Combination Therapy for GVHD Prophylaxis with Etanercept and Extracorporeal Photopheresis: Results of a Phase II Clinical Trial

Abstract

Reduced intensity conditioning (RIC) regimens minimize early toxicity after allogeneic hematopoietic cell transplant (HCT) by placing greater reliance on establishing a graft-versus-leukemia effect (GVL). Because graft-versus-host disease (GVHD) and GVL are tightly linked, inhibition of T-cell populations that cause GVHD may lead to an unintended increased risk of relapse in the RIC setting. Although not completely understood, etanercept and extracorporeal photopheresis (ECP) are thought to ameliorate GVHD without direct T cell inhibition. We hypothesized that adding these two agents to a standard GVHD prophylaxis regimen of tacrolimus and mycophenolate mofetil (MMF) would improve survival by reducing GVHD related mortality without increasing relapse rates. Therefore, we conducted a prospective phase II clinical trial that incorporated tacrolimus, MMF, etanercept, and ECP as GVHD prophylaxis in 48 patients undergoing RIC unrelated donor transplant (URD HCT). The preferred RIC was fludarabine 160 mg/m² + busulfan 6.4-12.8 mg/kg +/- TBI 200 cGy. Etanercept 0.4 mg/kg (max dose 25 mg) was given subcutaneously twice weekly for eight weeks post-HCT and ECP was given for 12 treatments, starting weekly on day 28 weekly and tapering off by day 180. The median age of the study patients was 60 (18-71) years. Donors were 7/8 (n=14, 29%) or 8/8 (n=34, 71%) HLA-matched. All patients engrafted neutrophils at a median of 12 days. The cumulative incidence of grade II-IV acute GVHD at day 100 was 46%, but it was typically sensitive to initial steroid treatment (84% day 56 CR/PR rate). Overall survival at one year in this older, frequently mismatched unrelated donor setting was excellent (73%) due to low rates of non-relapse mortality (21%) and relapse (19%). However, this strategy was not effective at preventing a high incidence of chronic GVHD and late deaths led to a drop in two-year survival declined to 56% reflecting a high incidence of chronic GVHD.

Introduction

Allogeneic hematopoietic cell transplant (HCT) is increasingly used as a curative option for patients with hematologic malignancies as the expansion of reduced intensity conditioning regimens (RIC) has allowed for the transplantation of patients who are older and those that have more co-morbidities. According to the National Marrow Donor Program (NMDP) there was greater than a four-fold increase in the number of patients over the age of 60 receiving an unrelated donor (URD) HCT from 2005-2009 compared to 2000-2004.(1) Although utilization of transplant has increased, there has not been a reduction in the rates of GVHD, and the three-year survival remains unsatisfactory, around 35%.(1) The major reasons for treatment failure are relapse and graft-versus-host disease (GVHD), the latter of which leads to significant morbidity and mortality. As acute GVHD severity worsens, so does survival, as demonstrated by previous studies that show limited mortality following grades I-II acute GVHD, but mortality rates approach 75% with grade III-IV acute GVHD.(2) The large difference in outcomes can be explained by the higher rates of steroid-resistant GVHD seen with more severe GVHD.(3-5) However, GVHD severity cannot be predicted before transplant and as a result, prevention strategies are designed to limit its occurrence, rather than its severity.

Traditional GVHD prevention strategies rely on combinations of immunosuppressants that block T-cell expansion and cytotoxicity. A common GVHD prophylaxis regimen following reduced intensity conditioning (RIC) includes the calcineurin inhibitor tacrolimus, plus mycophenolate mofetil (MMF), and results in rates of GVHD grades II-IV of 54-79%.(6-8) However, further augmenting GVHD prophylaxis with additional agents that directly target T-cells, such as anti-thymocyte globulin, increases the risk of relapse, graft failure, and/or infection, which may be especially relevant in the context of RIC(9-11). We have previously explored whether we could offset the risks from GVHD and its treatment by adding agents to the prophylaxis regimen that had indirect effects on T-cell function. In our prior study, tumor necrosis factor alpha (TNF-α) blockade with etanercept was given from admission through day 56 post-HCT to a standard tacrolimus and methotrexate backbone in the myeloablative setting. This approach did not reduce the cumulative incidence of GVHD grade II-IV from 45%, but an exceptionally high proportion (93%) of those patients that required treatment for GVHD achieved a complete response (CR)(12), which compared favorably to the expected CR rates of 40-50%.(13, 14) These results led us to speculate that TNF-inhibition during the early post-HCT period increased the steroid-responsiveness of GVHD when it developed.

