PERSPECTIVES

COMMENTARIES

inform IO combination studies. On the Designing In and Around one extreme, has a consider- able history of use in , now with an accompanying rich database to mine Tolerability Considerations for regarding the effect of dose and scheduling on tolerability. In contrast, the clinical Immunotherapy Combinations database for IO agents is growing only just 1 now for an increasing list of indications. Mark Stroh The robustness of model-based predictions for IO combinations can be limited by the Over a century ago, paths diverged in the treatment of cancer: lack of available data, although systems the well-traveled path employed cytotoxic chemotherapy , approaches are beginning to show promise even with this limitation.3 while one of the roads less traveled included immunotherapies. Clinical investigation of an IO combina- Cancer immunotherapy is now a path to durable responses, how- tion, often with one combination partner ever not all patients benefit. Immunotherapy combinations prom- held at the approved dose and dose escala- ise responses for a larger proportion of patients, but tolerability tion for the investigational agent, can reveal information about the maximally tolerated can prove to be a barrier. Providing deep, durable responses to dose (MTD) for that combination. How- more patients requires us to successfully navigate emerging ever, especially in the early phase I setting, combination tolerability issues. we explore only a fraction of the possible dosing and scheduling possibilities for a given combination, and are left with only a partial understanding of the underlying With the recent clinical and commercial individual steps represent possible points of exposure–tolerability relationship for the success of checkpoint inhibitors, especially therapeutic intervention, and are affected combination (Figure 1). A further compli- those targeting CTLA4 and PD-L1/PD-1, not only by IO agents, but by a myriad of cating factor arises from the fact that has come a seemingly disproportionate other classes of anticancer drugs, spanning immune-related adverse events (irAE) can increase in clinical investigations in cytotoxic to targeted emerge following the first cycle of treatment immuno-oncology (IO), now numbering agents. For example, cytotoxic agents act with IO agents, and that management of over 1,000 trials.1 Immunotherapy combi- directly on cancer cell killing (the final step theseAEsmayinvolveholdingsubsequent nation trials comprise a substantial propor- of the cancer immunity cycle), while anti- doses and, consequently, reduced overall tion of this unprecedented investment, vascular endothelial growth factor (VEGF) dose intensity; especially given the unique suggesting that immunotherapies could can affect T-cell extravasation occurring in kinetics of irAE onset and resolution, careful become backbone across a wide the middle of the cycle. Accordingly, IO– selection of safety endpoints is important to meaningfully inform the exposure– array of malignancies. The so-called “cancer IO agent combinations represent only a 4 immunity cycle” partitions IO action into small fraction of the possibilities. Explora- tolerability analysis for IO. In instances where the IO combination is poorly tol- several conceptual steps and provides a tion of this expansive IO combination 2 erated, we are faced with the prospect of concise rationale for IO combinations. landscape will continue to act as a driver designing a combination treatment regi- The cancer immunity cycle commences for large numbers of clinical investigations men around the peaks and valleys of this first with release of cancer cell antigens and well into the future. underlying relationship with only a lim- proceeds through a series of steps that are The available clinical database across the ited understanding of the topography, critical to generating immunity to cancer possible combination partners for IO while maintaining sufficient levels and ultimately killing cancer cells. These agents can prove to be inadequate to fully required for efficacy.

1CytomX Therapeutics Inc., South San Francisco, California, USA. Correspondence: Mark Stroh ([email protected]) doi:10.1002/cpt.945

