`© 2001 American Society for Blood and Marrow Transplantation
`
`ASBMT
`
`Randomized Clinical Trial of Thalidomide, Cyclosporine,
`and Prednisone Versus Cyclosporine and Prednisone as
`Initial Therapy for Chronic Graft-Versus-Host Disease
`
`Mukta Arora, John E. Wagner, Stella M. Davies, Bruce R. Blazar, Todd Defor, Helen Enright,
`Wesley J. Miller, Daniel J. Weisdorf
`
`Blood and Marrow Transplant Program, Departments of Medicine and Pediatrics, University of Minnesota,
`Minneapolis, Minnesota
`
`Correspondence and reprint requests: Daniel J. Weisdorf, MD, Division of Hematology, Oncology, and Transplantation,
`Box 480 Mayo, 420 Delaware St SE, University of Minnesota, Minneapolis, MN 55455 (e-mail: weisd001@tc.umn.edu).
`
`Received November 11, 2000; accepted March 8, 2001
`
`ABSTRACT
`Chronic graft-versus-host disease (CGVHD) is a major cause of morbidity following allogeneic bone marrow trans-
`plantation. Thalidomide is active in salvage therapy for high-risk or resistant CGVHD. In a prospective randomized
`trial, we tested initial therapy with thalidomide. Patients with extensive CGVHD were randomized to receive either
`cyclosporine and alternate-day prednisone (n = 27, no-thalidomide [no-thal] group) or cyclosporine, prednisone,
`and thalidomide (200-800 mg/day; n = 27, thal group). Although most patients responded, initial therapy with thal-
`idomide did not improve control of CGVHD. Response rates were 83% versus 89% at 2 months (P = .7), 88% versus
`84% at 6 months (P > .8) and 85% versus 73% at 1 year (P = .5) in the thal and no-thal groups, respectively. Multi-
`variate analysis revealed related donor transplant (odds ratio [OR] = 11.3; P =.03) and de novo or quiescent onset of
`CGVHD (OR = 7.7; P =.04) to be significant predictors of good early response, whereas a platelet count of
`≥100,000/µL was a significant predictor of good response (OR = 10.4; P =.04) at 1 year. Survival for the thal and no-
`thal groups was similar at 1 year (66% versus 74%) and 2 years (66% versus 54%, P = .85). Multivariate analysis
`revealed progressive onset CGVHD (relative risk [RR] = 4.2; P =.01), unrelated donor (RR = 5.7; P < .01), sex mis-
`match (RR = 7.9; P < .01), and platelet counts of <100,000/µL (RR = 3.8; P = .01) as significant predictors of poorer
`survival. These data suggest that despite a high response rate (79% response and 53% complete response) and
`encouraging survival rates (70% at 1 year and 60% at 2 years), thalidomide offers no clinical benefit when incorpo-
`rated into initial therapy for CGVHD. The value of thalidomide as salvage therapy requires further study.
`
`KEY WORDS
`Chronic graft-versus-host disease
`
`• Thalidomide
`
`• Randomized clinical trial
`
`INTRODUCTION
`Chronic graft-versus-host disease (CGVHD) is a late
`complication of bone marrow transplantation (BMT) charac-
`terized by a connective tissue–like disease that usually, but not
`always, occurs more than 100 days following BMT. Clinically,
`CGVHD may be manifested by involvement of the skin, oral
`mucosa, gastrointestinal tract, eyes, liver, lungs, and joints
`[1-4]. CGVHD develops in 30% to 60% of transplantation
`survivors [5,6] and remains a major cause of morbidity and
`mortality following allogeneic BMT. The major cause of
`death in patients with CGVHD is infection due to continued
`immunodeficiency [4,7]. Poor prognostic factors that have
`been identified include thrombocytopenia (platelet count,
`<100,000/µL), increased age, lichenoid skin pathology, liver
`
`involvement, and progressive presentation of CGVHD [1,8]
`(CGVHD developing before resolution of acute GVHD).
`Prednisone alone or in combination with other immuno-
`suppressive agents has been the standard treatment of
`CGVHD [6,9]. Thalidomide has also been investigated as
`an immunosuppressive agent active in the therapy of
`CGVHD [5,10,11], primarily as a salvage agent in patients
`with CGVHD resistant to other therapies or in high-risk
`patients [10,11]. Thalidomide’s role in early therapy of
`CGVHD is not yet established. This study was designed to
`test the efficacy of initial CGVHD therapy with thalidomide
`in addition to cyclosporine and alternate-day prednisone in
`an open label, prospective randomized trial. We report an
`analysis of the response rate and survival in addition to a
`
`B B & M T
`
`265
`
`
`
`M. Arora et al.
