A cross-sectional population-based survey on a primary care-based Portuguese volunteer group of 590 individuals (pcb-Cohort) was carried out as previously described (Rosa et al., 2017). For this study, five primary health care centers, in the Aveiro district of Portugal, were randomly selected. In brief, participants completed a structured interview covering their respective lifestyles, and clinical history was collected. Next, cognitive evaluations and dementia screening tests, namely CDR, Mini-Mental State Examination (MMSE), the Geriatric Depression Scale (GDS), the Katz Activities Daily Living (ADL), and Instrumental Activities Daily Living (IADL), were performed on all 590 volunteers.

Clinical data from the study participants, compiled by physicians and health professionals, was accessed via collaboration with the medical staff at all sites and was thoroughly investigated. Clinical data available from clinical records were scored, such as information regarding the presence of other diseases, namely HYP, DYS, OA, CVD, DEP, GID, DM, and RESP. HYP was diagnosed after observing persistent elevation of systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg in several temporally distinct measurements. Similarly, DYS was diagnosed by measuring total cholesterol, HDL cholesterol, and triglycerides after a 12-h fasting period, with repeated analyses at a minimum interval of 4 weeks. DEP was diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria, in which the individual must have at least five symptoms of either a depressed mood or loss of interest or pleasure for 2 weeks. In the present study, CVD included cardio pathologies, cardio arrhythmias, myocardial infarction, acute coronary syndrome, coronary revascularization or other arterial revascularization procedure, ischemic stroke and peripheral arterial disease. GID comprised dyspepsia, esophagitis, gastritis, duodenitis, inflammatory bowel diseases, diverticulosis, diverticulitis and anusitis. Finally, RESP included the following chronic pathologies: allergic rhinitis, asthma, chronic obstructive pulmonary disease, restrictive pathologies, and sarcoidosis. For RESP, pathologies considered acute, self-limited, or infectious, such as tonsillitis and pneumonia, were excluded. All common diseases considered in the present study were scored positive based on the previous diagnosis with confirmation by a specialist in the reference hospital for the Aveiro Region. Common diseases unable to be confirmed were not included when scoring for the prevalence of the different pathologies among the study participants.

A total of 568 volunteers fulfilled the inclusion criteria and were processed for APOE allele and BIN1 rs744373 SNP genotyping. Procedures regarding this phase are described below.

For each volunteer, blood was collected into 3 tubes for whole blood, serum, and plasma (3+5+5 ml, respectively), according to standard procedures. Samples were immediately aliquoted and frozen at −80°C. Whole blood samples collected in EDTA tubes were available for genotyping of APOE for only 508 individuals (Figure 1). APOE genotyping was performed prior to BIN1. Meanwhile, three samples were no longer biologically available for genotyping (BIN1 population = 505). Therefore, only 505 samples (available for both APOE and BIN1) were considered for the present study.

APOE and BIN1 genotyping were carried out by direct blood PCR using a modified Phusion Hot Start II High-Fidelity DNA Polymerase protocol (Phusion Blood Direct PCR Master Mix; Thermo Fisher Scientific). For APOE, genetic polymorphic regions were amplified using 1–2 μl of blood and the primers APOE-Fw 5'-CGGGCACGGCTGTCCAAGGAG-3' and APOE-Rev 5'-CACGCGGCCCTGTTCCACCAG-3', yielding a fragment of around 300-bp, as previously described (Rosa et al., 2017). The PCR conditions were as follows: 98°C for 5 min; 35 cycles of 98°C for 1 s, 64°C for 5 s, and 72°C for 15 s; and a final extension step at 72°C for 1 min. For BIN1 (rs744373), PCR of the polymorphic regions was carried out using 2 μL of whole blood from each patient, 2 × Phusion Blood Direct Master Mix (Thermo Fisher Scientific), and 0.5 μM gene-specific primer BIN1-forward 5'- AAGACGGAGAGAGGAGGCAT-3' and BIN1-reverse 5'-CCATCTTCTTCTGCTCTCCCA G−3 ′, yielding a fragment of around 767-bp. The PCR conditions were: 98°C for 5 min; 35 cycles of 94°C for 1 min, 63°C for 30 s, and 72°C for 46 s; and a final extension step at 72°C for 5 min. Afterwards, PCR products were purified with sodium acetate (3M, pH 5.2), and Sanger sequencing was performed. Results were analyzed to determine the nucleotide polymorphisms and the respective APOE and BIN1 genotypes. For the study design, the number of volunteers genotyped was 505, which is a reasonable sample size given the population density of the Aveiro district (Schulz and Grimes, 2005; Rosa et al., 2017).

