Clinical and Immunophenotypic Characterization of CVID Patients Compared to a Non-CVID hID Cohort

In this study, a cohort of 52 hID patients was examined. 31 patients fulfilled the diagnostic criteria for CVID, whereas the remaining 21 participants constituted the non-CVID hID control group (Fig. 1A).

All cases of secondary hID were attributed to the iatrogenic effects of immunosuppressive drugs (rituximab, corticosteroids, cyclophosphamide, methotrexate, and leflunomide) administered for the treatment of either autoimmune conditions (spondylarthritis, rheumatoid arthritis, vasculitis) or hematologic disorders (chronic lymphatic leukaemia).

Clinical and demographic characteristics of CVID and non-CVID hID patients, are reported and compared in Table 2 The two populations were similar in terms of sex distribution, age, disease onset, and duration. Immunoglobulin replacement therapy was significantly more frequent among CVID patients. With respect to clinical presentation phenotypes, the CVID cohort demonstrated a significantly increased prevalence of infections, splenomegaly, pulmonary disorders, and autoimmune cytopenia (Fig. 1B).

Table 2 Sample demographics and clinical characteristics

We next assessed immunological parameters by flow cytometry comparing CVID vs. non-CVID hID patients in order to verify the consistency of diagnosis (Table S1 and Fig. 2).

Fig. 2

A Comparison of peripheral B-cell subset phenotypes between CVID and non-CVID humoral immunodeficiency (hID) patients. Black bars and gray bars indicate respectively CVID and non-CVID hID populations. B cell (CD19+ lymphocytes) subests are defined as follow: switched memory B cells = CD27 + CD21 + CD38 + IgM-IgD-; non-switched memory B cells = CD27+CD21+CD38+IgM++IgD+; CD21low B cells = CD27−CD21lowCD38lowIgM+IgD+; Naïve B cells: CD27−CD21+CD38+IgM+IgD++; Transitional B cells: CD27−CD21+CD38++IgM++IgD++; Plasmablasts: CD27++CD21+CD38+++IgD−B IgG values at onset and IgA, IgM and IgE values at last evaluation are compared between CVID (black bars) and non-CVID hID (gray bars) patients. C. Comparison of lymphoid cell populations (T-cells: total, CD4 + , CD8 + , g/d; NK-cells; NKT-like cells) between CVID and non-CVID hID patients. Black bars and gray bars indicate respectively CVID and non-CVID hID populations. The respective lymphoid population are indicated upon each column. Absolute and percentage count are represented respectively in the first and second raw. The histograms show the medians with 95% CI. P values are based on Mann–Whitney U test (p < 0,05:*; p < 0,01:**; p < 0,001:***). Abbreviations: Lymph.: lymphocytes

Accordingly, CVID patients displayed reduced levels of all immunoglobulin isotypes, in addition to a reduction in class-switched memory B cells (CD19+IgM−IgD− CD27+CD21+CD38+), which is considered an immunophenotypic hallmark of the disease. Notably, in CVID patients, there was a discernible decrease not only in switched memory (sm)-B cells but also in non-sm B cells (CD19+IgM++IgD+CD38+CD27+CD21+), thereby underscoring an impaired B cell activation. In addition, concerning T-cell phenotypes, CVID patients exhibited higher relative and absolute αβ CD8+ T-cell counts and a decreased frequency of αβ CD4+ T-cells (Fig. 2C), resulting in a reduced CD4/CD8 ratio in CVID vs. non-CVID hID patients (respectively: 1.91 [CI95% 1.56–2.26]; 2.51 [CI95% 2.11–2.91]; p = 0.0181, two-tailed Mann Whitney u test).

Taken together, these data confirmed that CVID patients have a distinct clinical immunophenotypic signature with altered T and NK cell frequencies along with B cell defect, while this is not the case in other ID presenting with hypogammaglobulinemia.