Extracorporeal photopheresis (ECP) is an immunomodulatory approach that appears to ameliorate GVHD without increasing relapse and infection rates.(15-21) ECP has not been well studied for GVHD prophylaxis, but in one study, ECP treatment prior to HCT, followed by cyclosporine, methotrexate and MMF, resulted in high rates of donor engraftment (98%), low rates of non-relapse mortality (NRM) at day 100 (11%) and unexpectedly low rates of acute GVHD (9%).(22) The effectiveness of ECP delivered post-HCT for the prevention of acute GVHD has not been studied.

Given the extensive data to support the importance of TNF-α in the initiation of the GVHD reaction,(23-25) and the potential for additional immune modulation via ECP therapy, we decided to study standard GVHD prophylaxis (tacrolimus and MMF) in combination with etanercept and ECP in older patients receiving a reduced intensity conditioning unrelated donor HCT. Our hypothesis was that augmenting GVHD prophylaxis in this way would reduce the incidence of steroid refractory GVHD and NRM, yet preserve the graft-versus-leukemia effect, thereby leading to improved overall survival (OS). To further study this question, we explored whether a novel endpoint of steroid-free, relapse-free survival at six months post-HCT could serve as a surrogate endpoint for long-term survival in a GVHD prophylaxis study.

Methods

Study Population

The study was designed to target patients at high risk for non-relapse mortality following unrelated donor HCT. Patients who lacked related donors were eligible if they were older than 50 years or if they had co-morbid conditions that precluded intensive conditioning regimens regardless of age. Single allele level HLA mismatches (i.e. 7/8 matches) of the HLA-A, B, C, and DRB1 loci by high resolution typing were allowed when peripheral blood (PBSC) or bone marrow (BM) were used. Cord blood units were required to match for at least 4/6 loci (intermediate resolution typing for the HLA-A and B loci; high resolution typing for HLA-DR). Patients with infections not responsive to treatment or who were unlikely to tolerate the fluid shifts associated with ECP were ineligible. The protocol and informed consents were approved by the Institutional Review Board at the University of Michigan. All patients gave informed consent per the Declaration of Helsinki.

Study Design

The study was conducted as an open label, non-randomized phase II clinical trial (registered at ClinicalTrials.gov; NCT00639717). Pre-HCT conditioning was selected according to institutional practice given the underlying disease, previous therapy, and comorbidities provided the regimen was recognized as reduced intensity or reduced toxicity according to the established literature.(26) The preferred regimen was fludarabine + intravenous busulfan.(27-30)

The GVHD prophylaxis schema is shown in Figure 1 and consisted of a standard backbone supplemented by the investigational agents. The backbone consisted of the widely used regimen of tacrolimus and mycophenolate mofetil (MMF).(31, 32) Tacrolimus was begun on day -3 prior to HCT, titrated to a therapeutic trough level of 8-12 ng/ml, and tapered by 25% per month starting 56 days post-HCT in the absence of GVHD. Cyclosporine was substituted for patients who could not tolerate tacrolimus. MMF 10 mg/kg/dose (max dose 1 gm) was administered either orally or intravenously every 8 hours from day 0 through day 28.

Figure 1. Graft versus Host Disease Prophylaxis Schema.

