An N-of-1 Trial of Itacitinib for a Patient with Aplastic Anemia Associated with a Gain-of-


 An 18-year-old man presented with aplastic anemia, and exome sequencing identified a germline gain-of-function variant in the gene STAT1. Treatment with itacitinib, an investigational selective Janus Kinase 1 (JAK1) inhibitor, resulted in prompt recovery of hematopoiesis. An exhausted CD8+ T cell population and myeloid populations enriched for an interferon-y signature correlated with disease activity. Patient bone marrow sections displayed increased phospho-STAT1 staining, as did other idiopathic aplastic anemia cases, suggesting a shared pathophysiologic mechanism. This study describes the success and mechanism of a molecularly targeted therapy with potential implications for the treatment of aplastic anemias and other autoimmune disorders.

Page 4 patients with more than 100 different STAT1 GOF variants have been reported. 4,6,7 Aplastic anemia 78 has been described, as have numerous other autoimmune cytopenias, and severe aphthous 79 stomatitis is frequent. 4,6,8,9 GOF variants lead to hyper-phosphorylation of STAT1 in response to 80 stimulation, increasing STAT1-dependent transcription. 7,10,11 The p.Ala267Val variant identified in our 81 patient is absent from large population databases, but has been identified in more than 10 individuals 82 in multiple families with chronic mucocutaneous candidiasis, and shown to have a GOF effect. 6, 12 We 83 thus considered the p.Ala267Val variant as causal for our patient's recurrent aphthous ulcers, thrush, 84 and aplastic anemia. 85

Itacitinib Trial 86
As STAT1 activation is heavily dependent on JAK signaling, the JAK inhibitors ruxolitinib and 87 tofacitinib have been trialed in STAT1 GOF patients. Successful and unsuccessful cases have been 88 reported for the treatment of alopecia, diabetes, thrush, fungal infections, and autoimmune 89 cytopenias. 8, [13][14][15][16][17][18] Ruxolitinib and tofacitinib inhibit both JAK1 and JAK2 with the potential for anemia 90 and thrombocytopenia, since JAK2 is downstream of the erythropoietin and thrombopoietin receptors. 91 In considering a JAK inhibitor, we thus sought to avoid JAK2 inhibition. 92 The investigational drug itacitinib is a potent JAK1 inhibitor with an IC50 of 3.2 nM and selectivity for 93 JAK1. 19 Itacitinib has been studied in myeloproliferative neoplasms, graft-versus-host disease, and 94 autoimmune disorders. 95 The patient remained dependent on red blood cell transfusions, and he was thrice admitted to the 96 hospital for febrile neutropenia. Preparations were underway for hematopoietic stem cell 97 transplantation when, based on genotyping results, we enrolled the patient in a single patient 98 expanded use trial of itacitinib (https://clinicaltrials.gov/ct2/show/NCT03906318). On day 169 after 99 presentation, he received his weekly transfusion and initiated itacitinib 300mg daily. The following 100 week, and for all subsequent visits after the initiation of itacitinib, laboratory values demonstrated 101 resolution of neutropenia and anemia, and the patient became red blood cell transfusion independent 102 (Fig. 1C). A subsequent bone marrow biopsy demonstrated a return of trilineage hematopoiesis.
Increased STAT1 signaling as measured by phospho-STAT1 (pSTAT1) staining, which had been 104 present at time of diagnosis, resolved after treatment (Fig. 1D). The patient's thrush and oral ulcers 105 persisted but at significantly decreased frequency and severity. Clinically, fatigue and weight loss 106 resolved, and he returned to schooling. He completed 20 months of itacitinib therapy without any 107 adverse events before electing to self-discontinue. Aplastic anemia has not recurred. 108

