Abstract
The use of targeted small-molecule therapeutics and immunotherapeutics has been limited to date in pediatric oncology. Recently, the number of pediatric approvals has risen, and regulatory initiatives in the United States and Europe have aimed to increase the study of novel anticancer therapies in children. Challenges of drug development in children include the rarity of individual cancer diagnoses and the high prevalence of difficult-to-drug targets, including transcription factors and epigenetic regulators. Ongoing pediatric adaptation of biomarker-driven trial designs and further exploration of agents targeting non-kinase drivers constitute high-priority objectives for future pediatric oncology drug development.
Significance: Increasing attention to drug development for children with cancer by regulators and pharmaceutical companies holds the promise of accelerating the availability of new therapies for children with cancer, potentially improving survival and decreasing the acute and chronic toxicities of therapy. However, unique approaches are necessary to study novel therapies in children that take into account low patient numbers, the pediatric cancer genomic landscape and tumor microenvironment, and the need for pediatric formulations. It is also critical to evaluate the potential for unique toxicities in growing hosts without affecting the pace of discovery for children with these life-threatening diseases.
Introduction
Today, more than 80% of children with cancer survive for at least 5 years after diagnosis, the vast majority of whom are cured (https://seer.cancer.gov/archive/csr/1975_2015/). However, this success has not been uniform, and many survivors of pediatric cancer face lifelong sequelae of therapy (1). Although survival for children with the most common pediatric cancer, acute lymphoblastic leukemia, has improved from 10% to 90% through a series of clinical trials (2), little improvement in survival has been achieved in certain other childhood malignancies. These include high-grade gliomas and metastatic sarcomas, which remain incurable in the majority of patients. Further, as risk stratification has improved, defined subsets of patients with very high-risk molecular features of otherwise highly curable histologies have been identified (2). Even among children with cancers for which outcomes have improved, few patients who relapse are cured (3). Finally, although all pediatric cancers are rare by adult cancer definitions, some tumors including most carcinomas and melanoma occurring in pediatric populations have proved too rare to be successfully studied in clinical trials. Yet, in aggregate, these rare tumors comprise 11% of newly diagnosed malignancies in children, highlighting the need for novel trial designs to improve outcomes for these patients (4).
As therapy has been intensified, the late effects of cytotoxic chemotherapy in children have become increasingly apparent. These include risks of cardiomyopathy from anthracyclines and chest radiotherapy and secondary malignancies from radiotherapy, alkylating agents, and topoisomerase inhibitors, as well as numerous others. Among childhood cancer survivors, the cumulative incidence of a chronic health condition is higher than 70% by 30 years after the cancer diagnosis, with 42% cumulative incidence of severe, disabling, or life-threatening conditions (1). As the use of targeted agents has increased, unique toxicities of these agents in the growing host have become apparent (5–8). Today, drug development in pediatric oncology focuses on improving survival for those patients with largely incurable tumors as well as attempts to use novel agents to allow therapeutic substitution to improve the quality of survivorship.
Until the last decade, pediatric cancer drug development has focused on identification and intensification of cytotoxic therapy and improved supportive care. More recently, attention has turned to targeted small molecules and immunotherapy (3). Still today, relatively few children with cancer are treated with targeted agents as standard of care in the frontline/curative-intent setting, with notable exceptions including those with Philadelphia chromosome–positive acute lymphoblastic leukemia (Ph+ ALL) and chronic myelogenous leukemia (CML) treated with ABL-class inhibitors (9), acute promyelocytic leukemia (APML) treated with all-trans retinoic acid (ATRA; ref. 10), B-cell lymphomas treated with anti-CD20 antibodies (11), high-risk neuroblastoma treated with anti-GD2 antibodies (12), and, more recently, tropomyosin receptor kinase (TRK) fusion cancers treated with TRK inhibitors (13). Particularly for children with ALL, however, this is changing, with frontline trials evaluating the addition of either targeted small molecules or immunotherapeutics for the majority of patients (NCT02723994, NCT03959085, and NCT03914625).
Traditionally, dose-finding phase I trials of targeted therapies in children followed standard 3+3 or rolling 6 designs originally developed for cytotoxic agents. These designs evaluated 3 to 6 patients per dose level and often included a larger number of dose levels starting well below the dose equivalent to the already defined dose in adults (14, 15). Based on data showing a tight correlation of the pediatric and adult MTDs of kinase inhibitors, more efficient designs with fewer dose levels have been developed (16, 17). Pediatric dosing in early-phase trials demonstrates that postpubertal adolescents achieve similar drug levels as their adult counterparts (18). There are increasing data to suggest that these patients could safely be enrolled on adult trials, possibly enrolling at a dose level behind the current adult cohort where no dose-limiting toxicities occurred or in a separate dosing cohort (19). In addition, dose-finding studies have shown that pediatric recommended phase II doses (RP2D) in younger children are usually within 30% of the body surface area–adjusted approved adult dose, limiting the need for extensive dose finding (16, 17). Despite this, most pediatric phase I studies still enroll patients with a broad range of histologic diagnoses and do not include biomarker selection, limiting the ability to evaluate efficacy in populations of greatest interest (3).
Here, we review the current FDA/European Medicines Agency (EMA)–approved targeted drugs for the treatment of pediatric cancers and their development pathway. We also provide a perspective on the challenges of pediatric drug development moving forward, including the regulatory landscape, rarity of specific cancer subtypes in children, the relatively limited number of targetable mutations identified in pediatric cancers, difficulties in drugging non-kinase fusion oncoproteins, and the unique considerations of studying drugs in a developing host. We also highlight opportunities to overcome these challenges and continue to advance pediatric drug development in the era of targeted therapies including small-molecule kinase inhibitors, immune-based therapies, and, more recently, epigenetic therapies.
