On the issue of transmissibility of Alzheimer disease

Abstract

Results from recent experiments with rodents imply that Alzheimer disease might be inducible by seeding Aβ peptides into recipient animals. In respect to this new experimental data, public health aspects as well as epidemiological data have to be reevaluated. In this article, the available experimental and epidemiological data are reviewed.

Keywords: :

• Alzheimer
• transmissibility
• prion
• dementia
• epidemiology

Introduction

Considerable progress has been made to elucidate the complex pathogenesis of Alzheimer disease (AD) especially during the last two decades. Yet many questions still remain unanswered. The neurodegenerative character of AD has virtually been undisputed. However, as a classic example for a protein misfolding disease, it has been hypothesized that pathogenetic similarities between AD and prion diseases might exist.^1,2 The concept of “seeded proliferation of misfolded proteins” in Alzheimer disease is already part of the current scientific paradigm.³ This implies that AD might harbor the potential for being communicable. Latest experiments of Jucker, Eisele and collegues,⁴ Morales et al.⁵ as well as Clavaguera et al.⁶ have shed new light on the possibility that certain neuropathological hallmarks of AD such as Aβ or tau aggregation might be inducible. However, the results of these experiments have to be interpreted with care and cannot be generalized since they were performed using highly susceptible mouse models. They do not necessarily imply that a similar process occurs in humans. Yet, the idea of AD “inducibility” (and furthermore “transmissibility”) is being discussed in the broad public.^7-10

This is not the first time this topic has been debated.⁷ Yet the question arises whether human AD can also be induced exogenously, e.g., by accidental or common contact with AD-typical pathological protein aggregates. This issue has to be discussed also on the basis of evidence from epidemiological studies which were performed using classical approaches.^2,7,8,11 But are the available epidemiological studies on disease risk factors valid in this context? Did they take the potentially long lag time between the assumed pathological event and clinical disease onset into account?

Beforehand, we need to define precisely what to understand by the terms “inducible” and/or “transmissible.” Broadly categorized, three levels of induction of pathological changes on protein level can be distinguished: First, conversion of conformations restricted to within a cell’s cytosol (e.g., yeast prions); Second, conversion of conformations within the secretory system (e.g., extracellular proteins, Aβ) or secretion and uptake of cytosolic proteins (e.g., tau, synuclein, SOD1) including cell-to-cell spreading usually within one organ (CNS); Third, true (classic) transmissibility between individuals (e.g., prion diseases).^9,12 The main difference between the second and third is that for the third category, specific (receptor-based) uptake and neuroinvasive mechanisms have to be present. In our opinion, “infectivity” and “transmissibility” should only be used referring to the third category. Otherwise, “inducibility” appears to be the term of choice and mainly refers to cell-to-cell spreading.

Moreover, it is essential to note that a possible “induction/transmission” of neuropathologically detectable, Alzheimer-like lesions does not necessarily mean “induction/transmission” of clinically symptomatic disease. Even in brains of non-demented subjects AD-like lesions have been found.^13,14 Therefore, it can be considered a big step from induction of sole neuropathological alterations on one side and induction of clinically relevant Alzheimer dementia on the other.

Due to the newly emerging evidence it must be critically discussed whether AD-typical protein aggregates could be inducible or not—and if so, which implications this might have in an epidemiologic context. Here we review the available evidence from published literature regarding that matter. References for this work were selected systematically through searches of Medline with search terms such as “Alzheimer,” “transmission,” “infectivity,” “epidemiology,” “risk factors,” “induction” and “prion disease.” Titles and abstracts were analyzed and assessed for relevance. Reference lists of relevant studies were searched to identify additional literature not covered by the database search. Since epidemiological evidence is scarce, all relevant literature was sought to be included into this review. A meta-analysis approach was not considered useful due to the lack of utilizable epidemiological data on AD and potential transmissibility.

Experimental Clues

Early experiments particularly from the era before sensitive detection of prions was possible (first anti-human PrP antibodies became available in 1987¹⁵) reported a so-called transmissibility of AD-neuropathology in a few cases.^16-20 Given the possibility that prion diseases might not have been recognized or not been defined adequately, this “transmissibility” may have been mistakenly observed due to unrecognized Creutzfeldt-Jakob disease (CJD) or even both, concurrent CJD and AD in the same individual.¹⁶ In any case, no reasonable explanation for these results has been found so far.

