ABSTRACT
This retrospective cohort study aimed to identify clinical and laboratory predictors of relapse in patients with Still's disease. A total of 94 patients diagnosed by Yamaguchi or Fautrel criteria across two rheumatology centers between 2012 and 2024 were evaluated, of whom 87 were eligible for analysis. Thirty‐one patients (35%) experienced relapses during follow‐up. Relapse was associated with higher baseline C‐reactive protein (CRP > 100 mg/L), elevated neutrophil counts, and lower albumin levels. Composite ratios such as ferritin/albumin, CRP/albumin, and neutrophil/albumin were also significantly higher in the relapse group. Clinically, a joint‐dominant disease phenotype and initial use of conventional synthetic disease‐modifying antirheumatic drugs (csDMARDs) were more common among patients who relapsed. In multivariate logistic regression, high CRP levels, joint‐dominant phenotype, and csDMARD use emerged as independent predictors of relapse. These findings highlight the predictive value of widely available biomarkers and baseline clinical presentation in stratifying relapse risk in Still Disease. Early recognition of high‐risk patients using these factors may guide more targeted treatment approaches, support earlier initiation of biologic therapies, and ultimately improve long‐term disease outcomes.
Keywords: biomarkers, C‐reactive protein, disease‐modifying antirheumatic drugs, joint phenotype, relapse, Still's disease
Study Highlights
- What is the current knowledge on the topic?
-
○Relapse occurs in a substantial proportion of patients with Still's disease and contributes to increased morbidity and treatment burden. However, reliable predictors of relapse at the time of diagnosis remain incompletely defined.
-
○
- What question did this study address?
-
○This study evaluated clinical features and multiple inflammatory biomarkers to identify predictors of relapse in patients with Still's disease and to assess their predictive performance.
-
○
- What does this study add to our knowledge?
-
○High CRP levels, joint‐dominant disease phenotype, and first‐line csDMARD use were identified as independent predictors of relapse. In addition, several inflammatory ratios were associated with relapse in univariate analyses. This study simultaneously evaluated a broad range of clinical and laboratory parameters and identified simple predictors applicable in routine clinical practice.
-
○
- How might this change clinical pharmacology or translational science?
-
○These findings suggest that readily available inflammatory markers, particularly CRP, together with clinical phenotype may help stratify relapse risk at diagnosis and support earlier therapeutic decision‐making in patients with Still's disease.
-
○
1. Introduction
Still's disease (SD), formerly named as adult‐onset Still's disease (AOSD), is a systemic autoinflammatory disorder of unknown etiology, typically affecting young adults and considered to have a multigenic background. In addition to typical features such as high fever, salmon‐colored rash, lymphadenopathy, and joint involvement, the disease may also result in severe complications, including pulmonary involvement, cardiac involvement and macrophage activation syndrome (MAS) [1, 2].
SD can present in three main clinical patterns. The monocyclic form is characterized by a single episode with systemic manifestations and occasional joint involvement. The intermittent form shows recurrent flares with both systemic and articular features, while the chronic form is dominated by persistent arthritis accompanied by intermittent systemic symptoms, often requiring long‐term treatment [3]. More recently, an additional classification has been proposed, dividing patients into four subgroups: the “juvenile/transient type,” with early onset, rash, myalgia, sore throat, and splenomegaly; the “low acute phase response type,” with only mild inflammatory markers; the “hyperferritinemic type,” associated with markedly elevated ferritin levels; and the “systemic activation type,” which represents the most severe form with intense inflammatory activity. Furthermore, there is increasing recognition that systemic juvenile idiopathic arthritis (sJIA) and Still's disease may represent a spectrum of the same condition [4].
Relapse represents a major clinical challenge in SD, as systemic and joint symptoms reappear after remission, complicating disease management. Previous studies have shown that relapse risk is particularly associated with high ferritin levels, elevated CRP, pro‐inflammatory cytokines such as IL‐18, and various systemic disease activity scores. These findings suggest that relapse is a complex process closely linked to underlying inflammatory burden [5, 6, 7]. Relapses are often triggered by infections, stress, hormonal changes, vaccinations, or discontinuation of immunosuppressive therapy. They are characterized by a hyperinflammatory “cytokine storm”, driven by excessive production of IL‐1β, IL‐6, TNF‐α, and IL‐18, along with natural killer (NK) cell dysfunction and macrophage activation [8].
Between 30% and 50% of patients follow a relapsing–remitting course, which is associated with reduced quality of life and increased risk of severe complications, including MAS. Relapse has also been linked to higher mortality rates, highlighting the importance of early recognition and timely initiation of appropriate treatment [5, 9]. In polycyclic and chronic forms, relapses occur more frequently and contribute to greater morbidity. Therefore, identifying clinical and laboratory predictors of relapse at diagnosis is essential for optimizing management strategies in high‐risk patients [5, 10, 11, 12]. The aim of this study is to identify clinical and laboratory markers associated with relapse in Still's disease, evaluate features that increase relapse risk, and contribute to improving risk stratification at diagnosis.
2. Methods
2.1. Enrollment
Patients were identified from the electronic medical record systems of two major rheumatology centers in Ankara: Gazi University Hospital and Ankara University Hospital. The cohort included individuals diagnosed with Still's disease (SD) between January 2012 and August 2024 according to either the Yamaguchi or Fautrel classification criteria [13, 14]. The patient selection process is illustrated in Figure 1. In line with national healthcare regulations, only patients aged 18 years or older were evaluated in adult rheumatology clinics. Written informed consent was obtained from all participants prior to inclusion, and individuals who did not provide consent were excluded from the study. Ethical approval was obtained from the Ankara University Faculty of Medicine Human Research Ethics Committee on October 25, 2024, under the application number 2024000624–1 (2024/624). This study was conducted in accordance with the principles of the Helsinki Declaration.
