Lack of Survival Improvement in Patients with Large Cell Lymphomas of T-Cell Phenotype: A SEER Analysis Comparing Outcomes According to Phenotype and Year of Diagnosis.

Background: The majority of studies suggest that among patients with aggressive non-Hodgkin lymphomas (NHL), T-cell phenotype (TCP) confers a poor prognosis. The worse outcomes in TCP may have become more accentuated since the introduction of rituximab, which has been associated with improved survival in patients with diffuse large B-cell lymphoma (DLBCL). The purpose of our study was to evaluate changes in NHL-survival (NHLS) according to immunophenotype and diagnostic era in a large sample of tumor registry cases.

Methods: We analyzed SEER-reported cases aged 19 or older, diagnosed with NHL during the period from 1992 to 2003, and with available information on covariates including gender, race, site of disease, and stage at presentation. Lymphomas with ICD-0-3 codes 9675, 9680, or 684 and B-cell immunophenotype were classified as DLBCL, while TCP cases were identified by codes 9675, 9680, 9684, or 9702 and T-cell immunophenotype. NHLS was analyzed separately for cases diagnosed in 1992–1997 (era 1), and 1998–2003 (era 2), using SAS v9.1. We report 5-year Kaplan-Meier estimates and log rank comparisons, by cell type and era, adjusted for age and SEER summary stage. Since SEER data does not include IPI, we also analyzed two subgroups: patients under age 60 with Ann Arbor stage I/II disease (very good prognosis), and patients with age 60 and older with stage IV (poor prognosis).

Results: Based on 7,359 study cases (6,526 DLBCL; 833 TCP) from era 1 and 18,099 (16,858 DLBCL; 1,241 TCP) from era 2, NHLS was significantly better in patients with B-cell v T-cell phenotype (p<0.001 both eras). Five-year rates were 54.4% v 42.9% in era 1, and 61.3% v 42.2% in era 2. Improvement for DLBCL in era 2 compared with era 1 was also significant (p<0.001), while there was no change for TCP (p=0.708). For the subgroup with very good prognosis, outcomes were comparable for DLBCL and TCP in era 1 but significantly different in era 2 (p=0.531 and p<0.001). The outcome for DLBCL patients with good prognosis improved over time (p<0.001) whereas similar benefit was not seen in TCP patients (p=0.273). Five-year NHLS rates were 75.2% v 73.8% in era 1, and 84.7% v 66.6% in era 2. In the poor prognosis subgroup, TCP was associated with worse outcome in both eras (p=0.011 and p=0.001). For these patient population, improvement in NHLS over time was not statistically significant for either DLBCL (p=0.076) or TCP (p=0.858). Five-year rates for DLBCL versus TCP were 30.6% v 22.2% in era 1, and 35.5% v 25.3% in era 2.

Conclusions: The survival difference between DLBCL and TCP increased significantly after 1997 and is most likely attributable to widespread use of rituximab for B-cell lymphomas during the period 1998–2003. In DLBCL, the improved outcome was seen mainly in patients with good prognosis. In contrast, there were no significant advances in the treatment of T-cell lymphomas. Novel therapies are urgently needed in patients with TCP.