Current FDA/EMA-Approved Targeted Drugs for Pediatric Cancer
The dual challenges of the high costs of bringing a drug to market and the smaller number of pediatric patients with cancer compared with adults with cancer limit investment in this area. In this context, few drugs have achieved FDA or EMA approval specifically for the treatment of pediatric cancers: anti-GD2 monoclonal antibodies for neuroblastoma, and clofarabine and the chimeric antigen receptor T-cell (CAR-T) tisagenlecleucel for ALL. These drugs were developed for two of the most common pediatric cancers; development of agents specifically for the treatment of less common pediatric cancers has not yet been successful. Additional investment in preclinical work is necessary to expand the range of pediatric cancers for which novel therapies are ready to advance to the clinic. Further, additional incentives are necessary to encourage pharmaceutical companies to conduct clinical trials in rare pediatric cancer populations.
The vast majority of targeted therapies studied in children with cancer have been repurposed drugs initially developed for the treatment of adult cancers. Early-phase clinical trials of repurposed adult cancer drugs in children with relapsed/refractory solid tumors without biomarker selection have demonstrated very little efficacy, with a recent meta-analysis demonstrating a 4% objective response rate and less than 4-month median progression-free survival, substantially worse than has been observed in recent pooled analysis of adult phase I trials showing a nearly 20% response rate (3, 20). There has not been a similar formal meta-analysis of phase I trials in pediatric hematologic malignancies. However, several recent pediatric phase I trials including several leukemia trials have demonstrated markedly greater response rates among children selected to have the target of interest, highlighting the importance of this approach (13, 21–26). By definition, development of repurposed drugs in children lags behind that in adults, with the median interval of 6.5 years between the first studies in adults and those in children (27).
Prior to 2017, only 4 targeted therapies received a pediatric label indication, whereas 11 have achieved a U.S. FDA label since 2017 (Table 1). Several features of these successful drugs are notable. First, in all cases except for pembrolizumab, the disease or biomarker of interest within childhood cancer had been identified prior to the initiation of pediatric clinical trials. This resulted in enrichment or exclusive enrollment of this population within the first studies. Second, exceptional response rates were observed early in drug development, with 11 of 15 of these eventually approved agents having an initial response rate greater than 50% in the targeted population in pediatric phase I or II trials (Table 1). Given the smaller number of children with cancer, successful development of targeted therapies with modest activity has not been possible, highlighting the need to prioritize early-phase trials of potential transformational agents and design studies to provide early go/no-go decisions based on the presence of robust activity within the targeted patient population. A final challenge to pediatric drug development has been the cost and effort involved in seeking a pediatric label. Because of this, some drugs that have shown high response rates in the setting of academic or cooperative group trials [e.g., crizotinib for anaplastic large cell lymphoma (ALCL) and inflammatory myofibroblastic tumor (IMT); ref. 25] remain off-label uses in pediatrics.
Targeted therapies approved by the FDA and/or EMA for pediatric cancer
Regulatory Initiatives to Augment Pediatric Oncology Drug Development
Until recently, the greatest challenge to pediatric drug development was access to investigational agents. Economics and the prevailing myth that an early bad outcome in a child might derail a drug may have discouraged the pharmaceutical industry from sponsoring pediatric trials or providing drug supply for investigator-initiated trials. Indeed, recent data demonstrate a disparity in rates of industry sponsorship of pediatric oncology trials, where only 16% of trials including minors were industry-sponsored, compared with trials for adults with cancer (33%) or trials in other pediatric specialties (30%; ref. 28). Over the past two decades, the FDA in the United States and the EMA in Europe have passed legislation aimed at accelerating the development of drugs for children. Approaches have included both incentivizing and mandating sponsors to submit a pediatric development plan at the time of a new drug or biological license application. Early incentives introduced by the 2002 U.S. Best Pharmaceuticals for Children Act (BPCA; Public Law 107–109) granted a 6-month patent exclusivity extension to sponsors willing to develop appropriate drug formulations and complete pediatric clinical studies as outlined by the FDA. Soon after, the Pediatric Research Equity Act (PREA; 21U.S. Code 355B) codified requirements to perform such studies for agents undergoing FDA review for an adult indication. The EMA took a similar compulsory approach with the Pediatric Regulation [Regulation (EC) No 1901/2006 of the European Parliament and of the Council of December 12, 2006, on medicinal products for pediatric use] but required a more comprehensive Pediatric Investigation Plan (PIP) earlier in the drug-development process, that had to be completed by the time the first marketing authorization was granted. The European regulation combined the obligation to submit the PIP for all drugs requesting marketing authorization with the incentive of a 6-month patent extension for drugs completing their PIP successfully, and included voluntary PIPs which allowed companies that might have been granted waivers to access the incentives if they committed to pediatric development.
However, the lack of harmonization across different regulatory agencies and, in particular, the differences in the timing and scope of the EMA PIP and FDA PREA requirements further complicated pediatric academic–industry partnerships and the drug-development process (29). Both mandatory programs were also undermined by the provision of automatic waivers for drugs addressing orphan diseases or histologies rarely occurring in children [section 505B(a)]. For targeted agents directed toward growth pathways relevant to adult malignancy, sponsors were released from pediatric development responsibilities based on the fact that these adult histologies (e.g., lung or prostate cancer) did not occur in children. According to one European review, from June 2012 to June 2015, 147 oncology class waivers were granted for 89 drugs, 48 (54%) of which had a mechanism of action of substantial interest to pediatrics (30). As a corollary, although 126 drugs gained initial FDA approval for an oncologic diagnosis over the 20-year period from 1997 to 2017 in the United States, only 6 of these agents included a pediatric indication (27).
Newer legislation seeks to close these loopholes. In Europe, the EMA published a revised class waiver list in 2015 that came into effect in 2018. In the United States, a recent amendment to PREA [Title V of the FDA Reauthorization Act of 2017 (FDARA; Public Law 115–52)], called the RACE (Research to Accelerate Cures and Equity) for Children Act, requires sponsors to submit an initial Pediatric Study Plan when the molecular target of an agent under evaluation is “substantially relevant to the growth and progression of a pediatric cancer.” To date, the Relevant Molecular Target List (RMTL) developed by members of the NCI, FDA, and the pediatric oncology community contains more than 200 molecular targets involved in cell lineage, tumor growth, and the microenvironment relevant to pediatric malignancy (https://www.fda.gov/about-fda/oncology-center-excellence/pediatric-oncology). To facilitate the appropriate preclinical development of RMTL drugs, the Foundation for the National Institutes of Health (FNIH) recently announced a public–private partnership to foster industry and academic collaboration.
