Gemcitabine, dexamethasone, and cisplatin in patients with recurrent or refractory aggressive histology B-cell non-Hodgkin lymphoma

A Phase II study by the National Cancer Institute of Canada Clinical Trials Group (NCIC-CTG)




Gemcitabine has been shown to have activity as a single agent in lymphoma and, when combined with cisplatin, is effective therapy for a number of solid tumors. The authors wished to determine the response rate and toxicity of gemcitabine, dexamethasone, and cisplatin for recurrent or refractory non-Hodgkin lymphoma (NHL).


Patients with recurrent or refractory diffuse large B-cell NHL or variants (REAL classification), measurable disease, and one previous chemotherapy regimen were eligible. Treatment consisted of gemcitabine 1000 mg/m2 intravenously (i.v.) on Days 1 and 8, dexamethasone 40 mg orally on Days 1–4, and cisplatin 75 mg/m2 i.v. on Day 1 (GDP), every 21 days as an outpatient. The primary end point was a response after two cycles. Patients could then proceed to stem cell transplantation (SCT) or receive up to six treatment cycles.


Fifty-one eligible patients were evaluable for toxicity and response. The median age of the patients was 57 years (range, 18–84 years) and most had diffuse large-cell lymphoma. After 2 cycles, there were 8 complete responses (CR; 16%) and 17 partial responses (PR; 33%). There was an overall response rate (RR) of 49% (95% confidence interval = 37–63%). The RR afer completion of all protocol chemotherapy (including those who received > 2 cycles of GDP) was 53% (11 CR, 16 PR). Grade 3 and 4 neutropenia occurred in 33% and 39% of patients, respectively. Grade 3 and 4 thrombocytopenia occurred in 24% and 4% of patients, respectively. Seven patients (14%) experienced febrile neutropenia. Of the 35 patients < 66 years, 22 (63%) proceeded to SCT.


GDP is an active regimen in B-cell NHL and can be administered with acceptable toxicity to outpatients. A Phase III trial comparing GDP with standard cisplatin-based chemotherapy is now ongoing through the National Cancer Institute of Canada Clinical Trials Group. Cancer 2004. © 2004 American Cancer Society.

For patients with recurrent, aggressive non-Hodgkin lymphoma, second-line or salvage chemotherapy followed by high-dose chemotherapy and autologous stem cell transplantation (ASCT) improves progression-free and overall survival compared with standard-dose chemotherapy alone, and is the treatment of choice for younger patients with chemotherapy-sensitive disease.1, 2 Several multiagent chemotherapy regimens have been used to reduce the burden of lymphoma before transplantation.3–8 Many of these regimens include cisplatin or carboplatin together with high-dose dexamethasone, and are often associated with substantial hematologic and nonhematologic toxicity. Treatment with these regimens produces responses in 39–69% of patients,4–8 but often requires hospitalization either to administer the therapy or to manage resulting toxicities. Because of cumulative bone marrow toxicity, the ability to collect sufficient numbers of hematopoietic stem cells to support subsequent high-dose chemotherapy may be impaired.9, 10 Ultimately, only 20–30% of patients with recurrent or refractory aggressive lymphoma initially considered for SCT are alive and disease free 3–4 years later.3, 5 For elderly patients, and for those for whom initial salvage treatment is not effective, chemotherapy is given to palliate symptoms. Less toxic and more effective regimens are needed for these populations.

Gemcitabine (difluorodeoxycytidine) is a pyrimidine antimetabolite that has broad activity in solid tumors such as pancreatic, lung, bladder, and breast carcinoma. Responses to single-agent gemcitabine were reported in 20% of patients with heavily pretreated diffuse large B-cell lymphoma.11 Significant activity has also been reported in patients with advanced-stage mycosis fungoides and peripheral T-cell lymphoma involving skin.12 In vitro studies have demonstrated synergy between cisplatin and gemcitabine against cell lines and primary cultures from patients with leukemia and lymphoma.13, 14 Randomized trials comparing the combination of gemcitabine and cisplatin with other regimens in patients with lung and bladder carcinoma have demonstrated comparable or superior outcomes, including less treatment-related toxicity.15–17

We chose to evaluate the combination of gemcitabine and cisplatin, with the addition of dexamethasone (GDP), to determine whether this regimen could be effective and less toxic than currently used salvage regimens, with a view to proceeding to Phase III testing if these results showed sufficient promise.


