Next Article in Journal
Preprocedural Substrate Visualization and Image Integration Based on Late Enhancement Computed Tomography for Ventricular Tachycardia Ablation in Non-Ischemic Cardiomyopathy
Previous Article in Journal
Post-Coarctectomy Pseudoaneurysm with Recurrent Coarctation Treated with Open Surgery: A Comprehensive Literature Review and Case Report
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JCM
Volume 14
Issue 16
10.3390/jcm14165802
jcm-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Phenotyping Fatigue Profiles in Marfan Syndrome Through Cluster Analysis: A Cross-Sectional Study of Psychosocial and Clinical Correlates

by
Nathasha Samali Udugampolage
1,2,*,†,
Jacopo Taurino
2,†,
Alessandro Pini
2,
Edward Callus
3,4,
Arianna Magon
5,
Gianluca Conte
5,
Giada De Angeli
6,
Miriam Angolani
2,
Giulia Paglione
1,6,
Irene Baroni
6,
Pasquale Iozzo
1 and
Rosario Caruso
4,*
1
Department of Biomedicine and Prevention, University of Rome “Tor Vergata”, 00133 Rome, Italy
2
Cardiovascular Genetic Centre, IRCCS Policlinico San Donato, 20097 San Donato Milanese, Italy
3
Clinical Psychology Service, IRCCS Policlinico San Donato, 20097 San Donato Milanese, Italy
4
Department of Biomedical Sciences for Health, University of Milan, 20133 Milano, Italy
5
Health Professions Research and Development Unit, IRCCS Policlinico San Donato, 20097 San Donato Milanese, Italy
6
Clinical Research Service, IRCCS Policlinico San Donato, 20097 San Donato Milanese, Italy
*
Authors to whom correspondence should be addressed.
†
These authors contributed equally to this work.
J. Clin. Med. 2025, 14(16), 5802; https://doi.org/10.3390/jcm14165802 (registering DOI)
Submission received: 24 July 2025 / Revised: 9 August 2025 / Accepted: 12 August 2025 / Published: 16 August 2025
(This article belongs to the Section Cardiovascular Medicine)
Browse Figures
Review Reports Versions Notes

Abstract

Background/Objectives: Fatigue is a highly prevalent and burdensome symptom among individuals with Marfan syndrome (MFS), yet its heterogeneity and underlying psychosocial and clinical correlates remain underexplored. This study aimed to identify and characterize distinct fatigue-related profiles in MFS patients using a data-driven cluster analysis approach. Methods: A cross-sectional study was conducted involving 127 patients with MFS from a specialized connective tissue disorder center in Italy. Participants completed self-reported measures of fatigue severity (Fatigue Severity Scale, FSS), depressive symptoms (Patient Health Questionnaire-9, PHQ-9), and insomnia (Insomnia Severity Index, ISI). The body mass index (BMI) and clinical data were also collected. A t-distributed stochastic neighbor embedding (t-SNE) analysis was performed to reduce dimensionality, followed by hierarchical clustering (Ward’s method), exploring solutions from k = 2 to k = 10. The optimal cluster solution was identified based on silhouette scores and clinical interpretability. Results: Three distinct clusters emerged: (1) a cluster characterized by low fatigue with minimal psychological and sleep-related symptoms (younger patients, lower PHQ-9 and ISI scores), (2) a cluster characterized by moderate fatigue with moderate psychological and sleep-related symptoms (intermediate age, moderate PHQ-9 and ISI scores), and (3) a cluster characterized by high fatigue with elevated psychological and sleep-related symptoms (older patients, higher PHQ-9, ISI, and FSS scores). Significant differences were observed across clusters in age, BMI, depressive symptoms, insomnia severity, and fatigue levels (all p < 0.05). Conclusions: Our findings highlight the heterogeneity of fatigue experiences in MFS and suggest the importance of profiling patients to guide personalized interventions. This approach may inform precision medicine strategies and enhance the quality of life for individuals with this rare disease.

1. Introduction

Marfan syndrome (MFS) is a rare, heritable connective tissue disorder characterized by the systemic involvement of the cardiovascular, musculoskeletal, and ocular systems [1,2,3,4]. Advances in medical and surgical management, particularly in cardiovascular care, have substantially improved survival rates and life expectancy for individuals with MFS [5,6,7]. However, despite these clinical advancements, many patients continue to report reduced quality of life, largely due to the burden of chronic symptoms that extend beyond life-threatening complications [8,9,10,11].
Individuals with MFS commonly experience a wide spectrum of symptoms that affect daily functioning and overall well-being [1,2,12,13]. Musculoskeletal manifestations such as joint hypermobility, scoliosis, pectus deformities, and chronic pain are prevalent and often contribute to physical limitations and activity restrictions [14]. Ocular complications, including lens dislocation and myopia, can impair vision and further limit daily activities [15]. Additionally, patients may encounter respiratory difficulties, gastrointestinal issues, and neurological symptoms, each adding to the cumulative burden of the condition [1,2,3,16,17,18]. While cardiovascular complications remain the primary focus due to their life-threatening nature, these multisystem symptoms collectively impact patients’ independence, social participation, and psychological health [19].
Among the various symptoms experienced by individuals with MFS, fatigue stands out as a highly prevalent and burdensome yet often under-recognized issue [8]. Recent studies have reported that up to 88% of adults with MFS experience significant fatigue, highlighting its widespread impact within this population [8,20,21,22,23]. Fatigue in MFS is not merely a feeling of tiredness but represents a profound and persistent sense of physical and mental exhaustion that interferes with daily activities. It has been associated with reduced work ability, decreased physical activity levels, impaired social participation, and an overall diminished quality of life [16,17]. Despite these far-reaching effects, fatigue is frequently overshadowed by the focus on life-threatening cardiovascular complications, leaving it under-addressed in clinical practice [13,14]. Nonetheless, patients consistently describe fatigue as one of the most limiting and distressing symptoms they face, underscoring the urgent need for better recognition and targeted management strategies [24].
Fatigue in MFS is increasingly recognized as a complex, multidimensional phenomenon that extends beyond purely physical limitations [21,22]. While connective tissue abnormalities and musculoskeletal impairments contribute to physical fatigue, psychological factors, such as depressive symptoms, anxiety, and reduced self-efficacy, are hypothesized to play a substantial role in modulating fatigue severity and perception [8]. Sleep disturbances, including insomnia and poor sleep quality, have also been identified as key contributors, potentially exacerbating fatigue and further diminishing daily functioning [8]. Moreover, clinical variables such as disease duration, body mass index, and comorbidities may interact with psychological and lifestyle factors, creating diverse fatigue experiences among individuals with MFS [9,25]. Importantly, this variability underscores the heterogeneity of fatigue in this population; not all patients are affected equally, nor do they share the same underlying contributing factors. Identifying these potential correlates is essential to uncovering distinct fatigue phenotypes and informing personalized management strategies [8,9,19].
Despite the growing recognition of fatigue as a significant burden in MFS, current research has primarily focused on describing its prevalence and exploring individual associations with clinical or psychosocial factors [8,26,27]. While these studies have provided valuable insights into potential contributors, they often rely on average group-level analyses that overlook important inter-individual differences. Consequently, patients with MFS are frequently viewed as a homogeneous group, masking the presence of distinct subgroups with unique fatigue profiles and underlying correlates [8].
To date, no studies have systematically classified individuals with MFS into meaningful subgroups based on an integrated set of psychosocial and clinical variables, such as depressive symptoms, insomnia severity, and other disease-related factors [8,9,19]. There is a critical need for approaches that move beyond simple comparisons and instead identify clinically relevant phenotypes, which could guide more personalized assessment and management strategies. For these reasons, this study aims to identify and characterize distinct fatigue phenotypes in individuals with MFS using a data-driven cluster analysis approach, integrating fatigue severity, depressive symptoms, insomnia, and clinical variables to uncover psychosocial and clinical profiles to inform tailored assessment and management strategies. In addition, based on prior evidence [20,28,29,30,31,32,33], we hypothesized the following: (a) Fatigue in MFS is a heterogeneous construct, with individuals clustering into discrete phenotypes differing in severity and psychosocial burden [20,31]. (b) Higher fatigue severity would be associated with greater depressive symptoms, more severe insomnia, and a higher BMI [28,29,32]. (c) Demographic and clinical characteristics, such as age and disease duration, would further distinguish these phenotypes [30,31,32].

