To modulate or to skip: De-escalating PARP inhibitor maintenance therapy in ovarian cancer using adaptive therapy.

Toxicity and emerging drug resistance pose important challenges in poly-adenosine ribose polymerase inhibitor (PARPi) maintenance therapy of ovarian cancer. We propose that adaptive therapy, which dynamically reduces treatment based on the tumor dynamics, might alleviate both issues. Utilizing in vitro time-lapse microscopy and stepwise model selection, we calibrate and validate a differential equation mathematical model, which we leverage to test different plausible adaptive treatment schedules. Our model indicates that adjusting the dosage, rather than skipping treatments, is more effective at reducing drug use while maintaining efficacy due to a delay in cell kill and a diminishing dose-response relationship. In vivo pilot experiments confirm this conclusion. Although our focus is toxicity mitigation, reducing drug use may also delay resistance. This study enhances our understanding of PARPi treatment scheduling and illustrates the first steps in developing adaptive therapies for new treatment settings. A record of this paper's transparent peer review process is included in the supplemental information.

Lung adenocarcinomas without driver genes converge to common adaptive strategies through diverse genetic, epigenetic, and niche construction evolutionary pathways.

Somatic evolution selects cancer cell phenotypes that maximize survival and proliferation in dynamic environments. Although cancer cells are molecularly heterogeneous, we hypothesized convergent adaptive strategies to common host selection forces can be inferred from patterns of epigenetic and genetic evolutionary selection in similar tumors. We systematically investigated gene mutations and expression changes in lung adenocarcinomas with no common driver genes (n = 313). Although 13,461 genes were mutated in at least one sample, only 376 non-synonymous mutations evidenced positive evolutionary selection with conservation of 224 genes, while 1736 and 2430 genes exhibited ≥ two-fold increased and ≥ 50% decreased expression, respectively. Mutations under positive selection are more frequent in genes with significantly altered expression suggesting they often "hardwire" pre-existing epigenetically driven adaptations. Conserved genes averaged 16-fold higher expression in normal lung tissue compared to those with selected mutations demonstrating pathways necessary for both normal cell function and optimal cancer cell fitness. The convergent LUAD phenotype exhibits loss of differentiated functions and cell-cell interactions governing tissue organization. Conservation with increased expression is found in genes associated with cell cycle, DNA repair, p53 pathway, epigenetic modifiers, and glucose metabolism. No canonical driver gene pathways exhibit strong positive selection, but extensive down-regulation of membrane ion channels suggests decreased transmembrane potential may generate persistent proliferative signals. NCD LUADs perform niche construction generating a stiff, immunosuppressive microenvironment through selection of specific collagens and proteases. NCD LUADs evolve to a convergent phenotype through a network of interconnected genetic, epigenetic, and ecological pathways.

Modeling non-genetic information dynamics in cells using reservoir computing.

Virtually all cells use energy-driven, ion-specific membrane pumps to maintain large transmembrane gradients of Na+, K+, Cl-, Mg++, and Ca++, but the corresponding evolutionary benefit remains unclear. We propose that these gradients enable a dynamic and versatile biological system that acquires, analyzes, and responds to environmental information. We hypothesize that environmental signals are transmitted into the cell by ion fluxes along pre-existing gradients through gated ion-specific membrane channels. The consequent changes in cytoplasmic ion concentration can generate a local response or orchestrate global/regional cellular dynamics through wire-like ion fluxes along pre-existing and self-assembling cytoskeleton to engage the endoplasmic reticulum, mitochondria, and nucleus.

Mathematical Model-Driven Deep Learning Enables Personalized Adaptive Therapy.

Standard-of-care treatment regimens have long been designed for maximal cell killing, yet these strategies often fail when applied to metastatic cancers due to the emergence of drug resistance. Adaptive treatment strategies have been developed as an alternative approach, dynamically adjusting treatment to suppress the growth of treatment-resistant populations and thereby delay, or even prevent, tumor progression. Promising clinical results in prostate cancer indicate the potential to optimize adaptive treatment protocols. Here, we applied deep reinforcement learning (DRL) to guide adaptive drug scheduling and demonstrated that these treatment schedules can outperform the current adaptive protocols in a mathematical model calibrated to prostate cancer dynamics, more than doubling the time to progression. The DRL strategies were robust to patient variability, including both tumor dynamics and clinical monitoring schedules. The DRL framework could produce interpretable, adaptive strategies based on a single tumor burden threshold, replicating and informing optimal treatment strategies. The DRL framework had no knowledge of the underlying mathematical tumor model, demonstrating the capability of DRL to help develop treatment strategies in novel or complex settings. Finally, a proposed five-step pathway, which combined mechanistic modeling with the DRL framework and integrated conventional tools to improve interpretability compared to traditional "black-box" DRL models, could allow translation of this approach to the clinic. Overall, the proposed framework generated personalized treatment schedules that consistently outperformed clinical standard-of-care protocols.

