Joseph Pierro, MD

As cancer patients are surviving longer, there is growing interest in better understanding the potential for cardiovascular complications associated with cancer therapies. Indeed, over four decades, clinical data have shown that oncology treatments-from chemotherapies to targeted immunotherapies-can negatively impact heart health. Risks include ischemia, vascular disease, cardiomyopathy, myocarditis, hypertension, heart failure, arrhythmias, QT prolongation, and thrombosis.

 Often, patients face a tradeoff of the benefits and risks of treatment. Herceptin, for example, revolutionized the treatment of breast cancer, but also produced congestive heart failure in many patients.

	Thus, the FDA requires cardiac safety monitoring in oncology trials. And experienced clinical development teams understand and apply the evolving body of clinical research and therapeutic-level recommendations to clinical trial protocols in order to monitor for these cardiac conditions. Yet, the imaging required to monitor for cardiotoxicities during clinical development can, at times, be treated as a perfunctory task.

	The fact is, for cardiac safety trials, as in all clinical studies involving imaging, it matters how images are taken, how they are analyzed, and how the process is monitored. The quality and capabilities of the imaging management system-the process and tools used to collect, share, interpret, and store images-has an appreciable impact on patient safety as well as trial costs and timelines.

	Imaging is commonly used as a biomarker for safety in oncology development and, in many trials, is the primary efficacy endpoint. Echocardiography is used to assess symptomatic and asymptomatic cardiac dysfunction in oncology patients and to grade the severity of the condition; the results are surrogate markers for cardiac safety.

	The availability of high-resolution and sensitive imaging techniques allows for the identification of potential pathological changes earlier (e.g., decline in left ventricular ejection fraction (LVEF), acute myocarditis, and valve leaflet thickening).

	Such assessments are performed in order to:

		Make go/no-go decisions for continued drug development

		Assist in identifying high-risk subgroups within the population (for example, subjects with low ejection fraction or ventricular volume)

		Screen subjects for protocol eligibility

		Monitor patients for treatment-emergent changes, which may necessitate the need for dose adjustment or treatment discontinuation

	The study team will recommend the monitoring frequency based on an understanding of the drug’s mechanism of action, the intended patient population(s), the drug toxicity profile, and a review of prior clinical and non-clinical data.

	Early detection and diagnosis of cardiotoxicity is, of course, important, as the goal is to be able to medically intervene, prevent delayed effects, and improve outcomes for the patients. And, should the investigational product need to be re-engineered in the lab, it is better to determine this as early as possible in the development path.

	Increasingly, echocardiography is the primary assessment tool for cardiac safety assessments, given its advantages of wide availability, lower cost, and the improved detection three-dimensional methods afford.

	A critical factor in patient care and data integrity

	Several factors can impact the quality of data produced through echocardiography, as with other imaging modalities.

	First, the sonographer’s skill, experience, and knowledge can influence the quality of the image. Second, there can be variations in how the image is acquired, such as the display or resolution, which are related to the equipment itself. And, third, how individual readers interpret the images can-and does-vary. Studies conducted since 1947 have measured diagnostic discordance in the 25% to 40% range.

	Indeed, wide variability in both quantitative and qualitative echocardiography assessments is recognized in recent guidance publications offered by both the European Society of Cardiology (ESC) and the American Society of Echocardiography (ASE).

	Errors and inconsistencies can have a direct impact on patient care, as well as on the chance of regulatory approval for the investigational product. Thus, there’s a need for a comprehensive approach to ensuring that image acquisition techniques are standardized, that images are appropriately collected to the highest quality standard across multiple sites, and that the images are evaluated consistently to support the study protocol endpoints. That’s why imaging management systems and active reader management practices are critical. Additionally, the recent ASE Echocardiography Report recommends the use of a centralized echo-reading laboratory in multicenter clinical trials.

	Imaging management technology: Added transparency, reduced risk

	Today’s imaging technology platforms can help reduce the chance of human error, speed the assessment process, increase objectivity and consistency, and improve patient safety. A single system is used to collect the source data (which is input directly by the clinical trial site staff), manage image analysis, report on the results, and archive records. Because the data are always contained in the same system, there are no delays or errors caused by transferring it between different platforms to perform various operational tasks. This, in fact, eliminates virtually 100% of transcription errors. Furthermore, a consistent process is applied using standardized procedures in a common viewing platform.

 Imaging results are collected at the clinical site and are immediately available at the clinical institution, so safety signals can be identified at once for clinical decision-making. They are also immediately available to independent or blinded readers (BICR) so that they can make their independent assessments. The results can also be shared with other stakeholders such as sponsors, cardiac advisory boards, and data and safety monitoring committees.

Visibility to all study data in near real time.

 Trial managers and medical directors have total visibility to all study imaging data, throughout the imaging life cycle. They can:

		Measure the degree of discordance between site-based and blinded, central readers

		Track the number of cases requiring adjudication

		Monitor for selection bias in adjudicated cases

	Such insight into reader workloads and performance makes it possible to quickly address issues around reader drift, variability, and bias, thus minimizing their impact. If, for example, it becomes clear from the metrics that a reader’s approach has changed, retraining can be provided so that the reader’s performance is consistent with others in the reader pool.

 The system delivers reminders and gives users automatic prompts so that images pass through the agreed-upon workflow from reader to reader to adjudicator as efficiently as possible. The process is not dependent upon a project manager to keep it on track.

 Image analysis software can direct a reader through the analysis of each imaging time point and even pre-process and segment anatomical structures of interest in lockstep with the study’s imaging charter and image evaluation protocol (IEP). This minimizes protocol deviations and ensures that each reader’s unique bias does not creep into the analysis process by focusing the reader on targeted endpoints whose workflows are outline in the trial-specific IEP. Artificial intelligence (AI) can be used to augment human assessment. Using AI in this way can reduce read times by as much as 50% and the need for adjudication by 20%. It would, in the process, increase speed and reduce costs.

