Abstract
Androgen deprivation therapy (ADT) for prostate cancer is associated with an increased risk of dementia, including Alzheimer’s disease (AD). The mechanistic connection between ADT and AD-related cognitive impairment in patients with prostate cancer remains elusive. We established a clinically relevant prostate cancer–bearing AD mouse model to explore this. Both tumor-bearing and ADT induce complex changes in immune and inflammatory responses in peripheral blood and in the brain. ADT disrupts the integrity of the blood-brain barrier (BBB) and promotes immune cell infiltration into the brain, enhancing neuroinflammation and gliosis without affecting the amyloid plaque load. Moreover, treatment with natalizumab, an FDA-approved drug targeting peripheral immune cell infiltration, reduces neuroinflammation and improves cognitive function in this model. Our study uncovers an inflammatory mechanism, extending beyond amyloid pathology, that underlies ADT-exacerbated cognitive deficits, and suggests natalizumab as a potentially effective treatment in alleviating the detrimental effects of ADT on cognition.
INTRODUCTION
In 2023, the American Cancer Society estimated that there were approximately 268,490 new cases of prostate cancer and 34,700 deaths from prostate cancer in the United States (1). Androgen deprivation therapy (ADT), which reduces the levels of androgen hormones, prevents prostate cancer cells from growing (2), thereby effectively improving the life span and quality of life for patients with prostate cancer (3). Because of the well-documented survival benefit in a wide breadth of clinical scenarios, ADT has been widely used globally for prostate cancer treatment (4–6). Now, one-half of all men with prostate cancer receive ADT after diagnosis (4). However, in the long term, ADT increases the risk of osteoporosis, obesity, and diabetes (7, 8). Likewise, from clinical studies of >300,000 patients with prostate cancer, the risk for dementia, including Alzheimer’s disease (AD), in patients treated with ADT, is 1.5- to 2.5-fold higher than the general population followed for up to 10 years (9–14), raising a severe concern regarding application of ADT for elderly patients with prostate cancer.
AD is the most common form of dementia, and the risk for AD markedly increases with aging. It is estimated that ~10% of people older than 65 live with AD in the United States (15). Coincidentally, prostate cancer is also more likely to occur in elderly men at the age of 65 or older. Therefore, uncovering the mechanisms underlying the association between ADT and AD holds strong translational significance. The high heterogeneity of prostate cancer and AD, together with the potentially complex interaction among ADT, prostate cancer, and AD, poses major barriers to clinical mechanistic studies. Few, if any, experimental studies in the literature have examined how ADT affects AD onset and progression in patients with prostate cancer. Thus, understanding how ADT affects AD in patients with prostate cancer defines a currently unmet medical need, which is necessary for developing effective therapeutic strategies to minimize the negative effect of ADT on cognition and thus significantly benefiting the large population of patients with prostate cancer.
Animal models have served as valuable tools to examine the role of androgens on aging and AD. Previous studies have mainly focused on the effect of androgen depletion by gonadectomy on β-amyloid (Aβ) metabolism and suggest androgens acting as a negative regulator of Aβ generation and accumulation (16–18). While the rodent models used in these studies are highly informative for testing how androgen regulates Aβ, their applicability to address the impact of ADT on AD is limited, as these animal models do not involve tumors. Consequently, the contribution of prostate cancer to the ADT-AD interaction cannot be investigated using these models. Because both prostate cancer and ADT lead to altered immune responses (19, 20), which may significantly affect AD progression (21), we argue that, to be more relevant to clinical settings, the presence of prostate cancer is essential for testing the effect of ADT on AD. In addition, previous studies have solely used gonadectomy to achieve androgen depletion, and none of the ADT drugs used in humans have been assessed in this context. Thus far, aside from generic Aβ metabolism, previous mechanistic studies have not yet led to the identification of druggable targets specific to the ADT-AD interaction.