The investigational agents were administered on an overlapping schedule such that patients received at least one study agent continuously through the first 180 days post-HCT. Etanercept was given subcutaneously twice weekly (0.4 mg/kg, max dose 25 mg) at least 72 hours apart, from day 0 to day 56 for a total of 16 doses. To mitigate risk of infectious related complications, etanercept was held for persistent bacteremia (>1 positive blood cultures on separate days) until appropriate antimicrobial treatment was started; hemodynamic instability until resolved for 24 hours; newly diagnosed fungal or mycobacterial infection until 7 days of effective treatment were administered; rising cytomegalovirus viral copy number despite appropriate treatment, or fever ≥102F daily for >5 days until afebrile (< 100.5) for 24 hours. Etanercept was not held for uncomplicated fever <102F. Missed doses of etanercept were not made up unless due to patient error. Failure to receive all 16 doses did not require removal from the study.

For practical reasons, extracoporeal photopheresis could not be administered prior to white blood cell engraftment. Institutional minimum criteria for hematocrit (>28%) to ensure an adequate separation of red blood cells, white blood cells and plasma and for platelet count (>35,000/μL) to mitigate risk of bleeding when heparin was used for anticoagulation during ECP treatments were followed. To minimize the need for transfusions to meet these criteria, once weekly ECP was scheduled to begin no earlier than day +28 (± 5 days) for 7 treatments until day +70, then tapered to every other week × 2 treatments, then monthly until discontinuation at day +180 (2-3 treatments). Depending on the start and end dates, patients received 11-12 total ECP treatments. All patients received ECP treatment with the Therakos™ UVAR XTS® photopheresis system; according to the manufacturer's instructions. A treatment consisted of collection of leukocytes, methoxypsoralen incubation, photoactivation, and reinfusion of activated cells via intravenous access. ECP could be given more often at the discretion of the treating physician in the event of GVHD and remained on study to be assessed for primary and secondary endpoints. Transfusion and/or cytokine support was administered when necessary to allow ECP treatment. To mitigate risks associated with ECP delivery, treatments were not given within 72 hours of a positive blood culture, within 24 hours of fever >100.5F or hemodynamic instability, within 12 hours of clinically significant bleeding, or on the same day as surgical procedures except bone marrow biopsy and other low bleeding risk procedures. Missed treatments were made up within one week. Patients who received <2 ECP treatments were deemed inevaluable, removed from the study, and replaced. Physicians were allowed, but not required, to treat GVHD with increased frequency of ECP treatments.

Infections

Infections were identified based on positive blood or body fluid cultures or by detection of viral DNA or RNA in plasma or body fluid by quantitative PCR. Severity was graded according to Blood and Marrow Transplant Clinical Trials Network (BMT CTN) Technical Manual of Procedures, Version 3.0.(33) Grade 3 infections were defined as bacteremia with deep organ involvement; severe sepsis with bacteremia; fasciitis requiring debridement; pneumonia requiring intubation; brain abscess or meningitis without bacteremia; Clostridium difficile toxin positive stool with toxic dilation; fungemia including Candidemia; Pneumocystis jiroveci pneumonia; proven or probable invasive fungal infections; disseminated infections with Histoplasmosis, Blastomycosis, Coccidiomycosis or Cryptococcus; severe varicella zoster virus infection; cytomegalovirus (CMV) infection with end-organ involvement; Epstein-Barr virus (EBV) associated post-transplant lymphoproliferative disorder; adenovirus with end-organ involvement; lower tract respiratory viruses; any viral encephalitis/meningitis; end-organ toxoplasmosis.

GVHD Scoring and Treatment

A limited number of observers scored acute GVHD according to the modified Glucksberg criteria.(34) The diagnosis of GVHD was made clinically, but confirmatory biopsies of affected organs were usually obtained. GVHD was graded weekly until day +100 or resolution of acute GVHD, whichever occurred later. Scores were reviewed by an independent investigator and discrepancies were adjudicated by a study investigator (CK), when necessary. Grade II-IV acute GVHD was initially treated with 2 mg/kg/day of methylprednisolone or prednisone equivalent, then tapered at the discretion of the treating physician. Patients continued study treatment after development of acute GVHD except where toxicities precluded doing so. Treating physicians were permitted to increase ECP frequency or add additional treatments at their discretion. Chronic GVHD was assessed according to the National Institutes of Health Consensus criteria.(35)

