Emerging biomarkers for predicting response to cancer immunotherapy

DR LOVE: Welcome to Oncology Today — Predictors of Response to Cancer Immunotherapy, this is medical oncologist Dr Neil Love. For this program, I met with Professor Solange Peters from the Lausanne University Hospital in Switzerland. In addition to this audio podcast, there is also a video component. To begin, Professor Peters talked about emerging biomarkers to predict responses to cancer immunotherapy.

PROF PETERS: It’s my great pleasure to discuss with you today the topic of the emerging biomarkers to predict responses to cancer immunotherapy. We have been in cancer immunotherapy mainly today targeting 2 different distinct synapses, one which is quite important in the priming phase of the immune response, CTLA4, and the very important synapse in the effector phase of the immune response, a PD-1/PD-L1 signaling pathway.

Of course, we’ve been using many compounds, and in some diseases, like lung cancer, we are very fortunate to have almost tested all of these drugs, ipilimumab and tremelimumab as anti-CTLA4, and the growing list of anti-PD-1 and anti-PD-L1, that you can see on the left-hand side of this slide. And of course you’ve been probably hearing about the huge development into this new compound, anti-PD-1/anti-PD-L1, which I don’t think will change the paradigm of treatment, but might broaden and allow for better accessibility to this strategy globally in the future, which is very positive for our patients.

This being said, we have been seeing very variable activity of these compounds, anti-PD-1 and anti-PD-L1, when they are used as monotherapy across cancers. And as you can see here, even if some diseases are characterized by a nice response, like Hodgkin disease and bladder cancer, you can see that roughly less than half of the patients respond to this monotherapy. And that’s one of the reasons why we need to develop biomarkers for this specific strategy, because not all of the patients respond and not all of the patients benefit, which also means that in terms of sustainability we need to enrich the treatment population further for benefit.

Of course we need to avoid harming patients. It’s of course about toxicities. But it’s also because sometimes making the wrong choice is preventing your patients to be exposed to the right compound, for example chemotherapy for some of them. And of course we have to struggle with the financial burden of the compounds.

Think about biomarkers, and I invite you to go through this wonderful review. There’s a growing list of potential candidates that you can try to classify. The first very obvious category, which I call the essentials, is about PD-L1 and TMB, that I will comprehensively discuss later on. And there is a growing list of potential candidates looking at the host, the T-cell repertoire, the HLA composition, as well as the famous microbiome. You can think about the oncogenic contexture of the tumor itself, characterizing various levels of efficacy of immunotherapy. And probably less studied today, the microenvironment of the cancer cells, which can also represent some kind of predictive biomarker. Of course beyond PD-L1 and TMB none of them are for decision, but just for research to date.

But let’s start with the famous essentials, the TMB. So if you think about PD-L1, probably the paradigm of PD-L1 was developed in non-small cell lung cancer. We have been first of all trying to look at the equivalence of various platforms and assays, which we have been able to show to be quite equivalent. And we have been showing across almost all trials a predictive ability of PD-L1 when measured on tumor cells using one of these assays. We’ve been showing also that you can use fine needle aspiration cytology that unfortunately PD-L1 is dynamic and heterogeneous, but you have to deal with that and use it the way it is. And of course we all know it’s a continuum, always opening the debate about the borders of any threshold you would like to define.

But of course this is lung cancer, and the performance of PD-L1 is not the same across all diseases. And on the left-hand side you have this nice shoulder of the PD-L1 as a predictive biomarker on the ROC curve at more or less 50% of expression of tumor cells, when on the right-hand side you can see melanoma, that you have no shoulder. So that PD-L1 doesn’t show to have a threshold at which you can define the PD-L1 as a biomarker. But maybe just have a look on the left-hand side, the non-small cell lung cancer PD-L1, with this shoulder at 50% of PD-L1 on tumor cells. It’s quite an interesting to continue to use. I will show you some curves just on the next slide.

Thinking about PD-L1 as a biomarker, one has to understand that it’s not always measured the same way across diseases. In lung cancer we use tumor cell staining for PD-L1, and you can see on the upper schema here that this tumor cell staining for PD-L1 has a prevalence which is very variable from one disease to the other ones. But in other disease types, like bladder cancer and breast cancer, you use PD-L1 staining not on the tumor cell but on the immune cells, which unfortunately was shown to be less reproducible across pathologies. So adding complexity but also an accuracy kind of discussion.

And to add some complexity for you to know, there’s a third scoring system which was developed, so combined positivity score, called CPS on the left-hand side, which is counting basically on all the cells being positive for PD-L1 confounding tumor cells and immune cells. So these 3 kinds of paradigms, tumor cells, immune cells, CPS, can be used differently across tumor types. Just keep it in mind to be sure that you understand that. And this table that you can stop a little time on is summarizing for you across disease what has been used, immune cells, tumor cells, CPS, and what thresholds have been used in which indication, just telling you that if your pathology tells you PD-L1 is positive you need to make sure that you are discussing about the same topic and the same biomarker.

This being said, I wanted to show you this curve. Remember lung cancer, PD-L1 50% on tumor cells has been shown to be a strong biomarker. The upper panel is a long-term follow up of pembrolizumab versus chemo in patients with more than 50% PD-L1 on tumor cells. You can see here a doubling of the 5-year survival, up to 32%, just with pembrolizumab. And this has been validated with not only pembro, but cemiplimab and atezolizumab, which just kind of proves the fact that this biomarker is a very solid one in lung cancer.

