|Home | About | Journals | Submit | Contact Us | Français|
S.A. responsible for conduct of study, data collection, data analysis, and manuscript writing. S.J.H. responsible for study design, data analysis, and manuscript writing. R.L. responsible for data collection. R.M.W. and P.W.L. conducted statistical analysis and edited manuscript. R.S.N. and K.G.B. involved in critical editing of manuscript. L.J.J, G.G.L., R.Lowsky, D.B.M., J.A.S., W-K.W. involved in final approval of manuscript.
Autologous hematopoietic cell transplantation with augmented BCNU-regimens is effective treatment for recurrent or refractory Hodgkin lymphoma (HL), however BCNU-related toxicity and disease recurrence remain challenges. We designed a conditioning regimen with gemcitabine in combination with vinorelbine in an effort to reduce the BCNU dose and toxicity without compromising efficacy. In this phase I/II dose escalation study, the gemcitabine maximum tolerated dose (MTD) was determined at 1250 mg/m2, and a total of 92 patients were treated at this dose to establish safety and efficacy. The primary endpoint was the incidence of BCNU-related toxicity. Secondary endpoints included 2-year freedom from progression (FFP), event-free survival (EFS), and overall survival (OS). Sixty-eight patients (74%) had one or more previously defined adverse risk factors for transplant (stage IV at relapse, B symptoms at relapse, greater than minimal disease pre-transplant). The incidence of BCNU-related toxicity was 15% (95% confidence interval, 9% to 24%). Only 2% of patients had a documented reduction in diffusing capacity of 20% or greater. With a median follow-up of 29 months, the FFP at 2 years was 71% and the OS at 2 years was 83%. Two-year FFP was 96%, 72%, 67%, and 14% for patients with 0 (n=24), 1 (n=37), 2 (n=23), or 3 (n=8) risk factors, respectively. Regression analysis identified PET status pre-transplant and B symptoms at relapse as significant prognostic factors for FFP. This new transplant regimen for HL resulted in decreased BCNU toxicity with encouraging FFP and OS. A prospective, risk-modeled comparison of this new combination with other conditioning regimens is warranted.
High dose chemotherapy and autologous hematopoietic cell transplantation (AHCT) is an effective treatment for patients with recurrent Hodgkin lymphoma (HL). Randomized controlled trials have shown improved freedom from progression (FFP) with high dose BEAM (carmustine 300 mg/m2, etoposide, cytarabine, melphalan) and AHCT over conventional salvage chemotherapy in chemosensitive patients [1-2]. The German Hodgkin Lymphoma Study Group/European Bone Marrow Transplant Registry (GHSG/EBMT) randomized trial of BEAM-AHCT vs Dexa-BEAM showed a 3-year FFP of 55% vs 34%, respectively . Attempts to improve on FFP in AHCT have included further intensification of salvage therapy before transplant [2-7], or intensification of the transplant conditioning regimen itself either with high dose sequential therapy [8-9] or with augmented carmustine (BCNU)-based regimens [10-16]. With increasing doses of BCNU from 300 mg/m2 to 600 mg/m2, however, the incidence of pulmonary toxicity increases to 35% and higher [17-20]. When oral lomustine (CCNU) was substituted for BCNU in the conditioning, the interstitial pneumonitis incidence was as high as 63% . Further, BCNU toxicity is likely under-reported because symptoms of fever, fatigue, nausea, poor appetite and weight loss are often not attributed . Although it typically responds rapidly to corticosteroid therapy, the dose-related BCNU syndrome can be potentially life-threatening [22-24]. The high dose BCNU regimen most commonly reported in the literature is CBV (cyclophosphamide, carmustine 300 mg/m2, and etoposide) . As previously reported by the Stanford group , our variation of the CBV regimen, which utilized high-dose BCNU at a maximum of 550 mg/m2, was associated with early (within 100 days post-transplant) and late (approximately 6 months post-transplant) treatment-related deaths, primarily respiratory (4% early respiratory deaths and 7% late respiratory deaths).
