Singular cases of Alzheimer’s disease disclose new and old genetic “acquaintances”

Alzheimer’s disease (AD) is the most common age-related degenerative dementia. From a clinical perspective, AD typically displays an amnestic syndrome of the hippocampal type that can be associated with various cognitive or behavioural deficits during disease evolution [1]. Atypical forms of AD present with relative preservation of memory at onset and generally occur at an earlier age. They include posterior, logopenic and frontal variant of AD [1]. According to the revised international criteria, at least one biomarker of in vivo Alzheimer’s pathology must be positive: a cerebrospinal fluid (CSF) profile consisting of decreased amyloid-β 1–42 (Aβ₄₂) together with increased total tau (T-tau) or 181-phopshorylated tau (P-tau) concentrations, or an increased retention on amyloid tracer PET (AMY-PET) [1]. In addition to the use of single CSF markers, the combination of multiple CSF markers in the form of ratios further increases the diagnostic accuracy [2]. AD is usually sporadic, with age of onset most often being > 65 years, thus qualifying for late onset AD (LOAD). In no more than 5% of all patients, a positive familial history for dementia or a clear-cut autosomal dominant pattern of inheritance can be found. These familial AD cases (FAD) arise before age 65 more frequently than sporadic cases, hence the definition of early onset AD (EOAD) [3]. Approximately 50% of FAD patients carry a mutation in presenilin 1 (PSEN1), presenilin 2 (PSEN2) or amyloid-β protein precursor (APP) genes, with more than 350 variants collectively identified so far [4, 5]. However, some of them are not pathogenic or their significance remains uncertain, as they may qualify as genetic risk factors or disease-modifying alterations.

Here, we describe a case series of cognitive disorders with in vivo biomarker positivity for Aβ deposition, showing various clinical atypical aspects together with peculiar genetic features. We disclosed two known missense variants, one in PSEN1 and the other in PSEN2, and a novel silent variant in PSEN2. Moreover, additional variants in dementia-related genes have been identified by next generation sequencing (NGS). Our results are intriguing as they raise the question of the role of genetic risk burden in AD.

Alzheimer’s disease is mainly distinguished in a typical presentation with hippocampal amnestic syndrome and atypical forms with different cognitive or behavioural deficits.

In this paper, we describe a series of 6 unrelated patients affected by dementing syndromes characterized by one or more “atypical” features including age at onset, clinical presentation and disease progression rate. Case 5 presented a complex syndrome indicative of bvFTD with parkinsonism and additional atypical features. The severity of clinical picture and the high levels of CSF tau might suggest the possibility of a prion disease. However, the long course, the MRI features, the neuroimaging findings (parieto-temporal atrophy and hypometabolism) and CSF Aβ₄₂ reduction made presenile AD the most likely diagnosis. Case 6 was classified as possible CBS, a clinical syndrome with different underlying pathological substrates [11, 12]. In vivo AD pathophysiological biomarkers and ¹⁸FDG-PET hypometabolic pattern suggested an underlying AD pathology (CBS-AD), in agreement with the results of a recent combined ¹⁸FDG-PET/neuropathological study [13]. Notably, in all patients, the in vivo AD pathophysiological biomarkers supported the diagnosis of probable AD. Indeed, these biomarkers should always be looked for, together with the downstream degenerative topographical biomarkers (¹⁸FDG-PET, MRI), in atypical dementia cases.

The results of genetic analyses were, in our opinion, very interesting. The variant found in cases 1 and 2, PSEN1 Glu318Gly, was first identified in patients with EOAD [14]. Studies performed to define its effects on amyloid-β metabolism gave conflicting results [15, 16], and association studies were inconclusive [16, 17]. The variant disclosed in case 3, PSEN2 Arg71Trp, probably involved in protein stability and signalling pathways [18], has been found in patients with EOAD or LOAD, as well as in healthy subjects and Parkinson’s disease dementia [19, 20], and only in one large AD family it seemed to clearly segregate with the disease [21, 22]. It is possible that, by interacting with other factors, PSEN1 Glu318Gly and PSEN2 Arg71Trp increase disease risk and modulate clinical phenotype. PSEN2 Ser236Ser, present in case 4, is a silent variant whose pathogenicity is not predictable.

