Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Cancer Therapy: Clinical
`
`Clinical
`Cancer
`Research
`
`A Randomized Phase II Crossover Study of
`Imatinib or Rituximab for Cutaneous Sclerosis
`after Hematopoietic Cell Transplantation
`Sally Arai1, Joseph Pidala2, Iskra Pusic3, Xiaoyu Chai4, Samantha Jaglowski5,
`Nandita Khera6, Jeanne Palmer6, George L. Chen7, Madan H. Jagasia8, Sebastian A. Mayer9,
`William A. Wood10, Michael Green11, Teresa S. Hyun4, Yoshihiro Inamoto4, Barry E. Storer4,
`David B. Miklos1, Howard M. Shulman4, Paul J. Martin4, Stefanie Sarantopoulos11,
`Stephanie J. Lee4, and Mary E.D. Flowers4
`
`Abstract
`
`Purpose: Cutaneous sclerosis occurs in 20% of patients with
`chronic graft-versus-host disease (GVHD) and can compromise
`mobility and quality of life.
`Experimental design: We conducted a prospective, multicenter,
`randomized, two-arm phase II crossover trial of imatinib (200 mg
`daily) or rituximab (375 mg/m2 i.v. weekly  4 doses, repeatable
`after 3 months) for treatment of cutaneous sclerosis diagnosed
`within 18 months (NCT01309997). The primary endpoint was
`significant clinical response (SCR) at 6 months, defined as quan-
`titative improvement in skin sclerosis or joint range of motion.
`Treatment success was defined as SCR at 6 months without cross-
`over, recurrent malignancy or death. Secondary endpoints included
`changes of B-cell profiles in blood (BAFF levels and cellular sub-
`sets), patient-reported outcomes, and histopathology between
`responders and nonresponders with each therapy.
`
`Results: SCR was observed in 9 of 35 [26%; 95% confidence
`interval (CI); 13%–43%] participants randomized to imatinib
`and 10 of 37 (27%; 95% CI, 14%–44%) randomized to ritux-
`imab. Six (17%; 95% CI, 7%–34%) patients in the imatinib arm
`and 5 (14%; 95% CI, 5%–29%) in the rituximab arm had
`treatment success. Higher percentages of activated B cells
`(CD27þ) were seen at enrollment in rituximab-treated patients
`who had treatment success (P ¼ 0.01), but not in imatinib-treated
`patients.
`Conclusions: These results support the need for more effective
`therapies for cutaneous sclerosis and suggest that activated B cells
`define a subgroup of patients with cutaneous sclerosis who are
`more likely to respond to rituximab. Clin Cancer Res; 22(2); 319–27.
`Ó2015 AACR.
`
`Introduction
`
`Cutaneous sclerosis associated with chronic graft-versus-host
`disease (GVHD) can severely affect mobility and quality of life
`
`1Department of Medicine/Division of Blood and Marrow Transplant,
`Stanford University School of Medicine, Stanford, California. 2Blood
`and Marrow Transplantation, H. Lee Moffitt Cancer Center and
`Research Institute, Tampa, Florida. 3Division of Oncology, Section of
`Bone Marrow Transplantation, Washington University School of Med-
`icine, St. Louis, Missouri. 4Clinical Research Division, Fred Hutchinson
`Cancer Research Center, Seattle, Washington. 5Division of Hematol-
`ogy, Internal Medicine, Ohio State University, Columbus, Ohio. 6Divi-
`sion of Hematology/Oncology, Mayo Clinic, Scottsdale, Arizona. 7Ros-
`well Park Cancer Institute, Buffalo, New York. 8Department of Hema-
`tology and Oncology, Vanderbilt University Medical Center, Nashville,
`Tennessee. 9Mayer Weill Cornell Medical College, New York, New York.
`10Department of Hematology/Oncology, University of North Carolina,
`Chapel Hill, North Carolina. 11Division of Hematologic Malignancies
`and Cellular Therapy, Department of Medicine, Duke Cancer Institute,
`Duke University, Durham, North Carolina.
`
`Note: Supplementary data for this article are available at Clinical Cancer
`Research Online (http://clincancerres.aacrjournals.org/).
`
`Corresponding Author: Mary E.D. Flowers, Fred Hutchinson Cancer Research
`Center, 1100 Fairview D5-290, Seattle, WA 98109-1024; Phone: 206-667-5160;
`Fax: 206-667-1034; E-mail: mflowers@fhcrc.org
`
`doi: 10.1158/1078-0432.CCR-15-1443
`
`Ó2015 American Association for Cancer Research.