Gemcitabine and vinorelbine are active drugs in patients with HL with mechanisms of action distinct from alkylating agents [26-32]. We hypothesized that these drugs would allow for reduction of the BCNU dose in conditioning, thereby reducing early and late adverse effects of this agent. The combination of gemcitabine and vinorelbine on days 1 and 8 was taken from solid tumor experience [33-38]. In this phase I/II study our goals were to reduce the BCNU dose from 550 mg/m2 to 350 mg/m2 in an effort to reduce the risk of pulmonary toxicity while simultaneously adding gemcitabine and vinorelbine in an effort to maintain or improve efficacy. We report the phase I/II experience of utilizing this five-drug regimen for the treatment of 92 patients with relapsed or refractory HL.
Eligibility criteria included: histologically proven, recurrent or refractory HL confirmed at Stanford University; age ≤ 70 years; ECOG performance status 0-2. Adverse risk factors have been previously defined  as: 1) stage IV disease at relapse, 2) constitutional “B” symptoms at relapse, and 3) failure to achieve minimal disease (single lymph nodes ≤ 2cm or > 75% reduction in a bulky tumor mass or bone marrow involvement ≤ 10%) at transplant. High risk HL patients were defined as those who had one or more of the above risk factors. All patients signed informed consent for the study approved by the Institutional Review Board at Stanford University School of Medicine. Pre-transplant testing included routine staging with medical history, physical examination, computed tomography (CT) with or without positron emission tomography (PET), bone marrow aspiration and biopsy with cytogenetics, baseline assessment of cardiac ejection fraction and pulmonary function tests (PFTs). Patients with ejection fraction <40% and diffusion capacity corrected for hemoglobin < 55% were not eligible for the study.
The first seven patients all had high risk features and were enrolled in the phase I dose escalation study to determine the maximum tolerated dose (MTD) of gemcitabine. In phase II, all risk category patients were accepted.
In phase I, gemcitabine was dose-escalated at planned doses of 1250 mg/m2, 1500 mg/m2, and 1800 mg/m2, in combination with vinorelbine 30 mg/m2 on days -13 and -8, followed by a lowered dose of BCNU (10 mg/kg and capped at 350 mg/m2, as compared to our standard dose of 15 mg/kg, capped at 550 mg/m2) on day -6, etoposide 60 mg/kg on day -4, and cyclophosphamide 100 mg/kg on day -2. Cohorts of three high risk patients proceeded at each gemcitabine dose level. Any grade III or IV non-hematologic toxicity constituted an adverse event.
The phase II portion of the study enrolled additional patients at the gemcitabine MTD.
Patients’ peripheral blood hematopoietic cells (PBHC) were mobilized from their salvage chemotherapy, or from cyclophosphamide 4 grams/m2 with G-CSF (10 μg/kg per day), or from G-CSF alone per the treating physician's discretion. No specific recommendation was given for salvage chemotherapy prior to AHCT, however 40 patients received DHAP (dexamethasone, cytarabine, cisplatin) , 35 patients received ICE (ifosfamide, carboplatin, etoposide) , and 17 patients received other chemotherapy combinations, including 2 with a gemcitabine combination. Leukaphereses were performed until ≥2 × 106 CD34 cells/kg were collected. The apheresis product was cryopreserved per institutional practice and infused on transplant day 0.
Patients received prophylactic antimicrobial treatment including ciprofloxacin, fluconazole, acyclovir and trimethoprim-sulfamethoxazole. Low dose intravenous heparin (100 u/kg/d) as a continuous infusion was used for prevention of sinusoidal obstruction syndrome (SOS), formerly referred to as hepatic veno-occlusive disease . Patients received G-CSF beginning on day +6 after AHCT.
Clinically significant BCNU-related toxicity was defined as a non-infectious syndrome of one or more of the following: low-grade fever, dyspnea, fatigue, nausea, poor appetite, weight loss, dry cough, pulmonary infiltrates, or a decrease in diffusing capacity of up to 20% from pretransplantation levels that, in the judgment of the treating physician, required corticosteroid therapy in the first 100 days post-transplant. The BCNU toxicity syndrome was assessed in all patients and categorized as either predominantly pulmonary or gastrointestinal (nausea, poor appetite and weight loss) , based upon symptoms. Pulmonary function testing was repeated whenever possible for symptomatic patients suspected of having the BCNU toxicity syndrome.
