Concomitant neurodegenerative pathologies contribute to the transition from mild cognitive impairment to dementia

The aged brain frequently exhibits multiple pathologies, rather than a single hallmark pathology (pure pathology [PurP]), ranging from low/intermediate levels of additional pathology (LowP) to mixed severe pathology (mixed SevP). We investigated the frequency of PurP, LowP, and mixed SevP, and the impact of additional LowP on cognition.


INTRODUCTION
Age In addition, cerebrovascular disease (CVD) and CVL may be the neuropathological correlate of neurological disease. However, these hallmark lesions are not mutually exclusive and in brains of elderly individuals the presence of only one characteristic pathology, that is, a pure pathology (PurP), is the exception: the majority of brains show multiple pathologies, a condition referred to as cerebral multimorbidity. 1,2 The degree of cerebral multimorbidity in neurologically impaired individuals ranges from one main disease with low/intermediate level additional pathology (additional LowP) with a low likelihood of causing clinical symptoms on its own, for example, AD with minor CVL, to cases in which the hallmark pathologies of two (or more) diseases are so severe that any one of these could independently cause cognitive impairment, 1 for example, AD and DLB, which can be categorized as mixed severe pathology (mixed SevP) and would be diagnosed as Mixed AD/DLB. 3 Data from large autopsy studies show that additional LowP and mixed SevP together are seen in up to 74% of brains of elderly people 2,4-6 and suggest that the presence of additional pathologies (either additional LowP or mixed SevP) is associated with a greater risk of dementia or accelerated cognitive impairment 7-15 due to possibly lowering the burden of major pathology necessary for clinical symptoms, for example, the presence of CVL in AD lowers the threshold at which AD pathology causes clinical dementia. 16 Another example of the clinical impact of cerebral multimorbidity is the presence of limbic predominant TDP-43 protein aggregates, a condition recently termed limbic-predominant age-associated TDP-43 encephalopathy neuropathological change (LATE-NC). LATE-NC has been suggested to cause a distinct disease (i.e., LATE 17 ) but is more commonly additional LowP, present in up to 50% of individuals over 80 years in age 17 and highly prevalent in AD (74% 18 ) where its presence is associated with accelerated cognitive decline. 19 The terminology used to describe to cerebral multimorbidity varies with different studies defining it as presence of more than one pathology, 8 or combined diagnoses, 6 or additional pathologies 13,20 .Few studies have differentiated between "mixed dementia" (mixed SevP) and additional low-severity concomitant pathology (additional LowP) 5 or classified comorbidity based on the severity of the co-pathologies. 7,12,15 The Brains for Dementia Research (BDR) program was started in 2008 in the UK to address the shortage of banked post mortem brain Manchester, Newcastle upon Tyne, and Oxford), which implemented a prospectively agreed protocol for brain sampling and standardized neuropathological assessment.
We used the BDR cohort to investigate the neuropathological frequency of common age-associated neurodegenerative pathologies and CVD in a large cohort, and distinguished between PurP, a single main disease with additional LowP, and mixed SevP. We also analyzed the impact of additional LowP on the rate of cognitive decline and the severity of dementia.  Table S1 in supporting information.

Clinical assessment and diagnosis and apolipoprotein E (APOE) genotype
During life, clinical assessments were conducted by a trained psychologist or research nurse. Baseline assessments were conducted face to face, with annual follow-up assessments over the next 1 to 5 years.
This study was inclusive of two clinical assessments performed by BDR; Clinical Dementia Rating (CDR; range: 0-3) 22   The neuropathological classification criteria are provided in Table 1.

Neuropathological assessment and diagnosis
Briefly, for AD pathology, full-blown disease was defined by "high AD neuropathological change," 23 which included cases with Thal Aβ phase 4/5, 26 Braak stage V/VI, 27 and CERAD stage for neuritic plaques B/C. 28 DLB cases fell into the McKeith stage of either limbic or neocortical LBD, and PD cases had brainstem LBD. 24 Because the LATE-NC staging criteria 29 17 We categorized CVD pathology according to the VCING criteria, which categorize cases without any or with only mild cerebrovascular pathology both as having "low likelihood that CVD contributed to cognitive impairment," and therefore, this category was not considered at all in our study, as it includes cases without any CVD. Cases in the VCING category of "moderate" were classified as having additional LowP CVD and those in the category of "high" were classified as having a PurP or mixed SevP diagnosis with CVD.

Statistics
We    Intermediate AD neuropathological change (IM AD-NC) was diagnosed in 9.9% closely followed by LBD in 9.7%. CVD and Mixed AD/CVD were diagnosed in 5.4% and 2.8%, respectively; 21.1% of cases were classified as having LowP only. Within this group, the neuropathological criteria for definite primary age-related tauopathy (PART; 32 Braak stage I-IV, Thal Aβ phase 0) were met in 34 cases and for "possible" PART (Braak stage I-IV, Thal Aβ phase 1-2) in 64 cases.
All neuropathological stages stratified by neuropathological diagnosis are presented in Table S2. APOE genotype stratified by disease group is presented in Table 3.

