Correlation of Local Interleukin-8 with Immunoglobulin A against Gardnerella vaginalis Hemolysin and with Prolidase and Sialidase Levels in Women with Bacterial Vaginosis

Abstract

Mucosal immune system activation may represent a critical determinant of adverse consequences associated with bacterial vaginosis (BV), such as sexual human immunodeficiency virus transmission, upper genital tract infections, postsurgical infections, and adverse pregnancy outcomes. Concentrations of sialidase, prolidase, and anti-Gardnerella vaginalis hemolysin (Gvh) immunoglobulin A (IgA) were higher in vaginal fluids of 75 fertile women with BV, compared with concentrations in vaginal fluids of 85 healthy control subjects. Interleukin (IL)-8 levels were positively associated with anti-Gvh IgA response and inversely correlated with high levels of prolidase and sialidase in women with BV. IL-8 concentration was strongly associated with leukocyte count in both healthy and BV-positive women. The absence of leukocytes in most women with BV likely is due to lack of IL-8 induction. Parallel impairment of innate and adaptive mucosal immune factors, likely through microbial hydrolytic effects, may allow for the ascent of microorganisms to the upper genital tract and may facilitate viral infections.

Bacterial vaginosis (BV) is the main vaginal syndrome afflicting fertile and pregnant women [1–5]. BV is a polymicrobial disorder characterized by an overgrowth of several anaerobic or facultative bacteria (mainly Gardnerella vaginalis, Prevotella species, Bacteroides species, Mobiluncus species, gram-positive cocci, and Mycoplasma species) and by a reduction or absence of lactobacillus colonization [1]. Although BV is associated with several adverse outcomes, such as upper genital tract infections, pelvic inflammatory disease, endometritis, postsurgical infections, increased susceptibility to human immunodeficiency virus (HIV) infection, preterm delivery (PTD), and low birth weight (LBW), many basic questions regarding the pathogenesis of BV remain unanswered [1–3, 6–11]. In spite of the massive vaginal exposure to anaerobic microorganisms that alter the normal vaginal ecology, inflammatory signs are usually faint and, for this reason, the term “vaginosis,” not “vaginitis,” was adopted for this syndrome [1]. Some researchers found increased levels of proinflammatory cytokines, such as interleukin (IL)-1 and IL-8, in vaginal fluid of women with BV [12–15]. Such findings are not in line with the clinical observation of scarcity of inflammation. The reasons for the absence of inflammatory signs in most women with BV remain unexplained.

Recently, a mucosal IgA response against the G. vaginalis hemolysin (Gvh) has been found in vaginal fluids of women with BV [16, 17]. Such adaptive antigen-specific immune response is inversely correlated with the presence in the vaginal fluid of sialidase, a microbial enzyme that is associated with BV [18–22].

These results suggest that a subset of women with BV may be exposed to more-aggressive microbial colonization, characterized by high microbial enzymatic activity and diminished anti-Gvh IgA levels. To date, no study has examined the relationship of microbial enzyme activity [22, 23], an antigen-specific IgA response, and proinflammatory chemokine levels in vaginal fluid of BV-positive women. Chemokines are a family of low-molecularweight proinflammatory cytokines that stimulate recruitment of leukocytes. IL-8 is a CXC chemokine specific for neutrophil recruitment that is released by host cells after exposure to various microbial pathogens and is considered to have a major impact in the activation of immune effector cells against the invading microorganisms [24]. In this study, we examined the relationship of anti-Gvh IgA response and prolidase and sialidase levels with factors of the local innate immune response, specifically, IL-8 level and leukocyte count.

Patients and Methods

Study population. Nonpregnant women 19–50 years old were recruited during routine gynecologic examinations for Pap smears in 2 clinics (located in Udine and Trieste) in northern Italy, from December 1999 through May 2001. All women were white, 85% had ⩾13 years of education, and all were eligible to have a Pap test performed (i.e., they did not have bleeding or major lower vaginal tract inflammatory signs). No study participant had a severemedical condition, including malignancies, and none was positive for yeast vaginitis, Trichomonas vaginalis, Neisseria gonorrhoeae, or Chlamydia trachomatis. Further exclusion criteria were partial or total hysterectomy and physiological menopause. All enrolled women denied having had sexual intercourse or using douches or vaginal suppositories in the last 3 days and having used antibiotics in the last 2 weeks. Women with BV were enrolled by clinical evaluation of the presence of all 4 Amsel criteria [4]: typical adherent homogeneous discharge, vaginal pH >4.5, amine odor after addition of 10% KOH to the vaginal specimen, and clue cells on the wet smear. The healthy control group consisted of women without vaginal symptoms in the last 2 months, no evidence of vaginal pathologic condition, vaginal pH <4.5, and no other Amsel criteria present at the time of visit. The healthy and the BV-positive groups did not differ in contraceptive use or menstrual cycle phase. All enrolled women had normal Pap smear results (i.e., no cervical intraepithelial neoplasia).

