Neuropathologic Assessment of Dementia Markers in Identical and Fraternal Twins

Abstract

Twin studies are an incomparable source of investigation to shed light on genetic and non‐genetic components of neurodegenerative diseases, as Alzheimer's disease (AD). Detailed clinicopathologic correlations using twin longitudinal data and post‐mortem examinations are mostly missing. We describe clinical and pathologic findings of seven monozygotic (MZ) and dizygotic (DZ) twin pairs. Our findings show good agreement between clinical and pathologic diagnoses in the majority of the twin pairs, with greater neuropathologic concordance in MZ than DZ twins. Greater neuropathologic concordance was found for β‐amyloid than tau pathology within the pairs. ApoE4 was associated with higher β‐amyloid and earlier dementia onset, and importantly, higher frequency of other co‐occurring brain pathologies, regardless of the zygosity. Dementia onset, dementia duration, difference between twins in age at dementia onset and at death, did not correlate with AD pathology. These clinicopathologic correlations of older identical and fraternal twins support the relevance of genetic factors in AD, but not their sufficiency to determine the pathology, and consequently the disease, even in monozygotic twins. It is the interaction among genetic and non‐genetic risks which plays a major role in influencing, or probably determining, the degeneration of those brain circuits associated with pathology and cognitive deficits in AD.

Introduction

Twin‐cohort studies have historically represented an important source of analysis for investigations that aim to distinguish between genetic and non‐genetic components of diseases 13, 14, 36, 50, 53, 63. Studies on twins, either identical or non‐identical, represent indeed a natural and powerful model to estimate the relative impact of genetic and non‐genetic factors contributing to the clinicopathologic discrepancies often observed in neurological diseases, and especially, in neurodegenerative diseases. Even among twins with ascertained common and identical genetic risks, such as monozygotic (MZ) twins, clinicopathologic discordances can be observed 9, 71, 73. Longitudinal studies on twins, especially those including autopsy programs, are of extreme relevance to evaluate the levels of clinicopathologic agreement between subjects with identical genetics, initial environment in common, and shared genomic imprinting. Moreover, twin studies can help to identify risk factors for neurodegenerative diseases linked to aging, and specifically, for sporadic Alzheimer's disease (AD), nowadays the most frequent form of dementia worldwide, with enormous social and economic implications 30.

Earlier and more recent clinicopathologic correlation studies on AD 21, 22, 32, 33, 37, 56, 68, as well as series of neuroimaging studies 55, 64, 70, 72, have supported the notion that the histopathologic hallmarks of AD, mainly β‐amyloid neuritic plaques (Aβ‐NP) and tau neurofibrillary tangles (tau‐NFT), and the associated cognitive deficits in AD, are not linearly correlated. Often, even for a definite autopsy‐confirmed diagnosis of AD in subjects with a clinical diagnosis of AD, only probabilistic systems of classification, combining clinical and pathologic data can provide a sufficient level of consistency between observed brain pathology and history of dementia. Some of these combinational systems are, for example, the NIA‐Reagan Criteria 75 and the more recent “ABC” system 49.

This complex scenario reflects the fact that sporadic AD is, most probably, the result of mutual interactions among genetic predisposition, selective neuronal vulnerability and environmental influences 11. The twin‐study design provides unique tools to shed light on these major investigative challenges 20.

Studies of dementia focusing on clinicopathologic correlations and neuropathologic concordance between twins in a pair, using longitudinally collected clinical data, and neuropathologic postmortem findings, are largely lacking. This situation is mainly due to difficulties in collecting brains from both members of the same twin pair. Beyond a few case reports in the pre‐immunohistochemistry era 36, 63, very few neuropathologic investigations of twins with dementia have been published 17, 18, 59. Since then, newer clinical 47 and pathologic 31 diagnostic criteria of AD have been proposed. Furthermore, well‐established immunohistochemistry protocols and specific antibodies are now available 1, 4, 5.

The aims of this investigation were:

• (i) Evaluate if a higher clinical and pathologic concordance was present in MZ compared with dizygotic (DZ) twin pairs;
• (ii) Evaluate the neuropathologic concordance for AD pathology in each MZ and DZ twins pair, and observe if the neuropathologic concordance for the histopathologic hallmarks of AD (Aβ‐NP and tau‐NFT) could change using alternatively silver‐modified Bielschowsky stain or immunohistochemistry methods;
• (iii) Perform clinicopathologic correlations in octogenarian and older twins;
• (iv) Evaluate the impact of ApoE genotypes on AD pathology in MZ and DZ twins;
• (v) Assess the co‐occurring brain pathologies in these octogenarian and older twins.

To address these aims, the following steps were necessary:

• (i) Perform systematic neuropathologic assessments of both twin brains, either MZ or DZ, when one or both twins received a clinical diagnosis of dementia, and make definite diagnosis of AD following the most updated pathologic criteria for AD 31;
• (ii) Test the association between clinical and neuropathologic findings in MZ and DZ twins and their educational level, last Mini Mental Status Examination (MMSE) score, age at death, age at dementia onset, difference between twins in age at dementia onset, difference between twins in age at death and neuropsychiatric symptoms;
• (iii) Correlate clinical and neuropathologic data with specific ApoE genotypes, that is ApoE4 and non‐ApoE4 carrier twins;
• (iv) Assess the co‐occurring brain pathologies, histological markers of other types of dementia, such as: Lewy body dementia (DLB), Frontotemporal Dementia (FTD) and Vascular dementia (VaD) in older MZ and DZ twin brains using immunohistochemistry.

Material and Methods

The Swedish twin registry

This twin‐autopsy cohort was part of the autopsy program of the Study of Dementia in Swedish Twins 27, a derivative study on dementia based on the Swedish Twin Registry (STR). The STR started in the late 1950s to study smoking and alcohol consumption as risk factors for cardiovascular diseases and cancer 40. In 1984, a study on aging: the Swedish Adoption/Twin Study of Aging (SATSA) began 25. Soon after the SATSA study was launched, an ancillary Study of Dementia in Swedish Twins was proposed. The study became known as SALZA 28, 29. In 1998, another study on dementia was launched, HARMONY 27, that screened the entire STR.

Subject selection criteria

An electronic review of data records for all studies of aging twins from STR was performed. Main selection criteria were: (i) brain autopsy for both twins in the pair and (ii) one or both twins of the pair diagnosed with dementia before death. Among a total of 140 individual twin autopsies available, there were seven pairs of twins satisfying the selection criteria. One pair was evaluated as part of the SALZA study. Six pairs were evaluated by HARMONY assessment teams. Three of the pairs who participated in HARMONY had prior longitudinal data from the OCTO‐Twin study in which twin pairs aged 80 and older were followed longitudinally until death 44.

Clinical, cognitive and psychiatric assessment

Clinical diagnoses of dementia followed DSM‐III‐R 7 or DSM‐IV 8 criteria, reflecting the year in which the diagnosis was made. A category of “questionable dementia” was added for individuals who did not fulfill one of the first three diagnostic criteria, but did exhibit either cognitive impairment or functional disability in conjunction with memory problems. At the time of these clinical assessments, the clinical entity of mild cognitive impairment (MCI) was not as well characterized and defined as nowadays 54.

AD diagnoses were made according to NINCDS/ADRDA criteria 46. Briefly, each subject was diagnosed at a consensus conference attended by the clinician that examined the subject and the neuropsychologist or nurse that performed the cognitive assessments and met the family. A detailed description of the clinical evaluation methods used for each subject enrolled in the study is available in Gatz and Pedersen 28. However, for the present study, all previous clinical diagnoses based on NINCDS/ADRDA criteria were revised using the newer NIA‐AA clinical criteria 47. For MCI, VaD, FTD, DLB, Parkinson's disease dementia (PDD), the following clinical criteria were used respectively: MCI NIA‐AA criteria 6; NINDS‐AIREN criteria 60; Nomenclature for neuropathologic subtypes of frontotemporal lobar degeneration: consensus recommendations 41, and Nomenclature and nosology for neuropathologic subtypes of frontotemporal lobar degeneration: an update 42; Diagnosis and management of dementia with Lewy bodies: third report of the DLB consortium 45; and Diagnostic procedures for Parkinson's disease dementia: recommendations from the movement disorder society task force 24. Psychiatric symptoms were evaluated in HARMONY twins using the neuropsychiatric inventory (NPI) 23. In addition, we describe other relevant clinical diagnoses, and causes of death, as available from medical records and autopsy reports.

