Moving Forward in Cervical Cancer: Enhancing Susceptibility to DNA Repair Inhibition and Damage, an NCI Clinical Trials Planning Meeting Report

Abstract

Cervical cancer is the fourth most common cancer in women worldwide, and prognosis is poor for those who experience recurrence or develop metastatic disease, in part due to the lack of active therapeutic directions. The National Cancer Institute convened a Cervical Cancer Clinical Trials Planning Meeting in October 2018 to facilitate the design of hypothesis-driven clinical trials focusing on locally advanced, metastatic, and recurrent cervical cancer around the theme of enhancing susceptibility to DNA repair inhibition and DNA damage. Before the meeting, a group of experts in the field summarized available preclinical and clinical data to identify potentially active inducers and inhibitors of DNA. The goals of the Clinical Trials Planning Meeting focused on identification of novel experimental strategies capitalizing on DNA damage and repair (DDR) regulators and cell cycle aberrations, optimization of radiotherapy as a DDR agent, and design of clinical trials incorporating DDR regulation into the primary and recurrent or metastatic therapies for cervical carcinoma. Meeting deliverables were novel clinical trial concepts to move into the National Clinical Trials Network. This report provides an overview for the rationale of this meeting and the state of the science related to DDR regulation in cervical cancer.

Cervical cancer (CC) remains the fourth most common cancer of women worldwide, causing 500 000 deaths in 2018 (1). The incidence in the United States of 13 170 cases of invasive disease per year is lower relative to that in developing nations (295 989 new cases per year). Premalignant and early-stage disease remains highly curable, whereas the prognosis for women with advanced invasive, recurrent, or de novo metastatic disease remains poor. There were 4170 deaths in the United States because of invasive CC in 2018 (2,3).

There has been long-standing interest in identifying targetable molecular pathways for CC, stymied by the unrevealing results of the CC Tumor Cancer Genome Atlas (4). The National Cancer Institute (NCI) convened a Cervical Cancer Clinical Trials Planning Meeting (CTPM) in October 2018 titled “Moving Forward in Cervical Cancer: Enhancing Susceptibility to DNA Repair Inhibition and to DNA Damage.” This topic was selected to better understand and leverage the molecular causes and effects of the underlying human papillomavirus (HPV) infection: dysregulation of p53 and pRb, which are required for normal cell cycle and DNA repair functions. The primary objective of this meeting was to facilitate the design of clinical trials focused on the role of DNA damage and repair (DDR) in the treatment of CC. Highlighted below is the state of CC and of the HPV-driven molecular science as it relates to CC.

Current Approach to Treatment

The treatment of early-stage CC, International Federation of Gynecology and Obstetrics (FIGO) stages IA-IB2 (5), is surgical, with lymph node dissection considered in patients at sufficient risk for nodal disease. Definitive radiation therapy (RT) is reserved for patients who are unable to undergo surgery or have locally advanced disease. Open surgery is preferred over minimally invasive surgery for those women undergoing radical hysterectomy given recent findings indicating worse overall survival with minimally invasive surgery (6,7).

The role of adjuvant therapy depends on the presence of adverse pathologic risk factors. Adjuvant RT is recommended for findings of stromal invasion, larger tumor size, and/or lymphovascular invasion. Adjuvant RT with concurrent weekly cisplatin (CCRT) is indicated for women with pathologic lymph node involvement, positive margins, and/or parametrial invasion (8,9).

Locally advanced CC, FIGO stages IB3-IVA is treated with CCRT and brachytherapy (8,10–13). Treatment has been optimized with advanced imaging such as positron emission tomography (PET) and computed tomography (CT), and magentic resonance imaging (MRI). Advances in RT techniques, notably intensity-modulated RT and image-guided RT, have led to reduction in the risk of CCRT-related serious adverse events. This approach can yield a 3-year disease-free frequency, which appears to translate into cure in 40% (stage IVA) to 75% (stage IIB) of patients. Improvement is needed to prevent both local and distant disease recurrence, particularly among women with node-involvement.

Treatment for women with metastatic, persistent, or recurrent CC whose disease is not amenable to potentially curative pelvic exenteration, is palliative, and survival outcomes are poor. Cisplatin was established as an active single agent in 1981 (14). Response rates and progression-free survival improved when combined with a second agent such as paclitaxel; however, overall survival was not improved with a variety of tested doublets (15–17). Next was the addition of bevacizumab to combination chemotherapy in GOG 240, resulting in a 3.7-month improved overall survival (hazard ratio = 0.71, 97.6% confidence interval = 0.54 to 0.95, 1-sided P = .0035) and subsequent US Food and Drug Administration approval (18). The only other drug approval for recurrent CC is pembrolizumab, using the KEYNOTE-158 single-arm study reporting a 12% response rate (19).

