|Home | About | Journals | Submit | Contact Us | Français|
While in last four decades significant strides have been made in understanding the biology of graft-versus-host disease (GvHD) and its prevention, little is known about the different populations of lymphocytes and the changes in response to treatment for this condition. BMT-CTN 0302 was a randomized phase II clinical trial in the Blood and Marrow Transplant Clinical Trials Network that assessed the efficacy of combination therapy with steroids plus 1 of 4 other agents: pentostatin, mycophenolate mofetil, etanercept, or denileukin diftitox, in patients with acute GvHD. Patients enrolled in the study had their blood analyzed by flow cytometry on days 0, 14, and 28 of therapy to enumerate the number of total lymphocytes, T cells, B cells and lymphocytes expressing activation markers. Baseline lymphocyte total counts and the subpopulations were similar in the four treatment arms. Responding patients at day 28 had a smaller decrease in their total CD45+ cells (p=0.005) compared to non-responding patients. In univariate analysis those who developed chronic GvHD had significantly more CD8+ cells at day 14 compared to those without it (p =0.005). There was no significant association between baseline lymphocyte counts and survival. In univariate analysis, among patients with higher lymphocyte counts at days 14 and 28, there was a trend towards better survival at day 180 although this did not reach the pre-determined threshold for significance. Among the four different treatment arms, we found no significant differences in lymphocyte total or subpopulation counts and no notable influence on outcomes.
Acute graft-versus-host disease (aGvHD) is a common complication of allogeneic blood and marrow transplantation (BMT)1 characterized by an immune attack of the donor cells against host tissues, typically leading to skin rash, diarrhea and/or hyperbilirubinemia. The usual treatment involves the administration of immunosuppressive agents, particularly steroids. This treatment induces frequent clinical responses, but flares are common as steroids are tapered. The balance between effector and regulatory T cells plays a major role in the development of GvHD and its resolution. The impact of GvHD treatment on these defined lymphocyte populations and the association of baseline and post-therapy phenotyping and response to GvHD therapy are uncertain. To better understand the most promising pharmacological strategies for treatment of GvHD, understanding changes in the lymphocyte regulatory and effector T cell compartments may be important 2. CD25 and CD 69 expression on CD4+ and CD8+ T cells reflects their cellular activation status and can be a useful marker for identifying the association with GvHD3,4. Clearly, CD4+CD25+ and CD8+CD25+ T cell phenotype do not fully characterize effector/regulatory phenotypes, but understanding their relative dynamics (percentage and total numbers) in the setting of GvHD pre and post therapy may provide important insights toward our understanding of T cell immune recovery after GVHD. We hypothesized that agents tested in this trial (pentostatin, denilieukin diftitox, etanercept and mycophenolate mofetil) could have different effects on lymphocyte populations and that these changes could correlate with clinical outcomes. We anticipated better response to therapy and potentially less infection in patients who maintained higher numbers of CD4+ T cells after completion of therapy. Also, it was unknown if B cells (CD 20+) would correlate with or contribute to the response to GVHD therapy in these patients.
Despite their use for primary therapy of aGvHD over decades, there is little information on the changes in different lymphocyte populations in response to steroid based or combination agent aGvHD therapy. BMT-CTN 0302 was a randomized phase II clinical trial in the Blood and Marrow Transplant Clinical Trials Network that assessed efficacy of combination therapy with steroids plus 1 of 4 other agents: pentostatin, mycophenolate mofetil (MMF), etanercept or denileukin diftitox, based on previous studies demonstrating activity of these agents in steroid refractory aGvHD5–9. Within the BMT-CTN 0302 trial, serial blood samples were collected from study participants and assessed for lymphocyte subsets by flow cytometry. We examined differences following each of these treatments and assessed whether changes in lymphocyte populations correlated with GVHD response, development of chronic GvHD (cGvHD) and survival.