2. Materials and Methods

2.1. Design

This study employed a cross-sectional design conducted at a single center, specifically a reference center for MFS in Italy [3]. The study was designed and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines to ensure transparency and methodological rigor [34].

2.2. Sample Size

A Monte Carlo simulation analysis was performed to explore the relationship between sample size and silhouette score, a key indicator of cluster separation quality [35,36]. We simulated cluster solutions ranging from k = 2 to k = 10, varying the sample size from 50 to 250 individuals in increments of 25. For each combination of sample size and k, 1000 iterations were conducted to ensure robust estimates of the silhouette score distribution. The simulation was performed under the hypothesis of moderately separated clusters (Cohen’s d = 1), reflecting a realistic yet meaningful differentiation between potential phenotypes [8].
The results (Figure 1) demonstrated that a sample size of approximately 125 patients represented the optimal balance, achieving adequate silhouette scores and ensuring stable and interpretable clustering, particularly for solutions with k < 5. Beyond 125 patients, silhouette improvements plateaued, indicating diminishing returns with further increases in sample size. This sample size was therefore chosen as it offers sufficient power to identify distinct fatigue-related phenotypes while maintaining practical feasibility in the context of a rare condition such as MFS.

2.3. Setting and Eligibility Criteria

This study was conducted at a single specialized center for heritable connective tissue disorders in Italy, which adopts a multidisciplinary approach integrating cardiologists, clinical geneticists, nurse practitioners, and psychologists [19]. This center is dedicated to providing comprehensive clinical care for patients with MFS, facilitating regular follow-ups in an outpatient setting. Eligible participants were identified from clinic records and invited to participate during routine follow-up visits or contacted by phone or email. To achieve the required sample size for optimal silhouette scores (n = 125), all eligible patients at the center (n = 170) were invited, as a response rate of approximately 75% was anticipated based on prior experience [8,9,16,18,37]. Ultimately, 127 patients agreed to participate, in line with the anticipated response rate.
Inclusion criteria were age ≥ 18 years, a confirmed diagnosis of MFS, and proficient Italian language skills (speaking, reading, and writing). MFS diagnoses were established according to the 2010 revised Ghent nosology, incorporating major criteria such as aortic root dilation, ectopia lentis, specific systemic features, and, when available, confirmation of the FBN1 gene mutation [4]. Patients with cognitive impairments were excluded to enhance internal validity as such impairments could confound self-reported measures of fatigue and related psychosocial variables. Cognitive impairment status was determined based on routine assessments documented in medical records within the last two years.

2.4. Procedure

In this study, 127 individuals agreed and completed the data collection process, resulting in a response rate consistent with expectations. Participants were asked to complete a series of self-report questionnaires administered through a secure web-based platform [38], which was specifically designed to ensure compliance with the General Data Protection Regulation (GDPR) and guarantee confidentiality and data privacy [39].
To ensure data accuracy and reliability, a validation process was implemented for self-reported clinical and anamnestic information. This included a cross-check by a clinician who had access to each patient’s electronic medical records, allowing the verification of self-reported data against documented medical history and clinical assessments. This approach strengthened the overall validity and robustness of the collected dataset.

2.5. Measures

Sociodemographic and clinical data were collected, including sex (male, female, or other), age (in years), educational level (primary school, lower secondary school, higher secondary school, or university), and the number of years since diagnosis. The body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m2).
Medical status was documented with a focus on cardiovascular involvement, including aortic root dilatation, aortic aneurysm, aortic dissection, and valvular pathologies, as well as other comorbidities such as scoliosis and visual impairments. Information on ongoing pharmacological treatments (e.g., beta-blockers, angiotensin receptor blockers) and mental health treatments (e.g., insomnia medications) was also collected.
Fatigue severity was assessed using the Fatigue Severity Scale (FSS), a validated self-administered instrument designed to evaluate the impact of fatigue on daily functioning [40]. The FSS consists of nine statements rated on a 7-point Likert scale (1 = strongly disagree to 7 = strongly agree). The final score is calculated as the mean of the item scores, with higher scores indicating more severe fatigue. A score of ≥4.67 suggests clinically significant fatigue, with a reported sensitivity of 82% and specificity of 87% for identifying disabling fatigue [40].
Depressive symptoms were measured using the Italian version of the Patient Health Questionnaire-9 (PHQ-9) [41], which assesses the presence and severity of depressive symptoms over the previous two weeks. Each of the nine items is scored from 0 (not at all) to 3 (nearly every day), resulting in a total score ranging from 0 to 27. Scores are interpreted as follows: 0–4 = no depressive symptoms, 5–9 = minimal depression, 10–14 = mild depression, 15–19 = moderate depression, and ≥20 = severe depression. Higher scores reflect greater symptom severity [41].
Insomnia severity was evaluated using the Italian version of the Insomnia Severity Index (ISI) [42], a 7-item self-report scale assessing the severity and impact of insomnia symptoms. Each item is rated on a 5-point Likert scale (0–4), with total scores ranging from 0 to 28. Scores are interpreted as follows: 0–7 = no clinically significant insomnia, 8–14 = subthreshold insomnia, 15–21 = moderate clinical insomnia, and 22–28 = severe clinical insomnia. Higher scores indicate greater insomnia severity [42].

2.6. Data Analysis

Data distributions were preliminarily inspected to verify completeness, identify outliers, and detect potential data entry errors [43]. No missing data were reported, as all self-reported measures were collected through mandatory fields in the online platform, ensuring complete datasets.
To reduce dimensionality and enhance the visualization of patient-level variability, t-distributed stochastic neighbor embedding (t-SNE) was performed [44]. Age, the BMI, and the fatigue severity score were used for t-SNE. The t-SNE analysis was conducted with two output dimensions (dims = 2), a perplexity value of 20 (balancing local and global structure), and a maximum of 500 iterations to ensure convergence. Perplexity is a key parameter in t-SNE, interpreted as a smooth measure of the effective number of nearest neighbors, which helps balance the focus between preserving local and global data structure in the resulting low-dimensional embedding. An initial principal component analysis (PCA) step was applied as part of the standard t-SNE preprocessing pipeline to improve computational efficiency and avoid local minima [45]. The resulting two t-SNE components effectively standardized and integrated information from variables with different measurement scales, enabling subsequent clustering analyses in a lower-dimensional space that preserves neighborhood relationships.
Once the two-dimensional t-SNE coordinates were obtained, hierarchical clustering using Ward’s minimum variance method was applied [46]. Solutions ranging from k = 2 to k = 10 clusters were explored. Cluster validity was evaluated through the average silhouette score, which quantifies cohesion and separation between clusters, and by assessing clinical interpretability. The optimal cluster solution was identified based on a combination of silhouette score trends, dendrogram inspection, and the clinical relevance of patient profiles [46,47]. After selecting the most informative solution, clusters were visualized in the t-SNE scatterplots, and dendrogram cuts were assessed to confirm the appropriateness of the partitioning.
Subsequently, inferential comparisons between clusters were performed to describe and characterize each group. For continuous variables (e.g., age, BMI, years since diagnosis), normality was preliminarily assessed. In cases of non-normal distributions, non-parametric Kruskal–Wallis tests were used, while ANOVA was applied when the assumptions were met. For categorical variables (e.g., sex, education level, depressive and insomnia categories), Pearson’s chi-square tests were employed to test for differences across clusters. To further explore the reviewer’s suggestion regarding possible linear or monotonic relationships between age, symptom burden, and fatigue, exploratory bivariate associations among age, years since diagnosis, BMI, fatigue severity (FSS), depressive symptoms (PHQ-9), and insomnia severity (ISI) were examined using Spearman’s rank correlation coefficients with Benjamini–Hochberg’s false discovery rate correction (α = 0.05) [48]. In addition, partial Spearman correlations were computed to assess the associations of fatigue with depressive symptoms and insomnia while controlling for age [49]. Finally, to investigate variability within older individuals, participants in the highest age tertile were subdivided into low- and high-fatigue groups based on the FSS clinical cut-off (4.67) and compared on psychosocial and clinical variables using the Mann–Whitney U test.
All statistical analyses were conducted in R (R Core Team, 2024. R: A Language and Environment for Statistical Computing, Version 4.5.0), using a two-tailed significance level set at α = 0.05.