Frequency of PurP, additional LowP, and mixed SevP
Overall, 22.7% of cases were classified as PurP, 69.9% as additional LowP, and 7.5% as mixed SevP. Figure 1 illustrates each neuropathological diagnosis and the proportionate associated additional LowP, highlighting the high proportion of cases within each neuropathological diagnostic group that have associated additional LowP and the complexity of multimorbidity. Table 4

Association of additional LowP and age at death
Age was highly associated with an additional LowP diagnosis (χ 2 , df 5 ,

Clinicopathological correlations
Cases without CDR or MMSE data were excluded from further clinicopathological analysis. CDR scores were recorded in 508 cases (75.8% of cohort). Overall mean last CDR assessment to death was 11.10 (± 11.6) months. Frequency of CDR and last assessment to death intervals are presented in Table 5. Two hundred thirty-two cases (34.6% of cohort) had more than one MMSE score and time interval(s) between  in Data S1A in supporting information). Odds ratio (OR) data are presented in Table 6. Briefly, increasing Braak NFT stage was associated with an almost 4-fold increase and Thal Aβ phase and McKeith stage with an almost 3-fold increase in the odds of CDR > 1 versus CDR < 0.5.

Effect of main pathological stage, APOE genotype, sex, disease duration, and years of education on the odds of dementia
No neuropathological stage predicted the transition from CDR 0 to CDR 0.5 (model: P = 0.112). Increasing Braak NFT stage increased the odds of transitioning from CDR 0.5 to CDR 1 by 75% and the odds of having CDR 2 versus CDR 1 increased more than 2.5 times with increasing neuritic plaque density. Finally, the odds for transitioning from CDR 2 to CDR 3 were increased 57% by Braak NFT stage. APOE genotype status was not found to be significantly associated with the transition from CDR 0 to CDR 0.5 (4/4, P = 0.102; 3/4, P = 0.316; 3/3, P = 0.994; 2/4, P = 0.243; 2/3, P = 0.935; 2/2, P = 0.379). APOE genotype did not significantly influence the OR of neuropathological stages.
A longer disease duration (i.e., survival) was associated with being in the CDR < 0.5 group compared to CDR > 1 group (OR = 0.985, 95% CI: 0.976-0.995), but disease duration did not significantly influence OR of neuropathological stages. Sex did not have a significant effect on the odds of being demented (P = 0.218) and no effect on OR of neuropathological stages. The number of years of education was significantly higher in the no-cognitive-impairment group compared to the dementia group (P = 0.003; mean values no-cognitive-impairment, 13.30 ± 3.8 years; dementia, 12.23 ± 3.41 years); however, due to the marginal difference, as expected this did not significantly influence the odds of being CDR > 1 versus CDR < 0.5 (P = 0.185) or the OR of the neuropathological stages.

Additional LowP on the odds of dementia
Forward enter linear regression was used to determine whether the presence of any 1 or 2+ additional LowP significantly contributed to the odds of having dementia (CDR < 0.5 vs. CDR > 1), a categorical increase in CDR score, or the ORs of neuropathological stages ( Table 6).
The findings are summarized in Data S1B and

Clinical impact of additional LATE-NC and CVD pathology in AD
The addition of low-level LATE-NC and CVD pathology has been implicated as an important clinical influence on AD. We investigated the influence of additional LATE-NC or CVD on age at death, final MMSE scores, rate of cognitive decline, and disease duration.

Additional CVD pathology in AD
We selected for neuropathological cases diagnosed as having PurP AD (as above), AD with additional CVD (n = 13; mean age at death 83.62 [± 6.56]) and Mixed AD/CVD (no additional pathologies; n = 7; mean age at death 85.71 [± 6.75]).
We compared age at death and clinical scores between the groups, but no differences were found (P < 0.099). However, disease duration was significantly shorter in AD with additional CVD than in PurP AD (P = 0.038).