Analyses of the data collected at each phase were carried out, blind to the data from the other stages, using the Statistical Package for the Social Sciences (SPSS) version 26 (Marôco, 2021). Categorical variables were assessed through the examination of frequencies. In contrast, continuous variables were evaluated by the generation of descriptive methods (means, standard deviations) to investigate the differences in the group (CDR, cognitive performance, depression groups, APOE allele carriers, BIN1 G+ carriers vs. normal groups).

Regarding the multivariate analysis, logistic regression was used for the dichotomous dependent variables (risk allele G of BIN1 rs744373 variant and APOE carriers of ε2 or ε4), considering the socio-demographic and cognitive characteristics, as well as the diseases scored for each volunteer. The reference group included non-carriers of the risk allele; that is, for APOE ε4, the reference group was the one that lacked this allele; for APOE ε2, the reference group was the one that did not have this allele; and for BIN1 the reference group was the one that did not have the risk allele G. The Odds Ratio was calculated. This ratio, if >1, indicates a risk factor, and if < 1 indicates a protective factor. In the present study, this methodology was used to identify the risk/protective factors concerning both BIN1 and APOE.

A two-sided statistical test was carried out for each analysis, and a p-value < 0.05 was considered statistically significant. P-values between 0.05 < and < 0.1 were considered a trend.

In the study population, the most predominant APOE haplotype was the ε3ε3, representing 75% of the individuals, followed by ε3ε4 (16.1%) and ε2ε3 (6.1%). It is noteworthy that ε2 and ε4 have opposite effects, but in the pcb-Cohort only 1.4% of the participants had the ε2ε4 haplotype (Figure 2). Similarly, only 1.4% of cases were ε4ε4, being one of the least prevalent haplotypes. As for ε2ε2 haplotype, it was absent in the study population. Regarding BIN1, the most prevalent genotype was the AA (53.2%), followed by AG (41.4%) and the GG genotype, representing only 6.3% of the individuals.

A similar analysis by alleles shows that only 7.5% of the study population had the protective ε2 allele of APOE, while more than twice as many individuals (18.8%) carried the allele that confers risk of developing AD (allele ε4). In turn, the G allele of the BIN rs744373 variant, which confers risk of developing LOAD, was present in almost half of the study population (47.7%).

Relevant characteristics, namely socio-demographic and cognitive deficits within the pcb-Cohort and associations with APOE and BIN1 rs744373, were analyzed and are summarized in Table 1. Overall, no significant associations were evident for ε3 carriers (data not shown). In contrast, there is a skewed prevalence for the stratified genotypes among these characteristics, where APOE ε4 carriers associated with cognitive deficits, while APOE ε2 and BIN only associated with age.

In the pcb-Cohort, women were more prevalent (70.3%). Our results do not show any significant gender effect associated with either APOE ε4 (p = 0.346), APOE ε2 (p = 0.225), or BIN1 (p = 0.614). Although not significative (p = 0.088), our results also showed a decrease in the mean age between non-carriers (68.1 years) and carriers of the APOE ε2 allele (65.4 years). Moreover, in the age group < 65 years ε2 carriers were more frequent than non-carriers (52.6% vs. 33.2%), while in the age group ≥ 65 years ε2 carriers were less frequent than non-carriers (47.4% vs. 66.8%), reaching statistical significance (p = 0.015). Regarding BIN1, there was a significant decrease (p = 0.034) in the mean age of individuals with the G allele (67.0 years) compared to non-carriers (68.7 years). However, when grouping individuals by age (< 65 years and ≥65 years), there were no significant differences in the frequency of the BIN1 risk allele (p = 0.446).

