Abstract
Introduction
Overall survival (OS) of B-cell lymphoma patients, including diffuse large B-cell lymphoma (DLBCL), has markedly improved since the development of rituximab. Rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) have been established as the standard initial therapy for DLBCL patients over past 10 years. However, investigation of risk factors for new prediction models for DLBCL patients in the rituximab-era has been limited. Thus, the International Prognostic Index (IPI) has been used over past 20 years for risk stratification of lymphomas, and although it retains some usefulness, the revised IPI model cannot identify the poor prognosis group with 5-year OS of < 40%. This study aimed to verify risk factors for survival and responses to R-CHOP or R-CHOP-like therapy in DLBCL patients, and to construct a new prediction model to supplant the IPI model.
Methods
A historical cohort study comprised patients with de novoDLBCL who received R-CHOP or R-CHOP-like therapy as initial therapy at our institutions between January 2002 and July 2013. A retrospective analysis was performed to identify the pre-therapy risk factors for survival and lymphoma progression. Transformed DLBCL and intravascular large B-cell lymphoma patients were excluded.
Results
Study comprised 254 patients (145 males, 109 females). Median age at diagnosis was 70.0 years (range 20–96 years). Response rates in patients to first regimen of therapy were as follows: 80.7%, CR (n = 205); 9.1%, PR (n = 23); 1.2%, SD (n = 3); and 9.1%, PD (n = 23). Ann Arbor clinical stage was as follows: 42 (16%) patients. stage I; 60 (24%), stage II; 47 (18%), stage III; and 105 (41%), stage IV. B symptom was observed in 40 (16%) patients and a bulky mass in 38 (15%). The revised IPI indicated 23 (9%) very good, 88 (35%) good, and 143 (56%) poor risk cases. In 235 patients, we categorized 80 cases as germinal center B-cell-like (GCB) type and 155 cases as non-GCB type using Hans’s algorithm of immunohistochemistry. Autologous peripheral blood stem cell transplantation was performed in 18 patients. Initial therapy that included radiation was performed in 29 patients. 5-year OS was 59.3% and 5-year progression-free survival (PFS) was 48.7%. According to the revised IPI, 5-year OS was stratified as 100%, very good; 69.2%, good; and 46.4%, poor (P< 0.001). Based on univariate analysis using Log-rank tests, significant poor prognostic factors for OS were as follows: B symptom positivity (P = 0.003), hemoglobin< 12.0 g/dL (P< 0.001), albumin < 4.0 g/dL (lower limit of normal; P< 0.001), C-reactive protein (CRP) > 0.6 mg/dL (P< 0.001), soluble IL-2 receptor (sIL-2R) > 2000 U/mL (P< 0.001), and non-GCB type (P = 0.03) along with all IPI factors including age ≥ 61 (P< 0.001), Ann Arbor clinical stage ≥ 3 (P = 0.005), lactate dehydrogenase (LDH) > 220 IU/L (upper limit of normal; P< 0.001), ECOG performance status ≥ 2 (P< 0.001), and number of extranodal disease sites ≥ 2 (P = 0.02). In multivariable analysis for 235 patients using Cox proportional hazards model including 11 previously noted factors, age (P = 0.003), CRP (P = 0.04), hemoglobin (P< 0.001), and albumin (P = 0.02) were significant. Step-down procedure in multivariable analysis also identified these 4 factors, age [hazard ratio (HR) 2.84, P = 0.002], CRP (HR 2.10, P = 0.003), hemoglobin (HR 2.36, P< 0.001), and albumin (HR 2.16, P = 0.008), as significant in 254 patients. These 4 factors were used to define the ACHA (Age-CRP-Hemoglobin-Albumin) score. ACHA scores stratified 5-year OS as 100%, 82.1%, 64.3%, 51.4%, and 22.3%, and 5-year PFS as 93.9%, 71.2%, 51.5%, 37.3%, and 14.6% as 0 (n = 34), 1 (n = 46), 2 (n = 48), 3 (n = 66), and 4 (n = 60) of the ACHA scores, respectively (OS, P< 0.001: PFS, P< 0.001).
Conclusions
Although predictive power of this scoring system should be validated using another data set, ACHA scoring factors at diagnosis stratified expected survivals. ACHA was constructed without Ann Arbor staging; thus, computed tomography scanning and bone marrow examination were not required for risk stratification. Moreover, this predictive scoring system could extract poor prognosis patients with 5-year OS < 30% and patients highly likely to be resistant to R-CHOP or R-CHOP-like regimen with 5-year PFS < 20%. New therapy regimens that improve OS for such patients with very poor predicted prognosis are required.
Disclosures:
No relevant conflicts of interest to declare.
Novel prognostic scoring system for diffuse large B-cell lymphoma