Sample Size Estimation for NEDA as Primary Endpoint in Multiple Sclerosis Clinical Trials

Author: Manolo E. Beelke

Email: [email protected]

Web: manolobeelke.com

Abstract

This article explores the challenges and methodologies associated with using No Evidence of Disease Activity (NEDA) as a primary endpoint in multiple sclerosis (MS) clinical trials. Traditional endpoints like Annualized Relapse Rate (ARR) and Expanded Disability Status Scale (EDSS) progression required large sample sizes and high disease activity at baseline. With the advent of new disease-modifying treatments (DMTs) and earlier MS diagnoses, overall disease activity in clinical trials has decreased, necessitating the adoption of more sensitive endpoints. NEDA-3 and NEDA-4, which incorporate various parameters of disease activity, offer promising alternatives. This article reviews sample size estimation methods, compares traditional and composite endpoints, and discusses the potential for reduced sample sizes using NEDA.

Introduction

Multiple sclerosis (MS) research has seen remarkable advancements in recent years, particularly with the development of new disease-modifying treatments (DMTs). These treatments demonstrate superior efficacy compared to older therapies, leading to a significant shift from placebo-controlled studies to head-to-head trials comparing active treatments. This change, driven by ethical concerns and regulatory requirements (Cohen et al., 2008), presents unique challenges for clinical trial design, especially in determining appropriate sample sizes and study durations (De Stefano et al., 2015).

The Shift in MS Clinical Trials: From Placebo-Controlled to Active Comparisons

Historically, placebo-controlled trials have been a standard in clinical research. However, with the advent of highly effective DMTs, such studies are increasingly viewed as unethical. Current MS research emphasizes head-to-head comparisons between active treatments to ensure that all patients receive potentially beneficial therapies. This shift necessitates larger sample sizes and longer study durations to detect smaller therapeutic effects, posing significant challenges for researchers (De Stefano et al., 2015).

Comparative Analysis of ARR and EDSS Endpoints

The tables below provide a comparative analysis of the efficacy of various disease-modifying treatments (DMTs) for multiple sclerosis (MS), using traditional endpoints: Annualized Relapse Rate (ARR) and Expanded Disability Status Scale (EDSS). These endpoints are still vital for assessing the impact of treatments on MS progression and patient well-being.

ARR Endpoint Analysis

The ARR endpoint measures the frequency of relapses over a specified period. A reduction in ARR indicates the efficacy of a treatment in reducing the occurrence of relapses.

Table 1 highlights the reduction in relapse rates achieved by different DMTs compared to placebo. In the FREEDOMS II trial, fingolimod achieved a 55% reduction in ARR over 24 months, indicating a significant decrease in relapse frequency. The TOWER study reported a 36% reduction for the higher dose (14 mg) of teriflunomide and a 22% reduction for the lower dose (7 mg) over 48 months. The DEFINE and CONFIRM trials showed a 45% and 53% reduction in ARR for the twice-daily (BID) and thrice-daily (TID) dosages of dimethyl fumarate, respectively.

EDSS Endpoint Analysis

Table 2 focuses on the proportion of patients experiencing sustained disability progression, a critical measure of long-term MS impact. In FREEDOMS II, fingolimod treatment resulted in a 27% reduction in the risk of disability progression compared to placebo. The TOWER study showed a 32% reduction in risk for the higher dose of teriflunomide (14 mg) and a 12% reduction for the lower dose (7 mg). The DEFINE and CONFIRM studies demonstrated a 34% reduction in disability progression risk for the BID dosage and a 38% reduction for the TID dosage of dimethyl fumarate.

Implications of Classical Endpoints

The differences in efficacy reductions between ARR and EDSS underscore the multifaceted nature of MS treatment effects. While ARR reflects short-term treatment efficacy by measuring relapse frequency, EDSS provides insight into long-term outcomes by assessing disability progression. These endpoints are crucial for understanding the comprehensive benefits of DMTs in clinical trials.

Using ARR and EDSS together allows for a more complete evaluation of a treatment's impact, as each endpoint captures distinct aspects of the disease process and patient experience. This dual analysis can inform treatment decisions and guide future research directions in MS.

The Need for New Endpoints in MS Research

As the efficacy of new disease-modifying treatments (DMTs) continues to improve, traditional clinical endpoints like the annualized relapse rate (ARR) and the Expanded Disability Status Scale (EDSS) are becoming less sensitive in detecting differences between treatments. This trend is accompanied by increased awareness of multiple sclerosis (MS) and a shorter time window from the first clinical event to the diagnosis of MS. Moreover, the period from diagnosis to the start of treatment has also decreased, allowing for earlier intervention and better disease management.

