Doxorubicin-loaded nanoparticles for patients with advanced hepatocellular carcinoma after sorafenib treatment failure (RELIVE): a phase 3 randomised controlled trial

Service d’Hépatologie et Gastroentérologie, Hôpital de la Croix-Rousse, Lyon, France (Prof P Merle MD); Hôpital Haut Lévèque, Pessac, France (Prof J-F Blanc MD); Centre Hospitalier Universitaire (CHU) de Saint-Étienne, Saint-Étienne, France (Prof J-M Phelip MD); Hôpital Paul Brousse, Villejuif, France (G Pelletier MD); CHU de Nancy—Hôpital Brabois, Nancy, France (Prof J-P Bronowicki MD); CHU Nantes-Hôpital Hôtel Dieu, Nantes, France (Y Touchefeu MD); CHU de Montpellier—Hôpital Saint-Eloi, Montpellier, France (Prof G Pageaux MD); Hôpital de la Conception, Marseille, France (Prof R Gerolami MD); Centre Hospitalier Regional Universitaire de Strasbourg— Hôpital Civil, Strasbourg, France (F Habersetzer MD); CHU d’Amiens, Hôpital Nord, Amiens, France (Prof E Nguyen-Khac MD); Istituto Romagnolo Ricerca e Cura dei Tumori, Meldola, Italy (Prof A Casadei-Gardini MD); Université Catholique de Louvain Saint-Luc, Brussels, Belgium (Prof I Borbath MD); CHU de Nice—Hôpital Archet, Nice, France (Prof A Tran MD); Universitätsklinikum Hamburg–Eppendorf, Hamburg, Germany (Prof H Wege MD); Department of Oncology, Ain Shams University Hospitals, Cairo, Doxorubicin-loaded nanoparticles for patients with advanced hepatocellular carcinoma after sorafenib treatment failure (RELIVE): a phase 3 randomised controlled trial


Introduction
The treatment of hepatocellular carcinoma follows guidelines based on the Barcelona Clinic Liver Cancer (BCLC) staging system. 1 Surgical resection, trans plantation, and thermoablation are potential curative therapies for patients with earlystage hepatocellular carcinoma, whereas chemoembolisation is recom mended as a palliative option for intermediatestage hepatocellular carcinoma. For patients with advanced hepatocellular carcinoma, or with intermediate stage disease that is no longer a candidate for chemo embolisation, systemic strategies based on oral tyrosine kinase inhibitors such as sorafenib as firstline treatment 2,3 and regorafenib as secondline treatment 4 provide a clinically significant improvement in overall survival. Lenvatinib is non inferior to sorafenib as firstline treatment 5 and cabozantinib is efficacious as secondline or thirdline treatment. 6 Ramucirumab, a monoclonal anti body targeting vascular endothelial growth factor receptor 2, has shown efficacy in the subgroup of patients with hepatocellular carcinoma who have concentrations of alphafetoprotein of at least 400 ng/mL after sorafenib treatment failure, 7 whereas ramucirumab did not show any significant benefit in nonselected patients with hepatocellular carcinoma. 8 All other systemic drugs in phase 3 trials showed no efficacy in this population. [9][10][11][12][13][14][15][16] Further, radio embolisation with yttrium90 has not shown superiority to sorafenib in a randomised phase 3 study in patients with advanced hepatocellular carcinoma or in those with intermediate hepatocellular carcinoma in whom chemo embolisation has failed. 17,18 The best observed overall survival in a systemic setting was for sorafenib followed by regorafenib (median 26·0 months [95% CI 22·6-28·1]). 19 More effective systemic therapies are needed to increase overall survival of patients with advanced hepatocellular carcinoma.
To date, no phase 3 trials of cytotoxic chemotherapy for hepatocellular carcinoma have shown signs of efficacy. Doxorubicin was a potential candidate, but the administration of free doxorubicin is associated with high morbidity in cirrhosis; 20 it also did not show any additive or synergistic effects when added to sorafenib. 21 Doxorubicinloaded nanoparticles in the liver overwhelm the efflux pumps encoded by multiple drug resistance genes. 22,23 A phase 1-2 trial suggested a potential benefit of doxorubicinloaded nano particles on overall survival of patients with hepato cellular carcinoma, although the trial was prematurely stopped because of lung toxicity associated with doxorubicin loaded nanoparticles injected by the hepatic arterial route. 24 Preclinical data from Wistar rats showed that this lung toxicity was reduced when doxorubicinloaded nanoparticles were infused over 2 h. 24 Thus, here, we assessed the efficiency of doxorubicinloaded nano particles admin istered by a 6 h intravenous infusion in patients with hepatocellular carcinoma after failure of sorafenib therapy.

