East Horizon Archives -

Parkinson’s Disease Through a Statistical Lens

Parkinson’s disease — a progressive movement disorder of the nervous system — affects more than 1.1 million people in the US (and over 11 million globally), with an estimated 90,000 new diagnoses each year, making it the second-most common neurodegenerative disease after Alzheimer’s disease.1,2 The prevalence and rise of Parkinson’s disease has led to robust investment in understanding and treating this disorder.3

Here, we provide a brief overview of Parkinson’s disease and discuss common endpoints used in clinical trials with an illustrative case study on how those endpoints may be analyzed.

An introduction to Parkinson’s disease

Parkinson’s disease is a progressive movement disorder of the nervous system.4 It causes nerve cells (neurons) in parts of the brain to weaken, become damaged, and die, leading to symptoms that include problems with movement, tremor, stiffness, and impaired balance. As symptoms progress, people with Parkinson’s disease (PD) may have difficulty walking, talking, or completing other simple tasks.

The rate of PD progression and the particular symptoms differ among individuals. The four primary/hallmark symptoms of PD are tremor, rigidity, bradykinesia, and postural instability.

Other problems related to PD may include mental and emotional health problems, speech changes, dementia or other cognitive problems, pain, and fatigue.

On and Off states/periods

The On state is when PD medications are effective and motor and non-motor symptoms are controlled. The Off state is when PD symptoms return between medication doses or in the morning before the first dose.

Measuring Parkinson’s disease severity: Two evaluation methods

MDS-UPDRS: Evaluating motor and non-motor symptoms

The MDS-UPDRS (Movement Disorder Society–Unified Parkinson’s Disease Rating Scale) was developed to evaluate various aspects of PD, including daily non-motor and motor experiences and motor complications.5, 6

It is the most frequently used outcome in clinical trials, though it can also be employed in the clinical setting. It consists of four parts with 50 items in total, with each item rating the impairment with scores from 0 (normal) to 4 (severe). A patient’s global impairment is calculated as the total sum of these scores, with a higher score indicating greater impairment. Missing values might be imputed by the worst-case value of 4 (severe) if sufficient items are scored, otherwise the total score is set to missing. Each part can be analyzed separately as well.

MDS-UPDRS:

Parts of the MDS-UPDRS can be assessed during the ON and OFF state to evaluate the differences between those two states.

PDQ-39: A patient-reported health status questionnaire

The PDQ-39 (Parkinson’s Disease Questionnaire) is a 39-item patient-reported measure that assesses Parkinson’s disease–specific health-related quality of life.7, 8

It requires the patient to grade how often he/she experienced difficulties over the past month. Each item is scored on a scale from 0 (never) to 4 (always or cannot do at all, if applicable), with lower scores indicating better status. Items are grouped into eight dimension subscales.

PDQ-39:

PDQ-39 subscale scores range from 0 to 100, with 0 representing perfect health for the dimension and 100 representing worst health for the dimension. A PDQ-39 total score — the PDQ-39 Summary Index (PDSI) — can be computed as the mean of the eight PDQ-39 subscale scores providing an overall score reflecting the impact of Parkinson’s on quality of life.

In case of missing values, a possible approach is to impute missing values with the mean of the available subscale items, if the number of missing values is smaller than 50% within the subscale.

LED (Levodopa Equivalent Dose)

The dose of antiparkinsonian medication is standardized to the LED in mg based on predefined conversion rates.

A confirmatory Parkinson’s study: Statistical analysis and adaptive design

Our team partnered with a large biotech and biomedical engineering company to conduct the statistical analysis of a multi-center, open-label (one-arm) adaptive confirmatory study that used a device providing deep brain stimulation for Parkinson’s patients. The efficacy and futility boundaries of the adaptive design were computed using Cytel’s East Horizon™ platform.

The study had the following endpoints:

Primary endpoint: MDS-UPDRS (part III)
Secondary and exploratory endpoints: Other parts of MDS-UPDRS, PDQ-39, Clinical Global Impression of Change (CGI), Schwab and England ADL (Activities of Daily Living), antiparkinsonian medication use

Statistical analysis and its challenges

MDS-UPDRS (part III) score, PDQ-39, and antiparkinsonian medication use were analyzed using the paired t-test and CGI was analyzed using the non-parametric Wilcoxon signed-rank test. The Schwab and England ADL scale was analyzed with an ANOVA.

The first challenge was to understand the differences between the Off and On states. We also had to deal with missing data. It was decided that the missing values on visit level would be imputed by the worst response observed among all participants (primary analysis), with sensitivity analyses employing the baseline observation carried forward (BOCF) and the multiple imputation (MI) using Markov chain Monte Carlo (MCMC) methods.