These initiatives aim to expedite pediatric access to drugs under development for the treatment of adult cancers. Beginning in 2012, the U.S. Congress introduced the priority review voucher as a powerful new incentive through the Creating Hope Act, currently up for reauthorization. As opposed to the above legislation which seeks to encourage the study of drugs being developed for adult cancers, this act seeks to incentivize drug development specifically for pediatric cancer and other rare pediatric diseases. The voucher is awarded to a sponsor who successfully obtains FDA approval for a rare pediatric indication and can be either used to expedite the FDA review of a future drug or sold on the open market to raise capital. Successful development of both dinutuximab and tisagenlecleucel resulted in the granting of such vouchers to their sponsors. In the case of dinutuximab, the sale of the priority review voucher was valued at $350 million (31).
Although these changes successfully address issues of early access to investigational products, with greater interest in pediatric trials to fulfill regulatory requirements, they also pose new challenges. In medical oncology, there may be a value proposition for both patient and industry around biosimilars (32), and the market can support several agents in a class each with subtle differences in efficacy, toxicity, and resistance mechanisms. Given the small numbers of patients available for early-phase pediatric trials, however, there is little enthusiasm to repeat similarly designed trials using multiple agents with the same mechanism of action when the first demonstrated only modest activity. For example, single-agent phase I trials have been conducted for multiple VEGFR-blocking agents and antiangiogenic tyrosine kinase inhibitors (TKI) including bevacizumab, aflibercept (NCT00622414), sorafenib, sunitinib (NCT00387920), pazopanib (NCT00929903), cediranib (NCT00354848), lenvatinib (NCT02432274), and cabozantinib (NCT01709435), each demonstrating sporadic responses and a signal of stable disease in sarcoma, such that there is little knowledge to be gained by additional single-agent phase II trials of agents in class. Similarly, there can be contemporaneous trials of single agents all in the same therapeutic class such as the multiple single-agent pediatric anti–PD-1 inhibitor trials currently or recently enrolling despite the lack of early results for immune checkpoint inhibitors in pediatrics. Numerous competing trials of agents in the same class may compromise accrual of children with rare diseases, delaying or preventing the ability to define efficacy within the pediatric population, as has occurred with pediatric studies of BRAF inhibitors for melanoma (33, 34).
Coordinating drug development in children across multiple agents in the same class raises a number of challenges, including a lack of clear standards to determine which agent is the “best in class” and an adequate mechanism to distribute the financial burdens and incentives across competing pharmaceutical companies. Several initiatives are ongoing to overcome these challenges by aligning key stakeholders (Fig. 1). The Pediatric Cluster is a series of regular teleconferences between the FDA, EMA, Japan's Pharmaceuticals and Medical Devices Agency, Health Canada, and Australia's Therapeutic Goods Administration (https://www.fda.gov/science-research/pediatrics/international-collaboration-pediatric-cluster). In 2018, these regulators began generating nonbinding comments which are provided to drug sponsors after these meetings. Early inclusion of adolescents (12 years and older) on phase I trials has been encouraged by several working groups particularly for trials targeting known drivers of their disease or disease-specific mutations (19, 35). In parallel, many pharmaceutical companies have hired pediatric oncologists and more actively engage in early interactions with academia. The Accelerate Platform has held a series of multistakeholder pediatric strategy forums which include involvement from the EMA and FDA, drug companies, patient advocacy groups, and academic pediatric oncologists. To date, these meetings have focused on coordinating development across high-priority pediatric targets including ALK inhibitors, mature B-cell malignancies, immune checkpoint inhibitors, acute myeloid leukemia (AML), epigenetic modifiers, and CAR-T cells (36, 37).
Key stakeholders contributing to rational pediatric cancer drug development. IND, investigational new drug.
Ideally, collaboration should begin even earlier in the drug-development process. The rarity of pediatric cancers does not allow for routine clinical testing of all potential new agents, thereby creating a greater reliance on prioritization based on scientific rationale and preclinical testing. For targeted therapies, it is important to assess the frequency and dependence on the target or pathway in the common malignancies of childhood. Although the new U.S. Childhood Cancer Data Initiative may provide new in silico tools that can inform pediatric cancer drug development, ultimately, pharmaceutical companies may have limited access to relevant tissue or appropriate model systems for testing, and therefore collaboration with independent academic labs should be encouraged. In addition, the U.S. Pediatric Preclinical Testing Consortium has a long track record of providing testing of novel agents in established models, and similar work is being done in Europe by Innovative Therapies for Children with Cancer (ITCC; ref. 38). New public–private partnerships such as that sponsored by the FNIH may continue to bridge this gap. Although there is no predefined level of preclinical evidence needed to move an agent into the pediatric oncology clinic, preclinical testing can help to prioritize or deprioritize specific new agents or drug classes. The required level of evidence needed to move a drug forward depends upon the available histology and biomarker-specific data from adult patients, the agent's toxicity profile, and the level of unmet medical need in the potential pediatric target population.
With regard to unmet clinical need, potential for benefit, and tolerance for risk, parent and patient advocacy organizations can provide the patient voice in early go/no-go discussions and as such have become an integral part of the pediatric drug-development process. Their contribution to multi-stakeholder initiatives is invaluable, and they can provide insight about facilitating access to novel therapies, help ensure equipoise in randomized trials, facilitate research forums, fundraise to bridge specific needs in the field, and advocate with lawmakers, regulators, and health technology assessment bodies. As one recent example, passage of the U.S. RACE for Children Act requiring sponsors to submit pediatric study plans based on relevance of the molecular target was achieved with the support of numerous advocacy groups.
