Immunological abnormalities associated with hereditary haemorrhagic telangiectasia


  • A. Guilhem,

    Corresponding author
    1. CHU de Montpellier, Service de Médecine Interne A, Hôpital Saint Eloi, Montpellier, France
    • Correspondence: Alexandre Guilhem, MD, Service de Médecine Interne A, Hôpital Saint Eloi 80, avenue Augustin Fliche 34295, Montpellier cedex 5, France.

      (fax: +33 4 67 33 72 91; e-mail:

    Search for more papers by this author
  • C. Malcus,

    1. Hospices Civils de Lyon, Laboratoire d'immunologie, Hôpital E Herriot, Lyon, France
    Search for more papers by this author
  • B. Clarivet,

    1. CHU de Montpellier, Unité de Recherche Clinique et Epidémiologique, Hôpital La Colombiere, Montpellier, France
    Search for more papers by this author
  • H. Plauchu,

    1. Hospices Civils de Lyon, Centre National de Référence pour la Maladie de Rendu-Osler, Service de Génétique, Hôpital Louis Pradel, Bron, France
    Search for more papers by this author
  • S. Dupuis-Girod

    1. Hospices Civils de Lyon, Centre National de Référence pour la Maladie de Rendu-Osler, Service de Génétique, Hôpital Louis Pradel, Bron, France
    Search for more papers by this author



Hereditary haemorrhagic telangiectasia (HHT) is a genetic disorder related to mutations in one of the coreceptors to the transforming growth factor-β superfamily (ALK1 or endoglin). Besides the obvious vascular symptoms (epistaxis and arteriovenous malformations), patients have an unexplained high risk of severe bacterial infections. The aim of the study was to assess the main immunological functions of patients with HHT using the standard biological tests for primary immunodeficiencies.

Design, setting and subjects

A prospective single-centre study of 42 consecutive adult patients with an established diagnosis of HHT was conducted at the National French HHT Reference Center (Lyon). Lymphocyte subpopulations and proliferation capacity, immunoglobulin levels and neutrophil and monocyte phagocytosis, oxidative burst and chemotaxis were assessed.


Innate immunity was not altered in patients with HHT. With regard to adaptive immunity, significant changes were seen in immunological parameters: primarily, a lymphopenia in patients with HHT compared with healthy control subjects affecting mean CD4 (642 cells μL−1 vs. 832 cells μL−1, < 0.001), CD8 (295 cells μL−1 vs. 501 cells μL−1, < 0.0001) and natural killer (NK) cells (169 cells μL−1 vs. 221 cells μL−1, < 0.01), associated with increased levels of immunoglobulins G and A. This lymphopenia mainly concerned naïve T cells. Proliferation capacities of lymphocytes were normal. Lymphopenic patients had a higher frequency of iron supplementation but no increase in infection rate. Lower levels of immunoglobulin M and a higher rate of pulmonary arteriovenous malformations were found amongst patients with a history of severe infection.


Patients with HHT exhibit immunological abnormalities including T CD4, T CD8 and NK cell lymphopenia and increased levels of immunoglobulins G and A. The observed low level of immunoglobulin M requires further investigation to determine whether it is a specific risk factor for infection in HHT.


Hereditary haemorrhagic telangiectasia (HHT), also known as Rendu–Osler–Weber disease, is an autosomal dominant vascular disease with an approximate prevalence of 1 in 5000–8000. The main clinical features include mucocutaneous telangiectasia (responsible for epistaxis) and arteriovenous malformations (AVMs) involving the lungs, brain, liver or gastrointestinal tract in most cases [1, 2]. Haploinsufficiency of ENG, ALK1 or SMAD 4 is known to cause the HHT phenotype. All of these genes are implicated in the cellular pathway of the transforming growth factor-beta superfamily. BMP9, one of about thirty members of this family, has recently been identified as the natural ligand of ALK1 [3]. The exact pathogenesis of HHT is not yet fully elucidated, but dysfunction of the BMP9/ALK1/endoglin signalling pathway is likely to play a role [4, 5].

