Chromogranin-A production and fragmentation in patients with Takayasu arteritis

Fifty-one patients fulfilling the American College of Rheumatology (ACR) criteria for TA [27] were evaluated in 2013 at the San Raffaele Scientific Institute in Milan. All patients were extensively evaluated to exclude TA mimics, as previously described [10]. We performed a cross-sectional analysis of disease biomarkers between April and August 2013, excluding nine patients that did not attend a visit in that period or had moderate to severe renal or heart failure, and carried out a longitudinal evaluation of arterial involvement as assessed by serial radiological imaging follow-up (see below). The final sample consisted of 42 consecutive patients with a median age of 46 years (range 23 to 66 years). Twenty age-matched women served as healthy controls (HCs). All subjects gave written informed consent for participation in the study and the Institutional Review Board of the San Raffaele University Hospital (Comitato Etico dell’Ospedale San Raffaele, Milano, Italy) approved the study protocol (protocol “Autoimmuno-mol”, PI Angelo Manfredi). This study was conducted in accordance with the Declaration of Helsinki.

Systemic activity of TA was evaluated according to National Institutes of Health (NIH) criteria [5]. Arterial hypertension was defined in accordance with international guidelines. Cardiac involvement was defined as significant aortic regurgitation or presence of systolic or diastolic dysfunction. Three inflammatory reactants were evaluated: erythrocyte sedimentation rate (ESR), and concentration of C-reactive protein (CRP) and of pentraxin-3 (PTX3). ESR was assessed by the Westergren method, CRP by a latex-enhanced immuno-turbidimetric assay (ADVIA Chemistry System, Bayer AG, Leverkusen Germany), and plasma PTX3 by ELISAs as previously described [10]. Four sandwich ELISAs were used to detect CgA-derived polypeptides (Additional file 1: Figure S1) [15]: (1) 439 ELISA, that specifically detects CgA₄₃₉; (2) 436/439 ELISA, that detects the peptides with the N-terminal domain and the C-terminal region, with or without the 436–439 residues; (3) 436/439 + FRs-ELISA, that detects peptides with the N-terminal domain and the central region; (4) 76 ELISA, that detects VS-1. CgA_tot was computed summing the results of 436/439 + FRs-ELISA and 76 ELISA, thus including all the CgA-derived polypeptides containing the N-terminal region. Given that both CgA₄₃₉ and VS-1 have anti-angiogenic properties, we pooled these two polypeptides by summing the ranks of the respective peptides in our sample. The levels of CgA-FRs were computed subtracting the results of 436/439 + FRs ELISA and 436/439 ELISA. To study the regulation of CgA processing, we evaluated the ratios of CgA₄₃₉/CgA_tot, CgA-FRs/CgA_tot and VS-1/CgA_tot. There were no samples missing for CRP, PTX3 and CgA or derived peptides assessment. Three samples were missing for ESR. Samples were analysed in a random order and in a blinded manner.

