Abstract
Alzheimer’s disease (AD) is a neurodegenerative disease characterized by extracellular amyloid-β (Aβ) peptide aggregates, forming amyloid plaques, and intracellular deposits of phosphorylated tau. Neuroinflammation is now considered as the third hallmark of AD. The majority of clinical trials tested pharmacological strategies targeting amyloid, tau, and neuroinflammation, with disappointing results overall. In parallel, innovative strategies exploring other pathways and approaches are being tested. In this article, we focus on the rationale and preliminary preclinical evidence for a novel application to AD of a widely used therapeutic strategy for oncological and benign conditions: low-dose radiation therapy (LD-RT). LD-RT has shown to be effective against systemic amyloid deposits, as well as against chronic inflammatory diseases, and could thus be able to modulate amyloid load and neuroinflammation in AD. The anti-amyloid effect could be possibly mediated by the LD-RT action on the β-sheet structure of amyloid fibrils, by breaking H-bonds, and depolymerize glucoaminoglycans which are highly radiation-sensitive molecules associated with amyloid fibrils. The anti-inflammatory effect could be linked to the decrease of leukocytes-endothelial cells interactions and to the stimulation of the release of anti-inflammatory molecules. One preclinical study has observed a dramatic reduction of amyloid plaques 4 weeks post-RT, more important with fractionated protocols at low doses than hypofractionated single dose treatments, associated with modulation of inflammatory and anti-inflammatory cytokines and cognitive improvement. Ongoing Phase I clinical trials will test the ability of LD-RT to hold these promises.
INTRODUCTION
Alzheimer’s disease (AD) is a neurodegenerative disease which involves complex interactions between amyloid-β (Aβ) peptides, tau, and neuroinflammation. Indeed, the observation of Aβ plaques by Alois Alzheimer is at the origin of the amyloid cascade hypothesis. It is now believed that production and aggregation of Aβ peptides lead to abnormal tau hyperphosphorylation, inflammation, synaptic alterations, and neuronal damage, linked to the cognitive alterations observed in AD patients. This cascade hypothesis generated dozens of clinical trials targeting amyloid, all of which failed [1]. Since decades, the amyloid cascade is controversial, mainly for the lack of association between amyloid deposition and symptoms, which are rather correlated with the presence of tau aggregates in the neocortex [2]. Unfortunately, trials targeting tau have also failed so far [3]. The third hallmark of AD, neuroinflammation, is at the center of current AD research. Neuroinflammation in AD is characterized by the presence of reactive astrocytes and activated microglial cells, which are mainly observed around amyloid plaques. Nevertheless, neuroinflammation could appear even before amyloid plaque formation and cognitive symptoms [4, 5], making neuroinflammation an interesting biomarker for early detection and an interesting therapeutic target with disease-modifying potential. It is often proposed that neuroinflammation is a protective mechanism at disease onset and in the earliest phases, which then becomes deleterious contributing to AD progression. A deeper understanding of its role is necessary. Prospective studies on the use of anti-inflammatory drugs showed no beneficial effects [6]; nonetheless, clinical trials are ongoing at all phases (source: Clinicaltrials.gov).
During the last decade, alternative strategies were also tested. Natural molecules have been explored by numerous studies. For example, curcumin stirred the curiosity of the research field as it presents antioxidant, anti-inflammatory, and neuroprotective properties and as it has been shown to lower Aβ in animal models of AD (see [7] for a detailed review). However, the majority of the natural molecules studied have either limited capacity to cross the blood-brain barrier or a fast metabolism and elimination, complicating their use for AD therapeutic purposes. The involvement of the microbiota in AD has also been actively discussed over the last years. Indeed, the gut microbiota changes with aging and particularly in AD. More pro-inflammatory bacteria and less anti-inflammatory bacteria are observed, which could be associated with neuroinflammation. Some data showed also a link between the gut microbiota flora and Aβ clearance or tau pathology (see [8] for a comprehensive review). The study of the gut microbiota is still in the pre-clinical stage but its modulation could represent a novel interesting approach [9]. Some studies also focus on non-invasive neuromodulation technics, using for example transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS). These alternative strategies have already proven their abilities to modulate and improve the efficacy of neuronal signal transmission and integration in other disorders such as depression [10]. Consequently, they could help to compensate for the neuronal network disruption observed in AD patients [11, 12], which may be linked to cognitive impairment. More studies are clearly needed to investigate its potential in AD and to define future clinical guidelines [13]. Five clinical trials using TMS or tDCS are actually ongoing (Source: clinicaltrials.gov).
