Quantitative videocapillaroscopy correlates with functional respiratory parameters: a clue for vasculopathy as a pathogenic mechanism for lung injury in systemic sclerosis

A cross-sectional study was carried out on SSc patients from the Vall d’Hebron Hospital cohort. All patients fulfilled LeRoy et al.’s SSc classification criteria [17], and accomplishment of the 2013 ACR/EULAR classification criteria was also assessed [18]. The study was approved by the Ethics Committee for Clinical Research (PG(AG)4/2015), and all patients provided written informed consent for their participation. NVC was consecutively conducted in 152 patients over a 12-month period. PFTs and echocardiography were performed within a 3-month period. Thirteen patients were excluded with fewer than eight explored nailfold fields, and another five patients for being lung transplant recipients. Finally, 134 SSc patients were selected for the analyses. The disease onset was described as the date of first symptom attributable to SSc including Raynaud’s phenomenon (RP). All clinical or parameter data were collected at the time of NVC. Study quality was assessed by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cross-sectional studies [19].

Cutaneous subsets were defined as previously [20], according to the extent of skin thickening: limited cutaneous SSc (lcSSc), if skin sclerosis was distal to the elbows and knees or the face; diffuse cutaneous SSc (dcSSc), if skin sclerosis was extended proximally to the elbows or knees; and sine scleroderma SSc (ssSSc), if there was no skin involvement. Manifestations collected included RP, telangiectasias, past or current digital ulcers (DU), calcinosis and past history of scleroderma renal crisis (SRC).

Interstitial lung disease was defined as radiologic evidence of interstitial findings on HRCT examined by an expert thorax radiologist [21]. PAH was defined as mean pulmonary arterial pressure (mPAP) ≥ 25 mmHg with pulmonary artery wedge pressure (PAWP) ≤ 15 mmHg, and pulmonary vascular resistance (PVR) > 3 Wood units in right heart catheterization (RHC) [22, 23]. Cardiac involvement was defined as past or current pericardial effusion, left ventricular ejection fraction (LVEF) < 50%, macrovascular or microvascular ischemic heart disease with no cardiovascular risk factor (CVRF), conduction abnormalities, diastolic dysfunction with no CVRF or mitral regurgitation with no CVRF [24].

Musculoskeletal disease included arthritis, tendon fiction rubs, joint contractures and myositis. Gastrointestinal SSc disease was established if any of the following were present: oesophageal dysmotility, gastric antral vascular ectasia, gastric or bowel dysmotility, intestinal bacterial overgrowth, intestinal pseudo-obstruction or anal sphincter dysfunction.

Antinuclear antibodies (ANA) were evaluated by indirect immunofluorescence (IIF) assay using HEp-cell line 2. Anticentromere antibodies (ACA) were described by IIF, and anti-topoisomerase I antibodies were determined by enzyme-linked immunosorbent assay. A commercial line blot assay was performed to detect anticentromere proteins A and B, anti-RNA pol III and anti-PM/Scl antibodies (EUROLINE Systemic Sclerosis (Nucleoli) Profile (IgG), Euroimmun, Germany) according to the manufacturer’s instructions.

Complete PFTs including the percentage of the predicted forced vital capacity (FVC) and the percentage of the predicted diffusion capacity of the lung for carbon monoxide (DLCO) were accomplished using MasterLab equipment (MasterLab, Jaegger, Germany), following the ATS/ERS recommendations [25]. Echocardiography was performed by an experienced echocardiographer using a Vivid E9 system (General Electric Vingmed, Horten, Norway), according to the consensus guidelines [26]. Right ventricular systolic pressure (RVSP), tricuspid regurgitation velocity (TRV) and the presence of other echo pulmonary hypertension (PH) signs were measured. In order to avoid loss of data information about patients with no measurable TRV, the variable ‘TRV ≥ 2.9 m/s or other echo PH signs’ was created [23].

Nailfolds were examined using an Optilia Digital Videocapillaroscope (Optilia Instruments AB, Sollentuna, Sweden) with 200× lens magnification and an LED lamp by the same operator (AS-O) who was blinded to clinical patients’ conditions. Each patient was acclimatized for at least 20 min at room temperature of 20–24 °C before the examination. NVC was conducted with a contact adapter and immersion oil dropper, studying the middle nailfold from the second to fifth fingers of both hands. Optipix Lite software (Optilia Instruments AB) was used for visualization of consecutive 1-mm-wide images of first-line capillaries. Two images were taken on each finger, although when a finger was not possible to evaluate additional pictures were taken on other fingers. Patients with fewer than eight images were excluded from the study. Quantitative and qualitative analyses of the images were performed blind to clinical data by investigator AS-O. Inter-observer reliability and intra-observer reliability were also explored with a calculated sample size of 45 patients. This sample was randomly selected and blindly assessed once by investigator AG-DC to calculate the inter-observer reproducibility. Afterwards, to estimate the intra-observer repeatability, a second round was independently performed by investigators AS-O and AG-DC with a minimal time interval of 2 weeks between both measurements.

Quantitative NVC was carried out according to previous definitions as follows: capillary density, number of capillaries measured in the distal row following the 90° method [27]; enlarged capillary, an increase in capillary diameter > 20 μm; giant capillary, a capillary dilatation > 50 μm; microhaemorrhage, distal haemosiderin deposits; tortuous, curled capillaries with no cross; and neoangiogenesis, branching, bizarre, bushy, disorganized, ramified or arborized capillaries (Fig. 1) [14, 28, 29].