	Active reader management: Added efficiency and effectiveness

	An imaging core lab will be concerned with all of the controllable factors that impact data quality, from data collection to image review and data analysis. The best practices of an imaging core lab include:

		Ensuring that the imaging endpoints support the protocol

		Selecting a limited number of independent readers based on their training and experience

		Managing the image assessment-starting with training on the system and standardizing the imaging protocol and reading criteria

		Monitoring reader performance periodically during the study to lower variability and adjudication rates (readers are asked to re-read cases to identify performance drift)

		Reporting on imaging status and reader performance with recommendations for discussion and intervention if needed

	The lab’s ability to monitor reader performance and oversee imaging progress is dependent upon the capabilities of the imaging management software.

	Ensuring the validity of cardiac imaging data points in oncology clinical trials is difficult, and the challenge must be addressed early in study planning to both protect patients and the integrity of the trial itself. Sponsors should:

		Look to both positive and negative results in early trials to determine the potential predictive value of cardiac endpoints.

		Identify patients at risk for cardiotoxicity and determine how to balance survival benefits with reduced cardiac risk relative to a patient’s quality of life. (There is a higher tolerance for adverse events in oncologic treatments, and the benefit/risk balancing point is different than in other disease states.)

		Clearly define the protocol-required cardiac assessments and measurement parameters to be obtained during the echocardiogram. This will help the clinical team understand patients’ clinical course and inform decisions on treatment adjustments that may be needed to maintain or restore normal cardiac function. The protocol should include details on the number of imaging findings required, together with information based on statistical analysis to assist investigators in managing patients.

		Establish a baseline of cardiac performance prior to treatment, perform serial assessments during the trial, and conduct a follow-up assessment post-therapy. 

		Pay particular attention to the technologies and processes used to ensure accuracy in how images are taken and interpreted.

	Questions to ask your imaging partner in cardiac safety trials

		Which system is used to collect imaging endpoints?

		How quickly are the data available? How easy is the analysis?

		What is the expected rate of reader discordance?

		What is the optimal number of independent readers for the trial?

		What has the adjudication rate been on similar trials?

		What process is used to monitor the timeliness and accuracy of reads?

		What process/technology is in place for creating an audit trail and ensuring compliance?

		What systems and training are in place to ensure that sites know how to follow the imaging protocol?

	Ensuring that imaging in cardiac safety clinical trials is managed expertly and with the benefit of the latest technology and careful oversight has broad implications for patient safety as well as trial cost and timing. Overall error rates can be reduced by as much as 20%, read times can be reduced by up to 50%, and adjudication rates can be cut
	by 20%.

		Blaes, AH, Thavendiranathan, P, Moslehi J. Cardiac Toxicities in the Era of Precision Medicine: Underlying Risk Factors, Targeted Therapies, and Cardiac Biomarkers; 2018 

 (Print ISSN: 1548-8748 38; Electronic ISSN: 1548-8756) published by the American Society of Clinical Oncology, Inc. (“ASCO”), volume 38, p764-774	

		Zamorano JL, Lancellotti P, Rodriguez Muñoz D, Aboyans V, Asteggiano R, Galderisi M, et al. 2016 ESC Position Paper on cancer treatments and cardiovascular toxicity developed under the auspices of the ESC Committee for Practice Guidelines: The Task Force for cancer treatments and cardiovascular toxicity of the European Society of Cardiology (ESC)

Articles

How new technology can impact cardiac imaging in oncology clinical trials and have broader implications for patient safety.

The Benefits of Advanced Imaging Management Systems

	As cancer patients are surviving longer, there is growing interest in better understanding the potential for cardiovascular complications associated with cancer therapies. Indeed, over four decades, clinical data have shown that oncology treatments-from chemotherapies to targeted immunotherapies-can negatively impact heart health. Risks include ischemia, vascular disease, cardiomyopathy, myocarditis, hypertension, heart failure, arrhythmias, QT prolongation, and thrombosis.

 Often, patients face a tradeoff of the benefits and risks of treatment. Herceptin, for example, revolutionized the treatment of breast cancer, but also produced Congestive Heart Failure in many patients. 

	Thus, the US Food & Drug Administration (FDA) requires cardiac safety monitoring in oncology trials. And, experienced clinical development teams understand and apply the evolving body of clinical research and therapeutic-level recommendations to clinical trial protocols in order to monitor for these cardiac conditions. Yet, the imaging required to monitor for cardiotoxicities during clinical development can, at times, be treated as a perfunctory task. 

	The fact is, for cardiac safety trials as in all trials involving imaging: it matters how images are taken, how they are analyzed, and how the process is monitored. The quality and capabilities of the imaging management system-the process and tools used to collect, share, interpret, and store images-has an appreciable impact on patient safety as well as trial costs and timelines.

	Cardiac imaging: essential in oncology trials

	Imaging is commonly used as a biomarker for safety in oncology development and in many trials is the primary efficacy endpoint. Echocardiography is used to assess symptomatic and asymptomatic cardiac dysfunction in oncology patients and to grade the severity of the condition; the results are surrogate markers for cardiac safety.

	The availability of high-resolution and sensitive imaging techniques allows for the identification of potential pathological changes earlier (e.g., decline in Left Ventricular Ejection Fraction (LVEF), acute myocarditis, and valve leaflet thickening). 

	Such assessments are performed in order to: 

		Assist in identifying high-risk subgroups within the population (for example subjects with low ejection fraction or ventricular volume) 

		Monitor patients for treatment-emergent changes which may necessitate the need for dose adjustment or treatment discontinuation. 

