Transactive DNA Binding Protein 43 Rather Than Other Misfolded Proteins in the Brain is Associated with Islet Amyloid Polypeptide in Pancreas in Aged Subjects with Diabetes Mellitus

A link between diabetes mellitus (DM) related islet amyloid polypeptide (IAPP) and Alzheimer’s disease (AD) related amyloid-β (Aβ) has been suggested in epidemiological and clinical studies. In 2017, proof for existing interaction between type 2 DM and AD on a molecular level was provided based on research carried out in experimental animal models. We assessed aging-related neurodegenerative lesions, i.e., misfolded proteins, associated with dementia such as hyperphosphorylated τ (HPτ), Aβ, α-synuclein (αS), and phosphorylated transactive DNA binding protein 43 (pTDP43) seen in the brain and IAPP seen in the pancreas in subjects with and without DM applying immunohistochemical techniques. HPτ in the brain and IAPP in the pancreas were observed in most subjects. The prevalence and the extent of all misfolded proteins increased with age but this increase was not influenced by DM. Interestingly the extent of misfolded proteins in the brain was higher in non-diabetics when compared with diabetics in demented. A significant correlation was observed between HPτ, Aβ, αS, and pTDP43, whereas IAPP showed no association with HPτ, Aβ, and αS. In subjects with DM, the extent of pTDP43 in brain correlated with the extent of IAPP in pancreas. Thus, there is no evidence of a link between AD-related pathology and DM in humans, whereas an association was found between pTDP43 and IAPP in DM. TDP43 is ubiquitously expressed in all organs but whether TDP43 is phosphorylated in other organs in DM or whether the phosphorylation of TDP43 is influenced by glucose metabolism is yet unknown.


M. Leino et al. / TDP43 and IAPP in Aged Subjects with DM
these proteins is regionally predictable; thus, a reliable assessment of the extent of the pathologies can be carried out [2,6,7]. All these proteins can be seen simultaneously in one and the same subject, i.e., mixed pathologies [8,9].
In diabetes mellitus (DM), deposition of misfolded protein, i.e., islet amyloid polypeptide (IAPP), is observed in the pancreas [10,11]. IAPP is secreted by the islet cells, and the deposition of IAPP is associated with type 2 DM and in line with AD, type 2 DM is more common in the aged [10,12]. To our knowledge, there are no reports with human subjects indicating that aging in itself would be associated with deposition of IAPP in the pancreas. Thus, in aging misfolded proteins can be seen in both brain and pancreas as soluble and as insoluble aggregates [13]. The insoluble aggregates have been referred to as "hallmark" lesions of a certain disease and they can readily be visualized as immunoreactivity (IR) as is seen in Fig. 1, in the postmortem tissue, brain and pancreas, applying immunohistochemical (IHC) techniques [2,6,7,14].
Epidemiological studies have shown that there is an association between DM and dementia and thus DM has been suggested being a risk factor for AD [15][16][17]. It has been suggested that AD related and DM associated "pathologies" might influence each other and that a cross talk and/or cross-feeding might exist between the aggregation prone proteins [18].
In 2017, Moreno-Gonzales and colleagues reported that there is indeed a link between AD and type 2 DM, i.e., misfolded IAPP produced in pancreas in type 2 DM promoted AD pathology (A␤) by cross-seeding [19]. This was observed both in cell culture and in transgenic animals. Based on the above a question arises as to whether the misfolded protein in the pancreas, i.e., IAPP might influence the protein aggregation in the brain, i.e., ␤A, HPτ, ␣S, and pTDP43 and vice versa in humans.
Contrary to the above, neuropathological observations carried out on large human cohorts have not been able to show any link between AD and DM [20,21]. Moreover, recently in a clinical setting a lack of association was reported between biomarkers of neurodegeneration and type 2 DM [22].
The objective of this study was to investigate further the presumed association between the misfolded proteins in DM and neurodegeneration. We assessed the incidence and the extent of the insoluble aggregated proteins at the end stage, i.e., postmortem.
We assessed pathologies such as HPτ, A␤, ␣S, and pTDP43 observed in the brain tissue and the IAPP observed in the pancreas. The primary objective was to look for if any signs of association, i.e., cross talk and/or cross-seeding between these proteins could be detected. In addition, we looked for signs of colocalization of misfolded proteins in brain and pancreas.

