2.1 Patients

Children and young adults (aged ≥2 to ≤18 years in Phase 1, and ≥2 to ≤21 years in Phase 2) with a histologically or cytologically confirmed diagnosis of recurrent or refractory malignant solid tumor (excluding hepatoblastoma and lymphomas), including primary CNS tumors, were eligible for Phase 1. Patients were required to have adequate organ function and a Karnofsky Performance Status (KPS; patients >16 years of age) or Lansky play score (LPS, patients ≤16 years of age) of ≥50. In Phase 1, patients with evaluable disease (non-target lesions only) or measurable disease (target and/or non-target lesions) per Response Evaluation Criteria in Solid Tumors (RECIST) version (v) 1.1 or Response Assessment in Neuro-Oncology (RANO), as appropriate, were eligible. Phase 2 was limited to patients with a diagnosis of EWS, RMS, or HGG (excluding patients with diffuse intrinsic pontine glioma) with measurable disease and ≤2 prior VEGF/VEGFR-targeted therapies (but no prior VEGF/VEGFR-targeted therapy in combination with an mTOR inhibitor).

This study was conducted in collaboration with the Children's Oncology Group (COG) in accordance with the Good Clinical Practice guidelines of the International Council for Harmonization and ethical principles originating from the Declaration of Helsinki. All patients and/or their parent(s) or legally authorized representative(s) signed an informed consent/assent form before any study-specific procedures were performed.

2.2 Study Design and Treatments

Study 216/COG-ADVL1711 (NCT03245151) was a multicenter, open-label, single-arm, Phase 1/2 trial of lenvatinib in combination with everolimus, each administered orally once daily in 28-day treatment cycles. Lenvatinib was provided as hard capsules, and everolimus was administered as a dispersible tablet. Younger patients who were unable to swallow tablets received an oral extemporaneous solution of lenvatinib. The Phase 1 portion used a rolling-six design [25] with dose escalation in sequential cohorts. Dose-limiting toxicity (DLT) assessment for dose escalation was limited to the first cycle; therefore, the “treatment phase” was defined as 4 weeks of therapy in Phase 1. Patients with nonprogressive disease could subsequently receive up to 24 cycles of treatment. The initial dose level (dose level 1; DL1) for lenvatinib was 11 mg/m² (the maximum dose was 18 mg/day), representing approximately 80% of the single-agent maximum tolerated dose (MTD) of lenvatinib in pediatric and young adult patients with solid tumors (14 mg/m²) [21]. The initial dose of everolimus was 3 mg/m², which is 66% of the FDA-approved dose for adults and pediatric patients with tuberous sclerosis complex-associated subependymal giant cell astrocytoma [26]. The maximum everolimus dose was 5 mg/day (equivalent to the adult dose when used in combination with lenvatinib for advanced RCC [19]). A single de-escalation to dose level −1 (DL –1), consisting of lenvatinib 8 mg/m² plus everolimus 3 mg/m², was incorporated as a contingency for toxicity at DL1. The Phase 2 portion was initiated at RP2D in patients with a diagnosis of EWS (Cohort 1), RMS (Cohort 2), or HGG (Cohort 3). In Phase 2, the “treatment phase” was defined as 16 weeks of therapy. Patients could subsequently proceed with treatment in an “extension phase” for up to 24 cycles. All patients could receive study treatment until disease progression, lack of clinical benefit (per investigator's judgment), unacceptable toxicity, withdrawal from the study for any reason, or termination of the study by the sponsor. The study design is shown in Figure S1.