`
`comparison of the frequency of complications including
`infections in patients treated with and without thalidomide.
`
`METHODS
`Eligibility Criteria
`Patients who underwent allogeneic BMT at the Univer-
`sity of Minnesota between September 1993 and April 1999
`and developed extensive CGVHD were eligible for inclusion
`in the trial. Extensive CGVHD was defined as involvement
`of 1 or more of the following organ systems: generalized skin
`involvement (≥50% body surface area), liver involvement
`(bilirubin, ≥3 mg/dL), positive Schirmer’s test, histologically
`proven CGVHD of the oral mucosa, lung dysfunction with
`bronchiolitis obliterans, and gastrointestinal involvement
`with malabsorption and/or weight loss unexplained by other
`etiologies. High-risk CGVHD was defined as the presence
`of 1 or more of 3 risk factors: progressive onset, platelet
`count of <100,000/µL, and bilirubin level of ≥3 mg/dL.
`Fifty-four patients were enrolled and randomized. Patients
`were excluded if they had acute complications resulting in a
`life expectancy of <1 month, were aged <2 or >60 years, had
`an inability to take oral medications, had peripheral neuropa-
`thy, or were pregnant or intended to become pregnant.
`Patients were instructed about the hazards of birth defects
`associated with prenatal exposure to thalidomide, and sexu-
`ally active patients were required to use effective contracep-
`tion while taking thalidomide. Before randomization,
`patients were stratified into cohorts according to whether
`they had received related donor or unrelated donor marrow.
`Patients were randomly assigned to the thalidomide (thal)
`arm or no-thalidomide (no-thal) arm. Blocked randomiza-
`tion using block sizes of 4 and 6 was carried out. All patients
`or their parents/guardians gave signed informed consent
`according to guidelines approved by the University of Min-
`nesota Institutional Review Board. Thalidomide was sup-
`plied by Celgene Corporation (Warren, New Jersey) under
`terms of a Food and Drug Administration Investigational
`New Drug approval issued to the investigators (D.J.W.).
`
`Treatment Plan
`Beginning at the time of randomization, all patients
`received initial therapy with high-dose methylprednisone.
`Methylprednisone was given at a dose of 15 mg/kg as an
`intravenous injection weekly for 8 weeks. Patients assigned
`to the thal arm received prednisone, cyclosporine, and thal-
`idomide from the time of randomization. Those assigned to
`the no-thal arm received prednisone and cyclosporine. The
`initial dosage of thalidomide was 50 mg 4 times a day in
`adult patients and 0.75 mg/kg 4 times a day in pediatric
`patients. Dosage was increased as tolerated at 2 to 4 week
`intervals to a maximum of 200 mg 4 times a day in adults
`and 3 mg/kg 4 times a day in pediatric patients. Thalido-
`mide levels were not monitored. Prednisone was given at a
`dosage of 0.5 mg/kg by mouth on alternate days and cyclo-
`sporine was started at 6.25 mg/kg by mouth twice daily (or
`1.5 mg/kg intravenously twice daily) with the dosage being
`modified to maintain trough levels >200 ng/mL. In patients
`with no response to the initial 8 weeks of therapy or with
`disease progression after initial stabilization or response,
`methylprednisone was given at a dosage of 15 mg/kg per
`
`266
`
`day intravenously for 5 days, and oral prednisone was
`increased to 1 mg/kg per day by mouth for 6 weeks, then
`tapered over a 3-month period to 0.5 mg/kg per day on
`alternate days. Therapy was continued until 9 months fol-
`lowing the last clinical evidence of active CGVHD, followed
`by a taper over 2 to 3 months. In persistent or refractory
`disease, crossover to the thal arm was permitted, but only
`after 6 months of assigned therapy. All subjects randomized
`were analyzed as intention-to-treat in their assigned treat-
`ment groups.
`
`Measurement of Response
`Patients were evaluated for response to therapy at
`2 months, 6 months, and 1 year from randomization.
`Because patients did not always return to the transplantation
`center on the scheduled dates, window periods for evalua-
`tion were used to maximize the completeness of evalua-
`tion. For the 2-month visit, the evaluation window period
`was between 1 and 3 months from randomization. For the
`6-month and 1-year visits, window periods of 4 to 9 months
`and 9 to 16 months, respectively, were used. All patients
`included in the study were randomized at least 6 months
`prior to study termination. The median follow-up of surviv-
`ing patients was 22 months (range, 6-58 months).
