The Cancer-Alzheimer’s Disease Nexus: Exploring Relationships, Mechanisms, and Therapeutic Implications Workshop
On this page:
Agenda
9:00 a.m.: Introductions and Welcome: Eliezer Masliah, M.D., NIA and Damali Martin, Ph.D., MPH, NIA
9:20 a.m.: Meeting overview and logistics: Damali Martin
Session 1: Why the joint interest in cancer and Alzheimer’s disease?
Session Chair: Mary Ganguli, MBBS (MD), MPH, University of Pittsburgh
- 9:30 a.m.: Mary Ganguli, MBBS (MD), MPH, University of Pittsburgh
- 9:40 a.m.: Disentangling the cancer-dementia relationship: Lessons from an epidemiological paradox, Lindsay Kobayashi, Ph.D., MS, University of Michigan
- 10:10 a.m.: Epidemiology of the inverse comorbidity of dementia and cancer: Biostatistical challenges and opportunities, Michelle D. Shardell, Ph.D., University of Maryland
- 10:40 a.m.: AD and cancer as a systems level maladaptation to stressors: From mechanisms to diagnostics and treatment, Gabriela Chiosis, Ph.D., Memorial Sloan Kettering Cancer Center
- 11:10 a.m.: Q & A, Panel of Speakers
- 11:20 a.m.: BREAK
Session 2: Genes, Mechanisms, and Epidemiological Evidence
Session Chair: Marcel Salive, M.D., MPH, NIA
- 11:30 a.m.: Risk of ADRD by vascular diseases and tumor factors in long-term cancer survivors, Xianglin L. Du, M.B., Ph.D., MS, University of Texas
- 11:45 a.m.: The good and bad biases of the cancer-dementia association: An epidemiologic assessment of the literature, Monica Ospina-Romero, M.D., MAS, University of Wisconsin
- 12:00 p.m.: The molecular genetics of cancer and Alzheimer's disease, Kelly Nudelman, Ph.D., Indiana University
- 12:15 p.m.: Investigating the genetic relationship between Alzheimer's disease and cancer using GWAS summary statistics, Liming Liang, Ph.D., Harvard University
- 12:30 p.m.: Q & A, Panel of speakers
- 1:00 p.m.: LUNCH
Session 3: Cancer Chemotherapy and Cognitive Dysfunction
Chair: Todd Horowitz, Ph.D., BRP, DCCPS, NCI
- 1:45 p.m.: What can CRCD tell us about ADRD (and vice versa)- Lessons from the Thinking and Living with Cancer Study, Jeanne Mandelblatt, M.D., MPH, Georgetown University
- 2:00 p.m.: From neuroimaging to multi-omics: A systems approach to cognitive changes in cancer and Alzheimer’s, Andrew J Saykin, Psy.D., Indiana University
- 2:15 p.m.: Management of Cancer Related Cognitive Decline: State of the science on intervention strategies, Allison Magnuson, D.O., MS, University of Rochester
- 2:30 p.m.: Q & A, Panel of Speakers
- 2:45 p.m.: Discussion and wrap up of day 1
9:00 a.m.: Goals for the day’s activities, Damali Martin, Ph.D., MPH, NIA
Session 4: Meds, Mechanisms, and Drug Repurposing
Session Chair: Paul Grothaus, Ph.D., NIA
- 9:10 a.m.: Reversing Alzheimer’s disease gene network states by approved cancer and inflammatory drugs: From informatics to bench to clinical trial, Mark Albers, M.D., Ph.D., Harvard University
- 9:25 a.m.: Neuronal fate loss and metabolic transformation in age-equivalent Alzheimer patient neurons, Jerome Mertens, Ph.D., University of Innsbruck, Austria
- 9:40 a.m.: Microtubule-normalizing agents as potential therapeutics for AD and related neurodegenerative disorders, Kurt R. Brunden, Ph.D., University of Pennsylvania
- 9:55 a.m.: Q & A, Panel of Speakers
- 10:20 a.m.: BREAK
- 10:30 a.m.: World Café Breakout Session, Damali Martin and Camille Pottinger, NIA
- Participants will contribute ideas that address current gaps in the field and set future research directions. Each group will have a facilitator and note-taker.
- 12:30 p.m.: LUNCH (gather all responses)
- 1:30 p.m.: Report back & Presentation of responses, Camille Pottinger, MPH, NIA
- 2:00 p.m.: Open discussion of future directions, Dallas Anderson and Damali Martin, NIA
- 3:00 p.m.: Closing remarks and meeting ends, NIA staff
Executive Summary
The Cancer-Alzheimer’s disease nexus: Exploring relationships, mechanisms, and therapeutic implications workshop was held on October 18-19, 2022. This summary highlights findings and conclusions for each of the discussions.
Day 1: Tuesday, October 18 | Introductions, Welcome, and Meeting Overview
Eliezer Masliah, M.D., NIA, and Damali Martin, Ph.D., MPH, NIA
Dr. Martin opened the meeting at 9:03 a.m. ET and welcomed attendees. She introduced herself and her colleagues on the planning committee, then welcomed Dr. Masliah for his opening remarks.
Dr. Masliah thanked Dr. Martin, the meeting attendees, speakers, and workshop organizers. He expressed his excitement for the workshop, which has been in development for years and became possible due to advancements in research on aging, Alzheimer's disease (AD), cancer, and the links between them. He recalled an experience from his time as a postdoctoral researcher when he conducted an autopsy and found the signs of AD in one hemisphere of the brain and a glioblastoma in the other. This suggested to him that factors produced by the tumor affected, or were affected by, the neuropathology in the other brain hemisphere. Since then, Dr. Masliah has had a particular interest in the relationship between neurodegeneration and cancer and believes this meeting will be an important step in increasing the understanding of gaps and opportunities in these research fields.
Dr. Martin reviewed the purpose of the meeting—to understand the relationship between AD and cancer. Exploring this connection may offer unique opportunities to understand the mechanisms of both disease types and may also lead to the repurposing of cancer therapeutics for the treatment of AD. She then reviewed the agenda and meeting logistics, inviting attendees to type their questions for the speakers in the Zoom chat box and reminding them to keep themselves muted when not speaking. She then introduced Dr. Ganguli, the chair of the first session.
Session 1: Why the joint interest in cancer and Alzheimer’s disease?
Session Chair: Mary Ganguli, M.B.B.S. (M.D.), MPH, University of Pittsburgh
Disentangling the cancer-dementia relationship: Lessons from an epidemiological paradox
Lindsay Kobayashi, Ph.D., MS, University of Michigan
Dr. Kobayashi provided a high-level introduction to the cancer-AD nexus.
The Paradoxical Inverse Cancer-AD Association
The inverse relationship between cancer and AD has a high impact on the population. Since the 1970s, the prevalence of cancer survivorship has been dramatically increasing. This is especially true among adults aged 65 and older, who now comprise the largest share of cancer survivors in the United States.
Both cancer and cancer treatments are thought to impact the aging of several bodily systems, including cognitive functioning. Adult cancer survivors have higher prevalence of disability, chronic pain, comorbidities, anxiety, and depression than their age-matched cancer-free counterparts. Even survivors of childhood cancer experience incidence rates of comorbidities, frailty, and functional impairment that match those seen in the cancer-free population at much older ages.
Paradoxically, large population-based epidemiological studies have consistently identified an inverse relationship between the incidence of cancer and AD. Older cancer survivors appear to have lower dementia risk and better memory function than similarly aged cancer-free adults. These findings are especially surprising because cancer survivors experience a decrease in cognitive function compared with their cancer-free counterparts and some of the most common cancer treatments produce direct neurotoxic effects.
To understand this apparent contradiction, it is important for researchers to discern whether the inverse cancer-AD association is causal or the result of study biases. Two possible sources for a causal association are a shared common etiology and an unknown confounding factor that Dr. Kobayashi designated “Factor U.”
Findings from a Study on Long-Term Memory Aging
Dr. Kobayashi and her colleagues conducted a study using data from the U.S. Health and Retirement Study (HRS) to explore both possibilities. The analysis used longitudinal data from a nationally representative cohort of around 14,500 adults aged 50 and older who had no cancer history at the time of their enrollment in the study. Study participants were followed up via telephone interview every 2 years from 1998 to 2014. Each follow-up call captured any new cancer diagnosis as well as assessments of episodic memory.
The analysis compared the rate of memory change in the long term before cancer diagnosis, immediately at the time of diagnosis, and in the long term after diagnosis among cancer survivors to the rate of aging-related memory change over the same period in age-matched cancer-free individuals. The researchers used segmented linear mixed models adjusted for confounders, including social, behavioral, and health-related factors that could be common causes of cancer and dementia.
Findings from the analysis included:
- A significant memory advantage among individuals who would later go on to develop cancer
- An acute drop in memory function among participants with cancer around the time of their diagnosis, consistent with evidence from smaller, clinic-based studies of cancer survivors—a phenomenon known as “chemo brain” or cancer-related cognitive decline (CRCD)
- Better cognitive function among older cancer survivors compared with their cancer-free counterparts both before and after diagnosis.
These results support the long-term inverse association between cancer and AD and provide new evidence that the cognitive advantage that cancer survivors experience emerges in the years prior to their diagnosis and treatment. They also reinforce the short-term decline in memory that individuals with cancer experience at the time of diagnosis. This suggests that there may be a common inverse cause of cancer and dementia that arises before either disease state occurs, but it is still important to rule out the common sources of bias that occur in observational studies.
Potential Sources of Bias and How to Address Them
Dr. Kobayashi introduced a list of common sources of bias that could be contributing to the observed inverse cancer-AD association.
- Confounding by an unknown factor (“Factor U"): There could be a genetic or biological regulation between carcinogenesis and neurodegeneration.
- Confounding by known factors: Inappropriate statistical modeling may be used for known factors that influence the risk of both cancer and AD.
- Inappropriate control for downstream factors: Inappropriate modeling may be used for cancer-related outcomes that may influence AD risk.
- Diagnostic bias of AD status: Greater health surveillance among cancer survivors may influence the likelihood of AD diagnosis.
