Atezolizumab for children and young adults with previously treated solid tumours, non-Hodgkin lymphoma, and Hodgkin lymphoma (iMATRIX): a multicentre phase 1–2 study

Articles

Atezolizumab for children and young adults with previously treated solid tumours, non-Hodgkin lymphoma, and Hodgkin lymphoma (iMATRIX): a multicentre phase 1–2 study Birgit Geoerger, C Michel Zwaan, Lynley V Marshall, Jean Michon, Franck Bourdeaut, Michela Casanova, Nadège Corradini, Gianluca Rossato, Mufiza Farid-Kapadia, Colby S Shemesh, Katherine E Hutchinson, Francis Donaldson, Minlei Liao, Hubert Caron, Tanya Trippett

Summary

Background Atezolizumab is an inhibitor of PD-L1, which can lead to enhanced anticancer T-cell activity. We aimed to evaluate the safety, pharmacokinetics, and activity of atezolizumab in children and young adults with refractory or relapsed solid tumours, with known or expected PD-L1 expression. Methods iMATRIX was a multicentre, open-label, phase 1–2 trial of patients (aged <30 years) with solid tumours or lymphomas recruited from 28 hospitals in ten countries (USA, France, Italy, UK, Spain, the Netherlands, Denmark, Israel, Switzerland, and Germany). Eligible patients younger than 18 years received 15 mg/kg atezolizumab (maximum 1200 mg); patients aged 18–29 years received the adult dose (1200 mg) until disease progression or loss of clinical benefit. Co-primary endpoints were safety (assessed by incidence of adverse events) and pharmacokinetics (assessed by serum atezolizumab concentrations). Secondary endpoints included the proportion of patients achieving an objective response. This trial is registered with ClinicalTrials.gov, number NCT02541604. Findings Between Nov 5, 2015, and April 2, 2018, we screened 115 patients, 25 of whom did not meet the inclusion criteria. 90 patients, with a median age of 14 years (IQR 10–17), were enrolled. At the data cutoff (April 2, 2018), two patients remained on study treatment. 87 (97%) of 90 patients received at least one dose of atezolizumab at 15 mg/kg or 1200 mg and were evaluable for safety. Three patients were not treated owing to either poor clinical condition or withdrawal of consent. In the safety-evaluable population (n=87), the most common adverse events were pyrexia (36 [41%] patients) and fatigue (31 [36%]). The most common grade 3–4 adverse event was anaemia (19 [22%] patients). The most commonly reported serious adverse events were in the categories of infections and infestations; pyrexia was the only serious adverse event reported in more than two patients. 57 (66%) patients had at least one treatment-related adverse event (grade 1–4); fatigue was the most common treatment-related adverse event (17 patients [20%]). There were no fatal adverse events. Mean serum concentrations of atezolizumab were overlapping and comparable between children receiving 15 mg/kg and young adults receiving 1200 mg of atezolizumab every 3 weeks. Serum concentrations of atezolizumab were above the target exposure level in all patients. At 6 months, four patients (5%) achieved an objective response (all partial responses). Interpretation Although response to atezolizumab was restricted, atezolizumab was well tolerated with generally comparable exposure across populations. Our findings might help to define future development strategies for immune checkpoint inhibitors either by focusing research to specific disease subpopulations that exhibit greater benefit from immune checkpoint inhibitors, or by providing the means to identify therapeutic combination partners that augment T-cell infiltration and proliferation in so-called immune cold tumour microenvironments. Funding F Hoffmann-La Roche. Copyright © 2019 Elsevier Ltd. All rights reserved.

Introduction The proportion of children and young adults surviving with relapsed or refractory cancers is low, representing a high level of unmet medical need for which novel treatment strategies are required. PD-L1 expression has been reported sporadically in several paediatric cancers.1 About 9% of paediatric tumours are positive for PD-L1 (>1% threshold); Burkitt lymphoma, glioblastoma, and neuroblastoma are the most common PD-L1-positive tumour types.1 Atezolizumab is an immune checkpoint inhibitor which, by blocking PD-L1, leads to enhanced anticancer

T-cell activity.2,3 Atezolizumab is approved by the US Food and Drug Administration and the European Medicines Agency for the treatment of adults with previously treated, locally advanced or metastatic urothelial carci­ noma, small-cell lung carcinoma, or non-small-cell lung carcinoma (NSCLC);4,5 with locally advanced or metastatic triple-negative breast cancer in combination with nabpaclitaxel;4 and in combination with bevacizumab, paclitaxel, and carboplatin, in some NSCLCs.4 Data regarding the efficacy and tolerability of atezolizumab in children are absent. PD-L1 is highly expressed on

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Lancet Oncol 2019 Published Online November 25, 2019 https://doi.org/10.1016/ S1470-2045(19)30693-X See Online/Comment https://doi.org/10.1016/ S1470-2045(19)30777-6 Gustave Roussy Cancer Center, Department of Pediatric and Adolescent Oncology, Université Paris-Saclay, Villejuif, France (B Geoerger MD); Department of Pediatric Oncology, Erasmus MC-Sophia Children’s Hospital, Rotterdam, Netherlands (C M Zwaan MD); Princess Máxima Center, Utrecht, Netherlands (C M Zwaan); Paediatric and Adolescent Oncology Drug Development Unit, The Royal Marsden NHS Foundation Trust, London, UK (L V Marshall MB BCh); The Institute of Cancer Research, London, UK (L V Marshall); Service de Pédiatrie, SIREDO Center, Curie Institute, Paris, France (J Michon MD, F Bourdeaut MD); Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy (M Casanova MD); Haematology and Oncology Paediatric Institut, Léon Bérard Centre, Lyon, France (N Corradini MD); F Hoffmann-La Roche, Basel, Switzerland (G Rossato PhD, H Caron MD); F Hoffmann-La Roche, Mississauga, Canada (M Farid-Kapadia MD); Genentech, San Francisco, CA, USA (C S Shemesh PhD, K E Hutchinson PhD, M Liao MA); F Hoffmann-La Roche, Welwyn, UK (F Donaldson PhD); and Memorial Sloan Kettering Cancer Center, New York, NY, USA (T Trippett MD)

1

Articles

Correspondence to: Dr Birgit Geoerger, Gustave Roussy Cancer Center, Department of Pediatric and Adolescent Oncology, Université Paris-Saclay, Villejuif 94805, France birgit.geoerger@ gustaveroussy.fr

Research in context Evidence before this study We searched PubMed on Dec 4, 2018 for original and review articles using the search terms: “PD-L1”, “immune checkpoint inhibitors”, “atezolizumab”, “ipilimumab”, “nivolumab”, and “pembrolizumab”, without language or date restrictions. The expression of PD-L1 by malignant cells has been reported in many paediatric tumour types. In adult patients with refractory late-stage cancer, encouraging results have been reported in early-phase clinical trials targeting the PD-L1 and PD-1 pathway. Atezolizumab is an immune checkpoint inhibitor that targets PD-L1 and blocks interaction with its receptor PD-1. Retrospective analyses of seminal studies leading to the approval of atezolizumab in adults showed that longer and more frequent responses were observed in patients with tumours harbouring PD-L1 expression. Added value of this study To our knowledge, this trial is the first to characterise the safety, pharmacokinetics, and activity of atezolizumab in

tumour-infiltrating leukocytes and tumour-associated macrophages.6,7 Paediatric tumours are highly infiltrated with tumour-associated macrophages compared with adult tumours.8 As these macrophages have an important immunosuppressive role in solid tumours,9,10 suppression of tumour-associated macrophage activity with an antiPD-L1 antibody was of therapeutic interest for paediatric cancers at the time of study conception. This open-label phase 1–2 study aimed to investigate the antitumour activity of single-agent atezolizumab in children and young adults with recurrent or refractory solid tumours with known or expected PD-L1 expression.