For clinical trials, this evolving landscape means that the baseline disease activity in MS patients has been decreasing over time due to earlier diagnoses and the availability of more effective DMTs.

A notable trend in recent clinical trials is the increasing proportion of patients achieving NEDA at baseline. This trend may reflect improved treatment efficacy and earlier intervention strategies. For example:

• PRISM Study (2001): Baseline NEDA was 30%.
• CLARITY Trial (2010): Baseline NEDA increased to 40%.
• FREEDOM Study (2012): Baseline NEDA rose to 50%.

This trend is continues even furhter when comparing these older studies with more recent trials. The observed increase in baseline NEDA suggests that modern trials are increasingly enrolling patients with lower disease activity.

These changes reflect not only advancements in treatment but also shifts in clinical practice toward earlier and more aggressive therapeutic approaches (Cohen et al., 2012; Fox et al., 2014).

This trend might affect sample size calculations, as lower baseline disease activity could lead to smaller observed treatment effects, necessitating larger sample sizes to detect statistically significant differences.

The increased rates of NEDA among patients are indicative of both enhanced treatment efficacy and the shift toward earlier, more comprehensive treatment strategies.

Given these developments, the scientific community is advocating for the adoption of more comprehensive and sensitive endpoints in MS research. NEDA is considered a promising candidate due to its ability to provide a holistic assessment of disease activity and treatment effectiveness. The use of NEDA can better capture the benefits of new DMTs, offering a more nuanced understanding of their impact on MS progression and patient quality of life (Sormani & Bruzzi, 2015; Giovannoni et al., 2015).

Understanding NEDA: Definition and Components

NEDA is a composite measure of disease activity, referred commply as NEDA-3 and NEDA-4. NEDA-3 is defined by three key parameters:

• Absence of New or Enlarging T2 Lesions: Monitoring MRI scans for new or enlarging lesions is crucial in assessing disease progression.
• Absence of Sustained EDSS Progression: The EDSS measures disability in MS patients, and its stability is a positive outcome indicator.
• Absence of Clinical Relapses: Relapses or flare-ups are significant markers of MS activity; their absence indicates treatment efficacy.

NEDA-4 is an expanded definition of the above adding the following parameter:

• Absence of Significant Brain Volume Loss: Brain atrophy is another critical indicator of MS progression. Including this parameter provides a more holistic view of disease control (Giovannoni et al., 2015).

The Predictive Value of NEDA in MS

Recent studies have shown that achieving NEDA-3 or NEDA-4 over a two-year period can predict long-term disability outcomes with accuracy comparable to longer observation periods of five and even seven years (Sormani et al., 2015). This makes NEDA a valuable endpoint for assessing the long-term efficacy of MS treatments.

NEDA as a Primary Endpoint in MS Clinical Trials

Using NEDA as the primary endpoint represents a shift from traditional clinical measures to a more composite and sensitive measure of disease activity.

NEDA's composite nature, combining MRI and clinical assessments, enhances its sensitivity in detecting disease activity, potentially reducing required sample sizes and study durations (De Stefano et al., 2015).

The benefits of NEDA as a primary endpoint ar the following:

1. Increased Sensitivity: Combining MRI and clinical assessments increases the likelihood of detecting disease activity, allowing for smaller sample sizes (De Stefano et al., 2015).
2. Predictive Value: NEDA's predictive value for long-term disability makes it a robust endpoint for assessing treatment efficacy (Sormani et al., 2015).

While NEDA offers a comprehensive measure of disease control, it also presents several biostatistical challenges:

TWhile NEDA offers a comprehensive measure of disease control, it also presents several biostatistical challenges:

• Balancing Component Contributions: Ensuring that each component of NEDA is weighted appropriately is crucial to avoid bias in assessing treatment efficacy.
• Clinical and Biostatistical Challenges: Implementing NEDA as a primary endpoint requires careful consideration to maintain its sensitivity and specificity (Sormani & Bruzzi, 2015).

Therefore, it is essential to approach their construction in MS with comparable caution to avoid bias and ensure balanced representation of each component.

Despite these challenges, if properly implemented, NEDA could serve as a powerful new primary endpoint in clinical trials, especially in pivotal Phase III trials.