- Competing risks bias: Cancer increases the risk of mortality, which decreases AD risk.
- Selective survival bias: Individuals who survive cancer may have protective factors that also reduce their risk of AD.
Best practices for avoiding these biases include:
- Adjusting for common causes of cancer and the aging outcome under investigation, based on prior evidence and theory about associations. Factors may include age, sex, education, race/ethnicity, other sociodemographic factors, as well as lifestyle, environmental, health history, and social factors.
- Avoiding adjusting for factors downstream of cancer incidence, such as treatment, subsequent health conditions, or behavior changes.
- Considering how diagnostic practices may affect findings (if using medical records or claims data).
- Accounting for competing risks bias. When using data from longitudinal cohort studies, survival analysis should be conducted using techniques that account for mortality as a competing risk, and when using case-control data, incidence density sampling should be used.
- Accounting for selective survival bias. Prevalent and incident cancer cases should be separated when using data from longitudinal studies, and techniques such as weighting and sensitivity analyses should be used to mitigate bias.
Conclusions and Future Directions
Cognitive resilience support can help the aging of older cancer survivors, which is important in both clinical practice and public health policy.Further research is needed to confirm the causes of the inverse cancer-AD association. This may be achieved through triangulation across complementary, diverse data sources and disciplines, including clinical data and cohorts, registry data, genomic/biomarker data, and large population-based cohorts.
Epidemiology of the inverse comorbidity of dementia and cancer: Biostatistical challenges and opportunities
Michelle D. Shardell, Ph.D., University of Maryland
Dr. Shardell presented on methods for overcoming bias from truncation by death and selective survival in studies of the cancer-AD nexus. The right method for any given situation will depend on three factors: the study’s aim, its baseline, and when cancer is assessed.
Method 1: Explicitly State the Research Question
Mitigating selective survival bias and truncation by death begins with matching a study’s analysis approach to its aims. Examples of possible aims include identifying the:
- Association between cancer and AD among survivors. The strategy for implementing this aim is known as weighted independence estimating equations (W-IEE). In this approach, both independence and weighting are essential components of accurate analysis partially conditioned on study participants being alive, and mortality does not equal data missingness.
- Association between cancer and AD as though the entire cohort had survived. This can be achieved with mixed-effects models, which compute slopes by imputing values for decedents. In this approach, mortality and data missingness are the same.
- Causal effect of cancer on AD. With this approach, mortality and data missingness are not the same, and the analysis is not conditioned on vital status. This may permit investigators to consider the “road not taken” regarding potential outcomes. Conditioning on covariates in completers-only analysis may remove bias, and under some assumptions, this can be interpreted as a principal stratum causal effect.
Method 2: Know the Timeline
A second factor in mitigating immortal time bias and truncation by death is ensuring that cancer status at baseline is defined appropriately. Some possibilities include:
- Cancer status determined over the course of follow-up (e.g., last time assessed)
- Cancer status defined in terms of time after enrollment (e.g., early vs. late diagnosis)
- Cancer status defined in terms of reported diagnosis during at least one assessment after enrollment.
Participants may enter a study with an existing cancer diagnosis, receive a diagnosis during the study period, be diagnosed after the study has concluded, or never have cancer at all. Additionally, the first three conditions (cancer diagnosis at enrollment, mid-study cancer diagnosis, diagnosis after study conclusion) each include a period before diagnosis that may be identified as either cancer free or, more accurately, pre-cancer.
These distinctions are extremely important when examining the relationship between cancer and the time to onset of other conditions like AD to avoid immortal time bias. One solution is to treat cancer status as a time-varying covariant exposure.
Method 3: Understand the Assumptions of Analytical Methods
Investigators must be alert for many different pitfalls and forms of bias.
- In the HRS time-to-AD analysis mentioned by Dr. Kobayashi, competing risks may have prevented AD diagnosis. One strategy that has been proposed for mitigating that risk is using composite endpoints, but such use would depend on the research question.
- Other examples involve using binary or categorial data in which death after baseline is treated as one of many outcome categories. This could be achieved using a joint model.
- Cancer status can be treated as a time-varying exposure with methods such as inverse probability weighting and nested case-control design.
Conclusions
Threats to validity can be overcome only when investigators know what they are. To prevent immortal time bias and bias due to truncation by death, investigators should:
- Be very explicit about the aims of their research question.
- Know their timeline and when health conditions are assessed relative to that timeline.
- Understand the underlying assumptions of their analytical methods.
AD and cancer as a systems level maladaptation to stressors: From mechanisms to diagnostics and treatment
Gabriela Chiosis, Ph.D., Memorial Sloan Kettering Cancer Center
Dr. Chiosis presented the mechanistic commonalities between cancer and AD and began by noting that the human body is made up not only of individual molecules but of the connections between them. Like a human being, no cell, organ, molecule, or tissue exists in isolation; they all exist in an environment that is shaped by interactions with other individuals and other factors. Long-term changes to this environment can pathologically rewire its homeostasis and result in disease—a stressor maladaptation. A stressor maladaptation, in the context of stressor to phenotype, is modulated by the duration, intensity, and controllability of a specific stressor or stressors, as well as the genetic makeup, environment, and life history. Thus, individuals and their internal structures (e.g., cells, tissues, and organs) experience stressors and respond to them in different ways.
Protein-protein interaction (PPI) networks are critical in the stressor-to-phenotype context, as they encode and execute the flux of information linking stressors to phenotype at the cellular level. Because cells do not function alone, the effect of PPI networks reverberates and extends beyond individual cells into tissues, organs, and whole-organism levels. Consequently, the paradigm of therapy development needs to shift from the unitary protein approach to a goal of targeting context-specific, stressor-mediated PPI networks for disease control.
Chaperones and Epichaperomes
Transient, multiprotein complexes are important facilitators of cellular functions. This includes the chaperome, an abundant protein family comprising chaperones, co-chaperones, adaptors, and folding enzymes—dynamic complexes that regulate cellular homeostasis. Under normal circumstances, chaperome complexes act as folding machines, facilitating dynamic, short-lived interactions. But in response to chronic stressors, a small fraction of chaperones changes into long-lived oligomeric complexes, or epichaperomes. This maladaptation to stressors causes thousands of proteins to aberrantly organize inside cells, negatively affecting cellular phenotypes.
If researchers can identify how and which proteins are impacted by these formations, they may be able to better understand cell-specific vulnerabilities to chronic stressors. Developing a modality to detect these changes also could yield treatment modalities by way of disrupting the changes and reassigning epichaperome components to their prior functions.
Epichaperomes nucleate on major chaperones, and their composition is context dependent. For example, by coopting the proteins HSC70 or HSP60, HSP90 may trigger the pathologic changes seen in cancer, AD, and Parkinson’s disease. Other chaperones may be involved in the formation of these structures in inflammation or infectious disease. Thousands of proteins may be impacted.
The significant commonality between AD, cancer, and Parkinson’s disease is that they all involve a misregulation of protein pathways. If the pathological rewiring of proteins could be disrupted, function could possibly be restored via a reversal of phenotype. Data from several models of neurodegenerative disease, from cellular to animal, support this idea.
The development of drug candidates to disrupt epichaperomes may have an important role in the treatment of a broad spectrum of diseases. The target is the aberrant complement of PPIs within a specific disease context, i.e., a cell-specific interactome dysfunction, which may be corrected through epichaperome disruption.
Differentiating epichaperomes from chaperones is possible because a chaperone is both structurally and dynamically distinct. An epichaperome as a unit is as distinct from a chaperone as another protein altogether, which makes epichaperomes optimal for both drug targeting and imaging of dysfunctional PPI networks.
Targeting epichaperomes could lead to a precision medicine approach to detecting and treating complex diseases, including cancer and neurodegenerative disorders, as companion diagnostics for both patient selection and precise measurements of target engagement that are readily available for use in the clinic and as biomarkers for use in patient selection. Treating a patient at a dose and schedule regimen based on target engagement rather than maximal tolerated dose is yet another consideration to factor into individualized precision medicine treatment.
To this end, Dr. Chiosis’ team has developed several small molecules, some of which are drug candidates and some of which are diagnostic. Early clinical testing has demonstrated their efficacy in patients with cancer, amyotrophic lateral sclerosis (ALS), and AD.
Session Chair: Marcel Salive, M.D., MPH, NIA
Risk of ADRD by vascular diseases and tumor factors in long-term cancer survivors
Xianglin L. Du, M.D., Ph.D., MS, University of Texas
Dr. Du presented background on the relationship between dementia risk factors and cancer that was the catalyst for studies of complex interactions between tumor factors, cancer therapies, vascular disease, and Alzheimer's disease and related dementias (AD/ADRD) risk.
Retrospective Cohort Studies in Long-Term Cancer Survivors
The studies aimed to (1) evaluate long-term risks of ADRD associated with cardiovascular disease (CVD), stroke, hypertension, and diabetes in long-term survivors of colorectal, breast, and prostate cancers; (2) quantify effects of tumor factors and cancer therapies on ADRD risk; and (3) compare ADRD risks in cancer survivors versus Medicare beneficiaries without cancer aged 65 and older. Data were obtained for 17 SEER areas from SEER-Medicare linked datasets for cancer cohorts and a 5 percent non-cancer Medicare cohort. These SEER areas account for about 28 percent of the U.S. population. The study populations included 246,686 women with breast cancer, 210,809 men and women with colorectal cancer (CRC), 351,571 men with prostate cancer, and 545,449 men and women without cancer. All subjects were free of ADRD at baseline to ensure populations were at risk for the study outcomes.
Covariates included sociodemographic factors (age, race/ethnicity, marital status), tumor factors (stage, size, grade, hormone receptor status, receipt of chemotherapy and radiation therapy), comorbidity, year of diagnosis, and SEER areas.
Colorectal Cancer Findings
The crude 26-year cumulative incidence of total ADRD in people with CRC was higher in those with CVD, stroke, hypertension, or diabetes versus without. After adjusting for sociodemographic and tumor factors, the risk of developing ADRD was significantly higher in patients with CVD, stroke, hypertension, or diabetes versus persons without these conditions.