Vaginal sample collection. Samples were collected from the posterior fornix or lateral vaginal wall (avoiding cervical mucus) of nonbleeding women with an Ayre's spatula, using a nonlubricated speculum, to perform amine odor evaluation and pH determination by paper strip (Merck) and to obtain a smear, which then was Gramstained. Vaginal fluids were retrieved bywashing with 10 mLof sterile saline. Samples were centrifuged at 700 g for 10 min at 4°, and 1-mL aliquots of supernatant were stored immediately at −80°.

Gram-stained smear evaluation. Vaginal flora, clue cells, and leukocyte counts were evaluated on the Gram-stained smear. The mean number of each parameter was determined by evaluation of 5 different fields under oil immersion (magnification, ×1000). Three different bacterial morphotypes—lactobacilli, Gardnerella-like species (including G. vaginalis, Bacteroides species, Prevotella species, and Porphyromonas species), and Mobiluncus species—were quantitatively evaluated according to the Nugent score method [5]. The group of healthy control subjects had Nugent scores of 0–2. The clinically enrolled women with BV were all positive for clue cells and had Nugent scores of 5–10. If there were .30 leukocytes per field, the results were arbitrarily ranked as 40 for statistical calculations.

Anti-Gvh IgA. Anti-Gvh IgA levels were evaluated in the vaginal fluids, according to a procedure described elsewhere [17], using the purified Gvh for coating of the microtiter wells. Intra-assay coefficients of variation were <10%, and interassay coefficients of variation were <15%. As in previous studies [17–19], a cutoff of 380 milliunits optical density (mOD) at 405 nm was adopted for the anti-Gvh IgA response (this cutoff was calculated previously for a large healthy reference control group [19]). To define semiquantitatively the anti-Gvh IgA response, values below the threshold of 380 mOD were considered to be “no anti-Gvh IgA response,” and values ⩾760 mOD (2 times the cutoff) were considered to be “high anti-Gvh IgA response.”

Sialidase levels. Levels of sialidase were determined by incubation of 100 µL of the vaginal fluid sample with the substrate 2-(3′-methoxyphenyl)-N-acetyl-D-neuraminic acid (Sigma-Aldrich), as described elsewhere [18]. After the addition of 4-aminoantipyrine and potassium ferricyanide, the absorbance was read at 492 nm. Specific activity was expressed as nanomoles of methoxyphenol produced, compared with a standard curve of pure methoxyphenol. Intra-assay coefficients of variation were <2%, and interassay coefficients of variation were <16%. On the basis of previous observations, high sialidase levels were defined as ⩾5.00 nmol methoxyphenol [19].

Prolidase levels. Prolidase levels were determined by incubation of 100 µL of the vaginal fluid sample with the substrate L-proline-p-nitroanilide (Sigma-Aldrich) [19]. A duplicate sample without the prolidase substrate served as sample blank. Absorbance (mOD) was read at 405 nm. Intra-assay coefficients of variation were <10%, and interassay coefficients of variation were <14%. On the basis of previous observations, high prolidase levels were defined as those ⩾2000 mOD [21]. For any of the assays described above, each sample giving an optical absorbance of ⩾2000 mOD was diluted 1:1 with sterile saline, and the resulting value was doubled to give the final result.

IL-8 assay. IL-8 was quantified in the vaginal fluid by sandwich-type ELISA, using commercial capture and conjugated detection antibodies. The concentration of human IL-8 (in pg/mL) was determined by interpolation with the standard curve. The vaginal fluid samples were tested in duplicate; if the reading was outside the standard curve, the sample was serially diluted to obtain values in the range of the reference curve. Measurements were performed according to the manufacturer's instructions (Central Laboratory of the Netherlands Red Cross Blood Transfusion Service). The lower detection limit for human IL-8 was 8 pg/mL. Samples below this limit were given the value 0.0 pg/mL for data analysis. All samples except 2 (98.8%) had detectable levels of IL-8. Low, medium, and high levels of IL-8 were defined as those below the 25th percentile, between the 25th and the 75th percentile, and above the 75th percentile values, respectively.