3. Results

3.1. Sample Characteristics

A total of 127 patients with MFS were included in the analysis (see Table 1). The mean age was 38.8 years (SD ± 14.1), and 52% of the participants were female. Most patients had completed higher secondary school (37.8%) or university education (37.0%), and the median time since diagnosis was 13 years (IQR, 7–23 years). Regarding professional status, the majority were active workers in an office setting (55.9%), while 33.1% were unemployed. The median BMI was 21.3 kg/m2 (IQR: 19.1–23.4). Cardiovascular comorbidities were reported in 80.3% of participants, with 76.4% taking cardiovascular medications and 8.7% reporting hypertension. Other frequent comorbid conditions included scoliosis (68.3%), visual impairments (44.1%), and thyroid dysfunction (11.5%). The mean PHQ-9 score was 6.0 (SD ± 4.5), indicating generally mild levels of depressive symptoms. Specifically, 40.9% of patients reported no depressive symptoms, while 42.5% experienced minimal depressive symptoms. The mean ISI score was 7.0 (SD ± 4.1), with 61.4% of patients reporting no clinically significant insomnia. Fatigue severity, assessed via the FSS, had a mean score of 3.9 (SD ± 1.7), with 31.5% of participants meeting the threshold for clinically relevant fatigue (FSS score ≥ 4.67). Over the previous two weeks, 67.2% of patients reported experiencing fatigue occasionally, and 25.2% reported experiencing fatigue daily. In terms of fatigue pattern, the majority (65.5%) described fatigue mainly in the afternoon, 21.0% in the morning, and 13.4% throughout the entire day.

3.2. t-SNE

The two-dimensional t-SNE visualization (Figure 2) provides a projection of the sample based on age, BMI, and fatigue severity (FSS score), enabling the exploration of latent data structures prior to clustering. This approach enables the optimization of input information into standardized, low-dimensional variables that serve as carriers of relevant clinical and psychosocial information for the clustering procedure. In this scatterplot, individual points are colored by sex (blue for male and red for female) and sized according to FSS scores, with larger points indicating higher fatigue severity. Ellipses were overlaid to represent the distribution of males and females, highlighting the overall spread of each sex within the t-SNE space. Additionally, the wide variation in point sizes across both ellipses highlights substantial within-group heterogeneity in fatigue severity, underscoring the importance of identifying subgroups with distinct fatigue-related phenotypes, which could be determined through cluster analysis.

3.3. Hierarchical Clustering

Hierarchical clustering was performed using the Ward method on the t-SNE components to identify distinct fatigue-related phenotypes among individuals with MFS. Figure 3 displays the distribution of participants projected in the two-dimensional t-SNE space, with color coding indicating cluster membership for solutions ranging from k = 2 to k = 10.
Figure 4 presents the corresponding dendrograms for each k, along with the calculated silhouette widths as an indicator of cluster separation quality. The silhouette width values progressively decreased from k = 2 (0.622) to k = 10 (0.471), indicating that more granular solutions offered slightly reduced internal cohesion. After examining all solutions, k = 3 was ultimately selected as the most appropriate choice. This solution represented a balance between acceptable silhouette width (0.543), sufficient granularity to capture inter-individual variability, and clinical interpretability. The chosen k = 3 solution enabled the identification of three clinically meaningful fatigue profiles, facilitating the subsequent analysis of psychosocial and clinical correlates. This approach supported a nuanced understanding of fatigue heterogeneity in MFS, aligning with the study’s aim to inform tailored assessment and management strategies.

Optimal Cluster Solution

After examining all solutions, the three-cluster (k = 3) configuration was identified as the most suitable, as it balanced silhouette width and clinical interpretability. Table 2 summarizes the detailed characteristics and statistical comparisons among clusters.
Cluster 1 (N = 49) included the youngest participants, with a mean age significantly lower than the other clusters (22.6 ± 3.1 years, p < 0.001). This group also exhibited the lowest median BMI (20.0, IQR 18.4–21.4, p < 0.001) and the shortest disease duration (median 8.6 years, IQR 6.6–10.0, p = 0.0027). PHQ-9 scores were significantly lower (mean 3.5 ± 2.6, p = 0.016), indicating fewer depressive symptoms, and ISI scores were also the lowest (mean 4.3 ± 3.7, p = 0.029), reflecting minimal insomnia. Fatigue severity was the lowest among clusters, as reflected in the FSS score (mean 2.7 ± 0.8, p < 0.001), and no individuals in this cluster presented with clinically relevant fatigue (p < 0.001).
Cluster 2 (N = 32) showed an intermediate clinical profile, with a mean age of 36.4 ± 5.9 years and a median BMI of 21.8 (IQR 19.5–23.6). Disease duration was moderate (median 12.1 years, IQR 9.7–15.0). Participants had intermediate levels of depressive symptoms (PHQ-9 mean 5.8 ± 3.7) and insomnia severity (ISI mean 6.7 ± 4.5). Fatigue severity (FSS mean 3.5 ± 1.1) and the proportion of individuals with clinically relevant fatigue (25.1%) were also intermediate, demonstrating statistically significant differences compared to both Cluster 1 and Cluster 3 (all p-values < 0.05 where indicated).
Cluster 3 (N = 46) comprised the oldest participants (mean age 53.5 ± 6.1 years, p < 0.001), with the highest BMI (median 23.9, IQR 22.5–25.6, p < 0.001) and the longest disease duration (median 15.6 years, IQR 13.7–19.0, p = 0.0027). This group reported the highest levels of depressive symptoms (PHQ-9 mean 7.2 ± 3.6, p = 0.016) and the most severe insomnia (ISI mean 8.8 ± 4.8, p = 0.029). Fatigue severity was markedly higher (FSS mean 4.9 ± 1.0, p < 0.001), and two-thirds of participants (67.4%) exhibited clinically relevant fatigue (p < 0.001), indicating a substantial psychosocial and clinical burden. No statistically significant differences were observed in sex distribution across clusters (p = 0.137), suggesting that sex did not contribute to cluster differentiation in this cohort.
Based on the integrated analysis of fatigue severity, depressive symptoms, insomnia, and clinical variables, the three clusters were labeled to reflect their empirical characteristics. Cluster 1, comprising younger patients with the lowest fatigue scores, minimal depressive symptoms, and lower levels of insomnia, was designated as the “low fatigue with minimal psychological and sleep-related symptoms” profile. Cluster 2 represented an intermediate group, characterized by moderate levels of fatigue, depressive symptoms, and insomnia, and was labeled as the “moderate fatigue with moderate psychological and sleep-related symptoms” profile. Cluster 3 included older patients exhibiting the highest levels of fatigue, depression, and insomnia, along with higher BMI values, and was labeled as the “high fatigue with elevated psychological and sleep-related symptoms” profile. This naming convention aims to reflect the observed data patterns without inferring unmeasured psychological constructs.
To further illustrate the separation and internal coherence of the identified fatigue-related phenotypes, we plotted the three clusters using a two-dimensional t-SNE scatter plot (Figure 5). Each point represents an individual patient, with color indicating cluster membership and point size reflecting the individual FSS score. Ellipses were added to capture the approximate 95% confidence region for each cluster, emphasizing the compactness and distinctiveness of the groups in the reduced-dimensionality space. This graphical representation confirms that the clusters maintain clear separation and limited overlap, supporting both their statistical validity and clinical interpretability. The t-SNE visualization thus provides an intuitive depiction of how the clusters diverge with respect to age, the BMI, and fatigue severity, strengthening confidence in the identified fatigue-related profiles.

