Abstract
Prognostic models for primary diffuse large B-cell lymphoma (DLBCL) resistant to or relapsing after initial therapy with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) have not been well-established. A refined categorization of normalized lactate dehydrogenase (LDH) (the ratio to the upper limit of normal) was recently shown to predict prognosis of newly-diagnosed patients with DLBCL better than mere dichotomization of LDH into ‘normal’ and ‘abnormal’ groups. To define the prognostic impact of the LDH ratio and various other factors in order to develop a new prognostic model system, we retrospectively analyzed 31 patients with refractory/relapsed DLBCL initially treated with R-CHOP at our hospital from 2002 to 2013. Median age at the start of salvage therapy was 63 years. Median time from diagnosis to relapse was 349 days. In univariate analysis, LDH, short time from initial diagnosis to relapse (TTR), and low absolute lymphocyte count at relapse (ALC-R) were significant factors for reduced progression-free survival, and the LDH ratio was a more powerful factor than LDH abnormality. Furthermore, an LDH ratio >3 and short TTR were independent prognostic factors in multivariate analysis. We built a new prognostic scoring system, namely the time, LDH, and lymphocyte count (TLL), based on TTR, LDH ratio, and ALC-R, which is able to separate patients into three risk groups with 2-year PFS of 100%, 68.6% and 4.8%, respectively, and which also predicts outcome of autologous stem cell transplantation at the start of salvage therapy. The present study indicates that the LDH ratio is an important predictor of survival for patients with refractory/relapsed DLBCL and proposes the TLL index as a useful prognostic tool.
Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin's lymphoma all over the world (1). Rituximab with cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) has been established as the standard initial therapy for DLBCL, with about 60% of cases cured with this frontline therapy (2, 3). However, a significant number of patients experience relapse and need effective salvage therapies. Thus, treatment for patients with refractory or relapsed DLBCL after R-CHOP is still a crucial problem.
For untreated patients with DLBCL, prognostic models, such as the International Prognostic Index (IPI) and revised IPI, are well established (4, 5). More recently, a refined categorization of normalized lactate dehydrogenase (LDH) (the ratio to the institutional upper limit of normal) was revealed to provide better risk prognostication than the mere dichotomization of LDH into ‘normal’ versus ‘abnormal’ and was incorporated into National Comprehensive Cancer Network-IPI (6). Furthermore, risk-adapted treatment strategies, such as an upfront high-dose chemotherapy followed by autologous stem cell transplantation (ASCT) for high-risk patients (7), have been tried. However, prognostic models for relapsed/refractory DLBCL have not been well established, and risk-oriented treatment strategies have not been conducted.
Before the rituximab era, the IPI at relapse (IPI-R) was shown to correlate with the response to salvage therapies and survival for patients with relapsed/refractory aggressive lymphoma by the PARMA trial (8), which also demonstrated superiority of high-dose chemotherapy followed by ASCT over conventional chemotherapy. Subsequently, time from initial diagnosis to relapse (TTR) and absolute lymphocyte counts at the time of relapse (ALC-R) have been shown to predict clinical outcome for patients with DLBCL in first relapse (9, 10). Although prior rituximab exposure before salvage therapy has been reported to be an important predictor of outcome (11, 12), rituximab-naïve patients are currently almost absent. In this regard, prognostic factors for patients with DLBCL which relapsed after initial therapy containing rituximab are not well established (11, 12). It should be also noted that no report has addressed the significance of categorized LDH ratio for prediction of outcomes for patients with refractory/relapsed DLBCL.
Patient characteristics.
After salvage therapies, transplant-eligible patients with refractory/relapsed DLBCL proceed to high-dose chemotherapy followed by ASCT in general practice. Although the response to salvage therapies and disease status before ASCT have been reported to predict outcome of ASCT, there are limited data on correlation between prognostic factors at the start of salvage therapy and ASCT outcome (10-12).
In the present study, we retrospectively analyzed the prognostic factors at the start of salvage therapy for patients with DLBCL refractory to or in first relapse after initial R-CHOP therapy, and created a simple prognostic index by combining the LDH ratio, TTR, ALC-R for prediction of outcome of these patients including those treated with ASCT.