FIGURE 1.
Flow diagram of patient inclusion and analysis. Flow diagram showing the selection of patients with Still's disease included in the study. A total of 95 patients diagnosed and treated between January 2012 and August 2024 were screened. Eight patients were excluded due to unavailable relapse outcome data, and 87 patients were included in the final analysis.
2.2. Data Collection
Demographic characteristics, clinical manifestations, laboratory findings, treatment history, relapse episodes, and medication discontinuation records were collected for all patients. Patients with stable disease were routinely monitored every 3–6 months, while those requiring closer follow‐up due to medical necessity were seen at shorter intervals.
2.3. Definitions and Parameters
Patients who experienced one or more relapses during follow‐up were classified as the relapse group, whereas those without relapse were classified as the non‐relapse group.
Relapse was defined as either the appearance of a new systemic or articular manifestation that was not present at the previous visit, a clear worsening of an existing but previously stable clinical finding, or the occurrence of a flare requiring a change in treatment. In addition to these clinical changes, inflammatory reactivation was required, demonstrated by a ≥ 50% increase in at least one acute phase reactant (CRP, ESR, or ferritin) compared with the prior assessment.
Because infections are common in Still's disease, all suspected relapse episodes were evaluated for infectious etiologies through clinical assessment, routine laboratory tests, cultures, ELISA and PCR‐based assays, and imaging studies when indicated. Relapse was diagnosed only when new or worsening clinical findings were not fully explained by an isolated infectious process. However, episodes in which an infection triggered secondary MAS meeting MAS diagnostic criteria were classified as relapse events, as MAS represents a disease‐related hyperinflammatory state rather than a simple infection.
Macrophage activation syndrome (MAS) was diagnosed based on the treating physician's clinical judgment supported by compatible clinical and laboratory findings. In line with the commonly referenced “4H rule,” [15] MAS was suspected when the following features were present: hyperferritinemia, hepatopathy (elevated liver enzymes), hematologic abnormalities, and hyperinflammation. Advanced MAS related biomarkers such as soluble IL‐2 receptor (sCD25), NK‐cell activity, or HScore components were not routinely available during the study period and therefore could not be incorporated into the analyses.
Patients were also categorized according to their predominant clinical presentation at baseline. Individuals presenting primarily with systemic features such as fever, rash, hepatosplenomegaly, lymphadenopathy, or serositis were classified as having a systemic‐dominant phenotype. Those with predominantly musculoskeletal symptoms, especially arthritis, were categorized as joint‐dominant. In cases where systemic and articular features coexisted, classification was based on the category judged by the treating rheumatologist to represent the principal burden of disease at the time of diagnosis. Because of potential overlap, some degree of phenotypic misclassification is possible; this is addressed in the Limitations section.
For corticosteroid‐sparing purposes, conventional synthetic disease‐modifying antirheumatic drugs (csDMARDs) such as methotrexate, cyclosporine, leflunomide, and azathioprine were initiated according to national reimbursement policies and routine clinical practice. Biologic DMARDs (bDMARDs) including IL‐1, IL‐6, and TNF inhibitors were generally reserved for refractory or relapsing cases. Combined DMARD therapy was defined as the concurrent use of csDMARDs and bDMARDs. Diagnostic delay was calculated as the time between symptom onset and confirmed diagnosis.
Parameters assessed in this study were grouped into clinical, biochemical, hematological, and mixed categories. Clinical parameters included fever, weight loss, rash, lymphadenopathy, hepatomegaly, splenomegaly, arthritis, number of affected joints, and patient and physician visual analog scale (VAS) scores. Biochemical parameters included erythrocyte sedimentation rate (ESR), C‐reactive protein (CRP), fibrinogen, ferritin, albumin, triglycerides, liver enzymes (ALT, AST, ALP, GGT), creatinine, and ratios such as ferritin/albumin, fibrinogen/albumin, BUN/albumin, and CRP/albumin. Hematological parameters included hemoglobin, leukocyte, neutrophil, lymphocyte, monocyte, and platelet counts, along with derived indices including the systemic immune‐inflammation index (SII = neutrophil × platelet/lymphocyte), systemic inflammation response index (SIRI = neutrophil × monocyte/lymphocyte), platelet‐to‐lymphocyte ratio (PLR), mean platelet volume (MPV)/lymphocyte, and MPV/platelet. Mixed parameters included AST/lymphocyte, lymphocyte/CRP, monocyte/HDL, neutrophil/albumin, platelet/albumin, CRP/lymphocyte ratio, neutrophil/ferritin, and the Still's Activity Score (SAS) [12].
2.4. Analysis
Continuous and categorical variables were analyzed using appropriate statistical methods. The normality of continuous variables was evaluated with the Kolmogorov–Smirnov test. Normally distributed variables were compared using the independent samples t‐test, whereas non‐normally distributed variables were assessed with the Mann–Whitney U test. Categorical variables were expressed as percentages and compared using the chi‐square or Fisher's exact test as appropriate. Cut‐off values were determined by receiver operating characteristic (ROC) curve analysis, with thresholds selected at the point corresponding to 80% specificity. To improve clinical applicability, cut‐off values were expressed as whole numbers rather than decimals. Logistic regression analyses were performed to identify independent predictors of relapse, and highly correlated parameters were not included simultaneously in the regression models. A two‐sided p‐value < 0.05 was considered statistically significant. Statistical analyses were conducted using SPSS version 25.
3. Results
3.1. Patient Characteristics and Clinical Parameters
Out of the 95 patients diagnosed with Still Disease and followed up in both centers, 7 patients were excluded from the cohort due to missing information regarding relapse status. As a result, a total of 31 patients (35%) were classified in the relapse group, while 56 patients (65%) were classified in the non‐relapse group.
