Common variable immunodeficiency disorders (CVID) comprise the largest clinically relevant group of primary immunodeficiency disorders (PID) with an estimated prevalence of 1:25.000 (1). According to the European Society for Immunodeficiencies (ESID) criteria, patients present with relevant IgG and IgA +/- IgM deficiency together with reduced class-switched memory B cells and/or impaired specific antibody response to vaccination. There is substantial clinical heterogeneity in CVID patients as a result of infectious as well as non-infectious complications. In particular, non-infectious manifestations contribute significantly to disease morbidity and mortality and may present as immune cytopenia, lymphoproliferation, granulomatous disease, interstitial lung disease, enteropathy, autoimmunity, malignancy, or other signs of immune dysregulation (2).

The clinical burden and mortality of CVID-related liver diseases is high (3–6), however, data on prevalence vary substantially between 9% and 79% and depend greatly on the defining inclusion criteria (7). Considering elevated liver enzymes, approximately 50% of CVID patients were found to present with any sort of hepatic manifestation (8, 9), however, histologically proven liver involvement was reported to occur in only 9% of patients (10).

In recent publications, approximately 4-6% of adult CVID patients were found to suffer from clinically relevant portal hypertension (4, 11), presenting with non-malignant ascites, esophageal varices, or other signs of increased portal pressure and expressing an elevated mortality rate as high as 41% (4).

Most CVID patients with portal hypertension were reported to suffer from additional non-hepatic organ manifestations, such as autoimmune cytopenia, enteropathy, granulomatous disease, interstitial lung disease, and lymphoproliferation (4), but it remains unclear if these overlapping non-infectious manifestations follow common pathomechanisms. While other non-infectious complications may occur early or even before CVID diagnosis, portal hypertension is regarded as a late organ manifestation (4). Although the majority of patients with portal hypertension show signs of nodular regenerative hyperplasia (NRH) in histopathology (7, 12, 13), there is significant heterogeneity and potential risk factors remain incompletely understood. Due to the parallel appearance of nodular regenerative hyperplasia together with other histopathological features, Crotty et al. coined the term NHR-like changes (NRH-LC), which will be used in this article (13).

On a cellular level, hepatic disease in CVID is associated with (CD8+) T cell infiltration, while only a few B cells are detectable in the liver tissue of affected patients (12, 13).

In order to avoid delayed diagnosis of portal hypertension, screening strategies for early detection are important (14). Splenomegaly and elevated liver stiffness were proposed to be associated with portal hypertension (4), however, changes in these parameters can result for multiple reasons and invasive measurement of the hepatic venous pressure gradient (HVPG) may be required to establish the diagnosis.

In the present study, we retrospectively analyzed the occurrence of portal hypertension in a single-center cohort of 196 adult CVID patients. We identified 11 patients (5.6%) who presented with non-malignant ascites, esophageal varices, or gastropathy as clinical manifestations of significant portal hypertension. We compare clinical and immunological data between affected (CVID+PH) and non-affected patients and report on liver histology as well as on treatment of portal hypertension. In addition, we provide longitudinal laboratory and imaging data of eight affected patients before developing and while suffering from portal hypertension.

We identified 11 patients with symptomatic PH. Symptom onset occurred with a median delay of 7 years (IQR: 4.5 years; mean delay 10 years) and ranging from 3 years to 36 years after establishing CVID diagnosis. In our cohort, 9/11 patients were female (see Table 1 ). The mean observation period for all patients was 12 years (IQR: 2 years).

In the present study, we report the clinical, immunological, and histological findings of 11 patients with significant PH as hepatic involvement in a cohort of 196 CVID patients. Relevant PH was defined by the presence of at least one of the following clinical manifestations: non-malignant ascites, esophageal varices, or PH gastropathy. In our cohort, 5.6% of all adult CVID patients met these defining criteria, thereby confirming previous data on the prevalence of this hepatic complication (4, 11). The clinical burden of PH in our cohort was high. The occurrence of spontaneous bacterial peritonitis (SBP) and gram-negative sepsis shortly after TIPS implantation highlights the increased risk for infections in this patient population (17) and advocates for the early use of antibiotic prophylaxis. The mortality rate in patients with CVID+PH was higher than in unaffected CVID patients (18% vs 6%), but lower than the previously reported frequency of 28%-42% (4, 5), which may be related to the shorter observation period. Although reported as a late-onset complication, it is important to communicate that both fatal cases in our CVID+PH group occurred at a very young age (patients were deceased at 29 and 34 years).

We confirm a high association with other non-infectious manifestations in CVID+PH patients, most significantly with the occurrence of GLILD, other non-pulmonary granulomatous lesions, splenomegaly, and lymphadenopathy (4, 5, 9, 18, 19). The higher frequency of affected female patients (82%) and the shorter interval of symptom onset of PH in our cohort differ slightly from previous reports.