3.4. Exploratory Associations Between Age, Psychosocial Factors, and Fatigue

Age showed a moderate positive correlation with the BMI (ρ = 0.51, p < 0.001) and weaker but statistically significant associations with fatigue severity (ρ = 0.22, p = 0.014) and insomnia severity (ρ = 0.18, p = 0.042). Fatigue severity correlated strongly with depressive symptoms (ρ = 0.62, p < 0.001) and moderately with insomnia severity (ρ = 0.41, p < 0.001), but not with the BMI (ρ = 0.04, p = 0.634).
Partial correlation analyses controlling for age confirmed that fatigue severity remained strongly associated with depressive symptoms (ρ = 0.60, p < 0.001) and moderately with insomnia severity (ρ = 0.39, p < 0.001), indicating that these relationships are largely independent of age.
To further explore variability within older participants, those in the highest age tertile (≥47 years) were stratified into low-fatigue (FSS < 4.67) and high-fatigue (FSS ≥ 4.67) groups. Within this subgroup, high-fatigue individuals reported significantly higher PHQ-9 scores (9.53 ± 4.56 vs. 5.00 ± 3.76, U = 90.5, p = 0.0008) and a trend toward higher ISI scores (9.00 ± 4.24 vs. 6.67 ± 3.03, U = 158.5, p = 0.087), whereas the BMI and years since diagnosis did not differ meaningfully (both p ≥ 0.66).

4. Discussion

This study aimed to identify distinct fatigue-related phenotypes among patients with MFS, leveraging advanced clustering approaches to capture the heterogeneity of clinical presentations. Through the integration of t-SNE for dimensionality reduction and hierarchical clustering, we successfully delineated three clinically meaningful clusters: one characterized by low fatigue with minimal psychological and sleep-related symptoms, a second with moderate fatigue and intermediate levels of depressive and insomnia symptoms, and a third with high fatigue and elevated psychological and sleep-related symptoms.
Fatigue is increasingly recognized as a significant and debilitating symptom in hereditary connective tissue disorders such as MFS, profoundly impacting patients’ functional status and quality of life [8,15,22,25,27]. Our findings highlight that fatigue in MFS is not a uniform experience, but rather a complex and multifaceted phenomenon influenced by demographic, psychological, and clinical factors [44,50]. Importantly, by applying a data-driven clustering strategy supported by t-SNE, we were able to generate low-dimensional representations of patients’ characteristics, thus optimizing the interpretability of the clusters while maintaining the richness of the underlying data [51]. This methodological approach enabled the identification of homogeneous subgroups with clear clinical profiles, offering new insights into the variability of fatigue experiences in MFS and paving the way for more personalized care strategies.
Our findings are consistent with the previous literature highlighting fatigue as a highly prevalent and impactful symptom among individuals with MFS [8,24]. Prior studies have reported fatigue prevalence rates ranging from 30% to over 80% in patients with MFS, often emphasizing its significant detrimental effect on daily functioning and health-related quality of life [8,20,22,52]. Moreover, fatigue has been shown in previous published research to correlate with psychological distress, particularly depression and anxiety, and to be exacerbated by sleep disturbances such as insomnia [8,24]. The identification of three distinct clusters in our analysis reinforces the notion of fatigue as a multifactorial phenomenon in MFS, supporting existing hypotheses that fatigue severity is shaped not only by physical factors but also by psychological and lifestyle-related components. A notable novelty of our work is the clear emergence of a cluster characterized by low fatigue and minimal psychological and sleep-related symptoms. While previous studies have suggested the existence of individuals with better fatigue adaptation, this specific subgroup has not been explicitly characterized in a data-driven manner. The identification of this resilient cluster underscores the potential for protective psychological or behavioral factors in modulating fatigue burden and highlights the importance of personalized supportive interventions [53,54,55,56,57].
The three identified clusters provide meaningful insights into the heterogeneity of fatigue experiences among patients with MFS. Cluster 1 was characterized by younger age, lower PHQ-9 and ISI scores, and a lower BMI. These features suggest the presence of potential protective factors, such as fewer comorbidities, higher physiological reserve, and possibly greater psychological resilience [58]. This subgroup might benefit from standard monitoring, focusing on maintaining physical and psychological well-being to preserve their favorable fatigue status. Cluster 2 showed intermediate levels of fatigue severity alongside higher rates of depressive symptoms and subthreshold insomnia. This indicates a subgroup where psychosocial vulnerability plays a prominent role, despite relatively lower physical burden [59]. In this profile, the early identification of emotional distress and tailored psychosocial support, including psychological counseling and sleep interventions, could help prevent progression to higher fatigue states. Cluster 3 included older patients with higher BMI, longer disease duration, and the highest levels of depressive symptoms and insomnia severity. This profile reflects the cumulative burden of physical comorbidities and psychological distress [60]. Patients in this group are likely to require more intensive, multidisciplinary management strategies that address both somatic and mental health dimensions of fatigue.
The identification of distinct fatigue profiles among individuals with MFS has important clinical and practical implications. Profiling patients into homogeneous subgroups allows clinicians to move beyond a “one-size-fits-all” approach and tailor interventions to the specific needs of each profile [61]. For example, patients in the “low fatigue and minimal psychological and sleep-related symptoms” group (Cluster 1) may primarily benefit from regular monitoring and general lifestyle counseling to maintain their favorable status. In contrast, those in the “moderate fatigue with moderate psychological and sleep-related symptoms” group (Cluster 2) could require targeted psychological support, such as counseling, cognitive-behavioral therapy, or sleep hygiene interventions, to address emotional distress and subthreshold insomnia that may exacerbate fatigue [62]. Finally, the “high fatigue with elevated psychological and sleep-related symptoms” (Cluster 3) highlights a subgroup in need of intensive, multidisciplinary care strategies that simultaneously address comorbid physical conditions, psychological distress, and severe fatigue [63].
In addition to the cluster-based characterization, exploratory analyses were conducted to clarify further the relationships among age, psychosocial burden, and fatigue severity. Spearman correlations showed that fatigue severity was most strongly related to depressive symptoms (ρ = 0.62, p < 0.001) and, to a lesser extent, insomnia severity (ρ = 0.41, p < 0.001), whereas its association with age was weaker (ρ = 0.22, p = 0.014). Partial correlations controlling for age confirmed that both depressive symptoms (ρ = 0.60, p < 0.001) and insomnia severity (ρ = 0.39, p < 0.001) remained significantly associated with fatigue, indicating that these relationships are largely independent of chronological age. To examine variability within older participants, those in the highest age tertile (≥47 years) were stratified by fatigue status using the clinical FSS cut-off. High-fatigue older adults reported substantially higher depressive symptom scores and a trend toward higher insomnia scores, while the BMI and disease duration did not differ. These findings suggest that even among older individuals, lower levels of depressive symptoms and potentially better sleep quality may serve as protective factors against severe fatigue. This underscores the importance of addressing modifiable psychosocial factors in targeted intervention strategies, especially for high-risk subgroups, even though these hypotheses from explorative analyses require further empirical testing in future studies.
This study has several limitations that should be acknowledged. First, its cross-sectional design precludes establishing causal relationships between fatigue and the associated clinical or psychosocial factors identified across clusters. Longitudinal studies are necessary to confirm whether these profiles remain stable over time and to investigate the trajectories of fatigue severity and its associated factors. Second, the reliance on self-reported measures for key variables such as fatigue severity (FSS), depressive symptoms (PHQ-9), and insomnia (ISI) introduces potential reporting and recall biases. Although these instruments are validated and widely used, self-report data may be influenced by individual perceptions or social desirability. Third, while our sample was carefully selected and achieved a high response rate, it consisted of patients from a single specialized center in Italy, which may limit the generalizability of findings to other populations or healthcare settings. Additionally, there is a potential for selection bias as patients who agreed to participate may be more motivated or health-conscious, potentially leading to an underestimation of fatigue severity in the broader MFS population. Future studies incorporating objective measurements (e.g., actigraphy for fatigue, clinical interviews for psychiatric symptoms), broader sampling, and longitudinal follow-up would help validate and expand on these findings.

5. Conclusions

This study identified and characterized three distinct fatigue-related profiles in individuals with MFS using an integrative, data-driven cluster analysis approach. The combination of t-SNE dimensionality reduction and hierarchical clustering allowed for the identification of clinically meaningful subgroups that differ in terms of age, psychosocial vulnerability, and fatigue severity. These findings highlight the heterogeneity of fatigue experiences in this population and underscore the importance of personalized assessment and management strategies tailored to individual needs. Future research should confirm these profiles longitudinally and explore tailored interventions to improve quality of life and clinical outcomes in patients with MFS.

