|Home | About | Journals | Submit | Contact Us | Français|
Patients with diffuse large B-cell lymphoma (DLBCL) who do not achieve a complete response to front-line combination chemotherapy are often offered high-dose therapy and autologous hematopoietic cell transplantation (AHCT). However, the efficacy of this therapy in this patient population has been addressed in only a few published reports. We retrospectively analyzed the outcomes of patients with a diagnosis of de novo DLBCL who underwent AHCT at our center between 1988 and 2002, and identified 43 consecutive patients who had not achieved a CR before AHCT, although most showed at least a partial response (PR) to either induction or subsequent salvage chemotherapy. A total of 15 patients received a conditioning regimen that included high-dose chemotherapy with fractionated TBI (FTBI), whereas 28 patients received high-dose chemotherapy only. All autografts were treated ex vivo with MoAbs and complement in an effort to remove any residual malignant B cells. A total of 33 (77%) patients achieved a CR after AHCT. With a median follow-up of 7.3 years, the 5-year OS was 69% and EFS was 59%. Four patients died from non-relapse mortality. By univariate analyses, the following characteristics did not significantly impact OS: disease stage at diagnosis, age-adjusted IPI (International Prognostic Index) score, age ≥40 years, earlier radiotherapy and the use of FTBI in the conditioning regimen. These results confirm the long-term efficacy of AHCT for patients with DLBCL after induction failure.
Approximately 45% of patients with diffuse large B-cell non-Hodgkin’s lymphoma (DLBCL) with advanced disease at diagnosis achieve a CR with anthracycline-containing front-line chemotherapy, although this number has recently increased by approximately 25% since the introduction of rituximab.1–3 An achievement of less than a CR to front-line chemotherapy, hereafter designated as induction failure, is associated with a median survival of <15 months if treated only with conventional chemotherapy.4,5
The outcomes of DLBCL patients who fail induction therapy have been described previously.4,6–11 A majority of patients with induction failure will show chemosensitivity to platinum-containing salvage regimens, but few will experience long-term survival without autologous hematopoietic cell transplantation (AHCT).12,13 Similar to the observations with relapsed disease, the best outcomes are seen in those patients who show chemosensitivity to salvage chemotherapy before AHCT. However, an interpretation of these studies is limited by the inclusion of patients with heterogeneous lymphoma histologies, the use of varied response criteria and the inclusion of partial responders with truly chemorefractory patients. Earlier reports have also lacked long-term follow-up.
In this report, we describe the outcomes of 43 patients with de novo DLBCL who failed to achieve a CR in response to front-line anthracycline therapy and who underwent AHCT with B-cell-purged autografts. Most patients had disease that was deemed chemosensitive before AHCT.
The Stanford Blood and Marrow Transplantation database was screened for patients with a histological diagnosis of non-Hodgkin’s lymphoma (NHL), who underwent AHCT from the inception of the transplantation program in 1988 to December 2002, at which time ex vivo Ab and complement purging was discontinued. Among the 591 consecutive patients fitting these criteria, we identified 156 patients who carried a diagnosis of de novo DLBCL. All patients had biopsy specimens that were reviewed by the Pathology Department at Stanford University, Stanford, CA, USA and were classified according to the International Working Formulation (Group F), REAL (Revised European American Lymphoma) or WHO (World Health Organization) classification systems, and were without mixed histology or evidence of transformation from a low-grade lymphoma. Stage of disease at diagnosis was assigned according to the Ann Arbor System. Age-adjusted IPI was assigned according to the method of Shipp et al.14
Response to induction using anthracycline-containing chemotherapy was assessed by a computed tomography (CT) scan with or without gallium scanning and BM biopsy. Positron emission tomography (PET) scanning was not widely available during the study period and was thus not routinely used in the majority of patients described here. Response criteria were as follows: complete response indicated an absence of all measurable disease, partial response (PR) indicated a >50% reduction in the sum of the product of diameters of all lesions without the appearance of new lesions and stable disease indicated less than a partial response, but without evidence of progressive disease. From among the 156 patients with de novo DLBCL, 46 (29%) patients were identified as having induction failure, that is, only achieving PR or stable disease. Of these 46 patients, 3 were not included in this analysis: one who was treated with high-dose sequential therapy and two who received post transplantation rituximab as part of a separate protocol.15,16 All patients were treated after providing a written informed consent under a research protocol approved by the Stanford University Medical Center Administrative Panel on Medical Human Subjects and according to the Declaration of Helsinki.