Adjusted hazard ratios (HRs) of dementias by age, sex, race/ethnicity, and marital status in patients with CRC indicated that the risk of dementia, regardless of type, increased with age, as many other studies have shown. Dementia risk for women with CRC was slightly higher compared with men. Non-Hispanic blacks were at highest risk, followed (in descending order of risk) by Hispanics, Non-Hispanic Asians/Pacific Islanders, and Others. Unmarried women and men were at slightly higher risk for dementia. In older patients with CRC, a significant dose-response relationship was observed between an increasing number of these vascular diseases and the risk of all types of dementia. When treating age as a continuous variable, AD and ADRD risk increased nearly 8 percent for each 1-year age increase.
Investigators conducted analyses to control for potential reverse causation and immortal time biases. Adjusted HRs for ADRD were compared for the following three groups: all AD cases, excluding AD cases that occurred less than 1 year after baseline, and excluding AD cases that occurred fewer than 5 years after baseline.
Breast Cancer Findings
Among women with breast cancer, CVD, stroke, hypertension, and diabetes were associated with a significantly higher risk of developing any ADRD combined. The risk of ADRD was higher in black women and lower among Asian/Pacific-Islanders than among white women.
The crude 26-year cumulative incidence of ADRD among women with breast cancer was higher in those with CVD, stroke, hypertension, or diabetes versus without. After adjusting for confounders, women with breast cancer who also had CVD, stroke, hypertension, or diabetes had significantly higher risks of developing ADRD. Women aged 80–84 years and 85 years and older had five- and seven-fold higher risks for AD than those aged 65–69. Compared with white women, black women had a significantly higher AD risk, and Asians/Pacific-Islanders had a significantly lower AD risk.
Prostate Cancer Findings
ADRD risk for men with prostate cancer was consistent with that found for colorectal and breast cancers—that is, higher among those with vascular disease.
Other Variables: Tumor Characteristics, Treatment Modalities, Comorbidity Scores
Although no associations were observed between dementia and other cancer characteristics, higher tumor stage was associated with increased risk of dementia in all three cohorts (CRC, breast cancer, and prostate cancer).
Findings for chemotherapy and radiotherapy were mixed. In CRC and breast cancer, chemotherapy was associated with a significantly lower risk of dementia. In contrast, chemotherapy was associated with a slightly increased risk of dementia in prostate cancer. Radiation was not significantly associated with risk of any dementia. No significant associations were observed between androgen deprivation therapy (ADT) and risk of dementia in prostate cancer.
Dementia risk also increased with increased comorbidity scores.
Preliminary Findings: Cumulative Incidence among Medicare Beneficiaries without Cancer
Dr. Du presented results of a preliminary analysis of data for Medicare beneficiaries without cancer, noting limitations due to potential selection bias and confounders.
Similar associations between vascular disease and risk of dementia were observed; those with CVD, stroke, hypertension, or diabetes had significantly higher risk for dementia (AD and ADRD). Risk of dementia increased with advanced age; women were significantly more likely to develop dementia than men. No significant differences in risk were observed between black and white populations; however, lower dementia risk was observed among Asians/Pacific Islanders.
To control for potential reverse causation bias and immortal time bias, investigators conducted analyses, one of which excluded cases that occurred fewer than 5 years after baseline. Patterns of dementia risk were consistent with those for cases that were excluded.
An analysis of cumulative incidence of AD in three cancer survivor cohorts (breast, CRC, prostate) and non-cancer subjects aged 65 years or older showed higher incidence among women with breast cancer, higher incidence among those with vascular disease, and similar incidence between non-cancer subjects and CRC and prostate cancer survivors. Similar patterns were observed for cumulative incidence of ADRD.
In terms of adjusted HRs of ADRD in cancer patients versus non-cancer subjects aged 65 years or older, cancer patients had significantly higher HRs for dementia risk.
Dr. Du noted limitations of these preliminary results due to potential selection bias and confounders.
Conclusions
- In both cancer and non-cancer cohorts, vascular diseases were significantly associated with higher risk of AD/ADRD.
- Higher tumor stage was associated with higher risk of ADRD.
- Radiotherapy was associated with a lower risk of ADRD.
- Chemotherapy and ADT findings were mixed, with low and high risks of ADRD across different cancers.
- Higher risks of AD and ADRD were observed in three large cancer cohorts versus a non-cancer cohort. No inverse cancer-AD association was observed based on preliminary results. This may be due to different background risks in cancer cohorts (e.g., cancer-related depression and anxiety, therapy) and other confounding factors (e.g., smoking, education, physical therapy).
Future Directions
- Explore further the detailed multiple interactions between cancer therapies (dose, duration, and adherence) and vascular diseases (and their treatments).
- Examine comorbidities and medication control (e.g., antihypertensive drugs) associated with ADRD in both cancer and non-cancer cohorts.
- Conduct studies in prospective cohorts of cancer and non-cancer individuals with similar backgrounds to confirm a potential cancer-AD inverse association.
The good and bad biases of the cancer-dementia association: An epidemiologic assessment of the literature
Monica Ospina-Romero, M.D., MAS, University of Wisconsin
Dr. Ospina-Romero provided an overview of the biases to consider when conducting or evaluating analyses that support or oppose the cancer-AD association.
Two meta-analyses published in 2014 and 2015 reported that cancer survivors have a 40 percent lower incidence of AD-type dementia than individuals with no history of cancer. One hypothesis proposed for this inverse association is that cancer influences AD. Given that patients receiving cancer treatment frequently report cognitive symptoms, cancer therapy may be a mediator for this association. Uncovering the type of bias that gives rise to the inverse association could illuminate causes of AD.
Good biases are those in which unknown factors “U” are in the nonmediating pathway between cancer and AD. These “good biases” could reveal etiological mechanisms of AD. The bad biases are consequences of study limitations or methodological errors and unlikely to improve understanding of AD etiology.
Good Bias
Good biases include confounding bias and selective survival bias. A confounding bias is one in which an unknown, common cause “U” increases cancer risk and simultaneously prevents AD; “U” is potentially an inverse regulator between carcinogenesis and neurodegeneration. In selective survival bias, the “U” factor promotes cancer survival and prevents AD. The cohort is enriched over time by cancer survivors who have this unknown factor “U.”
Bad Bias
Bad biases include confounding by known but unadjusted factors (e.g., an investigator failed to adjust for known common risk factors) ; inappropriate control for factors influenced by cancer (e.g., adjustment for cancer treatment or comorbidities after cancer diagnosis); diagnostic bias of AD status (e.g., a cancer diagnosis influences the AD diagnosis); and competing risk bias (e.g., an investigator estimated cumulative incidence proportions in unadjusted analyses).
Systematic Review and Meta-Analysis
Dr. Ospina-Romero described a systematic review and meta-analysis of 22 cohort and case-control studies designed to examine two questions:
- Does an association exist between cancer and subsequent AD risk?
- What biases (good or bad) could explain the cancer-AD association?
The review included articles published in any language in PubMed, Embase, and PsycINFO databases up to September 2, 2020, reporting on a longitudinal cohort or case-control study design that met the following criteria: did not require mortality data to ascertain the outcome; collected history of cancer at baseline (prevalent cancer) or incident cancer over the follow-up period; included a comparison group of individuals with no cancer history at baseline or no cancer history at each follow-up assessment; and reported incident AD or dementia outcomes. Studies were included in the meta-analysis that reported a measure of association (risk ratio, hazard ratio, incidence rate ratio, odds ratio [OR]) and 95% confidence interval (CI). For the statistical analysis, fixed-effects and random-effects meta-analytic models were used with a random effects meta-regression to assess the influence of each type of study bias on the pooled estimate.
Results of the random-effects model for pooled cancer-AD risk estimates included the following:
- Pooled OR from three case-control studies: 0.75
- Pooled HR from nine cohort studies
- All cancer types: 0.81
- Prostate cancer: 0.99 (no association observed)
- Breast cancer: 0.93
- Colorectal cancer: 0.88
- Non-melanoma skin cancer: 0.89.
Methodological study biases included bias from handling of potential confounders (i.e., missing adjustment for age, sex, or educational level [12 out of 22 studies], adjusted for factors influenced by cancer); diagnostic bias (i.e., cognitively impaired individuals not excluded at baseline, cancer status might influence AD diagnosis [15 out of 22 studies], competing risks, estimated cumulative risk); and survival and related biases (i.e., prevalent cancers not separated from incident cancers, cancer type that raises subsequent mortality risk, high percentage of missing data, restrictive inclusion and exclusion criteria).
Meta regression estimates were calculated for studies that did not have a particular type of bias, where a zero HR represents a bias with a null association; a bias with a negative HR supports an inverse association; and a bias with a positive HR supports a positive association. All estimates for studies without that specific bias supported an inverse association. Meta regression estimates of differences in HR were calculated for studies with each type of bias. Missing adjustment for age, sex, and educational level moved the estimate toward the null value, as did cancer status that might influence AD diagnosis. Prevalent cancers not separated from incident cancers moved away from the null, making the inverse larger in magnitude.
Conclusions
Meta-analysis results showed an 11 percent lower risk of AD in cancer survivors compared with those with no cancer history. This is a smaller percentage than the finding from the 2014 study (40%).
Bias-adjusted meta regressions suggested that competing risk bias and bias from handling potential confounders did not explain the inverse association. Selective survival bias and confounding bias by an unknown factor “U” may explain the observed inverse cancer-AD association.
Dr. Ospina-Romero noted a limitation of the meta-analysis: although the 22 studies were subject to multiple methodological biases, meta regression models only adjust for one bias at a time.
Is there sufficient evidence to rule out diagnostic bias? A recent simulation study showed that delayed dementia diagnosis could explain the inverse association:
- Non-differential diagnosis delay introduced substantial bias in cancer types with a high mortality rate (lung cancer was used in the model).
- For differential diagnosis delay for all cancer types combined, scenarios with a dementia underdiagnosis rate of 20 percent or more (an additional 4.5-month delay) could explain the inverse association.
