Elucidating Mechanisms of Resistance to ADT/PI3K-AKT Blockade in PTEN-Deficient mCRPC - Akash Patnaik

April 20, 2023

Andrea Miyahira and Akash Patnaik discuss a Clinical Cancer Research publication that explores a new approach to disrupt immuno metabolic pathways within an aggressive variant form of prostate cancer. Dr. Patnaik provides background on the current state of metastatic castrate-resistant prostate cancer and the biologic underpinnings of aggressive variant prostate cancer, focusing on the PTEN PI3-kinase pathway. The goal is to find a better way to target this molecular subset of aggressive variant prostate cancers, which typically have poor prognosis and therapeutic outcomes.


Akash Patnaik, MD, PHD, MMSc, Medical Oncologist, Physician-Scientist, Assistant Professor of Medicine, the University of Chicago Comprehensive Cancer Center

Andrea K Miyahira, PhD, Director of Global Research & Scientific Communications, The Prostate Cancer Foundation

Read the Full Video Transcript

Andrea Miyahira: Hi everyone, I'm Andrea Miyahira and I'm the senior director of Global Research and Scientific Communications at the Prostate Cancer Foundation. Today, I'm joined by Dr. Akash Patnaik. Dr. Patnaik is a physician scientist and medical oncologist specializing in genitourinary cancers. He leads the Translational Research Program in prostate cancer at the University of Chicago Comprehensive Cancer Center. Dr. Patnaik's laboratory and clinical trial efforts are focused on the tumor immune interface and the development of novel precision medicine strategies to activate innate immunity against prostate cancer. Dr. Patnaik is also the recipient of the 2010 PCF Young Investigator Award in two PCF Challenge awards in 2016 and 2022. Dr. Patnaik's group has recently published the paper Reversal of Lactate and PD-1 Mediated Macrophage Immunosuppression Controls Growth at the PTEN/p53 Deficient Prostate Cancer in Clinical Cancer Research. Dr. Patnaik, thank you for joining us today. I look forward to learning more about this study.

Akash Patnaik: Thank you Andrea, for the kind introduction and it's great to be here. As Andrea mentioned, my translational research program is very interested in developing strategies to turn on innate immunity as a strategy to activate immunotherapy and prostate cancers. What I'm going to discuss today is a paper that was recently published in Clinical Cancer Research talking about a novel strategy to disrupt immuno metabolic pathways within an aggressive variant form of prostate cancer, i.e. PTEN/p53 deficient prostate cancers.

As way of background, metastatic castrate resistant prostate cancer remains an incurable disease. While we've had success with incremental benefits and survival benefits of multiple hormonal therapies that target the AR pathway, as well as more recent precision medicine therapies that target DNA repair pathways that led to the FDA approval of PARP inhibitors, mCRPC or Metastatic Castrate Resistant Prostate cancer remains an incurable disease. More recently in the last several years, we've developed a deeper understanding of the biologic underpinnings of mCRPC. In particular, one of the pathways that is important in aggressive variant prostate cancer is the PTEN PI3-kinase pathway and PTEN loss hyperactivates PI3 kinase signaling and is frequently found in about 50 to up to 75% of mCRPC patients. Typically, carries a poor prognosis and poor therapeutic outcomes to hormonal therapies, chemotherapies, and immune checkpoint inhibitors. This is going to be the focus of the rest of my talk is how best to target this molecular subset of aggressive variant prostate cancers.

One of the first experiments that we did in a genetically engineered mouse model of PTEN/p53 deficient prostate cancer. This is conditional knockout of these genes within the prostate that leads to very aggressive tumors that mimic features of human disease. We treated these mice with androgen deprivation therapy and observed that similar to patients these aggressive variant prostate cancer murine model also exhibits limited responses to androgen deprivation therapy. This was typically a strategy where we would treat established tumors with androgen deprivation therapy and then monitor tumor development over the course of several weeks using non-invasive MRI. As shown here, we saw a response rate of about 16% at four weeks, which is relatively low and similar to what we see in patients where there's an initial response to androgen deprivation followed by rapid development of castrate resistant disease. This credential that this mouse model as a potential tool to investigate novel immuno-oncology therapeutic strategies in this particular subset of prostate cancer.

