Transforming the Early Drug Development Paradigm at the National Cancer Institute: The Formation of NCI's Experimental Therapeutics Clinical Trials Network (ETCTN)

The National Cancer Institute's (NCI) early therapeutics program has substantially contributed to developing anticancer agents, having filed more than 900 Investigational New Drug (IND) applications since the 1950s (1). By the mid-2000s, however, NCI recognized that its predominant drug development model—using single-arm trials answering a single question at a single research site—was inefficient, costly, and poorly positioned for the era of precision medicine, where greater collaboration among institutions is needed to identify and enroll patients with specific molecular profiles (2–5). In March 2014, NCI overhauled its early therapeutics program and launched the Experimental Therapeutics Clinical Trials Network (ETCTN), a highly integrated network emphasizing team science (6). The creation of the ETCTN was part of NCI's larger efforts to modernize and streamline the Institute's clinical trials enterprise, which also included the transformation of NCI's late-phase trial cooperative group program into the National Clinical Trials Network (NCTN).

The NCI's vision for the ETCTN was a consolidated, national network of academic cancer centers to facilitate accrual to early-phase trials of agents that have been approved by NCI for collaborative development between NCI and industry. These studies address molecularly defined patient subsets to address clinical and scientific questions not evaluated by the private sector. The ETCTN relies on project teams of basic, translational, and clinical investigators who devise and execute drug development plans, with a major emphasis on including early career investigators alongside senior mentors. NCI provides centralized systems to construct protocols, enroll patients, collect and manage data, report adverse events, and bank tumor tissue, all of which allow multiple sites to function as a network. This perspective provides a brief overview of the ETCTN, summarizes its successes and challenges over its first grant funding cycle, and discusses the program's future directions.

The ETCTN is housed within NCI's Cancer Therapy Evaluation Program (CTEP) in the Division of Cancer Treatment and Diagnosis (DCTD). It is funded by cooperative agreements with 12 lead academic organizations (LAOs) that are responsible for scientific leadership and accruing patients. Among LAOs, there are 37 grant principal investigators (PIs) who are international leaders in cancer drug development. Each LAO has its own subnetwork: the main member (MM; i.e., LAO grantee), other institutions within the MMs' financial and governance structure, and affiliated organizations (AO) external to the MM (AOs, N = 28; median = 2 per LAO; range: 1–5). LAOs oversee the activation and conduct of NCI's early-phase therapeutic trials, comply with required policies and regulations, and activate and support accrual to each other's studies (1). Each LAO has an annual minimum accrual benchmark goal per grant year (GY): 50 patients for phase I trials and 72 for phase II.

Prior to the ETCTN, the early drug development process was funded via UO1 grants for phase I trials and NO1 contracts for phase II trials. In 2014 with the ETCTN, funding switched to UM1 grants for phase I trials but continued with the NO1 contracts until their scheduled end date; at that point, phase II trials were funded via supplements to the UM1 grants. With the second ETCTN grant cycle, there will be a single UM1 grant for phase I and II trials, and U24 grants for pharmacokinetics. The annual approved budget for the ETCTN UM1 cooperative award program, which included funds to complete the legacy phase I and phase II programs, was approximately $20M. There were additional contract costs for centralized clinical trial support services.

The ETCTN supports a robust portfolio of trials in partnership with LAOs, provides access to novel agents from industry, acts as an honest broker for drug combination studies (7, 8), and assumes regulatory responsibility for the trials as IND holder. NCI supports two pathways for new protocol development within the ETCTN: project teams and unsolicited trial letters of intent (LOIs). LOIs submitted by early career investigators (<7 years from oncology training) have a competitive advantage in the prioritization process to decide which studies to perform. Unsolicited LOIs are most typically received by investigators within the ETCTN and less commonly received by those outside the network, though all will be considered for review.

CTEP has a broad portfolio of anticancer agents available through agreements between NCI and industry (7). CTEP serves as sponsor for these agents' IND applications. Novel agents are added to the portfolio through NCI's Experimental Therapeutics (NExT) program pipeline (1). NExT provides resources to researchers developing therapies for unmet needs—agents enter NExT at various points in the discovery and development pipeline, and when ready for clinical evaluation, a project team is formed.