Overall, the disappointing results obtained across different strategies underline an important issue. As complex interactions between amyloid, tau, and neuroinflammation are at play in AD, it seems illusory to treat AD by targeting only one feature and letting the other deleterious factors progress. In this hypothesis article, we will develop the rational to propose low-dose radiation therapy (LD-RT) as treatment which could have an effect both on the amyloid load and neuroinflammation.
RADIATION THERAPY: AN INNOVATIVE TREATMENT AGAINST AD?
RT: A short introduction on principles and doses
Radiation therapy (RT) is a well-known treatment in oncology. RT is commonly based on the delivery of megavoltage ionizing radiations by linac accelerators. Radiation interacts with molecules in the cell leading to direct or indirect ionizations. Direct ionization consists of single or double strands DNA breaks and is more frequently observed using high linear energy transfer (LET) radiations. Indirect ionization is via hydroxyl free radicals formation after water ionization and it is most commonly observed in RT treatments using low LET irradiations. Delivered doses differ based on the pathology and treatment intent. Standard fractionation, the most commonly used regimen, consists of delivering daily fractions of 1.8 or 2 Grays (Gy) in order to allow healthy cells to reoxygenate and repair between treatments, reducing therefore the incidence of side effects. Although RT is commonly used to treat oncological diseases and dose escalation up to 80 Gy has been demonstrated to increase the local control rates for some tumors, RT can also be delivered at lower doses, in the order of 5 to 10 Gy with daily or weekly fractions≤2 Gy, to treat benign degenerative inflammatory disorders [14].
The effects of ionizing radiations on proteins and amyloidosis
RT appears to be an efficient treatment for pathologies linked to peripheral amyloid deposition (Table 1). Indeed, RT showed favorable results for its use in the treatment of tracheobronchial amyloidosis (TBA), an uncommon disorder characterized by amyloid deposition in the airways. Three case reports described a successful treatment of TBA using fractionated RT schedules with a median dose of 20 Gy. Other cases of respiratory function improvement after RT for TBA were reported by four other teams, with a complete response and no recurrence in 7 patients [18]. A retrospective study validated the success of this treatment and the absence of recurrence during 7 years following RT [19]. Encouraging studies for RT as a safe treatment of peripheral amyloid pathologies were also described in the case of laryngeal amyloidosis [20], nasopharyngeal amyloidosis [21], primary parenchymal amyloidosis [22], orbital amyloidosis [23, 24], periocular light chain amyloidosis [25], and primary bladder amyloidosis [26].
Clinical evidence of LD-RT efficacy on peripheral amyloid pathologies
Gy, grays, fx, fractions, RT, radiation therapy.
The mechanism of action of RT on systemic amyloid deposits is not well understood. One of the most established effects of RT is the induction of DNA damages through a double strand break, leading to cell death. However, the effect on systemic amyloid load cannot be explained only by the destruction of plasma cells, supposed to be at the origin of systemic amyloid deposits, because the cellular component is not prominent when deposits are already formed. RT could impact others mediators involved in amyloid production or degradation. Moreover, it has been proposed that RT affects the β-sheet structure of amyloid fibrils, by breaking H-bonds, and depolymerizes glucoaminoglycans which are highly radiation-sensitive molecules associated with amyloid fibrils [27] (Fig. 1). Importantly, systemic amyloidosis and AD plaques share the same cross-β-sheet fibril structure consisting of misfolded and/or misassembled proteins. Consequently, all these studies showing LD-RT efficacy on peripheral pathologies linked to amyloid deposits offer reasonable and encouraging arguments to explore LD-RT impact in AD. Moreover, the ability of RT to overcome the blood-brain barrier, which can limit or thwart the efficacy of many pharmaceutical agents, makes this treatment an interesting new approach.