Deep reinforcement learning identifies personalized intermittent androgen deprivation therapy for prostate cancer.

The evolution of drug resistance leads to treatment failure and tumor progression. Intermittent androgen deprivation therapy (IADT) helps responsive cancer cells compete with resistant cancer cells in intratumoral competition. However, conventional IADT is population-based, ignoring the heterogeneity of patients and cancer. Additionally, existing IADT relies on pre-determined thresholds of prostate-specific antigen to pause and resume treatment, which is not optimized for individual patients. To address these challenges, we framed a data-driven method in two steps. First, we developed a time-varied, mixed-effect and generative Lotka-Volterra (tM-GLV) model to account for the heterogeneity of the evolution mechanism and the pharmacokinetics of two ADT drugs Cyproterone acetate and Leuprolide acetate for individual patients. Then, we proposed a reinforcement-learning-enabled individualized IADT framework, namely, I$^{2}$ADT, to learn the patient-specific tumor dynamics and derive the optimal drug administration policy. Experiments with clinical trial data demonstrated that the proposed I$^{2}$ADT can significantly prolong the time to progression of prostate cancer patients with reduced cumulative drug dosage. We further validated the efficacy of the proposed methods with a recent pilot clinical trial data. Moreover, the adaptability of I$^{2}$ADT makes it a promising tool for other cancers with the availability of clinical data, where treatment regimens might need to be individualized based on patient characteristics and disease dynamics. Our research elucidates the application of deep reinforcement learning to identify personalized adaptive cancer therapy.

Small extracellular vesicles from surviving cancer cells as multiparametric monitoring tools of measurable residual disease and therapeutic efficiency.

Although conventional anti-cancer therapies remove most cells of the tumor mass, small surviving populations may evolve adaptive resistance strategies, which lead to treatment failure. The size of the resistant population initially may not reach the threshold of clinical detection (designated as measurable residual disease/MRD) thus, its investigation requires highly sensitive and specific methods. Here, we discuss that the specific molecular fingerprint of tumor-derived small extracellular vesicles (sEVs) is suitable for longitudinal monitoring of MRD. Furthermore, we present a concept that exploiting the multiparametric nature of sEVs may help early detection of recurrence and the design of dynamic, evolution-adjusted treatments.

Experimental and phylogenetic evidence for correlated gene expression evolution in endometrial and skin fibroblasts.

Gene expression change is a dominant mode of evolution. Mutations, however, can affect gene expression in multiple cell types. Therefore, gene expression evolution in one cell type can lead to similar gene expression changes in another cell type. Here, we test this hypothesis by investigating dermal skin fibroblasts (SFs) and uterine endometrial stromal fibroblasts (ESFs). The comparative dataset consists of transcriptomes from cultured SF and ESF of nine mammalian species. We find that evolutionary changes in gene expression in SF and ESF are highly correlated. The experimental dataset derives from a SCID mouse strain selected for slow cancer growth leading to substantial gene expression changes in SFs. We compared the gene expression profiles of SF with that of ESF and found a significant correlation between them. We discuss the implications of these findings for the evolutionary correlation between placental invasiveness and vulnerability to metastatic cancer.

Testing Adaptive Therapy Protocols Using Gemcitabine and Capecitabine in a Preclinical Model of Endocrine-Resistant Breast Cancer.