There was no significant difference between the groups in terms of gender, age at diagnosis, diagnostic delay, or mortality rates. However, the follow‐up duration was significantly longer in the relapse group compared to the non‐relapse group (median 94.7 months vs. 52.9 months, p = 0.02). No significant difference was found between the time to relapse of the relapse group and the median follow‐up duration of the non‐relapse group (40.6 (17.1–86.7) months vs. 34.6 (15.8–93.9) months; p: 0.89) (Table 1).
TABLE 1.
Baseline sociodemographic and clinical characteristics of still disease patients according to relapse status.
| Non‐relapse group (n:56) | Relapse group (n:31) | p | |
|---|---|---|---|
| Sociodemographic data | |||
| Female Gender n (%) | 40 (71.4) | 19 (61.3) | 0.33 |
| Age at diagnosis, years, mean (±SD) | 37.5 (13.6) | 37.6 (12.9) | 0.98 |
| Diagnostic delay. months, median (25th–75th percentile) | 1 (0.4–3.2) | 1.5 (1–7.2) | 0.13 |
| Follow‐up duration, months. median (25th–75th percentile) | 52.9 (19.6–104.4) | 94.7 (41.7–142) | 0.02 |
| Patients with comorbidities n (%) | 23 (41.1) | 16 (51.6) | 0.34 |
| Never smokers n (%) | 31 (67.4) | 21 (75) | 0.57 |
| Patients with MAS n (%) | 5 (8.9) | 2 (6.7) | 0.71 |
| Patients with exitus n (%) | 1 (1.8) | 2 (6.7) | 0.24 |
| Clinical parameters | |||
| Fever (> 38°C), n (%) | 54 (98.2) | 30 (100) | 0.64 |
| Weight loss (> 10% in 6 months), n (%) | 35 (70) | 21 (77.8) | 0.46 |
| Presence of skin rash (Typical–Atypical), n (%) | 40 (75.5) | 20 (66.7) | 0.91 |
| Presence of lymphadenopathy, n (%) | 38 (69.1) | 18 (62.1) | 0.34 |
| Hepatomegaly, n (%) | 19 (35.2) | 8 (27.6) | 0.32 |
| Splenomegaly, n (%) | 21 (38.2) | 4 (13.8) | 0.01 |
| Arthritis, n (%) | 18 (32.7) | 11 (36.7) | 0.71 |
| Number of joints with arthritis, median (25th–75th percentile) | 2 (2–4) | 2 (2–8.5) | 0.77 |
| Patient VAS, mean (±SD) | 5.6 (2.1) | 6.4 (1.5) | 0.37 |
| Physician VAS, mean (±SD) | 5.1 (2.2) | 6.1 (1.0) | 0.28 |
Note: Bold values indicate statistically significant differences (p < 0.05).
Clinically, there was no observed difference between the groups regarding fever, weight loss, rash, lymphadenopathy, and arthritis at the time of diagnosis. However, splenomegaly was reported more frequently in the non‐relapse group (38.2% vs. 13.8%, p = 0.01) (Table 1).
3.2. Laboratory Parameters
The laboratory parameters of patients with and without relapse are presented in Table 2. At the time of diagnosis, CRP levels were significantly higher in the relapse group compared to the non‐relapse group (165 vs. 101.5 mg/L; p < 0.01), while albumin levels were lower (2.9 vs. 3.4 g/dL; p < 0.01). Consequently, the ferritin/albumin (1399.1 vs. 388.2; p = 0.05) and CRP/albumin (58.4 vs. 29.6; p < 0.001) ratios were also higher in the relapse group. No significant differences were found in fibrinogen, sedimentation rate, ferritin, or kidney and liver function tests.
TABLE 2.
Baseline laboratory parameters and inflammatory markers of still disease patients according to relapse status.
| Non‐relapse group (n:56) | Relapse group (n:31) | p | |
|---|---|---|---|
| Biochemical parameters | |||
| ESR, mm/h, mean (±SD) | 73.0 (28.3) | 81.5 (23.0) | 0.19 |
| CRP, mg/L, median (25th–75th percentile) | 101.5 (65.7–152) | 165 (126–236) | < 0.01 |
| Fibrinogen, mg/dL, median (25th–75th percentile) | 510 (420–730) | 440 (401–530) | 0.20 |
| Ferritin, ng/mL, median (25th–75th percentile) | 1363 (502–4302) | 4442 (563–11,582) | 0.12 |
| Albumin, g/dL, median (25th–75th percentile) | 3.4 (2.9–3.9) | 2.9 (2.4–3.4) | < 0.01 |
| Triglycerides, mg/dL, median (25th–75th percentile) | 123 (97–219) | 162 (117–251) | 0.36 |
| Liver function test abnormality, n (%) | 36 (69.2) | 17 (68) | 0.91 |
| Serum creatinine, mg/dL, median (25th–75th percentile) | 0.65 (0.58–0.75) | 0.68 (0.57–0.84) | 0.51 |
| Ferritin/Albumin ratio, median (25th–75th percentile) | 388.2 (99.1–1258.3) | 1399.1 (190.7–4494.4) | 0.05 |
| Fibrinogen/Albumin ratio, median (25th–75th percentile) | 116.6 (2.8–232.0) | 2.9 (1.8–151.4) | 0.15 |
| BUN/Albumin ratio, median (25th–75th percentile) | 3.7 (2.4–5.0) | 4.5 (3.2–5.8) | 0.27 |
| CRP/Albumin ratio, median (25th–75th percentile) | 29.6 (15.4–44.8) | 58.4 (31.7–104.3) | < 0.001 |
| Hematological parameters | |||
| Hemoglobin, g/dL, mean (±SD) | 11.4 (1.9) | 11.4 (2.3) | 0.93 |
| Leukocytes, ×103/μL, median (25th–75th percentile) | 13.7 (10.3–19.8) | 19.6 (13.9–27.7) | 0.02 |