All patients were regularly monitored with magnetic resonance angiography (MRA) and vascular ultrasonography (US). Two patients had contraindications to MRA and underwent computed tomography angiography (CTA). MRA was performed with a 1.5-T magnetic resonance whole body scanner (Achieva Nova Master; Philips Medical Systems, Best, The Netherlands) with phased-array head and neck or phased-array thoracic dedicated coils. Morphologic sequences [proton-density weighted (PD) black blood turbo spin echo (TSE): field of view (FOV) 260 × 152; acquisition (Acq) matrix = 260 × 264; reconstruction (Recon) matrix = 528; Acq voxel MPS = 1.00/0.95/6.00; Recon voxel MPS = 0.49/0.49/600; echo time/repetition time (TE/TR) = 20/2 beats; electrocardiogram (ECG) triggered; in expiratory breath hold of approximately 10 sec] were performed to evaluate vessel wall thickening. First-pass MRA dynamic sequences targeted on thoraco-abdominal aorta and supra-aortic trunks (FOV 450 × 390; Acq matrix = 352 × 200; Recon matrix = 670; Acq voxel = 1.28/1.90/3.00; Rec voxel = 0.67/0.67/1.5; slice thickness = 1.5 mm; TR/TE/α = 4.6/1.35/40; acquisition time = 21 sec) were performed during contrast media infusion (Gadovist®: 1 nmol/ml Bayer Pharma, Berlin, Germany). High-resolution (HR) sequences were performed before and after contrast administration [coronal HR three-dimensional (3D) fast field echo (FFE): FOV 360 × 276; Acq matrix = 684 × 521; Rec matrix 880; slice thickness = 0.8 mm; TR/TE/flip angle (FA) = 6.3/2.1/20; acquisition time = 3:28 min]. These sequences allow the evaluation of vessel wall thickness up to 1 mm. CTA was performed with a Brilliance CT 64-channel scanner (Philips Medical Systems, Best, The Netherlands) with administration of 130 ml of Ultravist 370 (Bayer Healthcare LCC, Leverkusen, Germany) at a flow rate of 5 ml/sec. US was performed with iU22 Matrix Ultrasound system (Philips Medical Systems). US and CTA evaluations assessed carotid, subclavian, abdominal and femoral districts, while MRA was focused on thoraco-abdominal or cervico-thoracic regions according to disease involvement. Lesions were evaluated by two radiologists with an expertise in cardio-vascular imaging, blinded to the clinical status of patients. Variables evaluated for each lesion at MRA and CTA included lesion width, length, residual lumen and contrast enhancement. Since gadofosveset trisodium was retired, the presence or absence of vessel wall enhancement was scored by two radiologist experts in cardio-vascular imaging, blinded to the clinical status of patients. Enhancement data were available for 30 patients. At US, lesions width, residual lumen, peak systolic blood flow velocity and the presence of the halo sign were assessed. Imaging follow-up occurred yearly or, in case of relapses, until disease enters the tardive phase, clinical and morphological stability for at least 3–5 years.

Progression in arterial involvement was defined as appearance of novel lesions or as increase in width and/or length and/or percentage of luminal stenosis of established vasculitic lesions at follow-up. Two patients were evaluated with imaging only at a single time point, and progression over time could not be assessed. Progressive enlargement of ectasias/aneurysms may be related to biomechanical factors independent of disease activity. Therefore, it has not been included in the definition of progression of arterial involvement. Similarly, progression due to concurrent atherosclerosis was not taken into account. Confounder atherosclerotic lesions were identified and excluded on the basis of typical sites (arterial branches), eccentric appearance of the lesions, short and focal lesions, inhomogeneous content and irregular luminal border on US, the presence of parietal calcifications on the luminal side of the lesion and the absence of halo signs or adventitial thickenings on US.

We expressed scalar variables as median values and ranges. Mann-Whitney U test was used to compare biomarkers between patients with TA and controls, or between various subgroups of patients with TA stratified according to the presence or the absence of therapy with PPIs, steroids and immunosuppressive agents, arterial hypertension, wall enhancement, vascular progression and active disease. Multivariate analysis with multi-factor analysis of variance (ANOVA) was performed to verify the relationship between stratifying variables and plasma levels of CgA fragments. Plasma levels of CgA₄₃₉, CgA-FRs and VS-1, their ratio to CgA_tot and the anti-angiogenic CgA potential were used as dependant variables of the analysis. Five candidate factors were considered in the model on the basis of their clinical relevance and of their potential involvement: therapy with PPIs, presence of arterial hypertension, vascular progression, therapy with prednisone and therapy with immunosuppressive agents. Considering the sample size, we set the optimal number of factors in the model at four to avoid over-parameterization and loss of statistical power. Given the evidence of the impact of therapy with PPIs on plasma CgA levels [28] and the association between treatment with PPI and with steroids in our sample (p < 0.001), the latter was excluded from the final model, resulting in a four-way ANOVA. Kruskal-Wallis test was used to compare values of biomarkers within various classes of arterial involvement [29]. Spearman rank correlation coefficient was calculated for the correlation analyses. Associations were evaluated with χ² test and Fisher exact test.

A two-tailed p value less than 0.05 was considered to represent statistically significant differences, and p values less than 0.10 were shown in the tables. Statistical analysis was performed with IBM SPSS Statistics, version 20 (IBM Corp., Armonk, NY, USA).

