Serum growth differentiation factor-15 levels in Behçet’s syndrome

Abstract

Behçet’s syndrome (BS) is a heterogeneous systemic vasculitis characterized by chronic inflammation and immune dysregulation. Growth differentiation factor-15 (GDF-15), a member of the transforming growth factor-β superfamily, has been implicated in chronic inflammatory and immune stress responses. This study aimed to evaluate serum GDF-15 levels in patients with BS and to investigate their relationship with disease activity and clinical parameters. This cross-sectional study included 49 patients diagnosed with BS according to International Study Group criteria, as well as a control group of 33 healthy individuals. Demographic characteristics, clinical features, laboratory parameters, treatments, and disease activity scores were recorded. Serum GDF-15 levels were measured using a commercial enzyme-linked immunosorbent assay. Statistical analyses were performed to compare groups and to evaluate associations between GDF-15 levels and clinical and laboratory variables. In the unadjusted comparison, serum GDF-15 levels were significantly higher in patients with BS than in controls (P = .016). However, GDF-15 levels were not associated with disease activity scores. In multivariable logistic regression adjusting for sex and BMI, GDF-15 was no longer an independent predictor of BS (adjusted P = .53), whereas sex (OR = 5.61, P = .006) and BMI (OR = 0.86, P = .024) remained independently associated with the group, indicating that the unadjusted GDF-15 difference largely reflects demographic confounding rather than a disease-specific effect. No significant correlations were observed between GDF-15 levels and classical acute-phase reactants. This study underscores the importance of recognizing negative biomarker findings to prevent overinterpretation of biomarker utility. Although serum GDF-15 levels were higher in patients with BS in the unadjusted comparison, this difference was no longer statistically significant after adjustment for sex and BMI. GDF-15 levels were also not associated with disease activity, limiting their utility as a disease-specific biomarker. Further longitudinal studies are needed.

1. Introduction

Behçet’s syndrome (BS) is a systemic vasculitis of the autoinflammatory type that can affect arteries and veins of all sizes. It is primarily characterized by recurrent oral, genital, and intestinal ulcers; skin lesions; uveitis; and parenchymal brain lesions.^[1] In Turkey, prevalence rates are 20 to 420 per 1,00,000 people; in other Asian countries, they are 2.1 to 19.5; in Southern Europe, they are 1.5 to 15.9; in Northern Europe, they are 0.3 to 4.9; and in the United States, they are 5.2.^[2,3]

The etiopathogenesis of BS has not yet been clarified. Its typical geographical distribution suggests a genetic predisposition, and its association with HLA-B51 is noteworthy. A comprehensive meta-analysis has shown that individuals carrying the HLA-B51 gene, which is the dominant variant of the HLA-B5 antigen, have a fivefold increased risk of BS compared to those who do not carry it. It also explains that the HLA-B51/B5 allele is one of the strongest known disease-gene associations in terms of the risk of developing BS.^[4]

In BS, non-HLA common variants in the IL10, IL23R and IL12RB2 loci; the signal transducer and activator of transcription 4 gene; and the endoplasmic reticulum aminopeptidase 1 (ERAP1) gene all play a role. ERAP1 encodes proteins involved in antigen presentation and plays a key role in inflammatory cell recruitment and activation (CCR1-CCR3 genes).^[5,6] These proteins include ERAP1 and the CCR1-CCR3 chemokine receptor family, which are important for inflammatory cell recruitment and activation. Infectious agents and microbiomes, such as the herpes simplex virus-1, the bacterium Staphylococcus aureus, and the bacterium Streptococcus sanguinis, have been identified as potential environmental factors.^[7,8]

From an immunological perspective, the etiology of BS is believed to be caused by the activation of the natural and adaptive immune systems, neutrophils, and T cells; the polarization of proinflammatory cytokines; and the loss of immune tolerance.^[9]

The presence of multisystemic involvement in BS requires a multidisciplinary management approach. Treatment is urgent due to the possibility of mortality and organ sequelae in severe cases. Therefore, understanding its etiology and pathogenesis is important for early diagnosis and and treatment. However, there are currently no biomarkers that enable the early diagnosis of BS. This has increased interest in developing biomarkers for diagnosing BS.

