Critical importance of correctly defining and reporting secondary endpoints when assessing the ethics of research biopsies

Introduction

Research biopsies have been increasingly incorporated into early clinical trials of oncology agents with the goal of identifying predictive biomarkers for efficacy or pharmacodynamic biomarkers that will inform drug development. While predictive biomarkers may be used to define eligibility (and thus have the potential for direct patient benefit), more often they are investigated as secondary or exploratory endpoints to be correlated with efficacy. In contrast, pharmacodynamic biomarkers are only used to gain scientific insights into mechanism of action.

The American Society of Clinical Oncology (ASCO) and others are increasingly concerned about the ethics of research biopsies, especially those that are a mandatory condition of enrollment on a trial.^1,2 With the exception of biopsies used to define eligibility, these biopsies have no benefit to patients and, thus, are only ethical if they are supported by strong scientific justification. ³ Additional concerns center on research participants’ risk, discomfort, and inconvenience from undergoing research biopsies, and the potential for informed consent forms to mislead research participants about the scientific importance of biopsies and the possibility of direct personal benefit.^4–10 There are similar ethical concerns with nontherapeutic invasive research procedures included in clinical trials of other areas of medicine. ¹¹

In 2019, ASCO published an ethical framework for the inclusion of research biopsies in oncology clinical trials. ¹ While the focus of the framework is clinical trials for patients with cancer, the principles may apply to other areas of medical research. The framework lays out guidelines for the inclusion of biopsies based on whether the biopsies are an optional or mandatory condition of enrollment in the trial, the level of participant risk, and the potential scientific contribution of the biopsy. Assessments of the potential scientific contribution of biopsies are informed by the type of endpoint that the biopsy supports, with primary endpoints having expected scientific utility, secondary endpoints having potential scientific utility, and exploratory endpoints and biobanking having unknown scientific utility. Higher levels of participant risk are acceptable under the framework when a biopsy is likely to answer an important scientific question, such as a primary trial endpoint. Only lower levels of participant risk are acceptable when the scientific contribution is less likely, such as with secondary or exploratory endpoints.

The definitions of primary, secondary, and exploratory outcome measures (synonymous with “endpoints”) are included in the Code of Federal Regulations (42 CFR Part 11). The primary outcome measure is the endpoint of greatest importance and is the one that is usually used in a power calculation supporting the sample size of the trial. In contrast, secondary outcome measures are endpoints that are part of a prespecified analysis plan and not specified as an exploratory endpoint. Exploratory endpoints are all endpoints that are neither primary nor secondary. Furthermore, under the Food and Drug Administration Amendments Act of 2007 Section 801 (FDAAA 801), the methods and results for primary and secondary outcome measures in most clinical trials must be publicly reported via the website, ClinicalTrials.gov.

Implementation of ASCO’s ethical framework on research biopsies has been challenging. ¹² Investigators and sponsors may incorrectly classify biopsy endpoints as secondary when the science and study design only supports a designation of exploratory. This could be due in part to challenges in interpreting ASCO’s framework. Other reasons for incorrectly classifying biopsy endpoints may reflect confusion in the research community about the definition of a secondary endpoint, which according to the CFR requires a “prespecified analysis plan.” There are also strong incentives for the research community to justify the inclusion of research biopsies in clinical trials. Inaccurate characterization of biopsies is problematic because it may lead to an incorrect assessment of their scientific utility, thus, distorting the application of ASCO’s framework, complicating protocol evaluation by institutional review boards (IRBs), and leading to situations where participants are misled during the informed consent process and asked to assume an unacceptable level of risk. This misclassification may also put sponsors and investigators at risk for violations of Section 801 of the Food and Drug Administration Amendments Act of 2007 (FDAAA 801).