The challenges of addressing the relationship among prostate cancer, ADT, and AD by clinical studies, coupled with the absence of clinically relevant animal models, have prompted us to establish a clinically relevant tumor-bearing AD model to examine the effects of ADT for prostate cancer on AD-related pathology and cognitive deficits. Employing this model, we identified blood-brain barrier (BBB) damage and subsequent peripheral immune cell infiltration into the brain, rather than Aβ deposition, as the primary driving factor behind AD-related neuroinflammation and cognitive decline in response to ADT at the early stage. Guided by these novel mechanistic discoveries, our further translational studies indicated that natalizumab (NAT), an Food and Drug Administration (FDA)–approved treatment for multiple sclerosis (MS) that inhibits immune cell infiltration, may serve as an effective therapy to mitigate the risk of AD-related cognitive impairment in patients with prostate cancer undergoing ADT and even in patients suffering other disease conditions that result in BBB damage and immune cell infiltration.
RESULTS
ADT accelerates AD-related cognitive deficits in tumor-bearing AppKI mice
In human prostate cancer, PTEN and its relevant signaling genes are the most frequently altered genes (22–24). Up to 70% of patients with advanced prostate cancer exhibit loss of PTEN function or activation of the phosphatidylinositol 3-kinase/AKT pathway (25). In mice, nearly all prostate-specific Pten knockout (KO) male mice develop prostate cancer with aging (26). Homozygous prostate-specific Pten-deficient male mice respond to androgen withdrawal (26) in a way similar to that observed in human prostate cancer with ADT (27). Thus, the Pten KO PTEN-CaP8 prostate cancer cell line derived from this model recapitulates prostate cancer features and responses to ADT treatment in human patients. We engrafted this murine Pten KO prostate cancer cell line with a well-established AD model, AppNL-G-F/NL-G-F knock-in (referred to as AppKI) mice, to generate a tumor-bearing AD model to examine the effect of ADT on prostate cancer growth and AD-related pathological and behavioral changes.
Antiandrogens and chemical castration are two major methods of ADT in the clinic. Antiandrogens can enter cells to prevent the binding of testosterone to the receptor proteins, due to their higher affinity for the androgen receptor (AR). A CYP17A1 inhibitor, abiraterone acetate (ABA), is a second-generation antiandrogen that impairs AR signaling by inhibiting adrenal and intratumoral androgen biosynthesis in vivo (28). Chemical castration can be achieved by luteinizing hormone (LH)–releasing hormone agonists and antagonists, which both inhibit the formation of LH in the pituitary gland to lower the amount of testosterone. Degarelix is a synthetic peptide derivative drug that reduces LH and follicle-stimulating hormone, which ultimately causes testosterone suppression and increases patient survival in clinics. ABA and degarelix are commonly used in the treatment of patients with prostate cancer, and the effectiveness of combined treatment with these two drugs has also been demonstrated in clinical trials (29, 30) (Clinical Trial Phase II: NCT06060587) and animal models (31–33). Therefore, we opted to perform ADT in our mouse model using surgical castration followed by combined treatment with ABA and degarelix. This approach ensures more effective androgen deprivation, as surgical castration alone does not completely eliminate androgens in the body, given that other sources, such as the adrenal glands, continue to produce androgens in small quantities.
Our goal is to study the effect of cancer and ADT on the development of AD-related pathology and cognitive impairment. Because aging can potentially affect the outcomes of both prostate cancer and AD phenotypes, we designed the timeline of tumor injection, surgery, and ADT treatment as depicted in Fig. 1A, guided by the following considerations. We implanted prostate cancer cells at 8 weeks of age, immediately after the completion of prostate development and before the onset of Aβ deposition in the AppKI model (34, 35). At this age, tumor cells also engraft efficiently. Two weeks after tumor cell implantation, when the injected tumor cells had formed a solid tumor, we performed castration surgery. From this age, testosterone remains at a stable high level, making it suitable for examining the role of this hormone in the development of AD-related phenotypes. After a 2-week recovery period from surgery, we started ADT drug treatment following a regimen that resembles the clinical setting. Cognitive behavior in mice was then assessed at 20 weeks of age. Our observations indicate that AppKI mice begin to exhibit cognitive impairments between 5 and 6 months, providing a sensitive timeframe for detecting differences among various groups.