Author Contributions

Conceptualization: N.S.U. and J.T. Methodology: N.S.U., J.T., and R.C. Formal Analysis: R.C. and A.M. Investigation: N.S.U., J.T., A.P., R.C., E.C., A.M., M.A., G.C., G.D.A., P.I., G.P., and I.B. Data Curation: N.S.U., J.T., A.P., R.C., E.C., A.M., M.A., G.C., G.D.A., P.I., G.P., and I.B. Writing—Original Draft Preparation: N.S.U. and J.T. Writing—Review and Editing: all authors. Visualization: R.C. and A.M. Software and Analytics: R.C. and A.M. Supervision: R.C., A.P., J.T., and N.S.U. Project Administration: R.C., A.P., J.T., and N.S.U. Funding Acquisition: R.C., J.T., and N.S.U. A.P. and R.C. acted as the guarantor for this manuscript. All authors have read and agreed to the published version of the manuscript.

Funding

This study was partially supported by “Ricerca Corrente” funding from the Italian Ministry of Health to IRCCS Policlinico San Donato.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board (Ethics Committee) of Ospedale San Raffaele (protocol code 32/INT/2021; date of approval: 16 July 2021).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study through secure digital means integrated into the online data collection platform. As the study was conducted entirely online, written (paper-based) consent was not required; instead, participants provided explicit electronic consent before accessing and completing the questionnaires, in accordance with ethical standards and data protection regulations.

Data Availability Statement

Data are available upon reasonable request. The datasets generated and/or analyzed during the current study are not publicly available due to privacy or ethical restrictions but are available from the corresponding author upon reasonable request.

Acknowledgments

The authors would like to thank all the participants of the study for their valuable time and contribution. During the preparation of this manuscript, the authors used ChatGPT 4.0 (OpenAI) solely for the purposes of grammar checking and editing text drafted by the authors. The authors have reviewed and edited the output and take full responsibility for the content of this publication.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
MFSMarfan Syndrome
BMIBody Mass Index
FSSFatigue Severity Scale
PHQ-9Patient Health Questionnaire-9
ISIInsomnia Severity Index
SDStandard Deviation
IQRInterquartile Range
GDPRGeneral Data Protection Regulation
STROBEStrengthening the Reporting of Observational Studies in Epidemiology
PCAPrincipal Component Analysis
t-SNEt-Distributed Stochastic Neighbor Embedding

