Globalization of Alzheimer's disease clinical trials

Alzheimer's disease (AD) therapies are increasingly being tested in global clinical trials. A search of ClincalTrials.gov revealed that of 269 currently active trials, 28% are currently being conducted in the United States; the majority of trials and the majority of trial sites are ex-US. The US has the largest number of trial sites of any single country; cumulatively, nearly half of all sites are outside the US. The US conducts more trials in all phases of drug development but has a greater proportion of phase 3 trials. The increasing importance of global participants in clinical trials emphasizes the importance of considering the ethnic and international factors that may influence trial outcome. The International Conference on Harmonization guidelines divide ethnic factors that may affect drug development into intrinsic and extrinsic influences. These include language, cultural factors, educational levels, the general level of health and standard of care, as well as nutrition and diet. Ethnic influences on pharmacokinetics are known for some metabolic pathways. The biology of AD may also differ among the world's populations. The frequency of the apolipoprotein e4 allele, a major risk factor for AD, differs internationally. Genetic variations might also affect inflammatory, excitotoxic, and oxidative components of AD. Diagnostic standards and experience vary from country to country. Levels of practitioner training and experience, diagnostic approaches to AD, and attitudes regarding aging and AD may differ. Experience and sophistication with regard to clinical trial conduct also vary within and between countries. Experience with conducting the necessary examinations, as well as the linguistic and cultural validity of instrument translations, may affect trial outcomes. Operational and regulatory aspects of clinical trials vary and provide important barriers to seamless conduct of multiregional clinical trials. Collection and testing of biological samples, continuous provision of drug substance, and protection of the integrity of supply lines may be difficult in some international circumstances. Attention to these potential influences on clinical trials will determine the success of global drug development programs and the utility of global trials for developing new AD therapeutics.

Alzheimer's disease (AD) is a growing global epidemic crisis. Currently, 35.6 million people are living with dementia worldwide; the number is expected to increase to 65.7 million in 2030 and 115.4 million in 2050 [1]. Th ese staggering numbers require the development of meaningful interventions that will delay the onset, slow the progression or improve the symptoms of patients suff ering from this progressive neurodegenerative disorder. Many promising agents show effi cacy in preclinical models of AD and must be tested for effi cacy in double-blind, placebo controlled clinical trials.
More AD clinical trials are conducted in the United States (US) than in any other single country, but most trials now have a signifi cant number of international sites and there are numerous compelling reasons to consider incorporating international sites in AD trials. Most clinical trials fail to recruit the necessary numbers of patients within the anticipated timeframe [2]. Recruitment must extend globally in order to facilitate patient recruitment and enable trials to be completed in acceptable periods. Th e presence of large numbers of patients, many of whom are treatment naïve, enhances the attractive ness of international clinical trial sites. In addition, trials conducted outside of the US are typically less expensive, providing another motivation for ex-US drug development. Th ere are an increasing number of well trained scientists internationally, and laws and regulatory provisions make using foreign sites an attractive alternative [3]. Finally, most patients with dementia live in the developing world [4] and the growing size of international pharmaceutical markets [5] attracts sponsors to conduct studies in countries where pharma ceutical sales may be pursued. With the growth of globalization of clinical trials, there are now more studies conducted collectively outside of the US than within the US [6]. Globalization of trials is an increasing trend and will continue into the foreseeable future [7].
Th ere are many challenges to reliable drug development in global settings and some of the recent failures in AD drug development may be attributable in part to inadequate attention to these issues [8]. Here we discuss the aspects of drug development to be considered when implementing global clinical trials for AD therapies.

