A prospective cohort study comparing long-term outcomes with and without palifermin in patients receiving hematopoietic cell transplantation for hematologic malignancies

Abstract

Background:

The incidence of debilitating oral mucositis (OM) can be as high as 99% after myeloablative conditioning regimens preparing patients with hematologic malignancies for hematopoietic cell transplantation (HCT). Palifermin (KGF) is a recombinant human keratinocyte growth factor that reduces incidence and duration of severe OM. Long-term safety of KGF, however, has not been well established.

Methods:

In this long-term, prospective, matched cohort study, patients who received KGF (cases) and underwent autologous or allogeneic HCT for hematologic malignancies during 2006 – 2013 were 1:1 matched to patients who did not receive KGF (controls). The primary outcome was overall survival (OS). Other outcomes were disease relapse, new malignancies, pancreatitis, renal failure requiring dialysis, pulmonary complications, cataract surgery and acute and chronic graft-versus-host disease (GVHD).

Results:

The analysis population consisted of 2191 matched pairs with a wide range of diseases and donor types and receiving diverse conditioning and GVHD preventive regimens, representing contemporary practice patterns. The median follow-up was 8 years (range, 1 – 12.5 years). In multivariate analyses, the probabilities of overall survival (relative risk [RR] 1.01 [95% CI 0.91 – 1.12]), relapse (RR 1.06 [95% CI 0.94 – 1.18]), new malignancies (0.89 [95% CI 0.67 – 1.18]), and cataract surgery (RR 1.05 [95% CI 0.74 – 1.50]) were not statistically significantly different between cases and controls.

In univariate analyses, no increased risks were observed for renal failure requiring dialysis, pancreatitis, acute GVHD, chronic GVHD, interstitial pneumonitis/acute respiratory distress syndrome/idiopathic pneumonia syndrome, or bronchiolitis obliterans/cryptogenic organizing pneumonia/bronchiolitis obliterans organizing pneumonia among cases compared to controls.

Conclusions:

This long-term prospective safety cohort study demonstrated that the KGF group had no increased risk of overall mortality, relapse, new malignancies, or any other key outcome. The broad inclusion criteria allow the results to be generalized to contemporary practice for patients with a wide range of diseases and receiving a wide range of HCT conditioning regimens and graft sources from diverse donor types.

Graphical Abstract

Probability of overall survival

Cases: exposed to palifermin (KGF). Controls: not exposed to KGF.

Introduction

Conditioning agents used prior to hematopoietic cell transplantation (HCT) cause oral mucositis (OM) in 75 to 99% of patients ^1–3. Severe cases of OM are associated with extreme discomfort, result in lengthy hospitalization, increase the need for total parenteral nutrition (TPN) and patient-controlled analgesia (PCA), and increase the risk of infections ^4–6. Additionally, OM is an expensive clinical complication and has been reported to lead to more than $70,000 of additional hospital charges among HCT patients ⁷.

Palifermin (Kepivance®) is a recombinant human keratinocyte growth factor (KGF) that stimulates growth of epithelial cells and promotes healing. It is indicated to decrease the incidence and duration of severe OM in patients with hematologic malignancies receiving myelotoxic therapy in the setting of autologous hematopoietic stem cell support administered at a dose of 60 μg/kg/day for 3 consecutive days before and 3 consecutive days after myelotoxic therapy for a total of 6 doses ⁸. Clinical studies have demonstrated that KGF is effective in reducing the risk of severe OM ^9–23. Several reviews also support the use of KGF ^{1,4,8,14,17,18,24}.

KGF may reduce rates of infections in patients who receive high-dose conditioning with either BEAM (carmustine, etoposide, cytarabine and melphalan) or busulfan-thiotepa ²⁵.

Short-term adverse events observed with use of KGF were mild to moderate in severity ^{12,15,17,20,23}.

Since the KGF-receptor is expressed on multiple epithelial tissues, the potential for stimulation by KGF to cause changes that might lead to adverse effects such as malignant transformation was a concern. Alternatively, given the protective effect of KGF for mucosal tissues, malignant cells could possibly be spared the anticancer effects of the transplant procedure or of other pretransplant or posttransplant anticancer therapy (e.g., planned maintenance therapy), and therefore, increased risk of post HCT disease relapse was also a concern. A previous study also demonstrated a greater proportion of patients in the KGF than in the placebo group developed cataracts: 48% (25/52) versus 29% (4/14) (difference of 17% [95% CI: −11, 46]) though the difference was not statistically significant ²⁶.

