Clinical Trials Innovation J u n e 2 0 1 6
sponsored by
PharmTech.com Cover image: Getty images/artvea TOC Table of contents I Clini nn 12 25 6 3 o c va al Trials Trials al t By Moe Alsumidaie Model? Compatible More A What’s Centricity: Patient TrialClinical S Jürgen Hummel, Ashley Simmons By Dirk Reitsma, MD, Austin Combest, A O We I Can By Kenneth A. Getz PoisedDesigns A to A By Dr. Paul Ingram S through I through upply Chain Transformation Chain upply i daptive Designs daptive A of doption ncology Trials Using o n mprove nnovation ite Centricity VS Centricity ite daptive Trialdaptive
ccelerate Supply Chain Transformation through Innovation By Dr. Paul Ingran
Video An important measure of success of any Demand Led Supply clinical supply chain strategy is whether or not clinical sites have the necessary study Click to view medications at the ready when needed video to avoid delays in new patient starts or continuance of therapy. The complex nature of some clinical trials can create Sponsored challenges which can place considerable Managing Change stress on the supply chain. The difficulty in ensuring the continuity of supply Click to view PDF increases as variables are introduced into the mix. These variables can include blinded studies, those with multiple arms or dynamic designs such as adaptive studies, fluctuations in patient recruitment, insufficient or interrupted supply of the IMP or comparator drugs, and more. Forecasting and demand simulations are useful for anticipating when and where stock is likely to be needed to formulate a kit production and
3 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
distribution strategy. Additional stock cause of supply delays and shortages can over and above the forecasted amount be eliminated instead of just managed needed is commonly added to create and the focus can shift from managing a generous buffer inventory to help inventory to managing demand. By manage any deviations and hopefully using a demand led approach the supply avoid supply disruptions or shortages. chain becomes much more flexible and However, even with forecasting and responsive and requires a significantly building out a sizeable buffer inventory, smaller buffer stock. clinical sites may still not always have the appropriate kits on hand to dispense to Stage 1: Advance Preparation of patients. Why? Bright Stock The answer is surprisingly simple. Prior to the start of the study the Nearly all clinical supply models in investigational medical product and any use today are linear and rely upon comparator drugs are processed up to the advance production of finished or the point of primary packaging. These partially-finished patient kits prior to primary packaged drugs are uniquely the study start. Depending upon the numbered but remain unlabeled. And supply strategy employed, large initial because the drugs are not packaged lots of finished kits and periodic resupply into patient kits at this stage, they can be shipments may be sent to clinical sites pooled for use across multiple protocols based upon anticipated demand or if applicable. Once released by Quality, partially-finished kits may receive final they are tracked in a central inventory labeling and be shipped to clinical sites management system as bright stock, but on an as-needed basis. In either scenario, are physically distributed to regional GMP drug product must be committed (e.g., clinical packaging facilities closer to the packed into) a patient kit before it is clinical sites. actually needed. And therein lies the crux of the problem – a linear supply chain Stage 2: Secondary Packaging and lacks the flexibility to effectively address Distribution unforeseen inconsistencies between Once clinical sites indicate the need for a supply and demand. small supply of seed stock or there is an In order to create a supply chain that actual patient demand, the appropriate is flexible and responsive, it must first regional packaging facility picks the be broken. By splitting the supply chain required bright stock from its inventory into two independent stages the root and performs secondary packaging and
4 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
labeling to produce only the type and By interrupting the traditionally linear quantity of patient kits requested. The supply chain at the primary packaging finished kits are released, packed and stage instead of the secondary packaging shipped out to the clinical site in a matter stage, the supply chain under a demand of days. led approach is transformed into one that Not only does the demand led is flexible, efficient and most importantly, approach mean that clinical sites receive able to get study drugs to clinical sites the kits they need quickly, it greatly quickly. reduces the risk of treatment delays due to insufficient inventory.I n addition, slower recruiting clinical sites are not faced with Dr Paul Ingram, is Global Director for the challenge of storing an overabundance Strategic Business Development at Catalent Pharma Solutions. His career has focused on of inventory or potentially returning Medical Device and Pharmaceutical Research or updating the labeling on expiring and Development. Paul obtained his PhD in inventory. Delayed commitment of the Pharmaceutical Development from Strathclyde University, Scotland, and has over 15 years of study drugs greatly reduces the amount experience in clinical supplies, working both of buffer stock needed from upwards of in Phase 1 units and trial supplies service 200% or more common in today’s linear companies including Quintiles (Aptuit), Fisher Clinical and Catalent. supply models to less than 20%.
5 | June 2016 | Pharm Tech Adoption of Adaptive Trial Designs Poised to Accelerate Adaptive trial designs have the potential to transform success rates, but require new operating strategies and practices. By Kenneth A. Getz
Free Webinar Risk-based monitoring, patient centricity, Supply Model Selection and investigative site management reform (e.g., selection, training, and accreditation) Click to launch webinar are widely cited today as critical areas that will transform drug development performance, cost, and inefficiency. Adaptive trial designs are one area that has received scant attention but that holds the near-term potential to have a far more transformational impact.