Subjects
During a six-year long period (from 2010 to 2015), an autopsy including a neuropathological investigation was carried out on 615 subjects, age ranging from 48 to 102 years, at Uppsala University Hospital (Fig. 2). The medical records revealed that 80 of the deceased subjects (13%) had a clinical diagnosis of DM, age ranging from 48 to 95 years at death. In 74 (age range 48 to 87 years at death) of these 80 subjects, in addition to the brain tissue the pancreatic tissue was available for this study. An age and gender matched non-DM control group, including 74 subjects (49 to 87 years at death), was separated from the available cohort of 535 autopsy cases.
All procedures performed in the study involving human participants were in accordance with the ethical standards of the institutional and national research committee. The subjects included and/or their relatives had given their consent for the use of the tissue, and the study has been authorized by the regional Ethics Committee of Uppsala, Sweden # 2011/286. The study was carried out in accordance with the 1964 Helsinki declaration and its amendment or comparable ethical standards.

Sampling of the tissues
At autopsy, the brains were weighed, fixed in excess of 4% buffered formaldehyde for a few days, and cut into 1-cm-thick coronal slices. Grossly notable lesions were noted as present or not on all fixed coronal slices. Following a standardized protocol, 16 brain specimens were sampled in each case, including areas such as the frontal, temporal, parietal, precentral, occipital cortices, the gyrus cinguli, and the striatum; basal forebrain including the amygdala, thalamus, and the anterior and posterior hippocampus; midbrain including the substantia nigra; pons including the locus coeruleus, medulla, vermis, and the cerebellar cortex.

Processing of tissue samples
Seven m thick sections of the paraffin embedded sampled brain tissue (16 blocks/case) was stained with hematoxylin-eosin (HE) stain to assess the microscopic lesions such as infarcts that were noted to be seen or not.
Pancreatic tissue was sampled at autopsy from the body of the organ and fixed in formalin for at least one week. Three m thick sections of the paraffin embedded material were stained applying HE stain to assess the representability of the tissue.
A representative section of the pancreas with notable IAPP/IR was stained with Congo-red stain to confirm the apple-green birefringence under polarized light (Fig. 1).

Assessment of stained sections
All assessments were carried out in light microscopy, in magnifications x40-x400. The brain sections to be stained with the antibodies listed above were selected based on the requirements of the recommended staging protocols. For HPτ, the Braak stage (0, a,b, I-VI) [2]; for A␤, the Thal phase (0-5) [7,23]; and for ␣S, the Brain Net Europe stage (0-6) [6] were given. The type 1 or type 2 cerebral amyloid angiopathy was assigned as recommended by Thal and colleagues [24]. The pTDP43/IR was looked for and assessed in a dichotomized manner as seen or not seen in the section of the amygdala, medulla, and the hippocampus. All lesion types, i.e., intracytoplasmic inclusion and neurites were assessed. The semi-quantitative extent of pTDP43 was given as (0) not seen, (1) seen in one region, (2) seen in any two regions, and (3) seen in all three regions. The IAPP/IR was assessed by counting the IAPP/IR patches in ten randomly selected fields in magnification x100. A mean value of IAPP/IR was calculated for each case, i.e., sum of IR patches divided by the number of assessed fields.
The IAPP/IR was also sought after on a section obtained from the cortex. Cases that displayed a high load of A␤ (n = 28) in the brain were chosen for this purpose. The A␤ was also looked for on a section obtained from the pancreas. Cases that showed a high load of IAPP/IR (n = 15) were selected for this purpose.

