Christopher Wallis:  Hello, and thank you for joining us for this UroToday Journal Club.  Today we are discussing the recently published DANUBE trial examining durvalumab alone and durvalumab plus tremelimumab versus chemotherapy in previously untreated patients with unresectable, locally advanced, or metastatic urothelial carcinoma, a phase 3 trial. This is the citation of this recently published paper led by Dr. Powles and colleagues. So by way of background, for patients who are eligible, the current standard of care is first-line treatment with a combination of platinum-based chemotherapy for patients with advanced and metastatic urothelial carcinoma.

However, with this approach, disease progression is common and the median overall survival remains less than a year and a half unless there's interest in alternative regimes. For patients who are cisplatin-ineligible, immune checkpoint inhibitors have shown benefit in terms of objective response rate based on single-arm, phase II studies. Additionally, more recently, two trials showed the benefit of using anti-PD-L1 agents in first-line therapy for metastatic urothelial carcinoma. In the IMvigor130 trial, a combination of atezolizumab plus platinum-based chemotherapy improved progression-free survival compared to chemotherapy alone. In the JAVELIN Bladder 100 trial, the use of avelumab maintenance following chemotherapy improved overall survival compared to the best standard of care.

And so it's important to note that in each of these trials, the treatment approach relies on the chemotherapeutic backbone with the augmentation using immunotherapy approaches. However, durvalumab has been approved for platinum-refractory disease based on data from open-label phase one and phase two trials. Tremelimumab has also shown activity in patients who have advanced bladder cancer with progression following chemotherapy. This trial sought to examine the use of these two agents in the upfront setting. This is an overview of the study schema of the DANUBE trial, this is an open-label multi-center phase III trial enrolling in 23 countries and 224 centers.

So, patients with untreated unresectable locally advanced or metastatic urothelial carcinoma were randomized into one of three arms; durvalumab monotherapy, durvalumab plus tremelimumab, or standard of care chemotherapy, which is gemcitabine with cisplatin or carboplatin on the basis of the patient's renal function. There were a number of endpoints that were examined, but two were key co-primary endpoints, which was the examination of overall survival and the comparison of durvalumab versus chemotherapy along with PD-L1 high patients, and the comparison of durvalumab plus tremelimumab versus chemotherapy among the entire intention to treat population.

So we will dig into this in a little more detail.  These are more specifically the inclusion criteria.  This was adult patients with histologically or cytologically confirmed urothelial carcinoma of either the renal pelvis, ureter, or bladder, and these patients had to have either unresectable locally advanced disease or metastatic disease, and they could not have received previous first-line chemotherapy.  In order to monitor response, they had to have measurable lesions, per RECIST criteria, and a life expectancy of greater than 12 weeks with good performance status. In order to allow stratification, they had to have tissue available for PD-L1 testing as well as adequate marrow, liver, and renal function to allow for use of these agents.

Patients were excluded if they had previous systemic immunotherapy, have had radiotherapy in the last month, autoimmune or inflammatory disorders, or hepatitis B, C, or HIV, which would preclude the use of immunotherapy approaches.  Patients were further excluded if they had uncontrolled intercurrent illnesses or symptomatic and untreated brain metastasis. Finally, patients who had received adjuvant or neoadjuvant treatment for locally advanced disease with progression within the last six months of therapy were also excluded.

So patients were randomized in one-to-one-to-one randomization with stratification according to cisplatin-eligibility, PD-L1 status, and the presence of liver or lung metastasis which was done by either or both versus neither. And we are taking an open-label approach with the three treatment approaches as described previously. Dose reductions for the immunotherapy agents were not allowed, interruptions were allowed with delays for management of both IO related and non-IO related adverse events. There was trial mandated assessment with imaging every eight weeks until progression in full.  When there was evidence of progression, confirmation is required within four weeks. Biomarker assessment, as we described follows PD-L1 expression and so this was assessed using the Ventana assay with a high PD-L1 defined in three ways as detailed here.