To date, however, no conclusive preclinical or clinical observations suggest that administration of KGF to patients with hematologic malignancies reduces responses to therapy, increases rates of recurrence or progression of underlying hematologic malignancies, increases the incidence of new malignancies, or reduces overall survival (OS). Further, the randomized studies had stringent inclusion criteria enrolling patients with few disease types and focused on select set of conditioning regimens. These studies also had modest sample sizes (enrolled between 20–100 subjects) and were designed to evaluate short-term events (follow-up periods ranging from 2 weeks to 2 years) ^{12,15,19–23}.

A previous randomized study enrolled 662 patients who were participants of several short-term randomized studies for continued long-term follow-up ²⁷. The overall median follow-up in the study was 8.2 years. The study observed comparable long-term safety outcomes between KGF- and placebo-treated patients undergoing autologous HCT for hematological malignancies as measured by OS, progression-free survival and incidence of secondary malignancies. However, the study was not prospectively designed as a non-inferiority study, and the parent studies were stratified for the factors influencing the incidence of OM. Thus, the researchers acknowledged the potential lack of adjustment for disease type or other important prognostic factors for disease progression and survival in this patient population.

A previous retrospective study of pediatric patients who underwent HCT found similar outcomes for patients who were treated with KGF and for matched controls: This study found similar rates of overall survival, disease-free survival, neutrophil recovery, and GVHD 2.5 years after transplantation ²⁸.

The FDA-required a Phase IV commitment by the manufacturer with a request to collaborate with the CIBMTR. Therefore, to fully evaluate long-term safety of KGF, this prospective study was designed to evaluate the risk for late, infrequent, adverse events across a wide range of hematologic malignancies and conditioning regimens.

Methods

Study design and size

This was a long-term, prospective, observational cohort study, designed to accrue 4800 patients, 2400 receiving KGF therapy (cases) and 2400 matched controls not receiving KGF therapy (controls), who underwent autologous and allogeneic HCT and were reported to the Center for International Blood and Marrow Transplant Research (CIBMTR).

The patient enrollment plan assumed a loss to follow-up rate that would increase over the course of the study, with a cumulative loss of no more than 20% over 10 years, so that at least 1,900 patients in each cohort would be followed until death or the end of the study period.

Setting

We included children and adults who underwent first autologous or allogeneic hematopoietic stem cell transplantation (HCT) for hematologic malignancies in standard clinical care from 171 transplant centers within the CIBMTR network in the United States; Canada; Europe; Asia; Australia/New Zealand; Middle East; Africa; and Central/South America.

The CIBMTR is a research affiliation between the National Marrow Donor Program® (NMDP)/Be The Match® and the Medical College of Wisconsin. It facilitates critical research through medical, scientific, and statistical expertise; a network of more than 330 participating centers worldwide; an outcomes database with clinical data on 575,000 patients; and a biospecimen repository ²⁹. Centers participating in this had to agree to provide supplemental data on eligible patients beyond the comprehensive clinical data routinely collected by CIBMTR.

Accrual to the study began on January 1, 2006 and continued until the full study population of 2,400 matched pairs was reached in November 2013. Subjects’ data were collected through Sept. 2, 2019.

Participants

CIBMTR centers must prospectively report all HCT recipients via a Pre-TED (Transplant Essential Data) registration form. Cases and controls were selected from registered patients in participating centers. Additionally, cases and controls were selected based only on the first transplantation and whether or not the patient received KGF for that transplant.

The CIBMTR Pre-TED form was modified to request information regarding the planned use of KGF therapy. All responses of “yes” were eligible for selection to the study as exposed cases. Final classification of patients into case vs. control was based on the information of actual use or non-use of KGF on the Day 100 Form post-transplant Comprehensive Report Form. A patient was considered to be a case if he/she received at least one dose of the drug.

One author (W.S.) reviewed all pathology reports for new malignancies while being blinded to the treatment arm assignment.

The study was subject to Institutional Review Board (IRB) approvals and patients and/or guardian(s) provided informed consent prior to research participation.

Matching process

Controls were selected to match on the following variables: recipient age, transplant year, transplant type, underlying disease and disease stage, donor type, region, and use of TBI in the conditioning regimen. The matching factors were chosen to reduce confounding and were approved by regulatory agencies.

Outcomes

Primary outcome

The primary outcome was overall survival, defined as time to death. Death from any cause was considered an event. Surviving patients were censored at time of last follow-up.

Secondary outcomes

Relapse: time to onset of first hematologic or clinical recurrence of the malignancy for which the HCT was performed. Death in remission was treated as a competing risk.

New malignancy: time to development of a first new malignancy post HCT. Death without development of new malignancy was treated as a competing risk.