`Response to therapy was graded as complete response
`(CR), defined as resolution of all signs and symptoms of
`CGVHD. Partial response (PR) was defined as improve-
`ment in 1 or more organs of involvement and no evidence of
`worsening in any organ. Flare was defined as PR or CR fol-
`lowed by worsening of CGVHD to a severity less than that
`at baseline evaluation. No response was defined as either
`progression of CGVHD to worse than at baseline evalua-
`tion or no improvement in CGVHD after 6 months of ther-
`apy. Prevalence of CGVHD was defined as the proportion
`of all patients with active CGVHD among surviving
`patients. Improvement or worsening of disease was deter-
`mined through both subjective and objective criteria. Sub-
`jective criteria were symptomatic changes in cough, dyspnea,
`anorexia, nausea, vomiting, diarrhea, arthralgia, or dry eyes.
`Objective criteria included physical exams of skin and oral
`mucosa, weight measurements, liver function tests, pul-
`monary function tests, Schirmer’s test, biopsies, and radio-
`logical studies. Patients who had no response or flare before
`they died were included at subsequent time points as nonre-
`sponders or treatment failures.
`
`STATISTICAL ANALYSIS
`The study was designed as a prospective randomized
`trial. The expected high incidence of complications caused
`by thalidomide (chiefly constipation and somnolence) pre-
`cluded blinding of the study drug, so no placebo therapy
`was used. However, data collection, grading of response,
`and, importantly, statistical analysis were performed without
`knowledge of treatment-group assignment. Analysis of
`response and survival were performed before any examina-
`tion of side effects to maintain the blinding during analysis.
`
`Prestudy Sample Size and Power Projections
`The primary endpoint of the study was response to
`treatment; the secondary endpoint was overall survival. To
`
`
`
`Thalidomide as Initial Therapy for CGVHD
`
`detect a 20% difference in response (presuming a response
`rate of 65% in the control arm) with 80% power at α of
`0.05, a sample size of 134 patients (half randomized to each
`arm) was needed. However, after accrual of 54 randomized
`patients, this blinded interim analysis was performed.
`Because both treatment arms had response rates higher than
`projected, with only a 12% difference in response between
`the 2 treatment arms, recalculation of the sample size to
`detect a >20% difference would have necessitated an added
`enrollment of 104 patients. This study size was unworkable,
`so the study was closed.
`
`recipient age. Acute GVHD (grade III-IV) was excluded
`from the model because of colinearity between progressive
`onset and acute GVHD.
`
`Complications After Transplantation
`Complications studied included hypertension, hyper-
`glycemia requiring treatment, constipation, somnolence,
`seizures, neuropathy, thrombotic thrombocytopenic purpura
`(TTP), avascular necrosis, and infections. Pearson’s chi-
`square test was employed to compare the proportion of
`patients with complications in the 2 treatment groups.
`
`Response to Therapy
`Pearson’s chi-square test was employed to compare the
`proportion of subjects who responded to therapy with those
`who did not (complete and partial response versus no
`response and flare) at 2 months, 6 months, and 1 year after
`randomization. The Fisher exact test was used when the
`expected cell count was less than 5.
`
`Infections Following BMT
`To account for multiple events, density incidence was
`used to describe the total rate of infections. Density inci-
`dence was defined as the total number of infections per
`1000 patient-days. Statistical comparison of the density
`incidence was done by using the Mantel-Haenszel chi-
`square test [14] for person-days data.
`
`Predictors of Response
`Thirteen potential predictors were evaluated. These
`included treatment arm, recipient age at transplantation, sex
`of recipient and donor, cytomegalovirus serologic status of
`recipient and donor, type of transplant (allogeneic sibling
`versus unrelated donor), HLA mismatch, prior presence and
`clinical grade of acute GVHD, onset of CGVHD (de novo,
`progressive, quiescent), organ involvement with CGVHD
`(eyes, mouth, skin, lungs, gastrointestinal tract, liver), white
`blood cell count, platelet count, and serum bilirubin and
`alanine aminotransferase levels at baseline. Pearson’s chi-
`square test was employed in the univariate analyses to com-
`pare the proportion of subjects with response to therapy
`within each category of potential predictors. Multivariate
`logistic regression was used to evaluate the independent
`effect of study variables on treatment response. A stepwise
`regression with forward selection was used. A variable had
`to be significant at the 0.10 level before it could enter into
`the model and had to be significant at the 0.15 level for it to
`remain in the regression model.
`
`Survival
`Patient survival was determined using the Kaplan-Meier
`[12] estimation with 95% confidence intervals derived from
`standard errors. Patients were censored at the date of last
`contact. Comparison of survival between the 2 treatment
`groups was carried out using Kaplan-Meier plots and log-
`rank tests [12].
