Cerebrospinal Fluid Biomarkers in Alzheimer’s Disease: An Invaluable Tool for Clinical Diagnosis and Trial Enrichment

INTRODUCTION

Alzheimer’s disease (AD) is the most common neurodegenerative disorder, affecting around 35 million people worldwide. Currently, available treatments for this disorder aim to reduce symptoms, and do not have detectable effects on disease progression. Evidence indicates that there is a long preclinical and prodromal phase before the full-blown syndrome appears.

Therefore, ideal disease-modifying pharmacological treatments should be administered as early as possible, that is, before the neurodegenerative process becomes too severe and widespread [1]. This is one of the most compelling reasons for searching an early AD signature based on cerebrospinal fluid (CSF) biomarkers. However, even if disease-modifying therapies are lacking, the advantages of an accurate diagnosis justify the use of advanced diagnostic technology [2]. In fact, an accurate early diagnosis of AD before the onset of dementia is vital to ensure that patients receive timely and appropriate personalized care, including counseling and planning, avoiding the use of inappropriate medications and ancillary investigations. Furthermore, it allows the eventual implementation of appropriate steps to prevent unsafe behaviors also allowing the patients to decide and manage legal issues, being still competent. Besides its utility in clinical practice, the availability of a CSF based toolbox for the early diagnosis of AD is pivotal to the development of better strategies for patient recruitment in research studies and clinical trials.

CSF BIOMARKERS IN ROUTINE DIAGNOSTIC WORK-UP

The core AD CSF biomarkers include total tau (T-tau), phosphorylated tau (P-tau), and the 42 amino acid peptide (Aβ₄₂) of amyloid-β. These proteins reflect key pathogenic aspects of the disease, i.e., neuronal and axonal degeneration, phosphorylation of tau with tangle formation, and aggregation and deposition of the Aβ₄₂ peptide into plaques [3].

After being validated in several studies and meta-analyses [4], CSF biomarkers entered the AD diagnostic criteria. In the International Working Group IWG-2 criteria [5], CSF biomarkers have a pivotal role for the diagnosis of prodromal AD, together with amyloid PET, because of their high diagnostic performance.

The National Institute on Aging– Alzheimer’s Association (NIA-AA) criteria for mild cognitive impairment (MCI) due to AD [6] and dementia due to AD [7] allow for the assessment of the likelihood of being correctly diagnosed with both amyloid and (neuronal) injury biomarker, with positive cases having the highest likelihood. Although the two criteria sets are based upon different approaches and terminology, most patients who meet the IWG-2 criteria will also meet the NIA-AA criteria and vice versa [8].

The NIA-AA criteria for MCI due to AD consider both amnestic and non-amnestic MCI as possible prodromal stages of AD-type dementia. Several studies showed that CSF biomarkers predicted accurately AD dementia in subject with amnestic MCI [9–11].

In recent years, several studies have pointed out the utility of including Aβ₄₀, the most abundant variant of Aβ isoforms, in the CSF signature of AD. Even if CSF Aβ₄₀ is relatively unchanged in AD, the CSF Aβ₄₂:Aβ₄₀ ratio has been suggested to have stronger diagnostic accuracy for AD when compared with CSF Aβ₄₂ alone [12, 13]. Accordingly, many reports show that the CSF Aβ₄₂:Aβ₄₀ ratio is more closely related to what is observed with PET amyloid imaging [14]. Furthermore, a recent finding also suggests that the use of the CSF Aβ₄₂:Aβ₄₀ ratio rather than CSF Aβ₄₂ alone could contribute to reduce inter-laboratory variation in Aβ values, thus favoring the general use of CSF AD biomarkers in routine practice [15].

A recent paper outlined a strategic roadmap for bridging the gaps until an early AD diagnosis based on biomarkers (i.e., CSF or imaging) will be completely integrated in clinical practice [16]. The roadmap is built upon the development and use of biomarkers for screening and delivery of personalized care in oncology, a discipline offering a unique perspective, due to the most advanced stages of implementation of biomarkers for prevention, diagnosis, and treatment. As in the framework developed for oncological patients in 2001 by Pepe and colleagues [17], the roadmap includes five phases characterized by one or two primary aims, as well as several secondary objectives. According to this approach, we can see that CSF biomarkers for AD are at an advanced stage of development [18].

A major drawback for CSF biomarkers is that the measurements obtained with currently available manual immunoassays are sufficiently stable and comparable only when used in experienced laboratories with well-established quality control procedures. Important recent advancements are represented by automated assays [19, 20], which in the near future will significantly increase precision by minimizing operator errors, potentially allowing a greater diffusion of CSF analysis also in non-research centers.

Nevertheless, standardized protocols for controlling pre-analytical and analytical factors remain a priority. We need to reassess the cutoff values for all immunoassays by using a suitable reference (preferably neuropathology). Despite large evidence from studies of clinical assay development and observations on retrospective studies using longitudinal data available in repositories, there is still modest evidence in prospective diagnostic accuracy studies and none in disease burden reduction studies.

NEW CSF BIOMARKERS TO EMPOWER THE DIAGNOSTIC PERFORMANCE OF THE AD BIOCHEMICAL SIGNATURE

The CSF Alzheimer signature is represented by increased total tau (T-tau) and phosphorylated tau (P-tau), together with reduced Aβ₄₂ and Aβ₄₂:Aβ₄₀ ratio. However, several observations in the last years have shown that there is room to improve this well established and in-use toolbox. A recent meta-analysis by Olsson and colleagues [4] found that, within the large number of CSF biomarkers studied so far, neurofilament light protein is also strongly associated with AD. Other molecules not directly reflecting AD core pathology, namely, neuron specific enolase (NSE) [21], a neuron-enriched enzyme of the glycolytic pathway, visinin-like protein 1 (VLP-1), a calcium-sensor protein found in the neuronal cytoplasm [22], heart fatty acid binding protein (HFABP), an intracellular fatty acid transport protein also expressed in neurons [23], and YKL-40 a marker of activated microglia and astrocytes [24, 25], may add accuracy for diagnosing AD. These molecules could be promising candidates as prognostic markers, as well.