The transforming growth factor (TGF)-β superfamily cytokine growth differentiation factor-15 (GDF-15) has attracted attention in BS as a marker that plays a role in chronic inflammatory pathways, such as rheumatoid arthritis (RA).^[10] Initially defined as a factor secreted by activated macrophages, GDF-15 plays an important role in cell growth and differentiation.^[11] GDF-15 is also known as macrophage inhibitory cytokine-1, nonsteroidal anti-inflammatory drug-inducible gene-1, placental transforming growth factor-beta, prostate-derived factor, and placental bone morphogenetic protein.^[10,12] GDF-15 is abundant in the placenta, but its physiological expression is almost undetectable in most somatic tissues in humans. Elsewhere, GDF-15 is typically induced under stressful conditions to maintain cellular and tissue homeostasis. However, moderate increases in GDF-15 blood levels are observed with age. Elevated GDF-15 levels are primarily associated with pathological conditions, including inflammation, myocardial ischemia, and particularly, cancer.^[13]

Studies have been conducted to determine whether there is a connection between inflammation and rheumatological diseases. Several studies have investigated GDF-15 levels in rheumatologic diseases such as systemic lupus erythematosus (SLE), systemic sclerosis (SSc), antiphospholipid syndrome (APS), idiopathic inflammatory myopathy (IIM), RA, and ankylosing spondylitis (AS). However, studies on BS are limited and conflicting. Given this information, our study aimed to demonstrate the relationship between GDF-15 and BS.

2. Materials and methods

A total of 82 participants were included in this cross-sectional study. The patient group consisted of 49 consecutive adult patients who attended the Rheumatology Outpatient Clinic of Malatya Turgut Özal University between June and September 2025 and fulfilled the International Study Group criteria for Behçet’s disease. The control group consisted of 33 healthy adult volunteers recruited during the same study period from individuals presenting to the hospital and healthy hospital employees. Control participants had no known systemic inflammatory, autoimmune, malignant, chronic renal, or chronic liver disease. No individual or frequency matching for age or sex was performed during participant recruitment. Our study was conducted in accordance with the Declaration of Helsinki.

The minimum required sample size was calculated using G*Power version 3.1 (Heinrich Heine University Düsseldorf, Germany). Assuming a 2-sided independent samples t test, a type I error rate (α) of 0.05, a statistical power (1−β) of 80%, and an effect size (Cohen d) of 0.70, the minimum required sample size was estimated to be 34 participants per group (68 participants in total).^[14] The present study included 49 patients with BS and 33 healthy controls. Thus, the patient group exceeded the calculated target, whereas the control group was 1 participant below the planned sample size.

Our study is cross-sectional and observational. We compared patients’ demographic characteristics (age, gender, body mass index, smoking status and alcohol consumption), laboratory findings and GDF-15 levels. We recorded the current treatments of the BS group (colchicine, immunosuppressants and biological agents) and we performed a subgroup analysis. All patients and control group participants were over the age of 18. Those under 18 years of age, pregnant or breastfeeding women, and individuals with malignancy, chronic renal, or liver failure were excluded from the study. The Ethics Committee of Malatya Turgut Özal University Rectorate Scientific Research Ethics Committee approved this study on May 21, 2025, under the number E-30785963-020-304401. All participants signed an informed consent form before participating in the study.

Blood samples taken from patients and the control group were centrifuged at 3000 RPM to determine GDF-15 levels. The resulting sera were stored at −40°C in 1.5-mL tubes until analysis. GDF-15 concentrations in the serum samples were determined using a commercial enzyme-linked immunosorbent assay (ELISA) test kit (catalog no. E0037Hu, BT Lab, Shanghai, China).

The serum samples were thawed under the appropriate conditions prior to analysis. Measurements were performed at the Research Laboratory of the Faculty of Medicine at Malatya Turgut Özal University using the Human GDF-15 ELISA Test Kit procedure exactly as specified. Serum GDF-15 concentrations were expressed in ng/L.

The test’s measurement range was 10 to 3000 ng/L. Intra-assay reproducibility was evaluated using the coefficient of variation (CV) and was determined to be <5%. Samples were analyzed individually according to the manufacturer’s recommendations.

Disease activity was assessed using the Yosipovitch Behçet’s disease activity classification, a pragmatic clinical categorization tool based on mucocutaneous, ocular, vascular, neurologic, and articular manifestations.^[15] This tool was preferred because it allowed rapid classification of disease severity using the variables available in our cohort (Table 1).

Table 1.

