ATLANTIS: An adaptive multi-arm phase II trial of maintenance targeted therapy after chemotherapy in patients with metastatic urothelial cancer

Background Metastatic urothelial cancer (UC) is the eighth most common cause of cancer death in the UK. Standard first-line treatment, for most patients, is cytotoxic chemotherapy. Although UC is initially sensitive to chemotherapy, relapse is almost inevitable after which outcomes are poor, with median overall survival 8 months. There is, therefore, an urgent need for novel therapies to improve outcomes for this patient group. Methods ATLANTIS is a randomised phase II, adaptive design trial of maintenance therapy in biomarker defined subgroups of patients with advanced UC. The primary end-point is progression-free survival (PFS) and the study involves over 30 UK cancer centres. Discussion ATLANTIS is the first study in the UK to employ a precision medicine approach to UC patients for maintenance treatment. Agents with positive efficacy signal will proceed to randomised phase III trials to confirm activity of novel biologically stratified therapies in UC.

treatment is currently limited to patients with high PDL-1 expression who are not suitable for platinum-based chemotherapy 6,7 . There are though, still a majority of patients with advanced UC who do not derive significant benefit from immune checkpoint inhibitors and whose subsequent treatment options are very limited. Second-line chemotherapy may be used, but response rates are low and benefit compared to best supportive care is uncertain, although recent data suggest that the combination of docetaxel and the vascular endothelial growth factor targeted antibody ramucirumab may provide modest benefits in selected patients in the second line setting 8 .
There is therefore a clear need for new effective treatments for patients with advanced UC. The molecular heterogeneity of urothelial cancer suggests that patients may be well served by a precision-medicine approach. Testing new drugs in combination with first line chemotherapy is often challenging due to toxicity of combinations in this patient group 9 .
Studies in the second line setting have historically been limited by the high symptom burden and poor prognosis for patients. Therefore, maintaining clinical benefit after first line chemotherapy may be an attractive way to improve outcomes for patients with advanced UC.
Maintenance therapy after first line chemotherapy is an opportunity for novel drug development in advanced UC. This was demonstrated in the UK NCRN LAMB trial, which completed accrual in 2013 having screened 520 patients and randomised 221 patients with epidermal growth factor expressing urothelial tumours from over 40 UK sites between lapatinib and placebo (NCT00949455). In this trial, lapatinib did not prolong survival or progression free survival in any of the defined subgroups, confirming observation alone as the standard of care in this patient group 10 . The study did reinforce the proof of concept that such a trial would be acceptable to patients and could be successfully delivered in the UK.

Methods And Study Design
ATLANTIS is a multi-centre randomised phase II signal-searching trial of targeted novel agents in biomarker-defined subgroups using an adaptive design (Figure 1). Multiple novel agents will be used in parallel and patients will be entered into ATLANTIS subgroup studies dependent on tumour biomarker profile. The control arm for each comparison will be placebo and comparison will be double-blind where possible.

Trial Population
The target population in ATLANTIS is patients with metastatic or locally advanced urothelial cancer (T4b and/or N1-3 and/or M1) who are not being considered for radical therapy. Patients must have achieved an objective response or stable disease, according to local radiology review, after 4-8 cycles of first-line chemotherapy. Patients must be able to start maintenance treatment within the study at least 3 but no more than 10 weeks after completion of their first line of chemotherapy for metastatic or locally advanced disease. Biomarker analysis of archival tissue to determine ATLANTIS biomarker defined subgroups can occur any time after the diagnosis of UC, as long as appropriate consent has been taken.