Another more challenging aspect was understanding and programming the antiparkinsonian medication use (analyzed as secondary endpoint), which is calculated in LED. For this task, a close collaboration with the sponsor’s medical experts was needed to define the conversion factors and handle correctly special cases of medication combinations.

An adaptive design with four interim analyses

The study was designed to include four interim analyses and one final analysis, using the Lan-DeMets group sequential method with the O’Brien-Flemming α-spending function and Pocock β-spending function. The O’Brien-Fleming boundaries preserve a nominal significance level at the final analysis that is close to that of a single test procedure, so it is very conservative for the earlier interim analysis.9 The Pocock β-spending function uses approximately equal cutoffs for each analysis.

The efficacy and futility boundaries were computed via Cytel’s EAST software, which is integrated into the East Horizon™ platform. For the interim analyses, the efficacy and futility boundaries had to be recalculated based on the actual sample sizes.

Final takeaways

Parkinson’s disease is a lifelong and progressive, degenerative multiple-symptom disease that affects millions worldwide. The treatment is highly individualized and depends on the disease stage and severity of motor and non-motor symptoms. When symptoms become bothersome, current therapies primarily focus on symptom management, with pharmacological options such as levodopa and dopamine agonists forming the cornerstone of care. For those whose symptoms don’t respond well to medication in later stages, advanced options like deep brain stimulation (DBS) offer hope, which can provide relief for tremors and reduce dyskinesias.

The adaptive design of the case study offered a flexible, efficient, and ethical approach without compromising the validity and integrity of the study, which is implemented in the East Horizon™ platform that offers a comprehensive tool for trial design during all stages of development.

Blending Power and Flexibility: How AI-Generated R Code is Reshaping Clinical Trial Design

In today’s fast-evolving clinical research landscape, designing robust and efficient trials is more critical than ever. As statistical designs grow in sophistication, biostatisticians are increasingly relying on both commercial platforms and open-source tools to meet unique modeling needs. But this hybrid approach also comes with challenges, particularly for those new to advanced simulation software or lacking programming experience.

At Cytel, we’ve been exploring how artificial intelligence (AI) can help bridge this gap. At the 2025 Joint Statistical Meetings (JSM), we will present on our latest innovation: AI-powered R code generation for clinical trial design, a feature embedded in our East Horizon™ platform. This assistant, called RCACTS (R Coding Assistant for Clinical Trial Simulation), represents a significant step forward in making custom trial design faster, more accessible, and more reliable.

Why talk about this now? The open-source imperative

While commercial clinical trial design software offers rapid design development through validated and user-friendly workflows, it doesn’t always address the full complexity of real-world problems. Trial statisticians often face challenges in areas such as oncology, rare diseases, and adaptive designs that require tailored statistical tests, unique outcome generation models, or alternative randomization techniques.

This is where open-source tools like R become invaluable. R allows statisticians to write custom code to simulate complex trial designs, perform Bayesian analyses, or integrate evolving regulatory guidance. Over the years, a vibrant ecosystem of R packages has emerged, offering a high degree of transparency, flexibility, and academic rigor.

Yet this flexibility comes with trade-offs: code development can be time-consuming, error-prone, and requires significant programming expertise. As a result, many biostatisticians find themselves switching between validated commercial workflows and custom R functions, leading to a process that is often fragmented and inefficient.

Recognizing this, Cytel’s East Horizon platform has introduced R integration points, enabling users to inject custom code directly into validated simulation workflows. This integration delivers the best of both worlds: the speed and structure of commercial software with the creativity and control of open-source.

Enter AI: Speed, simplicity, and smarter coding

Our next logical question was: can AI make this process even easier?

The answer, increasingly, is yes. With recent advances in generative AI, particularly large language models (LLMs), it’s now possible to assist in the generation of R code for simulation-based design tasks. At Cytel, we’ve harnessed OpenAI’s GPT-4o via API, securely deployed within Microsoft Azure, to create RCACTS, a coding assistant purpose-built for biostatisticians.

Unlike generic AI tools that produce standalone R scripts, RCACTS generates R code specifically tailored for the East Horizon simulation engine. It ensures that the generated functions:

Match expected input/output structures,
Include pre-defined parameters as shown in our internal statistical package CyneRgy,
Are immediately ready for integration and testing within a live trial design workflow.

With RCACTS, users can simply describe what they want in plain English and receive functioning R code that can be integrated into East Horizon.