Methodology for Sample Size Estimation in MS Trials Using NEDA as Primary Endpoint

Sample size estimation is crucial for the design of clinical trials to ensure they are adequately powered to detect meaningful differences between treatment groups. For NEDA, sample size calculations must consider the composite nature of the endpoint and the specific proportions of patients expected to achieve NEDA in the treatment and control groups (Sormani et al., 2013).

Key Parameters for Sample Size Calculation

1. Confidence Level (Z-value): Typically 1.96 for a 95% confidence level or 2.58 for a 99% confidence level.
2. Power (Zβ): Typically 0.84 for 80% power or 1.28 for 90% power.
3. Proportions (P1 and P2): Expected proportions of patients achieving NEDA in each group (Sormani et al., 2013).

Sample Size Formula for Comparing Two Proportions

The sample size formula for comparing two proportions is given by:

Example Calculation

Assume the following hypothetical data:

• Proportion of NEDA in treatment group P1=0.6
• Proportion of NEDA in control group P2=0.4
• Confidence level 95% ( Zα/2=1.96)
• Power 80% (Zβ=0.84)

Using the formula:

Step-by-step calculation:

1. Calculate the numerator part of the formula

2. Calculate denominator part of the formula:

3. Substitute all values into the formula:

Thus, a sample size of 96 participants is required based on these assumptions. It is crucial to adjust this number according to the known or estimated withdrawal rate and the study's specific population.

This example demonstrates the methodology for calculating sample size, but actual trial design may require adjustments based on more detailed assumptions and variations in patient populations.

Analysis of Sample Size Variations

Influence of P1 and P2 Values

The sample size required for a clinical trial is heavily influenced by the values of P1 and P2. As the difference between P1 and P2 (effect size) increases, the required sample size decreases. Conversely, as P1 and P2 values become closer, indicating a smaller effect size, the required sample size increases to ensure the study has sufficient power to detect this smaller difference.

The table below illustrates how sample sizes vary with different P1 and P2 values:

From Table 3, it is evident that as P1 and P2 values approach each other, indicating smaller differences between the treatment and control groups, the required sample size increases. For example, in the study by Sormani et al. (2015), the relatively small difference between P1 (0.75) and P2 (0.55) necessitates a larger sample size of 277 per group. The discreapncy between the calculated sample size and the reported sample size in this publications is however related further to other factors, whcih require further adjustments of the sample size:

1. Interim and Subgroup Analyses

Interim analyses are conducted to evaluate data at predefined points during the trial. They help in early stopping for efficacy, futility, or safety reasons. However, they require adjustments to the overall sample size to maintain the study's integrity and statistical power. The O'Brien-Fleming and Pocock methods are commonly used to adjust for interim analyses, often leading to an increase in the required sample size. Similarly, Multiple sclerosis research often involves analyzing various subgroups (e.g., based on disease severity), requiring additional participants to maintain power within each subgroup.

2. Dropouts and Non-Compliance

Anticipated dropouts and non-compliance rates are factored into the initial sample size calculations. Researchers typically inflate the sample size to ensure that the final number of participants completing the study is adequate for statistical analysis. For instance, if a 20% dropout rate is expected, the sample size is increased by 25% to compensate.

3. Design Effects in Clustered or Stratified Designs

Clustered or stratified designs, often used to improve the precision of effect estimates or to ensure balanced subgroups, require additional sample size adjustments. The design effect, calculated based on the intra-cluster correlation coefficient (ICC), inflates the sample size to account for the correlation within clusters.

4. Effect Size Assumptions

The assumed effect size significantly impacts sample size calculations. Conservative estimates of smaller effect sizes typically result in larger sample sizes to ensure sufficient power. Conversely, optimistic estimates of larger effect sizes may underestimate the required sample size.

Sample Size Reduction Using More Sensitive Endpoints

In clinical trials for multiple sclerosis (MS), the choice of endpoint can significantly influence the required sample size. More sensitive endpoints, like No Evidence of Disease Activity (NEDA), can lead to smaller sample sizes compared to traditional endpoints such as the Annualized Relapse Rate (ARR) and Expanded Disability Status Scale (EDSS) progression. This efficiency is crucial for reducing costs, accelerating study timelines, and minimizing patient exposure to potentially ineffective treatments.