Primary and Secondary Endpoints

We reasoned that a novel endpoint (alive, in remission, and on no more than low dose steroids -4mg methylprednisolone or equivalent for treatment of acute GVHD) at 6 months post-transplant could serve as an acceptable surrogate for long-term survival for patients undergoing reduced intensity URD. We calculated that a sample size of 48 evaluable subjects would provide 80% power to detect an improvement in the proportion of patients who met all these criteria for success from 40% (the historical rate in patient who met study eligibility criteria at our center) to 60% with an alpha level of 0.05. Secondary endpoints included the incidence of acute GVHD grades 2-4, the incidence of steroid-refractory GVHD, the incidence of chronic GVHD, NRM and overall survival.

Statistical Analysis

Continuous patient characteristics were summarized with medians and ranges, and categorical patient characteristics were summarized with proportions. Overall survival was estimated using Kaplan-Meier methods and the cumulative incidence of GVHD, relapse, and non-relapse mortality were estimated using Gray's method. All analyses were done in the statistical package R, Version 3.0.1.

Results

Study Feasibility and Population

The study enrolled 52 patients from April 2, 2009 to July 11, 2012 in order to obtain 48 patients who received at least 2 ECP treatments and were therefore evaluable. Thus, the study treatment plan was feasible in 92% of all consented patients. Per study design, four patients were removed from study treatment secondary to failure to receive ECP; primary graft failure (1), non-compliance (1), severe veno-occlusive disease with thrombocytopenia that precluded initiating ECP (1), and early relapse (1). These patients did not receive any ECP treatments and were excluded from this analysis. Table 1 shows the baseline demographic and clinical characteristics of the 48 evaluable patients. By design, the study population was composed of patients at high risk for NRM. The primary risk factors were advanced age (median age 60 years, range 18-71 years), high (n=28, 58%) or intermediate (n=16, 34%) HCT comorbidity index(36, 37), and frequent use of HLA-mismatched donors (n=14, 29%). All but two patients received peripheral blood stem cells; there was one bone marrow transplant and one cord blood transplant.

Table 1. Patient clinical characteristics.

Median age of recipient (range)	60 years (18-71)
Female	22 (46%)
Diagnosis
Acute myeloid leukemia	19 (40%)
Myelodysplastic syndrome	12 (25%)
Non-Hodgkin lymphoma	8 (17%)
Acute lymphoblastic leukemia	4 (8%)
Myeloproliferative disorder	3 (6%)
Chronic lymphocytic leukemia	1 (2%)
Multiple myeloma	1 (2%)
Disease status
Low	19 (40%)
Intermediate	18 (37%)
High	11 (23%)
Comorbidity Index
Low (0)	4 (8%)
Intermediate (1 or 2)	16 (34%)
High (≥ 3)	28 (58%)
HLA matching *
8/8 HLA matched unrelated donor	34 (72%)
7/8 HLA mismatched unrelated donor	13 (28%)
Cytomegalovirus status
R- D-	18 (38%)
R- D+	2 (4%)
R+ D-	17 (35%)
R+ D+	11 (23%)
CD34+ count (106 cells per kg)*	5.45 (0.7 – 9.8)

•Excludes 1 cord blood recipient that received two 5/6 HLA mismatched cord blood units.

Engraftment, chimerism, and toxicities

We did not find evidence that etanercept or ECP affected neutrophil or T-cell engraftment. The median time to neutrophil engraftment of 12 days (range 8 to 26 days) was consistent with expectations.(38-40) The median donor chimerism for CD3+ T-cells at day 30 was 93% (range 48% to 100%) and increased to 100% at one year (range 91% to 100%). Exact date of platelet engraftment was difficult to determine based on need for platelets transfusions in order to perform ECP in some patients.

