Increased Risks of Lymphoma and Death Among Patients With Non–Hepatitis C Virus–Related Mixed Cryoglobulinemia

Background  Data on essential mixed cryoglobulinemia (MC) are scarce, and most date back to studies before 1989 (ie, before the discovery of hepatitis C virus [HCV] infection). Our objective was to describe the spectrum of MC in the era of HCV infection.

Methods  Retrospective study from a single university hospital's database of 1434 patients who tested positive for MC between January 1989 and December 2003.

Results  One hundred thirty-three patients (9%) with persistent MC without HCV were included in the study. Sixty-five of 133 patients who fulfilled the criteria for MC vasculitis were compared with 118 patients with HCV-related MC vasculitis. The patients without HCV had increased frequencies of renal involvement and B-cell non-Hodgkin lymphoma (NHL), lower gammaglobulin levels, and higher death rates. Twenty-three of the patients had B-cell NHL (primarily of the lymphoplasmocytic and marginal zone types), and 8 patients had Sjögren syndrome. In multivariate analysis, a cryoglobulin level higher than 0.6 g/L (odds ratio [OR], 1.44) and the presence of MC vasculitis (OR, 4.3) and hypogammaglobulinemia (OR, 6.7) were independently associated with B-cell NHL. After a mean follow-up of 49.4 months, 18 (14%) of 133 patients had died, primarily of sepsis. In multivariate analysis, age at diagnosis older than 60 years (OR, 1.06) and renal involvement (OR, 5.20) were independently associated with death.

Conclusion  Patients with non–HCV-related MC vasculitis have a poor outcome and have a 4-fold increased risk of developing B-cell NHL.

Cryoglobulinemia is defined as the presence of circulating immunoglobulins that precipitate with cold temperature and resolubilize with rewarming. After immunochemical typing, cryoglobulins are sorted according to the classification by Brouet et al¹: type I cryoglobulinemia comprises single monoclonal immunoglobulins, and types II and III are mixed cryoglobulinemias (MCs), with a monoclonal component in type II and only polyclonal immunoglobulins in type III. Type I cryoglobulins are usually associated with lymphoproliferative disorders, while MCs (type II or III) are associated with connective tissue diseases, lymphoproliferative disorders, and chronic infections. When MC is found in the absence of well-defined disease, the syndrome is designated as essential MC.

Mixed cryoglobulinemia is a systemic vasculitis characterized by the proliferation of B-cell clones producing pathogenic IgM with rheumatoid factor (RF) activity. The clinical manifestations range from MC syndrome (purpura, arthralgia, and asthenia) to more serious lesions with neurologic and renal involvement.² Since the discovery of hepatitis C virus (HCV) infection in 1989,³ it has become clear that HCV is associated with most cases of MC. Among patients with MC, 60% to 90% are infected with HCV.⁴

Data on essential MC are scarce, and most date back to studies before 1989 (ie, before the discovery of HCV infection), with most of the data about the therapy and outcome of non–HCV-related MC derived from small studies or from isolated case reports.^5-7 Because cryoglobulin-producing B cells are primarily monoclonal, MC can be viewed as a benign B-cell lymphoproliferative condition. Findings from many studies suggest a link between HCV-related MC and B-cell non-Hodgkin lymphoma (NHL).^8,9 No study has focused on the relationship between non–HCV-related MC and B-cell NHL, to our knowledge.

In this study, we analyzed the characteristics and outcomes of 133 patients with persistent MC without HCV at a single center (Hôpital La Pitié-Salpêtrière, Paris, France). Sixty-five of 133 patients who fulfilled the criteria for MC vasculitis were compared with 118 patients with HCV-related MC vasculitis. Logistic regression analysis was used to assess the variables associated with B-cell NHL and with death.