2.3 Study Assessments and Endpoints

DLT and safety assessments included monitoring and recording all adverse events (AEs) per the Common Terminology Criteria for Adverse Events version 4.03. DLTs attributable to the investigational agents were to include Grade 4 thrombocytopenia or neutropenia; any Grade ≥2 arterial thromboembolic events; any nonhematologic Grade 4 toxicity with the exception of fever <5 days; Grade 3 or 4 febrile neutropenia; any Grade 3 nonhematologic toxicity with exceptions of nausea, vomiting, diarrhea, or headache <3-day duration; weight loss; fever or infection <5 days; liver enzyme elevation (aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transferase, bilirubin, or alkaline phosphatase), amylase, lipase, or triglycerides that improve to Grade ≤1 or baseline within 7 days; proteinuria unless confirmed within 72 hours; any Grade 2 toxicity persisting for ≥7 days and considered sufficiently medically significant or intolerable by patients despite optimal supportive care; and any dose interruption or reduction due to toxicity resulting in administration of <75% of the planned dosage of lenvatinib and/or everolimus. Dose-limiting hypertension was defined as Grade 4 hypertension with confirmed systolic or diastolic blood pressure >25 mmHg above the 95th percentile for sex, age, height/length; or elevated diastolic blood pressure (i.e., >95th percentile for age) not controlled by a single antihypertensive medication within 14 days of study drug use. Investigator-determined response assessments were based on RECIST v1.1 for all tumor types except HGG, where assessments were based on RANO (see Supporting Information Methods for details). Serial blood samples for plasma concentrations of lenvatinib and whole blood concentrations of everolimus were collected for pharmacokinetic (PK) analyses.

For Phase 1, the primary endpoints were MTD, RP2D, and safety/toxicity. PK was a secondary endpoint, and provisions were made to ensure that data were available from at least six patients younger than the age of 6 years. For Phase 2, the primary endpoint was the objective response rate (ORR) at Week 16. ORR was defined as the sum of complete response (CR) and partial response (PR). For both phases, secondary endpoints included ORR at the time of data cutoff, disease control rate (proportion of patients who had the best overall response [BOR] of CR, PR, or stable disease [SD; duration of ≥7 weeks since the first dose of the study treatment]), clinical benefit rate (CBR; defined as the proportion of patients who had a BOR of CR, PR, or durable SD [duration ≥23 weeks since the first dose of the study treatment]), and duration of response (DOR).

Safety was evaluated in all patients who received at least one dose of the study drug (safety analysis set). In the safety analysis set, PK analyses were conducted in all patients with at least one measurable post-dose plasma concentration of lenvatinib and an adequately documented dosing history. Plasma concentrations of lenvatinib were determined using validated liquid chromatography with tandem mass spectrometry. The lower limit of quantitation for lenvatinib was 0.250 ng/mL in human plasma. Phase 2 efficacy endpoints were evaluated in all patients with measurable disease at baseline and underwent at least one post-baseline efficacy assessment (evaluable analysis set; additional details are included in the Supporting Information Methods).

2.4 Statistical Methods

Dose escalation utilized a rolling-six design [25] in which the MTD was considered exceeded, and dose escalation stopped if more than 33% of patients (≥2/6) experienced DLTs. If two DLTs at a given dose level were of different AE classes, at least one did not appear to be dose related, and both were readily reversible, then expansion of the cohort could be considered. RP2D was to be declared based on the totality of safety data observed during Phase 1. Secondary safety endpoints were summarized descriptively. ORR was assessed only in patients with measurable disease at screening/baseline. Patients with evaluable disease were analyzed separately for BOR. None of the patients treated in Phase 1 were included in Phase 2 cohorts. In Phase 2, a 10+10 Simon's optimal two-stage design was used for each cohort (10 evaluable patients per stage; see Supporting Information Methods for additional details) with the goal of detecting an ORR of 20% (88% power to detect a 20% increase in response rate with one-sided alpha = 0.07 assuming a null response rate of 5% and alternative response rate of ≥25%). If ≥1 responses were observed after the first 10 patients were treated, enrollment in the cohort would continue for up to 20 patients total. If no objective responses (CR/PR) were observed in the first 10 patients, enrollment was ceased, and the therapy was deemed inactive for the cohort. For an expanded cohort, if ≤2 objective responses were observed, the lenvatinib/everolimus combination was deemed insufficiently active for that cohort. ORR (at Week 16 and the cutoff date) and CBR were summarized descriptively, with 95% confidence intervals (CIs) determined using the Clopper–Pearson exact method. DOR was analyzed using the Kaplan–Meier method. Database lock occurred after all patients had withdrawn from the study and were no longer in survival follow-up.

