Articles Cediranib in patients with alveolar soft-part sarcoma (CASPS): a double-blind, placebo-controlled, randomised, phase 2 trial Ian Judson, James P Morden*, Lucy Kilburn, Michael Leahy, Charlotte Benson, Vivek Bhadri, Quentin Campbell-Hewson, Ricardo Cubedo, Adam Dangoor, Lisa Fox, Ivo Hennig, Katy Jarman, Warren Joubert, Sarah Kernaghan, Antonio López Pousa, Catriona McNeil, Beatrice Seddon, Claire Snowdon, Martin Tattersall, Christy Toms, Javier Martinez Trufero, Judith M Bliss Summary Background Alveolar soft-part sarcoma (ASPS) is a rare soft-tissue sarcoma that is unresponsive to chemotherapy. Lancet Oncol 2019; 20: 1023–34 Cediranib, a tyrosine-kinase inhibitor, has shown substantial activity in ASPS in non-randomised studies. Published Online The Cediranib in Alveolar Soft Part Sarcoma (CASPS) study was designed to discriminate the effect of cediranib from May 31, 2019 http://dx.doi.org/10.1016/ the intrinsically indolent nature of ASPS. S1470-2045(19)30215-3 See Comment page 901 Methods In this double-blind, placebo-controlled, randomised, phase 2 trial, we recruited participants from The Institute of Cancer 12 hospitals in the UK (n=7), Spain (n=3), and Australia (n=2). Patients were eligible if they were aged 16 years or Research and The Royal older; metastatic ASPS that had progressed in the previous 6 months; had an ECOG performance status of 0–1; life Marsden NHS Foundation expectancy of more than 12 weeks; and adequate bone marrow, hepatic, and renal function. Participants had to have Trust, London, UK no anti-cancer treatment within 4 weeks before trial entry, with exception of palliative radiotherapy. Participants were (Prof I Judson MD); Clinical Trials and Statistics Unit, The randomly assigned (2:1), with allocation by use of computer-generated random permuted blocks of six, to either Institute of Cancer Research, cediranib (30 mg orally, once daily) or matching placebo tablets for 24 weeks. Treatment was supplied in number- London, UK (J P Morden MSc, coded bottles, masking participants and clinicians to assignment. Participants were unblinded at week 24 or sooner if L Kilburn MSc, L Fox PhD, K Jarman BSc, S Kernaghan BSc, they had progression defined by Response Evaluation Criteria in Solid Tumors (version 1.1); those on placebo crossed C Snowdon MSc, C Toms DPhil, over to cediranib and all participants continued on treatment until progression or death. The primary endpoint was Prof J M Bliss MSc); The Christie percentage change in sum of target marker lesion diameters between baseline and week 24 or progression if sooner, NHS Foundation Trust, assessed in the evaluable population (all randomly assigned participants who had a scan at week 24 [or sooner if they Manchester, UK (M Leahy FRCP); The Royal Marsden NHS progressed] with target marker lesions measured). Safety was assessed in all participants who received at least one Foundation Trust, London, UK dose of study drug. This study is registered with ClinicalTrials.gov, number NCT01337401; the European Clinical (C Benson MBChB); Trials database, number EudraCT2010-021163-33; and the ISRCTN registry, number ISRCTN63733470 recruitment is Chris O’Brien Lifehouse, complete and follow-up is ongoing. Sydney, NSW, Australia (V Bhadri MBBS, C McNeil PhD, Prof M Tattersall MD); Findings Between July 15, 2011, and July 29, 2016, of 48 participants recruited, all were randomly assigned to cediranib Newcastle upon Tyne Hospitals (n=32) or placebo (n=16). 23 (48%) were female and the median age was 31 years (IQR 27–45). Median follow-up was NHS Foundation Trust, 34·3 months (IQR 23·7–55·6) at the time of data cutoff for these analyses (April 11, 2018). Four participants in the Newcastle upon Tyne, UK (Q Campbell-Hewson MBChB); cediranib group were not evaluable for the primary endpoint (one did not start treatment, and three did not have their Hospital Puerta de Hierro, scan at 24 weeks). Median percentage change in sum of target marker lesion diameters for the evaluable population Madrid, Spain (R Cubedo PhD); was −8·3% (IQR −26·5 to 5·9) with cediranib versus 13·4% (IQR 1·1 to 21·3) with placebo (one-sided p=0·0010). The University Hospitals Bristol most common grade 3 adverse events on (blinded) cediranib were hypertension (six [19%] of 31) and diarrhoea NHS Foundation Trust, Bristol, UK (A Dangoor MD); (two [6%]). 15 serious adverse reactions in 12 patients were reported; 12 of these reactions occurred on open-label Nottingham University cediranib, and the most common symptoms were dehydration (n=2), vomiting (n=2), and proteinuria (n=2). Hospitals NHS Trust, One probable treatment-related death (intracranial haemorrhage) occurred 41 days after starting open-label cediranib Nottingham, UK (I Hennig MD); in a patient who was assigned to placebo in the masked phase. Princess Alexandra Hospital, Brisbane, QLD, Australia (W Joubert FRACP); Hospital de Interpretation Given the high incidence of metastatic disease and poor long-term prognosis of ASPS, together with la Santa Creu i Sant Pau, the lack of efficacy of conventional chemotherapy, our finding of significant clinical activity with cediranib in this Barcelona, Spain disease is an important step towards the goal of long-term disease control for these young patients. Future clinical (A López Pousa MD); University College London trials in ASPS are also likely to involve immune checkpoint inhibitors. Hospitals NHS Foundation Trust, London, UK Funding Cancer Research UK and AstraZeneca. (B Seddon PhD); and Hospital Miguel Servet, Zaragoza, Spain (J Martinez Trufero MD) Copyright: © 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. *Mr Morden died in September, 2017 Introduction affects young people, with a median age at presentation of Alveolar soft-part sarcoma (ASPS) is rare, accounting for 25 years and most patients younger than 30 years at less than 0·5% of all soft-tissue sarcomas. It predominantly diagnosis.1 ASPS commonly involves the lower limb, with www.thelancet.com/oncology Vol 20 July 2019 1023 Articles Correspondence to: Prof Ian Judson, ICR – Clinical Research in context Trials and Statistics Unit, Division of Clinical Studies, The Institute Evidence before this study Added value of this study of Cancer Research, Sutton, Before undertaking this study, the available data concerning ASPS has a high metastatic potential, but usually slow disease Surrey, SM2 5NG, UK the activity of the experimental drug cediranib (previously progression, and sometimes spontaneous disease arrest and Ian.Judson@icr.ac.uk AZD2171) in the treatment of alveolar soft part sarcoma even, rarely, regression can occur. These characteristics make (ASPS) consisted of an index case of extended response in progression-free survival an unreliable endpoint for this disease. ASPS in a phase 2 trial, a phase 2 study of cediranib in the By undertaking a placebo-controlled randomised trial with treatment of gastrointestinal stromal tumours and soft tumour size as the primary endpoint, we ensured that the activity tissue sarcomas including ASPS, and a phase 2 single-arm of cediranib could be measured reliably. To our knowledge, this study, which was ongoing at the time the CASPS trial was study is the only randomised trial to be reported in ASPS so far. being developed and has since shown activity in ASPS. Implications of all the available evidence Other studies involving ASPS, ongoing and completed, Although the precise molecular targets of cediranib in ASPS are were identified using the ClinicalTrials.gov website, searching not known, this study has confirmed the value of this drug in the for “alveolar soft part sarcoma”; PubMed, searching for treatment of advanced ASPS. The relative importance of publications in English between Jan 1, 2000, and