Pedro Barata: Hello, and thank you for taking the time to be here with us today. My name is Pedro Barata. I'm a GU oncologist out of University Hospital, Simon Cancer Center in Cleveland, Ohio. Today we're going to have an international conversation, a very exciting one in my opinion. I have with me today Dr. Michele Maffezzoli. He's a medical oncologist affiliated with the University of Parma in Italy, and also spending extra training as a clinical research fellow at Portsmouth University in the UK and did amazing work around the GRANT score. We'll talk a little bit about that today. Michele, thank you so much for joining us.

Michele Maffezzoli: Thank you. Thank you for the nice introduction and nice to meet you and thank you for this opportunity to explore further our work.

Pedro Barata: Of course. And I know you have a couple of slides to help us go through this.

Michele Maffezzoli: Yeah, I can-

Pedro Barata: While you pull them up, just to say, obviously your work and your paper got a lot of buzz, I think very fair to say that. So some folks are already aware of that. We're going to be talking a little bit about papillary RCC. You tell us how you validated a GRANT score. So please take a second and pull up the slides and tell us, walk us through that.

Michele Maffezzoli: So I would like to share with you our work on the GRANT score. This is a paper we developed together and published on technology and cancer research and treatment. The title of the paper is the External Validation of the Grade, Age, Nodes and Tumor, the GRANT Score for Patients with Surgically Treated Papillary Renal Cell Carcinoma. So this is a topic which is usually forgotten in our basic clinical practice. So prognostic models for surgically treated RCC. But these models are very important for us as a clinicians for patient counseling, individualizing surveillance, especially for a patient with high risk or low risk subgroups, and also identify potential candidates for adjuvant treatment. For example, the keynote prognostic model is well recognized to select patient for adjuvant immunotherapy, but for papillary renal cell carcinoma, this is more challenging. This is because most of the prognostic model existing are derived from studies which are largely including patient with clear cell histology, which has a different prognosis compared to the papillary one.

And also the relative infrequency of the papillary RCC complicates the creation of a specific stratification tool. So in this setting, the GRANT score is one of the models suggested by the European guidelines to predict the prognosis of surgically treated papillary RCC together also with the Leibovich score and with the VENUSS. And the aim of the present study was just to validate the GRANT score in this large cohort of papillary renal cell carcinoma patient using a three group risk stratification. So different, a little bit different from the original validation, which was a five group risk and to explore the accuracy in predicting cancer specific survival. So we collected clinical and pathological data from patients with papillary renal cell carcinoma who underwent radical and partial nephrectomy from 1984 to 2015. We assigned according to the GRANT score, one point for each of the risk factor that you can see in the table.

So grade major than two, age major than 60 years, T stage major than 3b, and node positive. And according to this score, we've stratified the patient into risk groups based on this score. So we calculated the cancer specific survival and see index to measure the accuracy of the model. And also since the proportional assumptions was violated, we used the restricted mean survival time measured up to 10 years to better assess the differences between groups. So these are the results. So [inaudible 00:04:38] almost 1,942 patients in total. Medium follow up time was 64.6 months. Most of the patients belong to the group one, which is the favorable one, 81% of patient. Group two, 14% of patient. And group three, just 4% of patients. At 60 months, the cancer specific survival rates were significantly different between the three risk groups. So it was 93% in the group one, the favorable one, 60% in the group two and 26% in the group three.

Similar differences were seen also in the restricted mean survival time up to 10 months, which was almost more than nine years in the group one, 76 months in the group two, and 56 months for the group three. That's significant, and there was a statistically significant difference between these two, these groups. The C index of the model was 0.73, which is a good value actually. So suggesting a good accuracy of the model and it is quite consistent with the previous validation and also with the validated models. So based on the results we validated further the GRANT score. The GRANT score was originally developed in the ASSURE trial cohort and then externally validated huge cohort of patient within the SEER database. This is a further external validation specific for papillary renal cell carcinoma patient. The GRANT score effectively stratified this patient into these three risk groups, demonstrating a good accuracy and predicted the cancer-specific survival quite consistent with the other validated models as the VENUSS or the Leibovich.