Adaptive therapy, an ecologically inspired approach to cancer treatment, aims to overcome resistance and reduce toxicity by leveraging competitive interactions between drug-sensitive and drug-resistant subclones, prioritizing patient survival and quality of life instead of killing the maximum number of cancer cells. In preparation for a clinical trial, we used endocrine-resistant MCF7 breast cancer to stimulate second-line therapy and tested adaptive therapy using capecitabine, gemcitabine, or their combination in a mouse xenograft model. Dose modulation adaptive therapy with capecitabine alone increased survival time relative to MTD but not statistically significantly (HR = 0.22, 95% CI = 0.043-1.1, p = 0.065). However, when we alternated the drugs in both dose modulation (HR = 0.11, 95% CI = 0.024-0.55, p = 0.007) and intermittent adaptive therapies, the survival time was significantly increased compared to high-dose combination therapy (HR = 0.07, 95% CI = 0.013-0.42, p = 0.003). Overall, the survival time increased with reduced dose for both single drugs (p < 0.01) and combined drugs (p < 0.001), resulting in tumors with fewer proliferation cells (p = 0.0026) and more apoptotic cells (p = 0.045) compared to high-dose therapy. Adaptive therapy favors slower-growing tumors and shows promise in two-drug alternating regimens instead of being combined.

Immunomodulatory and pro-oncologic effects of ketamine and isoflurane anesthetics in a murine model.

INTRODUCTION: Volatile and intravenous anesthetics may worsen oncologic outcomes in basic science animal models. These effects may be related to suppressed innate and adaptive immunity, decreased immunosurveillance, and disrupted cellular signaling. We hypothesized that anesthetics would promote lung tumor growth via altered immune function in a murine model and tested this using an immunological control group of immunodeficient mice. METHODS: Lewis lung carcinoma cells were injected via tail vein into C57BL/6 immunocompetent and NSG immunodeficient mice during exposure to isoflurane and ketamine versus controls without anesthesia. Mice were imaged on days 0, 3, 10, and 14 post-tumor cell injection. On day 14, mice were euthanized and organs fixed for metastasis quantification and immunohistochemistry staining. We compared growth of tumors measured from bioluminescent imaging and tumor metastasis in ex vivo bioluminescent imaging of lung and liver. RESULTS: Metastases were significantly greater for immunocompromised NSG mice than immunocompetent C57BL/6 mice over the 14-day experiment (partial η2 = 0.67, 95% CI = 0.54, 0.76). Among immunocompetent mice, metastases were greatest for mice receiving ketamine, intermediate for those receiving isoflurane, and least for control mice (partial η2 = 0.88, 95% CI = 0.82, 0.91). In immunocompetent mice, significantly decreased T lymphocyte (partial η2 = 0.83, 95% CI = 0.29, 0.93) and monocyte (partial η2 = 0.90, 95% CI = 0.52, 0.96) infiltration was observed in anesthetic-treated mice versus controls. CONCLUSIONS: The immune system appears central to the pro-metastatic effects of isoflurane and ketamine in a murine model, with decreased T lymphocytes and monocytes likely playing a role.

Testing Adaptive Therapy Protocols using Gemcitabine and Capecitabine on a Mouse Model of Endocrine-Resistant Breast Cancer.

Highly effective cancer therapies often face limitations due to acquired resistance and toxicity. Adaptive therapy, an ecologically inspired approach, seeks to control therapeutic resistance and minimize toxicity by leveraging competitive interactions between drug-sensitive and drug-resistant subclones, prioritizing patient survival and quality of life over maximum cell kill. In preparation for a clinical trial in breast cancer, we used large populations of MCF7 cells to rapidly generate endocrine-resistance breast cancer cell line. We then mimicked second line therapy in ER+ breast cancers by treating the endocrine-resistant MCF7 cells in a mouse xenograft model to test adaptive therapy with capecitabine, gemcitabine, or the combination of those two drugs. Dose-modulation adaptive therapy with capecitabine alone increased survival time relative to MTD, but not statistically significant (HR: 0.22, 95% CI 0.043- 1.1 P = 0.065). However, when we alternated the drugs in both dose modulation (HR = 0.11, 95% CI: 0.024 - 0.55, P = 0.007) and intermittent adaptive therapies significantly increased survival time compared to high dose combination therapy (HR = 0.07, 95% CI: 0.013 - 0.42; P = 0.003). Overall, survival time increased with reduced dose for both single drugs (P < 0.01) and combined drugs (P < 0.001). Adaptive therapy protocols resulted in tumors with lower proportions of proliferating cells (P = 0.0026) and more apoptotic cells (P = 0.045). The results show that Adaptive therapy outperforms high-dose therapy in controlling endocrine-resistant breast cancer, favoring slower-growing tumors, and showing promise in two-drug alternating regimens.