Yosipovitch’s classification of Behçet’s disease activity

Mild	Oral ulcer Mild genital ulcer Symptoms related to the skin
Moderate	Less than 3 arthritis Recurrent genital ulcer Mild anterior uveitis
Severe	Anterior and posterior uveitis Major vascular occlusion Arterial aneurysm Recurrent arthritis in more than 3 joints

The categorical data from the variables included in the study were summarized using numbers and percentages. The normality of the quantitative data was assessed using the Shapiro–Wilk test. Quantitative data with a non-normal distribution were summarized using the median (minimum–maximum), while quantitative data with a normal distribution were summarized using the mean ± standard deviation. The Mann–Whitney U test was used to compare 2 independent groups when data were not normally distributed. The Kruskal–Wallis test was used for comparisons involving more than 2 groups. The chi-square test (with Yates’ correction) and Fisher exact test were used for intergroup comparisons of categorical variables. Univariate logistic regression analyses were performed to evaluate the associations between individual variables and BS group status. To determine whether this association was independent of demographic confounders, a multivariable logistic regression model (group as the dependent variable) and a complementary ANCOVA (log-transformed GDF-15 as the dependent variable) were additionally performed, adjusting for sex and body mass index (BMI). Two participants were excluded from the multivariable logistic regression and ANCOVA analyses because BMI data were unavailable. Therefore, these analyses were performed on 80 participants with complete data. Odds ratios (OR) and 95% confidence intervals (CI) were calculated. The effect size calculations used the r-value for the Mann–Whitney U test and the epsilon-squared (ε²) value for the Kruskal–Wallis test. These values were interpreted according to Cohen criteria: for the Mann–Whitney U test, r: 0.1 = small, 0.3 = medium and 0.5 = large effect; for the Kruskal–Wallis test, ε²: 0.01 = small, 0.06 = medium and 0.14 = large). We examined the presence of relationships between variables using the Spearman Rho correlation coefficient. A P < .05 value was considered statistically significant in the statistical analyses performed. All analyses were performed using IBM SPSS Statistics 26.0 for Windows (New York).

3. Results

BS patients made up 59.8% (n = 49) of the study group, while the control group made up 40.2% (n = 33). Among the BS patients, 51% were male (n = 25) and 49% were female (n = 24). Among the control group, 12.1% were male (n = 4) and 87.9% were female (n = 29). The average age was 44.48 years (±12.86). All patients included in the study were over the age of 18. Height and weight data were available for 81 participants. The mean body weight was 72.91 ± 11.84 kg, the mean height was 164.48 ± 9.25 cm, and the mean body mass index (BMI) was 26.99 ± 4.31 kg/m². BMI differed significantly between the groups (P = .006, r = 0.31), with a higher median BMI in the control group than in the BS group (28.85 vs 26.3 kg/m², respectively). This difference represented a moderate effect size.

Smoking and alcohol consumption data were available for 81 participants. The vast majority of participants (96.3%, n = 78) reported not using alcohol, while only 3.7% (n = 3) reported using it. Regarding cigarette use, 76.5% (n = 62) of participants did not smoke, while 23.5% (n = 19) did. There was no statistically significant difference in alcohol use between the control group and the BS patient group (P = 1.000). Of the control group, 96.88% (n = 31) did not consume alcohol, compared to 95.92% (n = 47) of the BS patient group. Similarly, no statistically significant difference was found between the groups in terms of cigarette use (P = .589). Eighty-one percent of the control group were nonsmokers (n = 26), compared to 73.47% of BS patients (n = 36). Both groups exhibited similar alcohol and cigarette use characteristics. Hemoglobin levels were significantly higher in BS patients (median: 14.1 g/dL) than in the control group (median: 12.9 g/dL; P = .005, r = 0.31). The sedimentation rate was significantly higher in the control group (median: 12 mm/h) than in BS patients (median: 8.5 mm/h; P = .012, R = 0.30). This finding is likely influenced by sex distribution and hemoglobin differences between groups. BS patients had a statistically significantly higher alanine aminotransferase (ALT) level (median: 22 U/L) than the control group (median: 17 U/L; P = .006, r = 0.31). Triglyceride levels were significantly higher in BS patients (median: 126.5 mg/dL) than in the control group (median: 88 mg/dL; P = .017, r = 0.42). Moderate effect sizes were found for BMI, hemoglobin (HGB), sedimentation, ALT, and triglycerides (r = 0.30–0.42). There were no statistically significant differences between the groups in terms of white blood cell (WBC; P = .788), neutrophil count (P = .545), lymphocyte count (P = .403), platelet count (P = .679), C-reactive protein (CRP; P = .251), glucose, total protein (P = .893), albumin (P = .415), aspartate aminotransferase (AST; P = .203), cholesterol (P = .124), or low-density lipoprotein (LDL) levels (P = .278). (Tables 2 and 3).

