Read the Full Video Transcript

Petros Grivas: Hello, I'm Dr. Petros Grivas. I'm a Medical Oncologist at the Seattle Cancer Care Alliance. I'm an Associate Professor of University of Washington, and Fred Hutchinson Cancer Research Center. I'm very excited today to host Dr. Ali Khaki, who's an Assistant Professor at Stanford University. He was a MNT for our team here at the University of Washington until recently, and he moved a little bit South from here. Ali, welcome.

Ali Khaki: Thank you, Petros, and I'm pleased to be here.

Petros Grivas: Fantastic to talk to you Ali. And it's really amazing, your contributions in the field already have been accumulated and they're robust. So, let's talk a little bit about the recent work that we have published and also presented at ASCO GU 2021. And I will start with a poster that Dr. Makrakis, another Greek, worked on and presented at the ASCO GU under your and my mentorship. And this specifically, I think, is an interesting work, has to do with potential association between different clinical pathologic factors in relation to response to checkpoint inhibitor in patients with advanced urothelial cancer and several demographic factors we think of. Of course, the other big question is the site of metastasis. Does this play a role? Does lymph node only disease, versus liver metastasis, bone mets, does this correlate with response survival with checkpoint inhibitors in advanced disease? Do you want to comment a little bit on that?

Ali Khaki: Yeah. Thanks for asking, we've been very fortunate that over the last couple of years, we've developed this registry of just around 1000 patients with advanced urothelial cancer or metastatic urothelial cancer, who've been treated the immune checkpoint inhibitors. And in the past, we've used that database to ask and answer questions about outcomes related to performance status, as well as variant histology. And this year at GU ASCO, Dr. Makrakis had this poster about where we investigated the outcomes related to metastatic sites. And so naturally, I think that there's a lot of interest in whether different metastatic sites could have different responses to immune checkpoint inhibitors. With some thoughts that things like lymph node disease could have favorable outcomes in sites like liver metastases, which is a known poor prognostic factor from previous prognostic models, like the Belmont model, as being something that might have worse outcomes. But we wanted to systematically investigate this with the registry that we've developed.

Petros Grivas: Why don't you tell us a little bit more about the results of the study, Dr. Khaki?

Ali Khaki: Okay. Yeah. The way we designed the study is, we wanted to test a few different models based on our hypotheses. And so we developed four separate models looking at metastatic sites in different ways. And so the first model, we classified patients as having lymph node only disease, or having other disease. And there we saw that patients who had lymph node only disease had better response rates, longer progression free survival, and longer overall survival, compared to people who had other non lymph node disease.

In a second model, we were interested in looking at bone mets, liver metastases, or the combination. And so there we developed a four factor model where we looked, we classify patients as having neither bone or liver, having bone but not liver, having liver but not bone, or having the combination of both bone and liver. And what we saw there, is that patients who had neither bone nor liver, they did the best. Patients who had bone or liver had intermediate outcomes. And to give you a sense, neither bone or liver had response rates around 30%. Bone only was 17%, and liver only was 22%. And then the combination of bone and liver metastases had a response rate of a 9%.

And similarly, when you look at overall survival is 14 months for neither bone or liver, eight months for bone only, five months for liver only, and two months for patients who have bone and liver metastases. We were hypothesizing that liver only would be a poor prognostic factor. But we also know that bone only, and in the combination of bone and liver, seem to portend worse prognosis.
I think those were the most significant findings. We also interrogated the data in a couple of other ways, looking at classifying patients as having visceral versus non visceral, as well as having a more complex six factor model. But I think that my two take homes were that lymph node only disease seems to be a favorable risk factor, in terms of response rates and survival, with immune checkpoint inhibitors, and then bone metastases, liver metastases, and the combination of the two, portend worst prognosis.

Petros Grivas: I think this is relevant data, and this is real world practice. As you mentioned, more than a thousand patients from Europe and the US, 25 centers. I know you and me work hard to build that base, and it's great to see the results of that. And as you said, it seems that patients with lymph node only disease appear to do well, as we know from clinical trials. Relatively well with checkpoint inhibitors in advanced disease, but those with liver and bone mets, especially the combination, have very poor outcomes, illustrating the need, unmet need, for new therapies. And now we have more options in addition to chemotherapy immunotherapy, we have a form of adult in, and we covered afatinib in some patients, and new agents coming up. Maybe just do some sacituzumab govitecan down the road.