Serious infections in patients with HHT are clinically relevant events, documented in many case reports [6-8]. In a retrospective study, about 13% of patients in an HHT cohort had a history of severe infection, in particular cerebral abscesses or staphylococcal osteoarthritis. This risk of infection was linked to mechanical factors, such as pulmonary AVMs or prolonged nasal packing for epistaxis [9].

However, endoglin and, to a lesser extent, ALK1 were found to be expressed by mononuclear cells [10] and lymphocytes [11]. Moreover, in a functional comparative study of patients with HHT, decreases in polymorphonuclear leucocyte (PMN) and monocyte oxidative burst and phagocytosis were demonstrated [12]. Based on these findings, we have hypothesized that HHT could be associated with intrinsic immunological dysfunction.

In order to obtain an overview of immunity in HHT, we carried out a transversal examination of several immune parameters in an HHT population using the same biological assays as used for characterization of primary immunodeficiencies.

Patients and methods


We conducted a prospective study of consecutive adult patients at the National French HHT Reference Center (Lyon, France) between February and October 2008. Patients were included during a routine visit, which was scheduled several months in advance according to the French HHT guidelines. The French National Reference Center provides specialized medical care to a large cohort of regional patients and exceptionally to extra-regional patients requiring aggressive and highly specialized treatments (e.g. hepatic graft).

The diagnosis of HHT was based on the Curaçao criteria [13] and confirmed by genetic analysis of ALK1, ENG and SMAD4 mutations. All patients had at least one thoracic computed tomography scan and hepatic echography assessment, according to the French guidelines for HHT diagnosis and treatment. Cerebral or gastrointestinal tract investigations were only performed in symptomatic patients. Patients underwent a medical examination before blood samples were collected. Written informed consent, according to French national ethical guidelines, was obtained from all patients.

Assessment of innate immunity

Phagocytosis, oxidative burst and chemotaxis were evaluated using PHAGOTEST®, PHAGOBURST® and MIGRATEST® kits, respectively, according to the procedures recommended by the manufacturer (ORPEGEN Pharma, Heidelberg, Germany) and using a Cytomics FC 500 flow cytometer (Beckman Coulter, Villepinte, France).

PHAGOTEST® was performed with heparinized whole-blood samples to determine the percentage of PMNs and monocytes in which phagocytosis of opsonized fluorescein isothiocyanate (FITC)-labelled Escherichia coli was observed. PHAGOBURST® was used in heparinized whole blood to assess the percentage and the mean fluorescence intensity (MFI) of cells producing reactive oxygen species after stimulation with opsonized unlabelled E. coli using dihydrorhodamine 123 as a fluorogenic substrate. The percentage of positive PMNs and monocytes was determined by counting the number of cells with an MFI of >9 and >3, respectively.

Using leucocyte-rich plasma isolated from heparinized whole blood by spontaneous sedimentation, MIGRATEST® measured the number of PMNs that migrated actively through a plastic membrane (pore diameter of 3.0 μm) towards the chemoattractant N formyl-methionine-leucine-phenylalanine (fMLP), compared with spontaneous migration without fMLP. The chemotactic index was calculated by dividing the number of cells that migrated towards fMLP by the number of cells that migrated in the absence of fMLP. Further description of these kits is provided in the Supporting Information.

For all experiments (PHAGOTEST®, PHAGOBURST® and MIGRATEST®), a control group was used comprising 15 blood samples anonymously obtained from healthy donors through the French blood bank (Lyon France). Subjects were selected so that the mean age of the control group matched that of the study group. For PHAGOBURST®, the MFI cut-off values of cell positivity were established at two standard deviations below the mean of values in the control population.

Assessment of adaptive immunity

Serum immunoglobulin (Ig) G, IgA and IgM levels were determined (in g L−1) by kinetic immunonephelometry on an IMMAGE Immunochemistry System® (Beckman Coulter, Hialeah, FL, USA).

