Late-Relapse Diffuse Large B-Cell Lymphoma Frequently Represents Recurrence of the Original Disease, and Demonstrates Evidence of Superimposed Clonal Heterogeneity and Clonal Evolution

BACKGROUND: Diffuse large B cell lymphoma (DLBCL) is the most common non-Hodgkin lymphoma. While most relapses occur within 2 years, a small proportion of patients present with late relapse (LR) after 5 years. As there are very few studies addressing the pathobiology of LR-DLBCL, the aim of this study is to further characterize the clinical, pathologic and molecular features of these neoplasms.

METHODS: A retrospective analysis of all patients with DLBCL treated at Johns Hopkins Hospital between 1984 and 2013 was performed. Patients with low-grade lymphoma at any time-point were excluded. Disease-free intervals (DFI) of 5 years or greater were designated as LR. Five paired diagnostic (D) and relapse (R) samples were available for further studies. DNA was extracted from formalin fixed paraffin embedded tissue. IGH gene rearrangement status was determined by PCR. SNP microarray was performed, and copy number variations (CNV) were defined as loss or gain of signal over at least 2 megabases. Targeted next generation sequencing (NGS) using a cancer hotspot panel was also performed. Variant calls were generated using Torrent variant caller and a laboratory-developed analysis pipeline.

RESULTS: One hundred thirty-three patients with relapsed DLBCL were identified. Forty-three (32.3%) patients were diagnosed in the pre-rituximab era. One hundred fourteen (85.7%) patients had early relapse (ER) with 99 (74.4%) patients recurring within 2 years. Nineteen (14.3%) patients had LR (mean 7.9 years; median 7.3 years; up to 15.6 years). There were no significant differences in age at diagnosis, race, staging marrow status, or overall survival (OS) in ER versus LR patients. Extra-nodal presentation at diagnosis (89.5% vs. 65.8%; p = 0.04) and extra-nodal-only disease over time (73.7% vs. 48.2%; p = 0.04) were more common in LR cases. Both groups had similar rates of recurring at a different site from the original disease (79.3% vs. 89.5%; p = 0.30).

The average DFI was 7.1 years in the 5 LR patients selected for additional studies. IGH gene rearrangement analysis demonstrated identical D and R IGH clones in 3 cases (Table 1). Patient 4 showed 2 rearranged alleles at D with only 1 persisting at R. Patient 5 had lymphomas with unique IGH rearrangements. SNP microarray data demonstrated the presence of clonal heterogeneity in all but 1 sample (4 of 5 at D; 5 of 5 at R). Among the 4 patients with clonally related IGH gene rearrangements, there was only partial overlap in CNVs (approximately 40% on average) between the D and R lymphomas. The average CNVs was similar in the D and R samples (16.6 vs. 18.6 respectively; p = 0.75). Chromosomes 2, 3, 6, 9, and 17 were frequently altered, and CNVs involving the BCL-6, CDKN2A, TP53, and MYC loci were also commonly seen; but there was no systematic difference between the CNVs identified at D and R.

NGS showed a variety of mutations, but no consistent pattern of mutations acquired at R. There was a nonsense mutation in exon 2 of CDKN2A in the R sample in patient 1, and both D and R samples showed the same copy-neutral loss-of-heterozygosity of 9p encompassing the CDKN2A gene. In addition, missense mutations of TP53 were detected in patients 4 (only at R) and 5 (only at D).

CONCLUSIONS: This study demonstrates that LR-DLBCL is an uncommon phenomenon with most cases representing recurrence of the original disease. LR patients have similar OS as ER patients, and the only clinical factors segregating LR from ER are higher rates of extra-nodal presentation and extra-nodal-only sites of disease. Although most paired D and R cases share IGH clones, there is clear evidence of clonal heterogeneity with clonal evolution over time. This suggests that DLBCL may contain minor subclones not susceptible to chemotherapy, which persist subclinically acquiring additional mutations over time eventually generating clinically-evident relapse. In rare cases, the late “relapse” may occur as an unrelated lymphoma that arises spontaneously or secondary to the mutagenic effects of chemotherapy. The precise mechanism of this long latency is yet unclear, and requires further investigation.