Eligibility and Evaluation of Patients

Patients were eligible for the current Phase II trial if they had diffuse large B-cell, large-cell immunoblastic, anaplastic large-cell (B-cell or null-cell type), primary mediastinal large B-cell, or T-cell–rich B-cell lymphoma18 and had disease recurrence after, or were refractory to, one previous anthracycline-containing chemotherapy regimen. A reference pathologist at each center confirmed the histologic diagnosis of lymphoma by reviewing initial biopsy material and/or biopsy specimens obtained at the time of disease recurrence or progression. Patients with a diagnosis of transformed or composite lymphoma were excluded.

At least one site of disease had to be measurable in two dimensions using clinical examination, computed tomography (CT) scans, or magnetic resonance imaging scans. All patients had to be > 16 years, have an Eastern Cooperative Oncology Group performance status (PS) of 0–2,19 normal bone marrow function (absolute neutrophil count ≥ 1.5 × 109/L, platelet count ≥ 100 × 109/L), serum creatinine level < 140 mmol/L, aspartate or alanine aminotransaminase (AST or ALT) level < 2.5 × the upper limit of normal (ULN), and bilirubin level < 1.5 × ULN. Patients were excluded if they met the following criteria: central nervous system involvement by lymphoma; a known positive antibody test for the human immunodeficiency virus; previous treatment with gemcitabine, cisplatin, or high-dose chemotherapy and SCT; an active uncontrolled infection; or significant cardiac dysfunction. The human experimentation committee or research ethics board of each participating center approved the current study, and written informed consent was obtained from all patients.

Baseline Assessments

Before study entry, all patients underwent a complete history and physical examination, including evaluation of PS. Laboratory studies included a complete blood count (CBC), biochemistry profile (including monitoring of levels of creatinine, AST or ALT, bilirubin, albumin, and lactate dehydrogenase), and bone marrow aspiration and biopsy. Radiologic examinations, including a chest radiograph and CT scans of the chest, abdomen, and pelvis, were required within 3 weeks of starting treatment. Gallium scanning was recommended but not required.


Chemotherapy consisted of gemcitabine 1000 mg/m2 intravenously (i.v.) over 30 minutes on Days 1 and 8, dexamethasone 40 mg orally in divided doses on Days 1–4, and cisplatin 75 mg/m2 infused i.v. over 60 minutes on Day 1. Treatment cycles were repeated every 21 days. One of the dexamethasone doses was administered i.v. before chemotherapy to prevent emesis, and the oral protocol treatment dose was reduced accordingly. To reduce the risk of cisplatin-induced nephrotoxicity, a specific hydration protocol was recommended consisting of prehydration with 250 mL of 5% dextrose in saline given i.v. over 30 minutes, and 250 mL of 20% mannitol in normal saline was infused i.v. over 1 hour concurrently with the administration of cisplatin. Cisplatin was mixed in 500 mL of normal saline and administered over 1 hour followed by an additional 500 mL of normal saline. Patients were requested to drink six to eight glasses of fluid over the rest of the day after completion of chemotherapy. An antiemetic protocol appropriate for highly emetogenic chemotherapy was suggested (e.g., ondansetron or dolasetron plus dexamethasone). While on study, patients were not permitted to receive other cytotoxic, monoclonal antibody, or radiotherapy.

Assessment of Outcomes

After completing two cycles of therapy, all patients underwent disease reassessment by physical examination and CT scans. Other CT scans or X-rays were performed as indicated. Bone marrow biopsy was repeated after Cycle 2 if it was involved by lymphoma at baseline.

The primary end point of the current study was the response rate (RR) observed after two cycles of GDP. Responses were assigned according to international workshop criteria.20 Patients achieving a complete response (CR), unconfirmed CR (CRU), or partial response (PR) were considered to have satisfied the study end point of response. After two cycles of GDP, patients achieving one of the above response criteria, and those who were otherwise considered by their attending physician to be candidates for high-dose chemotherapy and SCT, were removed from the study and proceeded to stem cell mobilization. Patients with stable disease (SD) received additional cycles of GDP, or were removed from the study and given alternative salvage therapy at the discretion of the investigator. Patients with SD who were not candidates for SCT or patients who had any response after two cycles of GDP continued on the study for up to six cycles of treatment.