“Complicated” Clinical Phenotype in CVID Patients is a Function of Disease Duration

With specific regard to the CVID cohort, the 31 patients had a mean symptomatic disease duration of 13.7 years and exhibited a wide spectrum of clinical manifestations with patients displaying with one or multiple clinical features, as illustrated in Fig. 3. As shown in a direct comparison of frequency of clinical manifestations and relative odds ratios, we observed a clear association between the prevalence of splenomegaly, lymphadenopathy, and granulomatous manifestations. Indeed, the presence of granulomas strongly associated with splenomegaly and lymphadenopathy (LA), with an odds ratio of 12.9 for both conditions (p = 0.0217, Fisher’s exact test). Likewise, there was a significant correlation between the prevalence of LA and splenomegaly over the disease course (p = 0.0019, Fisher’s exact test).

Fig. 3

Features of clinical presentation of CVID patients with absolute frequencies and the relative odds ratio of the probability of their occurrences. Numbers in the lower left triangle (red text) indicate the absolute frequencies of occurrence/co-occurrence of symptoms among all CVID patients. Numbers in the upper right triangle (blue text) show the odds ratio for the probability of occurrence of the combined phenotypes. Significant associations are color-coded according to the legend (p values were based on two-sided Fisher’s exact test). Abbreviations: LA: lymphadenopathy

Overall at the time of analysis, the vast majority of patients (94%) exhibited an infectious phenotype, primarily characterized by upper respiratory tract infections including otitis and sinusitis (61%), followed by lower respiratory infections (45%, with 50% of these cases diagnosed as pneumonia), gastrointestinal infections (31%), and a minority of skin infections (7%), urinary tract infections (7%), central nervous system infections (1 viral encephalitis), osteomyelitis, and sepsis (CMV, C.jejuni.) and 3 cases of recurrent cutaneous herpes zoster.

Among autoimmune and inflammatory manifestations observed in 19 patients (61%), autoimmune cytopenia was the most frequent disorder observed (29%). Other autoimmune diseases included connective tissue disorders (systemic sclerosis, rheumatoid arthritis, Sjogren syndrome, undifferentiated connective tissue disorders), gastrointestinal conditions (IBD-like, celiac-like disease, autoimmune atrophic gastritis), dermatological disorders (discoid lupus erythematosus, psoriasis, lichen planus), and Kawasaki disease.

Malignancies were diagnosed in 6 patients (19%) with B cell lymphoma being the most prevalent (2 patients). Other oncologic conditions were gastric cancer, thyroid carcinoma, testicular seminoma, and uterine carcinoma.

Six patients exhibited granulomatous diseases (lymph nodes, lungs, and spleen), 3 patients with multiple tissue localization.

In view of the prolonged observation and the apparent progressively occurring additional symptoms, we next studied the time course of new clinical disease manifestations. Our assessment evaluated symptoms at the initial presentation (onset of clinical symptoms) compared to those observed at the last visit (after an average of 13.7 years, IC95% 8.8–18.5 years) and showed that the infectious phenotype was the predominant initial manifestation (77% at onset vs. > 90% at the last clinical evaluation). Other clinical manifestations gradually increased over time, with autoimmune diseases showing a six-fold rise in prevalence (Fig. 4A).

Fig. 4

A Clinical phenotypes in CVID: onset vs. last clinical evaluation. Values represent the absolute patient count for each specific manifestation. B Evolution of clinical phenotypes over time. This graph illustrates the progression of clinical manifestations from onset to the last visit. “Onset” indicates syndrome which were present at first evaluation. “Last evaluation” indicates add-on syndromes to the one in “onset”. Onset manifestations, if limited on time or controlled by therapy (e.g., resolved neoplasm, autoimmune disease without further exacerbations, single infectious episode), are not considered among the manifestations at the last assessment. The thickness of each line indicates the percentage of patients who developed the condition indicated by the arrow, respect to the baseline (“onset”) number. For instance, the purple arrow for asymptomatic patients (1 individual) represents 100% because 1 out of 1 patient (hence, 100%) displayed an autoimmune phenotype during the last assessment. It is possible that the sum of the percentages of arrows of the same colour exceeds 100%, as the same patient may have presented different clinical manifestations. The numbers within the boxes represent the absolute number of patients who exhibited that specific manifestation in that specific time