And of course when you define a biomarker and a threshold you have to show that the threshold is correct. And with the same compounds we have been showing that lower threshold, 5%, 25%, 1% to 49%, are all thresholds which are insufficient to predict the immunotherapy activity above chemo. So very strongly defined. And again, as I was telling before, it’s still a continuum, and if you think about lung cancer still, if you have 90%, it’s going to be better than 50%. So it’s really something. And here on this schema you can see the difference between 50% to 89% and more than 90%, and you can see how different it is.

So when you meet this patient with 100% PD-L1, of course you should give immunotherapy only. So a very interesting biomarker, and remember please about the complexity of PD-L1 across diseases.

So the second essential is TMB (tumor mutational burden). Keep in mind, again, that we define every disease according to the number of mutations that you find, the median number of mutations that you find in the tumor. And lung cancer, melanoma, bladder cancer, urothelial cancers, are all cancers characterized by a high number of mutations probably resulting from the aggression on DNA by smoking or UV light repetitive signatures. So you can see this classification according to the number of mutations.

And on the right-hand side this is about the prevalence of mutations which are called high TMB across diseases, more than 10 mutations per megabase. So you can every disease is more or less mutated. Okay? And it’s interesting to see, if you look at some very important diseases, that you can even look at the diseases according to the number of mutations and response rate to immunotherapy, which will be correlated, with a median number of mutations and a median response rate for renal cell carcinoma and non-small cell lung cancer, and a very high number of mutations and very high response rate for melanoma. So really something classifying your diseases.

Keep in mind, very important paradigms, PD-L1 and TMB are absolutely not correlated. So they are independent biomarkers. The second thing is this across disease study from Samstein and colleagues, very nice study, showing you that basically you can use TMB across diseases, and it will predict overall survival on immunotherapy, being used as a continuous variable or even with a binary cutoff. And if you look at the binary cutoff in this publication they really take the upper layer, the upper 20% or 10% in terms of number of mutations. But very interesting. And this holds true if you remove melanoma and lung cancer, so really telling you that this paradigm should be validated but it’s probably variable across all cancer types.

And on the right-hand side is the expansion of the previous slide, when even you look at rare diseases you can find a correlation between the number of mutations, it’s a Foundation Medicine platform, and the response rate that has been described under immunotherapy. Okay? This is how it should be.

However, it’s not as simple, again, and you probably don’t use it on a daily basis because of some difficulties in establishing it. First of all, what do we know about TMB? We use it as a surrogate of the visibility of the tumor by the immune system, as a surrogate of the neoantigen load. It is defined by the total number of somatic mutations. However, there is still no consensus about what mutations should be counted. Is it only single nucleotide variants, indels, silent mutations, nonsynonymous mutations? We still don’t agree about it in the community. And of course you can measure these mutations using very different platforms, whole-exome sequencing, whole-genome sequencing, or some panels we use in daily practice. I quoted Foundation Medicine, but we have dozens of these platforms which could be used to count mutations.

Of course, and this is a very nice, again, review that you can go through, telling you all the limitations and all the discrepancies we have speaking about TMB. Keep in mind that all the preanalytical will influence the number of mutations you count. But also of course the depth of the sequencing, the type of mutation you count, I would say a footprint of the panel you look at. And of course the way you will report these mutations. So we have no consensus, meaning that you can still not really use that.

Of course, like PD-L1, remember, dynamic, heterogenous, it’s the same problem with TMB. TMB has also been shown to be heterogeneous in terms of special distribution, whereas if you look at 2 different regions of the tumor you might classify differently the tumor in 20% of the cases, which is of course a struggle. And of course we have also been showing that basically the panels should be large enough to be able to correlate it with the whole-exome sequencing, but most of the panels we use are.

Interestingly too, and probably for the future, you can use the blood to measure the TMB. And the whole-exome sequencing, as well as large panels, are more and more applied on blood samples, on serum samples, on circulating tumor DNA, which has been shown to be correlated with the tissue. That is the future of course.

If you look at the data we have, and this is about TMB in non-small cell lung cancer, again, we have been looking back to all of these trials comparing immunotherapy to chemo, looking at TMB.

There are many platforms of different methodologies used to measure TMB. We have been showing that basically that here you have the high and low TMB being evaluated.

TMB is predictive of PFS and also OS, but to a lower extent, across most of these trials, showing that probably when you use immunotherapy only TMB will be used in the future. It looks to be even stronger when your immunotherapy is not only anti-PD-1 or anti-PD-L1, but is combined with a priming-phase stimulant, and anti-CTLA4. On the left-hand side it was ipilimumab and nivolumab. On the right-hand side it was tremelimumab and durvalumab. And here the TMB was strongly predictive of the benefit of the immunotherapy versus chemo. So probably even better when you have a dual immunotherapy.

Unfortunately, what we do in daily life in lung cancer, when you start combining with chemo the problem is that you dilute the efficacy of your biomarker because of course TMB doesn’t predict chemo, and you lose the predictive ability of your TMB. Keep it in mind it has been shown in all trials too.

So this is about TMB. Can you combine both? I told you they are not correlated. The idea is maybe yes, and in many trials you can see in all of these graphs the upper curve for PFS or for OS is a double positive, high PD-L1 and high TMB. And this might hold true for other biomarkers to come, the double/triple positives might be the best group for immunotherapy.

If I move on here to the next slide, I just wanted to give you here a little word of caution. Not all highly-mutated tumors respond well to immunotherapy. And this is very important to keep it in mind. It’s many publications from the Columbia group, but also Charlie Swanton in the UK, having shown that what drives immune response is what we call clonal mutation, a very early mutation in the oncogenic process, the one you will find probably in all lesions, right? We call them clonal or truncal mutations. And the late mutations, the heterogeneous mutations, will probably not result in a very high sensitivity to immunotherapy.