Current global distribution of Alzheimer's disease trials
A search of the ClinicalTrials.gov registry using the term ' Alzheimer's' , was refi ned to 'trials funded by industry' . Th is search yielded 475 studies as of 3 (11) are for evolving diagnostic devices or procedures.
Of the 269 studies meeting the search criteria, 76 (28%) are being conducted in the US whereas a majority of trials are occurring outside the US (Table 1). Western Europe is host to 43 studies (16%); other countries each had fewer trails but collectively accounted for a majority of trials. Forty-nine (18% of all trials; 64% of all US trials) are being conducted only in the US; 27 (10% of all trials; 35% of US trials) are multinational. Th e US has the largest number of trial sites (1,937 total for all sites in all trials; many sites participate in several trials and the total includes these sites for every trial), contributing 48% of all active trial sites for all trials. Europe had the second largest number of active trial sites (921; 23%). As in the total number of trials, there were more trial sites (52%) outside the US than within.
Th e global distribution of trials by phase is shown in Table 2. ClinicalTrials.gov lists 25 phase 1, 49 phase 2, 30 phase 3, and 15 phase 4 AD trials. Phase 1 trials may be under-represented in the data base since some foreign phase 1 trials are not conducted under an investigational new drug application [7]. Of the phase 1 trials listed, most are conducted in the US and Western Europe (20 of 25 are conducted in these two regions). Phase 2 and phase 3 trials are conducted worldwide; most countries except Russia have approximately equal numbers of phase 2 and 3 trials. Russia has a predominance of phase 2 AD trials (7 of 11 trials conducted in Russia). Th e US is involved in 52% of the listed phase 1 trials, 59% of phase 2 trials, and 73% of phase 3 trials.

Aspects to be considered in globalized Alzheimer's disease trials
Th e International Conference for Harmonization (ICH) guideline E5 [9] provides a framework for discussing ethnic infl uences in clinical trials relevant to globalized drug development programs. Th e guidelines divide ethnic infl uences into intrinsic and extrinsic factors. Table 3 provides a summary of major dimensions of drug develop ment to be considered when planning global AD clinical trials.

Host factors that may aff ect disease progression
Th e Human Development Index, which consists of life expectancy, education and per capita gross national income (as an indicator of standard of living), provides insight into diff erences in patient conditions throughout the world. Human Development Index scores vary from 0 (very poor) to 1 (very good). North America and Australia generally have values of 0.9 and above; Russia 0.7 to 0.749; China 0.65 to 0.69; India 0.55 to 0.59; parts of Africa may be as low as under 0.3 [10]. Studies of specifi c health indices across countries also provide examples of international variability. For example, a comparison of cardiovascular health of individuals included in the US Trial sites typically participate in more than one trial and may be represented several times in the 'Total number of locations by region' column.
National Health and Nutrition Examination Survey with the health measures of those included in the Nutrition and Health Survey in Taiwan revealed diff er ences between the two ethnic groups in left ventri cular hypertrophy, serum lipids, obesity and smoking status [11]. Such international diff erences could translate into diff erences in cerebro vascular co-morbidity, cogni tive reserve, and adverse event vulnerability. Health status measures could be incor porated in trial analyses to address variability introduced by international health disparities. Education levels also vary substantially from country to country. Th e average education level in the US is 12 years; Japan has an average education level of 9.5 years; United Kingdom 9.4; France 7.9; Italy 7.2; Th ailand 6.5; China 6.4; India 5.1; and Brazil 4.9 [12]. Many countries included in global clinical trials have less than half the educational level of US trial participants. In a validity study of the Chinese version of the Alzheimer's Disease Assessment Scale -Cognitive portion (ADAS-cog), for instance, 35% of the trial subjects had fi ve or fewer years of education [13]. Th is contrasts with the US, where participants have an average of 14 years of education in recent trials [14]. Individuals with lower levels of education tend to progress more slowly in the course of AD [15]. Persons with higher levels of education progress more rapidly, have higher levels of neuritic plaques at autopsy, and have more severe abnormalities on fl uorodeoxyglucose positron emission tomography and amyloid imaging than those with lower levels of educational achievement [16][17][18][19]. Placebo groups would decline more slowly and larger number of individuals would be required to show a drug-placebo diff erence in a poorly educated population compared to a well educated popula tion. Eff ects of education on disease progression in diff erent regions have not been investigated, and educational level should be considered as a potentially  impor tant source of variance in disease progression in global trials. Th e level of education in the proposed population would need to be known to model the magnitude of this eff ect for specifi c trials. Exercise levels may also vary among global populations and exercise has been shown to infl uence amyloid-beta protein abnormalities in cognitively normal older persons. Individuals with higher levels of exercise have lower levels of amyloid on amyloid imaging and higher (less abnormal) levels of amyloid levels in the cerebrospinal fl uid [20]. Th is infl uence should be considered when popu lations with markedly diff erent exercise levels are included in global prevention trials of agents targeting beta-amyloid mechanisms.

