We know that many NDs affect people as they age, so as people live longer, the impact of these diseases and burden of care they generate will increase alongside. This creates an even more pressing need for drug developers to increase their understanding of what causes these diseases, so they can develop new treatment approaches.
Alzheimer Disease (AD) – The Most Elusive ND
The most prevalent of these diseases, AD is an irreversible, progressive brain disorder that causes brain cells to degenerate and die. Despite extensive medical research efforts, AD remains the most intractable NDs; the failure rate for developing disease-modifying therapies for AD is 100%.
All eyes recently were on aducanumab, a drug targeted at amyloid beta, which was granted Priority Review by US Food and Drug Administration (FDA) in August 2020. Its future is now unclear since FDA’s Peripheral and Central Nervous System Drugs Advisory Committee voted on November 6, 2020 to recommend that FDA not approve aducanumab for the treatment of AD, because there is not substantial evidence of effectiveness.
Several factors have likely contributed to the failure of amyloid targeted drugs. They include late initiation of treatments, inappropriate drug dosages, incorrect selection of main treatment targets, and inadequate understanding of AD’s complex pathophysiology.2
Additional challenges include difficulty translating findings from preclinical animal models to humans, few biomarkers sensitive to therapeutic interventions, incomplete understanding of the disease pathology, different comorbidities converging in the elderly brain, and variability between and within patient populations based on their different comedications and genotypes.
Fortunately, despite this bleak history, there are still a significant number of drugs and targets under development for AD (see Figure 1).3
Figure 1
Clearly, a new approach is required. Many researchers are recommending that multi-target, combination treatments be explored rather than a monotherapy. That strategy acknowledges that no single factor causes AD; the symptoms are triggered by a combination of genetic, lifestyle and environmental factors.
What is Biosimulation, and Why is QSP the Right Approach for ND?
Biosimulation (also called model-informed drug development or MIDD) is a powerful technology used to conduct computer-based trials using virtual patients to predict how drugs behave in different individuals. Increasingly encouraged by regulators, biopharmaceutical companies use biosimulation software throughout drug discovery and development to inform critical decisions that not only save significant time and money but also help to advance drug safety and efficacy, improving millions of lives each year.
Built on first principles of biology, chemistry, and pharmacology with proprietary mathematical algorithms, biosimulation predicts how medicines and diseases behave in the body. It is used to conduct virtual trials to answer critical questions, such as: What will the human response to a drug be based on preclinical data? How will other drugs interfere or interact with this new drug? What is a safe and efficacious dose for children, the elderly, or patients with pre-existing conditions?
Quantitative systems pharmacology (QSP), the newest and fastest growing biosimulation technology, is being heralded as the requisite fresh approach for ND drug development.4 QSP combines computational modeling and experimental methods to examine the mechanistic relationships between a drug, the biological system, and the disease process. QSP integrates quantitative drug data with knowledge of the drug’s mechanism of action. Therefore, it can facilitate the evaluation of complex, heterogeneous diseases, such as ND.
Figure 24
One of QSP’s primary strengths lies in the fact that it focuses on entire biological systems instead of individual drug targets. It allows researchers to take a step back and look at the big picture view of the impact that the disease and different drug combinations are having on the body.
That all-encompassing approach is particularly important because NDs patients, especially those with AD, tend to be elderly and often have comorbidities, requiring them to take multiple medications at the same time. They will also require individualized treatment (personalized medicine).
QSP modeling is particularly well suited for this application because it allows many quantitative and predictive models to be integrated, forming a single, multiscale simulation platform. In this case, QSP science, methods, and technology can be combined with immunology and inflammation insights from immuno-oncology research, pertinent biomarkers, and data from genetic, cellular, and neural scales.
The resulting NDs simulator could then be integrated with a physiologically based pharmacokinetic (PBPK) simulator, allowing hypotheses to be tested in virtual patient populations with different demographics, ethnicities, and co-morbidities. This aspect is particularly important because the move toward combination therapies would necessitate evaluating more drug combinations than could practically or ethically be tested in clinical trials with real patients.