DISCUSSION
It has become increasingly clear that the aged human brain is characterized by the coexistence of multiple neurodegenerative pathologies ranging from minimal additional LowP to mixed SevP. Previous autopsy studies have not clearly defined and differentiated between additional LowP and mixed SevP, and the frequency of true additional LowP and the impact this has on cognition has been unclear. In the present clinicopathological autopsy study, by analyzing the severity of additional pathologies in common age-related neurodegenerative diseases, we have been able to capture the full complexity of multimorbidity and to show that even low amounts of additional pathology, which might have been considered clinically irrelevant, have a statistically significant impact on cognitive decline and the clinical syndrome.
Our findings indicate that only 22.7% of cases were considered a PurP, which is much lower than other community-based and large consortia clinicopathological studies that report a frequency of PurP ranging between 40% and 50% 6,13 . This discrepancy is likely to be due to the more stringent criteria applied in our study to identify cases with additional LowP that would otherwise be classified as a PurP. The AD and IM AD-NC groups had the highest rate of PurP at 44.5% and 62.2%, respectively, which is similar to findings in previous autopsy studies. 7,5,6,33  These findings are in agreement with previous clinicopathological studies that found that a clinical diagnosis of MCI was associated with a comorbid diagnosis at autopsy 7 and the number of additional pathologies was associated with clinical dementia 11 or cognitive decline. 12,15 This suggests the presence of additional LowP lowers the threshold for overt cognitive decline, perhaps by lowering brain reserve 37  All neuropathological lesions, with the exception of CVD, were associated with cognitive decline; in particular Hpτ-, Aβ-, and α-synrelated pathology were associated with up to a 3-fold increase in the odds of dementia, in agreement with previous clinicopathological studies. 8,45,46 Our study provided novel information regarding the specific neuropathologies that significantly contributed to the progression and severity of the clinical dementia, namely Hpτ pathology in the conversion from MCI to mild dementia and moderate to severe dementia, and neuritic plaques in the transition from mild to moderate dementia.
This highlights the impact of Hpτ, and subsequent AD-associated neuropathological change, on cognitive function, as has been previously recognized. 47 Perhaps surprisingly, no case was classified as pure CVD, that is, CVD without any LowP, in contrast to previous studies that reported the frequency of pure CVD between 2% and 11%. 5 Additionally, a diagnosis of Mixed AD/CVD was present in only 2.8% of cases-lower than reported frequencies within community-based clinicopathological studies from the United States 13 and the UK 4 but in line with previous reports from the Vienna consecutive autopsy series. 5 Furthermore, CVD was not associated with cognitive decline or an increased risk of cognitive impairment or dementia, contrary to other large clinicopathological studies (for reviews please see Kapasi and Schneider 48 and Kapasi et al. 49 ) but in agreement with a previous autopsy study. 11 This study also found that the addition of LowP CVD in AD did not impact clinical scores when compared to pure AD in contrast to previous studies. 4,16,50,51 However, this study did reveal that individuals with AD and additional CVD had a shorter disease duration, suggesting accelerated disease progression. These differences in prevalence and clinical contribution may be affected by selection bias, as exclusion criteria for BDR recruitment includes major stroke, and the use of the VCING criterion for the neuropathological assessment of CVD limits CVD assessments to low, moderate, or high likelihood of contributing to cognitive impairment. However, the VCING criteria reflect a validated neuropathological assessment of CVD in relation to the predicted probabilities of vascular cognitive impairment; therefore, our study may reflect a truer representation of the prevalence of clinically relevant CVD within this UK cohort. On the other hand, VCING criteria are relatively crude and do not have the accuracy of neuropathological criteria used for the assessment of neurodegenerative proteinopathies. The detailed assessment of CVD is challenging, as CVD-associated brain damage does not progress in a stereotyped topographical manner and therefore large areas of the post mortem brain would need to be assessed to get a complete picture of CVDassociated brain damage. In addition, post mortem delay may result in autolytic changes that may mask microscopic hypoxic tissue damage.
Hence, the use of VCING criteria may lead to an underestimation of the contribution of CVD to cerebral multimorbidity.
The exact pathomechanisms of cerebral multimorbidity are still poorly understood, but it is assumed that both age-associated failures of basic cellular mechanisms and protein-protein interactions play a crucial role (please see Spires-Jones et al. 52 for review). The accumulation of misfolded proteins and CVD/CVL in the human brain are clearly associated with advanced age; dysfunction of the complex and interrelating systems of basic cellular homeostatic regulation, DNA damage repair, autophagy regulation, and oxidative stress response are all associated with cellular dysfunction in aging, and some individual genetic variability, leaving cells vulnerable to further insults. Due to the complexity and heterogeneity of neuropathological lesions in the aged brain, future classification should move away from rigid categorization of neurodegeneration into distinct disease subtypes only (e.g., AD, LBD) and be inclusive of the presence, severity, and location of LowP. This will provide a more precise picture of neurodegeneration in general and unravel subtle clinicopathological phenotypes and may be transferable to future biomarkers and intra vitam diagnosis allowing for accurate clinical trials design and interpretation, as well as targeted and personalized therapeutics.

CONCLUSIONS
More than three quarters of aged individuals have multiple brain

CONFLICTS OF INTEREST
The authors report no declarations of interest.