Potential effects of low dose RT in AD brain. A) Effects of low dose radiation therapy (LD-RT) at cellular level. LD-RT could increase anti-inflammatory molecules released by microglia and astrocytes and decrease nitric oxide (NO) and reactive oxygen species (ROS) production by microglia, astrocytes or neurons. LD-RT could also impact circulating cells responsible for the production of pro- and anti-inflammatory signals in the blood vessels which are then released in the brain parenchyma. This overall decrease of inflammatory signals could offer a better survival environment to neurons and could influence the production of Aβ peptides by neurons, microglia or astrocytes. Phagocytosis functions of microglia or astrocytes could also be improved with the decrease of their reactive state. B) Molecular impact of LD-RT. LD-RT could decrease amyloid load directly by breaking H-bonds involved in the maintain of β-sheet structures of oligomers, fibrils and plaques. 1: anti-inflammatory effects. 2: anti-amyloid properties of LD-RT.
The effects of ionizing radiations on inflammatory cells
RT also interacts with inflammatory cells. When delivered at higher doses, RT, particularly in the brain, induces inflammation, mainly via reactive microglial cells, endothelial cells, and disruption of blood-brain barrier integrity. This radiation-induced inflammation, probably in response to double-strand breaks, leads to the secretion of multiple pro-inflammatory molecules and induces high stress which can participate to neuronal damage and finally to radiation-induced brain injury [28, 29]. However, low dose irradiation modulates different gene pathways associated with stress response control, DNA repair, neuronal signaling, and anti-inflammatory response [29].
The anti-inflammatory effects of LD-RT are known since decades on peripheral organs [30–32]. LD-RT has been used for the treatment of benign and chronic inflammatory and degenerative diseases with excellent results [31]. In addition to anti-inflammatory effects, analgesic effects were observed in many patients. For example, LD-RT was used to treat plantar fasciitis with a removal of pain in 70% of patients [33]. Mechanisms involved in anti-inflammatory effects of LD-RT may implicate endothelial cells, leukocytes, and macrophages with a decrease in leukocytes-endothelial cells interactions (reduction of adhesion molecule expression and cell adhesion) [34–36], oxidative stress (reduction of nitrite oxide and reactive oxygen species production) [37, 38], and a stimulation of anti-inflammatory molecules (such as Nuclear factor kappa-light-chain-enhancer of activated B cells pathway, Interleukine-10 (IL-10), and Transforming growth factor-β1) [39, 40].
Less is known about LD-RT and inflammation in the brain. The first evidence of a protective effect of LD-RT on the adult brain has been shown in 1994. With a dose at 100 cGy and lower, Yamaoka and collaborators observed a reduction of free radical production, known to be involved in oxidative stress [41]. Other studies, using gamma or X-ray irradiation at low doses, confirmed the reduction of oxidative stress on neural cell cultures or mouse models of Parkinson’s disease (see [42] for more details). Moreover, it has been also shown that low-dose gamma irradiation decreased neuroinflammation in two mouse models of multiple sclerosis [43, 44]. For example, it blocked the production of pro-inflammatory cytokine production like IFN-gamma and IL-6 and the proliferation of CD68+ cytotoxic T cells [43]. The response of astrocytes and microglial cells, the main actors of inflammation in the brain, has not been specifically studied and has yet to be deciphered. Consequently, mechanisms of anti-inflammatory effects of LD-RT still need to be explored in the brain but a reduction in oxidative stress and neuroinflammation levels may crucially influence cell function and neuronal survival (Fig. 1). More specifically to AD, an anti-inflammatory outcome may also modulate Aβ phagocytosis by microglia and astrocytes, as microglial phagocytic capacity might be reduced in the activated microglia phenotype [45].