Patients and Methods
This is a single-institution retrospective analysis of consecutive patients with DLBCL undergoing salvage chemotherapy after initial treatment with R-CHOP at the Tokyo Medical and Dental University Hospital, performed with approval of our Institutional Review Board (M2000-2157).
Patients. All patients underwent salvage chemotherapy with curative intent at our hospital. Inclusion criteria were (i) diagnosis of DLBCL (according to the WHO classification), (ii) in first relapse or refractory to initial therapy with R-CHOP with/without radiotherapy. Patients with meningeal or central nervous system involvement were excluded. Overall, 31 patients diagnosed with relapsed or refractory DLBCL who received salvage therapy between 2002 and 2013 were included in the study.
Re-staging and prognostic factors. All patients were re-staged with positron-emission tomography combined with computed tomography or computed tomography. Laboratory data and prognostic factors were re-assessed at the start of first salvage therapy.
Treatment. Patients mainly received CHASE (cyclophosphamide, high-dose cytarabine (Ara-C), methylprednisolone, etoposide) (13) or ESHAP (etoposide, methylprednisolone, high-dose Ara-C, cisplatin) (12) regimen with rituximab as salvage therapy. The decision to proceed to peripheral blood stem cell (PBSC) harvest/ASCT was physician's choice. Autologous PBSCs were collected during hematological recovery after the second or third chemotherapy course. The target cell dose was 2×106 CD34+ cells/kg in a PBSC harvest. Conditioning regimen for ASCT included 200 mg/m2 ranimustine on day −8 and −3, 300 mg/m2 carboplatin from day −7 to −4, 500 mg/m2 etoposide from day −6 to −4, and 50 mg/kg cyclophosphamide on day −3 and −2 (MCEC) as previously described (14).
Definition of response criteria. Response were classified according to the International Working Group criteria (15). Disease status at salvage therapy was classified as relapse, progression, or primary refractory. Primary refractory included patients who had shown response less than partial remission (PR) with/without progressive disease (PD). Response to salvage therapy was assessed by conventional diagnostic methods after the second or third chemotherapy course.
Statistical analysis. Progression-free survival (PFS) was defined as the time from the start of salvage therapy to disease progression or death. Overall survival (OS) was defined as the time from the start of salvage therapy to death or date of last follow-up for patients who were alive and censored.
The Kaplan–Meier method was used to estimate PFS and OS. Differences in survival between two groups were analyzed by the log-rank test. Cox proportional hazard regression analysis was performed for factors included in IPI [age, performance status (PS), Ann Arbor stage (CS), extranodal disease, LDH], ALC-R, and TTR at the start of salvage therapy. All reported p-values are two-sided, and values of p<0.05 were considered significant. Receiver operating characteristic (ROC) and area under the curve analyses were used to determine the ALC-R cutoff value. All analyses were carried out with the EZR software (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical computing) (16).
Results
Patients' characteristics. Patient characteristics at diagnosis and when judged as the first relapse/refractory status are summarized in Table I. The median age of the cohort at the time of first relapse/refractory status was 63 years (range=30-85 years). Nineteen patients (61.3%) had achieved CR after R-CHOP with/without radiotherapy. Median time from diagnosis to relapse was 349 days.
Univariate analysis for chemosensitivity.
Response to first salvage therapy. Patients received mainly R-ESHAP (48.5%) or R-CHASE (22.5%) as the first salvage therapy (Table I). The overall response rate (ORR: CR plus PR) for the whole cohort was 67.7% and that for those aged >70 years and ≤70 years was 71.4% and 66.7%, respectively. The factor significantly affecting the ORR was TTR (≤12 months vs. >12 months, p=0.0005), while LDH (≤ULN vs. >ULN, p=0.0661; LDH ratio ≤ULN vs. 1-3 ULN vs. >3 ULN; p=0.0751) marginally affected the ORR. The ORR to first salvage therapy tended to be lower in patients with ALC-R <1100/μl than that in those with ALC-R ≥1100/μl (p=0.106) (Table II).
