Targeting Hedgehog Signaling with Glasdegib in Patients with Refractory Sclerotic Chronic GVHD: A Report of Two Phase I/II Trials

1 Regular article 1 2 Targeting Hedgehog Signaling with Glasdegib in 3 Patients with Refractory Sclerotic Chronic Graft vs. 4 Host Disease: A Report of Two Phase I/II Trials 5 6 Running head: Glasdegib in patients with sclerotic cGVHD 7 8 Eduardo Rodríguez-Arbolí1‡*, Catherine J. Lee,2* Teresa Caballero-Velázquez1‡, 9 Carmen Martínez3‡, Clara García-Calderón1‡, María Reyes Jiménez-León1‡, María 10 Aránzazu Bermúdez-Rodríguez4‡, Lucía López-Corral5‡, Ana Triguero3‡, Lynn Onstad6, 11 Mitchell E. Horwitz7, Stefanie Sarantopoulos7, Stephanie J. Lee6**, José Antonio Pérez-12 Simón1‡** 13 14 * Co-first authors; ** Co-corresponding authors. 15 ‡ On behalf of Grupo Español de Trasplante Hematopoyético y Terapia Celular (GETH-TC). 16 17 1. Department of Hematology, Hospital Universitario Virgen del Rocío, Seville Biomedicine 18 Institute (IBiS/CSIC), University of Seville, Seville, Spain. 19 2. Utah Blood and Marrow Transplant Program, Division of Hematology and Hematologic 20 Malignancies, Huntsman Cancer Institute, University of Utah, Salt Lake City, UT. 21 3. Hematopoietic Stem Cell Transplantation Unit, Hematology Department, Institute of 22 Hematology and Oncology, IDIBAPS, Josep Carreras Research Institute, Hospital Clínic, 23 Barcelona, Spain. 24 4. Hospital Universitario Marqués de Valdecilla, Santander, Spain. 25 5. Department of Hematology, Hospital Clínico Universitario de Salamanca (CAUSA/IBSAL), 26 Salamanca, Spain. 27 6. Clinical Research Division, Fred Hutchinson Cancer Center, Seattle, WA. 28 7. Division of Hematologic Malignancies and Cellular Therapy, Department of Medicine, Duke 29 University School of Medicine, Duke Cancer Institute, Durham, NC. 30 31 32 Address for correspondence: 33 34 Stephanie J. Lee, MD MPH 35 Clinical Research Division, Fred Hutchinson Cancer Center; 36 2 1100 Fairview Ave N, D5-290; Seattle, WA 98109-1024, USA 37 Phone: 206 667 5160; Fax: 206 667 1034; E-mail: [email protected] 38 39 José Antonio Pérez-Simón, MD PhD 40 Department of Hematology, Hospital Universitario Virgen del Rocío, Seville 41 Biomedicine Institute (IBiS/CSIC), University of Seville 42 Avda. Manuel Siurot s/n 41013, Seville, Spain 43 Phone: +34 955 01 3260; Fax: +34 955 01 3265 ; E-mail: 44 [email protected] 45 46 Version: May 26, 2023 47 48 Conflict of interest: 49 S.J.L.: Consultant fees from Mallinckrodt, Equillium, Kadmon; research funding from 50 Amgen, AstraZeneca, Incyte, Kadmon, Pfizer, Syndax; provision of study medication 51 from Janssen; steering committee for a clinical trial, Incyte. 52 J.A.P.S: Honoraria for educational activities and/or conferences and/or participation in 53 advisory boards and/or travel grants and/or funding for research activities from Pfizer, 54 JNJ, BMS, Jazz, Alexion, Novartis, Gilead, Amgen, AbbVie. 55 E.R.A.: Travel and conference funding from Jazz Pharmaceuticals, Abbvie, Gilead. 56 All other authors report no conflicts of interest relevant to this article. 57 58 59 60 61 62 3 63 64 65 66 67 68 Translational relevance statement 69 Prior studies have revealed Hedgehog (HH) signaling activation in dermal fibroblasts as 70 a key biological hallmark of sclerotic chronic graft-versus-host disease (scGVHD). The 71 results from these two phase I/II trials provide proof of concept of the actionability of 72 the HH signaling pathway in this clinical setting. Treatment with glasdegib, a selective 73 Smoothened inhibitor, was associated with clinically meaningful and sustained 74 improvements in a range of standardized response measures in patients with heavily 75 refractory scGVHD. Tolerability was constrained by the frequent emergence of on-76 target toxicities, but prolonged treatment was feasible in some patients. Taken together, 77 our data support the use of glasdegib as novel targeted therapeutic option for scGVHD 78 patients not responding to established therapies. Notably, our analyses failed to identify 79 fibroblast-independent immunomodulatory effects upon HH signaling inhibition. 80 81 82 83 4 84 85 86 87 88 89 Abstract 90 Purpose: 91 Sclerotic chronic graft vs. host disease (scGVHD) is characterized by progressive skin 92 fibrosis and frequent refractoriness to available therapies. Aberrant activation of 93 Hedgehog signaling in dermal fibroblasts has been implicated in scGVHD. Here, we 94 report the results of two phase I/II studies (NCT03415867, GETH-TC; NCT04111497, 95 FHD) that evaluated glasdegib, a SMO antagonist, as a novel therapeutic agent in 96 refractory scGVHD. 97 Patients and Methods: 98 Adult patients with active scGVHD after ≥1 (FHD) or ≥2 (GETH-TC) lines of therapy 99 were enrolled. Primary endpoints were dose-limiting toxicity (DLT) and maximum 100 tolerated dose (MTD) in the GETH-TC trial, and safety and tolerability measures in the 101 FHD trial. Glasdegib was administered once daily in 28-day cycles. Responses were 102 scored per 2014 NIH cGVHD criteria. Correlative studies were performed to evaluate 103 the role of fibroblast-independent immune mechanisms on clinical activity. 