3.1 Patient Disposition and Baseline Characteristics

In Phase 1, 23 patients met eligibility criteria and were treated at dose level 1 (DL1) or −1 (DL −1). In Phase 2, 41 patients met eligibility criteria and received ≥1 doses of study treatment (EWS, n = 10; RMS, n = 20; HGG, n = 11, with one patient not evaluable for tumor response). At the time of database lock (November 14, 2022), all patients had discontinued study treatment. Additional details on patient disposition are shown in Figure S2.

Baseline demographic and disease characteristics are shown in Table 1 and Table S1. In Phase 1, most patients were White (60.9%), had a KPS/LPS of ≥80 (82.6%), were <17 years of age (91.3%, including 30.4% <6 years of age), and were female (52.2%). In Phase 2, most patients were White (75.6%), had a KPS/LPS of ≥80 (83%), were <17 years of age (61%), and were male (53.7%). Patients in both Phase 1 and Phase 2 were heavily pretreated with a median (range) number of prior regimens of two (1–7) and two (1–10), respectively. Of patients in Phase 1 and Phase 2, 78.3% and 85.4% had received ≥2 prior regimens, respectively, and 21.7% and 14.6%, respectively, had received prior VEGF-targeted therapy (Table 1).

3.2 DLTs (Phase 1)

Three patients were initially enrolled at DL1 (i.e., lenvatinib 11 mg/m² plus everolimus 3 mg/m²); two of these patients had Grade 3 AEs that were initially thought to meet DLT criteria (proteinuria, n = 1; headache, n = 1). Five patients were subsequently enrolled at DL −1 (lenvatinib 8 mg/m² plus everolimus 3.0 mg/m²) with no observed DLTs. On re-review, one of the potential DLTs (Grade 3 headache) reported at DL1 did not meet the definition of a DLT (for details, see Supporting Information Results), and hence, additional enrollment reopened at DL1. Of three additional patients enrolled at DL1, one patient had DLTs of Grade 3 hypertriglyceridemia and Grade 4 hypercholesterolemia. The DLTs resulting in dose modifications during Cycle 1 are summarized in Table 2. Because the two patients treated at DL1 had DLTs from different classes of AEs (proteinuria and lipid abnormalities), the cohort was expanded to 12 patients with no additional DLTs observed. Hence, DL1 (lenvatinib 11 mg/m² plus everolimus 3 mg/m²) was determined to be the MTD and RP2D. After the RP2D was defined, six additional patients were enrolled at DL1 to obtain evaluable PK data from at least six patients aged 2 to <6 years. No DLTs were observed in these six patients.

3.3 Safety

Treatment-emergent AEs (TEAEs) and treatment-related TEAEs observed in Phases 1 and 2 are presented in Table 3, Tables S2 and S3, and Supporting Information Results. Two patients (8.7%), both receiving DL1 in Phase 1, discontinued treatment with lenvatinib (4.3%), everolimus (8.7%), or both (4.3%) due to a treatment-related TEAE (Patient 1: hypercholesterolemia and hypertriglyceridemia; Patient 2: tendon rupture). Treatment-related TEAEs of any grade occurred in all patients treated in Phase 1; the most common treatment-related TEAEs (>40% of patients) were hypertension (60.9%), hypothyroidism (52.2%), hypertriglyceridemia (47.8%), abdominal pain (43.5%), and diarrhea (43.5%). Grade ≥3 severity treatment-related TEAEs were observed in 13 (56.5%) patients. Treatment-related pneumothorax was observed in two patients treated at DL1. The most common Grade 3 treatment-related TEAEs (occurring in three patients each) were hypertriglyceridemia and proteinuria.