Both etanercept and ECP were generally well tolerated. No deaths were attributed to study treatment. An early lymphoma relapse at day 99 post-HCT was considered possibly related to study treatment because the etanercept package insert describes the development of lymphoma in non-transplant patients treated with the drug. Two patients who experienced chest pain without EKG changes during their 6th and 8th ECP procedure respectively were removed from study treatment as a precaution. No etiology for the chest pain was identified. No other toxicities were considered to be ECP related. Two early deaths (sudden death at home on day 151, five days after an ECP treatment and a fatal intracranial hemorrhage on day 55 from profound thrombocytopenia eight days after ECP treatment) were deemed unrelated to study treatment based on their timing. Other CTC grade 3-4 toxicities are summarized in Table 2 but none of these toxicities were considered related to the study interventions by the investigators or after review by the University of Michigan Cancer Center Data and Safety Monitoring Board.

Table 2. All CTC grade 3-4 non-hematologic toxicities.

Event	Number
Cardiac events	5 (2 episodes atrial fibrillation, 1 each syncope, orthostatic hypotension, NSTEMI)
Deep vein thrombosis	4
Altered mental status secondary to calcineurin inhibitor	3
Musculoskeletal abnormality	3 (1 each – daptomycin associated rhabdomyolysis, trauma associated ankle pain, steroid myopathy)
Electrolyte abnormality	2
Acute cholecystits	1
Depression	1
Diverticulitis	1
Evan's syndrome	1
Headache	1
Hematemesis	1
Renal calculi	1

We considered the possibility that the addition of etanercept and ECP would increase infection risk and therefore patients were carefully monitored for signs and symptoms of infection at least weekly. The majority of patients (34/48, 71%) developed at least one infection that required treatment during the 180 day study period (Table 3). While the exact contribution of etanercept and ECP to the risk of infection is difficult to quantify, the incidence of grade 3 infections observed on study was similar to that reported in the literature(12, 41), and similar to the rates of infection that occurred in non-study patients during approximately the same time as the study period (unpublished data). Two patients died from infections – one from HHV6 encephalitis on day 83 and one from fungal pneumonia (Aspergillus fumigatus and Rhizopus) on day 88. We observed twelve Grade 3 non-fatal infections in 7 (15%) patients, 4 of who were also on systemic steroid treatment for GVHD at the time, which is a well-known major risk factor for post-HCT infection.(42, 43) Grade 3 viral infections developed in three patients (all CMV end organ disease, complicated in one case by simultaneous human herpesvirus 6, adenovirus and EBV infection). There were four grade 3 bacterial infections (one each of Streptococcus mitis sepsis, Corynebacterium striatum septic joint hardware, Achromobacter pneumonia, Staphylococcus aureus pneumonia). In addition to the fatal fungal pneumonia, there were two cases of non-lethal grade 3 fungal pneumonia (one case of Aspergillus fumigatus and one probable case based on radiographic findings and positive galactomannen result from BAL). Study treatment was held per study design for severe infections and then resumed in patients who had an appropriate response to therapy. All other infections were either grade 1 or 2 and resolved with appropriate antimicrobial therapy.

Table 3. Infectious Complications.

Type of Infection	Grade I	Grade II	Grade III
Bacterial1	21	17	4
Viral2	10	22	6
Fungal	--	--	2
Acid fast bacilli3	--	3	--

¹Grade 1: Bacteremia with coagulase negative Staph or Corynebacterium, uncomplicated urinary tract infections; Grade 2: Bacteremia other than grade 1 and without severe sepsis.

²Grade 1: HSV mucositis, asymptomatic CMV viremia with appropriate viral load decline with therapy; Grade 2: EBV reactivation treated with Rituximab, documented viral upper respiratory infection, symptomatic CMV or CMV not responding to appropriate therapy within 14 days, BK viremia or viruria requiring therapy or surgical intervention.

³Grade 2: AFB recovered from BAL or bacteremia.

ECP appeared to have little impact on transfusion requirements. Study subjects received a median of three packed red blood cell transfusions and one pooled platelet transfusion during the 5 months of ECP administration.