Table 2.

Comparison of laboratory parameters among groups

Variable	Control: median (minimum-maximum)	BS median (minimum–maximum)	P	r
GDF-15 (ng/L)	212.9 (84–2399.01)	258.49 (33.77–1510.18)	.016 *	0.27
WBC (103/µL)	7.82 (4.34–11.5)	7.36 (4.49–16.14)	.788	0.03
HGB (g/dL)	12.9 (8.6–17.9)	14.1 (10.2–16.7)	.005 *	0.31
Neutrophil (103/µL)	4.56 (2.47–9.5)	4.01 (1.58–11.13)	.545	0.07
Lymphocyte (103/µL)	2.29 (0.51–4.17)	2.39 (0.59–6.33)	.403	0.09
PLT (103/µL)	276 (36–489)	284 (165–456)	.679	0.05
Sedimentation (mm/h)	12 (4–30)	8.5 (1–29)	.012 *	0.30
CRP (mg/L)	0.3 (0.1–9.7)	0.4 (0.1–8.5)	.251	0.13
Glucose (mg/dL)	89 (74–125)	91 (41–452)	.903	0.01
Total protein (g/dL)	7.4 (5.6–9.2)	7.5 (4.7–8.4)	.893	0.02
Albumin (g/dL)	4.20 (3–5.1)	4.3 (2.5–5.2)	.415	0.09
ALT (U/L)	17 (8–48)	22 (11–52)	.006 *	0.31
AST (U/L)	20 (13–35)	23 (7–68)	.203	0.14
Cholesterol (mg/dL)	181 (89–254)	210 (132–250)	.124	0.27
LDL (mg/dL)	107.8 (45.8–159.3)	124.4 (75.1–166)	.278	0.19
Triglyceride (mg/dL)	88 (42.9–284)	126.5 (78–341)	.017 *	0.42

Bold values indicate statistically significant results (P < .05).

The Mann–Whitney U test was used.

Smoking and alcohol data were missing for 1 participant.

ALT = alanine aminotransferase, AST = aspartate aminotransferase, CRP = C-reactive protein, GDF-15 = growth differentiation factor-15, HGB = hemoglobin, LDL = low-density lipoprotein, PLT = platelet, r = effect size, WBC = White blood cell.

^*P < .05.

Table 3.

Results of univariate logistic regression analysis for Behçet’s syndrome

Variable	B	SE	Wald	P	OR (95% CI)
BMI (kg/m2)	−0.179	0.063	8.012	.005 *	0.836 (0.739–0.946)
GDF-15 (ng/L)	0.000	0.001	0.240	.624	1.000 (0.999–1.001)
HGB (g/dL)	0.336	0.153	4.788	.029 *	1.399 (1.036–1.889)
Sedimentation (mm/h)	-0.078	0.039	4.010	.045 *	0.925 (0.857–0.998)
ALT (U/L)	0.083	0.034	5.973	.015 *	1.086 (1.017–1.161)
Triglyceride (mg/dL)	0.012	0.007	3.565	.059	1.012 (1.000–1.025)

Bold values indicate statistically significant results (P < .05).

ALT = alanine aminotransferase, B = regression coefficient, BMI = body mass index, BS: Behçet’s syndrome, CI = confidence interval, GDF-15 = growth differentiation factor-15, HGB = hemoglobin, OR = odds ratio, SE = standard error.

^*P < .05.

Serum GDF-15 levels were significantly higher in patients with BS compared with healthy controls. The median GDF-15 level was 258.49 ng/L in the BS group and 212.9 ng/L in the control group, representing a statistically significant difference (P = .016). Despite partial overlap in distributions, the overall shift toward higher GDF-15 levels in patients suggests increased systemic stress in BS (Fig. 1).

Figure 1.

Comparison of serum GDF-15 levels between patients with Behçet’s syndrome and healthy controls. Data are presented as median (minimum–maximum). Comparisons were performed using the Mann–Whitney U test. GDF-15 = growth differentiation factor-15.