Challenges of Pediatric Drug Development and the Opportunities they Create
All Pediatric Cancers Are Rare Diseases
Pediatric cancers occur in approximately 16 of 100,000 children between 1 and 18 years (39), which as a whole meets the European Union and U.S. definitions of rare or orphan diseases (40). Pediatric tumors include many histologically and biologically disparate tumors presenting unique challenges around patient numbers. As focus shifts increasingly to targeted therapy, conduct of studies of targeted agents can be challenging in the pediatric population. Identifying and selecting for subpopulations with specific mutations or biomarkers presents challenges around sample size and the power to detect an effect. Novel trial designs and approaches must be employed to obtain safety and preliminary efficacy data efficiently and from as few patients as possible. The majority of pediatric phase I dose-finding studies utilize either the 3+3 or rolling 6 trial designs. These conservative trial designs, although easy to conduct, are less likely to determine the true MTD and treat more patients at suboptimal dosing (41–43). Utilization of adaptive model trial designs such as the continuous reassessment method or Bayesian optimal interval can provide better dose-finding accuracy using data from prior studies and thereby improving efficiency (41, 43, 44). In addition, these model-based designs are able to account for size-based dosing in non–pediatric-friendly formulations such as pills (43, 45).
Historically, phase I studies in pediatrics have been agnostic of tumor type for the dose-finding portion of the study. This approach allows for increased enrollment numbers and decreased time to determine the RP2D. However, this method may limit opportunities for the desired patients with the targeted biological changes in their tumors to enroll. Novel trial designs can enrich for the targeted population while maintaining efficiency. One such design permits enrollment of tumor regardless of mutational status; however, patients with the desired tumor mutations targeted by the study drug can enroll either at the current dose-escalation dose or, if there are no available slots, at one dose level below the current dose-escalation level. This design is currently being employed in at least two pediatric trials (NCT03654716 and NCT03936465). This enriches for the desired population while continuing to efficiently determine the RP2D and MTD. The Pediatric MATCH trial (NCT03155620) employs an alternative approach, incorporating dose finding and efficacy evaluation within a pediatric histology-agnostic basket trial. In subprotocols for which the adult MTD and RP2D are not well defined, a limited dose-finding phase is implemented (NCT03213678 and NCT03698994). Such early-phase trial designs allow for efficient determination of pediatric dosing while concurrently enriching for targeted tumor subtypes which are most likely to respond. In addition to increasing the chance to benefit enrolled patients, early efficacy signals in targeted tumor types on these studies may augment enthusiasm for patients and physicians to enroll, and, if early evidence of activity is seen, for academia and pharma to continue to pursue pediatric development of the agent.
Umbrella and basket protocols allow for the application of precision medicine in a tumor-agnostic approach. This type of trial can increase accrual when compared with a histology-specific biomarker-positive design while increasing the chance of benefit to individual patients and providing early evidence of efficacy when compared with enrolling unselected cohorts of patients (46). Notably, the pediatric phase I/II study of dabrafenib successfully utilized such a design, whereas the pediatric phase I study of vemurafenib failed to accrue while enrolling only a cohort of patients with BRAF-mutated melanoma (33, 34). Current pediatric trials such as the Pediatric MATCH (NCT03155620) and the European ESMART trial (NCT02813135) enroll patients with multiple relapsed/refractory pediatric cancer types and then assign patients to one of several treatment arms based on defined molecular alterations. Umbrella protocols such as these improve directed enrollment for tumor-specific mutations to respective targeted agents and in some instances permit subsequent assignment to another therapy arm if initial treatment is not successful.
For agents of interest for diseases affecting young children, drug-development plans should incorporate pediatric-friendly formulations early in the development process to allow rapid initiation of pediatric drug development as soon as adult MTD is determined. When multiple drugs in the same class are in development, the availability of a pediatric-friendly formulation can influence which agent(s) are selected for studies in children. Conversely, the selection of the agent(s) for studies in children should be early enough to allow for the development of pediatric formulations if none exist given the time and expense required. Oral formulations developed for fixed dosing in adults are often too large to allow for appropriate body surface area–based dosing for smaller children and limit the eligible population to those who can swallow pills. Also, multiple-strength tablets or a liquid formulation are often needed to allow for appropriate dosing during the dose-escalation phases, accounting for children with different body-surface areas, and allow suitable dose increases and reductions. Delays in developing a pediatric-friendly formulation restrict enrollment on pediatric trials to older and larger children, inadequately representing younger patients and some pediatric histologies that occur almost exclusively in this age group. This point is highlighted by the pediatric development of larotrectinib, a selective TRK inhibitor. TRK-fusion cancers occur in very young children, with the majority of pediatric cases in patients under 2 years of age, including glioma, fibrosarcoma, and congenital mesoblastic nephroma. A pediatric-friendly formulation of larotrectinib was developed early, permitting near-simultaneous agent development in both pediatric and adult populations. For the pediatric study, almost 30% of the enrolled patients were less than 2 years of age, and almost half of the responders had infantile fibrosarcoma (13). As a result, an age- and histology-agnostic global approval was obtained in 2018. Without the pediatric formulation, these young patients would not have been included and the pediatric data would not have fulfilled the needs of children with TRK-fusion cancers. Notably, the TRK inhibitor entrectinib is FDA-approved only for children 12 years and older despite an ongoing pediatric trial of this drug (NCT02650401), and pediatric development has been slowed by delayed availability of a pediatric-friendly formulation (26).
As pediatric cancer continues to be parsed into genomically defined subtypes of already rare tumors, the importance of collaboration becomes even more critical. Pediatric cooperative groups have been integral in pediatric oncology for several decades. However, further international collaboration is crucial as targeted therapy is developed for molecular subtypes of pediatric tumors. Large multinational trials allow opportunity for enrollment of increased numbers of patients of a specific rare molecular subtype to garner enough power to be able to detect a difference in this population (47). As discussed above, the rarity of specific subsets of pediatric cancer also requires coordination between regulators and sponsors when there are multiple agents in the same class seeking pediatric safety and efficacy data.