Ali Khaki: Yeah. And it was exciting at GU ASCO to see the EV-301 data that cemented that as a good third line option, after platinum chemotherapy and checkpoint inhibitors.

Petros Grivas: I agree with you, definitely highlight of ASCO GU was the EV-301, if you do a cohort two. The other poster that Dr. Makrakis presented answered, tried to answer, a different question. But irrelevant to that, patients with advanced urothelial cancer who received checkpoint inhibitor, does it matter if they have the primary tumor in place, or not? Does the history of prior resection of the primary tumor, radical surgery, Radical nephrectomy, or radical cystectomy, does it influence, does it correlate with the response and survival with immunotherapy in advanced disease? Can you comment on this?

Ali Khaki: Yeah. This was a very interesting and exciting study. In GU cancers more broadly, there's a long history of the role of removal of the primary tumor, for example, in renal cell carcinoma, in portending better outcomes. But, with immune checkpoint inhibitors, you could hypothesize that maybe having your primary tumor in place may present novel antigens that would help immune response. It might actually be beneficial to have in place. So this was a really unmet, unanswered question within the literature. There are some small retrospective series that suggest that prior surgery is beneficial, but in general, there's not much data on this. We're excited to use our database to ask this question.

And what we did is we, again, used the same registry that I mentioned before. And we classified patients as having a prior history of radical surgery or not. Keep in mind that everyone in our database had other metastatic disease, or advanced urothelial cancer that was locally advanced and unresectable. And so none of these patients would have had the surgery in the setting of immune checkpoint inhibitors at all, haven't had a history of it in the past.

And then again, we compared outcomes in overall response rate, progression-free survival overall, survival between the populations, history of surgery or not. And here we saw that patients who had a history of prior surgery had similar outcomes to those who did not have a history. So the response rates, progression-free survival and overall survival in general, were not different between the two populations.

Petros Grivas: I think this is interesting because as you mentioned in other tumor types, have been metastatic kidney cancer, it seems that resection of the primary tumor may be correlating with better outcomes. But it's interesting to see this analysis. And one of the first studies, I've actually looked at this specific question.
Another question that another mentee of us presented at ASCO GU, Dr. Esagian, another Greek, I would say, he did a great job with presenting data answering the following question. Does the location of the primary tumor impact or is correlated with response and survival, with immune sequence in advanced disease. So lower tract bladder, for example, versus upper tract disease. Is there any difference there? And it appears that there is no significant difference. Could you comment a little bit more?

Ali Khaki: Yes. As you stated, this is the next study that we looked at, it was presented at GU ASCO, and actually the publication has already been published for this in the British journal, BJUI. And what we looked at here was, similar to the other studies, we used the same database, and our exposure of interest here was the site of primary disease. And we classified upper tract as being ureter or renal pelvis, and lower tract as being bladder or urethra. There's very few urethral cases. And we compared the same outcomes, overall response rate, progression-free survival, and overall survival between the two populations. And what we found here is that in general, the outcomes were not different between the two populations.

The way we model this is, we want to listen three different ways. We modeled it as a full total population, and then we did two stratified analysis. One was based on treatment line, first-line or salvage therapy. And then a second was based on histology, pure histology, or mixed histology.

And the one thing that was provocative was that patients who had mixed histology urothelial cancer, those with upper tract mixed histology seemed to have worse outcomes, compared to those with lower tract mixed histology. This was the one significant finding in this analysis. And that was their overall response rate, as well as progression-free survival, was worst in those, in lower overall response rates, shorter progression-free survival, when those who have mixed histology upper tract, compared to those with mixed histology lower tract. The overall survival was also numerically shorter in those with mixed histology upper tract, but it was not statistically significant.

This is interesting because the data before, in clinical trials and otherwise, has been quite mixed for upper tract versus lower tract urothelial cancer. There's data from IMvigor210, the phase two of atezolizumab. And this is by an ineligible arm where the response rates with atezolizumab were high as 39% in upper tract disease. But in the other Phase two trials in immune checkpoint inhibitor space, it's ranged from about 11 to 22%.

