Results

Demographic and clinical characteristics of the studied populations

Afifty behçet patients 6(12%) females and 44(88%) males with a mean of age 33.44 ± 7.8 years; another 50 controls as 40(80%) males and 10(20%) females with a mean of age 33.86 ± 4.0 were included in the present study. No significant difference was reported in neither age nor sex between groups with p > 0.05.

Symptoms were diagnosed and recorded for all patients including oral and/or genital ulcer and organ involvement (skin, musculoskeletal, gastrointestinal tract (GIT), ocular, central nervous system (CNS), and vascular manifestations) (Table 1).

Variable | Behçet patients(N = 50)
Clinical symptoms
Oral ulcer | 23 (46%)
Genital ulcer | 6 (12%)
Organ involvement
Skin manifestations | 15 (30%)
Musculoskeletal manifestations | 30 (60%)
GIT manifestations | 11 (22%)
Ocular manifestations | 32 (64%)
CNS manifestations | 16 (32%)
Vascular manifestations | 6 (12%)
Immunosuppressants
Azathioprine | 20 (40%)
Cyclosporin | 20 (40%)
Hydroxychloroquine | 4 (8%)
Drugs
Steroids | 39 (78%)
Anti TNF | 10 (20%)
Medication patterns
Colchicine | 38 (76%)
BDCAF Patient’s Index Score: 3.08 ± 1.85
Score (Inactive: Active Behcet Disease) | 0.80 ± 0.42:3.64 ± 1.61 (p = 0.004*)

In addition, treatment and medication patterns have been recorded. Patients undertook immunosuppressants such as Azathioprine, Cyclosporin, and Hydroxychloroquine, covering 20 (40%), 20(40%), and 4(8%), respectively. While 39(78%) of the patients were under the treatment of Steroids, only 10 (20%) were using anti-tumor necrosis factor (Anti-TNF). For other medication patterns, 38 (76%) of the patients were using Colchicine (Table 1).

Patient’s Index Score BDCAF, had been evaluated where all BD patients had a mean and standard deviation of 3.08 ± 1.85. When patients were subclassified into inactive and active Behçet disease according to a significant difference as regards activity of the disease, with p = 0.004 (Table 1).

CircRNA-FUNDC1 and TNF-α serum biomarkers levels for Behçet disease and healthy control groups

CircRNA-FUNDC1 and TNF-α were assessed and comparatively analyzed` between BD patients and controls. Statistically significant variations were identified between the two groups with respect to both biomarkers, where log2 CircRNA-FUNDC1 and TNF-α were higher in patients group compared to controls (p < 0.0001) for both biomarkers (Table 2; Fig. 1).

Serum biomarkers | Behçet patients(N = 50) | Healthy control(n = 50) | p-value
Log2 circRNA-FUNDC1(FC) | 8.99 ± 4.76 | 1.01 ± 0.03 | < 0.0001*
TNF-α (pg/mL) | 30.73 ± 16.14 | 5.23 ± 3.2 | < 0.0001*

Fig. 1: Serum levels of the study biomarkers between Behçet patients and healthy controls (a) expression level of log2 circRNA-FUNDC1 (FC), (b) serum concentration level of TNF-α.

No significant difference was reported between the expression level of circRNA-FUNDC1 in serum and clinical data among the BD patients, including (gender, clinical symptoms, either oral or genital ulcer) where p > 0.05. Even no significant difference was detected as regards organ involvement (skin, GIT, ocular, or vascular), where p-value > 0.05. However, a significant difference was detected between log2 circRNA-FUNDC1 in patients with negative musculoskeletal and CNS manifestations, with higher expression levels of log2 circRNA-FUNDC1 than in positive patients with manifestations respectively p = 0.007 and 0.0037 (Table 3).