In addition to supporting dose selection and optimization for combination therapies, this paired QSP/PBPK simulator could be used to investigate different therapeutic modalities, such as small molecules, biologics, antisense oligonucleotides, and gene therapy. It would also enable protein biomarker levels to be assessed in different parts of the body, including cerebrospinal fluid (CSF) and plasma.
Regulatory Perspective on QSP
During the past five years, QSP has progressed from an academic discipline to a regulatory requirement. QSP adoption has moved faster than previous modeling discussions with global regulatory agencies; it took far longer for PBPK modeling to reach a similar level of acceptance.
FDA no longer asks sponsors to justify the value of QSP but is instead requesting their assistance in establishing industry best practices for applying it. Regulators and sponsors are working together to determine how best to evaluate complex QSP models.
For example, Professor van der Graaf cochaired FDA Industry QSP Exchange Workshop with Raj Madabushi, Ph.D., on July 1, 2020. At that point, FDA had already received 22 submissions with QSP data in their review packet in 2020. Out of those applications, neuroscience indications were second only to oncology. A wide range of cases were presented. QSP was employed to support translation, product differentiation, and utility of biomarkers; address nuanced drug development aspects; inform dose selection and dosing; provide insights into currently accepted endpoints; provide a better understanding of the underlying pathophysiology; and identify druggable pathways.
Further underscoring QSP modeling’s acceptance, about 15% of applications for the MIDD Pilot Program included in Prescription Drug User Fee Act (PDUFA) VI are QSP focused.
FDA also noted at the workshop that while there were consistent views, there were some different perspectives, and it would be useful to develop a common understanding or framework for ascertaining decision-risk and model-influence. The presence of multiple approaches to model development and model calibration point to the need to establish industry best practices.
They would also help solve the current challenge of model reproducibility, which could limit regulatory acceptance and damage QSP’s credibility in the industry. For example, in a paper published in 2019, only 30%5 of the models could be replicated. In a more recent paper, the reproducibility jumped to 50%,6 but that is still below expectations for robust software. That issue could be addressed by learning from developers of commercially available biosimulation software, such as Phoenix, NONMEM, Gastro-Plus, and Simcyp.
QSP and AD
AD’s defining characteristics include the buildup of amyloid plaques and neurofibrillary tangles in the brain consisting of abnormal tau protein, which usually stabilizes the neuronal cytoskeleton. A growing number of neuropathology, experimental preclinical, and genetic studies, together with positron emission tomography (PET) imaging in live patients, have also identified other contributing processes.
Genetic and clinical evidence suggest that amyloid pathology is an important driver of clinical outcomes for AD patients.4 But biological differences between short and long forms of the Aβ protein have been documented, and soluble Aβ oligomers, cerebral amyloid angiopathy, and clearance through the glymphatic system also seem to play a role. Recent results using amyloid-modulating agents to treat AD produced substantial target engagement, but no effect on patients’ cognitive clinical scales, underscoring the difficulty of translating scientific knowledge into therapeutic advances. In addition, the appearance of amyloid related imaging abnormalities (ARIA) at the blood-brain barrier can complicate treatment with amyloid-modulating agents.7
Interactions with the glutamatergic and nicotinic neurotransmitter system and neuronal firing activity are also dependent upon the specific form of the amyloid peptide, which could explain some of the unexpected clinical observations.8,9
Imaging data also suggest that tau pathology is a strong predictor of clinical outcomes.10,11 A recent report indicates that individual cognitive decline is associated with tau species’ tendency to drive aggregation and seeding.12 Furthermore, there is increasing evidence to support the concept of prion-like progression along well-defined axonal trajectories with tau pathology.13
AD vascular pathology negatively affects proper metabolic functioning of neurons and possibly reduces clearance of misfolded proteins.14 Data from recent genetic and proteomic studies also indicate that neuroinflammation plays an important but complex role in clearance of misfolded proteins and neuroprotection.13
Furthermore, large cohort neuropathology studies have shown that proper neuronal functioning is even more important because synapse loss is strongly correlated with reduction in cognitive functioning.15 Therefore, the contributions of neurotoxicity and mitochondrial dysfunction should also be considered.15.