Statistical methods
The IBM SPSS statistics software (IBM Corporation, Armonk, NY) was used. Statistical differences were analyzed using non-parametric Mann-Whitney U-test when two cohorts and Kruskall-Wallis test when more than 2 cohorts were compared, and the correlation was assessed by Spearman's correlation test.

Demographics of the subjects
The age at death of the selected study cohort, including 74 subjects with DM and 74 non-diabetics ranged from 48 to 87 years. Out of this cohort 16% had displayed cognitive impairment during life. The brain weight ranged from 988 to 1810 grams and the post mortem delay ranged from 10 to 336 hours. Demographics of the included subjects are given in Tables 1 and 2.

Brain weight
The brain weight decreased with age and differed significantly between the age groups (p < 0.05) in the whole cohort. This significant decrease was also noted for non-diabetics but lost in DM. A significant difference was noted between the youngest and middle aged subjects (p < 0.005) but only in non-diabetics ( Table 2).

The incidence of tissue alterations
The incidence of protein alterations increased with age in the whole cohort of 148 subjects (Table 2). HPτ/IR increased from 84 to 100%, A␤/IR from 16 to 54%, ␣S/IR from 3 to 25%, pTDP43/IR from 29 to 51%, and IAPP/IR from 90 to 100%. Only 3% of the  subjects lacked any IAPP/IR in their pancreas, and a high extent of IAPP/IR was observed in 8% of the subjects. The incidence of protein alterations was not influenced by DM. The incidence of vascular lesions both micro-and macroscopic seemed to increase with age and infarcts were more common in subjects with DM, 49% versus 32% ( Table 2).

The extent of altered proteins in the brain and pancreas and DM
The extent of HPτ/IR in the brain assessed applying the regional distribution of the misfolded protein increased with age. All of the youngest subjects (≤65 years) were in Braak stages 0-II, whereas most of the older subjects (≥66) were in Braak stages II-IV (Table 3). DM did not influence the extent of HPτ/IR.
The extent of A␤/IR in the brain assessed applying the regional distribution of the misfolded protein increased with age. Most of the youngest subjects (≤65 years) did not display any A␤/IR in the brain tissue, whereas phase 4-5 was primarily seen in the oldest subjects (≥76). DM did not influence the extent of A␤/IR ( Table 4).
The extent of ␣S/IR in the brain assessed applying the regional distribution of the misfolded protein increased with age. ␣S/IR was seen in 18% of the subjects, and most of these subjects were older than 65 years. In 8% of the whole cohort the highest stages of ␣S pathology (BNE stage 5 and 6) was seen. DM did not influence the extent of ␣S/IR (Table 5).  The extent of pTDP43/IR in the brain while assessing three commonly affected regions (amygdala, hippocampus, and medulla) increased with age. In 43% of the whole cohort pTDP43/IR was observed. In the youngest age group, 29% of subjects were affected with pTDP43 pathology and in 8% of the whole cohort all three assessed regions were affected. DM did not influence the extent of pTDP43/IR ( Table 6).
The extent of IAPP/IR in pancreas while applying a semiquantitative assessment strategy increased with age when all subjects were included ( Table 2). The increase was close to significant (p = 0.07) for the whole cohort but significant when comparing the youngest subjects with oldest (p < 0.05). When only diabetics were included the increase in the extent of IAPP/IR with age was significant (p = 0.021) and when the youngest group was compared with the middle-aged or the oldest group (p < 0.05 and p < 0.01 respectively). The mean extent of IAPP/IR did not  -diabetic 32  72  3 3 6 13 3  diabetic  31  81  3 7 3  3 3  All  all  148  82  4 3 3  5 3  non-diabetic 74  82  4 1 3  7 3  diabetic  73*  82  4 5 3  3 3 differ significantly between DM and non-diabetic. In 57% of the subjects, the mean number of IAPP/IR patches was 4 or less and in 8% the mean number was more than 10 (Table 7). Fifteen percent of the subjects with DM had a high extent of IAPP/IR in their pancreas (IAPP/IR >10.1) when compared with 1% of the non-diabetic subjects. This difference was noted also for different age groups, i.e., 0 versus 7%, 0 versus 25%, and 3 versus 10% respectively.