Neuropathologic assessment

A total of 14 post‐mortem brains were examined. All brains were stored and examined for standard neuropathologic assessment at the Brain Bank at the Karolinska Institutet, Karolinska University Hospital at Huddinge, Stockholm, Sweden. Each autopsy was obtained in accordance with Swedish law. All neuropathologic assessments were performed by neuropathology‐investigators (IN, DI) blind of the cognitive and zygosity status of the subjects.

At gross examination, all brains were free of macroscopic cerebrovascular lesions (extra‐ or intra‐parenchymal hemorrhages, softening or tissue discolorations), parenchymal or vascular malformations, aneurysms or evident neoplastic lesions. No regional atrophy was observed on the surface of the cerebrum or cerebellum in any brain. The main cerebral vessels (circle of Willis) presented various degrees of atherosclerotic disease severity, ranging from no signs of atherosclerosis to 50% of arterial occlusion on one or both middle cerebral arteries and basilar artery. After weighing, the entire brain was fixed in 10% buffered formaldehyde for at least 2 weeks, and then cut coronally. From the left hemisphere, tissue blocks were dissected from the following cerebral regions: middle frontal gyrus (MFG), frontal prepolar gyrus, orbitofrontal gyrus, parietal cortex (PC; lobulus inferior), medial temporal gyrus (MTG), occipital cortex (OC; including area striata and prestriata), anterior cingulate gyrus (ACG), pre‐ and post‐central gyrus (ppCG), basal ganglia (BG; including thal and subthal), basal forebrain (BF), anterior hippocampus (AH), posterior hippocampus (PH, at the level of the corpus geniculate lateralis), mamillary bodies and hypothalamus (MAM), vermis, cerebellar cortex and nucleus dentatus (CRBL), mesencephalon [MES, including the Substantia Nigra (SN)], medulla oblongata, medulla spinalis, amygdala (AMY).

Tissue blocks were processed and embedded in paraffin, cut 7 μm‐thick and stained with hematoxylin and eosin (H&E) for standard neuropathologic evaluations. The histologic assessment of cerebrovascular pathologies focused mainly on the detection of lacunes (loss of parenchymal tissue detectable at gross examination), microinfarcts (infarcts detectable at microscopy only, not detectable at gross examination), microhemorrhages (hemorrhages detectable at microscopy only, not detectable at gross examination) and arteriovenous malformations. The cerebrovascular assessment was performed on H&E stain in all 20 cerebral regions.

Neuropathologic criteria and double CERAD and Braak scoring

To meet previous and more recent pathologic criteria for the diagnosis of definiteAD, each brain was assessed using CERAD 48 and NIA‐AA pathologic criteria 31. For NIA‐AA pathologic criteria, the “practical approach” was used 49. In addition to silver‐modified Bielschowsky's method 74, CERAD age‐related plaque and NFT‐Braak scores 15, were also obtained using immunohistochemistry protocols for β‐amyloid (plaques) and tau (NFT) pathology. Established protocols for assessment of AD pathology from BrainNet Europe (BNE) recommendations were followed 1, 5. This double pathologic scoring aimed to verify the level of consistency of neuropathologic AD changes between two different methods of staining.

The co‐occurring brain pathologies were assessed using immunohistochemistry protocols for: Lewy bodies (LB), TAR‐DNA binding protein‐43 (TDP‐43), phosphorylated‐TAR‐DNA binding protein‐43 (pTDP‐43), p62 protein (p62) and Ubiquitin (Ubiq) intraneuronal inclusions. In addition, for LB pathology, the staging/typing of LB‐related α‐synuclein pathology by BNE Consortium was also considered 3, 16. Furthermore, vascular alterations were also evaluated following some of the recommendations from Alafuzoff 2.

To optimize the amount of neuropathologic data available for the immunohistochemistry assessment, a restricted subset of 10 cerebral regions out of 20 available was considered. This subset included most of the regions necessary to satisfy the pathologic criteria used. This subset included the following cerebral regions: MFG, PC, MTG, OC, ACG, PH, CRBL, MES (including SN), PONS, AMY.

Briefly, after microtomic serial cutting, a 7 μm‐thick section, randomly chosen, was collected for each specific immunostain for each cerebral region. The primary antibodies used for this investigation are summarized in Table 1. Vector Elite Kit and Vector secondary antibodies (VECTOR Laboratories, Burlingame, CA, USA) were used for visualization of antibody reactions. Counterstain with hematoxylin was performed before sections were coverslipped.

Table 1.

Primary of antibodies used for the immunohistochemistry assessment of the twin brains

Antibody	Pretreatment	Dilution	Host	Source
Anti‐β amyloid (1–40)	99% Formic acid	1:1500	Rabbit	Donation from Dr. Jan Naslund
Anti‐β amyloid (1–42)	99% Formic acid	1:1500	Rabbit	Donation from Dr. Jan Naslund
Anti‐tau (clone AT8)	No	1:400	Mouse	Innogenetics, Gent, Belgium
Anti‐α‐synuclein	99% Formic acid	1:2000	Rabbit	Chemicon International, Temecula, CA, USA
Anti‐Ubiquitin	No	1:300	Rabbit	Dako, Glostrup, Denmark
Anti‐p62	Autoclaved 1 hour by 120°C in Diva Decloaker	1:1000	Guinea pig	Progen, Heidelberg, Germany
Anti‐TDP‐43	Autoclaved 1 hour by 120°C in Diva Decloaker	1:1000	Rabbit	Proteintech, Chicago, IL, USA
Anti‐phosphorylated TDP‐43	Autoclaved 1 hour by 120°C in Diva Decloaker	1:2000	Mouse	Cosmo Bio, Tokyo, Japan

The table describes the main features of the antibodies used to assess the presence of Alzheimer's disease pathology and other co‐occurring pathologies in a series of 10 cerebral regions from seven pairs of twin brains. Anti‐βamyloid antibody (both for 1–40 and 1–42) was a generous gift from Dr. Jan Naslund 51.

Neuropathologic concordance criteria

The neuropathologic concordance between twins of the same pair, for AD pathology, was considered full (neuropathologic concordance = yes) if identical CERAD age‐related plaque or NFT‐Braak scores were assessed in each twin.

Neuropathologic concordance rates, for AD pathology, were also calculated across all twin pairs, and, separately, for MZ and DZ twin pairs. In addition, neuropathologic concordances were assessed, separately, on silver‐modified Bielschowsky stain and immunohistochemistry.

ApoE genotyping

ApoE genotype was available from blood samples of all subjects. For newer samples, genotyping was performed using the Illumina GoldenGate assay system on Illumina BeadStation 500GX equipment (Illumina, Inc., San Diego, CA, USA), after all samples were subjected to whole genome amplification 10. Older samples were genotyped using Dynamic Allele Specific Hybridization 57. Twin pairs were clustered in three ApoE groups: ApoE2 carriers (group1: twin set‐1,‐2), no ApoE2 or ApoE4 carriers (group2: twin set‐3,‐4) and ApoE4 carriers (group3: twin set‐5,‐6,‐7).

Results

Demographic and clinical findings

Data on demographics, zygosity, sex, educational level, age at death, age at dementia onset, interval of time between diagnosis of dementia and death, difference between twins in age at dementia onset, difference between twins in age at death and cause of death are summarized in Table 2.

Table 2.