GVHD

The median day of onset of GVHD II-IV was 47 days (range 9-201) and GVHD occurred prior to the initiation of ECP in ten patients. The cumulative incidence of grade II-IV GVHD by 6 months post-URD HCT ranges from 50 to 59%,(44, 45) and as expected the cumulative incidence in this study fell into that range (57%, 95% CI 42%-71%; Figure 2A), with 19% experiencing severe GVHD (overall grade III-IV) by 6 months (95% CI 8-30%). However, as hoped for, we observed a predominance of moderate GVHD (overall grade II; 20/29), and these cases of GVHD were highly steroid sensitive (day 56 CR/PR = 84. Given the steroid sensitive nature of the GVHD, it is not surprising that the median steroid dose reduction by day 56 was 88% (range 65-94%). Steroid refractory (SR) GI GVHD is the primary cause of NRM after HCT.(46) The cumulative incidence of SR GI GVHD was 23% by 6 months post-HCT (95% CI 13-38%; Figure 2B), which may explain the low incidence of NRM at one year (21%, 95% CI 9-32%; Figure 2C). Notably, attenuation of GVHD severity did not appear to impact the GVL effect given the low cumulative incidence of relapse at 1 year was 19% (95% CI 8-30%; Figure 2D). Thus, low rates of lethal acute GVHD and relapse contributed to excellent one-year survival in this high-risk population (73%; 95% CI 61-87; Figure 2E). Unfortunately, the apparent survival benefit was not durable as by 2 years survival declined to 56% (95%CI 44-72%).

Figure 2.

A) Cumulative incidence of acute GVHD at day 100 was 46% (95% CI 32-60) and at D180 was 57% (95% CI 42-71). B) Cumulative incidence of steroid refractory GI GVHD at D180 was 23% (95% CI 13-38%). C) Cumulative Incidence of non-relapse mortality at one year was 21% (95% CI 14-27%) and at 2 years 27% (95% CI 14-40). D) Cumulative Incidence of relapse at one year was 19% (95% CI 8-30) and at 2 years was 21% (95% CI 9-32). E) Overall Survival at one year was 73% (95% CI 61-87) and at 2 years was 56% (95% CI 44-72). F) Cumulative Incidence of chronic GVHD at one year was 42% (95% CI 27-56) and at 2 years was 58% (95% CI 44-73%).

Chronic GVHD was a major contributor to deaths that occurred after one year (Table 4). The median day of onset of chronic GVHD was 209 days (range 99-1248), which was shortly after the study intervention completed. The cumulative incidence of moderate-to-severe chronic GVHD according the NIH consensus criteria(35) at 1 and 2 year was 42% (95% CI 27-56) and 58% (95% CI 44-73) respectively (Figure 2F).

Table 4. Causes of death after one year post-HCT.

Day of Death post-HCT	Primary Cause	Secondary Cause
419	cGVHD	Fungal pneumonia
457	cGVHD	Fungal pneumonia
563	cGVHD	Failure to Thrive
795	cGVHD	Pulmonary Failure
915	cGVHD	Pulmonary Embolism

The proportion of patients who were alive, in remission, and not requiring steroid therapy for treatment of acute GVHD (defined as more than 4mg methylprednisolone or equivalent) at 6 months post-URD HCT was 58% (n=28), which was below the 60% required to demonstrate a statistically significant improvement over historical control rate of 40% to support our novel primary end point. When use of steroids for any cause was used to redefine endpoint failure, the rate of success fell to 42% (n=20), reflecting a high proportion of patients receiving treatment for chronic GVHD at 6 months post-HCT. The proportion of patients who had discontinued steroid treatment was 57% at one year post-HCT and 60% at two years post-HCT.

Discussion

The majority of GVHD-related deaths occur in allogeneic HCT recipients who develop steroid-refractory or steroid-dependent GVHD.(5, 47) Older and/or infirm recipients, especially of unrelated donor HCT, are particularly susceptible to the toxic effects of steroid treatment and in need of new GVHD prevention strategies that can decrease steroid exposure without further suppressing T-cell activity which could compromise GVL effect and increase the incidence of graft failure and severe infections. The 73% one-year survival in the high risk study population of older and infirm patients, who underwent unrelated donor HCT, often mismatched, compares favorably to the 53% one year survival reported for a large series of recently transplanted patients >60 years facilitated by the National Marrow Donor Program.(1) Unfortunately, the possible benefit from the investigational approach tested in this study appeared to be relatively short-lived as a significant number of deaths continued to occur after one year post-HCT.