`
`Predictors of Mortality
`Potential factors associated with effects on mortality
`were studied. The Kaplan-Meier product limit method [12]
`was used to compare survival in the subsets, and the Cox
`regression model [13] was used to assess the independent
`effect of the predictors on survival as well as any potential
`confounding of the effect of the randomized treatment. A
`stepwise regression with forward selection was used with a
`significance level of P = .10 being needed to enter into the
`model and a significance level of P = .15 to remain in the
`model. Recipient age was not included in the model because
`of nonproportionality, and the results were stratified by
`
`RESULTS
`The clinical characteristics of the 54 patients are shown
`in Table 1. There were 27 patients in each treatment group.
`Patients randomized to receive thalidomide were slightly
`older than those who did not receive it (median age, 44 versus
`37 years; P = .01), but there were no significant differences in
`pretransplantation diagnoses, types of transplant donors,
`conditioning regimens, prophylaxis for acute GVHD, pres-
`ence or stage of acute GVHD, or pattern of onset and organ
`involvement with CGVHD. Overall, 27 patients had high-
`risk CGVHD. Nine patients (5 receiving thalidomide and
`4 receiving no thalidomide) had progressive onset of
`CGVHD. Six patients (3 in each group) had hyperbilirubine-
`mia, and 22 patients (10 in the no-thal group and 12 in the
`thal group) had significant thrombocytopenia at the time of
`diagnosis of CGVHD. Eight of the 27 patients (4 in the thal
`group, 4 in the no-thal group) had 2 high-risk factors,
`whereas 3 patients (1 in the thal group, 2 in the no-thal
`group) had all 3 high-risk factors at diagnosis of CGVHD.
`
`Response to Therapy
`As shown in Table 2, the response to combination
`immunosuppressive therapy was evaluated at 2 months,
`6 months, and 1 year from randomization. Of the
`54 patients enrolled, the number of patients evaluable for
`response was 51 at 2 months, 49 at 6 months, and 42 at
`1 year. Three patients in the thal group died within 1 month
`of starting therapy and were unevaluable. Two patients in
`the no-thal group relapsed and died and thus had no
`response assessed at 6 months and 1 year. Two additional
`patients in the thal group died and were not evaluable at
`1 year. Five patients (2 in the thal group and 3 in the no-thal
`group) have not as yet completed 1 year of therapy.
`Complete and partial responses were observed in 86%,
`85%, and 79% of the patients at 2 months, 6 months, and
`1 year, respectively. At 1 year, 22 patients (52%) had a CR.
`Response rates were similar in both treatment groups at all
`3 time points. Clinical response rates (CR + PR) in the thal
`and no-thal groups were similar at 2 months (83% versus
`89%, P = .7), 6 months (88% versus 84%, P > .8) and 1 year
`
`B B & M T
`
`267
`
`
`
`M. Arora et al.
`
`Table 1. Clinical Characteristics at Study Entry*
`
`Thal Group, No-Thal
`n (%)
`Group, n (%) P
`
`27 (50)
`
`12 (44)
`12 (45)
`2 (7)
`1 (4)
`
`Patient Demographics
`No. of patients
`Diagnosis
`Acute leukemia
`CML
`Other malignancies
`Nonmalignant Diseases
`Age, y
`Median (range)
`<20
`20-29
`30-39
`≥40
`Donor/recipient sex mismatch
`Male recipient with female donor
`Others
`Donor/recipient CMV
`sero-status (either or both +)
`Type of transplant
`18 (67)
`Sibling donor
`9 (33)
`Unrelated donor
`3 (17)
`HLA mismatch (related donor)
`HLA mismatch (unrelated donor) 1 (11)
`Conditioning regimens
`TBI + cyclophosphamide
`Others
`Prophylaxis for acute GVHD
`Methotrexate + cyclosporine
`T-cell depletion
`Others
`Acute GVHD
`Grade II-IV acute GVHD
`Grade III-IV acute GVHD
`
`44 (17-60)
`1 (4)
`2 (7)
`7 (26)
`17 (63)
`
`5 (19)
`17 (63)
`17 (63)
`
`25 (92)
`2 (8)
`
`20 (74)
`5 (19)
`2 (7)
`
`18 (67)
`7 (26)
`
`27 (50)
`
`12 (44)
`11 (41)
`3 (11)
`1 (4)
`
`37 (12-50)
`5 (19)
`5 (19)
`5 (19)
`12 (43)
`
`4 (15)
`19 (70)
`19 (70)
`
`16 (59)
`11 (41)
`1 (6)
`4 (36)
`
`23 (85)
`4 (15)
`
`22 (81)
`4 (15)
`1 (4)
`
`20 (74)
`6 (22)
`
`.39
`
`.01
`
`>.8
`
`.77
`
`.77
`
`.60
`.31
`
`.66
`
`.50
`
`.55
`.75
`
`(85% versus 73%, P = .5). In patients with high-risk disease,
`response rates were similar over time and did not differ in
`the 2 treatment groups (Table 2), although a few more
`patients in the no-thal group had either no response or flare
`of CGVHD at their 6-month and 1-year evaluations.