Who benefits? Everyone from newcomers to experts

One of the major advantages of this AI-enhanced workflow is lowering the barrier to entry. For a new user unfamiliar with Cytel’s R integration or syntax requirements, writing compatible code from scratch can be daunting. RCACTS significantly reduces the learning curve by providing validated function templates, sensible defaults, and clear parameterization, all supported by generative AI.

At the same time, experienced statisticians benefit by spending less time on repetitive coding tasks, debugging, or remembering function signatures. This allows them to focus on higher-level design questions, such as: What analysis method is most robust? How sensitive is the design to different outcome distributions? What dropout patterns pose the greatest risk?

Our assistant supports a wide range of trial design elements:

Simulating patient responses: Binary, Continuous, Time-to-event, and Repeated-measure endpoints.
Analyzing simulated data: Statistical analysis for these endpoints.
Randomization: Flexible randomization of patients across treatment groups.
Enrollment and dropout modeling: Custom mechanisms for realistic patient enrollment and dropout scenarios.
Treatment selection: Supporting multi-arm multi-stage (MAMS) trial designs.

Balancing innovation with responsibility

Of course, like any AI solution, there are caveats. AI-generated code must be carefully reviewed for correctness, appropriateness, and regulatory readiness. RCACTS includes a built-in testing functionality to flag structural or syntactic errors, but statistical validation remains the user’s responsibility. Also note that all data interactions adhere to Azure OpenAI’s stringent data protection policies to ensure security and compliance.

There’s also a broader concern: will over-reliance on AI limit the creativity and deep statistical thinking that define our profession? At Cytel, we view AI not as a replacement for expertise, but as a tool to amplify it. Our goal is to give statisticians more time and mental space to explore, iterate, and innovate rather than reduce them to prompt engineers.

Looking ahead

The future of clinical trial design lies in intelligent integration: combining the strengths of validated commercial tools, flexible open-source frameworks, and AI-powered coding assistance. With East Horizon and RCACTS, we believe we’re building the blueprint for this future, with a platform that supports both scientific rigor and operational speed.

As the field continues to evolve, biostatisticians will need tools that not only keep up with complexity but also support creativity, collaboration, and efficiency. AI-generated R code, embedded within a powerful simulation engine, is one such tool and is already transforming how we approach design flexibility in clinical trials.

Catch us at JSM 2025 to learn more about how AI is transforming the future of clinical trial design within Cytel.

Interim Decision-Making in Clinical Trials: A Focus on Sample Size Re-Estimation and Population Enrichment

In the evolving landscape of clinical trial design, flexibility and efficiency have become essential for success. Sample size re-estimation (SSR) and population enrichment — both adaptive trial design methods — use interim data to make informed mid-trial adjustments. While they address different aspects — SSR focusing on how many patients to enroll and population enrichment focusing on which patients to include — both approaches aim to optimize trial outcomes, reduce unnecessary exposure, and make better use of limited resources.

This blog explores how these two methods work, their statistical underpinnings, and how they can be used to build more ethical, targeted, and cost-effective trials.

Sample size re-estimation

Sample size re-estimation is a type of clinical trial design adaptation in which the sample size can be reassessed at an interim look, based on accumulated data. Over the years, this method has grown in popularity for several reasons:

SSR designs address variability in an observed treatment effect when the treatment shows some promise, but the effect size is not as pronounced as originally expected.
SSR designs produce more ethical trials, as they limit the number of patients exposed to treatment until sufficient efficacy evidence is collected.
These designs provide flexibility in trial implementation in cases of hard-to-recruit patient populations or rare disease.
They allow for gatekeeping of investment for biotech companies who may undergo additional scrutiny to justify additional R&D spend.
They limit the pursuit of relatively small treatment effects that may not be clinically meaningful.

The CHW and CDL statistical methods for SSR

Following the seminal work on adaptive interim analysis by Bauer and Kohne (1994) and others, Cui, Hung, and Wang proposed a method that is today widely accepted in the field of biostatistics, combining statistics with pre-specified weights to preserve Type I error now known as CHW (1999). An alternative method proposed by Chen, DeMets, and Lan (2004) and known as CDL, provides an alternative to the use of the weighted statistic in a confirmatory two-arm, two-stage design where the sample size of the second stage is increased based on an unblinded analysis of the data at the first stage.

Both CHW and CDL are accepted by regulatory bodies such as the FDA in cases where such an adaptation is deemed appropriate. The CHW method applies a lower weight to the contributions of the second stage of the design relative to those of the first stage, and the CDL method permits the use of conventional statistics for testing the primary endpoint at the end of the study while still preserving Type I error.