So that’s why it’s very important to make sure that probably for TMB the older sample you have is the best to define TMB and not the most recent one. And also keep in mind that we will need some time to count the heterogeneity of mutations to make sure that TMB is accurately measured.

How are we going to move? I told you TMB cannot be used today in the daily practice, and the reason is the harmonization that we need to implement for TMB, and this is ongoing. There are 3 big groups, the Friends of Cancer Research, but also the German Group, and the TMB Working Group, which are all working on harmonization processes in order to make us use all the same definition of TMB in the future. And this will help with supporting the development of these biomarkers.

But in the US, TMB can be used to prescribe, and that’s quite important to keep it in mind, based on the KEYNOTE-158 Phase II study. The KEYNOTE-158 study is looking at high TMB. High TMB is defined as more than 10 mutations per megabase, specifically using the Foundation Medicine platform. And now you have understood that I need to be very accurate on that because of all the divergence you might have using another test, depending on the various definitions. In using this platform and this definition, a certain percent of the patients that have been looked at across diseases had a high TMB. So certain persons is still a large population of patients.

And I just want you to look at the table on the right-hand side, maybe just look at the partial responses. Using pembrolizumab in high TMB across diseases results in 25% response in high-TMB patients, and only 5% response rate in the non-high-TMB patients, a very dramatic difference, right?

Here is the waterfall plot, and on the right-hand side you can see the PFS on the upper graph, and the overall survival on the lower graph, really proposing to the community that TMB might help to guide treatment selection, at least in late lines of treatment, across diseases. And this has been FDA approved and looks like to be a very reliable way to prescribe pembrolizumab in a certain proportion of the patients.

Very importantly too is I have told you that some of the mutations which are not clonal may not give rise to a high immunosensitive phenotype. Some mutations look like to be giving rise to a very immunosensitive phenotype, and these are the ones which affect the DNA repair and the replication of the genes. And this you can see here, some of this lung cancer data, again, where some of these DNA repair genes being mutated leads to the generation of supersensitive tumors to immunotherapy, probably through the activation of what we call the STING pathway. And all the biological hypothesis that you can find in many publications has been showing how this activation of the STING pathway might be per se above the mutation count, a very strong predictor of immunotherapy sensitivity.

Why do I speak about it? It’s because there is a very specific circumstance of, I would say, somatic genetic characteristics, which is related to a high immunotherapy phenotype sensitivity. On this graph it summarizes the sum of the mutations and context, and you can see here on the left-hand side, on the green part, what are these genome-wide mutations which might characterize immunosensitivity.

And here is the next topic I’d like to cover with you. It’s what we call the mismatch repair, which we were seeing in green on the last slide. Mismatch repair genes are more than 1, there are 4 main ones, that you can see on this slide, which are DNA repair genes. When you lose the function of these genes it will result in alteration in the size of microsatellites and will result in what we call into microsatellite instability. This is a characteristic which was known for a while, for a long time, to be associated with the pure risk of developing a colorectal cancer. And this has been shown recently to be not only characterized by a change in the size of microsatellites, but also an accumulation in general of mutations and an increased in the tumor infiltrating lymphocytes seen in the tumor. So not only mutations, but also other definitions of immunosensitivity.

That’s the reason why we now characterize, even more than before, not only colorectal cancer between microsatellite instable or stable, but also other disease types, as I will show you in the next slide.

How frequent is the microsatellite instability across cancer types? So it’s not very frequent. You can see on this graph that you can find microsatellite instability as a rare event in many cancer types and as a frequent event in uterine endometrial carcinoma, and mainly in GI malignancies, being colon, being stomach, and being rectal adenocarcinoma. But again, keep in mind, it can rarely be found in other disease entities like lung cancer, like renal cancer, like bladder cancer, but only in a 1-digit kind of number percentage of patients.

But still, if you look at microsatellite instability across disease type, look at this waterfall plot with pembrolizumab treatment only. This is an amazing waterfall plot across diseases, with an interesting PFS and an amazing overall survival. So it might be a rare event, but it’s a very interesting event for immunotherapy.

This is not only about pembrolizumab. You can see on these tables that nivolumab was also evaluated as being a potential drug to be used in microsatellite instability, as well as nivolumab and ipilimumab, with even a higher response rate of 55%. So today pembrolizumab, as well as nivolumab and ipilimumab, were both granted approval for the treatment of any type of tumor with microsatellite instability, and particularly, of course, colorectal cancer.

I just wanted to show you the most recent data, which was establishing the relevance specifically in advanced colorectal cancer. What you can see here the comparison of pembrolizumab versus investigator’s choice chemotherapy in naïve patients, in patients of course with microsatellite instability, so really is evidence 1 Level that we’d like to have. And this is the PFS curve, where you can see that in MSI colorectal cancer pembrolizumab is better than chemotherapy up front, with a very amazing nonproportional hazard ratio to be seen there.

And of course, like TMB and like PD-L1 to a lesser extent, how to measure microsatellite instability needs a consensus, which now exists. I invite you to go through some of these articles defining the consensus we gave as ESMO recommendations on how to measure microsatellite instability in a reproducible and accurate, manner which can be done in the daily practice elsewhere in the world.

