Peri-transplant palifermin use and lymphocyte recovery after T-cell replete, matched related allogeneic hematopoietic cell transplantation

Abstract

A previously conducted randomized, double-blind, placebo-controlled study from 2000-2003 tested peri-transplant palifermin (KGF) in 100 recipients of T-replete matched related donor allogeneic hematopoietic transplant (MRD allo-HCT). The use of KGF in preclinical transplant models, including an autologous non-human primate model, has been associated with improved immune reconstitution. Therefore, we investigated whether palifermin treated patients (n = 69) had improved absolute lymphocyte counts (ALCs) at days 30, 60 and 100 after transplant compared to placebo treated patients (n = 31). No statistically significant difference in ALC was noted at these time points. Additionally, there was no difference in ALC between patients who received low (240μg/kg) vs. high (720μg/kg) dose palifermin. Patients with a day 30 ALC of >600 ×10⁶/L had a trend towards improved progression-free survival (49% vs. 29%, p=0.07), lower transplant-related mortality (18% vs. 35%, p=0.1) and grade II-IV aGvHD (31% vs. 47%, p=0.14), but this was not influenced by palifermin therapy. We conclude that following myeloablation and a T-replete MSD allogeneic graft, peri-transplant palifermin does not accelerate early lymphocyte recovery. Studies combining palifermin with other agents (as in pre-clinical models) may be required to improve immune reconstitution after allo-HCT.

Allogeneic hematopoietic cell transplantation (allo-HCT) is often the only curative option for people with otherwise fatal hematologic malignancies. As the number of allo-HCT procedures continues to increase [1], it is increasingly clear that a major obstacle to success is delayed immune recovery, which puts atients at risk for a wide variety of opportunistic infections [2-7][8]. Additionally, rapid early lymphocyte recovery may serve as a surrogate predictor of better transplant outcomes. Robust recovery of absolute lymphocyte counts (ALC) early after transplantation is associated with improved survival following autologous, sibling, unrelated bone marrow, peripheral blood and umbilical cord blood transplantation [9-15]. There is a clear need to develop strategies to accelerate and improve immune reconstitution (IR). Several novel approaches have been successfully tested in preclinical animal models and early human clinical trials. These include pre-transplant androgen ablation, keratinocyte growth factor (KGF), and a p53 inhibitor or post-transplant administration of IL-7, IL-15, growth hormone, or insulin-like growth factor-1 [16-20].

KGF is a potent growth factor belonging to the fibroblast growth factor family that stimulates epithelial growth and differentiation, without affecting the non-epithelial cells that lack the FGFR2-IIIb receptor. In the thymus KGF is produced locally by mesenchymal cells and acts on thymic epithelial cells (TECs) [18, 21]. Through its trophic effects on TECs, KGF facilitates normal thymopoiesis which is required for naïve T cell generation [22]. In murine allo-HCT models, KGF treatment improved thymopoiesis and enhanced T cell IR [18, 23]. In a nonhuman primate model of autologous hematopoietic cell rescue following myeloablation, peri-transplant KGF administration led to the preservation of thymic architecture for up to 12 months after HCT [24]. In the same study, KGF treated animals had higher frequencies of naïve T cells compared to the control group. Another preclinical study which combined KGF with androgen depletion led to enhanced thymopoiesis characterized by a broad V-beta T cell repertoire [25].

To assess the GvHD protective potential of KGF, a phase I/II randomized, placebo controlled trial of recombinant human KGF (palifermin) was performed from 2000-2003 in 100 patients undergoing T cell replete, MRD allo-HCT following myeloablative conditioning [26, 27]. The focus of this study was to determine the impact of KGF on GVHD. Detailed IR was not studied within the trial. Because ALC is associated with transplant outcomes [9-15]. we questioned whether there might be differential effects of KGF on early ALC recovery. Here, we report our retrospective analysis using data from our earlier trial to assess the effects of peritransplant use of palifermin on ALC recovery at following HCT. This is the first report analyzing the results of palifermin and lymphocyte recovery in humans.