558 VOLUME 103 NUMBER 4 | APRIL 2018 | www.cpt-journal.com PERSPECTIVES

the diseased site. An early example of direct administration of IO agents comes from the first use of intravesical Bacillus Calmette– Guerin (BCG) immunotherapy in bladder cancer patients in 1976.8 Intravesical BCG immunotherapy is now indicated for carci- noma in situ of the bladder, and is under investigation in combination with other agents, including the PD-L1 inhibitor atezo- lizumab (ClinicalTrials.gov Identifier: NCT02792192) which is administered intravenously. Additional recent investiga- tions of intratumoral IO combinations also include IPI administered with interleukin-2 (IL-2) in advanced melanoma.9 However, unlike these previous examples, not all Figure 1 Hypothetical tolerability of a given immuno-oncology (IO) combination as a function of tumors are as accessible for direct adminis- exposure (e.g., area under the curve, maximum concentration, etc.) of combination partners 1 tration, and local administration to tumors and Drug 2 (upper blue surface). Red shaded areas represent sampling of this underlying relation- ship in a phase I combination study. Since this underlying relationship is not necessarily known from does not preclude at least some leakage back monotherapy development of the combination partners, identification of a tolerated dose and sched- to the systemic circulation. ule can require an extended exploration of this underlying relationship across multiple trials. Restrict- Given the strengths and limitations that ing the action of IO combination agents to the tumor holds promise to flatten this relationship to come with direct administration, a comple- resemble the cyan surface, resulting in improved tolerability of IO combination by design. mentary approach is to consider safer sys- temic IO combinations. Several approaches Development of the ipilimumab (IPI) 1 and NIVO was administered q2w and dose are under development to better restrict nivolumab (NIVO) combination provides escalated in a 313 design, the current clini- IO combination action to the tumor fol- an important case example of enhancing tol- cal database now includes both the 1 mg/kg lowing systemic administration. One exam- erability through exploration of scheduling and 3 mg/kg dose levels for both IPI and ple comes from combination of IO with a and dosing of drugs in the combination. In NIVO and a collection of schedules for IPI nanoparticle paclitaxel (PTX) formulation. the phase III investigation CheckMate 067, spanning q3wx4 to every 12 week (q12w) As mentioned previously, chemotherapy represents a promising combination part- melanoma patients received IPI 3 mg/kg administration. A subset of this clinical data- ner with checkpoint inhibitors by generat- and NIVO 1 mg/kg every 3 weeks for base has been pooled to inform exposure– ing antigens following cancer cell killing. 4 doses (q3wx4) followed by NIVO alone response (E-R) analyses and support the dos- However, working against this possible syn- 3 mg/kg q2w; patients were further ran- ing regimen for this combination in domized to receive NIVO alone 3 mg/kg 7 ergy is the hallmark immunosuppression melanoma. that comes with high-dose chemotherapy. q2w and IPI alone 3 mg/kg q3wx4. Patients 1 In the example of the IPI NIVO com- Albumin-conjugated paclitaxel (nab-PTX) receiving IPI1NIVO had improved survival bination, we note the substantial effort benefit relative to IPI alone; however, the provides a possible avenue for reducing this that can be required to design around pos- risk. Preclinical data suggest the distribu- frequency of Grade 3-4 AEs was 59% for sible safety considerations with IO combi- tion of PTX is altered following nab-PTX the combination vs. 21% and 28% for nations. Leveraging historical monotherapy administration relative to following admin- NIVO and IPI alone, respectively, with gas- data may not fully inform dosing in combi- istration of PTX (as Taxol), with elevated trointestinal events as the most common nation, and sampling the underlying dose- PTX levels in the tumor for nab-PTX. Grade 3-4 select adverse event. Further, 39% schedule-tolerability relationship for IO Population pharmacokinetic analysis of 1 of patients receiving IPI NIVO experi- combinations in the clinic can involve mul- nab-PTX suggests faster distribution and enced adverse events of any grade leading to tiple trials to yield sufficient cross-sections 5 slower elimination, respectively, of PTX discontinuation. Safety guidelines were fol- of this underlying relationship. This begs following nab-PTX vs. PTX administra- lowed to manage immune-mediated AEs in the question of how we can design better tion in cancer patients, and that albumin part through the use of immune-modulating tolerability into combinations. levels are a significant covariate on PTX agents. Multiple dose levels and schedules of Restricting the action of the IO and/or ; it is unclear if this effect would IPI and NIVO have been explored in combination agent to the tumor provides impact relative levels of intratumoral PTX addition to the IPI1NIVO regimen of one route for greater tolerability by design, in patients with hypoalbuminemia.10 Clini- CheckMate 067 as summarized elsewhere.6 since poor tolerability can arise from unnec- cal data further demonstrate reduced rates Starting first with the initial phase Ib investi- essary systemic exposure to a drug. Perhaps of neutropenia for nab-PTX, and nab- gation where IPI was initially fixed to the the most conceptually simple means to this PTX does not require steroid pretreatment approved 3 mg/kg q3wx4 dose and schedule end comes from direct administration to due to PTX-associated hypersensitivity. In

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 103 NUMBER 4 | APRIL 2018 559 PERSPECTIVES

