Conceptual aspect

A basic conceptual issue is the definition of a complex disease like AD. Historically, AD has been regarded as a clinicopathological disorder with a clinical picture of dementia as well as specific AD neuropathology. With the proposed new lexicon and definition of AD by Dubois et al. (2007) and the revised NIA–AA criteria (Jack et al., 2011), there is a distinction of the clinical manifestations from the underlying AD neuropathology. Two essential differences from the criteria of AD by McKhann et al. (1984) are the incorporation of biomarkers and formalization of the different stages of disease in the new diagnostic criteria (Jack et al., 2011). Thus, “asymptomatic at risk state for AD” (Dubois et al., 2007) and “preclinical AD” (Sperling et al., 2011) have become “disease entities.” This has raised a fundamental question on the concept of what is a “disease.” According to the World Health Organization (WHO), a “disease” is associated with malfunction and suffering (Giaccone et al., 2011). Paradoxically, “asymptomatic” and “preclinical” AD, conditions when a person has no symptoms or signs, are now designated as “disease” in these new AD criteria.

The new criteria of AD rely heavily on biomarkers of amyloidosis and neurodegeneration. It is hypothesized that amyloid biomarkers may be abnormal 10–20 years before the onset of symptoms, while abnormalities in biomarkers of neurodegeneration occur later (Jack et al., 2011). In reality, we do not know with certainty that people with AD neuropathology will develop cognitive impairment and dementia in the future. Despite growing evidence that a positive amyloid PET scan with Pittsburgh compound B or low CSF A Beta 42 predicts cognitive decline or dementia, the relationship for progression of mild cognitive impairment (MCI) to dementia and for normal to MCI is probabilistic rather than deterministic (Blacker, 2010). We have to be mindful of the pitfalls of labeling at-risk status defined by biomarkers as disease, since many factors might influence the development of cognitive impairment or dementia. In particular, compensatory mechanisms and cognitive reserve should be given due consideration. For instance, a recent study on 249 subjects found that education had a positive effect on late-life cognitive ability, and it counteracts the deleterious effects of cerebrovascular disease and AD as assessed by neuroimaging biomarkers (Murray et al., 2011).

In addition, a major challenge in the new framework of the diagnosis of AD is the multiple brain pathologies frequently encountered in dementia. Around 40% of patients with AD have significant cerebrovascular disease, and around 90% of patients with dementia with Lewy bodies (DLB) have concurrent AD pathology (Ballard et al., 2011). This is particularly the case in very old age where multiple pathologies are frequent (Wharton et al., 2011); indeed, there is a decrease in the correlation between AD pathology and clinical symptoms with older age (Wharton et al., 2011). Further complicating the picture is that about 20%–30% of people with the neuropathology of AD at postmortem did not have dementia in life (Neuropathology Group of MRC CFAS, 2001).

AD is a heterogeneous condition, and subtypes may be better defined by clinical phenotypes rather than biomarkers. None of the existing biomarkers has been shown to have good criterion validity. For instance, Oksengard et al. (2010) tested the Dubois 2007 criteria and found that they lacked accuracy for predicting AD. The Dubois criteria were valid in only 55% of patients clinically diagnosed as AD. In addition, only 7.3% of prodromal AD subjects were identified from a group of 96 patients with subjective cognitive impairment and/or MCI in this study.

Technical aspect

Biomarkers are biological parameters that reflect disease-related pathophysiological processes. An ideal biomarker should be (1) sensitive, specific, reliable, reproducible, and inexpensive; (2) able to detect a fundamental feature of the disease and reflect some basic pathophysiological processes; and (3) obtained through simple non-invasive procedures (Humpel and Marksteiner, 2009). In addition, it has been recommended that an ideal biomarker of AD should indicate the actual presence of AD, not merely increased risk, and it should change with disease severity (Klunk, 1998). Unfortunately, none of the existing biomarkers for AD satisfies all these criteria.

Before the adoption of biomarkers in clinical use, standardized procedures for biomarker assessment need to be developed. It is noteworthy that all biomarkers are continuous measures and hence there should be agreement on the interpretation of findings and optimal cut-off points. Currently, most AD biomarkers suffer from poor standardization and reproducibility across centers (Frisoni et al., 2011). In addition, the duration of follow-up of abnormal biomarkers in the literature is insufficient for us to know how accurately they will predict the progression of preclinical dementia to MCI or dementia.