Another thing which is very important for these studies is that is very easy to use, which is very crucial when we talk about our clinical practice because which the GRANT score in this case is able to combine a good accuracy and, of course, easiness of use, which is one of the future important to effectively apply in clinical practice prognostic model. Of course this score was validated upon retrospective data, so future model should be prospectively validated. And we are also working with the ECOG-ACRIN group and Naomi Haas and Professor Michael Carducci to validate this model within the PROSPER trial in a new cohort of patient treated with perioperative nivolumab. But yes, we believe that future models should also incorporate further information apart from the pathological and classical, traditional clinical information. So molecular and genetic information like the ctDNA would be important in the future. And so that's it. So in conclusion, the GRANT score is a good prognostic model validated for papillary RCC and it can combine easiness of use and accuracy to be effectively applied in clinical practice. Thank you.

Pedro Barata: Michele, it was amazing presentation. Congratulations. Very good work.

Michele Maffezzoli: Thank you.

Pedro Barata: Super important. And so it's good to see that you're basically showing us this with the C index of 0.73, actually, I think it compares in a very strong way, I guess with the validation of scores that we use in clear cell. So I think it has a good performance to predict what patients with papillary RCC are going to do over time, as you said. So it's very important work and congratulations for that.

Michele Maffezzoli: Thank you.

Pedro Barata: I guess as you presented this and sounds like you already give us a few ideas where you might go trying to see the performance of the score in other important data sets like PROSPER and others. And so the question that I have for you is, one comment I'm going to make is people in clinical practice, a lot of folks like you and I, I mean, a lot of folks don't use prognostic scores in clinical practice for an unclear cell because they would say, "Well, whatever happens happens and there's no adjuvant treatment approved as of yet, as of now." And so the question is, Michele, where do we think we can go to change that, to apply, as you said, easy to apply because these factors are available to all of us to start using these more. How do you think we can implement that in clinical practice today? Does that change the schedule of your scans when we are monitoring these patients? Or what do you think is going to be applied today before getting maybe an adjuvant treatment in the near future?

Michele Maffezzoli: So yes, this is a very challenging question actually because yes, we have these models that can be easily applied in the clinical practice, but actually if you have a low-risk patient, you know that the survival would be long. If you have an high-risk patient, you know that it can really relapse in a short time. So the challenging would be for the intermediate risk, especially for papillary carcinoma where we don't have actually an adjuvant treatment. So in the future, I think that the key would be to implement the information we have for this patient, so hopefully, KIM-1 is one of the promising biomarker for disease relapse. ctDNA is one is really a good information to have.

Still challenging a little bit because most of the patients are ctDNA negative after surgery. But I think this would be the case. So first of all, having more information, more information about possible biomarkers in the sample, which are easy to collect and can be combined within a prognostic score. And secondly of course we have to test some drug in the adjuvant setting also for this cohort of patient because actually papillary RCC are 10, 15% of all RCC. So they are uncommon but not that uncommon. So we need further trial to assess also adjuvant treatment for these patients.

Pedro Barata: No, that's perfect, Michele. And I agree with you. That was actually my next question. Do you have plans to add molecular information to the score perhaps and maybe get it even better? And you mentioned KIM-1, ctDNA, I don't know if you think it's feasible to add things like RNA six gene expression signature or any DNA alterations, for example, to that to kind of come up. Or even AI, because you did have on your last slide, the information about AI. I am wondering if digitized pathology on these tumors would allow us to get even better. Do you have any thoughts on that?

Michele Maffezzoli: Well, I completely agree with you. So even the machine learning would be a prospective future for these prognostic models in patients. We are unfortunately, this database, the data in this data set were quite limited, so we couldn't further validate, we couldn't further collect other biomarkers. We are actually working on different biomarkers, especially in the cholesterol metabolism, also in the different blood values like the Hb/RDW ratio as inflammatory index of the tumor presence. So we are trying also to collect this data in surgically treated patient to further implement this prognostic score, which are less expensive actually than the ctDNA. But yes, we are collecting these blood-based markers together also with the KIM-1.

Pedro Barata: Got it. No, that's fantastic. And at least while you're talking, I was thinking what studies are testing therapies in adjuvant for unclear cell, including papillary or specifically for papillary RCC. And I think there's a cooperative group effort that is working to that with [inaudible 00:14:38]. And then I think there's an agent study also with immunotherapy for papillary RCC. That's the last time I checked into that. So maybe there'll be opportunities to work together in the future. Michelle, thank you so much for taking the time. This is amazing. Again, congratulations. Very good work. And I'm glad we had the chance to sit down with you and tell us more about it and give us the details. So I'm looking forward to hear more about updates on this and other exciting projects. And so thank you for taking the time.

Michele Maffezzoli: Thank you. Thank you very much for your time and for this opportunity again.