Read the Full Video Transcript

Marzia Del Re: Thank you for your invitation. It's a great privilege to be here today in Milan and with so many great scientists. I'm a clinical pharmacologist, so my presentation will mainly focus on predictive biomarkers of resistance and response to therapy in a scenario of heterogeneity in prostate cancer.

Just to start, I would like to highlight some of the key points about tumor heterogeneity. We know that prostate cancer is characterized by clinical, morphological and molecular points of view of high heterogeneity. This heterogeneity is mainly determined by genetic and epigenetic alterations.

Technological improvements like next-generation sequencing, comprehensive genomic profiling, artificial intelligence and liquid biopsy, teach us that tumor heterogeneity is very complex. We can have interpatient heterogeneity that is,diversity between different patients. We also have intratumoral heterogeneity in the primary tumor, and we also have heterogeneity between different metastatic sites, and we talk about inter-tumor heterogeneity.

When we talk about tumor heterogeneity, we include epigenetic expression, post-translational, morphological phenotypic heterogeneities, and all contribute to disease progression and the clinical manifestation of a solid tumor. We will talk about prostate cancer, but these main characteristics are shared between all solid tumors.

When we talk about tumor heterogeneity, now we are also able to identify what could be a model of clonal progression of prostate cancer. We know that the tumor is originally composed of different cell clones characterized by different molecular alterations. During progression, there is an evolution of different clones that are partially separated, and they go to constitute the primary tumors, with different and distinct genomic driver changes. What we have when the tumor progresses and metastasizes under the selective pressure of systemic therapies, is that the metastatic dissemination of a tumor subclone, accumulates with further genetic and phenotypic alterations, and this will build a different tumor under a molecular profile, very different from the primary tumor.

We know that prostate cancer is highly heterogeneous, and we also know that, from the morphological point of view, because we know that the prostate tumor may have different Gleason scores, so this is a representation of a radical prostatectomy specimen with two distinct tumor foci. You can see that there are different Gleason scores in these different sections, and this is one of the most clear representations of tumor heterogeneity talking about morphological heterogeneity.

At the basis of this morphological heterogeneity, there is molecular heterogeneity, and you can see here in this clonal and subclonal heterogeneity in tumor tissue, you talk about PTEN previously. Now I'm showing you how this tumor is able to show a subclonal loss of PTEN in a subset of cancer glands. You can see the differential expression of PTEN in the same lesion. This is mainly one of the reason why we had the Ipatasertib trial fail, maybe tumor heterogeneity, is one of the causes of this failure.

What we know is that the clonal evolution of all these molecular alterations is selected by the treatment pressure. We have the initial presentation of a tumor mainly characterized by different clones with different allelic fractions. We have the recurrence of the disease and the resistance to therapy. As I told you before, there is an expansion of the resistant clones that are able to proliferate under the selective pressure of the treatment.

This is the schematic representation of what actually a pharmacological treatment is able to do. We have at the baseline the coexisting heterogeneous cancer cell clones. What the treatment that you are administering to your patients is doing, is to kill the sensitive cancer cells and to allow the proliferation and the differentiation of resistant cells. What we know about prostate cancer is that the number of clones increases with the increasing of the PSA values.

I will focus my presentation on a few molecular drivers. In prostate cancer, there are a lot of mutations other than these, but I just selected these mutations for timing reasons and also for their mainly clinical significance in terms of therapies. I will focus on androgen receptor alterations that are the tumor driver and also predictive biomarkers of both response and resistance to treatment. There is PTEN deletion. You already talked about PTEN a lot. There is TP53 mutation that I used to put together with RB1 deletion because together, they represent a predictive biomarker of resistance to treatment, and there are the DNA repair genes like BRCA1 and ATM and so on.

Talking about the androgen receptor, I cannot escape from talking about the strange, unlucky story of the androgen receptor variant seven. This biomarker was published in the New England Journal of Medicine and was quite an unlucky story because we had a lot of evidence from the literature of the predictive role of this variant in prostate cancer as the resistance to hormonal treatment. We had evidence on circulating tumor cells, evidence in exosomes, in tissues, in RNA, but we still don't use these biomarkers in clinical practice.

Maybe the factor that we are still lacking is to give AR-V7 a cutoff. The study only mainly focused on the presence or absence of AR-V7. Maybe a cutoff will help to stratify patients as really a resistant or responders to therapies. But the androgen receptor is also a predictive biomarker of resistance to treatment in the sense of over expression of the receptor. We know that patients who have been exposed to hormonal treatment, in this case it's enzalutamide, over-expressed the androgen receptor, and this is one of the main pharmacological mechanisms of resistance of solid tumors. They over-express the target of the therapies.