Correlation between altered proteins, the age of the subject and diabetes
Overall the highest correlations were observed between misfolded proteins seen in the brain (Table 8). High and significant correlation was observed in the 148 subjects between the age of the subjects and HPτ/IR, age and A␤/IR, HPτ and A␤/IR, HPτ and pTDP43/IR, A␤ and pTDP43/IR and A␤ and ␣S/IR. This association was preserved in the 74 non-diabetic subjects, whereas it decreased or was lost, i.e., was insignificant in subjects with DM. Extent of IAPP/IR showed weak but significant correlation with age in the 148 assessed subjects. This significant correlation was also observed for the subjects with DM. A negative correlation (r = -0.3, p < 0.05) was observed between IAPP/IR and A␤/IR in cognitively unimpaired subjects without DM. A significant correlation was detected between IAPP/IR and pTDP43/IR in subjects with DM (r = 0.2, p < 0.05) and when only the 12 subjects with DM and that displayed extensive IAPP/IR (>9) were included the significance became higher (r = 0.8, p < 0.005). A significant correlation was also observed in cognitively unimpaired subjects with DM (r = 0.3, p < 0.05). This significant correlation was lacking    HPτ, hyperphosphorylated tau (extent Braak stage [2]); A␤, amyloid-␤ (extent Thal phase [7,23]); pTDP43, phosphorylated transactive DNA binding protein (extent: immunoreactivity to be seen in one, two or three regions, i.e., regions medulla, amygdala, hippocampus); ␣S, ␣ synuclein (extent BNE stage [6]) IAPP, islet amyloid polypeptide (extent: the mean number of immunoreactive patches in ten randomly selected fields); DM, subjects with diabetes; nonDM, subjects without diabetes; CI, subjects with cognitive impairment; nonCI, subjects without cognitive impairment. Spearman correlation test 1 p < 0.05, 2 p < 0.01, 3 p < 0.0001, r values when p < 0.0001 are given in bold.
in demented subjects with DM and was negative (r = -0.6, p < 0.05) in demented subjects without DM.

Dementia and diabetes
There were 24 subjects that had displayed cognitive impairment during life, 10 of them were subjects with DM (Table 9, cases 15 to 24). The most common cause of death was cardiovascular complication followed by bronchopneumonia, and DM did not influence the cause of death. The mean age at death (not significant) and the brain weight (p < 0.05) were lower in the group of cognitively impaired with DM when compared to the non-diabetics, and the mean extent of IAPP/IR (not significant) was higher in subjects with DM when compared with the non-diabetic group. The extent of all pathologies (HPτ, A␤, ␣S, pTDP43) was higher (not significant) in subjects without diabetes. In both the DM and in the non-diabetic group some subjects displayed sparse and some extensive IAPP/IR. There were no significant differences in the definite diagnosis of neurodegeneration between subjects with or without DM ( Table 9).

Colocalization of Aβ and IAPP
IAPP/IR was not observed in the 28 sections of the frontal cortex with a high load of A␤/IR, and A␤/IR was not observed in the 15 pancreas sections with a high load of IAPP/IR.

DISCUSSION
Here we found that the incidence and the extent of age related misfolded proteins seen in the brain or the pancreas increased with age. Noteworthy, this increase was not influenced by DM. Thirteen percent Table 9 Proteinopathy seen in the brain and pancreas in 24   of our autopsy cohort of 615 subjects, age ranging from 48 to 102 years had DM, based on the medical records. In 93% of subjects with DM, tissue was available for this study. The percentage of 13% diabetics is well in line with the reported incidence of 10 to 20% of DM observed in Europeans over 65 years of age [25].