The phenotypic characterization of lymphocyte subsets was performed using whole blood and standard immunofluorescence and flow cytometry technology. The following monoclonal antibodies were used for staining: CD4-FITC, CD8-phycoerythrin (PE), CD45RA-PE, CD19-PE, CD3-FITC, CD16-PE, CD56-PE, CD25-PE-cyanin-5 (PC5) and CD127-PE (from Beckman Coulter, Villepinte, France), and CD27-FITC and HLA-DR-FITC (from BD Biosciences, Le Pont De Claix, France). Cells were analysed using a Cytomics FC 500 flow cytometer with cxp software (Beckman Coulter, Hialeah, FL, USA). A specific single platform based on lyse/no wash four-colour flow cytometry was used for quantification of total lymphocytes and CD3+, CD4+ and CD8+ T lymphocytes. T, T CD4, T CD8, natural killer (NK) and B lymphocytes were defined according to the following respective phenotypes: CD3+, CD4+, CD3+CD8+, CD16+CD56+CD3− and CD19+. Naïve CD4 and CD8 T lymphocytes were defined as CD45RA+CD4+ and CD27+CD45RA+CD8+, respectively. Regulatory T lymphocytes (Tregs) were assessed by counting the CD4+CD25+CD127− population. CD4+HLA-DR+ and CD8+ HLA-DR+ cells were considered as activated CD4 and CD8 T cells, respectively.

The results of all these tests (IgG, IgA and IgM levels and lymphocyte subset absolute values and proportions) were compared with those of a control group of 42 healthy subjects. We have access to a local database of results obtained from healthy subjects of both sexes within a large age range, and this database is used in our laboratory to determine normal ranges for interpretation of the test results. The control group was retrospectively selected based on the recorded results of subjects matched to the patients by age and gender.

T lymphocyte function was studied using a standard in vitro proliferation method. Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized whole blood by density gradient centrifugation and then stimulated for 2 days with 10 μg mL−1 phytohaemaglutinin A (PHA), or for 5 days with 10, 5 or 1 μg mL−1 tetanus toxoid (TT) or tuberculin purified protein derivative (PPD). Next, 2 μCi [3H]-thymidine was added for the final 18 h of incubation. The level of [3H]-thymidine uptake, reflecting the degree of lymphocyte proliferation, was assessed using a liquid β scintillation counter (TopCount NTX®, PerkinElmer, Courtaboeuf, France) and expressed in mean counts per min (cpm). The TT or PPD concentration giving the highest proliferation value was recorded, and the stimulation index (mean cpm in the presence of stimulator/background mean cpm) was calculated. Mean cpm of ≥70 000 for PHA, and a stimulation index ≥3 with mean cpm of ≥10 000 for TT and PPD were considered normal values.

For all tests, we only considered whether or not proliferation criteria were fulfilled in the patient group and we did not use a control group.

Clinical characteristics

Data regarding HHT symptoms, medical history, treatments, biological measurements (haemoglobin level and creatininaemia) and clinical status at the time of the study were collected. In order to determine a link between immune parameters and infectious susceptibility, we considered a history of severe infection as a hallmark of relative immunodeficiency. Infectious events that required hospitalization were classified as severe. Any other infectious event was classified as nonsevere. Unclear causes of lymphopenia were investigated in medical records by establishing concomitant conditions amongst the following, in accordance with previously reported criteria [14]: malnutrition, the presence of or recent treatment for solid cancer or lymphoma, immunosuppressant or corticosteroid therapy, HIV infection, systemic lupus erythematosus, splenomegaly, renal insufficiency and granulomatous disease (sarcoidosis and Wegener's granulomatosis).

Statistical analysis

Immunological results are presented as box plots with individual values. Comparisons between groups were made using the Mann–Whitney U or Student's t-tests (with or without Welch correction) as appropriate. Analyses were performed using prism 5.03 (GraphPad Software, La Jolla, CA, USA).