Between January 1989 and December 2003, 17 035 patients underwent testing for circulating cryoglobulins in the immunology department at Hôpital La Pitié-Salpêtrière (Figure). In 1434 patients, a cryoglobulin level higher than 0.05 g/L was detected on at least 2 occasions. Among them, 1301 (91%) were infected with HCV and were excluded from the study. We included the 133 remaining patients without HCV (9%) and with persistent circulating cryoglobulins for more than 6 months. For each patient, the following data were collected: sex, outcome, hemogram findings, presence of sicca syndrome, gastrointestinal tract involvement, immunological features, age at diagnosis of MC, dates of MC vasculitis and NHL diagnoses, general symptoms (fatigue or recent weight loss), neurologic involvement (polyneuropathy, multiple mononeuritis, or central nervous system), cutaneous involvement (Raynaud phenomenon, purpura, or distal ulcers), renal involvement (proteinuria, hematuria, or serum creatinine level >1.6 mg/dL [>140 μmol/L]), rheumatologic involvement (arthralgia, arthritis, or myalgia), and cause (infections, essential MC, connective tissue diseases, or hematologic disorders).

Among 133 patients with persistent MC, 65 patients had palpable purpura or histologically proven MC vasculitis (based on renal biopsy findings in 37 patients, skin biopsy findings in 24 patients, and nerve biopsy findings in 6 patients) (totals 67 because some patients underwent both skin and kidney or nerve biopsy). and fulfilled the criteria for MC vasculitis.² The characteristics of these 65 patients were compared with those of 118 HCV-infected patients who fulfilled the criteria for MC vasculitis.² All patient data were recorded at Hôpital La Pitié-Salpêtrière.

Cryoglobulins were measured and characterized by immunochemical analysis as previously described¹⁰; cryoglobulin levels had to be higher than 0.05 g/L on at least 2 occasions (during >6 months) for patients to be considered MC positive. Cryoglobulins were classified according to the classification by Brouet et al¹ as type II MC, which includes a monoclonal component, or as type III MC, which includes only polyclonal immunoglobulins. All patients with circulating MC underwent systematic testing for HCV using 2 third-generation immunoassays (Monolisa anti-HCV Plus; Sanofi Diagnostic Pasteur, Marne la Coquette, France; and Axsym HCV version 3.0; Abbott Laboratories, Les Ulis, France) and by serum HCV RNA testing (Amplicor HCV test; Roche Diagnostics, Neuilly/Seine, France). Patients with HCV were excluded. Antinuclear antibodies were determined by indirect immunofluorescence on HEP2 cells (ImmunoConcepts, Sacramento, Calif). Antiextractable nuclear antigen antibodies were assessed using enzyme-linked immunosorbent assay (ELISA) (Bindazyme, Saint Egreve, France). Anti–double-stranded DNA antibodies were detected by ELISA (Biomedical Diagnostics, Marne la Vallé, France), antineutrophil cytoplasmic antibodies by indirect immunofluorescence (Menarini; Inova Diagnostics, San Diego, Calif), and IgG and IgM anticardiolipin antibodies by ELISA (CARDIO-LISA, Biomedical Diagnostics). Rheumatoid factor, complement, and serum gammaglobulin levels were determined according to standard procedures.

Connective tissue diseases were diagnosed based on standard criteria for Sjögren syndrome (SS),¹¹ systemic lupus erythematosus,¹² rheumatoid arthritis,¹³ systemic sclerosis,¹⁴ antiphospholipid syndrome,¹⁵ Behçet disease,¹⁶ and mixed connective tissue disease.¹⁷ Hepatitis B virus surface antigen was analyzed by ELISA. Anti–HIV-1 p24 antibodies were analyzed by ELISA using HIV-1 p24 coating. For confirmation, Western blots were applied. Other infections were diagnosed using standard methods. The diagnosis of NHL was made on the basis of evidence of bone marrow, nodal, or extranodal lymphoproliferative disease with pathological features that were compatible with the World Health Organization classification of neoplastic diseases.¹⁸ Clinical staging of patients with lymphoma included physical examination, computed tomographic imaging of the thorax and abdomen, and biopsy specimens of bone marrow and malignant tissues. Other hematologic diseases were diagnosed using standard criteria. Finally, essential MC was considered when autoimmunity, infections, and hematologic diseases were not found.