Variable | Log2 circRNA-FUNDC1(FC) | p-value | TNF-α (pg/mL) | p-value
Sex | Female (N = 6) | 8.77 ± 2.32 | 0.087 | 40.75 ± 18.69 | 0.49
Male (N = 40) | 9.02 ± 5.02 | 29.35 ± 15.51
Clinical symptoms |  | 
Oral ulcer | Negative (N = 27) | 9.73 ± 4.12 | 0.12 | 28.06 ± 13.78 | 0.041 *
Positive (N = 23) | 8.12 ± 5.39 | 33.85 ± 18.36
Genital ulcer | Negative (N = 44) | 9.48 ± 4.63 | 0.87 | 30.89 ± 15.74 | 0.34
Positive (N = 6) | 5.37 ± 4.46 | 29.51 ± 20.53
Organ involvement |  | 
Skin manifestations | Negative (N = 35) | 8.52 ± 4.50 | 0.47 | 27.41 ± 14.50 | 0.025 *
Positive (N = 15) | 10.08 ± 5.33 | 38.46 ± 17.60
Musculoskeletal manifestations | Negative (N = 20) | 10.0 ± 6.23 | 0.007 * | 31.38 ± 13.87 | 0.24
Positive (N = 30) | 8.32 ± 3.43 | 30.28 ± 17.72
GIT manifestations | Negative (N = 39) | 9.14 ± 5.21 | 0.19 | 30.52 ± 15.33 | 0.27
Positive (N = 11) | 8.47 ± 2.81 | 31.43 ± 19.58
Ocular manifestations | Negative (N = 18) | 9.07 ± 3.97 | 0.66 | 31.34 ± 16.40 | 0.84
Positive (N = 32) | 8.95 ± 5.22 | 30.38 ± 16.25
CNS manifestations | Negative (N = 34) | 9.13 ± 5.54 | 0.037 * | 27.22 ± 15.75 | 0.88
Positive (N = 16) | 8.70 ± 2.53 | 38.17 ± 14.80
Vascular manifestations | Negative (N = 44) | 9.06 ± 5.0 | 0.28 | 30.60 ± 16.45 | 0.45
Positive (N = 6) | 8.47 ± 2.69 | 31.58 ± 15.01
Immunosuppressants |  | 
Azathioprine | No (N = 30) | 7.47 ± 3.21 | 0.019 * | 33.73 ± 15.80 | 0.72
Yes (N = 20) | 11.26 ± 5.81 | 26.21 ± 15.99
Cyclosporin | No (N = 30) | 9.74 ± 5.62 | 0.049 * | 27.50 ± 16.11 | 0.08
Yes (N = 20) | 7.87 ± 2.84 | 35.55 ± 15.34
Hydroxychloroquine | No (N = 46) | 9.08 ± 4.93 | 0.28 | 29.34 ± 15.03 | 0.040 *
Yes (N = 4) | 8.0 ± 2.29 | 46.55 ± 22.54
Drugs |  | 
Steroids | No (N = 11) | 6.82 ± 3.16 | 0.035 * | 28.84 ± 14.61 | 0.38
Yes (N = 39 | 9.60 ± 4.99 | 31.25 ± 16.69
Anti TNF | No (N = 40) | 9.57 ± 4.83 | 0.063 | 30.88 ± 16.17 | 0.89
Yes (N = 10) | 6.68 ± 3.89 | 30.09 ± 16.88
Medication patterns |  | 
Colchicine | No (N = 12) | 8.16 ± 5.51 | 0.86 | 28.59 ± 13.96 | 0.22
Yes (N = 38) | 9.25 ± 4.55 | 31.39 ± 16.89

In addition, the level of log2 circRNA-FUNDC1 was reported higher in BD patients who received Azathioprine as an immunosuppressant than in patients who did not, with p = 0.019. While patients who did not take Cyclosporin had higher levels of log2 circRNA-FUNDC1 than those who took Cyclosporin (p = 0.049). In addition, there was a significantly higher level of log2 circRNA-FUNDC1 in patients who received Steroids than in those who did not, with p-value = 0.0035 (Table 3).

The concentration level of TNF-α was higher in patients having oral ulcers than those without ulcers, with a significance of p = 0.041. As for organ manifestations, there was only a significant difference as regards patients with skin manifestations involvement than those who had negative skin issues, with p = 0.025.

In the present study, TNF-α concentration levels were higher in patients receiving Hydroxychloroquine as an immunosuppressant, with p = 0.040 (Table 3).

After applying the FDR correction for multiple comparisons, the previously observed associations between circRNA-FUNDC1 and the associated clinical variables, as well as between TNF-α expression levels and the associated clinical variables, were no longer remained statistically significant after adjustment (all q-values > 0.05); therefore, these findings should be considered exploratory and hypothesis-generating.

Association of circRNA-FUNDC1 and TNF-α with disease activity

Based on the BDCAF Patient’s Index Score, patients were subclassified into inactive and active BD, and after evaluating the circRNA-FUNDC1 and TNF-α levels in serum accordingly, it was noticed that as disease activity gets worse, the levels of both biomarkers increase. For the expression of circRNA-FUNDC1 (FC) and the concentration level of TNF-α, tended to elevate in active BD groups than in active BD patients (Fig. 2).