The disease status of all patients was evaluated prior to transplantation utilizing CT or PET-CT imaging and bone marrow exam. Post-transplant restaging with CT or PET-CT imaging was performed routinely at 3 months, 6 months, 1 year, 18 months, 2 years and annually thereafter until year 5 after transplantation. A bone marrow biopsy was performed once at 3 months post-transplant and thereafter as clinically indicated. Response to therapy was assessed according to the revised response criteria for malignant lymphomas . PET response followed the guidelines put forth by Hutchings  and Gallamini et al , which categorized patients as positive, negative or minimal residual uptake (MRU, defined as standardized uptake value of 2.0 to 3.5). Patients with PET results showing MRU were considered PET negative for the purposes of analysis.
Radiation therapy was administered to selected patients following transplantation based on previously published criteria . Post-transplant radiation therapy was typically administered 2-3 months after transplantation.
The goal of the phase I portion of the study was to define the regimen-related toxicity and to determine the MTD for gemcitabine. The primary endpoint for the phase II portion of the study was the incidence of BCNU-related toxicity. Secondary endpoints included 2-year freedom from progression (FFP), event-free survival (EFS), and overall survival (OS) at the gemcitabine MTD. Probabilities of FFP, EFS, and OS over time were estimated with the product-limit method of Kaplan and Meier . Progression of HL was the only event defined in FFP, with censoring at time of non-relapse death. Disease progression and death from any cause defined events in the calculation of EFS. All calculations were made from the date of transplantation.
The Fisher's exact test or Pearson chi-square test was used to examine correlations across prognostic factors. Cox regression was used in both univariate and multivariate analyses for each of the outcome variables (FFP, EFS, OS). The Wald test was used in testing hypotheses on covariates. Results from correlation and univariate analyses guided the selection of variables in the multivariate stepwise regression analyses. The log rank test was used to compare survival curves.
From September 2001 to March 2008, 114 patients were prospectively screened for enrollment on the trial at Stanford University Medical Center. Eighteen patients were deemed ineligible due to diffusion capacity for carbon monoxide (DLCO) corrected for hemoglobin <55% (2 patients), history of recent pulmonary embolus or radiation pneumonitis (2 patients), active uncontrolled infection (2 patients), known allergy to etoposide (1 patient), a history of grade 3 hemorrhagic cystitis with cyclophosphamide (1 patient), ≥ grade 2 sensory or motor peripheral neuropathy from prior vinca alkaloid (2 patients), prior malignancy (3 patients), and inadequate peripheral blood hematopoietic cell collection (5 patients).
In phase I, three patients received gemcitabine at 1250 mg/m2 without dose-limiting toxicity. Four patients proceeded to gemcitabine 1500 mg/m2, where the dose-limiting toxicity was reached in 3 of the patients based on grade 3-4 elevated liver transaminases and a symptom complex of fever, headache, and skin toxicity. These 4 patients treated at gemcitabine 1500 mg/m2 were excluded from phase II analysis. A total of 92 patients were treated at the gemcitabine MTD of 1250 mg/m2, including the first three patients in phase I and an additional 89 patients in phase II, to determine the safety and efficacy for the phase II analysis.
The pre-transplant characteristics for the 92 patients are shown in Table 1. Thirty-six patients (39%) were characterized as primary induction failures. Primary induction failure was defined as progression of disease during induction treatment, or initiation of second-line treatment, or response of < 60 days duration . Fifty-six patients (61%) had previously achieved a CR (duration greater than 1 year in 22 patients and ≤ 1 year in 34 patients). Before transplantation, most patients received cytoreductive chemotherapy with the goal of achieving a minimal disease state. Typical cytoreduction consisted of two to three cycles of combination chemotherapy.
