Abstract
Introduction
Familial Mediterranean fever (FMF) is the most common monogenic autoinflammatory disease. Colchicine is the first-line treatment, yet 5–10% of patients are resistant, increasing the risk of complications like amyloidosis. In 2023, Batu et al proposed the Turkish Paediatric Autoinflammatory Diseases (TURPAID) score to predict colchicine resistance in paediatric FMF at diagnosis. Its utility in broader populations is unknown. We assessed its performance in paediatric and adult FMF patients from the international Juvenile Inflammatory Rheumatism (JIR) cohort.
Methods
We retrospectively analysed 236 genetically confirmed FMF patients treated with colchicine for ≥6 months. Patients were classified as colchicine-sensitive (CoS) or colchicine-resistant (CoR) based on the initiation of biologic therapy, which served as an operational definition of resistance, and matched for age and sex. The TURPAID score (range 0–4; resistance threshold ≥2) was retrospectively applied. Receiver operating characteristic (ROC) curves were used to assess predictive value.
Results
A TURPAID score ≥2 was observed in 89% of paediatric and 76% of adult CoS patients. Mean scores were significantly higher in paediatric-onset FMF. ROC analysis showed poor discrimination in both paediatric and adult groups (area under the curve=0.6). Clinical features and attack patterns varied by age. The genetic component (1.5 points for MEFV exon 10 mutations) contributed to overclassification, reducing predictive accuracy.
Conclusion
The TURPAID score did not effectively predict colchicine resistance in the JIR cohort. Its limited generalisability may stem from age-related differences, recall bias and excessive genetic weighting. Genetic results should be a prerequisite and not a determinant of colchicine resistance prediction scores in FMF.
Keywords: Familial Mediterranean Fever, Hereditary Autoinflammatory Diseases, Therapeutics, Amyloidosis
WHAT IS ALREADY KNOWN ON THIS TOPIC
The Turkish Paediatric Autoinflammatory Diseases (TURPAID) score was proposed to predict colchicine resistance at Familial Mediterranean fever (FMF) diagnosis, but its performance outside the original paediatric cohort and in genetically confirmed FMF populations remained unknown.
WHAT THIS STUDY ADDS
In a large multinational cohort, the TURPAID score showed poor discrimination between colchicine-sensitive and colchicine-resistant patients, regardless of age. The strong weighting of MEFV exon 10 biallelic mutations led to systematic overclassification and limited predictive value.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Future prediction tools should avoid genetic overweighting and rely on prospectively collected clinical and biomarker data to define colchicine resistance more accurately.
Introduction
Familial Mediterranean fever (FMF) is the most common monogenic autoinflammatory disease in the world.1 It primarily affects populations of Mediterranean descent, including those of Turkish, Armenian, Arab and North African origin. FMF is a recessive disease classically associated with class IV or V variants in the MEFV gene, encoding pyrin, a key regulator of the innate immune response. Pathogenic variants in MEFV lead to inappropriate activation of inflammatory pathways, resulting in recurrent self-limited episodes of fever and serosal inflammation. Clinically, patients typically present with febrile episodes associated with abdominal, joint, chest or testicular pain, accompanied by elevated inflammatory markers. FMF is classically a paediatric-onset disease, with the majority of patients developing symptoms before the age of 4.2 Early diagnosis in childhood is common in endemic regions, allowing timely initiation of colchicine therapy. However, in some cases, diagnosis is delayed until adulthood.3 4 Adult-diagnosed FMF can present atypically, sometimes without classic fever episodes, making diagnosis more difficult. These atypical presentations in adulthood may contribute to underdiagnosis or misdiagnosis, leading to prolonged periods of uncontrolled inflammation.
