REPORT OF 104 PATIENTS FROM THE SPANISH COOPERATIVE GROUP GEL/TAMO
One hundred and four patients with low-grade(9 patients), intermediate grade(31 patients)and high-grade(64 patients)non-Hodgkin's lymphoma received autologous bone marrow transplantation (ABMT). Disease status at transplant was first complete remission (CR) in 46 patients, second CR in 14 patients, third CR in 7 patients, chemosensitive disease in 16 patients and chemoresistant disease in 21 patients. Estimated 5-year disease-free survival (DFS) for all 104 patients was 49% (95% confidence interval (CI), 36-63%) with a median follow-up of 24 months. Five-year relapse relapse rate for 80 evaluable patients was 26% (95% CI, 14-44%). The 8-year DFS and relapse for the 46 patients transplanted in first CR were 75% (95% CI, 63-82%) and 15% (95% CI, 7-33%) respectively, with a median follow-up of 27 months (range, 13 to 104) and a median time to relapse of 5 months (range, 4 to 20). In the univariate analysis, variables correlated with DFS were performance status at ABMT, disease status at ABMT, LDH level at ABMT, failure to achieve CR at diagnosis, front-line chemotherapy (1 vs 2 or more regimens) and Working Formulation. Variables correlated with relapse were disease status at ABMT, preparative regimen and Coiffier's index at diagnosis. Multivariate analysis showed that performance status was the only independent predictor of DFS and that disease status at ABMT was the best relapse predictory variable. In patients transplanted in first CR, the variables correlated with DFS were stage at diagnosis and performance status at ABMT. LDH level at diagnosis was the only variable correlated with relapse in patients transplated in first CR. The overall mortality was 45% (47/104). Twenty-eight patients (27%) died because of lymphoma and the remaining 19 (18%) as a result of toxicity. Among patients in first CR, 5 (10.8%) died in relapse and the other 5 because of toxicity. This series confirms the fact that the outcome of ABMT is better when performed early in the course of the disease. It is also noteworthy that well-known prognostic parameters at diagnosis which identify patients not likely to be cured with conventional chemotherapy (namely , stage and LDH level) are also associated with a poor outcome after transplant in patients in first CR.
Over the past decade substantial progress has been made in the treatment of patients with aggressive non-Hodgkin's lymphoma (NHL). Current combination chemotherapy regimens are associated with complete response rates of 60-80% and 50-60% of patients become long-term disease-free survivors (1-4) . However, patients who achieve only partial response to front-line therapy and those with refractory or relapsed disease are rarely cured with conventional salvage treatment (5,6) . In a review of 29 chemotherapy trials involving about 700 young persons with lymphoma that failed to achieve an initial remission or suffered a relapse, the 2-year disease-free survival rate was less than 5% (7) .For these patients high-dose chemotherapy and autologous bone marrow transplantation (ABMT) is the only potentially curative treatment modality (6,8-11) . Previous analysis strongly suggests that certain subgroups of patients with NHL are more likely to benefit from this approach. Patients with chemoresistant disease have less than 15% long-term disease-free survival following ABMT (8,9,12) . In contrast, patients with chemosensitive disease are considerably more likely to benefit from ABMT with long-term disease-free survival (DFS) ranging from 25% to 50% (8,9,13) . However, the optimal timing and prognostic factors for ABMT in NHL have not been completely defined yet.
In this report we present the experience of the Spanish Cooperative Group GEL/TAMO with ABMT as treatment for patients with NHL and we emphasize the identification of prognostic factors that were associated with outcome after AMBT.