| Neutrophils, ×103/μL, median (25th–75th percentile) | 10.2 (7.1–18.3) | 17.6 (11.6–25.6) | 0.01 |
| Lymphocytes, ×103/μL, median (25th–75th percentile) | 1.2 (0.8–1.8) | 1.35 (0.95–1.88) | 0.62 |
| Monocytes, ×103/μL, median (25th–75th percentile) | 0.6 (0.4–0.85) | 0.5 (0.41–0.9) | 0.86 |
| Platelets, ×103/μL, mean (±SD) | 309 (149) | 307 (145) | 0.95 |
| NLR (neutrophil‐to‐lymphocyte ratio), median (25th–75th percentile) | 8.4 (4.4–13.6) | 11.2 (7.8–25.6) | 0.09 |
| SII (systemic immune‐inflammation index), median (25th–75th percentile) (calculated as [neutrophils × platelets]/lymphocytes) | 2327 (1367–4603) | 3704 (1756–6698) | 0.27 |
| SIRI (systemic inflammation response index), median (25th–75th percentile) (calculated as [neutrophils × monocytes]/lymphocytes) | 5.15 (2.63–9.08) | 5.63 (2.36–12.42) | 0.42 |
| PLR (platelet‐to‐lymphocyte ratio), median (25th–75th percentile) | 8.5 (5–14) | 11.2 (7.8–25.6) | 0.15 |
| MPV/Lymphocyte ratio, median (25th–75th percentile) | 7.4 (5.5–11.0) | 7.6 (4.8–10.8) | 0.78 |
| MPV/Platelet ratio, median (25th–75th percentile) | 0.028 (0.019–0.052) | 0.025 (0.020–0.041) | 0.82 |
| Mixed parameters | |||
| AST/Lymphocyte ratio, median (25th–75th percentile) | 31.3 (15.8–98.4) | 35.7 (16.9–101.9) | 0.97 |
| Lymphocyte/CRP ratio, median (25th–75th percentile) | 0.012 (0.007–0.025) | 0.008 (0.004–0.018) | 0.07 |
| Monocyte/HDL ratio, median (25th–75th percentile) | 0.018 (0.010–0.027) | 0.020 (0.009–0.055) | 0.47 |
| Neutrophil/Albumin ratio, median (25th–75th percentile) | 3.4 (2.0–5.3) | 5.6 (3.4–8.7) | 0.006 |
| Platelet/Albumin ratio, mean (±SD) | 92.4 (48.3) | 105.9 (52.2) | 0.26 |
| CRP/Lymphocyte (CRP/Ly) ratio, median (25th–75th percentile) | 1.91 (0.08–84.0) | 17.5 (0.1–148.5) | 0.56 |
| Neutrophil/Ferritin ratio, median (25th–75th percentile) | 0.68 (0.005–6.72) | 0.57 (0.002–10.07) | 0.85 |
| Percentage of patients with severe disease according to SAS score, n (%) | 9 (16.4) | 7 (26.9) | 0.20 |
Note: Bold values indicate statistically significant differences (p < 0.05).
Hematologically, leukocyte (19.6 vs. 13.7; p = 0.02) and neutrophil (17.6 vs. 10.2; p = 0.01) levels were significantly higher in the relapse group, while no marked differences were observed in other hematological parameters. The neutrophil/albumin ratio was also higher in the relapse group (5.6 vs. 3.4; p = 0.006), whereas the lymphocyte/CRP ratio was lower at a borderline significance level (0.008 vs. 0.0012; p = 0.07).
The distribution of key inflammatory biomarkers according to relapse status is illustrated in Figure 2.
FIGURE 2.
Distribution of inflammatory biomarkers according to relapse status in adult‐onset Still's disease. Box plots showing CRP levels (A), neutrophil count (B), albumin levels (C), ferritin/albumin ratio (D), CRP/albumin ratio (E), and neutrophil/albumin ratio (F) in patients with and without relapse. Boxes represent the interquartile range, the horizontal line indicates the median, and whiskers represent 1.5× the interquartile range. Group comparisons were performed using the Mann–Whitney U test, and p‐values are indicated above the corresponding comparisons.
Receiver operating characteristic (ROC) curve analysis was performed to evaluate the discriminative performance of predictors identified in the univariate analysis. CRP > 100 mg/L, ferritin > 5000 ng/mL, neutrophil count > 10,000/μL, joint‐dominant phenotype, and first‐line csDMARD use were evaluated individually. The AUC values were 0.68 for CRP > 100 mg/L, 0.61 for ferritin > 5000 ng/mL, 0.67 for neutrophil count > 10,000/μL, 0.66 for joint‐dominant phenotype, and 0.66 for first‐line csDMARD use.
Variables that remained significant in the multivariate analysis (CRP > 100 mg/L, joint‐dominant phenotype, and first‐line csDMARD use) were subsequently combined into a multivariable prediction model. ROC analysis of this combined model demonstrated improved discriminative performance for predicting relapse (AUC = 0.84). The ROC curves for these predictors and the combined model are presented in Figure 3.
FIGURE 3.
Receiver operating characteristic (ROC) curves for relapse prediction in Still's disease. ROC curves showing the discriminative performance of CRP > 100 mg/L, joint‐dominant phenotype, first‐line csDMARD use, and the combined prediction model. The combined model includes CRP > 100 mg/L, joint‐dominant phenotype, and first‐line csDMARD use. The diagonal line represents the reference line.