Last but not least exciting biomarkers that everybody speaks about, the microbiome. There is a significant microbial contribution in the total makeup of our cellular composition and function, right? What was described by The Economist as “Microbes maketh man”. Probably yes. And it has been shown in immunotherapy in many publications now that the microbiome you have in the gut, some of this bacteria, we defined in green, a very high response rate to immunotherapy. And some of them, in red, will define a poor response to immunotherapy.

And of course you usually have a mix of them. And what was shown today is a diversity of the microbiome is probably the best biomarker for a positive response to immunotherapy. So diversity of your microbiome. And you can see in the middle graph with the high diversity performing way better than the intermediate of low diversity of the microbiome. And on the other hand, the other way around, if you treat a patient, and here’s this non-small cell lung cancer patient, with a lot of antibiotics you will reduce the diversity of the microbiome, and the response to immunotherapy, anti-PD-1 or anti-PD-L1, will be lower.

So microbiome, very exciting, very simple paradigm. What do we know today? So a broad diversity in microbiome is associated with better outcome with IO. The presence or the absence of certain bacterial strains will be specifically associated with response or nonresponse to IO. The use of probiotics, and here I speak about the one you buy in the supermarket, the one you consume probably at breakfast, has been roughly shown to reduce the diversity, probably even more if you abuse these kinds of compounds. So be careful. You shouldn’t drink too much of them.

The use of antibiotics is negatively influencing the diversity. The fibers; remember the fibers. So fibers are associated with a more diverse microbiome, probably explaining how they can prevent colorectal cancer and influence the outcome under immunotherapy. And last but not least, probably, the proton pump inhibitors might also result in the reduction in the microbiome diversity and a lower response rate to immunotherapy.

But of course it’s a moving field. What can you do? We can of course use fecal transplants, meaning you can use a microbiome put into caps of patients who respond well to immunotherapy. We can design microbial consortia and give them to patients under immunotherapy. You can try to use good diet and good supplements that would be medically led, and I would say built, in order to support immunotherapy. And of course you can try to target, by antibiotics, the wrong strains for immunotherapy. And all of these trials are, you will be surprised, ongoing now in specific centers, particularly on the West Coast of the US.

Case: A man in his late 50s with metastatic pleomorphic carcinoma of the lung and a high tumor mutational burden (TMB) attains a complete response to pembrolizumab

PROF PETERS: This is about the biomarkers. I’d like to discuss with you PD-L1, TMB, and microbiome, which probably are the closer to our daily practice and of course PD-L1 being already there.

I’d like to present you a case, just to end the story of biomarkers. This was my patient, of course, with lung cancer, my specialty. 59-year-old man diagnosed with a very bad tumor, which is a pleomorphic carcinoma, usually very aggressive, of the right upper lobe. The initial stage was cT3, a very big tumor, with probably lymph node invasion at least at the hilum, but probably in the mediastinal ipsilateral part too, and metastases which could be seen in the bone and in the adrenal.

The patient received first cisplatin and gemcitabine, which was at the time, in 2016, the standard of care, for 3 cycles. At restaging we were a little disappointed to see only a partial response, and you will see the picture here for you. The PET/CT you can see this huge tumor with some metastatic pattern which remains quite stable after 3 cycles of chemotherapy.

This being said, we decided because of the very, very big tumor, to do an upper lobe resection, that we could potentially help this patient in terms of symptoms, and to irradiate the remaining adrenal and bone lesions in order to try to treat this patient as an oligometastatic patient. So we did that, with some very aggressive treatment.

Unfortunately, and as is sometimes observed, at the next restaging after surgery there was a subsequent progression in the bone and in the lymph nodes. So the oligometastatic had become even more metastatic. You can see here that there was a lesion, for example, in the sacrum, a lesion paravertebral, and of course some nodal. In the thorax you can see some nodal invasion that was not seen before.

So of course some kind of failure of our strategy. So we went back to the pathologist asking for PD-L1, now it was time, and TMB. We are very kind of progressive in the way we try to implement TMB in many Swiss centers. We could find a very low PD-L1, 1% is not satisfactory, but a very high TMB of 15 mutations per megabase, which might be frequently observed in these poorly differentiated tumors.

So basically based on that we did this bet of giving pembrolizumab monotherapy in this patient who was not so much responsive to chemo. We used the usual regimen of 200 mg every 3 weeks, and this is what we observed. Unfortunately, after 2 cycles this patient presented with a dramatic immunotherapy-induced pneumonitis. As you can see here, the picture was impressive, the saturation went very low, and the patient as doing very badly with an ECOG performance status of 3 to 4, only in the bed because of pneumonitis.

So we had to go with very high dose of prednisone, as you can see, for a long period of time, covered in the counterpart by large-spectrum antibiotics. At that time we decided to stop immunotherapy, demand that we discontinue immunotherapy based on that, and the patient recovered.

And what is interesting now is restaging 2 months after complete resolution of symptoms has shown a complete response. And what is even more interesting is we met this patients in December last year. This patient is still today in complete response after 2 cycles of pembrolizumab and a massive pneumonitis related to this immunotherapy. So just showing that based on this paradigm, probably on TMB, we could define a memory response against neoantigens, which still is lasting in these patients.

With this, it’s my great pleasure to thank you for your kind attention.

Role of TMB in determining response to immune checkpoint inhibitors; FDA approval of pembrolizumab for TMB-high (≥10 mut/Mb) solid tumors and implications for clinical practice

DR LOVE: So just reflecting on the last case that you presented, with the low PD-L1 but TMB of 15, if you kind of look at the literature and try to find patients like that, low PD-L1, TMB in the range of 15, what would you say globally the response rate is?