Converting genetic information to non-equilibrium cellular thermodynamics in enzyme-catalyzed reactions.

Living systems use genomic information to maintain a stable highly ordered state far from thermodynamic equilibrium but the specific mechanisms and general principles governing the interface of genetics and thermodynamics has not been extensively investigated. Genetic information is quantified in unitless bits termed "Shannon entropy", which does not directly relate to thermodynamic entropy or energy. Thus, it is unclear how the Shannon entropy of genetic information is converted into thermodynamic work necessary to maintain the non-equilibrium state of living systems. Here we investigate the interface of genetic information and cellular thermodynamics in enzymatic acceleration of a chemical reaction S+E→ES→E+P, where S and E are substrate and enzyme, ES is the enzyme substrate complex and P product. The rate of any intracellular chemical reaction is determined by probability functions at macroscopic (Boltzmann distribution of the reactant kinetic energies governed by temperature) or microscopic (overlap of reactant quantum wave functions) scales - described, respectively, by the Arrhenius and Knudsen equations. That is, the reaction rate, in the absence of a catalyst, is governed by temperature which determines the kinetic energy of the interacting molecules. Genetic information can act upon a when the encoded string of amino acids folds into a 3-deminsional structure that permits a lock/key spatial matching with the reactants. By optimally superposing the reactants' wave functions, the information in the enzyme increases the reaction rate by up to15 orders of magnitude under isothermal conditions. In turn, the accelerated reaction rate alters the intracellular thermodynamics environment as the products are at lower Gibbs free energy which permits thermodynamic work Wmax=-ΔG. Mathematically and biologically, the critical event that allows genetic information to produce thermodynamic work is the folding of the amino acid string specified by the gene into a 3-dimensional shape determined by its lowest energy state. Biologically, this allows the amino acid string to bind substrate and place them in an optimal spatial orientation. These key-lock are mathematically characterized by Kullback-Leibler Divergence and the interactions with the reaction channel now represent Fisher Information (the second derivative Kullback-Leibler divergence), which can take on the units of the process to which it is applied. Interestingly, Shannon is typically derived by "coarse graining" Shannon information. Thus, living system, by acting at a quantum level, "fine grain" Shannon information.

Logistic tumor-population growth and ghost-points symmetry.

The observed time evolution of a population is well approximated by a logistic function in many research fields, including oncology, ecology, chemistry, demography, economy, linguistics, and artificial neural networks. Initial growth is exponential at a constant rate and capped at a limit size, i.e., the carrying capacity. In mathematical oncology, the carrying capacity has been postulated to be co-evolving and thus patient-specific. As the relative tumor-over-carrying capacity ratio may be predictive and prognostic for tumor growth and treatment response dynamics, it is paramount to estimate it from limited clinical data. We show that exploiting the logistic function's rotation symmetry can help estimate the population's growth rate and carry capacity from fewer data points than conventional regression approaches. We test this novel approach against a classic oncology database of logistic tumor growth, achieving a 30% to 40% reduction in the time necessary to correctly estimate the logistic growth rate and carrying capacity. Our results will improve tumor dynamics forecasting and augment the clinical decision-making process.

Updating the Definition of Cancer.

Most definitions of cancer broadly conform to the current NCI definition: "Cancer is a disease in which some of the body's cells grow uncontrollably and spread to other parts of the body." These definitions tend to describe what cancer "looks like" or "does" but do not describe what cancer "is" or "has become." While reflecting past insights, current definitions have not kept pace with the understanding that the cancer cell is itself transformed and evolving. We propose a revised definition of cancer: Cancer is a disease of uncontrolled proliferation by transformed cells subject to evolution by natural selection. We believe this definition captures the essence of the majority of previous and current definitions. To the simplest definition of cancer as a disease of uncontrolled proliferation of cells, our definition adds in the adjective "transformed" to capture the many tumorigenic processes that cancer cells adopt to metastasize. To the concept of uncontrolled proliferation of transformed cells, our proposed definition then adds "subject to evolution by natural selection." The subject to evolution by natural selection modernizes the definition to include the genetic and epigenetic changes that accumulate within a population of cancer cells that lead to the lethal phenotype. Cancer is a disease of uncontrolled proliferation by transformed cells subject to evolution by natural selection.