Dr. Ospina-Romera suggested that brain autopsy studies might be useful for addressing questions about diagnostic bias. She summarized findings from two brain autopsy studies:
- Results from the Religious Orders Study and Rush Memory and Aging Project (ROSMAP) showed lower burden of tau neurofibrillary tangles in those with a history of cancer. This counters the diagnosis delay.
- A community-based cohort in Kentucky with neuropathological confirmation found that cancer diagnosis was associated with a lower burden of AD pathology and less cognitive impairment.
The molecular genetics of cancer and Alzheimer's disease
Kelly Nudelman, Ph.D., Indiana University
Epidemiological studies have shown a significant inverse association between cancer and AD. To date, most studies of the molecular mechanisms underlying the inverse association have compared data from neurodegenerative studies with data from cancer studies. For example, a 2014 study compared transcriptomic data from AD, Parkinson’s, and schizophrenia studies with data from studies of three cancer types (CRC, lung, prostate). Notably, most of the genes upregulated in neurodegenerative disease and downregulated in cancer are linked to organismal systems or environmental information processing categories. Genes that are downregulated in neurodegenerative disease and upregulated in cancer fall into cellular and genetic information processing categories.
Hallmarks of Cancer and AD
Many of the molecular mechanisms, risk factors, and characteristic pathways or hallmarks of cancer that contribute to oncogenesis or progression are also important in AD—common underlying risk factors, inversely regulated mechanisms, and functions with complex mechanisms. Pathways that operate in the same direction for both diseases include genomic instability and increased inflammation. Pathways that operate in the inverse direction include cell death (i.e., cancer resists, AD increases), cell proliferation (i.e., cancer sustains, AD involves abnormal signaling), growth suppression (cancer evades, AD upregulates), and immunity (cancer avoids destruction, AD increases activation). Complex pleiotropic mechanisms include cell adhesion or contact inhibition and angiogenesis.
Two-Hit Model
Recently, it has been proposed that late-onset AD (LOAD) follows the two-hit model that is widely accepted in cancer. The two-hit model involves an early insult or germline mutation (e.g., childhood stress, chemical exposure) followed by a second hit that initiates the disease process. Common risk factors (i.e., demographic, behavioral, exposure) contribute to this second hit. Once the disease process starts, hallmark pathways come into play.
Cancer History in AD
Dr. Nudelman described a study on cancer history in AD that used data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). She outlined some limitations to investigation of cancer history in AD, particularly using older cohorts of people with good data on AD and other neurodegenerative diseases. Collecting cancer history information generally has not been a priority for many of these studies, and fewer longitudinal studies have genetic data. Despite these limitations, the study provides a unique perspective on and positions investigators to advance research in this area with larger studies.
ADNI collects self-reported information on cancer history. Out of 1,609 subjects, 421 reported a single cancer incidence; 82 reported multiple cancer incidences, the majority of which were NMSC and another type of cancer. The incidence was similar to national cancer incidence rates. The inclusion of individuals with NMSC in the dataset offered a unique opportunity to avoid survival bias because NMSC typically is not lethal and does not always require treatment.
In the ADNI cohort, about 26 percent in the AD diagnostic group had a history of cancer. People in the AD diagnostic group were significantly less likely to have a history of cancer at baseline. A subgroup analysis showed that people in the AD diagnostic group were significantly less likely to have a history of NMSC. Cancer history was associated with delayed AD age of onset. An additive effect was observed in that people who reported more than one cancer incidence had a later age of onset compared with people with one prior cancer incidence or no prior incidence reported.
A voxel-wise analysis showed that people with a cancer history had lower gray matter (GM) density in frontal regions compared with those with no cancer history. These regions of lower gray matter density are consistent with what is seen in studies of chemotherapy-related cognitive decline. While there may be an inverse association between cancer and AD, it does not appear to be influenced or driven by this difference in gray matter density that is seen in cancer survivors.
Dr. Nudelman described further exploration of this gray matter finding using data from the Religious Orders Study/Memory Aging Project (ROSMAP) study. ROSMAP has longitudinal data in an older cohort who were healthy at baseline, some of whom later developed neurogenerative diseases; available data include cancer history and neuropathology. ROSMAP participants with cancer history had lower odds of AD and fewer neurofibrillary tangles compared with those with no cancer history, although they had similar levels of amyloid plaques.
The study included data for a subset of 1,448 ROSMAP subjects with age of death and cancer history as well as microRNA (miRNA) and gene expression data. Because numbers were somewhat small, investigators calculated weighted co-expression gene networks, which looks for patterns of co-expressed genes across individuals and collates them into gene networks or eigengene modules. Investigators looked at differences between networks by AD clinical diagnosis and cancer history.
- One eigengene module correlated positively with cancer history and AD diagnosis.
- Four eigengene modules correlated negatively with cancer history and AD diagnosis.
- No modules correlated inversely for either category in this dataset.
- One miRNA module (M1) correlated positively with AD clinical diagnosis and negatively with cancer history.
The top modules significantly associated with these traits (miRNA_M1 and Gene_M1) were run through Gene Ontology to see which biological pathways were significantly enriched in these networks. The most significantly enriched gene pathway was the metal-binding pathway of metallothionein. The hub miRNA hsa-let-7i-3p from miRNA_M1 targets the hub gene from Gene_M1, Metallothionein 1A. Network analysis suggests that risk for cancer and AD may be associated with metallothionein binding of metals, a pathway previously linked separately to cancer and AD.
Conclusions
Gene and miRNA regulatory networks play complex roles in cancer and AD.
Study of cancer history in patients with AD can further elucidate mechanisms of disease that may be unclear in studies comparing across different disease populations.
Further investigation is needed in larger, more diverse populations that have data on cancer history and progression and neurodegenerative disease.
Investigating the genetic relationship between Alzheimer's disease and cancer using GWAS summary statistics
Liming Liang, Ph.D., Harvard University
Dr. Liang introduced an approach that uses genome wide association studies (GWAS) as a tool for understanding the link between cancer and AD. Genetic mutations are not subject to reverse causation. Finding the genetic components shared by AD and cancer can help establish the causal association between the two diseases.
The natural direction of genetic effect provides a unique opportunity to integrate multiple intermediate omics into a disease’s causal diagram. Very large-scale GWAS for diseases and functional data can provide more precise estimates of genetic effect (e.g., facilitate causal inference-based Mendelian randomization [MR] and risk prediction based on polygenic risk modeling). Summary statistics-based analyses overcome the hurdle of data sharing.
Dr. Liang used asthma to elucidate a pipeline for investigation of disease-disease cross-trait genetics. Genetic links were employed to eliminate environmental confounding and connect different disease traits.
Asthma Subtypes and Risk Factors
Asthma is a highly heterogeneous disease. Investigators examined shared and distinct genetics between asthma subtypes (e.g., allergic versus non-allergic) and between asthma and asthma risk factors (e.g., allergic diseases, obesity, mental health disorders). Genetic information enabled one to distinguish whether the diseases or traits share common genetic factors or undergo a causal pathway, with one being the upstream effect of the other. For example, an MR analysis identified a gene shared by asthma and anxiety as a horizontal pleiotropic effect rather than a causal effect. On the other hand, there is strong evidence to suggest that obesity has a causal effect on asthma (a vertical pleiotropic effect).
This approach is readily extensible to other diseases, as demonstrated by Dr. Liang’s studies of AD and cancer, AD and metabolic traits, COVID-19 and non-allergic asthma, and COVID-19 and obesity. Dr. Liang’s team published a summary of their approach, including organization of GWAS data, estimation of overall genetic relationships using genetic correlations within GWAS data, and narrowing to individual variants and their functions. In addition, they compared advantages and disadvantages of various genome wide cross-trait analysis methods. For example, many MR methods are based on different assumptions to avoid potential biases in causal effect inferences; it is critical to understand these assumptions and apply them appropriately to genetic datasets.
Genetic Correlation Between AD and Cancer
Dr. Liang outlined steps for exploration of the genetic correlations between AD and cancer.
- Use genome wide genetic correlation to estimate the association between AD and cancers.
- Estimate genetic overlap in specific functional categories (e.g., exome regions, epigenetic markers).
- Identify the cross-disease associations at the level of individual genetic variants and extend functional consequences.
A meta-analysis of GWAS data from the International Genomics of Alzheimer’s Project (iGAP) and the NCI Genetic Associations and Mechanisms in Oncology (GAME-ON) Initiative included 54,162 samples after quality control. An examination of genetic correlation between AD and each cancer type identified breast and lung cancers as having positive correlations with AD. Functional annotation-specific genetic correlations between AD and each cancer type were calculated. From AD-related genetic variants, breast cancer- and lung cancer-related single nucleotide polymorphisms (SNPs) were scanned. A cross-phenotype meta-analysis identified AD SNPs shared with more than one cancer type. Top SNPs seemed to have stronger associations with one type of cancer than with the other cancer.
Dr. Liang outlined scenarios from studies that applied other methods for examining associations between cancer and AD; for example, correlation due to tissue-specific genetic effects versus correlation due to gene expression.
Genetic Correlation Between AD and Metabolic Traits
A study that evaluated the genetic correlation between AD and metabolic traits used GWAS data to estimate the genetic relationship. Fasting glucose (FG) and fasting insulin (FINS) had substantial magnitude of genetic correlation with AD in opposite directions. High-density lipoprotein (HDL) also had a significant genetic correlation with AD. No substantial correlation was observed between AD and obesity traits (e.g., body mass index [BMI], waist-hip ratio, type 2 diabetes).
Genetic Correlation Between AD and Stroke
Cancer is known to have an association with certain types of stroke. An analysis of AD (LOAD and AD-by-proxy) and stroke GWAS studies found an interesting negative genetic correlation overall for AD and different subtypes of stroke; however, when looking at stroke as the causal exposure and AD as the outcome, cardioembolic stroke (CES) had a positive effect on AD.