The purpose of this article is to review the implementation challenges to ASCO’s framework, including challenges related to the definitions of secondary endpoints, the scientific and regulatory framework, and the incentive structure that encourages inclusion of biopsies. Principles of research stewardship require that the clinical trials community correctly articulate the scientific goals of any research biopsies, especially those that are required for the patient to enroll on a trial and receive an investigational agent. Furthermore, it is important to sufficiently justify the characterization of secondary (as distinguished from exploratory) endpoints, protect the interest of research participants, and report accurate and complete information to ClinicalTrials.gov and the published literature. Overall, these activities further the goal of conducting “informative” research that is designed, conducted, and reported in a way that advances medical science. ¹³

Nomenclature challenges

Establishing a clear definition of secondary endpoints is critical to the implementation of ASCO’s framework and the accurate description of the scientific goals of research biopsies and the evaluation of their risks and benefits. ASCO’s ethical framework, ClinicalTrials.gov, and the National Cancer Institute (NCI) all use different terminology, which creates confusion.

ASCO’s ethical framework uses the term “secondary endpoint,” which it defines as having the potential to increase scientific knowledge based on a well-defined technical and statistical analysis plan. The framework states that biopsies that support secondary endpoints are embedded in trial protocols with clear scientific plans to generate knowledge based on analyses of the biopsied tissue, whereas 42 CFR Part 11 simply requires a prespecified analysis plan. These statistical analysis plans do not need to have the level of statistical power that would be expected for a primary endpoint but should provide a reasonable expectation of meaningful results. There should be sufficient evidence from prior studies to support the importance of the specific hypothesis being tested.

In contrast, NCI classifies biopsies by the type of biomarker study they support. Relevant here are biopsies that support integrated biomarkers, which are defined as those in which the “examination is prospectively planned, samples collected, and analysis is retrospective.” ¹⁴ All biomarkers assessed for secondary outcomes/endpoints are “integrated.” However, if there is not a predefined statistical analysis plan, then biopsies included in integrated biomarker studies do not meet the ASCO or CFR definitions for secondary outcome measures/endpoints. For example, some integrated biomarkers (e.g. whole genome sequencing) are only being assessed for exploratory endpoints.

Establishing the rigor for a predefined analysis plan for secondary endpoints is critical to justify the risks involved in the relevant research biopsy and to ensure high-quality translational research. It is straightforward to include basic descriptive statistical methods or hypothesis tests for endpoints utilizing biopsy specimens in protocols even for small sample sizes, such as 5–10 paired biopsies (i.e. before and after drug administration). However, simply providing statistical analysis language does not imply that meaningful evidence will be generated, especially when sample sizes are small. To be considered a secondary (vs exploratory) endpoint a minimally sufficient analysis plan should include justification for the sample size with support for the detectable effect sizes or, when the objective does not include hypothesis testing, that the sample size will provide a reasonable likelihood of achieving the objective. This will generally require preliminary data from the endpoints of interest (e.g. assay measures of similar specimens) with estimates of variance (including intraindividual variability for paired experiments) of the endpoint(s).

If these preliminary data do not exist, then it is not possible to obtain reliable estimates of variability. The analyses are then exploratory and the power calculations that support them are performative—they simply provide what may be “detectable” without any support that a significant or insignificant results would be meaningful (clinically or otherwise). For example, sample size justifications that provide the number of standard deviation units that are detectable based on a fixed sample size should be considered exploratory; these do not consider prior knowledge of meaningful effect sizes or variability. Practical issues should ideally also be addressed in statistical analysis plans and sample size calculations for biopsy-based data analyses, such as the chance of missing data due to patients leaving trials prior to planned biopsies, patients declining to undergo optional biopsies, potential for non-useable tissues, and problems with assays or processing/storage of biopsies. This will help ensure that at the end of the trial there is high likelihood of achieving robust data from biopsies that will generate sufficient evidence to have justified the risks to subjects.

Scientific and regulatory controversies

There are scientific controversies about the use of both predictive biomarkers and pharmacodynamic and mechanistic biomarkers. These controversies complicate implementation of ASCO’s framework and result in the inclusion of many biopsies of questionable scientific value in clinical trials.