Status at transplant referred to the response to cytoreductive chemotherapy just prior to transplantation. Eighty-five patients (92%) responded to treatment before transplant conditioning. Complete remission or a minimal disease state was achieved in 53 patients (58%). Only seven patients had stable or progressive disease. With regard to the previously defined 3 adverse risk factors for prognosis , 24 patients (26%) had no risk factors, and 68 (74%) had one or more risk factors.
PET status at the time of transplantation was recorded in 77 patients. Thirty-eight percent (29/77) of patients were PET positive at time of transplant, 55% (42/77) were PET negative, and 8% (6/77) had MRU.
The median time to absolute neutrophil count (ANC) >500/μl was 10 days (range 8-17). The median time to untransfused platelet count of 20,000/μl was 15 days (range 8-34).
Twenty-two patients received post-transplantation involved field radiation. The post-transplant radiation dose ranged from 1200-3600 cGy, depending on previous radiation treatment and normal tissue tolerance. Fourteen patients received post-transplant regional irradiation (modified mantle or inverted Y) to a total dose of 3600 cGy, while the remaining 8 patients had variable doses to a single radiation field.
Three patients (3%) suffered early transplant-related mortality. The causes of early death were fungal infection in two patients (day +12 and day +15) and severe SOS in one patient (day +34). Fever and mucositis were the most common grade 3 toxicities, followed by transaminase elevations, skin rash, headache and culture positive infections. Grade 4 toxicities occurred in only 2 patients and in both cases were related to elevated transaminases. Table 2 lists grade 3-4 toxicities.
The incidence of BCNU-related toxicity in this study was 15% (14 of 92 patients, 95% confidence interval (CI) 9% to 24%). Table 3 shows the BCNU toxicity characteristics. Only two patients had a documented decrease in DLCO of ≥20%, although 8 others had one or more respiratory symptoms, of whom 3 did not have formal pulmonary function testing. All patients recovered from the BCNU-toxicity symptoms with corticosteroid therapy, which was administered at an initial dose of 1mg/kg/day with a taper of 10 mg per week as tolerated. BCNU-related toxicity was not more prevalent in patients with a history of chest irradiation. There was one late pulmonary death (day +532) from adult respiratory distress syndrome (ARDS).
The median follow-up of the entire group of 92 patients is 29 months (range 8-86 months). Figure 1 illustrates the Kaplan-Meier survival curves for the entire population. Freedom from progression (FFP) at 2 years was 71% (CI 61% to 81%) and EFS at 2 years was 67% (CI 57% to 77%). The actuarial 2-year OS was 83% (CI 75% to 91%). The 2-year non-relapse mortality for the entire study was 6%. Causes of non-relapse mortality included candidemia, pulmonary fusariosis, severe SOS, pulmonary embolus and ARDS. There were 16 deaths, 11 from relapse and 5 from non-relapse mortality. Twelve patients with relapse are still alive. Of the seven patients with advanced disease status at transplant (one with stable disease and six with progressive disease-Table 1), three patients relapsed within 4 months of transplant and subsequently died; one patient died from complications of SOS at 1 month post-transplant and was too early for disease follow-up, and three patients are alive and in CR at last follow-up.
Figure 2a illustrates the Kaplan-Meier plot of FFP according to the number of risk factors present. The 2-year FFP was 96%, 72%, 67%, and 14% for 0, 1, 2, and 3 risk factors, respectively, P < 0.0001.