There's been a lot of interest in targeting the molecular consequences of PTEN deficient aggressive variant prostate cancers. This really goes back over a decade with work that was published by Brett Carver and Charles Sawyers showing a reciprocal feedback of PI3 kinase and androgen receptor signaling in PTEN deficient and prostate cancers. If you block one pathway, you hyper activate the other and vice versa. If you concurrently target these two pathways you may see more of a tumor cell-intrinsic cell death.

This was then translated to a clinical trial combining intensified androgen deprivation therapy. In this case, testosterone depletion plus a CYP17 inhibitor, Abiraterone, in combination of AKT inhibitor. As you could see here, this is two particular plots looking at both the progression-free survival in both PTEN loss, as well as P 10 wild-type patients. As you can see in the PTEN loss patients there was a modest improvement in progression-free survival with the combination of AKT inhibitor with intensified ADT.

However, this modest improvement was in no way a game changer. Clearly there's a critical unmet need to develop strategies to push the survival and the progression-free survival of these patients even further. We decided to take a reverse translational approach and essentially design a co-clinical trial in our PTEN/p53 deficient genetically engineered mouse model of prostate cancer. Similar to what we did in the clinical trial that I just showed you published by De bono, and colleagues, we treated our murine model with a androgen deprivation therapy and combination with the PI3 kinase inhibitor. We found that similar to what we saw in patients there was a partial anti-cancer response in about 25% of patients. This was a pretty modest response.

Then when we went on to actually profile both the in vivo tumors for Ki-67, which is a marker proliferation, as well as a proliferation index in tumor derived cell lines from these GEM models. What we discovered was that while there was a decrease in proliferation in vivo in the murine model there was actually no change in proliferation in the tumor derived cell lines. Suggesting a potential tumor cell extrinsic mechanism by which these drugs are actually working in vivo.

That led to our hypothesis that androgen deprivation therapy in combination of PI3 kinase inhibitors reprograms the tumor microenvironment in PTEN deficient prostate cancer resulting in partial tumor control. In order to understand this at a deeper level we treated our mice and vivo and then did some immunological profiling studies. As shown here, what we discovered was that androgen deprivation therapy actually increases the influx of tumor associated macrophages into the tumor microenvironment in these PTEN deficient tumors that was not further enhanced with the addition of PI3 kinase inhibitor. However, when you give the combination, you do see an increase in macrophage activation with the combination as suggested by the MHC class 2 markers. This suggested that there's in increase influx induced by ADT and activation induced by PI3 K inhibitor treatment.

However, we also noticed that the combination did also result in an increase and influx of PD1 high tumor associated macrophages, which was also elicited by ADT alone. It's a bit of a double edge sword. We're bringing in myeloid cells and macrophages. These myeloid cells are getting more activated with PI3 kinase inhibitor. Then you know, PD1 influx, high levels of PD1 actually drives a potential resistance phenotype was what we hypothesized based on seeing this data.
We then went to doing some co-culture assays to assess for functionality of these tumor associated macrophages, i.e phagocytosis. What we discovered was that the PTEN/p53 deficient gem tumors when we sorted the tumor associated macrophages and co-cultured them with tumor derived cell lines the AC1 and SC1 cells, these are derived from the gem tumors. We did the co-culture. We did find that the combination of androgen deprivation therapy plus a PI2 kinase inhibitor resulted in a significant increase in phagocytosis of the induced by these treatments on the macrophages above and beyond what we saw with androgen deprivation alone.

Androgen deprivation alone had some effect on phagocytosis. However, this was all done in specific macrophage subsets characterized in different macrophage subsets. We identified that the macrophage subsets that were MHC class 2 high, i.e. more activated macrophages, actually showed a more significant increase in cytosis induced by this combination relative to the inactivated or MHC class 2 low macrophages. We then, this is shown in the right two panels here, the high MAC class 2 high macrophages showed an increase in phagocytosis not observed in the MAC class 2 low or inactivated macrophages.
We then ask the question, if PD1 is actually being expressed on these macrophages is it actually driving immunosuppression and a limited responsiveness to the ADT, PI3 kinase inhibitor combination?