Colchicine remains the cornerstone of FMF treatment. It has shown strong efficacy in reducing the frequency and severity of inflammatory flares and, most importantly, in preventing inflammatory (AA) amyloidosis—a serious and potentially irreversible complication of chronic systemic inflammation.5 However, approximately 5–10% of patients are considered to be resistant to colchicine, either because inflammatory symptoms persist despite adequate dosing or because intolerance limits its use.6 Identification of colchicine resistance is critical as it guides the need for increased monitoring and potential escalation to biologic therapies, such as IL-1 inhibitors, to avoid long-term organ damage. According to the current European Alliance of Associations for Rheumatology (EULAR) recommendations,7 treatment with colchicine should be initiated as soon as a clinical diagnosis of FMF is made. Compliant patients with active disease despite the maximum tolerated dose of colchicine can be considered non-respondent or resistant, and alternative biologic therapies are indicated for these patients.
In 2023, Batu et al8 proposed a predictive scoring system at the time of diagnosis for colchicine resistance based on the Turkish Paediatric Autoinflammatory Diseases (TURPAID) paediatric cohort in Turkey. This score incorporates clinical and genetic parameters and predicts colchicine resistance when the total score is ≥2, with a maximum of 4 points. The development of such a predictive tool is an important step towards personalised management of FMF. However, the external validity of this score outside the original cohort remains unknown, particularly in different clinical settings or in adult patients.
The aim of our study was to evaluate the applicability of the TURPAID predictive score in a real-world, multinational setting by testing it in both paediatric and adult FMF patients enrolled in the Juvenile Inflammatory Rheumatism (JIR) cohort. Our aim was to assess its performance across age groups and to determine whether it can reliably predict colchicine resistance in wider clinical practice.
Patients and methods
TURPAID score
The TURPAID score combines clinical and genetic variables to predict colchicine resistance at the time of FMF diagnosis, based exclusively on disease manifestations observed before diagnosis. The score includes the following items: age at symptom onset ≤3 years, attack frequency prior to diagnosis ≥1 per month and presence of chest pain—each contributing 0.5 points; arthritis before diagnosis contributes 1 point; and homozygosity or compound heterozygosity for exon 10 MEFV mutations contributes 1.5 points. The maximum total score is 4. In the original validation cohort, a score ≥2 predicted colchicine resistance with a sensitivity of 93.5% and a specificity of 53.8%.
JIR cohort
Data were obtained from the JIR cohort, a multinational, multicentre registry designed for the collection of standardised clinical data on patients with inflammatory rheumatic diseases. The registry is approved by the French National Commission on Informatics and Liberties (approval number 914677, granted in March 2015).
All participants provided informed consent or non-opposition (depending on the legislation of the patient’s country) for the use of their anonymised data to be used for research purposes in compliance with applicable data protection regulations.
To assess whether the TURPAID score could be applied to patients in the JIR cohort, we extracted from the database the same variables that constitute the score: age at symptom onset ≤3 years, attack frequency prior to diagnosis ≥1 per month, presence of chest pain or arthritis before diagnosis, and homozygosity or compound heterozygosity for exon 10 MEFV mutations.
Patients
To be included in the present study, patients had to fulfil the 2019 Eurofever/PRINTO group classification criteria9 with a confirmatory MEFV genotype and to receive colchicine treatment for at least 6 months.
Colchicine resistance was defined by biologic DMARD (bDMARD) prescription, while colchicine-sensitive (CoS) patients did not receive any bDMARDs. In this cohort, all biologic therapies consisted exclusively of IL-1 inhibitors, and none were prescribed for comorbid conditions, ensuring that biologic initiation reflected FMF-related disease activity rather than alternative indications. Sensitive and resistant patients were randomly matched based on age and sex.
Statistical analysis
Descriptive analyses are presented using proportions, medians, means, SDs and IQRs as appropriate. Categorical variables were compared by using the χ2 test or Fisher’s exact test as appropriate.
Receiver operating characteristic (ROC) curves were generated to assess the ability to predict colchicine using the TURPAID score (≥2 indicating resistance). The area under the curve (AUC) and 95% CIs were calculated. Statistical analysis was performed with EasyMedStat (V.3.38; ).
Results
Among patients with FMF from the JIR cohort, 118 individuals with colchicine-resistant (CoR) FMF were identified. For each CoR patient, one CoS control was randomly selected, matched by age and sex, resulting in a total of 236 patients included in the analysis (118 CoS and 118 CoR individuals). Overall, 72% of patients had received a diagnosis during childhood. General characteristics of both groups are detailed in online supplemental table 1.