Population factors that may infl uence drug pharmacokinetics
Diff erences in body size would be expected to contribute to diff erences in brain exposure levels when patients of diff erent sizes are given similar doses of drugs. Th e average height in Sweden, for example, is 1.815 meters, whereas the average height in the Philippines is 1.634 meters, a 20 centimeter diff erence with corresponding anticipated diff erences in drug exposures [21]. In Japan, where body size tends to be smaller than in Western nations, two-thirds of new molecular entities approved between 2003 and 2005 had standard doses that diff ered from those approved in the US and Europe [22].
Obesity likely may aff ect drug metabolism and distribution. Obesity is more common in Western populations. Lipophilic drugs will have disproportionately large volumes of distributions in obese and overweight individuals [23]. Such diff erences may aff ect brain exposures and both effi cacy and adverse event occurrence.
Polymorphisms that may infl uence drug metabolism, such as those governing cytochrome P450 enzymes, may also vary substantially across ethnic groups and could contribute importantly to diff erences in drug metabolism, central nervous system drug exposure, and drug response [24,25]. Th ere are more slow metabolizers in Asian popula tions, possibly leading to higher exposure levels with contrasting effi cacy and side eff ect profi les among participants in global trials. Concomitant approved drugs taken by patients in trials may also aff ect trial results through drug-drug and drugdisease interactions. Diff erent approaches to managing medical conditions prevail globally and average doses in clinical use may vary. Japanese physicians, for example, tend to use small doses of several medications to manage medical illnesses; European doctors tend to gradually titrate single medications upward; and US physicians are more likely to press the dose upward until toxicity occurs and then titrate downward [23]. Th ese practice patterns result in smaller doses of medications in Japan, inter mediate doses in Europe, and higher doses in the US. Th is creates an opportunity for diff ering drug-drug inter actions across regions.
Some compounds may be less sensitive to ethnicrelated infl uences and are better candidates for global clinical trials. Table 4 summarizes the characteristics of compounds less sensitive to ethnic infl uences.
Biological therapeutic agents may be subject to somewhat diff erent global infl uences than small molecule drugs. Asians, for example, have higher rates of exposure to infections and a diff erent complement of circulating antibodies; they are less likely to have been included in national immunization campaigns and therefore have more naïve immunological backgrounds from an iatrogenic perspective [23].

Population factors that may aff ect Alzheimer's disease biology
Genetic diversity also creates diff erences in the biology of AD relevant to clinical trials. Th e apolipoprotein E ε4 allele (APOE ε4) varies from a high frequency of 0.221 in Oceania to a low frequency of 0.068 in India [26]. APOE ε4 carrier status infl uences the rate of progression of both cognitive and functional decline in patients with mild cognitive impairment and prodromal AD [27,28]. Ethnic diff erences in receptor sensitivity and pathway modu lation have been insuffi ciently explored but also could aff ect drug effi cacy in diverse populations. Th is variability across ethnicities and regions could infl uence rate of disease progression and aff ect the sample size needed to observe a drug-placebo diff erence with an effi cacious agent.
Head size and brain size refl ect cognitive reserve and have an important interaction with disease state. Head size refl ects brain size and neuronal number. Smaller head size has been associated with more rapid progression of AD [29]. Similarly, smaller head size is associated with greater cognitive impact of brain atrophy compared to persons with larger heads [30]. Persons with smaller body size have correspondingly smaller head sizes and patients from countries with smaller citizens might progress more rapidly in trials. More comprehensive modeling of this eff ect is needed and collection of empirical data is warranted.
Diet and nutrition might also contribute to diff erences in therapeutic response, side eff ect vulnerability, and disease progression. Populations with high levels of antioxidants in the diet might reduce oxidative injury, while those with high levels of salt or animal fats may be more vulnerable to hypertension, cardiovascular disease and central nervous system co-morbidity.