Population enrichment

Population enrichment is a clinical trial design adaptation that allows for the use of data from an ongoing clinical trial to adjust the sample size of the entire study population, or a promising subpopulation based on a specific biomarker or other characteristics. At the outset, the overall trial population is enrolled in the study, regardless of biomarker status or other subgroup attribute. At the time of an interim analysis, a decision can be taken to either continue enrollment of the overall population, a subgroup of the population that is showing promise, or terminate the entire study for futility. Restricting enrollment to a specific subgroup enriches the data collected for this subpopulation.

There are several benefits for this adaptation, including:

Optimizing resource allocation by enriching promising subpopulations while avoiding continued investment in less-successful subpopulations.
It allows investigators the opportunity to examine a larger population while reducing the risk of trial failure or unnecessary drug exposure due to heterogeneity among the study’s subpopulations.
At the same token, it increases the probability of success of a study, by increasing the sample size of promising subgroups.

How to model SSR and population enrichment

Both CDL and CHW methods for sample size re-estimation and population enrichment, are adaptations that can be modeled using Cytel’s East Horizon™ platform. Find out more by booking a product demonstration.

Final takeaways

Sample size re-estimation and population enrichment approaches are powerful adaptations in the biostatistician’s toolbox for advanced, cost-effective, and ethical clinical trial design. They empower sponsors to allocate R&D resources more appropriately towards promising treatments, while limiting exposure of patients to potentially ineffective or harmful treatments.

Metrics to Assess Clinical Trial Design Strength

The probability of success of a study is a critical metric in assessing the viability of a study design. In simple trial designs, probability of success can be defined as the study’s statistical power. However, more nuanced definitions of success are available, including some that incorporate assumptions of multiplicity (multiple study arms or multiple endpoints), and uncertainties about the true underlying treatment effect. Assurance is one such Bayesian concept, considering potential variability in treatment effect assumptions.

What is study power?

It is the conditional probability of rejecting your null hypothesis given an assumed treatment effect.

What is assurance?

One way of defining assurance is the expected power across different treatment effect assumptions. Assurance is especially useful when there is uncertainty around the treatment effect. Rather than calculate power based on a single assumption, we can calculate the power across a series of assumptions, and assurance is the average of the power across all scenarios.

In addition, if some information is available about the likelihood of each of these treatment effect scenarios prevailing, it can be incorporated in this calculation to produce a more realistic expectation of results. In this case, each treatment effect scenario is assigned a likelihood, and this likelihood is included in the assurance calculation. This process gives more weight to the scenarios that are more likely to be the true treatment effect.

Cytel has incorporated both concepts — power and assurance — into its East Horizon^TM platform. As well-established metrics in clinical trial design practice, these are widely used in the trial design process. In addition, Cytel has developed two related concepts to power and assurance, performance score and robustness score, to enhance the process of design, offer two additional metrics for assessment, and to elevate the role of the statistician as a strategic thinker in clinical development practice. These two metrics are also embedded in Cytel’s software.

What is the performance score?

The performance score is a linear weighted function that allows the statistician to integrate strategic priorities (reduction in sample size, reduction in duration, and increase in probability of success) into the clinical trial design process. The statistician can assign weights representing the importance of each strategic priority in assessing a study design, and the resulting score provides an estimate of the relative desirability of a design based on these priorities. Thus, the performance score provides an additional metric by which to assess the viability of a study design.

The weighted function of these performance criteria is:

What is the robustness score?

Like the concept of assurance, the robustness score is the average of the performance scores across different treatment effect assumptions. If there is uncertainty about the treatment effect, the robustness score takes that potential variability into account.

Here too, if some prior information about the likelihood of treatment effect scenarios is available, it can be incorporated in the calculation of the robustness score to produce a more nuanced expression of robustness, based on the strategic priorities set by the drug development team. If no prior information is available, each scenario automatically gets assigned equal likelihood.

These four concepts together, power, assurance, performance, and robustness scores, are key tools in the statistician’s toolbox for clinical trial design. Crucially, the latter two metrics are a shift in the statistician’s mindset from a purely scientific consideration of trial design, to incorporating strategic thinking in the design process. In addition, the scoring mechanism allows statisticians to identify a multitude of designs that satisfy basic statistical criteria and choose from among those the best-suited design based on strategic priorities. Finally, the score is a powerful communication tool, anchoring the statistician at the center of the discussion about tradeoffs in priorities in the design selection process with a wider cross-functional team.