PROF PETERS: It’s difficult to answer that because of this lack of definitive definition of TMB. But of course when you start to describe such high TMBs you can be sure that whatever you use as a definition you probably look at the range of tumor with a very high load of neoantigen, and without doing too much details, probably looking at immunogenic tumors. So with this patient you probably reach, what was described in the nice KEYNOTE-158 trial, you probably might expect in this patient a response rate between 20% and 40% of the patients more or less. I would say 20% to 40%.

But more importantly, the fact that TMB’s a surrogate of neoantigen this is also potentially predictive of memory T cells, of the capability of getting in this patient a long-term benefit. So of course we need more data and more homogeneous definition of TMB. But I think this patient — and that’s why you are very lucky in the US with this FDA-granted approval for high TMB, that’s exactly what we need to demonstrate treating more patients because they are rare. So this absolute long-term benefit is related to this neoantigen surrogate.

DR LOVE: I’ve got to say, when I saw that FDA approval come through, and then I started to ask people about it, it was kind of a little difficult for me to figure out exactly how it would apply. I mean you have some situations where you already have a lot of data available on the effect of immunotherapy, say lung cancer for example, where you have a paradigm already established but with PD-L1 for example, but you have cases like this where maybe it’s relevant. But there are a lot more cancers where we just don’t have a lot of data period, sarcomas, whatever.

So in a situation like — that was the first thing I thought about, unusual tumors, you run out of options. Half the time people are considering immunotherapy anyhow, just because whatever, it’s on TV, and the patients want it. And at least now you have maybe a way to pick the patients. So how do you see this really playing out clinically in these various kinds of tumors?

PROF PETERS: So first of all, remember these people from Samstein who were showing that this TMB is the defining benefit from immunotherapy across cancer types. Even if you remove the usual suspects, lung cancer and melanoma, across cancer types as being a continuous variable using threshold, it predicts efficacy. Though probably think about this TMB using used to prescribe pembrolizumab across diseases, not always being the front-line strategy, but it can be a second or third line once you have exhausted other treatment options.

By the way, very often in some centers you need time to have the result of TMB because it might not be the first biomarker you look at. So maybe we need to think about this paradigm as being used maybe in front line in some diseases, like lung cancer, but also maybe as second and third line in other disease entities. But please keep in mind that the aim of such a strategy is not to improve 1 month overall survival, but to potentially get for a small proportion of these patients something like corresponding to a control of metastatic disease. So I understand the idea thought about offering it to patients. And again, remember it’s a certain percent of the patients, with a small group across diseases, offering them the opportunity of potentially getting long-term control of a deadly disease.

As prescribers I think we probably have the duty to try and in that circumstance, provided it is reimbursed and accessible, to prescribe pembrolizumab there. But the other duty you have is to make sure that it works. So after 2 or 3 cycles make sure that this paradigm of 30% to 40% of them responding is the context of your patient. But still, we were giving chemo to some patients with a way lower response rate until today, right? So I think we need to revisit that also in perspective of what you have been doing in the last decades.

DR LOVE: And of course really what we’re looking for is not response rate but prolonged response rate; triple, homerun, whatever.

Assessment of TMB as a biomarker for use of immunotherapy; variation of TMB over time

DR LOVE: What do we know about, first of all, liquid assessment of TMB, and also TMB over time? I would guess that it doesn’t change with tumor bulk, but I don’t know. You tell me. What do we know about TMB over time?

PROF PETERS: So very nicely, the TMB measured in the serum looks like to correlate quite well with the tissue TMB. And of course both have flaws. Tissue TMB very often meets the limitation of not having enough tissue. And this is really met. If you don’t really go for it, maybe half of your patients will not have enough tissue for measuring TMB. But keep in mind that it’s the same for blood. In potentially 20% of the patients you can consider the patients to be called nonshedders, or insufficient shedders, meaning you don’t find enough circulating DNA to do the TMB on that.

So both of the consulting might be met. You might not have the TMB measurement in the tissue, but you also might not have it in the blood. But this being said, they correlate quite well, and probably what you have been showing, the one taken in the blood, the blood TMB, is strongly usable and accessible for the predictive ability in order to prescribe and decide for immunotherapy. So if you are asking me what I bet for the future, it’s really probably to use more of the serum, the blood, to look at TMB. Because of feasibility, reproducibility.

I don’t think there is very important variation over time. And again, remember this clonal TMB. The mutations which gives rise to mounting a T-cell response are the ones which are truncal and clonal. So probably the earliest to get it, being in tissue or in blood, will be your TMB value for the future. So I don’t think there’s any value in repeating that measurement. Take the earliest you have, and you will look at your truncal mutations, okay?

DR LOVE: So for people like me who aren’t too good with pathways and stuff, can you go back through what the truncal thing was? The slide looked very interesting.

PROF PETERS: So basically when you look at the tumor and the metastasis resulting from the tumor, initially you have the chance to sequence multiple metastases, as well as the primary tumor. There are some mutations which you will find in each of the lesions, which probably were in the primary and did spread over in all the lesions. They are truncal. They are even probably part of the oncogenic process. They are probably oncogenes, right, which have led to the malignant phenotype.

On the other hand, by this divergence of the clones there are some mutations you will only find in 1 or 2 of the lesions, the late lesions. These ones have not been shown to be able to result in any immune response. The ones resulting in immune response are the ones which are in the majority of the lesions, if not all, which have composed the oncogenic signature of the tumor. So the ones you will find wherever you look at. And this can be measured very often by the type of mutations, but also the quantity of mutations, the quantification of these mutations in 1 tumor sample.