Survival and response to first salvage chemotherapy. After a median follow-up for surviving patients of 23 months (range=8-103 months), 16 patients were still alive (51.6%), while 74% of the patients experienced relapse or progression at a median time of 8 months (range=0.6-30.1 months) after the administration of first salvage therapy. The 2-year PFS and 2-year OS were 27.4% and 46.8%, respectively, for the whole patient cohort. Patients who were sensitive to first salvage therapy had significantly better PFS (57.1% vs. 0%) and OS (100% vs. 13.3%) at 1 year than other patients (Figure 1A and B, respectively). Univariate and multivariate analysis of survival. To test previously reported prognostic factors, Cox proportional hazards regression analysis was performed. In univariate analysis, TTR, LDH, LDH ratio, and ALC-R were identified as significant predictors for PFS (Table III). Importantly, the categorized LDH ratio provided better risk prognostication than the mere dichotomization of LDH as normal vs. abnormal. Thus, multivariate analysis was performed for LDH ratio, ALC-R and TTR. Short TTR and the LDH ratio (>3 ULN) were significantly associated with poorer PFS also in multivariate analysis (Table III).
Impact of the response to the first salvage therapy on progression-free (A) and overall (B) survival of patients with relapsed/refractory diffuse large B-cell lymphoma. CR, Complete remission; PR, partial remission.
Univariate and multivariate analysis for progression-free survival.
The time, LDH and lymphocyte count (TLL) prognostic scoring system. The three prognostic factors identified were then integrated into a new prognostic scoring system. This TLL scoring system used a maximum of 5 points for TTR (>12 months, 0 points; ≤12 months, 2 points), the categorized LDH ratio (≤1, 0 point; >1-3, 1 point; >3, 2 points) and ALC-R (≥1100/μl, 0 point; <1100/μl, 1 point). Three risk groups were classified based on Kaplan–Meier curves for 2-year PFS: Low (0 point, N=3), intermediate (1 point, N=7), and high (≥2 points, N=21). The 2-year PFS after first salvage therapy for patients in these groups was 100%, 68.6% or 4.8%, respectively (Figure 2A; p<0.0001). Similarly, the 2-year OS for patients was 100%, 83.3%, or 24.7%, respectively (Figure 2B; p=0.0081). In this prognostic model, risk categories correlated not only with ORR after first salvage therapy (low: 100%, intermediate: 100%, high: 47.6%), but also with PFS for patients who responded to the first salvage therapy (Figure 2C; p=0.013).
ASCT. Adequate numbers of PBSCs were collected for 13 out of 16 patients for whom the harvest was attempted and 12 patients proceeded to ASCT. All patients who received ASCT were aged ≤70 years and, except for one patient, had achieved PR/CR after first salvage therapy.
Patients treated with ASCT had a significantly better 2-year PFS (47.6% vs. 14.0%) and 2-year OS (72.2% vs. 27.8%) compared to patients not receiving ASCT (Figure 3A, B; p=0.0031, p=0.0397, respectively). When limited to the high TLL group, patients treated with ASCT still had significantly better 2-year PFS and tended to have better 2-year OS as compared to those without ASCT (Figure 3C and D; p=0.0368 and p=0.156, respectively). Importantly, it was also demonstrated that the TLL scoring model discriminated PFS and OS of patients who received ASCT (Figure 4A and B).
Discussion
Although IPI-R was reported as a useful prognostic tool for patients with relapsed DLBCL in the PARMA study, Panizo et al. recently reported that revised IPI at relapse is not useful in patients with relapsed/refractory DLBCL after immunochemotherapy (17). Moreover, Porrata et al. showed that, although IPI-R is useful, LDH had more impact than IPI-R in patients with relapse (11). In accordance with this, we found in this study that no factors other than LDH in IPI-R showed any statistical significance in prediction of PFS (Table III). Furthermore, we showed that the categorized LDH ratio, particularly that of >3, more strongly correlated with PFS in univariate and multivariate analysis (Table III). This is in agreement with the recent report that the LDH ratio, which should reflect tumor burden and aggressiveness, provided better prognostication than mere LDH abnormality in patients with untreated DLBCL (18). However, to the best of our knowledge, this is the first report to demonstrate the prognostic importance of the LDH ratio for relapsed/refractory DLBCL previously treated with the standard initial R-CHOP therapy.