Approaches to Alzheimer's disease diagnosis that may diff er among countries
Attitudes toward aging dictate the extent to which cognitive decline is considered abnormal and may aff ect rates of referral for evaluation and trial participation. Asian countries place less emphasis on the importance of memory in aging, and AD is less likely to be perceived as an abnormality [4]. In countries where aging and dementia are not distinguished, referral for cognitive assessment is typically later in the disease course when cognitive decline is more severe; consultation may be triggered by behavioral changes rather then cognitive or memory complaints [4,31]. Dementia diagnosis will be delayed in these settings and behavioral disturbances are more common in conjunction with AD diagnoses.
In some countries, the high frequency of hypertension, cardiovascular, and cerebrovascular disease will increase the proportion of patients with mixed dementia syndromes or comorbid central nervous system pathology [32].
Clinical phenomenology may be aff ected by international setting. For example, rates of apathy as measured by the Neuropsychiatric Inventory (NPI) are lower in Asian compared to US populations [33], likely attributable to diff erences in expectations of activity levels and the value placed on serenity in elderly persons.
Clinicians vary widely in the extent to which they use standardized rating scales and collect systematic information on AD in their clinical practices [34]. Diff erences in practice may translate into diff erences in understanding the variability in presentations of AD and their impact on diagnosis.

Aspects of clinical trial instrumentation that may aff ect trial outcomes
Nearly all widely used clinical trial instruments for AD were developed in North America, and they may not be well suited for use in other countries, especially those with diff erent cultural heritages. Using the same instru ments in all sites of a global clinical trial is required for data uniformity and analysis, but the quality of the data collected may be aff ected by diff ering cultural and national infl uences. Elderly in many developing countries have low levels of reading and health literacy, poor testtaking skills, and higher levels of test-related anxiety [4]. If skills improve and anxiety declines with multiple test sessions, both the active treatment group and the placebo group will exhibit substantial improvement, possibly obscuring a drug-placebo diff erence.
Th ere have been numerous local translations of clinical trial instruments, often with little attempt at harmonization or consideration of the impact of the diff erences on trial outcomes. In the Hindi version of the Mini-Mental Status Examination, for example, 16 of the items were changed to make it culturally appropriate [4]. Even within the US, substantial administration and scoring variances have been documented for the ADAS-cog [35] and the NPI [36]. Th ese diff erences are exaggerated in global trials.
Normative data for trial instruments are often lacking in cultures where clinical trials are conducted and cutoff scores for study entry as well as expected performances for healthy elderly persons may be diffi cult to extrapolate from Western experience. Th is can result in entry of noncompromised patients into prodromal AD and AD dementia trials because of culturally and educationally appropriate low scores.
Trial data for the Clinical Dementia Rating (CDR), Activities of Daily Living Scale, and the NPI are collected from caregivers. Th e role of the care partner may be more straightforward in Western countries than in many developing nations. Patients tend to live at home in extended families with multiple caregivers. Many families, even those of modest fi nancial means, employ domestic maids who may speak diff erent dialects or even diff erent languages from the patients [4]. Th ese indivi duals are responsible for the majority of the daily care of the aff ected person. How best to collect information under these circumstances and how best to defi ne 'caregiver' can be challenging. Many indigenous persons have extremely simple life styles unencumbered by utilities and conveniences; the usual trial instruments will not eff ectively quantify functional changes in these circumstances.
Biological measures may also be diff erentially acceptable across international trial sites. Asian patients are reported to be more reluctant to undergo invasive procedures and less likely to participate in trials including lumbar puncture [4].