The patients who underwent AHCT before 1 April 1994 received BM as the hematopoietic cell source, whereas patients transplanted after this date received mobilized peripheral blood hematopoietic cells. This shift reflected the change in practice at that time within the field of HCT (hematopoietic cell transplantation). Those who received peripheral blood grafts underwent mobilization with CY (4 gm/m2) or VP-16 (2 gm/m2) and G-CSF (10 mcg/kg/day). The autografts were purged as previously described.17 Briefly, nucleated cells were isolated by discontinuous Percoll gradients, washed and resuspended in media containing 1% human albumin. The washed cells were incubated for 30 min at 4 °C in a cocktail of MoAbs directed against B-cell Ags, including anti-CD9 (clone J9), anti-CD10 (clone 4–35), anti-CD19 (clone J149) and anti-CD20 (clone 1F5). DNAse was added at a concentration of 100 U/ml. Rabbit complement was then added at a dilution of 1:8 or 1:16. After an incubation of 45 min at 37 °C, the cells were pelleted by centrifugation, resuspended in media and the process was repeated. The cells were then resuspended in 70% autologous plasma supplemented with 10% dimethyl sulfoxide and frozen in a controlled-rate freezing chamber.
Patients were treated with fractionated TBI (FTBI) in 10 equal fractions to a total dose of 1200 cGy, VP-16 (60 mg/kg) and with CY (100 mg/kg). BCNU (15 mg/kg up to a maximum of 550 mg/m2) or CCNU (15 mg/kg orally) was substituted for FTBI in those patients who had received earlier radiotherapy to the chest or pelvis, or if excessive morbidity with FTBI was anticipated on the basis of age >50 years or poor performance status. Chemotherapy doses were determined according to the adjusted ideal body weight, defined as ideal body weight plus 50% of the difference between ideal body weight and actual body weight.
Neutrophil engraftment was defined as the first of three consecutive days after transplantation with an absolute neutrophil count (ANC) of ≥500/mm3. Platelet engraftment was defined as the first day with a platelet count of ≥20 000/mm3 and no platelet transfusions within the previous five days.
Restaging CT scans and BM biopsies were performed at 3, 6 and 12 months after AHCT, and then annually.
Outcome was analyzed with respect to response, OS, EFS and freedom from progression (FFP). Time to an OS event was defined as time from HCT to death from any cause. The EFS event time was defined as time from HCT to death from any cause or progression of disease. In FFP analysis, disease progression was defined as an event with censoring at the time of death, without disease progression. Two patients who achieved CR after AHCT subsequently relapsed and proceeded to non-myeloablative allogeneic transplantation. These patients were censored from analysis of OS, EFS and FFP at the time of second HCT. The Kaplan–Meier method was used to estimate actuarial OS, EFS and FFP. Cox proportional hazards regression models were used to identify the predictors of survival. Significance (P≤0.05) was determined from the Wald χ2-test.
Patient characteristics are listed in Table 1. There were 24 men and 19 women, ranging in age from 19 to 64 years, with a median of 43 years. A total of 12 (28%) patients had stage I–II disease, 5 (12%) had stage III disease and 26 (60%) had stage IV disease. A total of 28 (65%) of 43 patients had extranodal disease, 11 of whom had more than one extranodal site of involvement. A total of 10 (23%) patients had BM involvement, 17 (40%) had bulky disease and 26 (60%) experienced B symptoms. The age-adjusted NHL IPI score was low risk in 20 (47%) patients, high risk in 14 (32%) patients and not available in 9 (21%) patients.
Treatment characteristics are presented in Table 2. All patients were initially treated with an anthracycline-containing regimen: 29 (70%) patients received CHOP, 6 (14%) received rituximab–CHOP and 8 (16%) received other regimens. A total of 22 of 43 patients achieved a PR with a median of six cycles (range 2–12 cycles). The remaining 21 patients had stable disease after a median of six cycles of chemotherapy (range 2–8 cycles). A total of 13 patients did not receive salvage chemotherapy, and proceeded directly to BM harvest (n=2) or mobilization chemotherapy (n=11) and AHCT, including four patients with stable disease. Of the 30 patients who received salvage chemotherapy, 25 received a platinum-containing regimen. One patient who did not receive rituximab as a part of induction therapy received rituximab as a part of salvage therapy. All patients who received salvage chemotherapy achieved a PR. The median time from mobilization chemotherapy to transplantation was 32 days (range 27–42 days). Patients were not routinely restaged after mobilization. A total of 10 (23%) of the original 43 induction failure patients required involved field radiation therapy for disease control before mobilization chemotherapy.
The median time from diagnosis until AHCT was 8.8 months (range, 2.3–15.3 months). The majority of patients (n = 26, 60%) received BCNU, CY and VP-16, whereas the remainder received FTBI (n = 15, 35%), or CCNU, CY and VP-16 (n = 2, 5%) as the preparative regimen. Six patients (14%) received a BM graft. Of the 37 patients who received a peripheral blood graft, 32 received CY as mobilization chemotherapy and 5 received VP-16. All patients received B-cell-purged products. A total of 10 (23%) of the 43 patients received involved field radiation therapy after AHCT. Of these, eight had consolidative radiotherapy that had been planned before AHCT to sites of earlier bulky disease. The remaining two patients received radiotherapy as a component of a non-myeloablative conditioning regimen for salvage allogeneic HCT.