3.3. Treatment Comparisons
The differences in treatment are presented in Table 3. The duration of treatment was significantly longer in the relapse group compared to the non‐relapse group as expected (94.7 vs. 40.6 months; p = 0.002). Additionally, the duration of oral steroid use was higher in the relapse group (73.4 vs. 30.5 months; p = 0.02).
TABLE 3.
Treatment characteristics and therapeutic strategies in still disease patients according to relapse status.
| Non‐relapse group (n:56) | Relapse group (n:31) | p | |
|---|---|---|---|
| Treatments | |||
| Total treatment duration (months), median (25th–75th percentile) | 40.6 (17.1–86.7) | 94.7 (41.7–142) | 0.002 |
| Initial treatment duration (time to relapse in relapsing patients, follow‐up duration in non‐relapsing patients), months, median (25th–75th percentile) | 40.6 (17.1–86.7) | 34.6 (15.8–93.9) | 0.89 |
| Number of patients using oral steroids, n (%) | 53 (94.6) | 31 (100) | 0.19 |
| Duration of oral steroid use (months), median (25th–75th percentile) | 30.5 (16.9–83.8) | 73.4 (24.9–127‐4) | 0.02 |
| Number of patients using csDMARDs, n (%) | 38 (67.9) | 31 (100) | < 0.00 |
| Duration of csDMARD use (months), median (25th–75th percentile) | 34.9 (20.5–83.5) | 13.0 (7.4–61.4) | 0.25 |
| Number of patients using bDMARDs, n (%) | 12 (21.4) | 13 (41.9) | 0.04 |
| Duration of bDMARD use (months), median (25th–75th percentile) | 7.6 (1.2–35.9) | 13.5 (5.6–24.1) | 0.5 |
| Patients receiving first‐line monotherapy with steroids, n (%) | 21 (37.5) | 5 (16.1) | 0.03 |
| Patients receiving first‐line csDMARDs, n (%) | 25 (44.6) | 24 (77.4) | 0.003 |
| Patients receiving first‐line bDMARDs, n (%) | 1 (1.8) | 0 (0) | 0.64 |
| Patients receiving first‐line combination therapy, n (%) | 7 (12.5) | 2 (6.5) | 0.3 |
Note: Bold values indicate statistically significant differences (p < 0.05).
The use of csDMARDs was more frequent in the relapse group (100% vs. 67.9%; p < 0.001), and the usage rate of bDMARDs was also significantly higher in this group (41.9% vs. 21.4%; p = 0.04). First‐line solo steroid use was lower in the relapse group (16.1% vs. 37.5%; p = 0.03), while csDMARD use as first‐line therapy was significantly higher (77.4% vs. 44.6%; p = 0.003). However, there was no significant difference between the groups regarding first‐line use of bDMARDs or combined DMARD therapy.
3.4. Regression Analysis
In the univariate regression analysis, several factors were found to be independently associated with relapse. These factors included splenomegaly (OR: 0.25; CI: 0.07–0.84), CRP > 100 mg/L (OR: 5.25; CI: 1.58–17.42), neutrophil count > 10,000/μL (OR: 3.84; CI: 1.25–11.82), ferritin > 5000 ng/mL (OR: 3.63; CI: 1.28–10.30), albumin < 3 g/dL (OR: 3.36; CI: 1.23–9.15), joint‐dominant phenotype (OR: 3.97; CI: 1.54–10.22), first‐line solo steroid use (OR: 0.32; CI: 0.10–0.96), and first‐line csDMARD use (OR: 4.25; CI: 1.57–11.47) (Table 4).
TABLE 4.
Univariate/Multivariate analysis of factors predicting relapse in still disease patients.
| Parameter | Univariate odds ratio (95% CI) | p | Multivariate odds ratio (95% CI). Model 1 | p | Multivariate odds ratio (95% CI). Model 1 (backward selection) |
|---|---|---|---|---|---|
| Splenomegaly | 0.25 (0.07–0.84) | 0.02 | 0.32 (0.06–1.54) | 0.15 | |
| Diagnostic delay | 1.010 (0.984–1.037) | 0.46 | |||
| ESR (erythrocyte sedimentation rate) | 1.012 (0.994–1.032) | 0.19 | |||
| CRP > 100 mg/L | 5.25 (1.58–17.42) | 0.007 | 9.06 (1.42–57.58) | 0.01 | 15.57 (2.97–81.64) |
| Neutrophils > 10.000/μL | 3.84 (1.25–11.82) | 0.019 | 1.63 (0.37–7.15) | 0.51 | |
| Ferritin > 5000 ng/mL | 3.63 (1.28–10.30) | 0.015 | 1.54 (0.34–6.97) | 0.57 | |
| Fibrinogen | 0.997 (0.993–1.001) | 0.17 | |||
| Albumin | 0.40 (0.19–0.86) | 0.02 | |||
| Albumin < 3 g/L | 3.36 (1.23–9.15) | 0.003 | |||
| NLR | 1.008 (0.981–1.035) | 0.57 | |||
| PLR | 1.004 (0.978–1.030) | 0.77 | |||
| Neutrophil/Albumin ratio | 1.111 (0.994–1.241) | 0.06 | |||
| SAS score | 0.519 (0.12–2.24) | 0.38 | |||
| Articular phenotype (vs. Systemic phenotype) | 3.97 (1.54–10.22) | 0.004 | 10.05 (2.22–45.46) | 0.003 | 9.53 (2.21–41.06) |
| First‐line monotherapy with steroids | 0.32 (0.1–0.96) | 0.04 | |||
| First‐line csDMARD use | 4.25 (1.57–11.47) | 0.004 | 10.13 (2.00–51.18) | 0.005 | 9.89 (2.17–44.99) |
Note: Bold values indicate statistically significant differences (p < 0.05).