To better understand dynamics of symptoms/syndromes over time in CVID patients, we elaborated data according to first presentation and subsequent syndromic evolutions for all patients and we drew the evolution of clinical phenotypes from the initial presentation to the final clinical assessment within the cohort (Fig. 4B). This analysis showed that, despite the onset, clinical manifestations have an exceptionally broad spectrum of evolution. Indeed, among the 24 patients who initially displayed an infectious phenotype, nearly 60% subsequently exhibited at least one additional clinical manifestation over time, including autoimmunity, malignancies, and granulomatous diseases. Notably, patients with an infectious onset and no additional complicating syndromes, had a significantly lower disease duration compared to those that developed autoimmunity, malignancies, or granulomatosis (means: 7.5 years [IC 95%, 1.1–13.9 years]; 18.43 years [IC95% 10.6–26.3 years], respectively; p = 0.0246, two-tailed Mann–Whitney U test).

Therefore, these observations challenge the routine clinical categorization of CVID patients based on specific phenotypes (e.g. complicated vs. non-complicated).

CMV Viraemia is Detected in a Proportion of CVID Patients

As for CMV or EBV, viraemia with at least one of these two viruses was identified in 25% of the patients overall. CMV replication was observed in 16% of CVID and in 0% of non-CVID hID patients (p = 0.05, Fisher’s exact test). Conversely, EBV replication was equally distributed between the two populations. The mean CMV-DNA load was 380.9 IU/mL (range: 67–1382 IU/ml), while the mean EBV-DNA load was 2550.5 IU/mL (range:107–15288 IU/mL). No patients displayed clinical symptoms or laboratory abnormalities indicative of acute infection. CMV DNA detection in blood was not related to immunosuppressive medications. Specifically, among the CMV DNA-positive CVID patients, 4 out 5 were not treated with immunosuppressive drugs, while one patient was solely receiving low-dose of oral prednisone (5 mg/day) at the time of sampling.

Regarding EBV viraemia, 36% of the patients had either received or were currently undergoing immunosuppressive therapy. No differences were observed in terms of sex and age distribution between CMV and non-CMV patients (p = 0.637; p = 0.598, respectively), as well as between EBV and non-EBV patients (p = 0.318; p = 0.788, respectively). In addition, all patients tested negative in HHV6, HIV, HCV, and HBV RT-PCR blood testing.

Overall, patients with CMV viraemia had all a “complicated phenotype”, and a mean disease duration of 16.2 years, superimposable to the one of the “complicated phenotype” patients (18.4) years and clearly distinct from patients initially displaying a simple infectious clinical phenotype and no CMV viraemia (7.5 years). Further, CMV viraemic patients represented 24% of the patients with complicated phenotype (5 out of 21).

This suggests that, within a clinical evolution of CVID clinical phenotypes, CMV viraemia is present in only a subset of patients with complicated phenotype and higher disease duration.

Reduced NK-cell Activation and Circulation of Inflammatory Precursors in CVID Patients with Active Viral Replication

Following the observation of virus circulation (EBV or CMV) in 26% of CVID patients, we addressed the question whether this lack of control could be associated with clinical and flow-cytometric phenotypes. To this end, we further studied the CVID cohort according to the presence/absence of viraemia (CMV only, EBV only, either CMV or EBV) (supplementary Tables S2-S4).

All these 3 groups were homogeneous in terms of demographic characteristics and disease duration. We observed, a strong association between CMV and EBV replication (OR = 11.5, p = 0.0376, Fisher’s exact test), with 3 patients displaying both CMV and EBV viraemia. Splenomegaly was the prevalent clinical finding in CMV and CMV/EBV patients (p = 0.0047 and p = 0.0316, respectively). No association was observed between immunoglobulin levels or B cell subpopulations and lack of viral control. There was a trend towards increased CD3+CD8+ T-cell frequencies in viraemic patients and reduced CD3+CD4+ T-cells, resulting in low CD4/CD8 ratio (Fig. 5).