Statistical analysis. Examined parameter values were not normally distributed; thus, nonparametric statistical methods were used for the analyses. Spearman's rank correlation was used to examine the relationship between immune factor and enzymatic activity concentrations in vaginal fluid and vaginal leukocyte counts in the smear; 1-tailed significance of Spearman's ρ coefficient (ρ_s) was reported. The Mann-Whitney U test was used to compare levels of vaginal factors between healthy and BV-positive women and between subgroups of patients with BV. P <.05 was considered to be statistically significant. The software package SPSS (Statistical Package for Social Sciences; SPSS Institute) was used for data analyses.

Results

Overall, 99% (158/160) of the women had detectable levels of IL-8 in their vaginal fluid. The median level of IL-8 was 1028.5 pg/mL, and the 25th and 75th percentiles were 350.7 pg/mL and 3094.0 pg/mL, respectively.

Comparison of diagnostic, microbial, and immune factors in healthy and BV-positive women. The median and interquartile range (25th and 75th percentiles) values for pH; Nugent score; prolidase, sialidase, anti-Gvh IgA, and IL-8 concentrations; and leukocyte counts found in vaginal fluid specimens of 85 healthy and 75 BV-positive nonpregnant women are shown in table 1. These parameters were compared between allBV-positive women and healthy control subjects, using the Mann-Whitney U test. As expected, BV-positive women had significantly higher occurrences of BV diagnostic markers (i.e., pH and Nugent scores). Prolidase and sialidase levels and anti-Gvh IgA responses were significantly higher in BV-positive women. Results indicated that, overall, BV-positive women had a median IL-8 concentration of 1438 pg/mL, whereas healthy control women had a median IL-8 concentration of 828 pg/mL; however, the difference of IL-8 concentrations was not statistically significant (P = .345). The leukocyte counts in the vaginal fluid smears were not significantly different for BV-positive women (median, 5.1 leukocytes/ field), compared with those for healthy women (median, 4.0 leukocytes/field; P = .549).

Table 1.

Bacterial vaginosis (BV) diagnostic parameters, enzymatic activities, and immune factors in vaginal fluid from healthy and BVpositive women.

Correlation of IL-8 concentration with other local immune factors and microbial activities. Within-group analyses indicated that IL-8 was strongly associated with the leukocyte count in both healthy (ρ_s, 0.524; P <.001) and BV-positive women (ρ_s, 0.634; P <.001). In BV-positive women, IL-8 was strongly associated with anti-Gvh IgA (ρ_s, <0.607; P <.001), whereas it was inversely correlated with prolidase (ρ_s, 20.337; P = <.002) and sialidase (ρ_s, 20.214; P = .033) levels.

Correlations of anti-Gvh IgA and leukocytes with microbial enzymes in BV-positive women. Anti-Gvh IgA levels in vaginal fluid of women with BV were positively associated with leukocyte counts (ρ_s, 0.560; P <.001), whereas they were inversely correlated with prolidase (ρ_s, −0.511; P <.001) and sialidase (ρ_s, −0.282; P = .007) levels. The leukocyte counts in the vaginal fluid smears of BV-positive women were inversely correlated with prolidase (ρ_s, µ0.395; P <.001) and sialidase (ρ_s, −0.258; P = .013) levels. Sialidase and prolidase levels were positively correlated with each other (ρ_s, 0.351; P = .001).

Analysis of BV subsets on the basis of IL-8 concentrations. In table 2, we compared BV-positive women with high, medium, and low IL-8 concentrations. High (n = 21) and medium (n = 35) IL-8 responders were not significantly different for any of the above-described parameters, with the exception of leukocyte count (P = .002). Sialidase levels were higher in medium, compared with high, IL-8 responders, although this difference did not reach statistical significance. However, when low IL-8 responders (n = 19) were compared with high IL-8 responders, significant differences were observed. Specifically, low IL-8 responders had 3.3-fold higher prolidase levels (P = .001), nearly 7-fold lower anti-Gvh IgA levels (P <.001), and 25-fold lower leukocyte counts (P <.001). All such comparisons remained statistically significant when low and medium IL-8 responders were compared.