In the case of neuroimaging, there is variability between machines and even longitudinal variability within machines. Hippocampal atrophy is not linear and, hence, discrimination between normal subjects and MCI subjects may be poor. Normal hippocampal volume can vary a lot and this has prevented universally accepted norms (Frisoni et al., 2011). Volumetric change is only seen after excessive neuronal loss and imaging biomarkers may give opposite results in volumetric and tracer studies. Information provided by biomarkers may be conflicting and do not fit in the theoretical framework of dynamic biomarkers, which suggests that biomarkers of A β-amyloid are indicative of initiating events while biomarkers of neurodegeneration become dynamic later (Jack et al., 2011).

Turning to CSF, there is a similar variability in the collection methods, standardization between laboratories, stability of chemicals, and platforms for analysis. Hort et al. (2010a) reported on the use of CSF biomarkers in the diagnosis of dementia across 23 European countries, Israel, and Georgia, and found that each laboratory had its own norms in half of the countries.

In addition, CSF changes are not specific. Aβ is decreased, whereas total tau (T-tau) is increased in AD, as well as other types of dementia, such as vascular dementia, Creutzfeldt–Jakob disease, frontotemporal dementia, and DLB. Phosphorylated tau (p-tau) is increased in AD and DLB, although different epitopes of p-tau may be more specific, including p-tau 231 and p-tau 181. Furthermore, the diagnostic performance of CSF biomarkers decreases with age (Mattsson et al., 2012).

Albeit the risk is small but lumbar puncture is an invasive procedure with potential side effects. Blood biomarkers are non-invasive and would be more acceptable to subjects. However, the currently available blood biomarkers are not yet ready for use in the diagnosis of AD. For example, plasma levels for β-amyloid (1–42) are unstable and influenced by various drugs. In addition, it is not sensitive and specific for early diagnosis of AD (Humpel and Marksteiner, 2009).

Definitional aspect

The new criteria of AD appear to offer a simple and straightforward way of making a diagnosis with the use of biomarkers in addition to the clinical picture, and this is very appealing to clinicians. However, there are practical difficulties in operationalizing the criteria.

In the case of MCI due to AD (Albert et al., 2011), the proposed criteria for MCI are as follows: (1) Concern regarding a change in cognition, (2) impairment in one or more cognitive domains, (3) preservation of independence in functional abilities, and (4) not demented. The authors stated that cognitive test scores in people with MCI are typically 1–1.5 standard deviations (SD) below the mean for their age- and education-matched peers. On a practical level, it is unclear whether clinicians should use 1 SD or 1.5 SD, and the number of cognitive tests to be used per cognitive domain. It is important to recognize the wide cultural variation in the applicability of cognitive testing. Assessment of cognitive function is influenced by cultural factors and literacy levels. Many older adults in developing countries have low literacy levels and low acceptance of formal cognitive testing (Chiu and Lam, 2007). Another limitation is that norms are not well established for the oldest-old population even in developed countries.

Ultimately, the decision of making a diagnosis of MCI rests on “clinical judgment.” While this is acceptable practice for experts and experienced clinicians, these criteria may result in a massive over-diagnosis in the field.

Utilitarian aspect

Despite the exponential studies on the use of biomarkers, the advances appear to be associated with only incremental gain on the accuracy of diagnosis above a good history, clinical examination, and cognitive testing. Interestingly, a recent study on the discriminative utility of different biomarkers and cognitive markers examined their ability to predict conversion from MCI to AD. The authors concluded that cognitive markers were more robust predictors of conversion than most biomarkers (Gomar et al., 2011). Schmand et al. (2011) reported that neuropsychological tests and structural MRI scans were most accurate at identifying AD and that CSF and fluorodeoxyglucose (FDG) PET were less accurate and did not significantly add to accuracy of neuropsychological tests or MRI. This was especially so for those older than 75 years, where presumably multiple pathologies were operating. This study did not include amyloid imaging.

An important result of early diagnosis should be early treatment. However, at present there is no effective disease-modifying treatment. Although a very early diagnosis of AD may facilitate advance planning for the future of the individual, the predictive values of the new criteria of preclinical AD and MCI due to AD are uncertain and there may be misdiagnosis. These factors seriously undermine the clinical utility of very early or preclinical diagnosis of AD. One might argue that lifestyle strategies including exercise, mental activities, and social activities as well as legal planning should be implemented for people who are in the preclinical or MCI stage to prevent dementia. However, these lifestyle preventative strategies and contingency planning could be and should be encouraged universally anyway, for brain health, as well as physical and mental health.

Social and cultural aspects

Much of the development and discussion on the use of biomarkers in the diagnosis of AD are by experts in developed countries, and the views and perspectives from researchers and clinicians in developing countries are missing.

According to a report of Alzheimer's Disease International (ADI, 2009), 57.7% of people with dementia lived in low- and middle-income countries in 2010, rising to 70.5% in 2050. With the majority of people suffering from dementia residing in developing countries, it is essential to consider the perspectives of people from these countries. Three factors regarding the use of biomarkers are important globally, but are particularly pertinent in developing countries: availability, affordability, and acceptability.