angiogenesis inhibition and immunomodulation are the subject Dec 31, 2018, using the search terms “tyrosine kinase of active investigation and translational research from this study inhibitor”, “anti-angiogenic agent”, and “alveolar soft part will be published in due course. Confirmation of the activity of sarcoma”; via presentations at international meetings, cediranib in a randomised, placebo-controlled trial will provide a and personal communications. Other drugs with reported sound basis for future research with this and other drugs for ASPS. activity in ASPS include sunitinib, pazopanib, and anlotinib. a slight predominance in women and a high incidence of (HIF1-α) via inhibition of the prolyl hydroxylase metastatic disease at diagnosis.1 Although metastases are responsible for its degradation,7 raising the possibility intrinsically indolent, the long-term outlook is poor.2 that the lactate transporter MCT1 could be a therapeutic Lieberman and colleagues2 report that only 15% of patients target for inhibition of tumour angiogenesis.8 with no metastases at diagnosis remained metastasis free Cediranib is a receptor tyrosine-kinase inhibitor, the after 20 years of follow-up, with a median metastasis-free targets of which include the VEGF receptors VEGFR1, period of 6 years and median survival after development VEGFR2, and VEGFR3; KIT; and platelet-derived growth of metastases of 2 years. If patients presented with factor receptors. After observation of a prolonged partial metastases, median survival was 3 years, compared with remission with cediranib in a patient with locally 11 years for patients who were metastasis free at diagnosis, advanced and metastatic ASPS treated in a phase 2, and survival tended to worsen with increasing age.2 hypertension management study (NCT00264004),9 an Unusually for a soft-tissue sarcoma, in addition to lung opportunity arose to treat a further six patients with metastases, ASPS also metastasises to brain and bone.3 ASPS who were treated in a cediranib pharmacodynamic Histologically, the disease is characterised by uniform study of patients with soft-tissue sarcoma, most of whom polygonal cells arranged in a pseudoalveolar pattern had gastrointestinal stromal tumours (D8480C00046).10 separated by vascular septae, and molecular studies4 have Strong evidence of the clinical activity of cediranib shown a characteristic non-reciprocal translocation, t(X;17) against ASPS, in terms of durable partial remissions and (p11·2;q25), resulting in the ASPSCR1–TFE3 fusion gene disease stabilisation, led to further studies including a that replaces the N-terminal portion of TFE3 in a manner single-arm, phase 2 trial of cediranib in ASPS by the US consistent with transcriptional deregulation.4 National Cancer Institute.11 ASPS cells have periodic acid-Schiff (PAS)-positive Other tyrosine-kinase inhibitors have also been shown precrystalline granules that contain monocarboxylate to have activity in ASPS. A direct antitumour effect has transporter 1 (MCT1) and its chaperone basigin (CD147).5 been shown with sunitinib mediated by platelet-derived In a genetically engineered mouse study6 that used the growth factor receptor β (PDGFRB), VEGFR2, and RET, ASPSCR1–TFE3 gene to drive oncogenesis, the mice with five partial responses in nine patients and median developed tumours in the brain and orbit—ie, the cranial progression-free survival on treatment of 17 months vault—a region known to have the highest lactate (range 2–33).12 In adults, a retrospective study reported concentrations in the mouse. Metabolic studies6 showed one complete response and seven partial responses in that ASPS cells in this model used lactate as an energy 30 patients treated with pazopanib, with a median source. Lactate is imported via MCT1 and is converted progression-free survival of 13·6 months (range 1·6–32·2) directly to pyruvate for entry into the citric acid cycle. In and only little activity with trabectedin.13 In 40 patients addition to supplying energy, lactate acts to stimulate cell with ASPS and a rearrangement of the TFE3 gene treated proliferation and angiogenesis because of the excess with the serine/threonine-protein kinase Haspin pyruvate, which upregulates hypoxia-inducible factor 1-α homolog ALK1 and hepatocyte growth factor receptor 1024 www.thelancet.com/oncology Vol 20 July 2019 Articles (MET) inhibitor crizotinib, one patient had a partial restricted eligibility criteria. Patients with brain response and 35 had stable disease as their best response, metastases were eligible if their disease was controlled with 1-year progression-free survival of 37·5% (95% CI with a stable dose of corticosteroid or a non-enzyme- 22·9–52·1).14 Anlotinib has also been reported to have inducing anticon­ vulsant. Exclusion criteria included activity in a prospective basket study15 in which six of history of gastro­ intestinal disorder likely to impair 13 patients with ASPS had partial responses and the absorption of cediranib; poorly controlled hypertension; median progression-free survival was 21 months. any severe or uncon­ trolled comorbidity—eg, active Pazopanib is the only multi-targeted tyrosine-kinase infection; prolonged QT interval—ie, QTc ≥480 msec inhibitor approved for second-line or further-line treat­ (using Bazetts correction) or family history of long QT ment of all soft-tissue sarcomas. Preliminary reports of syndrome; substantial recent haemorrhage (>30 mL activity with immunomodulatory drugs have been bleeding or episode in previous 3 months); major presented in the past 2 years.15,16 thoracic or abdominal surgery in previous 14 days; recent The Cediranib in Alveolar Soft Part Sarcoma (CASPS) history of thrombosis; pregnant or breastfeeding women; trial aimed to assess the efficacy of cediranib in the unwillingness to use adequate birth control measures; treatment of ASPS. The double-blind, placebo-controlled anticancer treatment in the previous 4 weeks, with design was chosen because of the unusual biology of this the exception of palliative radiotherapy; known hyper­ cancer, which, although having strong metastatic sensitivity to any excipient of cediranib; history of other potential, is characterised by indolent metastatic tumour malignancy except cancer in situ unless individual had growth and periods of spontaneous stabilisation, making been disease free for more than 2 years and with tissue single-group uncontrolled studies difficult to interpret.3,12 diagnosis of ASPS from target lesion; and any other Ethical challenges exist in placebo groups, which we concomitant anticancer therapy except steroids. A full list mitigated by having a 2:1 randomisation favouring of exclusion criteria is in the appendix (pp 27–28). See Online for appendix active treatment, restricting the no-treatment period to Previous treatment with cediranib was added as an 24 weeks, and allowing crossover to active treatment on exclusion criterion and approved as a protocol disease progression or, if no progression, after week 24. amendment on the advice of the joint Independent At entry, patients were required to have progressed in the Data Monitoring and Steering Committee (IDMSC) on previous 6 months, hence 6 months was chosen as the Sept 24, 2014. period for comparison between cediranib and placebo. Patients provided written, informed consent before A preliminary report of CASPS was presented at the 2017 enrolment. The study protocol is in the appendix Annual General Meeting of the American Society of (pp 9–75). The study was approved by the South West Clinical Oncology.17 London Research Ethics Committee 4 (REC reference 10/H0806/118) in the UK; the Clinical Research Ethics Methods Committee of Hospital de la Santa Creu i Sant Pau, Study design and participants