Clinical characteristics are presented as tables.The chi-squared, Fisher's exact, Student's t (with or without Welch correction), Spearman's rank and Mann–Whitney U-tests were used as appropriate. Multivariate analysis was performed by taking into account parameters with a P value of <0.2 in the univariate analysis or factors known to increase the risk of infection in HHT. All tests were performed using statistical analysis Software version 9.2 (SAS Institute, Cary, NC, USA).

A P value of <0.05 was considered statistically significant.


A total of 42 consecutive adult patients were included in the study. Patient characteristics are shown in Table 1. The studied sample of patients was rather similar to the entire cohort followed in our centre in terms of age, sex ratio, HHT mutations and localization of AVMs. Only four patients were relatives (two sets of brothers). All patients underwent at least one thoracic assessment by computed tomography. Overall, a history of 22 severe infectious diseases was recorded in 13 (31%) patients, including: cutaneous or soft tissue infection (= 6), infectious arthritis or osteitis (= 4), cerebral abscess (= 3), spondylodiscitis (= 3), pneumonia (= 3) and others (prostatitis, tuberculosis and diverticulitis = 1 each). Furthermore, 43% of patients reported at least one nonsevere infectious disease (data not shown). Nine patients had a haemoglobin level below 120 g L−1 (<100 g L−1 in one case), five of whom were not receiving iron supplementation.

Table 1. Characteristics of patients with HHT (= 42)
  1. n-p/n-s, number of patients with confirmed AVM/number of patients screened for this type of AVM; ID, infectious disease.

Age in years, mean (min–max)50.8(19–82)
Female gender, n (%)23(54.8)
Identified mutation, n (%)
 ALK1 22(52.4)
 ENG 13(31.0)
 SMAD4 2(4.8)
 None identified5(11.9)
Frequency of epistaxis at the time of study, n (%)
Iron supplementation (yes), n (%)20(47.6)
Pulmonary AVM, n-p/n-s (%)16/42(38.1)
Cerebral AVM, n-p/n-s (%)5/32(15.6)
Hepatic AVM, n-p/n-s (%)16/38(42.1)
Gastrointestinal tract AVM, n-p/n-s (%)8/21(38.1)
Number of severe IDs, n (%)
Presence of an unclear cause of lymphopenia, n (%)3(7.1)
Haemoglobin level, g L−1, mean (min–max)134 (94–175)
Creatininemia, μmol L−1, mean (min–max)78.3(52–115)
Lymphocyte count, cells mm−3, mean (min–max)1358.3(369–2868)

Phagocytic and oxidative burst activities of PMNs and monocytes, and chemotactic activity of PMNs

Phagocytic activity and oxidative burst values were within the normal ranges for most patients, as determined by values in the control subjects (Fig. 1). The mean values were not statistically different, and the flow cytometry response profiles were similar to the reference population. Moreover, there were no significant differences in chemotactic activity of PMNs for fMLP between patients with HHT and control subjects. The retrospective power to detect a significant difference in oxidative burst of PMNs was 96.37% (HHT population: mean 99%, standard deviation 2%; control subjects: mean 96%, standard deviation 4%).

Figure 1.

Polymorphonuclear leucocyte and monocyte phagocytosis (a) and oxidative burst (b) and PMN chemotaxis index (c) in 42 patients with HHT (HHT) and 15 healthy control subjects (HC) matched in mean age. Results in HHT (white boxes) and HC (grey boxes) are presented as box plots and individual values. Comparisons between groups were performed using the Mann–Whitney U-test or Student's t-test.

