Safety and efficacy of rituximab-based first line treatment of chronic GVHD

Abstract

Initial therapy of chronic GVHD (cGvHD) has not changed for over three decades, despite limited efficacy and long-term toxicity. We have previously shown in a small pilot study the feasibility of rituximab-based first-line therapy of cGVHD. To better assess safety and efficacy, we now evaluate 69 patients that received rituximab as part of their initial treatment. Median follow-up for surviving patients was 47 (11–81) months. Resolution of cGVHD occurred in 49 patients with median time to IST discontinuation of 349 (138–920) days. The cumulative incidence (CI) of cGHVD resolution was 41%, 69 and 77% at 1-, 2- and 3-years, respectively. No systemic corticosteroids were used in 27 patients, and 67% received ≤ 10 mg/kg cumulative exposure. Overall survival (OS) at 1-, 2- and 3-years following cGVHD diagnosis was 87, 79 and 77% respectively; corresponding rates of non-relapse mortality (NRM) were 10%, 16 and 19%. The probability of being alive and free of cGVHD at 1-, 2-, and 3-years was 36, 55, and 57% respectively. This study demonstrates the feasibility and efficacy of rituximab-based first-line cGVHD treatment. This approach demonstrates significant activity and avoids long courses of corticosteroids in most patients.

Introduction

Chronic graft-versus-host disease (cGVHD) is an important cause of late morbidity and mortality after allogeneic hematopoietic cell transplantation (alloHCT), occurring in up to 60% of long-term survivors [1–3]. The treatment of cGVHD often continues for years [4], and is a major cause of death in long-term survivors after alloHCT [5, 6]. Fur- thermore, cGVHD can significantly impact quality of life [7–10].

Conventional immunosuppressive therapy (IST) for cGVHD, using systemic corticosteroids which are tapered slowly and used with a calcineurin inhibitor, has remained the standard-of-care approach for decades despite limited efficacy and long-term toxicity. With such an approach, patients usually remain on IST for a median of 2–3 years depending on stem cell source and other factors [11]. In addition to the detrimental effects of IST therapy on immune function, prolonged corticosteroid use has multiple other adverse effects such as avascular necrosis, glucose intolerance, hypertension, weight gain, changes in body habitus, cataracts, osteoporosis, myopathy, and disturbances of mood and sleep. Unfortunately, the prolonged duration of cGVHD treatment makes it a leading cause of infection, morbidity, and late treatment-related deaths [12–16]. Therefore, new approaches to cGVHD treatment are clearly needed to achieve early control of cGVHD manifestations and facilitate tolerance.

Alternative approaches to the treatment of cGVHD may be guided by recent understanding of its pathophysiology. There is now considerable evidence implicating B cells in the development and maintenance of chronic GVHD [17–19], which has led to the hypothesis that an anti-CD20 monoclonal antibody may be an effective for the treatment of cGVHD. Subsequently, clinical trials have confirmed the efficacy of rituximab in treatment of steroid-refractory cGVHD [20–24]. Furthermore, there is evidence that the use of rituximab during the early post-transplant period may reduce the risk for cGVHD [25, 26].

We previously published the first clinical experience incorporating rituximab into the initial treatment of cGVHD [27]. That paper summarized the outcomes of a 25-patient phase II study with a median f/u time of 27 months. Rituximab-based first-line treatment of cGVHD was felt to be feasible, safe and effective. It avoided the need pro- longed exposure to systemic corticosteroids, permitted earlier discontinuation of immunosuppression, and resulted in low rates of treatment-related morbidity and mortality. Following completion of this study, rituximab was incor- porated into the upfront treatment of cGVHD at our insti- tution. To better define the safety and efficacy of this strategy, we now present our updated experience with prolonged follow-up of 69 consecutive patients treated at our institution with a rituximab-based first-line treatment approach.