The paradox of cooperation among selfish cancer cells.

It is traditionally assumed that during cancer development, tumor cells abort their initially cooperative behavior (i.e., cheat) in favor of evolutionary strategies designed solely to enhance their own fitness (i.e., a "selfish" life style) at the expense of that of the multicellular organism. However, the growth and progress of solid tumors can also involve cooperation among these presumed selfish cells (which, by definition, should be noncooperative) and with stromal cells. The ultimate and proximate reasons behind this paradox are not fully understood. Here, in the light of current theories on the evolution of cooperation, we discuss the possible evolutionary mechanisms that could explain the apparent cooperative behaviors among selfish malignant cells. In addition to the most classical explanations for cooperation in cancer and in general (by-product mutualism, kin selection, direct reciprocity, indirect reciprocity, network reciprocity, group selection), we propose the idea that "greenbeard" effects are relevant to explaining some cooperative behaviors in cancer. Also, we discuss the possibility that malignant cooperative cells express or co-opt cooperative traits normally expressed by healthy cells. We provide examples where considerations of these processes could help understand tumorigenesis and metastasis and argue that this framework provides novel insights into cancer biology and potential strategies for cancer prevention and treatment.

Intermetastatic and Intrametastatic Heterogeneity Shapes Adaptive Therapy Cycling Dynamics.

Adaptive therapies that alternate between drug applications and drug-free vacations can exploit competition between sensitive and resistant cells to maximize the time to progression. However, optimal dosing schedules depend on the properties of metastases, which are often not directly measurable in clinical practice. Here, we proposed a framework for estimating features of metastases through tumor response dynamics during the first adaptive therapy treatment cycle. Longitudinal prostate-specific antigen (PSA) levels in 16 patients with metastatic castration-resistant prostate cancer undergoing adaptive androgen deprivation treatment were analyzed to investigate relationships between cycle dynamics and clinical variables such as Gleason score, the change in the number of metastases over a cycle, and the total number of cycles over the course of treatment. The first cycle of adaptive therapy, which consists of a response period (applying therapy until 50% PSA reduction), and a regrowth period (removing treatment until reaching initial PSA levels), delineated several features of the computational metastatic system: larger metastases had longer cycles; a higher proportion of drug-resistant cells slowed the cycles; and a faster cell turnover rate sped up drug response time and slowed regrowth time. The number of metastases did not affect cycle times, as response dynamics were dominated by the largest tumors rather than the aggregate. In addition, systems with higher intermetastasis heterogeneity responded better to continuous therapy and correlated with dynamics from patients with high or low Gleason scores. Conversely, systems with higher intrametastasis heterogeneity responded better to adaptive therapy and correlated with dynamics from patients with intermediate Gleason scores. SIGNIFICANCE: Multiscale mathematical modeling combined with biomarker dynamics during adaptive therapy helps identify underlying features of metastatic cancer to inform treatment decisions.

Adaptive therapy for ovarian cancer: An integrated approach to PARP inhibitor scheduling.

Toxicity and emerging drug resistance are important challenges in PARP inhibitor (PARPi) treatment of ovarian cancer. Recent research has shown that evolutionary-inspired treatment algorithms which adapt treatment to the tumor's treatment response (adaptive therapy) can help to mitigate both. Here, we present a first step in developing an adaptive therapy protocol for PARPi treatment by combining mathematical modelling and wet-lab experiments to characterize the cell population dynamics under different PARPi schedules. Using data from in vitro Incucyte Zoom time-lapse microscopy experiments and a step-wise model selection process we derive a calibrated and validated ordinary differential equation model, which we then use to test different plausible adaptive treatment schedules. Our model can accurately predict the in vitro treatment dynamics, even to new schedules, and suggests that treatment modifications need to be carefully timed, or one risks losing control over tumour growth, even in the absence of any resistance. This is because our model predicts that multiple rounds of cell division are required for cells to acquire sufficient DNA damage to induce apoptosis. As a result, adaptive therapy algorithms that modulate treatment but never completely withdraw it are predicted to perform better in this setting than strategies based on treatment interruptions. Pilot experiments in vivo confirm this conclusion. Overall, this study contributes to a better understanding of the impact of scheduling on treatment outcome for PARPis and showcases some of the challenges involved in developing adaptive therapies for new treatment settings.