As neuroinflammation could appear even before formation of amyloid plaques or symptoms appearance [4, 5], the potential anti-inflammatory effect of LD-RT could be interesting at the earlier phases of the disease, in preclinical subjects, to stabilize the cellular environment and reduce the neuronal suffering and death, helping to maintain cognitive abilities. In this case, a longitudinal follow-up of inflammatory markers could be highly interesting and possible with TSPO imaging, as well as by quantifying inflammatory molecule levels in the plasma. At the same time, if LD-RT can reduce the amyloid load in the brain, as it is observed for peripheral amyloidosis, this treatment approach could be interesting even in advanced AD stages, with the objective to stabilize or improve the cognitive performances of patients. Nevertheless, the optimal AD candidates for this therapeutic approach remain to be defined, with future preclinical and clinical data eagerly awaited to better understand the best treatment strategy.
The neurotoxicity of ionizing radiations
The effects of dose escalated high-dose (HD) RT on the brain have been studied in oncological patients. RT at high doses may induce brain atrophy, deficits in neurogenesis, and cognitive decline, and some reports have identified an increased risk for AD after RT treatment for brain tumors [28, 48]. Consistent with this hypothesis, Lowe et al. observed that LD-RT induce a downregulation of molecular pathways link to learning, memory, and cognitive functions. Over the 15 pathways modulated by LD-RT, six have a high degree of concordance with pathways modulated in AD brains [46]. Importantly, they showed that LD- and HD-RT do not affect the same molecular networks, suggesting that mechanisms involved in the cellular responses to LD-RT and HD-RT are different. Another study evaluated in a ApoE deficient mutant mouse the impact of chronic low-dose rate irradiation (1 mGy/day or 20 mGy/day) over a 300 days period, with cumulative doses ranging from 0.3 Gy to 6 Gy [49]. Changes in the phosphoproteome associated with synaptic plasticity, calcium-signaling, and metabolism in the hippocampus, pathways also related to AD, were observed. However, neurogenesis or cell death was not influenced at these doses. In agreement with these results, it appears that hippocampal atrophy may be observed after a HD-RT treatment but not after LD-RT [50, 51]. Consequently, it is imperative to differentiate negative effects of LD- and HD-RT on cognitive functions. Furthermore, it has been demonstrated that LD-RT may induce early transcriptional alterations but does not induce amyloid plaques and neurofibrillary tangles, the two essential hallmarks of AD, in wildtype mice [52, 53].
Potential side effects from brain RT are strongly dependent on the total delivered radiation dose. RT schedules commonly used for the treatment of peripheral amyloidosis or inflammatory benign diseases deliver lower doses compared to those proposed in the clinical practice for whole brain radiation therapy (WBRT) of multiple brain metastases (ranging from 30 Gy in 10 daily fractions to 40 Gy in 20 fractions) or focal brain irradiation (60 Gy in 30 fractions for malignant glioblastoma for example). A dose more than three times lower is consequently presumed to be associated with minimal risk of long-term side effects, especially if critical brain structures, namely the hippocampus, are spared using modern RT techniques with image guidance [54]. Obviously, use of hippocampal sparing RT techniques remains debatable when LD-RT should be applied as a possible treatment strategy for AD. As the hippocampus is one of the main brain structures involved in AD, dose sparing of this region remains questionable. However, in patients with mild AD, amyloid load and symptoms seems to be related not only to the hippocampi but also to the cortex regions [2, 55]. Moreover, some publications have demonstrated that even at low doses (i.e., 2 Gy), RT can reduce the cell survival in culture of neural stem cells and can alter the neurogenesis in the hippocampus of rats [54]. In adult patients with benign or low-grade brain tumors, a 7.3 Gy RT dose delivered to 40% of the hippocampus was associated with long-term alterations in list-learning delayed recall [56]. In the RTOG 0933 single-arm phase II trial, hippocampal–sparing WBRT was associated with improved memory and quality of life preservation as compared to historical series [57]. Consequently, inclusion or not of this structure in the low-dose RT volume should be clearly evaluated. In fact, a comprehensive balance between the potential treatment efficacy and the risk of long-term cognitive decline induced by WBRT has to be determined.