As we alluded to a variety of endpoints with two co-primary endpoints, each being overall survival among subgroup comparisons of the three different trial agents. Key secondary endpoints included overall survival, progression-free survival, time from randomization to the second progression, objective response rate, and duration of response. The authors sought to enroll 1005 patients in a one-to-one-to-one ratio expecting that nearly two-thirds would have high PD-L1 status. They used a multiple testing procedure with an alpha-exhaustive recycling strategy. In this, they split their alpha for their co-primary endpoints with 0.035 allocated to durvalumab versus chemotherapy in the high PD-L1 subgroup, and the remaining 0.015 allocated to overall survival comparison in the durvalumab plus tremelimumab versus chemotherapy comparison among the intention to treat population.  For the hypothesis that was rejected, the alpha was recycled on to further comparisons.

They are presented at two-point interim analyses.  The first assessing overall response rate and duration of response in the cisplatin-ineligible subgroup, the second assessing overall response when 80% of the data for the primary endpoint was available. They undertook a number of prespecified subgroup analyses including stratification factors of cisplatin eligibility, PD-L1 status, and liver or lung metastases.  As well, according to demographics, other clinical factors, disease characteristics, and prior treatments including the use of BCG for nonmuscle invasive disease and prior use of adjuvant or neoadjuvant systemic therapy.

I will now pass it over to Dr. Klaassen to take us through the results and implications of this trial.

Zachary Klaassen:  Thanks, Chris. So there were 1,378 patients that were screened and ultimately 1,032 that were randomly assigned.  As Chris previously mentioned, this was one-to-one-to-one randomization and which is essentially equal between the three treatment groups, 346 for durvalumab monotherapy, 342 for durvalumab plus tremelimumab therapy, and 344 for the standard of care chemotherapy. So this is the baseline characteristics, as you can see, it's a big table.  I split it into two sides. Looking on the left, the column on the far left is the variables, next is the durvalumab monotherapy, to the right of that is durvalumab plus tremelimumab, and the far right is chemotherapy. So between these three groups, as you can see, pretty standard for bladder cancer patients.  The median age is 67 to 68 years of age, the majority of these patients were not surprisingly, males, nearly three quarters.  Also not surprising, the majority of these patients were caucasian or white at about three quarters amongst the three groups.

And if you go down to the primary tumor site, about 75 to 80% had bladder cancer with about 12 to 16% having renal pelvis cancer.  A majority of these patients were metastatic at more than 90% of all the patients included, and most commonly, 80% roughly had visceral metastasis. Moving over to the right, you can see that even for the bladder cancer population, these patients had good performance status.  About half of them were ECOG 0 and the other half were ECOG 1.  About 10% of these patients were categorized as anemic, about two-thirds of these patients had a Bajorin risk factor score of one.  And important for this trial, about 60% of these patients had high PD-L1 expression, and just over half of the patients at about 56, 57% were deemed cisplatin eligible.

So, as Chris mentioned there were Co-primary endpoints. This is the first Kaplan Meier curve looking at overall survival specifically in the high PD-L1 population. durvalumab is in red and chemotherapy is in blue.  You can see there was no difference between these two treatments with a hazard ratio of 0.89, 95% confidence interval of 0.71 to 1.11. Looking at the subgroup analysis for this group, this is the durvalumab versus chemotherapy in that high PD-L1 population, a quick scan of the forest plot essentially shows that nearly all of these subgroups cross the hazard ratio of one, so no difference between durvalumab and chemotherapy.  There are several interesting signals, although with relatively a small sample size, take your attention to the upper tract, renal pelvis, or ureteral urothelial carcinoma, it did favor durvalumab, but as I mentioned with small sample size.

This is the other Co-primary endpoint, which is overall survival in the intention to treat population looking at durvalumab plus tremelimumab which you can see in green and chemotherapy in blue, and once again, a hazard ratio of 0.85, 95% confidence interval of 0.72 to 1.02 and no difference between durvalumab plus tremelimumab versus chemotherapy alone. Another forest plot, looking at overall survival in this group in the intention to treat population, similar to the previous forest plot, a lot of these subgroup analyses crossing the hazard ratio of one, so no difference between the two.  You can see possibly a signal here for men or women less than 65 years of age favoring durvalumab plus tremelimumab.  I'm working our way down here to the PD-L1 high status, there was a significant hazard ratio of 0.75, 95% confidence interval of 0.60 to 0.94.