Adaptive trial designs are preplanned, typically through the use of trial simulations and scenario planning where one or more specified clinical trial design elements are modified and adjusted— while the trial is underway—based on an analysis of interim data. Most transformational approaches receiving attention today aspire to reduce cost and development duration. Adaptive trial designs could potentially improve success rate. With less than 20% of drugs that enter clinical testing ultimately receiving regulatory approval, modest
6 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
increases in success rates have been a clinical trials to increase significantly over coveted yet elusive goal. the next several years. Current adoption low and later stage “Study terminations due to futility” is In late 2012, the Tufts Center for the the most common simple adaptive design Study of Drug Development (Tufts CSDD) used and it is becoming widely adopted conducted a study among 12 major throughout the industry. Sponsor pharmaceutical companies. Based on companies unanimously agreed that early in-depth interviews, Tufts CSDD probed terminations due to futility are relatively current adoption rates and their impact easy to implement and should become on drug development study economics standard practice in Phase II and Phase III and durations. As a follow-up to the studies, across all therapy areas. Although in-depth interviews, in February 2013, companies view sample size re-estimation Tufts CSDD hosted and facilitated a as a relatively simple adaptive design, the roundtable in Boston to gather a more adoption of this approach appears to be comprehensive assessment of industry- well below that of futility stopping. wide adaptive trial design practices and The adoption rate of more perspectives. Forty senior executives sophisticated adaptive design representing 31 companies participated approaches—dose-finding and seamless in the roundtable. Executives represented Phase II/III studies—appears low at less cross-functional viewpoints from clinical than 10% of clinical trials. Despite the research and development, biostatistics, fact that sophisticated adaptive designs project management, and clinical during exploratory phase clinical trials operations. Perspectives from the Food could have the greatest impact on and Drug Administration (FDA) and the improving later phase success rates, European Medicines Agency (EMA) Tufts CSDD found low use of adaptive were also represented in the roundtable dose finding and extremely low use of meeting. This study was funded by an seamless Phase II/IIII studies. unrestricted grant from Aptiv Solutions. The results of this study suggest that Areas of resistance overall, simple adaptive designs are being Although the concept of adaptive trial used on approximately one out of five designs has been discussed and explored (20%) later stage Phase III clinical trials. by pharmaceutical and biotechnology Sponsor companies report, however, that companies for decades, adoption has they expect the adoption of adaptive been slow for a variety of reasons. Tufts trial designs in earlier exploratory phase CSDD interviews revealed that internal
7 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
organizational resistance appears to and inform clinical teams prior to be the primary factor limiting more committing to pivotal late-phase studies. widespread adoption. Regulatory agency The agencies are concerned that early receptivity to the use of adaptive trial development approaches are not designs does not appear to be a barrier efficiently detecting failures prior to Phase to adoption. Agency clarity to sponsor III or that decisions taken in exploratory companies on its position regarding the development are sub-optimal leading use of adaptive designs appears to be to unnecessary Phase III failure. Using lacking. adaptive trial designs to perform Phase II Clinical teams and operating functions dose response assessments, for example, perceive enrollment and logistical could dramatically improve dose selection factors—specifically delays and in Phase III clinical trials. disruptions in trial execution, patient Reported and estimated impact participation, and distribution of clinical In theory, adaptive trial designs may supplies—as major barriers to adoption. help to reduce the number of protocol Sponsor companies express concerns amendments. In a 2012 study conducted around how to monitor data without by Tufts CSDD, we found that sponsor introducing bias; the lack of adaptive organizations spend approximately half- trial design experience among both a-million dollars in direct costs and an internal development teams and external additional 60 days to implement each contract research organizations; gaps protocol amendment. In our interviews on in infrastructure and technology to adaptive trial design adoption, however, implement more sophisticated adaptive several sponsors have indicated that designs; and the limited capacity not all amendments will go away since of independent data monitoring certain country-specific health authorities committees. will still require protocol modifications Senior pharmaceutical company after pre-planned adaptations have been executives perceive that regulatory affairs approved. functions are risk averse to adopting Early study terminations due to futility sophisticated adaptive design approaches and sample size re-estimation could due to their belief that more clarity from save up to a hundred million dollars in regulatory agencies is needed. direct and indirect costs annually for a Regulatory agencies, in contrast, sponsor company depending on clinical appear highly receptive to exploratory trial scope, when the trial is actually phase adaptive trial designs to challenge terminated, and on the sponsor’s overall
8 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
implementation of this adaptive approach functions. Cross-functional education and across the development portfolio. support, aided by senior management A global top 20 drug development encouragement, will go far in increasing company that has been applying simple general awareness and stimulating adaptive trial designs to Phase II and usage. Sponsor organizations that have Phase III studies across its portfolio for established strong proponents among five years reported saving more than $70 therapy area heads, medical experts, million each year through the adoption of and clinical operations professionals in simple adaptive dose-finding studies in addition to their statistical functions— Phase II and sample size adjustments and have been far more effective in futility stopping in Phase III. promoting successful adoption. Sponsor companies recognize that Sponsor organizations higher up the the greatest potential value created by adoption curve did so by starting with adaptive trial designs will come from simple and straightforward adaptations improvements in late-stage success rates. (e.g., sample size re-estimation and futility Even modest improvements in success analyses). Simple designs applied across rates for new molecular entities (NME) the portfolio appear to ease the company and new biologic entities (BME) represent into implementing adaptive designs billions of dollars in increased revenue and to facilitate a smoother transition to potential for research sponsors. sophisticated approaches. Companies In the immediate term, adaptive that have applied sophisticated and trial designs are already offering cross- complicated adaptive designs to a few functional teams direct insights into select studies early in the organization’s study design through scenario planning adoption process have had a lot of and trial simulation prior to finalizing difficulty promoting more widespread the protocol. Rigorous upfront planning usage. is forcing organizations to challenge The adoption of more sophisticated protocol feasibility prior to placing the adaptive designs in exploratory protocol in the clinic. development may necessitate the use of new technology solutions and operating Ramping up adoption practices and will require organizations In most sponsor organizations, awareness to re-evaluate their existing clinical and knowledge about the usage and operating processes. the potential impact of adaptive trial Clinical operations teams will need to designs has been limited to statistical focus particular attention on obtaining