Between June 1983 and April 1991, 104 patients with NHL underwent ABMT in 15 Spanish Hospitals. Patients were classified according to the Working Formulation for Clinical Usage (14) as low-grade in 9 patients, intermediate grade in 31 patients and high-grade in 64 patients. The number of cases in each histologic group was as follows: follicular small-cleaved-cell 5, follicular mixed 2, follicular large-cell 1, diffuse small-cleaved-cell 4, diffuse mixed 11, diffuse large-cell 13, immunoblastic 23, lymphoblastic 28, small-noncleaved-cell 8 and unclassified 9. Patients with lymphoblastic lymphoma have been previously reported (15) . Thirty patients had B-cell malignant lymphoma and 22 had T-cell malignant lymphoma; immunophenotype could not be determined in 52 patients.Several chemotherapy regimens were used at diagnosis. Patients with low-grade lymphoma received usually CVP (6 patients) (16) . Patients with intermediate-grade lymphoma or immunoblastic malignant lymphoma were treated with CHOP (35 patients) (17) , MACOP-B (14 patients) (1) , ProMACE-CytaBOM (16 patients) (18) or COMLA (3 patients) (19) . The remaining patients with high-grade malignant lymphoma (lymphoblastic and small-noncleaved-cell) were treated with aggressive combination chemotherapy regimens such as LSA2-L2 (25 patients) (20) and 77-04 (5 patients) (21) . Tumor response after induction chemotherapy was: CR 83 patients (80%), partial response (PR) 14 patients (13,5%) and failure 7 patients (6,5%). Of the 83 patients who achieved CR, 68 received one regimen as initial therapy and 15 two regimens. Thirty seven out of the 83 patients who achieved initial CR, relapse before ABMT. Patients who did not reach CR and those who relapsed were treated with salvage therapy consisting of a different regimen from that used initially or with MIME (8 patients) (22) or DHAP (10 patients) (23) . Patients characteristics at diagnosis are shown in Table-1 .
Disease status at the time of ABMT was: first CR in 46 patients, second CR in 14 patients, third CR in 7 patients, chemosensitive disease (SD) in 16 patients (7 patients had a PR and the other 9 had a sensitive relapse) and chemoresistant disease (RD) in 21 patients (14 patients had primary refractory disease and 7 had a resistant relapse). Seventy-one patients (68%) received a single regimen as therapy at diagnosis while the remaining 33 patients (32%) received two or more. A single regimen as front-line therapy was received by 38 patients transplanted in 1st CR, 12 in 2nd CR, 6 in 3rd CR, 9 with SD and 6 with RD. By contrast, 2 or more regimens were received as initial treatment by 8 patients autografted in 1st CR, 2 in 2nd CR, 1 in 3rd CR, 7 with SD and 15 with RD. The median interval from CR to ABMT in patients transplanted in first CR was 4 months (range, 1-32). Any histologic change was observed in patients with active disease at ABMT. Patients characteristics at ABMT are shown in Table-1 .Forty six patients were transplanted in 1st CR. Twenty-six of them had a Coiffier index of 3 at diagnosis. Fourteen patients had an index of 2, six of which needed two different regimens as initial therapy to achive CR. There were six patients, autografted in 1st CR, in which the Coiffier index could not be applied because the LDH level at diagnosis was not known. However, they fulfilled two of the other three criteria included in the Coiffier index, and two of them needed two different initial treatment regimens to achieve CR.
Bone marrow harvest was performed under general anesthesia and cryopreserved in 71 patients, or filtered and stored at 4-10 ºC in citrate-phosphate-dextrose (CPD) for reinfusion within the following 56 h. in the remaining 33 patients. The median number of nucleated cells infused in the patients who received non-cryopreserved rescue marrow was 2.54 x 108/kg (range 1.1 to 10.7). When the rescue bone marrow was cryopreserved, the median numbers of mononucleated cells and infused CFU-GM were 0.49 x 108/kg (range, 0.15-5.9) and 1.8 x 104/kg (range, 0.1-6.37), respectively.
The conditioning regimen was cyclophosphamide (60 mg/kg per 2 consecutive days) and total body irradiation (TBI) (9-12 Gy in 1 to 4 fractions) in 64 patients. The remaining 40 patients were treated with BEAM (17 patients) (24), BEAC (14 patients) (13) or CBV (9 patients)(25) regimens. Mesna was used for prevention of hemorrhagic cystitis. Colony stimulating factors were used to stimulate hematological recovery of the bone marrow in 14 patients. The rescue bone marrow was treated "ex vivo" with monoclonal antibodies and complement in 6 patients and with mafosfamide in 1 case.