DR LOVE: Wow. That’s really interesting. I mean I don’t know, maybe there are autopsy studies that have been done. But when you look at multiple metastases you’re saying you might see different TMB levels in different metastases?

PROF PETERS: Yeah, yeah. So unfortunately, again, if you look at the primary and metastases, probably 30% of them will have, if you classify more than 10/less than 10, 30% of them will have different results in the metastases as compared to the primary tumor. And you will guess that very often you see more mutations in a late lesion because they accumulate over time, though you usually find more mutations in the metastases.

So that’s why I say if you can, use an old archival sample. And if you can, use a sample of the primary. This will probably define to you the most accurate number for the TMB.

Correlation between TMB or PD-L1 level and smoking status; role of POLE mutations and the STING pathway in the immune response

DR LOVE: I was fascinated by one of the slides you showed in terms of TMB and response, breaking out non-small cell lung cancer versus smokers and nonsmokers. Can you talk a little bit about that? What you think it means? And also PD-L1 levels in smokers and nonsmokers.

PROF PETERS: In all of these randomized trials with large numbers, in forest plots we have been showing that immunotherapy works better in smoker patients, obviously. And on the other hand, it doesn’t work very well in never smokers. So what is it correlated with? So very probably, and it has been shown in many series now, the TMB, the tumor mutational burden, is correlating with the smoking habit. And they are even called signatures. So these are the signatures qualifying the aggression which happens at the DNA level.

This same signature can be found in some types of breast cancer, the APOBEC signature. It can be found in melanoma with the UV light signature, and in smokers with a smoking signature. This is repetitive changes in the nucleotide which are related to the aggression that you perform on your genetic information. And of course this happens in smoking habit, in bladder cancer, in lung cancer, as being the best examples.

So this is correlated. What about PD-L1?

You’ve been seeing that PD-L1 is absolutely not correlated with TMB. You can have a very high TMB and a low PD-L1, and a very high PD-L1 and a low TMB. And PD-L1 shouldn’t be thought to be correlated with smoking habit. It can be completely divergent to smoking habit and can be also representing some other mechanism of immune reaction to the environment,

PD-L1 is more characterizing that something immune is happening in your tumor, what we called in the past the interferon gamma signature. And it doesn’t mean that it’s about smoking or about mutations,

So smoking correlates with neoantigens correlated with TMB, but not with PD-L1.

DR LOVE: So just to add in another piece of the puzzle that you mentioned, briefly, that I hear a lot of oncologists asking about, which is POLE. What it is, how you pick it up, what the implications are, and where it fits in with the spectrum with, for example, MSI high.

PROF PETERS: So it’s difficult to explain, but all POLE is a good example that there are many of these DNA repairing mechanisms which can be altered, can be changed by a mutation. POLE is one. So what happens when your DNA repair is not functional, or insufficiently functional? You accumulate mutations. It’s the same thing with mismatch repair, when you have a mismatch repair what is happening, you are accumulating mutations, particularly affecting microsatellite, but you are accumulating mutations.

So all of these deficiencies in DNA repair is just about accumulating visible mutations, so increasing the TMB. What is interesting, on the STING pathway, is in addition to accumulating mutations there is an intrinsic mechanism in the cell which signals to the immune system that the cell is abnormal. So in addition you have an added layer of immunostimulation which happens through the STING pathway. But basically DNA repair, if it is deficient, leads to a higher number of neoantigens.

DR LOVE: And again, can you just kind of go a little bit more into what the STING pathway is?

PROF PETERS: Yeah. So the STING pathway is one of the reactions which happens in the cell when some DNA alteration is seen, right? This is a way to signal to the same, usually some apoptotic mechanism, but also to signal to the surroundings of the cell that something abnormal is happening. So it’s a very strong stimulant for all the mechanistic around the immune response to happen, or potentially if only it would be possible for the cell to die. But as we all know, cancer cells usually resist apoptosis quite well.

Rationale for the investigation of PARP inhibitors in combination with immunotherapy

DR LOVE: So of course any time we hear the phrase DNA repair we think about PARP inhibitors, and I know I see trials combining IOs and PARP inhibitors. But is that in any way tied into these pathways that you’re talking about? How does that — like if you introduce a PARP inhibitor into a patient with a DNA repair issue, how does it affect response to immunotherapy? We know maybe it’ll help the tumor, but does it affect the response to immunotherapy?

PROF PETERS: To try to make this story not too complex, the PARP pathway, I would say the PARP machinery, is here to compensate when you have a DNA repair deficiency. So it’s kind of the salvage solution you have. So when you have a DNA repair you start to accumulate mutations, and the PARP is here to try to still catch them, to try to still repair. So we call it synthetic lethality. If in addition the PARP is deficient, then it’s a catastrophe. The cell might die, or in case of cancer cells it might become even more sensitive to immunotherapy and intervention because more mutations accumulates, right?

So it’s really a salvage strategy to repair the genetic information. And if you have the second hit on the PARP, being genetically defined or being defined by a drug which will inhibit the PARP, you add on the layer effect, and on this path which might still lead to cancer cell deaths, right?

DR LOVE: So it sounds like there’s maybe a translational basis then for combining PARP inhibition and IOs.

PROF PETERS: Yeah. Absolutely. A little bit like we use PARP inhibition in tumors which have already DNA repair deficiency, like BRCA1. We have a DNA repair deficiency, and then you add the PARP inhibitor to make the cell die. So you could do the same thing, yes, hoping really for more mutations to be visible. Absolutely.