An association between APOE ε2 and cognitive features of the study population was not observed. In contrast, the frequency of individuals with moderate to severe cognitive deficits based on the CDR scores (CDR ≥ 1) was more significant in ε4 carriers (p = 0.038), and the mean in CDR scores was also significantly higher in carriers (1.8 score) compared to non-carriers (1.1 score). Regarding the GDS scale (GDS ≥ 5), we observed an association between APOE ε4 carriers and high GDS scores (p = 0.037). Additionally, our results showed a higher association trend between the ε4 allele of APOE and ADL (p = 0.083), with an increase in ADL-dependent cases among ε4 carriers. No associations between APOE and neither the MMSE nor the IADL were found.

Regarding BIN1, the frequency of the G risk allele increased in individuals with CDR scores equal to 0.5 (34.1% in non-carriers compared to 40.2% in carriers), contrary to the other two groups (CDR = 0 and CDR ≥ 1). Still, this result was not significant (p = 0.193). Furthermore, no significant associations with the remaining cognitive tests were identified.

Regarding socio-demographic characteristics, our results showed no association between age and APOE ε4 or BIN1. Nevertheless, a lower association trend between age and APOE ε2 when comparing non-carriers to carriers (Table 2) was observed. The group of individuals over 65 years old showed 47% reduced odds of having the ε2 allele compared to the younger age group (% of 1–0.53; OR = 0.534; 95% CI 0.256–1.112; p = 0.094).

As for the cognitive characteristics, our results showed an association between APOE ε4 and cognitive deficits (Table 2). Carriers of the ε4 allele were 2.5 times more likely to have a CDR score above 1 (OR = 2.527; CI 95% = 1.089–5.865; p = 0.031). Likewise, our results showed a tendency for a higher association between depressive states and a positive GDS score among ε4 carriers (OR = 1.649; CI 95% = 0.974–2.791; p = 0.063).

The multivariate analysis of the studied diseases showed results aligned with the bivariate analyses (Supplementary Table 1). DYS was statistically significant among the studied diseases for both APOE ε4 and BIN1. APOE ε4 to carriers were 1.8 times more likely to have DYS (OR = 1.804; CI 95% = 1.040–3.129; p = 0.036) than non-carriers. Likewise, carriers of the BIN1 G allele had a 44% reduced risk of having DYS compared to non-carriers (% of 1–0.56; OR = 0.558; 95% CI 0.367–0.847; p = 0.006).

Our results also showed that carriers of the G allele of BIN1 had a 44% reduced risk of having RESP compared to non-carriers (% of 1–0.56; OR = 0.556; 95% CI 0.331–0.934; p = 0.026). On the contrary, although not reaching statistical significance, there was a 48% reduced risk of having RESP diseases among carriers of the ε4 allele of APOE, compared to non-carriers (% of 1–0.52; OR = 0.515; 95% CI 0.240–1.105; p = 0.088).

DM was another disease showing a trend to be associated with both APOE and BIN1. Despite not reaching statistical significance, individuals with the ε2 allele had a 66% reduced risk of having DM compared to non-carriers (% of 1–0.34; OR = 0.342; 95% CI 0.098–1.195; p = 0.093). Conversely, when comparing carriers of the risk allele G of BIN1 rs744373 to non-carriers, the first showed a higher association trend with a greater risk of having DM (OR = 1.491; CI 95% = 0.928–2.398; p = 0.099).

Interestingly, APOE and BIN1 seem to influence each other. According to the logistic regression, the presence of the APOE ε4 allele significantly increased by 1.7 times the odd of having the G allele of BIN1 rs744373 (p = 0.030) and vice versa (p = 0.029). Moreover, without reaching statistical significance, we also observed a similar trend for a higher association between APOE ε2 and BIN1.

In the pcb-Cohort, our results show a higher percentage of ε2 carriers in the age group < 65 years and a lower rate in the age group ≥65 years, compared to non-carriers (p = 0.015). Data suggest that younger generations have a higher frequency of this allele than older generations. APOE ε2 has been associated with longevity (Shinohara et al., 2020). However, contrary to our results, previous studies showed higher frequencies of APOE ε2 in elderly individuals and centenarians compared to younger populations (Cauley et al., 1993; Sebastiani et al., 2019). Nevertheless, it is essential to mention that the APOE ε2 variant represents the most recent variant of this risk gene (Fullerton et al., 2000), suggesting that selective pressures contributed to the evolution and global distribution of human APOE alleles over time (Huebbe and Rimbach, 2017). Thus, this preliminary observation may relate to genetic variations of the younger population. Still, more comprehensive studies need to be carried out in this respect, particularly in larger cohorts, to evaluate potential generational effects on allele frequencies. Of note, APOE ε2 is not only a protective gene but might also increase the risk of certain cerebrovascular diseases and neurological disorders (Li et al., 2020), which could also contribute to the reduced frequency of APOE ε2 carriers among ≥65 years old individuals in our study population.