To efficiently research AD drugs, a QSP simulator needs to be developed that can integrate β-amyloid aggregation and spatiotemporal tau progression with biochemical and imaging biomarkers and link that information to clinically calibrated modules that assess functional outcomes.
Fortunately, QSP models have already been developed for trans-neuronal tau spreading, post translation tau modification, tau synaptic dysfunction, and amyloid/tau aggregation. In addition, a QSP model that links the impact of molecular processes such as levels of toxic β-amyloid and tau to neuronal firing activity of well-defined microcircuits that generate a calibrated functional clinical scale outcome has been developed and validated. A QSP platform for simulating neuroinflammation has also been created that can integrate with the PBPK simulator to perform computer-based trials on virtual patients.
Case Study – Asyn for Parkinson Disease
A common theme with many NDs is the formation and propagation of misfolded proteins, as exemplified by tau in AD. Working with GSK, we developed a QSP model of Parkinson disease pathogenesis and propagation, including α-synuclein (Asyn), specifically focused on modulating Lag 3 receptors.16 This model was developed to select targets not just based on biology, but also pharmacology and dose.
Figure 3 shows misfolded Asyn in the middle of the biological knowledge map of the model, together with various feedback loops.16 Positive loops are shown in blue. For example, mitochondrial damage leads to oxidative stress, which creates more misfolded Asyn. Similarly, misfolded Asyn increases dopamine levels leading to more Asyn expression. Other effects, such as the decline in proteosomal/lysosomal clearance with ageing, can also be taken into account. In this case, Asyn aggregation was modeled with QSP from monomers to dimers, multimers and ultimately to misfolded proteins. This approach is also being used with amyloid and tau.
Figure 3
The model then has to be put into physiological context in the brain. The model is informed by clinical reality, as provided by brain tissue samples from the University of Amsterdam. The resulting model simulates the spread of Asyn over time throughout the brain. It shows how much of the brain stays healthy or gets infected with misfolded Asyn. The output of the model is shown below with the healthy brain in blue and the diseased brain in yellow.
There are several ways in which Asyn levels can be modulated using hypothetical antibodies—monomers, oligomers, fibrils, or target of interest, such as Lag 3. The impact of increasing doses of these hypothetical antibodies is demonstrated in Figure 4 from left to right. Realistic concentrations at the neuronal level are derived from PBPK simulations. By targeting oligomers and Lag 3, the deterioration can be slowed. Targeting the fibrils accelerates the deterioration, suggesting that is not a good target. But, if the Asyn monomers can be removed (top row), it would not only slow the spread, in principle, it could even reverse it. This QSP model suggest that not all Asyn is the same.
Figure 4
The QSP model enables the effectiveness of blocking selected targets in the pathway to be tested. This model can also be used to guide target and dose selection.
Conclusion
Focusing on the complex pathologic pathways of ND, together with a better understanding of the relationships between the many mechanisms involved in the pathophysiology and progress of these diseases, will allow us to better support successful therapies. By integrating preclinical and clinical knowledge about misfolded proteins and their impact on neuroinflammation, QSP should be able to advance ND drug discovery and development. Of course, managing these multiple factors to identify the optimal combination therapy requires the ability to evaluate an immense number of scenarios. QSP is the most appropriate technology for this application because it can leverage big data to enable the understanding of disease pathophysiology, and the dynamic interactions between drug(s) and the biological system. It allows therapeutic strategies to be tested systematically and the combination with the greatest synergy in silico to be identified, before investing in long and expensive clinical trials.