Genomic Profiling of CC

HPV is well established in the etiology of CC. Its E6 and E7 proteins dysregulate p53 and pRb activity, respectively. Numerous somatic mutations have been identified in subsets of CC, including in PIK3CA, PTEN, TP53, STK11, and KRAS (20–24). Ojesina and colleagues (25) performed whole exome, transcriptome, and whole genome sequencing of 115 CC specimens, of which 79 were squamous cell and 24 adenocarcinomas. Substitutions were found in MAPK1 in 8%, leading to E322K protein mutation. Inactivating mutations were found in HLA-B in 9% as well as mutations in EP300 (16%), FBXW7 (15%), NFE2L2 (4%), TP53 (5%), and ERBB2 (6%). Somatic ELF3 mutations (13%) and CBFB mutations (8%) were found in the adenocarcinomas.

The CC Tumor Cancer Genome Atlas molecularly characterized 178 primary CC samples, most of which were early stage and may offer a skewed view of the breadth of the genomic events of invasive disease (4). This study confirmed the presence of the previously described CC mutations and identified new genes of interest, including ARID1A, SHKBP1, ERBB3, CASP8, HLA-A, and TGFBR2. Amplifications in the immune target genes CD274 (encoding PD-L1) and PDCD1LG2 (encoding PD-L2) were discovered. Averages of 88 somatic copy number alterations per tumor were reported. With the exception of HER2 mutation or amplification, and modulation of the PI3K/PTEN/AKT and MAPK pathways, none of these mutations were identified as immediately actionable.

DNA Damage as a Target

There are both endogenous and exogenous sources of DNA damage. Endogenous sources include DNA replication stress, cell metabolism yielding reactive oxygen species (ROS), and normal mutational error; exogenous sources include radiation, viral infection, hypoxia and other microenvironmental changes, and chemical exposures, such as chemotherapy (26). The fate of an injured normal cell is senescence or death if the damage cannot be repaired. Thus, DNA damage can be direct or indirect. One mechanism of indirect damage is through inhibition of DNA repair pathways. There are an estimated 55 000 DNA single-strand breaks within the 3.2 billion bases in the human genome in normal cells on a daily basis. The cell has more than 6 major mechanisms for DNA repair at different stages of the cell cycle. Some of these may be less available to the malignant cell depending on its genomic alterations.

Radiation and the CC Genome

Radiation causes double-strand DNA breaks (DSB) through both direct and indirect DNA damage, the majority of which is via generation of free radicals leading to ROS. ROS generation can be reversed by presence of glutathione or other reducing agents (27). Cisplatin is a known radiosensitizer acting by creating interstrand DNA crosslinks leading to DNA DSB because of topographical restrictions.

Insufficient repair of DNA damage leads to cell death (28). Cells initiate DDR signaling following DNA damage, leading to cell cycle arrest, repair, cell senescence, or death. Cells normally monitor for DNA damage and, when identified, initiate DDR signaling. The first step in DDR is usually induction of p53 to cause cell cycle arrest, which then allows repair. Cells undergo senescence or death when they cannot successfully repair or if they are overwhelmed with damage (Figure 1) (29). Inadequate repair of DNA damage can also propagate mutation into subsequent cell generations.

Figure 1.

Simplified representation of the DNA damage response pathway (29). ATM = ataxia telangiectasia mutated kinase; ATR = ATM and Rad3-related kinase; CHK1/CHK2 = cell cycle checkpoint proteins.

Promotion of ROS, and thus DNA damage, may provide therapeutic opportunities in the treatment of CC. This may be achieved through increased glucose metabolism in cancer cells, inhibition of glycolysis and ROS, and inhibition of glutamine metabolism. Glutaminase is an amidohydrolase that converts glutaminate to glutamine and is an integral component of glutathione. A novel glutaminase inhibitor has multiple therapeutic mechanisms and preclinical data to support testing in CC treatment (27,30). Glutathione, or other reducing agents, can promote reversal of radiation or platinum-induced DNA damage or the damage can be stabilized by oxygen following indirect DNA damage. A fresh approach to evaluating radiosensitization and the agents used may be needed to improve the results in treating locally advanced CC with RT.