9 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
expertise in planning and forecasting adept at publicizing quantitative measures logistics and resource requirements of the impact of adaptive trial design use. under more flexible study designs. At this Adaptive trial designs represent a time, most sponsors have set very rigid critical opportunity for pharmaceutical performance and cost variance thresholds and biotechnology companies making it challenging for operating staff to transform drug development to entertain more variable capacity and performance, efficiency and—most resource planning practices. Clinical importantly—success rates. Beyond teams will also need to work more closely building awareness, as always, the with investigative site personnel to ensure challenge will be implementing that patient safety and expectations are successful adoption of simple adaptive maintained under more adaptive clinical designs leading to more sophisticated trial conditions. approaches. Active and careful measurement of cost savings from the use of adaptive designs (e.g., futility analysis) is an essential Kenneth A. Getz MBA, is ingredient in convincing organizations— the Director of Sponsored particularly senior management—to Research at the Tufts CSDD and Chairman of CISCRP, continue to support and extend adoption. both in Boston, MA, e-mail: Companies that have led successful [email protected] implementations have been particularly
10 | June 2016 | Pharm Tech
We Can Improve Oncology Trials Using Adaptive Designs By By Dirk Reitsma, MD, Austin Combest, Jürgen Hummel, Ashley Simmons
Free Webinar Today’s rich oncology pipeline— Leveraging Forecasting Models to Optimize Clinical accounting for nearly 25% of agents in Trial Supply Management clinical development—promises much Click to launch webinar needed advances in cancer therapy.1 That promise dims in the face of other discouraging statistics: only 7% of oncology agents entering Phase I clinical trials gain marketing approval2 while only 34% of Phase III oncology trials achieved statistical significance in primary endpoints.3
The cost, time, and numbers of patients required to conduct conventional oncology clinical trials continue to escalate. The complex demands of evaluating new targeted therapies add to this burden. Novel methodologies are available that make trials more efficient and informative so that precious resources of patients, time, and money are invested in studies with the greatest chances of success. Adaptive trial design offers opportunities for improvement by shortening the time needed to
12 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
answer key research questions, reducing in oncology drug development. Its the number of patients needed for positive impact can be seen in the evaluation, and improving the quality groundbreaking I-SPY 2 breast cancer of decision-making to increase overall trial, which uses adaptive design to success rates. The use of adaptive designs streamline identification of active drugs also raised scientific and regulatory and predictive biomarkers.5 I-SPY 2 questions that slowed adoption by the (“Investigation of Serial Studies to Predict biopharmaceutical industry. A growing Your Therapeutic Response with Imaging body of experience culminated in the U.S. and Molecular Analysis”) suggests a Food and Drug Administration’s (FDA) model for new, adaptive design-based 2010 draft guidance, Adaptive Design approaches to advance the oncology Clinical Trials for Drugs and Biologics, drug development process. which details adaptive approaches and encourages their use.4 Traditional Design: Poor information FDA defines an adaptive study as leads to poor performance one that “includes a prospectively Traditional designs contribute to high planned opportunity for modification failure rates and escalating costs because of one or more specified aspects of the answers to pivotal research questions are study design and hypotheses based on obtained only at the end of the trial. Trials analysis of data (usually interim data) using fixed designs rely on assumptions from subjects in the study.” Five adaptive that may be found to be incorrect at the designs—including blinded sample size end of the study. Faulty assumptions used re-estimation and halting early for lack of in underpowered Phase I and Phase II utility—are cited as “well-understood.” trials yield poor information on which to FDA encourages drug developers to base decisions about Phase III designs use these approaches for all studies. where the impact of failure is greatest Seven “less well-understood” designs— due to the large number of patients and including unblinded applications that time involved. The cumulative effects of use interim estimates of treatment effect the traditional approach are low overall for endpoint selection and sample size success rates and high costs (Table 1). re-estimation—should be reserved Advancing oncology drug evaluation for exploratory studies while more depends on: 1) selecting the best drug experience is gained. candidates; 2) identifying and eliminating This regulatory underpinning supports failures as early as possible; and 3) wide application of adaptive design designing trials to identify the right dose,
13 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
Key Performance Metrics change throughout the trial. Instead of Average Cost per Patient: Oncology vs. All Rx categories making pivotal decisions with limited (2011)6 Phase II: $73,000 (vs. $36,000) information before a trial, adaptive Phase IIIa: $57,000 (vs $47,500) Phase IIIb: $66,000 (vs $47,000) designs use accruing information to Overall Success Rates (1993-2004)2 obtain relevant data that inform and 7.1% of Phase I oncology entries were approved 19.0% of Phase I entries in all Rx categories were approved improve critical decisions. Data are Phase Ill Success Rates (2003-2010)3 analyzed continuously or at designated 34.0% of trials achieved statistical significance in primary endpoints interim points, and results are used to Source: Reitsma et al. Table 1. Performance measures in oncology trials. shape future design parameters such as doses, disease indications, or populations for the right disease, in the right patients being studied. Using this flexible as early as possible. With thousands of approach, the trial becomes a learning potential drugs awaiting development— tool that applies evolving knowledge to and with relatively few of these likely to drive subsequent decisions. demonstrate efficacy—earlier information and better-focused evaluation are critical Roles of Bayesian statistics, to improving success rates. Adaptive trial simulation, and biomarkers designs are especially well suited to this Adaptive designs can incorporate more purpose. than one adaptation in a trial and may address a number of research questions Incremental decision-making simultaneously. A single trial can be improves research outcomes designed to evaluate multiple dose Adaptive designs leverage accumulating regimens, indications, drug combinations, data to modify trials as they progress, and even multiple drugs. making better decisions at each For example, a seamless Phase II-III sequential step. Adaptive approaches use breast cancer trial might include adaptive early findings to improve the next phase approaches to stop early for futility, in a flexible process that can accelerate assess dose response, drop or add timelines, reduce costs, and generate arms, change the proportion of patients the most knowledge from the smallest randomized to each arm, and enrich the number of patients. patient population with subjects most Traditional designs use a probabilistic likely to respond. Table 2 lists eight statistical approach. Decisions regarding adaptive settings commonly used in drug dosage, randomization, and sample size development and particularly relevant for are made in advance and usually do not oncology trials.6