	The study team will recommend the monitoring frequency based on an understanding of the drug’s mechanism of action, the intended patient population(s), the drug toxicity profile, and a review of prior clinical and non-clinical data. 

	Early detection and diagnosis of cardiotoxicity is, of course, important, as the goal is to be able to medically intervene, prevent delayed effects, and improve outcomes for the patients. And, should the investigational product need to be re-engineered in the lab, it is better to determine this as early as possible in the development path.  

	Increasingly, echocardiography is the primary assessment tool for cardiac safety assessments, given its advantages of wide availability, lower cost, and the improved detection three-dimensional methods afford. 

	Imaging: a critical factor in patient care and data integrity

	Several factors can impact the quality of data produced through echocardiography, as with other imaging modalities. 

	First, the sonographer’s skill, experience, and knowledge can influence the quality of the image. Second, there can be variations in how the image is acquired, such as the display or resolution, that are related to the equipment itself. And third, how individual readers interpret the images can-and does-vary. Studies conducted since 1947 have measured diagnostic discordance in the 25 percent to 40 percent range.

	Indeed, wide variability in both quantitative and qualitative echocardiography assessments is recognized in recent guidance publications offered by both The European Society of Cardiology (ESC) and The American Society of Echocardiography (ASE).

	Imaging management technology: added transparency, reduced risk

	Today’s imaging technology platforms can help reduce the chance of human error, speed the assessment process, increase objectivity and consistency, and improve patient safety. A single system is used to collect the source data (which is input directly by the clinical trial site staff), manage image analysis, report on the results, and archive records. Because the data are always contained in the same system, there are no delays or errors caused by transferring it between different platforms to perform various operational tasks. This, in fact, eliminates virtually 100 percent of transcription errors. Furthermore, a consistent process is applied using standardized procedures in a common viewing platform. 

Imaging results are collected at the clinical site and are immediately available at the clinical institution, so safety signals can be identified at once for clinical decision making. They are also immediately available to independent or blinded readers (BICR) so that they can make their independent assessments. The results can also be shared with other stakeholders such as sponsors, cardiac advisory boards, and Data and Safety Monitoring Committees. 

Visibility to all study data in near real time. 

Trial managers and medical directors have total visibility to all study imaging data, throughout the imaging life cycle. They can:					

				Monitor intra/inter reader variability

				Measure the degree of discordance between site-based and blinded, central readers

				Track the number of cases requiring adjudication

				Monitor for selection bias in adjudicated cases

	            Such insight into reader workloads and performance makes it possible to quickly address issues around reader drift, variability, and bias, thus minimizing their impact. If, for example, it becomes clear from the metrics that a reader’s approach has changed, re-training can be provided so that the reader’s performance is consistent with others in the reader pool. 

.  The system delivers reminders and gives users automatic prompts so that images pass through the agreed-upon workflow from reader to reader to adjudicator as efficiently as possible. The process is not dependent upon a project manager to keep it on track. 

. Image analysis software can direct a reader through the analysis of each imaging time point and even pre-process and segment anatomical structures of interest in lockstep with the study’s imaging charter and image evaluation protocol (IEP). This minimizes protocol deviations and ensures that each reader’s unique bias does not creep into the analysis process by focusing the reader on targeted endpoints whose workflows are outline in the trial specific IEP. Artificial Intelligence (AI) can be used to augment human assessment. Using AI in this way can reduce read times by as much as 50 percent and the need for adjudication by 20 percent. It would, in the process, increase speed and reduce costs.  

	Active reader management: added efficiency and effectiveness 

		Ensuring that the imaging endpoints support the protocol 

		Monitoring reader performance periodically during the study to lower variability and adjudication rates (readers are asked to re-read cases to identify performance drift) 

		Reporting on imaging status and reader performance with recommendations for discussion and intervention if needed 

	The lab’s ability to monitor reader performance and oversee imaging progress is dependent upon the capabilities of the imaging management software. 

	Ensuring the validity of cardiac imaging data points in oncology trials is difficult, and the challenge must be addressed early in trial planning to both protect patients and the integrity of the trial itself. Sponsors should: 

		Look to both positive and negative results in early trials to determine the potential predictive value of cardiac endpoints. 

		Identify patients at risk for cardiotoxicity and determine how to balance survival benefits with reduced cardiac risk relative to a patient’s quality of life. (There is a higher tolerance for adverse events in oncologic treatments, and the benefit/risk balancing point is different than in other disease states.)  

		Clearly define the protocol-required cardiac assessments and measurement parameters to be obtained during the echocardiogram. This will help the clinical team understand patients’ clinical course and inform decisions on treatment adjustments that may be needed to maintain or restore normal cardiac function. The protocol should include details on the number of imaging findings required, together with information based on statistical analysis to assist investigators in managing patients. 

		Establish a baseline of cardiac performance prior to treatment, perform serial assessments during the trial, and conduct a follow-up assessment post therapy.  

		Pay particular attention to the technologies and processes used to ensure accuracy in how images are taken and interpreted (See Box:  Questions to Ask Your Imaging Partner). 

		Which system is used to collect imaging endpoints? 

		How quickly are the data available? How easy is the analysis? 

		What is the optimal number of independent readers for the trial? 

		What has the adjudication rate been on similar trials? 

		What process/technology is in place for creating an audit trail and ensuring compliance? 

		What systems and training are in place to ensure that sites know how to follow the imaging protocol? 
		 

	Ensuring that imaging in cardiac safety trials is managed expertly and with the benefit of the latest technology and careful oversight has broad implications for patient safety as well as trial cost and timing. Overall error rates can be reduced by as much as 20 percent, read times can be reduced by up to 50 percent, and adjudication rates can be cut by 20 percent.