In inoculation experiments by Baker and collegues,²¹ primates injected intracerebrally with brain homogenate from eoAD (early onset AD) patients developed Aβ plaques (but no neurofibrillary tangles) while controls did not. The induction of congophilic angiopathy—a form of Aβ aggregation in cerebral blood vessels—was observed, too. In 1994, Brown et al.²² reviewed transmission experiments regarding various human CNS diseases performed with non-human primates including chimpanzees as the closest relatives presumably with the lowest “species barrier.” It was concluded that only CJD but no other degenerative brain diseases would be transmissible and lead to clinical symptoms. However, it has not been reported, whether the recipient brains were thoroughly (re)investigated with sensitive detection methods for Aβ or phosphorylated Tau (pre)tangle formations.

Induction of cerebral amyloidoses appears to be possible by intracerebral “seeding” as has been demonstrated by Kane et al., who injected brain homogenate of AD patients into transgenic mice, which in return developed profuse Aβ plaque formations²³ or by DeGiorgio, who implanted tissue samples from AD patients’ meninges or skin into mice cortices.²⁴ These mice developed plaques consisting of APP, Aβ, cathepsin D, apolipoprotein E and ubiquitin. However, those plaques were present only for a short period of time (one month post implant). Melanie Meyer-Luehmann and colleagues succeeded in inducing AD-neuropathology in transgenic mice by intracerebral injection of diluted human AD brain extracts in a concentration- and time dependent manner.²⁵ This seeding could be counteracted by Aβ immunodepletion or immunization of the hosts. Moreover, Langer et al. were able to demonstrate, that especially soluble Aβ seeds are very potent inducers of cerebral amyloidosis and might thereby be mediators of the spreading pathology.²⁶

All these experiments involved central inoculation or seeding. During the past few years, Eisele and team have been investigating, whether induction of cerebral amyloidosis might be achievable by peripheral inoculation of Aβ.⁴ After oral, intravenous, intraocular and intranasal inoculations did not show any effect,¹⁰ Eisele and collegues decided to inoculate Aβ rich brain extracts from mice into the peritoneum of APP23 transgenic mice.⁴ These rodents, which were predisposed to the formation of plaques, developed Aβ plaques. This implies, that cerebral amyloidosis can be accelerated if not even induced by peripheral inoculation of Aβ seeds, possibly analogous to prion diseases.^27-29 Morales and colleagues took a further step and were able to show that this process is reproducible in animals, which without inoculation, would not develop Aβ plaques.⁵ It is hypothesized, that some kind of cell-to-cell transfer of pathogenic Aβ molecules must have been present in this scenario. The results of various immunization studies could be used as an argument in favor of this cell to cell transfer.^30,31 Moreover, the publication of data from experiments in which peripheral seeding is attempted by blood transfusion is expected in the near future.

Epidemiologic clues

Several attempts have been made during the 1980s and 90s to investigate the potential transmissibility as well as other risk factors of AD with classic epidemiological methods.^16,32 Early studies focused on the identification of potentially infectious agents. Like in other diseases of unknown origin, slow virus infections were suspected to play a role in AD pathogenesis. However, only very few studies provided evidence for an association of AD with Herpes simplex and Herpes zoster infections, whereas the vast majority was not able to demonstrate any proof.^33-42 (Table 1)

Table 1. Overview of case control studies, in which risk factors for developing AD were examined

Study	Method	N (patients)		N (controls)		Examined factor categories potentially relating to induction or transmission of AD	Results regarding factors potentially related to AD induction/transmission
Amaducci 1986^33	Case control	116		116 (hospital based), 97 (population based)		viral	no association
Heyman 1984^39	Case control	40		80 (community based)		environmental, viral	no association
Graves 1990^38	Case control	183		130 matched pairs (friends, non blood relative)s		viral	no association
Bohnen 1994^43	Case control	252		252 matched pairs		blood transfusions	no association
Broe 1990^44	Case control	170		170 matched pairs			no association
O’Meara 1997^45	Case control	326		330 matched pairs		blood transfusions	no association
Breteler 1991^32	Pooled re-analysis of 8 case control studies	Exposure frequencies		Exposure frequencies		blood transfusions, viral	no association
		blood transfusions	71/523	blood transfusions	112/562
		herpes simplex	94/314	herpes simplex	78/295
		herpes zoster	100/1103	herpes zoster	105/1110
		poliomyelitis	5/451	poliomyelitis	7/467
		encephalitis/meningitis	7/1081	encephalitis/meningitis	5/1167
Zuo 2010^48	Case control	26		161		spine surgery	no association

Selected were those possibly associated with AD induction by a potentially transmissible pathogen (e.g., viral infections, blood transfusion, cns surgery). Therefore, studies regarding anesthesia, peripheral surgery, life style factors, genetics, etc. which are more abundant are not shown.