THE ROLE OF CSF BIOMARKERS FOR TRIAL ENRICHMENT

Bapineuzumab and solanezumab Phase III trials in mild to moderate AD ended up with negative results. One possible reason for that is the inclusion of participants with unlikely AD pathology [26].

PET sub-studies of bapineuzumab and solanezumab trials classified the patients that were amyloid-negative (Aβ-) based on amyloid PET imaging and demonstrated that more than 20% of patients diagnosed with AD based on clinical criteria were Aβ-, with higher proportions of Aβ- among APOE ɛ4 non-carrier and mild dementia patients [27]. As expected, Aβ- subjects did not demonstrate the same rate of cognitive decline typically observed in AD. These findings, along with other observations, show the basic need of β-amyloidosis markers, either PET amyloid imaging or CSF Aβ levels, for the purpose of trial enrichment.

As reported above, PET amyloid imaging or CSF Aβ levels are used in the new NIA-AA criteria for evidentiating brain β-amyloidosis. Accordingly, many ongoing or planned trials are using these amyloid biomarkers as enrichment to catch prodromal AD cases.

Coric and colleagues [28] reported the results of a randomized, placebo-controlled phase II clinical trial that prospectively enriched a study population with prodromal AD defined by CSF biomarker criteria and MCI symptoms. The study failed to demonstrate clinically meaningful pharmacodynamic effects of avagacestat but met its clinical trial enrichment aims.

CSF biomarkers have been now reported in the inclusion criteria of Phase I, Phase II, and Phase III clinical trials, with enrichment strategy pursued in several manners (Table 1).

In trials on AD populations, several definitions are used to list CSF AD markers in the inclusion criteria: to meet NIA-AA criteria; to have a CSF profile consistent with AD pathology; CSF Aβ₄₂ under a certain cut-off depending of the target population; to lie below/above of Aβ₄₂/tau cut-offs. In trials including MCI patients, participants are required to meet NIA-AA criteria for MCI due to AD.

The use of CSF Aβ₄₂ and tau proteins as inclusion criterion for clinical trials in patients with AD has been endorsed by the European Medicines Agency (EMA). The EMA released two qualification opinions, in April 2011 and February 2012, stating that a pathological signature based on low CSF Aβ₄₂ and high t-tau levels in patients with MCI is useful for identifying those who are at risk of developing AD dementia. In addition, given the high sensitivity and moderate specificity, EMA concluded that the CSF biomarker signature based on a low Aβ₄₂ and a high T-tau is useful for the enrichment of clinical trial populations [29]. The FDA has also released draft guidance on clinical trials in patients in the predementia stage of AD. According to this guidance, FDA supports the concept of enriching trial populations with patients most likely to progress to dementia, using both clinical and biomarker-based criteria. However, the need for an assessment of sensitivity and specificity in identifying patients who do have actual AD in clinical trials, as well as for the validation methodologies (e.g., selection of appropriate cut-points, quantification of assay variability), does not allow FDA to formally endorse CSF biomarkers as definite diagnostic tool, at this time.

To date, the FDA has issued a letter of support to Coalition Against Major Diseases encouraging the further use and study of CSF analytes as exploratory prognostic biomarkers for enrichment in clinical trials targeting the pre-dementia stage of the disease [30].

At present, the use of CSF markers as a screening tool and enrichment criterion is feasible and recommended, due to the availability of recent development of reference standard procedures and materials. As compared to amyloid PET imaging, CSF markers are less costly and have a comparable accuracy. In the near future, we need to establish standard cutoffs to be used as inclusion criteria. In order to implement an effective strategy, we should handle the issue of between-site variability for CSF biomarker measurements. We also need to consider if previously measured CSF biomarkers might be considered as valid for inclusion.

DATA SHARING AND INDIVIDUAL PATIENT DATA META-ANALYSES

There is high interest among researchers in sharing data and protocols. Such an option allows the scientific community to comprehensively reanalyze previously collected data, encourage new interpretations, and promote research collaborations as well as enhanced transparency.

The use of electronic data capture methods consistently simplifies the task of data collection and has the potential to standardize many aspects of data sharing.

A trend toward increased sharing of neuroimaging data has emerged in recent years [31], and the CSF markers research field should follow the same path. Besides clinical trials, as a methodological approach in clinical research setting, data harmonization according to international standard formats should be constantly applied in order to make these data available to the scientific community, as in the ADNI experience (http://adni.loni.usc.edu).

Another aspect related to the sharing of raw-data is the conduct of individual patient data (IPD) meta-analysis, which is the gold standard for summing-up evidence. Despite the increasing availability of studies addressing many clinical issues, an intensive use of IPD meta-analyses is lacking. The IPD meta-analysis of Jansen and colleagues [32] provided interesting results suggesting a 20- to 30-year interval between the first sign of amyloid positivity and the onset of dementia.

CONCLUSIONS

CSF biomarkers entered the diagnostic criteria for AD. However, there are some steps to take in order to fully implement the use of CSF biomarkers in the AD routine diagnostic work-up and as strategy for enriching trial populations. We need to increase the general awareness of the importance of early diagnosis, the collaboration within the scientific community by promoting data sharing practices, and to encourage IPD meta-analyses.

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/17-9910).