1
|
Naik K, Goyal RK, Foschini L, Chak CW, Thielscher C, Zhu H, Lu J, Lehár J, Pacanoswki MA, Terranova N, Mehta N, Korsbo N, Fakhouri T, Liu Q, Gobburu J. Current Status and Future Directions: The Application of Artificial Intelligence/Machine Learning for Precision Medicine. Clin Pharmacol Ther 2024; 115:673-686. [PMID: 38103204 DOI: 10.1002/cpt.3152] [Citation(s) in RCA: 2] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/16/2023] [Accepted: 11/28/2023] [Indexed: 12/18/2023]
Abstract
Technological innovations, such as artificial intelligence (AI) and machine learning (ML), have the potential to expedite the goal of precision medicine, especially when combined with increased capacity for voluminous data from multiple sources and expanded therapeutic modalities; however, they also present several challenges. In this communication, we first discuss the goals of precision medicine, and contextualize the use of AI in precision medicine by showcasing innovative applications (e.g., prediction of tumor growth and overall survival, biomarker identification using biomedical images, and identification of patient population for clinical practice) which were presented during the February 2023 virtual public workshop entitled "Application of Artificial Intelligence and Machine Learning for Precision Medicine," hosted by the US Food and Drug Administration (FDA) and University of Maryland Center of Excellence in Regulatory Science and Innovation (M-CERSI). Next, we put forward challenges brought about by the multidisciplinary nature of AI, particularly highlighting the need for AI to be trustworthy. To address such challenges, we subsequently note practical approaches, viz., differential privacy, synthetic data generation, and federated learning. The proposed strategies - some of which are highlighted presentations from the workshop - are for the protection of personal information and intellectual property. In addition, methods such as the risk-based management approach and the need for an agile regulatory ecosystem are discussed. Finally, we lay out a call for action that includes sharing of data and algorithms, development of regulatory guidance documents, and pooling of expertise from a broad-spectrum of stakeholders to enhance the application of AI in precision medicine.
Collapse
Affiliation(s)
- Kunal Naik
- Office of Clinical Pharmacology, Office of Translational Sciences, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | - Rahul K Goyal
- Center for Translational Medicine, University of Maryland School of Pharmacy, Baltimore, Maryland, USA
| | | | | | | | - Hao Zhu
- Office of Clinical Pharmacology, Office of Translational Sciences, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | - James Lu
- Modeling & Simulation/Clinical Pharmacology, Genentech Inc., South San Francisco, California, USA
| | | | - Michael A Pacanoswki
- Office of Clinical Pharmacology, Office of Translational Sciences, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | - Nadia Terranova
- Quantitative Pharmacology, Ares Trading S.A. (an affiliate of Merck KGaA, Darmstadt, Germany), Lausanne, Switzerland
| | - Neha Mehta
- Office of Clinical Pharmacology, Office of Translational Sciences, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | | | - Tala Fakhouri
- Office of Medical Policy, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | - Qi Liu
- Office of Clinical Pharmacology, Office of Translational Sciences, Center for Drug Evaluation and Research, US Food and Drug Administration, Silver Spring, Maryland, USA
| | - Jogarao Gobburu
- Center for Translational Medicine, University of Maryland School of Pharmacy, Baltimore, Maryland, USA
| |
Collapse
|
2
|
Naqash AR, Floudas CS, Aber E, Maoz A, Nassar AH, Adib E, Choucair K, Xiu J, Baca Y, Ricciuti B, Alessi JV, Awad MM, Kim C, Judd J, Raez LE, Lopes G, Nieva JJ, Borghaei H, Takebe N, Ma PC, Halmos B, Kwiatkowski DJ, Liu SV, Mamdani H. Influence of TP53 Comutation on the Tumor Immune Microenvironment and Clinical Outcomes With Immune Checkpoint Inhibitors in STK11-Mutant Non-Small-Cell Lung Cancer. JCO Precis Oncol 2024; 8:e2300371. [PMID: 38330261 PMCID: PMC10860998 DOI: 10.1200/po.23.00371] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/26/2023] [Revised: 11/05/2023] [Accepted: 01/02/2024] [Indexed: 02/10/2024] Open
Abstract
PURPOSE Non-small-cell lung cancer (NSCLC) with STK11mut has inferior outcomes to immune checkpoint inhibitors (ICIs). Using multiomics, we evaluated whether a subtype of STK11mut NSCLC with a uniquely inflamed tumor immune microenvironment (TIME) harboring TP53 comutations could have favorable outcomes to ICIs. PATIENTS AND METHODS NSCLC tumors (N = 16,896) were analyzed by next-generation sequencing (DNA-Seq/592 genes). A subset (n = 5,034) underwent gene expression profiling (RNA-Seq/whole transcriptome). Exome-level neoantigen load for STK11mut NSCLC was obtained from published pan-immune analysis. Tumor immune cell content was obtained from transcriptome profiles using the microenvironment cell population (MCP) counter. ICI data from POPLAR/OAK (n = 34) and the study by Rizvi et al (n = 49) were used to