Barcelona (12/070 [R]), in Spain; and by the Metro In this multicentre, double-blind, placebo-controlled, South Health Service District Human Research Ethics randomised, phase 2 trial, participants were recruited Committee, QLD (HREC/12/QPAH/10), and the Sydney from 12 hospitals in the UK (n=7), Spain (n=3), Australia Local Health District Human Research Ethics Committee (n=2). Eligible patients were women or men aged 16 years (HREC/12/RPAH/26), in Australia. The Clinical Trials or older, with a histologically confirmed diagnosis of and Statistics Unit at The Institute of Cancer Research, ASPS. A tumour block was required for central review London, UK (ICR-CTSU), had overall responsibility for and confirmation of the presence of t(X;17)(p11·2:q25) trial coordination with two international trials groups translocation. Patients were required to have measurable (Grupo Español de Investigación en Sarcomas [GEIS], metastatic disease that had progressed according to the Madrid, Spain, and Australasian Sarcoma Study Group Response Evaluation Criteria in Solid Tumors (RECIST) [ASSG], Melbourne, Australia) having responsibility for version 1.1 in the previous 6 months; an ECOG perfor­ regulatory and ethics submissions, monitoring, and mance status of 0–1; life expectancy of more than safety reporting within their respective countries. Safety 12 weeks; and adequate bone marrow, hepatic, and renal and efficacy data were reviewed regularly by the IDMSC. function (absolute neutrophil count >1·5 × 10⁹ per L; A Trial Management Group was responsible for the day- platelet count >100×10⁹ per L; serum bilirubin to-day running of the trial. The ICR-CTSU undertook all <1·5 × upper limit of normal [ULN], unless proven central statistical monitoring, interim, and final analyses. Gilbert’s syndrome; alanine transaminase or aspartate transaminase <2·5 × ULN or <5 × ULN if liver metastases Randomisation and masking present; serum creatinine <1·5 × ULN or creatinine Participants were randomly allocated (2:1) to cediranib or clearance >50 mL per min). The restrictive criterion of placebo by computer-generated random permuted block progression in the previous 6 months was based on the method (block size of six) derived by ICR-CTSU, and the known indolent nature of the disease. CASPS thus random­ isation sequence centrally was generated differs from other ASPS studies that do not have such centrally at the ICR-CTSU. Because of the small trial www.thelancet.com/oncology Vol 20 July 2019 1025 Articles size, we did not use any stratification factors. Both the Outcomes participants and clinicians were masked to treatment The primary endpoint was percentage change in the sum allocation until week 24 or until disease progression if of the longest diameters (or shortest if nodal disease) of this occurred sooner.Treatment was supplied in number- target marker lesions, measured at 24 weeks (or at coded bottles, masking participants and clinicians to progression if sooner) from the date of treatment assignment. assignment. Protocol-defined secondary endpoints were the proportion of participants with an objective response Procedures at week 24, defined as the proportion of participants Depending on treatment allocation, participants having either a partial or complete response at the end of received cediranib or matching placebo 30 mg orally blinded treatment; best response up to week 24, defined as once daily, for the first 24 weeks of the study. At best response at any point during blinded treatment; best 24 weeks, or sooner if the patient had confirmed disease percentage reduction of target marker lesion size during progression according to RECIST, participants were blinded treatment; progression-free survival defined as unmasked. Participants allocated to placebo crossed time from random assignment to disease progression over to open-label cediranib and continued on treatment (defined by RECIST version 1.1) or death from any cause; until disease progression or death. Participants the proportion of patients alive and progression free at allocated to cediranib who had not progressed by 12 months; overall survival, defined as time from random 24 weeks continued on cediranib until progression or assignment to death (any cause); and the safety and death. Participants could withdraw from the trial at any tolerability profile of cediranib in all participants. For point and for any reason. progression free survival, participants who were alive and Clinical assessments, including physical examination, progression-free were censored at the date of last known symptom review, and routine blood and urine follow-up. Participants with non-RECIST confirmed investigations, took place at 2 and 4 weeks, then once progression (eg, radio­logically confirmed but lesions not every 4 weeks until week 24, every 8 weeks up to 48 weeks, measured according to RECIST, or participants with then every 12 weeks until progression or treatment clinical evidence of progression only) were censored at the discontin­uation. Assessments for participants crossing date of their progression. over from placebo to cediranib were recommenced Exploratory endpoints included radiological responses similarly to in the first 24 weeks for the original cediranib using Choi criteria, identification of predictive angio­ group. Tumour assessments (CT or MRI if indicated) genesis markers of response, to describe changes in were done at baseline, every 8 weeks until week 48, and angiogenesis markers and expression of angiogenesis then every 12 weeks until disease progression or death. regulatory genes in peripheral blood and optional tumour Blood pressure was monitored at least weekly for the first biopsies, and to explore changes in circulating endothelial 4 weeks, then monthly up to 24 weeks and thereafter as cells and other rare cells events, including potential specified in the protocol—ie, every 8 weeks up to sarcoma circulating cells. These exploratory endpoints 48 weeks and every 12 weeks thereafter. Masked radiology will be reported elsewhere once these data are available. review was not planned as part of this study, although translational imaging studies are planned. Statistical analysis Toxicity was assessed using National Cancer Institute Our study size was informed by previous studies of Common Terminology Criteria for Adverse Events participants with metastatic ASPS. In the previously (CTCAE) version 4 on the same schedule as the clinical mentioned cediranib pharmacodynamic phase 2 study assessments. Coding was done by use of the Medical (D8480C00046),10 of 36 participants (ten of whom had Dictionary for Regulatory Activities (version 14). soft-tissue sarcoma and 26 had a gastrointestinal stromal Treatment-related toxicities of grade 3 or worse, or tumour), six participants with metastatic ASPS had a repeated episodes of grade 2 toxicities not responding to mean decrease in tumour size at 8 weeks of 25%, with a adequate supportive measures were managed initially coefficient of variation of 19%. The second scan at with dose interruptions of 2–5 days, with reintroduction 16 weeks showed that four of six participants had a more on resolution at the same dose. Longer interruptions than 30% decrease in tumour size—ie, a partial response. up to 14 days were permitted for chronic problems We assumed that a smaller effect and greater coefficient such as nausea, diarrhoea, and palmar-plantar erythro­ of variation than these results might be observed in a dysaesthesia syndrome if refractory to supportive larger trial; therefore, we determined that 36 participants treatment—eg, antiemetics or loperamide. If toxic effects would be required to detect a 20% decrease in the sum of continued, a dose reduction to 20 mg daily was per­ the