IAPP immunoreactivity in pancreas in aged
Numerous reports have indicated that aggregation of IAPP in the pancreatic tissue is a phenomenon related to DM [10,11]. In 1990, it was reported that altered IAPP aggregates were seen in the pancreas in up to 77% of DM when compared to 7% in non-diabetic subjects [26]. In this study, we assessed the IAPP/IR in the endocrine part of the pancreas (the islets of Langerhans) applying IHC techniques. In many of the cases, the tissue was altered due to the postmortem delay, but the IHC technique readily visualized the IAPP/IR. The youngest subject (48 years old) had DM and already displayed IAPP/IR in pancreas, whereas lack of IAPP/IR was registered in only five subjects, age ranging from 50 to 71 years. Thus, contrary to the report by Clark and colleagues, we observed that almost all our subjects (97%), independent of clinical diagnosis of DM, displayed IAPP/IR in their pancreas. The discrepant result when comparing our results with Clark and colleagues might be related to selection bias, methods used and to the population studied (Scandinavians versus Pima Indians). Contrary to our expectations, three of the subjects with DM lacked IAPP/IR and this might be due to selection bias; only one section obtained from the body of pancreas was investigated. Surprisingly, almost all aged non-diabetics in our study displayed IAPP/IR and this might be explained by the recent observations that a large number of adults in the U.S. have undiagnosed type 2 DM, and 5 to 10% of the European population display impaired glucose regulation [27,28].
Not previously reported to our knowledge, we observed that the extent of IAPP/IR increased with age, and this increase was noted for the whole cohort of 148 subjects and for the 74 subjects with DM. Thus the increase of extent of IAPP/IR in the pancreas is an age-and DM-related phenomenon, i.e., IAPP/IR was significantly higher in the oldest group when compared to the youngest. This observation is in line with the previous reports indicating that the prevalence of impaired glucose regulation seems to increase with age [28]. The mean number of labeled islets of Langerhans in the pancreas did not differ significantly between the subjects with DM and nondiabetics, but a high extent of IAPP/IR was observed more often in subjects with DM. The reports indicating that there is a substantial number of undiagnosed subjects with type 2 DM or with impaired glucose regulation is intriguing [27,28]. The question arises as to whether the pathology seen by us postmortem is a sign of a prodromal stage of DM.

Misfolded proteins in the brain
The prevalence of aggregation prone proteins in the brain, i.e., HPτ, A␤, ␣-S, and pTDP43, increased with age. The most common alteration to be observed was HPτ followed by A␤, pTDP43, and ␣S. Thus, our results here are in agreement with previous reports [1]. This prevalence did not seem to be significantly influenced by DM. It was not only the prevalence but also the extent of HPτ, A␤, ␣-S, and pTDP43 that increased with age. The Braak stage of HPτ, the Thal phase of A␤, and the BNE stage of ␣S increased with age for the whole study population. When assessed separately for DM and non-diabetics, no major differences in the patterns of the age-related changes were seen. These results are in line with previous reports [20]. A similar trend was also observed for pTDP43 pathology as assessed here, i.e., number of brain regions affected. Thus, our results imply that DM does not influence the prevalence or the extent of altered proteins to be seen in the aging brain.
In addition to altered proteins, the prevalence of vascular lesions increased with age in both DM and non-diabetic subjects, and vascular lesions were found to be more frequent in DM. This observation is in line with previous reports indicating that an association exists between the extent of vascular lesions and DM [29].