DR LOVE: I was doing a session with Rob Coleman, and we were talking about this MEDIOLA study presented at ESMO in, it was actually ovary, where they combined PARP and IO, and there was a tie into, I don’t know if you’re familiar with those data, but it looked like I think the patients with the DNA repair did better. Have you seen that?

PROF PETERS: Yeah. So basically what is quite interesting in ovarian cancer is many of these tumors, a large proportion of them as compared to other cancer types, are characterized already by some kind of recombination deficiency, so DNA repair deficiency, which means they already accumulated mutations. So in these tumors if you shut down the PARP that’s a catastrophe for the cancer cells, which might immediately die from that. We call it, again, synthetic lethality.

But if on top of that you add immunotherapy you might really result in a very, very strong signal against these cancer cells, right? So you really accumulate all the mechanisms which make any kind of compensation, any kind of survivor from the cancer cells, impossible.

DR LOVE: So a couple other maybe side questions. Microsatellite status, like what’s the satellite? What is the satellite? Where is it? I never thought about it, but like what’s it mean?

PROF PETERS: The satellites are sequencing which are found in the chromosome at the end of genes which are very repetitive. And by the way, they are here to protect the integrity of the gene between these microsatellites because it helps that this gene is very, I would say, consistently reproduced by cell division and consistently accurately expressed as proteins, right? So when you start to decrease the microsatellites, reduce them because they are instable, the risk is to damage the gene itself and to result in mutations in protein, right? So by this mismatch repair you create microsatellite instability, which after a certain period of time will result in mutations being observed by a wrong translation of the genes.

DR LOVE: But again, it’s like the satellite is an actual entity?

PROF PETERS: Yeah. They are repetitive sequencing in the genes. They are small repetitive stretches of DNA.

DR LOVE: Interesting.

Efficacy of cemiplimab versus chemotherapy for PD-L1-high advanced non-small cell lung cancer (NSCLC)

DR LOVE: Another question, I saw your slide there about cemiplimab.

PROF PETERS: It’s data presented at the ESMO

They were reproducing the KEYNOTE-024 data with pembrolizumab in high PD-L1, more than 50%. Basically it’s a knockoff trial, right? They were reproducing the same thing, more than 50% PD-L1 on tumor cells, cemiplimab versus chemotherapy in advanced non-small cell lung cancer. They show a hazard ratio which is even a little better than the one with pembrolizumab, but of course with a shorter follow up. The reason is in that trial they only, you will be interested with that, they only selected smoker patients. They got rid of the usual 15% we have who are never smokers. So one of the inclusion criteria was only a previous history of active time with cigarettes.

PROF PETERS: So that’s why the results have a slightly better hazard ratio as compared to KEYNOTE-024.

PROF PETERS: Which we understand of course.

DR LOVE: I actually wrote down on my pad “cigarettes” because I saw that little icon you put up there. I was curious of the cigarettes. So that’s really interesting. Just out of curiosity, are there data on chemo plus cemiplimab?

PROF PETERS: They have data on chemo/cemiplimab, yes. This trial is ongoing, right?

Diversity of the gut microbiome as a potential predictor of response to immune checkpoint inhibitors

DR LOVE: Well of course we have to talk about the microbiome, for sure. And maybe you can just talk a little bit and kind of summarize the bottom line right now in terms of what we know from a research perspective, but also whether there’s anything there that’s clinically relevant, issues about antibiotics and stuff. Anything that right now, maybe it’s not perfect science, but that clinicians might want to think about?

PROF PETERS: It’s not perfect science, but as you can see, right, it’s very reproducible. There’s a few data. There are a few very big publications, but they all show the same thing, right, is some bacteria are promoting the presentation of neoantigens to the immune system, so promoting immunotherapy, and some do the other way around. So basically it looks like a little complex to go bacteria by bacteria now because there’s hundreds of them, right? But what might be used today is the concept of diversity of the microbiome. It’s a little bit like mothers and kids in the garden, right. When you had the kids you would always ask did they put anything in the mouth because — and remember my grandmother was telling me, yeah, it’s good for her immunity. So she was right. It’s good for her immunity. So the more diverse or more strange things you eat, right, the more diversity you’ll have, crude vegetables and so on, and the more potentially you might have a strong immune response.

So the diversity is what you can use. So what does it mean in practice? Everything which abusively reduces the diversity, which means giving antibiotics for nothing, giving proton pump inhibitors for nothing. I don’t know about the US, but in Europe we use a lot of proton pump inhibitors without strong indications, right? Same thing for antibiotics. We use antibiotics sometimes without a strong indication. So all of this should be avoided.

Everything which really minimizes the diversity and probably more in the US, but these probiotics that you can buy in supermarkets, some of your colleagues probably do not drink 1 per day, but 5 or 6, which probably is not good. Remember these little kind of cubes, you can keep them in your car, in the warmth for 12 hours, and they are always good, meaning that these bacteria are very numerous but very resistant too. So it’s probably not good for the diversity of your microbiome at the end of the exercise, right?

So all these kinds of things should be taken into account. And probably the fibers, the famous fibers that we speak about since decades, is also a way to increase the diversity of the bacteria which can attach to that, right, and potentially increase the diversity. So that’s what we know today. What will be tomorrow, probably caps. Caps with bacteria inside. Will it be a composition of bacteria coming from a patient composing 2 caps and given to another patient? Yes, maybe to start. But I think in the future it will be a nice combination of various strains, very well selected, of bacteria. And this is ongoing today.