To minimize the effects of confounding factors, splenomegaly, CRP > 100 mg/L, neutrophil count > 10,000/μL, ferritin > 5000 ng/mL, joint‐dominant phenotype, and first‐line csDMARD use were included in the multivariate analysis. CRP > 100 mg/L (OR: 15.57; CI: 2.97–81.64), joint‐dominant phenotype (OR: 9.53; CI: 2.21–41.06), and first‐line csDMARD use (OR: 9.89; CI: 2.17–44.99) were found to be significantly associated with relapse (Table 4).
4. Discussion
In this study, we evaluated clinical and laboratory predictors of relapse in Still's disease. Patients who experienced relapse had longer follow‐up durations overall; however, the time to first relapse was comparable to the observation time of the non‐relapse group, suggesting that relapse frequency was not solely driven by follow‐up length. Splenomegaly was more common in the non‐relapse group, whereas baseline CRP, leukocyte, and neutrophil levels were significantly higher among relapsing patients. Ratio‐based inflammatory indices such as CRP/albumin, ferritin/albumin, and neutrophil/albumin were also elevated at baseline in the relapse group. From a treatment perspective, both the duration of therapy and the use of csDMARDs and bDMARDs were higher in the relapse cohort. In multivariate analysis, CRP > 100 mg/L, joint‐dominant phenotype, and first‐line csDMARD use emerged as the strongest predictors of relapse.
Consistent with previous cohorts, relapse risk increased with longer follow‐up [5], and although relapses typically occur within the first 3 years, later flares have also been described [5, 11]. The relapse rate of 35% in our cohort aligns with long‐term studies reporting rates approaching 40% [5, 16]. The relatively lower MAS (8%) and mortality (3.4%) rates may be explained by shorter diagnostic delays and early initiation of therapy, as reported in earlier cohorts [5, 16].
Differences in relapse risk between phenotypes support the concept that systemic and joint‐dominant subtypes follow distinct disease trajectories [17, 18, 19]. Splenomegaly, a feature more often associated with systemic disease, appeared protective against relapse, whereas joint‐dominant phenotype was strongly associated with recurrent disease activity. Nonetheless, phenotype assignment remains challenging because systemic disease frequently includes articular manifestations, and early treatment may modify clinical presentation. Although we used predominant baseline features to classify phenotypes, misclassification remains possible, and this may have influenced the observed association between joint‐dominant disease and relapse.
Laboratory analyses demonstrated that relapsing patients had a stronger baseline systemic inflammatory response reflected by elevated CRP, leukocyte, and neutrophil counts and lower albumin levels. Although ferritin/albumin, CRP/albumin, and neutrophil/albumin ratios showed significant differences in univariate analyses, none remained independent predictors in multivariate models, suggesting that these composite markers reflect general inflammatory burden rather than providing incremental predictive value. The ferritin/albumin ratio has previously been linked to mortality in systemic inflammatory diseases [20, 21], while CRP/albumin and neutrophil/albumin ratios are recognized markers of systemic inflammation [22, 23, 24, 25, 26]. Their elevation in our relapse group indicates that high inflammatory load at diagnosis may still have prognostic relevance, even if not independently predictive.
Several clinical and treatment‐related variables were associated with relapse. High CRP, hyperferritinemia, neutrophilia, and hypoalbuminemia were common among relapsing patients, while splenomegaly was protective. These findings are consistent with earlier studies: Meng et al. identified leukocytosis ≥ 13 × 109/L and insufficient steroid exposure as predictors of relapse [9]; Kong et al. described associations with leukocyte count ≥ 30 × 109/L, ESR > 100 mm/h, and ferritin ≥ 1500 ng/mL [16]; and Kang et al. showed that arthritis, lymphadenopathy, and higher ferritin and leukocyte levels were more frequent in relapsing patients [5]. CRP was also higher at baseline among relapsing patients in Kang's study, consistent with our results.
Overall, our findings reinforce the central role of CRP and ferritin as markers of inflammatory burden in Still's disease and highlight the prognostic significance of joint involvement. Importantly, our cohort evaluated a wide range of parameters simultaneously, making this one of the more comprehensive analyses reported to date. Despite the broad biomarker assessment, CRP consistently emerged as the most reliable independent predictor, reaffirming its relevance in clinical monitoring and risk stratification.
Recent therapeutic recommendations prioritizing IL‐1 and IL‐6 inhibitors as first‐line agents underscore the limitations of csDMARDs in controlling systemic inflammation [27, 28]. Kernder et al. demonstrated that first‐line biologics are associated with markedly improved event‐free remission and fewer complications, irrespective of phenotype [29]. In our cohort, however, national reimbursement restrictions necessitated csDMARD use as first‐line therapy in many patients. The association between first‐line csDMARD use and relapse is therefore likely driven by confounding by indication, reflecting physician selection constrained by treatment accessibility rather than true causal risk. Nonetheless, the heightened relapse risk in joint‐dominant patients with elevated CRP, hyperferritinemia, or hypoalbuminemia suggests that earlier transition to biologic therapy may be beneficial in specific high‐risk subgroups.
This study contributes to refining relapse risk stratification in Still's disease by identifying clinical and laboratory markers associated with recurrence. Patients with a joint‐dominant phenotype and/or markedly elevated inflammatory parameters, especially ferritin > 5000 ng/mL or CRP > 100 mg/L, may require closer monitoring and consideration of earlier biologic therapy. Although retrospective design and treatment limitations restrict generalizability, CRP remained the most robust predictor of relapse, underscoring its clinical utility in guiding long‐term disease management.