104 5 Results: 105 Twenty (GETH-TC) and 15 (FHD) patients were recruited. Treatment-emergent grade 106 (G) ≥2 adverse events (AEs) in the GETH-TC trial included muscle cramps (85%), 107 alopecia (50%) and dysgeusia (35%). Two patients experienced a DLT (G3 muscle 108 cramps), and the MTD was established at 50 mg. G3 muscle cramps were the most 109 frequently reported AE (33%) in the FHD trial. At 12-months, the skin/joint scGVHD 110 overall response rate was 65% (all partial responses) in the GETH-TC trial and 47% (6 111 partial responses, 1 complete response) in the FHD cohort. No immune correlates of 112 response were identified. 113 Conclusions: 114 Glasdegib demonstrated promising responses in patients with refractory scGVHD, but 115 tolerability was limited by muscle cramping. 116 117 118 119 120 121 122 123 124 6 125 126 127 128 129 Introduction 130 Chronic graft-versus-host disease (cGVHD), a late complication after allogeneic 131 hematopoietic stem cell transplantation (alloSCT), is a pleomorphic condition in which 132 immune-mediated clinical manifestations in multiple organs usually coexist (1–3). 133 Sclerotic cGVHD (scGVHD), a distinctive phenotypic variant that affects up to 20% of 134 patients requiring systemic treatment for cGVHD, is characterized by progressive skin 135 fibrosis and frequent refractoriness to available therapies(4–6). In advanced stages, 136 scGVHD can lead to chronic skin ulceration, joint contractures and pulmonary 137 restriction, often resulting in major disability, prolonged requirements of 138 immunosuppressive therapy, and severely impaired quality of life. Yet, despite an 139 expansion of the cGVHD armamentarium in recent years(7–10), the development of 140 effective therapeutic options for steroid-refractory scGVHD remains a largely unmet 141 need(11,12). 142 Although our mechanistic understanding of cGVHD-induced fibrosis is still far from 143 complete, prior studies have revealed Hedgehog (HH) signaling activation in dermal 144 fibroblasts as a key hallmark in the pathogenesis of scGVHD. HH signaling, a 145 developmental morphogen pathway tightly regulated in adult tissues, promotes fibrosis 146 7 in scGVHD through the differentiation of resting fibroblasts into metabolically active, 147 collagen-releasing myofibroblasts(13,14). Notably, blockade of HH signaling has been 148 shown to prevent the onset of sclerosis and to halt the progression of established 149 sclerotic features in murine models of scGVHD(14). In addition to its direct role in the 150 modulation of fibroblast activity, HH signaling is also known to be involved in a 151 number of immune regulatory circuits relevant to cGVHD biology. For instance, 152 overactivation of the HH pathway in thymocytes reduces the intensity of the TCR 153 signal, thereby interfering with TCR-mediated positive selection and allowing the 154 escape of autoreactive cells from clonal deletion(15). Likewise, SHH modulates 155 lymphocyte activation and cytokine production in peripheral CD4+ T lymphocytes, 156 cytotoxic T lymphocyte function, proliferation in germinal center B cells, and M2 157 polarization in macrophages(16–19). Whether aberrant HH activation in immune cell 158 subpopulations might therefore contribute to scGVHD onset and maintenance, and 159 account to certain extent for the clinical activity of SMOi in this setting, remains 160 unexplored. 161 HH signaling is amenable to therapeutic intervention through the inhibition of 162 smoothened (SMO), a transducer protein with a central role in this pathway. Upon 163 binding of HH ligands to the patched homolog-1 (PTCH1) membrane receptor, SMO is 164 released to direct the downstream activation of the HH signaling transcriptional 165 program through the stabilization of GLI family zinc finger transcription factors(20–166 22). Prominent HH signaling dysregulation across cancer types(23) has propelled the 167 development of SMO inhibitors (SMOi), with approved agents in clinical use for the 168 treatment of basal-cell carcinoma and acute myeloid leukemia(24–26). 169 Here, we report the results of two independent phase I/II clinical trials that evaluated 170 glasdegib, an orally bioavailable, selective, small-molecule SMO antagonist(27,28), as a 171 8 novel therapeutic agent in patients with refractory scGVHD. Our clinical data was 172 complemented by correlative studies conducted with the aim of determining if 173 fibroblast-independent immune modulation might contribute to glasdegib activity in this 174 patient population. 