diameters of target marker lesions at 24 weeks mitted. Treatment with cediranib was to be discontinued between placebo and cediranib with a coefficient of permanently for gastrointestinal perforation, wound variation of 25%, 80% power, and a one-sided significance dehiscence, severe haemorrhage, or severe uncontrolled level of 5%. We calculated this sample size for a two- hypertension. Abnormal thyroid function was treated sample t test of the log transformation of the sum of the with L-thyronine as appropriate. diameters of target marker lesions using the sampsi 1026 www.thelancet.com/oncology Vol 20 July 2019 Articles command in Stata. We did a formal interim analysis after The safety analysis population included all treated 18 participants (12 assigned to cediranib and six assigned participants (ie, all those who received at least one dose to placebo) had 24 weeks of follow-up data (or had disease of study drug or placebo), and we summarise the worst progression if sooner) to determine early activity of grades of adverse events during the blinded treatment cediranib. Recruitment to the study was not halted while phase. We did no formal comparisons of safety between the interim analysis was undertaken. treatment groups because of the small patient numbers. We used descriptive statistics to show the baseline We present here any adverse event that was reported in at demographic information of the participants who were least 10% of participants in either treatment group, or in enrolled. We compared the primary endpoint between one or more participants for grade 3 and worse events. cediranib and placebo groups using the Wilcoxon-Mann- We also report on the number of serious adverse Whitney test. We chose to present the data as medians reactions to cediranib. and IQRs and to use non-parametric tests a priori We also did an unplanned, exploratory analysis to because of the uncertainty around whether or not the explore the effect of previous tyrosine-kinase inhibitor data would be normally distributed. We used waterfall therapy on progression-free survival. plots to graphically present the primary endpoint result. Post hoc, we analysed duration of response (calculated The principal analysis population for the primary as time from first response, complete or partial response, endpoint was the evaluable population, defined as all to date of progression), and the proportion of participants randomly assigned participants (the intention-to-treat in each group who had a clinical benefit at 24 weeks for [ITT] population) who had a scan at week 24 (or a scan at completeness and to enable comparison with other progression if earlier) with target marker lesions studies. measured. We did a sensitivity analysis of the primary Analyses are based on a database snapshot taken on endpoint and of the best percentage change in the sum April 11, 2018. We did all analyses with Stata (version 13.1). of the longest diameters of target marker lesions at This study is registered with ClinicalTrials.gov, number 24 weeks based on this population, but excluding two NCT 0133740; on the ISRCTN registry, number ISRCTN participants who had received cediranib before entering 63733470; and on the European Clinical Trials database, the study (an exclusion criterion added after these number EudraCT 2010-021163-33. two participants were enrolled). We report binary endpoints (objective response, best Role of the funding sources response, and clinical benefit) as proportions and The funders of the study had no role in study design, compared them between treatment groups using Fisher’s data collection, data analysis, data interpretation, or exact test (one sided). The population analysed for best writing of the report. The corresponding author had full response also included participants who had a scan access to all data in the study and final responsibility for before 24 weeks, but were not evaluable for the primary the decision to submit for publication. endpoint. For survival-related endpoints, we used the ITT population; thus, we still analysed participants Results allocated to placebo who subsequently crossed over as Between July 15, 2011, and July 29, 2016, 48 patients belonging to the placebo group. We plotted Kaplan-Meier recruited from 12 hospitals in three countries (n=31 from curves for the two treatment groups and compared them seven hospitals in the UK, n=9 from three hospitals in with the log-rank test. We calculated hazard ratios (HRs) Spain, and n=8 from two hospitals in Australia) were and 90% CIs from Cox proportional hazards regression enrolled. More participants were recruited than were models for progression-free survival, with HRs of less originally planned under the recommendation of the than 1 favouring cediranib. HRs were not calculated for IDMSC to account for the four participants who were not overall survival because of the failure of the proportional evaluable for the primary endpoint and two participants hazards assumption that is required to use Cox who had received cediranib previously. 32 participants regression, as anticipated before the analysis. All analyses were randomly assigned to cediranib and 16 to placebo were unadjusted. Similar to the primary endpoint, we did (figure 1) and baseline characteristics were well balanced a sensitivity analysis for progression-free survival and between the treatment groups (table 1). Median age at overall survival, excluding two participants who had baseline was 31 years (IQR 27–45), with most participants received cediranib before entering the study. We planned in the 20–29 years age group. More than half of two additional sensitivity analyses for progression-free participants were diagnosed within 2·5 years of trial survival: one to include participants who had non- entry, the most common site of disease was the upper leg RECIST confirmed progression, and the other was a or groin and 11 (23%) of 48 participants had known brain landmark analysis, also including participants who had metastases at trial entry. 20 (42%) participants had non-RECIST confirmed progressio,n but censoring previously received tyrosine-kinase inhibitor treatment, participants 26 weeks after random assignment to including two (4%) with previous cediranib. provide insight into progression-free survival in the A planned interim analysis was done of a snapshot of absence of crossover. data taken on Feb 4, 2014, and did not meet the stopping www.thelancet.com/oncology Vol 20 July 2019 1027 Articles 28 participants were disease progression (n=21), adverse 48 patients recruited events (n=2), patient choice (n=4), and death (n=1). Median time on open-label cediranib was 41·6 weeks 48 enrolled (IQR 21·0–67·0) for the participants from the cediranib group and 40·3 weeks (12·2–108·6) for those from the placebo group. In the evaluable population (44 [92%] of 48 enrolled 32 randomly assigned to cediranib 16 randomly assigned to placebo patients), we found a significant difference in the median percentage change in the sum of the diameters of target 1 did not start treatment* marker lesions at 24 weeks of −8·3% (IQR –26·5 to 5·9) in the cediranib group compared with 13·4% (1·1 to 21·3) in the placebo group (one-sided p=0·0010; figure 2). 