A critical expectation of LAOs is to lead and/or participate in multidisciplinary, multi-institutional scientific project teams to design initial drug development plans for agents approved by the NExT program (3). CTEP and LAOs colead teams; members apply via a streamlined project team member application that includes basic, translational, and clinical extramural scientists and early career investigators. Participation requires phone meetings over an average of 6 to 8 weeks and culminates in a drug development plan to submit to NCI's Investigational Drug Steering Committee (IDSC) for review. After IDSC evaluation, team members submit LOIs to conduct the proposed studies in the plan. LOIs follow CTEP's protocol development processes set forth by NCI's Operational Efficiency Working Group (OEWG) guidelines (9).

Investigators submit unsolicited LOIs to examine novel hypotheses that are supported by preclinical data for any agent in the CTEP portfolio (10). These LOIs cannot compete with studies put forward by a project team and may be submitted only after the approval of a team's plan. ETCTN investigators submit unsolicited LOIs that follow the OEWG guidelines (9, 10) for review by CTEP's Investigational Drug Branch.

Because biomarker assay development is central to the ETCTN's mission, NCI implemented several mechanisms to ensure that assays are analytically validated and fit-for-purpose for their intended use and that allow for the banking and accession of tissues for future research (see Table 1).

Agents under development in the ETCTN are chosen through a peer-review mechanism through NCI's Developmental Therapeutics Program and then approved by the NCI DCTD Senior Advisory Committee, so that agents under development in the ETCTN are selected independently of ETCTN program staff. The drug development activities of the ETCTN are reviewed by the external IDSC that makes recommendations of all initial drug development plans formulated by drug project teams. Both ETCTN grantees and NCI program staff meet with the IDSC quarterly to review proposals from the project teams.

As part of the larger effort to streamline NCI's clinical trial enterprise described earlier, NCI recognized the need to develop and implement standardized procedures, processes, and tools to support centralized clinical trial oversight—from LOI concept review through clinical trial activation, patient accrual, auditing, monitoring, closing a trial, and patient safety oversight. To do so, NCI developed a centralized infrastructure to support all ETCTN and NCTN clinical trials; Table 1 provides a summary of the components of this infrastructure.

Table 2 presents the metrics showing the growth of ETCTN's trial portfolio over its first five GYs. A total of 334 LOIs were submitted (range: 49–80 per year), with the LOI approval rate increasing substantially from 20% in GY1 to 57% by GY5. Of the LOIs, almost half (42%) were led by early-career PIs. Since GY1, 19 project teams were initiated; project teams required an average of 7.8 months between approval for project team and first LOI submitted for that project team's focus. From GY1 to GY5, 102 ETCTN protocols were activated, with the last three GYs averaging 27 trials activated per year. In addition, organizations increased the number of trials opened at their sites from an average of <1 trial activated in GY1 to an average of 10 activated in GY5. In five GYs, 3,570 patients were accrued by ETCTN organizations (Fig. 1). For GY1 and GY2, most accrual was to legacy trials (i.e., those trials activated in NCI's previous early development program that transitioned into the ETCTN). Accrual lagged for the ETCTN trials through GY2 as few had yet to open at sites due to the lead time required to approve and activate new protocols. As more ETCTN trials opened, accrual increased substantially with ETCTN trials accounting for the majority of accruals by GY3 (vs. legacy trials) and 97% of all accruals by GY5. With respect to productivity of the ETCTN network, more than 80 agents have an active CTEP agreement and at least 100 manuscripts related to ETCTN activities were published by the end of its first grant cycle. More information on the agents can be found at: https://ctep.cancer.gov/industryCollaborations2/agreements_agents_table.htm.

The ETCTN program differed substantially from the predecessor program; therefore, it was important to continually seek input from ETCTN investigators about their satisfaction with the ETCTN and identify obstacles to opening and accruing to trials across the network. Two methods were employed: (i) an online satisfaction survey fielded annually during the first three GYs (May: 2015–2017); and (ii) 60-minute phone interviews conducted with grant PIs in January 2017 (N = 35; of 37) midway through the first grant funding cycle.