Methods

Study design and participants

See Online for appendix

2

iMATRIX was a phase 1–2, multicentre, open-label study designed to evaluate the safety, tolerability, pharma­ cokinetics, pharmacodynamics, immunogenicity, and preliminary activity of atezolizumab in children and young adults with relapsed or refractory solid tumours and lymphomas. A gated stepwise study design was used (appendix p 10). Patients were recruited from 28 hospitals in ten countries: the USA (seven centres), France (three), Italy (five), the UK (five), Spain (three), the Netherlands (one), Denmark (one), Israel (one), Switzerland (one), and Germany (one). Key inclusion criteria were patients younger than 30 years with solid tumours, Hodgkin lymphoma, or non-Hodgkin lymphoma, with known or expected PD-L1 pathway involvement and for whom previous treatment had proven ineffective, or for whom no curative standardof-care treatment options existed. A life expectancy of at least 3 months was required as was a Lansky playperformance scale (aged <16 years) or Karnofsky performance status (aged ≥16 years) score of at least 50.

children and young adults with relapsed or refractory solid tumours for whom there is no available effective standard treatment. Our trial provides data on the PD-1L inhibitor that are complementary to two large clinical trials on two PD-1 inhibitors. Our results suggest that the safety and pharmacokinetic profile of atezolizumab in children is similar to that seen in adults. Similar to other reports on PD-1-targeting antibodies, response to atezolizumab in children and young adults with classical solid tumours is restricted. Implications of all the available evidence These findings provide important data on atezolizumab activity in children and young adults and broaden our knowledge in this rare disease population. Our findings might also help to inform future development of checkpoint inhibitors as monotherapy or in combination with other anticancer therapies in children and young adults.

Histological or cytological confirmation of tumour type was also required for study inclusion. At the start of the study, the role of PD-L1 expression was a major discussion point for adult cancers and no clear recommendation had been given. Because cutoffs for determining PD-L1 positivity or negativity with available assays had not yet been established (and PD-L1 positivity cutoffs might be different for the various tumour types enrolled in this study), we analysed PD-L1 tumour expression retrospectively. Although we anticipated that most enrolled patients would be younger than 18 years, inclusion of young adults up to 30 years of age was permitted because these patients with paediatrictype tumours, who are treated in paediatric oncology facilities, often have scarce options for participation in clinical research. We excluded patients with primary CNS tumours or CNS metastases, with the exception of atypical teratoid rhabdoid tumour without brain stem involvement. Treatment with high-dose chemotherapy and haem­ atopoietic stem-cell rescue within 3 months before starting study drug was not permitted. Previous allogeneic hematopoietic stem-cell transplantation or previous solid-organ transplantation was also not permitted. Full inclusion and exclusion criteria are available in the study protocol (appendix pp 75–81). An atypical teratoid rhabdoid tumour cohort was added after a confirmed objective tumour response to atezolizumab was seen in a patient with an extracranial malignant rhabdoid tumour. Cohorts of patients with atypical teratoid rhabdoid tumour and malignant rhabdoid tumour were added after the start of the study as part of a protocol amendment (appendix pp 25–26). These cohorts were included after a clinically signifi­ cant response was seen in a patient in the

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Articles

non-rhabdomyosarcoma soft-tissue sarcoma cohort. In addition, since know­ ledge on immune checkpoint inhibitor activity in paediatric solid tumours (including ultra-rare tumour subtypes) at the time of study initiation was scarce, a study cohort without known documented PD-L1 expression was included to investigate these rare solid tumours. We used revised Response Criteria for Malignant Lymphoma11 to assess disease in patients with Hodgkin lymphoma or non-Hodgkin lymphoma, International Neuroblastoma Response Criteria12 for neuroblastoma, Response Assessment in Neuro-Oncology13 for atypical teratoid rhabdoid tumour, and Response Evaluation Criteria in Solid Tumours (version 1.1)14 for all other tumour types. Because conventional response criteria might not always be adequate to characterise the antitumour activity of immunotherapeutic agents such as atezolizumab, which can produce delayed responses that can be preceded by initial apparent radiological progression, including the appearance of new lesions, we used modified response criteria in the study as secondary efficacy measures to account for the possible appearance of new lesions and to allow radiological progression to be confirmed at a subsequent assessment. These included immune-modified RECIST (version 1.1) and immunerelated response criteria for patients with neuroblastoma, Hodgkin lymphoma, and non-Hodgkin lymphoma. Patients’ tumours were retrospectively analysed for PD-L1 expression with techniques and thresholds consistent with adult studies. The current study is part of the agreed European Paediatric Investigational Plan EMEA-001638-PIP01 and US Written request (IND number 124026) voluntarily submitted by the Sponsor in the context of treatment of malignant neoplasms. This study was done in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice, and was approved by the Institutional Review Board and Ethics Committee according to local law and regulations in each participating country. Written, informed consent was obtained before participation in the study. An independent data monitoring committee monitored patient safety throughout the study.

Procedures Patients received atezolizumab (Roche Products, Grenzach-Wyhlen, Germany) by intravenous infusion on day 1 of each 3-week cycle until disease progression or loss of clinical benefit. Patients younger than 18 years received 15 mg/kg atezolizumab (maximum 1200 mg). Patients aged 18–29 years received the adult flat dose of 1200 mg atezolizumab. We used a bodyweight-adjusted dosing strategy based on linear atezolizumab pharma­ cokinetics in the dose range of 1−20 mg/kg,15 which includes the approved dose in multiple cancer types (1200 mg; equivalent to a bodyweight-based dose of 15 mg/kg, assuming an adult bodyweight of 80 kg)