`
`Predictors of Response
`We analyzed 13 clinical factors as potential predictors of
`response at 2 months and 1 year (Table 3). Similar findings
`were observed at 6 months (not shown). Use of thalidomide
`was not associated with more frequent responses. In univari-
`ate analysis, more frequent response to therapy was seen in
`recipients of related donor transplants and in patients with
`de novo/quiescent onset of CGVHD. Recipients of related
`donor transplant had more frequent early responses (94%
`versus 72%, P = .08). More frequent early responses were
`also seen in patients with de novo and quiescent onset of
`CGVHD (91% versus 63% at 2 months, P = .06 and 84%
`versus 40% at 1 year, P = .05, respectively). Involvement of
`the skin also identified patients with more frequent
`responses at 2 months and 1 year.
`In the multivariate analysis (Table 4), related donor
`transplant (odds ratio [OR] = 11.3; P = .03) and de novo or
`quiescent onset of disease (OR = 7.7; P = .04) were signifi-
`cant predictors of more frequent early (2 months) response.
`A platelet count of >100,000/µL was a significant predictor
`of more frequent late (1 year) response (OR = 10.4; P = .04).
`Involvement of skin was an independently significant pre-
`dictor of more frequent early as well as late responses (OR =
`9.7, P = .03 and OR = 25, P < .01, respectively). Randomiza-
`tion to thalidomide had no significant impact on response to
`therapy at any time point.
`Of note, patients with lung involvement also showed a
`high response. Eight of 10 patients in the thal group and
`7 of 9 patients in the no-thal group showed a clinical
`response at 1 year.
`
`Survival
`Thalidomide therapy for CGVHD did not lead to
`improved survival. After a median follow-up of 22 months
`(range 6-58 months), it was determined that at 2 years thal-
`idomide-treated patients had survival rates similar to those
`of the control cohort (Figure 1) (range, 66% and 54%,
`respectively, P = .85) with no deaths observed in surviving
`patients beyond 2 years. Patients with high-risk disease had
`an actuarial survival of 53% (range, 33%-73%) 1 year after
`transplantation and 44% (range, 21%-67%) 2 years after
`transplantation. Therapy with thalidomide did not alter sur-
`vival time in the higher-risk subgroup (2-year survival rate
`of 26% versus 28%, respectively, in the high-risk thal and
`no-thal groups, respectively, P = .5).
`Overall, 19 patients died, 10 in the no-thal group and
`9 in the thal group (P = .8). Three patients in the thal group
`died within 1 month of starting therapy because of infec-
`tious complications. Four patients in the thal group and
`8 patients in the no-thal group died with severe unrespon-
`sive CGVHD. Two patients relapsed, both in the no-thal
`group. One patient in the thal group developed progressive
`multifocal leukoencephalopathy and 1 other patient in the
`thal group with severe continuing pancytopenia died of bac-
`terial and fungal infections after 1 year of therapy.
`
`Characteristics of CGVHD
`Onset of CGVHD
`De novo
`Quiescent
`Progressive
`Time to treatment (days after
`transplant) median (range)
`Organ involvement
`Skin
`Oral
`Liver
`Gastrointestinal
`Lungs
`Eyes
`High-risk CGVHD†
`Standard risk
`Platelet count,
`median (range)
`WBC count,
`median (range)
`Bilirubin ≥3 mg/dL
`Alanine aminotransferase
`>250 units/dL
`
`4 (15)
`6 (22)
`19 (70)
`16 (59)
`4 (15)
`5 (19)
`190 (63-794) 194 (75-812) >.8
`
`>.8
`
`21 (78)
`16 (60)
`21 (78)
`24 (89)
`9 (33)
`12 (44)
`17 (63)
`21 (78)
`14 (52)
`9 (33)
`9 (33)
`13 (48)
`13 (48%)
`14 (52)
`14 (52%)
`13 (48%)
`118 (17-352) 105 (39-384)
`
`5150 (2100- 3900 (1400-
`15,000)
`8400)
`3 (11)
`3 (11)
`4 (15)
`5 (19)
`
`.14
`.27
`.4
`.23
`.16
`.26
`.8
`
`.8
`
`.18
`
`>.8
`.71
`
`*P values reflect chi square or the Fisher exact test comparisons between
`treatment groups. CML indicates chronic myelogenous leukemia; CMV,
`cytomegalovirus; TBI, total body irradiation; GVHD, graft-versus-host
`disease; plts, platelet count; WBC, white blood cell count.