31 started treatment 16 started treatment Best median percentage change in the sum of the diameters of target marker lesions by 24 weeks was 3 did not undergo scan at –15·7% (IQR –26·3 to –2·4) with cediranib and 1·2% 24 weeks (–2·4 to 10·9) with placebo (one-sided p<0·0001; 1 due to early appendix p 2). discontinuation caused by toxicity Similar results were obtained in the sensitivity analysis 2 due to other reasons† in the cediranib-naive population (42 [88%] of 48 participants): median percentage change in the sum 28 evaluable for primary analysis 16 evaluable for primary analysis of the longest diameters (or shortest if nodal disease) of target marker lesions at 24 weeks was –4·4% (IQR –26·3 to 6·0) with cediranib versus 14·4% 1 had previous cediranib 1 had previous cediranib (1·1 to 22·6) with placebo (one-sided p=0·0019) and best percentage change was –15·0% (IQR –26·3 to –2·4) with cediranib versus 1·3% (–2·5 to 11·8) with placebo (one- 27 in cediranib naive population‡ 15 in cediranib naive population sided p=0·00010). The proportion of participants with an objective Figure 1: Trial profile *Found to have a tumour infiltrating their heart after random assignment. †One patient was subsequently found to response, based on best response during the blinded be ineligible because of unconfirmed progression in the 6 months before trial entry. ‡One patient was subsequently treatment phase, was 19%, with six of 31 participants on found to be ineligible because of unconfirmed progression in the 6 months before trial entry, but was included in cediranib having a partial response at any timepoint analyses. during the masked treatment phase. However, at week 24, three of these participants no longer met RECIST criteria for the primary endpoint (appendix p 57), thus criteria for partial response, and thus the proportion of the study continued until the primary analysis. Of participants with an objective response at week 24 was 32 participants randomly assigned to cediranib, one (3%) 11% (three of 28). No participants in the placebo group did not start treatment and three (6%) did not have their had a partial response during the blinded treatment scan at 24 weeks (one of whom discontinued due to phase (one-sided p=0·072, cediranib vs placebo). 14 (50%) toxicity). Median time on blinded treatment was of 28 participants in the cediranib group and seven (44%) 23·9 weeks (IQR 20·1–24·6) for participants assigned to of 16 in the placebo group had stable disease at 24 weeks; cediranib and 22·2 weeks (IQR 9·6–23·8) for those thus the proportion of participants with clinical benefit assigned to placebo. Of those who started treatment, at 24 weeks was 61% (17 of 28) for cediranib and 44% 14 (45%) in the cediranib group and four (25%) in the (seven of 16) for placebo (post-hoc analysis). Of the seven placebo group had at least one dose modification, dose participants in the placebo group who were stable at delay, or missed dose during blinded treatment because 24 weeks, three had received no treatment in the of adverse events. 11 (35%) of 31 who were given blinded 6 months before randomisation, one had surgery to cediranib did not continue to open-label treatment; the primary disease site, and another three were on eight (26%) because of disease progression, one (3%) another tyrosine-kinase inhibitor until a month before because of haematemesis, one (3%) because of randomisation. Of the 14 patients randomly assigned to palmar-plantar erythrodysesthesia and fatigue, and cediranib with stable disease or better at 24 weeks before one (3%) because of general poor tolerance and planned therapy, three had no previous treatment; eight had local surgery. All participants allocated to placebo went on to therapy (surgery or radiotherapy, or both) only to primary have at least one dose of open-label cediranib. Of 36 (75%) or metastatic sites, or both; two had local therapy and participants who started open-label cediranib, eight (five another tyrosine-kinase inhibitor; and one had another from the cediranib group and three from the placebo tyrosine-kinase inhibitor only. Median duration of group) were still on treatment at the data cutoff for this response in the six patients in the cediranib group who analysis. Reasons for discontinuation in the remaining had a partial response during the blinded treatment 1028 www.thelancet.com/oncology Vol 20 July 2019 Articles Cediranib group Placebo group Cediranib group Placebo group (n=32) (n=16) (n=32) (n=16) Sex (Continued from previous column) Male 17 (53%) 8 (50%) Previous treatment for ASPS Female 15 (47%) 8 (50%) Any localised treatment Age at baseline, years No 7 (22%) 5 (31%) Median 30·3 (26·8–44·7) 33·1 (29·3–43·5) Yes 25 (78%) 11 (69%) <20 0 1 (6%) Surgery 21 (66%) 11 (69%) 20–29 15 (47%) 5 (31%) To the primary disease site 19 (59%) 10 (63%) 30–39 8 (25%) 4 (25%) For metastatic disease 10 (31%) 4 (25%) 40–49 4 (13%) 5 (31%) Radiotherapy 21 (66%) 7 (44%) 50–59 4 (13%) 0 To the primary disease site 12 (38%) 7 (44%) 60–69 0 1 (6%) Any systemic treatment ≥70 1 (3%) 0 No 19 (59%) 7 (44%) ECOG performance status Yes 13 (41%) 9 (56%) 0 16 (50%) 9 (56%) Chemotherapy 5 (16%) 4 (25%) 1 16 (50%) 7 (44%) Tyrosine-kinase inhibitor 12 (38%) 8 (50%) Time since original ASPS diagnosis, years Crizotinib 5 (16%) 5 (31%) ≤2·5 18 (56%) 8 (50%) Sunitinib 1 (3%) 2 (13%) >2·5–5 4 (13%) 6 (38%) Axitinib 2 (6%) 0 >5–7·5 2 (6%) 0 Pazopanib 2 (6%) 0 >7·5–10 2 (6%) 0 Cediranib 1 (3%) 0 >10 6 (19%) 2 (13%) Cediranib plus dovitinib 0 1 (6%) Site of primary disease Sunitinib plus pazopanib 1 (3%) 0 Upper leg or groin 14 (44%) 7 (44%) MET inhibitor (ARQ197) 1 (3%) 0 Upper limb 6 (19%) 2 (13%) HDAC inhibitor (PXD101) 0 1 (6%) Lower leg or foot 3 (9%) 3 (19%) Data are n (%) or median (IQR). ASPS=alveolar soft part sarcoma. Trunk 4 (13%) 2 (13%) MET=hepatocyte growth factor receptor. HDAC=histone deacetylase. Buttock 3 (9%) 0 Table 1: Baseline demographic and clinical characteristics and previous Pelvis 1 (3%) 2 (13%) treatments received for ASPS prior to trial entry Cranial or facial 1 (3%) 0 Synchronous metastases at trial entry Yes 5 (16%) 0 cediranib group, 14 from the placebo group) had No 27 (84%) 16 100%) had RECIST-confirmed progression or death before Brain metastases at trial entry progression. We found no evidence of a significant Yes 8 (25%) 3 (19%) difference in progression-free survival between the No 24 (75%) 13 (81%) treatment groups (unadjusted HR 0·82, 90% CI 0·47–1·43; p=0·28; figure 3), although, this analysis (Table 1 continues in next column) was probably confounded by the crossover to cediranib. 12-month progression-free survival was 38·7% (95% CI phase was 16·0 months (IQR 15·7–26·0); all six patients 22·0–55·2) for cediranib and 34·4% (12·7–57·5) for have subsequently progressed (post hoc). Of the placebo. In a sensitivity analysis excluding those 32 participants assigned to the cediranib group, two (6%) participants with previous cediranib treatment, a who had stable disease up to week 24 subsequently similar result was achieved (unadjusted HR 0·76, 90% achieved a partial response (one by week 32, the other by CI 0·43–1·35; one sided p=0·21). We expect that with week 40). Notably, at the time of primary analysis with a extended follow-up, the progression-free survival curves minimum of 6 months follow-up after random will increasingly converge due to the proportional effect assignment to treatment, four (25%) of 16 participants in of crossover to active therapy in those patients who the placebo group had not progressed. Percentage change were originally allocated to placebo. The first planned in the sum of the diameters of target marker lesions additional sensitivity analysis included three (6%) of from randomisation to beyond week 24 is shown in the 48 participants (one in the cediranib group and two in appendix (pp 3–4). the placebo group) who had non-RECIST confirmed At the time of data cutoff for this analysis (median progression (appendix p 5). This analysis showed no follow-up of 34·3 months [IQR 23·7–55·6]), the follow- evidence of a difference between the two groups up period was dominated by time on open-label (HR 0·73, 90% CI 0·43–1·24; one-sided p=0·16). The cediranib and 43 (90%) of 48 participants (29 from the second planned sensitivity analysis, the landmark www.thelancet.com/oncology Vol 20 July 2019 1029 Articles 50 Cediranib 8·3 weeks 40 Placebo Cediranib (patient progressed*) 40 Placebo (patient progressed*) 15·0 weeks 35 11·0 weeks 16·0 weeks 30 7·6 weeks Percentage change in sum of target marker lesions (%) 25 11·3 weeks 20 7·9 weeks 15 7·6 weeks † 7·7 weeks 10 5 0 7·3 weeks –5 6·9 weeks –10 –15 –20 –25 –30 –35 –40 –45 † –50 Patient Figure 2: Percentage change in sum of target marker lesions from baseline to week 24 (or progression if sooner) in all evaluable participants (n=44) Each bar represents one patient. Where the number of weeks is given, it indicates the timepoint at which progression occurred for those who did not reach the 24 week assessment. *Patients who progressed had either progression of non-target lesions or appearance of new lesions despite a less than 20% decrease in the sum of target marker lesions. †Patient received cediranib before trial entry. 