Patients and methods
Patient characteristics

Institutional Review Board (IRB) approval was granted for this retrospective review of 69 consecutive patients who received rituximab as part of their first-line treatment for cGVHD, treated between April 2011 and February 2017. The cohort included the 25 patients treated on our pre- viously published phase 2 study [27]. Following the com- pletion of this study, the addition of rituximab to first-line treatment of cGVHD became our institutional standard-of- care practice. Outside of the clinical trial, rituximab use required appropriate insurance approval which became the major driver of when rituximab was started. The diagnosis of chronic GVHD was made by NIH consensus criteria [28]. Eligible patients were those with those that began rituximab within 50 days of a first episode of systemic IST- requiring cGVHD. Patients in which rituximab was given as second-line therapy due to early cGVHD progression were not eligible. Patients were excluded from this analysis if they had disease relapse prior to cGVHD diagnosis or those that received rituximab prior to the diagnosis of cGVHD.
Baseline characteristics were prospectively recorded in our institutional database, and events (occurrence and grading of initial episode of cGVHD, cGVHD response assessment, IST start and stop dates, systemic corticosteroid exposure, disease relapse, death, and cause of death) were entered into the database in real time. Assessments were performed by a dedicated GVHD nurse specialist at baseline and then at least monthly in patients with active GVHD and included medical history, physical examination, laboratory evaluation, documentation of concomitant medications, and completion of institutional GVHD assessment form. These data were retrospectively extracted from the database at the time of analysis.

Treatment plan

All patients required initiation/continuation of systemic IST (e.g. tacrolimus, sirolimus) and received rituximab induc- tion (375 mg/m2 weekly × 4). In general, if cGVHD occurred while on GVHD prophylaxis, the systemic IST was either changed (tacrolimus-to-sirolimus or vice-versa) or an additional agent was added (e.g. mycophenolate mofetil [MMF]). Stable or responding patients were eligible for rituximab maintenance (375 mg/m2 every 3 months × 4), given on the start of months 4, 7, 10, and 13 of treatment. Maintenance was planned in all 25 patients treated on our previously published phase 2 study and was continued in all stable or responding patients [27]. Outside of the clinical trial, maintenance was offered and continued at the discre- tion of the treating physician. Systemic corticosteroids were utilized in short courses for symptom control at physician discretion. Typically, methylprednisolone (MP) was started at 0.5-1.0 mg/kg/day and tapered approximately every 4 days, with the intention of discontinuing steroids in 3– 4 weeks. There were no restrictions on the use of topical therapies for ocular, oral and skin involvement.

Definitions and Study endpoints

Primary outcomes analyzed were overall survival (OS), disease-free survival (DFS, survival without evidence of active malignancy after transplantation), current chronic GVHD-free survival (CFS, a “dynamic” endpoint reflecting survival without evidence of active moderate-to-severe cGVHD at the most recent GVHD evaluation), failure-free survival (FFS, survival without relapse or need for second- line cGVHD treatment), CI of cGVHD resolution (defined as resolution of all reversible manifestations of cGVHD and discontinuation of IST) and CI of NRM. Secondary objec- tives were to estimate cGVHD recurrence after IST dis- continuation and the requirement for and cumulative exposure to systemic corticosteroids. Cumulative corticos- teroid exposure was measured as total mg/kg of MP or its equivalent during the entire course of cGVHD treatment. In addition, cumulative corticosteroid exposure was also expressed as a function of cGVHD follow-up duration (mg/ kg/month of MP or its equivalent, with months defined as the time from initial cGVHD treatment until relapse, death, or last GVHD follow-up). Acute GVHD was classified as clinically significant (grades 2–4) or severe (grades 3–4) [29]. Chronic GVHD was classified as mild, moderate, or severe by National Institutes of Health consensus criteria [28].

Statistical analysis

The Kaplan-Meier method was used to estimate prob- abilities of OS and FFS. The endpoints for FFS were treatment change, relapse and non-relapse mortality. These three endpoints were considered as competing risks in estimating the cumulative incidence of each type of failure.

Results
Characteristics of the study cohort

Patient characteristics are summarized in Table 1. The median age of patients was 54 years (range 26 -74) and 52% were male. Transplant type was matched unrelated donor, matched sibling donor and haploidentical donor in 36%, 45 and 19% respectively. PBSC was the stem cell source in 81% of patients, and conditioning was myeloablative in 54%. The median follow-up among survivors was 47 months (range, 11–81) from cGVHD onset.

The median time from transplantation to initial systemic treatment of cGVHD was 214 days (range, 74–538). The median time to rituximab initiation from the start of sys- temic IST for cGVHD was 7 days (range, 1–49). Forty-nine (71%) of patients received rituximab ≤ 14 days from the HCT-CI hematopoietic cell transplantation-comorbidity index, ALL acute lymphoblastic leukemia, AML acute myeloid leukemia, MDS myelodysplastic syndrome, MPN myeloproliferative neoplasm, CML chronic myelogenous leukemia, NHL non-Hodgkins lymphoma, HL Hodgkins lymphoma, CLL chronic lymphocytic leukemia, DRI disease risk index, NMA nonmyeloablative, MA myeloablative, RIC reduced intensity conditioning, BM bone marrow, PBSC peripheral blood stem cells, CMV cytomegalovirus serostatus start of cGVHD therapy. No patients received rituximab as part of second-line cGVHD treatment.