During treatment, all patients had a CBC including a leukocyte differential count and underwent monitoring of levels of serum creatinine, AST or ALT, alkaline phosphatase, and serum bilirubin on Day 1 of each treatment cycle. The CBC and differential were repeated on Day 8. Treatment cycles were delayed by 1 week for granulocytopenia < 1.0 × 109/L or thrombocytopenia < 100 × 109/L. If the Day 8 granulocyte count was < 0.5 × 109/L or the platelet count was < 50 × 109/L, the physician omitted the Day 8 gemcitabine dose, or delayed treatment and administered this dose 1 week later, provided the granulocyte count was then > 0.5 × 109/L and the platelet count was > 50 × 109/L. If isolated granulocytopenia of 0.5–0.9 × 109/L was observed on Day 8, the physician had the choice of administering the full dose of gemcitabine and initiating therapy with filgrastim, or administering gemcitabine with a dose reduction of 25%. If isolated thrombocytopenia of 50–99 × 109/L was observed on Day 8, the dose of gemcitabine was reduced by 25%. Primary prophylaxis with colony stimulating factors was not permitted. After one dose delay or attenuation for neutropenia, patients were to receive filgrastim with subsequent treatment cycles to maintain dose intensity. For patients who experienced an episode of febrile neutropenia, filgrastim was recommended in subsequent treatment cycles.

Toxicities were graded using the National Cancer Institute Common Toxicity Criteria (Version 2.0). Doses of cisplatin and gemcitabine were reduced by 25% if any Grade 3 nonhematologic toxicities (except nausea, emesis, and alopecia) were observed. Doses of cisplatin were reduced by 25% if the serum creatinine level was 140–199 mmol/L. Protocol therapy was discontinued with the occurrence of any Grade 4 nonhematologic toxicities or for elevations of serum creatinine levels ≥ 200 mmol/L.

Statistical Methods

The primary end point of the current multicenter trial was the objective RR after two cycles of therapy. It was assumed that this regimen would not be of further interest, and additional Phase III testing not warranted, if the observed RR was < 40%. To minimize the average sample size and ensure that we would continue to a Phase III trial with high probability if the true RR is > 50%, and with lower chance to a Phase III trial if the true RR is < 30%, a 2-stage design was used for patient accrual.21 In the first stage, 22 patients were to be entered. If six or fewer responses were observed, the study would be stopped and it would be concluded that the regimen was inactive. If ≥ 13 responses were observed, it would be concluded that the GDP regimen was active. If 7–12 responses were observed, the second stage of the protocol would be completed and another 22 patients would be enrolled. If ≥ 19 responses were observed among 44 patients (i.e., a RR of 43%), it would be concluded that GDP is an active regimen in this patient population, with a power of 85% and a Type 1 error of 4.4%.



Between February 2001 and May 2002, 53 patients with recurrent or refractory aggressive B-cell lymphoma were enrolled from 9 centers in Canada. Two patients were subsequently found to be ineligible after local pathology review. A total of 51 patients were evaluable for toxicity. Two patients died during Cycle 1 and could not be evaluated for response, but are included in the denominator used to calculate the RR.

The pretreatment characteristics of the 51 eligible patients are shown in Table 1. The majority of patients had diffuse large B-cell lymphoma. All but three patients had been treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) at the time of initial diagnosis. Two patients had received cyclophosphamide, vincristine, and prednisone with etoposide instead of doxorubicin, and one patient was treated with doxorubicin (Adriamycin, Adria Laboratories, Columbus, OH), bleomycin, vinblastine, and dacarbazine (ABVD). The patient treated with ABVD was initially diagnosed with Hodgkin disease, but at the time of disease progression was found to have T-cell–rich, B-cell lymphoma on pathology review. Five patients had received rituximab in combination with CHOP and 18 patients had previously received radiotherapy to involved sites of disease. Twenty-two patients achieved a CR with initial chemotherapy at diagnosis, and 14 had disease progression during or within 3 months of initial treatment. Using the international prognostic index22 (IPI) to evaluate the prognostic risk of patients at the time of study entry, 21 patients (41%) had 0 or 1 risk factor and 30 patients (59%) had ≥ 2 risk factors.

