|Home | About | Journals | Submit | Contact Us | Français|
Graft-versus-host disease (GVHD) causes most non-relapse mortality (NRM) following alternative donor (unrelated and mismatched related) hematopoietic cell transplant (HCT). We previously showed that increases in day +7 TNF-receptor-1 (TNFR1) ratios (post-transplant day +7/pre-transplant baseline) after myeloablative HCT correlate with outcomes including GVHD, NRM and survival. Therefore, we conducted a phase II trial at two centers testing whether the addition of the TNF-inhibitor etanercept (25 mg twice weekly from start of conditioning to day +56) to standard GVHD prophylaxis would lower TNFR1 levels, reduce GVHD rates, and improve NRM and survival.
Patients underwent myeloablative HCT from a matched unrelated donor (N=71), one-antigen mismatched unrelated donor (N=26) or one-antigen mismatched related donor (N=3) using either total body irradiation (TBI)-based conditioning (N=29) or non-TBI-based conditioning (N=71).
Compared to historical controls, the increase in post-transplant day +7 TNFR1 ratios was not altered in patients who received TBI-based conditioning, but was 40% lower in patients receiving non-TBI-based conditioning. The latter group experienced relatively low rates of severe grade 3-4 GVHD (14%), one-year NRM (16%), and high one-year survival (69%).
These findings suggest that (1) the effectiveness of TNF-inhibition with etanercept may depend on the conditioning regimen, and (2) attenuating the expected rise in TNFR1 levels early post-transplant correlates with good outcomes.
Allogeneic hematopoietic cell transplantation (allo-HCT) is an important therapeutic option for a variety of malignant and nonmalignant conditions. One barrier to increased utilization of allo-HCT is the inferior outcomes when donors other than HLA-matched siblings are used 1,2_ENREF_1_ENREF_1. Compared to matched related donors, recipients of matched or single antigen mismatched unrelated donor and mismatched related donor transplants are at a significantly increased risk of non-relapse mortality (NRM) 3,4. The major contributor to NRM is acute graft-versus-host disease (GVHD), which develops in 50 – 70% of recipients receiving these type of grafts despite standard immunosuppressive prophylaxis 5-8. Thus, novel GVHD prophylaxis strategies which successfully attenuate acute GVHD-related mortality without increasing other causes of NRM or relapse are needed.
Tumor necrosis factor-alpha (TNF-α) plays an important role in the inflammatory cascade that ultimately evolves to acute GVHD 9, and thus represents a potential target for pre-emptive treatment in the control of GVHD. We have previously shown that in the first week after myeloablative HCT the magnitude of change in TNF receptor 1 (TNFR1) ratio (post-transplant day +7/pre-transplant baseline), a surrogate for TNF-α, strongly correlates with important transplant outcomes such as GVHD, NRM, and overall survival (OS) 10. We reported that 161 patients who underwent myeloablative unrelated donor HCT followed by GVHD prophylaxis consisting of the widely utilized regimen of a calcineurin inhibitor and mini-methotrexate experienced a near doubling of TNFR1 levels at day +7 post-transplant (median 1.84x baseline, mean 2.4x baseline). This increase likely reflected the combination of conditioning-induced tissue damage together with release of TNF-α by activated components of the immune system 5,11.
Given that TNF-α amplifies the early alloreactive response in allo-HCT 12, we investigated whether TNF-inhibition could attenuate this pathway. Etanercept consists of two recombinant human TNFR (p75) monomers fused to the Fc portion of human immunoglobulin (Ig) G1, binds to TNF-α and renders it inactive 13. We tested the hypothesis that addition of etanercept to standard GVHD prophylaxis would lower TNFR1 ratios and thereby reduce severe GVHD and subsequent NRM, improving OS in patients following transplants from unrelated and partially matched related donor transplants.
Patients older than 1 year of age who were candidates for a myeloablative allo-HCT were eligible for inclusion. Donors and recipients were required to match for 7/8 or 8/8 Human Leukocyte Antigen (HLA)-A, -B, -C, and –DRB1 loci. Patients with an 8/8 HLA-matched related donor were not eligible. Mid-resolution DNA typing was performed for all class I loci. Allelic typing by high-resolution DNA sequencing defined a match at DRB1. Patients with progressive malignancies were ineligible. Patients with uncontrolled infection despite treatment with appropriate anti-infectives were also ineligible. The protocol and informed consents were approved by the Institutional Review Board at the University of Michigan and Loyola University Medical Center. All patients or their legal guardian signed informed consents in accordance with the Declaration of Helsinki.