In addition to AppKI mice, we included wild-type (WT) C57BL/6 mice with sham surgery and vehicle treatment as controls for comparison. All mice were randomly assigned to four groups: Group 1 (vehicle + sham surgery in WT mice), group 2 (vehicle + sham surgery in AppKI mice), group 3 (tumor + vehicle + sham surgery in AppKI mice), and group 4 (tumor + castration surgery + ADT in AppKI mice). We evaluated tumor size every week along the course of treatment. Tumor growth was slower in group 4 given castration surgery combined with ADT than that of group 3 without castration surgery and ADT (Fig. 1B, bottom). The size of tumor engrafts in group 4 was smaller than that in group 3 at 20 weeks of age (Fig. 1B, top). In addition, after 1 week of ADT, we measured blood testosterone level and validated the effectiveness of ADT (Fig. 1C). These results confirm the ADT-mediated tumor suppression of prostate cancer in our tumor-bearing AD mouse model.
In human prostate cancer patients, ADT impairs cognitive functions. We first sought to determine whether the detrimental effect of ADT on cognition could be recapitulated in our tumor-bearing AD model. We accessed associative learning in a passive avoidance paradigm. At 20 weeks of age, group 4 tumor-bearing AppKI mice treated with ADT, but not the other three groups, showed cognitive deficits, as manifested by no notable difference in latency to dark between the two test days (Fig. 1D). The log ratio of latency on day 2 over day 1 in group 4 was significantly lower than that in group 1 WT mice (Fig. 1E). Compared to other groups, group 4 was associated with a less percentage of successful avoidance (defined by the ratio of latency on day 2 over day 1 ≥ 3; Fig. 1F). Only 38.5% of group 4 mice displayed successful avoidance, whereas 75% of group 1, 50% of group 2, and 55.6% of group 3 mice did so (Fig. 1F). These data suggest that ADT accelerates the onset of AD-related cognitive deficits.
In addition, we evaluated the general activity of mice in an open field and their innate anxiety in elevated zero maze (EZM) tests. No difference was observed among the groups in total distance traveled or the percentage of time they spent in the center area of the open field (fig. S1). However, in EZM, the group 4 mice spent a significantly less amount of time in the open zone (Fig. 1G) and more time in the closed zone (Fig. 1H) compared to group 2 AppKI sham control mice. These data suggest that ADT treatment also increases anxiety-like behaviors in tumor-bearing AppKI mice.
ADT alters peripheral adaptive and innate immune responses in tumor-bearing AppKI mice
To gain insights into potential mechanisms underlying the detrimental effect of ADT on cognition in AD, we conducted detailed characterization of our tumor-bearing AD model mice. Given that both tumor and ADT can modulate immune responses, we first investigated changes in immune responses and inflammation in tumor-bearing AppKI mice with ADT treatment. We started by examining the innate and adaptive immune responses in the peripheral blood. We analyzed the number and relative abundance of different types of immune cells by flow cytometry in all groups of mice at 20 weeks of age. As shown in Fig. 2 (A and B), CD3+ T cells were markedly increased in group 4 tumor-bearing AppKI mice treated with ADT compared to the other three groups. CD4+ T cells were likely to be increased by ADT in group 4 compared to groups 2 and 3, but not different compared to group 1 (Fig. 2, C and D). CD8+ T cells in groups 2, 3, and 4 were lower than that in group 1, but not different among groups 2, 3, and 4 (Fig. 2E). Furthermore, CD8+ T cells with activated cytotoxic molecule Granzyme B were likely to be decreased in groups 3 and 4 compared to groups 1 and 2, but no statistical significance was found (Fig. 2F), whereas CD8+ T cells with exhausted marker programmed cell death protein (PD-1) were markedly increased in group 4 than in the other three groups (Fig. 2, G and H). The number of regulatory T (Treg) cells in group 4 was higher than that in groups 1 and 2, but compared to group 3, was significantly decreased (Fig. 2I). B cells were increased in group 4 compared to groups 1 and 2, but not significantly different compared to group 3 (Fig. 2J)….