Lymphocyte populations in HHT

A significant reduction in lymphocyte count was observed in the HHT population (mean 1358 cells μL−1 vs. 1825 cells μL−1 in the control population, < 0.001) (Fig. 2a,b). In total, 18 subjects (43%) were lymphopenic according to the standard threshold of 1200 cells μL−1, and lymphocyte count was <600 cells μL−1 in five patients. Lymphopenia of patients with HHT mainly affected T cells (mean 961 cells μL−1 vs. 1383 cells μL−1 for control subjects, < 0.0001) and to a lesser extent NK cells (mean 169 cells μL−1 vs. 221 cells μL−1 for control subjects, < 0.01). The B lymphocyte count was unaffected. Mean CD4 and CD8 lymphocyte counts were both significantly lower in the HHT population (CD4: 642 cells μL−1 vs. 832 cells μL−1 for control subjects, < 0.001; CD8: 295 cells μL−1 vs. 501 cells μL−1 for control subjects, < 0.0001) as were the mean proportions of naïve subpopulations (CD4: 37% vs. 48%, < 0.01; CD8 37% vs. 52%; < 0.01). Treg and activated T lymphocyte percentages were not altered, although in some patients, the proportions of DR+CD4+ and DR+CD8+ T lymphocytes were clearly higher.

Figure 2.

Absolute counts of total, CD3+, CD4+, CD3+CD8+, CD19+ and NK lymphocytes (a) and percentages of naïve and activated CD4+, CD8+ and CD25+CD127-CD4+ lymphocytes (b) in 42 patients with HHT (HHT) and in age- and gender-matched healthy control subjects (HC). Results in HHT (white boxes) and HC (grey boxes) are presented as box plots and individual values. Comparisons between groups were performed using the Mann–Whitney U-test or Student's t-test, and P-values are shown.

Next, we analysed the characteristics of the 18 lymphopenic patients, compared with those of the nonlymphopenic HHT subjects. The results are shown in Table 2. Despite the small size of the study population, it is clear that iron intake at the time of the study was about two times more common in the lymphopenic group and reached the significance threshold. Moreover, this group had a trend towards more frequent epistaxis and contained a larger proportion of men. Proportions of subjects with a history of infection and the causes of lymphopenia were not different between the two populations. Untreated pulmonary AVM was not more common amongst lymphopenic subjects (data not shown). Lymphopenia was not significantly greater in the nine subjects with haemoglobin levels below 120 g L−1 than in the others patients (median 1101 cells mm−3 vs. 1298 cells mm−3, = 0.26, Mann–Whitney U-test).

Table 2. Comparison between lymphopenic and nonlymphopenic patients with HHT
CharacteristicsLymphopeniaNo lymphopenia P
  1. n-p/n-s, number of patients with confirmed AVM/number of patients screened for this type of AVM.

  2. P-values were calculated using chi-squared, Fisher's exact or Student's t-tests, as appropriate.

Total number of patients, n1824 
Age (years), mean (min–max)51.83 (20–74)49.96 (19–82)0.69
Female sex, n (%)7 (38.9)16 (66.7)0.07
Identified mutation, n (%)
 ALK1 10 (55.6)12 (50.0)0.87
 ENG 6 (33.3)7 (29.2) 
 SMAD4 0 (0)2 (8.3) 
 None identified2 (11.1)3 (12.5) 
Frequency of epistaxis at the time of study, n (%)
 Daily10 (55.6)6 (25.0)0.07
 Weekly/monthly5 (27.8)15 (62.5) 
 Rarely3 (16.7)3 (12.5) 
Iron supplementation at the time of the study, n (%)13 (72.2)7 (29.2)0.006
Pulmonary AVM, n-p/n-s (%)6/18 (33.3)10/24 (62.5)0.58
Cerebral AVM, n-p/n-s (%)2/14 (14.3)3/18 (16.7)1.00
Hepatic AVM, n-p/n-s (%)6/16 (37.5)10/22 (45.5)0.62
Gastrointestinal tract AVM, n-p/n-s (%)4/10 (40.0)4/11 (36.4)1.00
Number of severe infectious events, n (%)6 (33.3)7 (29.2)0.77
Other disease-causing lymphopenia, n (%)0 (0)3 (12.5)0.25
Haemoglobin level, g/L mean (min–max)129.75 (94–175)137.09 (110–172)0.29

There were no significant collinear variations between lymphocyte count and age, haemoglobin level or creatininaemia (Spearman's rank test; data not shown).

