AB0380 MACHINE LEARNING CAN PREDICT GIANT CELL ARTERITIS RELAPSE AFTER GLUCOCORTICOID TAPERING

Background:

To date reliable biomarkers and risk factors for relapsing giant cell arteris (GCA) after glucocorticoid (GC) tapering are still lacking.

In an increasing number of social and clinical scenarios, machine learning (ML) is emerging as a promising tool for the implementation of complex multi-parametric decision algorithms. A ML approach allows to handle complex non-linear relationships between patient attributes that are hard to model with traditional statistical methods, merging them to output a forecast or a probability for a given outcome.

Objectives:

To assess whether ML algorithms can predict GCA relapse after glucocorticoid tapering.

Methods:

GC-naïve GCA patients who presented to 4 tertiary care centers between January 2015 and January 2019, who underwent GC therapy and regular follow up visits for at least 12 months were retrospectively analyzed and used for training and validation (through 10-fold cross-validation) of n.2 ML algorithms, namely Decision Trees (DT) and Random Forest (RF).

Test of the algorithms was carried out GCA patients referred to the same centers from March 2019 to September 2020 whose data was longitudinally recorded during the 12 months after presentation.

Demographic, clinical an laboratory characteristics (Erythrocyte Sedimentation Rate (ESR) and C Reactive Protein (CRP) levels) were gathered.

The outcome of interest was the GCA relapse within 12 months after induction of remission, during GC tapering.

The accuracy of the algorithms in both validation and test phases was assessed.

Results:

The training and validation dataset consisted of n.85 GCA patients (59 female, 69.4%) with mean age 73.8 (±8.7) years at presentation. They were treated with 27.1 (±17.4) mg prednisone (PDN) equivalent at first visit. During GC tapering 34 of them (40%) experienced a disease relapse within 12 months. The test dataset consisted of n.22 patients (14 female, 63.4%) with mean age 75.5 (±8.7) years at presentation, who underwent GC induction therapy with a mean dose of 30.3 (±17.3) mg PDN equivalent. Nine of them (40.9%) had a GCA flare during GC tapering, within 12 months. Accuracy of DT and RF in predicting the outcome of interest on the training dataset was 68.3% and 73.4% respectively. On testing datasets DT and RF accuracy was 57.1 and 72.4%, respectively.

As shown in Figure 1, the most important patient attributes for RF forecast were found to be CRP and ESR baseline levels as well as age and symptom duration (months) at first visit.

Conclusion:

RF algorithm can predict GCA relapse after glucocorticoid tapering with fairly good accuracy. To date this is one of the most accurate predictive modeling for such outcome. This ML method represents a reproducible tool executable on computers as well as mobile devices and capable of supporting clinicians in GCA patient management.

References:

[1]Hellmich B., Agueda A., Monti S., et al. 2018 Update of the EULAR recommendations for the management of large vessel vasculitis Annals of the Rheumatic Diseases 2020;79:19-30.

[2]Venerito, V., Angelini, O., Cazzato, G. et al. A convolutional neural network with transfer learning for automatic discrimination between low and high-grade synovitis: a pilot study. Internal and Emergency Medicine 2021.

Disclosure of Interests:

None declared

Clinical phenotypes of Behçet's syndrome in a large cohort of Italian patients: focus on gender differences

Objective: The aim of this study was to investigate the clinical phenotypes of Italian patients with Behçet's syndrome (BS) according to gender. BS is a rare chronic multisystemic disorder with a wide spectrum of clinical manifestations. Human leucocyte antigen (HLA)-B51, gender, and ethnicity have been suggested as factors that could influence the clinical manifestations in BS patients. To date, few data assessing gender differences in Italian BS patients are available in the literature.Method: We retrospectively evaluated a group of Italian patients seen consecutively at our dedicated tertiary centre from 1 January 2000 to 31 May 2018. Demographics, clinical features during follow-up, and HLA status were obtained from a review of medical records and analysed in male and female groups.Results: In total, 285 [168 male (M) and 117 female (F)] patients were eligible for the study. Males had papulopustolar lesions, posterior uveitis, and deep venous thrombosis more often than females (83.3% M vs 46.2% F, 36.9% M vs 18.8% F, and 8.3% M vs 0.9% F, respectively; p < 0.01). Erythema nodosum (59.0% F vs 41.1% M; p < 0.01) and arthralgia (52.1% F vs 31.6% M; p < 0.01) were more frequent in females. No differences were found in HLA-B51 status (59.2% M vs 59.0% F).Conclusion: In our Italian cohort, BS was slightly more prevalent in males. Some gender-related differences were observed when comparing male and female cohorts. The data also confirmed that BS tends to be less aggressive in Italian female patients.