References

  1. Warnink-Kavelaars, J.; De Koning, L.; Rombaut, L.; Alsem, M.; Menke, L.; Oosterlaan, J.; Buizer, A.; Engelbert, R.; on behalf of the Pediatric Heritable Connective Tissue Disorders Study Group. Heritable Connective Tissue Disorders in Childhood: Increased Fatigue, Pain, Disability and Decreased General Health. Genes 2021, 12, 831. [Google Scholar] [CrossRef] [PubMed]
  2. Warnink-Kavelaars, J.; De Koning, L.E.; Rombaut, L.; Menke, L.A.; Alsem, M.W.; Van Oers, H.A.; Buizer, A.I.; Engelbert, R.H.H.; Oosterlaan, J.; Pediatric Heritable Connective Tissue Disorder study group. Heritable Connective Tissue Disorders in Childhood: Decreased Health-related Quality of Life and Mental Health. Am. J. Med. Genet. A 2022, 188, 2096–2109. [Google Scholar] [CrossRef]
  3. Marelli, S.; Micaglio, E.; Taurino, J.; Salvi, P.; Rurali, E.; Perrucci, G.L.; Dolci, C.; Udugampolage, N.S.; Caruso, R.; Gentilini, D.; et al. Marfan Syndrome: Enhanced Diagnostic Tools and Follow-up Management Strategies. Diagnostics 2023, 13, 2284. [Google Scholar] [CrossRef]
  4. Loeys, B.L.; Dietz, H.C.; Braverman, A.C.; Callewaert, B.L.; De Backer, J.; Devereux, R.B.; Hilhorst-Hofstee, Y.; Jondeau, G.; Faivre, L.; Milewicz, D.M.; et al. The Revised Ghent Nosology for the Marfan Syndrome. J. Med. Genet. 2010, 47, 476–485. [Google Scholar] [CrossRef] [PubMed]
  5. Chiu, H.-H. An Update of Medical Care in Marfan Syndrome. Tzu Chi Med. J. 2022, 34, 44–48. [Google Scholar] [CrossRef]
  6. Sama, C.; Fongwen, N.T.; Chobufo, M.D.; Ajibade, A.; Roberts, M.; Greathouse, M.; Ngonge, A.L.; Adekolu, A.; Hamirani, Y.S. Frequency of Cardiac Valvulopathies in Patients With Marfan Syndrome: A Systematic Review and Meta-Analysis. Cureus 2024, 16, e54141. [Google Scholar] [CrossRef]
  7. Janjua, S.; Lew, A.; Sokol, S.; Jalali, P.; Andres, U.; Olski, D. Early Cardiac Intervention Decreases the Mortality Risk Linked With Marfan Syndrome. Cureus 2025, 17, e82910. [Google Scholar] [CrossRef]
  8. Udugampolage, N.; Taurino, J.; Bassotti, A.; Pini, A.; Caruso, R.; Callus, E.; Magon, A.; Conte, G.; De Angeli, G.; Paglione, G.; et al. Exploring Fatigue in Marfan and Hypermobile Ehlers-Danlos Syndromes: An Analytical Cross-Sectional Study in Two Italian Healthcare Centres. BMJ Open 2025, 15, e087298. [Google Scholar] [CrossRef]
  9. Udugampolage, N.; Caruso, R.; Magon, A.; Conte, G.; Callus, E.; Dellafiore, F.; Pittella, F.; Arrigoni, C.; Taurino, J.; Pini, A. Self-Care Behaviors and Their Individual-Level Determinants in Italian Adults with Marfan Syndrome: A Single-Center Cross-Sectional Study. Appl. Nurs. Res. ANR 2024, 78, 151821. [Google Scholar] [CrossRef] [PubMed]
  10. Clark, N.L.; Kainth, G.S.; Johnson, M.; Rangan, A.; Kottam, L.; Swainston, K. Psychological Interventions to Improve Pain, Fatigue, Anxiety, Depression, and Quality of Life in Children and Adults with Hypermobility Spectrum Disorders and Ehlers-Danlos Syndrome: A Systematic Review. Rheumatol. Int. 2023, 44, 41–55. [Google Scholar] [CrossRef]
  11. Peters, K.; Kong, F.; Horne, R.; Francomano, C.; Biesecker, B. Living with Marfan Syndrome I. Perceptions of the Condition. Clin. Genet. 2001, 60, 273–282. [Google Scholar] [CrossRef] [PubMed]
  12. Yuan, S.-M.; Jing, H. Marfan’s Syndrome: An Overview. Sao Paulo Med. J. (Rev. Paul. Med.) 2010, 128, 360–366. [Google Scholar] [CrossRef]
  13. Zeigler, S.M.; Sloan, B.; Jones, J.A. Pathophysiology and Pathogenesis of Marfan Syndrome. Adv. Exp. Med. Biol. 2021, 1348, 185–206. [Google Scholar] [CrossRef] [PubMed]
  14. Connolly, H.M.; Niaz, T.; Bowen, J.M. What Is Marfan Syndrome? JAMA 2023, 329, 1618. [Google Scholar] [CrossRef]
  15. Spencer, M. Marfan Syndrome. Nursing 2024, 54, 19–25. [Google Scholar] [CrossRef]
  16. Udugampolage, N.; Caruso, R.; Magon, A.; Conte, G.; Callus, E.; Taurino, J.; Pini, A. Describing Post-Traumatic Stress Disorder and Its Associations with Depression, Anxiety and Insomnia: A Descriptive Study in Italian Adults with Marfan Syndrome during the COVID-19 Third Wave. BMJ Open 2022, 12, e067024. [Google Scholar] [CrossRef]
  17. Vanem, T.T.; Rand-Hendriksen, S.; Brunborg, C.; Geiran, O.R.; Røe, C. Health-Related Quality of Life in Marfan Syndrome: A 10-Year Follow-Up. Health Qual. Life Outcomes 2020, 18, 376. [Google Scholar] [CrossRef]
  18. Udugampolage, N.; Caruso, R.; Panetta, M.; Callus, E.; Dellafiore, F.; Magon, A.; Marelli, S.; Pini, A. Is SF-12 a Valid and Reliable Measurement of Health-Related Quality of Life among Adults with Marfan Syndrome? A Confirmatory Study. PLoS ONE 2021, 16, e0252864. [Google Scholar] [CrossRef]
  19. Panetta, M.; Bianchetti, A.; Udugampolage, N.S.; Taurino, J.; Caruso, R.; Pini, A.; Callus, E. Discussing Psychological and Psychotherapeutic Support for Patients with Marfan Syndrome (MFS) and Their Family: An Example of a Structured Program in Italy. Front. Psychol. 2023, 14, 1176692. [Google Scholar] [CrossRef] [PubMed]
  20. Bathen, T.; Velvin, G.; Rand-Hendriksen, S.; Robinson, H.S. Fatigue in Adults with Marfan Syndrome, Occurrence and Associations to Pain and Other Factors. Am. J. Med. Genet. A 2014, 164, 1931–1939. [Google Scholar] [CrossRef]
  21. Maher, C.; Crettenden, A.; Evans, K.; Thiessen, M.; Toohey, M.; Watson, A.; Dollman, J. Fatigue Is a Major Issue for Children and Adolescents with Physical Disabilities. Dev. Med. Child Neurol. 2015, 57, 742–747. [Google Scholar] [CrossRef] [PubMed]
  22. Velvin, G.; Johansen, H.; Østertun-Geirdal, A.; Bathen, T. Fatigue in Patients with Syndromic Heritable Thoracic Aortic Disease: A Systematic Review of the Literature and a Qualitative Study of Patients’ Experiences and Perceptions. Orphanet J. Rare Dis. 2023, 18, 119. [Google Scholar] [CrossRef] [PubMed]
  23. Voermans, N.C.; Knoop, H.; Bleijenberg, G.; Van Engelen, B.G. Fatigue Is Associated with Muscle Weakness in Ehlers-Danlos Syndrome: An Explorative Study. Physiotherapy 2011, 97, 170–174. [Google Scholar] [CrossRef] [PubMed]
  24. Baylow, H.E.; Esfandiarei, M.; Ratiu, I. Voice Symptoms and Quality of Life in Individuals With Marfan Syndrome: A Cross-Sectional Study. J. Voice Off. J. Voice Found. 2024, 38, 1254.e1–1254.e10. [Google Scholar] [CrossRef]
  25. Voermans, N.C.; Knoop, H.; van de Kamp, N.; Hamel, B.C.; Bleijenberg, G.; van Engelen, B.G. Fatigue Is a Frequent and Clinically Relevant Problem in Ehlers-Danlos Syndrome. Semin. Arthritis Rheum. 2010, 40, 267–274. [Google Scholar] [CrossRef]
  26. Cognitive-Behavioural Therapy For Insomnia (CBT-I) Across The Life Span: Guidelines and Clinical Protocols for Health Professionals, 1st ed.; Baglioni, C., Espie, C.A., Riemann, D., European Sleep Research Society, European Insomnia Network, European Academy for Cognitive Behavioural Therapy for Insomnia, Eds.; Wiley: New York, NY, USA, 2022; ISBN 978-1-119-78513-2. [Google Scholar]
  27. van Andel, M.M.; Graaumans, K.; Groenink, M.; Zwinderman, A.H.; van Kimmenade, R.R.J.; Scholte, A.J.H.A.; van den Berg, M.P.; Dickinson, M.G.; Knoop, H.; Bosch, J.A.; et al. A Cross-Sectional Study on Fatigue, Anxiety, and Symptoms of Depression and Their Relation with Medical Status in Adult Patients with Marfan Syndrome. Psychological Consequences in Marfan Syndrome. Clin. Genet. 2022, 102, 404–413. [Google Scholar] [CrossRef]
  28. Andonian, C.; Freilinger, S.; Achenbach, S.; Ewert, P.; Gundlach, U.; Kaemmerer, H.; Nagdyman, N.; Neidenbach, R.C.; Pieper, L.; Schelling, J.; et al. Quality of Life in Patients with Marfan Syndrome: A Cross-Sectional Study of 102 Adult Patients. Cardiovasc. Diagn. Ther. 2021, 11, 602–610. [Google Scholar] [CrossRef]
  29. Sharp, H.E.C.; Critchley, H.D.; Eccles, J.A. Connecting Brain and Body: Transdiagnostic Relevance of Connective Tissue Variants to Neuropsychiatric Symptom Expression. World J. Psychiatry 2021, 11, 805–820. [Google Scholar] [CrossRef] [PubMed]
  30. Fletcher, E.K.S.; Fischer, A.L.; Manocha, R.H.K. Depressive Symptoms Are Highly Prevalent and Associated with Fatigue and Pain Catastrophizing in the Hypermobility Spectrum Disorders and Hypermobile Ehlers Danlos Syndrome: A Cross-Sectional Study. Rheumatol. Int. 2025, 45, 128. [Google Scholar] [CrossRef]
  31. Trieste, L.; Cannizzo, S.; Palla, I.; Triulzi, I.; Turchetti, G. State of the Art and Future Directions in Assessing the Quality of Life in Rare and Complex Connective Tissue and Musculoskeletal Diseases. Front. Med. 2022, 9, 986218. [Google Scholar] [CrossRef]
  32. Cavalier, I.; Kamga, L.T.; Morin, L.; Bal, L.; Faivre, L.; Jondeau, G.; Tran, V.-T.; Milleron, O. Determinants of Fatigue in Patients with Marfan Syndrome: A Study Using PROMS. Orphanet J. Rare Dis. 2025, 20, 359. [Google Scholar] [CrossRef]
  33. Behrens, M.; Gube, M.; Chaabene, H.; Prieske, O.; Zenon, A.; Broscheid, K.-C.; Schega, L.; Husmann, F.; Weippert, M. Fatigue and Human Performance: An Updated Framework. Sports Med. 2023, 53, 7–31. [Google Scholar] [CrossRef]
  34. von Elm, E.; Altman, D.G.; Egger, M.; Pocock, S.J.; Gøtzsche, P.C.; Vandenbroucke, J.P.; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for Reporting Observational Studies. J. Clin. Epidemiol. 2008, 61, 344–349. [Google Scholar] [CrossRef]
  35. Milligan, G.W. A Review Of Monte Carlo Tests Of Cluster Analysis. Multivar. Behav. Res. 1981, 16, 379–407. [Google Scholar] [CrossRef] [PubMed]
  36. Cervan, D.; Coronado, A.M.; Luyo, J.E. Cluster-Based Stratified Sampling for Fast Reliability Evaluation of Composite Power Systems Based on Sequential Monte Carlo Simulation. Int. J. Electr. Power Energy Syst. 2023, 147, 108813. [Google Scholar] [CrossRef]
  37. Udugampolage, N.S.; Pini, A.; Magon, A.; Conte, G.; Callus, E.; Taurino, J.; Caruso, R. COVID-19 Vaccine Hesitancy in Italian Adults with Marfan Syndrome: Insights from a Secondary Analysis of a Cross-Sectional Study. Vaccines 2023, 11, 734. [Google Scholar] [CrossRef] [PubMed]
  38. Harris, P.A.; Taylor, R.; Minor, B.L.; Elliott, V.; Fernandez, M.; O’Neal, L.; McLeod, L.; Delacqua, G.; Delacqua, F.; Kirby, J.; et al. The REDCap Consortium: Building an International Community of Software Platform Partners. J. Biomed. Inform. 2019, 95, 103208. [Google Scholar] [CrossRef]
  39. Janßen, R.; Kesler, R.; Kummer, M.; Waldfogel, J. GDPR and the Lost Generation of Innovative Apps; National Bureau of Economic Research: Cambridge, MA, USA, 2022; p. w30028. [Google Scholar]
  40. Siciliano, M.; Chiorri, C.; De Micco, R.; Russo, A.; Tedeschi, G.; Trojano, L.; Tessitore, A. Fatigue in Parkinson’s Disease: Italian Validation of the Parkinson Fatigue Scale and the Fatigue Severity Scale Using a Rasch Analysis Approach. Parkinsonism Relat. Disord. 2019, 65, 105–110. [Google Scholar] [CrossRef]
  41. Costantini, L.; Pasquarella, C.; Odone, A.; Colucci, M.E.; Costanza, A.; Serafini, G.; Aguglia, A.; Belvederi Murri, M.; Brakoulias, V.; Amore, M.; et al. Screening for Depression in Primary Care with Patient Health Questionnaire-9 (PHQ-9): A Systematic Review. J. Affect. Disord. 2021, 279, 473–483. [Google Scholar] [CrossRef]
  42. Castronovo, V.; Galbiati, A.; Marelli, S.; Brombin, C.; Cugnata, F.; Giarolli, L.; Anelli, M.M.; Rinaldi, F.; Ferini-Strambi, L. Validation Study of the Italian Version of the Insomnia Severity Index (ISI). Neurol. Sci. 2016, 37, 1517–1524. [Google Scholar] [CrossRef] [PubMed]