Plasma samples collected pre-transplantation (baseline) and day +7 after allo-HCT were frozen for later analysis. TNFR1 levels were measured in duplicate using an enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s protocol (R&D systems, Minneapolis, MN). ELISA plates were read by a microplate reader (Bio-Rad, Hercules, CA).
The study was conducted at two institutions as an open label phase II clinical trial. All patients received a myeloablative conditioning regimen (Table 1) selected by the treating physician on the basis of underlying disease, age, degree of donor match, and disease status at the time of transplant.
GVHD prophylaxis consisted of tacrolimus initiated on day -3 (titrated to a goal level of 8-12 ng/mL) and mini-methotrexate administered at a dose of 5 mg/m2 IV on days +1, +3, +6, and +11. In the absence of acute GVHD, tacrolimus was tapered starting on day +56 post-transplant by 25% per month so that it was discontinued by day +180 post-transplant. The first dose of study drug, etanercept (0.4 mg/kg, maximum dose 25 mg), was administered subcutaneously within 24 hours of initiation of the conditioning regimen. The subsequent doses were administered twice weekly (at least 72 hours apart) through day +56 for a total of 18-19 doses depending on the length of conditioning regimen. Doses were not modified based upon hepatic or renal function. However, doses were held for persistent bacteremia, hemodynamic instability, or fever above 100.5°F for more than five days and not resta rted until resolution. Additionally, if the cytomegalovirus (CMV) viral load increased after 72 hours of anti-CMV treatment, etanercept was held until improvement in the viral load was documented. Additional doses of etanercept were not administered if more than 2 consecutive doses were held for any reason. Etanercept was discontinued and patients were replaced if early relapse of the underlying malignancy occurred within day +21.
Supportive care therapies were administered according to institutional guidelines. Anti-microbial prophylaxis included levofloxacin 500 mg once daily for prevention of bacterial infections; voriconazole 200 mg twice daily; acyclovir 400 mg twice daily; and sulfamethoxazole/trimethoprim or pentamidine for prevention of PCP. Pediatric patients received age/weight equivalent dosing of antibiotics. CMV DNA was monitored weekly by quantitative PCR 14 and pre-emptive therapy with anti-viral agents begun in the event of a positive assay. IVIg 400 mg/kg replacement therapy was given for IgG levels < 400 mg/dL. Total body irradiation (TBI)-treated patients at Loyola University Medical Center (N=9) also received palifermin for mucositis protection according to their institutional guidelines.
An infection was defined using the following criteria: one or more positive blood and/or fluid cultures, or the detection of DNA in the plasma by quantitative polymerase chain reaction (PCR). CMV disease was defined as an organ infected by CMV 15. Proven, probable, and possible invasive fungal infections were classified according to international consensus criteria up to day +180 or relapse 16.
Acute GVHD was scored weekly by the modified Glucksberg criteria 17. Biopsies were obtained of involved target organs to confirm the diagnosis of acute GVHD. Clinically significant acute GVHD was treated with methylprednisolone 2 mg/kg/day orally or IV. Etanercept was continued during therapy for acute GVHD. Complete response (CR) to therapy was defined as the resolution of all manifestations of GVHD (all target organs stage 0). Chronic GVHD was evaluated according to the NIH consensus criteria 18.
The study was originally designed to enroll 80 patients in order to have sufficient power to detect a 50% decrease in NRM and a 20% decrease in acute GVHD compared to historical rates at our centers. After it was recognized that etanercept was not benefiting patients receiving TBI-based conditioning, the protocol was amended to exclude patients receiving TBI-based conditioning. Accrual was extended to 100 patients to provide sufficient power to detect a 50% decrease in NRM in the non-TBI conditioned patients compared to historical NRM rates for these patients at our centers. Differences in median day +7 TNFR1 ratios between study and control patients were assessed using a Wilcoxon Rank Sum test. The control patients consisted of 161 previously reported patients who underwent myeloablative unrelated donor HCT at the University of Michigan between 2000 and 2005. They received GVHD prophylaxis consisting of the widely utilized regimen of a calcineurin inhibitor and mini-methotrexate 10. Study patients were significantly older (median age 47 vs 38 years, p=0.004), but otherwise were not statistically different from the control patients with respect to disease treated, match, and use of TBI . Overall survival was estimated with the methods of Kaplan and Meier 19. The cumulative incidence of NRM and acute GVHD were adjusted for competing risks and estimated using Gray’s method 20. The association of day +7 TNFR1 ratios with clinical outcomes was assessed with Cox regression (OS) and competing risks regression (NRM/acute GVHD) with relapse treated as a competing risk. Multivariate models included TNFR1 ratio and age at transplant as continuous variables and categorical variables for conditioning regimen (non-TBI/TBI), donor type (matched/mismatched) and baseline risk group (high/intermediate/low). Statistical significance was defined as a p-value less than 0.05.