Cumulative incidence of pancreatitis: time to onset of pancreatitis. Death without development of pancreatitis was treated as a competing risk.

Cumulative incidence of renal failure requiring dialysis: time to onset of renal failure requiring dialysis. Death without development of renal failure requiring dialysis was treated as a competing risk.

Pulmonary complications: the cumulative incidence of interstitial pneumonitis (IPn), acute respiratory distress syndrome (ARDS), idiopathic pneumonia syndrome (IPS), bronchiolitis obliterans (BO), cryptogenic organizing pneumonia (COP), and bronchiolitis obliterans organizing pneumonia (BOOP) was calculated. Time to onset of these pulmonary complications was estimated. Death without development of pulmonary complications was treated as a competing risk. The analysis for BO/COP/BOOP was restricted to allogeneic HCT recipients.

Cataract Surgery: time to cataract surgery. Death without undergoing cataract surgery was treated as competing risk. Cataract surgery was used as a surrogate for clinically burdensome cataracts.

Acute graft-versus-host disease (GVHD): the cumulative incidence of developing acute GVHD was calculated. Death without acute GVHD was treated as competing risk.

Chronic GVHD: the cumulative incidence of developing chronic GVHD was calculated. Death without chronic GVHD was treated as competing risk. Analyses for acute and chronic GVHD were restricted to allogeneic HCT recipients.

Statistical methods

With a final study cohort of 4382 patients (Figure 1), the study had 80% power (with alpha 0.05) to detect the following differences: 4.5% for survival and relapse, and 1.5% for new malignancy and cataract surgery.

Figure 1.

Disposition of participants

We completed multivariate analysis for overall survival, relapse, new malignancy, cataract surgery, acute GVHD, and BO/COP/BOOP. Univariate analyses only were performed for all other outcomes.

Potential imbalances of patient, disease and treatment characteristics between cases and controls were adjusted for using a propensity score approach. The propensity score is the probability of a particular patient receiving KGF treatment ^30–32.

Logistic regression was used to model the propensity score. Any risk factors with P < 0.01 were included in the final logistic regression model; these included Karnofsky performance score at transplant (KPS), patient race, year of transplant, conditioning regimen, GVHD prophylaxis, and use of planned use of granulocyte or granulocyte-macrophage colony-stimulating factor.

Using the above logistic regression model, the propensity score was calculated for each patient based on his/her characteristics. Equal propensity scores indicated patients with similar probabilities of being treated with KGF. The distributions of estimated propensity scores between exposed and unexposed were examined. Patients were divided into 4 to 5 strata based on quartiles or quintiles of propensity scores from pooled cohorts. Then, stratified tests were used in all comparisons. The median (range) of estimated propensity scores for KGF exposed (cases) and KGF unexposed (controls) are 0.51 (0.21 – 0.96) and 0.46 (0.21 – 0.93), respectively, with P < 0.0001.

The relative risk of overall mortality, relapse, new malignancy and cataract surgery for cases versus controls was estimated using Cox proportional hazards regression analysis; stratifying on the matched pair and considering the propensity score as a continuous covariate. The proportionality assumptions were tested by adding a time-dependent covariate; tests indicated that the proportionality assumptions were valid. The potential interactions between receiving KGF and propensity score were tested, and no interactions were detected.

Results

Participants

Figure 1 summarizes the flow of subjects. After exclusion of pairs that were later identified as mismatched by a prospectively defined process to review updates in the collected data or had a follow up of < 1 year, a total of 2191 matched pairs were included.

In our study, with a median follow-up of 8 years (1 – 12.5 years), the lost-to-follow-up rate was low overall and was comparable at 7% and 6% in cases and controls, respectively ³³.

Descriptive data

Overall, the median age was 51 years (range, < 1–80 years). The most common indications for HCT were acute leukemia (40%), followed by plasma cell disorder (28%), and non-Hodgkin lymphoma (13%). TBI was used in 41% of conditioning regimens. Fifty percent of the matched pairs underwent allogeneic HCT; the two most common donor types were HLA-identical siblings (37%) and well-matched unrelated donors (35%). Among allogeneic HCT recipients, most (88.5%) received myeloablative regimens (Table 1).

Table 1.