Descriptive analysis of CoS patients
Among CoS patients (table 1), those diagnosed in adulthood had a significantly later age at symptom onset compared with those diagnosed in childhood (mean age 8.5 vs 3.7 years; p<0.001). Only 7% of adult-diagnosed patients reported symptom onset at or before 3 years of age (p=0.001). The mean age at diagnosis for this subgroup was 33 years. No significant differences were observed between adult- and paediatric-diagnosed patients regarding the proportion of patients with a frequency of attacks ≥1 per month, average attack duration or associated clinical manifestations prior to diagnosis.
Table 1. Descriptive analysis of CoS FMF patients (n=118).
| Characteristics of control FMF patients | All CoS (n=118) | Paediatric diagnostic (n=85) |
Adult diagnostic (n=33) | P value paediatric versus adult |
|---|---|---|---|---|
| Female, n (%) | 67 (56.8) | 51 (60) | 16 (48.48) | 0.354 |
| Age at symptom onset, mean (SD) (years) | 5.03 (4.8) | 3.7 (3.3) | 8.46 (6.08) | <0.001 |
| Age at first symptoms ≤3 years, n (%) | 55 (46.6) | 48 (56.5) | 7 (21.2) | 0.001 |
| Age at diagnosis, mean (SD) (years) | 14.09 (13.8) | 6.76 (4.59) | 32.98 (11.45) | – |
| Attack duration, mean (SD) (days) | 2.81 (1.65) | 2.59 (1.12) | 3.39 (2.5) | 0.172 |
| Parental consanguinity, n (%) (data availability) | 12 (11.8) (102/118) | 9 (11.7) (77/85) | 3 (12) (25/33) | >0.999 |
| Homozygosity or compound heterozygosity for exon 10 MEFV variants, n (%) | 118 (100) | 85 (100) | 33 (100) | – |
| Frequency of attacks ≥1 per month, n (%) | 69 (58.5) | 53 (62.4) | 16 (48.5) | 0.244 |
| Clinical findings before diagnosis, n (%) | ||||
| Fever ELE Arthralgia Arthritis Abdominal pain Chest pain |
82 (69.5) 43 (36.4) 71 (60.2) 20 (16.9) 102 (86.4) 59 (50) |
59 (69.4) 28 (32.9) 55 (64.7) 13 (15.3) 76 (89.4) 47 (55.3) |
23 (69.7) 15 (45.5) 16 (48.5) 7 (21.2) 26 (78.8) 12 (36.4) |
>0.999 >0.292 0.16 0.427 0.143 0.101 |
| TURPAID score, mean (SD) | 2.44 (0.6) | 2.52 (0.59) | 2.24 (0.57) | 0.028 |
| Patients with TURPAID score ≥2, n (%) | 101 (85.6) | 76 (89.4) | 25 (75.8) | 0.079 |
Variables in bold are statistically significant.
CoS, colchicine-sensitive; ELE, erysipela-like erythema; FMF, Familial Mediterranean fever; TURPAID, Turkish Paediatric Autoinflammatory Diseases.
Notably, the TURPAID score was ≥2 in 89% of paediatric-diagnosed patients and in 76% of adult-diagnosed patients. The mean TURPAID score was significantly higher in patients diagnosed during childhood compared with those diagnosed in adulthood (2.52 vs 2.24; p=0.028).
Descriptive analysis of CoR patients
Among CoR patients (table 2), those diagnosed in adulthood had a significantly later age at symptom onset compared with paediatric-diagnosed patients (mean age 7.8 vs 3.8 years; p<0.001). Only 6% of adult-diagnosed patients reported symptom onset at or before 3 years of age (p<0.001). The mean age at diagnosis was 33 years in adult-diagnosed patients and 6.7 years in paediatric-diagnosed patients.