100 overall survival between the groups (figure 4). Overall Cediranib Median progression-free survival 10·1 months (IQR 5·3–19·0) survival estimates at 12 months were 90·3% (95% CI Placebo 72·9–96·8) for cediranib and 68·8% (95% CI Median progression-free survival 4·9 months (IQR 1·9–20·0) 40·5–85·6) for placebo (figure 4). Excluding participants 75 with previous cediranib treatment, the log rank p value Progression-free survival (%) Unadjusted HR 0·82 90% Cl (0·47–1·43); one-sided p=0·28 (one sided) was 0·42. The safety population included 47 (98%) participants 50 who had at least one dose of trial treatment. The side- effect profile was as expected for a VEGFR inhibitor. The most common adverse events on blinded treatment were 25 diarrhoea, hypertension, fatigue, and nausea (table 2). Most grade 3 adverse events on cediranib were hyper­ tension and diarrhoea, which were managed by dose reduction (adverse events that resulted in doses reductions 0 0 3 6 9 12 15 18 21 24 27 30 33 36 are not shown separately in table 2). No grade 4 adverse Time from randomisation (months) events or deaths occurred on cediranib during the masked Number at risk (number censored) phase. 15 serious adverse reactions were reported in Cediranib 32 (1) 27 (1) 19 (1) 16 (1) 12 (1) 11 (3) 7 (3) 5 (3) 4 (3) 2 (3) 1 (3) 0 (3) 0 (3) 12 participants (appendix p 8). Of these serious adverse Placebo 16 (1) 10 (1) 6 (2) 4 (2) 4 (2) 3 (2) 3 (2) 1 (2) 1 (2) 1 (2) 1 (3) 1 (3) 0 (3) reactions, 12 occurred on open-label cediranib. The most Figure 3: Progression-free survival common symptoms were dehydration (n=2), vomiting HR=hazard ratio. (n=2), and proteinuria (n=2). One (2%) of 48 participants died because of an intracranial haemorrhage 41 days analysis, showed a significant improvement in after starting open-label cediranib with no evidence of progression-free survival for participants allocated to cerebral metastases. This event was considered as the cediranib group (appendix p 6). At the time of probably related to treatment. All other deaths were analysis, 24 (50%) of 48 participants had died (16 in the unrelated to treatment and due to ASPS. cediranib group and eight in the placebo group). As An unplanned exploratory analysis to determine expected, we found no evidence of a difference in the effect of previous tyrosine-kinase inhibitors on 1030 www.thelancet.com/oncology Vol 20 July 2019 Articles progression-free survival was difficult to assess because 100 small numbers precluded formal comparisons (appendix p 7). 75 Discussion CASPS has confirmed the activity of cediranib in Overall survival (%) participants with advanced, metastatic ASPS, with a significant difference in the median percentage change 50 in sum of the diameters of target marker lesions at 24 weeks compared with placebo. The primary endpoint Cediranib was chosen to show the degree of tumour shrinkage 25 Median overall survival 27·8 months (IQR 18·8–63·2) occurring in response to cediranib without being Placebo Median overall survival 47·3 months (IQR 7·8–not reached) confounded by treatment crossover. This endpoint was considered a more reliable index of treatment effect than one-sided log-rank p=0·48 0 progression-free survival in light of known indolent 0 3 6 9 12 15 18 21 24 27 30 33 36 progression, spon­taneous stabilisation, and spontaneous Time from randomisation (months) Number at risk regression in this disease.12 Additionally, this uncon­ (number censored) ventional endpoint was considered more sensitive than Cediranib 32 (1) 31 (1) 30 (2) 27 (2) 27 (3) 25 (4) 22 (6) 14 (7) 13 (7) 12 (10) 8 (10) 8 (11) 6 (11) progression-free survival and thus required fewer Placebo 16 (0) 16 (0) 16 (0) 11 (0) 11 (0) 11 (0) 11 (2) 9 (3) 8 (4) 6 (4) 5 (5) 4 (5) 4 (5) participants to show a significant difference between the Figure 4: Overall survival two groups—an important con­sideration for a trial in Hazard ratio was not calculated for overall survival due to violation of the non-proportionality assumption. such a rare disease. The difference in median progression-free survival at the time of the analysis was not significant, with absolute values of 10·1 months (IQR 5·3–19·0) for cediranib and 4·9 months (1·9–20·0) Cediranib group (n=31) Placebo group (n=16) for placebo. Our findings here suggest that cediranib Grade 1–2 Grade 3 Grade 1–2 Grade 3 stabilises metastatic disease and produces objective Diarrhoea 26 (84%) 2 (6%) 6 (38%) 0 responses, with 19% of participants given cediranib Hypertension 20 (65%) 6 (19%) 9 (56%) 0 having a partial response in the first 24 weeks. Some Fatigue 16 (52%) 1 (3%) 6 (38%) 0 participants with stable disease and partial response had Nausea 12 (39%) 0 3 (19%) 0 long-lasting disease control, with a median duration of Dyspnoea 11 (35%) 0 2 (13%) 1 (6%) response of 16 months (IQR 15·7–26·0). Decreased appetite 9 (29%) 0 6 (38%) 0 We recognise that our analyses of secondary endpoints Arthralgia 9 (29%) 0 2 (13%) 0 such as progression-free survival and overall survival are Weight decreased 9 (29%) 0 2 (13%) 0 underpowered. However, the rarity and indolence of Headache 9 (29%) 0 4 (25%) 0 ASPS and the incorporation of crossover within the Hypothyroidism 9 (29%) 0 1 (6%) 0 randomised trial dictated the use of an unconventional primary endpoint and acceptance that demonstration of Cough 8 (26%) 1 (3%) 6 (38%) 0 superiority using more conventional criteria would not Abdominal pain 9 (29%) 0 4 (25%) 0 be possible. Progression-free survival and overall survival Pain in extremity 7 (23%) 1 (3%) 6 (38%) 0 data are reported to enable comparisons with other Constipation 8 (26%) 0 3 (19%) 0 drugs, none of which have yet been studied in a Mucosal inflammation 6 (19%) 1 (3%) 1 (6%) 0 prospective randomised trial (eg, sunitinib, pazopanib, Palmar-plantar erythrodysaesthesia 7 (23%) 0 0 0 axitinib, axitinib plus pembrolizumab, anlotinib). Most Stomatitis 5 (16%) 0 1 (6%) 0 data on other tyrosine-kinase inhibitors are derived from Back pain 5 (16%) 0 1 (6%) 0 reports on small single-centre studies or larger studies in Blood bilirubin increased 4 (13%) 0 1 (6%) 0 soft-tissue sarcomas of which ASPS comprised only a Dry skin 4 (13%) 0 0 0 small component. Toxicity was as expected for a drug Insomnia 4 (13%) 0 2 (13%) 0 of this class and was generally manageable with dose Lymphocyte count decreased 3 (10%) 1 (3%) 1 (6%) 0 interruptions or reductions, with few participants Upper respiratory tract infection 4 (13%) 0 1 (6%) 0 allocated to cediranib withdrawing due to treatment side- Asthenia 3 (10%) 1 (3%) 0 0 effects. Four participants withdrew subsequently because Neutrophil count decreased 4 (13%) 0 1 (6%) 0 of patient choice, two because of adverse events, and Vomiting 4 (13%) 0 1 (6%) 0 one died. One death due to a vascular event was deemed Rash 4 (13%) 0 1 (6%) 0 probably