Characteristics of patients who did and did not receive palifermin (KGF), and underwent transplant from January 2006 – December 2013

Variable No. (%)	Allo KGF No. (%)	Allo No KGF No. (%)	Auto KGF No. (%)	Auto No KGF No. (%)	Total KGF No. (%)	Total No KGF No. (%)
Number of patients	1104	1104	1087	1087	2191	2191
Number of centers	83	132	63	110	109	157
Patient-related variables
Age at transplant, median, years (range)	43 (<1 – 73)	44 (<1 – 72)	57 (8 – 79)	57 (14 – 80)	51 (<1 – 79)	51 (<1 – 80)
Sex
Male	630 (57)	609 (55)	612 (56)	658 (61)	1242 (57)	1267 (58)
Race
White	939 (85)	958 (87)	798 (73)	883 (81)	1737 (79)	1841 (84)
Karnofsky/Lansky score at transplant
≥ 90	682 (62)	695 (63)	574 (53)	648 (60)	1256 (57)	1343 (61)
Treatment region
US	1076 (97)	1056 (96)	1050 (97)	1053 (97)	2126 (97)	2109 (96)
Disease-related variables
Disease/Disease status a
AML, early	248 (22)	248 (22)	43 ( 4)	4 ( 4)	291 (13)	291 (13)
AML, intermediate	93 ( 8)	93 ( 8)	14 ( 1)	14 ( 1)	107 ( 5)	107 ( 5)
AML, advanced	143 (13)	143 (13)	0	0	143 ( 7)	143 ( 7)
ALL, early	204 (18)	204 (18)	3 (<1)	3 (<1)	207 ( 9)	207 ( 9)
ALL, intermediate	103 ( 9)	103 ( 9)	0	0	103 ( 5)	103 ( 5)
ALL, advanced	47 ( 4)	47 ( 4)	0	0	47 ( 2)	47 ( 2)
CLL	28 ( 3)	28 ( 3)	1 (<1)	1 (<1)	29 ( 1)	29 ( 1)
Other leukemia	17 ( 2)	17 ( 2)	0	0	17 (<1)	17 (<1)
CML, early	13 ( 1)	13 ( 1)	0	0	13 (<1)	13 (<1)
CML, intermediate	26 ( 2)	26 ( 2)	0	0	26 ( 1)	26 ( 1)
CML, advanced	6 (< 1)	6 (<1)	0	0	6 (<1)	6 (<1)
MDS	101 ( 9)	101 ( 9)	0	0	101 ( 5)	101 ( 5)
MPS	19 ( 2)	19 ( 2)	0	0	19 (<1)	19 (<1)
HL	3 (<1)	3 (<1)	143 (13)	143 (13)	146 ( 7)	146 ( 7)
MM	7 (<1)	7 (<1)	615 (57)	615 (57)	622 (28)	622 (28)
Amyloidosis	0	0	20 ( 2)	20 ( 2)	20 (<1)	20 (<1)
Other plasma cell disorder	0	0	15 ( 1)	15 ( 1)	15 (<1)	15 (<1)
JMML, JCML, JCMML	2 (<1)	2 (<1)	0	0	2 (<1)	2 (<1)
Marginal zone lymphomas	1 (<1)	1 (<1)	0	0	1 (<1)	1 (<1)
Follicular lymphoma (I, II, IIIa)	9 (<1)	9 (<1)	17 ( 2)	17 ( 2)	26 ( 1)	26 ( 1)
Follicular lymphoma (IIIb)	3 (<1)	3 (<1)	10 (<1)	10 (<1)	13 (<1)	13 (<1)
Mantle cell lymphoma	4 (<1)	4 (<1)	49 ( 5)	49 ( 5)	53 ( 2)	53 ( 2)
DLBCL	10 (<1)	10 (<1)	128 (12)	128 (12)	138 ( 6)	138 ( 6)
High-grade B-cell lymphoma	1 (<1)	1 (<1)	2 (<1)	2 (<1)	3 (<1)	3 (<1)
Waldenstrom’s macroglobulinemia	1 (<1)	1 (<1)	0	0	1 (<1)	1 (<1)
Other B-cell lymphoma	0	0	5 (<1)	5 (<1)	5 (<1)	5 (<1)
Rare T-cell lymphoma	5 (<1)	5 (<1)	0	0	5 (<1)	5 (<1)
T-cell lymphoma	7 (<1)	7 (<1)	20 ( 2)	20 ( 2)	27 ( 1)	27 ( 1)
NK lymphoma/leukemia	3 (<1)	3 (<1)	2 (<1)	2 (<1)	5 (<1)	5 (<1)
Transplant-related variables
Donor
Autologous	0	0	1087	1087	1087 (50)	1087 (50)
HLA-identical sibling	408 (37)	408 (37)	0	0	408 (19)	408 (19)
Identical twin	6 (<1)	6 (<1)	0	0	6 (<1)	6 (<1)
Mismatched related	4 (<1)	4 (<1)	0	0	4 (<1)	4 (<1)
Matched unrelated	389 (35)	389 (35)	0	0	389 (18)	389 (18)
Mismatched unrelated	121 (11)	121 (11)	0	0	121 ( 6)	121 ( 6)
Cord blood	176 (16)	176 (16)	0	0	176 ( 8)	176 ( 8)
Cord blood, recipient +	104 ( 9)	115 (10)	0	0	104 ( 5)	115 ( 5)
Cord blood, recipient −	68 ( 6)	61 ( 6)	0	0	68 ( 3)	61 ( 3)
Cord blood, recipient not reported	4 (<1)	0	0	0	4 (<1)	0
Not reported	1 (<1)	2 (<1)	0	0	1 (<1)	2 (<1)
Conditioning regimen, alloHCT
MAC - TBI/cy	429 (39)	472 (43)	N/A	N/A	429 (39)	472 (43)
