Memory B cell subsets and plasmablasts are lower in early than in long-standing Rheumatoid Arthritis

Peripheral blood samples from 122 RA patients and 30 healthy subjects, matched for age and sex, were collected for flow cytometric analysis. The ethical approval for the study was obtained from the Catholic University of the Sacred Heart Ethical Committee and all participants gave informed consent.

All RA patients fulfilled the 1987 and 2010 American College of Rheumatology (ACR) criteria for RA [16],[17].

Sixty-eight of the 122 RA patients had symptom duration less than 12 months, of which 25 (36.8%) subjects with symptom duration less than 3 months were defined as having “very early RA (VERA)” [18]. The remaining 43 patients had symptom duration from more than 3 months up to one year (early-RA: ERA). Both VERA and ERA patients were naïve to disease-modifying antirheumatic drugs (DMARDs) and/or anti-TNF drugs at the time of enrollment.

VERA and ERA patients were followed every month up to three months and every three months thereafter. At each visit the ACR/ European League Against Rheumatism (EULAR) core data set (erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), swollen joint count (SJC), tender joint count (TJC), physician and patient global assessment, pain, health assessment questionnaire (HAQ)) was registered and DAS was calculated [19].

At baseline, VERA and ERA patients were treated with methotrexate (up to 20 mg weekly) and, when necessary, with a low steroid dose for three months; then a combination with a TNF blocker (adalimumab 40 mg every two weeks, or etanercept 50 mg weekly) was started if patients did not reach at least a good response according to EULAR criteria (DAS44: DAS ≤2.4) [20]. At each visit, clinical improvement and remission were evaluated according to DAS [20],[21] and CDAI cut-off points [22].

The other 54 RA patients had a long-standing disease (LSRA; mean disease duration of 12.5 ± 7.9 years). LSRA patients were consecutively selected from RA out-patients who had an active disease (DAS > 2.4) despite being on treatment with one or more DMARDs and/or anti-TNF drugs.

PB samples were collected at baseline from 122 RA patients and at 6-month follow-up visit from 61 RA patients and immunofluorescence labelling for flow cytometric analysis of circulating B cell subpopulations was performed by incubating 100 μl of PB with anti-human monoclonal antibodies. Cells were stained with anti-human antibodies specific for CD45 (APC-A750), CD19 (APC-700), CD3 (ECD), CD56 (ECD), CD38 (PC5 or APC), CD27 (PC7) and FITC-conjugated IgD (Beckman Coulter, Marseille France). After 20 minutes of incubation, the samples were fixed with 200 μl of Reagent 1 (Beckman Coulter, Marseilles, France) for 15 minutes at room temperature (RT) in the dark. After washing, the pellet was incubated with 200 μl of Reagent 2 (Beckman Coulter, Marceilles, France) for 15 minutes at RT in the dark. The samples were then incubated with PE-conjugated ZAP-70 specific AB (clone SBZAP, Beckman Coulter, Marseilles, France) for 30 minutes at RT in the dark. After staining, the cells were washed and immediately analyzed on properly compensated 8 colors Navios flow-cytometer and data were elaborated with the Kaluza program (Beckman Coulter, Marseille, France). Lymphocytes were gated on the basis of forward-and side-scatter properties (confirmed through CD45 staining) and at least 10,000 CD19+ cells were analyzed. B cell subsets were evaluated through the expression of surface B cell markers according to IgD/CD27 classification [23].

The expression of ZAP-70 on CD19+ cells was measured according to the gating strategy published by Crespo et al. [24]. B cells were defined as CD56 and CD3 negative and CD19 positive cells. The percentage of ZAP-70 positive B cells was evaluated at 98% of ZAP-70 positivity of T-NK cells [25].