  43. Garlits, J.; McAfee, S.; Taylor, J.-A.; Shum, E.; Yang, Q.; Nunez, E.; Kameron, K.; Fenech, K.; Rodriguez, J.; Torri, A.; et al. Statistical Approaches for Establishing Appropriate Immunogenicity Assay Cut Points: Impact of Sample Distribution, Sample Size, and Outlier Removal. AAPS J. 2023, 25, 37. [Google Scholar] [CrossRef] [PubMed]
  44. Anowar, F.; Sadaoui, S.; Selim, B. Conceptual and Empirical Comparison of Dimensionality Reduction Algorithms (PCA, KPCA, LDA, MDS, SVD, LLE, ISOMAP, LE, ICA, t-SNE). Comput. Sci. Rev. 2021, 40, 100378. [Google Scholar] [CrossRef]
  45. Bajal, E.; Katara, V.; Bhatia, M.; Hooda, M. A Review of Clustering Algorithms: Comparison of DBSCAN and K-Meanwith Oversampling and t-SNE. Recent Pat. Eng. 2022, 16, e180122191239. [Google Scholar] [CrossRef]
  46. Ran, X.; Xi, Y.; Lu, Y.; Wang, X.; Lu, Z. Comprehensive Survey on Hierarchical Clustering Algorithms and the Recent Developments. Artif. Intell. Rev. 2023, 56, 8219–8264. [Google Scholar] [CrossRef]
  47. Spada, F.; Caruso, R.; Notarnicola, I.; Belloni, S.; De Maria, M.; Magon, A.; Conte, G.; Prendi, E.; Pata, X.; Duka, B.; et al. Analyzing Readiness for Interprofessional Education among Health Program Students Using Hierarchical Clustering. J. Interprof. Care 2025, 39, 391–398. [Google Scholar] [CrossRef] [PubMed]
  48. Benjamini, Y.; Hochberg, Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J. R. Stat. Soc. Ser. B Stat. Methodol. 1995, 57, 289–300. [Google Scholar] [CrossRef]
  49. van Aert, R.C.M. Meta-Analyzing Partial Correlation Coefficients Using Fisher’s z Transformation. Res. Synth. Methods 2023, 14, 768–773. [Google Scholar] [CrossRef]
  50. Aslanidis, Z.; Kotsiou, O.S. The Impact of Chronic Fatigue on Psychopathology in Outpatient Physiotherapy Patients. Musculoskeletal Care 2025, 23, e70116. [Google Scholar] [CrossRef]
  51. Deb, C.; Schlueter, A. Review of Data-Driven Energy Modelling Techniques for Building Retrofit. Renew. Sustain. Energy Rev. 2021, 144, 110990. [Google Scholar] [CrossRef]
  52. Krahe, A.M.; Adams, R.D.; Nicholson, L.L. Features That Exacerbate Fatigue Severity in Joint Hypermobility Syndrome/Ehlers-Danlos Syndrome—Hypermobility Type. Disabil. Rehabil. 2018, 40, 1989–1996. [Google Scholar] [CrossRef]
  53. Ng, K.; Kartoun, U.; Stavropoulos, H.; Zambrano, J.A.; Tang, P.C. Personalized Treatment Options for Chronic Diseases Using Precision Cohort Analytics. Sci. Rep. 2021, 11, 1139. [Google Scholar] [CrossRef]
  54. Kim, M.T.; Radhakrishnan, K.; Heitkemper, E.M.; Choi, E.; Burgermaster, M. Psychosocial Phenotyping as a Personalization Strategy for Chronic Disease Self-Management Interventions. Am. J. Transl. Res. 2021, 13, 1617–1635. [Google Scholar]
  55. Zein Abdin, Z.; Yin, H.; Giannis, C.; Hsieh, R.; Pickering, J.G.; Chandy, M. Harnessing iPSCs to Model Marfan Syndrome: Advancing Clinical Diagnosis and Drug Discovery. Curr. Treat. Options Cardiovasc. Med. 2025, 27, 20. [Google Scholar] [CrossRef]
  56. Jouini, S.; Milleron, O.; Eliahou, L.; Jondeau, G.; Vitiello, D. Is Physical Activity a Future Therapy for Patients with Marfan Syndrome? Orphanet J. Rare Dis. 2022, 17, 46. [Google Scholar] [CrossRef] [PubMed]
  57. Jouini, S.; Milleron, O.; Eliahou, L.; Jondeau, G.; Vitiello, D. Effects of a Personalized Home-Based Training Program among Patients Suffering from Marfan Syndrome: A Pilot Randomized and Controlled Study. Intractable Rare Dis. Res. 2021, 10, 263–268. [Google Scholar] [CrossRef]
  58. Öcalan, S.; Üzar-Özçetin, Y.S. The Relationship between Rumination, Fatigue and Psychological Resilience among Cancer Survivors. J. Clin. Nurs. 2022, 31, 3595–3604. [Google Scholar] [CrossRef]
  59. Ashley, F. Accounting for Research Fatigue in Research Ethics. Bioethics 2021, 35, 270–276. [Google Scholar] [CrossRef] [PubMed]
  60. Vaes, A.W.; Goërtz, Y.M.J.; Van Herck, M.; Beijers, R.J.H.C.G.; Van Beers, M.; Burtin, C.; Janssen, D.J.A.; Schols, A.M.W.J.; Spruit, M.A. Physical and Mental Fatigue in People with Non-Communicable Chronic Diseases. Ann. Med. 2022, 54, 2521–2533. [Google Scholar] [CrossRef]
  61. Sbitan, L.; Alzraikat, N.; Tanous, H.; Saad, A.M.; Odeh, M. From One Size Fits All to a Tailored Approach: Integrating Precision Medicine into Medical Education. BMC Med. Educ. 2025, 25, 90. [Google Scholar] [CrossRef] [PubMed]
  62. Rosen, P.V. Beyond One Size Fits All—Personalised Prevention Strategies Using Physical Activity: Editorial. Eur. J. Physiother. 2024, 26, 1. [Google Scholar] [CrossRef]
  63. Purgato, M.; Singh, R.; Acarturk, C.; Cuijpers, P. Moving beyond a ‘One-Size-Fits-All’ Rationale in Global Mental Health: Prospects of a Precision Psychology Paradigm. Epidemiol. Psychiatr. Sci. 2021, 30, e63. [Google Scholar] [CrossRef] [PubMed]
Jcm 14 05802 g001
Figure 1. Simulated relationship between sample size and silhouette score across different cluster solutions (k = 2 to 10).
Figure 1. Simulated relationship between sample size and silhouette score across different cluster solutions (k = 2 to 10).
Jcm 14 05802 g001
Jcm 14 05802 g002
Figure 2. The two-dimensional t-SNE projection of the sample colored by sex and sized by fatigue severity. Legend: Colors—blue = male, red = female. Point size—proportional to Fatigue Severity Scale (FSS) score. Ellipses—contain the approximate distribution of each sex group (the gray ellipse is the aggregated information).
Figure 2. The two-dimensional t-SNE projection of the sample colored by sex and sized by fatigue severity. Legend: Colors—blue = male, red = female. Point size—proportional to Fatigue Severity Scale (FSS) score. Ellipses—contain the approximate distribution of each sex group (the gray ellipse is the aggregated information).
Jcm 14 05802 g002
Jcm 14 05802 g003
Figure 3. t-SNE scatterplots displaying cluster solutions from k = 2 to k = 10. Note. Each plot shows participant distribution in the t-SNE space, colored by cluster membership.
Figure 3. t-SNE scatterplots displaying cluster solutions from k = 2 to k = 10. Note. Each plot shows participant distribution in the t-SNE space, colored by cluster membership.
Jcm 14 05802 g003
Jcm 14 05802 g004
Figure 4. Dendrograms from hierarchical clustering for k = 2 to k = 10.
Figure 4. Dendrograms from hierarchical clustering for k = 2 to k = 10.
Jcm 14 05802 g004
Jcm 14 05802 g005
Figure 5. A t-SNE scatter plot colored by cluster membership, sized by fatigue severity. Legend: Points are colored according to final cluster assignment (Cluster 1 = low-fatigue resilient profile, Cluster 2 = moderate-fatigue psychosocially vulnerable profile, Cluster 3 = high-fatigue burdened profile) and scaled by individual FSS scores to reflect fatigue severity. Ellipses represent the estimated 95% confidence boundaries of each cluster in the t-SNE space, illustrating the separation and cohesion of the identified profiles.
Figure 5. A t-SNE scatter plot colored by cluster membership, sized by fatigue severity. Legend: Points are colored according to final cluster assignment (Cluster 1 = low-fatigue resilient profile, Cluster 2 = moderate-fatigue psychosocially vulnerable profile, Cluster 3 = high-fatigue burdened profile) and scaled by individual FSS scores to reflect fatigue severity. Ellipses represent the estimated 95% confidence boundaries of each cluster in the t-SNE space, illustrating the separation and cohesion of the identified profiles.
Jcm 14 05802 g005
Table 1. Summary of sample characteristics.
Table 1. Summary of sample characteristics.
Characteristic N (%)
Sex
Male61 (48%)
Female66 (52%)
Age
Years (mean ± SD)38.75 ± 14.06
Primary school2 (1.6%)
Lower secondary school30 (23.6%)
Higher secondary school48 (37.8%)
University47 (37.0%)
Years since diagnosis, median (IQR)13 (7–23)
Profession
Active worker—office71 (55.9%)
Active worker—home8 (6.3%)
Occasional worker3 (2.4%)
Retired3 (2.4%)
Unemployed42 (33.1%)
Cardiovascular comorbidities 102 (80.3%)
Hypertension11 (8.7%)
Cardiovascular medications97 (76.4%)
Respiratory diseases18 (14.2%)
Other diseases104 (81.9%)
Visual impairments56 (44.1%)
Thyroid dysfunction12 (11.5%)
Neuropathies3 (2.9%)
Joint diseases8 (7.7%)
Scoliosis71 (68.3%)
Multiple conditions1 (0.96%)
BMI
kg/m2 (median; IQR)21.33 (19.11–23.41)
PHQ-9
Score (mean ± SD)6.01 ± 4.51
No depressive symptoms52 (40.9%)
Minimal depressive symptoms54 (42.5%)
Minor depression15 (11.8%)
Moderate major depression4 (3.1%)
Severe major depression2 (1.6%)
ISI
Score (mean ± SD)7.02 ± 4.13
No clinically significant insomnia78 (61.4%)
Subthreshold insomnia43 (33.9%)
Clinical insomnia (moderate)6 (4.7%)
FSS
Score (mean ± SD)3.88 ± 1.68
Clinically relevant fatigue40 (31.5%)
Fatigue in the last two weeks
Never9 (7.6%)
Sometimes80 (67.2%)
Every day30 (25.2%)
Fatigue pattern
In the morning25 (21.0%)
In the afternoon78 (65.5%)
All day16 (13.4%)
Legend: Summary of sociodemographic, clinical, and psychosocial characteristics of the included patients with Marfan syndrome (N = 127). Values are presented as numbers and percentages for categorical variables and as mean ± standard deviation (SD) or median with interquartile range (IQR) for continuous variables, as appropriate. PHQ-9 = Patient Health Questionnaire-9; scores range from 0 to 27, with higher scores indicating more severe depressive symptoms. ISI = Insomnia Severity Index; scores range from 0 to 28, with higher scores indicating more severe insomnia. FSS = Fatigue Severity Scale; scores range from 1 to 7, with higher scores indicating greater fatigue severity.
Table 2. Clinical, demographic, and psychosocial characteristics across the three identified clusters (k = 3).
Table 2. Clinical, demographic, and psychosocial characteristics across the three identified clusters (k = 3).
Characteristic Cluster 1
(N = 49)		Cluster 2
(N = 32)		Cluster 3
(N = 46)		p-Value
Age (mean ± SD)22.6 ± 3.136.4 ± 5.953.5 ± 6.1<0.001
Years since diagnosis (median, IQR)8.6 (6.6–10.0)12.1 (9.7–15.0)15.6 (13.7–19.0)0.0027
BMI (median, IQR)20.0 (18.4–21.4)21.8 (19.5–23.6)23.9 (22.5–25.6)<0.001
Sex: male20 (40.8)17 (53.1)24 (52.2)0.137
Sex: female14 (59.2)23 (46.9)29 (47.8)
PHQ-9 (mean ± SD)3.5 ± 2.65.8 ± 3.77.2 ± 3.60.016
No depressive symptoms18 (36.7)15 (46.9)19 (41.3)
Minimal depressive symptoms12 (24.5)16 (50.0)26 (56.5)
Minor depression4 (8.2)5 (15.6)6 (13.0)
Moderate major depression03 (9.4)1 (2.2)
Severe major depression01 (3.1)1 (2.2)
ISI (mean ± SD)4.3 ± 3.76.7 ± 4.58.8 ± 4.80.029
No insomnia22 (44.9)23 (71.9)33 (71.79)
Subthreshold insomnia11 (22.4)16 (50.0)17 (37.09)
Clinical insomnia (moderate)1 (2.0)1 (3.1)3 (6.5)
FSS (mean ± SD)2.7 ± 0.83.5 ± 1.14.9 ± 1.0<0.001
Clinically relevant fatigue09 (25.1)31 (67.4)<0.001
Legend: BMI, body mass index; IQR, interquartile range; SD, standard deviation; PHQ-9, Patient Health Questionnaire-9; ISI, Insomnia Severity Index; FSS, Fatigue Severity Scale.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Udugampolage, N.S.; Taurino, J.; Pini, A.; Callus, E.; Magon, A.; Conte, G.; De Angeli, G.; Angolani, M.; Paglione, G.; Baroni, I.; et al. Phenotyping Fatigue Profiles in Marfan Syndrome Through Cluster Analysis: A Cross-Sectional Study of Psychosocial and Clinical Correlates. J. Clin. Med. 2025, 14, 5802. https://doi.org/10.3390/jcm14165802