We assessed the PIs' satisfaction with the ETCTN portfolio using an annual online survey. Invitations were e-mailed to all PIs who had actively engaged in the ETCTN over the GY(s). The survey included 13 core questions and took approximately 15 minutes to complete. The survey was considered exempt by the NIH's Office of Human Subjects Research, and responses were confidential with identities unknown to NCI. Statistical testing used SAS (v. 9.4), and we report P values with a two-sample t test on mean differences (P < 0.05 was deemed significant).

Across all 3 years, we achieved high completion rates: year 1, 81.4% (105 completed of 129); year 2, 98.7% (152 completed of 154); and year 3, 92.4% (171 completed of 185). Using a five-point Likert scale (1 = not at all satisfied; 5 = very satisfied), we found statistically significant increases from year 1 to year 3 in the PIs' mean satisfaction with the number of trials in the ETCTN portfolio (year 1: 2.36; year 3: 3.16, P < 0.01); number of trials available to open at each respective site (year 1: 2.51; year 3: 3.11, P < 0.01); representation of therapeutic classes (year 1: 2.74; year 3: 3.20, P < 0.01); balance of scientific areas (year 1: 2.94; year 3: 3.28, P < 0.01); and integration of preclinical findings (year 1: 3.33; year 3: 3.59, P = 0.03). Similarly, there were statistically significant increases from year 1 to year 3 in the PIs' mean level of agreement (1 = strongly disagree; 5 = strongly agree), with the following statements: “Enough ETCTN trials activated to meet center's accrual goals” (year 1: 2.23; year 3: 2.52, P = 0.05) and “ETCTN opens up careers opportunities for junior investigators at our site” (year 1: 3.49; year 3: 3.78, P = 0.05).

Interviews with grant PIs solicited PIs' greatest challenges in opening and accruing to ETCTN trials and recommendations that might improve network performance. Interviews revealed five critical challenges: (i) PIs wanted greater incentives to lead new protocols and open other investigators' trials; (ii) PIs needed more information and sooner about the ETCTN trial portfolio to plan their own center's trial portfolio and inform disease teams; (iii) PIs saw the unsolicited LOI review process as unpredictable and frustrating, requiring time-consuming decisions over minor details; (iv) PIs were frustrated with NCI's Central Institutional Review Board (CIRB) for the ETCTN (12), believing that too many changes were asked of them and comments seemed to be outside the CIRB's purview; and (v) PIs wanted NCI to shorten and streamline the OEWG timeline to approve LOIs and activate protocols sooner.

Following interviews, NCI implemented several changes over 6 months: (i) provided additional incentives to grantees such as credits toward an LAO's accrual quota for activating new studies and for opening other LAOs' studies, and authorship potential to PIs who signed up as a trial champion at their site; (ii) developed communication tools to increase PIs' awareness of the ETCTN portfolio (e.g., CIRB-approved handouts for patients, monthly PI e-mail newsletters providing trial updates by disease, interactive lists of ETCTN active trials); (iii) created a standardized decision form for LOI disapprovals; (iv) implemented a contract mechanism to author all ETCTN protocols for PIs with an approved LOI to ensure high-quality documents in a shorter time frame; and (v) reviewed the OEWG processes to address redundant steps and reduce the number of revisions asked of the PI.

A major ETCTN goal was to evolve a well-connected network among its LAOs and investigators to support NCI's early drug development research portfolio. Two social network analyses were conducted that show that ETCTN investigators were highly connected professionally from the onset and became a well-connected network over time with regard to accruing to each organization's active trials.

To assess professional connectivity, a question was added to the annual online survey described above for GY1 and GY2. Investigators were provided a list of all ETCTN organizations and asked to mark those institutions where they had direct collaborations with other early-phase cancer researchers over the past year (e.g., co-PI, publication, conferences) and to quantify the number of people they collaborated with at each organization and the frequency of their collaborations. NodeXL (12) was used to analyze respondents' interorganization collaborations over the program's first 2 years.