175 176 177 178 179 Methods 180 Study design 181 Two independent, single-arm, open-label, phase I/II investigator-initiated studies 182 [NCT03415867, Spanish Hematopoietic Transplantation and Cellular Therapy 183 Cooperative Group (GETH-TC); NCT04111497, Fred Hutchinson Cancer 184 Center/Huntsman Cancer Institute/Duke University Medical Center (FHD)] were 185 conducted at 5 centers in Spain and 3 centers in the United States. 186 Primary endpoints were dose-limiting toxicity (DLT) and maximum tolerated dose 187 (MTD) in the GETH-TC trial, and safety and tolerability measures [i.e. type and 188 severity of adverse events (AEs), time on treatment, and reasons for discontinuation] in 189 the FHD trial. Secondary endpoints included best overall response rate (ORR) in 190 cGVHD and sclerotic manifestations, patient-reported outcomes, and immunological 191 correlates of response. Data cut-off was at 12 months from the time of glasdegib 192 initiation. 193 9 Glasdegib was administered orally once daily (OD) in continuous 28-day cycles. The 194 GETH-TC trial followed a dose-finding strategy, whereas the FHD trial employed a 195 fixed-dose design. In the GETH-TC study, the starting treatment dose of glasdegib was 196 50 mg (Level 1), with subsequent dose escalation/de-escalation stages planned at 25 mg 197 (Level -1), 100 mg (Level 2), 150 mg (Level 3), and 200 mg (Level 4) per a standard 198 3+3 escalation design. DLTs were defined as any treatment-related grade (G) ≥3 199 toxicity that was uncontrolled despite optimal medical management, excluding G ≥3 200 electrolyte abnormalities and ALT/AST elevations that returned to G ≤1 within 7 days. 201 The DLT-evaluable period spanned the first two cycles of treatment. An independent 202 data monitoring committee (DMC) was established to evaluate the emerging safety 203 profile before each dose escalation stage. Intra-patient dose re-escalation after a 204 toxicity-related dose reduction was not permitted. Patients in the FHD study received 205 glasdegib at an initial dose of 50 mg. Dose reduction or interruption was indicated in the 206 event of treatment-related toxicity G ≥3 or intolerable treatment-related toxicity of any 207 grade, at least until the AE resolved to G ≤2 or stabilized to an acceptable degree. Taper 208 and eventual withdrawal of concurrent immunosuppressive treatment was allowed in 209 the event of a clinical response that was maintained after cycle 2 in the GETH-TC trial, 210 or at any time in the FHD trial. Target sample size was 20-24 patients in the GETH-TC 211 trial and 20 patients in the FHD trial. AEs were graded according to the National Cancer 212 Institute (NCI) Common Terminology Criteria for Adverse Events (CTCAE), version 213 4.0 (GETH-TC) or 5.0 (FHD). 214 The study protocols were approved by the institutional review boards of the 215 participating centers. All patients provided written informed consent, and the studies 216 were conducted in accordance with the principles of the Declaration of Helsinki. Both 217 trials were registered at ClinicalTrials.gov. 218 16 severity scores was 3 (range 2-6) and 2.5 (2-5) points (Figure 1C-D). Overall, 13 356 patients (65%) obtained a skin/joint scGVHD PR in the GETH-TC trial, with a median 357 time of 2.1 (range 0.9-6.0) and 8.4 (2.1-12.0) months to first and best response. 358 Likewise, 7 (47%) patients had a skin/joint scGVHD response (6 PR, 1 CR) in the FHD 359 cohort, with a median time of 0.9 (range 0.9-2.8) and 2.8 (1.7-6.0) months to first and 360 best response (Table 3 and Figure 2A-B). At the 12-month data cut-off, median time 361 of skin/joint scGVHD response on study was 9.0 months (range 0.0-11.0) in the GETH-362 TC trial and 1.9 months (range 0.0-11.0) in the FHD trial (Figure 2A-B). Median time 363 on study for responder patients was 12.0 (2.1-12.0) months in the GETH-TC trial vs. 3.6 364 (0.7-12) months in the FHD trial. Similarly, median time on study for non-responding 365 patients was 5.6 (1.6-12.0) months in the GETH-TC trial vs. 2.7 (1.0-12.0) months in 366 the FHD trial. Of note, there were patients who achieved a response after prior treatment 367 with ruxolitinib (GETH-TC n=6, FHD n=4), and ibrutinib (GETH-TC n=1). 368 Other target cGVHD organs 369 Clinical responses were also seen across other cGVHD target organs. Mouth (11/16, 370 69%; 5/9, 56%), eye (5/16, 31%; 3/10, 30%), lungs (2/7, 29%; 3/7, 43%), esophagus 371 (6/6,100%; 1/1, 100%), upper gastrointestinal tract (1/3, 33%; 2/2, 100%), and lower 372 gastrointestinal tract (2/2, 100%; 3/3, 100%) responses were reported in the GETH-TC 373 and FHD trials, respectively. Organ-specific response rates are detailed in Table 3. 374 Corticosteroid sparing 375 Sixteen (80%) patients in the GETH-TC trial were receiving treatment with 376 corticosteroids (CS) at the time of study entry. Among these, thirteen patients (81%; all 377 skin/joint scGVHD responders and 2 non-responders) achieved a CS dose reduction. 