Malignant lymphoma in granulomatosis with polyangiitis: subtypes, clinical characteristics and prognosis

Introduction

Several autoimmune and inflammatory conditions, such as rheumatoid arthritis (RA) and primary Sjögrens’s syndrome (pSS), have repeatedly been linked to an increased risk of malignant lymphoma [1,2]. Certain inflammatory conditions are also associated with the development of specific lymphoma subtypes such as mucosa-associated lymphoid tissue (MALT) lymphoma in pSS and diffuse large B-cell lymphoma (DLBCL) in RA. The underlying mechanisms behind this association remain unclear. The highly increased risk of developing MALT lymphoma of the parotid gland in pSS indicates that local inflammatory processes can promote lymphoma development at the site of chronic inflammation [3]. In RA, an association between disease severity and risk of lymphoma has been shown.

Granulomatosis with polyangiitis (GPA), formerly Wegener’s granulomatosis, is a systemic small vessel vasculitis associated with the presence of anti-neutrophil cytoplasmic antibodies (ANCA) and characterized by granulomatous inflammation and necrotizing vasculitis of the airways and kidneys but possibly affecting any organ system. An increased risk of lymphoma in patients with GPA has been reported in several epidemiological studies [4,5]. However, very little is known about risk factors for lymphoma development in this group, possible relation to disease severity, treatment and lymphoma subtypes or the prognosis for the lymphomas. This is the first published study on GPA and lymphoma, giving detailed information on the GPA characteristics and possible risk factors for lymphoma and also lymphoma subtypes treatment and survival.

Material and methods

Identification of cases

We used the Swedish Patient Register and the International Classification of Disease (ICD) coding system to identify all individuals with a diagnosis of Wegener’s granulomatosis/GPA (ICD8: 446.2, ICD9: 446.E, ICD10: M31.3) registered as the main or contributory diagnosis between 1964 and 2012 (n = 3224). The register contains information on all in-patient care since 1964 with nation-wide coverage since 1987 and non-primary outpatient care since 2001. Through linkage to the Swedish Cancer Register, which has completeness of around 98% [6], we obtained information on all lymphoproliferative malignancies (ICD7:200–202) registered after the first discharge listing GPA between June 1973 and July 2011 (n = 92). The medical records were collected and the GPA diagnosis was evaluated using the European Medicines Agency (EMA) vasculitis classification algorithm [7] that incorporates the American College of Rheumatology (ACR) criteria [8], the Chapel Hill Consensus Conference [9], as well as surrogate markers. To confirm the lymphoma diagnosis, lymphoma biopsies were collected and re-classified according to the WHO classification [10] by one experienced hematopathologist (CS). Of the initial 92 patients, 69 were excluded from the study. Of these, 66 patients did not fulfill GPA criteria and 3 patients had been diagnosed with lymphoma before the onset of GPA.

Clinical data and analyses of lymphoma tissues

Detailed clinical data of the GPA disease and the lymphomas were collected from the medical records from the onset of GPA until death or 31 December 2015. The median follow-up time was 12 years (0.5–26). We retrospectively evaluated disease activity at GPA diagnosis using the Birmingham Vasculitis Activity Score (BVAS) version 3.0 [11] and the accumulated disease-related organ damage at the time of lymphoma diagnosis using the Vasculitis Damage Index (VDI) [12]. DLBCLs were classified as either germinal center (GC) type or non-germinal center (non-GC) subtype according to the algorithm of Hans et al. [13] by staining with CD10, Bcl-6 and IRF-4 antibodies. The presence of EBV in lymphoma tissue was analyzed using EBV-encoded RNA (EBER) in situ hybridization [14]. Overall survival from lymphoma diagnosis was calculated from the date of diagnosis according to the Cancer Register until 31 December 015 at the latest. Ethical approval was obtained from the Regional Ethical Review Board in Uppsala, Sweden.