Table 1 summarizes the demographic, clinical and laboratory characteristics of patients with TA (42 subjects, 39 women and three men) and of age-matched HCs (20 women). The median age at TA onset was 30 years (range 17–56 years). Thirty-seven (88 %) TA patients had a widespread diffuse arterial involvement (angiographic class II or V). Sixteen patients (38 %) had arterial aneurysms. Thirty-eight patients (90 %) were on treatment: 30 received steroids, 30 immunosuppressive agents (12 azathioprine, 11 methotrexate, four mofetil mycophenolate, two sirolimus, one cyclophosphamide), 16 tumour necrosis factor (TNF) blockers, two tocilizumab and one rituximab. Thirty patients were on treatment with proton-pump inhibitors (PPIs). Twelve patients (29 %) fulfilled the NIH criteria for active TA. Arterial wall enhancement was detectable in 16 % (5/30) and vascular progression in 22 % (9/40) of the patients. Twenty-two (52 %) patients had arterial hypertension. CRP and PTX3 concentrations were higher in patients with TA (2.6 mg/l, 0.1–40 mg/l and 5.5 ng/ml, 1.3–55 ng/ml, respectively) than in HCs (0.6 mg/l, 0.3–9.0 mg/l, p = 0.017 and 3.9 ng/ml, 1.4–6.5 ng/ml, p = 0.009 respectively).

The CgA system encompasses a family of variably processed polypeptides. We estimated total CgA (CgA_tot) concentration by assessing all the polypeptides containing the N-terminal region, i.e. by summing the results of 436/439 + FRs ELISA and 76 ELISA [15]. CgA_tot was higher in patients with TA than in HCs (2.36 nM, range 0.45 to 7.85 nM vs 0.98 nM, range 0.47 to 1.72 nM, p = 0.001; Fig. 1a and Table 1). Similarly, TA patients had higher plasmatic CgA-FRs and VS-1 than HCs (p = 0.001 and 0.020, respectively; Fig. 1b and Table 1). The inter-subject variability of the levels of CgA-derived polypeptides was high, especially in the TA group and for CgA₄₃₉ levels (Fig. 1a-b and Table 1), which had the highest coefficient of variation (not shown).

We evaluated disease-related and therapy-related variables. TA patients on PPIs (30/40, 75 %) had significantly more active disease (p = 0.009) and were more frequently treated with steroids than patients without PPIs (p < 0.001, Table 2). Levels of CgA_tot were higher in the group on PPIs. Specifically, patients on PPIs had higher concentrations of CgA-FRs and VS-1 (Fig. 1c and Table 2). High inter-patient variability in the levels of CgA_tot, CgA₄₃₉, CgA-FRs and VS-1 persisted after stratification for PPI treatment (Fig. 1c and Table 2). CgA₄₃₉/CgA_tot, CgA-FRs/CgA_tot and VS-1/CgA_tot ratios were similar in patients whether or not on PPI therapy (Fig. 1d and Table 2). These data suggest that PPIs apparently do not influence CgA processing and only partially account for the high inter-patient variability observed.

Twenty-two patients had arterial hypertension. Hypertensive patients were more often treated with non-biologic immunosuppressive agents and showed a trend towards higher creatinine levels (Table 2). Levels of CgA-FRs and VS-1 were higher in hypertensive TA patients (p = 0.030 and 0.024, respectively, Fig. 1e and Table 2). To take into account the possible confounding effect of PPIs, we repeated the analysis in TA patients on PPIs, observing significantly higher concentrations of CgA-FRs (p = 0.035) and a trend toward increased VS-1 (p = 0.087, Fig. 1f and Additional file 2: Table S1). Similar CgA₄₃₉/CgA_tot, CgA-FRs/CgA_tot and VS-1/CgA_tot were observed in the presence or the absence of arterial hypertension in the whole sample (Table 2) and after stratification for PPI therapy (Additional file 2: Table S1).

Plasma concentration of CgA-derived polypeptides did not reflect systemic inflammation (as measured by ESR and CRP) or local inflammation (as assessed by PTX3 levels [9] and/or by the presence of arterial wall enhancement at imaging). In particular, ESR and PTX3 did not correlate with the levels of CgA₄₃₉, CgA-FRs, VS-1, CgA₄₃₉/CgA_tot, CgA-FRs/CgA_tot and VS-1/CgA_tot in TA patients, even after stratification for PPI therapy (Table 3). CRP concentrations correlated with CgA-FRs but the correlation was lost after stratification for PPI treatment. Levels of CgA₄₃₉, CgA-FRs and VS-1 (Fig. 2a-b), and their ratios to CgA_tot were similar in patients with or without arterial wall enhancement at imaging.