`
`and is a major cause of disability and morbidity after allogeneic
`hematopoietic cell transplantation (HCT). A recent multicenter
`prospective study of 909 HCT recipients reported a 10% 2-year
`cumulative incidence of cutaneous sclerosis after HCT (1). The
`3-year cumulative incidence of cutaneous sclerosis was 20%
`among the largest reported retrospective study of 977 patients
`with chronic GVHD (2). Cutaneous sclerosis is often refractory to
`immunosuppressive therapy. Advanced cutaneous
`sclerosis
`causes joint contractures, chronic skin ulcers, pulmonary insuf-
`ficiency due to thoracic encasement, and other disabilities. Risk
`factors for cutaneous sclerosis among patients with chronic
`GVHD and the potential impact of cutaneous sclerosis on trans-
`plant outcomes have been reported (2–4). Use of a mobilized
`peripheral blood graft and total body irradiation in the transplant
`conditioning regimen were associated with an increased risk of
`cutaneous sclerosis (2, 3). No increased risk of overall mortality,
`nonrelapse mortality, or recurrent malignancy has been found in
`patients with cutaneous sclerosis compared with chronic GVHD
`patients without cutaneous sclerosis, but the development of
`cutaneous sclerosis was associated with longer time to withdrawal
`of immunosuppressive treatment for chronic GVHD (2).
`The pathogenesis of cutaneous sclerosis is not understood.
`Although cutaneous sclerosis has some clinical and histopatho-
`logic similarities with systemic sclerosis (SSc), some differences
`are noted. For instance, cutaneous sclerosis begins in the upper
`
`www.aacrjournals.org
`
`AAC-R
`
`319
`
`

`

`Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Arai et al.
`
`Translational Relevance
`
`Chronic graft-versus-host disease (GVHD) is a syndrome in
`which the contributions of inflammation, innate and adaptive
`cell-mediated immunity, humoral
`immunity, abnormal
`immune regulation, and fibrosis vary from one patient to the
`next. Cutaneous sclerosis is a form of chronic GVHD where
`fibrosis of skin and fascia predominate. In this multicenter,
`randomized, two-arm, phase II crossover trial of imatinib or
`rituximab for cutaneous sclerosis, there was a statistically
`significant (P ¼ 0.01) higher percentage of activated B cells
`(CD27þ) before treatment in the rituximab patients who had
`treatment success compared with those who did not, suggest-
`ing that activated B cells may be a good marker for patients
`with cutaneous sclerosis who will respond to rituximab. This
`relationship was not seen in imatinib-treated patients.
`Although the number of analyzed cases is small, this finding
`adds further evidence for the role of B cells in the pathogenesis
`of sclerosis in chronic GVHD.
`
`dermal layers and then extends more deeply, whereas SSc begins
`in the deeper skin layer and then extends toward the surface (5).
`Intimal hyperplasia is seen in both chronic GVHD and SSc, but
`capillary rarefaction and loss of endothelial-specific markers were
`not seen in chronic GVHD as they are in SSc (6). Still, the
`molecular stimuli for fibrosis could be similar in the two diseases.
`Stimulatory antibodies against the platelet-derived growth factor
`receptor (PDGFR) have been identified in patients with SSc and
`cutaneous sclerosis in chronic GVHD (7, 8). This observation has
`served as the rationale for testing imatinib, an inhibitor of signaling
`through PDGFR, as a treatment for cutaneous sclerosis. Imatinib
`has been reported to have clinical activity against sclerotic chronic
`GVHD (9–11). Another hypothesis is that dysregulated donor B-
`cell responses result in the sclerotic phenotype. Accumulating data
`suggest high levels of B-cell activating factor (BAFF) after allogeneic
`HCT promote the survival of allo- and autoreactive B cells and cause
`persistent activation of B-cell signaling pathways in chronic GVHD
`(12, 13). In patient B cells and in murine models, inhibition of B-
`cell signaling can prevent or reverse tissue injury caused by chronic
`GVHD (14, 15). Rituximab has broad immunoregulatory effects
`and has shown promising activity in patients with chronic GVHD as
`a B-cell–depleting therapy (16–19).
`In this prospective clinical trial targeting cutaneous sclerosis
`associated with chronic GVHD, we tested whether imatinib or
`rituximab could improve the clinical manifestations of cutaneous
`sclerosis.