Statistical analyses

Patients treated with CYC for GPA were divided into a high-dose CYC group (cumulated dose ≥25 g) and a low-dose CYC group (cumulated dose <25 g). The cutoff was based on the typical modern low-dose remission induction dose of CYC in GPA (<25 g) [15]. The 1- and 5-year survival rates after lymphoma diagnosis were compared between the groups using the Kaplan–Meier method, and differences were calculated using the log-rank test. Predictors for all-cause mortality were further evaluated by multivariate Cox regression, expressed as hazard ratios (HR) with 95% confidence intervals (CI). The model included cumulative CYC dose for GPA, sex, age at lymphoma, GPA disease duration at lymphoma diagnosis, BVAS at GPA diagnosis, VDI at lymphoma diagnosis, Ann Arbor staging [16] and Eastern Cooperative Oncology Group (ECOG) performance score. Statistical analyses were performed using GraphPad software (Prism, version 7, GraphPad Soft Inc., San Diego, CA, USA) and IBM SPSS Statistics Subscription software (IBM version 1.0.0.903, Armonk, NY, USA). p-Value <.05 was considered statistically significant.

Results

GPA disease characteristics and treatment

A total of 23 cases of malignant lymphoma developing after a diagnosis of GPA were identified. Table 1 shows characteristics of the study population and the treatment used for GPA. All except one patient, with localized upper and lower airway engagement, had generalized and severe GPA disease. The most common organ involvement was upper airways (n = 18) followed by pulmonary (n = 14) and renal involvement (n = 14). All GPA patients had been treated with corticosteroids and at least one other immunosuppressive drug (Table 1). CYC was used in 18 (78%) of the patients for remission induction in combination with steroids. Many (n = 11; 48%) were continuously given CYC as maintenance therapy for long periods. The median duration of CYC treatment was 28 months (range 1–170).

Table 1. Characteristics of 23 patients with granulomatosis with polyangiitis (GPA) complicated with lymphoma.

Study population	n = 23
Male, n (%)	17 (74)
Age at GPA diagnosis, years, median (range)	61 (39–83)
BVAS at GPA diagnosis, median (range)	18 (3–28)
ANCA positivity ever, n (%)	18 (78)
Treatments for GPA, ever
CYC,n (%)	18 (78)
CYC cumulative dose, median, grams (range)	42 (1.9–511)
Corticosteroids, n (%)	23 (100)
Other immunomodulating drugs, n (%)	17 (74)^a
Age at lymphoma diagnosis, years, median (range)	67 (47–89)
Time from GPA to lymphoma diagnosis, years, median (range)	12 (0–22)
VDI at lymphoma diagnosis, median (range)	3.5 (0–7)
Calendar year of lymphoma diagnosis
1970–1999, n (%)	6 (26)
2000–2008, n (%)	17 (74)
Site of lymphoma involvement
Extra-nodal disease, n (%)	18 (78)
Gastrointestinal tract/liver, n (%)	10 (43)
Bone marrow/blood, n (%)	7 (30)
Other, n (%)	5 (22)^b
Ann Arbor stage
Stage I–II, n (%)	5 (25)^c
Stage III–IV, n (%)	15 (75)^c
EBV-positive lymphoma tissue, n (%)	1 (6)^d

BVAS: Birmingham Vasculitis Activity Score; ANCA: antineutrophil cytoplasmic antibodies; CYC: cyclophosphamide; VDI: Vasculitis Damage Index; EBV: Epstein–Barr virus.

^aMethotrexate, azathioprine, mycophenolate mofetil, cyclosporine, intravenous immunoglobulin. ^bLung/pleura, central nervous system, skin, ear, nose and throat, skeletal. ^cThree patients with chronic lymphocytic leukemia not included. ^dIn 18 cases, EBV-related RNA in situ hybridization (EBER) was performed on lymphoma tissue, in 5 cases, the amount of tissue was too scarce for this analysis.

Lymphoma characteristics

The characteristics of the lymphomas and the distribution of lymphoma subtypes are shown in Table 2. Briefly, B-cell lymphoma predominated (83%), the single most common subtype was DLBCL. The majority of the lymphomas involved extranodal sites at the time of diagnosis, 75% were in Ann Arbor stage III–IV. Engagement of the upper airways was present in one case, a DLBCL in the hard palate. Apart from this, there was no other occurrence of lymphoma in for GPA typical sites of chronic inflammation.

Table 2. Lymphoma subtypes and outcome in 23 patients with granulomatosis with polyangiitis (GPA) complicated by lymphoma^a.