Patient characteristics are summarized in Table 1. A total of 100 patients participated and were transplanted in this study (April 2005 - November 2009). The median age was 47 years (range, 2-61 years) with 26% of the patients > 55 years of age at the time of transplantation. Seventy-one patients received a myeloablative non-TBI-containing conditioning regimen. These regimens were either busulfan-based (N=63) or BCNU-based (N=8). Twenty-nine patients received a myeloablative TBI-containing conditioning regimen.
There were no injection or allergic reactions related to etanercept. Etanercept was discontinued in one patient with primary graft failure following a 7/8-HLA mismatched unrelated donor bone marrow transplant. The patient subsequently engrafted following a second transplant from the same donor. All other patients engrafted neutrophils at a median of 12 days (range, 5-23 days). A total of 16 patients did not receive one or two doses due to persistent fever, a positive blood culture or inadvertent missed dose. In all but three patients, the missed doses occurred either after GVHD had developed or after day +28, precluding an analysis to determine a minimum number of doses needed to effect outcomes. Etanercept was discontinued early in five patients whose treatment was being held according to study design and the underlying reason, such as persistent fever, did not resolve before the third dose was held. Six patients died before receiving all of the planned study doses. All of these patients were included in the analysis.
Given the importance of TNF-α in the innate immune system 12,21, we monitored infectious complications in this trial. Stopping rules, which were never triggered, were in place in the event that fatal complications exceeded the expected rate observed in historical controls. Bacteremia was the most common type of infection in the first 100 days post-transplant. Fifty-nine patients developed a total of 79 bacteremia episodes. Gram-positive organisms, primarily coagulase negative staphylococci, accounted for 66 (83%) of all culture results. There were 14 gram-negative organisms isolated in the first 100 days post-transplant. Bacterial infections accounted for three deaths (3%). One patient developed fatal bacterial pneumonia while receiving etanercept. In two additional patients, bacterial septic deaths occurred five and six weeks after completion of etanercept and during treatment with high-dose corticosteroids for acute GVHD. To better assess the potential impact of etanercept on infection risk we compared the post-engraftment infection rates during etanercept administration (engraftment to day +56) and after etanercept discontinuation (day +56 to day +100). Following engraftment, the relative risk of infection was 1.3 times higher while receiving etanercept, compared to after discontinuation, but this difference was not statistically significant (p=0.36).
A total of 24 patients developed 29 viral reactivations, most often during treatment with corticosteroids for acute GVHD. The predominant virus was CMV (N=16). Lethal viral infections developed in three patients while receiving corticosteroids (CMV pneumonia N=2, HHV-6 encephalitis N=1).
Invasive fungal infections developed in three patients in the first 180 days post-transplant. Nine study patients had a pre-transplant history of invasive fungal infections (disseminated aspergillus N=7, liver candidiasis N=1, liver blastomycosis N=1). All infections were well-controlled with appropriate anti-fungal treatment at the time of study entry. Two of these patients developed radiographic evidence of progressive fungal pneumonia and died. One additional patient died of invasive Rhizopus infection on day +95 that developed while being treated with high-dose corticosteroids for acute GVHD.
In our previous study, the median day +7 TNFR1 level for recipients of myeloablative unrelated donor HCT was 1.84x baseline 10. Therefore, the significantly lower day +7 TNFR1 ratio of 1.34 (p<0.001, Figure 1A) that was observed on this clinical trial suggests that the addition of TNF-blockade to the GVHD prophylaxis regimen may have attenuated the expected rise in TNF levels. Unexpectedly, the effectiveness of TNF-blockade was confined to patients who received a non-TBI-containing conditioning regimen (Table 2). These patients experienced a significantly low day +7 TNFR1 ratio of 1.10 compared to 1.89 in patients who received a TBI-based conditioning regimen (p<0.001, Figure 1B). This finding stands out in contrast to our previous study where significant differences in TNFR1 ratios were not observed between non-TBI and TBI-treated patients 10. Furthermore, these findings cannot be explained on the basis of differences in baseline TNFR1 levels amongst study patients. The median baseline TNFR1 level in patients who received TBI-based conditioning was 1835 pg/mL, which was not significantly different than the median level of 1880 pg/mL in patients who received non-TBI-based conditioning. Moreover, the administration of palifermin as a radioprotectant to nine patients resulted in no apparent effect on the day +7 TNFR1 ratios in TBI-treated patients (2.1 vs 1.8, p=NS). Although patients with ALL were overrepresented and patients with AML/MDS were underrepresented in the TBI-treated patients, there were no significant differences in other patient characteristics, including age, gender, disease status at transplant, degree of HLA-match, or stem cell source between TBI- and non-TBI-treated patients in the current study.