treatment related because thromboembolic Nasopharyngitis 3 (10%) 0 4 (25%) 0 events are a known side-effect of VEGF receptor (Table 2 continues on next page) inhibitors. www.thelancet.com/oncology Vol 20 July 2019 1031 Articles with sunitinib monotherapy (NCT01391962) is currently Cediranib group (n=31) Placebo group (n=16) recruiting participants and has yet to report any results. Grade 1–2 Grade 3 Grade 1–2 Grade 3 This study does not have a no-treatment group, (Continued from previous page) precluding control for spontaneous stabilisation or Chest pain 4 (13%) 1 (3%) 4 (25%) 0 regression or variations in the rate of disease progression Proteinuria 3 (10%) 0 2 (13%) 0 between the groups. Dysphonia 3 (10%) 0 2 (13%) 0 In CASPS, the proportion of participants who had an γ-glutamyltransferase increased 3 (10%) 0 0 1 (6%) objective response at week 24 (19%) was lower than that Oedema peripheral 2 (6%) 0 2 (13%) 0 reported overall by Kummar and colleagues in a phase 2 Epistaxis 2 (6%) 0 2 (13%) 0 trial of cediranib in metastatic ASPS (35%),11 in which Dry mouth 2 (6%) 0 2 (13%) 0 participants had a slightly lower median age than in this Haemoptysis 2 (6%) 0 2 (13%) 0 study (27 years vs 31 years), but a similar proportion had Oropharyngeal pain 2 (6%) 0 2 (13%) 0 received antiangiogenic therapy previously (26% vs 33% Blood amylase increased 0 1 (3%) 2 (13%) 0 in CASPS). A key difference between the studies is that Pain 1 (3%) 0 2 (13%) 0 participants were not required to have had disease Anaemia 1 (3%) 0 3 (19%) 0 progression in the previous 6 months in Kummar and colleagues’ study. Although CASPS was open to Injection site haematoma 0 0 2 (13%) 0 participants aged 16 years and older (in Kummar and Pruritus 0 0 2 (13%) 0 colleagues’ study, the threshold was ≥18 years) only one Monoparesis 0 0 2 (13%) 0 patient younger than 20 years was recruited, showing the Lower respiratory tract infection 0 0 2 (13%) 0 predominance of participating centres only seeing adult Blood alkaline phosphatase increased 1 (3%) 1 (3%) 0 0 patients. Alanine aminotransferase increased 1 (3%) 1 (3%) 0 0 Undertaking a randomised trial in such a rare disease, Pyrexia 2 (6%) 1 (3%) 1 (6%) 0 which has an incidence of less than one per million Hypophosphataemia 2 (6%) 1 (3%) 0 0 people per year, was challenging. The decision to Amenorrhoea 1 (3%) 1 (3%) 0 0 undertake CASPS as a randomised trial was vindicated Partial seizures 0 1 (3%) 0 0 by the fact that seven (44%) of 16 participants in the Adverse events were graded according to Common Terminology Criteria for Adverse Events version 4. Adverse events placebo group had stable disease at 24 weeks, despite all occuring in at least 10% of patients (or one or more patients for grade 3 or worse events) are reported here. There were participants having documented progressive disease no grade 4 adverse events or deaths due to these causes. before study entry. At the time of primary analysis with a Table 2: Adverse events reported during the blinded treatment phase minimum of 6 months follow-up after random assign­ ment to treatment, four (25%) participants in the placebo The idea that inhibitors of angiogenesis, such as VEGF group had still not progressed. Considering these data, receptor inhibitors like cediranib, might be active against following up participants with ASPS for at least 3 months ASPS was prompted by the observations of dormancy and to confirm progressive disease before considering spontaneous stabilisation, and the belief that some form systemic therapy seems reasonable. The decision to of angiogenic switch might be responsible.18 An additional allow crossover to cediranib was made on ethical grounds explanation, which is not mutually exclusive, is that this to enable all participants in the study access to the drug behaviour could be due to immune surveillance, for which at some stage. This choice inevitably led to confounding supportive clinical data now exist from the use of in the interpretation of overall survival and long-term immunomodulatory drugs (eg, pembrolizumab).16 progression-free survival, as shown by the apparent Evidence of lactate as an energy source in ASPS,6 with convergence of the progression-free survival curves at consequent upregulation of HIF1α and VEGF, might 12 months. partially explain the activity of drugs that inhibit VEGF The findings of this study support the concept that receptors, although other targets could be involved. antiangiogenic therapy is active against advanced ASPS. Spontaneous regression could be mediated via the Whether or not the spectrum of receptor tyrosine kinases immune system. A study of the VEGF receptor inhibitor inhibited by cediranib is substantially different from that axitinib in combination with pembrolizumab investi­ of other drugs that are active in this disease (eg, sunitinib) gating immunomodulation in soft-tissue sarcomas is unclear. Based solely on hypertension incidence as a (NCT02301039) is ongoing.16 In participants with ASPS, surrogate marker in this and other studies, cediranib the proportion with an objective response was 45·5% appears to be a potent VEGF receptor inhibitor. (95% CI 18·1–75·4) and with a 3-month progression-free Comparative data on relative potency in vitro against survival estimate of 90·9% (95% CI 50·8–98·7). different receptor tyrosine kinases do not seem helpful To our knowledge, none of the molecularly targeted in predicting toxicity profiles, and hypertension is not a drugs discussed here have been studied as extensively as reliable efficacy biomarker.19 cediranib in ASPS or have shown superior activity. Given the high incidence of metastatic disease and A randomised phase 2 study comparing cediranib poor long-term prognosis of ASPS, together with the lack 1032 www.thelancet.com/oncology Vol 20 July 2019 Articles of efficacy of conventional chemotherapy, the con­ Australia was provided by the Australasian Sarcoma Study Group (ASSG) firmation of significant clinical activity with cediranib in and funding in Spain was provided by AstraZeneca, Spain, and the Grupo Español de Investigación en Sarcomas (GEIS) to cover this disease is an important step towards the goal of long- local costs. The Institute of Cancer Research Clinical Trials and Statistics term disease control for these young patients. Further Unit (ICR-CTSU) also receives programme grant funding from Cancer studies using cediranib in ASPS and other sarcomas, in Research UK (grant number C1491/A15955). We also thank the NIHR conjunction with other drugs with potential activity in CRN: Cancer and the UK National Health Service (NHS) for support in funding the NIHR Biomedical Research Centre at The Royal Marsden the disease, are warranted. and the ICR. We thank all the patients and their families who Contributors participated in this study, all staff involved at the 12 participating IJ was chief investigator; contributed to trial design, protocol hospitals, and the staff involved in the trial at the ICR-CTSU, ASSG, development, participant recruitment, data collection, data The Centre for Biostatistics and Clinical Trials at Peter MacCallum interpretation, and writing; and was a member of the Trial Management Cancer Centre (which manages Australian trial activities on behalf of Group. JPM contributed to trial design, protocol development, statistical ASSG), GEIS, and Sofpromed Investigación Clínica (which manages analysis, and data interpretation, and was a member of the Trial Spanish trial activities on behalf of GEIS). Finally, we thank the past and Management Group. LK contributed to statistical analysis, present colleagues on the Cediranib in Alveolar Soft Part Sarcoma data interpretation, and writing, and was a member of the Trial (CASPS) Trial Management Group, and