Himisha Beltran: Thank you for the invitation and for this fantastic meeting. When looking into the future for oligometastatic disease, I am quite optimistic that we will have made significant progress on refining our definitions for disease stratification. We will have nominated molecular biomarkers to help us better capture biologic disease states, which will ultimately help better define goals of care and improve patient selection. As we know, our current definitions based on number, volume, and or location and metastases are relatively crude estimates based on standard imaging. And these definitions and paradigms are further impacted and challenged by more sensitive imaging such as PSMA PET. I do believe that goals of care should be determined by biology, and understanding the spectrum of biologic subsets within the spectrum of oligometastatic disease has implications for all the different oligometastatic clinical presentations that we’ve heard a lot about.

For if we identified a patient with oligometastatic disease whose tumor truly had low metastatic potential, you would imagine goals would be different, delayed treatment, come off of therapy, or even achieve cure. So obviously we would do anything it takes for that individual. There is some data to suggest that the lesion, the primary tumor that metastasizes and is responsible for death, lethal prostate cancer, is not always the lesion that metastasizes. So I think that there are likely clues within those non-index lesions or in those metastases that could help us predict patterns of spread and the relative indolence of certain metastatic lesions. The primary tumor can also continue to seed metastases, which is one of the rationales for treating the primary in the setting of distant mets. STAMPEDE and HORRAD point to a greater benefit for those patients with low volume disease. And I would hypothesize is potentially due to a greater contribution of those primary non-index clones contributing to metastasis in this clinical situation.

Then there are patients with intermediate or heterogeneous disease. For instance, one lesion’s progressing and others are stable. Where we want them to combat heterogeneity with multi-modality therapy. And our goals might be to stay on the same systemic treatment or slow progression. When thinking about all the oligo-progressive disease when there might be different biologies across different metastases, there are a number of possibilities. It could be that multiple clones within the primary are responsible for metastasis and can explain a differential progression or resistance of one lesion over another. Or it could be that one clone was responsible for metastasis, and then tumors acquire additional mutations to make them more resistant or oncogenes to make them grow faster. Understanding these clonal evolutionary patterns have important implications for biomarker development. When we go back to that primary tumor and are trying to predict biology and to help guide us for what to actually look for in that metastatic lesion. And then there are patients that truly have systemic disease where local or focal therapy may have little or no impact. For these patients, we really need to focus on systemic therapy intensification.

There are a number of tools to try to capture systemic disease. What rate for the tumor, as well as the microenvironment. But just to note, there’s very little known about the microenvironment in oligometastatic disease. It’s important to define what we’re actually looking for. Do we just want to detect more cancer? Are we looking for specific alterations or metastasis markers that may predict behavior? These have different biologic and clinical implications. We do know that PSMA PET is highly sensitive in capturing tumor burden noninvasively. Cell-free DNA, at least in this CRPC setting has been showed to associate with not just the presence of bone mets, but the number of bone meds in the presence of visceral mets. But is this what we’re looking for to try to find the cancer … Detect more cancer? Or can we use molecular tools to try to predict metastatic behavior before we actually detect it or see it so we can intervene early?

As we know, what we see may not always be what’s underneath. And there are some lesions, even if there are few of them in a patient that have more aggressive biology. And similarly, there may be lesions that look scary, that are actually more indolent. Metastasis or metastasis seeding further leads to interpatient heterogeneity, which is difficult to capture by single-site biopsy. I think likely requires a more refined analysis of liquid and imaging biomarkers. So this is an example of a tissue, metastatic tissue sample and a cell-free DNA plasma sample lined together. You can look genome-wide at amplifications and deletions, and you can see there’s a concordance of alterations for many areas across the genome. But there are also areas of differences. And if you look at the end at 16q22 we scale in, we see there’s homozygous deletion in the tumor. But the average signal in the plasma is lower. So what this means is that there’s likely some clones have a homozygous deletion, and some clones do not. And we can do this genome-wide to try to recapitulate computationally the relative contribution of different clones in the circulation, and understand private and shared mutations and track different clones with time.

Most cell-free DNA studies have been in castration-resistant prostate cancer which is quite challenging in cases of low tumor burden such as oligomets. But I do think understanding resistance patterns will be quite interesting in the setting of oligometastatic and oligo-progressive disease. Just to illustrate this point, these are two patients. These are CRPC patients. On the left is a patient with … Data from a patient with bone only mets rising PSA. Each column is a time point, a cell free DNA time point. And you can see that the genomic alterations differ with time points. What this suggests is different clones in the circulation with potentially different competing frequencies or clonal competition. On the right you can see that patient, every single time point looks nearly identical suggesting dominance of a certain clone that may be driving progression in that case. And so understanding this in this setting of oligomets will be again, I think quite informative.

Circulating tumor cells can also capture phenotypic heterogeneity resistance mechanisms. And also lends a capability for single cell genomics to detect the emergence of clones and sub-clones. In two years we will also have more data on how we have altered the natural history of oligometastatic prostate cancer with our treatments. What happens, do they develop more mets? Do they develop widespread systemic disease? And was this impacted by our local or focal therapy? And understanding these progression patterns will help us further refine our biologic definitions. So I’m looking forward to the next APCCC in two years and hearing about all the defined clinical definitions, and nominated biomarkers. But I think we still have a lot of work to do as a field. The molecular tools do exist that can help us capture biologic subsets and tumor evolution patterns. But we need to study this and we need to test them in our clinical trials. And this will ultimately lead to better biomarkers and improved patient selection. Thank you.