Further, APOE ε4 carriers significantly associated with cognitive deficits (p = 0.038), contrary to ε2 carriers (p = 0.382). In fact, APOE ε4 emerged as a significant risk factor, increasing the susceptibility to develop dementia (OR = 2.527; p = 0.031). This result is in line with previous reports where APOE ε4 was associated with poor cognitive performance even in healthy individuals (Caselli, 2009; Wisdom et al., 2011). Other studies also reported that the ε4 variant increases the risk of developing AD, while the ε2 variant reduces AD risk (Loy et al., 2014). Moreover, in the present study, while executing the CDR test, APOE ε4 carriers presented worse memory performance and other cognitive alterations such as diminished orientation, judgment, and problem-solving skills. This result indicates that the ε4 allele correlates with memory problems and changes in a broader range of cognitive functions, corroborating published data (Yasuno et al., 2012). This conclusion was further strengthened by observing a potential association between the presence of the ε4 allele and ADL-dependent individuals (p = 0.083). Therefore, in the study population, APOE ε4 proved to be a risk factor for the existence of cognitive deficits, contributing to doubling the risk of having a score of CDR >1 (OR = 2.527; p = 0.031), often associated with dementia. Nevertheless, early identification and management of dementia in the primary care setting remain a challenge (Parmar et al., 2014), and dementia is often underdiagnosed by clinicians in primary care centers and underreported by patients and families (Amjad et al., 2018). Due to dementia-related symptoms being considered part of the normal aging process (Schulz and Grimes, 2005), an estimated 50% of primary care patients 65 years or older are not diagnosed with this syndrome (Iliffe et al., 2009). Thus, underdiagnosed dementia might contribute to the doubled risk of CDR≥1 among the APOE ε4 carriers' group, as was observed in this study group.

In our study population, APOE ε4 allele was also associated with GDS (p = 0.037), the test evaluating depression. In the multivariate analysis, we observed a trend showing that ε4 carriers are more likely to have a depressed state (OR = 1.649; p = 0.063). This observation is in line with previous studies (Wang et al., 2019). Reports have suggested that amyloid-associated depression may precede the onset of AD, particularly in APOE ε4 carriers (Sun et al., 2008; Qiu et al., 2016). Moreover, previous research reports that the ε4 allele may increase the likelihood of depression by about 4 times in women (Delano-Wood et al., 2008). In contrast, men do not show an association between DEP and this APOE allele (Delano-Wood et al., 2008). Other studies report no association between depressive states and APOE ε4 carriers (Locke et al., 2013). Thus, more research should be conducted on the interplay between DEP and this risk gene.

Regarding BIN1, in the pcb-Cohort the G risk allele of rs744373 SNP does not seem to be associated with either socio-demographic characteristics or cognitive deficits (Table 1). Thus, we could not replicate Seshadri et al. (2010) results, which showed a significant association between that variant and AD cases. Still, it is crucial to remember that in such an intricate and complex disease as AD, it becomes more challenging to replicate GWAS discoveries due to the heterogeneity of different populations. More studies are required to evaluate the role of this and other BIN1 variants in larger populations with different characteristics, to have a greater insight into the relationship between this risk gene and cognitive deficits that might result in dementia.