14 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
Common Adaptations adaptive designs are more complex • Stopping early (or late, i.e., extending accrual) with a conclusion of and depend heavily on simulations to superiority or futility • Adaptively assigning doses to more efficiently assess the dose- understand trial behavior, efficiencies, outcome relationship • Adding or dropping arms or doses and risks as inputs to inform and • Seamless phases of drug development within a single trial • Changing the proportion of patients randomized to each arm optimize trial design. Depending on • Adaptively identifying in on an indication or responder population the phase and design, regulators may • Changing accrual rate require submission of simulation results Source: Berry D., Nat Rev Clin Oncol 2012; 9: 199-207 Table 2. Eight common types of adaptations. to justify the scientific credibility of an adaptive trial4, particularly if the data Bayesian statistics in adaptive is intended to support a regulatory design. Adaptive designs often use approval. Specialized simulation software, Bayesian statistical methodology to such as FACTS, is available to assess key model complex scenarios. In Bayesian performance characteristics including approaches, statistical models require power, Type 1 error, bias, and average the formulation of a set of prior sample size.7 distributions for any unknown parameters, in addition to the parts of the model Biomarkers provide early information. based on the traditional probability Biomarkers are important in adaptive distribution of observations. Multiple designs to provide early measures sources of information are combined to of efficacy. Since early data may be make inferences, allowing researchers used to modify a trial as it progresses, to test assumptions based on both the traditional long-term oncology direct observations and additional endpoints of survival and progression- information on neighboring doses, free survival are of less benefit. To satisfy different populations, similar compounds, this purpose, biomarkers do not need preclinical modeling, genetic targeting, to be validated surrogates. Berry notes and historical data. Repeated analyses that early findings based on “auxiliary can be conducted within a study—and markers (that) might be correlated with, even across studies—using sequential and predictive for, the primary end point analysis techniques. Results can be used … may be incorporated into the trial to inform the design of the current trial. design to help guide the adaptive aspect of the design.”7 Useful markers might Simulation informs optimal design. include early clinical outcomes (such as While fixed designs depend on imaging, response, and progression), theoretical justification of trial behavior, serum markers, or molecular markers
15 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
from tumors via biopsies. In a provocative toxicity is observed, the lower dose is article, Verweij suggests that functional declared to be the MTD. target pharmacology studies followed by A 1999 analysis reported that when proof-of-concept studies could replace using the 3+3 method, “the probability traditional Phase I, II, and III trials, given of recommending the (correct) MTD at that early tumor shrinkage—as measured the end of the trial … never exceeds 44% by Response Evaluation Criteria in Solid and is most often closer to 30%.”9 Poor Tumors—still appears to be the most MTD identification is attributable to the reliable biomarker.8 tendency to select larger incremental “jumps” in order to observe toxicity more Improves Phase I dose determination quickly in fewer steps. The true MTD The primary goal in Phase I is to often resides in a smaller incremental determine maximum tolerated dose dose and is not observed. (MTD) for the experimental agent. Over- and under-estimation of the true MTD is a Adaptive CRM design. The Continual common problem in oncology trials, most Reassessment Method pinpoints the of which identify MTD using the “3+3” true MTD more precisely by efficiently method. An emerging adaptive approach, evaluating more dose levels. CRM called the Continual Reassessment models the probability of the MTD as a Method (CRM), yields more precise MTD function of dose at each dosage level and determination and increases the likelihood continuously refines it. The 3+3 method that the true MTD is used in Phase II. bases the next dose allocation (and, therefore, the level that will eventually Traditional 3+3 method. In the 3+3 be declared the MTD) on the last cohort method, dose escalation steps are of subjects, while ignoring the data from defined prior to the trial. A cohort of the previous cohorts. CRM uses all the three subjects receives the drug at a data to update the estimation of the MTD starting dose based on preclinical data. and to allocate the next patients, either If no toxicity is observed, another cohort in cohorts or continuously. The model is of three subjects is added and the dose frequently updated and improves with is escalated to the next level. If one of accruing data. three subjects experiences dose-limiting In the majority of cases, CRM yields toxicity, another three-patient cohort better estimation of the MTD and can is added at the same dose, and dose allow for more rapid progression through escalation continues. If any additional early dosing levels depending on the
16 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
operating characteristics and rules that Correct Selection of Maximum Tolerated Dose are established in the design. Although the CRM approach is more complex and requires high levels of modeling and simulation, experience has proved its value in identifying true MTD with a higher level of confidence. As shown in Figure 1 adapted from Parke, CRM is better than 3+3 at identifying the correct dose level in nine of the 10 scenarios presented. In Scenarios 1, 3, 4, and 6, CRM was substantially better, providing a Source: Parke T., Tesselia Technology Consulting, 2010 10% higher probability of identifying the Figure 1. Comparing the CRM and 3+3 methods in MTD than the 3+3 method. In Scenario identifying the maximum tolerated dose (MTD). 2, the CRM and 3+3 approaches yielded very similar results.10 Slow adoption of CRM. Despite current Additional CRM benefits.P arke cites literature demonstrating the superiority additional advantages of CRM: “Unlike of CRM in determining the MTD, most the 3+3, its operating characteristics can Phase I and Phase I-II oncology trials be easily optimized in light of the current continue to use the 3+3 method, likely circumstances, different levels of toxicity based on sponsor and investigator level can be targeted, different cohort sizes of familiarity. Our search using the key used and different levels of accuracy words “adaptive,” “Bayesian,” “CRM,” required before stopping, offering better “3+3” and “escalation” found a total of determination of the MTD at the cost of 12 Phase I and Phase I-II dose escalation greater sample size.”10 Seamless Phase trials published in The Oncologist I-II trials can be designed to allocate (four trials) and the Journal of Clinical subjects based on continuing information Oncology (eight trials) from August on both tolerability and efficacy, an 2012 through August 2013. All 12 trials approach that shortens timelines. Another used the 3+3 design, confirming the benefit is that patients involved in dose 2013 review by Ji and coworkers, which determination may continue to participate reported “... more than 95% of Phase in activity evaluation—an important I studies have been based on the 3+3 advantage from an ethical point of view. design.”10