The Quiet Pediatric Cancer Genome
Compared with most adult malignancies, most pediatric cancer genomes are characterized by a low tumor mutational burden (48, 49). Instead, pediatric tumors often harbor structural variants, such as copy-number alterations and translocations (see next section). When present, mutations recurrently observed in the same histology are thought to play an important role in the biology of the disease, particularly if few other genomic alterations are seen (e.g., SMARCB1 mutations in epithelioid sarcoma and malignant rhabdoid tumors, and BRAF mutations in melanoma, glioma, and Langerhans cell histiocytosis). These mutations may lead to specific vulnerabilities that may be exploited for therapeutic gain. For example, SMARCB1-inactivating mutations result in a dependency on EZH2 (50), leading to objective responses and recent regulatory approval for the EZH2 inhibitor tazemetostat for patients ≥16 years of age with epithelioid sarcoma (51). In addition, recent preclinical work suggests that SMARCB1-inactivating mutations result in derepression of endogenous retroviral elements, leading to highly inflamed tumors that may be responsive to immunotherapy despite their exceedingly low mutational burden (52).
Nevertheless, in many pediatric cancers there is not a clearly defined genomic target. In these cases, the field has moved toward targeting other vulnerabilities, including cell surface overexpression of specific antigens. As many pediatric cancers are embryonal cancers, developmentally regulated proteins with highly restricted expression may be present on the cell surfaceome and/or presented via the MHC, allowing for immunotherapy approaches that minimize “on-target/off-tumor” toxicities. Targeting the cell surfaceome has led to several pediatric regulatory approvals of monoclonal antibodies, bispecific T-cell engagers, and CAR-T cells. For example, the anti-GD2 chimeric antibody dinutuximab was shown to improve 2-year event-free survival by 20% in patients with newly diagnosed high-risk neuroblastoma when given during the postconsolidation phase of therapy (12). The bispecific T-cell engager (BiTE) antibody targeting-CD19 and CD3 blinatumomab has improved disease-free survival and rates of minimal residual disease response when combined with standard relapse chemotherapy in ALL (53). Another dual-affinity CD123 X CD3 antibody, flotetuzumab, is currently in phase I trials though the Pediatric Early Phase Clinical Trials Network for pediatric AML (NCT04158739).
Adoptive cell therapy with engineered autologous CAR-T cells against CD19 has demonstrated transformational activity in ALL, with response rates as high as 93% and 12-month event-free survival of 50%, leading to FDA approval of tisagenlecleucel in 2017 (54–56). Epitope loss and downregulation are secondary resistance mechanisms (57, 58) which are now being targeted clinically with dual-targeted CAR-T products against CD19/CD22 or CD19/CD20 (NCT03241940 and NCT03330691). Studies in adult patients are now evaluating the combination of checkpoint inhibition with CAR-T cell products to overcome resistance due to CAR-T cell exhaustion/senescence and failure of in vivo expansion, which can lead to relapses that retain CD19 expression (NCT03630159). Additional CAR constructs which target other cell-surface antigens expressed on leukemia and lymphomas are actively in development and are of interest for AML, including in cases of lineage switch after treatment of ALL with CD19 targeted CAR-T cell products (59).
Clinical trials of CAR-T cells targeting antigens on pediatric solid tumors have been relatively disappointing to date, highlighting the challenges of overcoming the immune-suppressive tumor microenvironment (60). Although challenged by the need to subdivide already rare patient populations into HLA-specific cohorts, studies of cells engineered to express T-cell receptors recognizing developmentally restricted self-antigens in an MHC-dependent manner have shown promise in early studies for solid tumors and remain under investigation (NCT03697824; ref. 61). Moreover, cell-surface expression of specific antigens may be heterogeneous within the same histology (e.g., GD2 expression in pediatric sarcomas; ref. 62), and clinical assays to select appropriate patients for therapies targeted to specific antigens are needed in these situations.
Another consequence of the quiet pediatric cancer genome may be fewer cancer neoantigens to generate an antitumor response in the context of immune checkpoint inhibition. Indeed, the clinical experience to date with this class of compounds in pediatrics has been disappointing. Expression of PD-1 and its corresponding ligands PD-L1 and PD-L2 is low on most pediatric tumors (63). Clinical responses to immune checkpoint inhibitors in children have largely been reported in pediatric tumors with higher PD-1 expression such as lymphoma or with increased tumor mutational burden as seen in melanoma, tumors with high microsatellite instability, or patients with congenital mismatch repair deficiency (64–66). However, multiple early-phase studies of immune checkpoint inhibitors in unselected pediatric patients have failed to demonstrate objective responses in the majority of patients enrolled (65–67). Although checkpoint monotherapy has been disappointing, there is interest in combination therapy of checkpoint inhibitors (NCT 03837899, NCT03130959, and NCT02304458) or combinations with chemotherapy, radiation, or other targeted radiotherapy (NCT02927769, NCT03445858, and NCT02914405) which could improve the “cold” immune environment of pediatric cancers through increased antigen presentation and T-cell activation (68, 69).
The lower tumor mutational burden generally seen in pediatric cancers compared with adult cancers may ultimately be advantageous in the context of drug development. Owing to fewer secondary genomic events, pediatric tumors may tend to be more genomically homogeneous, potentially reducing the potential for confounding through activation of other pathways by passenger mutations. Likewise, primary resistance mutations may be less likely to be present de novo in tumors that are less genomically complex and that have not evolved over many years. For example, the response rate to the TRK inhibitor larotrectinib in patients with TRK-fusion cancers is greater in children (94%) than in adults (71%) despite harboring the same biomarker (70, 71). Similarly, the response rates to crizotinib for ALK fusion–positive IMT (86%; ref. 25) and ALCL (88%; ref. 25) were numerically higher than for ALK-fusion lung cancer in adults (65%–74%; refs. 72, 73). However, it is worth noting that ALK inhibitors have had very dissimilar results in pediatric tumors harboring different genomic aberrations, such as ALK-mutated or ALK-amplified neuroblastomas which show more resistance to ALK inhibitors. This further highlights the need for pediatric-specific drug-development efforts which account for the common molecular alterations within common pediatric cancers.
