Blood pressure – the hematocrit and serum cholesterol levels change in patients with Parkinson’s disease long before the onset of motor symptoms


A research team led by Nagoya University in Japan has found that blood pressure, the hematocrit (the percentage of red blood cells in blood), and serum cholesterol levels change in patients with Parkinson’s disease long before the onset of motor symptoms.

This finding, which was recently published online in Scientific Reports, may pave the way for early diagnosis and treatment of the disease.

Parkinson’s disease, the second most common disease affecting the nervous system after Alzheimer’s disease, is caused by a deficiency in a neurotransmitter called dopamine.

It is known that more than half of all dopaminergic neurons are already lost in patients with Parkinson’s disease in the stage wherein they experience motor symptoms such as tremors, stiffness, and slowness of movement. In addition, previous studies have shown that non-motor symptoms, such as constipation, rapid eye movement sleep behavior disorder, impairment of the sense of smell, and depression, emerge in patients with Parkinson’s disease10 to 20 years before the onset of motor symptoms.

These results suggest that Parkinson’s disease develops decades before the onset of motor symptoms. “If we can detect biological changes in the patients’ bodies well before the onset of the motor symptoms, we can start medical treatments in an early stage,” says Professor Masahisa Katsuno of the Graduate School of Medicine at Nagoya University.

From this perspective, the research team led by Prof. Katsuno and Katsunori Yokoi, the lead author and graduate student at Nagoya University, focused on the results of general health checkups, which are carried out among individuals yearly in Japan.

The team analyzed multiple years of data from the checkups of 22 male and 23 female patients with Parkinson’s disease whose checkup results before the onset of motor symptoms were available. For comparison, the team also used data from the checkups of 60 male and 60 female healthy individuals who underwent checkups for at least four years.

The researchers first compared the baseline values of each checkup item between patients with Parkinson’s disease and healthy individuals separately by sex. In male patients, the weight, body mass index, hematocrit, total and low-density cholesterol levels, and serum creatinine levels were lower than those in healthy male individuals.

In female patients, the levels of blood pressure and an enzyme called aspartate aminotransferase were higher, while other items’ values were lower compared to those in healthy female individuals.

Next, the researchers examined longitudinal changes in the checkup items in patients with Parkinson’s disease before the onset of motor symptoms.

As a result, they found that in the premotor stage, blood pressure levels are increased in female patients, whereas total and low-density cholesterol levels and the hematocrit are decreased in male patients. Regarding other checkup items, no significant changes were observed.

“In this study, we found that blood pressure, hematocrit, and serum cholesterol levels are potential biomarkers of Parkinson’s disease before the onset of its motor symptoms,” says Prof. Katsuno.

“This finding indicates that general health checkups can help detect early signs of developing Parkinson’s disease.”

In this context, his team is now pursuing studies to identify individuals who are at high risk for the disease based on checkup examinees.

“We are also conducting clinical trials of medication in the individuals who are considered, based on their checkup data, to be at high risk for Parkinson’s, in an attempt to prevent the development of the disease in them.

Parkinson’s disease detection from 20-step walking tests using inertial sensors of a smartphone: Machine learning approach based on an observational case-control study

Parkinson’s disease (PD) [1] is a progressive neurological disease affecting the motor ability of the patient, by inducing involuntary tremor at rest, rigidity and slowness of movement [2]. Additionally, PD may cause difficulties in walking by impacting balance and increasing the frequency of falls or cause the inability to fluently walk through visual obstacles, also known as freezing of gait [3].

The symptoms may be suppressed by levodopa medication, which compensates the loss of dopamine in the central nervous system, but the effect of the medication varies on an individual level and is dependent on many factors. PD cannot be cured, but with a successful symptom management plan, the quality of life of the patient can be maintained for decades.

The Movement Disorder Society has developed the clinical diagnostic criteria for PD, which was last revised and published in 2015 [2]. The diagnosis consists of excluding any other diseases with similar symptoms and the detection of at least two out of three cardinal features of PD.

Currently the assessment of PD patients is based on subjective assessment by a neurologist, including physical tests and interviews of the patient, but not necessarily using any standard format of questionnaire or test sequence like the Unified Parkinson’s Disease Rating Scale (UPDRS) [4]. However, UPDRS is widely used in categorizing the symptoms during research studies, for example by Eskofier et al. [5] and Arora et al. [6].

As PD is a progressive disease, the patients regularly visit doctors to follow-up their health and to adjust the treatments if necessary. These follow-ups are based on subjective visual assessment by the neurologist and patient’s own description of the symptoms. However, visual assessment is prone to errors.