This study has several limitations. First, its retrospective design resulted in incomplete baseline data for some patients, particularly those diagnosed at external centers. In addition, the relatively limited sample size may have reduced the statistical power to detect weaker associations. Differences in follow‐up duration may have influenced relapse detection, although the time to first relapse was comparable between groups. Second, phenotype classification was based on predominant baseline clinical features, and overlap between systemic and articular manifestations may have led to some degree of misclassification. Third, several potential confounders—including treatment adherence, intercurrent infections, and the rate of glucocorticoid tapering—could not be systematically assessed. Furthermore, evaluation of macrophage activation syndrome (MAS) was limited because advanced biomarkers, such as soluble IL‐2 receptor (sCD25), NK‐cell activity, and components of the HScore, were not routinely available during the study period. The observed association between first‐line csDMARD use and relapse likely reflects confounding by indication related to restricted access to biologic therapies, which may limit the generalizability of our findings to settings with broader biologic availability. Finally, the retrospective structure of the dataset precluded formal time‐to‐event analyses such as Kaplan–Meier or Cox regression, and prospective longitudinal studies are needed to validate these findings.
Despite these limitations, our results remain clinically relevant. In addition, the inclusion of patients from two large tertiary rheumatology centers provides a well‐characterized real‐world cohort and enhances the clinical relevance of the findings. The predictors identified in this study, especially CRP levels and clinical disease phenotype, are routinely assessed at the time of diagnosis in patients with Still's disease. Therefore, these readily available parameters may help clinicians identify patients at higher risk of relapse in routine clinical practice and support closer monitoring and timely therapeutic adjustments when necessary.
5. Conclusion
In this study, high baseline inflammatory burden—particularly elevated CRP, ferritin, and neutrophil‐based indices along with a joint‐dominant clinical phenotype emerged as the strongest predictors of relapse in Still's disease. Splenomegaly appeared protective, whereas first‐line csDMARD use was associated with higher relapse risk, likely reflecting treatment limitations and confounding by indication. These findings highlight the importance of careful risk stratification at diagnosis and suggest that patients with a joint‐dominant phenotype or markedly elevated inflammatory markers may benefit from closer monitoring and consideration of earlier biologic therapy. Prospective studies are needed to validate these predictors and to inform more effective, individualized treatment strategies aimed at preventing relapse.
Author Contributions
İ.Y.Ç. and A.E. wrote the manuscript; İ.Y.Ç. and A.E. designed the research; İ.Y.Ç., Y.M., R.C.K., E.U., E.Ç.B., S.S., M.E.Y., and H.K. performed the research; İ.Y.Ç., R.C.K., and E.Ç.B. analyzed the data; A.E., A.A., M.T., M.A.Ö., and A.T. contributed new reagents/analytical tools.
Funding
The authors have nothing to report.
Conflicts of Interest
The authors declared no conflicts of interest.
Acknowledgments
The authors thank the medical and technical staff of Gazi University Hospital and Ankara University Hospital for their support in data collection and patient care. AI‐assisted language tools were used to improve the readability and language of the manuscript. The authors take full responsibility for the content of the manuscript.
Contributor Information
İbrahim Yahya Çakır, Email: ibrahimyahyacakir@gmail.com.
Yeter Mahmutoğlu, Email: dr.yetermahmutoglu@gmail.com.
Data Availability Statement
The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.
References
- 1. Giacomelli R., Ruscitti P., and Shoenfeld Y., “A Comprehensive Review on Adult Onset Still's Disease,” Journal of Autoimmunity 93 (2018): 24–36. [DOI] [PubMed] [Google Scholar]
- 2. Rao S., Tsang L. S. L., Zhao M., Shi W., and Lu Q., “Adult‐Onset Still's Disease: A Disease at the Crossroad of Innate Immunity and Autoimmunity,” Frontiers in Medicine 9 (2022): 881431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Ruscitti P., McGonagle D., Garcia V. C., et al., “Systematic Review and Metaanalysis of Pharmacological Interventions in Adult‐Onset Still Disease and the Role of Biologic Disease‐Modifying Antirheumatic Drugs,” Journal of Rheumatology 51, no. 5 (2024): 442–451. [DOI] [PubMed] [Google Scholar]
- 4. Ruscitti P., Cantarini L., Nigrovic P. A., McGonagle D., and Giacomelli R., “Recent Advances and Evolving Concepts in Still's Disease,” Nature Reviews Rheumatology 20, no. 2 (2024): 116–132. [DOI] [PubMed] [Google Scholar]
- 5. Kang J. H., “Risk Factors Associated With Relapse of Adult‐Onset Still Disease in Korean Patients,” Medicine (Baltimore) 99, no. 50 (2020): e23579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Maranini B., Ciancio G., and Govoni M., “Adult‐Onset Still's Disease: Novel Biomarkers of Specific Subsets, Disease Activity, and Relapsing Forms,” International Journal of Molecular Sciences 22, no. 24 (2021): 13320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Jung K. H., Kim J. J., Lee J. S., et al., “Interleukin‐18 as an Efficient Marker for Remission and Follow‐Up in Patients With Inactive Adult‐Onset Still's Disease,” Scandinavian Journal of Rheumatology 43, no. 2 (2014): 162–169. [DOI] [PubMed] [Google Scholar]