Zygosity, ApoE genotype and concurrent clinical diagnoses of seven pairs of twins

Twin ID	Zyg	Sex	Education	Age at death	Age at dementia onset	Interval dementia onset‐death	Difference between twins in age at dementia onset	Difference between twins in age at death	Cause of death
1a	MZ	M	5	98	85	13	7	0	Shock/Senility
1b		M	6	98	92	6			Septicemia/Thrombophlebitis lower limbs
2a	DZ	F	8	91	90	1	10	7	Stroke
2b		F	6	84	80	4			PD
3a	MZ	F	6	94	81.5	12.5	1	5	AD
3b		F	6	89	80.5	8.5			Chronic ischemic heart disease
4a	DZ	F	7	93	N/A	N/A	N/A	8	Chronic ischemic heart disease
4b		M	6	85	83.5	1.5			Prostate Cancer
5a	MZ	F	9	85	68.5	16.5	4.5	2	Atherosclerotic heart /ischemic disease/dementia
5b		F	6	87	73	14			Atherosclerosis/dementia /NPH
6a	DZ	M	6	79	71.5	7.5	1.5	0	Dementia
6b		F	6	79	70	9			Cerebrovascular disease/dementia
7a	DZ	M	6	88	88	0	5.5	2	Chronic ischemic heart disease
7b		M	6	90	82.5	7.5			Artery occlusions/dementia
	3MZ/4DZ	6M/8F	6.3 ± 1.0 (5–8)	88.5 ± 5.9 (79–98)	80.4 ± 7.6 (68.5–92)	7.7 ± 5.2 (1–16.5)	4.9 ± 3.3 (1–10)	3.4 ± 3.2 (0–8)

Education: years of formal school education attained. Age at dementia onset: age at diagnosis of dementia; interval dementia onset‐death: time between initial diagnosis of dementia and death; difference between twins in age at dementia onset: difference of the ages at dementia onset between twins in the same pair; difference between twins in age at death: interval of time between deaths of two twins in the same pair. Data about ages and time intervals are expressed in year. Abbreviations: AD = Alzheimer's disease; DZ = dizygotic; F = female; M = male; MZ = monozygotic; NPH: normal pressure hydrocephalus; PD = Parkinson's disease; Zyg = zygosity

Out of seven twin pairs, three were MZ and four were DZ. Among DZ pairs, two pairs were same sex (twin set‐2 and set‐7), while two were opposite sex (twin set‐4 and set‐6). Mean age at death and at dementia onset were closer in MZ (respectively, 3.5 ± 2.1 yearrs; 4.1 ± 3.0 years) than DZ twin pairs (respectively, 4.2 ± 3.8 years; 5.6 ± 4.2 years), with DZ twin pairs also having larger standard deviation (SD) values; however, and although intriguing, these findings did not reach statistical significance. Two out of seven twin pairs died the same year, that is, at an identical age, of respectively 98 (twin set‐1, MZ) and 79 (twin set‐6, DZ). These two twin pairs (set‐1 and set‐6) represented unexpected, very unique and stochastic events at clinical and neuropathologic level.

Data on ApoE genotypes, NINCDS/ADRDA and NIA‐AA clinical diagnoses and other relevant clinical diagnoses are shown in Table 3. All twins were diagnosed as having dementia using DSM criteria, except twin4a, assessed as cognitively normal at last cognitive evaluation. For this twin, however, the last cognitive evaluation was performed 5 years before the autopsy, making uncertain her final cognitive diagnosis. No twin was diagnosed as MCI, or as clinical probable DLB or FTD. Twin2b and twin3a were diagnosed with PDD, twin3b as possPD, possVaD and possFTD. In total, Twin2b, twin3a, twin4b were clinically diagnosed with PD and twin3b with parkinsonism.

Table 3.

Zygosity, ApoE genotype and concurrent clinical diagnoses of seven pairs of twins

Twin ID	Zyg	ApoE genotype	Clinical diagnosis (DSM‐IIIR/IV)	Clinical diagnosis (NINCDS/ADRDA)	Clinical diagnosis (NIA‐AA)	Other relevant clinical diagnosis
1a	MZ	23	Dementia	ProbAD	ProbAD	None
1b		23	Dementia	PossAD	PossAD	Hypertension
2a	DZ	23	Dementia	ProbAD	ProbAD	Ischemic stroke
2b		23	Dementia	Dementia in PD	Dementia in PD	Parkinson's disease (80), diabetes
3a	MZ	33	Dementia	ProbAD	ProbAD	Parkinson's disease
3b		33	Dementia	Dementia NOS (PossAD, PossVaD, PossFTD, possPDD)	PossAD	MI (at 86), epilepsy, hyperparathyroidism, renal failure, parkinsonism
4a	DZ	33	No Dementia	No dementia	No dementia	Hypertension
4b		33	Dementia	Poss AD	PossAD	Parkinson's disease, Prostate neoplasia
5a	MZ	44	Dementia	Hydrocephalus/PossAD	Hydrocephalus/PossAD	Dementia, gait disturbance, urinary incontinence, hypertension, pernicious anemia, GI disorders
5b		44	Dementia	Hydrocephalus/Dementia	Hydrocephalus/PossAD	dementia, gait disturbance, urinary incontinence, hypertension, pernicious anemia
6a	DZ	34	Dementia	ProbAD	ProbAD	Syncope, psychiatric disorder
6b		34	Dementia	PossAD	PossAD	AF (71), cerebral infarction (74)
7a	DZ	34	Dementia	VaD	VaD	Heart disease, diabetes, behavioral disorder, AVB
7b		44	Dementia	ProbAD	ProbAD	aortic aneurysm (81), atherosclerosis of extremities (87), melanoma
	3MZ/4DZ

Numbers between parentheses after other relevant clinical diagnosis indicate the age at the moment of the diagnosis. Abbreviations: AD = Alzheimer's disease; AF = atrial fibrillation; AVB = atrio‐ventricular block; Dementia NOS = dementia not otherwise specified; DZ = dizygotic; F = female; FTD = frontotemporal dementia; GI disorders = gastro‐intestinal disorders; M = male; MI = myocardial infarct; MZ = monozygotic; PD = Parkinson's disease; PDD = Parkinson's disease with dementia; Poss AD = possible AD; Prob AD = probable AD; VaD = vascular dementia; Zyg = zygosity

Good agreement between clinical diagnoses based on NINCDS/ADRDA and NIA‐AA clinical criteria was observed. The results showed, indeed, agreement for all five diagnoses of probAD, with twin3a having an adjunctive diagnosis of PD, and for eight cases diagnosed possAD. The remaining case was the twin who was not demented at last cognitive evaluation 5 years before autopsy (twin4a).

Neuropathologic findings

The mean brain weight at autopsy, across all subjects, was 1041.4 ± 101.7 g. An extensive neuropathologic assessment was performed on each brain, using histological (H&E), silver‐modified Bielschowsky's method (CERAD and Braak scores) and immunohistochemistry methods. In detail:

Twin set‐1

Twin a. No vascular lesions were observed in any of the 20 cerebral regions examined. Corpora amylacea were observed in hippocampal‐parahippocampal region. Thal phase: 2. Braak stage: IV. CERAD score: moderate. No LB pathology was observed in any considered region. Based on CERAD and NIA‐AA clinical criteria the diagnosis of probAD was confirmed after post‐mortem assessment. Ubiq inclusions were present in MFG, PC, ACG and with the highest density also in MTG; p62 inclusions had similar distribution to the Ubiq inclusions, although with higher frequencies, and were also present in PH, MES and AMY; TDP‐43 and pTDP‐43 inclusions were present across most of the examined regions.

Twin b. An isolated lacune was found in MTG. Marked perivascular enlargement in BG region was observed. No other vascular lesions were present in the remaining regions. Thal phase: 2. Braak stage: IV. CERAD score: moderate. No LB pathology was observed in any considered region. Based on CERAD and NIA‐AA clinical criteria, the diagnosis of possAD was changed in probAD after post‐mortem assessment. Ubiq inclusions were present in most of the examined regions, except in MFG, MES and CRBL; p62 inclusions showed comparable distribution as Ubiq inclusions, but were present also in MES; TDP‐43 and pTDP‐43 inclusions were present in PH and AMY.

The detailed pathologic scores for each type of co‐occurring pathology for twin set‐1 are summarized in Table 4 and Supporting Information Table S1. See also Figures 1 and 2.

Table 4.