Table 1. Patient Characteristics
  • CR: complete response; PR: partial response; CRU: complete response unconfirmed; IPI: international prognostic index; LDH: lactate dehydrogenase.

  • a

    Ann Arbor staging system.

No. of eligible patients51
Median age (yrs) (range)57 (18–84)
Stage at study entrya 
B symptoms11
 Diffuse large B cells40
 Immunoblastic cells3
 Anaplastic large cells3
 T-cell rich B cells3
 Mediastinal large cells2
Response to previous therapy: 
 No response4
Time to disease recurrence or progression (from end of initial therapy) (mos) 
 < 318
 > 1214
Previous radiation18
LDH at disease recurrence 
IPI risk factors at disease recurrence 

Protocol Treatment Received

A total of 139 cycles of GDP were administered during the current study. Three patients did not complete the planned two cycles of therapy—two because of fatal toxicity and one because of disease progression after the first cycle. Among the remaining 48 patients, 31 received 2 cycles, 4 received 3 cycles, 8 received 4 cycles, and 5 completed all 6 planned cycles of treatment.

Dose reductions occurred in 20 of 139 cycles (14%) and involved 14 patients. Dose delays occurred in 60 of 139 cycles (43%) and involved 35 patients. The majority of dose reductions and delays (60%) were due to hematologic toxicity. Filgrastim was given to 31 of 51 patients (61%) during ≥ 1 treatment cycle.

Response to Treatment

The primary end point of the current trial was the RR observed after completing two cycles of GDP. Of the 51 eligible patients, there were 8 CR (16%) and 17 PR (33%), for an overall RR of 49% (95% confidence interval [CI] = 37–63%). One patient with SD went on to achieve a PR, and three patients with a PR achieved a CR, after receiving additional cycles of GDP. The RR observed after completing all protocol therapy was 53% (95% CI = 40–67%), and included 11 CR and 16 PR. Among the 17 patients > 65 years, who would not usually be considered for high-dose chemotherapy and ASCT, the median time to progression from the start of GDP was 3.1 months (95% CI = 2.3–9.2) and the overall survival was 8.9 months (95% CI = 5.2–18.5).

Treatment-Related Toxicity

Hematologic toxicities related to treatment are shown in Table 2. The most common Grade 3 or 4 toxicity was granulocytopenia, which was observed in 32 patients (63%). However, only 7 patients (14%) experienced an episode of febrile neutropenia, for a frequency of 5.0% of all cycles (7 of 138). Grade 3 or 4 thrombocytopenia was observed in 14 patients (27%), and 9 patients (18%) required ≥ 1 platelet transfusion. Twelve patients (24%) required ≥ 1 red blood cell transfusion. Overall, 74% of patients did not require any blood product support during treatment with GDP.

Table 2. Hematologic Toxicity (Worst Ever by Patient; n = 51)
  • GI: gastrointestinal.

  • a

    National Cancer Institute Common Toxicity Criteria, Version 2.0.

Febrile neutropenia   7 1
Infection 23311
Melena/GI bleeding 2 1  
Epistaxis 5 1  
Petechiae/purpura 13   

Nonhematologic treatment-related toxicities are shown in Table 3. Nausea and/or emesis were the most common toxicities observed. Grade 2 or greater nausea and emesis occurred in 14 (27%) and 11 (22%) patients, respectively. Grade 1 or 2 ototoxicity occurred in 13 patients (25%) and was attributed to cisplatin. Other toxicities that have been observed with cisplatin-containing regimens were uncommon—≥ Grade 2 sensory neuropathy was seen in 5 patients (10%) and the most severe nephrotoxicity observed was Grade 2, which occurred in 3 patients (6%).

Table 3. Nonhematologic Toxicity (Worst Ever by Patient; n = 51)
  • AST: aspartate aminotransferase; ALT: alanine aminotransferase.

  • a

    National Cancer Institute Common Toxicity Criteria, Version 2.0.

  • b

    One fatal case.

Thrombosis/embolism  43b
Creatinine level73  
AST level4111
ALT level522 

Fatal toxicities were observed in 3 patients (6%). One patient died suddenly of a pulmonary embolism after completing four cycles of GDP. The two other treatment-related deaths occurred during the first cycle of therapy. A 76-year old male, who had all 5 IPI risk factors at the time of disease recurrence and had achieved only a PR to initial therapy with CHOP, died of complications of tumor lysis syndrome. A 69-year-old female, who had 4 IPI risk factors at disease recurrence, died of complications of neutropenic sepsis.

