Discussion
We and others have described a more than fourfold increase in lymphoma for patients with SLE compared with the age-matched and sex-matched general population.3 The reason for this is not fully understood, but both immunological abnormalities (which may be driven by genetic factors) and drug exposures (primarily cyclophosphamide) may play a role.19–21
Sixty per cent of the DLBCL in patients with SLE were non-GCB. In recent analyses of a population-based registry study of DLBCL (n=348) using the Hans algorithm, 41% were non-GCB and 59% were GCB.22 Our SLE sample, with 60% of cases being non-GCB, certainly tends towards the upper limit of what has been described in the general population. As in the general population, the non-GCB subtype in SLE cases was associated with shorter survival. The rate and pattern of incidence of dual expressers of MYC and BCL2 in our cohort is comparable with the general population, where dual expressors tend to be non-GCB types and are associated, in the general population, with poorer survival.23 24 One of these dual expresser SLE cases with disease localised to the lymph node had short survival (3 months) whereas the other had extranodal involvement only and a much longer survival (85 months).15 18 It must be acknowledged that the interpretation of our results is limited by the sample size. However, taken together, these results may suggest that the effect of SLE on overall DLBCL incidence and survival is unlikely to be specific for the biology of either subtype, as discussed in the following paragraph.
From a pathophysiological perspective, SLE itself potentially shares mechanisms with both GCB and non-GCB types. Non-GCB DLBCLs rely on the activation of the NF-κB and JAK–STAT pathways, both directly affected in SLE through derangements of A20, tumour necrosis factor superfamily (TNFSF4), TNF-α, CD79, CARD11 and interleukin-1 receptor-associated kinase 1 (IRAK1) activity as well as epigenetic modifications.21 25–29 Interestingly, in primary Sjögren’s syndrome, an autoimmune rheumatic disease at high risk for mucosa-associated lymphoid tissue (MALT) lymphoma, most MALT cases have either germline polymorphisms of TNFAIP3, related to the A20 protein important in NF-κB activation, or somatic alterations of the gene within the lymphoma tissue.30 Moreover, polymorphisms of TNFAIP3 are common to rheumatoid arthritis (yet another condition linked with lymphoma) and Hodgkin’s lymphoma.31 In previous genome-wide association analyses, our group was unable to confirm a strong relationship with the lupus-related TNFAIP3 single-nucleotide polymorphism (SNP) rs7749323 specifically for DLBCL, but this may be a sample size issue. In those analyses, the rs2205960 SNP, related to TNFSF4, was associated with an OR per risk allele of 1.07, 95% CI 1.00 to 1.16, p value 0.0549.32 The OR for the SLE interferon regulatory factor risk allele rs12537284 (chromosome 7q32, gene) was 1.08, 95% CI 0.99 to 1.18, p value 0.0765.
The STAT4 lupus risk SNP rs7582694 meanwhile was not clearly linked to DLBCL. Our interpretation is that TNFAIP3, TNFSF4 and possibly interferon pathways are of high interest as potential mediators of the risk of DLBC (particularly non-GCB type) in SLE. However, it will be interesting to see if emerging SLE treatments based on JAK kinase inhibition could ultimately modulate the risk of DLBCL in lupus populations.33
On the other hand, DLBCLs of the GCB type are often defined by PI3K/AKT/mTOR pathway which may be hyperactivated in SLE secondary to defective PTEN expression.5 6 34 Other evidence showed that cyclophosphamide, a medication used to treat lupus, is partly responsible for the increased incidence of lymphomas in patients with SLE, but it is unclear if it would favour development of a particular molecular subtype.35 Rapamycin and other mTOR inhibitors have the potential to block this central pathway in both autoimmune diseases like SLE and also various forms of cancer. The advent of this new approach in SLE represents another way that novel drug development could potentially modify some of the altered cancer risk in lupus.
Just over half of our DLBCL cases were diagnosed at a SLE duration beyond 10 years. Extranodal involvement was more common in patients with shorter SLE duration, and those cases with only extranodal involvement had the best survival. Those with both nodal and extranodal involvement had the worst survival, with nodal-only cases showing intermediate results. This is somewhat different than what is seen from population-based studies where extranodal involvement is typically a marker for worse survival.36 It is unclear why extranodal involvement of DLBCL (despite being also associated with BCL2 protein expression, itself a marker of poor prognosis22 37) would be associated with better survival, but the results may be driven by the association with shorter SLE duration (although our previous analyses suggested higher all-cause standardised mortality ratios in low-duration SLE). The results may alternatively reflect a type of detection bias (if, eg, extranodal DLBCL in SLE is picked up earlier than nodal-only disease). However, extranodal DLBCL presenting symptoms are often more subtle than nodal DLBCL, where systemic B-symptoms are common,38 often triggering investigation.
In general, chronic inflammation is associated with lymphoproliferative disorders and could establish an environment fertile to the development of DLBCL in both nodal and extranodal sites.39 We note that the ratio between nodal and extranodal involvement does not seem to be different between patients with SLE and the general population.
Summary
Sixty per cent of the DLBCL in patients with SLE were non-GCB. We believe that TNFAIP3, TNFSF4 and possibly interferon pathways are of high interest as potential mediators of the risk of DLBC (particularly non-GCB type) in SLE. The nodal and extranodal distribution was similar between patients with SLE and the general population, but extranodal disease occurred more often in patients with short SLE duration and was associated with longer overall survival. Although sample size limits the interpretation of our results, our findings suggest that the immunological alterations in patients with SLE influence the tumour biology of DLBCL. More research studying cancer in SLE will be key to understanding the complex interplay between cancer and the immune system, especially as emerging lupus treatments could have important effects on related pathways.