One-hundred and eighty patients were randomized to receive methylprednisolone at 2 mg/kg per day plus etanercept, MMF, denileukin diftitox, or pentostatin as reported by Alousi et al.5 Their median age was 50 years of age. Sixty-six percent received myeloablative BMT, grafts were peripheral blood (61%), bone marrow (25%), or umbilical cord blood (14%), and 53% were unrelated. Patients who received MMF for prophylaxis (24%) were randomized to a non-MMF arm. At randomization, aGVHD was grade I to II (68%), and III–IV (32%). One-hundred and forty subjects had immunophenotyping data from at least one time point. Complete response (CR) required resolution of all signs and symptoms of aGvHD in all organs without intervening salvage therapies. A partial response (PR) was an improvement of one stage in one or more organs without progression in any organ. Day 28 CR rates were etanercept 26%, MMF 60%, denileukin diftitox 53%, and pentostatin 38%. Corresponding 9-month overall survival was 47%, 64%, 49%, and 47%, respectively. Blood was obtained at baseline on day 0 of therapy for the GvHD (at the time of enrollment to the study), day 14 and day 28 of GvHD study treatment (blood was collected irrespective of response to GvHD therapy). Lymphocyte phenotyping by flow cytometry gated CD45+ cells (total lymphocytes) and enumeration of CD3+, 4+, 8+, 25+, 69+, 20+, and CD4+25+cells was performed at the transplant center and recorded as cells of each subset phenotype per μL of blood. These values were collected by the BMT-CTN data coordinating center and analyzed in conjunction with the clinical outcomes of response and the four randomly assigned treatments. Immunophenotyping and outcome data were analyzed in conjunction with the previously reported clinical study5 which included 180 subjects; 140 subjects had immunophenotyping data from at least one time point. The reasons for missing data include samples not obtained (54%), participant refused (33%), participant missed clinical visits (1%), participant died or too ill (5%), subset not assessed/performed/calculated by lab (5%), not part of flow cytometry (2%). The clinical endpoints of response (CR and PR) at day 28 and cGvHD were reviewed by the protocol team for all patients while still blinded to the treatment assignment and without knowledge of these immunophenotyping data. Support for this study was provided by grant #U10HL069294 from the National Heart, Lung, and Blood Institute, and the National Cancer Institute, along with contributions from Eisai Inc., Hospira Inc, Roche Laboratories Inc., and Immunex Corporation, a wholly owned subsidiary of Amgen, Inc. Additional support was provided by the National Institute of Allergy and Infectious Disease for the ancillary studies “Analysis of Serum Biomarkers Related to aGvHD Treatment Responsiveness” and “Pharmacogenetics of Steroid Responsiveness in aGvHD”. The reported analysis were performed independently of any of the study sponsors.
The primary objective of the study was to examine how the four randomly assigned agents influenced the circulating lymphocyte populations and how these changes correlated with clinical outcomes. Clinical outcomes analyzed were aGvHD response at Day 28, survival at 6 months and cGvHD by 9 months. Lymphocyte data at each time point were compared between the treatments using the Kruskal-Wallis test, after log10(x+1) transformation to induce normality. Linear mixed models were used to examine changes in lymphocyte populations over time and their relationship to treatment. Univariate case-control comparisons for each lymphocyte subpopulation were performed at each time point as well as a comparison of lymphocyte change from baseline using a Mann-Whitney test. Cases are defined as patients experiencing an event (day 28 CR or CR+PR, death within 6 months, or cGvHD by 9 months) while those alive at the same time points without the event were considered the controls. Logistic regression was used for the Day 28 GvHD response in multivariate analyses. Cox regression modeling was used to analyze overall survival in landmark analysis by treating lymphocyte counts at Day 14 or Day 28 as covariates in multivariate analyses. Cox regression model was used in landmark analysis for cGvHD by treating Day 28 lymphocyte subpopulations as covariates in multivariate analyses. Lymphocyte populations were explored using two approaches; as a continuous measurement or as binary covariates using the median as a cut-off. The assigned treatment arm and patient characteristics that might affect outcomes were considered in the multivariate analyses including graft type, donor type and aGvHD grade at onset. Correlations among the lymphocyte population were explored by using the Pearson correlation test. All P values were 2-tailed and were considered significant at P<.01 due to the large number of comparisons. Data were analyzed using statistical software SAS (v9, SAS Institute, Cary, NC).
Baseline lymphocyte total counts and the subpopulations were similar in the four treatment arms (Table 1). After treatment (day 14 and 28), there were modest differences for subpopulations of CD3+, CD4+, CD8 and CD25+ while total lymphocyte count (CD45+) was similar among the treatment arms, although none were statistically significant at p<0.01 (Table 2). Linear mixed models showed similar findings of modest differences in the same lymphocyte subpopulations, which were not statistically significant at p<0.01. No significant changes in lymphocytes with activation markers (CD25, CD69), those associated with regulatory T cell subsets (CD4+25+) or total B cells (CD20+) were noted after therapy or between four treatment cohorts.
In univariate analysis, there was a moderate association between lymphocyte count decline and GvHD response. Responding patients (CR/PR) at day 28 had a smaller decrease in their total CD45+ cells at day 28 (p=0.005) compared to non-responding patients. In multivariate logistic regression modeling of GvHD response, after adjusting for day 14 GvHD response, there was a significant effect of total lymphocyte change from baseline to day 28 (OR=4.63, CI=1.41 to 15.20, P-value=0.012) with a greater fall in lymphocyte counts associated with significantly lower likelihood of response. A box plot of CD45+ for Day 28 GvHD responder vs. non-responder is shown in Figure 1. No other lymphoid subsets were associated with GvHD responses (Supplemental Table 1A).