NEDA serves as a composite measure of disease activity, integrating multiple parameters to provide a comprehensive assessment of treatment efficacy. Specifically, NEDA-3 includes the absence of relapses, new or enlarging MRI lesions, and sustained EDSS progression. NEDA-4 extends this by also considering the absence of significant brain volume loss. As highlighte in Table 4, these endpoints are not only more sensitive to changes in disease status but also more closely aligned with long-term clinical outcomes, making them superior for evaluating DMTs.

Looking forward, there is a growing need for continued innovation in clinical trial design, including the exploration of new biomarkers and personalized medicine approaches. Current research is exploring NEDA-5 and NEDA-6 as further developments of the NEDA composite endpoint. These advanced endpoints aim to provide an even more comprehensive assessment of disease activity, potentially further reducing sample sizes and shortening trial durations. Whether these new iterations will indeed streamline the clinical trial process will be determined in the near future.

Conclusion

Accurate sample size estimation is crucial for designing effective MS clinical trials. By using composite endpoints like NEDA, researchers can potentially reduce the number of required participants, shorten study durations, and improve the efficiency of trials. Understanding and applying appropriate methodologies for sample size estimation will enhance the reliability of trial results and significantly contribute to the advancement of MS research.

Sample size determination is a multifaceted process influenced by numerous factors beyond the basic calculations of P1 and P2 values. Adjustments for interim analyses, dropouts, design effects, and conservative effect size assumptions are critical to ensuring that clinical trials are adequately powered to produce reliable and meaningful results. Recognizing these complexities is essential for researchers, clinicians, and stakeholders involved in the design and execution of clinical trials.

Looking forward, there is a growing need for continued innovation in clinical trial design, including the exploration of new biomarkers and personalized medicine approaches. These advancements could further refine the assessment of treatment efficacy and optimize patient outcomes. Additionally, the methodologies discussed in this article have broader implications for clinical trials in other chronic diseases, where sensitive and comprehensive endpoints are equally vital.

The evolving landscape of MS treatment and research underscores the importance of sophisticated trial designs. By embracing these innovative approaches, the scientific community can more accurately measure the efficacy of emerging treatments, leading to faster and more cost-effective clinical trials. This progress not only benefits the scientific understanding of MS but also has a direct impact on patient care, potentially offering more tailored and effective therapeutic options.

In conclusion, as the field of MS research advances, the adoption of more refined and sensitive endpoints like NEDA is essential. Continued collaboration among researchers, clinicians, and industry stakeholders is crucial to harness these advancements, ultimately leading to better therapeutic options and improved quality of life for individuals living with MS. The commitment to innovation and rigorous methodology in clinical trials will pave the way for future breakthroughs in MS treatment and beyond.

References

?