the CASPS Joint Independent Management Group. ML, VB, QC-H, RC, AD, IH, WJ, ALP, BS, MT, Data Monitoring and Steering Committee. CASPS is cosponsored in the and JMT contributed to participant recruitment and data collection, and European Union by The ICR and The Royal Marsden NHS Foundation were members of the Trial Management Group. CB and CM contributed Trust and in Australia by the Australasian Sarcoma Study Group. to participant recruitment and data collection. LF contributed to protocol References development, trial management, data collection, and data management, 1 Wang WL, Mayordomo E, Zhang W, et al. Detection and and was a member of the Trial Management Group. KJ contributed to characterization of EWSR1/ATF1 and EWSR1/CREB1 chimeric data collection and data management. SK and CT contributed to trial transcripts in clear cell sarcoma (melanoma of soft parts). management, data collection, and data management, and were members Mod Pathol 2009; 22: 1201–09. of the Trial Management Group. CS contributed to protocol development 2 Lieberman PH, Brennan MF, Kimmel M, Erlandson RA, and trial management, and was a member of the Trial Management Garin-Chesa P, Flehinger BY. Alveolar soft-part sarcoma. Group. JMB contributed to trial design, protocol development, statistical A clinico-pathologic study of half a century. Cancer 1989; 63: 1–13. analysis, data interpretation, and writing, and was a member of the Trial 3 Portera CA Jr, Ho V, Patel SR, et al. Alveolar soft part sarcoma: Management Group. All authors reviewed the manuscript prior to clinical course and patterns of metastasis in 70 patients treated at a submission. single institution. Cancer 2001; 91: 585–91. 4 Ladanyi M, Lui MY, Antonescu CR, et al. The der(17)t(X;17)(p11;q25) Declaration of interests of human alveolar soft part sarcoma fuses the TFE3 transcription IJ, JPM, LK, LF, KJ, SK, CS, CT, and JMB report grants and non-financial factor gene to ASPL, a novel gene at 17q25. Oncogene 2001; support in the form of study drug provision and distribution from 20: 48–57. AstraZeneca during the conduct of the study. IJ also reports personal 5 Ladanyi M, Antonescu CR, Drobnjak M, et al. The precrystalline fees from Eli Lilly, Bayer, Nektar, PharmaMar, GlaxoSmithKline, Ariad, cytoplasmic granules of alveolar soft part sarcoma contain and The Institute of Cancer Research outside of the submitted work. monocarboxylate transporter 1 and CD147. Am J Pathol 2002; JMB also reports grants and non-financial support from AstraZeneca, 160: 1215–21. Merck Sharp & Dohme, Puma Biotechnology, Clovis Oncology, 6 Goodwin ML, Jin H, Straessler K, et al. Modeling alveolar soft part and Janssen-Cilag; grants, non-financial support, and travel support sarcomagenesis in the mouse: a role for lactate in the tumor from Pfizer; and grants from Medivation, Novartis, and Roche outside of microenvironment. Cancer Cell 2014; 26: 851–62. the submitted work. All other authors declare no competing interests. 7 Lu H, Forbes RA, Verma A. Hypoxia-inducible factor 1 activation by aerobic glycolysis implicates the Warburg effect in carcinogenesis. Data sharing J Biol Chem 2002; 277: 23111–15. The Institute of Cancer Research, Clinical Trials and Statistics Unit 8 Sonveaux P, Copetti T, De Saedeleer CJ, et al. Targeting the lactate (ICR-CTSU) supports the wider dissemination of information from the transporter MCT1 in endothelial cells inhibits lactate-induced HIF-1 research it conducts and increased cooperation between investigators. activation and tumor angiogenesis. PLoS One 2012; 7: e33418. Trial data is collected, managed, stored, shared, and archived according to 9 Langenberg MH, van Herpen CM, De Bono J, et al. Effective ICR-CTSU Standard Operating Procedures to ensure the enduring strategies for management of hypertension after vascular quality, integrity, and utility of the data. Formal requests for data sharing endothelial growth factor signaling inhibition therapy: results from are considered in line with ICR-CTSU procedures with due regard given a phase II randomized, factorial, double-blind study of cediranib in to funder and sponsor guidelines. Requests are via a standard proforma patients with advanced solid tumors. J Clin Oncol 2009; describing the nature of the proposed research and extent of data 27: 6152–59. requirements. Data recipients are required to enter a formal data sharing 10 Judson I, Scurr M, Gardner K, et al. Phase II study of cediranib in agreement that describes the conditions for release and requirements for patients with advanced gastrointestinal stromal tumors or data transfer, storage, archiving, publication, and Intellectual Property. soft-tissue sarcoma. Clin Cancer Res 2014; 20: 3603–12. Requests are reviewed by the Trial Management Group (TMG) in terms 11 Kummar S, Allen D, Monks A, et al. Cediranib for metastatic of scientific merit and ethical considerations including patient consent. alveolar soft part sarcoma. J Clin Oncol 2013; 31: 2296–302. Data sharing is undertaken if proposed projects have a sound scientific or 12 Stacchiotti S, Negri T, Zaffaroni N, et al. Sunitinib in advanced patient benefit rationale as agreed by the TMG and approved by the alveolar soft part sarcoma: evidence of a direct antitumor effect. Independent Data Monitoring and Steering Committee as required. Ann Oncol 2011; 22: 1682–90. Restrictions relating to patient confidentiality and consent will be limited 13 Stacchiotti S, Mir O, Le Cesne A, et al. Activity of pazopanib and by aggregating and anonymising identifiable patient data. Additionally, trabectedin in advanced alveolar soft part sarcoma. Oncologist 2018; all indirect identifiers that could lead to deductive disclosures will be 23: 62–70. removed in line with Cancer Research UK Data Sharing Guidelines. 14 Schöffski P, Wozniak A, Kasper B, et al. Activity and safety of crizotinib in patients with alveolar soft part sarcoma with Acknowledgments rearrangement of TFE3: European Organization for Research We thank the investigator-sponsored study collaboration between and Treatment of Cancer (EORTC) phase II trial 90101 ‘CREATE’. AstraZeneca and the National Institute for Health Research (NIHR) Ann Oncol 2018; 29: 758–65. Clinical Research Network (CRN): Cancer for their support, with funding 15 Chi Y, Fang Z, Hong X, et al. Safety and efficacy of anlotinib, from AstraZeneca and for the study grant from Cancer Research UK a multikinase angiogenesis inhibitor, in patients with refractory (CRUK/10/021 grant reference C2130/A12118). Additionally, funding in metastatic soft-tissue sarcoma. Clin Cancer Res 2018; 24: 5233–38. www.thelancet.com/oncology Vol 20 July 2019 1033 Articles 16 Wilky BA, Trucco MM, Subhawong TK, et al. Axitinib plus 18 Holmgren L, O’Reilly MS, Folkman J. Dormancy of pembrolizumab in patients with advanced sarcomas including micrometastases: balanced proliferation and apoptosis in the alveolar soft part sarcoma: a single-centre, single-arm, phase 2 trial. presence of angiogenesis suppression. Nat Med 1995; 1: 149–53. Lancet Oncol 2019; published online May 8. http://doi.org/10.1016/ 19 Duffaud F, Sleijfer S, Litière S, et al. Hypertension (HTN) as a S1470-2045(19)30153-6. potential biomarker of efficacy in pazopanib-treated patients with 17 Judson IR, Morden JP, Leahy MG, et al. Activity of cediranib in advanced non-adipocytic soft tissue sarcoma. A retrospective study alveolar soft part sarcoma (ASPS) confirmed by CASPS (cediranib based on European Organisation for Research and Treatment of in ASPS), an international, randomised phase II trial (C2130/ Cancer (EORTC) 62043 and 62072 trials. Eur J Cancer 2015; A12118). Proc Am Soc Clin Oncol 2017; 35 (suppl 15): 11004 (abstr). 51: 2615–23. 1034 www.thelancet.com/oncology Vol 20 July 2019