, MD, Medical Director, Imaging, ERT and 

, MD, Vice President and Chief Medical Officer, Cardiology, ERT

Explore the growing interest in better understanding the potential for cardiovascular complications associated with cancer therapies.

Cardiac Imaging in Oncology Trials: The Benefits of Advanced Imaging Management Systems

	Over the last decade, many initiatives sponsored by various entities, including academic and clinical research centers [e.g., American Society of Clinical Oncology (ASCO), Friend of Cancer Research (FOCR)], pharmaceutical companies, patient advocacy groups, global regulatory and government agencies; have focused on efforts to streamline clinical trial eligibility and data collection as presented in a transformational ASCO 2011 campaign entitled “Accelerating Progress Against Cancer–ASCO’s Blueprint for Transforming Clinical and Translational Cancer Research.” Other important efforts are the 2012 passage of the Advancing Breakthrough Therapies for Patients Act which created expedited pathways for patient access to new drugs, and the 2014 21

 Century Cures Initiative that focused on modernizing clinical trials with digital tools to facilitate data sharing to achieve rapid development of drugs and devices for unmet medical needs. To parallel these efforts, the FDA updated their guidance documents and rolled out expanded access and breakthrough therapy designation programs by explaining how single-arm trials may be used to support approval. During the Conference on Clinical Cancer Research in 2015, Dr. Richard Pazdur, Director of FDA Office of Hematology and Oncology Products, expressed his strategic vision to improve navigation through the drug approval process within the expanded access system in order to improve the informed consent processes, the need for eligibility criteria justified by science to improve patient access and develop best practices for big data and data sharing.
	
	


	
	Clinical trials designed for regulatory approval are usually drug centric with restrictive eligibility criteria. The concept is to enroll a narrowly defined homogenous patient population by limiting variables that could impact the efficacy endpoint and provide a better chance for drug approval; albeit, a narrowly indicated population for labelling. A more patient centric approach has an expanded eligibility criteria where more patients are eligible for study participation, which results in faster study enrollment, exposes more patients to study drug treatment and data generalized to more patients.
	
	As oncology clinical study designs now center on selecting the right therapy for the right  patient in a variety of different adaptive designs, such as basket and umbrella trials, the trial endpoints and eligibility criteria have also changed along with the value of the data generated in early phase studies. Regulatory agencies have streamlined pathways to move drugs with enhanced activity on progression free survival (PFS) and response endpoints to ultimately reach patients faster by accelerating approval and breakthrough status programs. An example of a streamlined pathway is the accelerated approval of Keytruda® (pembrolizumab) based on the results from a single biomarker-based Phase IB study (KEYNOTE-001).
	
	As cancer care paradigms have evolved, so too have clinical trial designs evolved to include these complex strategies, striking the balance between having too permissive or too restrictive eligibility criteria that can impact study success and timely completion. A recent 

 (JCO) special article entitled “Modernizing Eligibility Criteria for Molecularly Driven Trials” by Edward S. Kim, MD et al

 reviewed trial eligibility requirements in the era of personalized medicine. The paper reported enrollment in oncology trials is poor, with only 5% to 8% of eligible patients participating in research studies-that number, as well as the complexity of eligibility criteria, has increased over time. For example, 13 eligibility categories were utilized in cytotoxic or non-molecularly targeted therapy protocols as compared to 32 eligibility categories were found in targeted therapy protocols, which has affected enrollment pace and the duration of the enrollment period. Kim pointed out, “many eligibility criteria are grandfathered into clinical trial protocols, not because they are appropriate for the trial, but because they were carried over from previous protocols written for similar study populations. This potentially leads to an additive effect in the number of criteria.” Following the practice can complicate the interpretation of study data and impact the applicability of the results to the intended population. The working group recommended study teams construct eligibility criteria in the context of: meeting scientific objectives, generalized to non-study populations, related to patient safety and drug toxicity concerns, and review and revise eligibility criteria on a periodic basis. These cooperative working groups plan to release several manuscripts later this year at the Fall Friends of Cancer Research (FOCR) meeting in November, which will include expanded template eligibility criteria for implementation across clinical protocols to accelerate and advance cancer care.


	
	Cancer heterogeneity plays a prominent role in determining which patients may benefit from  particular cancer medicines while many advanced technologies or treatment approaches have evolved to serve as biomarkers for use in drug development and assist in patient treatment. In-vitro diagnostic tests, companion diagnostics (e.g., assay to detect tumor PD-L1 expression in metastatic non-small cell lung cancer [NSCLC]) and molecular marker-based imaging tests may serve as biomarkers for treatment prediction, response assessment, determination of pharmacodynamics effect or as a safety endpoint. Adaptive clinical study designs, with subgroups enriched or stratified according to biomarker test results (i.e., marker positive or negative groups), are useful to characterize drug performance, and to modify study designs  when treatment is found to be futile or unsafe in the biomarker negative group. This approach was discussed in a recent Journal of the American Medical Association (JAMA) paper by Mark G. Kris, M.D. who reported matching cancer patients to targeted therapy-based on tissue biomarkers or plasma testing (e.g., digital or quantitative polymerase chain reaction [PCR] used to detect circulating tumor cells and cell free tumor DNA) have been shown to improve patient survival compared to patients assigned to non-matched targeted therapy

. More than 60 commercial products are available using these techniques (e.g., Biocept Oncotype IQ®, Liquid Genomics®). The advantages of using plasma-based testing over traditional tissue biopsy’s sampling (i.e., technique limitations include obtaining inadequate biopsy samples, potential procedural complications and costs) in future clinical studies include the ease of obtaining samples, faster reporting of results, and the ability to acquire serial samples with greater sensitivity to monitor therapy response and detect changes earlier, as compared to traditional CT or MRI-based imaging assessments

. As these tests gain more clinical validity, as represented by FDA’s recent approval on June 1, 2016 of the first blood-based test to detect EGFR mutations as a companion diagnostic for treatment with erlotnib [Tarceva®], clinicians and researchers will be able to use rapid testing when making treatment decisions for newly diagnosed or progressing patients.
	