Chronic GVHD characteristics are summarized in Table 2. The median number of sites involved by chronic GVHD was 4 (range, 1–7), and 83% of patients had involvement of ≥ 3 sites. The sites most frequently involved were the mouth (83%), skin (78%) and eyes (72%), fol- lowed by the upper GI tract (49%), liver (41%), lower GI tract (29%) and musculoskeletal (26%). Chronic GVHD was classified as de novo, interrupted, and progressive in 42%, 36%, and 22% of patients, respectively. Median NIH symptom score at the start of cGVHD treatment was 6 (range 2-11). Initial NIH cGVHD severity score was mild, moderate and severe in 16%, 51%, and 33% of patients, respectively.
All patients received rituximab induction (375 mg/m2 weekly × 4) and required systemic IST for cGVHD. A new IST agent was introduced at the start of cGVHD therapy in 58 (84%) of patients, which included tacrolimus ± MMF in 30 patients and sirolimus ± MMF in 21 patients. In 11 patients (16%), no new systemic IST was added and instead the existing IST agent was continued (Table 3). Planned rituximab maintenance (375 mg/m2 q3mo x 4) was used in 46 patients (64%). In these patients, 32 patients (70%) received all 4 planned doses, while shorter courses (0, 1, 2, or 3 doses) were noted in 2, 6, 3 and 3 patients, respectively.

Systemic corticosteroids were used sparingly with a median total cumulative dose of 6 mg/kg (range 0–62). Median (range) corticosteroid exposure, as a measure of corticosteroid use over GVHD follow-up time was 0.12 mg/ kg/month (range, 0–7.32). Corticosteroids were started at the initiation of cGVHD therapy in 34 (49%) patients, with a plan to taper off steroids in 3–4 weeks in responding patients. An additional 8 patients, who did not receive corticosteroids at initiation, did require them later in the course due to a flare of cGVHD. No systemic corticoster- oids were required at all during the course of treatment for cGVHD in 27 (39%) patients, and 67% of patients received ≤ 10 mg/kg cumulative corticosteroid exposure (Table 3).

Overall survival (OS) at 1- and 2- and 3-years following cGVHD diagnosis was 87, 79 and 77%, respectively; cor- responding rates of non-relapse mortality (NRM) were 10, 16 and 19%. The CI of relapse was 11% at 3 years. The probability of FFS (survival free of disease relapse or initiation of second-line treatment for cGVHD) was 75, 61 and 54%, respectively at 1-, 2-, and 3-years (Fig. 2); with corresponding rates of cGVHD treatment change of 9, 16 and 21% respectively. The probability of CFS (survival free of active cGVHD) was 36, 55, and 57%, respectively at 1-, 2- and 3-years (Fig. 3). CFS increased around 5 years due to resolution of recurrent cGVHD in three patients whose initial cGVHD resolved around 1 year and recurred around recurrent cGVHD included cGVHD resolution, ongoing cGVHD and death in 6, 3 and 1 patient respectively.

Univariate and multivariate analysis

We analyzed predictors of cGVHD response including age, gender, HCT-CI, stem cell source, regimen intensity, as well as cGVHD variables such as onset type, onset year, NIH cGVHD grade, NIH cGVHD score, prior acute GVHD, sites of involvement, and choice of systemic immunosuppressant (tacrolimus vs. sirolimus) (Table 5). In univariate analysis, the achievement of cGVHD resolution (i.e. time to discontinuation of IST) was negatively corre- lated with higher cGVHD NIH score and presence of oral or ocular involvement as sites of disease. In multivariate analysis, only higher cGVHD NIH score (HR 0.81 per 1 unit increment, p = 0.033) was maintained as a significant negative predictor of cGVHD resolution.

We also analyzed predictors of cGVHD recurrence in the subset of 49 patients who resolved their cGVHD (Table 5). In multivariate analysis, predictors of cGVHD recurrence were age ≥ 55 years (HR 5.45, p = 0.019) and HCT-CI ≥ 3 (HR 3.96, p = 0.042).