Thromboembolic events were observed in 6 patients in addition to the patient who suffered the fatal pulmonary embolism (a total of 7 patients [14%]). One of these patients was also found to have an asymptomatic abdominal aortic thrombosis on a CT scan during evaluation for a pulmonary embolus. Acute myeloid leukemia was observed in a 55-year-old woman 4 months after SCT, and 7 months after completing 2 cycles of GDP. At the time of the diagnosis of leukemia, bone marrow cytogenetics were normal. Rearrangement of the mixed-lineage leukemia (MLL) gene was not be detected in a sample of peripheral blood stem cells obtained before transplantation or in bone marrow cells obtained at the time of diagnosis of leukemia.

Of the 51 patients enrolled on study, 17 (33%) required hospital admission at some point during treatment with GDP. Hospitalization occurred during 26 of 139 cycles (19%) of protocol treatment, for a median of 9 days (range, 1–78 days): 11 patients required 1 admission, 4 patients required 2 admissions, and 2 patients required 3 and 4 admissions each. The reasons for admission were not always attributable to complications of treatment and included complications of disease, symptomatic disease progression, febrile neutropenia, emesis, and dehydration, as well as administrative problems preventing outpatient treatment. Only 9 cycles of protocol therapy (7%) were actually given in hospital.

Stem Cell Mobilization after Gemcitabine/Dexamethasone/Cisplatin Therapy

The patients in the current trial included those for whom SCT was planned if a response to GDP was observed, and those for whom a more prolonged course of therapy without transplantation was planned. Investigators were not asked to indicate at the time of patient registration whether SCT was intended. Allowing for this limitation in interpretation, there were 35 patients < 66 years for whom transplantation might have been considered a treatment option. Twenty-six of these patients (74%) successfully completed protocol therapy and proceeded with the intention of undergoing high-dose chemotherapy with stem cell support. One patient underwent syngeneic transplantation, and two experienced disease progression before stem cell collection could be completed. Autologous stem cell collection was attempted in the remaining 23 patients using a variety of mobilizing regimens, usually chemotherapy plus filgrastim, and was successful in 22 (96%). The median number of CD34-positive cells collected was 6.2 × 106/kg (range, 1.6–21.2 × 106), which was obtained with a median of 2 leukapheresis procedures (range, 1–4 procedures). Only one of these patients had a stem cell collection containing < 2.0 × 106 CD34-positive cells per kilogram. Twenty-one patients received high-dose chemotherapy and autologous stem cell infusion. The median time to neutrophil recovery > 0.5 × 109/L was 10 days (range, 8–24 days) and the median time to platelet recovery > 20 × 109/L was 13 days (range, 8–28 days). Two patients in whom stem cell collection was attempted did not undergo SCT: one because a sufficient number of stem cells could not be obtained and the other because of progression of lymphoma after stem cell collection. Overall, 22 of the 35 patients (63%) < 66 years underwent a transplantation procedure (21 ASCT and 1 syngeneic stem cell transplantation).


The results of the current multicenter Phase II study demonstrate that GDP is an active regimen for the treatment of patients with recurrent or refractory diffuse large B-cell lymphoma and variants. An overall RR of 53% was observed. This regimen was administered to outpatients using routine antiemetic prophylaxis and a specific intravenous hydration protocol, with manageable nausea and emesis and little cisplatin-induced neurologic or renal toxicity. Transfusion support with red blood cells (RBC) or platelets was required by only 26% of patients, and febrile neutropenia was experienced by only 14%, which compares favorably with other commonly used salvage regimens.3, 5, 8 Of the 35 patients enrolled who were ≤ 65 years, 63% proceeded to receive high-dose chemotherapy with stem cell support.