`
`Materials and Methods
`
`Participants
`Participants were enrolled at 11 institutions within the Chronic
`GVHD Consortium (NCT01309997). The protocol was IRB-
`approved at each site. Informed consent was obtained in accor-
`dance with the Declaration of Helsinki. Participants were enrolled
`in the study between March 2011 and June 2014, and the data
`were analyzed as of January 31, 2015.
`Eligible patients were children or adults diagnosed within the
`past 18 months with cutaneous sclerosis after allogeneic HCT,
`with no medication added for the treatment of GVHD within the
`
`past 4 weeks. Participants were receiving corticosteroids at a dose
`greater than required for treatment of adrenal insufficiency unless
`the physician documented why steroids were contraindicated, but
`documentation of steroid dependence or refractoriness was not
`required. Cutaneous sclerosis was defined as sclerotic skin, mor-
`phea-like involvement, myofascial involvement, or joint contrac-
`tures [a Vienna Skin Score (VSS) 2 in any area (ref. 20), or
`Photographic Range of Motion (P-ROM) score of 5 or less at the
`shoulders, elbows, or wrists, or a score of 3 or less at the ankles;
`ref. 21]. Exclusion criteria included treatment with imatinib
`within the previous 6 months for any indication, treatment with
`any monoclonal B-cell antibody therapy (e.g., rituximab, ofatu-
`mumab) within the previous 12 months for any indication, and
`concomitant treatment with extracorporeal photopheresis (ECP).
`Concomitant treatment with sirolimus was also not permitted
`initially because of potential interactions with imatinib, but this
`study exclusion was removed later.
`
`Study design
`The study was designed as a prospective, multicenter, open-
`label, randomized phase II trial of imatinib (200 mg daily by
`mouth, provided by Novartis) or rituximab (375 mg/m2 intra-
`venously weekly  4 doses, repeatable after 3 months, provided
`by Genentech) for the treatment of cutaneous sclerosis. Random-
`ization was stratified by center and baseline steroid dose (<30 mg/d
`vs. >30 mg/d).
`The primary objective of the trial was to determine the clinical
`response rate of cutaneous sclerosis after 6 months of initial
`therapy with either imatinib or rituximab. The primary endpoint
`was the significant clinical response (SCR) rate at 6 months,
`defined as a 2 or more point improvement on the VSS without
`worsening elsewhere or at least a 1-point improvement in the
`4-level P-ROM scale or a 2-point improvement in the 7-level
`scale without worsening elsewhere. Crossover to the other
`study arm was allowed at 6 months if cutaneous sclerosis did
`not improve, or earlier for cutaneous sclerosis progression or
`drug intolerance. Cutaneous sclerosis progression was defined
`as a 2-point or more worsening on the VSS or a 1-point
`worsening in the 4-level P-ROM scale or a 2-point worsening
`in the 7-level scale, although crossover was also allowed for
`clinical worsening not fulfilling these criteria. Treatment suc-
`cess was defined as SCR at 6 months without crossover to the
`other arm, recurrent malignancy or death.
`Secondary endpoints of the study included in this report are the
`following: (i) the cumulative incidence of treatment failure defined
`as failure to achieve an SCR at the 6 month assessment, crossover to
`the other arm, or stopping initial treatment due to toxicity, (ii) the
`proportion of patients able to decrease their daily corticosteroid
`dose to <50% of their enrollment dose, (iii) the proportion of
`patients with any body surface area (BSA) percentage decline in
`sclerosis without BSA increase in the percentage of higher grades of
`sclerosis elsewhere according to the VSS, (iv) correlation of changes
`in patient-reported outcomes with response, and (v) correlation of
`changes in skin biopsy histology and B-cell profiles in blood
`(cytokine and cellular subsets) between responders (SCR) and
`nonresponders with each therapeutic agent.
`Clinician assessments using the VSS (Supplementary Fig. S1),
`P-ROM (Supplementary Fig. S2), and NIH chronic GVHD con-
`sensus conference scoring system (22) and patient self-reported
`outcomes (SHAQ; refs. 23, 24), FACT-BMT, Short Form 36 (SF36;
`ref. 25), Lee symptom scale (26), and health activity profile (HAP;
`
`320
`
`Clin Cancer Res; 22(2) January 15, 2016
`
`Clinical Cancer Research
`
`

`

`Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Imatinib or Rituximab for Cutaneous Sclerosis Chronic GVHD
`
`ref. 27) were performed at study enrollment and months 3, 6, 9,
`12, and 18. Clinicians were also asked to qualitatively rate
`patients' response in skin and joint chronic GVHD at 6 months
`on an 8-point scale of resolved/very much better/moderately
`better (better), a little better/stable/a little worse (stable), and
`moderately worse/very much worse (worse).