We already have a lot of evidence about the androgen receptor over-expression. You can see here the pink curve that represents patients with androgen receptor overexpression. But today, since we increase our number of drugs, I would like to introduce another predictive biomarker related to the androgen receptor that is the point mutation of the androgen receptor. These point mutations may be acquired during treatment. They usually appear in the binding site of the androgen receptor and the drug, and what they do is to activate the receptor in some cases, but particularly some mutations are able to transform some drugs from antagonist to agonist. This would be very important and very interesting, also in terms of identifying a correct sequence strategy of treatments for those patients. You can see here, even if the number of patients is not very high, the green line represents the patients with the androgen receptor mutations treated with abiraterone or enzalutamide.

Another mechanism of resistance to hormonal treatment, I previously mentioned the RB1 mutations and TP53 because they can identify a particular mechanism of resistance that is the histological switch from adenocarcinoma to neuroendocrine histology. This is a very interesting study conducted on circulating tumor DNA, to recognize the transformation to castration-resistant neuroendocrine prostate cancer. You can see the molecular profile that is very different between the adenocarcinoma and the neuroendocrine histology with an increase in loss of genes, and increasing in the number of nonsynonymous single nucleotide variants. In particular, you can see here in this representation, there is a lack of mutation or focal gain of the androgen receptor, and there are many hypermethylated genes.

This was rucaparib, one of the main mechanisms of resistance is the loss of CDK12. I think this would be very interesting because CDK12 may be a predictive biomarker of resistance to PARP inhibitors because it leads to genomic instability and an increase in the neoantigen expression. And at the same time, CDK12 loss may be a predictive biomarker of response since this increased neoantigen expression may give sensitivity to PD1 inhibitors. You can see here that the CDK12 patients who had no response to PARP inhibitor, may have a higher PSA response rate to PD1 inhibitors, so very preliminary data, but that could be very interesting for prostate cancer.

I'm going to finish my presentation just highlighting a new concept and I would like to bring you this data, this very interesting data about the differential response of BRCA2 versus BRCA1 patients to PARP inhibitors. You can see here that apparently even if the number are very low, but patients with BRCA1 mutations respond less to PARP inhibitors. And I think that the conclusion and the results of this paper are very interesting because the authors say that the differences in response might be related to the concurrent alterations in the BRCA1 group.

I think this is a very interesting result because with the improvement in biological knowledge, with technological knowledge, we also learn to talk about concurrent mutations and not only mutually exclusive mutations. We were used to talking about mutually exclusive mutations until a few years ago.

What I would like to introduce with those data is the concept of gene dialogues because when we talk about concurrent mutations, we do not have to think of a functionally isolated role for each mutation. But we should start thinking about how these mutant genes are connected between each other. Because we know that from a biological point of view, genes are able to dialogue between each other. And today there are a lot of studies that use artificial intelligence, for example, to address the entity of this gene dialogue. You can see here that there are a few characteristics of a gene when it is mutated in a complex of other genes.

These characteristics are the degree or the connectivity or the betweenness of the mutant genes. How to translate these gene dialogues into your clinical practice and how to learn how to treat those patients, this is one of the studies that we are conducting in Pisa about the gene dialogues. This is the case of BRAF. The question was why if BRAF is a driver gene, why colorectal cancer does not respond to the same drug as melanoma patients? This is why, because of the gene dialogues between the tumor. These are the gene dialogues of the BRAF gene in colorectal cancer. These are the gene dialogues in melanoma patients, and these are the gene dialogues of BRAF in non-small cell lung cancer. Maybe one day earlier, I think, we will be able to use this information on how genes are able to dialogue between each other, to understand how to best treat our patients.

To conclude about tumor heterogeneity, primary prostate cancers are often multifocal with spacial and morphologically distant tumor morphology, which may show non-overlapping truncal genomic alterations, suggesting that multiple clonally distinct cancers can arise in a given patient. Intra-tumoral and inter-tumoral heterogeneity present within the prostate gland poses diagnostic and therapeutic challenges. Despite the multiclonality of primary cancer, therapeutic intervention seems to select for a single dominant clone. The development of novel technologies will allow us to navigate these challenges and refine approaches for translational research and ultimately improve our patient's care. Thank you for your attention.