17 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
Adaptive approaches in Phase II Adaptive Phase II designs can improve Phase III trials be instrumental in identifying the Improving dose-response evaluation. appropriate patient population for Phase Adaptive designs can be used to III evaluation. Identification of the right efficiently evaluate several active doses subpopulation can have a dramatic in Phase II without necessarily increasing impact on the number of patients the sample size. Evaluation of more active required in Phase III trials to demonstrate doses provides a better understanding efficacy. For example, suppose one half of the dose-response relationship, of subjects with non-Hodgkin lymphoma reducing the likelihood of failures due respond well to a drug, as measured by a to suboptimal dose selection in Phase 60% hazard ratio; the other half benefit by III. Ineffective or unsafe dose levels can only 10%. To show superiority in a Phase be discontinued early, and the majority III trial with all patients enrolled at 90% of patients can be allocated to the dose power, 530 events would be required. But levels most likely to be active. in a trial with the subpopulation of more positive responders, only 210 events Improving identification of target would be needed. populations. Increasing genomic knowledge of cancer subtypes is driving Halting for futility. Preplanned futility the need for efficient drug evaluation analysis based on interim data can be in targeted patient populations. The used to stop a study that is unlikely to milestone genetics study of breast meet its primary endpoint. Interim futility tumors published in 2012, for example, analysis also can allow developers to identified four distinct subtypes of breast continue a study with greater confidence cancer, suggesting targets for new drugs of success in Phase III. For example, a and better uses of existing drugs.11 As simple preplanned futility analysis was noted by Esserman and Woodcock, conducted in a Phase III multicenter study “The inability (or lack of explicit effort) to comparing a new therapy to standard of identify and incorporate specific disease care in patients with progressive and/or subtypes into trial design inhibits the recurrent non-resectable glioblastoma development of more cost-effective multiforme. The target sample size was drugs that target specific populations,” 323 randomized patients. Recruitment a dilemma that demands new clinical was difficult; after three years, only 137 trial designs that can address disease patients were randomized. An unblinded heterogeneity and complexity.5 interim futility analysis indicated that
18 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
the therapy was unlikely to demonstrate unblinded adaptations to maintain efficacy. Based on the analysis, the study power. Blinded approaches, which independent data monitoring committee FDA characterizes as generally well- recommended halting the trial. Early understood, compare interim findings to termination avoided unnecessary assumptions used in the planning of the exposure for approximately 180 subjects. study. For example, in studies that use an Halting early avoids Phase III failures event outcome such as response rate for that contribute significantly to the low the endpoint, a blinded examination of productivity and exorbitant cost of the overall event rate can be compared drug development, widely estimated at to assumptions used in study planning. $1.8 billion per approved drug. A 2013 If the comparison shows that actual Forbes analysis suggests that for large event rate is well below the assumption, biopharma companies—those that earn sample size can be increased. Such approval for eight to 10 new drugs over a blinded approaches also can be used in decade—the greater number of failures studies using time-to-event analysis and experienced in managing a large pipeline continuous outcome measures. Since result in an average cost of $5 billion per blinded approaches do not introduce approval.12 statistical bias or require statistical adjustments, they maintain Type 1 error Re-estimating sample size. Sample control. FDA recommends that they size is fixed in traditional designs, with “should generally be considered for most size based on initial assumptions about studies.”4 primary efficacy measures and the rate Unblinded approaches use interim and timing of patient withdrawal from analyses to estimate treatment effects. the study. This approach often results in Unblinded approaches allow initial under powering or overpowering. In the sample size to be increased if the size of first case, the study fails to show definitive the treatment effect is seen to be smaller results. In the second, the trial requires than anticipated, but is still clinically more subjects and time than necessary. relevant. In some cases, adaptations that Adaptive designs use interim data to re- address other elements of study design— estimate sample size as the trial proceeds, such as dose, population or study so sample size can be increased to ensure endpoint—could alter the study power adequate powering. and require re-estimation of sample The 2010 FDA draft guidance makes size. Changes in sample size based on a distinction between blinded and unblinded data analysis may cause an
19 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
increase in the Type 1 error rate, and a Seamless Phase II-III Trial statistical adjustment is necessary for the final study analysis. FDA considers unblinded approaches to be less well-understood and cautions researchers to be conservative when making changes based on early estimates of treatment effect, which
can be misleadingly large or small. Due Source: The National Academy of Sciences to concerns about Type 1 error and Figure 2. A seamless Phase II-III trial to evaluate operational bias, FDA suggests that two drugs alone and in combination. unblinded approaches be used primarily for studies in which the primary objectives questions in the same study. A simulated cannot be achieved using blinded Phase I-II oncology study designed by designs. Drug developers exploring these Huang and coworkers demonstrates the designs must show adequate control of efficiencies that can be gained using Type 1 error. seamless approaches.13 The authors designed a parallel Seamless adaptive designs improve Phase I-II study that combined dose trial efficiencies determination with efficacy assessment for Seamless designs use adaptations and two oncology agents when administered interim data to combine phases into in combination, and when administered a single study, reducing timelines and concurrently versus sequentially. The the number of patients required. These trial begins with an initial period of dose designs are especially useful in oncology escalation. Then patients are randomly studies because adaptations can address assigned to admissible dose levels that a wide variety of questions in the early are compared with each other. Bayesian (Phase II) stage to improve the later probabilities are used to adaptively assign confirmatory stage. Seamless designs more patients to doses with higher activity allow the long-term clinical endpoints levels. Combination doses with intolerable from subjects enrolled in an early phase toxicity are eliminated, while those with to be included in overall trial results. lower efficacy are temporarily closed. Seamless Phase I-II designs. Seamless The trial would be halted if the posterior designs can answer Phase I toxicity probability of safety, efficacy, or futility questions and early Phase II efficacy crosses a pre-specified boundary.