- 8. Conigliaro P., Priori R., Bombardieri M., et al., “Lymph Node IL‐18 Expression in Adult‐Onset Still's Disease,” Annals of the Rheumatic Diseases 68, no. 3 (2009): 442–443. [DOI] [PubMed] [Google Scholar]
- 9. Meng J., Chi H., Wang Z., et al., “Characteristics and Risk Factors of Relapses in Patients With Adult‐Onset Still's Disease: A Long‐Term Cohort Study,” Rheumatology 60, no. 10 (2021): 4520–4529. [DOI] [PubMed] [Google Scholar]
- 10. Jia J., Shi H., Liu M., et al., “Cytomegalovirus Infection May Trigger Adult‐Onset Still's Disease Onset or Relapses,” Frontiers in Immunology 10 (2019): 898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Seo J. Y., Suh C. H., Jung J. Y., Kim A. R., Yang J. W., and Kim H. A., “The Neutrophil‐To‐Lymphocyte Ratio Could Be a Good Diagnostic Marker and Predictor of Relapse in Patients With Adult‐Onset Still's Disease: A STROBE‐Compliant Retrospective Observational Analysis,” Medicine (Baltimore) 96, no. 29 (2017): e7546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kalyoncu U., Kasifoglu T., Omma A., et al., “Derivation and Validation of Adult Still Activity Score (SAS),” Joint, Bone, Spine 90, no. 1 (2023): 105499. [DOI] [PubMed] [Google Scholar]
- 13. Yamaguchi M., Ohta A., Tsunematsu T., et al., “Preliminary Criteria for Classification of Adult Still's Disease,” Journal of Rheumatology 19, no. 3 (1992): 424–430. [PubMed] [Google Scholar]
- 14. Fautrel B., Le Moël G., Saint‐Marcoux B., et al., “Diagnostic Value of Ferritin and Glycosylated Ferritin in Adult Onset Still's Disease,” Journal of Rheumatology 28, no. 2 (2001): 322–329. [PubMed] [Google Scholar]
- 15. Fardet L., Galicier L., Lambotte O., et al., “Development and Validation of the HScore, a Score for the Diagnosis of Reactive Hemophagocytic Syndrome,” Arthritis and Rheumatology 66, no. 9 (2014): 2613–2620. [DOI] [PubMed] [Google Scholar]
- 16. Kong X. d., Xu D., Zhang W., Zhao Y., Zeng X., and Zhang F., “Clinical Features and Prognosis in Adult‐Onset Still's Disease: A Study of 104 Cases,” Clinical Rheumatology 29, no. 9 (2010): 1015–1019. [DOI] [PubMed] [Google Scholar]
- 17. Mitrovic S. and Fautrel B., “Clinical Phenotypes of Adult‐Onset Still's Disease: New Insights From Pathophysiology and Literature Findings,” Journal of Clinical Medicine 10, no. 12 (2021): 2633. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Neau P. A., El‐Jammal T., Javaux C., et al., “The Spectrum of Still's Disease: A Comparative Analysis of Phenotypic Forms in a Cohort of 238 Patients,” Journal of Clinical Medicine 11, no. 22 (2022): 6703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Franchini S., Dagna L., Salvo F., Aiello P., Baldissera E., and Sabbadini M. G., “Efficacy of Traditional and Biologic Agents in Different Clinical Phenotypes of Adult‐Onset Still's Disease,” Arthritis and Rheumatism 62, no. 8 (2010): 2530–2535. [DOI] [PubMed] [Google Scholar]
- 20. Liu F. and Liu Z., “Association Between Ferritin to Albumin Ratio and 28‐Day Mortality in Patients With Sepsis: A Retrospective Cohort Study,” European Journal of Medical Research 28, no. 1 (2023): 414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Taşkin Ö., Yilmaz A., Soylu V. G., Demir U., and Inan F. Ç., “Ferritin / Albumin Ratio Could Be a New Indicator of COVID‐19 Disease Mortality,” Journal of Infection in Developing Countries 17, no. 1 (2023): 37–42. [DOI] [PubMed] [Google Scholar]
- 22. He Y., Tang J., Wu B., Yang B., Ou Q., and Lin J., “Correlation Between Albumin to Fibrinogen Ratio, C‐Reactive Protein to Albumin Ratio and Th17 Cells in Patients With Rheumatoid Arthritis,” Clinica Chimica Acta 500 (2020): 149–154. [DOI] [PubMed] [Google Scholar]
- 23. Cao X., Zhao M., Li H., et al., “Three New Inflammatory Markers C Reactive Protein to Albumin Ratio, Neutrophil to Lymphocyte Ratio, and Platelet to Lymphocyte Ratio Correlated With Relapsing Polychondritis Disease Activity Index,” Clinical Rheumatology 40, no. 11 (2021): 4685–4691. [DOI] [PubMed] [Google Scholar]
- 24. Hu Z., Wang J., Xue Y., et al., “The Neutrophil‐To‐Albumin Ratio as a New Predictor of All‐Cause Mortality in Patients With Heart Failure,” Journal of Inflammation Research 15 (2022): 701–713. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wang B., Li D., Cheng B., Ying B., and Gong Y., “The Neutrophil Percentage‐To‐Albumin Ratio Is Associated With All‐Cause Mortality in Critically Ill Patients With Acute Kidney Injury,” BioMed Research International 2020, no. 1 (2020): 5687672. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Wang X., Wang J., Wu S., Ni Q., and Chen P., “Association Between the Neutrophil Percentage‐To‐Albumin Ratio and Outcomes in Cardiac Intensive Care Unit Patients,” International Journal of General Medicine 14 (2021): 4933–4943. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Leavis H. L., van Daele P. L. A., Mulders‐Manders C., et al., “Management of Adult‐Onset Still's Disease: Evidence‐ and Consensus‐Based Recommendations by Experts,” Rheumatology (Oxford, England) 63, no. 6 (2024): 1656–1663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Fautrel B., Mitrovic S., Matteis A. D., et al., “EULAR/PReS Recommendations for the Diagnosis and Management of Still's Disease, Comprising Systemic Juvenile Idiopathic Arthritis and Adult‐Onset Still's Disease,” Annals of the Rheumatic Diseases 83 (2024): 1614–1627. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kernder A., Filla T., Friedrich R., et al., “First‐Line Biological Versus Conventional Synthetic Disease‐Modifying Antirheumatic Drug Therapy in Adult‐Onset Still's Disease: A Multicentre, Retrospective, Propensity Weighted Cohort Study,” Lancet Rheumatology 7 (2025): e415–e423. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.