In this way, the scoring mechanism embedded in East Horizon has transformed the software from a purely statistical design and analysis tool (albeit nuanced and powerful) to a clinical design strategy tool with a solid and potent statistical core. As more life sciences organizations adopt East Horizon and the advanced design tools it affords, we are seeing a gradual shift in the function of the statistician’s role within these organizations to a more central and consultative role in the trial design process.

Addressing Uncertainty in Survival Studies

As we have highlighted in prior blog posts, the ability to augment design characteristics with custom R code is especially relevant to the ever-evolving therapeutic area of oncology. As regulatory guidelines are routinely adjusted to comply with clinical practice and current research, oncology study simulations often require specific analysis approaches and/or patient outcome data generation methods to conform to changing evidence thresholds and to create more realistic simulated scenarios.

Defining parameters and addressing uncertainty in survival studies

As in all clinical studies, there is a degree of uncertainty in assessing the treatment effect in trials employing a survival endpoint. For these types of studies, the timing of a patient’s event is typically sampled from a distribution with known parameters such as an exponential distribution with a median time value for each arm in the trial. The assumptions employed in defining these parameters are based on some prior knowledge derived from previous studies, meta-analyses, or other experience of the clinical development team.

Why does this matter?

When prior data is scarce, both the assumed distributions and median values are highly uncertain, and may lead to trials that are more costly, longer in duration, and/or with a diminished probability of success. It is therefore important for product development teams to derive meaningful values for these inputs in the design stage of clinical studies.

Custom R coding for oncology designs

One approach to derisking such trials is to simulate patient data based on a distribution of possible median time values for each arm rather than one single value. This accounts for the fact that the true value is difficult to estimate before the trial begins and removes the need to select just one value. This approach also provides confidence in additional investment based on more realistic assumptions.

To employ this design approach, we propose using flexible R code in conjunction with Cytel’s East Horizon^TM platform to customize the way in which the data for each simulated patient is generated. We propose modifying the response generation’s algorithm to consider a distribution of true treatment effects rather than one single value assumption. The probability of success becomes more conservative but also more informative as the simulation is more realistic of the trials about to take place. This gives the product development team more confidence in trial execution and a better estimation of trial costs and length.

Want to learn more?

Watch J. Kyle Wathen and Valeria Mazzanti’s webinar “A Closer Look at Assurance: Sampling Patient Outcomes from Prior Distributions to Account for Uncertainty in Response Scenarios”:

Optimizing Interim Looks in Group Sequential Adaptive Study Designs

What are group sequential study designs?

Group sequential study designs include predetermined interim analyses (interim looks) in an ongoing clinical trial, to allow researchers the potential for stopping the trial earlier than the planned final analysis due to overwhelming evidence for success (efficacy), failure (futility), or safety concerns that arise from accumulating study data. Special considerations must be given to the preservation of Type-I error with the implementation of such interim looks, and several approaches have been developed over the years to control Type-I error, including those by Stuart Pocock, Peter O’Brien, and Thomas Fleming.

What are key considerations of group sequential designs?

There are several advantages for incorporating an interim look or looks in a study design, including the potential for more limited patient exposure, more efficient use of resources, time savings, and increased probability of success. Study design teams must weigh these considerations and agree on their strategic priorities before implementing group sequential design features. Specific points for consideration include the number and timing of interim analyses, and the stopping rules or thresholds used to declare early efficacy or futility.

Interim look timing

The timing of an interim look can be critical for the success of the group sequential approach. Performing the analysis too early may mean not enough information is available to make an informed decision; too late, and the benefits of the approach diminish significantly. Running extensive simulations across a variety of potential analysis time points can prove beneficial in selecting the optimal timeframe, balancing the team’s strategic priorities. Adding more than one interim look may prevail as a preferred approach, allowing for early stopping for futility only, with later look, or looks, focused on gains in early efficacy stopping (see schematic 1 below).

Schematic 1: A study with two interim looks: An early futility and later efficacy assessment

Early stopping rules

Setting the correct stopping rules for early efficacy and/or futility is also paramount in designing a robust clinical trial. If an early stopping threshold for futility is set incorrectly, it can lead to the termination of a promising treatment due to limited data. Conversely, setting a stopping rule for efficacy which is too aggressive, may lead to premature trial termination with inaccurate results. Here too, extensive simulation of trials with a variety of stopping rules for both efficacy and futility can help optimize these thresholds and the potential savings from these trial designs.

