|
Written by ActiGraph |
Clinical development is a challenging task. Clinical development in Parkinson’s disease (PD) faces additional challenges and pitfalls due to the high variability seen within and across patients. There is no doubt that there is a critical unmet need to improve measurement quality in PD trials as the field works to develop effective treatment options for people with PD. Wearable digital health technologies (DHTs) provide an opportunity to address this need to improve outcome measures by providing objective, continuous data to complement traditional outcome assessments.
In our July session of the Digital Health Monthly scientific webinar series, we were joined by two excellent speakers who shared their expertise on the topic of how we can address the need for better measures in PD trials:
- Dr. Fatta Nahab, the Executive Director of Clinical and Digital Development at Neuron23 reviewed the symptoms and natural history of PD, the challenges facing researchers, and discussed the opportunities and important considerations when using wearable devices for Parkinson’s disease research.
- Karen Krieger, MBA, Digital Health Strategy Consultant with PKG Health then gave more specific examples of measures collected from passive monitoring with wearable technology for Parkinson's disease clinical trials.
The Challenge of PD: Reviewing Signs, Symptoms, and Disease Progression
Most individuals first think of the cardinal or motor symptoms associated with Parkinson’s disease, such as tremor, walking difficulties, slowness of movement (bradykinesia), muscle rigidity, and balance problems. There are also non-motor symptoms that are common in PD, including cognitive impairment, mood disorders such as depression or anxiety, loss of sense of smell, sleep disturbances, cardiovascular/autonomic problems such as orthostatic hypotension, pain, constipation, urinary and sexual dysfunction. This wide variety of symptoms can be quite disabling to individuals with PD.
There is high variability across individuals regarding symptoms expression and the timing of symptoms onset. One challenge in developing treatments for this population is the time of diagnosis, which is typically soon after motor symptom onset. However, non-motor symptoms can often begin decades earlier, but it may be difficult to recognize the significance of these symptoms at the time. Therefore, intervening in PD as early as possible is a challenge, given that most individuals are diagnosed almost mid-way through the progression of the disease, and changing the long-term outcomes becomes more difficult.
Another challenge is that people with PD are a highly heterogeneous population. Consensus in the field is that PD is not a single disease, but a variety of biological mechanisms that can lead to a common phenotype. There is extensive symptom variability not just across individuals, but within individuals as well. Disease progression is not simply a worsening of symptoms, but the appearance of new symptoms, changes in existing symptoms, and/or the reduction of symptoms can occur. This biological variability in PD is a large part of the reason for measurement noise that challenges traditional PD measurement scales. Increasing the frequency of measures of functioning, using active (app-based) or passive (wearables) digital assessments is going to be able to better characterize the biological variability and improve the signal to noise ratio from the data.
In this highlight video, Dr. Fatta Nahab describes how symptom expression due to biological variability in Parkinson's disease creates measurement challenges.
The Path to Digital Endpoints in Parkinson's Disease
Currently, there are no regulatory endorsed digital endpoints for PD (like the SV95C for Duchenne muscular dystrophy endorsed by the EMA). The Digital Medicine Society has provided the V3+ framework, which has been extremely valuable to provide a systematic road map for validation of these tools; but this road map can be long, complex, and costly. Regulatory guidance indicates that the ideal digital endpoint can measure disease progression with higher signal-to-noise ratio than existing measures, reflects clinical meaningful concepts of interest, and matters to people with PD and relates to their quality of life / activities of daily living.
In practice, integrating a digital endpoint into the development pathway requires several important strategic considerations. For example, it can be helpful to articulate the goal of DHT use in your situation –
- Will this tool be used for signal finding?
- To help inform a go/no-go decision?
- Or will you use this endpoint as part of your registration strategy?
Answering these questions can help to balance the maturity of the digital outcome assessment with when and how to use a DHT in different phase(s) of clinical development.
Some examples of how digital measures can be implemented into PD clinical trials would be to determine if there is a reduction in a bradykinesia score in the treatment arm compared to a placebo group; or in the case of a unique or novel tremor drug, using an objective measurement such as a tremor score to assess if there is an improvement resulting from that particular treatment.
Digital Endpoints in Parkinson's Disease from Validated Algorithms
ActiGraph recently announced a partnership with PKG Health to improve outcomes in neurology trials, particularly movement disorders such as PD. As outlined above, more frequent data collection using passive measures can better capture fluctuations in PD symptoms; PKG Health has developed and validated several algorithms for digital measures relevant to PD which are now available in the ActiGraph platform, such as digital measures of bradykinesia, dyskinesia and tremor (see more details on our Cardinal Symptoms page). Here we’ll share more detail about how these digital measures have been validated from an analytical and a clinical standpoint.
The validation of the Bradykinesia Score (BKS) involved multiple expert neurologists and movement disorder specialists who observed approximately 80 Parkinson's disease patients performing the Purdue Pegboard test. This test evaluates finger movement between two dots on a board. The specialists found a high correlation between the measurements taken by the wearable accelerometer device and the clinical observations, demonstrating the device's high sensitivity and specificity in distinguishing normal movement from the slower movement characteristic of bradykinesia in Parkinson's patients. The PKG algorithm's results were correlated with the Unified Parkinson's Disease Rating Scale (UPDRS), and the BKS from PD participants differed from that of control participants1.