Table 2.

Bacterial vaginosis (BV) diagnostic parameters, enzymatic activities, and immune factors in vaginal fluid of subgroups of BV-positive women, by interleukin (IL)-8 level

Analysis of BV subsets on the basis of anti-Gvh IgA levels. In a further analysis (table 3), BV-positive women were subdivided into 2 groups on the basis of positive (⩾380 mOD) and negative (<380 mOD) anti-Gvh IgA levels (see Patients and Methods). The positive anti-Gvh IgA group (n = 26) had 1.7-fold lower prolidase (P = .005) and 2.3-fold lower sialidase levels than the negative anti-Gvh IgA group (n = 49). The innate immune factor concentrations were more elevated in women with positive anti-Gvh IgA levels. IL-8 concentrations were increased 6.6-fold (P <.001), and leukocyte counts were increased 14-fold (P <.001).

Table 3.

Bacterial vaginosis (BV) parameters, enzymatic activities, and immune factors in vaginal fluid of subgroups of BV-positive women, by anti-Gardnerella vaginalishemolysin (Gvh) IgA level.

Compared with healthy women, BV-positive women whowere positive for anti-Gvh IgA had higher concentrations of IL-8 (median, 2899 vs. 828 pg/mL; P <.001) and higher leukocyte counts (median, 15.7 vs. 4.0 leukocytes/field; P <.001). In contrast, BV-positive women who were negative for anti-Gvh IgA had lower IL-8 levels (median, 440 pg/mL; P = .157) and lower leukocyte counts (median, 1.1 leukocytes/field; P = .087), which was not significantly different from healthy women.

Comparison of subgroups among BV-positive women. In a more refined analysis, extreme subgroups of BV-positive women were compared (table 4). The group with extremely high anti-Gvh IgA levels (⩾760 mOD; see Patients and Methods) was compared with 2 subgroups of anti-Gvh IgA-negative women with high sialidase (⩾5.0 nmol methoxyphenol) or prolidase (⩾2000 mOD) levels. The anti-Gvh IgA-negative group with high sialidase levels (n = 25) had nearly 6-fold lower levels of IL-8 (P <.001), 15-fold lower leukocyte counts (P = .003), nearly 12-fold lower levels of anti-Gvh IgA (P <.001), and 6-fold higher prolidase levels (P <.001) than the group with high anti-Gvh IgA levels (n = 12). BV-positive women with no anti- Gvh IgA and high prolidase levels (n = 13) had 14.5-fold lower IL-8 levels (P <.001), 51-fold lower leukocyte counts (P <.001), and nearly 22-fold lower anti-Gvh IgA levels (P <.001) than did the group with high anti-Gvh IgA. This BV subgroup had 13-fold lower leukocyte counts (P , .001) and 3.5-fold lower IL-8 concentrations (P = .019) than did healthy women. Vaginal pH and Nugent score values were not different among the examined subgroups of BV-positive women, with the exception of the high sialidase group, which differed from the high anti-Gvh IgA group (P = .004, and P = .033, respectively; table 4).

Table 4.

Bacterial vaginosis (BV) diagnostic parameters, enzymatic activities, and immune factors in vaginal fluid of subgroups of BV-positive women with high anti-Gardnerella vaginalis hemolysin (Gvh) IgA levels (⩾760 milliunits optical density [mOD]), with no anti-Gvh IgA (<380 mOD) and high sialidase levels (⩾5 nmol), or with high prolidase levels (⩾2000 mOD).

Discussion

The present study identified a strong positive correlation between levels of the proinflammatory chemokine IL-8 and anti- Gvh IgA in vaginal fluids of nonpregnant women with BV. In addition, vaginal IL-8 was strongly associated with the number of leukocytes present in vaginal fluid in both healthy and BVpositive women. In our study, both IL-8 concentration and leukocyte count were not significantly greater in BV-positivewomen, compared with healthy control subjects. The lack of IL-8 increase and leukocyte recruitment in women with BV was correlated with vaginal prolidase and sialidase concentrations. We found a wide range of IL-8 concentrations in women positive for BV. High levels (⩾3094 pg/mL, 75th percentile) of IL-8 were found in 28% (21 of 75), and low levels (<351 pg/mL, 25th percentile) were found in 25% (19 of 75) of women with BV. Impairment of IL-8 induction amongwomenwith BV(as lowIL-8 responders group) is characterized by low number of leukocytes, low anti- Gvh IgA levels, and high prolidase levels.