ApoE genotypes, Lewy body pathology and co‐occurring brain pathologies in seven pairs of octo‐ and nonagenarian twins

Twin ID	ApoE genotype	LB pathology	DLB diagnosis	Ubiq	p62	TDP‐43	pTDP‐43
1a	23	Neg	No DLB	Mod in MTG	Mod in MFG, PC, MTG, ACG, MES, AMY	All Freq except in ACG, PH	Mod in MFG, PC, MTG, ACG, PH, MES, AMY
1b	23	Neg	No DLB	Freq in PH, PONS	Freq in PH, AMY	Mod in PH, AMY	Mod in PH, AMY
2a	23	Neg	No DLB	Neg	Freq in MES	Neg	Neg
2b	23	Sp in MES	Brainstem type	Neg	Neg	Neg	Neg
3a	33	Sp in AMY, ACG	Limbic type	Sp in PH, AMY	Freq in PC, MTG, OC, ACG, PH, PONS	Neg	Freq in AMY
3b	33	Neg	No DLB	Freq in PH	Freq in PH	Neg	Neg
4a	33	Neg	No DLB	Sp in AMY	Mod in MFG, AMY	Neg	Neg
4b	33	Sp in SN, PONS, AMY, PH, ACG	Limbic type	Mod in AMY	Mod in AMY	Neg	Neg
5a	44	Neg	No DLB	Freq in ACG, PH, AMY	Freq in PH, AMY	Freq in AMY	Freq in AMY
5b	44	Neg	No DLB	Sp in MTG, ACG, PH, MES	Freq in PH, AMY	Mod in AMY	Mod in AMY
6a	34	Neg	No DLB	All Freq except in OC, CRBL, MES, PONS	All Freq except in OC, CRBL, MES, PONS	Freq in AMY	Freq in AMY
6b	34	Freq in SN, mod in OC, PH, Sp in MTG, AMY	Cortical type	All Freq except in MFG, CRBL, PONS	All Freq except in MFG, CRBL, MES, PONS	Freq in AMY	Freq in AMY
7a	34	Neg	No DLB	Freq in MFG, PC, ACG, AMY	Freq in MFG, PC, MTG, AMY	Freq in AMY	Freq in AMY
7b	44	Sp in AMY, ACG	Brainstem type	Mod in PH, AMY	Freq in AMY	Mod in AMY	Freq in AMY

Lewy bodies (LB) and other intraneuronal lesions such as Ubiq (Ubiquitin), p62 (p62 protein), TDP‐43 (TAR‐DNA binding protein‐43), pTDP‐43 (phosphorylated TDP‐43) intraneuronal inclusions were scored as negative (Neg) if no intraneuronal inclusions were present across all cellular types of the entire cortical‐thickness section, as observed through light‐microscopic inspection performed with 40× and 100× oil immersion (if necessary) objectives. The frequency of the intraneuronal inclusions was defined as: sp = sparse if <2 neurons, mod = moderate if 3–6 neurons, freq = frequent if >6 neurons were found positive at immunohistochemistry for each considered antibody. The intraneuronal inclusion frequency was established on the microscopic field (at 40× objective) that contained the highest number of positive neurons with positive inclusions for each considered antibody across the entire section of each anatomical region. The table describes only the anatomical regions with highest frequency scores for each type of inclusion. For details, see the Supporting Information section.

Figure 1.

The figure shows sections of the hippocampi (posterior hippocampi) obtained from both twins set‐1: 98‐year‐old monozygotic (MZ) twins, homozygotic for ApoE2/3, died at the same age. Digital microphotographs were taken at crescent level of magnification using 10× (upper parts), 40x (lower parts) and 100× oil immersion (*) objectives: this latter objective was used to assess and photograph intraneuronal pathologies such as, for example, pTDP‐43 intraneuronal inclusions or Lewy bodies. Single sections of the hippocampus for each twin were separately immunostained for 1–42 β‐amyloid, tau, α‐syn (α‐synuclein), Ubiq (ubiquitin), p62, TDP‐43 and pTDP‐43 (phosphorylated TDP‐43).

Notice the low levels of AD and non‐AD pathologies: 1–42 β‐amyloid, tau, α‐syn, ubiq, pTDP‐43 pathology (shown in the figure), and TDP‐43 and p62 inclusions (not shown in the figure; see Supporting Information section) in the hippocampi of these almost centenarian twins.

The black bar in the corners of each image corresponds to a scale bar, respectively, of 100 μm (for 10× and 40× objectives), and 10 μ (for 100× oil immersion objective).

Figure 2.

The figure shows sections of the amygdala obtained from both twins set‐1: 98‐year‐old monozygotic (MZ) twins, homozygotic for ApoE2/3, died at the same age. Digital microphotographs were taken at crescent level of magnification using 10× (upper parts), 40× (lower parts) and 100× oil immersion (*) objectives: this latter objective was used to assess and photograph intraneuronal pathologies such as, for example, pTDP‐43 intraneuronal inclusions or Lewy bodies. Single sections of the amygdala for each twin were separately immunostained for 1–42 β‐amyloid, tau, α‐syn (α‐synuclein), Ubiq (ubiquitin), p62, TDP‐43 and pTDP‐43 (phosphorylated TDP‐43).

Notice the low levels of AD and non‐AD pathologies: 1–42 β‐amyloid, tau, α‐syn, ubiq, pTDP‐43 (shown in the figure) and p62 and TDP‐43 inclusions (not shown in this figure; see Supporting Information section) in the amygadala of these almost centernarian twins.

The black bar in the corners of each image corresponds to a scale bar, respectively, of 100 μm (for 10× and 40× objectives), and 10 μ (for 100× oil immersion objective).

Twin set‐2

Twin a. No vascular lesions were found in any of the 20 cerebral regions examined, except in AMY (old infarct). Thal phase: 0. Braak stage: I. CERAD score: none. No LB pathology was observed in any considered region. The CERAD and NIA‐AA clinical diagnosis of probAD was not confirmed after autopsy. None of the co‐occurring pathologic lesions considered in this study (Ubiq, TDP‐43, pTDP‐43 inclusions) was present, except for frequent p62 inclusions in MES and ACG (Table 4).

Twin b. No vascular lesions were found in any of the 20 cerebral regions examined. Thal phase: 0. Braak stage: 0. CERAD score: none. Sparse LB and Lewy neurites (LN) were found in MES, making a diagnosis of possDLB of brainstem type for McKeith's criteria, and of brainstem category, Braak stage 3 (PD‐Braak staging) 16, for the LB‐related α‐syn pathology staging/typing of BNE Consortium. This subject received a diagnosis of PD years before the appearance of the cognitive deficits, making possible a clinical diagnosis of PDD. This twin was confirmed having LB pathology compatible with diagnosis of PD and normal for AD pathology, at autopsy. None of the other co‐occurring pathologic lesions considered in this study (Ubiq, p62, TDP‐43, pTDP‐43 inclusions) were present.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐2 are summarized in Table 4 and Supporting Information Table S2.

Twin set‐3

Twin a. Lacunes were found in MFG, MTG, OC and BG; no other vascular lesions were observed in the remaining examined regions. Diffuse perivascular enlargement in most of the examined regions was observed. Thal phase: 2. Braak stage: V. CERAD score: frequent. Sparse LB and LN were observed in AMY and ACG, making a diagnosis of possDLB of limbic type for McKeith's criteria, and of limbic category as well (no AMY predominant), Braak stage 5 16, for the LB‐related α‐syn pathology staging/typing of BNE Consortium. This twin presented multiple neuropathologic diagnoses: definiteAD, DLB and VaD. CERAD and NIA‐AA clinical diagnoses of probAD was confirmed after post‐mortem assessment. Ubiq inclusions were present only in PH and AMY; by contrast, p62 inclusions were more frequent and present in most of the examined regions. pTDP‐43 inclusions were present in PH and AMY.

Twin b. Lacunes and multiple microinfarcts were found in CRBL, no other vascular lesions were observed in the remaining cerebral regions. Thal phase: 2. Braak stage: IV. CERAD score: moderate. No LB pathology was observed in any examined regions. Although a vascular pathology component was present, this twin was clinically diagnosed as dementia not otherwise specified and various clinical diagnoses could be made: possAD, possVaD, possFTD. The neuropathologic post‐mortem assessment, however, confirmed that this twin was probAD with vascular component contributing to dementia; presence of TDP‐43 and pTDP‐43 inclusions was excluded. Frequent Ubiq in PH and p62 inclusions were present in PH and other examined regions.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐3 are summarized in Table 4 and Supporting Information Table S3.

Twin set‐4

Twin a. No vascular lesions were observed in any of the 20 cerebral regions examined. Thal phase: 1. Braak stage: III. CERAD score: frequent. No LB pathology was observed in any examined regions. Although CERAD score was frequent, the patient did not have a diagnosis of probAD or possAD at the time of the last cognitive assessment (5 years before death). Due to the long interval between last cognitive assessment and autopsy, this twin was diagnosed as possAD based on CERAD pathologic criteria. Ubiq inclusions were present in AMY, while p62 inclusions were spread in most of the examined regions.