Among the relevant findings of NGS analysis, ABCA7 and SORL1 are well-known AD risk genes [23, 24]. The ABCA7 transporter is involved in Aβ clearance and its mutations accelerate amyloidosis in a mouse model of AD [25]. A strong association was demonstrated between ABCA7 variations and amyloidosis in AD patients [26]. A reduced expression of SORL1, promoter of the APP non-amyloidogenic pathway [27], has been demonstrated in human AD brains, and its genetic variants increase risk of both LOAD and EOAD [28]. In patient 1, we identified the ABCA7 Asp679Tyr and the SORL1 Thr833Ile variants. They had never been reported before but are predicted to be deleterious by in silico analyses, therefore possibly exerting a synergistic effect with the PSEN1 Glu318Gly variant in amyloidogenic process.

Patient 2, affected by LOAD with parkinsonism, harboured the Tyr239Cys variant in FUS, a gene implicated in ALS and FTD cases [29]. This variant is present in GnomAD with a very low frequency and is predicted to be deleterious by some in silico analyses.

In patient 5, we found the Iso196Val variant in DCTN1 gene. Several DCTN1 mutations have been described in association with ALS, degenerative parkinsonisms and Perry syndrome [30, 31]. Interestingly, our patient displayed some features of Perry syndrome at disease onset, such as personality change, and eating and sleep disturbances, while parkinsonism occurred thereafter. However, in vivo biomarkers more likely predicted amyloid-β rather than TDP-43 pathology, which is Perry syndrome’s substrate. Despite some evidence of pathogenicity from in vitro studies [32], DCTN1 Iso196Val variant has been reported both in patients and in several healthy controls, making it a possible risk factor rather than a causative mutation. This patient also presented the Ser10STOP variant in SORL1, which is a truncating variant absent in ExAc (Exome Aggregation Consortium, http://​exac.​broadinstitute.​org/​) and GnomAD databases, with a CADD score of 35: these types of variant are considered as definitely pathogenic and associated with a significant 12-fold increased AD risk, which is comparable with the APOE-ε4 homozygosity effect [33]. Rare pathogenic SORL1 mutations segregate with disease in LOAD families, and their pathological mechanism is likely to be haploinsufficiency [34].

In case 6, we found variants in other dementia-related genes. The novel Ala273Thr variant, predicted as damaging by in silico analysis, was identified in CSF1R. CSF1R mutations are causative of adult-onset leukoencephalopathy with axonal spheroids and pigmented glia [35], and have recently been reported in pathologically confirmed AD subjects [36]. Noteworthy, one of these cases exhibited a clinical picture very similar to that of our case. We can therefore hypothesize that rare variants of CSF1R may influence the susceptibility to AD, as already shown for other adult-onset leukodystrophy causative genes, such as TREM2 and NOTCH3 [37, 38]. Mutations in SERPINI1 are responsible for familial encephalopathy with neuroserpin inclusion bodies [39]. Though rapidly progressive dementia and myoclonus belong to the clinical spectrum of SERPINI1 mutations [40], the Ala280Thr variant found in patient 6 is predicted as tolerated by in silico analyses. Finally, the splicing mutation c.3529 + 5G > A identified in DCTN1 gene is predicted as potentially capable of altering the splicing site by in silico analyses; therefore, a possible pathogenic effect cannot be excluded.

Second, in this series of cases, it is also possible to highlight the very interesting aspects emerging from a wider than standard genetic analysis. We found the coexistence of more than one rare non-causative genetic variant in 4 out of 6 patients, suggesting an additive contribution of them to develop dementia, whereas each single variant may not be sufficient. This raises a crucial question: what is the role of these non-causative mutations that are increasingly found in different neurological disorders, particularly in dementias? One hypothesis is that they could act as risk or modifier factors to the disease. Further studies adding evidence from NGS data to the current knowledge will be necessary to support this hypothesis and to define the individual risk associated to each variant.