Fig. 5

Comparison of lymphoid cell populations of CVID patients according to active ( +) or absent (-) viral replication of CMV-only, EBV only, and CMV and/or EBV (CMV/EBV) groups. Except for the CD4/CD8 ratio, all data is expressed as a percentage, specifically, T-cells and NK-cells as a percentage of total lymphocytes, while CD4 and CD8 as a percentage of T-cells. The histograms show the medians with CI 95%. P values are based on two-tailed Mann–Whitney U test (p < 0,05:*)

To investigate whether failure to control CMV viraemia might be associated with a dysregulation in cell activation, we further evaluated by flow cytometric analyses the expression of CD69, CD38 and HLA-DR molecules on T-cells and NK-cells (Fig. 6A and B). The expression of activation markers on CD4+ and CD8+ T-cells was similar in CVID viraemic and non-viraemic patients (supplementary Table S2). The mean absolute count of NK-cells was slightly decreased in all viraemic patients, with a trend towards a reduction in NK CD56dim in the CMV group (p = 0.071). In CVID patients with CMV viraemia we observed a decrease in HLA-DR+ NK-cells (p = 0.048) without differences in CD69+HLA-DR− and CD69+HLA-DR+ cell frequencies (Fig. 6B).

Fig. 6

HLA-DR, CD38 and CD69 expression on T subsets and NK-cells. A Dot plots illustrate activation marker expression on CD4 + T-cells of both a CVID patient with CMV-only viraemia (CMV +) and a CVID patient without viraemia (CMV -). B Dot plots illustrate activation marker expression on CD8 + T-cells of both a CVID patient with CMV-only viraemia (CMV +) and a CVID patient without viraemia (CMV -). C. Dot plots illustrate HLA-DR and CD69 expression on NK-cells of both a CVID patient with CMV-only viraemia (CMV +) and a CVID patient without viraemia (CMV -). In the column bar graph, HLA-DR + NK-cell count between patients with active ( +) and absent (-) CMV replication was compared. The histograms represent the medians with CI 95%. P values are based on two-tailed Mann–Whitney U test (p < 0,05:*)

Finally, we analysed the circulation of inflammatory lymphoid precursors LIN−CD34+DNAMbrightCXCR4+ and CD34−LIN−CD16+CD7− precursors. Interestingly, the EBV/CMV viraemic group exhibited a reduction in both the absolute count of Lin−DNAMbright and of CD56−CD16+ precursors (Fig. 7 and supplementary Table S4). Both CMV and EBV groups displayed a trend towards decrease in these cells. When considering other clinical outcomes such as autoimmunity, splenomegaly, granulomatous disease, LA, atopy, malignancies, and lung involvement, there were no significant differences in the absolute count of these cells compared to their respective aviremic control groups. The absolute count of CD56−CD16+ precursors positively correlated with the absolute number of NK-cells, NK CD56dim, and HLA-DR+ NK-cells (Fig. 7), in line with their ability to generate in vitro functional NK-cell progenies [29]. Conversely, DNAMbright precursors did not exhibit any significant correlations with NK subsets. No correlations were identified between precursors and T-cell populations (total T-cells, CD4+, and CD8+ T-cells).

Fig. 7

1. Comparison of lymphoid precursors (Lin-DNAMbright and Lin-CD16 +) absolute count between patients with active and absent viral replication. A: Either CMV or EBV replication; B: CMV-only; C: EBV-only. Histograms represent medians with CI 95%. P values are based on Mann–Whitney U test (ns = not significant; p < 0,05:*). 2. Dot plots illustrate LIN-DNAMbright precursors of both a CVID patient with concurrent CMV and EBV viraemia (D) and a CVID patient without viraemia (E). 3. Correlation matrix of NK-cell subsets and lymphoid precursors absolute counts. The left figure depicts the correlation matrix, with each cell containing the Spearman r value between precursors and NK-cell subsets, while the right figure illustrates the adjusted p-values (by Benjamini–Hochberg correction) related to the same correlations using a color-coded legend. Abbreviations: Lin: lineage; Lymph: lymphocytes; Prec.: precursors

Taken together, these results show a possible involvement of NK-cells and inflammatory common lymphoid precursors in the lack of CMV control in a subset of CVID patients with prolonged disease course.