RF-IgM and RF-IgA (Orgentec Diagnostika GmbH, Mainz, Germany), ACPA (Axis Shield Diagnostics, Dundee, UK) and anti-mutated citrullinated vimentin (MCV) (Orgentec Diagnostika, Dundee, UK) were measured using commercial ELISA and performed according to the manufacturer’s instruction. The suggested cut-off levels were 20 U/ml for RF-IgM, RF-IgA and anti-MCV and 5 U/ml for ACPA.

Blood samples were collected from all patients at baseline and from VERA and ERA patients six months after starting therapy. Samples were immediately centrifuged and stored at −80°C until analysis. Plasma levels of IL-6 and BAFF were measured by ELISA (R&D Systems, UK). The sensitivity of the test was of 2.2 pg/ml for IL-6 and 2.4 pg/ml for BAFF.

Statistical analysis was performed using SPSS (SPSS version 16.0, Chicago, IL, USA) and Graph-Pad Prism statistical software (San Diego, CA, USA). Categorical and quantitative variables were recorded as frequencies, percentage, mean ± Standard Deviation (SD). The non-parametric Mann–Whitney U test was used to compare the continuous variables. Categorical variables were analyzed using χ² test or the Fisher’s exact test.

The Spearman rank correlation was used to evaluate the relationship between the different B cell subsets and between B cell subsets and inflammatory and clinical parameters and the Wilcoxon test was used to compare the B cell subpopulations and soluble biomarkers during follow-up in the VERA and ERA cohort. A receiver operating characteristic (ROC) analysis of the continuous parameters related to the response to therapy was performed by plotting the relationship between sensitivity, on the y-axis, and 1-specificity, on the x-axis, for different cut-off levels of test positivity. The area under the ROC curve (ROC AUC) provides a measure of the overall discriminative ability of a model. The ROC area and its standard error (SE) were estimated using the nonparametric approach. The optimal cut-off point was determined to yield the maximum corresponding sensitivity and specificity.

The variables related to response to therapy in VERA and ERA patients with a p ≤ 0.15 at the univariate analysis entered into a multivariate logistic regression model in which “DAS remission or CDAI remission at 24 week follow-up visit” was the variable to be explained. Results are expressed as the odds ratio (OR) and 95% confidence interval (95% CI). Statistical significance was defined as p <0.05.

There were no differences in demographic, clinical and immunological parameters between patients with VERA, ERA and LSRA. Each cohort had a moderate-severe disease at study entry and no difference was seen between groups (DAS: 3.7 ± 1.1 in VERA, 3.3 ± 1.0 in ERA and 3.7 ± 1.4 in LSRA patients) (Table 1).

At study entry, 4 VERA (16.0%) and 9 ERA (20.9%) patients had already been treated with low doses of corticosteroids (less than 10 mg of prednisone daily). Among patients with LSRA, 24 (44.4%) were taking DMARDs, while a combination therapy with anti-TNF was administered to 30 (55.6%) of them.

VERA and ERA patients showed similar percentages of CD19+ cells (VERA: 10.1 ± 4.5%; ERA: 9.8 ± 4.2%), gated on total lymphocytes, with controls (9.5 ± 2.6%) but significantly higher compared to LSRA patients (6.9 ± 4.2%, p = 0.002 vs VERA and p = 0.001 vs ERA). There were no differences in the absolute B cell number between the three analyzed cohorts (data not shown).

We analyzed the main B cell subsets, according to their IgD and CD27 expression [23]. The frequency and total count number of antigen inexperienced naïve B cells (IgD + CD27-), antigen experienced pre-switched memory B cells (IgD + CD27+), post-switched memory B cells (IgD-CD27+), double negative memory B cells (IgD-CD27-) and CD27 + CD38+ B cells from VERA and ERA patients were compared to LSRA patients.

