Analysis of prognostic factors in diffuse large B-cell lymphoma associated with rheumatic diseases

Introduction

Diffuse large B-cell lymphoma (DLBCL) is an aggressive non-Hodgkin's lymphoma with a median survival of less than 1 year in untreated patients.1 The relative risk of developing DLBCL varies between the different rheumatic diseases (RDs) and has been reported to be about 1.8 times higher in patients with rheumatoid arthritis (RA), 2 times higher in patients with systemic sclerosis (SSc), 6.2 times higher in patients with systemic lupus erythematosus (SLE) and 11 times higher in patients with primary Sjogren’s syndrome (pSS) than that in the general population.2 3

Chronic auto-antigen stimulation and inflammation, defining features of RDs, represent the major drivers of specific B-cell proliferation and the increase in frequency of their transformation that may promote lymphoma development.4 5 It is possible that RD-associated DLBCLs are a distinct subset within the category of ‘DLBCL’, exhibiting characteristic biological features and clinical behaviour. However, information regarding RD-associated DLBCL is limited.

The main purpose of this study was to comprehensively characterise RD-associated DLBCL and analyse its prognostic factors.

Methods

Results

We identified 27 patients who had an RD coexisting with DLBCL. Baseline clinical characteristics of the patients are shown in table 1. Twenty patients had pSS, three had SLE, two had RA and two had SSc. Secondary Sjogren’s syndrome was found in four patients: two with SSc, one with RA and one with SLE. The median age of the patients in our cohort at the time of DLBCL diagnosis was 59 years (range: 30–83 years). The female-to-male ratio was 26:1. The median time from the onset of RD symptoms to DLBCL diagnosis was 19 years (range: 0–38 years). Based on the original International Prognostic Index (IPI),8 11 (41%) patients were assigned to low and low-intermediate risk groups, while 16 (59%) patients were assigned to intermediate-high and high-risk groups. Bulk disease was detected in 6 (29%) of the 21 examined patients.

The IHC results of patients with RD-associated DLBCL are shown in table 2. Based on expression patterns, according to Hans’ algorithm, we classified 4 cases of DLBCL into the GCB subtype and 16 cases into the non-GCB subtype. EBV was detected in 2 (10%) of the 21 cases studied. The c-MYC/8q24 gene rearrangement was not found in any of the 19 examined cases.

According to the 2016 WHO classification of tumours of haematopoietic and lymphoid tissues, DLBCL cases in our cohort were classified as follows: DLBCL, not otherwise specified (17 cases); EBV-positive DLBCL, not otherwise specified (two cases) and T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL) (two cases). Six DLBCL cases could not be reclassified because of missing data.

After the lymphoma diagnosis, the median OS was 10 months (range: 0–238 months) and the 5-year OS rate was 46% (figure 1).

Request permission

Discussion

Autoimmune diseases (AIDs) are a heterogeneous group of more than 80 separate conditions. AIDs can be categorised widely as being mediated mainly by B-cell or T-cell responses, recognising some overlap.9 Most of the patients in our study had autoimmune conditions mediated by B-cell responses. A large pooled analysis from the International Lymphoma Epidemiology Consortium has shown that AIDs classified as primarily mediated by B-cell responses are associated with an increased risk of developing DLBCL.10

However, the diagnostic category of ‘DLBCL’ is heterogeneous in terms of genetics, morphology, virus positivity, primary localisation, biological behaviour and prognosis. The 2016 WHO classification of tumours of haematopoietic and lymphoid tissues recognises several distinct entities within this category characterised by unique clinical and pathological features, including THRLBCL, EBV-positive DLBCL, not otherwise specified, primary mediastinal (thymic) large B-cell lymphoma, intravascular large B-cell lymphoma, and others.7 Cases of DLBCLs that do not fulfil the criteria for any of these specific entities are referred to as DLBCL, not otherwise specified (formerly referred to simply as DLBCL). Most cases in our cohort of RD-associated DLBCL were DLBCL, not otherwise specified, consistent with the predominance of this variant in the diagnostic category of ‘DLBCL’ in the general population.

The mean age in our study corresponded to the mean age at the time of DLBCL diagnosis in the general population.11 Although DLBCL is slightly more common in men than in women in the general population,11 our cohort of patients with RD-associated DLBCL had a significant female predominance, which likely reflects the greater incidence of RDs identified in women. However, it cannot be ruled out that there may be sex differences in the risk of developing DLBCL in patients with RDs.

Klein et al hypothesised that continuing disease activity and immune stimulation were the most significant factors in the development of DLBCL in patients with RA.12 Our findings support this hypothesis because in all of our cases, DLBCL was diagnosed after the diagnosis of RD, and the average time from the onset of RD symptoms to the diagnosis of DLBCL was 19 years.

Prognostic factors predicting poor prognosis in DLBCL include high IPI, bulk disease and c-MYC/8q24 gene rearrangements.13 The addition of rituximab (R) to cyclophosphamide, adriamycin, vincristine, and prednisone (CHOP) or CHOP-like chemotherapy dramatically improved the outcome in patients with DLBCL.14 15 However, IPI based on age at lymphoma diagnosis, serum LDH concentration, ECOG performance status, Ann Arbor stage disease, and extent of extranodal involvement remain important prognostic factors in patients with DLBCL treated with rituximab and chemotherapy.16 17 In our cohort, the percentage of patients assigned to the intermediate-high and high IPI risk groups was higher than that in the general population (59% vs 43%).11 Since the IPI considers performance status, determining whether a poor performance state is caused by RD or DLBCL in RD-associated DLBCL cases is challenging. Perhaps the higher percentage of patients in the intermediate-high and high-risk IPI groups in our cohort can be explained by a poor performance status attributed to RD, but not to DLBCL. The incidence of bulk disease in our cohort was comparable with that in the general population of patients with DLBCL.1 IPI and bulk disease are based solely on clinical factors that do not reflect the pathobiology of DLBCL.

