SWOG 1314: A study of co-expression extrapolation (COXEN) with neoadjuvant chemotherapy for muscle-invasive bladder cancer

Bookmark and Share
Published: 10 Jun 2019
Views: 2020
Prof Thomas Flaig - University of Colorado Denver, Denver, USA

Prof Thomas Flaig talks to ecancer at the 2019 American Society of Clinical Oncology (ASCO) Annual Meeting about the randomised trial that examined chemotherapy regimens in muscle-invasive bladder cancer to assess the use of statistical biomarkers to predict which patients would respond to chemotherapy.

He describes the methodology behind this trial and the types of biomarkers that were used, along with their effectiveness. These included a biomarker for gemcitabine-cisplatin and one for MVAC.

Prof Flaig discusses future research opportunities, which include refining the use of the COXEN biomarker and correlating genetic information with this.

ecancer's filming has been kindly supported by Amgen through the ecancer Global Foundation. ecancer is editorially independent and there is no influence over content.


We’re delighted to present at this meeting SWOG 1314. This is a randomised study looking at two chemotherapy regimens in muscle invasive bladder cancer so it represents one of the only prospective datasets in this population comparing two chemotherapies. It was designed to study a statistical biomarker approach called COXEN to see if we could predict which patients are going respond to chemotherapy.

What sort of methods did you use?

COXEN is a mathematical design and interestingly enough it starts with in vitro data, so cell cultured data, which is humanised by correlating with human samples. So essentially you pull out from this in vitro cell culture data the relevant human markers then that signature, that gene expression signature, can be applied to an individual patient’s tumour to say will this patient respond to these chemotherapy agents.
This represents an important study for the bladder cancer community. It was a federally sponsored trial by the NCI and enrolled over 230 patients. There were two biomarkers we looked at, one was for gemcitabine cisplatinum, the other was for a chemotherapy called MVAC. The MVAC biomarker didn’t perform very well, the gemcitabine cisplatinum marker, for its individual arm, showed separation which did not quite meet statistical significance but if we put all the patients together treated with both regimens it had a p value of 0.02 in selecting those patients that would respond to chemotherapy with a pathologic response at the time of surgery.

I think there’s a lot of potential end facts from this work. One is the COXEN biomarker which we were primarily endeavouring to look at. This will give us a lot more opportunity now to take this data, refine it, perhaps at different thresholds, perhaps look at different ways of using that particular approach for biomarkers. In addition, again this represents one of the only prospective randomised trials in muscle invasive bladder cancer so an important set. We’re going to look at DNA expression, microRNA, we’re going to look at SNPs and a variety of other correlative biomarkers are in play. The hope from all of this is we can generate enough data to direct therapy for patients in the future.

How important is the use of a biomarker, particularly for this type of patients?

It’s interesting, for bladder cancer, muscle invasive, before patients go into surgery there are two acceptable regimens. There’s gemcitabine cisplatinum and dose dense MVAC. Right now providers basically just pick one and if you ask them why there wouldn’t be necessarily a systematic way that they would pick one over another. The hopes of this biomarker, and really all the other things we are doing, is can we have an approach where there is a thoughtful customised way of saying this patient should get chemotherapy regimen A and this patients gets chemotherapy B. That’s the real hope of this. It’s a step forward, we need to continue to work.