Elevated Methionine Flux Drives Pyroptosis Evasion in Persister Cancer Cells.

Induction of cell death represents a primary goal of most anticancer treatments. Despite the efficacy of such approaches, a small population of "persisters" develop evasion strategies to therapy-induced cell death. While previous studies have identified mechanisms of resistance to apoptosis, the mechanisms by which persisters dampen other forms of cell death, such as pyroptosis, remain to be elucidated. Pyroptosis is a form of inflammatory cell death that involves formation of membrane pores, ion gradient imbalance, water inflow, and membrane rupture. Herein, we investigate mechanisms by which cancer persisters resist pyroptosis, survive, then proliferate in the presence of tyrosine kinase inhibitors (TKI). Lung, prostate, and esophageal cancer persister cells remaining after treatments exhibited several hallmarks indicative of pyroptosis resistance. The inflammatory attributes of persisters included chronic activation of inflammasome, STING, and type I interferons. Comprehensive metabolomic characterization uncovered that TKI-induced pyroptotic persisters display high methionine consumption and excessive taurine production. Elevated methionine flux or exogenous taurine preserved plasma membrane integrity via osmolyte-mediated effects. Increased dependency on methionine flux decreased the level of one carbon metabolism intermediate S-(5'-adenosyl)-L-homocysteine, a determinant of cell methylation capacity. The consequent increase in methylation potential induced DNA hypermethylation of genes regulating metal ion balance and intrinsic immune response. This enabled thwarting TKI resistance by using the hypomethylating agent decitabine. In summary, the evolution of resistance to pyroptosis can occur via a stepwise process of physical acclimation and epigenetic changes without existing or recurrent mutations. SIGNIFICANCE: Methionine enables cancer cells to persist by evading pyroptotic osmotic lysis, which leads to genome-wide hypermethylation that allows persisters to gain proliferative advantages.

Back to basic: Trials and tribulations of alkalizing agents in cancer.

"Dysregulated" metabolism is a characteristic of the cancer cell phenotype. This includes persistent use of glycolytic metabolism in normoxic environments (Warburg effect) leading to increased acid production and accumulation of protons in the interstitial space. Although often thought to be disordered, altered cancer metabolism is the outcome of intense Darwinian selection and, thus, must have evolved to maximize cancer cell fitness. In an evolutionary context, cancer-induced acidification of the microenvironment represents a niche construction strategy to promote proliferation. Ecological advantages conferred on the cancer population included remodeling of the extracellular matrix to promote local invasion, suppression of potential competitive proliferation of fibroblasts, and suppression of host immune response. Preclinical data demonstrates that increasing the serum buffering capacity (through, for example, oral sodium bicarbonate and TRIS) can neutralize the acidic tumor microenvironment with inhibition local invasion and proliferation which can be synergistic with the effects of chemotherapy and immunotherapy agents. Here, we describe the proton dynamics in cancer and their influence on tumor progression and metastasis. Additionally, we will discuss targeting the tumor acidosis with alkalizing agents including our bicarbonate clinical trial results. Clinical Trial Registration: clinicaltrials.gov, identifier NCT01350583, NCT01198821 and NCT01846429.

A Phase 1b Adaptive Androgen Deprivation Therapy Trial in Metastatic Castration Sensitive Prostate Cancer.

Background: We hypothesize that cancer survival can be improved through adapting treatment strategies to cancer evolutionary dynamics and conducted a phase 1b study in metastatic castration sensitive prostate cancer (mCSPC). Methods: Men with asymptomatic mCSPC were enrolled and proceeded with a treatment break after achieving > 75% PSA decline with LHRH analog plus an NHA. ADT was restarted at the time of PSA or radiographic progression and held again after achieving >50% PSA decline. This on-off cycling of ADT continued until on treatment imaging progression. Results: At data cut off in August 2022, only 2 of the 16 evaluable patients were off study due to imaging progression at 28 months from first dose of LHRH analog for mCSPC. Two additional patients showed PSA progression at 12.4 and 20.5 months and remain on trial. Since none of the 16 patients developed imaging progression at 12 months, the study succeeded in its primary objective of feasibility. The secondary endpoints of median time to PSA progression and median time to radiographic progression have not been reached at a median follow up of 26 months. Conclusions: It is feasible to use an individual's PSA response and testosterone levels to guide intermittent ADT in mCSPC.