Determining the Role of Sex and APOE4 status on Psychosis in Alzheimer’s Disease

Abstract

Background

Psychosis occurs in approximately 41% of patients living with Alzheimer’s disease. Previous findings from our group based on analyses of a neuropathological cohort suggest that among AD patients with Lewy Body pathology, female APOE4 homozygotes are at significantly greater risk of psychosis. This study aims to replicate this finding in a clinical cohort using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset.

Methods

Our group used data from a sample of patients with AD in the ADNI database from the ADNI1, ADNI2, ADNI3, and ADNIGO studies. We defined psychosis status as experiencing hallucinations or delusions at one time point based on the Neuropsychiatric Inventory. We then used forward binary logistic regression to determine if sex and APOE4 status are predictors of AD + P.

Results

In total there were 204 participants who met the inclusion criteria, 133 of which were male, and 71 of which were female. Fifty-six patients were APOE4 non-carriers, 109 patients were APOE4 heterozygote carriers, and 39 were APOE4 homozygote carriers. In total, there were 59 patients with psychosis. When adjusting for mini mental state examination score, adjusted hippocampal volume, and age, we demonstrate that female APOE4 homozygotes have a significantly increased risk of psychosis compared to other groups (P = 0.0264, OR = 19.50).

Discussion

The results of our study demonstrate a significant association between psychosis risk and female APOE4 homozygotes, thus corroborating findings from a neuropathological cohort. The effects of APOE ε4 on psychosis risk are significant only in females, and not in males.

Introduction

Alzheimer’s disease (AD) is among the most prominent neurodegenerative diseases leading to severe cognitive decline in people over 65 years of age. ¹ AD is notably marked by the accumulation of the protein fragment amyloid-β (Aβ) into plaques ² and hyperphosphorylated Tau (pTau) protein molecules into neurofibrillary tangles (NFTs). ³ Research has determined that there is no singular cause of AD. Rather, it has many risk factors, such as older age, head injuries, diabetes, hypertension, smoking, obesity, and various environmental factors, such as lower education and air pollution. ⁴ AD risk also has genetic risk factors, the most prevalent of which is the APOE ε4 allele. The APOE gene codes for the apolipoprotein E (apoE), which is involved in lipid transport between tissues and cells. ⁵ Studies in humans and animals demonstrate APOE’s ability to affect Aβ clearance. APOE comes in 3 major alleles numbered 2 to 4. ⁶ Of these alleles, ε3 is the most common, and has the least known effect in regards to AD risk. ε2 is the least common, and has been observed to exert a protective effect against AD. On the other hand, carriers of the ε4 allele experience increased risk of AD compared to ε3 and ε2 carriers. ⁶ ε4 carriers display decreased age of AD onset, as well as increased risk of vascular damage, hyperlipidemia, hypercholesteremia, and greater cognitive deficits.^7,8 It is also worth noting that while the global frequency of the ε4 allele is around 13.7%, in AD patients, the frequency of the allele rises to around 40%. The specific mechanisms by which APOE4 increases AD risk are not fully known, however, APOE4 is a potential therapeutic target for treating AD. ⁸

Previous research has shown that some effects of the APOE4 allele are more pronounced in females than in males. For example, female carriers of the APOE4 allele are more likely to develop AD than male APOE4 carriers, and have worse cognitive deficits. ⁹ Previous research by our group also demonstrates that neuropathologically confirmed female AD patients homozygous for APOE4 carriers with Lewy body pathology are more likely to develop certain neuropsychiatric manifestations of AD than female APOE4 non-carriers, such as psychosis.^10-12 Importantly, this increased risk for APOE4 carriers is only observed in females, and not in males^10,11 based on analyses of data from a pathological cohort. Moreover, effects were accentuated in the presence of Lewy body pathology and not present among patients without Lewy body pathology.^10,11

Psychosis in AD (AD + P) is found in approximately 40-60% of AD patients, and is associated with accelerated decline in cognitive function and development of dementia. ¹³ Studies show that AD + P likely has a genetic component, since it has been demonstrated to be heritable. ¹⁴ In addition, newly established research criteria for psychosis in AD ¹⁵ suggest patients may suffer from delusions (persecutory/misidentification) and/or hallucinations, in addition to other symptoms of AD. Previous studies have found that sex and APOE4 status are significant predictors of AD + P, particularly in APOE4 homozygous females with AD.^10,11 However, as discussed these studies were limited to neuropathological cohorts. This study aims to further explore the association between sex, APOE ε4 status and risk of AD + P in a clinical cohort using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI).