`
`Baseline and change scores in patient-reported outcomes, lab-
`oratory markers, and histopathologic grades in skin biopsy sam-
`ples were compared between treatment arms and between sub-
`groups achieving treatment success versus those that did not have
`treatment success in each treatment arm.
`
`Laboratory correlates
`Whole blood samples were drawn into ethylenediaminetetraa-
`cetic acid (EDTA) and heparin-containing tubes at study enroll-
`ment and at 6 months after initial randomization to each treat-
`ment arm or at time of cross over, whichever occurred first. Plasma
`was separated from whole blood cells by centrifugation at 600 g
`and stored at 80C until first thaw and batch testing. Soluble
`BAFF was measured using a commercially available ELISA as
`previous described (28). Fresh blood in EDTA was shipped to
`the Sarantopoulos laboratory from the study sites and analyzed
`within 36 hours. Whole blood was processed for flow cytometry
`as previously described using antibodies directed at CD3, CD19,
`and CD27. Lymphocytes were gated by size using forward and
`side scatter criteria. A minimum of 50,000 lymphocytes were
`collected for all samples to ensure adequate subset analysis. Cells
`were analyzed using BD Canto and FlowJo 10 analysis software.
`
`Histopathology correlates
`Two 3-mm skin biopsies were obtained from participants at a
`leading edge of sclerosis at study enrollment and at 6 months after
`initial randomization to each treatment arm or at time of cross
`over, whichever occurred first. The sites were the same unless there
`was a clinical contraindication. All skin biopsy slides were stained
`with hematoxylin and eosin (H&E). Two pathologists (T.S. Hyun
`and H.M. Shulman) concurrently reviewed the slides with a
`double-headed microscope blinded to all clinical details, includ-
`ing treatment for GVHD, to reach a consensus about the sclerosis
`grade from 0 to 5 according to a previously published scale used to
`assess regression of sclerosis after autologous HCT for systemic
`sclerosis (29).
`
`Statistical design and analysis
`When the study was designed, no preliminary data were avail-
`able to estimate the response rate of cutaneous sclerosis associated
`with chronic GVHD using the NIH Consensus Diagnosis Criteria
`(22). Thus, a target enrollment of 74 patients was proposed so that
`70 patients could be evaluated for the primary endpoint (35 per
`arm). With 35 patients, the proportion of SCR could be estimated
`within approximately 15% of the actual response rate at 6 months
`(primary endpoint) after treatment with each agent, based on a
`95% confidence interval. Improvement would not be expected in
`the absence of effective therapy. All participants who received
`treatment with imatinib for at least 1 week or at least one dose of
`rituximab were evaluable for the primary endpoint.
`Overall responses of cutaneous sclerosis were assessed by the
`medical provider using semiquantitative measures (see Supple-
`mentary Figs. S1 and S2) and by patients using the SHAQ, a
`validated instrument for patients with SSc (23, 24). The response
`endpoint was calculated at 6 months by comparison of baseline
`and 6-month assessments. True discordance in response (improve-
`ment in one measure while worsening in the other) was considered
`progression. Cumulative incidences of treatment failure were
`estimated by standard methods.
`
`Results
`
`Of 72 patients enrolled in this study between March 2011 and
`June 2014, 35 were randomized to imatinib and 37 to rituximab.
`The patient flow diagram is shown in Fig. 1. Table 1 displays study
`participant characteristics. The median age was 56 years (range,
`19–77), 56% were male, and all had organs other than skin
`involved with chronic GVHD at study enrollment. The median
`time from chronic GVHD onset to study enrollment was 1 year
`(range, 0–3.8 years). The median follow-up among 54 surviving
`participants is 19.5 months (range, 5.3–47.5 months) from study
`enrollment.
`
`Safety and adverse events/infections
`Adverse events observed for treatment with imatinib or rituximab
`were similar to those reported for treatment of patients with chronic
`GVHD. The grade 3 to 5 toxicities reported to be possibly, probably,
`or definitely attributed to imatinib or rituximab are shown in the
`Supplementary Table. Most events were infectious in nature, pri-
`marily respiratory or skin infections, with 2 deaths each in the
`imatinib and rituximab arms potentially attributable to the study
`drug. All 4 deaths were due to respiratory complications. In the
`imatinib arm, the deaths were caused by aspergillus pneumonia and
`parainfluenza pneumonia. In the rituximab arm, the deaths were
`caused by Pneumocystis jirovecii pneumonia in a patient who was
`receiving Bactrim prophylaxis, and aspergillus pneumonia. One
`patient in the rituximab arm had a grade 3 infusional toxicity that
`resolved with additional medication. As expected, grade 3 to 4
`neutropenia occurred more frequently in the rituximab arm.