Schematic 2: Stopping boundaries for efficacy and futility: An interim look at 50% information fraction

A closer look at the benefits of implementing group sequential designs

Group sequential designs offer several key benefits in clinical trial practice:

Design trials that are more ethical: accurate decision rules for early stopping either for futility or efficacy can reduce the number of patients required for enrollment in a clinical trial and reduce unnecessary exposure of patients to potentially ineffective or harmful treatments.
Design trials with more efficient resource use: including interim looks in a study can lead to savings in both the timing and cost of clinical trials. Adaptive designs with interim analyses are shorter in overall average duration and average cost when compared to similar fixed study designs with no interim analyses. These savings are gained through the thoughtful implementation of early stopping rules.
Design trials with a higher probability of success: adaptive designs with interim analyses demonstrate and a higher average probability of success compared to fixed study designs. These benefits is especially pronounced when the true underlying treatment effect is clear at an early study stage (either beneficial or inefficacious).

Overall, interim analyses are an important feature in adaptive clinical trial design, and when well planned and executed, can lead to benefits and savings in clinical trial execution.

Group sequential designs now available in the East Horizon^TM platform

Cytel’s East Horizon platform now includes a Group Sequential module. This module offers statisticians the ability to compute and simulate single-arm and two-arm study designs with interim looks. The module allows users to select and optimize the number and timing of interim looks and the boundaries for efficacy and futility through advanced simulation and analysis tools.

Cytel’s East Horizon Group Sequential Module is the second in a series of six revamped cornerstone components of Cytel’s new cloud-based trial design platform. In combination with other platform components, the module provides statisticians with the tools needed for design, optimization, and selection of adaptive clinical trials with interim analyses.

Vaccine Efficacy Trials: Design Considerations and Simulation Tools

Vaccine efficacy (VE) trials play a critical role in assessing how well vaccines prevent infection or disease. These Phase 3 trials measure VE as the proportionate reduction in infection rates between vaccinated and unvaccinated groups. For decades, VE trials have been instrumental in the development of safe, life-saving vaccines, forming the cornerstone of public health policies. Their importance grew exponentially during the race to develop COVID-19 vaccines.

Designing robust VE trials is essential to generating reliable, actionable results. This is where tools like East Horizon^TM – Explore can make a significant impact by empowering researchers to design, simulate, and analyze these types of trials effectively.

Commonly used metrics for vaccine efficacy trials

Commonly used metrics to evaluate outcomes in vaccine efficacy trials include risk ratios, hazard ratios, and odds ratios:

Risk ratios: The risk of an event happening in one group vs. the risk of the same event happening in another group.
Hazard ratios: The relative risk of the complication based on comparison of event rates.
Odds ratios: The likelihood that an outcome will occur given a particular exposure vs. in the absence of that exposure.

These metrics enable researchers to evaluate outcomes with precision.

Special considerations for vaccine efficacy trials

VE trials have a unique set of characteristics that set them apart from other late-phase clinical trials. These characteristics include a large number of study participants, often in the tens of thousands of study subjects, specific follow-up and event requirements, unique testing rules, and super-superiority thresholds, which set a higher standard of efficacy for these products as they are targeted at a relatively small expected number of events. These features address the ethical, logistical, and public health demands of evaluating vaccines for healthy populations. Many of these aspects are shaped by regulatory guidelines (e.g., FDA, EMA) and global health priorities.

Some examples of these unique aspects include:

Fixed follow-up times: Standardized observation periods that ensure consistency in data collection and improve the reliability of trial results.

Targeted event counts & stopping boundaries: Setting target case numbers and stopping boundaries enhances trial efficiency by focusing resources on meaningful outcomes.
Unique testing methods for measuring vaccine efficacy

- 1 – Ratio of proportions: This approach compares infection rates between vaccinated and unvaccinated groups to estimate VE.
- 1 – Ratio of Poisson rates: Designed for time-to-event data, this method accommodates varying follow-up times among participants.

Super-superiority testing: Evaluate cases where vaccine efficacy significantly exceeds standard expectations.

Futility boundaries: Facilitate early termination of trials if interim results indicate the vaccine is unlikely to meet efficacy goals.

Generating and analyzing VE data using advanced simulation software

East Horizon – Explore enables precise trial design by simulating binary endpoints and time-to-event data and offers a powerful tool for analysis and data visualization. The solution allows users to model randomization and enrollment times, replicating realistic trial scenarios, and modeling enrollment schedules and infection incidence.

Binary and time-to-event endpoints allow biostatisticians to model infection risks and represent participants avoiding infection during the trial period. Additionally, East Horizon – Explore allows for effect measures and hypothesis testing at ease. Users can utilize either 1 – Ratio of proportions or Poisson rates as straightforward and industry-standard formulas.

Advanced continuity correction and R integration capabilities allow users to address potential Type-I error inflation for larger event rates, while enabling advanced customization through R code integration.