On the basis of antigen-specific immune response, BV-positive women demonstrating positive anti-Gvh IgA levels had nearly 7-fold higher IL-8 concentrations and 14-fold higher leukocyte counts, compared with women who were negative for anti-Gvh IgA. Our findings suggest that positive anti-Gvh IgA levels identify women with BV showing activated innate immunity. In fact, these women also had higher IL-8 concentrations and leukocyte counts, compared with healthy women. In contrast, BV-positive women with an impaired adaptive immune response (identified as negative anti-Gvh IgA levels) had high levels of both prolidase and sialidase.

The subgroup of BV-positive women characterized by elevated anti-Gvh IgA levels had nearly 4-fold higher median IL-8 concentrations and leukocyte counts, compared with healthy women. BV-positive, anti-Gvh IgA-negative women with high levels of sialidase had a concomitant lack of IL-8 induction and a 4-fold reduction in leukocyte recruitment, comparedwith healthy women. BV-positive, anti-Gvh IgA-negative women with high levels of prolidase showed a 3-fold reduction in IL-8 concentration and a 13-fold reduction in leukocyte count, compared with healthy women. Women with this profile thus constitute a subgroup of BV-positive women with more-severe innate factor impairment. Further studies will assess whether such parallel adaptive and innate immune response impairment is associated with adverse BV consequences, such as increased risk of sexual HIV acquisition, upper genital tract infections, and adverse pregnancy outcomes.

Our study suggests that the lack of leukocyte recruitment observed in some women with BV may be associated with the absence of IL-8 chemotactic effect. Overall, our findings provide evidence that a failure in the induction of the vaginal innate immune response occurs in a subset ofwomen with BV. In fact, immunologic boosting by microorganism factors is considered to be an important event in sustaining sufficient immunity to prevent pathogenic consequences of infection [25–29]. It is conceivable that the impairment of the innate immune response that we found in some women with BV, especially when accompanied by the failure of the adaptive immune response, could be responsible for allowing the ascension of the microbes to the upper genital tract and for failure of the local immune system to prevent sexual acquisition of HIV infection.

Previous studies demonstrated that IL-8 concentrations in cervicovaginal secretions are associated with the leukocyte counts [30] and that vaginal pathogens such as T. vaginalis [31], group B streptococci [32], and Candida albicans [33] are able to induce IL-8. We found that the proinflammatory cytokine increase not only induces innate immune response (through leukocyte recruitment) but also coincides with increased vaginal antigen-specific immune levels. It remains to be assessed whether such coupling of the innate and adaptive immune response is mediated by other cytokines [34]. Ongoing studies will clarify this issue. Indeed, IL-8 concentrations were found to be associated with IL-1 levels in vaginal secretions [35, 36], and increased levels of vaginal IL-1β and tumor necrosis factor-a were found in women with BV [14].

The present study differs from previously published studies of vaginal IL-8 concentrations in BV-positive women, because pregnant women were the subjects of the other studies [12, 30]. IL-8 is a potent chemoattractant for neutrophils but also is a pleiotropic cytokine with several functions, produced by many different kinds of cells in response to several stimuli. IL-8 plays physiological roles in gestation and parturition not related to infection [37–39]. In our opinion, the evaluation of IL-8 concentrations in nonpregnant women thus constitutes a more reliable model to assess the immunomodulatory effects associated with the anaerobic alteration of the vaginal ecology (i.e., BV). Our study points out that elevation of IL-8 concentration is associated with the increase of leukocytes in the vaginal secretion. It is likely that several mechanisms, not always connected to infection, regulate the levels of cytokines in the vaginal environment [40]. Different stimuli, including mechanical and endogenous ones, can trigger polymorphonuclear leukocyte accumulation and cytokine production, even in the absence of infection [41, 42]. It remains to be established whether the failure of the IL-8 response associated with high levels of enzymatic activities is due to a direct degradation of the chemokine by microbial factors, an impairment of IL-8 local production by damage of the producing cells, and/or absence of adequate cytokines' cascade effects [43].