`
`Clinical responses after initial treatment
`Disposition of study participants is shown in Fig. 1. Of 72
`participants, 61 were fully evaluable for the primary endpoint
`after initial randomization (30 in the imatinib arm and 31 in
`rituximab treatment arm) based on enrollment and 6-month
`clinician-reported data. Eleven patients did not have 6-month
`data available for the reasons detailed in Fig. 1.
`Clinical responses and other outcomes after initial randomi-
`zation to imatinib or rituximab are summarized in Table 2. SCR
`was observed in 9/35 (26%, 95% CI 13%–43%) participants
`randomized to imatinib and 10/37 (27%, 95% CI 14%–44%)
`randomized to rituximab. Among patients with SCR, 3 in the
`imatinib arm and 5 in the rituximab arm crossed over due to
`clinician-perceived lack of adequate response despite SCR. In 7 of
`these cases, improvement in one or more areas was recognized,
`but overall the response of the sclerosis was not deemed sufficient
`to continue on initial treatment. In one case, the patient was
`thought to have an SCR at 6 months but crossed over shortly
`thereafter when sclerosis worsened.
`Six (17%; 95% CI, 7%–34%) patients in the imatinib arm and 5
`(14%; 95% CI, 5%–29%) in the rituximab arm had treatment
`success defined as attaining an SCR without crossover, relapse or
`death at 6 months. Of the 35 participants randomized to imatinib, 7
`completed at least 6 months of treatment with imatinib, did not
`cross over to rituximab and remain alive; of these, two patients are
`
`www.aacrjournals.org
`
`Clin Cancer Res; 22(2) January 15, 2016
`
`321
`
`

`

`Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Arai et al.
`
`Enrollment
`
`I Randomized (n = 72)
`!
`
`I
`
`Allocation
`
`I
`
`l
`
`lmatinib (n=35)
`♦ Received allocated intervention (n= 35)
`♦ Did not receive allocated intervention (n= 0)
`
`Rituximab (n = 37)
`♦ Received allocated intervention (n = 35)
`♦ Did not receive allocated intervention (n= 0)
`
`I
`
`1•
`Discontinued intervention (n = 5)
`• Withdrew at 6 weeks (n= 1)
`♦ No response and toxicity at 2 mos (n = 1)
`Improved at 3 mos but withdrawn at 4 mos (n = 1)
`•
`• Withdrew for progression at 1 mo (n = 1)
`♦ Death at 3 mos (n= 1)
`
`Follow-up
`
`l
`'
`Discontinued intervention (n= 6)
`• Withdrew early from the study (n = 1)
`• SCR at 3 mos but died at 5 mos (n= 2)
`♦ Death at 3 mos (n= 1)
`♦ Baseline assessment only (n= 1)
`♦ No change at 3 mos then withdrew (n = 1)
`
`I
`I Evaluable for 6-mo study endpoint (n = 30) I
`
`Analysis
`
`l
`I Evaluable for 6-mo study endpoint (n= 31) I
`
`Figure 1.
`Disposition of trial participants.
`
`continuing treatment with imatinib. Of the 37 participants random-
`ized to rituximab, 10 completed one or two courses of treatment
`with rituximab, never crossed over to imatinib, and remain alive.
`The cumulative incidence of treatment failure defined as less
`than an SCR at the 6-month assessment or discontinuation of
`randomized treatment due to chronic GVHD progression or
`treatment intolerance within 6 months after initial randomi-
`zation was 65% (95% CI, 51%–83%) for patients in the
`imatinib arm and 58% (95% CI, 44%–77%) for the rituximab
`arm (Figure 2). Eleven patients (5 imatinib and 6 rituximab)
`could not be confirmed as either treatment success or treatment
`failure due to either early withdrawal for reasons other than
`cutaneous sclerosis progression or treatment intolerance, or
`lack of 6-month clinician-reported endpoint data.