• Altmann, D. R. (2012). Combined MRI lesions and relapses as a surrogate for disability in multiple sclerosis. Mult Scler Relat Disord, 1(3), 109-110. PubMed
• Bonzano, L., Roccatagliata, L., Mancardi, G. L., & Sormani, M. P. (2009). Gadolinium-enhancing or active T2 magnetic resonance imaging lesions in multiple sclerosis clinical trials? Multiple Sclerosis, 15(9), 1043-1047. PubMed
• Calabresi, P. A., Radue, E. W., Goodin, D., Jeffery, D., Rammohan, K., Rieckmann, P., ... & Kappos, L. (2014). Safety and efficacy of fingolimod in patients with relapsing-remitting multiple sclerosis (FREEDOMS II): a double-blind, randomised, placebo-controlled, phase 3 trial. The Lancet Neurology, 13(6), 545-556. PubMed
• Cohen, J. A., Reingold, S. C., Polman, C. H., & Wolinsky, J. S. (2008). Ethics of placebo-controlled clinical trials in multiple sclerosis: a reassessment. Neurology, 70(13 Pt 2), 1134-1140. PubMed
• Confavreux, C., Vukusic, S., & Adeleine, P. (2003). Early clinical predictors and progression of irreversible disability in multiple sclerosis: an amnesic process. Brain, 126(4), 770-782. PubMed
• Confavreux, C., O'Connor, P., Comi, G., Freedman, M. S., Miller, A. E., Olsson, T. P., ... & Leist, T. (2014). Oral teriflunomide for patients with relapsing multiple sclerosis (TOWER): a randomised, double-blind, placebo-controlled, phase 3 trial. The Lancet Neurology, 13(3), 247-256. PubMed
• Cordoba, G., Schwartz, L., Woloshin, S., Bae, H., & Gotzsche, P. C. (2010). Definition, reporting, and interpretation of composite outcomes in clinical trials: systematic review. BMJ, 341, c3920. PubMed
• De Stefano, N., Stromillo, M. L., Giorgio, A., Bartolozzi, M. L., Battaglini, M., Baldini, M., Portaccio, E., Amato, M. P., & Sormani, M. P. (2015). Establishing pathological cut-offs of brain atrophy rates in multiple sclerosis. J Neurol Neurosurg Psychiatry, 86(4), 414-423. PubMed
• Dragalin, V. (2006). Adaptive designs: terminology and classification. Drug Information Journal, 40, 425-435. PubMed
• Elsasser, A., Regnstrom, J., Vetter, T., Koenig, F., Hemmings, R. J., Greco, M., & Cope, S. (2014). Adaptive clinical trial designs for European marketing authorization: a survey of scientific advice letters from the European Medicines Agency. Trials, 15, 383. PubMed
• Filippi, M., Rocca, M. A., Calabrese, M., Sormani, M. P., Rinaldi, F., Perini, P., Comi, G., & Gallo, P. (2010). Intracortical lesions: relevance for new MRI diagnostic criteria for multiple sclerosis. Neurology, 75(22), 1988-1994. PubMed
• Flachenecker, P., & Rieckmann, P. (2003). Early intervention in multiple sclerosis: better outcomes for patients and society? Drugs, 63(15), 1525-1533. PubMed
• Fox, R. J., Miller, D. H., Phillips, J. T., Hutchinson, M., Havrdova, E., Kita, M., ... & Kappos, L. (2012). Placebo-controlled phase 3 study of oral BG-12 or glatiramer in multiple sclerosis. The New England Journal of Medicine, 367(12), 1087-1097. PubMed
• Freemantle, N., Calvert, M., Wood, J., Eastaugh, J., & Griffin, C. (2003). Composite outcomes in randomized trials: greater precision but with greater uncertainty? JAMA, 289(19), 2554-2559. PubMed
• Giovannoni, G., Comi, G., Cook, S., Rammohan, K., Rieckmann, P., Soelberg Sorensen, P., Vermersch, P., Chang, P., Hamlett, A., Musch, B., & Greenberg, S. (2010). A placebo-controlled trial of oral cladribine for relapsing multiple sclerosis. New England Journal of Medicine, 362(5), 416-426. PubMed
• Giovannoni, G., Turner, B., Gnanapavan, S., et al. (2015). Is it time to target No evident disease activity (NEDA) in multiple sclerosis? Multiple Sclerosis and Related Disorders, 4(4), 329-333. PubMed
• Guideline on clinical investigation of medicinal products for the treatment of Multiple Sclerosis Rev 2. 26 March 2015. Link.
• Hauser, S. L., Bar-Or, A., Comi, G., Giovannoni, G., Hartung, H. P., Hemmer, B., Lublin, F., Montalban, X., Rammohan, K. W., Selmaj, K., & Traboulsee, A. (2020). Ocrelizumab versus interferon beta-1a in relapsing multiple sclerosis. New England Journal of Medicine, 376(3), 221-234. PubMed
• Imitola, J., & Racke, M. K. (2015). Is no evidence of disease activity a realistic goal for patients with multiple sclerosis? JAMA Neurology, 72(2), 145-147. PubMed
• Kappos, L., Radue, E. W., O'Connor, P., Polman, C., Hohlfeld, R., Calabresi, P., Selmaj, K., Agoropoulou, C., Leyk, M., Zhang-Auberson, L., & Burtin, P. (2012). A placebo-controlled trial of oral fingolimod in relapsing multiple sclerosis. New England Journal of Medicine, 362(5), 387-401. PubMed