Lymphoma subtype, WHO classification	Number (%)	CYC ≥25 g, n (%)	CYC <25 g, n (%)
All patients	23 (100)	14 (61)	9 (39)
B-cell lymphomas	19 (83)	11 (79)^b	8 (89)^c
Diffuse large B-cell lymphoma	6 (26)	2 (14)^b	4 (44)^c
GC type	3 (50)^d	2 (33)^d	1 (17)^d
Non-GC type	3 (50)^d	0	3 (50)^d
Chronic lymphocytic leukemia	3 (13)	1 (7)^b	2 (22)^c
Mantle cell lymphoma	3 (13)	1 (7)^b	2 (22)^c
High-grade B-cell lymphoma, unclassifiable	4 (17)	4 (29)^b	0
Low-grade B-cell lymphoma, unclassifiable	1 (4)	1 (7)^b	0
B-cell lymphoma, unclassifiable	2 (9)	2 (14)^b	0
T-cell lymphomas	2 (9)	1 (7)^b	1 (11)^c
Peripheral T-cell lymphoma	1 (4)	0	1 (11)^c
High-grade T-cell lymphoma, unclassifiable	1 (4)	1 (7)^b	0
Others	2 (9)	2 (14)^b	0
Non-Hodgkin lymphoma, unclassifiable	1 (4)	1 (7)^b	0
High-grade malignant lymphoma, unclassifiable	1 (4)	1 (7)^b	0
Outcome
Overall survival after lymphoma diagnosis, months, median (range)	4 (1–66)^d	2 (0–35)^d	58 (2–84)^d
1-year overall survival, n (%)	9 (39)	3 (21)	6 (67)
5-year overall survival, n (%)	4 (17)	0	4 (44)

CYC: cyclophosphamide; GC: germinal center.

^aData are presented for all patients and patients divided into two groups based on the cumulative dose of cyclophosphamide given as GPA treatment. ^bPercentage of all lymphoma subtypes in the high-dose CYC group. ^cPercentage of all lymphoma subtypes in the low-dose CYC group. ^dPercentage of all diffuse large B-cell lymphoma; Four patients alive at end of follow-up excluded.

Epstein–Barr virus in the lymphoma tissue was present in only one of the 18 examined lymphomas, a high-grade T-cell lymphoma in an 86-year-old woman who died two months after the lymphoma diagnosis. She had been diagnosed with GPA 22 years before the lymphoma and had been exposed to a high cumulative CYC dose (130 g).

Lymphoma treatment and survival

Nineteen (83%) of the 23 patients died during the study period with a median survival time of 4 months (0–66). The cause of death was in most cases (n = 14; 74%) directly related to the lymphoma or its treatment. Seventeen of the patients (74%) received lymphoma treatment with curative intent. The most common treatment was cyclophosphamide, hydroxydaunorubicin, oncovin and prednisone (CHOP)-based regimens, (n = 13) in seven cases combined with rituximab (RTX) (R-CHOP).

Cyclophosphamide treatment for GPA and survival after lymphoma

Overall survival after lymphoma diagnosis was significantly shorter for patients exposed to high cumulative CYC doses (≥25 g) for GPA. The low-dose CYC group had a 1-year survival of 67% compared to 21% in the high-dose group (p = .020) and a 5-year survival of 44% while no patients were alive after 5 years in the high-dose group (p = .0008) (Supplementary Figure S1). In a multivariate model including factors of relevance for prognosis, exposure to high cumulative CYC doses (≥25 g) during the course of GPA was the strongest predictor for all-cause mortality after lymphoma diagnosis (adjusted HR 10.0; 95% CI: 1.5–68.6; p = .019) (Supplementary Table S1).

Discussion

This population-based study is the first to report detailed information on the lymphomas developing in GPA patients and information on the GPA-lymphoma patients’ phenotypes and treatments. The patients who developed lymphoma had severe GPA treated with potent immunosuppressive drugs during prolonged periods of time. Disease severity [17] and the cumulative burden of inflammatory activity [18] have been linked to an increased risk of lymphoma in patients with certain chronic inflammatory diseases. The contribution of immunosuppressive drugs to the development of malignancies, including lymphomas, in rheumatic diseases, has been debated and most extensively studied in patients with RA, so far without convincing evidence for increased risks linked to the drugs commonly used in RA [19]. Studies of risk factors for lymphoma development in patients with GPA are lacking, although an association between CYC treatment in GPA and other malignancies is well documented, most evident for non-melanoma skin cancer and urinary tract malignancy [20,21].

We found no clear association between GPA and any specific subtype of lymphoma. The distribution of lymphoma subtypes was comparable to that of the general population, the proportion of DLBCL, the most common subtype, was similar to that reported in the Swedish Lymphoma Register for the general population, with an equal proportion of the GC and non-GC subtypes. Based on a limited number of cases, we found MCL in 13% of the GPA patients compared to 4% reported in the Swedish Lymphoma Register. In a previous study, no increased risk for MCL was found in patients with various autoimmune diseases, GPA not included [22]. In one case report, an association between high cumulative CYC dose in a patient with GPA and the development of a MCL was suggested [23]. However, no such association between CYC doses and MCL could be confirmed in this study.

We were interested to investigate whether typical sites of chronic inflammation in GPA, such as the upper airways, were at particular risk for lymphoma development. In this study, only one lymphoma was situated in the upper airways, a DLBCL in the hard palate. Apart from this, we found no lymphomas in organs engaged by GPA.

Epstein-Barr virus has been shown to be highly involved in lymphomagenesis in conditions characterized by severe immune dysfunction such as HIV [24] or with the use of immunosuppression in organ transplanted patients [25] but also, to a lesser extent, during immunosuppressive treatment for autoimmune diseases [26]. In RA, the reports of EBV-positive lymphomas that regress after the withdrawal of immunosuppressive treatment with methotrexate support a role of EBV in some methotrexate-treated lymphoma patients [27]. In this study, we found no support for EBV as a significant contributor to the development of lymphoma in GPA patients despite the high degree of immunosuppression, bearing in mind that 5 of 23 examined lymphoma tissues could not be analyzed for EBV.

Survival after the diagnosis of lymphoma was markedly poor, notably though mainly in the group of patients exposed to high cumulative CYC doses. We found an association between cumulative CYC doses ≥25 g and worse prognosis which remained after evaluating the impact on all-cause mortality of several prognostic variables related to both GPA disease and lymphoma through multivariate analysis.

Possible explanations for an association between high-dose CYC and poor survival could be a negative effect of CYC on immunologic surveillance of malignant cells, suppression of lymphoma signs delaying lymphoma diagnosis or CYC induced bone marrow vulnerability increasing the risk of complications during chemotherapy. Many patients died during chemotherapy, often due to infectious complications and in several cases during periods of leucopenia, which could support this theory.

The strengths of the present study include the population-based setting, the verification of the GPA diagnosis, the detailed information on the GPA disease and treatments and cumulative CYC dose for all patients. The verification of the hematological malignancies through a reexamination of lymphoma tissues and re-classification according to WHO classification of all the cases is a further strength. A limitation is the small study population, a consequence of the relative rarity of both GPA and lymphoma.

In conclusion, this study emphasizes the need for awareness of signs of lymphoma in the long-term follow-up especially of CYC-treated severely ill GPA patients, and the risk for poor prognosis of lymphoma in these patients, especially when previously exposed to high cumulative doses of CYC. In view of the changes of treatment strategies for GPA in recent years, aiming at lowering doses of CYC and including rituximab for both induction and maintenance treatment, we might in the future see a change for the positive in the risk of lymphoma development and the characteristics and outcome of the lymphomas in these patients.

Authors’ contributions

A.K., E.H., K.H., and E.B. designed the study; E.H., A.K., and K.H. collected data; C.B. and C.S. contributed with preparation and analysis of lymphoma tissue; A.K., E.H., K.H., E.B., and G.E. analyzed the data; E.H., A.K., and E.B. wrote the paper, and all authors critically revised the manuscript and approved the final version.

Acknowledgments

We would like to thank Professor Johan Askling, Assistant Professor Fredrik Granath and Assistant Professor Karin E. Smedby at the Clinical Epidemiology Unit, Department of Medicine Solna, Karolinska Institutet, Stockholm, Sweden, for valuable advice and help with register linkages and bioinformatics Pascal Pucholt for statistical assistance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by grants from Uppsala Regional Research Council, the Swedish Cancer Society, the Swedish Society of Rheumatology, King Gustaf V’s 80-year Foundation, and the Lions Cancer Research Foundation in Uppsala, Sweden.