TNF-blockade did not alter the prognostic significance of a high day +7 TNFR1 ratio on outcomes. In etanercept-treated patients, the day +7 TNFR1 ratio, when treated as a continuous variable, significantly correlated with an increased risk of NRM within one year (HR 1.5, p=0.012) and with a decreased likelihood of survival (HR 1.5, p=0.027). Increased day +7 TNFR1 ratios were also associated with increased rates of grades 2-4 and 3-4 acute GVHD (HR 1.3, p=0.095 and HR 1.4, p=0.082), although not statistically significant.
The day +100 cumulative incidences of grades 2-4 and grades 3-4 acute GVHD were 45% and 18%, respectively (Figure 2). GVHD requiring treatment was primarily grade 2 acute GVHD (N=29) involving only the skin (N=18), upper GI (N=4), lower GI (N=2), or combined skin and GI (N=5). These patients had a high rate of CR to treatment (93%) within a median of 16 days. We expected to find a lower incidence of acute GVHD in the non-TBI-treated patients compared to TBI-treated patients based on the lower median day +7 TNFR1 ratio, but the cumulative incidences of grades 2-4 acute GVHD were the same (45%). There were twice as many cases of severe grade 3-4 GVHD in TBI-treated patients (28% vs. 14%, p=0.15) where TNF blockade was not as effective in attenuating the day +7 TNFR1 ratio, however this study lacked sufficient power to detect a statistically significant difference for this comparison. Nevertheless, TBI-treated patients were more likely to die from GVHD (p=0.04) and experienced higher 1-year NRM (50%) compared to non-TBI-treated patients (16%, p<0.001, Figure 3). All causes of 1-year NRM are provided (Table 3).
Allo-HCT from HLA-mismatched unrelated donors is associated with very high rates of acute GVHD and NRM. Given that over 25% of the study population fell into this very high-risk category, we analyzed GVHD and NRM outcomes for this specific population. The grade 3-4 GVHD rates for mismatched unrelated donor HCT were higher than those seen in the other patients (31% vs. 14%, p=0.04), but did not translate into significant differences in one-year NRM (35% vs. 22%, p=0.24).
The 1-year cumulative incidence of relapse for the entire study population was 18%, with similar relapse rates by conditioning regimen administered. The cumulative incidence of chronic GVHD at 1-year was 48% (Supplemental Figure 1). With a median follow-up of 15 months (range: 0.7-63 months), the 1-year OS for the entire study population was 62% (Supplemental Figure 2). There was a trend toward improved survival in patients who received a non-TBI-containing regimen (1-year OS 69% vs. 45%, p=0.06, Figure 3). Notably, the 1-year OS for HLA-mismatched, unrelated donor HCT recipients was not different from HLA-matched unrelated donor/mismatched related donor patients (54% vs. 65%, p=0.15), even when restricting the analysis to non-TBI treated patients only (61% vs. 71%, p=0.33).
In this study, the addition of the TNF-inhibitor etanercept to tacrolimus/methotrexate GVHD prophylaxis did not affect the overall risk of grades 2-4 acute GVHD. However, patients who received a non-TBI containing myeloablative unrelated or mismatched HCT experienced high rates of steroid-responsiveness in those who developed GVHD, low rates of NRM (16%), and good 1-year survival (69%), and this was associated with lower day +7 TNFR1 ratios. By contrast, there was no apparent effect of etanercept on day +7 TNFR1 ratios, NRM, or survival in patients who received TBI-containing conditioning regimens. Given the non-randomized study design, we cannot conclude with certainty that etanercept administration was responsible for the favorable outcomes in the non-TBI conditioned patients. However, when taken in light of recent observations from a large series of unrelated donor HCT recipients that showed no significant difference in NRM (31%) and 1-year survival (51%) between TBI and non-TBI conditioned patients 22, our findings suggest that etanercept may have an overall beneficial impact on the outcome following non-TBI HCT. The reasons for these observations are not clear.