A cross-trait meta-analysis identified 13 SNPs at the CELF1 locus with both AD and any stroke and the small vessel stroke subtype. The SNPs are associated with multiple genes; for example, C1QTNF4 is highly expressed in body fat and brain regions and has been positively related to brain fat distribution, BMI, risk-taking tendency, education, and CVD. There is a high genetic effect for this SNP related to both AD and stroke in these tissues.
Current Understanding and Future Steps
Current findings support:
- Significant genetic correlation for AD and certain cancer types.
- Significant genetic correlation for glucose metabolic traits.
- Complicated AD and stroke genetic correlation and causal effects, suggesting heterogenetic pathways that have different effect directions to these diseases.
Dr. Liang noted that statistical power for identifying polygenic signals from AD is limited.
Functional data and genetic effect could be useful to identify specific pathways shared between AD and cancers.
The analysis is being extended to metabolome and epigenome data in available cohorts, such as ADNI and Rush.
Session Chair: Todd Horowitz, Ph.D., Behavioral Research Program (BRP), Division of Cancer Control and Population Sciences (DCCPS), National Cancer Institute (NCI)
What can cancer-related cognitive decline tell us about ADRD (and vice versa)—Lessons from the Thinking and Living with Cancer study
Jeanne Mandelblatt, M.D., MPH, Georgetown University
Dr. Mandelblatt presented on the relationships between ADRD and CRCD. In their practice, oncologists hear daily reports from patients of difficulty thinking, multitasking, processing, and remembering, yet these experiences remain difficult to manage and quantify.
CRCD and ADRD share a risk factor—the APOE genotype—and affect similar areas of the brain and similar cognitive domains. They are also quite different: while ADRD is dramatic and disabling, CRCD may be more subtle and does not progress the way ADRD does.
The Thinking and Living with Cancer (TLC) Study
The NIA- and NCI-funded Thinking and Living with Cancer study aims to understand cognitive aging and the phenomenon of CRCD. The TLC study is a multisite prospective US study that uses data from older breast cancer survivors and demographically matched controls without cancer to evaluate cognition after breast cancer and treatment.
Study Design and Demographics
Newly diagnosed breast cancer patients (714) were enrolled after surgical interventions but before they began chemotherapy, hormonal therapy or HER-2 directed therapy. Neither of the patients or their matched controls (581) had any known cognitive disease. All volunteers were followed for up to 5 years. Assessments included neurocognitive test batteries, surveys, APOE testing, GWAS, inflammatory markers, magnetic resonance imaging (MRI) and actigraphy to capture sleep and physical activity. Participants in the study were, on average, healthier and better educated than other women their age.
Findings
At baseline—after surgery but before chemotherapy or hormonal therapy—breast cancer survivors scored lower on attention, processing speed, and executive function (APE) than their cancer-free counterparts. Survivors also reported higher levels of depression, fatigue, anxiety, and sleep disturbances at the time of enrollment.
Over a 2-year period, women who received chemotherapy with or without hormonal therapy had lower cognitive scores. When broken down by genotype, this trend persisted only among women who were APOE4 positive. APOE4-negative survivors who underwent chemotherapy did not demonstrate the same decline.
A preclinical extension of this work in human homozygous APOE knock-in mice yielded the same results: APOE4 appeared to create a predisposition toward chemotherapy-triggered cognitive decline. Gene expression analysis of the mice found increased expression of aging- and senescence-related genes after treatment in those who were APOE4-positive. The researchers hypothesized that DNA damage in inflammation and senescence pathways may lead to cognitive decline.
To test this hypothesis in a clinical setting, TLC researchers examined the relationship between C-reactive protein (CRP, a clinical marker of inflammation) and cognition. CRP predicts risk of mortality and cardiovascular disease and has been shown to be related to CRCD. The TLC researchers compared participants’ self-reported Functional Assessment of Cancer Therapy – Cognitive Function (FACT-Cog) scores with their CRP levels. They found that higher CRP levels predicted subsequent lower FACT-Cog scores—but only in survivors, not in healthy controls. These findings were significant even after controlling for age, race, study site, and comorbidities.
Other TLC work examined the relationship between cancer, immune markers, and cognition and found a significant association for interleukin (IL)-6 and a borderline significant relationship for IL-10. IL-6 is one of the most responsive and polymorphic immune markers, acting along a number of pathways.
Conclusions
These findings suggest that inflammation pathways may be implicated in both ADRD and CRCD and that aging processes may be a common factor.
Dr. Mandelblatt invited attendees to contact the TLC study team to use the dataset. Its cognitive scores have been harmonized with those of ROS and ADNI.
From neuroimaging to multi-omics: A systems approach to cognitive changes in cancer and Alzheimer’s
Andrew J Saykin, Psy.D., Indiana University
CRCD has been the subject of scientific interest for many years. The earliest studies focused on cytotoxicity in chemotherapy. Recent longitudinal prospective studies like the TLC study have advanced the field and provided more nuanced data.
Much is known about Alzheimer's disease, especially amyloid plaques and tangles, and biomarkers have been identified. The same is not true for CRCD. However, the same pathways—including neurodegeneration, synaptic loss, and vascular and immune changes, among many others—are the focus of research into both conditions. The same fundamental biological landscape is implicated in both; however, the effects may present in parallel or in an inverse relationship.
Multi-omic data are one way to advance the field. ADNI and other studies have been collecting rich longitudinal multi-omic data, which enables a systems biology perspective. This will allow researchers to understand disease mechanisms and identify novel therapeutic targets.
Imaging and Multi-Omics in TLC
The TLC study now is adding imaging and omics studies to its cohort of older women in the age range for elevated risk for AD and other neurodegenerative disorders. Indiana University is both a TLC site and an Alzheimer's Disease Research Center (ADRC), which has allowed researchers there to run the same multimodal neuroimaging protocol in both ADRC and TLC participants.
Initial analysis of images from breast cancer survivors over 12 months found regional reductions in frontal, temporal, and parietal gray matter on voxel-based morphometry; reduced FreeSurfer cortical thickness in prefrontal, parietal, and insular regions; and increased working memory-related functional MRI (fMRI) activation in the frontal, cingulate, and visual association cortices. Controls showed only reductions in fusiform gyrus on VBM and FreeSurfer temporal and parietal cortex thickness. Women with breast cancer showed higher estimated brain age and lower regional gray matter volume than controls at both time points. The cancer group showed a trend toward lower performance in APE at follow-up. There were no significant associations between brain imaging metrics and cognition or days on hormonal therapy. Larger samples and longer follow-ups are needed to validate these findings.
The Indiana team also is exploring brain networks and the connectome. A recent review article linked amyloid- and tau-misfolded proteins to connectome changes and the genes implicated in driving those changes.
Genetic Data in TLC
Dr. Nudelman and her colleagues used TLC data to identify genetic variants associated with CRCD, examining the interactions of SNPs with cancer diagnosis. The analysis revealed two genome wide significant signals in the APE domain: rs76859653 (chromosome 1) in the hemicentin 1 (HMCN1) gene and rs78786199 (chromosome 2) in an intergenic region. The HMCN1 gene codes for extracellular protein in the immunoglobulin superfamily. The closest gene to the relevant intergenic region is the FOXN2 gene, which has been studied in cancer and has a role in several important cancer-related biological processes as well as pathways related to immune function, inflammation, proliferation, and cell cycling in AD. After 1 year, survivors with these SNPs had lower scores than noncarriers and controls. This analysis should be repeated over a longer follow-up period and in a larger cohort.
A separate analysis revealed a signal related to learning and memory in the POC5 centriolar protein gene. This line of inquiry will be explored further. Another direction for genetic work has been polygenic risk scores in AD. Other scores that may have a role in the future include transcriptional risk scores and epigenetic scores. Metabolomics, lipidomics, and proteomics may also have scores that could be useful as biomarkers.
Future Directions and Conclusions
In the coming months, the TLC group will examine blood-based biomarkers for AD in the TLC cohort. These biomarkers could be useful for early detection and diagnosis, prognosis, longitudinal monitoring, and theranostics to provide precision interventions.
CRCD and AD are two complex disease areas associated with cognitive impairment in aging. Precision medicine is an important goal for therapeutics and prevention; systems biology and rapidly emerging biomarkers can contribute to achieving this aim. Genetic analysis and multimodal imaging in TLC already have uncovered important pathways and changes. Parallel studies of ADRD and CRCD show promise for elucidating common mechanisms or counterpoints between the conditions that could lead to therapeutic and preventive opportunities.
Management of cancer-related cognitive decline: State of the science on intervention strategies
Allison Magnuson, D.O., MS, University of Rochester
Dr. Magnuson presented on the management of CRCD and the state of the science on intervention strategies for older adults. There are more than 10 million older cancer survivors in the U.S., and as many as 35 percent of them report long-term cognitive symptoms. These symptoms can have profound practical consequences, including increased difficulty with activities of daily living and a compromised ability to live independently.
Research has found that older cancer survivors are interested in receiving treatments for CRCD, but there is currently no standard approach and no standard of care. Interventions have been explored, but much of this work has focused on cancer survivors, not older adults—and, in fact, older adults were severely underrepresented in many of these studies.
Behavioral Interventions
Multiple studies have focused on behavioral intervention strategies, which include cognitive rehabilitation-based approaches and physical activity- and exercise-based interventions. Cognitive rehabilitation-based approaches may include strategy or educational programs, computer-based cognitive training approaches, or a combination of the two.
A systematic review of 19 cognitive rehabilitation-based programs for CRCD found that all studies observed improvements in at least one cognitive measure. In most studies, patients reported fewer cognitive concerns or improved cognitive abilities after rehabilitation. There was also some evidence for intervention effect improving health-related quality of life.
There are some challenges in generalizing these data. The studies enrolled mostly younger patients, with a mean age in the late forties to mid-fifties, and many of the studies were focused on breast cancer patients, particularly in the survivorship setting, years after completion of treatment. The published literature also mainly includes smaller studies, although larger, more definitive studies are underway. The studies in the systematic review did differ significantly; some of the interventions were group-based while others were individualized, and outcome measures varied across studies. Two of the studies included in the review focused on older adult populations, but these were secondary analyses of cancer survivors enrolled in larger studies of cognitive interventions. The studies themselves were not cancer-specific, and the number of older adult cancer survivors was small; however, improvements in self-perceived cognition were noted.