`The proportion of patients at the 6-month visit able to decrease
`daily corticosteroids dose to 50% or less than the baseline dose
`was 26% (7/27) and 29% (9/32) among patients who could be
`evaluated in the imatinib and rituximab arms, respectively. The
`proportion of all patients at the 6-month visit with any percentage
`BSA decline (improvement) in total movable or nonmovable
`sclerosis without increase in the percentage of nonmovable scle-
`rosis was 47% (14/30) in the imatinib arm and 29% (9/31) in the
`rituximab arm. The proportion of patients at 6-months with
`increase (improvement) in the P-ROM in any joint without
`decreased (worsening) in other joints was 13% (4/30 evaluable
`patients) with imatinib and was 32% (10/31 evaluable patients)
`with rituximab.
`Clinicians' qualitative assessments of skin response at 6 months
`was 26% better, 52% stable, 11% worse, and 11% missing in the
`
`imatinib arm and 16% better, 54% stable, 16% worse, and 14%
`missing in the rituximab arm. For joints, clinicians reported 17%
`better, 54% stable, 6% worse, and 23% missing in the imatinib
`arm and 3% better, 73% stable, 3% worse and 21% missing in the
`rituximab arm.
`
`Clinical responses after crossover
`Among 18 patients who crossed over to the rituximab arm, 5
`experienced an SCR by 6 months after crossover, 2 have not yet
`been followed for 6 months, and 11 others either withdrew
`without response (n ¼ 2), died (n ¼ 1), or did not have an SCR
`(n ¼ 8), for a treatment success rate of 5 of 16 (31%) among those
`with at least 6 months of follow-up after crossover. Among 17
`patients who are alive and crossed over to the rituximab treatment
`arm, 10 patients have not required new treatment for cutaneous
`sclerosis at the time of this analysis. Among 23 patients who
`crossed over to the imatinib arm, 4 experienced an SCR by 6
`months after crossover, 2 have not been followed for 6 months,
`and 15 did not (4 withdrew without response, 2 withdrew due to
`toxicity, 3 died, 6 did not have an SCR), for a treatment success rate
`of 4/21 (19%) among those with at least 6 months of follow up
`after crossover. Among 14 patients who are alive and crossed over
`to the imatinib arm, 8 have not required new treatment for
`cutaneous sclerosis and 2 patients continue this treatment at the
`time of this analysis.
`
`Patient self-reported outcomes
`We evaluated whether sclerosis-related symptoms measured by
`the SHAQ standard disability index correlated with severity of
`
`322
`
`Clin Cancer Res; 22(2) January 15, 2016
`
`Clinical Cancer Research
`
`

`

`Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Table 1. Participant characteristics according to randomization
`
`Characteristic
`
`Patient age, median (range)
`Male patient, n (%)
`Female donor to male recipient, n (%)
`Advanced (high-risk) disease at transplantation, n (%)
`Conditioning regimen, n (%)
`Myeloablative
`Reduced intensity or non-myeloablative
`Graft source, n (%)
`Mobilized blood cells
`Bone marrow
`Cord blood
`Donor type, n (%)
`HLA fully matched related
`HLA fully matched unrelated
`HLA mismatched related or unrelated
`Time from transplantation to chronic GVHD, median (range), months
`Time from transplant to study enrollment, median (range), months
`Presence of GVHD sites involved at enrollment, n (%)
`Skin
`Eyes
`Mouth
`Liver
`Gastrointestinal tract
`Lung
`Joint or fascia
`Genital tract
`NIH global score at study enrollment, n (%)
`Moderate
`Severe
`Subcategory of chronic GVHD at enrollment, n (%)
`Classic
`Overlap
`Karnofsky score <80% at study enrollment, n (%)
`Prior grades II–IV acute GVHD, n (%)