T lymphocyte function in HHT

All tested samples showed a significant proliferation relative to laboratory normal values after stimulation with PHA (Table S1). The proliferation capacity was found to be normal in response to at least either TT or PPD stimulation in all blood samples, and a response to both was observed in 79% of subjects. Due to some cases of severe lymphopenia, the number of lymphocytes required for these tests could not be achieved for all subjects: only 38 blood samples were tested with PHA, 37 with TT and 35 with PPD.

Serum IgG, IgA and IgM levels

Mean serum levels of IgG and IgA were significantly increased in patients with HHT compared with the age- and sex-matched control population (IgG: 11.26 g L−1 vs. 9.33 g L−1, < 0.0001; IgA: 2.65 g L−1 vs. 1.88 g L−1, < 0.0001) (Fig. 3).

Figure 3.

Serum levels of immunoglobulins in 42 patients with HHT (HHT) and age- and gender-matched healthy control subjects (HC). Results in HHT (white boxes) and HC (grey boxes) are presented as box plots and individual values. Comparisons between groups were performed using the Mann–Whitney U-test or Student's t-test, and P-values are shown.

Analysis according to infectious history

Results of the univariate analysis of clinical and immunological data are showed in Table 3a and b; the multivariate analysis results are presented in Table S2a and b. As expected, the group with a history of infection was about 10 years older than those without previous infection in both analyses. Pulmonary AVM, a well-documented risk factor for infection in patients with HHT, was found to be significantly associated with history of severe infection in the multivariate analysis, with an odds ratio of 0.06 [95% confidence interval (CI) 0.01–0.55), but not in the univariate analysis. With regard to the immunological data, a significantly lower level of IgM was observed in patients with a history of severe infection, in the univariate as well as in the multivariate analysis (odds ratio of 0.06, 95% CI 0.01–0.55). The four patients in the present study with an IgM level <0.5 g L−1 (the threshold for IgM deficiency of our laboratory) had all experienced at least one severe infectious event.

Table 3. Comparison between HHT patients with or without a history of severe infection
 History of severe infection No history of severe infection P
(a) Clinical data
Total number of patients, n1329 
Age in years, mean (min–max)57.5 (33–82)47.7 (19–74)0.04
Female gender, (%)8 (61.5) 15 (51.7)0.55
Identified mutation, n (%)0.45
 ALK1 5 (38.5)17 (58.6) 
 ENG 6 (46.2)8 (27.6) 
 SMAD4 1 (7.7)1 (3.5) 
 None identified1 (7.7)3 (10.3) 
Epistaxis frequency at the time of study, n (%)0.74
Daily6 (46.2)10 (34.5) 
Weekly/monthly5 (38.5)15 (51.7) 
Rarely2 (15.4)4 (13.8) 
Iron supplementation, n (%)8 (61.5)12 (41.4)0.22
Pulmonary AVM, n-p/n-s (%)7/13 (53.9)9/29 (31.0)0.16
Cerebral AVM, n-p/n-s (%)2/12 (16.7)3/20 (15.0)1.00
Hepatic AVM, n-p/n-s (%)6/13 (46.2)10/25 (40.0)0.72
Gastrointestinal tract AVM, n+/n-s (%)1/6 (16.7)7/15 (46.7)0.34
Haemoglobin level, g/L, mean (min–max)131.7 (104–152)134.9 (94–175)0.67
Creatininaemia, μmol/L, mean (min–max)74.7 (52–100)80.2 (54–115)0.32
Lymphocyte count, cells/mm−3, mean (min–max)1245.3 (401–2178)1409 (369–2868)0.42
Parameter, median (Q25–Q75)History of severe infection No history of severe infection P
  1. n-p/n-s, number of patients with confirmed AVM/number of patients screened for this type of AVM; Q25, first quartile; Q75, third quartile.