In univariate analysis, patients with a higher CD8+ count at day 14 had a higher frequency of cGvHD: those who developed cGvHD had significantly more CD8+ cells at day 14 compared to those without cGvHD (p =0.005). In Cox regression modeling of cGvHD, the hazard ratio for risk of cGvHD for patients with CD8+ values above the median (≥100 cells/μL) was 1.88 (95% CI: 0.97–3.66), p=0.06 compared to those with lower CD8+ values. No other baseline risk factors had a significant association with cGvHD. Cumulative incidence of cGvHD by CD8+ value is shown in Figure 2. No other lymphoid subsets were associated with risks of cGvHD (supplemental Table 1B).
There was no significant association between baseline lymphocyte counts and survival, relapse, or CMV infections. In univariate analysis, among patients with higher lymphocyte counts at days 14 and 28, there was a trend towards better survival at day 180 although this did not reach the pre-determined threshold for significance (p<0.01). In Cox regression modeling of overall survival, after adjusting for GvHD response, there was a significant favorable effect of total lymphocyte counts above vs. below the median at day 14 (HR=2.90, CI=1.42 to 5.92, p=0.004), but not day 28 (p=0.53). Kaplan-Meier estimates of overall survival by Day 14 CD45+ counts above or below the median (value≥200/μL) are shown in Figure 3. It appears that higher CD69+ counts at day 14 (p=0.009) and day 28 (p=0.05) may correlate with better survival. No other lymphocyte subpopulations were associated with survival (supplemental table 1C).
We examined the influence of four different, randomly assigned GvHD therapies to cause distinct changes in lymphocyte populations that would help to understand or potentially predict the observed differences in therapeutic effects of each treatment. Among the four different treatment arms, we found no significant differences in lymphocyte total or subpopulation counts and no notable influence on outcomes.
We performed a detailed and prospective characterization of lymphoid populations through the initial course of acute GvHD therapy with steroids plus one of the 4 novel agents. While we failed to observe any significant differences in lymphocyte populations during the first 28 days among the 4 study groups, we did find that higher CD45+ lymphocyte counts at day 28 after initiation of therapy were associated with response, and a trend that higher CD8+ counts early after treatment appear to correlate with the subsequent development of cGvHD10. The results by Grogan et al. suggest that CD8+ T cells in patients with cGvHD are characterized by an increased level of activation and proliferation10, and it is possible that this population is present early on during an episode of acute GvHD.
We also observed a trend towards better survival among patients who had higher CD45+ lymphocyte populations at day 14 and day 28. These results are intriguing, and suggest perhaps that preservation of lymphocyte populations after acute GvHD therapy is crucial for immune recovery and protection against subsequent infections.
This study has its limitations as the number of patients was modest and we had incomplete data at some time points. Given the different mechanism of action for these drugs (MMF and pentostatin as lymphoid metabolic inhibitors, etanercept as TNF-alpha blocker, and denileukin diftitox as CD25+ directed lytic immunotoxin), one might have predicted differing influence on circulating lymphoid subsets. Somewhat surprisingly, each of the four agents tested caused similar changes in the lymphocyte counts studied in both total and subpopulation numbers. This is likely influenced and confounded by the fact that all four study cohorts were concurrently treated with high dose corticosteroids. Additionally, the phenotypic analysis used was limited and insufficient to accurately characterize functionality of T cell subsets. This is in part reflection of the feasibility to conduct multicolor flow cytometric analysis of specimens in the multi-institutional setting. While CD4+CD25+ and CD8+CD25+ T cell phenotypes are useful for identifying the association with GvHD3,4, they do not fully characterize effector/regulatory phenotypes. The addition of intracellular staining to forkhead box P3 (FoxP3) may have helped in better assessing regulatory T cells, however, some studies have found decreased number of CD4+CD25+Foxp3+ T cells in patients with GVHD while others have not11,12. On other hand our findings are consistent with previous study by Przepiorka et al. which also found that phenotypic changes in T cell subpopulations do not predict response to daclizumab given to patients with active GVHD13.
While this detailed examination of lymphocyte subpopulations did not clarify the mechanism of response or why the four study agents led to different clinical outcomes, we suggest that in future studies of GvHD therapy, ongoing evaluation of lymphocyte, serum or tissue biologic markers be analyzed to uncover important details of the immunobiology that can further refine clinical treatment approaches.
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.