`Prednisone dose at study enrollment, n (%)
`None
`<0.5 mg/kg daily
`0.5 mg/kg daily
`Other treatment of chronic GVHD at enrollment, n (%)
`Calcineurin inhibitor
`Sirolimus
`Mycophenolate mofetil
`Others
`Number of agents plus initial randomized agent, n (%)
`2
`3
`Time from onset of sclerosis to enrollment, median months (interquartile range)
`
`Imatinib or Rituximab for Cutaneous Sclerosis Chronic GVHD
`
`All patients (n ¼ 72)
`
`Initial treatment
`Imatinib (n ¼ 35)
`Rituximab (n ¼ 37)
`
`56 (19–77)
`40 (56)
`15 (21)
`16 (22)
`
`41 (57)
`31 (43)
`
`67 (94)
`3 (4)
`1 (1)
`
`24 (33)
`36 (50)
`12 (17)
`11 (0.4–82)
`29 (8–87)
`
`70 (99)
`47 (65)
`39 (54)
`23 (44)
`19 (26)
`26 (37)
`65 (90)
`11 (16)
`
`19 (26)
`53 (74)
`
`14 (19)
`58 (81)
`30 (44)
`31 (46)
`
`8 (12)
`43 (64)
`16 (24)
`
`36 (50)
`7 (10)
`6 (8)
`29 (40)
`
`56 (19–72)
`18 (51)
`7 (20)
`9(26)
`
`26 (74)
`9(26)
`
`32 (94)
`1 (3)
`1 (3)
`
`15 (43)
`12 (34)
`8 (23)
`11 (0.7–82)
`31 (8–87)
`
`34 (100)
`21 (60)
`20 (57)
`11 (46)
`10 (29)
`11 (31)
`31 (89)
`7 (20)
`
`11 (31)
`24 (69)
`
`8 (23)
`27 (77)
`14 (42)
`17 (55)
`
`6 (19)
`18 (58)
`7 (23)
`
`19 (54)
`2 (6)
`4 (11)
`14 (40)
`
`56 (21–78)
`22 (59)
`8 (22)
`7(19)
`
`15 (41)
`22 (59)
`
`35 (95)
`2 (5)
`0 (0)
`
`9 (24)
`24 (65)
`4 (11)
`11 (0.4–44)
`27 (14–61)
`
`36 (97)
`26 (70)
`19 (51)
`12 (43)
`9 (24)
`15 (42)
`34 (92)
`4 (11)
`
`8 (22)
`29 (78)
`
`6 (16)
`31 (84)
`16 (46)
`14 (39)
`
`2 (6)
`25 (69)
`9 (25)
`
`17 (46)
`5 (14)
`2 (5)
`15 (41)
`
`65 (90)
`7 (10)
`1.8 (0.5–5.7)
`
`30 (86)
`5 (14)
`1.6 (0.2–6.1)
`
`35 (95)
`2 (5)
`2.3 (0.6–4.1)
`
`cutaneous sclerosis by clinical findings and response to study
`treatment. The SHAQ score did not correlate with the percentage
`of total body surface with movable or nonmovable sclerosis using
`the VSS, but did correlate with total P-ROM (Spearman correla-
`tion coefficient 0.41, P ¼ 0.001). Compared with enrollment, 11
`evaluable patients had their SHAQ standard disability index
`decrease by at least 0.2 U, which is considered a clinically mean-
`ingful difference, but improvement in the SHAQ was not corre-
`lated with treatment success in either arm.
`Changes in other patient-reported outcomes were correlated
`with treatment arms and treatment success. The only significant
`difference at P < 0.01 was a median 10-point decrease (range 55
`to þ25) to P ¼ 0.001 for the Lee skin symptom scale for the
`imatinib arm. There were no differences in the other Lee subscale
`scores, the SF-36, FACT-BMT, or HAP for the imatinib arm, and no
`statistically significant changes for any of these scales in the
`
`rituximab arm. The correlation of changes in patient-reported
`outcomes and treatment success were evaluated for 28 patients in
`the imatinib arm (6 treatment successes) and 23 in the rituximab
`
`Table 2. Summary of overall clinical results
`
`Outcome
`
`Initial randomization
`Imatinib
`Rituximab
`n ¼ 35
`n ¼ 37
`
`9 (26)
`6 (17)
`
`Significant clinical response (SCR), n (%)
`Treatment success: SCR without crossover,
`relapse or death at 6 months
`29
`Treatment failure at 6 months, n
`No SCRa, n
`21
`Crossover to other arma, n
`18
`Not evaluableb, n
`5
`aTotals >100% because reasons are not mutually exclusive.
`bSee disposition of study participants shown in Fig. 1.
`
`10 (27)
`5 (14)
`
`32
`21
`23
`6
`
`www.aacrjournals.org
`
`Clin Cancer Res; 22(2) January 15, 2016
`
`323
`
`

`

`Published OnlineFirst September 16, 2015; DOI: 10.1158/1078-0432.CCR-15-1443
`
`Arai et al.
`
`C)
`
`ro
`ci
`
`c.q
`0
`
`N
`ci
`
`0
`ci
`
`0
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`Months
`
`Figure 2.