20 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
Applying this design to a combination number of patients and randomization chemotherapy trial for leukemia, the in Phase II are chosen adaptively. authors used simulations to compare Phase II results determine sample size the seamless Phase I-II approach to a in Phase III. Phase III may use interim conventional design with separate Phase analyses to halt early for either futility or I and Phase II trials. Results showed that expected success. Berry notes that the the Phase I-II design reduced sample Drug B-versus-control element during size was better powered and was more Phase II may be counted in the Phase III efficient in assigning more patients to comparison (i.e., inferentially seamless), or doses with higher efficacy levels.14 it may not be counted (i.e., operationally seamless). The entire trial must be Seamless Phase II-III designs. Larger simulated to control the Type 1 error rate. Phase II studies can increase the Like the use of CRM in dose probability of success in Phase III but also determination, the adoption of seamless increase research timelines and costs. In designs in oncology studies is slow. When many cases, Phase III success rates can be we broadened our key word search of improved and overall timelines reduced The Oncologist and the Journal of Clinical using a seamless Phase II-III design that Oncology to include all trials at any phase combines the learning-and-confirming of development, we found only three phases into a single study. The first published studies (all in Journal of Clinical stage generates information to guide the Oncology) that used adaptive designs confirmatory stage regarding decisions between August 2012 and August such as: whether to stop for futility; what 2013: two used adaptive randomization dose, regimen, endpoint, and responding strategies, while one was a seamless subpopulation to study; and whether to Phase II-III trial.14,15,16 evaluate the experimental drug alone or A 2012 survey conducted by the DIA in combination with another therapy. Adaptive Design Scientific Working Figure 2 shows a seamless Phase Group17 suggests a considerable increase II-III design for a trial to evaluate two in the use of adaptive design, particularly experimental drugs, alone and in compared to a previous survey conducted combination, as adapted by Berry from in 2008 (i.e., before the publication of the “A National Cancer Clinical Trials System draft FDA guidance). The survey of 16 for the 21st Century.”7 In this example, biopharma companies and CROs showed the single agent, Drug B, is selected in more enthusiasm overall for adaptive Phase II and continues into Phase III. The design within industry and academia, and
21 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
in particular an increase in the number of their impact on pathologic complete trials using designs described as less well response (pCR), a predictor of disease- understood in the draft FDA guidance free survival. Drugs considered successful (i.e., typically more complex adaptive in the screening trial are predicted to designs). The Tufts Center for the Study have an 85 percent likelihood of success of Drug Development also showed that, in a confirmatory, randomized trial of based on a roundtable discussion held in 300 patients with tumors that have the 2013 with 40 senior executives,18 across drug’s identified biomarker signature. The the industry simple adaptive designs ultimate goal is to evolve a new model (such as early stopping due to futility and to streamline clinical evaluation and sample size re-estimations) are used on accelerate regulatory approval pathways. approximately 20% of clinical trials and The first two “graduates” from the that the adoption of adaptive design in I-SPY 2 trial are veliparib in combination the exploratory drug development phase with carboplatin and standard is expected to increase significantly over neoadjuvant chemotherapy in the triple- the next several years. negative breast cancer subset, and neratinib in combination with standard Adaptive I-SPY 2 trial models a neoadjuvant chemotherapy in HER2+/HR- better research approach breast cancer. Details of the clinical results The potential of adaptive design to and predictive probability of success are advance oncology drug development shown in Tables 3 and 4. is evident in the groundbreaking I-SPY Bayesian Predictive Probability of Success for Veliparib 2 screening trial, a collaborative Phase Probability Predictive Probabil- II research platform sponsored by the Veliparib is ity of Success in 300- FDA and used by multiple industry and Signature Superior Patient Phase III Trial academic researchers. I-SPY 2 is designed All HER2- 92% 55% HER2-/HR+ 28% 9% to identify active experimental drugs for HER2-/HR- 99% 92% breast cancer, together with predictive Source: Reitsma et al. biomarkers.5,19 Table 3. The graduating arm is triple-negative (HER2-/ HR-) subset with a 93% Bayesian probability of success I-SPY 2 uses an adaptive design to in a 300-patient Phase III trial. simultaneously screen Phase II anticancer agents in women with stage 2 or 3 breast The graduating arm is triple negative cancer at risk for recurrence. Drugs (HER2-/HR-) subset with a 93% Bayesian are evaluated by class, using standard probability of success in a 300 patient and emerging biomarkers to measure Phase III trial.