`†High-risk CGVHD was defined as presence of either progressive
`onset of disease, bilirubin ≥3 mg/dL or platelet count of <100,000/µL.
`
`268
`
`
`
`Thalidomide as Initial Therapy for CGVHD
`
`Table 2. Patient Evaluability and Frequency of Response to Therapy*
`
`2-Month Follow-up
`No-Thal
`
`Thal
`
`All patients
`Evaluable patients
`CR, n (%)
`PR, n (%)
`CR+PR, n (%)
`NR+Flare, n (%)
`High-risk CGVHD patients
`Evaluable patients
`CR, n (%)
`PR, n (%)
`CR+PR, n (%)
`NR+Flare, n (%)
`
`24
`2 (8)
`18 (75)
`20 (83)
`4 (17)
`
`11
`1 (9)
`8 (73)
`9 (82)
`2 (18)
`
`27
`2 (7)
`22 (81)
`24 (89)
`3 (11)
`
`13
`1 (8)
`9 (69)
`10 (77)
`3 (23)
`
`P
`
`.7
`
`>.8
`
`6-Month Follow-up
`No-Thal
`
`Thal
`
`P
`
`Thal
`
`1-Year Follow-up
`No-Thal
`
`24
`4 (17)
`17 (71)
`21 (88)
`3 (12)
`
`11
`2 (18)
`8 (73)
`10 (91)
`1 (9)
`
`25
`7 (28)
`14 (56)
`21 (84)
`4 (16)
`
`12
`1 (8)
`7 (59)
`8 (67)
`4 (33)
`
`>.8
`
`.3
`
`20
`10 (50)
`7 (35)
`17 (85)
`3 (15)
`
`8
`3 (38)
`4 (50)
`7 (88)
`1 (12)
`
`22
`12 (55)
`4 (18)
`16 (73)
`6 (27)
`
`9
`3 (33)
`1 (11)
`4 (44)
`5 (56)
`
`P
`
`.5
`
`.1
`
`*P value represents Pearson’s chi-square comparisons between patients with CR+PR versus NR+Flare at 2 months, 6 months, and 1 year.
`CR indicates complete response; PR, partial response; NR, no response; CGVHD, chronic graft-versus-host disease.
`
`Table 3. Predictors of Response: Univariate Analysis*
`
`Predictor
`
`2-Month Follow-up
`No. of Evaluable
`Patients
`
`Responders, %
`
`51
`
`24
`27
`
`86
`
`83
`89
`
`P
`
`.7
`
`.4
`
`1-Year Follow-up
`No. of Evaluable
`Patients
`
`Responders, %
`
`42
`
`20
`22
`
`79
`
`85
`73
`
`80
`75
`
`P
`
`.4
`
`.7
`
`All patients
`Treatment group
`Thalidomide
`No thalidomide
`Age
`≥30 years
`<30 years
`Sex mismatch
`Male recipient:female donor
`Others
`Type of Transplant
`Unrelated donor
`Related donor
`Acute GVHD
`Grade III-IV
`Grade 0-II
`Onset
`Progressive
`Denovo + quiescent
`Organ involvement with CGVHD
`Skin
`None
`Oral
`None
`Eye
`None
`Lungs
`None
`Gastrointestinal
`None
`Liver
`None
`Platelets <100,000/µL
`Platelets ≥ 100,000/µL
`Bilirubin ≥3 mg/dL
`Bilirubin <3 mg/dL
`
`38
`13
`
`7
`44
`
`18
`33
`
`12
`39
`
`8
`43
`
`34
`17
`43
`8
`22
`29
`21
`30
`37
`14
`21
`30
`19
`32
`5
`46
`
`89
`77
`
`100
`84
`
`72
`94
`
`75
`90
`
`63
`91
`
`94
`71
`91
`75
`91
`83
`86
`87
`86
`86
`81
`90
`79
`91
`60
`89
`
`.5
`
`.08
`
`.3
`
`.06
`
`.03
`
`.2
`
`.7
`
`>.8
`
`>.8
`
`.4
`
`.4
`
`.13
`
`30
`12
`
`6
`36
`
`15
`27
`
`9
`33
`
`5
`37
`
`27
`15
`37
`5
`18
`24
`19
`23
`28
`14
`16
`26
`13
`29
`4
`38
`
`50
`83
`
`67
`85
`
`67
`82
`
`40
`84
`
`93
`53
`81
`60
`78
`79
`79
`78
`75
`86
`75
`81
`62
`86
`50
`82
`
`.1†
`
`.2
`
`.4
`
`.05†
`
`<.01†
`
`.3
`
`>.8
`
`>.8
`
`.7
`
`.7
`
`.1†
`
`.1†
`
`*Response refers to CR or PR at the time period shown. P values reflect chi-square tests of significance. GVHD indicates graft-versus-host dis-
`ease; CGVHD, chronic GVHD.