Clinical trial conduct practices that may vary among countries
Clinicians and trial sites vary widely with regard to their experience in conducting AD clinical trials. In a review of their proprietary database of 7,000 raters, one ratertraining company found that 57% of raters had more than 5 years' experience in administering the ADAS-cog and 27% had less than 1 year's experience [37]. Among raters in China included in a recent trial, 40% of raters had administered the ADAS-cog fewer than fi ve times. Lack of experience can lead to greater score variability and more diffi culty demonstrating a drug-placebo diff erence of an eff ective compound.
Clinical trial conduct includes training the sites on how to administer the assessment instruments and monitoring sites for appropriateness of trial data collection. Many investigator meetings focus on proper scoring of instruments based on videotaped interviews by experts. Investi gators and research assistants may have limited opportunity to administer the examinations under supervised circumstances. Reliability studies of the Clinical Dementia Rating show that at least fi ve administrations of the tool at each severity level are required to achieve inter-rater reliability [38]. Trial investigators are not trained to this level of reliability, and data collected exhibit large standard deviations and confi dence intervals.
Studies have also shown substantial variability in trialto-trial instructions for both the ADAS-cog [35] and the NPI [36]. Th is variability causes uncertainty among trial site personnel as they encounter the same scale administered in diff erent ways by varying sponsors. Th is increases score variability and undermines the ability to demonstrate a drug-placebo diff erence of eff ective agents.
Once trained, raters may not continue to rigorously adhere to scoring conventions and administration standards. In one international study with a ratings surveillance program in place, half of ADAS-cog raters and one-third of Mini-Mental Status Examination raters required some level of remediation after the conclusion of rater training [37]. Such programs are not in place for all trials and rater drift may aff ect outcomes.
Th ese studies demonstrate defi ciencies in training and test application in trials. Th ese problems are not unique to international trials but are amplifi ed in global trials where diff erences in language and trial site experience may be more marked and further undermine data reliability. Greater attention to rater training and rater monitor ing or use of centralized raters may assist in manag ing measurement-related variability in AD clinical trials.
Good Clinical Practice guidelines have been incorporated into trial procedures to a variable extent across international sites. Th is leads to variations in the care with which informed consent is obtained, diligence in application of inclusion and exclusion criteria, rigor of collecting information on concomitant medications and use of over-the-counter treatments, attention to timeframes of data collection, thoroughness of adverse event reporting, and completeness of data collection and recording [37]. Th is variable familiarity with and application of good clinical practice may aff ect data quality.
Protection of human subjects and obtaining informed consent through provision of complete information to patients and caregivers and insuring that they understand the procedures, risks, and expectations of clinical trial participation is an essential and required aspect of AD clinical trial conduct. In countries with less extensive histories of research conduct and investigations involving human subjects, there may be less familiarity with these procedures. Institutional review boards may be inexperienced and understaff ed [37,39]. Th is exposes the study to uninformed patient participation, lack of protection of human rights, unacknowledged confl icts of interest, and violations that may compromise study validity and acceptability.
Pharmacovigilance, both during trials and after drug approval in any country, provides critical data on drug safety and mechanisms must be in place to share this information globally in a timely way with appropriate translations and mechanisms for notifying local insti tutional review boards and patients.

Extrapolation of clinical trial data among regions
Guidelines regarding the proper sample size and statistical approaches to multi-regional trials are evolving. ICH guidelines [4] provide general instructions regarding extrapolation of data between regions and when bridging studies will be required. Tsou and colleagues [40] suggest that consistency of responses across ethnic groups in global trials is an important means of determining if responses can be extrapolated across regions. Ethnic groups or nations must comprise at least 20% of the trial population to make such analyses feasible [40]. Consistency in analyses can also be applied to determine if trials conducted in specifi c regions have outcomes consistent with reference trials conducted for regulatory approval in other countries [41].
Bridging studies are required by regulatory bodies when the validity of extrapolation of data from one region to another is uncertain. Bridging studies can vary from demonstration of safety in the new region, to doseresponse pharmacokinetic studies, to comprehensive clinical trials with requirements to demonstrate treatment effi cacy [42,43]. Drugs less sensitive to ethnic factors (Table 4) are less likely to require bridging studies. Bridging studies can be conducted simultaneously with pivotal trials in global drug development programs to reduce delay in seeking marketing authorization in several countries at study conclusion [42].