Given the demographic imbalance between the BS and control groups (sex distribution and BMI), a multivariable logistic regression model was performed with group (BS vs control) as the dependent variable and GDF-15, sex, and BMI as covariates to assess whether the GDF-15 difference was independent of these confounders. GDF-15 did not remain a statistically significant predictor of group after adjustment (adjusted P = .528), whereas sex (OR = 5.61, 95% CI 1.64–19.25, P = .006) and BMI (OR = 0.86, 95% CI 0.76–0.98, P = .024) were independently associated with group (Table 4). A complementary ANCOVA using log-transformed GDF-15 as the dependent variable was consistent with this finding: the group effect was not significant after adjusting for sex and BMI (F = 0.005, P = .945).

Table 4.

Results of multivariable logistic regression analysis for Behçet’s syndrome, adjusted for sex and body mass index

Variable	B	SE	Wald	P	OR (95% CI)
GDF-15 (ng/L)	−0.000	0.001	0.399	.528	1.000 (0.998–1.001)
Sex (male)	1.725	0.629	7.525	.006 *	5.612 (1.636–19.249)
BMI (kg/m2)	-0.149	0.066	5.116	.024 *	0.862 (0.757–0.980)

Bold values indicate statistically significant results (P < .05).

Model: Group (BS = 1, Control = 0) ~ GDF-15 + Sex + BMI, N = 80 (BS = 48, Control = 32). A complementary ANCOVA with log-transformed GDF-15 as the dependent variable, group and sex as fixed factors, and BMI as a covariate showed a concordant nonsignificant group effect (F = 0.005, P = .945). Multivariable analyses were performed on 80 participants with complete data because BMI values were unavailable for 2 participants.

ANCOVA = analysis of covariance, B = regression coefficient, BMI = body mass index, BS = Behçet’s syndrome, CI = confidence interval, GDF-15 = growth differentiation factor-15, OR = odds ratio, SE = standard error.

^*P < .05.

Upon examining the distribution of BS patients according to their disease activity index scores, it was found that most patients (57.1%, n = 28) had mild disease activity. 24.5% (n = 12) of patients had moderate disease activity, while 18.4% (n = 9) had severe disease activity (Table 5). Of the 9 patients with severe disease, 3 had ocular involvement, 3 had vascular involvement, 2 had neurological involvement, and 1 had gastrointestinal involvement. No statistically significant difference was found between the groups when GDF-15 levels were compared in BS patients according to disease activity index scores (P = .350, ε² = 0.002). The median GDF-15 value was 273.80 ng/L (range: 33.77–1510.18 ng/L) in patients with mild disease activity. In patients with severe disease activity, the median GDF-15 value was 260.88 ng/L (range: 229.43–359.53 ng/L; Table 6). A correlation analysis of selected variables in patients with BS revealed no statistically significant relationship between GDF-15 levels and age, weight, or height. However, a statistically significant negative correlation was found between GDF-15 levels and BMI (r = −0.272, P = .015). Given the demographic imbalance between groups, including higher BMI and female predominance in the control cohort, the inverse relationship between GDF-15 and BMI may reflect underlying confounding rather than a true biological association. Furthermore, sex-specific differences in inflammatory markers and body composition may further complicate the interpretation of BMI-related associations in this cohort.

Table 5.

Distribution of disease activity index scores in patients with Behçet’s syndrome

Index score	N	%
Mild	28	57.1
Moderate	12	24.5
Severe	9	18.4

Total number of Behçet’s syndrome patients: 49.

N = number, % = percentage.

Table 6.

Comparison of GDF-15 levels according to disease activity index scores in patients with Behçet’s syndrome

Index score	GDF-15 (ng/L) Median (minimum–maximum)	P	ε2
Mild	273.80 (33.77–1510.18)	.350	0.002
Moderate	238.84 (95.00–933.42)
Severe	260.88 (229.43–359.53)

The Kruskal–Wallis test was used. ε²: Epsilon-squared (effect size); 0.01 = small, 0.06 = medium, 0.14 = large effect.

GDF-15 = growth differentiation factor-15.