Twin b. No vascular lesions were observed in any of the 20 cerebral regions examined. Thal phase: 2. Braak stage: 0. CERAD score: moderate. Sparse LB pathology was found in ACG, PH, MES, PONS and AMY making a diagnosis of possDLB of limbic type for Mckeith's criteria, and of limbic category (no AMY predominant), Braak stage 5 16, for the LB‐related α‐syn pathology staging/typing of BNE Consortium. The diagnosis of possAD was changed in probAD + DLB because of the co‐presence of AD and LB pathology. Ubiq and p62 inclusions were present in AMY. Curiously, at immunohistochemistry, the very rare tau‐NFT and tau pre‐tangles were observed only in PH and AMY, without exactly following the predicted tau‐NFT progression proposed by Braak 15.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐4 are summarized in Table 4 and Supporting Information Table S4.

Twin set‐5

Twin a. No vascular lesions were observed in any of the 20 cerebral regions examined. Thal phase: 2. Braak stage: II. CERAD score: sparse. No LB pathology was observed in any examined regions. Marked generalized ventricular enlargement as for hydrocephalus was observed. Based on CERAD pathologic criteria, this twin was diagnosed as normal for AD pathology. However, Ubiq and p62 inclusions were present in different regions (Table 4). Positivity for p62 inclusions was present in all examined regions; while positivity for Ubiq inclusions was present mainly in the cortical regions. TDP‐43 and pTDP‐43 inclusions were observed in MTG and AMY.

The cognitive deficits present in this twin could be due to the hydrocephalus, or to the cortical intraneuronal inclusions, observed at autopsy, or both. Clinically, this patient presented the triad of dementia, gait disturbance and urinary incontinence at the first visit, and she continued to progress at each visit. The diagnosis of hydrocephalus was supported by a brain computed tomography (CT)‐scan showing widened ventricles, reabsorption and white matter atrophy. The EEG showed some abnormalities. At the first longitudinal visit, the consensus added a diagnosis of dementia of Alzheimer type (DAT) (possAD for CERAD), but stated that the hydrocephalus remained the primary diagnosis. The initial neuropsychological profile was not consistent with DAT, that is, with well‐preserved spatial ability and personal hygiene, memory span fairly good, no insight, marked disinhibition and aggression (as for frontal symptoms). Medical records showed also diagnosis of pernicious anemia, hypertension and gastrointestinal disorders (Table 3).

Twin b. Lacunes and microinfarcts were found in BG, BF, MAM, SN and AMY, no further vascular lesions were observed in the remaining cerebral regions examined. Thal phase: 2. Braak stage: IV. CERAD score: moderate. No LB pathology was observed in any examined regions. Marked generalized ventricular enlargement as for hydrocephalus was observed. This twin was diagnosed as probAD based on CERAD pathologic criteria. Positivity for p62 inclusions was observed in all examined cortical areas, except CRBL (with a very similar distribution to twin5a). Positive Ubiq inclusions were present in a minor number of regions (MTG, ACG, PH, MES) compared with p62 inclusions. No TDP‐43 or pTDP‐43 neuronal inclusions were present in this twin, except in AMY (Table 4).

The cognitive deficits in this twin could be due to hydrocephalus with vascular and AD pathology contributing to dementia. This twin also presented the classic clinical triad of dementia, gait disturbance and urinary incontinence at the first visit, and she continued to progress on the other visits. Longitudinally, the language was preserved, but by the fourth visit, dementia had progressed and communication was not possible. A brain CT‐scan showed low pressure hydrocephalus with possible blockage. The EEG showed clear pathology. Medical records showed also diagnosis of pernicious anemia and hypertension.

Moreover, cerebral amyloid angiopathy (CAA) positivity (1–42 β‐amyloid immunohistochemistry) was observed for both twins in different regions across the brain (Table 5).

Table 5.

Cerebrovascular lesions and CAA assessed in the twin brains

Twin ID	Zyg	ApoE Genotype	Vascular lesions	CAA‐assessment Congo red (MFG, OC, CRBL cortex only)	CAA‐assessment Aβ1–42 (all examined regions)
1a	MZ	23	Absent	Neg	Pos in MFG, MTG, OC
1b		23	Lacune in MTG	Neg	Neg
2a	DZ	23	Absent	Neg	Neg
2b		23	Absent	Neg	Neg
3a	MZ	33	Lacunes in MFG, MTG, OC, BG	Neg	Neg
3b		33	Lacunes and microinfarcts in CRBL	Neg	Neg
4a	DZ	33	Absent	Neg	Neg
4b		33	Absent	Neg	Neg
5a	MZ	44	Absent	Pos in MFG, OC	Pos in MFG, PC, MTG, OC, ACG, PH, cereb cortex
5b		44	Lacunes and microinfarcts in BG, BF, MAM, SN, AMY	Pos in MFG, OC	Pos in MFG, PC, MTG, OC, PH, cereb cortex
6a	DZ	34	Absent	Pos in MFG, OC, Cereb cortex	Pos in Pos in MFG, PC, MTG, OC, ACG, PH, AMY
6b		34	Microinfarcts in ppCG, AH, PH	Neg	Pos in PC, MTG
7a	DZ	34	Lacunes in MFG, OC, ppCG, BG; microinfarcts in AH, PH, CRBL, SN	Neg	Pos in MFG, PC, MTG, OC, ACG, PH, AMY
7b		44	Absent	Neg	Neg

The table shows the main vascular microscopic findings (lacunes, microinfarcts, micro‐hemorrhages) observed across all 20 cerebral regions selected for this investigation and analyzed with hematoxylin and eosin. In addition, cerebral amyloid angiopathy (CAA) was assessed by immunochemistry with 1–42β‐amyloid antibody (Aβ1–42), on the restricted set of 10 cerebral regions, the same was used for the immunohistochemistry assessment of this twin‐autopsy cohort. CAA was also evaluated on Congo red only on middle frontal gyrus (MFG), occipital cortex (OC) and cerebellar cortex (CRBL) only.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐5 are summarized in Table 4 and Supporting Information Table S5.

At our knowledge, p62 and Ubiq intraneuronal inclusions represent unspecific signals of pathology, probably linked to more complex subjacent neurodegenerative molecular mechanisms. However, because of the possible pathogenetic relevance of the association between the observed co‐occurring, non‐AD, pathologies (mainly p62, Ubiq, but also TDP‐43 and pTDP‐43 inclusions) and hydrocephalus, in a MZ twin pair, we cannot exclude and hypothesize common, or distinct, pathomechanisms between hydrocephalus and these non‐AD pathologies. Importantly, larger studies needed to confirm these initial observations.

Furthermore, these twins (twin set‐5), were not only MZ, but also homozygote for the ApoE4 (ApoE4/4). Intriguingly, however, these twins (twin set‐5), at a mean age of 86, showed lower CERAD scores if compared with other twin sets with only one ApoE4 allele, such as twin set‐6, with a mean age of 79 (that is, 7 years younger than twins set‐5), DZ twins, and heterozygote for ApoE4 (ApoE3/4); or such as twin set‐7, mean age of 89 (that is, 3 years older than set‐5) DZ twins, with one twin heterozygote for ApoE4 (ApoE3/4), and the other twin homozygote for ApoE4. Intriguingly, both twins of set‐5 were positive for all co‐occurring non‐AD pathologies assessed in this study, except for LB pathology.

At our knowledge, this is the first documented report of ultra‐octogenarian MZ twins, homozygote for ApoE4 (ApoE4/4), which received post‐mortem confirmation of normal pressure hydrocephalus associated with the classic clinical triad of dementia, gait disturbance and urinary incontinence, and extensive neuropathologic assessment for AD and non‐AD pathologies, using immunohistochemistry. We underline that these relevant, and possibly novel observations, need to be confirmed with future larger studies.

Twin set‐6

Twin a. No vascular lesions were observed in any of the 20 cerebral regions examined. Cerebral vessel walls showed β‐amyloid positivity (1–42 β‐amyloid), with higher degree in the more superficial cortical than intracerebral vessels. Thal phase: 2. Braak stage: V. CERAD score: frequent. No LB pathology was observed in any examined regions. Based on CERAD and NIA‐AA clinical criteria, the diagnosis of probAD was changed in definiteAD after post‐mortem assessment. Ubiq and p62 inclusions were present in most of the examined regions. TDP‐43 inclusions were observed in AMY, and pTDP‐43 inclusions in AMY, MTG, ACG and PH.