and evidence from previous studies16–18 of no clinically significant differences in pharmacokinetic profiles between adults and children. Given that the present study was, to our knowledge, the first paediatric investigation of atezolizumab, these properties (ie, pharmacokinetics, safety, and preliminary activity in the pediatric population) were unconfirmed and warranted clinical evaluation. Because of the known potential for pseudoprogression or tumour immune infiltration, patients were allowed to remain on study treatment after apparent radiographic progression, providing that the benefit–risk ratio was judged by the study investigator to be favourable. Furthermore, it was recommended that radiological progression be confirmed at a subsequent tumour assessment, and that progressive disease in responding patients be confirmed when feasible by a biopsy of the growing or new lesion. We evaluated safety according to National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0) throughout the study. Adverse events were recorded during the study at every study visit and for 30 days after the last dose of study treatment (or until the initiation of another anticancer therapy). We used a broad search strategy to capture infusion-related reactions using a prespecified infusion-related reaction and hypersensitivity grouping of MedDRA preferred terms. Study drug could be temporarily withheld if a patient experienced an adverse event that was considered to be related to atezolizumab; if withheld for more than 105 days, treatment would be discontinued. For pharmacokinetic assessments, atezolizumab concentration was measured in serum by ELISA (ICON Laboratories Services, Whitesboro, NY, USA). Trough serum concentrations (Cmin) of atezolizumab were determined before infusion on day 1 of cycles 2–4, 8, 12, and 16, then every eighth cycle thereafter. Maximum serum concentrations were determined after the infusion on day 1 of cycles 1 and 4. Data were analysed by age and bodyweight. We did an initial interim pharmacokinetic analysis on March 16, 2016, after the first five patients completed cycle 1, irrespective of age and tumour type. We did additional interim pharmacokinetic analyses on May 10, 2016, after five patients (aged <6 years) had completed cycle 1 and after a minimum of 20 patients had completed cycle 1. After these interim pharma­ cokinetic analyses and review, no dose modifications were deemed necessary to match adult exposures and to ensure an optimal safety profile. Pharmacokinetic and safety analyses were done across all tumour types and were centrally reviewed by the independent data monitoring committee. An initial response assessment was done by tumour type. An additional response assessment was planned, dependent on the number of patients meeting predefined activity thresholds (based on the proportion of historical control patients achieving an objective response; appendix p 1) at the initial assessment.

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

3

Articles

115 participants screened

25 excluded 25 did not meet inclusion criteria

90 enrolled 90 included in intention-to-treat population

(TC 2–3: ≥5% PD-L1-positive tumour cells; IC 2–3: ≥10% PD-L1-positive infiltrates), low expression (TC1 or IC1: ≥1% to <5% PD-L1-positive tumour cells or infiltrates), or no expression (TC0 or IC0: <1% PD-L1-positive tumour cells or infiltrates). PD-L1 expression on tumour samples from patients with non-Hodgkin lymphoma and Hodgkin lymphoma were analysed in the same manner as tumour samples from patients with solid tumours.

Outcomes 3 not treated

87 received atezolizumab 87 included in the safety-evaluable population

74 discontinued study 64 died 5 withdrew from study 4 lost to follow-up 1 other

13 on study at data cutoff (April 2, 2018) 2 on treatment 11 on follow-up

Statistical analysis

Figure 1: Trial profile

Radiographic tumour assessments were done locally by CT, MRI, ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) PET scan, or metaiodo­benzylguanidine scans at screening and at the end of cycles 2, 4, 6, and 8, and every fourth cycle thereafter, without central confirmation. Laboratory monitoring (haematology, serum chemistry, amylase, lipase, and thyroid functioning) was done locally at screening and at cycles 2, 4, 6, and 8, and every fourth cycle thereafter. We assessed the presence of anti-drug antibodies in response to atezolizumab using validated bridging ELISAs. Samples were analysed centrally at ICON Laboratories Services. Potential associations between treatment-emergent anti-drug antibodies and other clinically relevant outcomes, such as safety, were investigated. Changes in anti-tetanus antibody titres compared with before atezolizumab treatment were also assessed. For analysis of PD-L1 expression, representative tumour specimens collected from an intervention (surgical procedure or biopsy) before first dose of drug (fresh or archival), were required for entry into the study. Archival tumour tissue was dated to any timepoint before study entry (including diagnosis). Immunohistochemistry for PD-L1 was done on baseline (pre-treatment) tumour tissue samples. PD-L1 expression was categorised in a central laboratory by tumour cell (TC) and immune cell (IC) immunohistochemistry score using the validated SP142 antibody assay (Ventana; Tucson, AZ, USA). Patients were categorised as having high expression 4

The primary endpoints were the safety (assessed by incidences of adverse events) and pharmacokinetics (assessed by serum concentrations) of atezolizumab. Secondary endpoints were preliminary antitumour activity of atezolizumab (proportion of patients achieving an objective response, clinical benefit, and progression-free survival, including duration of objective response and overall survival) and immunogenicity of atezolizumab (frequency of treatment-emergent anti-drug antibodies relative to anti-drug antibody prevalence at baseline). Correlation between response to atezolizumab and PD-L1 tumour expression was a prespecified exploratory endpoint. We calculated the sample size needed for the study using historical control objective rates of response for each paediatric tumour type. We aimed to enrol a minimum of 40 patients across all tumour types, with at least two tumour-type cohorts enrolling a minimum of ten patients before study recruitment closure. No formal hypothesis testing was done. Safety analyses were done in the safety-evaluable population, which included all patients who received at least one dose of study drug. Pharmacokinetic analyses were done in patients who had received at least one dose of study drug and had at least one serum concentration result available at clinical data cutoff (Dec 7, 2017). Preliminary activity was assessed in the safety-evaluable population. Immunogenicity analyses were done in patients with at least one anti-drug antibody assessment. Correlation between PD-L1 expression and atezolizumab activity was assessed in patients who had available and analysable tissue samples. As interim analyses, we included an initial safety evaluation phase with predefined criteria for safety and pharmacokinetics. We used the cross-indication safety run-in phase to avoid enrolment of potentially unnecessary numbers of children. We did an initial interim pharmacokinetic analysis after the completion of cycle 1. Patient data were defined as missing if no post-baseline response assessments were available, and were defined as non-evaluable if all post-baseline response assessments were reported as not evaluable, or if assessment of stable disease occurred within 6 weeks from baseline. Statistics were descriptive; mean, median, and SD were calculated for continuous variables, and proportions

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Articles

Participants (n=90) Age, years

14 (10–17)

Age group

Participants (n=90) (Continued from previous column) Tanner staging‡

<2 years

2 (2%)

Stage 1

2 to <12 years

29 (32%)

Stage 2

7 (9%)

12 to <18 years

41 (46%)

Stage 3

14 (18%)

≥18 to <30 years

18 (20%)

Stage 4

8 (10%)

Stage 5

29 (37%)

Sex Male

49 (54%)

Female

41 (46%)

Race Black or African American White

Pre-menarchal

3 (3%)

Time from initial diagnosis, months

6 (7%)

Lines of previous therapy, n

52 (58%)

Multiple

2 (2%)

Unknown

27 (30%)

Weight, kg

45·7 (29·4–60·0)

Body-mass index, kg/m2

18·4 (15·6–22·8)

Distribution of tumour types Ewing sarcoma

11 (12%)

Neuroblastoma

11 (12%)

Non-rhabdomyosarcoma soft-tissue sarcoma

9 (10%)

Osteosarcoma

12 (13%)

Rhabdomyosarcoma

10 (11%)

Wilms’ tumour

10 (11%)

Hodgkin lymphoma

9 (10%)

Non-Hodgkin lymphoma

3 (3%)

Malignant rhabdoid tumour

3 (3%)

Atypical teratoid rhabdoid tumour

3 (3%)

PD-L1-positive*

4 (4%)

PD-L1-negative*

5 (6%)

Lansky or Karnofsky performance score† 100

41 (46%)

90

25 (28%)

80

13 (14%)

70

10 (11%)

60

Female reproductive status§ Post-menarchal

Asian

1 (1%) (Table 1 continues in next column)

were calculated for categorical variables. We used R (version 3.3.1) with additional CRAN packages for descriptive statistics. This trial is registered with ClinicalTrials.gov, number NCT02541604.