Because response to induction therapy had been reported to be significantly correlated with the success of autologous transplantation in recurrent and refractory HL by Stanford and other groups [9,12,45], we examined the 2-year FFP according to duration of initial chemotherapy remission (duration greater or less than 1 year or induction failure). Primary induction failure (IF) was associated with a significantly worse 2-year FFP of 48% (CI 30% to 66%) as compared to initial remission duration ≤ 1 year (90%, CI 79% to 100%; P < 0.0001), or > 1 year (80%, CI 62% to 98%, P= 0.05), however there was no significant difference in FFP outcomes between remission duration ≤ 1 year or > 1 year (P=0.25). Figure 2b shows FFP significantly inferior for IF patients versus non-IF patients (P=0.0003). The presence of B symptoms at relapse also significantly reduced two-year FFP (56%, CI 38% to 74%) versus without B symptoms (83%, CI 72% to 94%), P=0.008 (Figure 2c). Pre-transplant PET status was available in 77 of the 92 patients and allowed for further analysis. As shown in Figure 3a, FFP at two years was significantly lower for PET-positive (47%, CI 24% to 70%) versus MRU or PET-negative (87%, CI 76% to 98%) patients, P< 0.0001. A summary of survival statistics is provided in Table 4.
We next explored the significant factors of PET status, induction response, induction regimen and constitutional symptoms in multivariate analysis of 77 study participants. Pre-transplant PET-positive status and constitutional symptoms were significantly predictive of FFP (hazard ratios of 5.9 and 4.8) and EFS (hazard ratios of 3.3 and 4.3). ABVD primary therapy was of borderline significance (hazard ratio of 2.8). Table 5 summarizes these data.
As shown in Figure 3b, patients with both risk factors (PET-positive and symptomatic) had a 14% (CI 0% to 38%) two-year FFP compared to 86% (CI 67% to 100%), 82% (CI 63% to 100%) and 91% (CI 79% to 100%) two-year FFP in those who were PET-positive and asymptomatic, PET-negative and symptomatic or PET-negative and asymptomatic, respectively.
Although high dose chemotherapy and AHCT is the most effective treatment for patients with recurrent or refractory HL, disease recurrence in nearly half of patients requires new approaches [1-2,9,46-47]. We sought to reduce the toxicity of the conditioning regimen while maintaining efficacy to create an optimal platform for post-transplant therapeutics. In our previous experience, the pulmonary toxicity associated with augmented BCNU regimens (550 mg/m2) prohibited the introduction of post-transplant therapy in the first 100 days [12,19].
The present study was designed to evaluate the feasibility of a transplant regimen in which the combination of gemcitabine and vinorelbine allowed for a lowered dose of BCNU along with standard doses of etoposide and cyclophosphamide (GN-BVC). It was recognized that pretransplant GN-containing salvage regimens were being developed during this trial period with similar GN-dosing . Our toxicities and responses as a transplant regimen, compared to a pretransplant salvage approach with GN, however, might be expected to be different due to the administration close to BVC and thus, was a new experience.
The primary endpoint of the study was met in terms of demonstrating a reduced incidence of BCNU-related toxicity of 15% (CI 9% to 24%). A nominal rate of 35% was used as the historical control rate in the study design, which was a best estimate based on our own data and those of other studies using augmented BCNU dosing [10-13], and served to establish the study size for high probability to detect a toxicity reduction. It should be noted that a more stringent pulmonary definition of BCNU-related toxicity (DLCO decrease ≥20%) was applied in prior Stanford studies . In fact, only two patients had a documented DLCO decrease of at least 20% in the current study, suggesting an even more favorable result. Reducing the dose of BCNU from 15 mg/kg to 10 mg/kg in this regimen likely contributed to the reduced toxicity. Better supportive care in the current era of autografting [23,48-49], as well as earlier recognition and initiation of corticosteroid treatment for BCNU pneumonitis, has undoubtedly played a role in reducing BCNU-related toxicity and deaths. Our institution does not use prophylactic corticosteroids in the high dose BCNU regimens, however, this practice has allowed certain groups to decrease their incidence of interstitial pneumonitis while maintaining BCNU doses of 600 mg/m2 [23,49]. A specific pretransplant pulmonary function threshold that predicts for post-transplant pulmonary complications and mortality is not yet established , and this remains the case for BCNU toxicity. A corrected DLCO of 55% was used for entry onto the current study, which is the lowest threshold for our institution's AHCT regimens, and thus allowed more compromised patients to move forward to AHCT with less toxicity result.