Patients were nursed in single hospital rooms with laminar air flow or conventional reverse isolation procedures and received prophylactic antibiotics (either cotrimoxazole or ciprofloxacin and nystatine) and acyclovir if they were herpes simplex virus (HSV) seropositive. All blood components given after transplantation were irradiated to at least 15 Gy and were given as needed to maintain the hemoglobin level above 8 to 9 g/dl and the platelet count above 10 to 20 x 109/l. Empiric broad-spectrum antibiotics, antifungal therapy and parenteral hyperalimentation were used when indicated.
CR was defined as the disappearance of all clinical evidence of disease with normalization of radiographs and laboratory values that had been abnormal before therapy. PR and sensitive relapse were defined as a reduction of 50% or more in measurable disease for at least 1 month. Patients were defined as being in resistant relapse if their lymphoma progressed through their initial combination chemotherapy treatment, or if their disease showed less than a PR to salvage therapy.The major statistical end-points of this study were disease-free survival (DFS) and relapse rate. DFS was defined as the time from the day of ABMT until relapse, death from any cause or date last known alive. Relapse was evaluated in patients transplanted in CR and in those who achieved CR with ABMT.
Factors evaluated in relation with these end-points included the most important variables known at diagnosis or at ABMT. All analyses were made using the NCSS statistical package (26) . Event-time curves were plotted using the method of Kaplan & Meier (27) and differences between groups were evaluated by the log-rank test (28) . Confidence intervals were calculated according to Simon & Lee (29) . Variables found to be associated with DFS or relapse in univariate analyses were entered into a multivariate analysis. Multivariate analysis of prognostic factors was performed using the proportional hazard regression model (30) .
Eight patients died before 21st day post-ABMT and were not evaluable for engraftment. Marrow engraftment occurred in the remaining 96 patients. Median time to granulocyte recovery (=> 0.5 x 109/l) was 21 days (range, 10-103). There was no correlation between the number of cells infused and the time to WBC engraftment. Median time to stable platelet recovery (=> 20 x 109/l) was 26 days (range, 12-307 days). Median time to discharge was 32 days (range, 16-106).
Of the entire group of 104 patients, 53 (51%) remain in continued complete remission (CCR), 4 (4%) are alive after relapse, 15 (14%) have relapsed, 22 (21%) did not achieve CR with ABMT and 10 (10%) have died in remission. The median follow-up of the disease-free patients was 24 months (range 1 to 104 months). The median time to failure among the relapses was 4 months (range 1 to 26 months). Estimated 5-year DFS for all 104 patients was 49% (95% CI, 36-63%), with a median DFS of 26 months. The actuarial probability of relapse for 80 evaluable patients (67 patients transplanted in CR and 13 patients who achieved CR with ABMT) was 26% at 5 years (95%CI, 14-44%).DFS and relapse at 8-years for 46 patients transplanted in 1rt CR were 75% (95% CI, 63-82%) (Figure-1) and 15% (95% CI, 7-33%) respectively, with a median follow-up of 27 months (range, 13 to 104 months). The median time to failure among the 6 patients who relapsed was 5 months (range, 4 to 20 months). The DFS for patients autotransplanted in 2nd CR and 3rd CR was 62% (95% CI, 36-81%) and 29% (95% CI, 1-62%), respectively (Figure-1).
Of 37 patients transplanted with active disease, a CR was achieved with ABMT in 11/16 patients (69%) with sensitive disease and in 2 of 13 evaluable patients (15%) with chemoresistant disease (p=0.0041). The estimated DFS of over 2 years for patients autotransplanted with SD and RD was 33% (95% CI, 12-64%) and 5% (95% CI, 0-22%) (p=0.0002), respectively (Figure-1).