An increasing amount of research is focusing on cognitive interventions for addressing CRCD symptoms in older adults. Completed work includes the SeniorWISE (Wisdom Is Simply Exploration) cancer survivor subset and a subset analysis of the Cognitive Behavior Model of Everyday Memory.
Ongoing studies include:
- Promoting Physical Activity to Improve Cognitive Function in Older Adults Undergoing HSCT (hematopoietic stem cell transplant)
- Acupuncture for Cognitive Health in Older Survivors of Prostate Cancer
- Light Physical Activity for Brain Health in Older Adult Breast Cancer Survivors
- Mitigating CRCD in Older Adults with Breast Cancer.
Dr. Magnuson’s group now is evaluating a telehealth-based behavioral intervention called MAAT-G for older adults with cancer. The researchers adapted an existing intervention, the Memory and Attention Adaptation Training (MAAT), for older adults. The adaptation process included using the Contextual, Cohort-based, Maturity-Specific Challenge framework for cognitive behavioral therapy-based interventions for older adults, collaborating with older cancer survivor patient advocates to ensure the study was accessible, and conducting usability testing. The telehealth delivery model was selected based on prior input from patients, who expressed great interest in cognitive interventions but had barriers to attending additional in-person appointments. The flexibility of virtual, individualized sessions was appealing to them.
MAAT-G consists of 10 individualized weekly sessions with a clinical psychologist or trained research nurse delivered through a videoconferencing platform. The intervention occurs early in the treatment course, concurrent with active cancer therapy. The researchers now are conducting a randomized pilot study comparing MAAT-G to an active control condition. To date, they have enrolled 55 patients, and retention rates have been good. The study experience so far has provided useful information on how to best support older populations, including those with barriers to Internet or technology access, in telehealth-based intervention studies.
A recent systematic review of physical activity- and exercise-based interventions included 29 clinical trials with more than 3,000 participants in total. Most of these studies evaluated cognitive function only as a secondary outcome. Of the 29 studies, 12 observed improvement in perceived cognition, and 3 observed improvement in performance-based cognitive assessments. However, as with the cognitive rehabilitation-based studies, most patients included in the reviewed studies were younger, and the studies occurred in the survivorship setting, not pre- or mid-treatment.
Other Interventions
There are no established pharmacological interventions for CRCD. Research into this area is limited, and results have not provided any confirmatory results. As with other interventions, pharmacological studies addressing CRCD have not been conducted specifically in older populations.
Some emerging work is focusing on intervention studies in patients with dual diagnoses of cancer and mild cognitive impairment (MCI):
A Patient-Caregiver Behavioral Intervention for Older Adults with Cancer and MCI
Feasibility of MAAT-G in Older Cancer Survivors with MCI.
Other studies are evaluating communication-based interventions about cognition in the context of cancer. There are limited data and no standard approach for how clinicians should approach cancer management for these patients. For older adults with both cognitive impairment and cancer, decision making is more complex. It becomes even more important to incorporate a patient’s care partner into the decision-making process.
The Improving Communication in Older Cancer Patients and Their Caregivers (COACH) trial, which enrolled 541 older adults with cancer, demonstrated improved communication about aging-related concerns and patient and care partner satisfaction with communication. Dr. Magnuson’s group recently received funding to adapt this intervention specifically for patients with dual diagnoses of cancer and dementia.
Conclusions
Although there is currently no standard of care for the management of CRCD symptoms, emerging data suggest that behavioral interventions may be helpful. Further definitive studies are ongoing, and more data are needed in older adult populations. An important area of study will be populations of older adults with dual diagnoses of cancer and AD.
Discussion and Wrap-up of Day 1
Damali Martin, Ph.D., MPH, NIA
Dr. Martin commended the speakers and attendees for the day’s presentations and discussions. She acknowledged that Dr. Richard Hodes had joined the meeting and thanked him for taking the time to listen in on the workshop. She reminded attendees that the following day’s agenda would begin at 9 a.m. ET.
The session ended at 2:50 p.m. ET.
Day 2: Wednesday, October 19 | Goals for the day’s activities
Damali Martin, Ph.D., MPH, NIA
Dr. Martin opened the meeting at 9:03 a.m. ET, welcomed attendees, and presented a summary of key ideas from the previous day.
Session 1: Why the joint interest in cancer and AD?
Session 1 speakers Drs. Kobayashi and Shardell described best practices for studying AD and cancer and presented methods for overcoming biases. Recommendations included the following: adjust for common causes of cancer and the aging outcome under investigation based on prior evidence and theories about associations; avoid adjusting for factors downstream of cancer incidence, such as treatment and subsequent health conditions or behavioral changes; consider how diagnostic practices may affect findings when using medical record or claims data; and account for competing risk bias and selective survival bias. Attendees were advised to state the research questions explicitly, know the timeline, and understand the assumptions underlying analytical methods.
Dr. Chiosis described AD and cancer as systems-level maladaptations to stressors that have complex etiologies with external and internal stressors that negatively impact specific cells, tissues, and organs. A paradigm shift is needed to develop therapies that consider the protein-protein interactions that are critical in the stressor-to-phenotype complex and could serve as therapeutic targets. She described her work with epichaperome structures and the possibility of developing drugs that disrupt these structures, playing a role in the treatment of a broad spectrum of diseases.
Session 2: Genes, Mechanisms, and Epidemiological Evidence
Du, Ospina-Romero, Nudelman, Liang
In Session 2, emerging data from cancer and non-cancer cohorts indicated that vascular diseases were significantly associated with higher risk of AD/ADRD. Higher tumor stage was associated with increased risk of ADRD. Radiotherapy was associated with a lower risk of ADRD.
Presenters recommended further exploration of multiple interactions between cancer therapy dose, duration, and adherence and vascular diseases and their treatments; examination of comorbidities and medication control associated with ADRD in cancer and non-cancer cohorts; and conduct of studies in prospective cohorts to confirm the potential inverse cancer-AD association.
A meta-analysis showed an 11 percent lower risk of AD in cancer survivors than in those with no cancer history. Although competing risk bias and bias from handling potential confounders did not explain the inverse association between cancer and AD, selective survival bias and confounding bias by unknown factor “U” may do so.
Gene and miRNA regulatory networks play complex roles in cancer and Alzheimer’s disease. Study of cancer history in patients with AD can further elucidate mechanisms of disease that may be unclear in studies comparing across different disease populations. Further investigation is needed in larger, more diverse populations that have data on cancer history and progression and neurodegenerative disease.
Research findings support a significant correlation between AD and certain cancer types and between cancer and glucose metabolic traits. Functional data and genetic effects could be useful for identification of pathways shared by AD and cancer.
Session 3: Cancer Chemotherapy and Cognitive Dysfunction
<Mandelblatt, Saykin, Magnuson
In Session 3, Dr. Mandelblatt presented research on APOE4, which appears to create a predisposition toward chemotherapy-triggered cognitive decline. A preclinical extension of this work showed that doxorubicin impairs spatial learning in APOE4 mice compared with APOE3 mice. She suggested that transdisciplinary approaches may help to accelerate the discovery in the cancer-AD nexus.
Dr. Saykin described the importance of multi-omics toward precision medicine for therapeutic development, biosystems biology, and biomarkers. He is using TLC and ADRC GWAS and multi-omic data and imaging studies to identify neurogenetic pathways that will enable precision healthcare for AD, ADRD, and cancer-related cognitive disorders.
Dr. Magnuson spoke about emerging data from intervention strategies suggesting that these strategies may be helpful for management of CRCD and noted the need for further definitive studies, particularly in older adult populations. Enhancing communication about cognition in the context of oncology also will be important for the field.
Dr. Martin introduced Dr. Grothaus, chair of the next session.
Session 4: Meds, Mechanisms, and Drug Repurposing
Session Chair: Paul Grothaus, Ph.D., NIA
Reversing Alzheimer’s disease gene network states by approved cancer and inflammatory drugs: From informatics to bench to clinical trial
Mark Albers, M.D., Ph.D., Harvard University
Dr. Albers presented on off-label indications for U.S. Food and Drug Administration (FDA)-approved drugs.
Mechanistic Similarities and Differences in Cancer and AD
Age is the most important risk factor for cancer and AD, and mechanisms of aging might play a role for both. He noted that Dr. Mertens would talk about energy overload and suggested that drugs targeting metabolism may be protective against AD and cancer; for example, Dr. Albers’ team has been looking at metformin.
Inflammation mechanisms are different in cancer versus AD. Often, solid tumors induce an immunosuppressive microenvironment to prevent tumor-infiltrating lymphocytes and other mediators from acting. By contrast, neurodegenerative diseases increase regional neuroinflammation. Understanding how these two diseases modulate the immune system might promote crosstalk between them and lead to new therapeutic approaches for each disease (e.g., JAK inhibitors, RIG-I agonists).
Approaches for Drug Repurposing
Dr. Albers’ colleagues built the Drug Repurposing In AD (DRIAD) tool using gene expression profiles of Accelerating Medicines Partnership® Program for Alzheimer’s Disease (AMP-AD) brains to develop a predictor for early-, mid-, and late-stage AD. The AMP-AD dataset comes from neuropathologic brain banks (ROSMAP, Mayo, Mt. Sinai). The ML framework is publicly available on GitHub. A user assembles genes of interest, and the predictor tool reports whether there is a correlation with disease progression based on these profiles. Results were validated using gene-expression data related to aging and other known factors in AD.
Next, investigators examined what FDA-approved or investigational drugs do to the differentiated human neural cells in culture and the gene expression changes. They screened 80 kinase inhibitors, most of which are anti-cancer drugs. The top hit was ruxolitinib, an FDA-approved JAK inhibitor for hematologic oncology (e.g., lymphomas). Ruxolitinib was a hit in five out of six AMP-AD databases. The list of FDA-approved top hits includes multiple JAK inhibitors (e.g., including tofacitinib and varicitinib), which piqued investigator interest in the JAK family of kinases and their potential role in progression of AD.
