Selection patterns of B-cell receptors and the natural history of follicular lymphoma
- 1University of Freiburg
- 2Leiden University
- 3
Journal
British Journal of Haematology
ISSN
0007-1048
1365-2141
Open Access
bronze
Volume
175
Start page
972
End page
975
Follicular lymphoma (FL) has a paradigmatic stepwise pathogenesis with the balanced translocation (14;18)(q32;q21) as the founding event (Kridel et al, 2012). Most healthy individuals harbour t(14;18)-positive premalignant FL-like cells (FLLC) (Roulland et al, 2011). FLLC have a memory phenotype and have often undergone immunoglobulin class switch recombination (CSR) on the translocated IGH allele. Frank FL develops only after repetitive entry of FLLC into germinal centres (GC) and multiple GC passages (Sungalee et al, 2014). Paradoxically, FLLC fail to switch the functional IGH allele, suggesting dependence on IgM signalling at this stage. In contrast, many diagnosed FL cases have undergone CSR of their B-cell receptor (BCR) (Sachen et al, 2012; Scherer et al, 2015). Therefore, class-switched FL cells have apparently lost the characteristic dependency of FLLC on IgM signalling for recruitment into the GC. In addition, expression of activation-induced deaminase (AID) seems to be dissociated from somatic hypermutation (SHM) and aberrant SHM in class-switched FL (Scherer et al, 2015). We explored whether class switching indicated a more aggressive FL subgroup with reduced dependence on the GC microenvironment by analysing clinical outcome parameters in a well-characterized prospective cohort of 69 FL cases (Scherer et al, 2015) (Table 1). Patients not suffering from pruritus or B symptoms, without disease progression of more than 50% in the preceding 6 months, without life-endangering lymphoma infiltration, without compression of a vital organ by a tumour mass, without bulky disease (thoracic tumour mass >7·5 cm; other locations >5 cm), and without relevant cytopenia due to lymphoma bone marrow infiltration (Hb <100 g/l, neutrophils <1·5 × 109/l or platelets <100 × 109/l) were initially managed by watchful waiting. The local ethics committee approved the study, and all patients provided written informed consent. Patient follow-up was updated to less than 6 months prior to outcome analysis. While relatively low patient numbers may have led to insufficient power to detect potential significant differences, FL isotype was not associated with progression-free survival (PFS) after cytoreductive therapy (n = 48) or with the interval from diagnosis to first treatment (TTFT) in patients (n = 44) managed by watchful waiting (Fig 1A, B). To find a better biological surrogate than isotype for dependency on IgM-mediated GC signals, we reconstructed a 'primordial' unmutated IGH VDJ sequence from individual full-length IGH VDJ transcript sequences (Scherer et al, 2015) by an online tool (http://clip.med.yale.edu/sel/germline.php) using Somatic Diversification Algorithm (Volpe et al, 2006; Hershberg et al, 2008). In 65 cases with successful reconstruction of the primordial IGH VDJ sequence, selection pressure on diversifying IGH VDJ sequences was assessed with the 'Focused Test' (http://clip.med.yale.edu/sel/index.php) (Hershberg et al, 2008). Evidence for antigen-driven affinity maturation is indicated by a positive p value; a negative p indicates BCR structure preservation despite on-going SHM. In general, BCR structure preservation appears to be prevalent in FL (Zuckerman et al, 2010). In our series, 51 cases (78%) were categorized as preserving their BCR structure, but 14 cases (22%) had evidence for antigen-driven affinity maturation ('antigen selection' category). Antigen selection cases were more likely to express IgM than BCR preservation cases (Table 1). Among patients managed with watchful waiting, antigen selection cases had a markedly longer TTFT (median = 103 months) than FL with BCR preservation (median = 18 months; Fig 1C). This difference remained significant when only patients with FL International Prognostic index (FLIPI) scores of 0–2 were analysed (P < 0·01). Multivariate analyses (Cox regression model) demonstrated a large additional impact of antigen selection compared to BCR preservation (Hazard Ratio = 0·15, P = 0·011) when adjusting for FLIPI score. Of note, knowledge of the BCR selection category did not influence the decision to initiate treatment because this information was not yet available for any patient at start of therapy. After cytoreductive therapy, PFS was similar in both BCR selection groups (Fig 1D). The BCR selection categories were not discernibly biased with respect to type of regimen given (Table 1). At a median follow-up of 101 (range: 7–243) months after diagnosis, the median overall survival was not reached (Fig 1E) with an estimated 10-year survival of 88·5%. In the antigen selection category, only one death occurred in the oldest patient (aged 76 years at diagnosis) of the entire cohort. All four documented histological transformations occurred in the BCR preservation group. Comparison of transformation-free survival between BCR categories suggested superior outcome for antigen-selected cases (Fig 1F), but this difference was not significant. Stability of BCR selection categories over time was suggested in three FL cases with repeat biopsies taken at least 12 months after the original biopsy: These cases originally expressed IgM, did not undergo CSR, and remained within the BCR selection category assignment. Most FL have acquired N-glycosylation sites in complementarity determining regions through SHM (McCann et al, 2006). Glycosylation of these sites renders FL BCR susceptible to stimulation by lectins from opportunistic bacteria (Schneider et al, 2015). However, no prognostic significance has as yet been assigned to the acquisition of such N-glycosylation motifs. In our cohort, the presence of these motifs (Table 1) was not associated with BCR selection category, isotype or any of the tested outcome parameters (data not shown). This study identifies the selection pattern of the clonal BCR as emerging under on-going SHM as an apparently stable immunobiological feature of FL that appears to influence natural tumour progression prior to therapy. This risk for symptomatic progression is independent from the clinical risk estimated by the FLIPI score. Assignment to the prognostic BCR selection category can be accomplished by a diagnostic test composed of unbiased amplification of the IGH VDJ transcript and bioinformatic analysis of multiple clonally related IGH VDJ sequences. Simple determination of the lymphoma isotype cannot substitute for the algorithmic analysis. With respect to the IgM- and GC re-entry dependent progression model from FLLC to overt FL (Roulland et al, 2011; Sungalee et al, 2014), our findings suggest that the majority of clinically overt FL has lost the peculiar dependence on IgM expression for GC re-entry driven stepwise early tumour progression. Such FL cells can undergo AID-mediated CSR but remain otherwise dependent on their given BCR, whose structure is preserved despite on-going AID attack. A minority of FL appears to become clinically manifest at an earlier developmental stage when still undergoing BCR selection, possibly driven by autoantigen stimulation within the GC microenvironment. These cases predominantly express IgM in accordance with the Roulland-Nadel model and follow clinically a more benign initial course that translates into a longer interval from diagnosis to primary therapy. Autoantigen recognition can indeed be detected for approximately one quarter of FL BCR (Sachen et al, 2012). Our results may stimulate further efforts to elucidate antigen-dependent BCR signalling in FL pathogenesis. The study was supported by Deutsche Krebshilfe through grant number 108935 and departmental funds from the participating Departments. M.A.N. acknowledges partial support from FONDECYT through grant No 11140542. HV was the principal investigator and takes primary responsibility for the paper. FS, MvdB, CB-L, MDvM. performed experiments. KZ and KM recruited patients. SMK and LdW performed statistical analyses and gave advice. FS, HV, MAN designed the study and wrote the paper. No conflict of interest.
Name
10_1111-bjh_13901.pdf
Type
Main Article
Size
227.85 KB
Format
Adobe PDF
Checksum
(MD5):6ceeb3f107bb109bfed7ebc8a9204b86