Twin b. Microinfarcts were found in ppCG, AH and PH; no vascular lesions in the other remaining examined regions were observed. Thal phase: 2. Braak stage: V. CERAD score: frequent. LB pathology was found: frequent LBs in MES, moderate in OC and PH, sparse in MTG and AMY. This twin received a pathological diagnosis of DLB of cortical type for McKeith's criteria, and of cortical category, Braak stage 5 16, for the LB‐related α‐syn pathology staging/typing of BNE Consortium; and of definiteAD with cerebrovascular disease contributing to dementia. Ubiq and p62 inclusions were present in most of the examined regions (except PONS). TDP‐43 inclusions were found in AMY and pTDP‐43 inclusions were observed in PH and AMY.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐6 are summarized in Table 4 and Supporting Information Table S6. See also Figures 3 and 4.

Figure 3.

The figure shows sections of the hippocampi obtained from both twins set‐6: 79‐year‐old dizygotic (DZ) twins, heterozygotic for ApoE3/4, died at the same age, as the twins of set‐1. Digital microphotographs were taken at crescent level of magnification using 10× (upper parts), 40× (lower parts) and 100× oil immersion (*): this latter objective was used to assess and photograph intraneuronal pathologies such as, for example, pTDP‐43 intraneuronal inclusions or Lewy bodies. Single sections of the hippocampus for each twin were separately immunostained for 1–42 β‐amyloid, tau, α‐syn (α‐synuclein), Ubiq (ubiquitin), p62, TDP‐43 and pTDP‐43 (phosphorylated TDP‐43).

Notice the higher levels and variability of AD pathology (1–42 β‐amyloid and tau lesions), co‐occurring non‐AD pathologies (α‐syn, ubiq and pTDP‐43 lesions, shown in the figure) and TDP‐43 and p62 inclusions (not shown in the figure; see Supporting Information section) in this twin set (set‐6) compared with the lower and less variable levels of AD and non‐AD pathologies observed in twin set‐1 for the same region: the hippocampus.

The black bar in the corners of each image corresponds to a scale bar, respectively, of 100 μm (for 10× and 40× objectives) and 10 μ (for 100× oil immersion objective).

Figure 4.

The figure shows sections of the amygdala obtained from both twin set‐6: 79‐year‐old dizygotic (DZ) twins, heterozygotic for ApoE3/4, died at the same age, as the twins of set‐1. Digital mirophotographs were taken at crescent level of magnification using 10× (upper parts), 40× (lower parts) and 100× oil immersion (*): this latter objective was used to assess and photograph intraneuronal pathologies such as, for example, pTDP‐43 intraneuronal inclusions or Lewy bodies. Single sections of the hippocampus for each twin were separately immunostained for 1–42 β‐amyloid, tau, α‐syn (α‐synuclein), Ubiq (ubiquitin), p62, TDP‐43 and pTDP‐43 (phosphorylated TDP‐43).

Notice the higher levels and variability of AD pathology (1–42 β‐amyloid and tau lesions), co‐occurring non‐AD pathologies (α‐syn, ubiq and pTDP‐43, shown in the figure) and TDP‐43 and p62 inclusions (not shown in the figure; see Supporting Information section) in this twin set (set‐6) compared with the lower and less variable levels of AD and non‐AD pathologies observed in twin set‐1 for this other same region: the amygadala.

The black bar in the corners of each image corresponds to a scale bar, respectively, of 100 μm (for 10× and 40× objectives), and 10 μ (for 100× oil immersion objective).

Twin set‐7

Twin a. Lacunes in MFG, OC, ppCG, BG and microinfarcts in AH, PH, CRBL and MES were found; no other vascular lesions were observed in the remaining cerebral regions examined. Diffuse arteriolosclerosis, especially in the more anterior regions (MFG and PC) of the brain was observed. Thal phase: 2. Braak stage: V. CERAD score: frequent. No LB pathology was observed in any examined regions. Due to the relevance of the vascular pathology, already documented at clinical level and confirmed at autopsy, this case was diagnosed as VaD and definiteAD. Ubiq and p62 inclusions were present in most of the examined regions, while TDP‐43 and pTDP‐43 inclusions were observed in AMY only.

Twin b. No vascular lesions were observed in any of the 20 cerebral regions examined. Perivascular enlargement diffuse through all the examined regions (especially in the white matter) was observed. Thal phase: 2. Braak stage: V. CERAD score: frequent. Sparse LB and very rare LN were observed in ACG and MES, making a diagnosis of possDLB of limbic type for McKeith's criteria, and of limbic category (without positivity in AMY) for the LB‐related α‐syn pathology staging/typing of BNE Consortium. Based on CERAD and NIA‐AA clinical criteria the diagnosis of probAD was changed in definiteAD based on CERAD pathologic criteria. Ubiq and p62 inclusions were present in most of the regions, TDP‐43 and pTDP‐43 inclusions were present only in AMY.

The detailed pathologic scores for each type of co‐occurring pathologies for twin set‐7 are summarized in Table 4 and Supporting Information Table S7.

No cases of hippocampal sclerosis were observed in this twin sample after microscopic assessment on H&E stain at lower (5× objective) and higher magnification (20× objective).

Table 6 summarizes ApoE genotypes, CERAD age‐related plaque and NFT‐Braak scores, CERAD pathological diagnoses and NIA‐Reagan criteria stages for each twin. To note, based on the NIA‐Reagan criteria, the probability that dementia was caused by amyloid and tau lesions was low or intermediate in ApoE2 carriers, whereas intermediate/high in non‐ApoE2 carriers, with the very intriguing exception of the twin set‐5, MZ and homozygote for ApoE4; and where twin5a was normal for AD pathology and twin5b possAD based on the CERAD criteria. This twin set (set‐5) resulted to be one of the rare post‐mortem evidence in support of the coexistence of environmental factors able to modify the accumulation of AD pathology, even in presence of an established risk factor such as homozygosis for ApoE4. It is important to emphasize that both twins of set‐5 were diagnosed with hydrocephalus, and presented the classic clinical triad associated with hydrocephalus: dementia, gait disturbance and urinary incontinence. These clinical features make this twin set particularly intriguing because of the presence of ApoE4 in MZ twins, the almost absence of AD pathology, and the presence of other co‐occurring non‐AD pathologies.

Table 6.

ApoE genotypes, AD pathology scores, CERAD and NIA‐Reagan diagnosis of seven pairs of twins

Twin ID	ApoE genotype	CERAD age‐related plaques score	CERAD diagnosis	NFT‐Braak stage	NIA‐Reagan criteria
1a	23	B	ProbAD	IV	Intermediate
1b	23	B	ProbAD	IV	Intermediate
2a	23	0	Normal for AD	I	Low
2b	23	0	Normal for AD	0	Low
3a	33	C	DefiniteAD + VaD	V	High
3b	33	B	ProbAD + VaD	IV	Intermediate
4a	33	C	PossAD	III	Intermediate/High
4b	33	B	PossAD	0	Low/Intermediate
5a	44	A	Normal for AD	II	Low
5b	44	B	ProbAD	IV	Intermediate
6a	34	C	DefiniteAD	V	High
6b	34	C	DefiniteAD	V	High
7a	34	C	DefiniteAD + VaD	V	High
7b	44	C	DefiniteAD	V	High

The table shows the CERAD age‐related plaque scores (0, A, B, C) and NFT‐Braak stage (0‐VI) based on silver‐modified Bielschowsky stain. The NIA‐Reagan criteria define levels of probability that dementia was due to AD pathology as assessed at autopsy. CERAD and Braak scores were based on silver‐modified Bielschowsky's method.

Table 5 shows the vascular and CAA findings for all twin cohort. We observed a significant association between ApoE4 and CAA. However, and although other studies 52, 58 are in support of this possible significant association, larger studies are necessary to confirm these further observations found in this twin‐cohort study.

Table 7 shows the neuropathologic “ABC” scores and AD neuropathologic changes in each twin as required by the NIA‐AA criteria 49. Both, NIA‐Reagan criteria and “ABC” system are based on the combination of amyloid and tau pathology scores, but the “ABC” system seems to reduce the level of severity across the same cases compared with the NIA‐Reagan criteria.

Table 7.

“ABC” scores and levels of AD neuropathologic changes

Twin ID	ABC score			Level of AD neuropathologic change
1a	A1	B2	C3	Intermediate
1b	A1	B2	C3	Intermediate
2a	A0	B1	C0	Not
2b	A0	B0	C0	Not
3a	A1	B3	C3	Intermediate
3b	A1	B2	C3	Intermediate
4a	A1	B2	C3	Intermediate
4b	A1	B1	C3	Low
5a	A1	B1	C1	Low
5b	A1	B2	C3	Intermediate
6a	A1	B3	C3	Intermediate
6b	A1	B3	C3	Intermediate
7a	A1	B3	C3	Intermediate
7b	A1	B3	C3	Intermediate

The table shows “ABC” scores and levels of neuropathologic AD changes based on NIA‐AA pathologic recommendations for AD. A = amyloid stage based on Thal classification phases; B = NFT stage based on the Braak system; C = neuritic plaques score based on the CERAD system. Levels of AD neuropathologic changes are described as in NIA‐AA pathologic criteria 31, 49.