Study design
This multicentre, openlabel, randomised, controlled phase 3 trial was done at 70 sites in 11 countries in Europe, the USA, the Middle East, and North Africa. The trial was approved by each centre's ethics committee or institutional review board and complied with Good Clinical Practice guidelines, the Declaration of Helsinki, and applicable local laws. The protocol is available online.

Patients
Eligibility criteria were age of at least 18 years; hepatocellular carcinoma confirmed by pathological or noninvasive assessment according to the American Association for the Study of Liver Diseases or European Association for the Study of the Liver as per protocol criteria; 1

Research in context
Evidence before this study We searched PubMed for phase 3 randomised controlled studies of advanced inoperable hepatocellular carcinoma, published between Jan 1, 2000, and Feb 4, 2018, and in English. We used the search terms "hepatocellular carcinoma" AND "randomized trial" AND "chemotherapy" OR "doxorubicin". Our search showed that no cytotoxic chemotherapy has improved survival in a previous randomised controlled phase 3 trial in patients with advanced hepatocellular carcinoma, except the small study published by Lai and colleagues in 1988. However, although free doxorubicin was shown to be effective in terms of overall survival, this drug has never become the standard of care for advanced hepatocellular carcinoma because of the weak evidence (small sample size) and the high toxicity in patients with cirrhosis (sepsis, mucositis, and cardiotoxicity).

Added value of this study
The results of RELIVE show that treatment with doxorubicin-loaded nanoparticles, which overrides multiple mechanisms of drug resistance-related chemoresistance, did not result in a significant improvement in overall survival compared with best standard of care in patients with disease progression on sorafenib alone or with other subsequent systemic treatment lines. The secondary endpoints of progression-free survival, time to progression, disease control, and overall tumour response also showed no improvement.

Implications of all the available evidence
This phase 3 trial of doxorubicin-loaded nanoparticles demonstrates the absence of a benefit in overall survival for patients with advanced hepatocellular carcinoma in whom sorafenib treatment had failed. These findings are in contrast with preclinical and phase 1 or 2 clinical studies that had positive results with this treatment for hepatocellular carcinoma. The absence of more effective therapies is an unmet clinical need, but so far, chemotherapy has been clearly demonstrated to be ineffective and toxic in these patients.
For the study protocol see http://www.onxeo.com/site/ wp-content/ uploads/2019/03/20170727_ BA003_Protocol_Relive_v7.pdf past 5 years; HIV infection; hepatocellular carcinoma on transplanted liver; risk of variceal bleeding; previous cumulative dose of more than 300 mg/m² doxorubicin; ongoing immuno suppressive treatment; unstable medical or surgical conditions, particularly uncontrolled diabetes, that might disrupt study participation; un controlled systemic infection; life expectancy less than 2 months; receipt of an experimental drug in another clinical trial in the past 30 days; and unwillingness or inability to use two forms of contra ception for 6 months after final study drug administration. All patients provided written informed consent.

Randomisation and masking
Patients were randomly assigned to receive either 30 mg/m² doxorubicinloaded nanoparticles (30 mg/m² group) or 20 mg/m² doxorubicinloaded nanoparticles (20 mg/m² group) or standard care (1:1:1) using a computergenerated randomisation list prepared by the funder. This list was stratified by geographic region (Europe, USA, or Middle East and North Africa) using blocks (size 6). Investigators, patients, and the funder were unmasked to treatment assignment in this open label trial. However, independent central review as per RECIST, version 1.1, and data review by the data review committee before database lock were performed blindly. The assignment of number and code for patient identification ensured patient anonymity.