In this highlight video, Karen Krygier explains the analytical and clinical validation of PKG Health's accelerometer-based Bradykinesia Score (BKS).
The validation of the Percent Time Tremor score was based on data collected from a large cohort of Parkinson's disease patients using an accelerometer. The raw data from the accelerometer were analyzed in a spectrogram to identify the tremor frequency typically associated with Parkinson's disease, which ranges between 4 to 6 Hz for resting or re-emergent postural tremors. This was contrasted with dyskinesia, which generally occurs in the 1 to 4 Hz range. The validation process involved testing and retesting the data to establish a threshold that could reliably discriminate between the presence and absence of tremor. It was determined that a Percent Time Tremor score of 0.8% served as a critical threshold. A score greater than 1% was highly accurate in predicting that a patient had tremor, while a score less than 0.6% was predictive of the absence of tremor2.
Use Cases of Digital Endpoints in Parkinson's Disease Clinical Trials
There are several different use cases of digital endpoints to measure PD symptoms depending on the objective and design of a clinical trial. One use case is to capture bradykinesia and dyskinesia scores to help the identify and stratify a target patient population (whether that be primarily bradykinetic or dyskinetic). This may help as clinical trial sponsors look to enroll subjects into the study that meet the criteria for being able to detect a true response from their novel drug or device for treating PD. Another use case is looking at the change in symptom severity by comparing a change in state or symptom severity from a baseline measurement of bradykinesia and dyskinesia score to a follow up measurement made during the clinical study.
In symptomatic studies investigating if there's a change in symptoms more in an acute timeframe, potentially studies looking at severity in mild to advanced phases and fluctuation or variability in those symptoms, digital measures such as bradykinesia, dyskinesia, tremor, wearing off, immobility, and fluctuations can provide objective measures to demonstrate symptom change compared to baseline. In disease modifying studies taking place over a longer period, objective, digital measures of median and active bradykinesia, gait and walking, and/or nighttime sleep may be appropriate.
Here is one real world example of how PKG digital endpoints were implemented into a 200-patient clinical study. The primary endpoint for this study was a change in MDS-UPDRS, but secondary outcomes included the PKG bradykinesia and dyskinesia scores. The study had two arms, one where the PKG scores were used to guide clinical disease management from an uncontrolled state to a controlled clinical state, and that was compared to the other arm receiving standard of care. There was a statistically significant and clinically meaningful improvement in the MDS-UPDRS total score (the study's primary endpoint) when using PKG scores to guide therapy changes. The PKG bradykinesia score in those same patients over the baseline to five months was statistically significantly decreased too.
Considerations for Clinical Trial Operations
Implementing DHTs or digital endpoints into your PD clinical trial operations workflow starts with data collection from the study subject, who may receive the wearable directly from the clinical staff that's running the study or can be sent to their home. This will allow participants to wear the device in their home environment during activities of daily living and when they're potentially exhibiting fluctuations in their disease state.
The data that's captured by the wearable device is processed through a cloud environment, where raw data is processed through algorithms to obtain the measure of interest. This data output is provided to clinical trial sponsors where they can view patient-level scores and use time stamps to look at data collected during specific time periods of the study protocol, and his data is be used in the study data analysis.
When implementing DHTs in clinical trials, it's crucial to balance the benefits of comprehensive data collection against the burden on participants, sites, and other stakeholders. Regulatory considerations add another layer of complexity, as the status of vendors and their products must align with the specific countries involved in the trial, sometimes requiring additional regulatory submissions for the technology itself. Effective training for staff, CROs, and participants is essential, ensuring proper data collection and continuous quality monitoring throughout the trial. Additionally, managing timelines is critical to the success of the trial, requiring careful planning and oversight.
Due to the complexity of these operational considerations that integrating DHTs into your clinical development program can introduce, most companies favor a vendor or partnering approach. When considering a partner, the maturity across hardware, software, integration, security, endpoint development experience, and regulatory engagement experience are all very important aspects to vet.
Summary: How Digital Endpoints Can Add Value to PD Clinical Trials
The use of objective measurements made with wearable DHTs, as opposed to traditional rating scales, offers significant advantages in clinical research. These objective measures can enhance the accuracy of statistical sample size calculations, improve the identification and stratification of patients based on eligibility criteria, and ensure more precise characterization of study populations. Additionally, the convenience and ease of passive data collection in a patient's home environment allows for real-time monitoring of symptom fluctuations, providing a more comprehensive understanding of the patient's condition in a natural setting.
References
1. Griffiths RI, Kotschet K, Arfon S, Xu ZM, Johnson W, Drago J, Evans A, Kempster P, Raghav S, Horne MK. Automated assessment of bradykinesia and dyskinesia in Parkinson's disease. J Parkinsons Dis. 2012;2(1):47-55. doi: 10.3233/JPD-2012-11071.
2. Braybrook M, O'Connor S, Churchward P, Perera T, Farzanehfar P, Horne M. An Ambulatory Tremor Score for Parkinson's Disease. J Parkinsons Dis. 2016 Oct 19;6(4):723-731. doi: 10.3233/JPD-160898. PMID: 27589540.