Role of the funding source The study was designed by the funder (F Hoffmann-La Roche) and academic authors. F Hoffmann-La Roche and Genentech were the sponsors of the trial and were involved in the administration and conduct of study procedures, coordination of data collection, data analysis, and data interpretation. Authors employed by F Hoffmann-La Roche (GR, MF-K, FD, and HC) and by Genentech (KEH and ML) contributed to the writing and approval of the manuscript. Clinical data were

20 (26%)

Previous radiotherapy

19 (46%) 22 (54%) 26·4 (14·72–44·12) 6 (3–10) 58 (64%)

Previous surgery

78 (87%)

Previous systemic therapy

90 (100%)

Intention-to-treat population. Data are median (IQR) or n (%). The cohort with documented PD-L1 expression, based on local assessments, included fibrolamellar hepatocellular carcinoma, signet ring cell carcinoma, yolk sac tumour, and nasopharyngeal carcinoma. The cohort without documented PD-L1 expression included abdominal carcinoma, renal neuroendocrine tumour, renal cell carcinoma, midline epidermoid carcinoma, and yolk sac tumour. *PD-L1 status was confirmed retrospectively. †Karnofsky was used for patients aged >16 years and Lansky was used for patients aged ≤16 years. ‡Data available for 78 patients. §Data available for 41 patients.

Table 1: Baseline characteristics

collected by the academic authors and their research teams, and were interpreted by the authors and the funder. The corresponding author had full access to all of the data in the study and had final responsibility for the decision to submit for publication.

Results Between Nov 5, 2015, and April 2, 2018, we screened 115 patients, of whom 25 did not meet the inclusion criteria. 90 patients were enrolled and included in the intention-to-treat population (figure 1). 87 (97%) of 90 patients received at least one dose of atezolizumab at 15 mg/kg or 1200 mg and were evaluable for safety (figure 1). Three patients were not treated owing to either poor clinical condition or withdrawal of consent. 13 patients remained on the study at the data cutoff (April 2, 2018). 72 (80%) of 90 patients were younger than 18 years (table 1). Two patients younger than 2 years had malignant rhabdoid tumour. Most patients had metastatic disease and were heavily pretreated with a median of six lines of previous therapy (all had previous chemotherapy, most also had radiotherapy and surgery). Most of the patients received atezolizumab for fewer than 2 months (median 0·8 months, IQR 0·7–2·1) and received a median of two cycles (2–4). There were no dose reductions. The median follow-up time was 6·8 months (IQR 2·9–14·3); for patients who died before the cutoff date, this duration was the time from

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

5

Articles

Grade 1–2

Grade 3

Grade 4

Grade 5

Abdominal pain

3 (3%)

1 (1%)

0

0

Anaemia

3 (3%)

4 (5%)

1 (1%)

0

Cough

0

1 (1%)

0

0

Decreased appetite

6 (7%)

1 (1%)

0

0

Decreased lymphocyte count

4 (5%)

3 (3%)

0

0

Diabetic ketoacidosis

0

0

1 (1%)

0

Dyspnoea

0

1 (1%)

0

0

Fanconi syndrome

0

1 (1%)

0

0

0

0

0

Fatigue

17 (20%)

Febrile neutropenia

0

1 (1%)

0

0

Hypertension

0

1 (1%)

0

0

Hypophosphataemia

0

1 (1%)

0

0

Increased alanine aminotransferase

3 (3%)

2 (2%)

0

0

Increased amylase

0

2 (2%)

1 (1%)

0

Increased aspartate aminotransferase

4 (5%)

1 (1%)

0

0

Increased lipase

0

1 (1%)

0

0

Increased transaminases

0

1 (1%)

0

0

Influenza-like illness

2 (2%)

1 (1%)

0

0

Leukopenia

0

1 (1%)

0

0

Maculo-papular rash

1 (1%)

1 (1%)

0

0

Neutropenia

0

1 (1%)

1 (1%)

0

Papilloedema

0

0

1 (1%)

0

Pleural effusion

0

1 (1%)

0

0

Thrombocytopenia

1 (1%)

3 (3%)

0

0

Sixth nerve disorder

0

1 (1%)

0

0

Vomiting

1 (1%)

1 (1%)

0

0

Grade 1–2 events occurring in ≥10% of the population are shown; grade 1-2 events occurring in <10% of the population are also shown when there was a corresponding event at grades 3–5. All grade 3–5 events are shown. Events are listed alphabetically and were graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0) grade criteria. All counts represent patients; multiple occurrences of the same adverse event in one individual are counted only once at the highest grade.

Table 2: Treatment-related adverse events in the safety-evaluable population by grade (n=87)

enrolment to death or to the date on which patients were last known to be alive. At data cutoff (April 2, 2018), two (2%) of 90 enrolled patients remained on study treatment, 11 (12%) were still being followed up, 85 (94%) had discontinued treatment, and 26 (29%) patients were alive at data analysis. In the safety-evaluable population (n=87), 62 (71%) deaths were due to disease progression. Two additional patients with Hodgkin lymphoma died at days 129 and 330, respectively, after their last dose of atezolizumab; death was due to graft-versus-host disease and multiorgan failure after subsequent anticancer treatments (including other immune checkpoint inhibitors), and allogeneic stem-cell transplant. 20 (22%) of 90 patients in the intention-to-treat population had at least one major protocol deviation event, with 26 major protocol deviation events occurring overall before the clinical cutoff date (April 2, 2018). Identified major protocol deviations belonged to the following categories: eligibility criteria (inclusion and exclusion deviations), study drug administration, 6

assessments (for safety or efficacy assessments), visit timing, informed consent, and laboratory or endpoint deviations. Ten cohorts were enrolled by defined tumour type (table 1). Two further study cohorts including patients with mixed tumour types were categorised, following approval of the Medical Monitor, as with or without PD-L1 expression on tumour or immune cells, documented previously by immunohistochemistry. Atezolizumab was well tolerated. All adverse events (including grades 3 and 4) were similarly frequent across age groups and study cohorts (appendix pp 6–8). The most common adverse events were pyrexia (36 [41%] patients) and fatigue (31 [36%]); pyrexia was also the most common serious adverse event (four [5%] patients) and was the only serious adverse event reported in more than two patients. The most common grade 3–4 adverse event was anaemia (19 [22%] patients). The most commonly reported serious adverse events were infections and infestations. Ten (12%) patients had an adverse event of special interest in the infusion-related reaction events category; 16 (18%) patients had one such event in the hepatitis immune-related category and four (5%) in the pancreatitis immune-related category (appendix p 5). 57 (66%) patients had at least one adverse event considered to be related to the study drug; fatigue was the most common treatment-related adverse event (17 patients [20%]; table 2). Three patients withdrew from study treatment due to adverse events: one with grade 4 diabetic ketoacidosis and grade 3 renal Fanconi syndrome, one with grade 3 increased transaminases, and one with a grade 3 lung infection. There were no fatal adverse events. There were no clinically meaningful differences in system organ class adverse events by age (appendix p 5). Adverse events, serious adverse events, adverse events of special interest, and treatment-related adverse events by age categories are shown in the appendix (p 4), along with overall adverse events of special interest (appendix p 6). All adverse events, regardless of causality, are summarised in the appendix (pp 6–8). Most patients did not have a clinically relevant shift in any laboratory test parameter during the study (data not shown). Most shifts were sporadic events occurring in a small number of patients across all cohorts. Shifts in haematology laboratory tests were similar across cohorts and age groups. Mean serum concentrations of atezolizumab in patients aged 2–17 years receiving 15 mg/kg dose were generally consistent with those in young adults (aged ≥18 years) who received the flat 1200 mg dose (figure 2). Lower serum concentrations were seen in the two infants (aged <2 years) than in the older age groups. The observed serum Cmin concentration at steady state (ie, between cycles 4–8 based on a half-life of about 3 weeks) was above the target serum concentration of 6 μg/mL in all patients.