378 17 The median CS dose decreased from 0.22 mg/kg per day (range 0.02-0.49) (prednisone 379 equivalent) at baseline to 0.11 mg/kg per day (0.01-0.39; median change -52%). 380 In the FHD trial, 7 (47%) patients were on CS at baseline, of whom 6 (86%; all 381 skin/joint scGVHD responders and 3 non-responders) reduced their CS dose. Here, the 382 median CS dose decreased from 0.18 mg/kg per day (range 0.11-0.34) to 0.14 mg/kg 383 per day (0.09-0.45; median change -23%) (Table 3). 384 Patient-reported outcomes 385 In the GETH-TC trial, self-reported cGVHD global severity and skin tightness scores 386 were serially available (≥75% of scheduled visits) in 13 patients, 8 of whom were 387 responders per NIH physician-assessed response criteria. Seven (54%; 6 skin/joint 388 scGVHD responders) and 5 (38%; 4 skin/joint scGVHD responders) patients reported a 389 ≥2-point improvement lasting for ≥2 consecutive visits in cGVHD global severity and 390 skin tightness scores, respectively. 391 In the FHD trial, the median mLSS summary score at baseline was 21.2 points (range 392 2.8-26.1), and median change during follow-up was -3.1 points (range -18.3-11.4). Only 393 1 non-responder patient reported a clinically significant improvement ≥7 points. Two 394 patients had a ≥7-point mLSS worsening (1 responder and 1 non-responder). 395 Analyses of fibroblast-independent immunomodulatory effects of glasdegib 396 The baseline distributions of key immune cell subpopulations in patients and in healthy 397 controls are shown in Supplementary Table 4. TCR Vβ repertoire in scGVHD patients 398 was often characterized by clonotypic dominance resulting in decreased diversity 399 quantified by a Gini-like index (Figure 3A). Yet, treatment with glasdegib was not 400 associated with changes in TCR Vβ repertoire diversity over time (Figure 3B). 401 Similarly, no differences from baseline were detected in the distribution of circulating 402 18 T-cell subpopulations (i.e. naive, central memory, effector, or peripheral memory CD4+ 403 and CD8+ lymphocytes), or in the prevalence of regulatory-enriched T cells and 404 monocyte subsets (Figure 3C-E). Treatment with glasdegib also failed to show an 405 impact on the expression of co-stimulatory and co-inhibitory molecules (4-IBB, PD-1, 406 OX40, CTLA-4 and TNFRSF18) in the CD4+ and CD8+ compartments 407 (Supplementary Figure 1). No effect of glasdegib on downstream T-cell activation as 408 measured by the phosphoflow analysis of ERK, p38, STAT3, and STAT5 was detected 409 (Supplementary Figure 2). Finally, no changes were seen in the distribution of total, 410 naive, unswitched memory, and switched memory B-cells, nor in the expression of 411 BAFF (Supplemementary Figure 3). Overall, no differential patterns in the immune 412 profiles between glasdegib responders and non-responder patients were observed. As 11 413 out of 13 skin/joint scGVHD responders in the GETH trial achieved their first response 414 within 3 months of treatment with glasdegib, it was not possible to make any 415 meaningful inferences regarding potential differences in the immunophenotypic 416 repertoires between early responders and late responders in our analyses. 417 418 419 420 421 422 423 424 425 19 426 427 428 429 430 431 432 433 Discussion 434 These two prospective studies represent the largest reported cohort of patients with 435 scGVHD treated with a SMOi. With response rates of 65% and 47% in skin/joint 436 scGVHD after treatment with glasdegib, our findings provide proof of concept of the 437 clinical actionability of the HH signaling pathway in this hard-to-treat patient 438 population. Likewise, our data highlight the distinct challenges derived from the use of 439 SMOi in this setting, and more generally, the difficulties associated with the 440 performance of clinical trials in patients with scGVHD. 441 Notably, both of our studies encountered similar obstacles related to the emergence of 442 DLT-defining muscle cramps, which precluded dose escalation and resulted in study 443 closure in the GETH-TC and FHD trials, respectively. Muscle cramps, a well-444 characterized class-effect of HH signaling inhibitors, are thought to result from the non-445 canonical, SMO-mediated, paradoxical activation of muscle glucose uptake and Ca2+ 446 influx(33). Contrasting with the very high incidence of muscle cramps in our studies 447 20 (including G3 events in one third of the patients), muscle cramps of any grade were 448 experienced by 23% of patients in the experimental arm of the randomized trial of 449 glasdegib (100 mg daily) plus low-dose cytarabine in acute myeloid leukemia or high-450 risk myelodysplastic syndrome, with G3 events reported in only 5%(26). Thus, our data 451 suggest that the triggering threshold for SMOi-induced muscle AEs may be unusually 452 low in patients with scGVHD. Of note, the high baseline prevalence of muscle cramps 