  2. P-values were calculated using chi-squared, Fisher's exact, Student t or Mann–Whitney U-tests, as appropriate.

(b) Immunological data
Total lymphocytes1271(992–1516)1298(1057–1662)0.42
T lymphocytes787(445–1037)981(698–1100)0.30
T CD4 lymphocytes543(343–713)657(499–779)0.23
Naïve T CD4 lymphocytes39(22–55)38(32–58)0.95
Activated T CD4 lymphocytes8.6(7.8–13.8)7.3(6.3–10.1)0.08
T CD8 lymphocytes222(159–318)289(169–395)0.33
Naïve T CD8 lymphocytes27(22–47)39(18–48)0.58
Activated T CD8 lymphocytes6.2(3.1–12.5)4.9(2.3–12)0.57
Regulatory T lymphocytes6.3(4.9–8.4)6.6(5.7–7.9)0.99
NK lymphocytes169(83–253)135(87–207)0.69
B lymphocytes166(119–277)180(143–238)0.85
Immunoglobulin G11.9(10–12.2)10.9(9.49–12.7)0.94
Immunoglobulin A2.3(2–3.1)2.7(2–3.2)0.58
Immunoglobulin M0.6(0.5–0.9)1.2(0.8–1.4)0.002
Oxidative burst of PMNs98.8(97.6–99.7)99.3(98–99.7)0.95
Oxidative burst of monocytes92(91–94.1)94.2(89.7–97.5)0.55
Phagocytosis of PMNs99.4(98.2–99.6)99.5(97.6–99.6)0.58
Phagocytosis of monocytes97.3(94.9–97.6)96.7(95.6–97.3)0.60
Chemotaxis of PMNs6.4(5–8)5.5(4–7.6)0.25
TT stimulation index26.5(20.5–58.9)26.4(10.5–40.8)0.55
PPD stimulation index39.7(15.5–51.3)21.3(17.1–37.2)0.39


We did not observe any abnormalities of innate immunity mediated by PMNs and monocytes amongst patients with HHT. Nevertheless, immune deficiency was suggested by the high risk of bacterial, especially staphylococcal, infections and abscesses observed in these patients. Furthermore, Cirulli et al. have previously described reduced oxidative burst and phagocytic activity in HHT subjects [12]. We used exactly the same commercially available assay kits as these authors but in a larger patient population; therefore, the negative result in the present study was unexpected. A possible explanation for the discrepancy may be a difference in the state of iron repletion of the patients in the two studies; indeed, iron deficiency is known to alter both oxidative burst and phagocytic activity of PMNs [15, 16]. Iron deficiency is commonly associated with HHT and was not specifically assessed in these two studies. Nonetheless, in the present study, the mean level of haemoglobin was normal and the proportion of anaemic patients with HHT was relatively low. The immunological consequences of iron deficiency are probably minor. Moreover, chemotaxis of PMNs was not altered in these patients confirming our findings in favour of conserved granulocytic functions.

Our results highlight several abnormalities of adaptive immunity which, to our knowledge, have not been previously described. In our population of 42 patients, these abnormalities do not seem to be linked to a higher infection rate, but this needs to be confirmed in a larger cohort. External causes must be suspected first. Indeed, the link between iron intake and lymphopenia suggests the direct toxicity of iron. A high and prolonged intake of iron, as is the case for patients with HHT, most of whom suffer from chronic iron deficiency anaemia, could increase the level of oxidative stress [17, 18]. Analogous to exercise-induced lymphopenia, an excess of reactive oxygen species may result in lymphocyte DNA damage leading to lymphopenia [19, 20]. In the same way, the moderate rise in IgA and IgG could be related to a subclinical infection in some patients at the time of study. For example, recurrent rhinosinusitis, a complication of the frequent nasal packing in patients with HHT, could act as a chronic source of low-level inflammation. This could explain the high level of IgA and IgG, as well as the high number of activated lymphocytes found in some of the subjects in the present study.