In previous studies, the presence of high levels of vaginal sialidase activity was associated with low anti-Gvh IgA response and with mucosal IgA degradation in women with BV [18, 19]. Such an impairment of vaginal IgA likely is due to the hydrolytic action of an array of factors released by anaerobic bacteria [20, 44], and it is likely that, besides IgA, other proteins and host factors of themucosal defense can be compromised and/or degraded.

To our knowledge, no previous study has investigated the relationship of microbial enzyme activities (specifically, sialidase and prolidase levels) and IL-8 levels in vaginal fluid of women with BV. Both enzymes were found to be associated with BV in previous studies [18, 19, 22, 23]. Sialidases are enzymes released by several pathogens involved in urogenital diseases. Sialidases are specific for bacterial vaginosis and are correlated with high titers of anaerobes, mainly Bacteroides species and Prevotella species [22]. Persistent sialidase activity in pregnant women has been associated with increased risk of LBW and PTD [44]. A very recent study reported that sialidase activity at 17 weeks' gestation is significantly associated with LBW [21]. Sialidases exert their effect by cleaving the terminal sialic acid off various glycoproteins, such as mucins, fibronectins, and cell adhesion molecules. Changes in sialoglycoconjugates of vaginal host factors can potentially affect the local immune response through lifetime and biologic activity modulation of several sialylated immune mediators, including interleukins (e.g., interferon), immunoglobulins (especially IgA), proteinase tissue inhibitors, lactoferrin, and various cellular receptors [45]. In the present study, we confirmed that sialidase levels in vaginal fluids of women with BV are inversely associated with the mounting of a specific IgA vaginal response against the cytolysin of G. vaginalis [18, 19]. In addition, we demonstrated that high levels of sialidase are associated with reduced vaginal levels of innate immune factors.

Prolidases are proteolytic enzymes associated with BV [19, 23]. They can affect extracellular matrix components, such as mucins, and play roles in modulation of immune factors' activity [46]. We found that high levels of prolidase are strongly inversely correlated with the ability of the host to elicit a mucosal innate and antigen-specific IgA levels. Several microorganisms implicated in vaginal colonization (i.e., G. vaginalis, which is the main bacterium associated with BV, Mobiluncus species, and Peptostreptococcus species) are able to produce prolidases in vitro [20, 23]. In the present study, we found a strong association between prolidase and sialidase levels in women with BV. It is noteworthy that both sialidases and prolidases are enzymes potentially able to perturb the biofilm that protects the mucosal surface and to attack the cervical mucin plug that protects the uterine cavity in pregnancy [44]. We found that high levels of hydrolytic activity are associated with impairment of adaptive and innate immune responses in women with BV, as evaluated by anti-Gvh IgA and IL-8 concentrations. It is conceivable that, in BV-positive women showing high levels of antigen-specific response, most of the innate immune mediators are, in parallel, fully active.

An unequal risk for BV acquisition and/or recurrence could derive from different mucosal immune host abilities and/or capability of invading microbes to produce factors that inactivate the local immune defense. Recent evidence shows that individual genetic backgrounds can affect chemokine production [47]; future studies of host immune response factors may advance our understanding of BV pathogenesis and assess whether immune differences can account for the large differences observed in BV prevalence among ethnic groups and in interindividual BV disease course and adverse outcomes.

Acknowledgments

We thank Poul Thorsen (Danish Epidemiology Sciences Centre, Aarhus University, Aarhus, Denmark), for preliminary statistical evaluations; Jennifer Culhane (Jefferson University, Philadelphia), for critical reviewing of the manuscript; Nunziata Verdolina (Azienda Servizi Sanitari 4, Udine, Italy), for her assistance; and Marina Cudin (Department of Biomedical Sciences and Technologies, University of Udine, Udine), for laboratory technical help.

Presented in part: BV 2000, 3d International Meeting on Bacterial Vaginosis, Ystad, Sweden, 14–17 September 2000 (abstract 64).

Appropriate informed consent was obtained from the women enrolled in this study, and clinical research was conducted in accordance with guidelines for human experimentation of the authors' institutions.

Financial support: Ministero per la Ricerca Scientifica e Tecnologica of Italy; Regione Friuli Venezia Giulia grant; University of Udine; University of Trieste; Istituto Ricovero e Cura a Carattere Scientifico Burlo Garofolo Hospital.