Treatment-related data were recorded for patients with MC vasculitis or with B-cell NHL. The response to treatment was analyzed by comparing the variables at the initial evaluation and those at the end of follow-up. Among patients with MC vasculitis, the response to treatment was assessed by analyzing the evolution of the following signs: disappearance of serum cryoglobulins, skin involvement (absence of purpura), peripheral neuropathy (clinical or electrophysiological improvement at 2 successive examinations), and renal involvement (normalization of the serum creatinine level and disappearance of proteinuria or hematuria). A complete response of MC vasculitis was defined as an improvement in all baseline clinical manifestations and by the disappearance of serum cryoglobulins. A partial response of MC vasculitis was defined as an improvement in at least half of the baseline clinical manifestations, with persistence of serum cryoglobulins. All other patients were classified as nonresponders. Relapse was defined as the reappearance of clinical signs of MC vasculitis. Complete lymphoma remission was defined as the total disappearance of the lymphomatous mass and partial remission as a 75% reduction.

Quantitative variables were expressed as mean ± SD. Comparison between variables was assessed using Fisher exact test or χ² test. Continuous variables were dichotomized using median values (age at diagnosis and cryoglobulin level). Odds ratios (ORs) and 95% confidence intervals (CIs) were computed. Multivariate models were evaluated using multiple logistic regression analysis to assess the variables associated with B-cell NHL and with death. All variables with P<.10 in the univariate analysis were initially included. Variables were selected using a stepwise forward procedure based on the Akaike information criterion. In model checking, we examined potential interactions and collinearity, and goodness of fit was evaluated using the method proposed by le Cessie and van Houwelingen.¹⁹ All tests were 2-sided, with statistical significance set at P<.05. Analyses were carried out using SAS version 8 (SAS Institute Inc, Cary, NC).

The characteristics of 133 patients without HCV with persistent MC are given in Table 1. The mean age at diagnosis was 59.5 ± 16.6 years (age range, 20-97 years), and 87 patients (65%) were female. Clinical manifestations included asthenia in 77 patients (58%), arthralgia in 54 patients (41%), renal involvement in 53 patients (40%), purpura in 38 patients (29%), and peripheral neuropathy in 27 patients (20%). The mean cryoglobulin level was 0.8 ± 1.3 g/L, with 96 patients (72%) having type II MC. The mean gammaglobulin level was 10.3 ± 5.0 g/L, with 33 patients (25%) having hypogammaglobulinemia. Immunological disorders included low C4 complement level in 69 patients (52%), RF activity in 62 patients (47%), antinuclear antibodies in 57 patients (43%), low C3 complement level in 50 patients (38%), anticardiolipin in 28 patients (21%), extractable nuclear antigen antibodies in 21 patients (16%), DNA antibodies in 18 patients (14%), and antineutrophil cytoplasmic antibodies in 3 patients (2%).

Table 1 summarizes the characteristics of patients with type II and type III MC. Ninety-six patients (72%) had type II MC, comprising monoclonal IgMκ (75%), IgMλ (12%), IgGκ (8%), and IgGλ (5%). Compared with patients with type III MC, patients with type II MC were older and had more MC vasculitis and hematologic disease. The causative factors associated with MC comprised the following 3 disease groups: (1) Connective tissue diseases were present in 18 patients with systemic lupus erythematosus, in 15 patients with SS, in 5 patients with antiphospholipid syndrome, in 2 patients with rheumatoid arthritis, and in 1 patient each with mixed connective tissue disease, Behçet disease, sarcoidosis, systemic sclerosis, and Still disease. (2) Hematologic diseases were present in 18 patients with lymphoplasmocytic lymphoma, in 8 patients with marginal zone lymphoma, in 3 patients with multiple myeloma, in 2 patients with chronic lymphoid leukemia, and in 1 patient each with Castleman disease, angioimmunoblastic lymphadenopathy, cold agglutinin disease, and AL amyloidosis. (3) Infectious diseases were present in 5 patients with hepatitis B virus; in 3 patients with human immunodeficiency virus; in 2 patients each with tuberculosis, Streptococcus, and Plasmodium falciparum, and in 1 patient each with cytomegalovirus, Coxiella burnetii, and amebiasis. Thirty-six patients (27%) without underlying disease were considered as having essential MC.