In order to answer that question, we sorted tumor associated macrophages into these different subsets and essentially the four subsets that we characterize with the MHC class 2 high, PD1 high, and then MHC class 2 high, PD1 low, MHC class 2 low PD1 high, and then MHC class 2, low PD1 low. We wanted to look across the heterogeneity landscape of tumor associated macrophages in these in vivo tumor derived macrophages and see where PD1 blockade had its effect.

What we were struck to see is that the most, the immunosuppressive subsets that had high levels of PD1, both MHC class 2 high and MHC class 2 low, MHC class 2 low being your super inactivated macrophages, actually showed an enhancement phagocytosis with concurrently one blockade. Suggesting that this combination with the PI3 K, ADT doublet backbone might actually show improved responses in both genetically engineered mouse models and hopefully in patients down the road.
In order to get a little deeper into understanding mechanism for these combinatorial responses we tried to dissect it individually. We looked initially at the ability of PI3 kinase inhibition, which is known to perturb the Warberg Effect or aerobic glycolysis to test the hypothesis that maybe the PI3 kinase inhibitor is actually inhibiting lactate levels released from the tumor cells and that lactate is actually driving some aspect of macrophage immunosuppression. We did observe as has been known in the field that PI3 kinase does actually indeed inhibit lactate secretion also from our PTEN/p53 deficient prostate cancer cell lines, as well as vivo.

We then ask the question if indeed the lactate production is decreased from these tumor cell lines, does that actually alter a histone lactylation or post translational histone modification mechanism within tumor associated macrophages resulted in their activation. There was a recent paper demonstrating that histone lactylation can actually potentially drive immunosuppression in macrophages. We asked the question of whether PI3 kinase inhibitor treatment in vivo and in our ex vivo co-culture systems can actually suppress lactate and thereby suppress histone lactylation. We found that indeed when we added back lactate, which is a rescue experiment, we found that you can actually potently suppress phagocytosis.

We initially looked at the ability of copanlisib to increase phagocytosis, which it does. Then when we added back lactate as a rescue strategy we suppress phagocytosis and inversely increased histone lactylation. Suggesting that PI three kinase inhibitor treatment was working by reprogramming the tumor associated macrophages via lactate and histone lactylation. An interesting crosstalk between these tumor cells and macrophages.

We also went, as a proof of concept, to understand if this mechanism is important translationally, we actually went to our metastatic prostate cancer castrate resistant prostate cancer biopsy samples both within bone and lymph node. We, essentially, developed a aerobic glycolysis and a phagocytosis suppression signature in our bone mCRPC patients. We found that there was a interesting correlation between increased aerobic glycolysis and phagocytosis suppression within TAMs within the bone metastatic microenvironment.

We also looked in our lymph node samples and again looked at aerobic glycolysis and M1 polarization status and found an interesting inverse correlation in our lymph node samples, as well. Suggesting that this mechanism of lactate mediated immunosuppression within tumor associated macrophages is indeed preserved translationally in our human metastatic microenvironments and this could be relevant therapeutically as we think about translating these strategies into the clinic.
Lastly, for the androgen piece of it we were also interested in understanding of androgen deprivation can directly turn on macrophage activation. We, essentially, took tumor associated macrophages derived from our PTEN/p53 deficient murine models and treated with the singlet, doublets, and the triplets. As you can see here, the androgen deprivation alone, which is the AD or it actually resulted in increase in MHC class 2 in the presence of co-culture with either the AC1 cells or the SC1 cells. That was then not further increased by either adding on the PD 1 inhibitor or copanlisib and/or copanlisib. Suggesting that the antigen deprivation alone was sufficient to drive an aspect of macrophage activation.

Moving on to how we're putting this all together we hypothesized, and developed, and hopefully have convinced you, and certainly were able to convince the reviewers that this mechanism of lactate suppression induced by PI3 kinase inhibitors is actually extremely critical in regulating an aspect of macrophage biology, i.e. it's immunosuppressive properties. When you block lactate, you block histone lactylation, you enhance macrophage phagocytosis and concurrently with ADT or castration, which can also have an effect on macrophage activation you can get a more potent induction of phagocytosis.