`†Value was entered into subsequent multivariate analysis.
`
`B B & M T
`
`269
`
`
`
`M. Arora et al.
`
`Table 4. Predictors of Response: Multivariate Analysis*
`
`Predictor
`
`2-Month Follow-up
`OR (CI)
`
`Related donor versus URD transplant
`De novo + quiescent versus progressive disease onset
`Platelets ≥100,000/µL versus <100,000/µL
`Skin versus no skin involvement
`Treatment arm (no-thal versus thal)
`
`11.3 (1.2-109)
`7.7 (1.8-67)
`5.3 (0.5-52)
`9.7 (1.2-81.9)
`4.1 (0.4-43.8)
`
`P
`
`.03
`.04
`.2
`.03
`.28
`
`1-Year Follow-up
`OR (CI)
`
`8.0 (0.8-83.6)
`5.6 (0.3-93.6)
`10.4 (1.02-105)
`25 (2-250)
`0.7 (0.09-5.9)
`
`P
`
`.1
`.2
`.04
`<.01
`.6
`
`*Logistic regression showing the independent odds ratio (OR) and 95% confidence interval (CI) favoring complete response or partial response
`at the evaluation period shown. URD indicates unrelated donor; thal indicates thalidomide.
`
`Predictors of Mortality
`In univariate analysis, several factors were associated
`with poor survival. Type of transplant (unrelated donor ver-
`sus related donor, P = .05), grade III-IV acute GVHD (P =
`.05), sex mismatch (male recipient with female donor versus
`other sex combinations, P < .01), progressive onset of dis-
`ease (versus de novo and quiescent, P = .01), platelet count
`of <100,000/µL (P < .01), and bilirubin level of ≥3 mg/dL
`(P = .03) were each significant predictors of poor outcome
`(Figure 2).
`In the multivariate analysis (Table 5), progressive onset
`of disease (relative risk [RR] = 4.2; P = .01), unrelated donor
`transplant (RR = 5.7; P < .01), male recipient:female donor
`(RR = 7.9; P < .01), and platelet count of <100,000/µL (RR =
`3.8; P = .01) were independently significant predictors of
`poor survival. Treatment with thalidomide had no indepen-
`dently significant impact on survival.
`
`Prevalence of CGVHD
`Following the onset of randomized therapy, we evalu-
`ated the prevalence (persistence of active CGVHD in sur-
`viving patients) of CGVHD over time. Prevalence assessment
`incorporates both flares and persistence of CGVHD and as
`prevalence falls it reflects a clinical cure of CGVHD. The
`prevalence of CGVHD decreased to 82% at 6 months, 35%
`at 1 year, and 28% at 2 years from randomization. Resolu-
`tion of CGVHD occurred at the same frequency and pace
`in patients assigned to the thal or no-thal group, resulting in
`
`Figure 1. Survival after randomization to initial therapy of chronic
`graft-versus-host disease with and without thalidomide. P value repre-
`sents log-rank tests of significance between the 2 treatment groups.
`
`270
`
`prevalence of 81% versus 83% at 6 months, 36% in both
`groups at 1 year, and 25% versus 30% at 2 years, respec-
`tively (P > .8) (Figure 3).
`
`Complications
`The anticipated complications of thalidomide, constipa-
`tion (52% versus 11% in the no-thal group; P < .01), sleepi-
`ness (63% versus 0%, P < .01), and neuropathy (13 patients,
`48% versus 3 patients, 11%; P < .01) were more common in
`the thal group (Table 6). Six patients (22%) had significant
`neuropathy requiring discontinuation of thalidomide. Thal-
`idomide was continued in patients with nonprogressive
`paresthesias, although these patients were monitored fre-
`quently. Fewer patients in the thal group than in the no-thal
`group had hypertension (30% versus 56%; P = .05). Hyper-
`glycemia, TTP/hemolytic uremic syndrome, seizures, and
`avascular necrosis occurred with similar frequency in the
`thal and no-thal groups.
`
`Infections
`As shown in Figure 4, more frequent infections were
`observed in patients not receiving thalidomide. The density
`incidence of all infections showed a trend toward more
`infections in the no-thal group (6.6 per 1000 person-days
`versus 4.6 per 1000 person-days; RR for thal group = 0.7;
`P = .07). Bacterial infections were more frequent in the no-
`thal group across all time periods (P = .04). Across all time
`periods, similar incidence densities of viral and fungal infec-
`tions were seen in the 2 treatment groups.