Among 133 patients with MC without HCV, 29 patients (22%) had B-cell NHL. The characteristics and outcomes of patients with B-cell NHL are summarized in Table 2. Ninety-six patients (72%) had hypogammaglobulinemia. Twenty-three patients (79%) had MC vasculitis. Of 8 patients with marginal zone lymphoma, 4 patients with extranodal involvement had gastric mucosa–associated lymphoid tissue. Eighteen patients with B-cell NHL underwent chemotherapy (6 patients received chloraminophene, 5 patients received cyclophosphamide, 5 patients received CHOP [cyclophosphamide, doxorubicin, vincristine sulfate, and prednisone], and 2 patients received anti-CD20 monoclonal antibody). The response to chemotherapy was complete remission in 1 patient (6%) and partial remission in 14 patients (78%). In multivariate analysis (Table 3), a cryoglobulin level higher than 0.6 g/L (OR, 1.44; 95% CI, 1.02-2.00), the presence of MC vasculitis (OR, 4.3; 95% CI, 1.5-11.8), and the presence of hypogammaglobulinemia (OR, 6.7; 95% CI, 2.1-21.0) were independently associated with B-cell NHL.

After a mean follow-up of 49.4 ± 41.6 months, 18 deaths (14%) were noted from sepsis (n = 11), hemorrhage (n = 3), hemopathy (n = 2), and cardiovascular disease (n = 2). In multivariate analysis (Table 4), age at diagnosis older than 60 years (OR, 1.06; 95% CI, 1.02-1.11) and renal involvement (OR, 5.20; 95% CI, 1.20-22.53) were independently associated with death.

Sixty-five patients had MC vasculitis (mean age at diagnosis, 63.7 ± 15.5 years), 45 (69%) of whom were female (Table 5). The initial symptom of non–HCV-related MC vasculitis was purpura in 29 patients (45%), renal involvement in 21 patients (32%), arthralgia or arthritis in 18 patients (28%), and peripheral neuropathy in 5 patients (8%) (data not shown [some patients had more than 1 initial symptom]). Clinical manifestations included asthenia in 48 patients (74%), renal involvement in 41 patients (63%), purpura in 38 patients (59%), arthralgia or arthritis in 29 patients (45%), peripheral neuropathy in 20 patients (31%), Raynaud phenomenon in 17 patients (26%), sicca syndrome in 14 patients (22%), distal ulcers in 8 patients (12%), central nervous system involvement in 5 patients (8%), and gastrointestinal tract involvement in 5 patients (8%). Type II MC was present in 55 patients (85%). The immunological features were low C4 complement level in 49 patients (75%), RF activity in 43 patients (66%), low C3 complement level in 34 patients (52%), antinuclear antibodies in 25 patients (39%), anti-SSA in 11 patients (17%), and anticardiolipin antibodies in 5 patients (8%). The etiologic factors associated with non–HCV-related MC vasculitis were hematologic diseases (23 patients with B-cell NHL and 1 patient with multiple myeloma), connective tissue diseases (8 patients with SS and 1 patient each with systemic lupus erythematosus and Behçet disease), infectious diseases (1 patient each with human immunodeficiency virus, P falciparum, amebiasis, Streptococcus, and C burnetii), and essential MC (26 patients). The treatment of non–HCV-related MC vasculitis consisted of corticosteroid use in 61 patients (94%) (prednisone, 0.5 to 1 mg/kg of body weight per day, then tapered during several months), augmented in severe cases with immunosuppressive agents in 36 patients (cyclophosphamide or azathioprine sodium) and with plasmapheresis in 6 patients. Antiinfectious agents were used in 6 patients, alone (n = 4) or in combination with immunosuppressants (n = 2). A complete response of non–HCV-related MC vasculitis was observed in 16 patients (25%), 34 patients (52%) had a partial response, and 15 patients (23%) were nonresponders. A relapse occurred in 34 patients (52%) after a median of 13 months (range, 6-80 months). There were 13 deaths, 9 of which were from sepsis (6 from bacterial septicemia and 1 each from pneumococcal pneumonia, aspergillosis, and pneumocystosis). All 13 patients who died had renal involvement, with renal insufficiency in 10 patients, and none had a complete response to therapy.