Procedures
In both experimental groups, doxorubicinloaded nanoparticles were delivered by intravenous perfusion every 4 weeks with a maximum allowed cumulative dose of doxorubicin of 550 mg/m². Patients assigned to the 120 received allocated intervention (safety population) standard care control group received any systemic anticancer therapy (except sorafenib) according to the centre's practice and the decision of the principal investigator at that centre, being aware that any type of these systemic therapies had not shown efficacy in phase 3 trials at the time of randomisation. In all groups, patients received best supportive care. Treatment continued until disease progression as defined by RECIST, version 1.1, or clinical progression, death, unacceptable toxicity, withdrawal of consent by the patient, or decision by the principal investigator. Patients were followed up for tumour assessments every 8 weeks. Treatment could be continued beyond progression at the decision of the principal investigator. To prevent the occurrence of acute respiratory adverse events that we observed in our phase 1-2 trial, 24 perfusion of doxorubicinloaded nanoparticles was done over 6 h intravenously with safety measures (premedication with methyl prednisolone 32 mg orally and one antihistamine drug given 24 h and 1 h before perfusion and 24 h after perfusion). Respiratory symptoms and oxygen saturation were continuously monitored during the 6 h of perfusion: in case of dyspnoea or oxygen saturation decrease from 95% or more to 93% or less, the infusion rate was reduced by half (to a 12 h infusion) without changing the total dose; in case of persistence of dyspnoea beyond 1 h or oxygen saturation decrease to 90% or less, perfusion was immediately and definitively stopped. Safety was monitored continuously throughout the study and patients had safety assessments every 4week treatment cycle. Blood tests were assessed every 2 weeks. Adverse events were graded using National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.03.

Outcomes
The primary endpoint was overall survival, defined as the time from randomisation to death from any cause. Data are n (%) or median (IQR). ECOG=Eastern Cooperative Oncology Group. *The Child-Pugh system describes liver disease severity: patients are divided into classes A-C. †Patients may have more than one cause. ‡Patient can be counted more than once in case of multiple treatments. Secondary endpoints were safety, and overall survival for patients with Child-Pugh score A, progressionfree survival (defined as time from randomisation to radiological or clinical disease progression or death), the proportion of patients achieving an objective response (defined as a complete or partial response), and the proportion of patients achieving disease control (defined as complete response, partial response, or stable disease maintained for ≥8 weeks). Responses to treatment were assessed using RECIST (version 1.1), with independent central review. Safety was assessed by adverse events, laboratory abnormalities, vital signs, chest xray, and left ventricular ejection fraction by cardiac echography and electro cardiography. Exploratory endpoints were progressionfree survival and objective response as assessed by investigators and time to progression (time from randomisation to radiological or clinical disease progression) assessed by independent central review per RECIST version 1.1.

Statistical analysis
At study initiation in 2011, the initial sample size calculation was done on the basis of an estimated median survival of 6·6 months in the control group and 10·9 months in the experimental groups (hazard ratio [HR] 0·60); an accrual period of 36 months; and a one sided α level of 2·5%. The required sample size to achieve a 90% power was 130 patients per group (for the two tests of the two doses at a 5% level).
In 2016, Bruix and colleagues 4 published the results of a phase 3 trial of regorafenib, with a median overall survival of 7·8 months (95% CI 6·3-8·8) in the placebo group. The revised power of the estimated required sample size for our study in view of these new results and according to our accrual period of 54 months would be decreased from 90% to 58%.
Considering these results, study feasibility, and the need for results in this serious, lifethreatening disease, the statistical analysis plan was amended after validation by the US Food and Drug Administration (July 24, 2017) and signed off by the scientific committee (Aug 25, 2017) before database lock (Aug 28, 2017). The revised statistical analysis plan pooled the two doxorubicinloaded nanoparticle groups; assuming a median overall survival in the control group of 8 months, and aiming for a hazard ratio of 0·69, the required total sample size to achieve 85% power to compare the experimental groups with control (twosided α of 5%) was 348 patients (116 patients in the control group and 232 patients in the pooled experimental group). We expected recruitment to take 55 months, with 6 months of followup after the last inclusion (total followup 61 months), and around 10% of patients to be lost to followup. Thus the recalculated total sample size was 390 patients. The analysis was planned for when 285 events (deaths) occurred. For the primary efficacy endpoint of overall survival and the secondary endpoint of progressionfree survival, the groups were compared using a nonstratified logrank test. The HR for overall survival and its 95% CI were calculated using the stratified Cox model.
The primary analysis was done in the intentionto treat (ITT) population, defined as all patients who had been randomly assigned to a group; safety analyses included all patients who received at least one dose of the study drug. The study was overseen by an independent data safety monitoring committee. To  assess the primary endpoint of overall survival in the ITT population and the secondary endpoints of overall survival in the subpopulation of patients with Child-Pugh score A and progressionfree survival and objective response in the whole population and Child-Pugh score A subpopulation, we used a hierarchical sequential closedtest procedure to control the overall type I error rate of 5%, with the following sequence: overall survival in the ITT population, overall survival in the Child-Pugh A subpopulation, progressionfree survival in the Child-Pugh A subpopulation, and objective response in the Child-Pugh A subpopulation. If the closedtest procedure fails, all other analyses will be presented as exploratory.
We did a sensitivity analysis using a Cox model adjusting for predefined selected prognostic factors. We first analysed these prognostic factors in separate univariate analyses and then in multivariate analysis. More specifically, we tested each potential predictor in a Cox model in which the considered predictor was the only covariate included (treatment was not included in the model). We selected potential predictors with a p value    less than 0·10 for the multivariate analysis. We then included the selected predictors in a multivariate Cox model (treatment not included in the model) and further selected them with a backward selection procedure eliminating covariates with a p value above 0·10 in presence of the other covariates. We then introduced treatment and well known predictors (macroscopic vascular invasion, extrahepatic spread, Child-Pugh, hepatitis B virus infection, and alphafetoprotein) as an additional covariate in the reduced model obtained at the end of the backward selection procedure. We tested the covariate by treatment interactions by adding in a separate model all interactions corresponding to the finally retained covariates. We compared overall survival using a naive test based on the comparison of Kaplan-Meier estimates of survival. We compared proportions of patients achieving responses and disease control in the two groups using Fisher's exact test.
We did statistical analyses with the SAS software, version 9.4. This trial is registered with ClinicalTrials.gov, number NCT01655693.

