AB1133 AUTOMATED ASSESSMENT (AUTOCAPI) OF NAILFOLD CAPILLARY NUMBER VERSUS MANUAL COUNTING IN SYSTEMIC SCLEROSISPATIENTS WITH DIFFERENT CAPILLAROSCOPIC PATTERNS

Abstract

Background The nailfold videocapillaroscopy (NVC) analysis allows microvascular detection of possible markers of severity and progression in systemic sclerosis (SSc), as reduced capillary number, which has been associated with a high risk of developing disease complications (i.e. digital ulcers over time) (1-2). A recent study validated an automated counting system (AUTOCAPI) for the absolute nailfold capillary number in SSc patients by analysing the capillaroscopic images (3). Objectives To evaluate the performance of this automated software for absolute nailfold capillary number counting, in SSc patients with different NVC patterns of microangiopathy. Methods 183 random SSc patients collected at both Genova and Ghent Divisions of Rheumatology were enrolled (LeRoy 2001 or ACR 2013 criteria, mean age 55±13 year, mean disease duration 5.5±6.8 years) and classified by NVC in the following patterns: 28 ”not specific”, 37 “early”, 89 “active”, 29 “late”). 8 fingers for each patient were analysed, counting the number of nailfold capillaries manually and by the AUTOCAPI software (DS Medica, Italy) along a millimetre in each finger image. The mean capillary number value of the eight finger images was calculated. The software reliability was assessed through calculation of the intraclass correlation coefficient (ICC) between automatic and manual counting. Results The mean number of capillaries assessed by manual vs automatic counting was as follows: 5.23±1.7 vs 5.47±1.3 in the total group of SSc patients, 5.91±1.2 vs 6.87±1.2 in the ”not specific”, 7.23±1.4 vs 5.67±1.1 in the “early”, 4.67±1.1 vs 5.16±1.2 in the “active” and 3.72±1.5 vs 4.85±1.1 in the “late” pattern of microangiopathy. The higher standard deviation observed for automatic counting was 1.23 in the ”not specific” group. The following ICC’s were obtained respectively for total patients, ”not specific”, “early”, “active”, and “late” NVC patterns: 0.53, 0.51, 0.48, 0.50 and 0.66. The mean values for the manual versus automatic capillary counting assessed by the two centres in all SSc patients were respectively: 5.92±1.8 and 5.02±1.1 for Genova centre, and 4.71±1.5 and 5.83±1.4 for Ghent centre. The automatic counting confirmed that capillary number progressively reduces from ”early” to ”active” to ”late” NVC pattern of microangiopathy in SSc. Conclusion This study reports for the first time the good reliability of AUTOCAPI software in nailfold capillary number counting in SSc patients with different patterns of microangiopathy. The use of automated counting software need to standardize nailfold capillary assessment among different Rheumatological centres.