CERAD and Braak scores were based on immunohistochemistry method in this table, instead then on silver‐modified Bielschowsky's method as in Table 6.

Co‐occurring brain pathologies in twins

Table 4 shows all co‐occurring brain pathologies observed at autopsy and semi‐quantitatively assessed in each MZ and DZ twin. As for the co‐occurring brain pathologies considered in this investigation (LB, Ubiq, p62, TDP‐43 and pTDP‐43 inclusions), a very high frequency of co‐occurring, non‐AD, brain pathologies was observed in both MZ and DZ twins. Intriguingly, lower and higher frequencies of co‐occurring brain pathologies were found, in twin respectively, set‐2 (DZ and ApoE2/3 twins), and in twin set‐6 (DZ and ApoE3/4 twins).

In general, Ubiq and p62 positivity were very useful to confirm presence and type of those inclusions or lesions such as neuritic plaques, tau‐tangles, LB and LN, TDP‐43 and pTDP‐43 inclusions that were observed using other specific immunostain.

Clinicopathologic correlations

In general, there was 84.6% agreement between clinical and pathologic diagnoses for AD (11 out of 13 cases with clinical dementia were confirmed having AD pathology, with different levels of severity, at autopsy). Based on CERAD pathologic criteria, eight of 10 cases who were clinically diagnosed as AD (poss or prob) showed consistent AD neuropathology changes at autopsy, while two did not. Two cases, not diagnosed as AD, showed indeed AD neuropathology changes as well: one diagnosed VaD (who showed both AD and VaD at autopsy), and one diagnosed with dementia secondary to hydrocephalus (who also had confirmed hydrocephalus at autopsy). One case showed clinicopathologic concordance with respect to being clinically diagnosed as PD showing LB pathology (brainstem type), but not AD pathology. Finally, as for twin4a, although the last clinical diagnosis was considered uncertain because of the long interval between last cognitive assessment and death (5 years), the post‐mortem diagnosis was possAD.

Educational levels, expressed as number of attained years of school, did not show significant correlation with clinical diagnosis, mean age at death, at dementia onset or with neuropathologic AD changes. Age of onset was later in those with an age‐related CERAD plaque score of B (mean age of onset = 82.8) compared with those with an age‐related plaque score of C (mean age at onset = 78.7).

MMSE score (last score before death) and psychiatric symptoms descriptions as defined by NPI are described in Table 8. Mean MMSE score before death, excluding the twin who was not demented at last evaluation, differed by CERAD age‐related plaque score and by NFT‐Braak stage in the expected direction, although none of the differences were statistically significant.

Table 8.

Zygosity, ApoE genotype, last MMSE and neuropsychiatric symptoms in an autopsy‐cohort of seven pairs of identical and fraternal twins. The table shows main documented psychiatric symptoms in twin pairs based on the Neuropsychiatric Inventory (NPI) 22

Twin ID	Zyg	ApoE genotype	Last MMSE (years before death)	NPI assessment (years before death)	Psychiatric symptoms (NPI items)	Symptom type (based on the NPI items)
1a	MZ	23	16 (5)	5	Yes	Delusions, hallucinations, apathy, eating disturbance
1b		23	14 (1)	4	Yes	apathy, lability, sleep disturbance, eating disturbance
2a	DZ	23	13 (1)	1	None	None
2b		23	0 (1)	N/A	N/A	N/A
3a	MZ	33	0 (5)	5	Yes	Delusions, agitation, lability, sleep disturbance
3b		33	11 (1)	1	Yes	apathy
4a	DZ	33	29 (5)	N/A	N/A	N/A
4b		33	11 (0)	0	Yes	delusions, hallucinations, apathy, motor disturbance
5a	MZ	44	14 (6)	N/A	N/A	N/A
5b		44	8 (8)	N/A	N/A	N/A
6a	DZ	34	9 (5)	5	Yes	Delusions, hallucinations, agitation, anxiety, apathy, sleep disturbance
6b		34	9 (4)	5	Yes	delusions, hallucinations, anxiety, apathy, motor disturbance, eating disturbance
7a	DZ	34	10 (1)	N/A	N/A	N/A
7b		44	12 (3)	N/A	N/A	N/A
	3MZ/4DZ		11.3 ± 6.9 (3.2 ± 2.4)	3.5 ± 2.1 (0–5)

Abbreviations: F = female; DZ = dizygotic; M = male; MZ = monozygotic; Zyg = zygosity. Last MMSE = last score of the Mini Mental Status Examination (MMSE) (0–30); between parentheses, after the MMSE score: years from last MMSE and death (0–8). NPI (years before death): interval of time (in years) from last NPI assessment date and death; 0 = same year

NPI scores were available for eight out of 14 twins only. Hallucinations and delusions, more typical of DLB, were present in twin3a, twin4b and twin6b. These cases were indeed autopsy‐confirmed cases of DLB of limbic or cortical type 3, 45. Moreover, other twins (twin1a, twin1b, twin3b, twin6a) showing concurrent psychiatric symptoms (apathy, eating disturbance, agitation) were mainly positive for other co‐occurring pathologies: ubiq, p62, TDP‐43 or pTDP‐43 inclusions, but not for LB pathology. In addition, those pathologic inclusions were localized in cerebral areas (ie frontal cortex and amygdala) more frequently associated with psychiatric disturbances.

ApoE genotype findings

Mean ages ± SD at death for ApoE group1, 2 and 3 were, respectively, 92.7 ± 6.7 years, 90.2 ± 4.1 years and 84.6 ± 4.6 years. The groups did not differ significantly for mean age at death. Mean ages ± SD at dementia onset in ApoE group1, 2 and 3 were, respectively, 86.7 ± 5.3 years, 81.3 ± 1.5 years and 75.5 ± 7.8 years. Group 3 (ApoE4 carriers) showed mean age at dementia onset of about 10 years earlier than group1 (ApoE2 carriers); these differences, however, were not statistically significant.

As for β‐amyloid pathology scores, concordance rates within pairs was higher for group 1 (ApoE2) and group 3 (ApoE4) than group 2 (ApoE3). Higher scores for CERAD age‐related plaques and for NFT‐Braak were observed in ApoE4‐carrier twins, especially compared with ApoE2 twins. Intriguingly, two out of three MZ twin pairs (twin set‐1, set‐5) were negative for LB pathology, and were respectively non‐ApoE4 carrier (set‐1) and ApoE4 carrier (set‐5).

Importantly, ApoE4 carriers were also positive for TDP‐43 and pTDP‐43 intraneuronal inclusions, and the majority of non‐ApoE4 carriers were negative for TDP‐43 and pTDP‐43 intraneuronal inclusions. These latter findings seem to imply an important association of ApoE4 allele not only for AD pathology (mainly β‐amyloid plaques), but also for other brain pathologies.

Within‐pair concordance in neuropathologic scores

Table 9 shows pairwise neuropathologic scores: CERAD age‐related plaque scores, NFT‐Braak stage based on Bielschowsky stain and immunohistochemistry, separately. CERAD pathologic diagnoses, NIA‐Reagan criteria stages (Table 6) and “ABC” scores (Table 7) showed good concordance within pairs, independently of their zygosity. CERAD age‐related plaque scores, NFT‐Braak stage and cerebro‐vascular pathologies (ie lacunes and microinfarcts) showed lower within‐pair concordance. Using either Bielschowsky or immunohistochemistry stains, there was neuropathologic concordance for one of three MZ pairs and for two of four DZ pairs. NFT‐Braak scores showed lower concordance if compared with CERAD age‐related plaque scores concordance. This lower concordance rate for NFT‐Braak scores was confirmed with Bielschowsky stain and with immunohistochemistry for tau. There was no significant correlation between: mean age at death, mean age at dementia onset, mean difference between twins in age at dementia onset and at death, with respect to β‐amyloid plaques and NFT‐Braak scores. Ubiquitin, p62, TDP‐43 and pTDP‐43 intraneuronal inclusions were mostly concordant in both MZ and DZ twins.

Table 9.