The major prognostic parameters associated with a short DFS in univariate analysis were: Intermediate-grade histology, failure to achieve CR at diagnosis, front-line chemotherapy (>1 regimen), disease status at ABMT (3rd CR, SD, RD), performance status (=>1) and increased LDH level at ABMT (Table-2). . In addition to these parameters, the following variables were also included in a multivariable regression analysis for DFS duration: age-40 years, bone marrow involvement at diagnosis, Coiffier's index, number of extranodal involvement sites, and preparative regimen. Performance status was the only variable retained as an independent predictor of DFS (p=0.0001). Results were identical when patients with low-grade lymphomas were not included in the analyses.Table-2 gives the major prognostic parameters associated with a higher probability of relapse for evaluable patients in univariate analysis: disease status at ABMT (3rd, SD and RD), chemotherapy as preparative regimen, and groups 1 and 3 of Coiffier's index (there were only 3 patients in group 1). In addition to these variables, performance status at ABMT, bone marrow involvement at diagnosis, front-line chemotherapy, LDH level at diagnosis, age-40 years, number of extranodal sites involved at diagnosis were also included in a multivariate regression analysis. However, only disease status at ABMT retained significance for relapse (p=0.0001). Results were similar if patients with low-grade lymphomas were excluded from the analyses. In patients transplanted in 1st CR, variables correlated with DFS in the univariate analysis were stage at diagnosis (p=0.0328) and performance status at ABMT (p=0.0483). Patients transplanted in 1st CR with a higher LDH level than 600 U/l presented a higher relapse rate (p=0.0426).
The major acute toxicity was fever associated with neutropenia. One or more episodes of fever were registered in 95 patients. However, a clinically or microbiologically documented infection was demostrated in only 54/104 patients (52%). Other complications were: some grade of mucositis (48%), pulmonary or cerebral hemorrhage (4%), myocardiopathy (4%),veno-occlusive disease (3%), interstitial pneumonitis (2%) and severe hepatic toxicity (2%). The overall mortality and the causes of death according to disease status at ABMT are shown in Table-3.The causes of toxic death were: Infections in 9 patients, myocardiopathy in 4 patients, thrombocytopenic hemorrhage in 3 patients, veno-occlusive disease in 1 patient, hepatic toxicity in 1 patient and interstitial pneumonitis in 1 patient.
Eight procedure-related deaths were observed before 21st day post-ABMT. One patient was transplanted in 1st CR and died due to pneumonia at day 15, another patient was transplanted in 2nd CR and died of myocardiopathy at day 14 and a third patient was transplanted in 3rd CR and died of thrombocytopenic pulmonary hemorrhage at day 18. The remaining 5 patients were transplanted with resistant disease: 2 died of infection at days 12 and 16; two others died of pulmonary and cerebral hemorrhage at days 18 and 20, and the fifth patient died of myocardiopathy at day 17.
The largest experience of high-dose therapy with ABMT is in the setting of salvage therapy. Results in patients with refractory NHL show high rates of complete remission, short durations of the response and only a few sustained remissions (8,9,34) . In the present study, patients with refractory lymphoma who were autografted showed a CR rate of only 15% and a DFS of 5%, values which are similar to those previously reported (9,12) , these results confirming that ABMT should be carried out before the lymphoma becomes resistant to chemotherapy.
ABMT is more efficacious in patients with chemosensitive disease. In appropriately selected patients, both the rate and durability of response can be improved and the long-term survival is on the order of 40% (9,13,34) . This figure compares favourably with results obtained with conventional therapy where the long-term survival is less than 10% (11) , but is probably affected by patient selection biases. In this report, CR was achieved in 69% of the patients with sensitive disease and the DFS was 33%.
The use of ABMT as consolidation therapy for patients in first CR is controversial. Several theoretical considerations support an early use of ABMT in aggressive NHL. It has now been demonstrated that exposure to chemotherapy can lead to the development of drug resistance in some patients with NHL (35) . Moreover, it is probable that the dose response curve would be steepest in patients treated very early, before mechanisms of chemotherapy resistance develop or increase (36) . Several studies have been reported on patients who underwent high dose therapy and ABMT as consolidation therapy (37-41) . The DFS in these series varied from 65% to 88%, which is much higher than would have been expected with chemotherapy in these high risk patients. In the present study, the DFS and relapse rate for patients autotransplanted in first CR were 75% and 15% respectively.