`Cumulative incidence of treatment failure defined as discontinuation of
`randomized treatment due to chronic GVHD progression or treatment
`intolerance within 6 months after initial randomization or less than an SCR at
`the 6-month assessment according to treatment arm.
`
`arm (3 treatment successes) with 6-month patient-reported out-
`comes. Treatment success with imatinib was associated with
`improvement in the SF-36 vitality score (6.2 points vs. 3.1
`points, P ¼ 0.01) and the Lee lung symptom score (5.0 point
`improvement vs. 5.0 point worsening, P ¼ 0.005). There were
`no differences in the other patient-reported measures, includ-
`ing skin bother, and no differences correlated with rituximab
`treatment successes.
`
`Laboratory and histopathology correlates
`B cells play a role in the development of chronic GVHD, and
`CD27þ B cells are constitutively activated in patients with this
`disease (13, 30). For this study, a detailed analysis of plasma BAFF
`levels and B-cell phenotype was carried out at time of enrollment
`and 6 months after the initiation of therapy or crossover. Figure 3A
`shows activated B-cell percentages (CD27þ) at enrollment strat-
`ified by patients who had treatment success compared with those
`who did not in the imatinib and rituximab study arms. There was a
`
`statistically significant (P ¼ 0.01) higher percentage of CD27þ B
`cells at the time of enrollment in the rituximab patients who had
`treatment success (n ¼ 3) compared with those who did not. The
`absolute number of B cells was not statistically different among all
`of the treatment arms, but the power of the analysis in the
`rituximab arm was very limited due to sample size (Supplemen-
`tary Fig. S3). Figure 3B provides representative flow-cytometry
`analyses for the rituximab arm responders and nonresponders.
`Cells were initially gated for CD19 positivity, and from this
`population CD27-positive cells were plotted by granularity (side
`scatter). BAFF levels at time of enrollment were similar among the
`treatment arms (Supplementary Fig. S4), and were not correlated
`with steroid use in univariate analysis.
`At enrollment, the median histopathologic sclerosis score was 2
`(range, 0–5) in both the imatinib arm (n ¼ 32) and the rituximab
`arm (n ¼ 31) participants with evaluable biopsies. Histopatho-
`logic grading of sclerosis at enrollment was higher in the 6 patients
`who had treatment success with imatinib compared with the 22
`who did not (3.5 vs. 1.4, P ¼ 0.001). There was no difference in
`sclerosis histopathologic grade at enrollment in the 4 patients
`with treatment success on rituximab arm compared with the 22
`who did not. Treatment success was not correlated with change in
`the sclerosis grade on the posttreatment biopsies in either treat-
`ment arm, although numbers for analysis are very limited because
`of missing post treatment biopsies.
`
`Discussion
`
`Management of cutaneous sclerosis chronic GVHD is difficult
`(31, 32), with evidence limited to reports from uncontrolled
`single-arm studies of second-line or subsequent
`treatments
`(10, 11, 17, 33, 34), retrospective studies (19, 35), or reviews
`(36, 37). In the absence of a definitive understanding of the
`pathophysiologic mechanisms for development of cutaneous
`sclerosis, new treatments have relied on empirical testing of agents
`approved for other indications where inflammation, abnormal
`immune regulation, or fibrosis have been implicated as patho-
`genic mechanisms.
`We embarked on this study to estimate the efficacy of imatinib
`and rituximab in parallel arms in a multicenter study. This study
`applied stringent, semiquantitative metrics for the assessments of
`skin thickness and range of motion to evaluate success of either
`treatment at 6 months. The rates of SCR and treatment success
`were low (27% and 17%, respectively) at 6 months after initial
`treatment of cutaneous sclerosis with either imatinib or rituxi-
`mab. Our definitions of SCR were purposefully conservative to
`try to minimize observer bias and increase reliability, but we
`
`Figure 3.
`Proportion of CD27þ B cells at
`enrollment across treatment groups.
`A, percentage of CD27þ cells of the
`CD19þ parent gate in patients who
`later had treatment success
`(responder) or did not
`(nonresponder) to either imatinib or
`rituximab. B, representative flow
`cytometry showing baseline CD27
`expression of CD19þ (CD3) B cells.
`
`B
`
`Rituximab responders
`
`7.6%
`
`9.8%
`
`15.7%
`
`Rituximab nonresponders
`
`SSc
`
`-,
`
`3.4%
`
`3.1%
`
`0.18%
`
`■ ti-f
`
`P = 0.01
`
`A
`
`40
`
`30
`
`20
`
`10
`
`0
`
`Percent CD27+ B