Table 2. Descriptive analysis of CoR familial Mediterranean fever patients (n=118).
| Characteristics of CoR patients | All CoR (n=118) | Paediatric diagnostic (n=85) |
Adult diagnostic (n=33) | P value paediatric versus adult |
|---|---|---|---|---|
| Female, n (%) | 67 (56.8) | 34 (40) | 17 (51.52) | 0.354 |
| Age at symptom onset, mean (SD) (years) | 4.93 (4.81) | 3.84 (3.54) | 7.74 (6.41) | <0.001 |
| Age at first symptoms ≤3 years, n (%) | 52 (44.1) | 46 (54.12) | 6 (18.18) | <0.001 |
| Age at diagnosis, mean (SD) (years) | 14.14 (13.87) | 6.78 (4.64) | 33.08 (11.81) | – |
| Attack duration, mean (SD) (days) | 2.83 (1.2) | 2.65 (1.01) | 3.29 (1.53) | 0.008 |
| Parental consanguinity, n (%) (data availability) | 13 (12.9) (101/118) | 9 (12) | 4 (15.38) | 0.736 |
| Homozygosity or compound heterozygosity for exon 10 MEFV variants, n (%) | 118 (100%) | 82 (100) | 34 (100) | – |
| Frequency of attacks ≥1 per month, n (%) | 91 (77.1) | 71 (83.53) | 20 (60.61) | 0.013 |
| Clinical findings before diagnosis, n (%) | ||||
| Fever ELE Arthralgia Arthritis Abdominal pain Chest pain |
101 (85.6) 56 (47.5) 95 (80.5) 48 (40.7) 113 (95.8) 51 (43.2) |
74 (87.0) 41 (48.2) 67 (78.8) 35 (41.2) 81 (95.3) 40 (47.1) |
27 (81.8) 15 (45.5) 28 (84.9) 13 (39.4) 33 (97) 11 (33) |
0.56 0.947 0.607 >0.999 >0.999 0.253 |
| TURPAID score, mean (SD) | 2.73 (0.69) | 2.84 (0.692) | 2.45 (0.63) | 0.011 |
| Patients with TURPAID score ≥2, n (%) | 111 (94.1) | 82 (96.47) | 29 (87.88) | 0.095 |
Variables in bold are statistically significant.
CoR, colchicine-resistant; ELE, erysipela-like erythema; TURPAID, Turkish Paediatric Autoinflammatory Diseases.
Adult-diagnosed patients experienced longer attack durations (3.3 vs 2.65 days; p=0.008), while paediatric-diagnosed patients more frequently reported ≥1 attack per month (85.5% vs 60.6%; p=0.013). There was no significant difference regarding clinical findings before diagnosis.
A TURPAID score ≥2 was observed in 96.5% of paediatric-diagnosed patients and 87.9% of adult-diagnosed patients. The mean TURPAID score was also significantly higher in paediatric-diagnosed patients (2.84 vs 2.45; p=0.011).
Comparative analysis of CoS and CoR patients
The main clinical differences between CoS and CoR patients were observed in the paediatric subgroup, where CoR patients had a significantly higher frequency of attacks (p=0.003), as well as more frequent fever (p=0.009) and arthritis (p<0.001). In contrast, adult-diagnosed CoR patients more commonly reported arthralgia (p=0.004).
The mean TURPAID score was ≥2 across all patient groups, with no statistically significant differences in the proportion of patients reaching this threshold between CoS and CoR, regardless of age at diagnosis. Additional comparative data are presented in table 3.
Table 3. Comparative analysis of CoS and CoR familial Mediterranean fever patients.