Role of the funding source
The funder was involved in study design, data collection, analysis, and interpretation, and writing of the report. Data management was done by Lincoln Pharmaceuticals and Aixial and statistical analyses were done by Chiltern International, both supervised by eXYSTAT. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Results
541 patients were screened between June 15, 2012, and Jan 27, 2017, and 144 were excluded because they did not meet eligibility criteria. 397 patients were randomly assigned to either the 30 mg/m² group (n=133), the 20 mg/m² group (n=130), or the control group (n=134) and included in ITT analysis (figure 1). 376 (95%) patients started treatment (120 in the 30 mg/m² group, 122 in the 20 mg/m² group, and 134 in the control group) and comprised the safety population. Of the patients who started treatment, 113 (94%) of 120 receiving 30 mg/m² doxorubicinloaded nanoparticles, 118 (97%) of 122 receiving 20 mg/m² doxorubicinloaded nano particles, and 130 (97%) of 134 patients in the control group discontinued study treatment. The most common reason for discontinuation was radiological progression (77 [68%] of 113 in the 30 mg/m² group, 91 [77%] of 118 in the 20 mg/m² group, and 58 [45%] of 130 in the control). The mean treatment duration was 3·7 months (SD 3·9) in the 30 mg/m² group, 3·4 months (3·9) in the 20 mg/m² group, and 3·0 months (2·6) in the control group. In both experimental groups, the mean number of cycles was 4·6 (SD 4·0); 64 (26%) of 244 patients who received their allocated intervention delayed at least one treatment cycle, and 17 (7%) patients had at least one dose reduction; the mean doseintensity of doxorubicinloaded nanoparticles was 99% (SD 7). In the control group, 55 (41%) of 134 patients received only best supportive care whereas 79 (59%) were administered a systemic anticancer therapy considered as the best standard of care by the investigator, the most common of which was oxaliplatin plus gemcitabine in 37 (28%) patients (appendix p 12).
There was no significant difference in overall survival between the groups in the ITT population; median overall survival was 9·1 months (95% CI 8·1-10·4) in the pooled doxorubicinloaded nanoparticle group and 9·0 months (7·1-11·8) in the control group (HR 1·00 [95% CI 0·78-1·28], twosided p=0·99; figure 2A). In exploratory analyses, overall survival since start of sorafenib was also not significantly different between the groups, (HR 1·04 [95% CI 0·81-1·32], p=0·77). Overall survival in the Child-Pugh A subpopulation was also similar between groups: median overall survival was 10·1 months (95% CI 8·8-11·6) in the pooled experimental group and 10·7 months (7·2-12·4) in the control group (HR 0·95 [95% CI 0·72-1·26], p=0·74; In other exploratory analyses, time to progression by independent central review showed similar results, with an HR of 0·96 (95% CI 0·74-1·23; twosided p=0·74; appendix p 10) and a median time to progression of 2·3 months (95% CI 2·1-2·6) in the pooled group and 2·3 months (2·1-2·8) in the control group.   263 patients in the pooled doxorubicinloaded nano particles group and 72 [54%] of 134 in the control group; table 2). Review was not possible for 121 patients because of absence of imaging data (n=29), poor quality imaging or identification issues (n=11), and presence of baseline imaging but no followup imaging (n=81). Among those with available data, the proportion of patients achieving an objective response or disease control was similar in both groups, by both independent and investigator review (table 2). In the Child-Pugh A subpopulation, response were also similar between the groups (no patients achieved a complete response in either group, partial response in two  (table 3). These were deemed related to the study drug in 177 (73%) patients in the pooled doxorubicin group and 58 (43%) of the patients in the control group. Serious adverse events occurred in 74 (31%) patients receiving doxorubicinloaded nano particles and 48 (36%) in the control group, and were attributed to the study drug in 31 (13%) cases in the doxorubicin group and 13 (10%) in the control group. The most common grade 3 or 4 drugrelated treatmentemergent adverse events were asthenia (six [3%] of 242 patients) and neutropenia (25 [10%]) in the pooled doxorubicin group and asthenia (four [3%] of 134), neutropenia (eight [6%]), and thrombo cytopenia (ten [7%]) in the control group. Neutropenia was more frequent in those treated with doxorubicin loaded nanoparticles than in those in the control group; by contrast, thrombocytopenia was less common in those treated with doxorubicinloaded nanoparticles than in those in the control group (table 3).
Cardiorespiratory toxicity was rare and not severe for most patients treated with doxorubicinloaded nano particles. Asymptomatic decreases of left ventricular ejection fraction below 50% occurred in five (2%) of 242 patients; respiratory symptoms in 11 (5%); and oxygen desaturation in 31 (13%), which led to interruption or reduction of the speed of perfusion in 13 (5%) patients. Few