453 and myalgias among cGVHD patients may interfere with determining causality and 454 could worsen this muscle toxicity(34). However, the exclusion from the FHD trial of 455 patients with a significant history of muscle cramping did not appear to mitigate the risk 456 of intolerable cramping. Other on-target SMOi-related AEs such as dysgeusia and 457 alopecia, although not constituting DLTs, were common and could be expected to 458 impose a substantial burden on quality of life for patients undergoing long-term 459 treatment with glasdegib. Both pharmacological interventions and non-continuous 460 dosing schedules have been proposed for the management of SMOi-associated muscle 461 cramping when treating basal cell carcinoma(35). Among pharmacological 462 interventions, there is a rationale for the use of calcium channel antagonists based on the 463 pathophysiology of muscle cramps. In a pilot study, treatment with amlodipine was 464 associated with a decrease in the frequency of muscle cramps, but no changes in their 465 intensity or duration were detected(36). Other drugs have also been proposed for the 466 control of muscle cramps, although the level of evidence regarding their efficacy is 467 limited. These include antimuscarinic agents, anticholinergics, or baclofen(35). 468 Intermittent treatment regimens could represent another strategy to optimize SMOi 469 tolerability. In this regard, the results of a randomized phase 2 study evaluating two 470 intermittent treatment regimens with vismodegib suggest that therapeutic rest periods 471 could reduce the incidence of limiting toxicities in patients on long-term therapy(37). 472 21 These symptom palliation strategies were not assessed in our studies, but may have 473 improved the tolerability and net clinical benefit of glasdegib in the scGVHD 474 population. 475 Notwithstanding these major tolerability concerns, treatment with glasdegib was 476 associated with clinically meaningful and sustained improvements in a range of 477 skin/joint cGVHD standardized measures. Responses were also reported in other target 478 organs in which fibrosis might play an important role, such as the esophagus and the 479 gastrointestinal tract. Accordingly, and despite the frequent need for dose adjustments, 480 some patients opted for extended treatment with glasdegib beyond the first year. Also 481 notable from a efficacy standpoint, CS sparing was feasible in most patients. Although 482 response rates were moderately higher in the GETH-TC trial, differences might be 483 partially attributable to the longer exposure to glasdegib as compared to patients in the 484 FHD study. Moreover, median times to first and best response in the FHD trial were 485 shorter than in the GETH-TC trial. In this respect, it is worth noting that the median 486 times on study for both skin/joint scGVHD responder and non-responder patients were 487 significantly longer in the GETH trial. Additionally, standardized cGVHD assessments 488 were performed on a monthly basis in the FHD trial, whereas these were only scheduled 489 at the end of cycles 1, 3, 6, 9, and 12 in the GETH-TC trial. Therefore, longer follow-up 490 and sparser cGVHD assessments might have led to the capture of more late-responding 491 patients and deeper responses at later time points in the GETH-TC study compared to 492 the FHD trial. These results are encouraging given the heavily pretreated nature of these 493 two patients cohorts and the addition of glasdegib long after the initiation of the fibrotic 494 cascade, once the irreversibility of fibrotic changes could be anticipated to pose a 495 significant barrier to treatment efficacy. Targeting HH signaling earlier in the onset of 496 sclerosis might help optimise the efficacy and/or tolerability of SMOi. In fact, SMO 497 22 inhibition has been shown most efficacious when used as prophylaxis in preclinical 498 models(14). 499 Overall, our results are consistent with two previous studies that evaluated the use of 500 sonidegib (n=17) and vismodegib (n=6) in cGVHD. In the sonidegib study, a PR rate of 501 47% was reported in skin or sclerotic disease, though responses were not assessed per 502 NIH criteria. The trial was terminated early as a result of worsening quality of life and 503 cumulative toxicity burden not attributed to sonidegib. No DLT related to muscle 504 cramping was reported, but 3 patients experienced G3 myalgias that were not attributed 505 to sonidegib(38). Similarly, preliminary evidence of efficacy was observed in the 506 vismodegib trial in 5 patients who achieved a PR as determined per 2014 NIH response 507 criteria. Treatment-related AEs, including muscle cramps and dysgeusia, were common, 508 and the study was closed due to slow accrual(19). 