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Articles

1000

Mean atezolizumab serum concentration (µg/mL)

When analysed by bodyweight, the pharmacokinetic variability from the fixed-dose regimen in young adults was larger than the weight-based regimen in patients younger than 18 years. By contrast to weight-based dosing in children, in young adults, the fixed-dose regimen led to higher exposures in lighter patients and lower exposures in heavier patients. The distribution of exposures by weight-based dosing in the paediatric population substantially overlapped that of the fixed-dose regimen in young adults (appendix p 9). The activity of atezolizumab (ie, objective response, progression-free survival, duration of response, and overall survival) is summarised in the appendix (p 3); best overall response by tumour type is shown in table 3. Four (5%) of 87 patients had a partial res­ponse. Median treatment duration was 19·1 months (IQR 9·4–22·5) in patients with a partial response, 5·0 months (3·7–6·7) in patients with stable disease, and 0·8 months (0·7–2·1) in patients with disease progression. A 4-year-old boy with a malignant rhabdoid tumour of the left paraspinal soft tissue, recurrent after two lines of therapy, had a partial response associated with significant clinical improvement 3 weeks after the first atezolizumab infusion. The sum of the longest diameter of the tumour (RECIST 1.1) decreased by 38% at cycle 2. The tumour tissue had strong PD-L1 immunohistochemistry staining (TC3 or IC2). The patient received four cycles of treatment before discontinuing treatment to undergo surgery, and survived for a further 13·2 months. In a young adult patient with relapsed T-cell nonHodgkin lymphoma, a partial response was seen at cycle 16, with a maximum reduction in the sum of the longest diameter of the target lesion of 54% at cycle 24. After the data cutoff, a complete response was confirmed with a biopsy from the amygdala region (no residual tumour cells). The patient relapsed about 5 months after complete response was confirmed. Of the two patients diagnosed with Hodgkin lymphoma with a best overall response of partial response, one was a 17-year-old girl with multi-located disease who received 32 cycles of atezolizumab and experienced a maximum reduction in the sum of the longest diameter (−91%) at cycle 12. The duration of the response was 22 months. The second patient was a 9-year-old girl who had a maximum reduction in the sum of the longest diameter of 91% at cycle 6 and received 28 cycles (4 months) of atezolizumab when she experienced disease progression. At 6 months, as best overall response, 11 (13%) patients showed stable disease. The other 63 patients had disease progression as their best response during the study. As none of the cohorts met the prespecified minimum number of responders—ie, a best overall response of complete response or partial response (table 3)—none of the tumour type cohorts continued to the additional assessment phase. The frequency of post-baseline anti-drug antibodies for patients with at least one anti-drug antibody result was

100

Age

10

1

0

25

50

75

100

<2 years (n=2) 2 to <12 years (n=29) 12 to <18 years (n=38) ≥18 years (n=18) 125

150

Time since first dose (days)

Figure 2: Atezolizumab exposure Mean serum concentration–time profile of atezolizumab from cycles 1–8 in patients receiving atezolizumab 15 mg/kg or 1200 mg. Patients who received at least one dose of treatment are shown. Error bars are SD.

11 [14%] of 77. Anti-drug antibody status was not associated with adverse events or adverse events of special interest. The frequency of grade 3–4 adverse events was similar in anti-drug antibody-positive patients (seven [64%] of 11 patients) and anti-drug antibody-negative patients (37 [56%] of 66). The mean differences in atezolizumab serum concentrations in patients who developed an antidrug antibody response to atezo­lizumab were similar to, and within the variability and range of, those observed in patients who were anti-drug antibody negative. Analysis of activity by anti-drug antibody status was not possible given the low number of responders. Geometric mean concentrations of anti-tetanus antibodies at study completion or study drug discon­ tinuation remained consistent with concentrations measured before exposure to atezolizumab, and were similar across age groups (data not shown). Baseline tumour PD-L1 expression data were available for 70 patients (figure 3); baseline tumour PD-L1 expression data and response data were available for 63 patients. 11 (17%) of 63 patients with treatment response data had high PD-L1 expression (IC2–3); of whom four (36%) had a partial response, one (9%) had stable disease, and six (55%) had disease progression. The remaining 52 patients (83%) with low (five [10%] of 52) or no (47 [90%] of 52) baseline tumour PD-L1 expression had stable disease (seven [13%] of 52) or disease progression (45 [87%] of 52) as their best overall response. None of the patients with low or no baseline tumour PD-L1 expression achieved a partial response. The cohorts of patients with the greatest proportion of high PD-L1 tumour expression included Hodgkin lymphoma (figure 3; one of these patients did not have a best overall response assessment). The baseline archived tumour from the patient with malignant rhabdoid tumour and a partial response had high PD-L1 expression. In the cohort with documented PD-L1 expres­ sion, one patient (one [33%] of three) with concurrent best overall response data had high PD-L1 expression; this patient experienced disease progression.

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

7

Articles

Number of patients enrolled

Best overall response

Partial response

Stable disease

Number of patients eligible / number required for further expansion* Progressive Missing† disease

NE

Ewing sarcoma

11

0

2

9

0

0

Neuroblastoma

11

0

4

6

0

1

3/10

9

0

0

7

1

1

2/10

Non-rhabdomyosarcoma soft-tissue sarcoma

2/10

Osteosarcoma

10

0

0

10

0

0

3/10

Rhabdomyosarcoma

10

0

0

9

1

0

2/10

Wilms’ tumour

10

0

1

7

2

0

2/10

9

2

2

5

0

0

10/14 6/12

Hodgkin lymphoma Non-Hodgkin lymphoma

3

1

0

1

1

0

Malignant rhabdoid tumour‡

3

1

0

1

1

0

1/6

Atypical teratoid rhabdoid tumour

3

0

0

2

1

0

1/6

PD-L1-positive tumour§

4

0

0

3

1

0

NA

PD-L1-negative tumour¶

4

0

1

3

0

0

NA

87

4

10

63

8

2

··

Total

NA=not applicable. NE=non-evaluable (ie, response could not be measured). PD-L1 status was analysed retrospectively. *Number of patients required for the trial to undergo further expansion. The required number was not reached for any cohort. †No post-baseline response assessments were available. ‡The patient with a partial response in the malignant rhabdoid tumour cohort was initially enrolled into the non-rhabdomyosarcoma soft-tissue sarcoma cohort and subsequently triggered the opening of the malignant rhabdoid tumour-specific cohort; this patient was not considered in the count of number of patients required for further expansion. §Tumour types were fibrolamellar hepatocellular carcinoma, signet ring cell carcinoma, germ cell tumour (yolk sac), and nasopharyngeal carcinoma. ¶Tumour types were abdominal carcinoma, renal neuroendocrine tumour, renal cell carcinoma, and germ cell tumour (yolk sac). Patients were classified as NE if all post-baseline response assessments were reported as not evaluable, or if the assessment of stable disease was within 6 weeks from baseline or study entry.