22 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
The graduating arm is the HER2+/HR- subset, neratinib’s predictive probability subset with a 78% Bayesian probability of success was 78 percent at the time of of success in a 300 patient Phase III publication. trial. (Reprinted by permission from The benefits of theI -SPY 2 trial are the American Association for Cancer illustrated with the graduation of both Research.20) neratinib and veliparib. Development has been accelerated and focused on Bayesian Predictive Probability of Success for Neratinib the patient population with the greatest Probability Predictive Probability probable benefit from treatment with the Neratinib is of Success in 300-Patient Signature Superior Phase III Trial selected drugs, which leads to the greatest ALL 92% 44% likelihood of success in a pivotal Phase III HR+ 81% 40% trial. Interestingly, without participating HR- 89% 53% HER2+ 95% 73% in this collaborative trial, these agents HER2- 63% 20% may have been in competition following MP+* 91% 66% traditional drug development pathways HR-/HER2- 72% 34% with a lower probability of success HR-/HER2+ 94% 78% for each compound in a broader HR+/HER2+ 91% 65% HR+/HER2- 39% 12% population. Having graduated in unique
Source: American Association for Cancer Research20 patient subsets, the compounds are no Table 4. The graduating arm is the HER2+/HR- sub- longer competing for the same patient set with a 78% Bayesian probability of success in a 300-patient Phase III trial. population. This property of the I-SPY 2 trial enhances the development of multiple Each drug’s Bayesian predictive novel agents in breast cancer, which is probability of success is calculated for increasingly recognized as consisting of each unique patient subset until the many distinct sup-types of disease. threshold of 85 percent is met within any given subset. When 85 percent Conclusion probability of success is reached, Regulatory guidance recognizes the the accrual is stopped within this value of adaptive design, and emerging subpopulation and the drug graduates research models like I-SPY 2 demonstrate to a separate Phase III trial within the its great value in advancing oncology drug defined subpopulation. While the development. It remains for the biopharma published probability of Phase III success industry to implement and advance is greater than 85 percent for veliparib adaptive design as a fundamental clinical in the triple-negative breast cancer research methodology.
23 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
setting. Tessella Technology & Consulting White Paper 2010. Avail- able at: http://tessella.com/documents/comparison-crm-vs-33-oncol- Dirk Reitsma, MD, is Vice President, ogy-phase-1-sett.... Accessed November 20, 2014. Therapeutic Area Head, Oncology, Global (10) Ji Y, Wang SJ. Modified toxicity probability interval design: a safer and more reliable method than the 3+3 design for practical Phase I Product Development, PPD trials. J Clin Oncol 2013; 31(14): 1789-1791. Austin Combest, PharmD, BCOP, MBA, (11) The Cancer Genome Atlas Network. Comprehensive molecular por- is Senior Clinical Scientist, Global Product traits of human breast tumours. Nature 2012; 490: 61-70. (12) Desmond-Hellmann S. The cost of creating a new drug now $5 bil- Development, PPD lion, pushing big pharma to change. Forbes Pharma & Healthcare, Jürgen Hummel, is Statistical Science Director, August 11, 2013. Available at: http://www.forbes.com/sites/mat- Biostatistics, PPD thewherper/2013/08/11/how-the-staggering-... Accessed November Ashley Simmons, PharmD, is Associate Director, 20, 2014. (13) Huang X, Biswas S, Oki Y, et al. A parallel Phase I/II clinical trial Feasibility Strategy, PPD design for combination therapies. Biometrics 2007; 63: 429-436. (14) Smith DC, Smith MR, Sweeney C, et al. Cabozantinib in patients with advanced prostate cancer: results of a Phase II randomized dis- continuation trial. J Clin Oncol 2013; 31(4):412-419 (15) Garcia-Manero G, Jabbour E, Borthakur G, et al. Randomized open- References label Phase II study of decitabine in patients with low- or interme- (1) Citeline Pharma R&D Annual Review 2012. diate-risk myelodysplastic syndromes. J Clin Oncol 2013; 31(20): (2) DiMasi JA, Feldman L, Seckler A, et al. Trends in risks associated 2548-2553. with new drug development: success rates for investigational drugs. (16) Schmoll HJ, Cunningham D, Sombrero A, et al. Cediranib with Clin Pharmcol Thera 2010; 87(3): 272-277. mFOLFOX6 versus bevacizumab with mFOLFOX6 as first-line (3) Hay M, Rosenthal J, Thomas D, et al. Bio/BioMedTracker clinical treatment for patients with advanced colorectal cancer: a double- trial success rates study. BIO CEO & Investor Conference February blind, randomized Phase II study (HORIZON III). J Clin Oncol 15, 2011. Available at: http://insidebioia.files.wordpress.com/2011/02/ 2012; 30(29): 3588-3595. bio-ceo-biomedtracker-bio.... Accessed November 20, 2014. (17) Morgan CC, Huyck S, Jenkins M et al. Adaptive Design: Results (4) U.S. Food and Drug Administration. Guidance for industry: Adap- of 2012 Survey on Perceptions and Use. Therapeutic Innovation & tive design clinical trials for drugs and biologics. Rockville MD: Food Regulatory Science 2014; 48(4): 473-481. and Drug Administration, February 2010. Available at: http://www. (18) Tufts Center for the Study of Drug Development, “The Adoption and fda.gov/downloads/drugs/guidancecomplianceregulatoryinformati.... Impact of Adaptive Trial Designs,” R&D Senior Leadership Brief, Accessed November 20, 2014. 2013. Available at: http://csdd.tufts.edu/files/uploads/tuftscsddbrief- (5) Esserman LJ and Woodcock J. Accelerating identification and regula- 1final_new.pdf. Accessed November 20, 2014. tory approval of investigational cancer drugs. JAMA 2011; 306(23): (19) I-SPY 2 Trial. I-SPY 2 is a clinical trial for women with newly diag- 2608-2609. nosed locally advanced breast cancer. Available at: http://www.ispy2. (6) Berry, D. Adaptive clinical trials in oncology. Nat Rev Clin Oncol org. Accessed November 20, 2014. 2012; 9: 199-207. (20) Park et al. Neratinib plus standard neoadjuvant therapy for high-risk (7) Tessella and Berry Consultants FACTS™. The fixed and adaptive breast cancer: Efficacy results from the I-SPY 2 TRIAL; Proceedings clinical trial simulator. Available at: http://tessella.com/products/ of the 105th Annual Meeting of the American Association for Cancer fixed-adaptive-clinical-trial-simulator. Accessed November 20, 2014. Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; (8) Verweij J. Clinical trials in drug development: a minimalistic 2014. Abstract nr CT227. Available at: http://www.abstractsonline. approach. Curr Opin Oncol 2012; 24: (3)332-337. com/Plan/ViewAbstract.aspx?mID=3404&sKey=ed53.... Accessed (9) Parke T. A comparison of the CRM vs 3+3 in an oncology Phase 1 November 20, 2014.