	Together with traditional anatomic imaging techniques (e.g., X-ray, CT, MRI, or US), which are very good at visualizing pathology or the extent of disease in terms of the number, size and overall appearance of tumors; the advance of imaging biomarkers (while using  molecular tracers and advanced imaging techniques such as PET, SPECT, MRI or MR Spectroscopy) with greater sensitivity or better spatial resolution, provide a noninvasive tool to monitor anticipated drug related effects on tumors. Receptor level imaging biomarkers can target cell proliferation, apoptosis, glucose metabolism, amino acid or lipid synthesis, etc. and are useful across all phases of drug development to evaluate disease distribution, disease response and biological characterization. The following examples demonstrate how tumor evaluation with imaging biomarkers may be used in research.
	
	Evaluating tumor angiogenesis and new vessel formation-related treatment changes may be imaged with perfusion CT, dynamic contract enhanced DCE MRI as well as PET imaging. Drugs that alter tumor cellular density may be quantitatively measured with diffusion weighted MRI (DWI MRI), which has been shown to be a sensitive predictor of overall response. Receptor level imaging can be performed with a variety of PET or SPECT techniques: cellular proliferation may be demonstrated with F-18 FLT PET, MR spectroscopy or DWI MRI and apoptosis may be evaluated with 18F- Annexin labeled PET or DWI MRI. The FDA has approved several molecular biomarkers such F-18 FDG, C-11 Choline, F-18 FACBC and many other tracers such as F-18 FLT, F-18 FDHT, F-18 NaF. F-18 FMISO, F-18 FETNIM, etc. are available in academic labs as well as in many countries outside the USA.
	
	Biomarkers provide the diagnostic tools and information necessary for risk stratification, therapeutic response prediction and monitoring of response. Over the last decade there have been significant advances to qualify and validate the data in larger multicenter cohorts which have accelerated our understanding of their utility and relationship to traditional response measures in the cancer treatment paradigm, such as PFS or survival outcomes. Research efforts will continue to improve data collection methods and standardized protocols so that the interpretability of these methods can be applied to a greater number of patients and the data may play a more significant role in improving personalized medicine by selecting the appropriate drug for the right patient to achieve better overall outcomes.

		Modernizing Eligibility Criteria for Molecularly Driven Trials, Edward S Kim, et al J Clin Oncol, 33: 2815-2820.	

		Using Multiplexed Assays of Oncogenic Drivers in Lung Cancers to Select Targeted Drugs, Mark G Kris et al. JAMA. 2014 May 21; 311(19): 1998–2006. doi:10.1001/jama.2014.3741.	

		Analysis of Circulating Tumor DNA to Monitor Metastatic Breast Cancer, Sarah-Jane Dawson et al, N Engl J Med 2013; 368:1199-1209, March 28, 2013, DOI: 10.1056/NEJMoa1213261.

Over the last decade, many initiatives sponsored by various entities, including academic and clinical research centers have focused on efforts to streamline clinical trial eligibility and data collection. Oncology trial design is no exception as endpoints and eligibility criteria have also changed with the value of the data generated in early phase studies.

Transformative Directions in Oncology Clinical Trials

	One of the fundamental risks to a clinical research study is the selection of study endpoints. Other risks may include data quality issues, missing data collection, inadequate site selection or training, or differences between clinical interpretations performed at the clinical site and at the independent core laboratory. This last risk may lead to ineligible subjects being enrolled into the study or censoring information. In oncology trials, censoring occurs when there is a difference between the investigator’s determination of the time of disease progression versus the core laboratory’s interpretation, resulting in a form of missing data. For example, the clinical site determines the subject’s tumor has progressed, removing them from study treatment and subsequent imaging assessments; whereas, the expert reader at the core laboratory determines the subject’s tumor did not progress and the subject should have remained on study treatment. One method to reduce this risk is to employ an independent image review of clinical trial data as presented below. It is useful to emphasize that these issues and approaches are analogous to methods employed in other aspects of clinical research as highlighted in Dr. Kosov’s analysis of the difference between site evaluation and independent expert assessment of adverse events in three lung oncology trials and concluded that overall agreement is low.

	Core labs use expert physicians who specialize in a specific therapeutic area and have extensive clinical research experience to serve as independent reviewers. Typically, the core lab uses the smallest number of independent readers possible to reduce variability, but there are many reader models that can be employed to best meet the study requirements. The independent reviewer can confirm a patient’s eligibility for the study to determine if the site investigators are correctly enrolling patients that meet the study’s inclusion and exclusion criteria. The core lab collects and checks subjects’ data from sites and one reviewer will assess each subject’s case for all sites, or several reviewers may divide the caseload. Given the advanced technologies employed at core laboratories today, confirmation reports can be provided to the clinical sites within very short turnaround times of one to two days. Independently checking patients’ eligibility can alter the study’s screen failure rates and enrollment timelines. An independent reviewer can also confirm the presence or absence of disease, which directly impacts therapeutic response, the assessment of a potential safety finding or appearance of new disease or lesions. For example, during the bevacizumab study, it was reported that approximately 10% of subjects’ baseline radiographic scans were missing. Due to this missing data and a very high rate of discordance (50%) between the site investigator and independent reviewer’s assessment of disease response, the FDA reviewers did not have confidence in the Progression Free Survival (PFS) results, as presented within the FDA Advisory Board’s Briefing Document.