model progression-free survival (PFS), and a separate ICI-treated cohort (n = 53) from Dana-Farber Cancer Institute (DFCI) was used to assess time to treatment failure (TTF) and tumor RECIST response for STK11mutTP53mut versus STK11mutTP53wt NSCLC. RESULTS Overall, 12.6% of NSCLC tumors had a STK11mut with the proportions of tumor mutational burden (TMB)-high (≥10 mut/Mb), PD-L1 ≥50%, and microsatellite instability-high being 38.3%, 11.8%, and 0.72%, respectively. Unsupervised hierarchical clustering of STK11mut (n = 463) for stimulator of interferon-gamma (STING) pathway genes identified a STING-high cluster, which was significantly enriched in TP53mut NSCLC (P < .01). Compared with STK11mutTP53wt, tumors with STK11mutTP53mut had higher CD8+T cells and natural killer cells (P < .01), higher TMB (P < .001) and neoantigen load (P < .001), and increased expression of MYC and HIF-1A (P < .01), along with higher expression (P < .01) of glycolysis/glutamine metabolism genes. Meta-analysis of data from OAK/POPLAR and the study by Rizvi et al showed a trend toward improved PFS in patients with STK11mutTP53mut. In the DFCI cohort, compared with the STK11mut TP53wt cohort, the STK11mutTP53mut tumors had higher objective response rates (42.9% v 16.7%; P = .04) and also had longer TTF (14.5 v 4.5 months, P adj = .054) with ICI. CONCLUSION STK11mut NSCLC with TP53 comutation is a distinct subgroup with an immunologically active TIME and metabolic reprogramming. These properties should be exploited to guide patient selection for novel ICI-based combination approaches.
Collapse
Affiliation(s)
- Abdul Rafeh Naqash
- Stephenson Cancer Center, University of Oklahoma Health Sciences Center, Oklahoma City, OK
| | | | - Etan Aber
- Center for Immuno-Oncology, National Cancer Institute, NIH, Bethesda, MD
| | - Asaf Maoz
- Dana Farber Cancer Institute, Boston, MA
| | - Amin H. Nassar
- Department of Hematology/Oncology, Yale New Haven Hospital, New Haven, CT
| | - Elio Adib
- Dana Farber Cancer Institute, Boston, MA
| | - Khalil Choucair
- Department of Oncology, Barbara Ann Karmanos Cancer Institute, Wayne State University, Detroit, MI
| | | | | | | | | | | | - Chul Kim
- Department of Hematology and Oncology, Georgetown University, Washington, DC
| | - Julia Judd
- Fox Chase Cancer Center, Philadelphia, PA
| | - Luis E. Raez
- Memorial Cancer Institute//Florida Atlantic University (FAU), Miami, FL
| | - Gilberto Lopes
- University of Miami Miller School of Medicine, Miami, FL
| | | | | | - Naoko Takebe
- Developmental Therapeutics Clinic, National Cancer Institute, Bethesda, MD
| | - Patrick C. Ma
- Department of Hematology/ Oncology, Penn State Cancer Institute, Hershey, PA
| | - Balazs Halmos
- Medical Oncology, Albert Einstein College of Medicine, NY
| | | | - Stephen V. Liu
- Department of Hematology and Oncology, Georgetown University, Washington, DC
| | - Hirva Mamdani
- Department of Oncology, Barbara Ann Karmanos Cancer Institute, Wayne State University, Detroit, MI
| |
Collapse
|
3
|
Ibrahim EIK, Ellingsen EB, Mangsbo SM, Friberg LE. Bridging responses to a human telomerase reverse transcriptase-based peptide cancer vaccine candidate in a mechanism-based model. Int Immunopharmacol 2024; 126:111225. [PMID: 37988911 DOI: 10.1016/j.intimp.2023.111225] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/10/2023] [Revised: 11/10/2023] [Accepted: 11/10/2023] [Indexed: 11/23/2023]
Abstract
Therapeutic cancer vaccines are novel immuno-therapeutics, aiming to improve clinical outcomes with other immunotherapies. However, obstacles to their successful clinical development remain, which model-informed drug development approaches may address. UV1 is a telomerase based therapeutic cancer vaccine candidate being investigated in phase I clinical trials for multiple indications. We developed a mechanism-based model structure, using a nonlinear mixed-effects modeling techniques, based on longitudinal tumor sizes (sum of the longest diameters, SLD), UV1-specific immunological assessment (stimulation index, SI) and overall survival (OS) data obtained from a UV1 phase I trial including non-small cell lung cancer (NSCLC) patients and a phase I/IIa trial including malignant melanoma (MM) patients. The final structure comprised a mechanistic tumor growth dynamics (TGD) model, a model describing the probability of observing a UV1-specific immune response (SI ≥ 3) and a time-to-event model for OS. The mechanistic TGD model accounted for the interplay between the vaccine peptides, immune system and tumor. The model-predicted UV1-specific effector CD4+ T cells induced tumor shrinkage with half-lives of 103 and 154 days in NSCLC and MM patients, respectively. The probability of observing a UV1-specific immune response was mainly driven by the model-predicted UV1-specific effector and memory CD4+ T cells. A high baseline SLD and a high relative increase from nadir were identified as main predictors for a reduced OS in NSCLC and MM patients, respectively. Our model predictions highlighted that additional maintenance doses, i.e. UV1 administration for longer periods, may result in more sustained tumor size shrinkage.