Target affinity scoring was applied to all the kinase inhibitors and their response curves on the kinome overall. Relative affinity was calculated for members of the JAK family (JAK1, JAK2, JAK3, TYK2), which correlated with its ability to predict disease progression in the predictor tool. This was viewed as further validation that these drugs and targets might be interesting within AD.
This finding converged nicely with independent research on root causes of neuroinflammation in neurodegenerative disease. Previous research has shown that accumulation of double-stranded RNA (dsRNA) within neuronal cytoplasm is coincident with phosphorylated TDP-43 inclusions in C9orf frontotemporal dementia (FTD)/ALS brains. Because TDP-43 is prevalent in AD and Rush investigators have shown that TDP-43 inclusions lead to faster disease progression, investigators looked for the same phenomenon in AD brains. Indeed, multiplex imaging revealed dsRNA in cytoplasm overlapping with the phosphorylated TDP-43-bearing neurons but not tangle-bearing neurons nor amyloid plaques outside the cell. In the 20 AD cases studied to date, if TDP-43 is present, so is dsRNA; if TDP-43 is not present, neither is dsRNA. At the Harvard Brain Tissue Resource Center , about half of AD cases have TDP4-3 inclusions. This is important because cytoplasmic dsRNA is known to trigger the innate immune system and type 1 interferon signaling, which goes through this family of JAK kinases.
Investigators used a dsRNA-dependent protein kinase, PKR, to detect whether dsRNA is inert or biologically active in the cell. In the presence of dsRNA, PKR binds to it and auto-phosphorylates itself. An antibody against the phosphorylated form of PKR serves as a biosensor. In AD brains where TDP-43 pathology is observed, phosphorylated PKR is seen.
Innate immune signaling observed with aging, part of which might be induced by accumulation of dsRNA. Some transposable elements have palindromic sequences that can lead to production of cytoplasmic dsRNA.
Dr. Albers hypothesized that elevated type 1 interferon signaling arising from dsRNA should be observed in AD cases. At the suggestion of Mass General computational biologists, he looked at the AMP-AD dataset, where he found that interferon-stimulated genes were upregulated in different regions of AD brains, reaching genome wide significance and providing additional evidence of increased interferon signaling in AD brains that might be triggered by dsRNA. These observations were confirmed in the lab by recapitulating the biology in cultured human neural cells. Over a 48-hour period, dsRNA moved into differentiated cells and surrounded the nucleus, thus producing inflammation and eventual cell death. Induction of PKR and JAK family kinases leads to phosphorylation of STAT1, and type 1 interferon signaling is triggered by dsRNA. This is a dose-dependent effect: at lower doses, only inflammation is triggered but not neuronal death; at higher doses, inflammation is triggered, followed by neuronal death.
Using this assay for inflammation and neuronal death, investigators screened for this pathway using FDA-approved JAK inhibitors. Ruxolitinib rescues dsRNA-mediated neurotoxicity in a dose-dependent manner. Baricitinib and tofacitinib (JAK inhibitors identified by DRIAD) also rescue in a dose-dependent manner. Additional work with a mouse model confirmed these findings.
Neurodegenerative Alzheimer’s Disease and ALS (NADALS) Trial
Based on these two lines of evidence, investigators have launched a trial of baricitinib, a structural analog of ruxolitinib that is FDA approved for rheumatoid arthritis, COVID-19, and alopecia that has been on the market since 2014.
NADALS is an open-label, biomarker-driven basket trial in patients with subjective cognitive disorder (SCD), mild cognitive impairment (MCI), AD, and ALS and asymptomatic gene carriers of ALS genes. Cerebrospinal fluid collected via spinal tap will be used to detect elevated inflammation. The trial aims to precisely quantify exposure of baricitinib in sufficiently high concentration within the brain and CNS and movement of inflammatory biomarkers in a positive direction. Secondary outcomes include inflammatory and neural death biomarkers. Deep phenotyping of biofluids will inform refinement of the neuroinflammatory signature and create predictive biomarkers that could be useful in further clinical trials targeting this pathway.
Conclusions
- Mechanistic underpinnings of the potential inverse relationship between cancer and AD may, in part, arise from opposing effects on innate inflammation.
- DRIAD is a publicly available, machine-learning-based tool for evaluating gene lists toward drug repurposing in Alzheimer’s disease.
- FDA-approved JAK inhibitors significantly reversed gene expression patterns observed in the AD brain.
- ytoplasmic dsRNA is a possible cause of neuroinflammation associated with TDP-43 pathologic inclusions found in about half of AD and FTD patients.
- Human neural cell and mouse models demonstrated that cytoplasmic dsRNA is sufficient to evoke propagated neural cell death in a neuroinflammation setting.
- Baricitinib and other FDA-approved JAK inhibitors reverse neuroinflammation and rescue neuronal death.
Neuronal fate loss and metabolic transformation in age-equivalent Alzheimer patient neurons
Jerome Mertens, Ph.D., University of Innsbruck, Austria
Dr. Mertens outlined a multiple hit theory to approach the nongenetic nature of many AD cases. The first hit on the brain arises from an individual’s genotype, environment, and lifestyle, and progressive aging is the second hit. These hits meet at converging pathways and reinforce one another, leading to neuronal death and age-related disease.
Systems for Modeling Human Neuronal Aging and Age-Related Diseases
In the lab, investigators are trying to understand the differences between young, old, and diseased human neurons and distinguish between age-related necessities versus specific triggers of the disease. The ideal model system would be living neurons from young people, old people, and sporadic and familial AD patients for comparison in an unbiased investigation. Induced pluripotent stem cells (iPSCs)-derived neurons from AD patients appear to be ideal for modeling human neuronal aging and age-related diseases; however, the rejuvenation process that occurs early in differentiation creates young cells that may not be suitable for studying age-related phenomena.
Dr. Mertens’ team has developed directly converted induced neurons (iNs), which can be generated via stem cell technology in the laboratory (e.g., from patient-derived fibroblasts). Within a few weeks, these cells turn into functional neurons. Over the past 8–10 years, investigators have shown that these iNs are age-equivalent; they display adult identity and capture the biological age of donors. This was sufficient evidence for Dr. Mertens to use this approach to examine age-dependent diseases.
Dr. Mertens described his work with a cohort of fibroblasts from 34 individuals: 16 AD patients (13 sporadic and 3 familial) and 19 age-matched control donors. He acknowledged that the cohort size is a limitation and is working to increase these numbers. From these fibroblasts, investigators generated neurons in the dish that reflect marker expression of human neurons and become functionally mature. Transcriptome analyses revealed distinct AD iN-specific transcriptome signatures. Signatures from the model have a surprisingly high level of concordance with ROSMAP and single-cell datasets.
Among downregulated gene sets in AD iNs, Dr. Mertens observed mature neuronal gene sets that relate to neuronal activity, synaptic function, and functional maturation. Functional tests revealed that AD iNs have lower synaptic densities, less complex branching morphologies of neurons, and decreased spontaneous network activity.
Among upregulated gene sets in AD iNs, Dr. Mertens saw not only the expected pathways related to cellular stress and damage but also stem cell-signaling and cancer-related signaling pathways. From the cancer perspective, the most prominent pathways include P53 signaling, oncogenic signature, and Myc signaling targets. From the AD perspective, pathways include both positive and negative regulation of cell proliferation, which suggests a battle for identity within the cells: becoming mature, somatic, fully differentiated neurons vs. regressing toward the stem cell fate. These cells have partially lost their fully mature neuronal-differentiated phenotype. Control iNs had the highest similarity to the most mature stages of neuronal development and differentiation while the AD pathological cells seemed to be confused on the scale and showed similarities to immature neurons.
Functional tests indicate that AD iNs have elevated levels of reactive oxygen species (ROS) and lead to increased DNA damage (repair) signals, express non-neuronal glycolysis, and express cell cycle markers. Investigators never succeeded in observing a complete cell cycle.
ATAC-Seq [Assay for Transposase-Accessible Chromatin using sequencing] revealed a globally more open epigenetic landscape in AD iNs than in controls, a hallmark of immature stem and progenitor cells. Chromatin openness was enriched at hypo-maturity and de-differentiation genes.
Conclusions
Based on work with the limited iN cohort, Dr. Mertens concluded that AD iN cells are not only being damaged from the outside by plaques and tangles but also are activating and running a program of fate regression, fate loss, and de-differentiation, which is one of the major markers of cancer cells that have matured and then changed into something else.
- AD neuronal phenotypes are driven by a cellular program, leading to loss of identity.
- The hypo-mature state of neurons explains the loss of neuronal functionality.
- These findings provide a perspective on the “cell cycle re-entry hypothesis” in AD research.
- The data underscore the biological principle that damage triggers cellular de-differentiation or, possibly, re-differentiation.
- The cell biological relationship to cancer may point to strategies for intervention.
Further Exploration of Neuronal Fate Loss
Findings from the iN cohort raised Dr. Mertens’ curiosity about genetic and epigenetic changes and metabolism-controlled cell fate that led to cancer transformation.
The Warburg Effect is a metabolic mechanism that compromises mature cellular identity by switching from oxidative phosphorylation to glycolysis, which can disrupt energy currency in cells and lead directly to epigenetic modifier metabolites that support cancer transformation and proliferation. This activity is particularly interesting in the context of neurons, which, compared with other cell types, are highly specialized and rigid in their reliance on oxidative phosphorylation to support their high-energy demands.
No changes in PKM RNA expression abundance were observed in iN cells. PKM is regulated by alternative splicing. A tetrameric exon 9-containing PKM isoform promotes oxidative phosphorylation and pro-neuronal metabolism. A dimeric exon-10 isoform of PKM2 is a major promoter of Warburg-like glycolytic metabolism in cancer and stem cells.