Immunological Profiling 109
To understand the cellular mechanisms involved in the pathogenesis and resolution of this patient's 110 aplastic anemia, we performed comprehensive immunophenotyping. To confirm STAT1 GOF activity, 111 we used phospho-CyTOF to measure phospho-STAT1 in monocytes after stimulation with interferon-g 112 at timepoints pre-and post-itacitinib treatment. At all timepoints, the patient's monocytes had higher 113 levels of pSTAT1 than healthy control. Post-itacitinib, pSTAT1 levels decreased, though not to the 114 level of healthy control ( Fig. 2A). 115 The largest studies of STAT1 GOF patients have demonstrated increased frequencies of TH1 and 116 decreased frequencies of TH17 T cells. 20,21 Increased signaling through STAT1 biases T cell 117 differentiation into the TH1 phenotype, which is characterized by secretion of interferon-g. TH1 118 differentiation comes at the expense of STAT3-mediated TH17 differentiation, which is characterized 119 by secretion of IL-17A and cytokines critical for immunity to mucocutaneous Candida. It is 120 hypothesized that JAK inhibition may restore the balance between TH1 and TH17 differentiation by 121 dampening STAT1 signaling. 8 122 Interrogating the TH1 and TH17 pathways, we compared plasma cytokine levels pre-and post-123 treatment with itacitinib. This identified significant decreases in the TH1 cytokines interferon-g and 124 IL-12p40 following therapy, but no significant shifts in the TH17 cytokines IL-17A or IL-17F (Fig. 2B). 125 Of note, elevated serum interferon-g has been described in aplastic anemia. 22 126 Examining CD4+ T cells by post-stimulation intracellular cytokine staining, we found a near absence 127 of IL-17A-producing CD4+ T cells (TH17) in the patient compared to healthy controls; this did not Page 6 change with treatment (Fig. S2B). Frequencies of interferon-g-producing CD4+ T cells (TH1) did not 129 differ between healthy controls, the patient pre-itacitinib, or the patient post-itacitinib (Fig. 2C). Thus, 130 we did not observe evidence of itacitinib restoring a balance between STAT1-mediated TH1 and 131 STAT3-mediated TH17 frequencies. Moreover, while we observed other peripheral blood mononuclear 132 cells (PBMC) abnormalities including an absence of Treg cells, elevated frequencies of memory CD4+ 133 and CD8+ T cells, and reduced frequencies of T follicular helper (TFH) cells compared to healthy 134 controls (Fig. S2A), none of these frequencies changed significantly with itacitinib treatment. 135 Surprisingly, the patient's pre-itacitinib CD8+ T cells produced significantly less interferon-g than 136 healthy controls in vitro (Fig. 2D). After treatment with itacitinib, CD8+ interferon-g production 137 increased but not to the level of healthy controls. Pre-itacitinib memory CD8+ T cells also had a 138 striking increase in expression of the activation marker Programmed Death 1 (PD-1), which 139 decreased with itacitinib treatment (Fig. 2E). Thus PD-1+ CD8+ T cells-which are defective in their 140 ability to secrete interferon-g upon stimulation in vitro-closely correlate with aplastic anemia disease 141 activity. 142 Taken together, these findings suggest a model whereby during aplastic anemia, the patient's CD8+ 143 T cells are chronically activated against hematopoietic progenitors in vivo leading to an exhausted 144 phenotype characterized by impaired ability to secrete interferon-g in vitro and increased PD-1 145 expression. By inhibiting JAK1-driven activation, itacitinib relieved CD8+ T cell exhaustion. Consistent 146 with our findings, elevated frequencies of activated CD8+ T cells have been described in aplastic 147 anemia. 23 We did observe the seemingly paradoxical finding of increased serum interferon-g (Fig. 2B) 148 yet decreased interferon-g secretion by CD8+ T cells in vitro (Fig. 2D). One interpretation of these 149 results is that these CD8+ T cells are chronically secreting interferon-gamma in vivo, yet their 150 exhausted phenotype reduces their capacity to secrete interferon-gamma in vitro upon stimulation. 24

151
To explore the in vivo unstimulated cytokine effector function of these cells, we performed single-cell 152 transcriptional analyses. 153

Single-Cell Gene Expression Analysis 154
To understand the transcriptional programs governing our patient's aplastic anemia, we performed 155 single-cell transcriptional sequencing (scRNA-Seq), comparing PBMCs from healthy controls and our 156 patient at timepoints before and after itacitinib. Unsupervised clustering demonstrated that healthy, 157 pre-itacitinib, and post-itacitinib cells represent distinct clusters. Post-itacitinib samples clustered 158 between healthy and pre-itacitinib samples, suggesting progression from disease towards the healthy 159 state (Fig. 3A). 160 We then separately subclustered T cell-containing and myeloid populations for further analysis (Fig.  161   S3). T cell-containing cells from the patient and healthy controls subclustered into a CD8+ population, 162 NK population, and mixed CD4+ and CD8+ memory and naïve populations. Given the activated but 163 dysfunctional state of our patient's CD8+ T cells, we scored all CD8+ T cells for cytotoxicity, cytokine 164 effector function, and exhaustion ( Table S1). Each of these scores were elevated in the patient's pre-165 itacitinib cells compared to healthy controls, confirming their cytotoxic and exhausted phenotype (Fig.  166   3B). After treatment, exhaustion and cytokine effector scores decreased. Our finding of increased 167 transcriptional cytokine effector scores in the patient's pre-itacitinib CD8+ T cells supports the 168 hypothesis that these cells secrete higher levels of cytokines such as interferon-g in vivo, despite their 169 reduced capacity to secrete interferon-g in vitro upon stimulation. 170 Subclustering of myeloid populations identified five populations (Fig. 3C). Of note, "activated CD14+ 171 monocytes" and "C1Q+ monocytes" were found virtually exclusively in the patient's and not in the 172 healthy controls' cells. Both of these subclusters expressed STAT1 activation-induced genes such as 173 the complement genes C1QC, C1QB, C1QA or the interferon-inducible genes FAM26F or GBP1, 174 respectively (Fig. S3). Type I interferons (a, b, and others) and Type II interferon (only g) induce 175 overlapping but distinct transcriptional signatures. Since monocytes express high levels of the 176 interferon-g receptor, we scored each myeloid cell to assess for interferon exposure (Table S1). Both 177 Type I and II interferon scores in pre-treatment monocytes were higher than healthy control or post-itacitinib monocytes. Within pre-itacitinib monocytes, Type II scores were higher than Type I, 179 suggesting a primarily interferon-g induced state (p<10 -10 , Wilcoxon rank-sum). 180