In the pcb-Cohort, both the ε4 allele of APOE and BIN1 rs744373 are significantly associated with DYS. APOE ε4 allele considerably increased the risk of having DYS (OR = 1.804; p = 0.036). On the contrary, BIN1 rs744374 had a protective profile and is likely to prevent the disease by 50%, compared to non-carriers of the G allele (OR = 0.558; p = 0.006). Several studies have demonstrated associations between the different APOE haplotypes and plasma levels of lipids and lipoproteins (Sing and Davignon, 1985; Boerwinkle and Utermann, 1988; Mooijaart et al., 2006). APOE ε4 carriers usually have increased total cholesterol levels and low density protein (LDL)-cholesterol (Sun et al., 2014). Also, numerous common genetic variants have a combined effect on influencing plasma levels of HDL cholesterol (Spirin et al., 2007). Previous research suggested that higher plasma APOE and high-density lipoproteins (HDL) from early life might preserve cognitive functions in later life, especially in APOE ε4 carriers (Yasuno et al., 2012). A longitudinal study showed that APOE ε4 non-carriers with AD, have cholesterol metabolism dysfunction and functional harm with raised HDL-cholesterol levels, possibly due to lower availability of lipids to neuronal membranes (de Oliveira et al., 2017). Over the years, researchers have explored the interplay between DYS and dementia, however, the role of cholesterol in AD is still debatable. A study reported that protective variants of APOE against risk of AD also slow cognitive decline in patients with dementia, regardless of cholesterol variations, while therapy with lipophilic statins might benefit carriers of specific genetic variants (de Oliveira et al., 2020). The same was observed regarding two other protective variants, particularly among APOE ε4 carriers with AD (De Oliveira et al., 2022). To our knowledge, this is the first time that an association between BIN1 rs744373 and DYS has been reported. Further studies should elucidate the nature of this association and how important it might be for dementia.

Given that APOE ε4 and BIN1 are considered risk factors for AD and that previous studies have shown an association between respiratory diseases such as COPD and dementia (Villeneuve et al., 2012; Singh et al., 2014; Liao et al., 2015), one would expect that carriers of these risk variants would be more prone to RESP. Our data indicate a low association trend between APOE genotype and RESP diseases (p = 0.088). However, contrary to what we expected, the ε4 allele might decrease the likelihood of RESP (OR = 0.515). Similarly, BIN1 seems to be a protective factor for RESP as the G allele significantly reduces the risk of having RESP (OR = 0.556; p = 0.026). Further studies should address the mechanism by which APOE ε4 or BIN1 might influence the pathogenesis of RESP.

Our results also support a strong trend of APOE ε2 allele as a protective factor toward DM, with a reduction of type 2 diabetes odds by 66% in carriers of that allele (OR = 0.342; p = 0.093). Diabetic individuals have an increased risk of developing AD (Biessels et al., 2006), thus ε2 carriers may be less prone to both DM and AD. In a primary care setting, the screening and control of type 2 diabetes may aid in dementia prevention. Nevertheless, future studies in larger populations need to validate these observations.

Surprisingly, a higher association trend between the risk allele of BIN1 SNP rs744374 and DM was also detected (p = 0.099). It is important to emphasize that in the present study, while BIN1 showed a protective profile regarding DYS and RESP, in the case of DM, the rs744373 variant seems to be a risk factor (OR = 1.491), which also differs from the results of APOE ε2 (OR = 0.342). A previous study by Vacínová et al. (2017) reported no association between DM and the BIN1 SNP rs744373. Further, a recent study showed that, among ≥65 years old subjects with DM, the rs6733839 variant of BIN1 may contribute to individual changes in episodic memory performance (Greenbaum et al., 2016). Besides DM increasing AD risk by 2-fold (Mayeux and Stern, 2012), it is also a potential changeable risk factor for developing this type of dementia. Therefore, it might be altered or strengthened by other risk factors such as genetic causes (Lindenberger et al., 2008) and influenced by BIN1. Despite these findings, additional large-scale genetic studies in different populations are required to unravel the possible roles of BIN1 in the overlap between the two pathologies. Also, future studies should explore if the coexistence of the rs744373 variant and DM could contribute to cognitive deficits.

Overall, the present study represents an essential step in elucidating genomic contributions to dementia and how LOAD's two top risk genes might be associated with other diseases influencing the onset and development of cognitive impairment. There are, nonetheless, some limitations. First, the small sample size of this study may be underpowered to detect the minor effects of genetic variants. Likewise, since it was beyond this study's scope, we did not address gene-gene or gene-environment interactions, which would be an asset given that cognitive performance is multifactorial. Further, we were not able to detect minor effects/confounding factors.

For these reasons, and as mentioned above, it is imperative to reproduce and validate these results in other cohorts with larger populations and investigate the possible association of other BIN1 polymorphisms with cognitive deficits.