DNA Repair Dysfunction and the Tumor Microenvironment in Cervical Cancer

HPV and DDR

HPV alters DDR to promote its replication and to promote its viral life cycle. There are several mechanisms through which HPV manipulates DDR: 1) inhibition of innate immunity and DDR interactions, 2) modulation of the cell cycle to an S-phase state, 3) use of DDR to promote viral replication, 4) prevention of senescence, and 5) prevention of viral apoptosis (31). The many effects of HPV proteins are described in Table 1 (32) and Figure 2 (33). E6 and E7 are the most important of the HPV proteins with their binding and inhibition of p53 and pRb, respectively. The result of this is hijacking of the cell by HPV-activating DDR mechanisms for HPV self-preservation. This causes secondary inhibition of DDR; locks the cell in G1/S, allowing HPV replication; and prevents cell cycle regulatory functions that normally occur when DNA damage is present. HPV-related tumors are sensitive to DNA-damaging therapies, most notably radiotherapy and platinum-based chemotherapy regimens.

Table 1.

Mechanisms of action of HPV proteins^a

Protein	Interaction	Consequence
E1 helicase	Induces DSB to activate ATM	S/G2 arrest
E1/E2	Recruit DDR proteins	Viral replication
E7	Increased Rad51, BRCA1	Delayed RT repair
E6	Bind XRCC1	Impaired SSB repair
E6, E7, E5	Inhibit TGF-β signaling	OIS blocked
E7	Increased phosphor-ATM, CHK2	Viral replication
E7	Increased phosphor-ATM, CHK1	Viral replication
E7	pRb	Disrupt G1/S checkpoint
E6	p53	Disrupt G1/S, apoptosis
E7	FA pathway activation	Viral replication
E5	Decreased FasL and TRAIL	Inhibit ER-stress apoptosis

^aTable reproduced with permission from Dr M. Gillison (32). HPV proteins and DDR interactions. ATM = ataxia telangiectasia mutated kinase; CHK1/CHK2 = cell cycle checkpoint proteins; DDR = DNA damage response; DSB = double-strand breaks; ER = endoplasmic reticulum; FA = Fanconi anemia; HPV = human papillomavirus; OIS = oncogene-induced senescence; RT = radiation therapy; SSB = single-strand breaks; FasL = Fas ligand; TRAIL = tumor necrosis factor-related apoptosis-inducing ligand.

Figure 2.

Human papillomavirus (HPV) genome and functions of associated proteins. Figure reproduced with permission from Dr M. Gillison (33). URR = upper regulatory region of the viral genome.

Not all HPV-associated cancers are the same. Head and neck squamous cell carcinomas can be either HPV+ or HPV−; these are different in mutational signatures, copy number variations, and response to cetuximab (34). Mutations in genes, such as PIK3CA and PTEN, and ATM amplification and ATR deletion are found and may be associated with increased tumor mutational burden, making these HPV+ head and neck tumors potentially targetable with DDR and immune-oncology agents. Virus-host interactions in HPV+ tumors create this unique genetic profile that is markedly different than HPV− tumors but may be unique to head and neck squamous cell cancers (35).

Targeting the DNA Damage Response in Combination With Radiation

Both DNA repair and cell cycle checkpoints are vital to halt proliferation to allow repair during normal DDR activity, where they pause DNA synthesis and initiate DNA repair. Different DNA repair pathways are active at different times in the cell cycle. The mechanism in use is determined by when in the cell cycle the cell shuts down for correction of DNA damage, preservation of DNA integrity, and cell survival. This is a critical component of normal tissue response to ionizing radiation and other exogenous and endogenous forms of DNA damage. HPV disrupts the G1 checkpoint by E6/E7 protein inhibition of normal p53 and pRb function. This causes the tumor to behave as p53/RB mutant cancers where the G1 checkpoint is abrogated. This targets S or G2 phases as opportunities for selective sensitization to DNA damage (26). There are many agents that target proteins in the DDR pathways, at G1/S or G2, to prevent DNA repair and promote replication stress, resulting in tumor cell death in which activity has been shown in cancers where the p53 checkpoint is lost through mutation and potential for radiosensitization has been observed. These are potential resources for examination in CC.