Table 3: Best overall response by tumour cohort

r ou m tu

AT RT

er 2 Ot h

r

er 1 Ot h

ou

co

id do ab Rh

W

ilm

s’

ar

tu m

m

m co ar

yo s

Rh a

bd

om

te os Os

No

n-

a

a

TS SS

a RM nNo

a m to gk

in l

ym

ph om

a las ob Ne ur

Ho d

ym in l

gk Ho d

Ew

in

g’s s

ar

co

m

a

High 11% 100 17% 17% (1 of 9) 29% 25% 33% 90 (1 of 6) (1 of 6) (2 of 7) (1 of 4) (1 of 3) 80 100% 100% 67% 100% 100% 89% (4 of 4) (3 of 3) 70 (7 of 7) (8 of 9) (9 of 9) (2 of 3) 25% 60 83% 83% 89% 71% (1 of 4) 50 (5 of 6) (5 of 6) (8 of 9) (5 of 7) 40 67% 30 50% (2 of 3) 20 33% (2 of 4) 11% (1 of 3) 10 (1 of 9) 0

ph om

Samples with specified expression level of PD-L1 (%)

PD-L1 expression None Low

Figure 3: Baseline PD-L1 expression by study cohort Data are % (n of N). PD-L1 expression for 70 patients with available PD-L1 expression data. ATRT=atypical teratoid rhabdoid tumour. Non-RMS STS=non-rhabdomyosarcoma soft-tissue sarcoma. Other 1=other tumour types with documented PD-L1 expression. Other 2=other tumour types without documented PD-L1 expression.

Discussion iMATRIX was, to our knowledge, the first trial designed to evaluate the safety, pharmacokinetics, and preliminary activity of atezolizumab in children and young adults with various recurrent or refractory solid tumours. Mean serum concentrations of atezolizumab in children were consistent with those in adults.5 8

Atezolizumab was well tolerated and its safety profile in children was generally consistent with its known safety profile in adults,5 with no new safety signals or notable differences by age group or cohort. Shifts in haematology laboratory tests were consistent with commonly observed adverse events in previous atezolizumab studies. There were some differences in adverse event frequencies compared with adult safety data. Most notably, the frequency of infusion-related reaction events in the present study was higher than that in adults.4 This difference probably relates to the different methodologies used to identify these events; our study used a broader infusion-related reaction search strategy (a prespecified infusion-related reaction or hyper­sensitivity grouping of MedDRA preferred terms reported within 24 h of infusion) than those used for adult programmes (ie, MedDRA preferred terms infusion-related reaction and cytokine release syndrome). We also reported higher frequencies of adverse events of special interest in the immune-related hepatitis and pancreatitis categories in the current study than in an adult population. Most of these cases were grade 1 or 2 transaminase elevations (no events of hepatitis reported) or amylase or lipase elevations. Notably, two patients with Hodgkin lymphoma died due to graft-versus-host disease and multiorgan failure after allogeneic stem-cell transplant at days 129 and 330 after discontinuation from atezolizumab. There have been concerns about the prolonged risk of immunerelated effects after PD1 and PD-L1 checkpoint therapy,

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Articles

which need to be considered in the treatment strategies.19 Immune-related adverse events associated with immuno­ therapy can have a delayed onset and prolonged duration compared with adverse events associated with standard chemotherapy.20 Because of this, clinicians and patients need to be aware of the possible immune-related effects that might occur during treatment and the possibility that effects can appear late in the course of treatment or even years after treatment discontinuation.20 Consistent with data from most adult populations, a higher frequency of adverse events by anti-drug antibody status was not observed in our study. Adverse event frequencies did not differ between age groups or between cohorts. Because only two patients younger than 2 years were enrolled, comparisons of safety data in this age group with other age groups were not feasible. Our pharmacokinetic data showed that serum concentrations of atezolizumab in children (aged 2–17 years) are generally consistent with those in adults,5 showing the appropriateness of weight-adjusted dosing in children receiving atezolizumab. The observed Cmin was above the target therapeutic exposure level in all patients, and serum concentrations of atezolizumab were similar with the fixed dose in young adults or a weightadjusted dose in children. Exposure in infants younger than 2 years was about one-third of that seen in patients aged 2–17 years. When data were analysed by bodyweight in young adults the fixed-dose regimen led to higher exposures in lighter patients and lower exposures in heavier patients as expected. Although the findings in the paediatric population are evidence of appropriate dosing that is sufficient to support subsequent drug development in children older than 2 years, further research is needed to confirm the observed lower exposure in infants younger than 2 years than in older children and adults. Our study has several limitations. The protocol-defined criteria for study completion were met at the time of study closure; however, some of the tumour type cohorts did not meet the prespecified sample size of ten owing to a high number of non-responders. The numbers per cohort also reflect the rarity of the tumour types and the associated recruitment challenges, which might restrict the applicability of the study findings in patients with these particular tumour types. Interpretation and correlation of response data with baseline tumour PD-L1 expression was restricted by the low number of patient responses. However, high PD-L1 expression (defined as >5% of PD-L1-positive cells) was observed in tumours from all patients who showed some level of response, while other patients with high PD-L1expressing tumours did not respond to atezolizumab. This finding is consistent with previous studies of adults with NSCLC, urothelial bladder carcinoma, and triplenegative breast cancer treated with atezolizumab.21–23 In addition, the finding that only 17% of patients enrolled in the study with PD-L1 expression data exhibited high PD-L1 tumour expression is similar to results of a

previous study,1 which showed that PD-L1 was expressed in 39 (9%) of 451 evaluable paediatric tumours. A further study24 of 53 tumours with multiple tumour types cited no PD-L1 tumour expression in a range of untreated paediatric tumours; low expression was only seen in one case of rhabdomyo­ sarcoma. Notably, the numbers of children included in these previous studies were small. Our study was designed to include patients with potential PD-L1-positive tumours based on previous findings.1,24 Patients with tumour types with known or expected PD-L1 pathway involvement based on the literature or previous findings at the start of the study were included. To date, there is no validated assay to assess PD-L1 expression in paediatric tumours, and no common antibody or scoring method was used in these previous studies.1,24 Our study used the antibody and scoring methodology validated for PD-L1 detection in adult urothelial carcinoma and NSCLC from Ventana (SP142).21 At the time this study was designed, impeding the immunosuppressive activity of tumour-associated macrophages with an anti-PD-L1 antibody was of thera­ peutic interest for paediatric cancers when compared with treatment with anti-PD-1 therapies. Although higher concentrations of PD-L1-expressing tumour-associated macrophages in paediatric tumours than in adult tumours6,8 might be expected to enhance the activity of PD-L1-targeting therapies in younger patients, preclinical studies25 have shown that the presence of tumour-asso­ ciated macrophages in an immune suppressive tumour environment limits the antitumour activity of checkpoint inhibitors. We did not analyse PD-L1 expression on tumour-associated macrophages, so it is unknown whether that could help to explain why some patients do not respond to atezolizumab. Our study suggests that there was little response to single-agent atezolizumab in children and young adults with relapsed or refractory solid tumours, thus the study did not meet the criteria to proceed to a pivotal study for confirmation of efficacy. All patients had late-stage, mostly multi-metastatic disease reflecting high disease burden. Patients had received multiple lines of previous therapy; however, the average duration of atezolizumab treatment was short. Only four patients with high PD-L1 expression had partial responses: one with malig­ nant rhabdoid tumour, one with T-cell non-Hodgkin lymphoma, and two with Hodgkin lymphoma. Consistent with this, translocations or amplifications of chromo­ some 9p24 are common in patients with Hodgkin lymphoma and can lead to elevated expression of PD-L1 or PD-L2.26 9p24 variations are less common in nonHodgkin lymphoma.26 Notably, after the data cutoff date, the patient with T-cell non-Hodgkin lymphoma went on to achieve a complete response; this finding has not been reported previously in this disease. Factors including tumour immune cell infiltration, clonal neoantigens, tumour mutational burden, and immune checkpoint expression levels on tumour cells