Correlation between the misfolded proteins
Extent of all assessed misfolded proteins was significantly influenced by the age. A significant correlation was also observed between HPτ, A␤, pTDP43, and ␣S. Thus an association between AD related misfolded proteins and pTDP43 and ␣S/IR was observed. This might suggest that the AD related aggregation prone proteins might act as a seed to initiate aggregation of other proteins and thus this hypothesis should not be overlooked [18]. Contrary to the above, no correlation was observed between HPτ, A␤, and ␣S/IR in the brain and IAPP/IR in the pancreas. The extent of IAPP/IR did not show any association with AD-related pathology; moreover, a negative correlation was observed between IAPP and A␤/IR in cognitively unimpaired non-diabetics. Therefore, our results obtained in humans at the end stage, i.e., postmortem, do not support the epidemiological, clinical, or experimental data suggesting that a causative relationship exists between DM and AD, i.e., misfolded protein IAPP acts as a seed for misfolding of A␤, HPτ, and ␣S or vice versa [19]. Interestingly the extent of pTDP43/IR seen in as many as 43% of our cohort was significantly associated with IAPP/IR in subjects with DM. Noteworthy, when only 12 subjects with both DM and extensive IAPP/IR were included the correlation was strong (r = 0.8) and significant (p < 0.005). This correlation was also noted in the 64 cognitively unimpaired subjects with DM but was lacking in the 10 demented with DM. Surprisingly, a significant negative correlation between pTDP43 and IAPP/IR was observed for the 14 non-diabetic demented subjects (Table 8). Therefore, based on our finding, pTDP43/IR seems to be associated with IAPP/IR and DM. Recently it was reported that there is an interaction between IAPP and a single nucleotide polymorphism (SNP) on chromosome 12p12 (rs73069071) and this SNP has been shown to be associated with hippocampal sclerosis (HS) [30,31]. Parallel with this, an association between HS and pTDP43 has been proposed by many [32,33]. Furthermore, in TDP43 transgenic mice impaired insulin mediated glucose uptake has been described [34]. Thus, there is indeed evidence for interplay between IAPP and TDP43. TDP43 is ubiquitously expressed in all organs, binds to both DNA and RNA and has multiple functions including transcriptional repression, pre-mRNA splicing and translational regulation [35,36]. Whether phosphorylation of TDP43 is influenced by impaired glucose metabolism has not yet been studied and whether phosphorylation of TDP43 can be observed in peripheral organs in diabetes has not been looked for. It is reported that up to 50% of AD patient display concomitant pTDP43 pathology and whether subjects with AD and concomitant pTDP43 suffer from DM has not been looked for [37].

Colocalization of Aβ and IAPP
Already in 2010, it was reported that AD-related pathology was seen in pancreas in diabetics [38]. Following in 2013, it was reported that IAPP/IR was observed in the brains of demented subjects with type 2 DM [39]. We carried out the IHC stain on those cases that displayed an excess of the primary pathology (A␤ or IAPP) and applied mAb while staining for IAPP and A␤. We were unable to repeat any of the reported observations given above. There was no A␤/IR seen in the pancreas in 15 subjects with excess of IAPP/IR. Twelve of these subjects had DM, and 3 of these subjects had displayed cognitive impairment. There was no IAPP/IR seen in the 28 brain samples with excess of A␤/IR. Twelve of these subjects had DM and 13 were demented. The differing results are probably due to the methodology, both tissue and methodology related. Recently in 2017, cross-seeding and colocalization of IAPP and A␤ was reported to be seen in transgenic mice [19]. This finding was further highlighted by Ridler [40]. One factor that should be considered in the described experimental setting by Moreno-Gonzales and colleagues is the physiological increase of amyloid-␤ protein precursor (A␤PP) production seen in posttraumatic brains [41]. Thus, the inoculation of pancreatic aggregates into the brain (a trauma) as was carried out by Moreno-Gonzales and colleagues might cause a physiological increase in A␤PP production that in a setting of A␤PP transgenic mice might lead to A␤ aggregation.