Implementation challenges

Despite the publication of the ASCO framework in its Society journal, there has not been widespread uptake by investigators or pharmaceutical and biotechnology companies. Based on our experience, we suspect that a major reason is that research-grade tumor samples are valuable to companies for future investigations outside the trials being conducted; however, additional research is needed to confirm this opinion. Furthermore, the collection of samples provides opportunities for data mining that may lead to new directions for drug discovery and development, and in the best-case scenario, the identification of a reliable biomarker to predict efficacy of an investigational agent. Thus, many sponsors require that patients consent to one or more biopsies to gain access to the investigational drug. Institutions that wish to participate in these studies must allow such studies to proceed, regardless of whether the design is consistent with the ASCO framework.

Investigators are often supportive of the inclusion of biopsies, given the perception that interrogation of biopsy samples from patients enrolled in clinical trials will provide important insights and/or increase the publication value of study results. In fact, peer review committees may expect such analyses, without consideration of the plausibility of meaningful results. Investigators are also generally not concerned with the risks of biopsies because the same procedures are utilized in oncology care. For example, investigators generally understand that there is a significant risk of a pneumothorax with intrathoracic biopsies but recognize that this complication has a low mortality rate and can usually be managed with a chest tube thoracostomy. Furthermore, payers routinely cover the costs of complications from research procedures.

Finally, IRBs may be unaware of the ASCO framework and its implications for review and approval of early clinical trials. At academic medical centers with an NCI-designated Cancer Center, all clinical trials first undergo scientific peer review before IRB review. ¹⁹ In this context, IRBs may defer to scientific review, which may focus more on the rationale for the trial interventions and the importance of accessing investigational agents than the rationale for biospecimen collection. Furthermore, all clinical trials of investigational agents are subject to FDA review, leading to the potential misperception that the FDA has determined that any mandatory biopsy studies are scientifically appropriate.

Reporting to ClinicalTrials.Gov

FDAAA 801 and the 42 CFR Part 11 strengthened the legal requirement that sponsors register certain clinical trials of FDA-regulated products and report summary results to ClinicalTrials.gov. This is important in the context of biomarker studies because the results of these studies are frequently never published.^8,20–22 The information that must be reported includes the methods and results for all secondary endpoints, including their analyses plans. ²³ The reporting requirements do not cover exploratory endpoints. Nevertheless, miscategorization of exploratory endpoints as “secondary” in reports to ClinicalTrials.gov violates the FDAAA 801, as well as complicates assessments of the ethics of research biopsies under the ASCO framework.

Phase 1 trials are excluded from the registration and results requirements. Nevertheless, many early phase trials are reported to ClinicalTrials.gov in response to other policies and incentives intended to make trials more transparent. One such policy is the National Institutes of Health (NIH) policy on registering and reporting NIH-funded clinical trials in ClinicalTrials.gov. All NIH-funded awardees and investigators conducting clinical trials, funded in whole or in part by the NIH, are required to register the trial and submit summary results as specified in the FDAAA 801, including primary and secondary endpoints (but not exploratory endpoints). ²⁴ The policy is more expansive than those in the CFR and applies to all NIH-funded clinical trials regardless of study phase, type of intervention, or whether the trial is required under FDAAA 801. It applies to both extramural and intramural programs. This policy may result in many phase I trials being reported to ClinicalTrials.gov given the significant role of federal funding in early phase studies. ²⁵