Similar percentages and absolute count of B cell subsets were observed in VERA and ERA patients (Figure 1, Additional file 1: Table S1 and Additional file 2: Figure S1) and controls. The naïve B cell subset (IgD + CD27-) (both percentage and absolute count) was increased significantly in both VERA and ERA patients compared to patients with LSRA, whereas the percentage and absolute count of IgD-CD27- B cells was lower in VERA and ERA patients compared to LSRA (Figure 1B and 1D, Additional file 1: Table S1 and Additional file 2: Figure S1). Moreover, the frequency and number of CD19+/CD27 + CD38+ cells were lower in VERA and ERA patients compared to LSRA patients (Figure 2, Additional file 1: Table S1 and Additional file 2: Figure S1).

These data suggest that the early phases of RA are characterized by an expansion of naïve-activated B cell subset and a lower frequency and number of double negative memory B cells and plasmablasts compared to the later stages of the disease.

No differences were found in B cell subset distribution dividing patients according to demographic parameters, positivity for ACPA or RF, both IgM and IgA, and steroid therapy in VERA and ERA patients at baseline.

Finally, the B cell subset distribution was similar according to the different therapeutic strategies in LSRA patients (data not shown).

Plasma BAFF levels were significantly higher in both VERA (p = 0.002) and ERA patients (p < 0.001) compared to LSRA patients (Figure 3A), whereas plasma IL-6 levels where higher in the VERA cohort when compared to LSRA subjects (p = 0.03) (Figure 3B). No difference was seen in plasma BAFF and IL-6 levels between VERA and ERA patients.

In VERA and ERA patients, plasma BAFF and IL-6 levels at baseline were not influenced by demographic parameters and corticosteroid use. Moreover, plasma levels of IL-6 and BAFF were similar in the subgroup of LSRA, independently from the different therapeutic strategies (data not shown).

In VERA and ERA patients, plasma BAFF levels directly correlated with naïve B cell (IgD + CD27-, r = 0.31, p = 0.01), plasmablast (r = 0.26, p = 0.05) and memory IgD-CD27- B cell (r = 0.31, p = 0.01) frequencies, and inversely correlated with unswitched- (IgD + CD27+: r = 0.41, p = 0.001) and switched- (IgD-CD27+: r = 0.33, p = 0.01) memory B cell frequencies.

There were no significant correlations between plasma IL-6 levels and the different B cell subsets, in the three cohorts of RA patients (data not shown).

The percentage and absolute number of B cells expressing intracellular ZAP-70 were similar in the three analyzed cohorts of RA patients (Additional file 1: Table S1), but higher than in healthy controls (1.8 ± 0.9% , p = 0.001 vs VERA, p = 0.002 vs ERA and p = 0.04 vs LSRA).

As noted in an our recent work [25], ZAP-70+ B cells show a phenotype belonging to the memory pool [IgD + CD27+ (p = 0.003), IgD-CD27+ (p < 0.001) and IgD-CD27- (p < 0.001)] when compared to the ZAP-70 negative ones (Additional file 3: Figure S2).

As expected, the percentage of circulating ZAP-70+ B cells directly correlated with the percentages of plasmablasts (r = 0.32, p = 0.01) and of double negative memory B cells (r = 0.37, p = 0.002), in VERA and ERA patients.

Finally, both the percentage (r = 0.40, p = 0.002) and the number (r = 0.35, p = 0.01) of circulating CD19+/ZAP-70+ cells directly correlated with plasma BAFF levels.

VERA and ERA patients with a severe disease at baseline (DAS >3.7) showed lower percentages of CD19+ cells (8.4 ± 4.0%) compared to subjects with a low-moderate disease activity (DAS < 3.7) (11.2 ± 4.4%, p = 0.02) and higher percentages and numbers of CD19+/CD27 + CD38+ (CD19+/CD27 + CD38+ (%): 4.4 ± 5.9% vs 1.2 ± 1.1%, p = 0.004 and CD19+/CD27 + CD38+ cells/μl: 7.2 ± 15.2 vs 3.2 ± 4.7 cells/μl; p = 0.04). We found no correlation between disease activity at baseline and the other B cell subsets, both in number and percentage (data not shown).