In normotensive patients, VS-1 levels correlated positively with the concentration of CRP (Spearman coefficient 0.576, p = 0.008), and VS-1/CgA_tot correlated with ESR and PTX3 values (Spearman coefficient 0.684, p = 0.003 and 0.480, p = 0.032, respectively). CgA-FRs positively correlated with ESR (Spearman coefficient 0.570, p = 0.021; Table 3 and Additional file 3: Figure S2A-D). In hypertensive patients, CgA₄₃₉ and CgA₄₃₉/CgA_tot positively correlated with ESR (Spearman coefficient 0.517, p = 0.020 and 0.536, p = 0.015, respectively), while VS-1/CgA_tot negatively correlated with ESR and CRP (Spearman coefficient -0.561, p = 0.010 and -0.455, p = 0.033, respectively; Table 3 and Additional file 3: Figure S2E-H). These data suggest that arterial hypertension influences the link between the CgA system and inflammation in TA.

Similar concentrations of CgA₄₃₉, CgA-FRs and VS-1 (Fig. 2 and Table 4) and CgA₄₃₉/CgA_tot, CgA-FRs/CgA_tot and VS-1/CgA_tot (data not shown) were observed in patients with active versus inactive disease, defined based on NIH criteria [5]. Moreover, levels of CgA₄₃₉, CgA-FRs and VS-1 and their ratios to CgA_tot did not correlate with the number of arterial lesions as a measure of disease extent (Table 3). Concentrations of CgA₄₃₉, CgA-FRs and VS-1 were similar in patients regardless of the presence of vascular progression, defined as appearance of novel lesions or of increased thickness and/or length and/or percentage of luminal stenosis of established vasculitic lesions as assessed by imaging follow-up. We assessed the anti-angiogenic CgA potential by summing the ranks within the TA sample of CgA₄₃₉ and VS-1 [17]. We found that the anti-angiogenic CgA potential was unrelated to disease activity, to the number of involved vessels and to vascular progression. The group of patients on PPIs comprises eight out of the nine patients undergoing vascular progression. Vascular progression in patients on PPIs was associated with a significantly reduced anti-angiogenic CgA potential (p = 0.01, Table 4). The anti-angiogenic CgA potential was as well significantly reduced in hypertensive TA patients undergoing progression of the vascular damage (p = 0.05, Table 4).

Patients on prednisone had increased CgA-FRs (Additional file 4: Table S3). The difference was lost after stratification for PPIs (Additional file 4: Table S3) or for arterial hypertension (data not shown). Patients on conventional immunosuppressive agents had higher CgA-FRs, and the difference was maintained after stratification for PPIs (Additional file 4: Table S3) or for arterial hypertension (data not shown). Moreover, patients on immunosuppressive therapy had lower CgA-FRs/CgA_tot and a trend towards lower VS-1/CgA_tot (Additional file 4: Table S3), suggesting specific regulation of CgA processing. Absolute or relative concentrations of CgA and its peptides were similar in patients with or without TNF blockers or anticoagulant therapies (data not shown).

To verify the correlation between stratifying variables and plasma levels of CgA peptides, we performed a four-factor analysis of variance (ANOVA) (Additional file 5: Table S2). Plasmatic levels of CgA peptides were related with therapy with PPI, and the VS-1/CgA_tot ratio with treatment with immunosuppressive agents. The relationship between the presence of arterial hypertension and the plasma levels of CgA-FR and VS-1 did not reach the threshold set for statistical significance (p = 0.117 and 0.087, respectively), possibly because of the number of factors relatively to the sample size and of the association between arterial hypertension and therapy with immunosuppressive agents (p = 0.040, Table 2). Importantly, arterial progression correlated with reduced anti-angiogenic CgA potential (p = 0.032) at the four-way ANOVA analysis (Additional file 5: Table S2).