509 Beyond its direct influence on myofibroblast activity, HH signaling is involved in key 510 immune processes whose dysregulation might contribute to cGVHD pathogenesis, 511 encompassing TCR repertoire selection and deletion of autoreactive cells in the thymus, 512 B-cell homeostasis and M2 macrophage polarization(18,39–42). In this regard, our 513 extensive analyses failed to identify immunomodulatory effects or immune correlates of 514 response upon HH signaling inhibition with glasdegib. In contrast, treatment with 515 vismodegib has been associated with decreases in M2-like macrophages in skin biopsies 516 and circulating pre-germinal center and plasmablast-like B-cells, neither of which was 517 assessed in our trials(19). Additionally, as immune correlative analyses in the GETH 518 study were only performed in samples obtained within the first 3 months of treatment 519 with glasdegib, the possibility that immunomodulatory effects present at later time 520 points were not captured cannot be excluded. Outside the setting of cGVHD, recent 521 research has also implicated HH signaling activation in Th17 polarization in 522 23 inflammatory bowel disease, with vismodegib treatment or genetic ablation of Ihh in 523 CD4+ T cells greatly reducing disease severity in mouse models(43). Of note, the Th17 524 compartment was not evaluated in our correlative studies. Thus, it remains to be fully 525 elucidated whether fibroblast-independent HH-mediated immune mechanisms might 526 underlie some of the clinical activity of glasdegib. 527 This study has several limitations. First, no correlative studies were performed in skin 528 samples allowing for analyses of dermal fibroblasts activity. Accordingly, 529 pharmacokinetic/pharmacodynamic (PK/PD) data were not available. Therefore, no 530 firm inferences can be made on the impact of the use of glasdegib at a dose of 50 mg 531 daily on the degree of HH pathway modulation. However, marked (>80%) 532 downregulation of GLI1 expression in the skin has been observed following treatment 533 with glasdegib at doses ≥50 mg in phase 1 trials (28)(44). Taken together with the 534 treatment-emergent safety and efficacy outcomes hereby reported, these data would 535 indicate that a relevant biological effect can be achieved at the reduced doses employed 536 in our studies. Second, caution should be exercised when interpreting our results 537 considering the uncontrolled study designs, the partial overlap between SMOi-538 associated toxicities and cGVHD manifestations, and the inherent difficulties in 539 accurately assessing clinical responses in scGVHD(30). Still, the fact that both 540 independent studies yielded consistent outcomes supports the validity of our findings. 541 In recent years, the regulatory approvals of ruxolitinib, ibrutinib and belumosudil have 542 expanded the treatment options for cGVHD. While these drugs have demonstrated 543 activity in cutaneous cGVHD, response rates for sclerotic skin and joint disease have 544 not been explicitly reported(8,45,46). Interestingly, clinical responses were seen in our 545 studies in patients with prior exposure to ruxolitinib and ibrutinib. It is thus conceivable 546 that glasdegib and other SMOi could fill a niche for some patients not responding to 547 24 tyrosine kinase inhibitors, particularly if better tolerated dosing schedules are 548 determined. Looking further into the future of SMOi in this setting, the development of 549 molecules lacking non-canonical HH signaling activating properties could allow for a 550 more effective and better tolerated therapeutic blockade of this key pro-fibrotic pathway 551 in patients with scGVHD(33). 552 553 554 555 556 557 Author contributions: 558 S.J.L. and J.A.P.S. conceived the studies and obtained funding. E.R.A., C.J.L., C.M., 559 M.A.B.R., L.L.C., A.T., M.E.H., S.S., S.J.L., J.A.P.S. enrolled patients and collected 560 data. C.J.L., L.O., S.J.L., E.R.A., and J.A.P.S. analyzed and interpreted the data; T.C.V. 561 and C.G.C performed the flow cytometric analyses. E.R.A. wrote the manuscript draft. 562 All authors revised and approved the final version of the manuscript for submission. 563 Acknowledgements: Pfizer provided glasdegib, drug distribution and funding for 564 the FHD and GETH-TC trials. The authors thank Marcie Hall and Ángel Cedillo for 565 project management of the FHD and GETH-TC clinical studies, respectively. We 566 acknowledge the excellent work by the clinical research teams at the study sites, and the 567 patients and their families for participating in our research protocols. 568 569 25 570 571 572 573 574 575 576 577 578 References 579 580 1. Zeiser R, Blazar BR. Pathophysiology of Chronic Graft-versus-Host Disease and 581 Therapeutic Targets. N Engl J Med. 2017;377(26):2565–79. 582 2. MacDonald KPA, Hill GR, Blazar BR. Chronic graft-versus-host disease: 583 Biological insights from preclinical and clinical studies. Blood. 2017;129(1):13–584 21. 