And there's also the studies that have looked at the biology between the two different types of cancers that have been also mixed, where there are multiple studies that looked at the genomic landscape of upper tract disease and showed that they tend to have a lower tumor mutational burden, may have a T-cell depleted microenvironment, compared to lower tract disease. But at the same time, upper tract disease has a stronger association with Lynch syndrome and mismatch repair deficiency.

So there's been mixed data about whether these people, the patients, do better or worse with immune checkpoint inhibitors. And in our study, generally they did about the same, but there are these subpopulations that might do worse. And so it just underscores the need for more prognostic predictive biomarkers to help differentiate this further, because it seems like the heterogeneous population.
Petros Grivas: And talking about prognostic biomarkers. And by the way, you already point out this paper with Dr. Esagian, and our group is published at the BJUI that came up at the same time with the ASCO GU.

But speaking about prognostic biomarkers, you led the effort with the work we did to identify a new prognostic nomogram, a new prognostic model, in patients who receive immune checkpoint inhibitors in advanced urothelial cancer from that same database. It's really exciting because for years, we definitely rely on the shoulders of the giants right, as we say, Dr. Bajorin, with the Bajorin risk factors, and Dr. Bellmunt, with the Bellmunt risk factors. Now we have a new model and I think it's very exciting. Can you tell us a bit more about it?

Ali Khaki: This is some of the most exciting work that I was able to do as a fellow, so I'm happy to talk about this further. Using the same registry that we've been using, what we did here was, we set out to develop a new prognostic risk score in the immune checkpoint inhibitor era. The majority and dominant risk factors have all been developed in the setting of chemotherapy, before immune checkpoint inhibitors. So we hypothesized that there might be different prognostic factors that are important with immune checkpoint inhibitors. And so we had identified 18 co-variants that were all collected within our database.

And we did a stepwise approach of testing, with the interest being overall survival. So what we did is, we tested each of these 18 co-variants initially in a univariable Cox regression. And then we combined those with the strong relationship into a multi-variable model, to identify the factors that had the strongest association with overall survival.
We tried to do this for both the first line, as well as the salvage setting. We weren't able to find a strong model for the salvage setting, but found a good new model for the first line setting for immune checkpoint inhibitors.

The four factors we identified to have the strongest association with overall survival was ECOG performance status of two or greater, neutrophil to lymphocyte ratio greater than five, albumin, less than three and a half, and the presence of liver metastasis. They're all associated with shorter overall survival.

And so we assigned each one of these factors a score of one point, to come up with a new four point risk score. And then, to assess how good this risk score was, we calculated the C statistic. So the C statistic basically, is an estimate of how good your prognostic modeling is predicting the outcome of interest. And it ranges from 0.5 to one. Whereas, 0.5 is no better than a coin flip, and one is 100%, a perfect prediction. And our C statistic was 0.68, which is considered about moderate in its prognostic ability. And so we thought this was a pretty good outcome for our risk score.

Petros Grivas: It's really exciting to see this work coming out in European urologic oncology. It was published just a few weeks ago, and it's exciting to see, this massive amount of work. I think it can have implications, especially when we talk, in the patients, in the clinic about potential prognostic estimates, and also when we design clinical trials for example, in single-arm, phase two study, what is the estimated response rate, the effects or overall survival in patients getting checkpoint inhibitor? And if you want to use combinations, and you don't have a randomized cohort, you can use this data to guess estimate the outcome of those patients.

So congratulations, Ali. I can tell you, it has been such a great pleasure working with you in the last three years. And we look forward for more work together and more collaborations.

Ali Khaki: Thank you, Petros. It's been a pleasure to work with you, have you as a mentor, it's been such an honor, and I look forward to continue to do more, to better understand what's going on with bladder cancer in the field.

Petros Grivas: Same here. And now, you're already crossed in the side of another mentor, and you already have been mentoring junior colleagues, and will keep doing that, and try to contribute major advances in the fields, and help the patients and their families. Thank you so much today for your time, and thank the audience.

Ali Khaki: Thank you. Thank you. UroToday for allowing us to do this.