MAC - TBI/cy/others	165 (15)	160 (14)			165 (15)	160 (14)
MAC - TBI ± others	209 (19)	121 (11)			209 (19)	121 (11)
MAC - TBI	1 (<1)	0			1 (<1)	0
MAC - bu/cy	86 ( 8)	72 ( 7)			86 ( 8)	72 ( 7)
MAC - bu/cy/others	36 ( 3)	29 ( 3)			36 ( 3)	29 ( 3)
MAC - bu/mel	2 (<1)	4 (<1)			2 (<1)	4 (<1)
MAC - bu/mel/others	12 ( 1)	0			12 ( 1)	0
MAC - bu/flu ± others	41 ( 4)	75 ( 7)			41 ( 4)	75 ( 7)
MAC - bu ± others	30 ( 3)	3 (<1)			30 ( 3)	3 (<1)
MAC - flu/mel ± others	3 (<1)	5 (<1)			3 (<1)	5 (<1)
MAC - treosulfan-based	0	4 (<1)			0	4 (<1)
RIC - TBI/cy	1 (<1)	0			1 (<1)	0
RIC - TBI/cy/others	15 ( 1)	12 ( 1)			15 ( 1)	12 ( 1)
RIC - TBI	1 (<1)	1 (<1)			1 (<1)	1 (<1)
RIC - TBI200+bu/flu	0	7 (<1)			0	7 (<1)
RIC - TBI200+flu/mel	0	3 (<1)			0	3 (<1)
RIC - TBI/others	4 (<1)	14 ( 1)			4 (<1)	14 ( 1)
RIC - BEAM	1 (<1)	2 (<1)			1 (<1)	2 (<1)
RIC - flu/mel	37 ( 3)	33 ( 3)			37 ( 3)	33 ( 3)
RIC - flu/bu	9 (<1)	23 ( 2)			9 (<1)	23 ( 2)
RIC - Others	4 (<1)	3 (<1)			4 (<1)	3 (<1)
NST - TBI/flu/cy +−other	0	36 ( 3)			0	36 ( 3)
NST - TBI/flu	1 (<1)	16 ( 1)			1 (<1)	16 ( 1)
NST - flu/cy/+−others	1 (<1)	5 (<1)			1 (<1)	5 (<1)
NST - others	5 (<1)	3 (<1)			5 (<1)	3 (<1)
Not reported	11 (<1)	1 (<1)			11 (<1)	1 (<1)
Conditioning regimen, autoHCT
TBI/cy	N/A	N/A	16 ( 1)	26 ( 2)	16 ( 1)	26 ( 2)
TBI/cy/others			28 ( 3)	9 (<1)	28 ( 3)	9 (<1)
TBI			1 (<1)	0	1 (<1)	0
TBI/others			5 (<1)	15 ( 1)	5 (<1)	15 ( 1)
BEAM and similar			133 (12)	268 (25)	133 (12)	268 (25)
Bu/Cy/VP16			114 (10)	36 ( 3)	114 (10)	36 ( 3)
CBV or similar			3 (<1)	4 (<1)	3 (<1)	4 (<1)
Bu/Cy/Others(not VP16)			33 ( 3)	43 ( 4)	33 ( 3)	43 ( 4)
Bu/mel/Others			90 ( 8)	8 (<1)	90 ( 8)	8 (<1)
Melphalan			585 (54)	638 (59)	585 (54)	638 (59)
Melphalan + others			31 ( 3)	12 ( 1)	31 ( 3)	12 ( 1)
Thiotepa-based			10 (<1)	1 (<1)	10 (<1)	1 (<1)
Cy-based			2 (<1)	2 (<1)	2 (<1)	2 (<1)
Bu-based			33 ( 3)	22 ( 2)	33 ( 3)	22 ( 2)
Not reported			3 (<1)	3 (<1)	3 (<1)	3 (<1)
TBI
Yes	848 (77)	848 (77)	50 ( 5)	50 ( 5)	898(41)	898 (41)
No	256 (23)	256 (23)	1037 (95)	1037 (95)	1293 (59)	1293 (59)
GVHD prophylaxis
AutoHCT, none	0	0	1087	1087	1087 (50)	1087 (50)
CsA + MTX +/− other	118 (11)	116 (11)	0	0	118 ( 5)	116 ( 5)
CsA + other (not MTX)	158 (14)	170 (15)	0	0	158 ( 7)	170 ( 8)
FK506 + MTX +/− other	431 (39)	521 (47)	0	0	431 (20)	521 (24)
FK506 +/− other	325 (29)	242 (22)	0	0	325 (15)	242 (11)
T-cell depletion	17 ( 2)	29 ( 3)	0	0	17 (<1)	29 ( 1)
Other	55 ( 5)	26 ( 2)	0	0	55 ( 3)	26 ( 1)
Planned G-CSF, GM-CSF
No	664 (60)	744 (67)	564 (52)	296 (27)	1228 (56)	1040 (47)
Yes	438 (40)	359 (33)	523 (48)	731 (73)	961 (44)	1150 (52)
Not reported	2 (<1)	1 (<1)	0	0	2 (<1)	1 (<1)
Year of transplant
2006	51 ( 5)	76 ( 7)	127 (12)	135 (12)	178 ( 8)	211 (10)
2007	143 (13)	117 (11)	179 (16)	160 (15)	322 (15)	277 (13)
2008	215 (19)	296(27)	141 (13)	285 (26)	356 (16)	581 (27)
2009	214 (19)	256 (23)	159 (15)	203 (19)	373 (17)	459 (21)
2010	206 (19)	203 (18)	258 (24)	62 ( 6)	464(21)	265 (12)
2011	176 (16)	126 (11)	143 (13)	156 (14)	319 (15)	282 (13)
2012	93 ( 8)	24 ( 2)	76 ( 7)	73 ( 7)	169 ( 8)	97 ( 4)
2013	6 (<1)	6 (<1)	4 (<1)	13 ( 1)	10 (<1)	19 (<1)
Median follow-up of survivors, months	96 (13–148)	97 (13–147)	97 (12–150)	98 (12–148)	97 (12–150)	97 (12–148)