The time, normalized lactate dehydrogenase (LDH), and lymphocyte count (TLL) scoring system in risk stratification for patients with relapsed/refractory diffuse large B-cell lymphoma. Progression-free (A) and overall (B) survival of high-, intermediate (Int)-, and low-risk groups of patients. C: Progression-free survival of different risk groups of patients who, at least partially, responded to the first salvage therapy.
Impact of autologous stem cell transplantation (ASCT) on the progression-free (A, C) and overall (B, D) survival of the whole patient cohort (A, B) and high-risk group patients by the time, normalized lactate dehydrogenase and lymphocyte count (C, D).
TTR reflects aggressive tumor biology as well as chemosensitivity and has been reported to predict clinical outcomes in various lymphoid malignancies (19, 20). The PARMA trial demonstrated that TTR of less than 12 months correlated with poor prognosis in relapsed/refractory aggressive lymphoma (21), which was confirmed in patients with relapsed DLBCL in the rituximab era by the CORAL study (10). In accordance with these studies, we have demonstrated that TTR is the most powerful predictor of PFS as well as response to the salvage therapy in this study (Table II and III). On the other hand, ALC-R is a useful tool as a surrogate maker of host immunity at relapse, which is important for patients who receive immunotherapy. Thus, ALC-R was shown to associate with better response and clinical outcome in relapsed/refractory DLBCL (10). In accordance with this, ALC-R was identified as a significant prognostic factor for PFS in the present study (Table III).
Prognostic models, such as IPI-R/TTR and IPI-R/ALC-R, have been proposed for patients with relapsed DLBCL in previous reports (9, 10), which, however, have not been generated on clinical data of such patients experiencing relapse exclusively after standard R-CHOP therapy. Moreover, the prognostic significance of the LDH ratio was not evaluated in these studies and was not included in the previous prognostic models for patients with relapse of DLBCL. Based on the data on patients with relapsed/refractory DLBCL initially treated with R-CHOP, in this study we propose TLL as a new prognostic scoring system, consisting of three prognostic factors, TTR, the LDH ratio, and ALC-R, which reflect aggressiveness, chemosensitivity, tumor burden, and host immunity. This simple model was able to separate patients into three groups with distinct PFS and OS (Figure 2). Importantly, this scoring system also predicted outcome of subsequent ASCT at the start of salvage therapy (Figure 4). It was also revealed that although patients who underwent ASCT had better prognosis than patients who did not (Figure 3A and B), the patients with high TLL did not show any significant difference in OS with or without ASCT (Figure 3D). Furthermore, the PFS for patients with high TLL treated with ASCT, although better than without ASCT, was only about one year after starting salvage therapy (Figure 3C). Thus, alternative therapeutic strategies, including allogeneic stem cell transplantation, may be required for these patients.
The time, LDH, and lymphocyte count (TLL) scoring system in risk stratification for patients who received autologous stem cell transplantation (ASCT). Progression-free (A) and overall (B) survival of patients who received ASCT according to TLL risk. Int: Intermediate.
Data obtained by new technologies, such as gene-expression profiling and next-generation sequencing analyses, could be incorporated to generate new prognostic model systems for refractory/relapsed DLBCL in the future. However, these technologies are currently utilized only in limited clinical trials. The TLL prognostic model system we propose in this study is simple and applicable to general clinical practice. However, it should be noted that this model system, generated from a rather limited sample size, needs to be validated in an independent cohort in future studies.
Footnotes
Conflicts of Interest
The Authors declare that they have no conflict of interest in regard to this study.
- Received February 27, 2017.
- Revision received March 23, 2017.
- Accepted March 24, 2017.
- Copyright© 2017, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved