FormalPara Key Summary Points

The Centers for Medicare and Medicaid Services (CMS) national coverage determination (NCD) for anti-amyloid monoclonal antibodies (mAbs) under coverage with evidence development (CED) limits the coverage and reimbursement for all anti-amyloid therapies in Alzheimer’s disease (AD), regardless of approval status, to only patients who participate in CMS or National Institutes of Health (NIH)-approved clinical trials, prospective comparative studies, or registries under the CED.

However, recent reports raised concerns about the high rate of failure and operational challenges under CEDs in general, and specific discriminatory consequences of the current CED targeting anti-amyloid therapies for AD.

This commentary discusses clinical and scientific challenges to the effectiveness studies required by the current CED, including absence of a generally accepted approach to study design, lack of uniformity in clinical decision-making, inconsistency in care delivery, delay in patient detection and diagnosis, patient heterogeneity, deficiency in understanding the patient journey, increased operational burdens with decreased clinician participation, and a need for standardized clinical practice procedures and outcome assessments.

These challenges highlight the need for alternative approaches to the collection of community-level evidence in support of patient access to innovative medicine and advanced healthcare technologies.

Commentary

In April 2022, the Centers for Medicare and Medicaid Services (CMS) issued a Medicare national coverage determination (NCD) for Food and Drug Administration (FDA)-approved anti-amyloid monoclonal antibodies (mAbs) under coverage with evidence development (CED) [1]. This decision appears to ensure that coverage and reimbursement for newly FDA-approved or to-be-approved anti-amyloid therapies in Alzheimer’s disease (AD) are only granted to patients who participate in CMS or National Institutes of Health (NIH)-approved clinical trials, prospective comparative studies, or registries under the CED.

Recent reports have raised concerns about a high rate of failure to deliver purported results (i.e., the majority of NCD under CED have extended beyond 5 years) [2] and clinical practice-related operational challenges [3] associated with CMS issued CED requirements in general, as well as specific discriminatory consequences [4] of the current CED targeting anti-amyloid therapies for AD. We wish to discuss clinical and scientific challenges to the effectiveness studies required by the current CED. This article relies on existing evidence and expert opinion, and does not present original research conducted by the authors involving human participants or animals.

CMS stipulates that CED studies for anti-amyloid mAbs must address all the following questions: (1) Does the anti-amyloid mAb meaningfully improve health outcomes (i.e., slow the decline of cognition and function) for patients in broad community practice? (2) Do benefits and harms such as brain hemorrhage and edema, associated with the use of the anti-amyloid mAb, depend on characteristics of patients, treating clinicians, and settings? (3) How do the benefits and harms change over time? [1].

Most CED studies have not achieved the objective of demonstrating effectiveness in clinical practice settings [2]. Of the 27 therapies or diagnostics that have been subject to CED requirements since the program’s inception, the majority (77.8%; 21/27) are still incomplete after years of investigational efforts; among these CED studies, 90% (19/21) have continued for 5 years or more, including four that have been ongoing for over 15 years [2]. We believe that issues revealed in the literature and clinical observations from the Veterans Affairs (VA) healthcare system [5] may highlight specific challenges to the CED-required effectiveness studies, necessitating onerous scientific efforts by patients, caregivers, clinical investigators, and other stakeholders for little return.

The first challenge is settling on an appropriate study design that would address the inquiries of treatment effectiveness and sustained clinical benefits over time. For example, a case–control design would allow prospective comparisons, as required by the CED; however, assigning some patients to the control arm apparently presents ethical challenges (e.g., delayed access by patients or patient subgroups to an efficacious and safe therapy if approved as such by the regulatory agency), equity challenges (e.g., limited patient access to the studies due to geographic locations), as well as clinical/scientific challenges (e.g., confounding by indication, as would occur in naturally selected treatment vs non-treatment groups). While crossover design could help address confounding by indication, this design is highly challenging to operationalize in the practice setting with this disease indication, would substantially incur increased effort and cost, and would not address ethical and equity issues: (1) it is difficult to determine the proper time window for crossover as the treatment duration and effect time window may vary across patients, (2) it may require patient randomization at study outset that potentially disrupts normal patient flow at study sites hence altering the nature of community practice and rendering the study outcomes less generalizable to practice centers outside of the study, and (3) treatment would be delayed for patients initially assigned to the control arm, resulting in continued worsening of disease and health conditions. A randomized pragmatic clinical trial, in general, would be impractical for demonstrating therapeutic benefits with time-sensitive patient demands and longer treatment durations, because patients randomly assigned to the control arm would be motivated to immediately pursue switching to the active treatment arm. In order to preserve true characteristics of real-world clinical practice, therapeutic switches consistent with treatment guidelines and practice routines should not be prevented for the sake of research validity at the expense of patient benefits, which inevitably would lead to disruption of randomized effectiveness comparisons.

Practice inconsistency in case assignment is a major threat to effectiveness evaluation under the CED. In the VA healthcare system, close to 50% discordance was found between clinicians’ subjective assessments of AD severity and severity classification using Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA) test scores [5]. A clinician’s judgment typically considers psychiatric, behavioral, and functional factors as well as comorbidities beyond cognitive testing results. Notably, the low concordance was not only evident across the mild to severe spectrum of AD stages but also varied between specialty and general practice settings [5], making it even more difficult to collect meaningful clinical data on comparable assessment scales, ascertain patient eligibility for treatment, standardize follow-up visits, or conduct outcomes evaluation.

While standardization of clinical practice procedures and outcome measures may help address inconsistencies, clinician participation may be discouraged because of resource-intensive study requirements leading to costly disruption of practice routines [3]. Since the 2011 publication of research guidelines for mild cognitive impairment (MCI) and AD [6], there has not yet been an update accounting for the newer generation of therapeutics including anti-amyloid treatments. Therefore, it is not realistic to expect that real-world clinical practice will become more standardized in the near-term to enable consistent evaluations of effectiveness in CED studies.

Patient heterogeneity in AD is another challenge for CED studies. For example, in the VA healthcare system, comorbid mental health conditions are observed to be more common among patients with unspecified dementia relative to those with a documented clinical diagnosis of AD, supporting a hypothesis that mental health conditions may potentially mask or delay AD diagnosis. This is compounded by the understanding that certain mental health disorders may disproportionately affect disadvantaged patient subpopulations [7], leading to potential delay in AD diagnosis for those patient groups and hence inequitable access to appropriate treatment. In addition, some patients with AD lack adequate social support or informal care, resulting in faster disease progression and missed opportunities for participation in a CED study that is currently required for treatment coverage and reimbursement by CMS.

Practice inconsistency and patient heterogeneity can result in patients at different points along the disease trajectory within early AD being enrolled to or rejected from participating in a CED study targeted for MCI due to AD or mild AD. Enrolling a heterogenous patient group for study without random treatment assignment or stratification would make interpretation of comparative effectiveness outcomes difficult and less meaningful. Even if the same assessment tools were used for all patients, challenges would remain. For example, when the Clinical Dementia Rating Sum of Boxes (CDR-SB) is used, patients with scores ranging between 0.5 and 9.0 and confirmed amyloid status might be eligible for an FDA-approved anti-amyloid therapy. Differences among patients with the CDR-SB score range could be appropriately addressed in a randomized, well-controlled clinical trial; however, in clinical practice or in a randomized pragmatic trial, patient trajectories of disease progression might differ substantially because of patient heterogeneity and localized or individualized treatment protocols, clinical visits, and care processes in the community practice setting. Additionally, the CDR-SB is generally not used in routine patient care; introduction of the instrument to patient assessments is likely to induce additional clinical heterogeneity. Consequently, therapeutic outcomes in a pragmatic comparative study would likely differ as a result of factors beyond treatment effectiveness.

The potential complexity of implementing the current CED would inevitably lead to delayed access to the therapy by a large portion of eligible patients, including those who are traditionally underrepresented (e.g., various racial minority groups) and those with comorbidities (e.g., vascular disease) in order to pursue the rigor of comparative effectiveness studies. Even more regrettable, the current CED is likely to create a discriminatory reality [4] that will be confronted by all patients who have been suffering from this devastating disease that has historically had no other options for treatment beyond cholinesterase inhibitors, an N-methyl-d-aspartate (NMDA) antagonist, and nonpharmaceutical or supportive interventions. None of these treatments addressed underlying AD pathophysiology, delayed disease progression, or restored cognitive function. Reasonable benchmarks and guidance should be provided to ensure equal representation of the US population demographics in therapeutic access for any CED-required studies.

One example of an unexplainably stalled CED having significant impact on the daily practice of neurology is “FDG PET for Dementia and Neurodegenerative Diseases” [8]. Up to this date, the determination remains: “Medicare covers FDG-PET scans for either the differential diagnosis of fronto-temporal dementia (FTD) and Alzheimer’s disease (AD) under specific requirements; OR, its use in a Centers for Medicare & Medicaid Services (CMS)-approved practical clinical trial focused on the utility of FDG-PET in the diagnosis or treatment of dementing neurodegenerative diseases.” Despite 18 years having passed, the studies mandated by the FDG-PET CED remain incomplete, leaving the CED in place and significantly restricting PET services for individuals with neurological disorders and dementia. The prolonged and ambiguous duration of FDG-PET CED creates ongoing uncertainty in the clinical management of AD and consumes significant public resources that could have been allocated to other health technology innovations. This situation impedes evidence development and, ultimately, the improvement of patient outcomes.

Effectiveness studies required under CMS CEDs have been demonstrably unsuccessful [2, 3]. Clinical and scientific challenges in study design and operationalization are amongst the major barriers to successful CED implementation. In AD, these challenges are further aggravated by diverse clinical practices and by lack of advancements in drug development and clinical guidelines [9]. Attempting to address variability by adding standard clinical assessment processes and tools to existing practice will likely impose additional burdens on clinicians and support staff who are already under resource pressure, potentially leading to negative impacts on patient care and increased costs to clinical centers. Collectively, these issues represent substantial threats to the feasibility and validity of CED studies that require demonstration of effectiveness in the community practice setting. Consequently, the healthcare system readiness for adoption of a potential new generation of AD treatments such as anti-amyloid therapy is at risk.

In general, prospective comparative studies in community practice settings are not meant to be confirmatory for clinical benefits demonstrated by well-controlled, randomized clinical trials, but rather to be hypothesis-generating in nature. Ultimately, these studies may answer some of the sought-after questions for meaningful improvement in the community-based patient population but will likely do little to lessen the required burden of proof. CED-required studies are unlikely to achieve the scientific rigor conferred by phase 3 pivotal clinical trials conducted through drug development programs. Rather than aiming to confirm effectiveness, CED studies could instead be conceived to identify areas where clinical effectiveness may be confounded, masked, or attenuated in community practice settings, to detect additional safety signals, and to understand equitable patient benefits, thus serving to complement pivotal trial findings.