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

9

Articles

To request clinical study data see www. clinicalstudydatarequest.com

and immune cells can influence the response to immune checkpoint inhibitors.27 These factors might differ between adult and paediatric populations.28 Some encouraging responses to immune checkpoint inhibitors were seen in children with increased tumour mutational burden either due to therapy or germline mutations in the replication and repair genes;29–31 however, paediatric cancers tend to have few overlapping characteristics with adult cancers.30 These differences probably lead to different response patterns in paediatric populations. Similarly, studies of other immune checkpoint inhibitors in children have shown a low level of response. A phase 1 study32 of ipilimumab in 33 children with refractory solid tumours reported no responses, but stable disease in 18% of patients. Preliminary findings of single-agent nivolumab in osteosarcoma or Ewing’s sarcoma showed no antitumour activity.33 A study34 of pembrolizumab in children with brain tumours also showed no overall survival benefit. However, pembrolizumab has shown benefit in children with refractory Hodgkin lymphoma and a few rare tumours, such as adrenocortical carci­noma, mesothelioma, malignant ganglioglioma, epithelioid sarcoma, lymphoepithelial carcinoma, and malignant rhabdoid tumour.35 In summary, atezolizumab was well tolerated and exposure of atezolizumab was broadly comparable across populations. Overall, response to atezolizumab was restricted, which is consistent with other reports using PD-1-targeting antibodies in paediatric populations.32–35 To further understand other predictive factors for response and non-response to atezolizumab, we plan to analyse the exploratory genomic and transcriptomic data collected during this trial as a future biomarker-focused study. Our findings provide important information on single-agent atezolizumab activity in children and young adults with relapsed or refractory solid tumours. Coupled with future exploratory analyses, these data might help to define future development strategies for immune checkpoint inhibitors either by strengthening the research focus to specific disease subpopulations that exhibit greater benefit from immunotherapies (eg, based on neoantigen load) or by providing the means to identify therapeutic combination partners that augment T-cell infiltration and proliferation in immune cold tumour micro­environments. In this context, further clinical data regarding the correlation between the presence of tumour-associated macrophages and the efficacy of checkpoint inhibitors in children would be beneficial to evaluate combinations with therapies that deplete tumour-associated macro­ phages.26 Other approaches under investigation include immune checkpoint inhibitors combined with conven­ tional chemotherapies, targeted therapies, and other immunotherapies. Contributors BG, FB, HC, and TT designed the study. BG, FB, MC, KEH, CMZ, LVM, TT, JM, and FD recruited patients and collected data. BG, KEH, LVM, TT, FD, CSS, GR, MFK, and ML analysed the data. BG, MC, NC, GR,

10

HC, KEH, CMZ, LVM, TT, FD, CSS, MFK, and ML interpreted the data. BG and TT did the referencing. BG, KEH, TT, CSS, MFK, and ML created the tables and figures. All authors critically reviewed the manuscript and approved it for submission. Declaration of interests BG reports travel and accommodation expenses for attendance at a Roche-sponsored advisory board for atezolizumab. MC reports personal fees from Bayer, Lilly, Roche, Servier, and Tesaro outside the submitted work. CMZ reports grants from Pfizer, Jazz Pharma, and Celgene, and personal fees from Novartis outside the submitted work. LVM reports personal fees from Tesaro outside the submitted work. GR, MF-K, FD, and HC are employed by Roche. CSS, KEH, and HL are employed by Genentech. TT reports consultancy fees from Roche/Genentech for advisory board attendance, and previous protocol management activities from POETIC Data and Coordinating Center outside of the submitted work. JM, FB, and NC declare no competing interests. Data sharing Qualified researchers can request access to individual patient-level data through the clinical study data request platform. Further details on Roche’s criteria for eligible studies are available at https:// clinicalstudydatarequest.com/Study-Sponsors/Study-Sponsors-Roche. aspx. For further details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see https://www.roche.com/research_and_development/ who_we_are_how_we_work/clinical_trials/our_commitment_to_data_ sharing.htm. Acknowledgments We thank the patients who participated in this study, their families, medical teams, and all participating study sites. Investigators included: Kevin Bielamowicz (Arkansas Children’s Hospital, Little Rock, AR, USA), Gianni Bisogno (Azienda Ospedaliero-Universitaria di Padova, Padova Italy), Valerie Brown (Penn State Hershey Children’s Hospital, Hershey, PA, USA), Quentin Campbell-Hewson (Royal Victoria Infirmary, Newcastle-upon-Tyne, UK), Steven DuBois (Dana Farber Cancer Institute, Boston, MA, USA), Martin Elliott (Leeds General Infirmary, Leeds, UK), Franca Fagioli (Azienda Ospidaliera Universitaria Città della Salute e della Scienza di Torino, Turin, Italy), Didier Frappaz (Centrè Léon Bérard, Lyon, France), Soledad Gallego (Hospital Universitario Vall d’Hebron, Barcelona, Spain), Alberto Garaventa (Istituto Pediatrico di Ricovero e Cura a Carattere Scientifico Instituto Giannina Gaslini, Genoa, Italy), Nicolas Gerber (University Children’s Hospital Zurich, Zurich, Switzerland), Cynthia Herzog (MD Anderson Cancer Center, Houston, TX, USA), Thomas Klingebiel (Universitätsklinikum Frankfurt, Frankfurt, Germany), Norman Lacayo (Stanford University School of Medicine, Stanford, CA, USA), Anne-Marie Langevin (University of Texas Health Sciences Center at San Antonio/University Hospital System, San Antonio, TX, USA), Franco Locatelli (Ospedale Pediatrico Bambino Gesù, Rome, Italy), Jaume Mora (Hospital Sant Joan de Déu, Madrid, Spain), Lucas Moreno (Hosptal Infantil Universitario Niño Jesús, Madrid, Spain), Francis Mussai (Birmingham Women’s and Children’s NHS Fundation Trust, Birmingham, UK), Karsten Nysom (Rigshospitalet, Copenhagen, Denmark), Helen Rees (Bristol Royal Hospital for Children, Bristol, UK), Ash Shifra (Schneider Children’s Medical Center of Israel, Petach Tikva, Israel), and Isaac Yaniv (Schneider Children’s Medical Center of Israel, Petach Tikva, Israel). Third-party medical writing support, under the direction of the authors, was provided by Angela Rogers of Gardiner-Caldwell Communications (Macclesfield, UK) and was funded by F Hoffmann La-Roche. References 1 Majzner R, Simon JS, Grosso JF, et al. Assessment of programmed death-ligand 1 expression and tumor-associated immune cells in pediatric cancer tissues. Cancer 2017; 123: 3807–15. 2 von Pawel J, Bordoni R, Satouchi M, et al. Long-term survival in patients with advanced non-small-cell lung cancer treated with atezolizumab versus docetaxel: Results from the randomised phase III OAK study. Eur J Cancer 2019; 107: 124–32. 3 Socinski MA, Jotte RM, Cappuzzo F, et al; IMpower150 Study Group. Atezolizumab for first-line treatment of metastatic nonsquamous NSCLC. N Engl J Med 2018; 378: 2288–301.