Financial analysis: Beyond basic efficacy testing

East Horizon – Explore goes beyond traditional VE analysis by incorporating optional financial and operational modeling. Users may incorporate revenue and cost modeling, alongside traditional efficacy testing. For example, users can include variables related to potential market share and associated costs, based on expected treatment thresholds to generate an expected Net Present Value (eNPV) forecast. This option enables strategic decision-making with detailed financial forecasts tailored to vaccine development, which is especially sensitive to cost and market access pressures.

Tailored financial forecasts are particularly important for vaccines because they differ from other clinical products in key ways. Unlike therapeutic drugs, vaccines often require substantial upfront investment for large-scale manufacturing, face shorter market exclusivity periods, and must balance affordability with global accessibility. These unique challenges demand a specialized approach to financial modeling that ensures both economic viability and alignment with public health priorities.

Final takeaways

Vaccine efficacy trials are critical component of public health. A unique set of characteristics, however, set them apart from other late-phase trials, requiring special consideration. Advanced simulation software, like East Horizon – Explore, can help sponsors optimize trial designs and gain deeper insights into trial outcomes.

Interested in learning more?

East Horizon – Explore offers a comprehensive platform tailored for a variety of designs, including VE trials. The platform empowers researchers with flexible design capabilities, rigorous statistical methods, and decision-support tools. From robust VE analysis to financial modeling, Explore facilitates data-driven decisions that advance vaccine research and enhance public health outcomes.

The Importance of Simulation in Designing Fixed-Sample Clinical Trials

Fixed-sample clinical trial designs are a type of clinical trial in which the patient population and number of patients are set prior to the beginning of the trial. These traditional designs do not include adaptive elements, but their relative simplicity in approach does not imply they require any less rigor or attention to the statistical design.

Here, we discuss the value of fixed-sample designs as well as the role of a simulation-driven approach in avoiding inaccurate estimations of study outcomes and probability of success.

Oncology Clinical Trials: Design Trends in Biomarker Research

Oncology research has seen many changes and advances in recent decades, from new therapies in combination with backbone chemotherapy to novel treatments targeting malignancies, and compounds targeting specific disease biomarkers at the genetic mutation level. The latter approach has called to question large, relatively long clinical studies assessing the safety and efficacy of treatments against a large population defined at the tumor level. Rather, research at the subpopulation or biomarker level has garnered much more interest as targeted treatments are being developed.

This focus on subpopulations and biomarkers is changing how researchers approach clinical trials in oncology and helps resolve several issues with larger clinical trials. For example, treatment effects may be diluted in a heterogeneous population, possibly resulting in an underpowered study. Furthermore, a large trial in a heterogeneous population may place patients for whom the drug is ineffective at risk of serious adverse events. On the other hand, restricting enrollment to a target subgroup without sufficient evidence may deny a large segment of the patient population access to a potentially beneficial treatment. This blog post will briefly introduce two statistical approaches addressing the rise of more specific study populations: predefined subpopulation statistical analysis in the context of a larger trial population and population enrichment of the more promising subgroup within an ongoing study.

Subpopulation Analysis

Subpopulation testing and analysis is a phase III clinical trial design strategy in which a subset of the study population is selected based on patient characteristics that may be more likely to respond to the treatment under investigation. Identifying and analyzing specific subpopulations allows the researcher to explore whether a treatment leads to different effects in a pre-designated subpopulation. A subpopulation can be defined by any stratification characteristic such as gender or geography, and in oncology clinical trials, specific biomarkers identified within a study population.

This type of approach to clinical research has several significant benefits in Oncology studies:

A large trial in a heterogeneous population may place patients for whom the drug is ineffective at risk of serious adverse events.

In a heterogenous population, the treatment effect may be diluted, possibly resulting in an underpowered study.

Restricting enrollment to the targeted subgroup without sufficient statistical evidence of lack of efficacy in the non‐targeted subgroup may eliminate beneficial treatment options for patients.

Subpopulation analysis allows for treatment recommendations based on individual characteristics.

As with any novel adaptive design approach, subpopulation analysis requires several considerations at the design stage. These considerations include the specific definition of the subpopulations for analysis in the study, the appropriate timing for an interim analysis, the methods used for hypothesis testing and type-1 error preservation, and the sequence of hypothesis testing of the different subpopulations and/or the full study population.