^aDisease status definitions ³⁶: Early (low risk): AML, ALL and other leukemia: CR1; CML: CP1; CLL: CR and nPR; MDS: RA, RARS, RCMD, RCMD/RS, MDS unclassifiable, isolated 5q-syndrome; MPS: DIPSS low risk; HL: CR1, CRU1; MM/other plasma cell disorder: sCR, CCR, CR; Intermediate risk: AML, ALL and other leukemia: CR2, CR3+; CML: CP2, AP; CLL: PR, never treated and relapse; MPS: DIPSS intermediate risk; HL/NHL: CR2/+, CRU2/+, sensitive; MM/other plasma cell disorder: VGPR, PR, MR, SD; Advanced (high risk): AML, ALL and other leukemia: not in remission, never treated; CML: BP; CLL: SD, progression; MDS: RAEB, RAEB-T, RAEB-1, RAEB-2, CMML;MPS: DIPSS high risk; HD/NHL: PIF, relapse untreated, relapse resistant; MM/other plasma cell disorder: relapse, progression

Abbreviations: ALL, acute lymphoblastic leukemia; Allo, allogeneic; AML, acute myelogenous leukemia; AP, advanced phase; Auto, autologous; BEAM, BCNU (carmustine), Etoposide, Ara-C (cytarabine), Melphalan; BP, blast phase; Bu, busulfan; CBV, Cytoxan, BCNU (carmustine), VP-16 (etoposide); CCR, conventional complete response; CLL, chronic lymphocytic leukemia; CML, chronic myelogenous leukemia; CP1, first chronic phase; CR1, first complete remission; CRU1, first complete remission unconfirmed; Cy, cyclophosphamide; DIPSS, Dynamic International Prognostic Scoring System; DLBCL, diffuse large B-cell lymphoma; Flu, fludarabine; FK506, tacrolimus; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor; GVHD, graft-versus-host disease; HL, Hodgkin lymphoma; HLA, human leukocyte antigen; JCML, juvenile chronic myelogenous leukemia; JCMML, juvenile chronic myelomonocytic leukemia; JMML, juvenile myelomonocytic leukemia; KGF, palifermin; MAC, myeloablative conditioning; MDS, myelodysplastic syndrome; Mel, melphalan; MM, multiple myeloma; MPS, myeloproliferative syndromes; MR, minimal response; MTX, methotrexate; NK, natural killer; nPR, nodular partial remission; NST, non-myeloablative stem cell transplantation; PIF, primary induction failure; PR, partial response; RA, refractory anemia; RAEB, refractory anemia with excess blasts; RARS, refractory anemia with ringed sideroblasts; RCMD, refractory cytopenia with multilineage dysplasia; RCMD/RS, refractory cytopenia with multilineage dysplasia and ringed sideroblasts; RIC, reduced-intensity conditioning; sCR, stringent complete response; SD, stable disease; TBI, total body irradiation; VGPR, very good partial response.