In the AD iNs model, investigators observed an increased PKM2/1 splicing ratio and PKM2 protein and PKM2 enzyme activity. A study of PKM2 in ROSMAP AD patient brain bulk and single-cell transcriptomes revealed an increased PKM2/1 splicing ratio in the cortex and elevated levels of PKM2 total protein in neuron-rich regions of the AD frontal cortex.
In the AD iNs model system, mass-spec metabolomics revealed increased glucose uptake and flux into lactate. PKM2 is not a dominant active or inactive form of PKM1 in the cytoplasm; it expresses a nuclear localization site and moves through the nuclear pore into the nucleus, where it is known to collaborate with well-known cancer-related transcription factors such as STAT3, HIF1α, and β-Catenin, leading to stress and pro-apoptotic genes, glycolytic genes, de-differentiation genes, and oncogenes. In AD iNs, investigators observed nuclear translocation and kinase activity of p-PKM2 and increased chromatin accessibility at promoter regions of de-differentiation and cell death genes.
Investigators were struck by the presence of the pro-apoptotic genes. No basic apoptosis rate is observed in the AD iN system; however, when triggered with a BCL2 inhibitor, the AD iNs gain competency for apoptosis in response to death signals that mature neurons usually have switched off. To counteract this, investigators made use of advanced knowledge in the cancer field—compound Shikonin. The drug tetramerizes PKM2 and activates metabolic activity of the PKM2 isoform and inhibits nuclear translocation and nuclear functions. Shikonin not only rescues the mature metabolic profile but also reverses the apoptosis competency of the isomer neurons and globally shifts the pathological, AD-specific transcriptome signal back to a more control-like signature.
Conclusions
- Cancer-like PKM2 causes fate loss in AD neurons.
- PKM2 plays pathogenic metabolic loss of function and epigenetic gain of function roles.
- Neuronal de-differentiation leads to cell death competency of mature brain neurons partially driven by PKM2.
- Many mechanistic similarities and distinct differences were observed in the role of PKM2 in cancer.
- Pharmacological intervention designed to counteract early cancer stages of de-differentiation and transformation also may be a way to prevent neuronal AD-related fate loss in the brain.
Microtubule-normalizing agents as potential therapeutics for AD and related neurodegenerative disorders
Kurt R. Brunden, Ph.D., University of Pennsylvania
Dr. Brunden presented a mechanism-based drug discovery program that has been in process for a decade and uses an approach that has been used in development of anti-cancer drugs—a repurposing approach— focused on plaques and tangles, a key pathology observed in AD.
>Tau Pathology in Neurodegenerative Disease: Effects on Microtubule Structure and Function
Neuronal neurofibrillary tangles comprising the tau protein are a hallmark pathology of AD, with a high correlation between tau pathology, cognitive decline, and brain atrophy. Tau is an abundant microtubule (MT)-associated protein predominantly found in axons, where it appears to provide stabilization to more labile, distal portions of MTs and axons; blocks MT-severing enzymes; and may play a role in regulating MT motor protein interaction. In AD, tau becomes hyper-phosphorylated and disengages from MTs, increasing a cytosolic concentration that likely helps promote misfolding into the tangles and threads seen in AD and other tauopathies such as FTDs. Consequently, tau binding to MTs is reduced, which may affect their structure and function.
Evidence of MT abnormalities includes the following: a reduction of the stable MT marker acetyl-tubulin has been reported in AD neurons; the number and length of axonal MTs are decreased in AD neurons; and MT density and axonal transport are reduced in mouse models of tauopathy, with increased MT dynamicity.
aMT-Stabilizing Agents: Epothilone D (EpoD)
MT-stabilizing drugs such as Taxol (paclitaxel) and Taxotere (docetaxel) kill cancer cells by affecting MT dynamics in mitotic spindles during cell division. Might low doses of an MT-stabilizing drug help normalize MTs in the AD brain without causing side-effects seen in cancer treatment? Taxanes generally show poor BBB penetration. The epothilone class of MT-stabilizing compounds has full BBB penetration, with epothilone D (EpoD) showing good pharmacokinetic-pharmacodynamic (PK-PD) activity in the brain.
Bristol-Myers Squibb (BMS) developed EpoD and progressed it to Phase II trials but found it was not superior to existing anti-cancer agents and shelved it.
Dr. Brunden summarized an interventional study of EpoD in PS19 (Tau P301S) transgenic mice. PS19 mice overexpress the mutated form of tau that facilitates misfolding into tangles and show an age-dependent development of tau pathology that follows a pseudo-Braak stage. Nine-month-old mice at the equivalent of Braak stage 2 and 3 received low doses of EpoD over a 3-month period.
Doses were 30- to 100-fold lower than doses used in the Phase II cancer treatment studies. The team looked for evidence of efficacy (e.g., MT density, axonal dystrophy, behavioral testing, CNS pathology) and safety (e.g., no effect on cell division as marked by neutropenia, body/organ weight, behavior).
During the study period, untreated mice progressed to Braak stage 4 to 5 pathology and exhibited neuron loss evidenced by marked thinning of the hippocampus. Treated mice exhibited increased MT density, decreased axonal dystrophy, and reduction of tau pathology. EpoD prevented synapse and mossy fiber loss and improved cognitive performance. EpoD caused a slight dose-dependent improvement; when dosage reached 1 mg/kg neuron loss was inhibited.
Dr. Brunden observed that improving axonal transport seems to reduce further formation of tau pathology.
Following the mouse study, BMS conducted a Phase Ib Study of BMS-241027 (EpoD) in 40 AD patients. The multicenter randomized, double-blind placebo-controlled study outcomes included safety, CSF tau and neurofilament light chain (NfL), cognitive performance, and MRI. The study concluded with no reported safety issues and no demonstration of improved biomarker signal. The lack of a significant change in a disease biomarker in a 9-week trial is unsurprising given that typical AD disease-modifying trials are of 18–24 months duration. BMS largely exited AD research shortly after conducting this small trial.
MT-Stabilizing Agents: Triazolopyrimidine (TPD)
Dr. Brunden’s team decided to proceed on their own to identify additional MT-stabilizing agents that might have utility because they did not have rights to work with EpoD. This led them to a class of optimized, brain-penetrant, small-molecule triazolopyrimidine MT-normalizing compounds. TPD binds uniquely to MTs but elicits comparable effects to Taxanes and Epothilones. After demonstrating satisfactory brain PK-PD with a prototype TPD (CNDR-51657), they conducted a study with 9-month-old female PS19 tau transgenic mice with mild tau pathology. Effects similar to those in the EpoD mouse study were observed: reduction in dystrophic processes, increased MT density, and reduction in insoluble tau. There were no changes in neutrophil or white blood cell counts at efficacy doses, suggesting TPD had no negative effect on blood cell division.
Plaque-Associated MT Collapse: Enhanced Aβ Peptide Generation
A publication from the Vassar laboratory pointed to a plaque-associated MT collapse that disrupts axonal transport and leads to an accumulation of Aβ peptides. After confirming Vassar’s MT deficit finding, Dr. Brunden’s team reasoned that TPDs might normalize MT function, allowing more normal transport and, thus, reducing APP and BACE1 accumulation, and perhaps reducing secondary plaque biogenesis. This hypothesis proved true.
A study of 5XFAD mice at 1.5 months (when they begin forming plaques) treated with CNDR-51657 found a reduction of insoluble Aβ levels in the mouse brain. Ultimately, a reduction in total accumulation of fragments, APP itself, and BACE1 was observed. The team hypothesized that improving MT structure and function reduced APP and BACE1 accumulation and subsequent generation of Aβ and secondary plaques.
More recently, the team has conducted efficacy and tolerability testing of a second-generation TPD MT-normalizing compound. CNDR-51997 has shown efficacy in reducing plaque biogenesis in the 5XFAD mouse model in a dose-dependent manner. In doses up to 10mg/kg, no negative effects on body weight, organ weight, and blood cell division have been observed. Testing of this compound in the PS19 model is now underway.
Conclusions
- Low doses of MT-stabilizing agents such as EpoD and CNDR-51657 reduce tau pathology and axonal dystrophy in tau transgenic mice, with improved neuronal function and survival.
- MT-stabilizing TPD compounds such as CNDR-51657 and CNDR-51997 appear to correct MT collapse in plaque-associated dystrophic processes, leading to decreased APP and BACE1 accumulation with reduced Aβ plaque deposition.
- Doses required to elicit these improvements in mouse brain MTs are well tolerated.
- Dr. Brunden hopes to advance a lead TPD candidate for clinical testing in AD and/or related neurodegenerative diseases.
Breakout Sessions
Damali Martin, Ph.D., MPH, NIA, and Camille Pottinger, MPH, NIA
Dr. Martin described the purpose of the breakout sessions and invited Ms. Pottinger to describe the process for the breakout sessions.
Report-back and Presentation of Responses
Camille Pottinger, MPH, NIA
Ms. Pottinger summarized discussions from the breakout groups and their responses to three questions.
Future Directions
Dallas Anderson, Ph.D., MPH, NIA, and Damali Martin, Ph.D., MPH, NIA
Dr. Martin shared a list of next steps and potential future directions for this work:
- The meeting has been recorded. NIA will email attendees a link to the recording, which will be housed on the NIH website.
- A complete summary of the meeting will be written and shared with attendees.
- NIA will reach out to presenters to schedule a debrief meeting to discuss moving forward with a white paper.
- NIA will continue to communicate with the extramural community as the Institute develops further opportunities in cancer-AD research. This may include future convenings at conferences like the Alzheimer’s Association International Conference and the American Association for Cancer Research, as well as additional workshops.
Closing Remarks and Adjournment
NIA Staff
Dr. Martin invited her co-chairs to offer closing remarks. Drs. Anderson, Grothaus, Salive, and Horowitz offered their thanks to the meeting presenters, organizers, and attendees. Dr. Martin thanked Ms. Pottinger for her critical work in bringing the workshop together. Ms. Pottinger expressed her gratitude for the opportunity to collaborate with all who participated.
The meeting adjourned at 1:53 p.m. ET.
Contact Information
If you have questions about the meeting, contact Damali Martin (martinda@mail.nih.gov).