585 3. Williams KM, Inamoto Y, Im A, Hamilton B, Koreth J, Arora M, et al. National 586 Institutes of Health Consensus Development Project on Criteria for Clinical 587 Trials in Chronic Graft-versus-Host Disease: I. The 2020 Etiology and 588 Prevention Working Group Report. Transplant Cell Ther. 2021;27(6):452–66. 589 4. Skert C, Patriarca F, Sperotto A, Cerno M, Filì C, Zaja F, et al. Sclerodermatous 590 chronic graft-versus-host disease after allogeneic hematopoietic stem cell 591 32 ROCKstar Study. Blood. 2021;138(22):2278–89. 732 733 734 735 736 737 738 Figure legends 739 Figure 1. Waterfall plots showing the individual maximum change from baseline in P-740 ROM score (A, GETH-TC trial; B, FHD trial) and NIH skin and/or joint tightening 741 severity score (C, GETH-TC trial; D, FHD trial). The red line indicates the 2-point 742 improvement threshold applied to the definition of response. Skin and/or joint 743 tightening severity scores were not available for two patients in the FHD trial (baseline 744 sclerotic features score <3). 745 746 Figure 2. Swimmer plots showing the timeline of glasdegib treatment, dose reductions 747 and skin/joint scGVHD responses (A, GETH-TC trial; B, FHD trial). One patient in the 748 GETH-TC trial and 1 patient in the FHD trial achieved a response at their last 749 assessment. 750 751 33 Figure 3. TCR Vβ repertoire and immunophenotypic analyses of key immune cell 752 subpopulations. A) Representative clonogram of a scGVHD patient showing clonotypic 753 dominance resulting in decreased TCR Vβ repertoire diversity. B) Flow cytometric 754 quantification of the TCR Vβ repertoire using a Gini-like diversity index. A value of 0 755 indicates equal distribution of TCR Vβ usage, and a value of 1 represents complete 756 absence of diversity (C1D1: n=12 responders [R], n=7 non-responders [NR]; C2D1: 13 757 R, 6 NR; C4D1: 12 R, 4 NR). C) Fold change from baseline in the absolute count of 758 circulating regulatory-enriched T cells (C1D1: 12 R, 7 NR; C2D1: 12 R, 7 NR; C4D1: 759 12 R, 5 NR). D) Fold change from baseline in the absolute count of circulating classical 760 (left panel), intermediate (central panel) and non-classical monocytes (C1D1: 13 R, 7 761 NR; C2D1: 13 R, 7 NR; C4D1: 12 R, 5 NR). E) Distribution of circulating CD4+ and 762 CD8+ T-cell subsets (C1D1: 13 R, 7 NR; C2D1: 13 R, 7 NR; C4D1: 12 R, 5 NR). 763 Median + individual values (B-D) and mean + SEM (E) are shown. * Omnibus 764 Friedman test. In post-hoc tests, only a statistically significant difference between visits 765 C2D1 and C4D1 was detected (p=0.007).766 34 TABLE 1. Patient, disease and transplant characteristics. 767 768 Characteristic GETH-TC cohort 1 (50 mg) (n=3) GETH-TC cohort 2 (100 mg) (n=3) GETH-TC expansion cohort (50 mg) (n=14) All GETHTC patients (n=20) FHD patients (n=15) Age, median (range), years 50 (42-53) 39 (33-60) 47 (27-73) 47 (27-73) 64 (33-74) Female sex, n (%) 1 (33) 1 (33) 3 (21) 5 (25) 8 (53) Myeloablative transplant, n (%) 2 (67) 2 (67) 8 (57) 12 (60) 5 (33) Donor type, n (%) MSD 2 (67) - 6 (43) 8 (40) 6 (40) MURD 1 (33) 3 (100) 4 (29) 8 (40) 7 (47) MMURD - - 1 (7) 1 (5) 2 (13) Haploidentical - - 3 (21) 3 (15) - Transplant indication, n (%) AML 2 (67) - 6 (43) 8 (40) 5 (33) ALL - 2 (67) 2 (14) 4 (20) 3 (20) CML 1 (7) Lymphoma 1 (33) - 2 (14) 3 (15) 3 (20) MDS 2 (13) CMPN - 1 (33) 1 (7) 2 (10) 1 (7) Other - - 3 (21) 3 (15) - ECOG score, n (%) 0-1 2 (67) 2 (67) 11 (79) 15 (75) - 2 1 (33) 1 (33) 3 (21) 5 (25) Time from transplant to cGVHD, median (range), months 6 (3-19) 5 (4-11) 10 (4-59) 9 (3-59) 7 (3-65) Time from cGVHD to enrolment, median (range), months 71 (63-78) 52 (12-106) 28 (7-65) 31 (7-106) 30 (6-101) 35 Physician-assessed global severity score, median (range) 8 (8-9) 8 (5-9) 7 (6-9) 8 (5-9) 6 (3-9) Skin/joint tightening score*, median ( ran g e ) 8 (7-9) 8 (6-9) 8 (6-9) 8 (6-9) 7 (2-9) Organ involvement, n (%) Mouth Upper/lower GI tract Eyes Lung Liver Skin Joints/fascia 2 (67) 1 (33) 3 (100) 1 (33) 1 (33) 3 (100) 3 (100) 3 (100) 1 (33) 3 (100) 1 (33) - 3 (100) 3 (100) 11 (79) 4 (29) 10 (71) 6 (43) 2 (14) 14 (100) 14 (100) 16 (80) 6 (30) 16 (80) 8 (40) 3 (15) 20 (100) 20 (100) 6 (40) 4 (27) 10 (67) 5 (33) 0 15 (100) 13 (87) Organs involved, median (range) 5 (3-6) 5 (4-5) 5 (3-6) 5 (3-6) 4 (1-6) ≥ 4 organs involved, n (%) 2 (67) 3 (100) 11 (79) 16 (80) 9 (60) Prior treatment lines, median (range) 6 (5-10) 5 (3-10) 3 (2-6) 4 (2-10) 3 (1-16) Prior treatment regimen, n (%) Ruxolitinib 3 (100) 2 (67) 5 (36) 10 (50) 9 (60) Ibrutinib - - 1 (7) 1 (5) 1 (7) ECP 3 (100) 2 (67) 9 (64) 14 (70) 8 (53) Rituximab 2 (67) 1 (33) 3 (21) 6 (30) 3 (20) Imatinib 3 (100) 1 (33) 2 (14) 6 (30) 2 (13) mTOR inhibitor 1 (33) 1 (33) 6 (43) 8 (40) 7 (47) CNI 1 (33) 3 (100) 3 (21) 7 (35) 9 (60) Corticosteroids 3 (100) 3 (100) 14 (100) 20 (100) 13 (87) Belumosudil - - - - 1 (7) MMF - - - - 5 (33) Total nodal irradiation - - - - 2 (13) Dasatinib - - - - 2 (13) Nilotinib - - - - 1 (7) MTX - - - - 2 (13) 36 Ixazomib - - - - 1 (7) Azathioprine - - - - 1 (7) Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; cGVHD, chronic graft-versus-host disease; CMPN, chronic 769 myeloproliferative neoplasm; CNI, calcineurin inhibitor; ECOG, Eastern Cooperative Oncology Group; ECP, extracorporeal photopheresis; GI, 770 gastrointestinal; MMF, mycophenolate mofetil; MMURD, mismatched unrelated donor; MRD, matched related donor; MTX, methotrexate; 771 MURD, matched unrelated donor; scGVHD, sclerotic chronic graft-versus-host disease. 