Our results support the concept that the outcome of ABMT may be improved if it is performed early in the course of treatment while the disease is still sensitive to chemotherapy and the tumor burden is small. However, the optimal timing for ABMT has not been defined. In the GEL-TAMO experience, patients with refractory disease have a poor outcome which probably does not justify treating these patients with ABMT. On the other hand, our results of ABMT in patients with second CR are similar to those reported in patients with first CR and, therefore, it could be argued that ABMT is better suited for relapsed patients who achieve a second response. However, at least 50% of relapsed patients do not achieve a new CR and their outcome with salvage therapy, including ABMT, is very poor. Finally, the good results achieved in patients autografted in 1st CR do not prove that ABMT should be offered as primary therapy to all patients with aggressive NHL. However, they are sufficiently provocative to suggest that controlled trials should be performed in order to test the hypothesis that this treatment approach might yield the highest cure rate. In a preliminary randomized report of the Milan Group, results for ABMT were apparently better than those obtained with MACOP-B (42) .
The most important variable related with relapse was the disease status at ABMT. It is noteworthy that in the univariate analysis the preparative regimen was also significant. The relapse rate in patients treated only with chemotherapy was higher than that of those treated with chemoradiotherapy. The prognostic value of the preparative regimen disappears if patients with lymphoblastic lymphoma are not included in the analysis. This suggests that preparative regimens that include total body irradiation may be more erradicative in this particular lymphoma. The European Bone Marrow Transplant Registry (EBMT) suggests that there is no overall advantage to either TBI or a chemotherapy-only regimen (11) . However, the EBMT registry also suggests that TBI may be preferable in adults with T-lymphoblastic lymphoma who have not yet received cranial irradiation (11) . In order to determine whether some kind of preparative regimen is superior for the different types of NHL, it would be necessary to carry out randomized trials.
The identification of patients with a particularly poor outlook and high likelihood of relapse is the main goal. Several studies have been reported on prognostic factors associated with survival and cure in NHL (43-47) . Most prognostic indexes have been effective in identifying those patients for whom prognosis is poor (48) . In high risk patients both a lower CR rate and a higher relapse rate have been reported. Therefore, therapeutic approaches in high risk patients must be directed towards increasing the low initial CR rates as well as improving the consolidation therapy, with ABMT or other forms of chemotherapy, for complete responders who are at an increased risk for relapse.
Several reports have showed that poor prognostic factors retain their impact on survival in patients achieving a CR (2,43,44,47) . By using these prognostic factors it may be possible to identify a subset of patients with aggressive lymphomas who will do poorly in spite of the fact that they attain a CR with conventional chemotherapy. A noteworthy finding of this report is that well-known prognostic parameters at diagnosis which identify patients not likely to be cured with conventional chemotherapy (namely, stage and LDH level) are also associated with a poor outcome after transplant in patients in first CR.
In the experience of the EBMT, the single most important factor for disease-free survival in the multivariate analysis is the status at the time of ABMT (13) . In the present report, the performance status at ABMT is the only significant prognostic factor that influences both overall survival (data not shown) and DFS, but it is not a good predictor for CR duration. Thus, it may be mainly associated with early mortality. However, the disease status at the time of ABMT is the best relapse predictory variable in the multivariate analysis.
The mortality rate related to ABMT is an important issue. The greater antilymphoma efficacy of the procedure may be counteracted by a high transplant-related mortality. In the EBMT, the procedure-related death rate for ABMT in NHL was 14% (13) . This rate varies according to the status of the disease at the time of ABMT. Thus, in first remission patients the figure was 8%, it was 12% in patients with chemosensitive disease and 22% in those with chemoresistant disease (13) . In this study, the overall transplant-related mortality was 18%. The corresponding death rate in patients in first CR was 11%. The use of the hematopoietic growth factors such as GM-CSF and G-CSF in ABMT can help reduce infectious mortality. These agents may also allow an increased dose of chemotherapy in a nontransplant setting and thus obviate the need for autotransplant in many patients. However, more probably they will make ABMT safer and applicable to more patients (34,42) .
To define the place of ABMT as consolidation therapy in the treatment of NHL, further consideration must be given to the likelihood of relapse and toxicity after conventional chemotherapy and to the toxicity and effectiveness of ABMT. Randomised clinical trials should be carried out in order to establish whether ABMT increases the number of NHL patients who are cured.
REFERENCES