 An independent review to confirm patient eligibility would have identified site enrollment or data collection issues--which may represent a potential site selection bias (e.g., enrolling subjects without required baseline image sets) or this review would have indicated that the site investigators were not interpreting the study requirements appropriately (i.e., under- or over-reporting the presence or absence of disease to meet eligibility requirements). Confirmatory reviews are typically employed to strengthen the study endpoints that can have a direct impact on study’s success and may focus on interpretation of complex clinical laboratory or blood-based biomarker tests, verifying safety requirements, reporting pathological specimens, etc. to improve the quality of data and reporting from clinical trials. 

	Clinical protocol designs may include an independent review to confirm if a clinical endpoint has been met (e.g., safety finding for cardiac ejection fraction falling 20% from the baseline value) or to confirm disease status as determined by the site investigator (e.g., confirming tumor progression during the trial). The protocol design would require an independent confirmation of the site investigator’s diagnoses before the investigator is permitted to alter or discontinue patients’ study treatment. Clinical trial sponsors frequently apply independent confirmation as a risk-minimization strategy in oncology programs. Confirmation of response is extremely important in new immuno-oncology therapeutic trials where desirable treatment results may be obscured by tumor growth or flare effect before showing a reduction in the overall tumor size. The independent review would help to eliminate data censoring by the clinical sites.

	An expert independent review of data can reduce differences often observed in clinical trials involving large numbers of international clinical sites. Clinical sites utilize a number of clinicians who serve as investigators and sub-investigators who perform patient evaluations. These clinicians have varied training and clinical experiences which impacts their clinical decisions and diagnoses. The sheer number of clinical assessors influencing study data may cause greater variability in clinical interpretations across a wide range of clinical sites. Whereas, an independent review involves a small, select group of experienced specialty physicians capable of providing higher-quality data with less variability in interpretation. This expert review is very beneficial to clinical trials as regulatory bodies often focus on data discrepancies (e.g., differing imaging interpretations between site investigators and independent reviewers) when evaluating a drug or medical device for market.

	Clinical trial sponsors can take advantage of a panel of independent reviewers who are not associated with any of the clinical sites, to monitor site compliance with the study protocol and check for consistency across sites to ensure patient assessments are thorough and complete, as well as provide a real-time data review to improve overall data quality. The core lab checks the quality of data as it’s received from sites and works with sites to improve image quality, if necessary, to ensure the images are readable for the independent reviewers to assess. From a regulatory point-of-view, sponsors should take great care in describing the data review processes and how the data will be used and reported during the clinical trial in the study protocol

	Another example of operational study risk is the failure to have experienced evaluators at sites using the appropriate assessment criteria correctly during the course of the study. This can result in misinterpretations or misclassification of subjects (e.g., incorrectly deciding if a subject has an intervention related safety issue or treatment effect) which fundamentally leads to errors, data variability from site to site or discrepancies between site and central read results. The implementation of an independent review can minimize these issues and eliminate potential bias in sites’ interpretation, which may be based on clinical information, and may influence the disease assessment instead of the assessment being solely based on tests or imaging evaluations.

	Prospectively identifying and performing a risk-based assessment for the trial and implementing a plan to reduce those risks are important in a well-designed study protocol to produce high-quality data and improve overall study compliance. Clinical teams should focus on reducing study design complexity and crafting a well-designed protocol that is precise and carefully worded so that all clinical sites have a consistent understanding of the requirements and are able to generate high-quality data capable of standing up to the rigors of regulatory review for approval. When protocols require the use of novel clinical procedures, new response assessment methods, or employ clinical tests/scoring methods that are not routinely performed by investigators (i.e., investigators who lack experience in correctly applying these new methods), the study team should consider implementing an independent review methodology in the study.

	The added value of an independent review is that it will identify potential issues with clinical trial data early in the study and would allow the study team to monitor, intervene and reduce study risks. Study sponsors should select a risk-minimizing approach based on the importance of study endpoints and disease assessments for eligibility, safety and efficacy in reaching critical decisions for the clinical development program and understanding the level of data accuracy required to make those decisions. The independent review of data provides a rigorous, standardized process to reduce risks for the clinical trial. 

	Joseph Pierro, MD, is Chief Medical Officer of Biomedical Systems, and Kelley Atherton, MA, is a medical writer for Biomedical Systems.

		Maxin Kosov MD, PhD, et al. “Interexpert Agreement on Adverse Events Evaluation.” Applied Clinical Trials. Apr 21, 2016.  

http://www.appliedclinicaltrialsonline.com/interexpert-agreement-adverse-events-evaluation

		FDA Briefing Document Oncology Drug Advisory Committee Meeting, July 20, 2010. 

http://www.fda.gov/downloads/Adviso...rugsAdvisoryCommittee/UCM219224.pdf

The selection of study endpoints is one of many risks that a clinical research study can face. An independent review can help avert this by identifying potential issues within clinical trial data early in the study and allowing the study team to monitor, intervene and reduce study risks.

Independent Image Review as a Form of Risk-Based Monitoring

	In 1891, bone surgeon William B. Coley injected streptococcal organisms into a patient with inoperable cancer based on his belief that a patient’s immune system could be stimulated to attack the malignant tumor.  Dr. Coley and other physicians who used Coley’s Toxins reported excellent treatment results, and Coley is now known as the “Father of Immunotherapy.” [The Iowa Orthopedic Journal; Volume 26, page 154-158].