The mean of each capillaroscopic feature was calculated from the sum of consecutive images for each digit. Subsequently, the average values from eight fingers were added together and divided by the number of studied digits. The resulting value indicated the number of this capillaroscopic feature adjusted by each millimetre of the nailfold.

Qualitative NVC analysis of images was performed following capillaroscopy patterns described by Cutolo et al. (normal, early, active and late) [14].

In order to analyse inter-observer and intra-observer reliability, a sample size of 45 patients was calculated, based on two observers (AS-O and AG-DC), with a power of 80% to detect that an intra-class coefficient (ICC) of 0.7 was significantly higher than 0.45 [30, 31]. The value of 0.45 was arbitrarily selected as the lowest acceptable limit of reliability. Quantitative NVC features were evaluated as the number per millimetre and determined by the ICC, whereas qualitative NVC features were assessed by Cohen’s kappa coefficient. The measurement of observer agreement was defined as described in the literature: < 0.00 = poor; 0.00–0.20 = slight; 0.21–0.40 = fair; 0.41–0.60 = moderate; 0.61–0.80 = substantial; and 0.81–1.00 = almost perfect [32].

Demographic characteristics of the 134 SSc patients are presented in Table 1. The participants were mainly female (n = 113, 84.3%) and in the lcSSc subset (n = 88, 65.7%). More than 90% of the population fulfilled the 2013 ACR/EULAR classification criteria. ACA were the most common antibodies in 49 (36.6%) patients, followed by anti-topoisomerase I antibodies in 31 (23.1%) patients. Considering major organ involvement, digestive involvement was present in 110 (82.1%) patients, cardiac disease in 103 (76.9%) patients, ILD in 58 (43.3%) patients, musculoskeletal involvement in 40 (29.9%) patients and PAH in 11 (8.2%) patients.

The distribution of different cardiac manifestations was 76.9% of patients presenting cardiac conduction abnormalities, 57.5% mitral regurgitation (with no CVRF), 47% diastolic dysfunction (with no CVRF), 13.4% pericardial effusion, 8.2% microvascular ischemic heart disease (with no CVRF), 3.7% LVEF < 50% and 3% of patients macrovascular ischemic heart disease (with no CVRF). Among all patients, there were 19 (14.1%) with overlap features as follows: nine (6.7%) patients had dermatomyositis, seven (5.2%) patients had Sjögren syndrome and the other three (2.2%) patients had rheumatoid arthritis.

PFTs and echocardiography findings are presented in Table 2. The mean predicted FVC (± SD) was 80.8% (± 20.1), with a mean predicted DLCO of 66.2% (± 23.7). Regarding echocardiography, the TRV could be measured in 96 patients with a mean of 2.8 m/s (± 0.3), and no other signs of PH were found in the remaining 38 patients. Among 14 (10.4%) subjects with TRV higher than 2.9 m/s and/or with other echo PH signs, seven patients had been previously diagnosed with PAH by RHC. On the other hand, only 4 out of 120 patients with TRV lower than 2.9 m/s and no other echo PH signs had been formerly diagnosed with PAH by RHC.

A total of 2186 images were analysed from 134 patients, with a mean of 16.3 (± 5.0) images per subject and the median (IQR) of images for each finger was 2 (1–3). The median (IQR) of capillary density adjusted by each millimetre of nailfold was 5.44/mm (4.33–6.74) (Table 3). NVC findings according to the presence of ILD were compared. Median capillary density in ILD patients was 4.86/mm, significantly lower than patients with no ILD whose density was 5.88/mm (p = 0.005). Moreover, the number of capillaries with neoangiogenesis was higher in ILD patients (0.56/mm vs 0.31/mm, p = 0.005). Focusing on PAH, patients with this manifestation also showed greater frequency of capillaries with neoangiogenesis compared with patients with no PAH (0.70/mm vs 0.33/mm, p = 0.007). All prior comparisons remained statistically significant after Bonferroni correction (p < 0.008).

Concerning qualitative NVC analysis, a late pattern (27.6%) was more prevalent in patients with ILD (p = 0.006). In addition, a late pattern showed lower FVC and DLCO values compared with normal, early and active patterns (Table 4).

Age at NVC was inversely correlated with the number of enlarged capillaries (ρ = − 0.20, p = 0.01) and the quantity of microhaemorrhages (ρ = − 0.25, p = 0.003). The time elapsed from first symptom to NVC was negatively associated with the number of microhaemorrhages (ρ = − 0.25, p = 0.003) and positively with the number of capillaries with neoangiogenesis (ρ = 0.18, p = 0.03). Higher capillary density was positively correlated with the % DLCO (ρ = 0.26, p = 0.003) and negatively with the RVSP (ρ = − 0.21, p = 0.03). The number of capillaries with neoangiogenesis was inversely associated with % FVC (ρ = − 0.24, p = 0.004) and % DLCO (ρ = − 0.26, p = 0.002) and was positively associated with RVSP (ρ = 0.20, p = 0.04).

The inter-observer reproducibility for all quantitative NVC features was almost perfect with an ICC higher than 0.80 (Table 5), except for the number of capillaries with neoangiogenesis which displayed a substantial reproducibility (95% CI) of 0.77 (0.58–0.88). Qualitative Cutolo’s patterns also showed almost perfect reproducibility with Cohen’s κ of 0.83 (0.66–1.0), even when each pattern was analysed separately, with the only exception of an early pattern which had a substantial reproducibility with Cohen’s κ of 0.73 (0.42–1.0).

Regarding intra-observer repeatability, similar results were obtained for quantitative and qualitative variables, with an increment of neoangiogenesis repeatability up to 0.81 (0.56–0.90).