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

Articles

4

5 6 7 8

9 10 11 12 13 14 15 16 17 18 19

20

US FDA. Tecentriq (atezolizumab) highlights of prescribing information, revised June 2018. https://www.accessdata.fda.gov/ drugsatfda_docs/label/2018/761034s010lbl.pdf (accessed April 26, 2019). EMA. Tecentriq EPAR product information. https://www.ema. europa.eu/en/medicines/human/EPAR/tecentriq (accessed April 26, 2019). Chen DS, Irving BA, Hodi FS. Molecular pathways: next-generation immunotherapy–inhibiting programmed death-ligand 1 and programmed death-1. Clin Cancer Res 2012; 18: 6580−87. Berghoff AS, Ricken G, Widhalm G, et al. PD1 (CD279) and PD-L1 (CD274, B7H1) expression in primary central nervous system lymphomas (PCNSL). Clin Neuropath 2013; 33: 42−49. Vakkila J, Jaffe R, Michelow M, et al. Pediatric cancers are infiltrated predominantly by macrophages and contain a paucity of dendritic cells: a major nosologic difference with adult tumors. Clin Cancer Res 2006; 20: 49−54. Jaiswal S, Chao MP, Majeti R, Weissman IL. Macrophages as mediators of tumor immunosurveillance. Trends Immunol 2010; 31: 212–19. Ruhrberg C, De Palma M. A double agent in cancer: deciphering macrophage roles in human tumors. Nat Med 2010; 16: 861−62. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol 2007; 25: 579-86. Brodeur GM, Pritchard J, Berthold F, et al. Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment. J Clin Oncol 1993; 11: 1466−77. Wen PY, Macdonald DR, Reardon DA, et al. Updated response assessment criteria for high-grade gliomas: response assessment in neuro-oncology working group. J Clin Oncol 2012; 28: 1963−72. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009; 45: 228−47. Stroh M, Winter H, Marchand M, et al. Clinical pharmacokinetics and pharmacodynamics of atezolizumab in metastatic urothelial carcinoma. Clin Pharmacol Ther 2017; 102: 305–12. Dirks NL, Meibohm B. Population pharmacokinetics of therapeutic monoclonal antibodies. Clin Pharmacokinet 2010; 49: 633−59. Bai S, Jorga K, Xin Y, et al. A guide to rational dosing of monoclonal antibodies. Clin Pharmacokinet 2012; 51: 119−35. Xu Z, Davis HM, Zhou H. Rational development and utilization of antibody-based therapeutic proteins in pediatrics. Pharmacol Ther 2013; 137: 225−47. Ijaz A, Khan AY, Malik SU, et al. Significant risk of graft-versus-host disease with exposure to checkpoint inhibitors before and after allogeneic transplantation. Biol Blood Marrow Transplant 2019; 25: 94-99. Puzanov I, Diab A, Abdallah K, et al. Managing toxicities associated with immune checkpoint inhibitors: consensus recommendations from the Society for Immunotherapy of Cancer (SITC) Toxicity Management Working Group. J Immunother Cancer 2017; 5: 95.

21 Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet 2017; 389: 255–65. 22 Rosenberg JE, Hoffman-Censits J, Powles T, et al. Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, multicentre, phase 2 trial. Lancet 2016; 387: 1909–20. 23 Schmid P, Adams S, Rugo HS, et al. Atezolizumab and nabpaclitaxel in advanced triple-negative breast cancer. N Engl J Med 2018; 379: 2108–21. 24 Aoki T, Hino M, Koh K, et al. Low frequency of programmed death ligand 1 expression in pediatric cancers. Pediatr Blood Cancer 2016; 63: 1461–64. 25 Cassetta L, Kitamura T. Targeting tumor-associated macrophages as a potential strategy to enhance the response to immune checkpoint inhibitors. Front Cell Dev Biol 2018; 6: 38. 26 Merryman RW, Armand P, Wright KT, Rodig SJ. Checkpoint blockade in Hodgkin and non-Hodgkin lymphoma. Blood Adv 2017; 1: 2643–54. 27 Topalian SL, Drake CG, Pardoll DM. Targeting the PD-1/B7-H1 (PD-L1) pathway to activate anti-tumor immunity. Curr Opin Immunol 2012; 24: 207–12. 28 Park JA, Cheung NK-V. Limitations and opportunities for immune checkpoint inhibitors in pediatric malignancies. Cancer Treat Rev 2017; 58: 22–33. 29 Campbell BB, Light N, Fabrizio D, et al. Comprehensive analysis of hypermutation in human cancer. Cell 2017; 171: 1029-41. 30 Ma X, Liu Y, Liu Y, et al. Pan-cancer genome and transcriptome analyses of 1,699 pediatric leukemias and solid tumors. Nature 2018; 555: 371–76. 31 Gröbner SN, Worst BC, Weischenfeldt J, et al. The landscape of genomic alterations across childhood cancers. Nature 2018; 555: 321–27. 32 Merchant MS, Wright M, Baird K, et al. Phase I clinical trial of ipilimumab in pediatric patients with advanced solid tumors. Clin Cancer Res 2016; 22: 1364–70. 33 Davis KL, Fox E, Reid JM, et al. ADVL1412: Initial results of a phase I/II study of nivolumab and ipilimumab in pediatric patients with relapsed/refractory solid tumors—A COG study. Proc Am Soc Clin Oncol 2017; 35 (suppl 15): 10526. 34 Blumenthal DT, Yalon M, Vainer GW, et al. Pembrolizumab: first experience with recurrent primary central nervous system (CNS) tumors. J Neurooncol 2016; 129: 453–60. 35 Geoerger B, Kang HJ, Yalon-Oren M, et al. Pembrolizumab in paediatric patients with advanced melanoma or a PD-L1-positive advanced, relapsed, or refractory solid tumour or lymphoma (KEYNOTE-051): interim analysis of an open-label, single-arm, phase 1–2 trial. Lancet Oncol (in press).

www.thelancet.com/oncology Published online November 25, 2019 https://doi.org/10.1016/S1470-2045(19)30693-X

11