We tested ALCs after transplant in the palifermin and placebo treatment groups. As shown in figure 1a, the median ALC at days 30, 60 and 100 after transplant did not significantly differ in the palifermin vs. placebo arms (600 × 10⁶/L vs 600 × 10⁶/L; P = 0.26, 500 × 10⁶/L vs 600 × 10⁶/L; P = 0.86, 600 × 10⁶/L vs 700 × 10⁶/L, P = 0.19). Likewise, there was no influence of the palifermin dose and subsequent ALC recovery post-transplant ALC (figure 1b). As well, palifermin did not impact ALC based on the conditioning regimen (Bu/Cy vs. Cy/TBI), recipient age and the presence of aGVHD (data not shown). As previously reported, in this trial, palifermin use was not associated with shorter time to engraftment, aGvHD, survival or infectious complications after transplant. However, palifermin modestly decreased mucositis [26, 27].

Figure 1. Post-transplant ALC in palifermin and placebo groups.

A) Median and inter-quartile range of ALC for the placebo (open bars) and palifermin (shaded bars) at day 30, 60 and 100 after transplantation. The median ALC at the three time points in the two groups were 600 × 10⁶/L vs. 600 × 10⁶/L (P = 0.26), 600 × 10⁶/L vs. 500 × 10⁶/L (P = 0.86) and 700 × 10⁶/L vs. 600 × 10⁶/L (P = 0.19), respectively. B) Comparison of median and inter-quartile ranges of day 30, 60 and 90 ALCs at various doses of palifermin with placebo.

Based on prior studies [9-15], we also examined whether differential lymphocyte counts were associated with transplant outcomes. Patients with an ALC above the medin at D+30 (>600 ×10⁶/L) showed a trend for improved PFS (49% (95% CI, 32-65%) vs. 29% (95% CI 16-43%), p=0.07) (figure 2a). There was no association between D+30 ALC and disease recurrence (28% (95% CI 14-42%) vs. 23% (95% CI 10-36%), p=0.6) (figure 2b). There were trends toward less 2 year TRM (95% CI 18% (6-30%) vs. 35% (95% CI 20-50%), p=0.1) and grade II-IV aGVHD (31% (95% CI 17-45%) vs. 47% (95% CI 32-62%, p=0.14) in patients with an ALC>600 at D+30 after transplant (figures 2c and 2d). These rates were unaffected by KGF therapy.

Figure 2. Transplant outcomes based on ALC at day 30.

Outcomes were measured for patients above or below the median ALC (i.e, >600 × 10⁶/L or ≤ 600) at day 30. A) PFS at Day 30 was 49% vs. 29%, respectively (p=0.07). B) The cumulative incidence of relapse was 28% vs. 23%, respectively (p=0.6). C) 2-year TRM was 18% vs. 35%, respectively (p=0.1). D) Grade II-IV aGvHD was 31 % vs. 47%, (p=0.14).

In this first allotransplant palifermin dose escalation trial, using ALC as a marker for IR, we found that peri-transplant palifermin had no impact on ALC at day 30, 60 or 100 after allo-HCT. These results differ from rodent studies [22, 25] and a non-human primate study [24] where KGF was associated with protection of TECs from radiation-induced damage, resulting in improved thymopoiesis and peripheral IR after transplant. Numerous explanations may account for our the findings including suboptimal palifermin dose or dosing schedule, existing pre-transplant chemotherapy-induced TEC damage or a limited potency of this agent on human TECs. ALC is a relatively crude marker for IR and the lack of information on various T cell subsets, T cell receptor Vβ repertoire and newly formed T cells (i.e., thymic excision circles) limits our ability to detect any effects of palifermin on the naïve vs. homeostatically expanded T cells. In addition, only three patients were less than 18 years in this study, with the median age being 47. The lack of accelerated IR and higher ALC with palifermin could also be related to a reduced effectiveness of palifermin on atrophied thymic tissue, present in older subjects. Alternatively, a recent study by Chakraverty et al found that the degree of in vivo T cell depletion adversely affects ALC recovery post allo-HCT, underscoring the importance of graft-derived T cells in early ALC recovery [29]. Therefore it is possible that the early lymphocyte recovery after T cell replete HCT is not reflective of thymic output and may be more related to homeostatic proliferation of existing T cells. In this scenario, ALC may be unaffected by palifermin. In addition, significant aGVHD occurred in nearly half the patients on this study. This complication is associated with thymic injury and requires extended immunosuppressive therapy. Thus aGVHD (or its treatment) might have confounded any potential of palifermin to augment IR. However, there was no significant difference in ALCs between the palifermin and placebo treated groups, even after excluding patients with aGvHD (data not shown). However it is still possible that the GvHD prophylaxis itself (which were not used in the preclinical models) may have blunted any potential benefit of palifermin on lymphocyte recovery, considering that most GVHD prophylactic agents inhibit T cell production and expansion.

