Time Delay

An educational symposium on the importance of meaningful treatment outcomes in Alzheimer disease (AD) was sponsored by Novo Nordisk and presented at ISPOR Europe 2023. To gather insights from different stakeholders, 5 speakers from the pharmaceutical and life sciences industry (Novo Nordisk and IQVIA), academia (Karolinska Institutet, Sweden), and the Norwegian Medical Products Agency (NOMA), as well as a person living with AD, participated in this symposium titled “Time delay: Searching for a Meaningful Outcome of Disease Modification in Alzheimer Disease.” CURRENT PRACTICE IN ASSESSING TREATMENT OUTCOMES IN AD AD is a chronic, progressive neurodegenerative disorder. In its early stages, the decline in cognitive function can be notable, although some people living with AD display minimal or no symptoms despite underlying brain changes. However, in later stages, there is accelerated cognitive decline, which ultimately leads to total loss of independent function (Fig. 1).1–3FIGURE 1: Evolution of AD symptoms. AD indicates Alzheimer disease; MCI, mild cognitive impairment.The Clinical Dementia Rating (CDR) is an established measure that was originally developed to capture treatment effects in mild cognitive impairment (MCI) and mild AD. In addition, tools such as the Global Deterioration Scale and the National Institute on Aging and the Alzheimer Association (NIA-AA) Disease Framework can be used to characterize the different clinical stages of AD. Contemporary clinical trial designs and outcome assessments have also been developed to detect modest treatment effects on cognitive and functional decline. However, people living with early AD (MCI due to AD and mild AD dementia) show high variability in baseline cognitive and functioning ability, which makes it difficult to measure disease progression. Despite studies aiming to select homogeneous trial populations, traditional outcome measures such as Clinical Dementia Rating-sum of boxes (CDR-SB) may lack the sensitivity to detect moderate treatment effects in early AD. Thus, it is a challenge to interpret current outcome measures in terms of a patient-relevant treatment effect, such as slowing disease progression.1,4 A PATIENT’S PERSPECTIVE: WHAT IS IMPORTANT TO SOMEONE LIVING WITH AD? Birgitta Martensson was diagnosed with AD in 2017, aged 66 years. Birgitta still leads a relatively “normal” life in the early stages of disease, living alone, managing her own finances, seeing friends, and traveling. However, she understands that one day, she will not be able to make decisions for herself. Time is of the essence, and that’s why she believes an early diagnosis is crucial for people living with AD. With an early diagnosis, Birgitta has been able to plan for her future with the support of her family and friends. However, a late diagnosis is a loss of time, whereby people living with AD cannot make their own decisions regarding their treatment and lifestyle.“I want to make my own decisions for my future” —Birgitta Martensson Evaluating the value of disease-modifying therapies (DMTs) in early AD requires a thorough understanding of the priorities of people living with AD.5 Early detection of AD and associated symptoms is crucial to manage dementia-related impacts on daily functioning, to enable people living with AD and their care partners to plan for the future and seek the support they need. A systematic review conducted on behalf of Real-World Outcomes Across the AD Spectrum for Better Care (ROADMAP)6 provided insights into the priorities of people living with AD, wherein 32 outcomes considered important by those people and their care partners were identified across 7 domains. According to the review, outcomes typically assessed in clinical trials, such as cognition and daily functioning, were considered important by people living with AD, but other priorities were uncovered, too, including the importance of maintaining quality of life, independence, and identity or personality.7 Similarly, the recent what matters most study conducted by the Alzheimer Disease Patient and Caregiver Engagement group8 asked people living with AD what an “ideal treatment for AD” would offer them. People whose disease fell across the AD continuum reaffirmed the importance of disease modification to slow progression and improve memory, as well as remaining independent and able to perform daily activities.9 TIME DELAY: A NOVEL METHOD FOR ASSESSING OUTCOMES IN AD During the symposium, panelists noted that developing tools that can measure meaningful changes and treatment responses in people with early AD is crucial, both from a patient perspective and a health economics perspective. Julie Hahn-Pedersen, Novo Nordisk A/S, highlighted the fact that DMTs can offer people with early AD hope for early intervention and improved outcomes. From a health economics perspective, Linus Jönsson of the Karolinska Institutet stated that reimbursement decisions around DMTs in AD will represent some of the most important reimbursement decisions in coming years, due to their perceived high price point and the substantial prevalence of the disease. Furthermore, DMTs present an incentive to diagnose the disease earlier and qualify people for treatment, which is vital for people living with AD and their care partners. Thus, it is crucial to ensure that interventions are evaluated and implemented appropriately.“Reimbursement decisions around novel disease-modifying therapeutics in AD will represent some of the most important reimbursement decisions in coming years” —Linus Jönsson, Karolinska Institutet Traditionally, when reviewing the results of clinical trials in AD, the mixed method for repeated measures (MMRM) is used. This method models data with high unexplained variability, and few assumptions are made regarding the response. It focuses on potential treatment-induced changes in the outcome measure from baseline at a fixed time point, without considering what has happened before or what will happen after the point of observation, producing a vertical point difference.10 Thus, its results can be difficult to translate into meaningful and interpretable outcomes for people with AD and their care partners.11 The time delay method is a novel progression model for repeated measures that may assist in the quantification of treatment outcomes from potential DMTs that aim to slow AD progression. For example, it can be used to measure the time taken to reach a specified disease state, or the time to disease progression with a potential treatment versus placebo.11 The time delay method can be applied to clinical trial readouts. In a recent study by van Dyck et al12 in which lecanemab was assessed in people with early AD (N=1795), the mean CDR-SB score at baseline was ~3.2 in both the investigational (n=898) and placebo arms (n=897). At 18 months, using the MMRM, the adjusted mean change from baseline in the CDR-SB score was 1.21 in the investigational arm and 1.66 in the placebo arm (difference, −0.45; 95% CI, −0.67 to −0.23; P <0.001).12 Applying the time delay method to this readout shows that the investigational arm took 18 months (78.2 wk) to reach the same CDR-SB score as the placebo arm at 12.7 months (55.2 wk). This indicates a 5.3-month (23.0 wk) slowing of disease progression with lecanemab (Fig. 2A).FIGURE 2: Application of the time delay method to data from recent clinical trials. A, Data from Van Dyck et al (2023),12 demonstrating a 5.3-month (23.0-wk) delay in disease progression after treatment with a humanized IgG1 monoclonal antibody. B, Data from Sims et al (2023),13 demonstrating a 7.5-month (32.6-wk) delay in disease progression after treatment with a humanized IgG1 monoclonal antibody. *P<0.001; † P<0.0001. CDR-SB indicates Clinical Dementia Rating-sum of boxes.Sims et al13 compared the clinical benefits of donanemab in early AD with placebo. In the low-tau/medium-tau population (n=1182), the between-group difference in the least-squares mean change from baseline at 76 weeks was −0.67 (95% CI, −0.95 to −0.40) for CDR-SB, using the MMRM. Applying the time delay method to these data showed that the investigational arm took 17.5 months (76.0 wk) to reach the same CDR-SB score as the placebo arm at 10.0 months (43.4 wk), correlating to a 7.5-month (32.6-wk) slowing of disease progression (Fig. 2B).“The time delay method allows for the translation of treatment outcomes into time saved for potential disease-modifying agents that aim to slow the progression of AD” —Julie Hahn-Pedersen, Novo Nordisk When quantifying treatment outcomes using MMRM or the time delay method, clinicians and payers may refer to minimal clinically important differences (MCIDs). For people with MCI, the MCID referred to most commonly in the literature for the CDR-SB scale is 0.98, and for people with mild dementia, it is 1.63.14 Therefore, an 18-month trial requires the treatment to have an effect size between −0.98 and −1.63 to be considered clinically meaningful. Current MCIDs are unlikely to reflect what is meaningful to people with AD, especially people with MCI due to AD or mild AD dementia.14 For example, in the study by van Dyck et al,12 a point difference of −0.45 was observed in the adjusted mean CDR-SB score between the investigational and placebo arms at 18 months. Although this translates to a 5.3-month delay in disease progression, this