With these considerations in mind, rigorous planning and testing in the design stage of such a clinical trial is critical. Cytel’s East Horizon adaptive clinical trial design software offers a unique solution for the planning and testing of a clinical trial design that includes subpopulation analysis. In Cytel’s solution, hypothesis testing for the full and subpopulations can be performed using graphical multiple comparison procedures (gMPC) with a weighted Bonferroni procedure employed for closed testing. This method of hypothesis testing uses directed, weighted graphs where each node corresponds to a single hypothesis. A transition matrix is used as a complement to specify the weights and generate an intuitive diagram. Finally, a simple algorithm sequentially tests the individual hypotheses using the specified weights and hierarchies.

Population Enrichment

Population Enrichment is an adaptive clinical trial approach that includes the prospective use of any patient characteristic to obtain a study population in which detection is more likely than in the unselected population. There are two types of population enrichment: Prognostic Enrichment, in which a high-risk patient population is identified based on a biomarker, and Predictive Enrichment, in which the researchers identify a patient group more likely to respond to treatment. Some industry trends that have contributed to the popularization of this adaptive design method include the soaring costs of clinical trial execution, a move away from a “one-size-fits-all” approach to clinical development, and the rising interest in individualized medicine. This adaptive design approach has several benefits, including the identification of highly responsive patient populations, the efficient detection of a treatment effect in a smaller sample size, and the ability to identify beneficial treatments for a subgroup of patients that may have failed with a broader population in a more traditional study design.

Population enrichment can be seen as an extension of the sample size re-estimation (SSR) methodology, which we discussed in more depth in a previous blog post.

In the enrichment adaptive approach, a pre-specified number of subjects comprising the entire population, designated as cohort 1, is tested in an interim analysis, and a data monitoring committee reviews the results to assess efficacy or futility against predetermined thresholds. Suppose the analysis shows promising results for only a specific subpopulation of interest in the study, this population is “enriched” with additional patient enrollment in the remaining number of subjects of the study, designated as cohort 2, to enhance data collection for only this subgroup of interest and increase the overall probability of success of the study. As with any adaptive approach, this method has specific considerations, including closed testing with a p-value combination, the preservation of type-1 errors, and additional special considerations requiring attention in event-driven trials like most oncology ones.

Final Takeaways

Both subpopulation analysis and population enrichment are adaptive approaches to modern trial designs in oncology that offer great hope for researchers and patients alike. As the focus on specific patient populations narrows, these adaptive design types are gaining industry traction. Software-guided clinical trial design and simulation using tools such as East Horizon ensure adaptive elements are incorporated thoughtfully and are rigorously tested prior to trial launch.

Learn more about these approaches in our upcoming webinar ‘’Oncology Clinical Trials: Design Trends in Biomarker-Driven Research’’ with Boaz Adler and Valeria Mazzanti.

Discovery

Phase I-III Clinical Trials

Commercialization

Real-World Evidence Solutions

Clinical Trial Design

Trial Delivery

Advanced Analytics

Specialty Areas

Strategic Consulting

Beyond Functional Service Provider

Project-Based Analytical Solutions

Trial Design Software

Trial Implementation and Decision Support Software

LiveSLR® Software for Systematic Literature Reviews

Our Solutions

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

Quick Links

About Us

Quick Links

Insights

Quick Links

Careers

Quick Links

An introduction to Parkinson’s disease

Measuring Parkinson’s disease severity: Two evaluation methods

A confirmatory Parkinson’s study: Statistical analysis and adaptive design

Final takeaways

Why talk about this now? The open-source imperative

Enter AI: Speed, simplicity, and smarter coding

Who benefits? Everyone from newcomers to experts

Balancing innovation with responsibility

Looking ahead

Sample size re-estimation

The CHW and CDL statistical methods for SSR

Population enrichment

How to model SSR and population enrichment

Final takeaways

What is study power?

What is assurance?

What is the performance score?

What is the robustness score?

Defining parameters and addressing uncertainty in survival studies

Why does this matter?

Custom R coding for oncology designs

Want to learn more?

What are group sequential study designs?

What are key considerations of group sequential designs?

Interim look timing

Schematic 1: A study with two interim looks: An early futility and later efficacy assessment

Early stopping rules

Schematic 2: Stopping boundaries for efficacy and futility: An interim look at 50% information fraction

A closer look at the benefits of implementing group sequential designs

Group sequential designs now available in the East HorizonTM platform

Commonly used metrics for vaccine efficacy trials

Special considerations for vaccine efficacy trials

Generating and analyzing VE data using advanced simulation software

Financial analysis: Beyond basic efficacy testing

Final takeaways

Interested in learning more?

Subpopulation Analysis

Population Enrichment

Final Takeaways

Group sequential designs now available in the East Horizon^TM platform