Collapse
Affiliation(s)
| | - Espen B Ellingsen
- Department of Tumor Biology, Institute for Cancer Research, Oslo University Hospital, Oslo, Norway; Faculty of Medicine, University of Oslo, Oslo, Norway; Ultimovacs ASA, Oslo, Norway
| | - Sara M Mangsbo
- Department of Pharmacy, Uppsala University, Uppsala, Sweden; Ultimovacs AB, Uppsala, Sweden
| | - Lena E Friberg
- Department of Pharmacy, Uppsala University, Uppsala, Sweden.
| |
Collapse
|
4
|
Laurie M, Lu J. Explainable deep learning for tumor dynamic modeling and overall survival prediction using Neural-ODE. NPJ Syst Biol Appl 2023; 9:58. [PMID: 37980358 PMCID: PMC10657412 DOI: 10.1038/s41540-023-00317-1] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/20/2023] [Accepted: 10/23/2023] [Indexed: 11/20/2023] Open
Abstract
While tumor dynamic modeling has been widely applied to support the development of oncology drugs, there remains a need to increase predictivity, enable personalized therapy, and improve decision-making. We propose the use of Tumor Dynamic Neural-ODE (TDNODE) as a pharmacology-informed neural network to enable model discovery from longitudinal tumor size data. We show that TDNODE overcomes a key limitation of existing models in its ability to make unbiased predictions from truncated data. The encoder-decoder architecture is designed to express an underlying dynamical law that possesses the fundamental property of generalized homogeneity with respect to time. Thus, the modeling formalism enables the encoder output to be interpreted as kinetic rate metrics, with inverse time as the physical unit. We show that the generated metrics can be used to predict patients' overall survival (OS) with high accuracy. The proposed modeling formalism provides a principled way to integrate multimodal dynamical datasets in oncology disease modeling.
Collapse
Affiliation(s)
- Mark Laurie
- Modeling & Simulation/Clinical Pharmacology, Genentech, South San Francisco, CA, USA
- Department of Computer Science, Stanford University, Stanford, CA, USA
| | - James Lu
- Modeling & Simulation/Clinical Pharmacology, Genentech, South San Francisco, CA, USA.
| |
Collapse
|
5
|
Khandelwal A, Grisic AM, French J, Venkatakrishnan K. Pharmacometrics Golems: Exposure-Response Models in Oncology. Clin Pharmacol Ther 2022; 112:941-945. [PMID: 35286713 DOI: 10.1002/cpt.2564] [Citation(s) in RCA: 3] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/21/2021] [Accepted: 02/17/2022] [Indexed: 12/24/2022]
|
6
|
Kerioui M, Bertrand J, Bruno R, Mercier F, Guedj J, Desmée S. Modelling the association between biomarkers and clinical outcome: an introduction to nonlinear joint models. Br J Clin Pharmacol 2022; 88:1452-1463. [PMID: 34993985 DOI: 10.1111/bcp.15200] [Citation(s) in RCA: 5] [Impact Index Per Article: 2.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/13/2021] [Revised: 10/12/2021] [Accepted: 11/07/2021] [Indexed: 11/30/2022] Open
Abstract
Nonlinear joint models are a powerful tool to precisely analyze the association between a nonlinear biomarker and a time-to-event process, such as death. Here, we review the main methodological techniques required to build these models and to make inferences and predictions. We describe the main clinical applications and discuss the future developments of such models.