In Phase 2, three (7.3%) patients (one in RMS and two in HGG cohorts) discontinued treatment with lenvatinib (4.9%), everolimus (7.3%), or both (2.4%) due to a treatment-related TEAE (Patient 1: pancreatitis, increased lipase, and increased amylase; Patient 2: hypercholesterolemia and hypertriglyceridemia; Patient 3: hypertriglyceridemia, skin ulcer). Treatment-related TEAEs were reported in 38 (92.7%) patients. The most common treatment-related TEAEs (occurring in >40% of patients) were hypertriglyceridemia (n = 23, 56.1%) and proteinuria (n = 18, 43.9%). Treatment-related pneumothorax was observed in one patient (RMS cohort). The most common Grade ≥3 treatment-related TEAEs (occurring in seven [17.1%] patients each) were hypertriglyceridemia and lymphopenia.

Of the 34 deaths observed in Phase 2, there were eight within 28 days of the patient's last dose of the study drug. Seven patients died from disease progression, and one patient died due to disease-related encephalopathy that was considered unrelated to study treatment.

3.4 Efficacy

Secondary efficacy endpoints for Phase 1 are summarized in Table S4. Eighteen patients with measurable disease and five with evaluable disease were evaluated for BOR. No objective responses occurred at either dose level. At DL1, seven (38.9%) patients with measurable disease had a BOR of SD, and two (11.1%) patients with evaluable disease had a BOR of non-CR/non-PD (equivalent to SD in patients with measurable disease). The remainder had PD, except for one patient whose BOR was not evaluable because no post-baseline tumor assessment was available. At DL −1, a BOR of SD was reported in one patient with measurable disease; the remainder had PD. In Phase 1, a total of five patients (one treated at DL −1 and four at DL1) had durable SD (i.e., SD ≥23 weeks) with the following diagnoses: astrocytoma (DL −1), ependymoma, HGG, osteosarcoma, and Wilms tumor.

BOR and duration of treatment for Phase 2 patients are shown in Figure 1. In Phase 2, no objective responses (PR/CR) were observed in the EWS and HGG cohorts at Week 16, and further enrollment was stopped due to futility after Stage 1. In the RMS cohort, one patient in Stage 1 had a confirmed PR at Week 16, prompting cohort expansion to 20 patients. One additional patient demonstrated a PR at Week 16, resulting in an ORR of 10% for the RMS cohort (Table 4). However, the ORR for the RMS cohort did not meet the statistical threshold for activity. The DORs for the two patients with PRs were 2.10 and 2.76 months, respectively. The molecular characteristics of these two responders are included in the Supporting Information Results. An additional 13 (32.5%) patients had BORs of SD, which were similarly distributed across disease types (EWS n = 4, 40%; RMS n = 6, 30%; HGG n = 3, 30%). Patients with nonprogressive disease (n = 5) proceeded with treatment in the extension phase. The CBRs for patients with EWS, RMS, and HGG were 20%, 10%, and 0%, respectively (Table 4). Notably, two patients with EWS experienced SD lasting ≥23 weeks.

3.5 Pharmacokinetics

A summary of PK parameters of lenvatinib for Phase 1 is provided in Table 5. At DL −1, C_max and area under the curve (AUC) means were similar between Cycle 1 Day 1 (C_max: 240 ng/mL [standard deviation: 131]; AUC: 1230 ng*h/mL [standard deviation: 740]) and Cycle 1 Day 15 (C_max: 314 ng/mL [standard deviation: 150]; AUC: 1330 ng*h/mL [standard deviation: 521]). At DL1, both C_max and AUC were also similar between Cycle 1 Day 1 (C_max: 404 ng/mL [standard deviation: 121]; AUC: 1880 ng*h/mL [standard deviation: 549]) and Cycle 1 Day 15 (C_max: 448 ng/mL [standard deviation: 273]; AUC: 2140 ng*h/mL [standard deviation: 1160]). C_max and AUC were approximately dose-proportional between DL −1 and DL1 on both cycle days. Lenvatinib plasma concentration data for Phase 1 are shown in Figure S3. The DLTs experienced by patients at DL1 (i.e., lenvatinib 11 mg/m²) were not associated with high plasma concentrations of lenvatinib (Figure S4). Population PK analyses for everolimus were not performed because of the lack of efficacy for the combination.