(A) from Probody therapeutics (Pb-Txs): mono- clonal antibody prodrugs that are designed to activate preferentially in the tumor. A mask, which inhibits binding of the Pb-Tx in the healthy peripheral tissues, can be removed by tumor-associated proteases to release an active antibody. Nonclinical data and quantitative systems pharmacology mod- els alike suggest the distribution of Pb-Tx in the tumor and the systemic compartment is tunable with mask strength (Figure 2), lead- ing to an enhanced .12 A Pb-Tx directed against PD-L1 is now in phase I investigation (ClinicalTrials.gov Identifier NCT03013491) both in mono- (B) and combination, with several other IO Pb-Txs under development. Ex vivo

Parental mAb results in selected human tumor samples sug- 102 PbTx 1 gest no correlation of protease activity with PbTx 2 stage or grade of disease12 and that protease activity can be detected in most patients

100 (unpublished data). The emerging clinical Increasing

Plasma Drug (nM) dataset will permit identification of patient mask strength populations for further investigation, as well 0 5 10 15 20 25 30 as evaluation of any significant covariates on Time (Days) Pb-Tx and E-R. (C) In summary, as the era of IO monother- apy passes an inflection point in cancer 4 10 treatment, we enter a new era of IO combi- Increasing mask nations in our campaign to provide durable strength responses to more patients. This era of IO 102 combinations represents the intersection of multiple paths, some well-traveled and 100 some new, toward the ultimate goal of a

Tumor concentration(nM) cure. As responses become more durable 0 102030405060 for increasing numbers of patients, the tol- Time (Days) erability of IO combinations becomes ever Figure 2 (a) A Probody therapeutic (Pb-Tx) is a prodrug form of a monoclonal antibody (mAb), and is more important as we increasingly manage comprised of a parental mAb and a prodomain. The prodomain is comprised of a mask that inhibits cancer as a chronic disease. Through a binding to antigen, and a protease-cleavable substrate between the mask and the light chain of the comprehensive characterization of the dose mAb. The mask inhibits antigen binding in the periphery, and can be removed by tumor-associated proteases. (b) A quantitative systems pharmacology (QSP) model was developed and calibrated and schedule dependency of combination against monkey pharmacokinetic (PK) data using Pb-Txs directed against CD166, which is broadly tolerability, we can inform regimens for IO expressed in the tumor and periphery. The model predictions (solid lines) adequately described combination that minimize ; by observed monkey PK data (points) following a single dose of Pb-Txs having increasing mask strengths, and captured decreasing Pb-Tx systemic clearance with increasing mask strength, which restricting the action of IO combination is consistent with avoiding peripheral target. (c) Human projections suggested successively higher agents to the tumor, we can possibly levels of Pb-Tx in tumor following a single dose of Pb-Txs having increasing mask strengths at the improve tolerability by design. same dose level. ACKNOWLEDGMENTS We thank Applied BioMath for their scientific this way, tolerability is designed into the Just as with the nab-PTX example, where and technical contributions to the Pb-Tx QSP regimen, with possibly enhanced ability to we attempt to design in tolerability by model which was supported by CytomX Thera- peutics Inc. We also thank Sean A. McCarthy, solicit cancer cell kill at a reduced risk of controlling distribution of the combination 11 DPhil, Carol Verser, PhD, Rachel Humphrey, immunosuppression. Early clinical data partner, emerging technologies enable us to MD, Jason Sagert, PhD, and Michael Kava- of nab-PTX in combination with atezoli- modulate the distribution and action of the naugh, MD for helpful contributions and zumab are limited but encouraging. IO agent(s) as well. One example comes discussion.

