Analyzing Various Molecular Subtypes in Bladder Cancer

Roland Seiler, MD

As part of our coverage of the ASCO-sponsored 2017 Genitourinary Cancers Symposium being held February 16th to February 18th in Orlando, Florida, we are today speaking with Roland Seiler, MD, a clinician and researcher in the department of urology at the University Hospital of Bern in Switzerland. Dr. Seiler presented an analysis of how different bladder cancer molecular subtypes respond to neoadjuvant chemotherapy at the conference.

-Interviewed by Anna Azvolinsky, PhD

OncoTherapy Network: First, is there currently a standard of care for neoadjuvant therapy in muscle-invasive bladder cancer? And who are the patients that should receive neoadjuvant chemotherapy?

Dr. Seiler: So there is actually a standard of care that all patients with muscle-invasive bladder cancer that can receive a cisplatin-based regimen should be offered neoadjuvant chemotherapy for at least 3 cycles. The main drawback of this regimen is that only about 40% of patients show some sort of response, 60% of the patients still have muscle-invasive disease after this treatment and the survival benefit is limited, only a 5% to 7% improvement in 5-year overall survival. These are the main drawbacks and also why this treatment has not generally been accepted by all urologists and oncologists.

OncoTherapy Network: What do we know about the molecular subtypes of bladder cancer? How many are there or do we think there are? And besides distinct molecular features, do these subtypes have distinct clinical features as well?

Dr. Seiler: So, maybe about 6 years ago we started thinking about molecular characterization of bladder cancer and about 4 or 5 years ago, four different groups profiled muscle-invasive bladder cancers and analyzed gene expression and they saw, in four different independent datasets that there are actually molecular subtypes of bladder cancers that can be characterized into specific groups. These four different methods, they aligned with each other, so at a higher level, they showed the same characteristics, but then there were nuances that differed from one method to the other. The general idea of molecular subtype came about when people analyzed the biological characteristics of the epithelial layer. For example, investigating the basal part of the normal epithelium, you find characteristics of stemness, usually higher proliferation rate, expression of keratin 14 and keratin 5/6 and maybe not surprising, but there are bladder cancers that share these biological characteristics by having a higher rate of stemness, a higher proliferation rate.

Beside the molecular and biologic characteristics, there are actually clinical characteristics of these subtypes. So it is know that in a patient that undergoes a cystectomy without any chemotherapy with basal tumors, they usually have an inferior prognosis compared to patients with who undergo a cystectomy with luminal tumors only. That has been shown in different independent datasets and also different subtyping methods.

OncoTherapy Network: Can you tell us about the design of the study that you presented?

Dr. Seiler: So the idea to use molecular subtyping to predict molecular response to treatment is actually not new. The MD Anderson group, which used a subtyping method, already thinking that these molecular subtypes likely respond differently to cisplatin-based neoadjuvant chemotherapy. However, the datasets were of rather limited samples sites that didn’t allow for any final conclusions. So what we did is basically collect two of the largest cohorts to date of patients that received neoadjuvant chemotherapy and we profiled the tumor biopsies prior to neodjuvant chemotherapy and profiles gene expression analysis and used the expression data to assign these samples to these different molecular subtypes. And based on our findings we came up with a consensus of all of these four subtyping methods and grouped them into four subtypes that we called claudin-low, basal, luminal-infiltrated, and luminal. And we moved to investigate the biological characteristics of these labels, if you will, to train a classifier that you could apply to a single patient sample and that could be used for clinical daily practice.

OncoTherapy Network: Can you talk about the results of the study or what did it show?

Dr. Seiler: So, we could see that, based on the sizes of the datasets we currently have, that we could move these four molecular subtypes to a classifier that could be applicable to single patient sample. And we characterized the luminal and basal tumors on a higher level, but then within these two groups, we selected tumors that showed a kind of more intense interaction with the tumor microenvironment. These were the claudin-low tumors in the basal group and the luminal-infiltrated tumors in the luminal group. So finally we ended up having four classes and trained our classifier on that.

Then we used two independent cohorts to show that our classifier performs well when going through a single sample, so this is a test that could be applicable to a single patient. Also, we investigated the clinical characteristics of the subtypes that our classifier predicts and one of the findings was that the rate of pathological downstaging due to neoadjuvant chemotherapy. We could see that there was an enrichment for more advanced primary tumors at cystectomy particularly in claudin-low tumors and luminal-infiltrated tumors. So tumors that have a higher interaction with the microenvironment and that we think are more aggressive and more advanced tumors at cystectomy.

One of the drawbacks of our study is that the rate, the response of chemotherapy which is defined as absence of muscle-invasive disease at cystectomy was not related to one of our molecular subtypes. So the molecular subtypes that we predicted failed to be related to response to neoadjuvant chemotherapy. But then we further looked at patient outcomes and an interesting finding of our study is that we compared in the dataset from the public domain of patients that have muscle-invasive bladder cancer that did not receive chemotherapy compared to our dataset and we can see in the chemotherapy-free dataset what we would expect: that patients with luminal tumors have a better prognosis than those with basal tumors. And then we compared this outcome of the patients to our chemotherapy group, about 270 patients, and we could see that when receiving cisplatin-based chemotherapy, patients with basal tumors showed a dramatic improvement of prognosis when compared to patients with basal tumors that only underwent cystectomy. As for the 3 other subtypes, there was not a change in prognosis when treated with chemotherapy or not. However, the improvement of prognosis in patients with basal tumors was dramatic and impressive.

OncoTherapy Network: Thank you so much for joining us today, Dr. Seiler.

Dr. Seiler: Thank you.