Integration of the molecular features of DLBCL allows for further accurate prediction of disease outcome. Rearrangements in the c-MYC/8q24 gene were detected in 5%–15% of DLBCL cases in the general population.18 19 The presence of c-MYC/8q24 gene translocation to immunoglobulin partner genes is associated with unfavourable prognosis following R-CHOP treatment.20 To the best of our knowledge, there have been no studies on c-MYC/8q24 gene rearrangement in RD-associated DLBCL cases. Although the FISH study revealed one to two additional signals from the c-MYC/8q24 gene in six cases, none of the 19 examined cases in our cohort showed c-MYC/8q24 gene translocation.

Patients with compromised immune systems are more likely to have EBV-positive DLBCL than sporadic cases.21 22 However, the incidence of EBV-positive DLBCL varies significantly among patients with RDs (table 3).23–27 Since all studies in the analysed literature used in situ hybridisation for the detection of EBV in tumour tissues, such variability could be explained by the different cut-off scores of EBER-positive cells used to define DLBCL as EBV positive. A meta-analysis of 13 qualified studies showed that EBV-positive DLBCL had significantly worse OS and progression-free survival.28 In our cohort, only two (10%) cases were EBV positive, and in both cases the patients died of lymphoma progression after 1 month and 21 months from the diagnosis of DLBCL, respectively.

Although the prognostic value of DLBCL typing based on IHC algorithms has been inconsistent in patients treated with rituximab in addition to chemotherapy, the latest 2016 WHO classification of tumours of haematopoietic and lymphoid tissues recommend DLBCL typing for all cases.7 In the pre-R-CHOP era, patients with DLBCL with the GCB subtype had better prognosis than those with the non-GCB subtype.17 29 According to several studies, the addition of rituximab to chemotherapy has resulted in the nullification of the prognostic value of DLBCL typing based on IHC algorithms.17 30 31 In contrast, other studies have shown that Hans’ algorithm-based DLBCL typing retains predictive value in the rituximab era.32 33 Hans’ algorithm, as well as other IHC algorithms, shows a significant predominance of the GCB subtype of DLBCL in the general population.6 34 However, our findings, as well as those of other studies, have shown a predominance of the non-GCB subtype in RD-associated DLBCL (table 3).23–26 35

The role of RDs in DLBCL prognosis is still uncertain. The 5-year survival and median OS of patients with RD-associated DLBCL in our study were significantly lower than those of patients with DLBCL in the general population: 46% (our study) vs 60%–70%36 and 10 months (our study) vs 124 months37, respectively. Our findings are supported in part by other studies, although the results of a small number of studies published to date are not consistent. Kleinstern et al reported markedly shortened relapse-free survival and OS in eight patients with DLBCL in the presence of B‐cell-mediated AIDs compared with patients without AIDs.38 Similar results were obtained by Mörth et al in an analysis of a cohort of 39 patients with DLBCL and primarily B‐cell-mediated AIDs.39

In contrast, the study by Shih et al reported comparable OS in patients with non-Hodgkin’s lymphoma with and without pre-existing AIDs.40 However, this study reported outcomes for the entire AID group, and DLBCL represented only 18 of 34 cases of non-Hodgkin’s lymphoma. Results from the Surveillance, Epidemiology, and End Results database of 5926 elderly patients with DLBCL, of whom 270 had B-cell-mediated AIDs, showed no significant difference in OS in patients with B-cell-mediated AIDs compared with patients without a history of these diseases.4 However, this study found a tendency toward poor lymphoma-related survival in patients with DLBCL and SLE. A study using the Mayo Clinic database also showed no negative effect of autoimmune conditions mediated by B-cell responses on the prognosis of DLBCL, but there was also a trend toward inferior OS for DLBCL.9

The short OS in our cohort may be owing to several reasons. First, most patients in our cohort had the non-GCB subtype of DLBCL, and many of them received therapy in the pre-R-CHOP era. In a large Swedish cohort study that examined 22 patients with RA with DLBCL in the pre-R-CHOP era, the median OS was only 6 months.41 Second, an analysis of mortality in our cohort showed that most deaths (10 of 15 cases) occurred within the first few months after DLBCL diagnosis and were attributed to complications from chemotherapy. Our findings are consistent with those in the study of Mörth et al, which showed that patients with DLBCL and AIDs may be more prone to neutropenic fever than patients without concomitant AIDs, and that patients with neutropenic fever after their first course of treatment had poor OS.39 In contrast to our study findings, Mikuls et al showed that patients with RA and non-Hodgkin’s lymphoma had a lower risk of death as a result of lymphoma or its treatment, but were more than twice as likely to die from comorbid conditions than the non-RA lymphoma controls.42 However, in this study, DLBCL constituted only 43% of cases. Only two patients in our cohort had RA, and in both cases there was prolonged complete remission of the lymphoma.

There is a lack of knowledge in the literature regarding the biology of DLBCL arising in patients with RDs. Our findings agree with those of other researchers and suggest that DLBCL in the presence of RDs, in contrast to DLBCL in the general population, is more likely to be of the non-GCB subtype. Perhaps the poorer prognosis of patients with RD-associated DLBCL, established in some studies, is not only related solely to the biological features of DLBCL, but also to the fact that patients with long-standing severe RDs are predisposed to increased mortality and decreased tolerance to lymphoma treatment. Because the number of cases included in this study is limited, larger studies addressing this issue are needed to confirm this assumption.