560 VOLUME 103 NUMBER 4 | APRIL 2018 | www.cpt-journal.com PERSPECTIVES

AUTHOR CONTRIBUTIONS and CTLA-4. Cancer Res. 77(13 suppl.), 8. Fuge, O., Vasdev, N., Allchorne, P. & M.S. wrote the article. 4531 (2017). Green, J.S. Immunotherapy for bladder 4. Postel-Vinay, S. et al. Challenges of phase cancer. Res. Rep. Urol. 7, 65–79 CONFLICT OF INTEREST 1 clinical trials evaluating immune (2015). M.S. is an employee of CytomX Therapeutics, checkpoint-targeted antibodies. Ann. 9. Ray, A. et al. A phase I study of Oncol. 27, 214–224 (2016). intratumoral ipilimumab and interleukin-2 Inc., and owns CytomX stock. 5. Wolchok, J.D. et al. Overall survival with in patients with advanced melanoma. combined nivolumab and ipilimumab in Oncotarget 7, 64390–64399 (2016). VC 2017 ASCPT advanced melanoma. N. Engl. J. Med. 377, 10. Chen, N. et al. Pharmacokinetics and 1345–1356 (2017). of nab-paclitaxel 1. Cavnar, S., Valencia, P., Brock, J., 6. Roy, A. BMS experience in I-O combination in patients with solid tumors: disposition Wallenstein, J. & Panier, V. The immuno- dose selection: IPI 1 NIVO. FDA-AACR kinetics and pharmacology distinct from oncology race: myths and emerging Regulatory Science and Policy Workshop; solvent-based paclitaxel. J. Clin. realities. Nat. Rev. Drug Discov. 16, 83–84 Oncology Dose Finding Workshop Part III; Pharmacol. 54, 1097–1107 (2014). (2017). 2017; Washington, DC. 11. Soliman, H.H. nab-Paclitaxel as a potential 2. Chen, D.S. & Mellman, I. Oncology meets 7. Wang, X., Feng, Y., Statkevich, P. & Roy, A. partner with checkpoint inhibitors in solid : the cancer-immunity cycle. Characterizing exposure-response (E-R) tumors. Onco. Targets Ther. 10, 101–112 Immunity 39, 1–10 (2013). relationship of safety for nivolumab in (2017). 3. Milberg, O. et al. Abstract 4531: Systems combination with ipilimumab in patients 12. Desnoyers, L.R. et al. Tumor-specific pharmacology to predict cellular with previously untreated advanced activation of an EGFR-targeting probody biomarkers and optimize mono- and melanoma. J. Pharmacokinet. enhances therapeutic index. Sci. Transl. combination-therapy regimens: focusing on Pharmacodyn. 42, S37–S38 (2015). Med. 5, 207ra144 (2013). immune checkpoint targets PD-1, PD-L1

This Commentary addresses some of the Psychedelic Drugs as regulatory considerations and potential Therapeutics: No Illusions precedents. PRACTICAL BARRIERS About the Challenges Pragmatic barriers include: lack of intellec- tual property protection for composition Edward M. Sellers1 and Deborah B. Leiderman2 of matter for old drugs; developments costs in the hundreds of millions of dollars; diffi- culty of designing and conducting double- Interest in the potential therapeutic benefits of psychedelic blind trials with drugs that have easily agents has recently increased. In addition to psilocybin, a wide detected effects; unique clinical trial con- variety of agents with psychedelic properties have been proposed straints (e.g., limited appropriate treatment and partially tested. However, the challenges of obtaining settings, small number of experienced investigators); need for a site license to approval to market a restricted psychotomimetic agent are possess a controlled substance; and the formidable. perceived risk of such a product from a pharmaceutical company perspective. In addition, since for most potential therapeu- Recent clinical trials with psilocybin for entirely descriptive, attempts at improved tic applications a variety of effective drugs intractable anxiety and depression in patients exploratory studies (e.g., obsessive-compulsive exist, it is not obvious where drugs with a with life-threatening cancer, a Review article disorder, major depression, substance use dis- risk of behavioral toxicity would fit. and Commentary in this journal, have orders) have occurred in the past 20 years. renewed interest in the potential therapeutic While introducing elements of randomiza- PHARMACOLOGY AND MECHANISM usefulness of psychedelic agents.1–5 Medi- tion, blinding, and a control arm, these are OF ACTION cally, in addition to psilocybin, other generally not placebo-controlled studies. Psychedelic drugs are not identical in their substances with psychedelic properties, e.g., Research to understand the mechanism of mechanisms of action. Their basic pharma- ketamine, 3,4methylenedioxymethamphet- action of psychedelic agents as neurochemical cology is diverse and pleomorphic, includ- amine (MDMA), microdoses of lysergic probes is of scientific interest. The possibility ing complex and / acid diethylamide (LSD), and N, N- of resetting even part of the brain’s default antagonist actions on 5HT2A, 5HT2C, dimethyltryptamine (DMT), have been pro- network would have wide scientific, medical, 5HT1A, dopamine D2, trace amine associ- posed to treat psychiatric and other diseases. ethical, and social implications.4 The ate receptors 1, various transporters (e.g., Whilemostpasthallucinogenresearchwas obstacles to drug approval remain daunting. serotonergic, dopaminergic), intracellular

1Departments of Pharmacology and , and , University of Toronto Toronto, ON, Canada; 2CNS Drug Consulting LLC, McLean, Virginia, USA. Correspondence: E. Sellers ([email protected]) doi:10.1002/cpt.776

CLINICAL PHARMACOLOGY & THERAPEUTICS | VOLUME 103 NUMBER 4 | APRIL 2018 561