Table 5 summarizes the characteristics of 118 patients with HCV-related MC vasculitis, as well as the characteristics of 65 patients with non–HCV-related MC and 26 patients with essential MC. The patients without HCV had increased frequencies of renal involvement and of low C4 complement level, a lower gammaglobulin level, and a higher death rate.

Our study covers the complete spectrum of non–HCV-related MC, as the data represent different medical specialties at a university hospital. Between 1989 and 2003, 1434 patients tested positive for MC (ie, they had cryoglobulin levels >0.05 g/L on at least 2 occasions). Of these, 133 patients (9%) without HCV were included in this study. The causative factors associated with non–HCV-related persistent MC were connective tissue diseases (45 patients [34%]), hematologic diseases (35 patients [26%]), and infectious diseases (17 patients [13%]). Type II MC accounted for 72% (96/133) of our population without HCV and was closely associated with MC vasculitis and hematologic diseases. This is consistent with the data reported by Monti et al²⁰ in a retrospective multicenter study, although in their study population of patients with cryoglobulinemia HCV infection was investigated in only 24% of patients.

In our study, 29 patients (22%) had B-cell NHL, primarily lymphoplasmocytic lymphoma and marginal zone lymphoma histologic subtypes. These are prevalent types of B-cell NHL among HCV-infected individuals. However, individuals with HCV usually have primary extranodal localization.²¹ Data regarding the variables associated with NHL in patients with MC are scarce. In chronic HCV infection, the presence of cryoglobulinemia is an independent risk factor for lymphomas and may be considered an early marker of HCV-associated lymphoproliferation.²² In the present study, B-cell NHL was independently associated with a cryoglobulin level higher than 0.6 g/L and the presence of MC vasculitis and hypogammaglobulinemia. The presence of MC vasculitis and hypogammaglobulinemia increased the odds of developing B-cell NHL 4.2-fold and 6.7-fold, respectively. These variables are reminiscent of the predisposing variables for the occurrence of NHL in SS. Indeed, serum MC, skin vasculitis, and decreased serum immunoglobulins are the primary predictors of NHL in patients with SS.^23,24 Twenty percent of deaths among patients with SS are attributable to lymphoma.²⁵ The presence of palpable purpura and low C4 complement levels are the primary predictors of mortality.²⁵ These findings differ from those among patients with non–HCV-related MC, as B-cell NHL was not associated with increased mortality in the present study.

As in SS, low-grade marginal zone lymphoma seems to be common among patients with non–HCV-related MC vasculitis. All marginal zone lymphomas associated with chronic infection or autoimmunity exhibit biased immunoglobulin V gene use and somatic hypermutations.²⁶ The recent finding that mucosa-associated lymphoid tissue lymphomas stand out among other B-cell lymphomas as frequently expressing immunoglobulin V genes with strong homology to RF²⁷ underscores the links between chronic antigenic stimulation, autoimmunity, and the development of marginal zone–derived lymphoproliferations.