However, these macrophages are high in PD1, at least a subset of them, and those tumor associated macrophages can be concurrently activated with a triple combination of PI3 kinase, ADT, and PD1 blockade. Since most of this work was done mechanistically we went back into our in vivo murine models to test whether this triple combination can actually enhance immune responsiveness. Indeed we found a 60% response rate with the triple combination in these PTEN/p53 deficient gem mice in vivo.

What's particularly important here is that when we depleted macrophages with clodronate designated as the CL on the slide there, concurrent clodronate depletion completely abolished the antitumor response that we saw with the a triple combination. Suggesting that this triple combination response was entirely tumor cell extrinsic and dependent on a macrophage phenotype, presence of macrophages in the microenvironment.

The question that we were then interested in is why are these triple combination tumor mice responses limited to only 60% of the mice? We went back to our tumors from both responder and non-responder mice and did RNAC to try to an identify differential pathways that may predict for resistance to the triplet combination. Now, what we discovered was that the Wnt/beta-catenin signaling pathway emerged as a top hit in terms of a driving resistance in these PTEN/p53 deficient mice that were treated with a triple combination. We then went on to confirm by proteomic profiling of western blots and showed that indeed in the non-responders you see an up regulation of Wnt/beta-catenin signaling not present in the responders.
Putting this all together what we have demonstrated, and in the interest of time didn't get to show you all of the lactate restoration data, that 40% of these resistant mice can actually be rescued for lactate production using the Wnt/beta-catenin signaling pathways. The cancer cell is hardwired to preserve aerobic glycolysis. If you block PI3 kinase signaling you block lactate, but there are other signaling pathways that can, essentially, take over, and restore lactate production, and restore macrophage immunosuppression. That accounts for 40% of resistance in these mice. Similar to what we see in patients, not all of the patients respond to therapies that we test and there's always resistance. I think it's important to understand resistance in these murine models, so we can develop next generation smart IO combinations that can overcome resistance.

This is just my conclusion slide that co-targeting lactate and PD1 mediated macrophage immunosuppression combination of the ADT controls growth of PTEN/p53 deficient prostate cancer. We believe that precision medicine based combination, immuno-oncology combination, therapies will become a part of the future treatment paradigm for treatment of patients with advanced aggressive variant prostate cancers.

With that I'll end by thanking my team of investigators who really did an amazing job. This was great team science effort to really put this exciting story together. Both members of my laboratory team at the University of Chicago, collaborators at University of Miami, Mass General Hospital, the Garvan Institute in Australia, as well as collaborators of the NIH, Bristol Myers Squibb for providing the PD1 antibody in context of an international immuno-oncology network collaboration. Most importantly, I would like to thank the Prostate Cancer Foundation for actually funding this important study. We look forward to continuing to work with PCF on other projects and then other federal sources of funding that supported some of our co-investigators efforts. That I'll stop and I'm happy to take any questions. Thank you.

Andrea Miyahira: Thank you, Akash. That was such an interesting study. One of the most striking observations here is that when you add clodronate which depletes macrophages you limited all the activity of the triple combination of ADT plus PI3 kinase inhibition and anti PD1 in your mouse model. Your studies also suggest that the efficacy of ADT plus PI3 kinase inhibitors is not direct tumor cell death, but phagocytosis by macrophages. This suggests that even the efficacy of ADT alone depends on tumor phagocytosis by tumor associated macrophages.

Akash Patnaik: Yeah. That's a really good question, Andrea. What we have tried to do in the context of this work is the tumor associated macrophages are not a one size fits all. There are heterogeneous subsets. We broadly classified them into four distinct subsets in this paper. What we've demonstrated is that ADT can have a effect in inducing phagocytosis in the tumor associated macrophages that are MHC class 2 high, so the more favorable macrophages. But, in the extremely unfavorable or MHC class 2 low macrophages ADT doesn't seem to have any effect. It's unique to a subset of macrophages. That's why, I think, the triplet combination therapy takes advantage of targeting heterogeneous subsets of macrophages that may not be targetable by any individual strategy alone.