	Since that time, progress and innovation within oncology has rapidly accelerated from the basic treatment paradigm of finding the tumor, excising the tumor, and using doses of radiation or chemotherapy to the highest toxicity levels, to a paradigm that now includes immunotherapy and bio-genomics. For decades, physicians have known that several tumor types, such as melanoma, renal and lung; are able to generate an immune response with better clinical outcomes. Only several years ago, researchers proclaimed cancer immunotherapy as the scientific breakthrough of the year in Nature and Science magazine articles; thereby, establishing immunotherapy as one of the core components in the armamentarium to treat cancer. Recent initiative--such as National Cancer Moonshot, led by Vice President Joe Biden--will serve to further accelerate cancer research and improve patient care while focusing on breaking down clinical study barriers and sharing information. 

	The medical research community widely acknowledges that the oncology drug development process is slow and costly, typically taking 14 years from the time of discovery to drug approval and only 5% of drugs tested in Phase I clinical trials will ever reach approval.

	Traditional study designs may not be optimal for personalized therapy. The objective of this new era of precision medicine coupled with advancing technologies is enabling new trial designs that match patients to the right genomic- or immune-targeted drugs. Randomized clinical trial designs are utilized to remove potential biases and have been the gold standard for demonstrating safety and efficacy to regulatory bodies; however, design limitations may not result in efficient drug testing strategies. For example, if you enroll patients with breast cancer without first selecting the tumor and comparing several drug therapies, the efficacious drug will be the one that hits more common targets and other treatments will appear less effective due to this design limitation. Had these latter drugs been tested in a targeted or selected population, their efficacy potential could have been better evaluated. 

	Oncology trial design is moving from a drug-centric approach, which uses common features between patients in terms of a type of cancer or type of molecular aberration (e.g., anaplastic lymphoma kinase [ALK]) and places all patients on the same drug treatment, to a more customized patient-centric approach. It is this realization that even if all patients have one common aberration, the rest of the molecular portfolio will differ from patient to patient and the patient-centric approach or strategy with customized drug combinations targeted to the molecular profile selected for specific patients that fit the targeted drug regimen (i.e., molecular matching) and not solely based on a drug.

	Protocol designs have evolved from large master protocols that include multiple tumor types, multiple targets, multiple treatments and typically end up only including a small number of patients likely to have a response. This approach has led medical researchers to move toward designing smaller umbrella or basket trials with patients selected based on genomic targets and more likely to respond to targeted therapy. Umbrella trials, such as the SWOG LUNG-MAP trial, are based on including a single histology (e.g., lung, breast, colorectal) and matching patients with different molecular aberrations to various drug combinations using predefined algorithms. Basket trials, such as the National Cancer Institute-Molecular Analysis for Therapy Chose (NCI-MATCH) trial, matches patients based on the same molecular aberration profile (e.g., BRAF) and allows different tumor types to be enrolled; this is useful for finding signals related to the functionality of the aberration and treatment response irrespective of histology. It is important to realize that umbrella and basket trial designs can efficiently address multiple questions under the auspice of one protocol.

	Regulatory agencies, pharmaceutical companies and public groups have been actively working together to develop approaches that simplify the drug development process to enable patients earlier access to therapies and improved patient care. As these new trial designs focus on selecting the right patients, members of various oncology groups, such as the American Society of Clinical Oncology (ASCO), have met with the FDA to discuss simplifying clinical trials. Some of the proposed initiatives include lowering regulatory hurdles to allow earlier patient access to trials, accelerating approvals, improving informed consent requirement, use of biomarkers, requirements for safety and efficacy of therapeutic combinations, and steps to simplify compassionate use of therapy drugs. These groups plan to release several instructive white papers later this year and their cooperative efforts are critical to advancing cancer care as exemplified in the accelerated approval of Keytruda® (pembrolizumab) based the results of the KEYNOTE-001, a biomarker-based Phase IB trial.

	Another evident change in the oncology landscape is data sharing and the collaborative participation of multiple pharmaceutical companies and oncology groups to offer a variety of available targeted therapies in a single clinical protocol. Other innovative approaches from pharmaceutical companies and publicly sponsored clinical trial groups, such as the National Clinical Trial Network which is comprised of NRG Oncology Group, American College of Radiology Imaging Network (ACRIN), SWOG, ACTION (Alliance for Clinical Trials in Oncology), Children’s Oncology Group (COG), Eastern Cooperative Oncology Group (ECOG) and SWOG (formerly the Southwest Oncology Group); that are focused on generating new hypotheses that will define informative signals for identifying patient tumor characteristics and matching treatments according to patients’ genomic signatures and determining those patients most likely to benefit with improved outcomes.

	Collaborative efforts such as these are essential to helping oncologists identify patient subsets, using leukemia cell banks established by cooperative groups over the last 30 years, from patients who participated in past Cancer and Leukemia Group B (CALGB) trials (now part of the Alliance of Clinical Trials in Oncology). The data from these well-designed studies with properly defined study treatments and documented clinical outcomes have provided the basis for the identification and understanding of various genomic subtypes, and the impact of these subtypes on prognosis in hematological malignancies (Mrozek, et al. Blood 2007, 109; 431-448).

	Experience has shown that cancer is difficult to treat due to the fact that cancer does not represent a single disease, but represents multiple genomic subsets. In order to treat cancer properly, this heterogeneous group needs to be sorted by subtype and treated with targeted agents shown to have effectiveness to the aberration.

	The genomic complexity of cancer defines the role of big data analysis and the future requirements for access to a super-commuting power capable of analyzing huge amounts of data compiled across past and future trial databases to establish “benchmarks” of the effectiveness of new treatment combinations that can be compared to the standard of care or other experimental regimes under consideration.