No difference in GDF-15 levels was observed between genders (P = .530). The average disease duration for BS patients was 7.2 years, ranging from 1 to 22 years. No correlation was observed between disease duration and GDF-15 levels (P = .598). Twenty-one patients received only colchicine, 3 received only azathioprine, and 18 received colchicine in combination with azathioprine, cyclosporine, or corticosteroids. Six patients received anti-tumor necrosis factor (anti-TNF) therapy, and 1 patient had a history of cyclophosphamide treatment. In an exploratory subgroup analysis, patients receiving anti-TNF therapy or with a history of cyclophosphamide treatment (n = 7) had lower GDF-15 levels than the remaining patients (n = 41; median [minimum–maximum]: 251.33 [95.00–359.53] vs 260.88 [33.77–1510.18] ng/L, P = .032).

No significant associations were observed between GDF-15 levels and the evaluated laboratory parameters.

There was no statistically significant correlation between GDF-15 levels and any laboratory parameter. The highest correlation coefficient was observed for triglycerides (r = 0.350, P = .265), followed by neutrophils (r = 0.209, P = .158); however, neither association reached statistical significance. These findings suggest that GDF-15 levels were not significantly associated with hematological parameters, acute-phase reactants, metabolic parameters, or the lipid profile in patients with BS.

4. Discussion

This study demonstrated that serum GDF-15 levels were significantly higher in patients with BS compared with healthy controls. However, GDF-15 levels were not associated with disease activity or acute-phase reactants and were not independently associated with BS after adjustment for sex and BMI. These findings suggest that although GDF-15 is elevated in BS, it does not reflect active inflammatory burden or clinical disease severity. Therefore, GDF-15 should not be considered a diagnostic or disease activity biomarker in the clinical management of BS.

Although BS is associated with genetic and immunological abnormalities and is characterized by chronic and episodic inflammation, its etiology remains unclear. The uncertainty surrounding the etiology of BS has increased the need for various biomarkers and made identifying new inflammatory biological markers clinically important. Similar to the effects of TGF-β, GDF-15 has recently been thought to have an anti-inflammatory effect by reducing leukocyte migration to inflamed areas in many rheumatic diseases.^[16] This has sparked interest in its relationship with rheumatological diseases.

Several studies across rheumatological diseases suggest that GDF-15 is involved in chronic inflammation, tissue stress, and organ-specific damage. Elevated GDF-15 levels have been associated with increased disease activity in SLE and with pulmonary arterial hypertension in both SLE and SSc, indicating a link with vascular and cardiopulmonary involvement.^[17-19] In APS and İİM, GDF-15 has been proposed as a marker of subclinical atherosclerosis and myocardial involvement, respectively.^[20,21] In SSc, higher GDF-15 levels have been linked to skin fibrosis and interstitial lung disease, supporting its role in fibrotic processes.^[22] Similarly, in RA and spondyloarthritis, elevated GDF-15 levels have been correlated with inflammatory activity, CRP positivity, and structural damage.^[23,24] Collectively, these findings suggest that GDF-15 primarily reflects chronic immune stress and tissue remodeling rather than acute systemic inflammation.

The current evidence on GDF-15 in BS is limited and inconsistent. Karadeniz et al reported elevated levels of GDF-15 in patients with BS, with even higher levels observed in patients with vascular involvement. They also demonstrated a significant correlation between GDF-15 levels and femoral vein thickness.^[25] Consistent with these findings, our study also identified significantly higher levels of GDF-15 in patients with BS compared to the healthy control group. However, we did not perform subgroup analyses based on vascular involvement. Consequently, the potential contribution of vascular pathology to the increase in GDF-15 levels could not be assessed specifically in our cohort, which may partially explain the differences observed in the associations.

In the study by Sariyildiz et al, no difference in GDF-15 levels was observed between patients with BS and healthy controls. However, the authors reported positive correlations between GDF-15 levels and CRP, erythema nodosum, arthralgia, and peripheral arthritis.^[26] In contrast to these findings, our study demonstrated significantly elevated GDF-15 levels in patients with BS compared with controls. This discrepancy may be related to differences in study populations and sample size, as our cohort included a slightly larger number of participants, as well as potential variations in clinical characteristics and disease expression.

Our study found no correlation between GDF-15 levels and classic acute phase reactants, such as CRP and erythrocyte sedimentation rate. This supports the notion that GDF-15 reflects chronic inflammation and the immune stress response rather than acute inflammatory flare-ups in BS. The higher sedimentation values in the control group can be explained by the group’s predominance of females and lower hemoglobin levels, indicating the presence of anemia. This situation demonstrates that classic inflammation markers may not always accurately reflect the inflammatory burden in BS.