| Paediatric-diagnosis patients (n=170) | Adult-diagnosis patients (n=66) | |||||
|---|---|---|---|---|---|---|
| CoS (n=85) |
CoR (n=85) |
P value | CoS (n=33) |
CoR (n=33) |
P value | |
| Female, n (%) | 51 (60) | 51 (60) | >0.999 | 16 (48.5) | 16 (48.5) | >0.999 |
| Age at symptom onset, mean (SD) (years) | 3.7 (3.3) | 3.84 (3.5) | 0.984 | 8.46 (6.2) | 7.74 (6.4) | 0.626 |
| Age at first symptoms ≤3 years, n (%) | 48 (56.5) | 46 (54.1) | 0.877 | 7 (21.2) | 6 (18.2) | >0.999 |
| Age at diagnosis, mean (SD) (years) | 6.76 (4.6) | 6.78 (4.6) | 0.976 | 32.98 (11.6) | 33.08 (11.8) | 0.995 |
| Attack duration, mean (SD) (days) | 2.59 (1.1) | 2.65 (1.0) | 0.689 | 3.39 (2.5) | 3.29 (1.5) | 0.187 |
| Frequency of attacks ≥1 per month, n (%) | 53 (62.4) | 71 (83.5) | 0.003 | 16 (48.5) | 20 (60.6) | 0.458 |
| Clinical findings, n (%) | ||||||
| Fever ELE Arthralgia Arthritis Abdominal pain Chest pain |
59 (69.4) 28 (32.9) 55 (64.7) 13 (15.3) 76 (89.4) 47 (55.3) |
74 (87.0) 41 (48.2) 67 (78.8) 35 (41.2) 81 (95.3) 40 (47.1) |
0.009 0.061 0.061 <0.001 0.248 0.357 |
23 (69.7) 15 (45.5) 16 (48.5) 7 (21.2) 26 (78.8) 12 (36.4) |
27 (81.8) 15 (45.5) 28 (84.9) 13 (39.4) 33 (97) 32 (97) |
0.389 >0.999 0.004 0.181 0.054 >0.999 |
| TURPAID score, mean (SD) | 2.52 (0.6) | 2.84 (0.7) | 0.006 | 2.24 (0.6) | 2.45 (0.6) | 0.177 |
| Patients with TURPAID score ≥2, n (%) | 76 (89.4) | 82 (96.5) | 0.132 | 25 (75.8) | 29 (87.9) | 0.339 |
Variables in bold are statistically significant.
CoR, colchicine-resistance; CoS, colchicine-sensitive; ELE, erysipela-like erythema; TURPAID, Turkish Paediatric Autoinflammatory Diseases.
ROC analysis
ROC curves were used to evaluate the predictive value of the TURPAID score for colchicine resistance at diagnosis in both paediatric- and adult-diagnosed FMF patients (figure 1).
Figure 1. Receiver operating characteristic curves evaluating the performance of the TURPAID score to predict colchicine resistance at diagnosis in paediatric- and adult-diagnosed familial Mediterranean fever patients from the Juvenile Inflammatory Rheumatism cohort. AUC, area under the curve; Se, sensibility; Spe, specificity; TURPAID, Turkish Paediatric Autoinflammatory Diseases.
In paediatric-diagnosed patients, the AUC was 0.6 (95% CI 0.5 to 0.7). A TURPAID score ≥2 predicted colchicine resistance with a sensitivity of 78.8% and a specificity of 31.8%. The optimal threshold according to Youden’s index was 3.5, yielding a sensitivity of 11.8% and a specificity of 97.6%.
In adult-diagnosed patients, the AUC was also 0.6 (95% CI 0.5 to 0.7). At the ≥2 threshold, sensitivity and specificity were both 51.5%. The Youden-optimal cut-off was 3, corresponding to a sensitivity of 9.1% and a specificity of 97%.
Discussion
In this multicentric study, we aimed to assess the applicability of the TURPAID predictive score for colchicine resistance in a real-life international cohort of FMF patients from the JIR registry. To our knowledge, this is the first external validation attempt of this score, including both paediatric- and adult-diagnosed patients.
Our results show that although the majority of patients in all age groups had a TURPAID score ≥2, this threshold did not effectively discriminate CoS from CoR patients. ROC curve analysis yielded modest AUC values of 0.6 in both paediatric and adult groups, with relatively wide CIs. The relatively wide CIs observed further support the limited generalisability of the TURPAID score in this more heterogeneous, real-world population.
We also found differences in clinical presentation between children and adults. Paediatric-diagnosed patients more frequently reported frequent attacks, fever and arthritis, whereas adult-diagnosed patients more commonly experienced arthralgia and longer attack durations. These differences likely reflect both biological age-related variations and diagnostic delays. Adults diagnosed later in life may have had symptoms decades earlier and may not accurately recall early disease features. This recall bias likely contributes to the observed differences between paediatric- and adult-diagnosed patients and to the poor discriminatory performance of the TURPAID score in adults.