Many clinical trials are also registered in ClinicalTrials.gov to preserve sponsors’ and investigators’ ability to publish their results in top medical journals. The International Committee of Medical Journal Editors (ICMJE) member journals require trials to be registered in a public registry to be considered for publication. ²⁶ The registration must occur before the initiation of the trial and include the definitions of the primary and secondary outcome measures. ClinicalTrials.gov is one of the few registries that meet the requirements described in the policy. Importantly, if a sponsor or investigator voluntarily submits information to ClinicalTrials.gov, such as a submission intended to meet the ICMJE policy, the responsible party must then meet all the 42 CFR Part 11 requirements for registration and reporting, including information about all secondary outcome measures. Despite these requirements, research suggests that many trials are published in journals without meeting the ICMJE’s registration prerequisite.^27,28

Consequences of noncompliance with 42 CFR Part 11 are potentially serious. FDA may impose civil monetary or criminal penalties.^29,30 In addition, noncompliance could result in limitations to federal grant funding. The Department of Health and Human Services requires grantees to certify on their grant or progress report forms that they have made all required registration and results submissions to ClinicalTrials.gov. If this is not done, any remaining funding for a grant or funding for a future grant to such grantee will not be released. Investigators are given notice of noncompliance and allowed 30 days to correct the situation.

Despite remedies for noncompliance with 42 CFR Part 11, there has been little enforcement by FDA, NIH, or ICMJE journals. Many trials are not registered in advance of starting patient recruitment and fail to meet the results reporting requirements.^23,31,32 There are documented discrepancies between information reported to ClinicalTrials.gov, peer-reviewed publications, and/or FDA reviews, suggesting the submission of inaccurate information to ClinicalTrials.gov.^33–35 Many clinical trials also fail to have publicly available protocols and statistical analysis plans as required by the CFR, which hinders the interpretation of the statistical validity of all endpoints.^36–38 Moreover, there is evidence that many biomarker studies are conducted without any protocol or without adhering to one that may exist. ³⁹

While reasons for lack of compliance with reporting and results requirements may vary, collectively this reflects the failure of the research community to meet our ethical obligation to conduct research transparently. There are early signs that the FDA is increasing its enforcement and efforts to address the challenges to compliance. FDA issued its first notices of noncompliance with 42 CFR Part 11 in 2021. ⁴⁰ FDA is also participating in a Clinical Trials Transformation Initiative to explore challenges and solutions to increasing compliance with 42 CFR Part 11. ⁴¹ Parallel to these efforts is the development of reporting guidelines for journal publications to improve peer-reviewed publications. ⁴² Most recently, there has been Congressional interest in improving compliance with Ranking Member Frank Pallone, House of Representatives Committee on Energy and Commerce, sending a letter of concern to FDA Commissioner Califf and Acting NIH Director Tabak regarding the “lack of compliance by medical product sponsors with requirements to report certain clinical trial results information to the ClinicalTrials.gov database.” ⁴³ Investigators should be aware of compliance requirements for accurate definition and reporting of all clinical trials endpoints, including those for research biopsies. Beyond increased legal scrutiny, this is a foundational part of our goal and obligation to conduct high-quality translational science on a strong ethical footing and in partnership with our patient volunteers.

In conclusion, the research community has obligations to correctly articulate the scientific goals of research biopsies within clinical trials, to justify characterization of endpoints, to protect the interest of research participants, and to report accurate and complete information to ClinicalTrials.gov. The ASCO framework provides guidance for determining acceptable risks of research biopsies, but it can be misinterpreted when endpoints are incorrectly characterized. To address this critical issue, the research community should educate investigators, institutions, and IRBs regarding the implications of approving studies that include mandatory biopsies. It should also encourage the Office for Human Research Protections (the section of the Department of Health and Human Services charged with protection research participants) and/or FDA to develop federal guidelines clarifying best practices for research biopsies. For studies conducted under an investigational new drug (IND) application, FDA should consider all risks of a clinical trial, not just the risks of the administered drugs. This would have the benefit of encouraging sponsors to be cautious when including mandatory biopsies in their studies. In addition, sponsors, regulators, and journals should follow and promote the accurate registration and results reporting of studies. With attention to these issues, we can answer important questions to advance clinical care on a strong scientific and ethical foundation.