Forty-five out of 61 VERA and ERA patients (73.8%) were classified as good EULAR responders, while 32 (52.5%) were in DAS remission (DAS < 1.6) and 20 (32.8%) in CDAI-remission (CDAI ≤ 2.8) at 24 week follow-up visit. Seven VERA and ERA patients were lost during follow-up.

At 24 weeks of follow-up, 17 (27.9%) patients were in combination therapy with anti-TNF drugs, while the other 44 patients were in monotherapy with DMARDs (72.1%). The remission rate was similar in patients treated with DMARDs (DAS remission: 50.0%, CDAI remission: 27.3%) and anti-TNF drugs (DAS remission: 58.8%, p = 0.54; CDAI remission: 47.1%, p = 0.14) at 24 week follow-up visit.

The baseline disease characteristics and the B cell subset distribution were, first, analyzed in correlation with DAS remission at 24 weeks. By univariate analysis, younger age, lower inflammatory markers and lower disease activity, a higher frequency of naïve B cells (IgD + CD27-) and a lower frequency of switched-memory B cells (IgD-CD27+) were associated with DAS remission and subsequently included in the multivariate analysis. At the multivariate analysis, a younger age at diagnosis (OR(95% CI): 0.92 (0.87-0.97)) and a higher frequency of naïve B cells (IgD + CD27-) (OR(95% CI): 1.06 (1.01-1.11)) were significantly associated with DAS remission at 24 weeks of follow-up (Table 2). A more stringent definition of remission (i.e. CDAI remission) was evaluated in the same cohort of patients. At the multivariate analysis, “having VERA disease” (OR(95% CI): 8.89 (1.79-44.20)) and a low disease activity at baseline (OR(95% CI): 0.34 (0.14-0.83)) arose as significant predictors of CDAI remission together with a higher frequency of naïve B cells (IgD + CD27-) at baseline (OR(95% CI): 1.09 (1.02-1.17)).

When DMARD users were selected (n = 44), “having VERA disease” (OR(95% CI): 8.05 (1.16-55.99)) together with a higher frequency of naïve B cells (IgD + CD27-) at baseline (OR(95% CI): 1.09 (1.01-1.17)) were confirmed as the independent variables associated with CDAI remission.

We next performed ROC analyses for the circulating percentage of naïve B cells (IgD + CD27-) to predict remission (both DAS remission and CDAI remission) at the 24 weeks follow-up. The areas under the ROC curve (AUCs) were 0.743 ± 0.074 for DAS remission and 0.695 ± 0.092 for CDAI remission, which are at the fair to good threshold. Considering the CDAI remission, the sensitivity and specificity (cut-off value: 60.0%) were 75.0% and 53.0%, respectively.

Patients with circulating IgD + CD27- B cells at baseline > 60.0% had 4 times higher chances of reaching remission than patients with lower percentages [(OR(95% CIs): 4.2 (1.2-14.7)].

The multivariate analysis was not performed in the subgroup of anti-TNF treated patients due to the limited number of cases (n = 17).

Considering the subgroup of VERA and ERA patients treated with DMARDs, a reduction of the inflammatory biomarkers and disease activity was observed at 24 week follow-up visit, together with a significant decrease of RF-IgM and RF-IgA titers (Additional file 4: Table S2). No difference was seen in B cell subsets over time (both percentage and absolute number), both in patients reaching remission, defined as DAS < 1.6 or CDAI ≤ 2.8 and in those not responding to therapy. The low number of patients receiving anti-TNF agents forbids a definite conclusion, even though treatment with anti-TNF seemed to reduce the percentage of CD19+/ZAP70+ cells at 24 weeks (8.4 ± 9.5% at baseline vs 3.7 ± 2.6% at 24 weeks, p = 0.04) in VERA and ERA patients (n = 17).