Sebum RNA as a specific diagnostic tool for Parkinson's disease

A research group including Senior Associate Professor Shinji Saiki and Professor Nobutaka Hattori of the Juntendo University Graduate School of Medicine, and researchers from Kao Corporation and Preferred Networks, Inc. (PFN) have discovered that the sebum of patients with Parkinson's disease contains RNA which carries specific information related to the disease. Applying a machine learning model to the sebum RNA data of patients with Parkinson's disease revealed changes related to mitochondrial function in the patients of the disease, in contrast to healthy subjects. It is expected to result in early diagnosis by simple collection. The group had their results published in Scientific Reports.

There are about 150,000 patients with Parkinson's disease (PD) in Japan. PD is the second most common neurodegenerative disease in Japan. There is currently no cure for PD, but early diagnosis and treatment is crucial as it is a progressive disease. Specialized and complex examinations, such as MRI scans, are currently required for a diagnosis, highlighting the need for a more convenient diagnostic method. PD presents with motor symptoms such as muscle stiffness and tremors due to the loss of dopaminergic neurons in the substantia nigra of the brain. Recent studies have also revealed that non-motor symptoms such as depression and sleep disorders develop before the onset of motor symptoms.

Juntendo University has been researching biomarkers for the purpose of early diagnosis of PD to accurately reflect its progression with less invasive tests. Kao also revealed that the sebum contains abundant RNA information and established the new comprehensive analysis technique.

The hypothalamus, which is the center of the autonomic nervous system, commands the autonomic nerve cells of the spinal cord to regulate sebum, following which it is secreted by the sebaceous gland cells and moisturizes the skin. It has been reported that seborrheic dermatitis frequently occurs in PD patients and nerve endings, including those of the autonomic nerves, are reduced, which may change the composition of sebum. From this, the study group tested the possibility that sebum contains information related to PD.

Mild PD was specifically targeted in the study. Independent trials were conducted in two groups: Group 1 (7 untreated PD patients, 13 healthy subjects) and Group 2 (46 PD patients (including male patients taking antiparkinsonian medication), 50 healthy subjects), and the sebum RNA information of PD and healthy subjects were compared. The effects of the medication were examined in Groups 1 and 2.

A sample of sebum RNA can be obtained non-invasively by collecting sebum from the entire face with a single oil-blotting film. Information on 3768 genes in group 1 and 4685 genes in group 2 was obtained from the extracted sebum RNA as a result of comprehensively analyzing the RNA expression level using the next-generation sequencer. Among them, the researchers focused on approximately 200-400 RNAs with significant expression differences in PD patients to further investigate their functions.

The researchers found that, unlike in healthy individuals, the levels of multiple mitochondria related RNAs were modestly elevated in PD patients in both groups. Mitochondria are closely involved in changes in the substantia nigra neurons in PD. Following this, PFN constructed and verified a machine learning model to examine whether sebum RNA information can be used to determine the presence or absence of PD. Integrating the results of the analysis of groups 1 and 2 revealed that PD can be detected using sebum RNA, age, and gender information. Furthermore, it was revealed that by predicting the severity of PD using the same method and constructing a machine learning model by combining the predicted value with sebum RNA, age, and gender information, PD can be discriminated more accurately.

The ROC curve (sensitivity of the test) shows the results of the discrimination between PD patients and healthy subjects. The accuracy of the area under the curve increased to 0.793 when using sebum RNA, age, and gender information, and 0.806 when the severity prediction results were added. The researchers plan to add more cases in the future and verify whether differential diagnosis with other neurodegenerative diseases can be achieved with the aim of clinical application.