The anti-tumor efficacy of the conditioning regimen did not appear to be compromised in our cohort. With a median follow-up of 29 months, the 2-year FFP was 96%, 72%, 67%, and 14% for 0, 1, 2, and 3 risk factors, respectively. As a comparison, in our previous report , the 3-year FFP were 85%, 57%, 41%, and <20% for 0, 1, 2, and 3 risk factors, respectively. Patients with 1 or 2 risk factors appear to benefit with this new regimen over our historical experience. The distribution of risk factors was similar in the earlier and current studies . When looking at more universal risk factor groups, such as duration of first remission, our study results showed 2-year FFP of 48% for primary IF, 90% for remission duration ≤ 1 year, and 80% for remission duration > 1 year. As a comparison, the conditioning approach utilizing high dose sequential therapy followed by BEAM-AHCT, reported in the phase 2 Cologne trial by the GHSG , shows a freedom from second failure for risk groups having progressive disease versus early relapse (≤ 1 year) versus late relapse (> 1 year) of 41%, 62%, 65%, respectively, at a median follow-up of 30 months. The ongoing GHSG/EBMT randomized trial  is investigating the efficacy of high dose sequential therapy and BEAM-AHCT with standard DHAP followed by BEAM-AHCT, and results are awaited, however, this is limited to chemosensitive patients and questions remain on outcomes for chemorefractory patients prior to AHCT.
Subsequently, FDG-PET imaging has become incorporated in HL response assessment [40-42]. In the current study, pre-transplant PET status significantly correlated with FFP (P< 0.0001) and EFS (P= 0.003), and was of borderline significance with regard to OS (P= 0.07). Constitutional symptoms at relapse (P= 0.008) and induction failure (P= 0.0003) were also identified as significant adverse risk features. In multivariate analysis, PET-positive status and constitutional symptoms were independent risk factors for treatment failure. Further, we observed that the combination of constitutional symptoms and PET-positive status, representing 19% of patients with pre-transplant scans, conferred a very poor prognosis, with just 14% FFP at 2 years. In contrast, FFP at 2 years was >80% for patients with only 1 or neither risk factor. This observation should be viewed as hypothesis-generating and interpreted with caution given the small sample size.
In general, the literature is concordant regarding adverse risk factors prior to transplant for HL. B symptoms at relapse, extranodal disease, chemoresponsiveness before transplantation, and remission duration < 1 year [5, 46, 51-53] have been reported as independent predictors of long-term disease-free survival in primary refractory and relapsed HL patients. More recently, there are several reports indicating the prognostic significance of PET pre-transplant. The retrospective study by Jabbour et al  identified pretransplant PET status to be independently predictive for progression-free survival (PFS) and OS. In that study, 3-year PFS rates for PET positive and negative patients was 23% and 69%, respectively. Castagna et al  further showed that early PET after two cycles was predictive of survival after AHCT with 2-year PFS of 10% for PET positive vs 93% for PET negative patients (P<0.001). Of note, patients who achieve normalization of PET prior to AHCT with additional non-cross resistant second-line therapy had improved EFS in single institution study , but it is unclear if this strategy results in selection bias, meaningful tumor reduction for cure, or a combination of both.
In conclusion, the regimen of GN-BVC is associated with less pulmonary toxicity than the prior augmented BCNU-containing regimen with equal or greater efficacy. A prospective comparison of this new combination with other conditioning regimens using risk stratification is warranted. Whether using the previously established risk model or incorporating PET, about two-thirds of patients are alive and event-free at 2 years after transplant. A small group with multiple risk factors  in our former model or both symptomatic and PET-positive, had a dismal outcome. Targeting these higher risk patients for novel therapies should be the objective for future studies. Tandem AHCT  has been explored in a risk-adapted transplant strategy but requires further study. Other approaches could include antibody-drug conjugates, histone deacetylase inhibitors, immunomodulating agents and cell-based therapies which hold promise as peri-transplant therapeutics [58-61].
This work was supported in part by P01 CA49605; presented previously at ASCO 2005, abstract 6651 and ASH 2008, abstract 2194
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.
Conflict-of-interest disclosure: The authors declare no competing financial interests.