Finally, it should be noted that most published sequencing data from pediatric cancers are from diagnostic samples. Enrichment of potentially targetable molecular alterations has been noted at relapse, such as RAS–MAPK pathway activating alterations in neuroblastoma (74). However, in the NCI-COG Pediatric MATCH study, still only 29% of patients had actionable mutations identified in relapse tissue (75). Further evaluation of the evolution of the pediatric cancer genome over the course of therapy will be important to identify the frequency of the development of targetable alterations and whether the mutations that develop over the course of therapy are sufficiently clonal to serve as therapeutic targets. In addition, it will be especially important to collect and analyze relapse tissue and diagnostic tissue from patients with relapsed disease enrolling on targeted therapy trials to better characterize potential biomarkers. New approaches using cell-free DNA or circulating tumor DNA may provide a more feasible noninvasive means of detecting and monitoring mutational changes and burden during therapy (76).
Targeting Non-Kinase Fusion Oncoproteins
A substantial proportion of pediatric, adolescent, and young adult cancers are characterized by recurrent chromosomal translocations that generate fusion oncoproteins. These fusion oncoproteins can be broadly classified as kinase fusions or non-kinase fusions, depending upon whether or not the fusion oncoprotein harbors a kinase domain. Kinase fusions typically result in an overexpressed and constitutively active kinase that drives the cancer. Key examples include NPM–ALK fusions in ALCL or ETV6–NTRK3 fusions in infantile fibrosarcoma. In contrast, non-kinase fusions often result in an aberrant transcription factor that drives an abnormal gene-expression program that characterizes a specific cancer. Key examples include EWSR1–FLI1 fusions in Ewing sarcoma and PAX3–FOXO1 fusions in alveolar rhabdomyosarcoma.
Kinase fusions represent the minority of fusion oncoproteins in the pediatric population but are illustrative of important principles of drug development that might be applied to development of drugs targeting non-kinase fusions. First, targeting kinase fusions with selective kinase inhibitors has resulted in prompt, dramatic, and often durable responses that demonstrate the central role these fusions play as drivers of the disease (13, 25). Second, given the very high response rates associated with agents targeting these driver fusions, agents such as entrectinib and larotrectinib have received regulatory approval in adolescents and children based upon relatively small early-phase clinical trials. Third, the therapeutic index has typically been wide, due to both increased kinase selectivity and the extreme sensitivity of these cancers to kinase inhibition, with responses observed across dose levels in first-in-human or first-in-child trials. Fourth, primary resistance is exceedingly rare, but secondary resistance to monotherapy kinase inhibition has been reported (77–79), highlighting the potential role for combination strategies.
In this context, targeting non-kinase fusions may likewise provide important opportunities for developing drugs for the rare pediatric cancers driven by this type of fusion. To date, many of these cancers have been treated with nonspecific cytotoxic chemotherapy, with some success seen in patients with localized disease and little success in patients with metastatic or recurrent disease. Due to difficulties targeting aberrant transcription factors (rather than kinases), attempts at targeted therapy have largely focused on effectors downstream of the oncoprotein. For example, activity of the IGF1R pathway has been shown to be upregulated by the EWSR1–FLI1 fusion oncoprotein (80). Early-phase clinical trials of monoclonal antibodies directed against IGF1R demonstrated that approximately 10% of patients with relapsed Ewing sarcoma responded (81, 82). This experience highlights the modest activity associated with targeting just one of many downstream effectors of the fusion oncoprotein, rather than targeting the fusion oncoprotein itself. Indeed, a recent randomized phase III trial failed to show an improvement in event-free survival when an anti-IGF1R monoclonal antibody was added to conventional chemotherapy for patients with newly diagnosed metastatic Ewing sarcoma (83).
This type of disappointing outcome stands in stark contrast to the exciting results seen targeting kinase fusions and now motivates strategies to directly target non-kinase fusion oncoproteins. Key strategies under investigation in the laboratory and also now in the clinic include attempts to block the function of these oncoproteins, selectively reduce expression of the fusion oncogenes, or selectively degrade the fusion oncoprotein. As these non-kinase fusion oncoproteins are true drivers of their associated cancers, these strategies have shown promise in the laboratory (84–87), though proof of concept is yet lacking in the clinic. Likewise, as these oncoproteins should not exist outside of cancer cells, the therapeutic index of selective approaches to targeting non-fusion kinases is expected to be wide. Many of the cancers driven by non-kinase fusions affect adolescents and young adults, providing an opportunity for age-agnostic drug-development strategies that may facilitate accrual in these rare diseases. Some of the principles learned through the study of non-kinase fusions may extend to the evaluation of other oncoproteins that likewise serve as transcription factors driving abnormal gene expression programs (e.g., high-level MYCN amplification in neuroblastoma).
Assessing Long-Term Toxicities of Targeted Therapies in Growing and Developing Hosts without Halting Drug Development
Although most new anticancer drugs are more tolerable than conventional chemotherapeutics, they are not exempt from significant and severe toxicities in the short and long term. In the short term, toxicities of targeted therapies in children are similar to those identified in adults (17).
Only recently, efforts have started to improve our understanding of late toxicities of novel drugs in children with developing organs and systems (88). Novel drugs can have an impact on transcriptional pathways, intracellular signaling pathways, epigenetic functions, or the microenvironment, all of which are critical during growth and development of multiple organs. At the time of starting early-phase pediatric trials, potential developmental toxicities are often considered but are difficult to assess and predict. Examples are risk of growth plate closure with antiangiogenics and sonic hedgehog inhibitors or risk of neurocognitive toxicities of ALK and TRK inhibitors. Experience gained over the last two decades shows that, although these concerns must be addressed during the development of novel agents, particularly once drugs move into phase II or phase III trials, they should not halt or delay the initiation of pediatric early-phase trials.