The severity of symptoms of a PD patient changes on a daily basis due to medication intake, stress, or the overall health of the patient. Therefore, there is a need for methods to objectively assess PD symptoms for longer periods to maintain appropriate medication balance and the quality of life at home. Furthermore, up to one-fourth of diagnoses are incorrect when the analysis is only made based on the initial visit [7, 8]. Diagnostic accuracy improves during follow-ups, which also indicates that a longer period of monitoring time would help in reaching the correct diagnosis immediately. As wearable sensor development has been increasing in recent years, the option for objective assessment of patients at home without increasing the number of clinical visits has become more feasible.

Objective measurement methods have been studied and applied in detecting and monitoring PD. Wearable sensors have been used in PD to study various different motor symptoms, for example, tremor [9], rigidity, freezing of gait [10–12] and the risk of fall [13, 14] or fall detection [15, 16] both in the clinic and at home. For example, Del Din et al. [17] discovered that wearable sensors provide accurate information for the analysis of gait characteristics in free-living environments.

Schlachetzki et al. [18] studied the differences of gait properties in the clinic by conducting 10 meter walking tests for 190 PD patients and 101 age-matched controls by attaching inertial sensor units to both shoes. The difference between gait parameters in PD patients and controls was significant at moderate stages of the disease. These studies show that the changes in movement can not only be observed visually but also measured quantitatively with wearable sensors.

Machine learning approaches have been popular in many areas, and the use has also increased in the research in PD application area. Several machine learning studies featuring different symptoms have been conducted with a varying number of participants [5, 10, 12, 19–23]. Most of the studies implementing machine learning in assessing PD symptoms at the laboratory or clinic environment have collected a relatively small dataset (n = 5–20) of PD patients [5, 10, 19–21].

Some have also collected a group of healthy controls [12, 22]. There is a larger study by Klucken et al. [23], in which 92 subjects were used in the training phase and 81 subjects were used in an independent validation phase. Clinic and laboratory studies have reached classification results up to 96% in detecting freezing of gait [10, 12, 21]. There are several studies reaching to accuracies of 82–90% in detecting other symptoms [5, 19, 20, 23].

In addition to the clinic measurements, few home or free-living studies have been conducted as well. A small-scale test by Arora et al. [6] showed that Random Forest (RF) is an efficient classification method in detecting PD from controls using gait. 98% sensitivity and specificity were achieved in their study.

Larger datasets of several hundreds or thousands of participants have been collected by recruiting the subjects personally [24] or completely remotely using the subjects own cell phones to participate and collect the data [25, 26]. This implies that large datasets can be collected via smartphone with relatively small resources.

The selection of the dataset depends on the research question, for example Nguyen et al. [27] used the data of 6805 subjects with less balanced groups, whereas Mehrang et al. [28] used only 1237 subjects of the same dataset but with age-matched PD patients and controls. They discriminated PD patients from healthy controls with RF receiving accuracy, sensitivity and specificity of 70% each.

Further, the effects of taking levodopa medication have been detected with an area under the curve value of 0.7 using K-nearest neighbours [27] and accuracy of 71% using RF [26] classification methods. Body fixed sensors can also monitor for example, walk-to-sit and sit-to-walk transitions to discriminate healthy older adults from mild and severe PD patients (85–92% accuracy using Support Vector Machine -type classifier) [24].

The gaps remaining in the existing literature seem to be the following: a variety of symptoms has been assessed in clinic-based measurements, whereas most of the home measurements have currently only discriminated PD patients from healthy controls although some studies included the study of medication effects [27].

Also, some of the earlier studies are rather small in the number of participants, therefore statistically powerful datasets are needed to assess the symptoms at home using machine learning. Finally, machine learning does not have a single methodology that suits for every purpose. Therefore, testing several methods for the datasets before analyzing the details is important to find optimal methodologies for the type of data used in the study, and to gather more knowledge on which machine learning methods are the most useful in this application area.

The research question of this study was to determine, what are feasible ways of selecting the features and classifying the walking tests performed at the clinic [29]. We aimed to analyze different feature selection methods to find the most feasible features for detecting the differences in walking.

We also aimed to test the performance of different machine learning algorithms to find suitable methods to differentiate the PD group and control group from each other. These research questions aim to provide more knowledge on the use of smartphones in measuring PD patients, and to provide general information of the machine learning methodologies suitable for this application area.

This article is structured as follows: Material and Methods section provides the details of the dataset used in this study, and the methods used in the human activity recognition chain. Results section provides the numerical and graphical results of the feature selection and classification.

Discussion section analyzes the results compared to earlier state-of-art and discusses possible limitations of the study and future work. Finally, Conclusions section provides a summary of the study and our main conclusions.

reference link:

More information: Katsunori Yokoi et al. Longitudinal analysis of premotor anthropometric and serological markers of Parkinson’s disease, Scientific Reports (2020). DOI: 10.1038/s41598-020-77415-1


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.