Study Objectives and End Points
The principal research question is whether molecularly-targeted maintenance therapy after chemotherapy can delay the time to progression in molecularly-selected patients with advanced UC. ATLANTIS will thereby establish initial evidence of activity for the novel drug/biomarker combinations used in order to justify further phase III validation. A number of drugs will be tested, each compared to placebo or, where this is not feasible, observation alone. Treatment will be allocated based on molecularly defined subgroups of patients (where laboratory/clinical evidence to support such enrichment is clear) or in a manner that allows exploration of, or provides initial evidence for, predictive biomarkers.
Of note, it is anticipated that in most cases the biomarker for a particular arm of the trial will itself be experimental and not yet prospectively validated.
The primary end point is progression-free survival (PFS). This has been chosen as it is largely objective and the majority of patients with UC display progression in accordance with RECIST 1.1 criteria. PFS is also clinically meaningful as the progression after first line chemotherapy represents the transition to the lethal stage of the disease and often the requirement for further systemic therapy.
The secondary end points in ATLANTIS are overall survival (OS), response rate, maximum percentage decrease in measurable disease, safety and tolerability. Exploratory end points are progression free survival in biomarker-defined subgroups other than those used for selection. Exploratory research hypotheses are embedded in the primary purpose of the trial. As such, all patients must provide adequate tissue for biomarker analysis prior to participation in the trial. This tissue collection will also provide a bio-resource for future research relating to UC.
The SPIRIT figure for ATLANTIS trial is shown in Figure 2.

Trial analysis plan
The analysis for all subgroups in ATLANTIS will be on an intention-to-treat basis.
Progression-free survival will be compared between the novel agent and placebo in each study arm within the context of a Cox model incorporating the baseline minimisation factors. The p-value for the observed hazard ratio will be determined from this model. If the observed PFS difference in favour of the novel agent is statistically significant at the 10% 1-sided level, this will be deemed a clear signal that a subsequent phase III trial is warranted. A result significant at the 20% 1-sided level would require further evidence in terms of improvement such as reduction in size of measurable disease. The overall study data will be reviewed by an Independent Data Monitoring Committee (IDMC). The IDMC will review toxicity, treatment delivery and compliance data. Recommendations will take into account all available data as well as formal futility comparison for PFS when half of the required events have occurred.

Current biomarker-defined arms in ATLANTIS
The design of ATLANTIS allows for the addition of further biomarker-defined subgroups throughout the lifetime of the trial. At initiation, ATLANTIS was exploring a single drug (cabozantinib) but other comparisons have been added since the trial began, each with associated Investigational Medicinal Product (IMP), which are:

Cabozantinib
There is a wide body of pre-clinical evidence supporting the relevance of hepatocyte growth factor (HGF), the ligand for MET, and vascular endothelial growth factor (VEGF) as anti-cancer targets in patients with UC 11 . Cell line data shows that MET is associated with tumour proliferation and growth 12 . VEGF has also been demonstrated to be overexpressed in UC cell lines and associated with tumour proliferation 13 . The combination of VEGF and MET inhibition is therefore an attractive field in patients with UC. Cabozantinib is a multi-kinase inhibitor including MET and VEGF. It has been demonstrated to have significant activity and is licensed in Europe to treat medullary thyroid and renal cancers 14 . It has shown early signs of activity in an on-going phase II trial in patients with advanced UC 15 . There is a currently a lack of data on the expression of MET or VEGF as a biomarker for cabozantinib activity. For this reason, all patients in ATLANTIS and other on-going studies have not selected patients on the basis of MET expression. Patients in this study will therefore potentially be eligible for the cabozantinib/placebo arm if their tumour does not over-express any current ATLANTIS target biomarker or they are not deemed suitable to enter a biomarker-defined arm of the study.

Rucaparib
There is emerging evidence of a subgroup of patients with UC exhibiting a DNA repair deficiency phenotype resulting in defects in a variety of genes including BRCA1/2, BAP1, PALB2, FANCD2 and ERCC2 16,17 . These DNA repair gene defects predict for benefit following cisplatin-based chemotherapy in UC, implying that a switch to maintenance therapy strategy for PARP inhibition after prior chemotherapy may allow for enrichment of This data implies that BRCA mutation and/or HRD associated UC patient subgroups would be suitable for investigation of a PARP inhibitor. Rucaparib is an orally bioavailable small molecule inhibitor of PARP1, PARP2 and PARP3. Non-clinical evaluation of rucaparib has demonstrated potent inhibition of PARP enzymes and sensitivity to BRCA1/2 homozygous mutant cell lines. Patients in ATLANTIS whose tumour is positive for a composite 'HRD biomarker' of alterations to a list of relevant DNA repair genes and/or high percentage genomic loss of heterozygosity (LOH) at pre-screening will potentially be eligible to receive either rucaparib or placebo in double-blind fashion. LOH is a form of genomic alteration, associated with HRD, and is characterised by the loss of one copy of a gene or chromosomal region.