Discussion

This expanded analysis with prolonged follow-up confirms the efficacy of rituximab incorporation into first-line therapy for cGVHD. We demonstrated favorable response despite the reduced used of corticosteroids, allowing more rapid withdrawal of systemic IST. The safety of this approach is demonstrated by the low NRM (10% at 1-year) and favor- able OS (87% at 1-year). The finding that corticosteroids were avoided or used sparingly in the majority of patients challenges the long-held dogma that long courses of corti- costeroids are necessary for the treatment of cGVHD. Responses to rituximab-based IST were rapid with a CI of complete IST discontinuation of 41% by 1 year and 69% by 2 years following the start of cGVHD therapy.

These results compare favorably with the historical experience with corticosteroid-based treatment, where median duration of IST treatment for cGVHD is approxi- mately 3.5 years in patients transplanted with PBSC as the stem cell source [31]. In a recent analysis of NIH consensus criteria-defined cGVHD [32], the probability of dis- continuing IST at 2, 3, and 4 years following conventional cGVHD treatment was approximately 25, 45 and 60%. At 5 years, 32% of patients remained on IST. In contrast, the 2- year probability of IST discontinuation in our study was 69%. Furthermore, the 1- and 2-yr probability of FFS in our study of 75 and 61% compares favorably to the corre- sponding figures of 45 and 29%, reported following con- ventional cGVHD therapy in a recent large multicenter observational study [33].

The morbidity and mortality associated with cGVHD is caused both by immunodeficiency and organ dysfunction directly related to cGVHD, as well as by the toxicities of the immunosuppressive medications used treat it. Long courses of corticosteroids have long served as the backbone of cGVHD therapy. Randomized trials have failed to show any benefit from the addition of other agents (azathioprine, cyclosporine, thalidomide, mycophenolate mofetil, hydro- xychloroquine) to prednisone for the initial treatment of cGVHD [34–38]. Of these, only the addition of cyclos- porine demonstrated potential utility by decreasing pre- dnisone exposure.

Long-term corticosteroid use impairs immune function, and can therefore increase the risk of opportunistic infec- tions. Furthermore, patients are exposed to numerous other corticosteroid-associated toxicities, including avascular necrosis, hyperglycemia, cataracts, osteoporosis, bone loss, muscle atrophy, fluid retention, Cushingoid appearance and disturbances in mood and sleep. Thus, strategies to reduce corticosteroid exposure would be of significant benefit to patients with cGVHD.
Rituximab is an attractive agent for the upfront treatment of cGVHD due to its favorable toxicity profile, efficacy in steroid-refractory cGVHD, and ability to serve as a steroid sparing agent in other autoimmune diseases [39–44]. Clin- ical trials evaluating rituximab in autoimmune disease suggest that rituximab may be more efficacious when given early in the disease [45–49], prior to establishment of B- and T-cell abnormalities which may be less amenable to treatment [50]. In addition to its effects on B cell numbers, treatment with rituximab has widespread effects on immu- nity including downregulation of co-stimulatory molecules [51], upregulation of regulatory T-cell number and function [52], and reconstitution of normal B- and T-cell home- ostasis [39].

Recurrence of cGVHD remains an issue after rituximab- based IST for cGVHD, with a CI of recurrent cGVHD reaching 24% at 2 years following cGVHD resolution, requiring restarting IST. However, the inability to produce durable responses in some chronic GVHD patients does not appear to be unique to rituximab-based therapy, as similar or higher cGVHD relapse rates have been reported following conventional corticosteroid-based treatment (26-61%) [32, 53]. Chronic GVHD recurrence rates are higher following PBSC compared with BM transplants (39 vs. 5%) [53].

We recognize that this study has some important lim- itations. First, due to the retrospective nature of the analysis, we do not possess important data on immune reconstitution or the incidence of infections, hypogammaglobulinemia or the use of intravenous immunoglobulin. Second, informa- tion regarding the reasons for choice of therapy for initial cGVHD therapy is not available. Third, cGVHD response criteria as proposed for use in prospective clinical trials by the NIH consensus conference [54] could not be utilized in this analysis, although the utility of such response criteria have not yet been fully validated [55]. Fourth, we do not have any data on the quality-of-life or functional status of our patients. Lastly, we acknowledge that the secondary endpoint of the study, corticosteroid exposure, could be influenced by investigator bias.
Despite these limitations, this large single-institution retrospective analysis of 69 consecutive patients receiving rituximab-based first-line treatment for cGVHD
demonstrates that this approach is both safe and effective. It reduces the need for prolonged corticosteroid treatment while permitting earlier discontinuation of immunosup- pression, which may result in reduced treatment-related morbidity associated with cGVHD and its treatment. Pre- vention of GVHD recurrence remains an important issue, and future studies are needed to better predict which patients may require additional strategies to prevent the re- emergence of cGVHD.