A possible explanation for our findings in the TBI-conditioned patients is that high doses of radiation induced greater tissue injury and inflammatory cytokine release 23,24 than this dose and schedule of etanercept was able to effectively neutralize. If so, a more intensive etanercept dosing regimen may achieve better results, but further studies would be needed to evaluate the merits of such a strategy. Data from animal models demonstrate both TNF-dependent and TNF-independent pathophysiology, whereby TNF inhibition attenuates but does not completely eliminate GVHD 11. Accordingly, it is also possible that TNF-independent pathways are major contributors to GVHD and NRM in radiation-based conditioning regimens for which even highly effective TNF-inhibition will not significantly alter outcomes. Given the large difference in 1-year NRM based on conditioning regimen, we considered the possibility that TBI-conditioned patients experienced excess NRM. The 1-year NRM of 50% we observed in TBI-conditioned patients is similar to the 43-47% reported in a recently published registry series of over 1200 TBI-conditioned unrelated donor recipients 25. Therefore, although ineffective, compared to results reported in the literature, it does not appear that etanercept administration worsened outcomes in TBI-conditioned patients.
Because TNF-α is an important mediator of both innate and acquired immune responses12,21_ENREF_31, and infections represent an established risk of TNF-inhibitors 26, we paid particular attention to the possibility that etanercept administration could lead to an increased number or severity of infections in this immunocompromised study population. Infections within the first 100 days post-transplant were common in this study, as expected in patients undergoing high-risk allo-HCT 27. In the absence of a control population, we are limited in the conclusions we can draw regarding etanercept and bacterial complications. It is encouraging that the rate of bacteremia, as well as bacterial septic death (3%), observed in this study is in line with published rates 34,35. The incidence and severity of fungal and viral infections were also what would be expected in a high-risk population receiving standard GVHD prophylaxis 28-30_ENREF_29_ENREF_27. Importantly, invasive fungal infections were well-controlled in patients with a history of such infections or who were undergoing treatment at the time of HCT. The three fungal deaths observed within the first 180 days post-transplant are in line with published rates of 1-5% 31-33. Nonetheless, further data are needed to fully assess the safety of etanercept in the context of prior invasive fungal infections and allo-HCT, particularly in patients receiving TBI-based conditioning. Furthermore our trial design, which mitigated the risk of infection by monitoring for pathogens early and initiating preemptive treatments, may have offset any increased risk due to etanercept.
In summary, this study investigated the addition of etanercept given twice weekly for the first 8 weeks post-transplant to a standard GVHD prophylaxis regimen. Etanercept injections were well-tolerated and we did not identify any excess toxicity either during the two months of active therapy or the post-treatment observation period. Based on the lack of efficacy observed in patients receiving TBI-based conditioning, we would caution against the use of etanercept in this context. Conversely, in non-radiation based conditioning, etanercept appeared to attenuate the expected rise in TNFR1 ratios early post-transplant and these patients experienced good outcomes. Future approaches may build upon a TNF-inhibition platform by incorporating complementary strategies. One such strategy is to increase the number of regulatory T cells (Tregs), which inhibit GVHD while preserving GVL 34. An experimental GVHD model has demonstrated that extracorporeal photopheresis (ECP) induces Tregs 35. We are therefore currently testing this combination approach of TNF-inhibition and ECP in a prospective clinical trial in unrelated donor HCT.
Supported by grants from the National Institutes of Health (2P01CA039542 and 5P30CA046592). Amgen, Inc supplied study drug. SWC is the recipient of a St. Baldrick’s Scholar Award and the National Institutes of Health K23 Grant AI091623-01. We thank the patients, their families, and the clinical personnel who participated in this study; Jennifer Lay-Luskin and the BMT Program Team at the Clinical Trials Office at the University of Michigan and Loyola University Medical Center for outstanding data collection and management; and the BMT Program research nurses for research support without which this study would not have been possible.
Conflict of Interest Statement: There are no conflicts of interest to report.
Authors’ contributions: JEL, KC, TB, and JLMF designed and planned the study. TB was the study statistician. JW performed the cytokine assays. SWC and CK were in charge of data collection and quality assurance. All authors participated in writing the report.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.