Looking at the secondary endpoints overall survival in the intention to treat population, this is durvalumab versus chemotherapy, red for durvalumab, blue for chemotherapy, very similar curves between these two treatments, a hazard ratio of 0.99, and a non-significant 95% confidence interval of 0.83 to 1.17. As I mentioned on the previous forest plot, there was a difference between durvalumab plus tremelimumab versus chemo in the high PD-L1 population, I will remind everybody, this is a secondary endpoint of overall survival in this population, durvalumab in green, and chemotherapy in blue, the hazard ratio favoring durvalumab of 0.74, 95% confidence interval 0.59 to 0.93.

So this is a relatively busy slide, on the left, we have the intention-to-treat population looking at response, and on the right, we have PD-L1, sort of a general summary of the intention-to-treat population.  You can see here that the objective response was higher in chemotherapy, 49% compared to 36% for durvalumab plus tremelimumab, and 26% for durvalumab and these are similar trends between the cisplatin eligible and ineligible patients. What's interesting here, is that in looking at the best objective response for progressive disease, 53% for durvalumab monotherapy and 42% for durvalumab plus tremelimumab and only 18% for chemotherapy, so a strong indicator that chemotherapy is basically better disease control when you are looking at the best objective response. In terms of duration of response, responders who subsequently progressed or died, 82% for chemotherapy versus 67 for durvalumab plus tremelimumab compared to 62% for durvalumab monotherapy. Looking at the right side of the screen, the high PD-L1 population, you can see here that chemotherapy had an objective response of 48% compared to 47% for durvalumab plus tremelimumab and only 28% for durvalumab monotherapy.

And if you look down at the best objective response, you can see here that progressive disease, 51% durvalumab monotherapy compared to 34% for combo therapy, and 20% for chemotherapy. Also, a busy slide which I will summarize for you as well.  This is the treatment-related adverse events, so on the left, we have a host of various treatment-related adverse events. In this block of columns, you have durvalumab monotherapy, in the middle block of columns we have durvalumab plus tremelimumab and on the right we have chemotherapy.

So if we look at the very top row, this is the summary of any treatment-related adverse event.  You can see 42% grade 1-2 for durvalumab monotherapy, 47% for the combination therapy, and 30% for chemotherapy.  And looking at grade 3 and 4, 10% and 3% for monotherapy, up to 24% and 4% for durvalumab plus tremelimumab, So we are seeing an increase in toxicity when adding that tremelimumab to durvalumab.  And also for chemotherapy, not too surprising, grade 3 and 4 events, 46% and 14% respectively. In terms of most common adverse events for anybody receiving durvalumab tended to be an increase in lipase for chemotherapy, very well-known fatigue, neutropenia, anemia, all with the highest grade 3 and 4 adverse events.

So several discussion points from the DANUBE trial; this was a robust and mature randomized controlled trial that did not show a survival advantage compared to standard of care chemo in either of the experimental treatment groups. And so it is pretty convincing at least from the data we have right now, that immunotherapy is inferior to chemotherapy in the initial period of treatment. And as was recently published and also recently discussed in a previous Journal Club, the JAVELIN Bladder 100 trial showed that platinum-based chemotherapy followed by maintenance avelumab therapy is now the new standard of care, and this was published also by Dr. Powles in The New England Journal a couple of months ago. Interestingly, 60% of patients in the DANUBE trial had PD-L1 positive tumors, and so it leads us to think about whether PD-L1 expression alone may not be sufficient to identify patients who may benefit from PD-1 or PD-L1 inhibitors.

So in conclusion, the DANUBE trial did not show a survival advantage for durvalumab versus standard of care chemotherapy or for durvalumab plus tremelimumab versus standard of care chemotherapy. Some of the secondary endpoint results for the combo treatment suggest that there may be some activity for tremelimumab in this disease space, but certainly as we saw on the toxicity side, this comes with increased toxicity burden. And finally, further studies are needed to identify a potential role of immune checkpoint inhibitors alone or in combination, as first-line treatments for metastatic urothelial carcinoma. Thank you very much for your attention. I hope you enjoyed this UroToday Journal Club.