value would not be deemed clinically meaningful with the MCID values stated above. To be considered clinically meaningful, an 11-month delay in disease progression (−0.98) would need to be observed in people with MCI, and a near halt in disease progression (−1.63) in people with mild dementia.15 As such, additional research to establish MCIDs specifically related to time delay would provide crucial support for people with AD, to better understand treatment outcomes, especially in early disease stages. APPLYING THE TIME DELAY METHOD TO HEALTH ECONOMIC MODELING Where current trials do not offer extended follow-up periods beyond 18 months, there is a lack of data on the long-term effects of therapeutics for AD and their impact on institutionalization and healthcare costs. To apply the time delay method in such contexts, economic models such as Markov state transition models and discrete event (microsimulation) models can be used. These models help researchers forecast economic impacts by combining trial results with data from registries and other sources.16,17 Although discrete event models may offer flexibility in certain aspects, Markov state transition model frameworks are more commonly used in AD. Long-term treatment effects can be estimated in terms of time delay using the Markov state transition model, with the assumption that the effect on time delay for long-term natural history data will be the same as that observed in the clinical trial of interest. In fact, the same results are seen when implementing treatment effects as a time delay as when treatment effects are implemented through the conventional way of reduced transition probabilities to more severe states. Furthermore, the use of time delay data means less dependence on the choice of outcome measure used in the trial of interest and offers a more intuitive interpretation of results. However, there are limitations to the method. The method measures outcomes throughout the disease continuum, including mortality. Therefore, it is important to consider how mortality is handled in the model, as it increases complexity and may reduce the applicability of time delay data when conducting economic evaluations of DMTs.18 A potential limitation discussed during the symposium was the time delay method’s reliance on a surrogate endpoint (CDR-SB). Results of this measure are considered surrogate outcomes of outcomes unobservable in clinical trials (in this case, due to the short length of follow-up periods in relation to disease course). In discussing this limitation, Anja Schiel, a special advisor and lead methodologist in Regulatory and Pharmacoeconomic Statistics at NOMA, explained that the validity of surrogate outcomes has not been fully established, and they only indirectly predict treatment effect. To date, few surrogate outcomes have been shown to be true measures of tangible clinical benefit. Their use to assess the clinical benefit of health technology is neither widely accepted by HTA agencies, nor is it recommended in guidelines.19 Despite this limitation, it is important to recognize that endpoints may have different value for different stakeholders. For example, statisticians and regulators may not advocate the time delay method over traditional endpoint measurement methods, as both involve the same data, no MCID has been established for the time delay method, and the endpoint has not been validated for use in clinical practice or trials. However, physicians and researchers may consider the time delay method to be an additional, potentially more intuitive approach to measure treatment response in early AD, enabling treatment effects to be quantified such that they are interpretable by people living with AD and their care partners.“Endpoints have different values for different decision makers” —Anja Schiel, NOMA CONCLUSIONS The MMRM, an established method for assessing clinical trial results in AD, generates vertical point differences that can be challenging to translate into meaningful and interpretable outcomes for people living with AD and their care partners, particularly during early disease stages. With the advent of DMTs, there is a need to evaluate treatment benefits at a stage where people may display minimal or no symptoms. The time delay method, an innovative progression model for repeated measures, produces results that are easily interpretable, including by those living with the disease. As demonstrated using existing clinical trial readouts, this method can measure the time saved with a DMT versus placebo or natural disease progression. Additional evidence is required to validate this method from both a health economics and HTA perspective.