Outcomes

Overall survival

Overall survival at 8 years in the entire cohort was 42.4% (95% CI 40.8–43.9). For cases versus controls at 8 years, overall survival was 42.3% (95% CI 40.1–44.5) vs. 42.4% (95% CI 40.2–44.6) (P value = 0.959). In multivariate analysis, there was no difference between cases and controls (relative risk [RR] of mortality 1.01 [95% CI 0.91 – 1.12]) (Table 2 and Figure 2).

Table 2.

Multivariable analysis of overall mortality, relapse, new malignancy, and cataract surgery, for cases versus controls who underwent transplant between January 1,2006, and December 2013

KGF versus no KGF	RR (95% CI)	P
Overall mortality	1.01 (0.91 – 1.12)	0.7980
Relapse	1.06 (0.94 – 1.18)	0.3626
New malignancy	0.89 (0.67 – 1.18)	0.4243
Cataract surgery	1.05 (0.74 – 1.50)	0.7711

Abbreviations: KGF, palifermin

Figure 2. Probability of overall survival.

Cases: exposed to palifermin (KGF). Controls: not exposed to KGF.

Relapse

The cumulative incidence of relapse at 8 years in the entire cohort was 51.6% (95% CI 50.1–53.1). For cases versus controls at 8 years, relapse was 51.5% (95% CI 49.4–53.7) vs. 51.7% (95% CI 49.5–53.9) (P value = 0.906). In multivariate analyses, there was no difference between cases and controls (RR 1.06 [95% CI 0.94 – 1.18]) (Table 2).

New malignancy

The cumulative incidence of new malignancy at 8 years in the entire cohort was 8.8% (95% CI 7.9–9.7). For cases versus controls at 8 years, new malignancy was 7.4% (95% CI 6.3–8.6) vs. 10.1% (95% CI 8.8–11.5) (P value = 0.002). In multivariate analyses, there was no difference between cases and controls (RR 0.89 [95% CI 0.67 – 1.18]) (Table 2 and Figure 3, eTable 1 and eFigures 1–3).

Figure 3. Cumulative incidence of new malignancy.

Cases: exposed to palifermin (KGF). Controls: not exposed to KGF.

Cataract surgery

The cumulative incidence of cataract surgery at 8 years in the entire cohort was 6.9% (95% CI 6.2–7.8). For cases versus controls at 8 years, cataract surgery was 7.3% (95% CI 6.2–8.5) vs. 6.6% (95% CI 5.5–7.7) (P value = 0.352). In multivariate analyses, there was no difference between cases and controls (RR 1.05 [95% CI 0.74 – 1.50]) (Table 2).

Pancreatitis

The cumulative incidence of pancreatic cancer at 8 years in the entire cohort was 0.7% (95% CI 0.4–0.9). In univariate analysis, at 8 years there was no difference in the risk of pancreatitis between cases and controls (0.7% [95% CI 0.4–1.1] vs. 0.6% [95% CI 0.3–1]; P value = 0.7) (eTable 2).

Renal failure requiring dialysis

The cumulative incidence of renal failure at 8 years in the entire cohort was 8.9% (95% CI 8.1–9.8). In univariate analysis, at 8 years there was no difference in the risk of renal failure requiring dialysis between cases and controls (9.2% [95% CI 8–10] vs. 8.7% [95% CI 7.5–9.9]; P value = 0.56) (eTable 2).

Pulmonary complications: IPn, ARDS, IPS

The cumulative incidence of IPn/ARDS/IPS cancer at 8 years in the entire cohort was 9.1% (95% CI 8.2–9.9). In univariate analysis, at 8 years there was there was no difference in risk between cases and controls (8.7% [95% CI 7.6–9.9] vs. 9.4% [95% CI 8.2–10.6]; P value = 0.44) (eTable 2).