Nevertheless, an intrinsic cause directly due to ENG or ALK1 mutations could also explain the immunological findings. One of the key roles of the BMP9/ALK1/endoglin pathway seems to be the regulation of the CXCR4/SDF-1 chemotactic axis on endothelial cells during angiogenesis [21]. It has been demonstrated that endoglin and ALK1 are overexpressed on lymphocytes [11] and monocytes [10] after an antigenic stimulation, and that this overexpression is disturbed in HHT. Moreover, chemotactic abnormalities related to CXCR4/SDF1 on PBMCs in patients with HHT have been reported [22]. Therefore, as in endothelial cells, it is likely that a defect of endoglin and/or ALK1 overexpression will lead to dysregulation of the CXCR4/SDF1 chemotactic axis of HHT immune cells. This axis regulates the trafficking and homing abilities of lymphopoietic cells as well as thymic maturation and accumulation of these cells in inflamed tissues [23]. A CXCR4/SDF1-dependent dysregulation of immune cells could explain the lymphopenia observed in the present study and may also be linked to the high rate of infection reported in HHT.

Consistent with our results, it has previously been demonstrated that pulmonary AVM is a risk factor for infection, particularly brain abscesses, in HHT [9, 24]. The increased risk in the case of bacteraemia is probably due to the loss of the mechanical filter effect of the lung, which decreases after occlusion of the pulmonary AVM [25]. The small size of our sample most probably explains why we did not observe a significantly longer duration of epistaxis associated with an infection history, as previously described [9].

A low IgM level seems to be associated with the risk of infection observed in HHT to the same degree as the risk associated with pulmonary AVM. Secreted IgM antibodies are pentameric with a high capacity for agglutination and complement activation. There are two types of IgM antibodies, natural and immune. These antibodies are predominantly produced by the B1 lymphocyte subset in mice, but the existence of such a B subset in humans remains controversial [26]. IgM antibodies are the primary humoral component elicited by T-cell-independent antigens and are involved in the defence against viral, bacterial and parasitic pathogens [27, 28]. The clinical characteristics of selective IgM deficiency are not well defined, but case reports [29] and short series [30] report bacterial infections, especially with Staphyloccocus aureus, which is reasonably consistent with the severe infections observed in patients with HHT. Moreover, the loss of IgM-producing B cells seems to be correlated with disease level in common variable immunodeficiency [31]. The mechanism of the decrease in IgM amongst some of the patients with HHT in the present study is not easy to explain, and our population was too small to investigate this further.

The limitations of this study should be considered. First, iron status was not specifically investigated although it is known to interfere with many immune parameters and is frequently abnormal in patients with HHT. Secondly, most of our functional tests rely on highly immunogenic reactants resulting in strong immune responses. They may be inadequate to demonstrate in vivo dysfunction on a small scale as well as defective regulation by the cytokine environment. However, the advantage of these tests is that they are performed regularly in most academic immunology laboratories and have demonstrated usefulness for the exploration of primary immunodeficiency [32, 33].

In conclusion, the study of a large number of immunological parameters in patients with HHT has demonstrated a global lymphopenia that affects T and NK cells, with a greater impact on the naïve T lymphocyte population, associated with a rise in IgG and IgA levels. Lymphopenia seems to be linked to iron status and is independent of the risk of infection. Moreover, our data suggest that a low IgM level is a potential novel risk factor for severe infection. These findings provide a basis for more specific clinical and immunological studies to generate further understanding of the still poorly understood pathophysiology of HHT.

Conflict of interest statement

The authors have no conflict of interest to declare.


This work was supported by the Association Française Maladie de Rendu-Osler-Weber (AMRO France – HHT).


The authors would like to thank Dr S. Riviere and Dr M. Guilhem for their valuable proofreading of the manuscript.

Authors' contributions

CM, HP and SD-G designed and performed the study; BC and AG analysed the data; and AG and CM wrote the report.

All authors have seen and approved the final version of the manuscript.