Methods

The sample population was collected from 4 studies done by the Alzheimer’s Disease Neuroimaging Initiative, namely ADNI1, ADNI2, ADNI3, and ADNIGO (Figure 1). Study participants were identified based on the availability of data for mini-mental state examination (MMSE) scores, Neuropsychiatric Inventory Questionnaire (NPIQ) scores, hippocampal volume, age, sex, and APOE4 status. Psychosis subscores in the NPIQ were used to determine if the participant had experienced psychosis or not. If the patient scored positively for either hallucinations or delusions on one visit at minimum, they were considered psychotic.

Figure 1.

Breakdown of participant demographics by APOE status, sex, and psychosis status.

Analyses (ANOVA, post-hoc tests, t-tests) were completed in R version 4.3.1, to determine if there are significant differences in certain population demographics between the AD + P and AD − P groups, and a forward stepwise binary logistic regression model was used to determine if there is a correlation between psychosis status and APOE4 status to perform the cross-sectional study. The independent variables, in order of implementation into the model, were APOE4 status, sex, MMSE, age, and adjusted hippocampal volume. Variables are considered significantly associated with AD + P if their P-value is equal to or less than 0.05. Hippocampal volume was adjusted by performing intracranial volume normalization for each patient’s hippocampal volume. The odds ratio and confidence interval for APOE4 status were also determined, by exponentiating the beta coefficient.

Results

In total, there were 204 participants, mean age 74.7 (7.96), 34% female. Fifty-six patients were APOE4 non-carriers, 109 patients were APOE4 heterozygotes, and 39 were APOE4 homozygotes. These groups were further divided by sex, and then by psychosis status. In total, there were 59 participants with psychosis and 145 without psychosis. (See Table 1 – demographics)

Table 1.

Population Demographics of Psychosis and Non-psychosis Groups.

	Psychosis (P)	Non-psychosis (NP)	P-value	% P	% NP	SD
Males	33	100	N/A	16.2%	49.0%	N/A a
Females	26	45	N/A	12.7%	22.1%	N/A
APOE4 Homozygous	15	24	N/A	7.3%	11.8%	N/A
APOE4 Heterozygous	34	75	N/A	16.7%	36.8%	N/A
APOE4 Non-carrier	10	46	N/A	4.9%	22.5%	N/A
Average Age	76.83	73.81	0.007**	N/A	N/A	7.96
Average MMSE	23.12	23.39	0.505	N/A	N/A	2.30
Average NPIQ Scores	5.95	3.15	1.32 × 10−5****	N/A	N/A

*P-value is < .05; **P-value is < .01; ***P-value is < .001; ****P-value is < 10⁻⁵.

^aIf a variable is given an N/A, it means that the statistical value is not applicable.

In females (see Table 2), APOE4 homozygosity had a significant correlation with psychosis status (P value = 0.0242, OR = 19.42, CI = [1.72:332.24]), while in males, APOE4 homozygosity did not (P value = 0.097, OR = 5.60, CI = [0.84:54.57]) though there is a trend towards significance. Neither male nor female APOE4 heterozygotes had a significant correlation with psychosis status. Both male and female APOE4 homozygotes were more likely to develop psychosis compared to their respective non-carriers, though significance was only achieved in females.

Table 2.

Associations Between Psychosis Risk and APOE4 Status in Male and Female AD Patients.

Variable	P Value		Odds Ratio		Confidence Interval
	Male	Female	Male	Female	Male	Female
APOE4 Homozygotes	0.097	0.0242*	5.600	19.42	0.84:54.57	1.72:332.24
APOE4 Heterozygotes	0.456	0.144	2.035	5.208	0.35:17.12	4.23 × 10−8:0.036
Age	0.00317**	0.00436**	1.120	1.174	1.044:1.215	1.063:1.329
Adjusted Hippocampal Volume	0.827	0.643	1.000	1.000	1.000:1.001	0.999:1.001
MMSE	0.618	0.172	1.064	0.796	0.835:1.362	0.565:1.097