24 | June 2016 | Pharm Tech Clinical Trial Site Centricity VS Patient Centricity: What’s A More Compatible Model? By Moe Alsumidaie
Press Release With the introduction of mobile health Peel-ID (mHealth) technologies in the field of healthcare, Sponsors and CROs are looking Click to view into mHealth to design patient centric release clinical trials in order to reduce study visit costs and trial participation burden on patients. Nonetheless, with mHealth at its infancy, Sponsors and CROs at The Partnerships in Clinical Trials Conference expressed their encounters and challenges with the concept and execution of patient centered clinical trials. This article will evaluate concerns with patient centricity, and conceptualize how site-centered clinical trials may address these concerns, particularly with more complex studies.
Patient Centered Clinical Trials: Works for Some, but, Not All The concept of patient centric clinical trials (also
25 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
dubbed siteless trials, remote trials and are expressing concerns with the model, virtual trials), involves designing study as principal investigators (PIs) feel they do visits in combination with mHealth and not have much control over the patient, telehealth technological capabilities to and they are specifically concerned allow patients to conduct study visits with accountability (i.e., if a patient is from the convenience of their homes. harmed, the PI is ultimately accountable Subsequently, this convenience can offer for the patient’s safety). Additionally, benefits to Sponsors, such as reducing with clinical trials being a financially study visit costs, and minimizing subject restraining engagement at study sites, dropout rates. siteless models impact study sites, as During a round-table session on this sites generate less revenue from missing topic led by Kamyar Farahi, Clinical Trials in-person study visits. Lead at Janssen, the discussion favored The round-table discussion with that with the help of telemedicine and Farahi also touched on this topic and the virtual technologies, one can potentially consensus was that one of the limitations reduce the trial visit costs, travel burden of site-less clinical trial models could be to the patient, and potentially improve the impact on the business model of the patient and investigator interactions investigators, but with limited interactions during the course of a clinical trial. between the patient and the investigator While this model may work for less the patients could be less satisfied with complex trials, such as vaccine studies, his/her participation. or trials that require self-administered exams (i.e., mobile blood pressure, mail Breakthroughs in Telemedicine in blood test kits, urine exams, etc.), the Clinical Research siteless model poses limitations to more In the field of healthcare, advances in complex trials, such as those that require telemedicine are reaching new heights more complex study visit procedures, with telerobotics, where highly trained or study visit procedures that can only specialists are experimenting with be conducted by trained specialists with conducting complex medical procedures specialized equipment. remotely. Dr. Partho Sengupta of The Icahn School of Medicine at Mount Sinai Patient Centric Model: Concerns in New York in collaboration with Rush with Investigators University Medical Center in Chicago While siteless models offer benefits recently initiated a breakthrough clinical towards Sponsors and CROs, study sites trial to evaluate remote telerobotic
26 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
ultrasound examinations over the a 10,000 patient clinical trial pilot, where Internet. Through his computer in New Walgreens acted as the study site. York, Dr. Sengupta is able to control a Walgreens recently unveiled its patient robotic ultrasound arm to obtain cardiac recruitment services at the Partnerships echoimages of patients in Chicago. in Clinical Trials Conference, and may This capability can be particularly progress towards acting as remote study useful in implementing site centric clinical sites in clinical trials, especially with its trial models to distribute specialist access specialty pharmacy division, that wields to study sites globally. “Launching long- specialized facilities to handle a wide distance, tele-robotic ultrasound exams range of disease indications including between two major hospitals in two large oncology, cystic fibrosis, multiple cities is a sign that we may be able to sclerosis, rheumatoid arthritis, transplants, make waves in accelerating access to and much more. specialists to support global clinical trial site centric models,” said Sengupta. Is Site Centricity a More Compatible Model than Patient Centricity? What’s the Difference between Patient centric models tend to work Patient Centricity and Site best with less complex clinical trials and Centricity? relatively easy data collection methods. Patient centricity involves conducting However, this model includes limitations clinical trial study visits and providing for more complex study procedures study measurements directly from the that require specialized equipment, and patient and by the patient in the comfort expert medical specialists. Moreover, of their own homes and lives. Telehealth this model may introduce risks with and mobile health (i.e., Fitbit, Apple data quality, as patients may not adhere Watch, Scanadu, etc.) technologies to study regiments, or required data enable patients to report their outcomes collection methodologies (i.e., giving directly to study sites and Sponsors. up on taking their blood pressure if the Site centricity involves the process of device malfunctions on the first round, or engaging patients and conducting study if there are too many procedures). visits in local neighborhood clinics or With site centric models, although medical centers and allowing physicians can be more costly than patient centric to remotely monitor the patients and models, study teams can rely on local collect data via telehealth. For example, trained healthcare professionals to Novartis partnered with Walgreens to run collect required data according to
27 | June 2016 | Pharm Tech Adaptive Trial Improve Centricity Supply Chain Design Oncology Trials Models
study procedures (which may contribute fewer PIs/specialists. This method can be towards enhancing data quality compared less costly and more efficient compared to patient centered models), and with the to selecting/initiating numerous academic advances of telehealth and telerobotics, medical centers/involving several PIs, study teams can deploy hub and spoke and might be more acceptable by PIs models to execute global studies with compared to patient centric models.
28 | June 2016 | Pharm Tech