Network density (i.e., proportion of collaborations among all potential collaborations) indicated a high degree of professional collaboration from the onset: For GY1, 91% of all possible collaborations among organizations were detected, increasing to 97% for GY2, meaning nearly every organization was connected. Network centralization is the extent to which organizations are connected to multiple organizations—lower percentages signify multiple collaborations that are well distributed: centralization for both years was very low (GY1 at 11% and GY2 at 3.6%).

To assess accrual connectivity, we analyzed the degree to which organizations accrued to each others' trials for ETCTN's first 4 years. The annual number of active ETCTN trials (omitting legacy trials¹

Within its first 5 years, the ETCTN has become a highly collaborative network, activating more than 100 trials with the majority formulated via the project team process. Most of these studies are led by early-stage investigators, supported by senior mentors. Social network metrics indicate that the network is aligned for its scientific goals and for patient accrual to trials. Satisfaction measures indicate that network investigators increasingly approve of the ETCTN's scientific portfolio and key organizational challenges have been addressed. NCI's experience initiating the ETCTN teaches several lessons.

First, it was important to select sites with significant drug development expertise and where many investigators had long-standing interactions, likely increasing their willingness to consider a team science approach and enroll patients in trials led by others. Second, the ETCTN required substantial investments in a centralized infrastructure to allow real-time data capture of critical measures such as patient accrual and adverse events and timely interventions to change trial design when needed. Assuring active participation among early-career investigators and providing guidance for them through daunting, complex clinical trial development processes required additional resources. For example, the validation of predictive and prognostic biomarkers and their incorporation into early-phase clinical trials across a multisite network are enormously complex processes requiring substantive education of PIs (even those with high levels of drug development expertise) and sufficient resources. Finally, it was critical to provide mechanisms to clearly communicate and maintain high awareness of the network's portfolio (active and pipeline) among PIs because they need early notice to balance NCI's trials within their organizations' portfolio.

During the 2020–2025 award period, the ETCTN will continue to leverage NCI resources to enhance early drug development productivity and will focus on three priorities. First, NCI actively aims to address the need to recruit rare or uncommon molecularly defined subsets of patients through new requirements, such as requiring each organization to have a minimum number of identified key investigators responsible for disease-specific accrual and providing partial salary to support additional team members. Second, NCI hopes to ensure the collection of high-quality biopsy specimens, including providing partial salary support to include interventional radiologists and research pathologists on LAO teams. Finally, NCI will prioritize the use of biomarker assays to achieve precision medicine goals, with an increased use of NCI biomarker assay resources coordinated through the National Clinical Laboratory Network (see Table 1) while scaling back the UM1 biomarker assay development supplements and consolidating the ETCTN PK activities.

NCI's experience with the ETCTN shows that a national clinical trial network for early drug development is feasible but required a substantial investment of resources and talent. As testing new anticancer agents increasingly depends on the identification of biomarkers and patients with specific molecular characteristics, research systems need to adopt efficient, cost-effective, and sustainable strategies. The ETCTN model capitalizes on NCI's long-term support of a cancer center–based network of clinical researchers whose efforts can be marshaled by providing centralized services to efficiently and collaboratively design, implement, and complete early-phase oncology clinical trials.

J. Zwiebel has been an independent consultant to Boston Pharmaceuticals and Scandion Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: H.A. Massett, G. Mishkin, J.A. Moscow, J. Zwiebel, J.S. Abrams, S.P. Ivy

Development of methodology: H.A. Massett, G. Mishkin, J.A. Moscow, J. Zwiebel, S.P. Ivy

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H.A. Massett, J.A. Moscow, M. Kruhm, M. Montello, J. Zwiebel, S.P. Ivy

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H.A. Massett, G. Mishkin, J.A. Moscow, M. Steketee, J. Zwiebel, J.H. Doroshow, S.P. Ivy

Writing, review, and/or revision of the manuscript: H.A. Massett, G. Mishkin, J.A. Moscow, A. Gravell, M. Steketee, M. Kruhm, S.H. Han, M. Montello, J. Zwiebel, J.S. Abrams, J.H. Doroshow, S.P. Ivy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H.A. Massett, J.A. Moscow, A. Gravell, M. Kruhm, B. Barry, J.H. Doroshow, S.P. Ivy

Study supervision: H.A. Massett, J. Zwiebel