772 * Among those patients with skin features score=3 (n=20 in the GETH-TC trial; n=13 in the FHD trial). 773 774 775 776 777 778 779 780 781 782 37 TABLE 2. Summary of treatment-emergent adverse events 783 784 TEAEs GETH-TC trial FHD trial Event, n (%) ≥G2 G3 G4 G3-4 G3 G4 G3-4 Muscle cramps 17 (85) 6 (30) - 6 (30) 5 (33) - 5 (33) Alopecia 10 (50) - - - - - - Dysgeusia/hypogeusia 7 (35) - - - - - - ALT/AST elevation 5 (25) 2 (10) 0 2 (10) - - - Diarrhea 5 (25) 1 (5) 0 1 (5) - - - Dry eye 5 (25) 1 (5) - 1 (5) - - - Asthenia 4 (20) 1 (5) 0 1 (5) - Lymphopenia 4 (20) 1 (5) 0 0 - GGT elevation 3 (15) 1 (5) 0 1 (5) - Myalgia/myopathy 3 (15) 2 (10) - 2 (10) 2 (14) - 2 (14) Nausea/vomiting 3 (15) 0 0 0 - - - Neutropenia 3 (15) 1 (5) 0 1 (5) - - - CK elevation 2 (10) 1 (5) 1 (5) 2 (10) - - - Thrombocytopenia 2 (10) 1 (5) 1 (5) 2 (10) - - - Weight loss 2 (10) 0 - 0 - - - Hyponatremia - - - - 2 (13) - 2 (13) 38 Hypokalemia - - - - 2 (13) - 2 (13) Breast wound - - - - 1 (7) - 1 (7) 785 For the GETH-TC study, all ≥ G2 TEAEs in ≥ 10% of patients when at least 1 event was categorized as possibly related to the study drug are 786 included. For the FHD trial, all reported ≥ G3 TEAEs are included. Toxicity grading is based on CTCAE criteria v4.0. Abbreviations: ALT, alanine 787 aminotransferase; AST, aspartate aminotransferase; GGT, gamma-glutamyltransferase; CK, creatine kinase; TEAE, treatment-emergent 788 adverse event. 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 TABLE 3. Efficacy summary. 806 39 807 GETH-TC trial (n=20) FHD trial (n=15) Efficacy endpoint* CR PR CR + PR CR PR CR + PR Skin/joint scGVHD, n (%, 80% CI) Overall response** 0/20 13/20 (65, 48-79) 13/20 (65, 48-79) 1/15 (7, 0.7-24) 6/15 (40, 23-60) 7/15 (47, 28-66) Body surface area 0/20 9/20 (45, 29-62)) 9/20 (45, 29-62) 2/15 (13, 4-32) 1/15 (7, 0.7-24) 3/15 (20, 8-39) Sclerotic features 0/20 2/20 (10, 3-24)) 2/20 (10, 3-24) 1/15 (7, 0.7-24) 2/15 (13, 4-32) 3/15 (20, 8-39) Joints/fascia 3/20 (15, 6-30) 6/20 (30, 17-47) 9/20 (45, 29-62) 1/15 (7, 0.7-24) 2/15 (13, 4-32) 3/15 (20, 8-39) P-ROM score † Skin and/or joints tightening severity‡ 3/20 (15, 6-30) 0/20 9/20 (45, 29-62) 13/20 (65, 48-79) 12/20 (60, 43-75) 13/20 (65, 48-79) 1/13 (8, 0.8-27) 2/13 (15, 4-36) 4/13 (31, 14-52) 2/13 (15, 4-36) 5/13 (38, 20-60) 4/13 (31, 14-52) Other cGVHD target organs, n (%, 80% CI) Mouth 10/16 (63, 43-79) 1/16 (6, 0.7-22) 11/16 (69, 50-84) 4/9 (44, 21-70) 1/9 (11, 1-37) 5/9 (56, 30-79) Liver 0/3 0/3 0/3 - - - Esophagus 4/6 (67, 33-91) 2/6 (33, 9-67) 6/6 (100) 1/1 (100) 0/1 1/1 (100) Upper GI 1/3 (33, 3-80) 0/3 (33) 1/3 (33, 3-80) 2/2 (100) 0/2 2/2 (100) Lower GI 2/2 (100) 0/2 2/2 (100) 3/3 (100) 0/3 3/3 (100) Lung Eye 4/9 (44, 21-70) 4/16 (25, 11-44) 0/9 1/16 (6) 4/9 (44, 21-70) 5/16 (31,11-44) 3/7 (43, 17-72) 3/10 (30, 12-55) 0/7 0/10 3/7 (43, 17-72) 3/10 (30, 12-55) CS dose reduction, n (%, 80% CI) Overall Skin/joint scGVHD responder Skin/joint scGVHD no-responder Median percent change in CS dose from baseline, % (range) Overall Skin scGVHD responder Skin scGVHD no-responder 13/16 (81, 63-93) 11/11 (100) 2/5 (40, 11-75) -52 (-84, 0) -54 (-84, -15) 0 (-84, 0) 6/7 (86, 55-99) 3/3 (100) 3/4 (75, 32-97) -23 (-64, 315) -59 (-64, -11) -13 (-51, 315) 40 Abbreviations: CI, confidence interval; CR, complete response; CS, corticosteroids; GI, gastrointestinal; PR, partial response; P-ROM, 808 photographic range of motion; scGVHD, sclerotic chronic graft-versus-host-disease; 809 * The best response while on study is shown. 810 ** Skin/joint scGVHD response was defined as skin and/or joint score improvement (≥1 point for body surface area, skin features or joint/fascia 811 scores; ≥2 points for skin/joint tightening or P-ROM scores) in the absence of worsening in those scores or an increase ≥2 points in the global 812 severity score. 813 † One patient in the GETH-TC trial with a baseline P-ROM score = 24 points achieved CR (1-point improvement) and was included among the 814 responders. Two patients in the FHD trial had a baseline P-ROM score = 25 points and were excluded from the denominator. 815 ‡ Two patients in the FHD trial had a baseline sclerotic features score <3 and were excluded from the denominator. 816 ‖ Two patients in the GETH-TC trial with altered %FEV1 at baseline were not evaluable. 817 818 819