Another finding of this study was the lower GDF-15 levels observed in the small subgroup of patients receiving anti-TNF therapy or with a history of cyclophosphamide treatment. However, because this exploratory comparison included only 7 clinically heterogeneous patients, the finding should be interpreted cautiously. The cross-sectional design of the study also precludes causal inference regarding the effects of treatment on GDF-15 levels.

Importantly, the present study highlights the clinical relevance of negative biomarker findings. Demonstrating that a biomarker does not correlate with disease activity is valuable for preventing overinterpretation and the inappropriate clinical use of the biomarker. In this context, our findings suggest that GDF-15 should not be included in routine assessments of disease activity or diagnostic algorithms for BS. Rather than identifying a novel positive biomarker, this study is intentionally positioned as a negative biomarker study aimed at clarifying the clinical relevance of GDF-15 in BS. Importantly, these findings highlight the need for cautious interpretation of biomarker elevations in heterogeneous inflammatory diseases and reinforce the value of well-contextualized negative biomarker studies.

The marked sex imbalance between groups represents an important limitation, as GDF-15 levels have been reported to vary according to sex and body composition. Therefore, the observed differences should be interpreted with caution, as they may partially reflect demographic disparities rather than disease-specific effects. Healthy controls were recruited on a voluntary basis without prospective matching for age or sex. Although multivariable logistic regression and ANCOVA were used to adjust for sex and BMI, residual confounding cannot be completely excluded. In addition, disease activity was assessed using the Yosipovitch classification, which is less widely validated than commonly used Behçet’s disease activity instruments; therefore, the findings regarding disease activity should be interpreted cautiously. After adjustment, the between-group difference in GDF-15 lost statistical significance in both models (logistic regression: P = .53; ANCOVA: P = .95), while sex and BMI remained significant predictors. The relatively small sample size, single-center design, and cross-sectional methodology may further limit the generalizability of the findings and preclude assessment of longitudinal changes in GDF-15 levels. Nevertheless, the study reflects real-world clinical heterogeneity, and no consistent association between GDF-15 and disease activity was observed across the heterogeneous clinical spectrum of the cohort.

Additionally, organ-specific analyses were not performed. Nevertheless, the study provides meaningful evidence regarding the limited clinical utility of GDF-15 as a biomarker for BS. It also contributes to the existing literature by clarifying GDF-15’s nonspecific nature.

In summary, although serum levels of GDF-15 are elevated in patients with BS, these levels are not associated with disease activity or independent disease prediction. These results imply that GDF-15 reflects a nonspecific chronic immune stress response rather than clinically relevant disease activity. Further research is needed in the form of larger, longitudinal, and organ-specific studies to elucidate the role of GDF-15 in the context of BS.

5. Conclusions

In conclusion, although serum GDF-15 levels were higher in patients with BS in the unadjusted comparison, this association was no longer statistically significant after adjustment for sex and BMI (adjusted P = .53). GDF-15 levels were not associated with disease activity and were not independently associated with the presence of BS. These findings suggest that GDF-15 may reflect nonspecific systemic stress rather than serving as a reliable disease-specific or disease activity biomarker. Larger, well-balanced, and longitudinal studies are needed to clarify the potential role of GDF-15 in BS.

Acknowledgments

The authors used ChatGPT (OpenAI) for language editing and figure preparation during manuscript development. All scientific content, data analysis, interpretation, and final manuscript approval were performed by the authors.

Author contributions

Conceptualization: Mesude Seda Aydoğdu.

Data curation: Mesude Seda Aydoğdu, Cihat Uçar.

Formal analysis: Mesude Seda Aydoğdu, Sezgin Zontul.

Funding acquisition: Mesude Seda Aydoğdu.

Methodology: Mesude Seda Aydoğdu, Cihat Uçar.

Project administration: Elif Inanç.

Resources: Mesude Seda Aydoğdu, Servet Yolbaş.

Software: Elif Inanç.

Validation: Elif Inanç.

Visualization: Sezgin Zontul.

Writing – original draft: Mesude Seda Aydoğdu.

Writing – review & editing: Mesude Seda Aydoğdu, Zeynep Kaya, Sezgin Zontul, Servet Yolbaş.