Concordance rates for AD pathology based on silver‐modified Bielschowsky stain and immunohistochemistry

Twin ID	Zyg	ApoE genotype	CERAD score (Biel)	CERAD Concordance (Biel)	CERAD score (Immuno)	CERAD Concordance (Immuno)	NFT‐Braak (Biel)	NFT Concordance (Biel)	NFT‐Braak (Immuno)	NFT Concordance (Immuno)
1a	MZ	23	B	yes	C	yes	IV	yes	IV	no
1b		23	B		C		IV		V
2a	DZ	23	0	yes	0	yes	I	no	I	no
2b		23	0		0		0		0
3a	MZ	33	C	no	C	yes	V	no	VI	no
3b		33	B		C		IV		IV
4a	DZ	33	C	no	C	no	III	no	IV	no
4b		33	B		B		0		I
5a	MZ	44	A	no	A	no	II	no	III	no
5b		44	B		C		IV		IV
6a	DZ	34	C	yes	C	yes	V	yes	V	no
6b		34	C		C		V		VI
7a	DZ	34	C	yes	C	yes	V	yes	VI	yes
7b		44	C		C		V		VI
				57%		71%		43%		14%

The table shows the neuropathologic concordance (= yes), for AD pathology, based on the neuritic plaques (Aβ‐NP) frequencies (CERAD pathologic criteria), and neurofibrillary tangles (tau‐NFT) stages (Braak system), per each twin of each pair. No = absence of neuropathologic concordance for AD pathology.

β‐amyloid neuritic plaques (Aβ‐NP; CERAD pathologic criteria) and NFT‐Braak (tau‐NFT; Braak staging system) scores were assessed using two different methods of histological stain.

Biel = silver‐modified Bielschowsky stain; Immuno: immunohistochemistry methods. Neuropathologic concordances are described for each histological stain, separately. For each histological method, and for each type of lesion (neuritic plaques and tangles), neuropathologic concordance rates are expressed as percentage of “yes” per twin pair out the total number of twin pairs.

Discussion

In this study, we performed clinicopathologic correlations in octogenarian and non‐agenarian MZ and DZ twins previously diagnosed with dementia (except one subject). We performed these correlations using longitudinally collected clinical data and extensive neuropathologic assessment for both twins in the pair. In addition, we verified the presence of non‐AD co‐occurring brain pathologies and examined the neuropathologic concordance rates in these MZ and DZ twins. We attempted also to correlate specific ApoE genotypes, that is ApoE4, an established risk factor for AD, with the neuropathologic findings.

In general, we found a good diagnostic agreement between CERAD and NIA‐AA clinical criteria for AD. Moreover, the neuropathologic assessment was able to confirm the presence of AD pathology in most cases, but a wider spectrum of pathologic co‐diagnoses was found. This was due to the high frequency of co‐occurring brain pathologies and cerebrovascular disease observed at autopsy. We observed that the co‐occurring pathologies were correlated with the ApoE genotype. Specifically, ApoE4 carriers showed higher levels of AD and non‐AD pathologies. Moreover, we observed that ApoE4‐carrier twins had shorter lives compared with ApoE3 and ApoE2 carrier twins, and that they had earlier dementia onset as well. Although not statistically significant and larger confirming twin studies are needed, these observations in a twin cohort of MZ and DZ twins with definite post‐mortem pathologic assessment are novel, and support the hypothesis that ApoE4 is a risk factor for accelerating AD onset 35, and shortening survival 61. ApoE4 allele as risk factor for AD pathology deposition (mainly, β‐amyloid neuritic plaques) has been already described 12; however, ApoE4 influence on AD deposition using the most updated pathologic criteria for AD and non‐AD markers of dementias in MZ and DZ twins has never been observed before. In addition, the notion that the severity of AD pathology follows a gradient going from the least severe degree associated with ApoE2, and the most severe degree with ApoE4 seems to be confirmed also in this twin‐cohort sample. Importantly, however, although MZ ApoE4‐carrier twins (set‐5) were more concordant than DZ for β‐amyloid and tau pathology, they did not have higher levels of β‐amyloid or tau pathology per se. This latter finding supports the more general hypothesis that there are indeed other factors, probably environmental, interacting with genetic risk factors capable to determine, accelerate or delay the appearance of dementia 25, 26, 34, 39, 66, 67, 69.

As a new observation, in a twins sample, is that ApoE4 is not only more frequently associated with higher β‐amyloid scores, but also with higher frequency of other, non‐AD and co‐occurring brain pathologies, as well as with CAA. For example, twin set‐2 (DZ, ApoE2 carriers) showed almost absence of pathology (except for freq p62 inclusions in MES in twin2a, and sparse LB in MES in twin2b) at a mean age of 87.5; even with interval between the deaths of the twins equal to 7 years. Similarly, twin set‐1 (MZ, ApoE2 carriers), at a remarkable age of 98, that is, even later in life (14 years later compared with the youngest twin of twin set‐2, twin2b) was assessed with moderate scores of CERAD age‐related plaques and absence of LB pathology.

Neuropathologic concordance rates in MZ and DZ twins, assessed using different methods of stain (silver‐modified Bielschowsky and immunohistochemistry) are among the further unique findings of this investigation. These rates confirm that AD, independently on the histological stain methods used, and with the consequent variable scores obtained, is indeed a multifactorial disease where genetic risks (ie ApoE4) and non‐genetic risks interact with each other 62, 65, and probably for decades before the clinical phase of the disease. These data from MZ and DZ twins show, furthermore, higher concordance rate for β‐amyloid (CERAD age‐related plaque scores) compared with tau pathology (NFT‐Braak scores), in both MZ and DZ twins. Moreover, pathologic concordance rates were not significantly related to the interval between the deaths of the twins, or their zygosity, but rather to the ApoE genotype. Although we cannot exclude that other genetic, biological or environmental risk factors may influence the pathologic and clinical evolution of AD, our findings are in agreement with a previous twin study 18 and support the hypothesis that a complex interaction between genetic and non‐genetic risk factors 32, 39, 62, and deposition of insoluble AD pathology exists. Furthermore, this complex interaction does not exclude an increased risk of deposition for other non‐AD brain pathologies as well as an increased level of soluble amyloid or tau in forms of oligomers. Among the unique findings of this investigation, and although only one twin pair was available, we described an autopsy‐confirmed case of MZ twins with normal pressure hydrocephalus and multiple brain pathologies. These twins were homozygous for ApoE4. If there is an association between normal pressure hydrocephalus, presence of ApoE4 allele and altered risk of deposition of AD pathology and, very importantly, of other non‐AD pathologies, is not known. We emphasize that these possible novel findings need and deserve further and larger investigations 19, 43.

The twin‐autopsy cohort here considered did not differ demographically from other longitudinal cohort studies with assessment of dementia 11, and this supports its comparability with other general population cohort studies. However, the specificity of this twin‐autopsy study, consisted in the opportunity to observe clinical and pathologic findings and confirm the existence of a possible interaction between genetic and non‐genetic risk factors 38, 50 in a natural model of aging/dementia, where three pairs of subjects, the MZ twin sets, that shared an identical genetic background for 8–9 decades or more, or subjects that shared, at least initially, the same parental and environmental imprinting, as in both MZ and DZ twins of this cohort. We emphasize although, that future studies with larger samples of twin pairs with longitudinally collected clinical data and autopsy‐confirmed diagnoses are necessary to confirm these initial observations.

To our knowledge, this is the first clinicopathologic correlation study that evaluated a consistent number of clinically well‐characterized older twin pairs that received a pathologic definite post‐mortem diagnosis using the most updated clinical and pathologic criteria for AD, an extensive immunohistochemistry assessment, for AD pathology as well as for other non‐AD co‐occurring pathologies in MZ and DZ octogenarian and non‐agenarian twins. Further and larger investigations, however, are necessary to shed light on the possible interaction between non‐modifiable (genetic) and modifiable (environmental) factors 62 that could delay or forestall the manifestation of AD, as well as other dementias, even in presence of established risk factors such as ApoE4, or co‐occurring brain pathologies, as the clinicopathologic findings described in this study suggest.

Supporting information

Tables S1–S7. These tables describe the AD and non‐AD co‐occurring pathologies frequencies observed in 10 different brain areas of each MZ and DZ twin pair: from twin set‐1 (Table S1) to twin set‐7 (Table S7) for a total of 14 brains. The frequencies were assessed on immunohistochemistry.