The most frequent histological abnormalities in CVID-associated liver disease are nodular regenerative hyperplasia-like changes (3–5, 7, 8, 13, 14). In our cohort, histopathology of the liver was available for nine patients, mostly showing an NHR-LC pattern, but also (sinusoidal) lymphocytic infiltration, mild portal or porto-septal fibrosis, and epitheloid granulomas.

In CVID patients with liver fibrosis, Crotty et al. reported a distinct pattern of pericellular fibrosis (13). In general, liver fibrosis results from collagen remodeling and the accumulation of extracellular matrix (ECM) (20). ECM can further be subdivided into the endothelial basement, mainly consisting of laminin and collagen type IV, VIII, and X, and the interstitial matrix, which is composed of the fibrillar collagens type I, II, and III. The process of tissue remodeling may also be detected in systemic circulation. Changes in the pericellular basement membrane were reported to correlate with PRO-C4 and other markers in serum of mild-to-moderate alcoholic and non-alcoholic liver disease, while PRO-C3, as a marker of interstitial matrix collagen type III, appears to correlate with progressed stages of liver fibrosis (21–23). Interestingly, PRO-C4 was recently reported to be elevated in CVID patients with non-cirrhotic PH (4).

Immune dysregulation in CVID may result from B-cell (24–26) and/or T-cell dysregulation (27, 28). In CVID, intra-sinusoidal T lymphocytes were proposed to be involved in NRH-LC pathogenesis (13). In addition to hepatic T cell receptor clonality analysis in NHR suggesting targeting of sinusoidal endothelial cells, hepatocytes were also found to show a strong overexpression of IFN gamma mRNA, proposing a pathophysiological role of cytotoxic T lymphocytes in tissue (12). Histopathological work-ups in the current study also observed T lymphocytes in the majority of examined liver tissue. However, the precise mechanisms of T-cell-mediated liver fibrosis in CVID patients remain unclear. In addition, the primary cause of T lymphocyte migration to the liver of CVID patients also remains unknown and changes in the intestinal microbiome and recurrent bacterial translocation (5, 29) or rare viral infections have been proposed (30). Larger histological cohorts are required to further examine how different histological patterns in CVID patients are related to disease stages and possible pathomechanisms.

Immunological data in peripheral blood revealed reduced levels of naïve CD45RA+CD4+ T-cells in all CVID+PH patients, thereby confirming the recent observation of lower naïve CD4+ cells in CVID patients with PH and NRH-LC (17). We also observed a higher frequency of B-lymphocytopenia (EUROclass B-) in the CVID+PH patients than in the control group of the other 185 CVID patients (45% vs 4%; p-value < 0.00001). Lower levels of B-lymphocytes in patients with CVID-associated liver diseases have been previously reported (5, 8). Although there is an association between elevated levels of activated CD21low B cells and the occurrence of non-infectious manifestations (28, 31), we did not observe any significant differences in the B-cell subsets, in particular, we did not find a higher frequency of patients with Euroclass smB-21low or elevated levels of activated B cells (CD21lowCD38low). Of note, NHR was recently reported as a frequent late-onset hepatic complication in patients with X-linked agammaglobulinemia (XLA) (32), suggesting a T-cell-driven pathomechanism in this liver disease.

Early non-invasive detection of liver disease is key, however, apart from relatively unspecific changes in liver enzymes, GGT, and platelets, there are currently no established biomarkers for hepatic manifestations in CVID patients. Transaminases may be constantly or intermittently elevated (9), and our data also confirm that normal liver enzymes cannot rule out liver manifestation in CVID. Globig et al. proposed PRO-C4, a serum biomarker of liver fibrosis and collagen formation in CVID patients (4) and Lima et al. reported higher levels of beta-2-microglobulin in CVID patients with liver-spleen-axis abnormalities in comparison to those patients without hepato-splenic changes (5). In the present study, we confirm the association of low platelets and elevated GGT with PH in CVID patients.

Due to the multi-organ involvement with many non-infectious manifestations occurring simultaneously, common underlying pathomechanisms have been proposed in CVID patients (28, 31, 33). However, the frequent co-occurrence of different organ complications is hampering the identification of specific risk factors and biomarkers that could be related exclusively to the involvement of a single non-infectious manifestation. By analyzing longitudinal data in our patient cohort, we show for the first time the dynamics of different biomarkers (GGT, platelets, sIL-2R) and imaging (spleen size) in CVID patients both before the development and under the clinical manifestation of PH.

While the small sample size limits the interpretation, looking at longitudinal data might nevertheless allow us to get a slightly more granular glimpse of the development of portal hypertension in CVID. Eight longitudinally followed CVID+PH patients presented at the time of CVID diagnosis with a multi-organ combination of splenomegaly, lymphadenopathy, and GLILD. Over time, lymphadenopathy and GLILD remained stable in 7/8 patients, while imaging revealed a gradually increasing splenomegaly, and analysis of blood markers showed an elevation of sIL-2R in addition to a decrease in platelets and an increase of GGT.