Discussion
The absence of more effective therapies for hepatocellular carcinoma is an unmet clinical need; however, chemo therapy has been clearly shown to be ineffective and toxic in patients with advanced disease. In this phase 3 trial assessing doxorubicinloaded nanoparticles as subsequent line treatment for patients in whom sorafenib has failed, no difference was detected between doxorubicinloaded nanoparticles and control in terms of overall survival.
Survival in patients with hepatocellular carcinoma is influenced by many factors related not only to tumour burden but also to underlying liver conditions, and minor imbalances in prognostic factors can have a meaningful effect on overall survival. Nonetheless, baseline characteristics were similar between the groups and thus are not a possible explanation for these results. One explanation could be insufficient power, because patients with a better prognosis than in previous trials were enrolled. 9,13 Indeed, median overall survival in the control group was unexpectedly high in the whole population (9·0 months [95% CI 7·1-11·8]) as well as in the subpopulation of patients with Child-Pugh score A (10·7 months [7·2-12·4]). By contrast, median overall survival results in the Child-Pugh A populations of other phase 3 trials such as SHARP 2 (sorafenib) and RESORCE 4 (regorafenib) were lower: median overall survival was 7·9 months in the control group of both studies, and was 8·0 months in the cabozantinib trial. 6 Equivalent data were observed for overall survival in the control groups of other phase 3 trials of drugs being tested after failure of sorafenib: in the brivanib trial it was 8·2 months, 13 7·3 months in the trial of everolimus, 14 and 7·6 months in the REACH trial of ramucirumab; 8 overall survival in the control group in a phase 3 study of tivantinib was 9·1 months. 9 In the RELIVE trial, only threequarters of patients had previously received only sorafenib as systemic treatment whereas a quarter had received at least two lines of treatment (sorafenib plus one or more additional lines  Another possible explanation for the failure to detect a treatment difference in the RELIVE study is based on the fact that all the phase 3 trials used a placebo as the control group, whereas the control group in our trial was standard treatment, at the decision of each principal investigator. In our control group, although 55 (41%) of 134 patients received only best supportive care, 79 (59%) received a systemic anticancer therapy, of whom 37 (47%) were given gemcitabine plus oxaliplatin (GEMOX). It is possible that GEMOX might be of benefit for patients with hepato cellular carcinoma, thus concealing the potential benefit of doxorubicinloaded nanoparticles. Data from phase 3 randomised controlled trials using GEMOX in advanced hepatocellular carcinoma are still needed. In a prospective cohort study by Taïeb and colleagues, 25 overall survival with GEMOX was 12 months and in a phase 2 singlearm study of 32 patients by Louafi and colleagues, 26 it was 11·5 months. In addition, in a large multicentre retro spective study of 204 patients, overall survival with GEMOX was 11 months. 27 By contrast, overall survival with free doxo rubicin was only 4·9 months in a phase 2 randomised controlled trial in Asia. 28 Furthermore, the antitumour activity of doxorubicin loaded nanoparticles might not be strong enough to extend survival. This idea is supported by the negative findings of the secondary and exploratory endpoints such as progressionfree survival, objective response, and time to progression, as well as subgroup analyses, which all clearly show no difference between the 20 mg/m² and the 30 mg/m² groups. Furthermore, the 30 mg/m² group had more patients with drugrelated treatmentemergent adverse events than the 20 mg/m² group, thus demonstrating a dosedependent toxicity of doxorubicin.
Consistent with previously published data, clinically significant (observed in at least 10% of patients) drug related treatmentemergent adverse events of any grade attributable to doxorubicin were mostly asthenia, nausea, vomiting, and neutropenia, whereas thrombo cytopenia (probably due to gemcitabine) and par aesthesia (probably due to oxaliplatin) were observed in the control group. Although acute respiratory distress syndrome occurred in some patients due to intrahepatic arterial injection of doxorubicinloaded nanoparticles in the phase 1-2 trial, 24 no clinically significant pulmonary treatmentemergent adverse events were observed after 6 h of intravenous perfusion in RELIVE. Headache was more prevalent in the patients who received the nanoparticleloaded  Table 3: Treatment-emergent adverse events and drug-related treatment-emergent adverse events of any grade occurring in at least 10% of patients in either treatment group (safety population) doxorubicin, which might be specific to the nanoformu lation, since this is not commonly reported with free doxorubicin, but is reported with other forms of nano formulation of doxorubicin such as liposomal doxo rubicin. 29 Causes of three deaths (in the pooled experimental group) considered by investigators and the data safety monitoring board to be treatment related were not unusual for this patient population (interstitial lung disease, lung infection, and peritoneal haemor rhage). Of note, only about a tenth of patients in the pooled doxorubicinloaded nanoparticle group withdrew from the trial prematurely because of drug toxicity.
In conclusion, this first phase 3 study of doxorubicin loaded nanoparticles in patients with advanced hepato cellular carcinoma who have already been treated with sorafenib did increase overall survival. The results of our trial could inform the design of future studies in this patient population.