FRI0569 SERUM AMYLOID A: ASSESSMENT OF REFERENCE VALUE AND COMPARISON OF SERUM CONCENTRATION IN HEALTHY SUBJECTS AND PATIENTS WITH BEHÇET SYNDROME

Background:

Serum amyloid A (SAA) is a family of acute-phase reactants. The rise of SAA concentration in blood circulation is a clinical marker of active inflammation in several auto-inflammatory diseases, including Behçet syndrome (BS). Despite its practical and analytical advantages, SAA measurement by ELISA has been mainly used as a research tool rather than for the routine laboratory testing due to the lack of a robust reference data in the literature.

Objectives:

Using the recommended procedures of the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC), we aimed to develop the SAA reference interval for a well-defined Italian healthy population (HC). Secondly, we compared the SAA serum concentration between HC and patients with BS.

Methods:

Sera specimens were collected from adult healthy blood donors after rule out the exclusion criteria (inflammatory disorders, ongoing infections, pregnancy and breastfeeding, obesity, using oral contraceptives, use of any medication, or consumed of alcohol), and from unselected BS patients fulfilling the International Study Group (ISG) classification criteria. Serum SAA concentrations were detected and quantified with a commercial solid phase sandwich enzyme-linked immunosorbent assay (Human SAA ELISA kit, IBL International GmbH, Hamburg, Germany) used on automated analyzer (Immunomat, SERION Diagnostic, Alifax, Polverara (PD), Italy) according to the manufacturer’s protocol. Statistical analysis and data normalization of HC SAA values were carried out to determine the reference cut off. In the second step of the study, HC and BS patients were stratified in two groups according to the cut-off value.

Results:

We recruited 141 HC (84 M and 57 F; mean age, 44.5±13.2 years) and 63 BS patients (39 M and 24 F mean age, 45.3±13.2 years) assayed for SAA. The reference cut-off was calculated as 225 ng/ml. No statistically significant differences were found between males and females when SAA means were compared, suggesting that not gender-partitioned reference range is recommended for this analyte. After the stratification according to the cut-off value (group 1: < 225 ng/ml and group 2: > 225 ng/ml), we found 53/63 (84.1%) BS patients and 133/141 (94.3%) HC with concentration less than cut-off value, respectively. We identified 10/63 (15.9%) BS patients and 8/141 (5.7%) HC within the second group. The difference was statistically significant (p=0.0177; OR: 3.14, 95% CI: 1.17-3.38).

Conclusion:

This study allowed to define a widely accepted reference cut-off for the SAA detected by ELISA, responding to an unmet need of laboratory medicine. We found a statistically significant higher frequency of BS patients compared with HC when SAA values is higher than cut-off (225 ng/ml). This preliminary data could add significant information for better clarify the role of SAA as biomarker of inflammation and in guidance of clinical practice. Further studies will be required to stratify SAA values in relation to disease activity of BS.