The role of juvenile animal studies and their ability to predict toxicity in the pediatric oncology population has not been fully elucidated yet. For example, a review of 29 recently developed anticancer medicines found that severe toxicities arising in the first-in-child trials could not be prevented by conducting standard juvenile animal studies (89). Another review of drugs that received European approval and had Pediatric Investigational Plans showed that approximately 1 in 3 oncology medicines with PIPs had conducted juvenile animal studies, contributing to a better characterization of safety in some cases (90). Although these studies have proved useful in other pediatric disciplines, there has been general reluctance from pharma and academia to systematically incorporate these studies prior to initial clinical trials of new drugs for children with cancer. On the other hand, such studies have been supported by the EMA and on a case-by-case basis by the FDA (91). More data are needed to fully ascertain the capacity of juvenile animal studies to identify, predict, and hopefully prevent potential developmental toxicities in children.
At present, pediatric phase I and II trials provide limited opportunities to identify long-term toxicities because they are mostly conducted in patients with relapsed and refractory cancers with short life expectancy (92, 93). Even in highly successful trials with long-term responders, the relatively small and heterogeneous patient population, lack of randomized controls, and potential late effects of prior therapies limit the ability to evaluate long-term toxicities (13, 25, 34). Hence, the main challenge (and opportunity) arises at the time of bringing these drugs up front into first-line collaborative international trials by academic groups such as the Children's Oncology Group (COG) and the European Society for Pediatric Oncology where late effects can be systematically evaluated, particularly for tumor types occurring in younger children.
Potential late toxicities can affect virtually all organs. However, to date very few have been reported. Table 2 shows potential toxicities in the growing/developing host associated with novel agents according to target organ. In many cases, the impact of these toxicities would also be influenced by the patient's age and their developmental status. The following examples illustrate how these toxicities can be evaluated and risks mitigated.
Potential long-term toxicities of new targeted therapies in the developing host
Linear Growth with Imatinib and Other TKIs
Imatinib and other ABL-kinase inhibitors were among the earliest targeted anticancer agents evaluated in pediatrics in the early 2000s. Some reports have highlighted growth impairment, particularly in prepubertal children. Shima and colleagues reported a decrease in height in more than 70% of children treated with imatinib, which was more pronounced in children starting imatinib before pubertal age (5). Bone length and bone density were reduced after long-term exposure in preclinical studies. The mechanism of this toxicity has not been fully elucidated and could be related to off-target effects on IGF1 or PDGFR pathways (6). Subsequently, early- phase trials of new ABL TKIs, such as dasatinib and nilotinib, have included growth measurements, circulating biomarkers, and determinations of bone mineral density in all patients. The initial analysis from the dasatinib phase II trial reported bone growth and development toxicities in 5 of 113 patients treated (4%); all were grade 1–2 (47). In the nilotinib pediatric phase II trial, no impact on growth was reported (94); however, longer follow-up including analyses of growth biomarkers and bone mineral density studies will be required. A pediatric trial of bosutinib (NCT04258943), another more potent ABL TKI, has recently opened, with the potential advantage of reducing the risk of growth impairment (7).
For other TKIs, particularly those with an antiangiogenic profile, abnormalities in the growth plates have been reported in a small but relevant proportion of patients. In the phase I trial of pazopanib, out of 25 patients included, 1 had physeal widening and 3 had growth plate widening (95). As some of these agents move forward and are being evaluated in frontline studies, there is a greater need to continue growth plate monitoring (96, 97). Although a pooled analysis of children receiving bevacizumab showed no negative effects on growth velocity, height, or body mass index (98), longer follow-up of patients treated with these drugs is needed to determine long-term effects on growth and development.
Neurocognitive and Visual Outcomes with Crizotinib and Other TKI
Crizotinib, an ALK inhibitor approved for ALK-positive non–small cell lung cancer, showed promising preclinical and clinical activity in children with ALK-mutated neuroblastomas and ALK-fused ALCL and IMT (25). Furthermore, the recently reported pediatric phase I trial of lorlatinib, a third-generation ALK inhibitor, by the New Advances in Neuroblastoma Therapy consortium showed that a proportion of patients developed neurocognitive toxicities. Grade 1 events such as anxiety, cognitive disturbance, concentration, or memory impairment were seen in pediatric patients, and grade 1–4 events were noted in patients older than 18 years, particularly in a patient with an undiagnosed preexisting psychiatric condition (99). Given the known role of ALK in the developing embryonic and neonatal brain (100) and the occurrence of ocular and neurologic toxicities in adult trials, enhanced monitoring of visual and neurocognitive outcomes was incorporated into the lorlatinib trial, which is now being moved forward to a frontline phase III trial by the COG. However, it is difficult to assess these outcomes in very young children with appropriate tools and scales and in the context of complex and aggressive multimodal therapy.
Risk of Secondary Malignancies and Immune Dysregulation with Engineered Cellular Therapies
The development of CAR-T cells has revolutionized outcomes for children with B-cell acute lymphoblastic leukemia (55), leading to the approval of tisagenlecleucel. CAR-T cells directed against CD19 cause profound and potentially life-long B-cell aplasia which requires chronic immunoglobulin replacement. Late infections, second malignancies, immune-related events, and neurologic and psychiatric events have been described (8). Among 86 recipients of adult CAR-T cells surviving for a year or longer, 13 developed a second malignancy, but no data on pediatric patients have yet been reported. Following the regulatory requirements, a follow-up study of patients treated with tisagenlecleucel was launched (NCT02445222) to address in-depth long-term toxicities, paying particular attention to secondary malignancies, neurologic disorders, exacerbation of immune conditions, or other emergent toxicities.