Regulatory factors that may vary among countries
Th e ICH guideline E5 provides a framework for con sidering ethnic infl uences in global clinical trials [9]. Th e purpose of this guideline was to minimize unnecessary duplication of trials, save human and material resources, and expedite the development of medications to benefi t patients [44]. Th is guidance is the major regulatory reference for sponsors and contract research organizations conducting international clinical trials.
For industry sponsors pursuing marketing authoriza tion in international markets, strategies and requirements diff er by country. Japan generally requires that 50 to 60% of patients in a trial be Japanese for the agent to be considered for regulatory approval there. China requires that 200 to 300 Chinese patients be included in a trial [23].
Th e US Food and Drug Administration (FDA) accepts foreign data in support of a marketing application for a human drug, biological product, or device as long as the data were collected in a study conforming to the ethical principles contained in the Declaration of Helsinki [45]. It is anticipated that the number of US patients in a pivotal trial comprising part of a New Drug Application (NDA) will permit analysis to determine if the agent was eff ective in the US population.
Data from international trial sites are increasingly included in marketing applications for drugs and biologics. In 2008, 80% of applications included at least some data from foreign clinical trials [2]. Th e rise in the number of trials conducted abroad has not been matched with FDA regulatory scrutiny. An Offi ce of the Inspector General review found that the FDA conducted inspections on only 0.7% of foreign clinical trial sites [46]. Th e results of these reviews suggested that international sites often do not meet expected clinical trial conduct criteria.

Clinical operations and sample handling that may aff ect global trials
Contractual relationships may vary from country to country and operational fl exibility is required for global clinical trials. Administrative and fi nancial arrangements also diff er among countries.
Biomarker capacity may vary markedly among global sites. Th e US, Canada, and Western Europe tend to be early adopters of new technology compared to other countries, and advanced imaging capability will typically not be as widely available as clinical expertise in global sites. Use of phantoms for image standardization and machine calibration may not be familiar to all sites, and sponsors of international trials will need to provide additional technical expertise to ensure quality [47]. Use of centralized reading of brain imaging is advantageous to minimize site-to-site variability in image interpretation.
Sample collection and handling may also pose challenges in global trials. China, for example, has restrictions on DNA and plasma export, requiring identifi cation and standardization of a reference laboratory in China in order to pursue trials in that country.
Maintaining supply line integrity can be diffi cult in international circumstances. Importation laws can slow movement of drugs across international borders and implementing Good Manufacturing Practices for local production can take valuable time and delay trial initiation.

Comment
Globalization of AD clinical trials is a consistent trend over the past decade and will continue into the fore seeable future [7]. Including international sites in AD trials allows timely patient recruitment, diversifi cation of trial participants (including patients more likely to be treatment naïve), conduct of trials with less expense and facilitation of marketing applications in participating countries. Global trials have advantages for patients, investigators, and sponsors. Patients get access to experimental agents and may experience a higher level of care than is otherwise available; medications may become available in international markets more rapidly. Investigators are able to participate in international collaborations, contribute to the advancement of therapies and scientifi c progress, increase skill levels of themselves and their teams, may augment income and may experience increased prestige and credibility. Sponsors benefi t from cost savings, speed of recruitment, and market access. Th ese advantages must be considered in the context of the many challenges to conducting clinical trials on a global scale, including uncertainties regarding diff erences in disease biology, the eff ect of ethnicity on drug pharmacokinetics and pharmacodynamics, and challenges to adequate human protections.
Attention to the factors discussed (Table 3) is necessary to ensure the integrity of data included in multiregional trials. Many issues that may aff ect global trials have not been studied in suffi cient detail to predict the magnitude of the eff ect on trial outcomes. Some issues such as trial conduct can be addressed by rigorous site training and scrupulous monitoring and oversight during the course of the trial. Central ratings and application of new assessment technologies, such as computerized data collection, may improve some aspects of data quality. Biological factors that may be related to ethnicity and may aff ect pharmacokinetics, pharmacodynamics, and disease biology are more diffi cult to address. More rigorous scientifi c investigation of these factors and their eff ects in trials is needed. Complex modeling may be needed to determine the eff ects of education and head size on disease progression, for example; such models may determine adjustments in sample size or trial duration needed for effi cacy trials.
Scrutiny of the factors aff ecting international trials will help avoid failed trials and facilitate the development of urgently needed new treatments for AD.

Conclusion
Intrinsic biological and extrinsic sociocultural factors may aff ect trial outcomes in AD drug development programs. Host, disease and trial conduct practices may all vary internationally. Cross-national comparison studies are few and the eff ect of regional variations has been inadequately explored. Attention to these factors is important to successful global trials.