Inhibitors of proteins active in regulating G1/2, such as PARP1/2, ATM, and those that regulate G2 differentially, CHK1, WEE1, and ATR, are all potential targets for radiosensitization of p53-defective cancers. Data suggest that this may occur via inhibition of homologous recombination, the high-fidelity repair pathway that is predominant in G2, but may also be active in G1 (36–38). Inhibition of these enzymes causes replication stress because of upregulation of CDK1 and cell progression through the cell cycle, propagating DNA injury and with exhaustion of nucleotides required for repair (39,40). These targets may sensitize to both chemotherapy for improved overall tumor control and radiotherapy for improved locoregional control. The WEE1 inhibitor has been most studied and demonstrates sensitization to both chemotherapy (gemcitabine) and radiotherapy (36,40).

Strategic combinations of DDR regulators may have a greater therapeutic window than when combined with targeted agents, chemotherapy, or radiotherapy (41). Clinical trials are underway exploring these combinations of DDR regulators across a variety of tumors, and early work has been initiated in CC. Such combinations may also take advantage of different microenvironmental elements to augment tumor damage.

Damaged DNA is released on cell death and can be found both in circulation and in the cell cytoplasm. Cytoplasmic DNA can trigger the cyclic GMP-AMP synthase–STING interaction, activating the type I interferon pathway to stimulate the immune system. Free DNA also may itself be immunogenic and recognized by the innate immune system (42–44). HPV E7 blocks STING and can promote suppression of the innate immune system response (45). Therapeutic immune modulation has been shown to collaborate with DDR regulators and radiotherapy in a variety of solid tumors, such as lung cancer. RT can increase the expression of PD-L1, which may improve activity of anti-PD1 or PD-L1 therapy (46,47). Poly-(ADP)-ribose polymerase (PARP) inhibitors cause DNA damage and replication stress and may increase PD-L1 expression (48). Studies are ongoing in other solid tumors to examine if combinations of radiation or other DDR targets and immuno-oncology therapies yield benefit over either agent, which may enhance the efficacy of PD-1 or PD-L1 inhibitors.

Targeting Hypoxia and the DNA Damage Response in CC

Hypoxia occurs as a spatial and temporal gradient across cells in all tissues and contributes to malignant progression, development of metastases, and resistance to therapy. This gradient may be steeper in cancer, where high tumor cell oxygen consumption depletes the oxygen supply. Tumor cells may also be more distant to the oxygen supply as a result of poorly designed vasculature. For these 2 reasons, among others, sections of tumors are often chronically hypoxic. Hypoxia has several implications in CC. Hypoxia is proapoptotic, which may result in selection for hypoxia-tolerant cells. It also activates cell survival pathways, including promotion of angiogenesis, glycolysis, and the unfolded protein response (49,50). Hypoxia modulates DDR through loss of G2 checkpoint inhibition and decreased expression of key proteins in homologous recombination DNA repair (51).

Radioresistance can be borne out of either acute or chronic hypoxia; hypoxia mitigation has been examined as a method of radiosensitization. Preclinical data demonstrated that chronic hypoxia with reoxygenation before radiotherapy resulted in increased radiosensitivity (52). Unfortunately, clinical results of hypoxia-targeting therapies have been disappointing (53–55). The randomized trial of cisplatin and radiotherapy with or without tirapazamine for locally advanced CC was negative (53), and several other studies showed no improvement in clinical outcomes with hypoxia-activated prodrugs (56). There are many potential hypotheses why these studies failed to show a benefit, including failure to properly select patients with hypoxic tumors including spatiotemporal variation in hypoxia, poor drug distribution and/or activation, and limited sensitivity to alkylating agents.

The CXCL12 (ligand) and CXCR4 (receptor) chemokine pathway plays a key role in the hypoxic microenvironment and is a promising target. It can be inhibited by plerixafor. Preclinical data demonstrate that the most optimal tumor response is when plerixafor is sequenced to follow CCRT, compared with cisplatin or radiotherapy alone, or concurrent with CCRT. Plerixafor was observed to attenuate late radiation enteric changes, which suggests that the agent may also protect intestinal crypt cells from radiation injury in animal models (57,58). Targeting hypoxia-dependent biology, including the CXCL12/CXCR4 pathway, may have more promise with RT than using hypoxia-activated prodrugs. Modifying the microenvironment to improve the therapeutic index of treatment, as what may be happening with CXCR4 inhibition, is a novel way to leverage DDR approaches in CC.