Nevertheless, data on long-term effects on neurocognitive function of WBRT are largely derived from treatments delivering high-dose schedules. For example, in a population of adult survivors of childhood medulloblastoma treated with WBRT with doses ranging from 23.4 to 36 Gy, including a boost of the posterior fossa up to 60 Gy, decline in working memory was mostly observed 10 years after treatment [58]. Noteworthy, prophylactic cranial irradiation has become the standard of care for selected groups of adult patients with small cell lung cancer and in pediatric leukemia patients to decrease local central nervous system relapses, with no, mild, or moderate cognitive impairment [58–61]. Therefore, even if low RT doses are supposed to be effective and sufficient to decrease the two major hallmarks of AD, ongoing and future studies are certainly required to evaluate the safety and efficacy of this therapeutic regimen, especially in the long-term.
RADIATION THERAPY FOR AD: PRECLINICAL EVIDENCE IN A MOUSE MODEL OF AD AND ONGOING CLINICAL TRIALS
A recent study on an AD mouse model described for the first time the beneficial effects of LD-RT in the AD brain [62]. Marples et al. observed a dramatic reduction of amyloid plaques 4 weeks post-RT. Interestingly, the reduction was more important with a fractionated protocol at low dose (5×2 Gy) than with larger single dose treatments (5 Gy, 10 Gy, 15 Gy). They also observed an increase of inflammatory molecules such as Macrophage inflammatory protein 2 or Interferon-gamma 24 h after the treatment but levels returned to baseline after 4 weeks. Other cytokine levels studied were not impacted by LD-RT. Moreover, IL-10, an anti-inflammatory cytokine, tended to increase with the low and fractionated schedules, showing that LD-RT did not induce neuroinflammation in the long-term and may even have an anti-inflammatory effect in the brain [63]. More importantly, memory deficits observed using the Morris water maze were restored 2 months post-RT in old APPswe/PS1dE9 mice [62].
Overall, these first results are encouraging and the potential of LD-RT for AD have to be further studied. However, RT impact on tau has, to our knowledge, never been explored. As this AD mouse model do not develop tau pathology, it would be also very interesting to study the impact of LD-RT on tau hyperphosphorylation and aggregation using other AD models. It seems also necessary to go further into the characterization of LD-RT impact on neuroinflammation. The effects of LD-RT on neuronal transmission and functions and on neurogenesis need also to be investigated.
In parallel to the refinement of pre-clinical studies, two collaborative clinical trials leaded by our team in the Geneva University Hospital, Switzerland and a team from William Beaumont Hospitals and Virgina Commonwealth University, US (NCT03352258, NCT02359864 and NCT02769000) are currently exploring the role of LD-RT in AD patients.
These Phase I trials aim to 1) assess the safety and toxicity/adverse events of a short course LD-RT on the brain and 2) to evaluate for the first time in the human brain the effectiveness of this fractionated protocol to reduce amyloid deposits in patients with a diagnosis of prodromal or mild or moderate AD, as measured with amyloid positron emission tomography. Table 2 summarizes the number of patients included, treatment dose and principal outcomes for both studies.
Current clinical trials using LD-RT treatment for AD patients
WBRT, whole brain radiation therapy, VMAT, volumetric modulated arc-therapy; 3D-CRT, three-dimensional conformal radiation therapy; PET, positron emission tomography.
CONCLUSIONS
Although substantial progress has been made over the last decades in the understanding of the pathogenesis of AD and in the development of therapeutic strategies, most Phase III clinical trials have failed. Is it sufficient to target one of the disease hallmarks to hope to improve or stabilize cognitive function? Radiation therapy at low dose offers the possibility to modulate at least two key AD factors, having proven its efficacy to reduce systemic amyloid and inflammation. Moreover, the first results obtained in a mouse model of AD are highly encouraging since LD-RT reduced amyloid plaques and inflammation in the brain and improved memory deficits. The two Phase I clinical trials currently ongoing will give us responses concerning the promises of LD-RT in AD.