Disclosure of Interests:

Teresa Carbone: None declared, Maria Carmela Padula: None declared, Vito Pafundi: None declared, Carlo Schievano: None declared, Nancy Lascaro: None declared, Angela Padula: None declared, Pietro Leccese: None declared, Salvatore D’Angelo Consultant of: AbbVie, Biogen, BMS, Celgene, Eli Lilly, MSD, Novartis, and UCB, Speakers bureau: AbbVie, BMS, Celgene, Eli Lilly, Novartis, Pfizer, and Sanofi

AB0018 TNFΑ RS1800629 POLYMORPHISM: WHAT ABOUT ITS ASSOCIATION WITH CLINICAL MANIFESTATIONS AND ANTI-TNFΑ THERAPY? DATA FROM A SERIES OF ITALIAN PATIENTS WITH BEHÇET SYNDROME

Background:

Tumor Necrosis Factor-alpha (TNF-α) is a pleiotropic cytokine with a critical role in the pathogenesis of Behçet syndrome (BS). Anti-TNF-α therapy is useful for patients with refractory, severe BS, in particular for ocular, central nervous system, and gastrointestinal manifestations. However, although biological treatment with anti-TNF-α agents are effective in BS, not all patients are definite responders. Non-responders patterns could be due to: alternative non-TNFα related pathway of inflammation; anti-drug antibodies presence or development; polymorphic alleles ofTNFαgene.TNFαrs1800629 (-308G>A) is a drug-response single nucleotide polymorphism (SNP) located within the gene promoter. Poor and conflicting data are currently available about the association of this polymorphism and clinical manifestations of BS, as well as about the responsivness to the TNFα blockers in BS patients [1-3].

Objectives:

Aims of this study were to investigate in a cohort of Italian patients with BS the frequency of rs1800629 genotypes and its association with clinical features and anti-TNFα therapy response.

Methods:

Consecutive patients with BS were recruited. Patients demographic and clinical data were collected by medical records and analyzed. Home-made specific primer pairs were used for rs1800629 coverage. gDNA was isolated and amplified using PCR. Good-quality amplicons were sequenced (Sanger method).In silicoanalysis was downstream performed using specific software for query-subject similarity analysis.

Results:

130 BS patients (64M:66F; mean age: 45.8±12.3 years) were included in the study. Patients predominant lesions were oral aphtosis (100%), eye involvement (86.2%), skin lesions (72.3%) and genital ulcers (57.7%).TNFαrs1800629 wild-type GG genotype was found in 106/130 BS patients (81.5%); the heterozygous genotype (GA) was identified in 24/130 patients (18.5%). No statistically significant differences were found in genotypes frequencies when the patients were stratified for presence and absence of each clinical manifestation (p>0.05), while statistical significant differences were found when the patients were compared for therapy (anti-TNFα drugs) response. In detail, 73/130 patients (56.2%) were treated with anti-TNFα agents. We found 16/73 (21.9%) non-responders patients (NRP). In NRP group, we identified 9/16 patients (56.3%) with GG genotype and 7/16 (43.7%) with GA genotype, while 8/57 (14.0%) responder patients showed GA genotype and 49/57 responder patients (86.0%) showed GG genotype (p=0.0093; OR: 0.21, CI: 0.06-0.729).

Conclusion:

Here we described a low frequency ofTNFαrs1800629 SNP-containing allele and the lack of association between SNP and BS clinical hallmark, as previously reported in literature [1-4]. We also found higher percentage of GG genotype in case of therapy response than GA genotype. The SNP is a promoter polymorphism that could affect the auto-inflammatory response and the therapy responsivness, as suggested by our preliminary data of pharmacogenomics. Analyses of a larger cohort of patients are need to confirm the study findings and to explain the SNP role as outcome predictor.

References:

[1]Touma Z et al. (2010). Arch Med Res 41(2):142-6;

[2]Vallet H et al. (2015). J Autoimmun 62:67-74.

[3]Zhang M et al. (2013). Mol Vis 19:1913-24.

[4]Ateş A et al. (2006). Rheumatol Int 26(4):348-53.

Disclosure of Interests:

Maria Carmela Padula: None declared, Pietro Leccese: None declared, Nancy Lascaro: None declared, Giusi Gaia Sorrento: None declared, Rosa Paola Radice: None declared, Antonina Rita Limongi: None declared, Teresa Carbone: None declared, Angela Padula: None declared, Giuseppe Martelli: None declared, Salvatore D’Angelo Consultant of: AbbVie, Biogen, BMS, Celgene, Eli Lilly, MSD, Novartis, and UCB, Speakers bureau: AbbVie, BMS, Celgene, Eli Lilly, Novartis, Pfizer, and Sanofi