Duration of therapy is often as yet undefined for new targeted therapies, an aspect with crucial impact on potential for long-term toxicities. As novel drugs are being incorporated into frontline therapy, the optimal duration of therapy should also be explored. Some targeted therapies will be given for a relatively short duration to facilitate further consolidation therapies, such as an ALK inhibitor facilitating remission in a patient with ALCL who then goes on to receive a hematopoietic stem cell transplant. For targeted agents developed against hematologic conditions, such as inotuzumab (anti-CD22), blinatumomab (anti-CD19 BiTE), or anti-CD19 CAR-T, it remains essential to define the indications for hematopoietic stem-cell transplantation after targeted therapies.
Likewise, a patient with an inoperable TRK-driven cancer could receive a TRK inhibitor until appropriate nonmutilating surgery can be conducted and then stop therapy. However, for most targeted drugs developed in early clinical trials, treatment is given until disease progression or unacceptable toxicity. In these cases, responding patients can often receive therapy for years, and the optimal timing to stop therapy remains unknown. When considering stopping therapy in responding patients, it will be important to define the likelihood of response if treatment rechallenge is needed for subsequent relapse for each targeted therapy and condition. Defining the shortest possible course of targeted drugs that minimizes long-term toxicities while having no impact on reducing survival outcomes remains a key challenge.
Toward a New Model to Address the Need for Long-Term Follow-Up
With numerous targeted agents being developed for pediatric cancers, some of them being taken to frontline therapy, it will be necessary to perform pooled analysis of long-term outcomes across all pediatric cancers, but also for drugs against a specific target (e.g., all patients treated with ALK inhibitors) or within a specific condition (e.g., all patients with neuroblastoma treated with novel agents). It is hence necessary to create an international data repository to collect information on long-term health in children who have received new anticancer therapies, across different targets, sponsors, and tumor types, independent of specific regulatory requirements for each drug. The Accelerate platform has recently launched a working group on this matter (https://www.accelerate-platform.org/wp-content/uploads/sites/4/2019/07/Summary_7th-ACCELERATE-Paediatric-Oncology-Conference_14-15-Feb-2019.pdf). Pancare in Europe (www.pancare.eu) and the Childhood Cancer Survivor Study in North America are leading efforts involving all stakeholders, including empowering cancer survivors, to identify and mitigate long-term toxicities.
Conclusions
Only a limited number of targeted therapies have achieved regulatory approval or widespread use in pediatric oncology. Given recent regulatory changes, a broad range of drug classes and agents are of interest and under current investigation (Table 3). Although the majority of currently approved targeted therapies for pediatric cancer are kinase inhibitors, agents are under investigation for a broader range of other targets, including epigenetic modifiers, antiapoptotic proteins, and immunotherapies against cell-surface proteins. New pharmaceutical strategies to directly target non-kinase oncoproteins, one of the most common oncogenic drivers in pediatric cancer, have shown promise in preclinical studies and have entered the clinic. Drug development for pediatric oncology presents a number of challenges including lower patient numbers compared with adult oncology, a generally quiet genome, and the need to develop strategies to target non-kinase fusion oncoproteins. However, these same challenges result in a number of opportunities including a mandate to coordinate drug development across national boundaries and sponsors, the potential for easier identification of true oncogenic drivers in pediatric cancer, and less acquired resistance. As novel therapies move to the frontline setting, it will be important to develop mechanisms to evaluate the long-term toxicities of these agents without halting new discoveries or use of promising agents in the clinic.
Mechanisms of action of selected targeted agents under investigation for the treatment of pediatric cancer
Disclosure of Potential Conflicts of Interest
T.W. Laetsch reports grants, personal fees, and nonfinancial support from Bayer, grants, personal fees, and nonfinancial support from Novartis, grants from Pfizer, personal fees and nonfinancial support from Loxo Oncology/Eli Lilly, and personal fees from Cellectis outside the submitted work. S.G. DuBois reports nonfinancial support from Loxo (travel), nonfinancial support from Salarius (travel), nonfinancial support from Roche (travel), and personal fees from Bayer outside the submitted work (Clinical trial support), other from Loxo Oncology (Clinical trial support), other from Ignyta (Clinical trial support), other from Merck (Clinical trial support), other from Amgen (Clinical trial support), other from Roche (Manuscript preparation), and other from Celgene (Manuscript preparation) outside the submitted work. M.E. Macy reports other from Johnson & Johnson (stockholder) outside the submitted work. L. Moreno reports other from Novartis [Consultancy role and Member of DMC (fees to institution)], other from Shionogi [Consultancy role and Member of DMC (fees to institution)], other from Incyte [Member of DMC (fees to institution)], other from Eusa Pharma [member of the Executive Committee of SIOPEN (European neuroblastoma research cooperative group) which receives royalties for the sales of dinutuximab beta], and other from Actuate Therapeutics [member of the Executive Committee of SIOPEN (European neuroblastoma research cooperative group) which receives royalties for the sales of dinutuximab beta] outside the submitted work.
Acknowledgments
T.W. Laetsch is supported by the Norma and Jim Smith Professorship of Clinical Excellence and the Eugene P. Frenkel, M.D., Scholarship in Clinical Medicine. J. Glade Bender is supported by the NIH/NCI grant CA 008748.
Footnotes
Cancer Discov 2021;11:545–59
- Received June 2, 2020.
- Revision received August 8, 2020.
- Accepted September 8, 2020.
- Published first December 4, 2020.
- ©2020 American Association for Cancer Research.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.
- 102.
- 103.
- 104.
- 105.
- 106.
- 107.
- 108.
Volume 11, Issue 3
Jump to section
- Article
- Abstract
- Introduction
- Current FDA/EMA-Approved Targeted Drugs for Pediatric Cancer
- Regulatory Initiatives to Augment Pediatric Oncology Drug Development
- Challenges of Pediatric Drug Development and the Opportunities they Create
- Assessing Long-Term Toxicities of Targeted Therapies in Growing and Developing Hosts without Halting Drug Development
- Conclusions
- Disclosure of Potential Conflicts of Interest
- Acknowledgments
- Footnotes
- References
- Figures & Data
- Info & Metrics