Single-Cell Epigenetic Analysis 181
To understand the epigenetic effects of the STAT1 GOF variant and itacitinib treatment, we performed 182 single-cell Assay for Transposase-Accessible Chromatin with Sequencing (scATAC-seq). Within 183 effector CD8+ T cells, we saw increased accessibility at the PD-1 locus, consistent with our 184 observation of increased PD-1 protein expression (Fig. 2E), and this accessibility decreased after 185 treatment ( Fig. 3D and S4). 186 Genome-wide, STAT1 motif accessibility pre-itacitinib was increased when compared to healthy 187 controls or post-itacitinib samples (Fig. 3E). These studies suggest that changes in accessible 188 chromatin correlate with STAT1-mediated autoimmunity and can be reversed with itacitinib. 189 pSTAT1 Bone Marrow Immunostaining 190 Immunostaining our patient's bone marrow sections, we found increased pSTAT1 staining pre-191 itacitinib, and this STAT1 activity resolved following itacitinib therapy (Fig. 1D). To determine whether 192 similar STAT1 dysregulation exists in patients with idiopathic (non-STAT1-mutated) aplastic anemia, 193 we analyzed bone marrow sections from other aplastic anemia patients. Compared to healthy donor 194 marrow in which no staining was detected, three of four marrows exhibited positive pSTAT1 staining 195 (Fig. 4). This raises the exciting possibility that STAT1 activation is a feature of a subset of aplastic 196 anemias for which pSTAT1 is a potential biomarker. 197

Discussion 198
Here we present a man with a history of oral ulcers inherited in an autosomal dominant pattern who 199 developed aplastic anemia. Exome sequencing identified a pathogenic GOF mutation in STAT1. Longitudinal immunophenotyping of samples unperturbed by confounding immunosuppressive 203 therapy revealed an expanded, activated, cytolytic, and exhausted memory CD8+ T cell population 204 that correlated with disease activity, as did plasma levels of interferon-g. After itacitinib treatment, 205 these abnormalities improved, with parallels to mouse and human models of idiopathic aplastic 206 anemia. 23,25 207 To our knowledge, this represents the first report of a targeted therapy for the treatment of 208 autoimmune aplastic anemia. This case also marks the first report of itacitinib for use in a primary 209 immunodeficiency. Given the high frequency of cytopenias in primary immunodeficiencies, itacitinib's 210 JAK1 selectivity may be of particular clinical benefit. 26,27 211 Establishing the causality of a therapy in a single-patient trial is challenging, but the co-incident timing 212 of itacitinib with near immediate hematopoietic recovery after 6 months of transfusion-dependence, as 213 well as the low expected rate of spontaneous recovery in aplastic anemia, 2 are evidence in favor of a 214 therapeutic effect of itacitinib in this case. 215 While STAT1 GOF is a rare condition, aplastic anemia is more common. Aplastic anemia can be 216 triggered by a spectrum of host and environmental factors, 2 but our patient's immunophenotypic 217 similarities to other cases of aplastic anemia raise the possibility of a shared downstream 218 pathophysiology for which JAK inhibitor therapy could be effective. Additionally, our finding of 219 increased bone marrow pSTAT1 staining in a majority of aplastic anemia cases may be a useful 220 biomarker to identify candidate patients. We hope these results spur larger trials to answer these 221 questions, particularly given the relative tolerability of JAK inhibitors compared to current standard 222 treatment modalities including hematopoietic stem cell transplantation or medical 223 immunosuppression. 224