`
`Compliance With Thalidomide Therapy
`Of 27 patients assigned to the thal group, 13 of 18 sur-
`viving have remained on therapy for 1 year. One other
`patient continues on therapy at 6 months. Three patients
`stopped thalidomide within 3 months because of intolerable
`side effects (constipation, sleepiness, and neuropathy). Three
`other patients stopped treatment between 4 and 6 months
`because of neuropathy. One patient stopped both thalido-
`mide and cyclosporine at 5 months after he developed TTP.
`After the initial 6 months of therapy without thalidomide,
`2 patients in the no-thal group crossed over to add thalido-
`mide along with continuing cyclosporine and prednisone.
`One of these patients had no further response and died
`15 months later of bacterial and fungal infections. The sec-
`ond patient started thalidomide and achieved a CR at 1 year.
`He continued on combination therapy until 17 months, but
`died of infections at 23 months after initial randomization.
`
`
`
`Thalidomide as Initial Therapy for CGVHD
`
`Figure 2. Survival after randomization to initial therapy of chronic graft-versus-host disease (CGVHD) (as in Figure 1). D, Cohorts are divided by
`number of high-risk factors at onset of CGVHD. Factors included those shown in A, B, and C plus sex mismatch.
`
`the no-thal group) and a response rate of 66% (88% in the
`thal group and 44% in the no-thal group) in patients with
`high-risk CGVHD at 1 year that is similar to that reported
`for previous studies of initial CGVHD therapy. Sullivan et
`al. reported response rates in patients receiving alternating-
`day cyclosporine and prednisone of 56% and 71% in high-risk
`and refractory patients, respectively [16]. Using thalidomide,
`Vogelsang et al. [11] observed 38% and 78% response in
`high-risk and refractory patients, respectively. Parker et al.
`[10] reported only a 20% response to thalidomide in refractory
`
`DISCUSSION
`Previous reports have suggested that thalidomide has
`substantial clinical activity in treatment of CGVHD,
`although it has usually been administered as salvage therapy
`[5,10,11,15]. The current trial is the first prospective ran-
`domized study of thalidomide as initial therapy in patients
`with CGVHD. Our study was designed to compare the rate
`and completeness of response as well as survival in both
`treatment groups. Previous studies of thalidomide or ran-
`domized assessments of cyclosporine and prednisone have
`been conducted on high-risk groups or in patients refractory
`to conventional treatment [10,11,16]. In the current trial
`testing the utility of thalidomide as initial therapy of
`CGVHD, we observed a high overall response rate of 79%
`in the entire cohort (85% in the thal group versus 73% in
`
`Table 5. Factors Associated With Poor Survival After CGVHD:
`Multivariate Analysis*
`
`Predictor
`
`Progressive onset
`Unrelated donor transplant
`Sex mismatch (male recipient:female
`donor versus others)
`Platelets <100,000/µL
`Treatment (thalidomide versus no
`thalidomide)
`
`RR (CI)
`
`4.2 (1.3-13.2)
`5.7 (1.8-17.7)
`7.9 (2.3-16.3)
`
`3.8 (1.3-10.8)
`0.62 (.38-1.71)
`
`P
`
`.01
`<.01
`<.01
`
`.01
`.4
`
`*Results are stratified by recipient age (not included in the analysis
`because of nonproportionality). Shown is the relative risk (RR) (95%
`confidence interval [CI]) of mortality following therapy for chronic graft-
`versus-host disease (CGVHD).
`
`Figure 3. Prevalence of chronic graft-versus-host disease over time in
`patients treated with immunosuppression including thalidomide (thal)
`(black bars) or no thalidomide (no thal) (gray bars). P > .8 for compar-
`isons at all time points.
`
`B B & M T
`
`271
`
`
`
`M. Arora et al.
`
`Table 6. Complications of Therapy*
`
`Complications
`
`Thal Group, n (%) No-Thal Group, n (%)
`
`P
`
`Hypertension
`Diabetes mellitus
`Constipation
`Sleepiness
`Seizures
`Neuropathy
`TTP/HUS
`Avascular necrosis
`
`8 (30)
`3 (11)
`14 (52)
`17 (63)
`1 (4)
`13 (48)
`2 (7)
`1 (4)
`
`15 (56)
`6 (22)
`3 (11)
`0 (0)
`0 (0)
`3 (11)
`1 (4)
`3 (11)
`
`.05
`.3
`<.01
`<.01
`>.8
`<.01
`>.8
`.6
`
`*The frequency of complications in both treatment groups is
`shown. Those requiring therapy (hypertension, hyperglycemia,
`seizures, thrombotic thrombocytopenic purpura/hemolytic uremic syn-
`drome [TTP/HUS]) or treatment modification (neuropathy) are noted.
`P values represent chi-square tests of significance.
`
`patients. However, this study used