Pulmonary complications: BO, COP, BOOP (allogeneic HCT recipients)

Among alloHCT recipients, the cumulative incidence of BO/COP/BOOP at 8 years was 5% (95% CI 4.1–5.9) The cumulative incidences of BO/COP/BOOP at 8 years in cases versus controls was 3.1% (95% CI 2.2–4.2) vs. 6.8% (95% CI 5.4–8.4) (P value < 0.001). (eTable 4). In multivariable analysis, the lower risk of these pulmonary complications among cases compared to controls was also observed (RR = 0.39 (95% CI 0.23 – 0.64); (eTable 3).

Acute and chronic GVHD (allogeneic HCT recipients)

Among alloHCT recipients, the 100-day cumulative incidences of acute GVHD grade 2–4, acute GVHD 3–4, and 1-year chronic GVHD in cases vs. controls were (36% [95% CI 33.2–39] vs. 40.7% [95% CI 37.7–43.6]; P value = 0.029), (17.5% [95% CI 15.3–19.9] vs. 19.2% [95% CI 16.9–21.7]; P value = 0.317) and (41% [95% CI 38–43.9] vs. 42.5% [95% CI 39.5–45.5]; P value = 0.478), respectively. The lower unadjusted incidence of 100-day acute GVHD grade 2–4 among cases compared to controls was not confirmed in multivariable analysis (eTable 3).

Discussion

Based on the fact that palifermin is an N-truncated human KGF with the proven ability to stimulate proliferation, differentiation, and survival of epithelial cells, there was a theoretical concern that KGF potentially could affect tumorigenesis or inhibit cytotoxicity of cancer treatments ^27,34. However, pre-clinical data demonstrated that KGF does not promote tumor growth in KGF-expressing cell lines or protect these tumor cells from the antitumor effects of several chemotherapeutic and biological agents ³⁵. Further, cells of the hematopoietic lineage do not express the KGF receptor as do cells of mesenchymal origin, and thereby it was not expected that secondary hematological malignancies would be promoted by the administration of KGF.

Although previous clinical studies demonstrated the efficacy of KGF to reduce severe OM, they were not designed to address key long-term safety outcomes ^12,19–23. Compared to this study, the RCTs and dose-escalation studies shown in eTable 6 included fewer patients, fewer diagnoses, and shorter follow-up. The largest followed 212 participants²⁰, while the longest had follow-up of 7 to 36 months²². Strengths of our study include the greater number of participants (more than 4,000), longer follow-up (median 8 years), and real-world evidence of safety across diverse age groups, hematologic malignancies, and conditioning regimens.

Our data demonstrate that over a median follow-up of 8 years there were no differences in survival, relapse, new malignancy or cataract surgeries among cases and controls. Similarly, based on univariate analyses, there are no increased risks for renal failure, pancreatitis, acute GVHD chronic GVHD or pulmonary complications associated with KGF. In multivariable analysis, the risk for the non-infectious pulmonary abnormalities of BO, COP, and BOOP was decreased among alloHCT recipients who received KGF versus controls. However, the absolute difference in incidence at 8 years of BO/COP/BOOP between cases and controls was quite small (4%), rendering this finding not clinically significant.

The current study was well powered and had sufficiently long follow-up to detect any clinically meaningful increase in risk of key post HCT outcomes. Furthermore, given the very broad inclusion criteria, the number of participating centers, and the clinicians’ discretion to prescribe KGF, the results are considered to be widely generalizable.

The study’s limitations are related to its observational design. Although matching and regression adjustment minimized confounder effects, it is not possible to determine the exact reason why a patient was prescribed KGF or not, as this was at the clinician’s discretion. Therefore, the existence of unmeasured confounders related to the clinician’s decision is unknown, though we think the possibility that these could influence these comparisons in a major way is small, given the overlapping curves for most of the outcomes. Additionally, the matching variables were chosen for their importance in predicting relapse and new malignancy, based on the main objectives of the study, and the same matching variables might not cover what is relevant for other objectives, which should be considered when interpreting these secondary endpoints. Finally, the data collection did not include details on the schedule or dose of KGF, nor was the study aimed at investigating oral mucositis outcomes.

In summary, KGF does not increase risk of key post-HCT long-term outcomes, such as mortality, disease relapse, new malignancies or cataracts. The results are consistent across different subgroups (e.g., adults and children). It can therefore be concluded that it is safe to use KGF to prevent oral mucositis after HCT.

Highlights.

• Long-term safety of palifermin (KGF) was confirmed

• There was no increased risk of overall mortality, relapse or new malignancies

• KGF safely prevents oral mucositis after hematopoietic cell transplantation