The median time to neutrophil and plt engraftment was 10 days (range 8–30 days) and 14 days (range 11–33 days), respectively. There were two deaths (4.7%) within 100 days after AHCT: one from interstitial pneumonitis and the other from hepatic failure. These patients were not evaluated for disease response. Eight patients failed to achieve a CR with a median time to disease progression of 144 days (range 81–189 days). A total of 33 patients (77%) achieved a CR after AHCT. Of these, seven had a relapse of disease with a median time to relapse of 1363 days (range 307–2416 days). With a median follow-up of 7.3 years (range 0.1–16.3 years), the 5-year OS was 69% (95% CI 55–83%) and EFS was 59% (95% CI 42–74%), with a median EFS of 6.7 years. The median OS has not been reached. The 5-year FFP rate was 63% (95% CI 48–78%) (see Figure 1). Causes of death after AHCT were related to relapse in 10 patients and 5 patients died from non-relapse mortality due to hepatic, cardiac or respiratory failure. One patient died of unknown causes. Two patients developed treatment-related myelodysplasia, including one patient who was diagnosed 16 years after transplantation and 10 years after salvage chemotherapy for recurrent disease after AHCT.
Univariate analysis for prognostic factors is presented in Table 3. Stage at diagnosis, before AHCT irradiation, age-adjusted IPI, number of earlier regimens before AHCT (≤3 vs 3), age ≥40 years and the use of FTBI in the preparative regimen were not significantly associated with worse outcomes.
In this report, we describe the efficacy of AHCT in patients with de novo DLBCL who failed to achieve a CR to front-line conventional chemotherapy. With a lengthy follow-up, we show an OS of 69% and an EFS of 59%, which compares favorably with earlier published reports of the outcomes of patients with induction failure, provided chemosensitivity is established.
The landmark Parma trial established the role of AHCT in patients with chemosensitive, relapsed disease, but the value of AHCT for patients with aggressive NHL who failed induction chemotherapy is less well defined.5 Previous studies have shown a poor prognosis for this particular population, with salvage therapy yielding an EFS of only <15% and a median OS of <15 months.4,5 In a randomized study, a prospective Italian multicenter study attempted to establish the optimal regimen for NHL patients who were induction failures.8 Patients were restaged after receiving four of six planned cycles of front-line therapy. Those who achieved a PR were randomized to receive salvage chemotherapy (n=27) or AHCT (n=22). The 3-year OS was 59% for the salvage group vs 73% for the AHCT group, with the corresponding PFS being 52 and 73%. Owing to the small sample size and lack of power, these differences were not statistically significant.
In an analysis from the Autologous Blood and Marrow Transplant Registry, Vose et al.11 identified 184 patients with diffuse aggressive NHL who never achieved a CR with conventional chemotherapy. A total of 96% of patients received subsequent salvage chemotherapy, with 60% being deemed chemosensitive and 28% being chemotherapy resistant before AHCT. A total of 12% were not assessable. For all patients, the 5-year OS and PFS were 37% and 41%, respectively, but patients who were chemoresistant at the time of AHCT had a 2.44-fold higher relative risk of death from relapse than chemosensitive patients (P = <0.0001), thus confirming the unequivocal prognosticator of chemosensitivity. Response after transplantation also affected survival, as OS was 61% for patients who obtained a CR after AHCT compared with only 11% for PR patients and non-responders. These data are similar to a report from the GEL-TAMO (Grupo de Espanol de Linformas/Trasplante Autologo de Medula Osea) registry of 114 DLBCL patients who did not achieve a CR after induction.18 The 5-year OS was 43% for all patients. For patients who were truly chemoresistant and who did not respond to salvage therapy, OS and EFS were 10 and 0%, respectively, compared with 61 and 71%, respectively, for PR patients. Unlike our analyses, both of these registry studies included chemoresistant patients, which may explain our superior survival incidences.
In a single institution retrospective analysis, the Memorial Sloan Kettering group reported the outcome of 85 patients with aggressive NHL who achieved less than a CR after front-line chemotherapy.4 All patients subsequently received ICE (ifosfamide–carboplatin–etoposide) salvage chemotherapy. In an intent-to-treat analysis, the 3-year OS and EFS were 25 and 22%, respectively, but when the patients who actually proceeded to AHCT (n=42) were analyzed separately, the 3-year OS and EFS were 52 and 44%, respectively. The patients who did not proceed to AHCT had a median OS of only 3.7 months. Thus, these results also validate the role of AHCT for this challenging patient population. The investigators note that the outcomes of their chemosensitive patients who proceeded to AHCT were similar to the Parma results, which confirm that once chemosensitivity is established, there is no significant difference in outcome after AHCT in patients who were induction failures or who had relapsed disease.
