What is the clinical need that led to exploring ctDNA-guided, response-adapted bladder preservation in muscle-invasive bladder cancer?

As you know, muscle-invasive bladder cancer has typically been treated with neoadjuvant regimens which could be chemo, chemo/IO and now maybe EV/pembro followed by radical cystectomy or chemoradiation. But radical cystectomy is associated with a lot of comorbidity and it affects the quality of life of patients and patients are always looking for other options to see if they could preserve their bladder.

Now, we know that about 35-60% of patients after undergoing neoadjuvant therapy they have a pathologic complete response, which raises the question whether a select group of patients could potentially avoid cystectomy. But we know historically that our assessment of response to neoadjuvant therapy has been sub-optimal. So typically we use a CT scan, cystoscopic assessments and biopsies but those have oftentimes missed residual disease in the bladder. So there has been a need to incorporate biomarkers that would help us to better predict which patients will be able keep their bladder intact safely.

RETAIN-1 and RETAIN-2 are trials essentially incorporating such an approach where patients receive neoadjuvant therapy and after completing neoadjuvant therapy they get assessment of their bladders. Patients also have their pretreatment tumour tissue sent for mutations. So the mutations that were tested were essentially mutations in genes like ATM, RB1 and ERCC2 which have been shown to correlate with cisplatin response. As part of these trials patients who have one of the mutations and have a complete response in their bladders are offered active surveillance. Patients who either don’t have the mutation or have residual disease in the bladder, they go on to receiving an intervention.

So we have the mature results for both of these trials available and we recently did an analysis of ctDNA that was collected in both the RETAIN-1 and RETAIN-2 trials, both at baseline and post-neoadjuvant therapy, to predict which patients could potentially… how informative was ctDNA in combination with the clinical staging to make appropriate decisions for a patient.

How did ctDNA perform as a biomarker for identifying patients who could safely avoid or delay radical cystectomy?

We found that ctDNA is extremely prognostic in terms of metastasis free survival and overall survival, both at baseline and post-neoadjuvant therapy. So patients who are ctDNA positive at baseline or post-neoadjuvant therapy have about an 11-fold higher risk of having metastasis compared to patients who are ctDNA negative. We also found that patients who are persistently ctDNA negative, so ctDNA negative at baseline and post-neoadjuvant therapy, they do really well. Patients who clear ctDNA, so they were positive at baseline and then became negative, they do a little bit worse than that, and patients who are persistently positive do much worse than that.

So this was one of the main findings of our study but then that led to another question about how could ctDNA be used in terms of bladder preservation and treatment decision-making after neoadjuvant therapy. So we stratified metastasis free survival in our patients based on the treatment options – cystectomy versus active surveillance and their post-neoadjuvant therapy ctDNA status. What we found is that patients who, again, were ctDNA positive who underwent cystectomy did really poorly but patients who were ctDNA negative and underwent cystectomy had very similar outcomes to patients who were ctDNA negative and went on to active surveillance. Which raises the question of whether these patients could have potentially also avoided cystectomy.

Then another key finding of our study was the fact that ctDNA is not so good at predicting local recurrence or residual disease in the bladder. So, among the patients who had cystectomy about 67% of clinical, so ypT2, disease were ctDNA negative. So the negative predictive value of ctDNA predicting ypT0 was only 32%, however, the positive predictive value was quite high.

Similarly in patients who went on to active surveillance, what we found is that among the patients who had local recurrence in their bladder, so among the 40 patients who went on to active surveillance in our study we had very few patients who had ctDNA positivity at the time of local recurrence. So this drives home the point that ctDNA is an extremely good biomarker for predicting systemic recurrence but not local recurrence.

How might ctDNA-guided strategies change the standard treatment approach for muscle-invasive bladder cancer?

I do think that ctDNA will need to be assessed in combination with clinical staging and not ctDNA by itself. In combination with clinical assessment of the bladder, patients who are ctDNA positive we should really be thinking about whether cystectomy is the right thing for them. One may consider maybe intensifying their systemic regimen or maybe offering them less radical strategies like chemoradiation as opposed to cystectomy. So that’s one important consideration.

Then the next message that I will give is exploratory, based off of this study. But patients who are ctDNA negative, if they have no residual disease in their bladder, they could potentially be offered active surveillance with possibly having maintenance immunotherapy also available to them.

What are the next steps before this approach can be widely adopted in clinical practice?

I would say that ctDNA would need to be assessed in prospective trials where it’s used as a decisional point to make any definitive results or conclusions about how ctDNA can be incorporated in these settings. However, we have seen a lot of data from various phase II trials, including the HCRN 16-257 trial for which ctDNA results were recently also published which showed very similar findings, as well as in NIAGARA we have the data. They’re all quite concordant and so I do think that this strategy will probably become applicable in clinical practice in the near future.

Is there anything else you would like to add?

I do think that now with the incorporation of newer strategies like possibly enfortumab vedotin and pembrolizumab may become an available option in the near future, I still think that the ctDNA information that we have learned from the RETAIN trials will be applicable to those strategies because between RETAIN-1 and RETAIN-2, despite the use of different neoadjuvant regimens, we thought that the results were very similar. So I’m hoping that it will even be applicable in future studies.