A key issue concerns the genetic weighting of the TURPAID score. The score attributes 1.5 points for homozygosity or compound heterozygosity for exon 10 MEFV variants. In our cohort, all patients carried biallelic pathogenic exon 10 variants, making the genetic component a major driver of total score. As described in previous studies, heterozygous patients typically exhibit milder inflammatory phenotypes, lower biomarker levels and very low rates of true colchicine resistance.10 In our CoS subgroup, nearly all patients nevertheless reached a TURPAID score ≥2, despite controlled disease activity, underscoring how the genetic component inflates scoring independently of clinical severity. This excessive weighting likely increases false-positive classifications.
Of note, the original TURPAID cohort displayed substantial genetic heterogeneity: only 62.6% of patients underwent MEFV sequencing, and a proportion of those tested carried non-classifying genotypes, meaning that some individuals could not be confidently classified as FMF according to current Eurofever/PRINTO criteria. Patients without these variants may not truly have FMF, and their lack of response to colchicine could reflect another entity, such as Undifferentiated Autoinflammatory Disease (USAID).11 Evaluating the score in a genetically confirmed population, therefore, provides a more stringent and biologically coherent assessment of its predictive accuracy.
For these reasons, we intentionally restricted our analysis to patients with biallelic exon 10 variants, ensuring a biologically consistent FMF population in whom the concept of colchicine resistance is clinically meaningful. This avoids the diagnostic uncertainty associated with including heterozygous or non-classified individuals and reduces genetic noise that would otherwise obscure the evaluation of the score.
An important methodological limitation is the pragmatic definition of colchicine resistance. In the absence of systematically collected data on adherence, maximally tolerated colchicine doses and longitudinal inflammatory biomarkers, we defined resistance by the initiation of biologic DMARD therapy. In real-life practice, escalation to maximally tolerated colchicine doses is inconsistently applied and often limited by intolerance or variability in physician practice patterns, making the consensus definition difficult to operationalise in large retrospective registries, as shown in a recent international study.12 Nevertheless, in the context of heterogeneous real-world care across participating countries, the initiation of biologics represents a feasible surrogate for clinically significant colchicine resistance. Although this surrogate definition is not perfect and has limitations, it reflects actual clinical decision-making and reliably identifies patients in whom colchicine failure is considered meaningful. Although our population was ethnically relatively homogeneous, variability in physician practice patterns and access to biologic therapies across centres may still have influenced the threshold for biologic initiation. As highlighted above, biologic initiation could theoretically reflect indications unrelated to colchicine resistance; however, in our cohort, all biologic therapies consisted exclusively of IL-1 inhibitors, with no use of anti-TNF agents or other biologic therapies for comorbidities such as spondyloarthritis, sacroiliitis or inflammatory bowel disease, ensuring that escalation to biologics reflected FMF disease activity rather than alternative indications.
Together, these factors must be considered when interpreting our findings. Although retrospective in nature, our data offer a rigorous assessment of the TURPAID score in a clinically and genetically well-defined FMF population. The structural limitations of the score—particularly its heavy reliance on the genetic variable—appear to be the primary contributors to its poor performance, rather than weaknesses inherent to our cohort or methodology.
In conclusion, this external evaluation of the TURPAID score in a large multinational FMF cohort with genetically confirmed diagnoses demonstrates that the score is not a reliable predictor of colchicine resistance in real-life settings, regardless of age at diagnosis. While the idea of making an early prediction is appealing from a clinical perspective, the significant weighting given to the genetic variable results in substantial overclassification, particularly in populations where pathogenic exon 10 variants are common. Differences in clinical presentation across age groups, the retrospective nature of the data collection and variability in clinical practice further limit the applicability of the original scoring system. Overall, our findings suggest that the TURPAID score, in its current form, is not generalisable beyond the cohort in which it was developed. Refinement of predictive tools for colchicine resistance will require prospective data, standardised definitions of resistance and balanced integration of genetic and clinical variables.
Supplementary material
Footnotes
Funding: The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Consent obtained directly from patient(s).