Collapse
Affiliation(s)
- Marion Kerioui
- Université de Paris, INSERM IAME, Paris, France.,Université de , Université de Nantes, INSERM SPHERE, UMR Tours, Tours, France.,Institut Roche, Boulogne-Billancourt, France.,Genentech/Roche, Clinical Pharmacology, Paris, France
| | | | - René Bruno
- Genentech/Roche, Clinical Pharmacology, Marseille, France
| | | | | | - Solène Desmée
- Université de , Université de Nantes, INSERM SPHERE, UMR Tours, Tours, France
| |
Collapse
|
7
|
Krishnan SM, Friberg LE, Bruno R, Beyer U, Jin JY, Karlsson MO. Multistate model for pharmacometric analyses of overall survival in HER2-negative breast cancer patients treated with docetaxel. CPT-PHARMACOMETRICS & SYSTEMS PHARMACOLOGY 2021; 10:1255-1266. [PMID: 34313026 PMCID: PMC8520749 DOI: 10.1002/psp4.12693] [Citation(s) in RCA: 11] [Impact Index Per Article: 3.7] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 04/09/2021] [Revised: 06/09/2021] [Accepted: 06/24/2021] [Indexed: 11/16/2022]
Abstract
The aim of this study was to develop a multistate model for overall survival (OS) analysis, based on parametric hazard functions and combined with an investigation of predictors derived from a longitudinal tumor size model on the transition hazards. Different states – stable disease, tumor response, progression, second‐line treatment, and death following docetaxel treatment initiation (stable state) in patients with HER2‐negative breast cancer (n = 183) were used in model building. Past changes in tumor size prospectively predicts the probability of state changes. The hazard of death after progression was lower for subjects who had longer treatment response (i.e., longer time‐to‐progression). Young age increased the probability of receiving second‐line treatment. The developed multistate model adequately described the transitions between different states and jointly the overall event and survival data. The multistate model allows for simultaneous estimation of transition rates along with their tumor model derived metrics. The metrics were evaluated in a prospective manner so not to cause immortal time bias. Investigation of predictors and characterization of the time to develop response, the duration of response, the progression‐free survival, and the OS can be performed in a single multistate modeling exercise. This modeling approach can be applied to other cancer types and therapies to provide a better understanding of efficacy of drug and characterizing different states, thereby facilitating early clinical interventions to improve anticancer therapy.
Collapse
Affiliation(s)
| | - Lena E Friberg
- Department of Pharmacy, Uppsala University, Uppsala, Sweden
| | - René Bruno
- Clinical Pharmacology, Roche/Genentech, Marseille, France
| | - Ulrich Beyer
- Biostatistics, F. Hoffmann-La-Roche Ltd, Basel, Switzerland
| | - Jin Y Jin
- Clinical Pharmacology Roche/Genentech, South San Francisco, California, USA
| | | |
Collapse
|
8
|
Benzekry S. Artificial Intelligence and Mechanistic Modeling for Clinical Decision Making in Oncology. Clin Pharmacol Ther 2020; 108:471-486. [PMID: 32557598 DOI: 10.1002/cpt.1951] [Citation(s) in RCA: 37] [Impact Index Per Article: 9.3] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/27/2020] [Accepted: 06/04/2020] [Indexed: 12/24/2022]
Abstract
The amount of "big" data generated in clinical oncology, whether from molecular, imaging, pharmacological, or biological origin, brings novel challenges. To mine efficiently this source of information, mathematical models able to produce predictive algorithms and simulations are required, with applications for diagnosis, prognosis, drug development, or prediction of the response to therapy. Such mathematical and computational constructs can be subdivided into two broad classes: biologically agnostic, statistical models using artificial intelligence techniques, and physiologically based, mechanistic models. In this review, recent advances in the applications of such methods in clinical oncology are outlined. These include machine learning applied to big data (omics, imaging, or electronic health records), pharmacometrics and quantitative systems pharmacology, as well as tumor kinetics and metastasis modeling. Focus is set on studies with high potential of clinical translation, and particular attention is given to cancer immunotherapy. Perspectives are given in terms of combinations of the two approaches: "mechanistic learning."
Collapse
Affiliation(s)
- Sebastien Benzekry
- MONC Team, Inria Bordeaux Sud-Ouest, Talence, France
- Institut de Mathématiques de Bordeaux, CNRS UMR 5251, Bordeaux University, Talence, France
| |
Collapse
|