	The value of interrogating large databases was exemplified in a recent presentation by Dr. Maria Schwaerderle during ASCO’s Annual Meeting in 2015. She presented an analysis of more than 570 Phase II studies, including  32,149 patients, demonstrating that personalized-targeted therapies were associated with better outcomes than cytotoxic agents, which in turn were better than non-personalized targeted therapies (median response rates of 30%, 12% and 4% respectively) and therefore, the implications on study design can be readily understood. (Schwaerderle, et al., Impact of Precision Medicine in Diverse Cancers: A Meta-analysis of Phase 2 Clinical Trials; JCO 2015, 61, 5997).

	As institutions and oncology groups (e.g., ASCO, COP, SWOG) continue open collaboration and sharing of advances in clinical treatment protocols, the NCI established the Genomic Data Commons as a means to expand the cancer knowledge network by merging genomic and clinical data from an international community of cancer researchers into a unified data repository with standardized mapping and nomenclature that enables data sharing and analysis across studies in support of precision medicine.

	Cancer can be viewed as a complicated pathologic process and trial design innovation will continue to be molecularly driven, focusing on understanding the genomic aberrations. Future trial designs will evaluate more combinations of targeted drugs as patients are found to have more genomic aberrations. Collaborative approaches will continue to transform treatments in oncology and improve patient quality of life and overall survival rates.

	Joseph Pierro, MD is Chief Medical Officer with Biomedical Systems. Donald Cooper is Senior Director, Imaging Systems, with Biomedical Systems.

Progress and innovation within oncology has accelerated to a paradigm that includes immunotherapy and bio-genomics. Collaborative approaches in the future will continue to transform treatments in hopes of improving patient quality of life.

Future Directions of Cancer Immunotherapy in Clinical Trials

	In 1949, Norman J. Holter and Joseph A. Generelli, building on previous work, described a “radiocardiogram” and a “radio-encephalograph” in the 

, thus introducing a new technology to medical science. The goal of these recordings was to “….inquire into how various bodily functions are modified by exercise, excitement, emotion, or other activity….”

	Since then, the “Holter” recording methods have been improved by advances in miniaturization and recording technologies. The state-of-the-art in recording media has gone from magnetic tape to digital storage, with some recordings being transmitted to central processing centers using Bluetooth technologies, making almost real-time analysis possible. And, in the last decade, technology has advanced to the point where multiple-day recordings are possible.

	Use of single-day recordings reached its zenith prior to the publication of the Cardiac Arrhythmia Suppression Trial (CAST) findings. CAST tested the hypothesis that suppression of ventricular arrhythmias as seen on Holter recordings in patients with myocardial infarction would decrease mortality. The CAST study showed an increased mortality with treatment as opposed to placebo.

 After the CAST findings were announced, however, the use of Holter recordings was dramatically reduced, as the need for finding simple ventricular arrhythmias was attenuated.

	Up to this point, single-day data yielded a false sense of safety or concern, possibly because, as DiMarco noted, of the considerable day-to-day variability of arrhythmia and myocardial ischemia.

 The evanescent nature of arrhythmias was described by Corday, who compared a 10-hour Holter recording with a “conventional ECG”.

 What we have come to know about “normal” has evolved over the years as recording times have increased.

	The traditional way to describe “normal” on Holter recordings is to explain the abnormalities encountered. This includes any single or run of ectopic beats, rhythms, and sometimes included an ST-segment analysis or, more rarely, a QT-interval analysis. Judgments were made based on these ectopic counts and waveform descriptions. Additionally, normal healthy subjects might manifest benign arrhythmias not seen in other populations. These confounders make any description of “normal” difficult.

	As technology improved, recording periods were sometimes prolonged to two or three days.  Now, technology has advanced even further and recordings can be taken over weeks or months. It is, once again, time to rethink “what we know.”

	One of the challenges for those who write clinical trial protocols is to collect enough data to ensure that evanescent arrhythmias are not considered “drug-related.” This will most likely be done in two ways: 1) over-collection of screening and baseline data, 2) redevelopment of normal ranges.

	Due to the expense of clinical trials, cost-cutting has become very important in the drug-development process. The risk of cutting costs is that we collect homogenous data, and infrequent, otherwise unimportant abnormalities might take on an unwarranted significance. Indeed, the goal is to avoid “problematic” data, or data that might be considered an adverse event. An alternative to collecting limited data is to over-collect baseline data. Certainly, if we have 30 days of Holter data at baseline, we might be in a better position to determine a subject’s “normal” profile.

	From long-term data sets, new normal ranges can be developed. If it’s known that sinus pause of three seconds or 5-beat runs of VT are not uncommon, perhaps seeing these events in the drug-development process would not be considered an adverse event warranting further, expensive follow-up. The logical extreme is that some good drugs might not be killed in the development process.

	To conclude, technology has advanced enough that Holter’s original vision can be expanded to include time as a new variable. As technology changes, so must the prism of our analysis. What we thought was abnormal might now be considered normal or at least a normal variant. We need to collect more robust data to answer these questions.

, PhD, is Chief Scientific Officer, Biomedical Systems; 

, MD, is Chief Medical Officer, Biomedical Systems

	1. Holter NJ, Generelli JA. Remote Recording of Physiologic Data by Radio. 

	2. The Cardiac Arrhythmia Suppression Trial (CAST) Investigators. Preliminary Report: Effect of Encainide and Flecainide on Mortality in a Randomized Trial of Arrhythmia Suppression after Myocardial Infarction. 

	3. DiMarco JP, Philbrick JT. Use of ambulatory electrocardiographic (Holter) monitoring. 

	4. Corday E, Bazika V, Lang TW, Pappelbaum S, Gold H, Bernstein H. Detection of Phantom Arrhythmias and Evanescent Electrocardiographic Abnormalities. 

With significant technology advances in ECG recording through the years, the need to expand on traditional monitoring baselines-and include new variables such as time when designing clinical trial protocols-is important.