Ethics approval: The JIR cohort protocol was approved by the French Ethics Committee (CCTIRS) on 21 April 2015 (decision number 14.302). The electronic form of the JIR cohort was approved by the National Commission of Data Processing and Liberties (CNIL) on 27 March 2015 (decision number DR-2015-218). All patients or their guardians were not opposed to inclusion and the storage of their medical data in the JIR cohort. They were informed that the collected data might be used for research studies in compliance with privacy rules. Participants gave informed consent to participate in the study before taking part.
Data availability statement
Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.
References
- 1.Georgin-Lavialle S, Savey L, Cuisset L, et al. French protocol for the diagnosis and management of familial Mediterranean fever. Rev Med Interne. 2023;44:602–16.:S0248-8663(23)01203-1. doi: 10.1016/j.revmed.2023.10.441. [DOI] [PubMed] [Google Scholar]
- 2.Savey L, Grateau G, Georgin-Lavialle S. Familial Mediterranean fever in 2020. Nephrol Ther. 2021;17S:S119–25.:S1769-7255(20)30031-6. doi: 10.1016/j.nephro.2020.02.013. [DOI] [PubMed] [Google Scholar]
- 3.Bourguiba R, Deshayes S, Amaryan G, et al. Diagnostic delays in familial Mediterranean fever: a Juvenile Inflammatory Rheumatism (JIR) cohort study. Rheumatol Int. 2024;44:3107–11. doi: 10.1007/s00296-024-05706-1. [DOI] [PubMed] [Google Scholar]
- 4.Rodrigues F, Savey L, Delplanque M, et al. Characteristics of familial Mediterranean fever after 65 years of age. Eur J Intern Med. 2025;132:150–2.:S0953-6205(24)00429-1. doi: 10.1016/j.ejim.2024.10.010. [DOI] [PubMed] [Google Scholar]
- 5.Goldfinger SE. Colchicine for familial Mediterranean fever. N Engl J Med. 1972;287:1302. doi: 10.1056/NEJM197212212872514. [DOI] [PubMed] [Google Scholar]
- 6.Ozen S, Kone-Paut I, Gül A. Colchicine resistance and intolerance in familial mediterranean fever: Definition, causes, and alternative treatments. Semin Arthritis Rheum. 2017;47:115–20.:S0049-0172(16)30330-4. doi: 10.1016/j.semarthrit.2017.03.006. [DOI] [PubMed] [Google Scholar]
- 7.Ozen S, Sağ E, Oton T, et al. EULAR/PReS endorsed recommendations for the management of familial Mediterranean fever (FMF): 2024 update. Ann Rheum Dis. 2025;84:899–909.:S0003-4967(25)00084-6. doi: 10.1016/j.ard.2025.01.028. [DOI] [PubMed] [Google Scholar]
- 8.Batu ED, Şener S, Arslanoglu Aydin E, et al. A score for predicting colchicine resistance at the time of diagnosis in familial Mediterranean fever: data from the TURPAID registry. Rheumatology (Sunnyvale) 2024;63:791–7. doi: 10.1093/rheumatology/kead242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gattorno M, Hofer M, Federici S, et al. Classification criteria for autoinflammatory recurrent fevers. Ann Rheum Dis. 2019;78:1025–32. doi: 10.1136/annrheumdis-2019-215048. [DOI] [PubMed] [Google Scholar]
- 10.Elhani I, Backes S, Kallinich T, et al. Inflammatory biomarker analysis confirms reduced disease severity in heterozygous patients with familial Mediterranean fever. RMD Open. 2024;10:e004677. doi: 10.1136/rmdopen-2024-004677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Macaraeg M, Baker E, Handorf E, et al. Clinical, Immunologic, and Genetic Characteristics in Patients With Syndrome of Undifferentiated Recurrent Fevers. Arthritis Rheumatol. 2025;77:596–605. doi: 10.1002/art.43065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Haslak F, Oner N, Elhani I, et al. Comparative analysis of global practices in the management of colchicine-resistant familial Mediterranean fever: a CliPS network analysis. RMD Open. 2025;11:e006097. doi: 10.1136/rmdopen-2025-006097. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information.