Previous studies show a positive correlation between early ALC and transplant with outcome measures [9, 14, 30]. While the clinical impact of ALC recovery was only a secondary aim of this study, we observed a trend towards improved PFS and lower TRM in patients above the median ALC at day 30, consistent with other reports [13-15, 30]. Interestingly, a lower day 30 ALC was associated with a trend towards more frequent grade II-IV aGvHD. Thus, it is possible that before becoming clinically evident, aGvHD is heralded by delayed recovery of ALC.

Delayed IR continues to be a major problem after allo-HCT. Based on this analysis, palifermin alone is unlikely to significantly improve post allo-HCT immune recovery, at least following T-cell replete allo-HCT. In recent years there have been significant advances in understanding the mechanisms behind delayed immune reconstitution and some novel interventions have been successful in pre-clinical models (reviewed in [28]). Perhaps strategies combining two or more agents, shown to be promising in pre-clinical models, may be needed to overcome the profound immune deficiency which follows allo-HCT.

Patients and methods

Patient and transplant characteristics

Patient and transplant characteristics have been reported previously [26, 27]. 100 patients were enrolled in this study and randomized to receive placebo (n=31) or palifermin (n=69). Two patients in the palifermin group did not undergo transplantation and were excluded. As shown in table 1, patient demographics and underlying diagnosis were well matched between the two groups. Median follow-up among survivors is 4.1 years (range: 2-6.1 years) for both groups.

Table 1.

Patient baseline demographics and disease characteristics

	N	Palifermin	Placebo	P
Center				0.91
Michigan	46	32	14
Minnesota	54	37	17
Gender				0.99
Male	58	40	18
Female	42	29	13
Median Age, y (range)		46 (7-65)	46 (7-63)	0.58
Diagnosis				0.08
AML	36	24	12
CML	15	7	8
MDS	12	6	6
NHL	14	13	1
ALL	9	8	1
Hodgkin's disease	1	1	0
Other malignancies	13	10	3

AML indicates acute myelogenous leukemia; CML, chronic myelogenous leukemia; MDS, myelodysplasia syndrome; NHL, non-Hodgkin lymphoma.

All patients received allo-HCT from HLA-identical sibling donors after myeloablative conditioning which included 120 mg of cyclophosphamide and 1320 cGy of fractionated total body irradiation (TBI) at University of Minnesota or 16 mg/kg of oral busulfan and 60 mg/kg of cyclophosphamide (Bu/Cy) at the University of Michigan. A calcineurin inhibitor (cyclosporine or tacrolimus) and methotrexate (15 mg/m² on day 1 and 10 mg/m² on days 3, 6 and 11) were used for GvHD prophylaxis.

Study design

This was a randomized, double-blind, placebo-controlled, dose-escalation study. Three cohorts receiving increasing palifermin doses were sequentially enrolled, with each cohort randomized to achieve balance within each study site and in each cohort. Stratification was based on conditioning regimen and patient age. A total of 98 patients completed this study, with a 1:2 randomization between placebo (n=31) and study drug (n=67). All treatment cohorts received 3 days of palifermin at 40 mg/kg (n=8) or 60 mg/kg (n=61) prior to the start of conditioning (days -11 to -9) and then received the same palifermin dose on 3 consecutive days weekly (first three days of a week) staring on day 0 and extending for 1 (n=18), 2 (n=14) or 3 (n=37) weeks after transplant. Thus, the total palifermin dose was 240 mg/kg in the lowest dose cohort and 720 mg/kg for the highest dose cohort. For the current analysis, absolute lymphocyte counts (ALCs) were extracted from the patient medical records on days 30, 60 and 100 (+/- 7 days) after obtaining institutional review board approvals at both study sites.

Statistical analysis

Statistical comparisons of the distribution of ALC over time were performed by examining the median, inter-quartile ranges and ranges of ALC at days 30, 60 and 100 post transplant and performing either a general Wilcoxon or Kruskal-Wallis test by a number of factors: KGF group (placebo versus KGF), placebo dose, presence of grade II-IV acute GvHD, conditioning and age.

Outcomes by ALC were analyzed by simply dividing cohorts by the median ALC and estimating the outcomes of the two groups using Kaplan-Meier estimates for overall survival (OS) and progression-free survival (PFS) (34) and the cumulative incidence (CI) for relapse, non-relapse mortality (NRM) and GvHD (35). Analyses were performed using SAS 9.2 (SAS Institute) and R 2.4 statistical software.