Dementia and diabetes
The overall incidence of dementia in developed countries has been reported as decreasing during the last few years, from 11.6% to 8.8% [42]. This decrease is considered as being attributed to social, behavioral, and medical factors. Out of our 148 subjects, 16% of the subjects displayed cognitive impairment. This incidence is somewhat higher than overall incidence probably due to a selection bias, i.e., reason to refer for autopsy varies. The distribution of dementia in our study cohort with or without DM was fairly even (10 and 14 subjects, respectively). There was no notable difference in the distribution of the definite diagnosis of dementia. One obstacle, however, is the lack of consensus criteria regarding vascular lesions to be assessed as being causative for dementia thus the vascular pathology might have been overlooked [43].
A link between cognitive impairment, i.e., dementia and DM has been suggested in epidemiological and clinical studies and in particular, it has been proposed that there is an association between DM and AD [17,44,45]. This presumption has been challenged by others suggesting that dementia in DM is primarily related to vascular alterations [20,29,[46][47][48]. Here, based on our results, assessing the hallmark lesions of AD and other dementias and the alteration related to DM, i.e., IAPP/IR, no association could be seen between DM and AD. The lack of a link between AD and DM is also supported by the observations that less AD related pathology has been reported to be seen in subjects with DM when compared to non-diabetic subjects [21,49,50]. Our results are in line with the above, as our demented subjects with DM displayed less misfolded proteins (not significant) in their brain when compared with nondiabetics. Further, in 2015, Moran and colleagues reported that no association could be found between DM and A␤ in the brain when assessed with 11 C Pittsburgh compound B or with cerebrospinal fluid A␤ 42 levels [22]. In line with the above, Kuo and colleagues reported in 2015 that the hazard ratio for dementia in DM was associated with comorbidities such as hypertension, hyperlipidemia, coronary artery, and/or kidney disease-all risk factors for vascular lesions in the brain [51]. In line with the above vascular alterations were seen in 49% of subjects with DM when compared with 32% in non-diabetics in our cohort.

Strength and weaknesses of the study
The strength of this study is that we have assessed the pathological alterations in the brain and in the pancreas that have been reported as being associated with clinical conditions such as DM and dementia.
The selection of cases was based on a known clinical condition, i.e., DM, the prevalence being in line with what is seen in the general population. An age and gender matched control group was selected from the original autopsy cohort. The prevalence of dementia was quite evenly distributed in non-DM and DM. We were able to assess objectively both the incidence and the extent of neurodegenerative alterations as well as the DM related aggregation of IAPP seen in the pancreas. One of the major weaknesses of the study is the assessment of vascular lesions limited to infarcts and cerebral amyloid angiopathy and the lack of consensus of vascular pathology to be assessed in the postmortem brain of demented. Thus, the vascular pathology might have been overlooked.

Conclusion
We found that the incidence and the extent of HPτ, A␤, pTDP43, ␣S, and IAPP increased with age, independent of DM. The definite diagnosis of dementia varied in both the DM and the non-diabetic cohort. We noted that two of the misfolded proteins were observed in virtually all of the subjects, i.e., HPτ in the brain and IAPP in the pancreas. The high incidence of HPτ pathology in the brains of the aged is well known, whereas the observation of IAPP in the pancreas in almost all of the subjects, even those not clinically diagnosed with DM, is intriguing. This might be related to a high number of aged displaying impaired glucose regulation and/or undiagnosed type 2 DM, as recently reported. A strong correlation was observed between HPτ, A␤, ␣S, and pTDP43 that is in line with the cross-seeding hypothesis, whereas lack of correlation between HPτ, A␤, ␣S, and IAPP argues against this molecular mechanism in development of these lesions. Therefore, we are unable to confirm the suggested link between AD and DM. Interestingly an association was observed between pTDP43 and IAPP/IR, not previously reported. The association IAPP/pTDP43 was strong and significant particularly in subjects with DM and high extent of IAPP/IR in pancreas. A link between IAPP/HS, TDP43/HS, and TDP43/glucose homeostasis have been recently suggested indicating that an association between these two protein alterations might be of significance [30][31][32][33][34]. TDP43 is ubiquitously expressed in all organs and whether phosphorylation of TDP43 is influenced by impaired glucose metabolism or pTDP43 is observed in peripheral organs in diabetes has not yet been studied. The reported association between AD and DM (epidemiological and clinical studies) might, based on our results, be related to the high incidence of pTDP43 seen postmortem in AD patients. Thus, pTDP43 should urgently be looked for in other organs in diabetics and in experimental models and the influence of glucose metabolism on phosphorylation of TDP43 should urgently be explored.