AMA Style

Udugampolage NS, Taurino J, Pini A, Callus E, Magon A, Conte G, De Angeli G, Angolani M, Paglione G, Baroni I, et al. Phenotyping Fatigue Profiles in Marfan Syndrome Through Cluster Analysis: A Cross-Sectional Study of Psychosocial and Clinical Correlates. Journal of Clinical Medicine. 2025; 14(16):5802. https://doi.org/10.3390/jcm14165802

Chicago/Turabian Style

Udugampolage, Nathasha Samali, Jacopo Taurino, Alessandro Pini, Edward Callus, Arianna Magon, Gianluca Conte, Giada De Angeli, Miriam Angolani, Giulia Paglione, Irene Baroni, and et al. 2025. "Phenotyping Fatigue Profiles in Marfan Syndrome Through Cluster Analysis: A Cross-Sectional Study of Psychosocial and Clinical Correlates" Journal of Clinical Medicine 14, no. 16: 5802. https://doi.org/10.3390/jcm14165802

APA Style

Udugampolage, N. S., Taurino, J., Pini, A., Callus, E., Magon, A., Conte, G., De Angeli, G., Angolani, M., Paglione, G., Baroni, I., Iozzo, P., & Caruso, R. (2025). Phenotyping Fatigue Profiles in Marfan Syndrome Through Cluster Analysis: A Cross-Sectional Study of Psychosocial and Clinical Correlates. Journal of Clinical Medicine, 14(16), 5802. https://doi.org/10.3390/jcm14165802

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop