Cluster and Multiple Correspondence Analyses in Rheumatology

Clinical phenotyping in patients with anti-synthetase antibodies using cluster analysis

Objectives

To characterize clinically distinct subgroups among unselected patients with anti-synthetase antibodies using cluster analysis.

Methods

This study evaluated patients with anti-synthetase antibodies registered to two independent cohorts; 106 consecutive patients from a prospective, single-centre cohort of the Scleroderma/Myositis Centre of Excellence (SMCE) were used as a derivation cohort and 125 patients from the Multicentre Retrospective Cohort of Japanese Patients with Myositis-Associated Interstitial Lung Disease (JAMI) were used as a validation cohort. Anti-synthetase antibodies were identified by RNA immunoprecipitation. A multiple correspondence analysis followed by hierarchical clustering was performed to aggregate the patients into homogeneous subgroups. Subsequently, a simple-to-use classification tree was generated using classification and regression tree analysis.

Results

Three clusters were identified in the SMCE cohort: cluster 1 (n = 48), the interstitial pneumonia with autoimmune features/amyopathic dermatomyositis cluster, associated with older age at diagnosis and a higher frequency of malignancy; cluster 2 (n = 46), the DM cluster, corresponded to a younger age at diagnosis with a higher prevalence of myositis, arthritis, DM pathognomonic rashes, mechanic's hands and fever; and cluster 3 (n = 12), the SSc cluster, characterized by chronic interstitial lung disease. There was no significant difference in overall survival or progression-free survival between the clusters. A simple classification tree using myositis and RP was created in the SMCE cohort. Clusters 1 and 2 were successfully reproduced and the classification tree demonstrated favourable performance in the JAMI cohort.

Conclusion

Patients with anti-synthetase antibodies were classified into three distinct phenotypes, indicating substantial heterogeneity within this patient group.

Risk factors for carriage of meningococcus in third-level students in Ireland: an unsupervised machine learning approach

The aim of this study was to examine the risk factors for pharyngeal carriage of meningococci in third-level students using an unsupervised machine learning approach. Data were gathered as part of meningococcal prevalence studies conducted by the Irish Meningitis and Sepsis Reference Laboratory (IMSRL). Pharyngeal swab cultures for meningococcal carriage were taken from each student once they had completed a single-page anonymous questionnaire addressing basic demographics, social behaviors, living arrangements, vaccination, and antibiotic history. Data were analyzed using multiple correspondence analysis through a machine learning approach.In total, 16,285 students who had a pharyngeal throat swab taken returned a fully completed questionnaire. Overall, meningococcal carriage rate was 20.6%, and the carriage of MenW was 1.9% (n = 323). Young Irish adults aged under 20 years and immunized with the meningococcal C vaccine had a higher MenW colonization rate (n = 171/1260, 13.5%) compared with non-Irish adults aged 20 years or older without the MenC vaccine (n = 5/81, 6%, chi-square = 3.6, p = .05). Unsupervised machine learning provides a useful technique to explore meningococcal carriage risk factors. The issue is very complex, and asked risk factors only explain a small proportion of the carriage. This technique could be used on other conditions to explore reasons for carriage.

Evaluating the Influence of Rain Event Characteristics on Rainfall Interception by Urban Trees Using Multiple Correspondence Analysis

Urban trees play an important role in the built environment, reducing the rainfall reaching the ground by rainfall interception. The amount of intercepted rainfall depends on the meteorological and vegetation characteristics. By applying the multiple correspondence analysis (MCA), we analysed the influence of rainfall amount, intensity and duration, the number of raindrops, the mean volume diameter (MVD), wind speed and direction on rainfall interception. The analysis was based on data from 176 events collected over more than three years of observations. Measurements were taken under birch (Betula pendula Roth.) and pine (Pinus nigra Arnold) trees located in an urban park in the city of Ljubljana, Slovenia. The results indicate that rainfall interception is influenced the most by rainfall amount and the number of raindrops. In general, the ratio of rainfall interception to gross rainfall decreases with longer and more intense rainfall events. The influence of the raindrop number depends also on their size (MVD), which is evident especially for the pine tree. For example, pine tree interception increases with smaller raindrops regardless of their number. In addition, MCA gives a new insight into the influence of wind characteristics, which was not visible using previous methods of data analysis (regression analysis, correlation matrices, regression trees, boosted regression trees). According to the nearby buildings, a wind corridor is sometimes created, decreasing rainfall interception by both tree species.

Prevalence of pectus excavatum (PE), pectus carinatum (PC), tracheal hypoplasia, thoracic spine deformities and lateral heart displacement in thoracic radiographs of screw-tailed brachycephalic dogs

Pectus excavatum, thoracic spine deformities, tracheal hypoplasia and lateral heart displacement are frequently described in brachycephalic dog breeds. Pectus carinatum is described sporadically, although the authors' observations demonstrate that it may occur in certain brachycephalic dog breeds. It was hypothesised that dogs of screw-tailed brachycephalic breeds carry a greater risk of these anomalies than normal-tailed brachycephalic breeds, and that there could a relation between the presence of pectus excavatum or pectus carinatum and thoracic spine deformities, tracheal hypoplasia and lateral heart displacement. During retrospective studies, these anomalies were identified in lateral and dorso-ventral radiographs of the thorax in brachycephalic dog breeds. A statistical analysis revealed that the frequency of pectus excavatum occurrence in screw-tailed and normal-tailed brachycephalic dog breeds is similar. The greatest risk of pectus excavatum occurrence is carried by two breeds: Maltese (60%) and English Bulldog (58%), while for pectus carinatum: Pug (41%) and French Bulldog (18%). Dogs of screw-tailed brachycephalic breeds carry a greater risk of kyphosis (p < 0.0001), tracheal hypoplasia occurrence (p < 0.0001), compared to "normal-tailed" breeds. The hypothesis concerning a relation between the presence of pectus excavatum or pectus carinatum and the other anomalies studied was not confirmed (p > 0.05). It was demonstrated that in dogs of brachycephalic breeds there was a greater risk of co-incidence between kyphosis of the thoracic spine and lateral heart displacement (p = 0.038), as well as kyphosis of the thoracic spine and tracheal hypoplasia (p = 0.003).

Phenotypes Determined by Cluster Analysis and Their Survival in the Prospective European Scleroderma Trials and Research Cohort of Patients With Systemic Sclerosis

Objective

Systemic sclerosis (SSc) is a heterogeneous connective tissue disease that is typically subdivided into limited cutaneous SSc (lcSSc) and diffuse cutaneous SSc (dcSSc) depending on the extent of skin involvement. This subclassification may not capture the entire variability of clinical phenotypes. The European Scleroderma Trials and Research (EUSTAR) database includes data on a prospective cohort of SSc patients from 122 European referral centers. This study was undertaken to perform a cluster analysis of EUSTAR data to distinguish and characterize homogeneous phenotypes without any a priori assumptions, and to examine survival among the clusters obtained.

Methods

A total of 11,318 patients were registered in the EUSTAR database, and 6,927 were included in the study. Twenty‐four clinical and serologic variables were used for clustering.

Results

Clustering analyses provided a first delineation of 2 clusters showing moderate stability. In an exploratory attempt, we further characterized 6 homogeneous groups that differed with regard to their clinical features, autoantibody profile, and mortality. Some groups resembled usual dcSSc or lcSSc prototypes, but others exhibited unique features, such as a majority of lcSSc patients with a high rate of visceral damage and antitopoisomerase antibodies. Prognosis varied among groups and the presence of organ damage markedly impacted survival regardless of cutaneous involvement.

Conclusion

Our findings suggest that restricting subsets of SSc patients to only those based on cutaneous involvement may not capture the complete heterogeneity of the disease. Organ damage and antibody profile should be taken into consideration when individuating homogeneous groups of patients with a distinct prognosis.

Complication patterns in patients undergoing venoarterial extracorporeal membrane oxygenation in intensive care unit: Multiple correspondence analysis and hierarchical ascendant classification

Background

Treatment by venoarterial extracorporeal membrane oxygenation (VA-ECMO) is widely used today, even though it is associated with high risks of complications and death. While studies have focused on the relationship between some of these complications and the risk of death, the relationship between different complications has never been specifically examined, despite the fact that the occurrence of one complication is known to favor the occurrence of others. Our objective was to describe the relationship between complications in patients undergoing VA-ECMO in intensive care unit (ICU) and to identify, if possible, patterns of patients according to complications.

Methods and findings

As part of a retrospective cohort study, we conducted a multiple correspondence analysis followed by a hierarchical ascendant classification in order to identify patterns of patients according to main complications (sepsis, thromboembolic event, major transfusion, major bleeding, renal replacement therapy) and in-ICU death. Our cohort of 145 patients presented an in-ICU mortality rate of 50.3%. Morbidity was high, with 36.5% of patients presenting three or more of the five complications studied. Multiple correspondence analysis revealed a cumulative inertia of 76.9% for the first three dimensions. Complications were clustered together and clustered close to death, prompting the identification of four patterns of patients according to complications, including one with no complications.

Conclusions

Our study, based on a large cohort of patients undergoing VA-ECMO in ICU and presenting a mortality rate comparable to that reported in the literature, identified numerous and often interrelated complications. Multiple correspondence analysis and hierarchical ascendant classification yielded clusters of patients and highlighted specific links between some of the complications studied. Further research should be conducted in this area.