However, elevated sIL-2R levels cannot be regarded as a specific marker for hepatic involvement. In addition to significantly higher levels of soluble IL-2 receptor (sIL-2R) in the CVID+PH patients of our cohort, we could also confirm elevated levels in patients with GLILD (manuscript in preparation) (34). Therefore, sIL-2R should be considered a general marker of T cell activation (35–37). Our longitudinal data from CVID+PH patients expand that spectrum and suggest that it may, in combination with other markers, also hint at progressive PH in CVID patients.

In addition to biomarkers for screening of hepatic manifestations, imaging plays a pivotal role. Our observation of mostly normal findings in conventional sonography, including normal velocity of portal venous flow in (82%), are in line with the so far largest single center analysis (4), however, other studies reported sonographic abnormalities in approximately 50%-66% of CVID patients with NRH-LC (12, 14) and detected portal vein enlargement in 25% of patients with primary antibody deficiency (11). Of note, significantly elevated HVPG (> 10 mmHg) was detected in 6/6 patients despite normal velocity in portal trunk, suggesting an insufficient sensitivity of this non-invasive examination for the detection of non-cirrhotic PH in CVID patients.

While our longitudinal data also indicate that increasing spleen size should raise high suspicion of the presence of PH, other causes should always be considered. This is exemplified by patient #5, where histology of splenic tissue after splenectomy due to massive splenomegaly with organ compression revealed in addition a nodular lymphocyte predominant Hodgkin Lymphoma.

Inspired by the applicability of transient elastography in other chronic liver diseases (38–40), attention has also been drawn to its diagnostic use for identifying hepatic manifestations in CVID patients (4, 18, 19). All the CVID+PH patients in this study with proven PH (HVPG ≥ 10 mmHg) expressed increased rates of liver stiffness, thereby confirming previous reports and suggesting the early use of this methodology in CVID cases with suspected liver involvement (4, 18). However, transient elastography is not widely available and harbors technical challenges, such as in obese patients. Additionally, liver stiffness is not exclusively influenced by liver inflammation, but also by prandial status, biliary obstruction, or cardiac output (41).

Our study has important limitations. With an estimated prevalence of 5%-6% of CVID patients suffering from PH, our sample size is underpowered and the median follow-up time is not sufficient for a solid evaluation of long-term morbidity and mortality of PH in CVID. In order to better understand the pathomechanisms of liver disease in CVID, future studies would certainly benefit from a standardized in-depth analysis of tissue samples, including detailed immunohistochemical staining of lymphocytes, and should also include a more profound assessment of pericellular fibrosis.

Currently, no specific EMA- or FDA-approved treatment for CVID patients with PH due to NRH-LC is available. Data on corticosteroids are variable (12, 42) and there are only a few reports on the use of anti-TNFa treatment (43, 44). In our cohort, patient #5 showed a good response to anti-TNFa treatment on granuloma formation, however, overall liver disease progressed. Outcome data after liver transplantation are also variable, with the re-occurrence of NRH-LC in the transplanted organ in CVID patients (45–48) as well as in patients without inborn errors of immunity (49). Despite the elevated morbidity of CVID patients with non-cirrhotic PH, liver transplantation remains a challenging option. First, this is because the clinical severity of liver disease is mainly assessed by the MELD score and most CVID patients with non-cirrhotic PH score too low for listing (50). Second, many centers are reluctant to accept CVID patients due to preexisting comorbidities. Two of our patients died following acute infection. The role of HSCT remains unclear but it was reported that the presence of NRH-LC may affect outcomes after stem cell transplantation (51).

Our single center analysis of PH in CVID patients could confirm the high degree of clinical burden and an estimated prevalence of 5%-6%. Our data support a role for transient elastography during early screening and reveal that non-invasive ultrasound-based assessment can be normal despite significantly elevated HVPG. In addition, our longitudinal data suggest that the combination of increasing spleen size, increasing sIL2-R and GGT values, and gradually reducing platelets should raise high suspicion of the presence of PH. Currently, consensus recommendations for diagnostics and management of liver disease in CVID patients are rare (14) and larger observational and interventional studies as well as international clinical guidelines are urgently needed.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving humans were approved by ethics commitee of Charité (EA4/112/19), Berlin, Germany. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

LH: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Supervision, Writing – original draft, Writing – review & editing. SS: Visualization, Writing – review & editing. KW: Writing – review & editing. MS: Writing – review & editing. CK: Writing – review & editing. CS: Writing – review & editing. AF: Conceptualization, Project administration, Writing – review & editing.