Andrea Miyahira: Okay. Very interesting. Another observation is that, well, we know the anti PD1 is normally thought to act by targeting T-cells. Did you see a role for T-cells in this combination therapy approach?

Akash Patnaik: Yeah. We did actually look at the role for T-cells and actually did T-cell depletion studies, which I didn't get a chance to show in the interest of time. We actually saw no impact of T-cell depletion on the antitumor response, but a profound effect impact of macrophage depletion on the antitumor response, which as shown in one of the figures in the paper was pretty striking.

Andrea Miyahira: Do you think that this specific role for macrophages versus T-cells is this a prostate cancer unique effect?

Akash Patnaik: Yeah. That's an excellent question. PTEN loss, as you know, is not only restricted to prostate cancer it actually predicts for aggressive disease and other tumor types, as well. Including melanoma where PTEN loss has been shown to actually be a driver of immunotherapy resistance in melanoma and other tumor types. We would anticipate that this lactate crosstalk mechanism driven by PI3 kinase signaling is likely going to be important in tumor cell macrophage crosstalk beyond prostate cancer. I think, there's some relevance. We certainly haven't tested this yet in other models, but this would be work that we'd want to consider for future efforts.

Andrea Miyahira: Okay. Super interesting. Another question is, what causes the macrophages to phagocytosis tumor cells? Are they expressing eat me signals or are there don't eat me signals that are modulated by these treatments?

Akash Patnaik: Yeah. Excellent question. We did look at eat me and don't eat me signals present molecules like CD47, CER-1 alpha axis, which is very well known axis studied by Irving Weissman and others at Stanford. We've looked at other eat me and don't eat me signals. The short answer is we didn't see a difference in the expression level of these molecules either on the surface of tumor cells or their counterparts in the macrophages. We think that this enhancement of phagocytosis might be independent of the canonical eat me, don't eat signals. There's certainly a possibility that there are other eat me, don't eat signals that have yet to be discovered that may actually be important in driving this phenotype. We would, obviously, need to do SHRNA screen to actually look for additional molecules, which, again, is an interesting idea and one that is worthy of pursuit down the road.

Andrea Miyahira: Okay, interesting. Is AR expressed and activated in the tumor associated macrophages and do you know if it is what programs it's regulating?

Akash Patnaik: Yeah. That's a very important question and one that is a central focus of efforts within my laboratory. We've actually found, at least in two different contexts, that AR and macrophages is actually very important in driving innate immunity. Actually, one of these projects is the focus of our Prostate Cancer Foundation Challenge Award that recently got funded looking at AR and innate immune inflammasome signaling through NLRP3. That's an area of ongoing investigation. The short answer is that there appears to be a direct role of AR in macrophages that's driving aspects of the native immunity.

Andrea Miyahira: Okay. Really interesting. How common is PD1 and PD-L1 expression observed in prostate tumors with PTEN loss, especially, in mCRPC?

Akash Patnaik: Yeah. In this particular study we did observe PD1 expression as well as PD-L1 in both macrophages, other immune cell types, as well. We looked at NRT, CD4, CD8. In terms of PTEN loss relative to non PTEN loss we did some early IHC studies and had not really seen a significant difference, but that was, again, a small number of patients. I think, that question the jury's still out, because the tumor microenvironment is complex and there is a dynamic interplay of these checkpoint molecules. Not just in tumor cells like PD-L1, across different cell types in the tumor microenvironment and similarly PD1. There's no evidence that I'm aware of at this point that PD1 and or PD-L1 is more highly expressed in PTEN deficient relative to non PTEN deficient. I think that needs to be looked at a bit more systematically.

Andrea Miyahira: I thought it was really interesting that you identified mechanisms of resistance and here you found activated Wnt signaling with a mechanism of resistance to the triple combination. Do you see Wnt pathway activating mutations in prostate tumors with PTEN loss or PI3 kinase activation? Is there any differences in biology or treatment response in these subsets?