By logistic regression analysis, the variables independently associated with death are age at diagnosis older than 60 years and renal involvement. The mean age at diagnosis of the patients who died was 13 years older than the mean age at diagnosis of those who survived. Renal involvement increased the risk of death almost 5-fold. There are few data on the natural history of MC. The overall 10-year survival after the diagnosis of MC vasculitis ranges from 50% to 90% in patients with renal involvement.^28,29 In historical series by Meltzer et al³⁰ and Gorevic et al,² renal involvement was the primary cause of death among patients with cryoglobulinemia. Consistent with our findings, in a cohort of patients with HCV-related MC, Ferri et al²⁹ recently reported lower survival rates among patients older than 60 years and among patients with renal involvement.

The large population with non–HCV-related MC vasculitis studied herein enabled a detailed analysis of their characteristics and outcomes, as well as a comparison with patients with HCV-related MC. Consistent with a previous study³¹ that included HCV-infected patients, the primary symptoms in our study associated with non–HCV-related MC vasculitis were purpura, renal involvement, and arthralgia. The clinical manifestations were asthenia, renal involvement, and purpura, which were found in more than 50% of patients with non–HCV-related MC vasculitis. The frequency of renal involvement (63%) was in the range of the 40% and 55% frequencies observed by Meltzer et al³⁰ and Gorevic et al,² respectively. In our study, the patients without HCV had higher frequencies of renal involvement and B-cell NHL and higher rates of death compared with patients infected with HCV. The main etiologic factors associated with non–HCV-related MC vasculitis are SS and B-cell NHL. Mixed cryoglobulinemia and SS share several characteristics, with the most striking being the proliferation of B-cell clones producing pathogenic IgM with RF activity and an increased risk for developing B-cell lymphoma (about 35-fold higher and 44-fold higher prevalence, respectively, compared with the general population).^22,32 Most of the data regarding the treatment of non–HCV-related MC vasculitis are derived from small uncontrolled studies.^4,6 The treatment of non–HCV-related MC vasculitis is similar to that of other vasculitides, with corticosteroid use and, in the most severe cases, with the use of immunosuppressive agents and plasmapheresis. According to previous reports,^2,28 the clinical remission rate is low. In the present study, a complete response was observed in only 25% of patients, and half of the patients experienced at least 1 relapse. Similarly, the outcome among 65 patients with non–HCV-related MC vasculitis was poor, with a 20% death rate after a mean follow-up of 4 years. Severe sepsis accounted for 69% of deaths.

In conclusion, patients without HCV represent approximately 10% of the population with MC. Patients with non–HCV-related MC vasculitis have a poor outcome and have a 4-fold increased risk of developing B-cell NHL.

Correspondence: Patrice Cacoub, MD, Department of Internal Medicine, Hôpital La Pitié-Salpêtrière, 83 Boulevard de l’Hôpital, 75651 Paris CEDEX 13, France (patrice.cacoub@psl.aphp-paris.fr).

Accepted for Publication: July 10, 2006.

Author Contributions:Study concept and design: Saadoun, Sellam, Piette, and Cacoub. Acquisition of data: Saadoun, Sellam, Ghillani-Dalbin, and Cacoub. Analysis and interpretation of data: Saadoun, Sellam, Crecel, and Cacoub. Drafting of the manuscript: Saadoun, Sellam, Crecel, and Cacoub. Critical revision of the manuscript for important intellectual content: Saadoun, Sellam, Ghillani-Dalbin, Crecel, Piette, and Cacoub. Statistical analysis: Saadoun, Sellam, and Crecel. Administrative, technical, and material support: Saadoun, Ghillani-Dalbin, and Cacoub. Study supervision: Saadoun, Piette, and Cacoub. Drs Saadoun and Sellam contributed equally to the study.

Financial Disclosure: None reported.

Acknowledgment: We thank Mathieu Resche-Rigon, MD, and Antoine Benard, MD, for discussions regarding the study.