Even in developed countries, there is a lack of availability of the high-tech investigations, including PET scans, MRI scans, and CSF biomarker essay, outside tertiary centers or research centers. In developing countries, these high-tech investigations are only available in a few specialized centers in big cities. For instance, according to the information from the Organization for Economic Co-operation and Development Health Data 2011 (OECD, 2011), there are 97.3 computed tomography scanners per 1 million population, and 43.1 MRI scanners per 1 million population in Japan, compared with 4.3 and 1.9 per million, respectively, in Mexico.

Apart from the lack of high-tech investigations, there is a paucity of specialists as well as infrastructure for doing the biomarkers. For instance, there is up to a 200-fold difference in the number of psychiatrists and up to a 100-fold difference in the number of neurologists in low-income countries compared with high-income countries (WHO, 2004; 2005).

Affordability of these biomarkers also varies significantly across countries. To quote an example, the gross national income per capita in current US$ was $47,350 in the United States compared with $1,260 in India in 2010 (The World Bank, 2012). Considering the low income of people living in developing countries, many of the biomarkers are hardly affordable, and the high costs of the investigations are prohibitive to their use.

Cultural acceptability is another factor to consider. In particular, while the use of CSF biomarkers is considered to be important in the diagnostic workup of dementia in many European countries (Hort et al., 2010a), lumbar puncture and the use of CSF biomarkers are not routine in the diagnosis of dementia for research or clinical purpose in many Asian countries (Chiu and Lam, 2007).

In Chinese communities, CSF is regarded as precious fluid, and there is a great reluctance in the general population to do a lumbar puncture unless the condition is very serious and potentially lethal. In addition, there are many myths about lumbar puncture prevalent in the lay community, including the belief that lumbar puncture is harmful to health and is very painful.

In developing countries, the priority in terms of dementia care is in raising awareness of dementia, as well as to enhance primary health care and other community services (ADI, 2009). It is difficult to conceive how the stretched resources in these countries should be diverted from care of people with dementia to expensive investigations. In the scenario in which biomarkers are incorporated in the diagnostic criteria of AD, this may end up with a much lower prevalence of AD in developing countries, as many sufferers will not undergo these investigations.

Even for developed countries, resources are not unlimited. If clinicians order biomarkers for all subjects with memory complaints, it would result in a dramatic rise in health expenditure, and possibly a marked increase in the diagnosis of AD. We must bear in mind that at present there is no effective treatment to modify disease progression, and there is no study that addresses the cost–benefit ratio of early diagnosis of AD. Therefore, it is uncertain if the additional costs of these biomarkers may be offset by the savings in the early detection of people at risk of AD.

Ethical aspect

There are major ethical concerns in the use of biomarkers in very early diagnosis of AD, given the uncertain specificity and sensitivity of these tests in the new criteria for the diagnosis of AD in the preclinical and MCI stages. What are the implications of a pre-symptomatic or preclinical diagnosis in the absence of disease-modifying treatments? What will be the consequences for insurance, work, driving, and family relationships? Will there be prejudice against the person in job opportunities and insurance? Will the person be stigmatized by his or her family and society?

Even if biomarkers can deliver earlier diagnosis, the advantages of knowing one's diagnosis earlier are imputed and not empirically proved. There are very few studies on the effects on subjects after disclosure of early diagnosis of dementia, but catastrophic reactions are a definite risk. In addition, there is a possible increased risk of suicide and euthanasia after disclosure of early diagnosis (Draper et al., 2010). In the absence of effective disease-modifying treatments, early diagnosis may do more harm than good to the subject, the family, and the society. The diagnosis of a person as preclinical AD is particularly problematic as it might lead to personal suffering and anguish for many years, long before any manifestation of dementia. In case of misdiagnosis, the consequences can be very serious for the person and frequently the family as well.

In an interesting study, Boustani et al. (2005) carried out a screening study of primary care patients aged 65 or older based on cognitive battery. Only half of the screen positive subjects consented to further diagnostic assessment. The authors speculated that fear of stigmatization, discrimination, loss of independence, as well as a perception that dementia is a severe untreatable disease had led to the low acceptance rate of early diagnosis.

As mentioned earlier, lifestyle preventative strategies could be implemented without early diagnosis of AD. Therefore, on balance, it appears that very early diagnosis of AD with the new criteria may not be associated with improved treatment or quality of life in the subjects with AD pathology. Another pertinent area is that we have to be cautious that the drive to diagnose more people with AD pathology is not fuelled by market considerations for treatment development or use of diagnostic procedures.