Iatrogenic risk factors were studied in a number of case-control studies. One of the most interesting topics was the patients’ history of blood transfusions. Numerous studies were conducted with different populations in several countries with various methods, yet no evidence for an association between prior blood transfusions and the risk of AD could be shown.^{32,33,39,43-45} However, these studies were prone to recall biases (subjects answers are affected by their memory) as well as to selection biases of the control group (error in choosing the group correctly) to some extent. (Table 1)

Important aspects were mainly discussed as limitations concerning issues such as case definitions, control of confounding factors (e.g., surgery, use of anesthetics) appropriate use of proxies or design of questionnaires.

The choice of control groups is crucial in all kind of case-control studies, since it is challenging to find a specific control population matching the population under observation in respect to potential confounding factors. In addition to the issue of recall bias, population-based controls will be generally less likely to have a history of e.g., blood transfusions than AD patients diagnosed at a hospital based reference center and included in any kind of study. Conversely, hospital-based controls will usually be more likely to have a history of blood transfusions due to the diseases, which they had initially been referred for to the medical facility. Selection bias will therefore generally play an important role, even more, since the expected effect sizes might be small. Breteler et al. discussed this issue in detail, but until now, no feasible solution has been developed.³² [When blood transfusions are mentioned here in context with AD, it has to be kept in mind that no study to date has provided evidence for AD transmissibility through transfusion of blood products. Epidemiologically, no association has been found (Table 1). Some data from experiments using mice models were reported on conferences, however, details of the experiments are not known and a final conclusion can be dawn after the publication only.]

Other iatrogenic risk factors have also been examined.^32,45-48 Among these, general anesthesia had been suspected to be associated with an increased risk for AD, since long-term cognitive decline was observed after anesthesia in various studies.^49-55 To discuss general anesthesia in this context is important, as it may act as a major confounder when researching surgery or (intraoperative) blood transfusions as risk factors for potential transmission of AD in epidemiological studies. Molecular mechanisms were hypothesized and especially volatile anesthetics were demonstrated to be associated with Aβ production, Aβ oligomerization, Tau formation and Tau phosphorylation as described thoroughly in a review by Papon et al.⁵⁶ Several observational studies were designed to substantiate these experimental theories with epidemiological data.^{32,47,48,57-60} Nonetheless, a recent systematic review concluded that, despite the amount of studies available, evidence for an association of anesthesia with AD is still inconclusive.⁵⁶ This exemplifies the difficulties in interpreting these kinds of studies.

Furthermore, head trauma and surgery have been shown to be associated with a higher risk of AD. Increased stress hormone levels were suggested as one possible causal factor.^47,61 Traumatic brain injury (TBI) may even trigger neuropathologically detectable abnormal tau and amyloid deposition.^62,63 One surely could argue that general anesthesia, TBI or transmissibility between individuals might be the causal factor beneath the observed association. This clarifies that although potential molecular mechanisms have been discussed, it is still not known which factors might be causal and which might only be confounders, since they appear highly correlated and collinear to each other.

In synopsis, for most of the investigated iatrogenic risk factors, inconclusive or no evidence of an association with AD was detectable. Specific issues such as endoscopy or surgery were not evaluated in detail. Is this absence of evidence based on a few case control studies conducted 20 years ago enough to conclude evidence of absence of associations between AD and iatrogenic risk factors in general? The only solution to answer this question is to set up new specifically designed epidemiologic studies.

Problems to address in epidemiological studies

Unfortunately, the application of classical epidemiological methods entails more significant limitations in a disease with unknown time of pathophysiologic onset than only selection biases. The dissociation between pathophysiologic and clinical onset—or, in terms of infectious diseases, unknown incubation times—with possibly more than 20–30 years is complicating matters severely, as has been demonstrated in various case-control studies in the prion field.⁶⁴ On one hand, differential misclassification in terms of recall bias might play an important role. On the other hand, non-differential misclassification might also be able to bias an existing effect against unity until it becomes undetectable, since detailed information such as contact patterns or other potential iatrogenic risk factors have to be remembered irrespective of a recall bias. Moreover, in newly arising diseases with unusual ways of transmission, these limitations regarding detection of infectivity are even more pronounced and appropriate study designs become even more crucial. Experiences from prion studies might be useful for further research in this respect.