Building on DNA Damage and Repair Inhibition With Novel Treatments

Most of the many DNA repair pathways are interdependent. For example, although homologous recombination repair requires the BRCA 1/2 pathway proteins, deoxyribonucleotides (dNTPs) are still necessary for synthesis of the complementary strands. Thus, maintenance of the dNTP pool remains a critical element of DNA synthesis and repair.

Ribonucleotide reductase (RNR), a key enzyme required for DNA synthesis and repair, is regulated by p53 in normal cells. HPV-infected cells lose p53 because of its downregulation by HPV-E6, functionally hindering normal p53 activity. This causes increased activity of RNR, creating additional dNTPs. RNR is induced by ATR and is more active at the site of DSB, where nucleotide demand peaks soon after the cell recognizes the presence of DNA damage. The RRM2 subunit of RNR has been shown to be upregulated in CC and is prognostic of a worse outcome (59). These findings support the rationale for combining triapine, a very potent and relatively nontoxic RNR inhibitor, with RT or CCRT for treatment of CC (GY006, NCT02466971).

Many new DDR-targeted agents that may prove efficacious when combined with chemotherapy, RT or CCRT, or other targeted agents are becoming available for use in CC. DNA-PKcs is a DNA-dependent serine/threonine-specific protein kinase that detects DNA DSBs and signals activation of nonhomologous end joining DNA repair; oral agents that inhibit DNA-PKcs are now available for clinical and preclinical examination. Inhibition of WEE1 leads increased replicative stress, promoting DNA DSB and pushing cells into mitosis with unrepaired DNA, leading to potential mitotic catastrophe. Similarly, ATR inhibition promotes replication stress, leading to double-strand breaks. Preclinical enhanced activity and potential to overcome PARP inhibitor resistance has been demonstrated with the combination of these agents with PARP inhibitors that prevent repair in G1 (60). The are many agents targeting key DDR components that may be applied to combination therapy for CC (61).

Novel Approaches Involving RT

Advancements in RT have focused on advanced imaging, novel applicators, and techniques to better identify and deliver radiation dose to the anatomic target. Data from retroEMBRACE, a multi-institutional retrospective trial examining 3-dimensional image-based brachytherapy, demonstrate that local control of disease is excellent and is improved with image-guided dosing (62). Most patients receive brachytherapy at the completion of CCRT, and there is increasing use of volume-based planning, most commonly to the high-risk clinical target volume. Patients with a high-risk clinical target volume larger than 30 cc are expected to benefit most from a hybrid intracavitary or interstitial applicator to treat the anatomic target with relative sparing of the adjacent organs at risk (63). The ongoing prospective trial, EMBRACE II, is examining delivery of MRI-based adaptive brachytherapy to 3-dimensionally defined targets (64). Results of PET and functional MRI obtained during CCRT provide information prognostic for disease progression (65–71). The marked heterogeneity in how tumors respond to standard CCRT also argues for biologic heterogeneity (72,73). The question of biology can be addressed by consideration of CCRT combination strategies and radiation sensitization for the brachytherapy phase of treatment.

Opportunities include adding systemic agent(s) to synergize with platinum, changing agents based on response to therapy, or adding or changing systemic agent(s) at the time of brachytherapy based on response to primary CCRT. Application of molecular data to define a nomogram or score, as was done in breast and prostate cancers, may be useful to identify patients with greatest risk who might benefit from escalation of therapy. Additional work needs to be done so that our “set it and forget it” approach to CC acknowledges the heterogeneity of CC and its response to therapy.

Novel Drug Combinations

Dysregulation of DDR is a result of HPV infection and is confirmed by the demonstration that the most active agents in CC are radiation and platinum-based therapies. Inhibiting DDR may increase tumor mutational burden, enhance the immunogenic properties of cancer cells, alter the tumor microenvironment, and stimulate neo-antigen production (74). Determination of optimal agents for combination use is the next horizon.

Vascular endothelial growth factor (VEGF) and VEGF receptor (VEGFR) inhibitors, immunotherapies, and PARP inhibitors have demonstrated cytostatic or cytotoxic activity, some in combination with DDR-regulating agents. A phase I-II study tested the PARP inhibitor, veliparib, combined with topotecan in recurrent or persistent CC in the second-line setting. A response rate of 7%, with 37% stable disease, was reported (75). Cisplatin, paclitaxel, and veliparib were evaluated in the GOG 76HH phase I study of 34 patients, all of whom received prior radiotherapy (76). This regimen yielded an objective response rate of 34%, including 7% complete responses, 28% partial responses, and stable disease in 41%; progression-free survival was 6.2 months and overall survival was 14.5 months. CC has been shown to be an angiogenic disease. GOG-0240 resulted in an overall survival benefit from 13.3 to 17 months from the addition of bevacizumab to chemotherapy (18). Cediranib is a potent oral inhibitor of VEGFR1-3 tyrosine kinases. The potential value of angiogenesis inhibition was further supported by the benefit observed from the addition of cediranib to carboplatin and paclitaxel in the CIRCCa randomized phase II trial (77).