Contributors
PM, PA, and BV conceived and designed the study. Principal investigators at each site enrolled patients. PM and JLeB collected the data. PM, BV, and JLeB analysed and interpreted the data. All authors participated in the drafting, review, and approval of the manuscript and the decision to submit for publication.

Declaration of interests
PM has received consultancy and advisory fees from Onxeo, Bayer, Lilly, Ipsen, BristolMyers Squibb (BMS), and Merck Sharp & Dohme (MSD). JFB has received consultancy and advisory fees from Onxeo, Bayer, Lilly, Ipsen, and BMS. JPB has received consultancy and advisory fees from Onxeo and Bayer. GP has received consultancy and advisory fees from Bayer, AbbVie, Novartis, and Gilead. AA has received consultancy and advisory fees from AbbVie, MSD, and Gilead. IW has received consultancy and advisory fees from Janssen, AbbVie, MSD, Marcyrl, Pharco, and Gilead. NSY reports grants from Penn State Cancer Institute, Halozyme, Boston Biomedical, Caris Life Sciences, Foundation Medicine, Pharmacyclics, EMD Serono, Onxeo, Regeneron, Momenta, Merck, Daiichi Sankyo, Eli Lilly, Novartis, Taiho, Bayer, Celgene, Lexicon, Incyte, Pfizer, and BMS. PA is an Onxeo shareholder and an author of a patent for nanoparticles loaded with chemotherapeutic antitumoral drug (US Patent and Trademark Office 9763874). All other authors declare no competing interests.

Data sharing
Data collected for this study (including individual participant data and a data dictionary defining each field in the set) will be made available to others after approval by the corresponding author. Additional, related documents will be also available (case report forms, statistical analysis plan, and informed consent forms) on request by the corresponding author and sponsor's approval. These data will be available after publication.