We speculate that the possible early survival advantage we observed may be related to the high incidence of steroid responsive GVHD, which was almost double that expected.(13, 14) Although all GVHD developed during or after etanercept treatment, in the small number of cases of early onset prior to day 28, the initiation of ECP can rightly be considered early treatment. It is possible that some of the observed steroid sensitivity may be related to the benefit of early treatment with ECP. Acute GVHD is a well-described risk factor for the development of chronic GVHD.(48) In this study, concurrent or shortly after completion of investigational therapy, the majority of patients developed chronic GVHD. Thus, it is possible that whatever reduction in steroid refractory acute GVHD achieved by our treatment strategy was offset by the development of steroid dependent chronic GVHD, which was the primary cause of most deaths that occurred after one-year post-HCT.

Our study design did not allow us to independently assess the potential benefit of augmenting GVHD prophylaxis with etanercept alone, as we previously studied(12), or ECP alone in this study population. However, late treatment failures were the major contributor to the lack of sustained improvement in long-term survival, raising the possibility that alternative ECP schedules might have a more durable impact on GVHD outcomes. More intensive ECP treatments have been shown to improve long-term survival in patients with severe acute GVHD,(17) therefore, one strategy for optimizing outcomes could include more frequent ECP treatments around the median time to acute GVHD development. Although we do not know if it would have been useful, ECP could have been started shortly after engraftment, up to two weeks earlier than the selected start of day 28 on this trial. Such an approach may have further attenuated acute GVHD. We are encouraged by the high rate of steroid responsiveness observed in our patients, but it is important to note that very few interventions that decrease acute GVHD incidence have translated into decreased incidence of chronic GVHD,(49, 50) suggesting that although acute GVHD is a risk factor for chronic GVHD, the pathogenic mechanisms are likely to be different, thus requiring distinct strategies. Although ECP has been used for therapy of acute and chronic GVHD, its value in prophylaxis of chronic GVHD has never been officially tested and thus remains unknown. Furthermore, on this trial, ECP stopped at day 180, which is close to the median time to chronic GVHD onset. Although speculative, it is possible a more prolonged course of treatment might impact the rate or severity of chronic GVHD that we were otherwise not able to alter in this study. It is important to note that our prophylaxis strategy did not appear to impact relapse rates, but further intensification of the number or duration of treatments might affect GVL. It is also possible that more intensive ECP schedules will not confer benefit and the desired long-term outcomes that we did not achieve in this study will remain elusive with alternative ECP strategies.

In this study the novel composite endpoint of steroid-free, relapse-free survival, at six months post-HCT we investigated did not prove useful as an early surrogate for overall survival. Indeed, despite failing to meet this endpoint, one-year survival was better than expected. Other surrogate marker such as GVHD-free, relapse-free survival have been proposed by others,(51) but have not been tested prospectively and are measured at 12 months post-transplant, which limits its utility as an early marker for long-term survival.

The treatment approach we tested was well tolerated and we did not find evidence that the addition of these agents led to excess infections, graft failures, or relapse. Furthermore, although our results suggest that the addition of etanercept and ECP to a standard GVHD prophylaxis regimen may increase the likelihood of response to treatment when acute GVHD develops, the benefit appears to primarily delay, not prevent, non-relapse mortality as we saw no favorable impact on chronic GVHD. Effective treatments that prevent chronic GVHD are needed. More intensive ECP schedules that start earlier, deliver more ECP treatments, or end later might produce the desired effects and are worthy of testing.

• Delivery of etanercept and ECP for GVHD prophylaxis was safe and feasible

• One year survival was excellent at 73% for high risk HCT patients

• Two year survival declined due to high incidence of chronic GVHD