Viral infection has been proposed to lead to greater benefit with immunotherapies. The KEYNOTE-028 study included 24 CC patients whose tumor expressed PD-L1, 90% who had received prior RT, and 63% of whom had received 2 prior lines of chemotherapy. Pembrolizumab elicited an overall response rate of 17% (78). Tumor expression of PD-L1 was observed in 84% of the 98 CC patients in KEYNOTE-158 (19). Its overall response was 12%, resulting in US Food and Drug Administration approval for PD-L1 positive CC. Immunotherapy combinations are currently underway for CC. A phase I study combining the PD-L1 inhibitor durvalumab with the PARP inhibitor olaparib or with cediranib included 2 CC patients receiving durvalumab with cediranib (79). The combination of durvalumab and cediranib yielded a response rate of 50% and was found to reduce clearance of cediranib, obviated by reducing the cediranib dose and administering the agent 5 days of every 7. Such combinations warrant further investigation in CC.

Charge to the CTPM and Outcomes

This CTPM was charged with investigating the role of DDR in CC and recommending clinical trial concepts for development. Two subgroups of investigators conferred over a 6-month period before the meeting to focus on trial designs to develop novel agents and combinations, including RT combinations, and trials for treatment of recurrent and metastatic disease. The concepts were to emphasize strong preclinical science and rationale, safety considerations, feasibility, and statistical design targeting a clinically meaningful effect size. Inclusion of early career investigators promoted the NCI’s mission of career development.

The CTPM delivered concepts for 4 clinical trials addressing unmet needs at different points in the CC life cycle, emphasizing DDR-regulating agents in combination with chemotherapy and/or RT. Two concepts were proposed for the locally advanced CC population, aiming to add promising agents with strong preclinical data either concurrently or sequentially to CCRT and brachytherapy.

The concepts further recognized the importance of second-line therapy and beyond. Discussion was focused on the real-world application of the results of GOG 240. Participants noted that most physicians capped or limited the number of chemotherapy-containing cycles because of toxicity. CTPM investigators proposed development of a concept testing a switch-maintenance approach, continuing bevacizumab maintenance as the control or reference arm, and building on maintenance to allow earlier discontinuation of chemotherapy and to mitigate toxicity. Approaches to leverage new agents and opportunities for women with high recurrence risk or who recur after chemotherapy or bevacizumab led to ideas currently under preclinical development.

Additional work is needed not only to pursue the next generation of therapies for CC but also to deliver and provide access to standard of care therapies for women with locally advanced CC whose disease is curable with these therapies. The CTPM executive committee acknowledges the hard work of all working group members and their National Clinical Trials Network Groups in supporting development towards making progress in addressing critical unmet therapeutic needs for women with CC.

Funding

The NCI CTPM was supported by the NCI Coordinating Center for Clinical Trials and was developed under the leadership of the NCI Gynecologic Cancer Steering Committee.

Notes

Role of the funder: The funder had no role in the writing of this commentary or the decision to submit it for publication.

Conflicts of interest: The following potential conflicts of interest were disclosed: Matthew M. Harkenrider, MD, one-time advisory board role with AstraZeneca; Merry Jennifer Markham, MD, institutional funds from Aduro, Lilly, Tesaro, Novartis, and VBL Therapeutics; Don S. Dizon, MD, institutional funds for clinical trials support from BMS, Tesaro, Kazia, Pfizer and personal compensation for role on DSMBs with AstraZeneca, Clovis, Tesaro, and Regeneron; Ritu Salani, MD, advisory board role with AstraZeneca, GSK, Iovance, and Clovis. The rest of the authors had no potential conflicts of interest to disclose.

Acknowledgments: The authors wish to acknowledge with gratitude the participants of the NCI Clinical Trials Planning Meeting held on October 25-26, 2018 in Rockville, MD, for their valuable contributions to pre-meeting working group discussions and in-meeting presentations and comments, from which the content and recommendations in this manuscript was generated. A list of meeting participants can be found in the Supplementary Materials.