Using Clinical Insights to Design Analytical Solutions in Parkinson's Disease Trials

Part 2 of our 3-part series on blinded data analytics in CNS Clinical Trials.

In clinical trials, particularly those involving complex neurological conditions like Parkinson's disease (PD), the quality of data can significantly impact study outcomes. 

While traditional statistical approaches have their place, we're finding that incorporating deep clinical insights into analytical solutions leads to more meaningful and accurate assessments.  

This approach is revolutionizing how we identify data concerns and ultimately improving trial success rates. 

The Challenge of Data Variability in Parkinson's Disease

At first glance, tracking a patient's progress in a Parkinson's disease trial might seem straightforward. Clinicians use validated rating scales like the Unified Parkinson's Disease Rating Scale (UPDRS) to measure symptom severity over time.  

When looking at a typical patient's UPDRS trajectory, researchers expect to see relatively stable measurements with gradual changes that reflect disease progression or treatment effects. 

However, when examining real-world data, we sometimes encounter puzzling scenarios. Consider "Subject A" (a patient) from one of our case studies.  

Looking solely at the overall UPDRS scores across multiple visits, nothing appears immediately concerning—the scores fluctuate between approximately 8 and 18 points, which could represent normal variability in a progressive condition. 

But is this the complete picture? Clinical understanding of Parkinson's disease suggests we should dig deeper. 

The Missing Piece: Lateral Predominance in Parkinson's Disease 

One of the cardinal features of Parkinson's disease is asymmetry of motor signs, also known as lateral predominance.  

In most patients, symptoms begin and remain more pronounced on one side of the body. This asymmetry typically persists throughout the disease course, even as symptoms progress. While the difference between sides may decrease over time, a complete reversal of the affected side would be highly unusual from a clinical perspective. 

This clinical insight provides a perfect opportunity to enhance our analytical approaches. By separating UPDRS scores by laterality (left versus right), we can uncover patterns that might be missed when looking only at total scores.

Revealing Hidden Data Concerns Through Clinical Insights 

When we applied this clinically informed analytical approach to "Subject A," (a patient), the results were striking. While the total UPDRS scores appeared reasonable, the laterality pattern showed dramatic and clinically implausible changes: 

• At early visits, the patient showed clear left-side predominance
• Midway through the trial, there was an abrupt switch to right-side predominance
• The laterality continued to swing back and forth between visits

These dramatic shifts in laterality are highly inconsistent with the known clinical presentation of Parkinson's disease. A typical PD patient's laterality chart should show consistent predominance on one side, with relatively minor fluctuations—exactly what we observed in our "Typical PD Patient" reference data. 

Expanding this analysis to an entire study site revealed similar concerns. While the site's overall UPDRS scores did not reveal any concerning patterns, the laterality patterns showed wild fluctuations across most patients recruited at the site, suggesting systematic issues with assessment techniques rather than isolated errors. 

Quantifying the Impact of Blinded Data Analytics 

To validate this approach, we conducted a comparative analysis across multiple sites and visits, comparing those with and without blinded data analytics (BDA) implementation. The results were compelling: To quantify the impact of our approach, we compared historical trials in PD without BDA implemented with those that implemented BDA methodologies.  

• In studies without BDA implementation, laterality changes affected over 5% of data compared to 3% in studies with BDA implemented (p < 0.001) More importantly, at the site level, almost 19% of sites were identified with significant presence of changes in lateral predominance in studies with BDA implemented compared to 6% of sites with BDA in place. (p = 0.003) These statistically significant improvements demonstrate the value of incorporating clinical insights into analytical approaches. 

From Insights to Implementation: Recommendations for Parkinson's Disease Trial Quality

Based on these findings, we've developed a comprehensive approach to data quality in Parkinson's disease trials that: 

1. Leverages clinical understanding: Uses disease-specific features like laterality to identify implausible patterns
2. Goes beyond standard metrics: Examines relationships between subscores that should follow established patterns
3. Provides clinically meaningful feedback: Helps researchers understand why certain patterns are implausible from a clinical perspective
4. Implements targeted training: Addresses specific misconceptions about disease presentation

Real-World Applications and Benefits

This clinically informed analytical approach delivers several key benefits: 

• Earlier detection of data concerns: Identifies issues that might be missed by conventional statistical methods
• More precise remediation: Targets training to address specific clinical misconceptions
• Improved signal detection: Reduces noise in the data by eliminating clinically implausible variability
• Enhanced scientific validity: Ensures that trial outcomes reflect true disease patterns and treatment effects 

Conclusion 

The integration of deep clinical insights into analytical solutions represents a significant advancement in clinical trial methodology.  

By understanding the clinical features of diseases like Parkinson's—not just their statistical presentations—we can develop more sophisticated approaches to data quality. 

This case example demonstrates that the most effective analytical solutions aren't built on statistical principles alone. They incorporate fundamental understanding of disease pathophysiology, progression patterns, and clinical presentation.  

As we continue to refine these approaches, we're creating a more robust foundation for successful clinical trials and, ultimately, more effective treatments for patients with neurological conditions. 

For researchers and sponsors planning Parkinson's disease trials, implementing these clinically informed analytical approaches can make the difference between robust, interpretable results and data that raises more questions than answers. 

This article is the second in our three-part series exploring Blinded Data Analytics in CNS Clinical Trials: 

Part 1: Impact of Data Concerns in Neurological Clinical Trials: Why Quality Matters 

About the Author

Dr. Alan Kott is the Practice Leader for Data Analytics at Signant Health, with both academic and industry experience in clinical trials. He has led the development of Signant’s Data Analytics Program, overseeing data analytics in over 200 clinical trials across multiple indications. Prior to joining Signant, Dr. Kott was an Assistant Professor at Charles University and a house officer in psychiatry at General Teaching Hospital in Prague. He holds a Medicinae Universae Doctor (MUDr.) from Charles University.

Dr. Alan Kott's presentation, "Application of BDA Methodologies in Clinical Trials Focus on Neurology," was supported by Petra Reksoprodjo (Director of Clinical Program & Performance, Operations) and Chris Murphy (Associate Director Clinical Service, Operations), and presented at ISCTM in Washington, D.C.