Previous in vivo and in vitro data have demonstrated that the Androgen Receptor (AR) is a
potential anti-cancer target receptor in patients with UC 20,21 . Unpublished data from tissue microarray in patients tested within the LAMB study demonstrated that 30% of UC tumours over-express the AR. This was also associated with poorer prognosis, raising potential clinical benefit of targeted agents against AR. Enzalutamide is a potent androgen receptor antagonist, which unlike earlier generations of anti-androgen therapies has no known agonistic effect on AR and is known to be active in men with prostate cancer who have previously failed conventional androgen deprivation therapy. The drug is well tolerated even with prolonged administration 22 . Patients in ATLANTIS whose tumour is found to over-express the AR on immunohistochemical analysis will potentially be eligible to receive either enzalutamide or placebo in double-blind fashion.

Discussion
Recent studies of second-line therapy, particularly immune checkpoint inhibitors, in patients with UC have demonstrated a clinical benefit for a small proportion of patients.
The emerging treatment landscape of urothelial cancer is also increasingly incorporating these therapies in first line clinical trials. There remains, though, a need to improve outcomes for patients in this challenging disease. ATLANTIS offers a novel approach to previous studies in this arena by offering biomarker-defined selection of maintenance therapy after chemotherapy in patients with UC. The primary research question of this signal-searching multi-arm phase II study is to investigate whether targeted maintenance therapy after chemotherapy can delay the time to progression in molecularly-selected patients with UC. This would therefore establish clinically relevant evidence of activity in molecularly defined subgroups for the novel drug used. The study design will investigate cabozantinib, enzalutamide and rucaparib in this setting, but also provide a generic framework that will allow new treatments to be introduced into the study in future with prospective stratification based around a molecular target.
There are currently no targeted therapies with proven activity in UC. Preclinical data suggests there are a number of eligible targets, but the number of randomised trials performed to test these hypotheses have been small. The reasons behind this remain unclear, but low levels of clinical activity in unselected patients in non-randomised phase II trials do not support progression to phase III trials in unselected patients with UC.
Another explanation is that combination targeted therapy and standard chemotherapy is poorly tolerated in this patient population. ATLANTIS will be a leading global study in the development of personalised targeted therapy for patients with UC.
Such a positive randomised phase II study would be a significant breakthrough in UC and may lead into randomised phase III trials in both the metastatic and adjuvant setting. If a cohort of ATLANTIS is positive, it would be anticipated to lead to a randomised phase III study of maintenance therapy in the appropriately selected UC population. There is also unmet need for adjuvant therapy in patients with high-risk muscle invasive bladder cancer and a positive signal from ATLANTIS would also support a randomised phase III trial in this patient group with appropriate molecular selection.
Patients in ATLANTIS are asked to consent to collection of surplus tissue for translational research. Translational research hypotheses are embedded in the primary purpose of this trial, the aim of which is to demonstrate efficacy of predictive and appropriate drug interventions in UC. As such, all patients must provide adequate tissue for biomarker expression prior to trial participation. This will also provide a bio-resource for future research in UC.
The ATLANTIS trial reflects an exciting innovation in front line precision cancer medicine for patients with advanced urothelial cancer. The study design, with biomarker-negative arm, allows all patients who enter pre-screening to potentially be able to take part. The adaptive design also allows maximal opportunity to detect an efficacy signal and rapid

Consent for publication
Not applicable

Availability of data and material
Not applicable Figure 1 Study design for patients in ATLANTIS trial.