The regimen combining dexamethasone, high-dose cytosine arabinoside (ara-C), and cisplatin (DHAP)8 was developed based on in vitro experiments demonstrating synergy between ara-C and cisplatin, even though the single-agent activities of these drugs were modest.23, 24 Based on an RR of 44% in Phase II testing in patients with recurrent or refractory aggressive histology lymphoma, DHAP became widely used as second-line therapy in this patient population and was the standard treatment used in the randomized trial reported by Philip et al.,1 who compared standard-dose therapy with high-dose therapy and autologous bone marrow transplantation. Eligibility for randomization in that trial required at least a PR to two cycles of DHAP, which was observed in 55% of enrolled patients.

Gemcitabine is an analog of ara-C but is more effectively taken up into cells, phosphorylated, and retained intracellularly.25 It has a self-potentiating mechanism of action, resulting in enhanced accumulation and prolonged retention within malignant cells.26 Gemcitabine may also inhibit DNA synthesis by inhibiting the activity of ribonucleotide reductase and decreasing the intracellular nucleotide pool concentrations.27 These properties may allow gemcitabine to be a more effective antilymphoma agent compared with high-dose ara-C. In addition, hematologic toxicities may be less severe with this agent.

Other platinum-containing second-line regimens have been reported, with significant activity observed.4, 5 Moskowitz et al.5 reported an RR of 66% using ifosfamide, carboplatin, and etoposide (ICE) in transplant-eligible patients with recurrent or refractory aggressive histology lymphoma. However, only 58% of the 163 patients in that study proceeded to transplantation, and the event-free survival for the entire cohort was 25% at 40 months. The primary objective of our trial was to determine the RR achieved with two cycles of GDP, which reflects the accepted practice in North America of proceeding to SCT if a patient has achieved an adequate tumour response after two cycles of salvage therapy. In the entire cohort of 51 patients, an objective response was observed after two cycles of protocol therapy in 49% of patients. In the 35 patients who were ≤ 65 years, tumor response and overall patient status were sufficient to permit 22 patients (63%) to proceed to high-dose chemotherapy and SCT. These results appear similar to those reported by Moskowitz et al.,5 using ICE, and by others using DHAP,6, 8 providing a basis to justify comparative testing. Comparisons of the RRs of second-line regimens are difficult because of differences in the patient and disease-related characteristics of the populations included in Phase II trials. Other than the trial reported by Philip et al.,1 which established high-dose therapy with stem cell support as the standard of care for younger patients with recurrent aggressive lymphoma, no other Phase III trials evaluating treatment strategies in this patient population have been published. The question of what constitutes the optimum salvage regimen has therefore not been adequately addressed.

Hematologic toxicity of second-line regimens employed before SCT is often substantial, and may interfere with subsequent attempts at stem cell mobilization. For example, in the initial report of the DHAP combination, 43 of 90 patients (48%) required hospitalization for management of febrile neutropenia or documented infection and 10 of 90 patients (11%) died,8 although that study was carried out before the availability of hematopoietic growth factors. Treatment-related mortality is lower with more recently reported regimens such as carmustine (BCNU), etoposide, ara-C, and melphalan (mini-BEAM; 4%)3 and ICE (0%),5 but hematologic toxicity requiring transfusion support, treatment delay, and hospitalization for management of infection remain significant problems. For example, red blood cell or platelet transfusions were required during 35% and 16.5% of cycles of ICE5 and during 60% and 78% of cycles of mini-BEAM,3 respectively. In contrast, RBC transfusions were given in 14% of cycles and platelets in 10% of cycles of GDP. Furthermore, stem cell mobilization was not impaired after GDP, with 22 of 23 patients (96%) achieving a target stem cell collection of 2.0 × 106 CD34-positive cells per kilogram, which compares favorably to 86% reported with ICE.5

In conclusion, GDP is an active regimen for patients with recurrent or refractory aggressive histology B-cell lymphoma, can be administered to outpatients, and does not interfere with the ability to harvest autologous stem cells for subsequent transplantation. Whether these results can be generalized to patients with other histologic types of lymphoma, such as peripheral T cells, transformed, or Burkitt-like lymphoma, requires further study. In addition, comparative trials are required to determine whether the efficacy and toxicity profile of GDP offers advantages over currently used second-line regimens. To evaluate these hypotheses, the National Cancer Institute of Canada Clinical Trials Group has initiated a randomized Phase III trial comparing GDP with DHAP in patients with recurrent or primary refractory aggressive histology lymphoma as second-line therapy before ASCT, followed by a second randomization posttransplant to maintenance treatment with rituximab 28 or observation.