Akash Patnaik: We found Wnt/beta-catenin signaling program activation as a rescue mechanism in these resistant mice. It appears to be happening at a signaling level. We are still doing genomics to understand if this is actually an activating mutation. We don't have the answers to that question yet. Wnt/beta-catenin alterations have been reported in mCRPC in about 10% of patients. In terms of concurrent alterations of PTEN and Wnt I've seen them on occasion and some of my metastatic biopsies done on patients for clinically actionable purposes in the clinic. I'm not aware that that's a frequent alteration. Although, certainly there'll be small numbers of patients that do have both alterations. Some of that work, at least in the murine context, we're hoping to get done in the coming weeks to months.

Andrea Miyahira: Okay. Yeah. I was wondering if that suggests that Wnt alterations might be a biomarker for patient could be less likely to respond to at least this treatment combination?

Akash Patnaik: Absolutely. Yep. Our prediction based on this work would be that if you were to be able to profile these tumors for Wnt/beta-catenin activation, since it's a tumor cell activation you could even look at, potentially, circulating tumor cells. If it's an activating mutation you could even do a liquid biopsy and look at circling tumor DNA and use that as an early marker of potential resistance to this combination therapy. Absolutely.

Andrea Miyahira: Thanks. I guess, looking more broadly do we know if Wnt activation PI3 kinase activation or any other lactate producing pathways are associated with non-response to check point for immunotherapy?

Akash Patnaik: Yep. That's an excellent question. There's a part two of this story that we're working on presently that is actually pretty close to being submitted where we are at. We've actually been able to target Wnt signaling in this murine model in the context of this combinatorial strategy to, essentially, shut lactate production back off. In the context of some of that work we found other signaling pathways that have, essentially, restored lactate production. We have now approaches to [inaudible 00:30:04]  across the board. This will be the subject of a second story that will be coming out at least in pre-print form hopefully soon. Stay tuned for that.

Andrea Miyahira: All right. Yeah. We're staying tuned.

Akash Patnaik: Yeah.

Andrea Miyahira: Well, what toxicities would you anticipate in patients with this triple combination?

Akash Patnaik: Yeah. It's a very important question clinically that we've been thinking about. We're actually participating in a couple of clinical trials at the University of Chicago looking at dual targeting of PI3 kinase or AKT inhibitors, in this case, with either in the hormone sensitive De Novo setting or in the castrate resistant setting. Certainly, there are known on-target toxicities of PI3 kinase inhibitors, hyperglycemia being a common one, because you perturbed glucose metabolism. Often will treat these patients with Metformin, which, hopefully, can abrogate some of the attenuated anti-cancer effects that you see with PI3 K inhibitors, because you're driving insulin IGF-1 signaling.

On target, hyperglycemia, fatigue, rash, these are common class effects with kinase inhibitors. Androgen deprivation therapy has non-overlapping toxicities for the most part, which include fatigue and hot flashes. Then PD1 blockade is associated with immune related adverse events, particularly autoimmune toxicities. There is a theoretical question of would you exacerbate any of the autoimmune toxicity [inaudible 00:31:44] combination of PI3 K inhibitors, because you're altering innate immunity. That's something we would really need to test in a phase one clinical trial. We certainly hadn't observed significant toxicity in the mice, but we need to do dose escalation studies in the phase one setting with these medicines. Particularly, the PI3 K inhibitor, and PD1, and be able to assess for toxicity. More to be done on that in the clinical setting.

Andrea Miyahira: Yeah. That leads perfectly into my next question. What are your next steps for translating these findings for patients?

Akash Patnaik: Yeah. We're in discussion with a couple of companies to look at bringing their assets together to target both PI3 kinase and the PD1 access. Those are ongoing discussions. Now that the paper is out there's certainly more momentum to get this to the next level, do a investigator initiated trial, collect biopsies before and after to try to understand if some of the resistance mechanisms that we've identified in mice is actually important in patients. We would be, hopefully, getting a study off the ground once we've gotten buy-in from companies. All of that's in the planning and discussion phase. Stay tuned on that, as well. Yeah.

Andrea Miyahira: Okay. Well this is a really interesting study. Thank you for coming and sharing this with us today.

Akash Patnaik: Thank you for having me. It was a pleasure.