As previously mentioned, our results compare favorably with the above mentioned studies and show the long-term success of this treatment approach. In contrast to those reports, our study was limited to patients with histologically confirmed DLBCL. Previous similar studies included diverse NHL subtypes, including transformed indolent disease, high-grade disease or T-cell histologies.4,8,11 Another distinct feature of our study is the lengthy follow-up of 7.3 years as other similar published studies only reach a follow-up duration of up to 4 years. Despite the long follow-up duration, our favorable outcomes have remained durable over time.
All of our patients received ex vivo purged autografts, which is another distinct feature of our study and also may have contributed to our notable outcomes. Our method of purging with a panel of MoAbs and complement has been shown to reduce the lymphoma tumor burden from peripheral blood autografts.17 Purging had no impact on the time to neutrophil and plt engraftment. A comparison of this with subsequent protocols at our institution using unpurged autografts showed comparable times to engraftment.
The impact of purging on survival after AHCT for NHL is controversial.19 Randomized trials that aimed at assessing the value of in vitro purging have shown no benefit, although some of these studies were limited by the inclusion of diverse lymphoma histologies and a variety of purging techniques.19 Of patients with DLBCL in CR mobilized with chemotherapy and G-CSF, approximately 50% will have apheresis products that show a PCR evidence of contamination with tumor cells.20 Vose et al.21 also reported that patients with DLBCL who received unpurged BM with PCR evidence of contamination by lymphoma had an inferior EFS compared with patients whose grafts were PCR negative for tumor cells. There are no randomized studies evaluating the impact of B-cell purging of autografts in DLBCL, and it is unlikely that such studies will ever be conducted given the labor and cost of producing clinical-grade antibodies, as well as the increasing use of rituximab for the purposes of in vivo purging.22
In addition, with rare exception, DLBCL patients are currently treated with rituximab as part of front-line therapy, salvage therapy or both. Rituximab has also been used for purposes of in vivo purging and has yielded PCR-negative grafts primarily in patients with follicular and mantle cell NHL.22–24 Thus, the advent of rituximab has nearly eliminated the need or interest in pursuing ex vivo purging with MoAbs or immunomagnetic columns, both of which are quite labor intensive. Only seven patients in our cohort received rituximab before AHCT, and thus the impact of rituximab in our cohort cannot be accurately assessed. A recent publication by Alousi et al.25 has shown that patients with metabolically active disease and who receive rituximab as a component of the mobilization regimen have a 5-year PFS of 46%. In contrast, similar patients who did not receive rituximab had a 5-year PFS of 10%.
Another possible explanation for our notable results is that our patients were assessed by CT and not by metabolic imaging with [18F] fluorodeoxyglucose PET scanning, which may have led to an underestimation of a true CR rate. CT scanning is unable to distinguish viable tumor from fibrotic or necrotic tissue, and is therefore likely to overestimate the presence of viable tumor. Juweid et al.26 compared the response assessment by CT with PET–CT in 54 patients with aggressive NHL, including 47 with DLBCL. A total of 31% of patients evaluated with CT achieved a CR, whereas restaging with PET–CT increased the CR rate to 65%. PET–CT eliminated the category of Cru (complete response, unconfirmed) and also reduced the percentage of PR from 35 to 22%. The 3-year PFS was significantly lower in patients scored as PR by CT–PET compared with those scored as PR by CT alone. Thus, it is possible that a percentage of our patients were actually in a CR after front-line therapy and thus were already in a favorable prognostic group.
In summary, we conclude that for de novo DLBCL patients, failure to achieve a CR to upfront combination chemotherapy does not necessarily confer a poor prognosis. Our favorable outcomes may partially be attributed to our homogenous population of DLBCL patients and to the use of B-cell purging. Our results showed that over half of such patients can be cured with AHCT, provided chemosensitivity is established either with induction and/or with salvage chemotherapy. The impact of our B-cell purging technique cannot be definitively assessed as this was not a randomized trial, but the current widespread use of rituximab during induction, salvage and/or mobilization is most likely to reduce risk of tumor cell contamination in mobilized autografts and may ultimately improve outcomes of this patient population. Thus, our results show the long-term benefit of this treatment strategy for patients with DLBCL who do not achieve a CR with induction chemotherapy.
This study was supported in part by grant, P01-CA 49605, from the National Institutes of Health, Bethesda, MD, USA.
Conflict of interest
The authors declare no conflict of interest.