) was defined as the CMV multiplied by the average SUV of that volume. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint At the time of six-month follow up imaging, patients were provided with their written diet log from the initial scan, and instructed to replicate the same high-fat/low carbohydrate diet and fast duration as closely as was feasible. Follow up scans were performed using the same protocol and PET scanner. The sarcoidosis metabolic lesion burden at baseline and post-treatment follow-up was determined using the whole-body PET segmentation tool LesionID® (MIM Software Inc, Cleveland, Ohio, U.S.A.). This tool allows semi-automatic determination of maximal and mean standardized uptake values (SUV) (SUV max and SUV mean , respectively), mean tumor volume (MTV, cm 3 ), and total lesion glycolysis (TLG, product of SUV mean and MTV) of individual sarcoidosis lesions, and subsequent automatic calculation of MTV and TLG for the whole body. In the absence of a dedicated inflammatory lesion lexicon, the oncology lexicon was used to define metabolic parameters for sarcoidosis burden assessment. The initial automatic segmentation (lesion contouring) performed by the software was based on a predefined threshold which was defined by placing a 3 cm diameter sphere in the right lobe of the liver. The threshold value was determined using 1.5 * liver SUV mean + 2 * liver standard deviation, as defined by PET Response Criteria in Solid Tumors (PERCIST)(7). All voxels above this value were segmented and volumetrically separate regions were made into single volumes of interest (VOIs). The automatic segmentation was subsequently corrected by consensus of 3 readers, a dual-boarded radiologist and nuclear medicine physician (15 years of experience) (D.P.), a board-certified cardiologist (B.D.Y.), and a board-certified dermatologist (W.D.). During the correction process, the readers rejected false-positive lesions (mostly attributed to physiological uptake or pathologic uptake deemed sarcoidosis-unrelated) and made any necessary additional edits (modifying segmentation to avoid physiological uptake). All regions were combined into a total lesion burden VOI that encompassed all approved lesions for that timepoint. All lesions were tracked across time at the follow-up study. On follow-up scan, we placed a liver reference region in the right lobe of the liver in order to calculate the threshold value specific to this timepoint. Tracked lesions were transferred from the baseline PET/CT to the follow-up PET/CT via a rigid fusion and redefined automatically using PET Edge+®, a hybrid intensity and gradient-based tool (8) . We adjusted the tracked lesion VOIs as described above, if necessary. Next, similarly to the baseline timepoint, the rest of the image volume, not including the tracked lesions, was segmented with the PERCIST value specific to the follow-up timepoint. We followed the same steps as described above on the baseline timepoint to approve and finalize a total lesion burden VOI for the follow-up timepoint. The software then calculated all metabolic parameters from the total lesion burden VOI based on final lesion contours created by semi-automatic segmentation and individually tracked lesions were compared across both timepoints. Baseline and follow-up studies were processed simultaneously for immediate automatic determination of treatment-related changes in the individual patients. Most patients were on therapy for their sarcoidosis at the outset of the study. Patients had the option to continue, change the dose, or discontinue other immunosuppressive medications at the outset, or during the trial. One patient was taking hydroxychloroquine and it was discontinued upon initiation of tofacitinib. Four patients were taking methotrexate at the outset, all elected to discontinue this medication upon initiation of the tofacitinib. Five patients were taking prednisone. Throughout the 6-month study period, patients were permitted to taper and/or discontinue the prednisone based on their symptoms and physical examination findings and the investigators' discretion. In one patient, prednisone was increased during the study period related to increased pain associated with pre-existing Achilles tendonopathy. Statistical analysis was performed using GraphPad Prism 9. Simple linear regression analysis and P value determination was performed. The 95% confidence interval bands were plotted in Prism. Line graphs and histograms were also plotted in Prism. Histology and immunohistochemistry (IHC) were performed on formalin-fixed, paraffin-embedded sections using standard methods. For IHC, antigen retrieval was performing using citrate buffer (pH 6.0) (Life Technologies). The All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint network analyses were utilized to compare differentially expressed gene lists generated using Seurat (scRNA-seq) or Partek Flow (bulk RNAseq). ggplot2 v3.2.1, ggrepl (v0.8.2) and dplyr (v1.0.2) were used to visualize the results by plotting Z-score versus p-value. Selected genes or pathways were labeled at the investigators' discretion. Heatmaps were generated using gplots (v3.0.4) heatmap.2 function. Raw values were scaled using the scale function. Manhattan clustering was utilized. For Cellphone DB (version 2.1.4), normalized counts and meta data were exported from Seurat and imported into CellphoneDB. The statistical_analysis command was used. Dotplots were generated using the dot_plot command. Blood samples were collected from participants at baseline and again after 6 months of therapy. Samples were compared to 11 healthy, roughly age matched controls (Table S4 ). Blood was collected in EDTA-coated tubes and centrifuged; the buffy coat was removed and stored separately from the plasma, which was aliquoted and stored at -80ºC. Plasma was analyzed using the Olink Explore 1536 panel. This is a multiplexed panel that quantifies 1536 protein analytes simultaneously using a proximity extension approach. Detection of individual proteins requires binding by two matched, oligo-barcoded antibody pairs for each target. When both antibodies are bound, the oligo tags come in close proximity to each other and hybridize forming a unique template for DNA polymerase-dependent extension. After PCR amplification, next generation sequencing is performed and processed by Olink (Boston, Massachusetts). Log-fold changes and p values were calculated for pre-vs post-treatment and pre-treatment vs healthy. The data was then imported into R to generate heatmaps and volcano plots using ggplot2 (v3.3.3), ggrepl (v0.9.1), gplots (v3.1.1) and dplyr (v1.0.3). All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint SUPPLEMENTARY TABLES Table S1 . Summary of adverse events. *This was a chronic, pre-existing condition. The Achilles was not significantly PET avid arguing against it representative a specific manifestation of sarcoidosis. **Patient requested this approach as it had worked well for him in the past. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Table S5 . Samples available on Gene Expression Omnibus. The Skindex-16 metric was administered at baseline and again after 6 months of tofacitinib. 13 All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Figure S2 . Summary of all cells analyzed using scRNA-seq in sarcoidosis and control skin. A. UMAP projection of scRNA-seq data showing clustering of all cells, colored by cell type. B. UMAP projection of scRNA-seq data in A, colored by condition/library. C. Violin plots showing cell-lineage markers used to classify cell types in A and B. D. Histograms showing contribution of each library/condition to each cluster. NK: natural killer cell, L-endo: lymphatic endothelium, Myofibro: myofibroblast. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Figure S3 . Analysis of T cells in scRNA-seq experiments. A. UMAP projections of T cell clusters from scRNA-seq experiments (corresponding to Figure 3A) showing relative expression levels of selected genes. B. Volcano plot showing the most differentially expressed genes between CD4+ T cells in sarcoidosis (clusters 2,7,12) versus CD4+ T cells in controls (clusters 0,5), corresponding to Figure 3A . C. Histogram showing selected predicted upstream regulators of CD4+ T cells in sarcoidosis (clusters 2,7,12) versus CD4+ T cells in controls (clusters 0,5) as determined by IPA. Significance cutoff of p<0.001 is shown by a dotted vertical line. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Figure S4 . Analysis of myeloid cells in scRNA-seq experiments. A. UMAP projections of myeloid cell clusters from scRNA-seq experiments (corresponding to Figure 3B ) showing relative expression of selected genes. Inset shows IL12B expression in cluster DC-S (DCs with a cDC1 phenotype in sarcoidosis). B. Volcano plot showing the most differentially expressed genes between macrophages in sarcoidosis (clusters 0,1,3,4,6,7,9,10,11) versus myeloid cells in controls (clusters 2,8,13). C. Histogram showing selected predicted upstream regulators in macrophages in sarcoidosis (clusters 0,1,3,4,6,7,9,10,11) versus myeloid cells in controls (clusters 2,8,13) as determined by IPA. Significance cutoff of p<0.001 is shown by a dotted vertical line. 17 in All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint 18 lls All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Figure S6 . Principal component analysis of gene expression data with various gene sets. Analysis of gene expression data from skin in complete responders (CR) and partial responders (PR) relative to healthy controls. A. Macrophage (Mac-SAR) and T cell (Th1 SAR) activation signature genes from scRNA-seq analysis ( Figure 3C ) were used to perform principal component analysis. B-F Cytokine response signature gene sets ( Figure 3J ) were used to perform principal component analysis. All rights reserved. No reuse allowed without permission. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 5, 2021. ; https://doi.org/10.1101/2021.07.01.21259700 doi: medRxiv preprint Primary antibody binding was detected and visualized using ImmPRESS peroxidase reagent kit (Vector) and diaminobenzidine (DAB) substrate (Vector) Total RNA was extracted using the RNeasy Fibrous Tissue Mini Kit (Qiagen) according to the manufacturer's instructions. Two healthy controls were also included (Table S2). RNA was submitted to the Yale Stem Cell Center and complementary DNA (cDNA) libraries were generated and 100-base pairs paired-end sequencing was performed on an Illumina HiSeq 4000 at Yale Stem Cell Center. Bulk RNA-seq data was initially processed using Partek Flow software Further processing to identify differentially expressed genes between control, pre-treatment, and during treatment groups was carried out using DESeq2 Single cell RNA sequencing (scRNA-seq) Biopsies were immediately biopsied and placed in to Dispase II (10 mg/mL) (Sigma) in RPMI with 2% FBS for 45 minutes at 37°C with shaking at 125 RPM. The dispase solution was placed on ice. The tissue was removed from the dispase solution After incubation in the liberase and the dispase solution was combined (to recover any additional cells from the dispase) and the mixture was triturated and filtered through a 100 μM sterile filter. Cells were washed with RPMI + 2% FBS and RBCs were lysed with ACK Lysing Buffer (Lonza) and washed again in RPMI + 2% FBS. Cells were stained with LIVE/DEAD Red viability dye (ThermoFisher) in RPMI + 0.1% FBS. Viable single cells were isolated using a BD FACS Aria cell sorter and sorted into cold RPMI + 2% FBS. Up to 10,000 cells (depending on the yield) were loaded onto a 10X Chromium Controller. Single cell libraries were prepared according to the manufacturer's instructions by the Yale Center for Genome Analysis (YCGA) Libraries were aggregated using cellranger aggr without normalization to generate a single cell-bygene matrix. The Seurat R package (v3.2.0) was used for further analyses. Droplets with ≤ 100 expressed genes were removed from the matrix. The NormalizeData command with a scaling factor of 10,000 was used to normalize counts. ScaleData was used to regress the data against the number of transcripts and center gene expression values. Principle component analysis (PCA) was performed using RunPCA. Cells were clustered using FindNeighbors and FindClusters. Clusters consisting of cells with low/null expression of GAPDH (non-cells) were removed from further analysis using the SubsetData command, resulting in 24,034 cells for analysis. The data was visualized by performing Uniform Manifold Approximation and Projection (UMAP) dimensional reduction. Cell-type assignments for each cluster were determined using canonical markers. For immune cell subsets, cell type assignments for each cluster were verified by comparing with ImmGen datasets(11). T cell and myeloid cell clusters were subsetted and re-analyzed separately using the approach described above Reliability and convergent validity of the cutaneous sarcoidosis activity and morphology instrument for assessing cutaneous sarcoidosis Measurement properties of Skindex-16: a brief quality-of-life measure for patients with skin diseases Quantitative interpretation of FDG PET/CT with myocardial perfusion imaging increases diagnostic information in the evaluation of cardiac sarcoidosis ASNC imaging guidelines for nuclear cardiology procedures : Standardized reporting of nuclear cardiology procedures 18)F-FDG PET/CT for the assessment of myocardial sarcoidosis Joint SNMMI-ASNC expert consensus document on the role of (18)F-FDG PET/CT in cardiac sarcoid detection and therapy monitoring From RECIST to PERCIST: Evolving Considerations for PET response criteria in solid tumors Evaluation of a novel hybrid intensity and gradient-based method for PET tumor segmentation using a Monte Carlo simulated NSCLC phantom. Annual Congress of the European Association of Nuclear Medicine Pathway Is Hyperactive in Hidradenitis Suppurativa and Contributes to Skin Infiltration and Destruction Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles Immunological Genome Project C. The Immunological Genome Project: networks of gene expression in immune cells Tofacitinib Treatment and Molecular Analysis of Cutaneous Sarcoidosis Treatment of Multiorgan Sarcoidosis With Tofacitinib The authors would like to thank G. Wang and the Yale Center for Genome Analysis as well as M. Zhong and the Yale Stem Cell Center sequencing core for assistance with scRNA-seq and bulk RNA-seq experiments, respectively. WD is supported by a Career Development Award from the Dermatology Foundation. CR is supported by research funding from the Chest Foundation and NIH/NHLBI (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Sarcoidosis patient 7 after treated with tofacitinib for six months All rights reserved. No reuse allowed without permission.(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.