Irrespective of the current lack of evidence, iatrogenic induction/transmission needs to be re-evaluated in the light of the recent experimental data. Case control studies with improved measures of detection of exposure (e.g., by using medical records of population based databases) as well as improved questionnaires (e.g., by breaking down intervals to a maximum of only a few years to allow for unknown incubation times) might be helpful. Prospective cohort studies with the power to detect even small effect sizes need to be designed despite the fact, that they will have to be continued for decades before a final conclusion can be drawn. In addition, studies evaluating the AD risk attributed to sub-populations at risk (e.g., nursing homes, physicians) might be of use to gain further insights. The only study available to date compares disease-specific mortality rates (SMR) of neurosurgeons with the general US population. It suggests neurosurgeons to have a more than 2-fold risk of death from AD.⁶⁵ Is the exposure to patient brain responsible for that finding or are there alternative explanations? First of all, careful assessment of potential biases is necessary. Neurosurgeons (pursuing a profession associated with high socioeconomic status) are less likely to die from other causes than the general population. A healthy worker effect might have influenced the study considerably. It is somehow conspicuous that AD was the only disease evaluated which featured a significantly higher risk whereas the SMRs for other neurodegenerative diseases (e.g., Parkinson) were not increased. Reasons other than a potential induction of AD appear more likely. As described above, exposition to volatile anesthetics must be taken into account. It has to be discussed, whether anesthesiologists would have been the adequate control group in a prospective study.

Almost all these considerations depend on the assumption that AD-typical neuropathology, if transmissible at all, resembles prion diseases in some way. The available epidemiological and experimental evidence is by far not good enough to either conclude that AD is transmissible or, on the other hand, to guarantee that certain ways of transmission are excluded to play a role in the development of some AD cases at all. Previous experiments were mostly performed using predisposed animals with short periods of follow-up. No valid conclusion about long-term effects can be drawn therefore. Moreover, it has to be asked: are there alternate ways of transmission apart from inoculation of brain material, CSF or blood? And if so, what would the public health implications be?

In summary, epidemiological evidence is extremely difficult to assess in this context. This is especially true, since case definitions and case detection rates have changed over time. Given the fact that a potential way of transmission is not only unknown but that no way of transmission can be excluded yet, careful assessment is imperative.

Public health implications

In the epidemic of non-communicable diseases AD plays an important role for morbidity and mortality as well as the associated costs. To date, AD is the 6th leading cause of death in the US and the projected costs in the US by 2050 are $1.1 trillion.⁶⁶

Knowledge about transmissibility is essential in all kinds of epidemics for prevention, diagnosis and treatment. If AD featured transmission patterns comparable with those of prion diseases, iatrogenic induction/transmission would play a major role for public health. Prevention would be first priority and might include measures such as extended sterilization methods for surgical instruments as well as identification of patients potentially posing a risk as well as patients at risk. From a public health perspective also alternate ways of transmission not evaluated yet must be considered. For prevention to work, biomarkers for early disease detection must be validated, independently of the ways of possible induction since the disease itself starts well before clinical onset. As Sigurdsson stated correctly almost 10 years ago, future research on therapies might also be limited as long as there is no convincing evidence against transmissibility of AD.⁸ This is especially true for vaccination studies and clinical trials using parts of β-amyloid.

Conclusion

In conclusion, with this short review, we want to encourage a broader discussion and modern epidemiological research in this context. Well-designed epidemiologic studies have to be initiated. Methodologically these studies must be constructed to address the epidemiologically challenging aspects of Alzheimer disease such as dissociation between first neuropathological alterations and clinical onset, uncertainties in early diagnosis as well as AD heterogeneity.

Abbreviations:
Aβ	=	amyloid beta
AD	=	Alzheimer disease
CJD	=	Creutzfeldt-Jakob disease
CSF	=	cerebro-spinal fluid
eoAD	=	early onset Alzheimer disease
rpAD	=	rapidly progressive Alzheimer disease

Acknowledgments

This work was supported by a Bundesministerium für Bildung und Forschung grant within the German Network for Degenerative Dementia (KNDD-2, 2012-2015, determinants for disease progression in AD, grant no. 01GI1010C), as well as JPND, EU-FP7 PRIORITY.

Authors’ contributions: C.S. was responsible for the initiation of the project, general conception and the composition of the final manuscript. A.K. provided parts of the section on epidemiology. C.K. provided parts of the section on experimental clues regarding AD transmissibility/inducibility. I.Z. was responsible for the initiation of the project, the critical review for contentual errors and writing parts of the conclusion. All authors contributed to the revision of the manuscript. The authors declare no conflicts of interest.