Piet van der Graaf, Ph.D., is senior vice president and head of QSP.
Hugo Geerts, Ph.D., is head of QSP modeling for neuroscience at Certara.
Phoenix and Simcyp are trademarks osif Certara L.P. All other trademarks and tradenames used herein are owned by their respective owner.
References
- Harvard Neurodiscovery Center. The Challenge of Neurodegenerative Diseases.
- Yiannopoulou KG, Anastasiou AI, Zachariou V, Pelidou SH (2019) Reasons for Failed Trials of Disease-Modifying Treatments for Alzheimer Disease and Their Contribution in Recent Research. Biomedicines
- Cummins J, Lee, G, Ritter,A, Sabbagh, M, Zhong , K (2019) Alzheimer’s Disease Drug Development Pipeline: 2019. Alzheimers Dement, 1016/j.trci.2019.05.008.
- Nijsen, M et al. Preclinical QSP Modeling in the Pharmaceutical Industry: An IQ Consortium Survey Examining the Current Landscape (2018). CPT Pharmacometrics Syst Pharmacol. 2018 7(3) 135-146.
- Sakai K, Boche D, Carare R, Johnston D, Holmes C, Love S, Nicoll JA (2014) Abeta immunotherapy for Alzheimer's disease: effects on apoE and cerebral vasculopathy. Acta Neuropathologica 128, 777-789.
- Geerts H, Spiros A (2020) Learning from amyloid trials in Alzheimer's disease. A virtual patient analysis using a quantitative systems pharmacology approach. Alzheimer's & dementia: the journal of the Alzheimer's Association.
- Geerts H, Spiros A, Roberts P (2018) Impact of amyloid-beta changes on cognitive outcomes in Alzheimer's disease: analysis of clinical trials using a quantitative systems pharmacology model. Alzheimer's research & therapy 10, 14.
- Johnson KA, Schultz A, Betensky RA, Becker JA, Sepulcre J, Rentz D, Mormino E, Chhatwal J, Amariglio R, Papp K, Marshall G, Albers M, Mauro S, Pepin L, Alverio J, Judge K, Philiossaint M, Shoup T, Yokell D, Dickerson B, Gomez-Isla T, Hyman B, Vasdev N, Sperling R (2016) Tau positron emission tomographic imaging in aging and early Alzheimer disease. Annals of Neurology 79, 110-119.
- Braak H, Braak E (1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathologica 82, 239-259.
- Dujardin S, Commins C, Lathuiliere A, Beerepoot P, Fernandes AR, Kamath TV, De Los Santos MB, Klickstein N, Corjuc DL, Corjuc BT, Dooley PM, Viode A, Oakley DH, Moore BD, Mullin K, Jean-Gilles D, Clark R, Atchison K, Moore R, Chibnik LB, Tanzi RE, Frosch MP, Serrano-Pozo A, Elwood F, Steen JA, Kennedy ME, Hyman BT (2020) Tau molecular diversity contributes to clinical heterogeneity in Alzheimer's disease. Nature Medicine.
- Gibbons GS, Lee VMY, Trojanowski JQ (2019) Mechanisms of Cell-to-Cell Transmission of Pathological Tau: A Review. JAMA neurology 76, 101-108.
- Mestre H, Mori Y, Nedergaard M (2020) The Brain's Glymphatic System: Current Controversies. Trends in Neurosciences 43, 458-466.
- Edison P, Brooks DJ (2018) Role of Neuroinflammation in the Trajectory of Alzheimer's Disease and in vivo Quantification Using PET. Journal of Alzheimer's disease : JAD 64, S339-S351.
- Bakshi, S, et al (2019) Mathematical Models of Parkinsons Disease CPT Pharmacometrics Syst. Pharmacol. (2019) 8, 77–86.
This article is licensed under the CC BY-ND 4.0 license, which permits copying and redistributing the article with attribution to AAPS Newsmagazine as the original publisher, and no derivatives of this work are allowed.