Fig. 2: Association of circRNA-FUNDC1 and TNF-α with disease activity.

Correlations between the serum levels of circRNA-FUNDC1 and TNF-α with age and BDCAF Patient’s Index Score of patients with Behçet’s disease

Results showed that there were no correlations found between circRNA-FUNDC1 and the concentration level of TNF-α (r=−0.23, p = 0.11), circRNA-FUNDC1 and age (r = 0.03, p = 0.79), circRNA-FUNDC1 and BDCAF Patient’s Index Score (r=−0.07, p = 0.61), age and TNF-α (r=−0.09, p = 0.52), TNF-α and BDCAF score (r = 0.23, p = 0.09) among Behçet patients.

circRNA-FUNDC1 and TNF-α: discriminatory performance between Behçet’s disease patients and healthy controls

The discriminatory performances of circRNA-FUNDC1 and TNF-α as biomarkers for Behçet’s disease were evaluated to discriminate patients from controls. The sensitivity for circRNA-FUNDC1 was 98% and the specificity was 100%, with a 99% accuracy (p < 0.0001) at a cutoff value of > 1.16 FC. For TNF-α, the sensitivity and specificity were 92% and 94%, respectively, with a 93% of accuracy at p < 0.0001 and a cutoff > 9.6 pg/mL. The receiver operating characteristic (ROC) curves of circRNA-FUNDC1 and TNF-α levels for serum of BD patients are shown in Table 4; Fig. 3.

Biomarker | Log2 circRNA-FUNDC1(FC) | TNF-α (pg/mL)
AUC | 0.99 | 0.95
Cut-off value | > 1.16 FC | > 9.6 pg/mL
Sensitivity | 98% | 92%
Specificity | 100% | 94%
Accuracy | 99% | 93%
95% CI a | 0.96 to 1.0 | 0.88 to 0.98
p-value | < 0.0001* | < 0.0001*

Fig. 3: ROC curve analysis for (a) log2 circRNA-FUNDC1 (FC) and (b) serum concentration level of TNF-α.

CircRNA-FUNDC1 showed excellent discriminatory performance in distinguishing Behçet’s disease patients from healthy controls in the present study and may represent a promising biomarker for Behçet’s disease. However, these findings should be interpreted with caution because disease control groups were not included. Further large-scale studies incorporating patients with other inflammatory and autoimmune diseases are required to validate the specificity and clinical utility of circRNA-FUNDC1 before it can be considered for routine diagnostic use.

In addition, regression analysis was used by which the models were intended to improve model stability in the presence of overlapping treatment regimens and unequal numbers of patients within treatment subgroups. Thus, medications that demonstrated significant associations in the univariate analyses where used in the models.

Hydroxychloroquine was independently associated with higher TNF-α levels (β = 17.20, p = 0.04). However, the model explained only 8.5% of the variability in TNF-α levels (Table 5).

(a) Model for TNF-α according to significant treatment variables
Variable | β | 95% CI lower | 95% CI upper | p-value | R² | Adjusted R²
Constant | 29.35 | 24.72 | 33.97 | < 0.001 | 0.085 | 0.066
Hydroxychloroquine | 17.20 | 0.84 | 33.55 | 0.04*
(b) Model for circRNA-FUNDC1 according to significant treatment variables
Variable | β | 95% CI lower | 95% CI upper | p-value | R² | Adjusted R²
Constant | 6.28 | 3.31 | 9.25 | < 0.001 | 0.179 | 0.125
Azathioprine | 3.17 | − 1.43 | 7.77 | 0.17
Cyclosporin | − 0.32 | − 5.07 | 4.42 | 0.89
Steroid | 2.02 | − 2.28 | 6.32 | 0.35

On the other hand after adjustment for azathioprine, cyclosporin, and steroid use, none of the medications remained independently associated with circRNA-FUNDC1 expression (p > 0.05). The model explained 17.9% of the variability in circRNA-FUNDC1 levels (Table 5).

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Regression={\text{ }}\beta 0+\beta 1X1+\beta 2X2+...\beta iXi$$\end{document}

Where β0 is the constant intercept, X1 and X2 denote circFUNDC1 expression level of medication variable(s), and β1-βi are the estimated coefficients for each.

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} Regression\; equation\; for\; circRNA - FUNDC1 & =6.282+3.17\left( {Azathioprine} \right) \hfill \\ & \quad - 0.32\left( {Cyclosporin} \right)+{\text{ }}2.02\left( {Steroid} \right) \hfill \\ \end{aligned}$$\end{document}