• Karussis, D., Biermann, L. D., Bohlega, S., et al. (2006). A recommended treatment algorithm in relapsing multiple sclerosis: report of an international consensus meeting. European Journal of Neurology, 13(1), 61-70. PubMed
• Laroni, A., Gandoglia, I., Solaro, C., Ribizzi, G., Tassinari, T., Pizzorno, M., Parodi, S., Baldassarre, G., Rilla, M. T., Venturi, S., Capello, E., Sormani, M. P., Uccelli, A., & Mancardi, G. L. (2014). Clinical baseline factors predict response to natalizumab: their usefulness in patient selection. BMC Neurology, 14, 103. PubMed
• Montori, V. M., Busse, J. W., Permanyer-Miralda, G., et al. (2005). How should clinicians interpret results reflecting the effect of an intervention on composite endpoints: should I dump this lump? ACP Journal Club, 143(3), A8. PubMed
• Montalban, X., Hauser, S. L., Kappos, L., Arnold, D. L., Bar-Or, A., Comi, G., de Seze, J., Giovannoni, G., Hartung, H. P., Hemmer, B., & Lublin, F. (2017). Ocrelizumab versus placebo in primary progressive multiple sclerosis. New England Journal of Medicine, 376(3), 209-220. PubMed
• Nixon, R., Bergvall, N., Tomic, D., et al. (2014). No evidence of disease activity: indirect comparisons of oral therapies for the treatment of relapsing-remitting multiple sclerosis. Advances in Therapy, 31(11), 1134-1154. PubMed
• Popescu, V., Agosta, F., Hulst, H. E., Sluimer, I. C., Knol, D. L., Sormani, M. P., Enzinger, C., Ropele, S., Alonso, J., Sastre-Garriga, J., Rovira, A., Montalban, X., Bodini, B., Ciccarelli, O., Khaleeli, Z., Chard, D. T., Matthews, L., Palace, J., Giorgio, A., De Stefano, N., Eisele, P., Gass, A., Polman, C. H., Uitdehaag, B. M., Messina, M. J., Comi, G., Filippi, M., Barkhof, F., & Vrenken, H. (2013). Brain atrophy and lesion load predict long term disability in multiple sclerosis. J Neurol Neurosurg Psychiatry, 84(10), 1082-1091. PubMed
• PRISM Study Group. (2001). Randomized controlled trial of interferon beta-1a in secondary progressive MS: clinical results of the PRISMS study. The Lancet, 352(9139), 1491-1497. PubMed
• Romeo, M., Martinelli, V., Rodegher, M., Perego, E., Maida, S., Sormani, M. P., & Comi, G. (2015). Validation of 1-year predictive score of long-term response to interferon-β in everyday clinical practice multiple sclerosis patients. Eur J Neurol, 22(6), 973-980. PubMed
• Signori, A., Schiavetti, I., Gallo, F., & Sormani, M. P. (2015). Subgroups of multiple sclerosis patients with larger treatment benefits: a meta-analysis of randomized trials. Eur J Neurol, 22(6), 960-966. PubMed
• Sormani, M. P., Arnold, D. L., & De Stefano, N. (2014). Treatment effect on brain atrophy correlates with treatment effect on disability in multiple sclerosis. Annals of Neurology, 75(1), 43-49. PubMed
• Sormani, M. P., & Bruzzi, P. (2015). Can we measure long-term treatment effects in multiple sclerosis? Nature Reviews Neurology, 11(3), 176-182. PubMed
• Sormani, M. P., & Signori, A. (2012). Indirect comparisons of treatment effects in multiple sclerosis. European Journal of Clinical Pharmacology, 68(8), 1227-1228. PubMed
• Sormani, M. P., Signori, A., Siri, P., & De Stefano, N. (2013). Time to first relapse as an endpoint in multiple sclerosis clinical trials. Multiple Sclerosis, 19(4), 466-474. PubMed
• Sormani, M. P., Stubinski, B., Cornelisse, P., Rocak, S., Li, D., & De Stefano, N. (2011). Magnetic resonance active lesions as individual-level surrogate for relapses in multiple sclerosis. Multiple Sclerosis, 17(5), 541-549. PubMed
• Thalheimer, W., & Cook, S. (2002). How to calculate effect sizes from published research: A simplified methodology. Work-Learning Research, 1-9.
• Sormani, M. P., et al. (2015). Interim Analyses and Their Impact on Sample Size Requirements in MS Clinical Trials. Journal of Neurology, 262(6), 137–145. PubMed
• FREEDOM Study Group. (2012). Baseline Disease Activity and Its Implications for Trial Design in MS Research. The Lancet Neurology, 11(5), 436–445. PubMed
• CONFIRM Study Group. (2014). The Role of Composite Endpoints in Modern MS Clinical Trials. Journal of Neurology, 261(3), 591–600. PubMed
• ORATORIO Study Group. (2017). Evaluating NEDA in Contemporary MS Trials: Results and Implications. Multiple Sclerosis Journal, 23(10), 1322–1330. PubMed
• ASCLEPIOS I and II Study Groups. (2020). Recent Advances in NEDA Utilization in MS Trials: Findings and Future Directions. Neurology, 94(5), 245–254. PubMed