**P‐value is less than .005

Discussion

Our results show that psychosis status was found to be significantly correlated with sex, age, and APOE4 status with adjusted hippocampal volume as a covariate. APOE4 status was determined to be a significant predictor of psychosis in AD, and having more APOE4 alleles increased the risk of developing psychosis. This finding is corroborated by previous papers that report similar results. ¹⁶ This suggests that APOE4 is a significant risk factor of AD + P. The analysis also finds that while APOE4 heterozygosity and homozygosity both increase the risk of developing psychosis, only APOE4 homozygosity significantly predicts psychosis status, and only among females. In addition, APOE4 status only achieved significance for female homozygotes when adjusted hippocampal volume was added as a modifying variable. This suggests that the hippocampus is additionally involved in the development of AD + P pathology, but the odds ratio for adjusted hippocampal volume suggests that there is no exacerbated hippocampal atrophy in AD + P compared to AD-P.

Previous studies have found that there are significant sex differences in AD + P. For example, a previous study by our group done using data from the NACC pathological cohort found that female APOE4 homozygotes were significantly more likely to develop AD + P than female APOE4 heterozygotes and non-carriers, but that this effect was not replicated in males. ¹⁰ No male group was significantly associated with AD + P risk. Males are also significantly less likely to develop psychosis compared to females when adjusting for MMSE and APOE4 status. It should be emphasized that this study occurred in a pathological cohort and findings were accentuated in the presence of Lewy body pathology so may not have direct relevance to the present study.

Possible mechanisms for the increased effect of APOE ε4 in females compared to males could be due to the action of estrogen, as it has been previously shown that psychotic episodes are more common during periods of low estrogen, such as menopause. ¹⁷ In addition, previous research by Ko et al demonstrates that short-term hormone replacement therapy (HRT) of estrogen and progesterone in women with chronic schizophrenia led to increases in cognitive performance on a variety of tests, however, the long term effects of estrogen HRT in treating psychosis are not well-studied. ¹⁸ It is important to note that estrogen has beneficial neural effects, and APOE is implicated in those effects. Previous literature also reports interactions between estrogen and APOE in the development of AD, as APOE protein expression can be induced by estrogen as demonstrated in animal models and cell culture. However, the effects of estrogen depletion on the expression and function of APOE and AD development are not fully known. ¹⁹ Thus, it could be possible that in menopausal women, lower estrogen levels are affecting APOE ε4 expression in a way that is correlated with AD + P risk.

While males typically have an earlier age of onset and greater incidence of schizophrenia compared to females, the involvement of testosterone and any potential associations with APOE ε4 are not well reported. Results are mixed, with some studies reporting lower testosterone in males with schizophrenia, ²⁰ or no difference between them and healthy controls. ²¹ Inflammatory markers were also considered as a possible mechanism to explain the association of APOE ε4 and AD + P, but no significance was found in either APOE ε4 homozygous or heterozygous males and females.^22,23

While our results are of interest there were some limitations that should be addressed. First, the diagnosis of AD was based on clinical data and not neuropathologically verified. Second, psychosis was based on NPI scores as opposed to formal criteria. Greater verification of both AD and psychosis status will greatly improve the validity of the study’s methodology. Finally, the sample was relatively small compared to other studies, in particular when adjusting for sex and APOE4 status. Our study is also limited in that we do not have enough data to determine which specific symptom or combination of symptoms of psychosis is more common, and if there are any differences in how each symptom is associated with APOE4 and sex. A previous publication by Naasan et al finds that hallucinations and other closely related symptoms are more common in people with Lewy body disease and Alzheimer’s disease, while delusions are more common in frontotemporal lobar degeneration. ¹² Additionally, they state that the particular combination of psychotic symptoms can be used in predicting underlying neurological pathologies. ¹² While our present study does not determine the occurrences of hallucinations and delusions in this much detail in the ADNI dataset, it is worth looking into for future research. Finally, a further limitation is that we did not adjust for the presence of other neuropsychiatric symptoms, such as depression or anxiety. These factors are outside the scope of this publication, and will be explored in future studies.

In summary our study demonstrated a significant effect of sex and APOE4 status in which female homozygotes with AD are at particular elevated risk of psychosis, thus corroborating previous findings from a neuropathological cohort.^10,11 Additionally, when adjusting for sex and APOE4 status, increased hippocampal volume appears to mediate this risk. Further studies are required to further delineate the mechanisms behind this effect.