DR LOVE: So could you maybe just track out a little bit about how fiber leads to diversity and how proton pumps reduce diversity?

PROF PETERS: Well, for the fibers, nobody knows. We don’t know if it’s really about the fibers increasing the bacteria kind of diversity or because of the nutriments which have fibers increase the diversity of what you eat. Because when you are asked to eat 5 fruits per day it’s very probable you will not eat 5 of the same fruits, right, so maybe to answer your question on diversity. So this hasn’t been explained.

For the proton pump, it’s probably because of the change of the general pH or your gut, right? So you modify probably how the whole phenomenon of digestion happens. So probably you interfere with the survival of some strains of bacteria.

DR LOVE: Of course everybody’s so interested in the biome, but I was kind of flashing back on a conversation I had with Keith Flaherty. I don’t know if you’ve ever worked with him, and he was talking about in the gut kind of the immune mechanisms that protect from the bacteria getting out of the gut. And he was talking about the same thing in the liver. Have you heard that kind of discussion? And how does that relate to this idea of the relationship between the microbiome and immunotherapy?

PROF PETERS: Well, it’s interesting because we’ve focused a lot of gut microbiome, which he will agree with me in terms of lung cancer immunotherapy looks like to be a strange concept, right? Why are you going to the gut? Probably because we have always been under evaluating the importance of the gut. The gut is probably the main factory for the immune system to be mounting against, I would say, aggressive factors coming from the outside. So it’s quite important that the huge concentration of the Peyer’s patches, right, which is the place where antigens meet the immune system or at least stimulate the T cells to travel in the body. So here’s a place where cytokines happen, and the whole, I would say, immune stimulation might start. Not only there, but probably very often there. So it’s really about the factory of the immune system.

But I agree with you that we need to move ahead with the microbiota, the microbiome, in other organs. And you probably have been reading that the initial publications about the colonies of bacteria present in the lung too, because it might be interesting to see how the bronchial tree can potentially itself generate various responses to immunotherapy and the various environments of macrophages, M1 and M2 and T cells, in the specific organs. So I think the next step we will have to understand if the microbiome diversity in the gut is superimposable with the microbiome in the lung and how each of them might specifically interfere with, for example, a lung cancer immunotherapy.

This is really ongoing, but you can guess that going to look at the microbiome in the lung is quite invasive, so it takes a little bit more of effort or circumstances to be able to get data about it.

Effect of PD-1/PD-L1 expression on the immune response; assays for the assessment of PD-L1 status

DR LOVE: So I want to rewind a little bit to the part of your talk where you were talking about PD-L1 assays and ask, I want you to imagine that you have an oncology fellow, the fellow’s about to start general oncology practice, and says, “What kind of PD-L1 assays are out there that I’m going to have to consider for lung cancer, melanoma, gastric cancer?” You had a great slide there. “How many different PD-L1 assays am I going to have to know about and utilize in my practice?”

PROF PETERS: Yeah. So the first thing I would tell to my fellow is keep in mind that the interaction of PD-1 and PD-L1 is switching off the immune response in the tumor. This key, this PD-1/PD-L1 interaction, can happen very often between the tumor cell and the lymphocyte, but can happen in different kinds of compositions, right, between the immune cells and the tumor cells.

So keep in mind that the importance of PD-L1 on the tumor cell versus on the immune cell, same thing with PD-1, might be variable in each tumor type. But this interaction, whatever the composition, will switch off the immune response and probably lead to the apoptosis of the T-cell component, right? So keep in mind that measuring PD-L1 can be a good predictive biomarker for immunotherapy, but it can and has to be measured differently, depending on the tumor type you look at.

And I would show him my table, which changes every month, which tells you how best to measure PD-L1 in every disease entity. Not only about the cell type, but also on the threshold, which has been shown to define a significant difference between benefit and no benefit, response and no response. And this is of course a moving field, right? But I think it just speaks to us about the diversity of the mechanism which can be mounted by a cancer in order to avoid the immune system.

DR LOVE: Yeah, I mean right now oncologists are stuck with trying to figure out what trial was it that got the indication? What does the FDA say? Which PD-L1 assay? What’s the cutoff? I was just flashing on this meeting in gastroesophageal cancer, and we must have spent a half an hour going through each different study, they did different PD-L1 assays in each one, and at the end of that I’m like I’m still not understanding what these people in practice are supposed to do. Are they going to get that assay in second line, but this assay in first line? Do you think that we’re moving towards more like 1 assay in the future?

PROF PETERS: Unfortunately not. I think we are measuring there the immune escape. And the way you measure this immune escape is not the same in each tumor type because I think it happens differently. That’s also true because remember the PD-L1 is not at all a biomarker in some cancer types. So meaning that the immune escape has various mechanisms, which can be described by differential expression of PD-L1 on different cells, but probably also by other things, TMB and probably, you know in the future we’ll look at TIM-3, we’ll look at LAG3.

There might be other checkpoints which also characterize this immune responsiveness or this immune escape.

So I think, unfortunately, by knowing more we might make this story more accurate but more diverse from one cancer type to the other cancer type. The more I move in oncology the more I think that it’s difficult to remain the oncologist of all cancer types. It’s not not feasible, but it’s very difficult, because the more we learn the more granular we can see the differences between each cancer type.

DR LOVE: 100% true, but there’s such an advantage, though, when you have general oncologists out there.

Activity of the anti-TIGIT antibody tiragolumab in combination with atezolizumab for PD-L1-high advanced NSCLC; potential role of neoantigen burden as a predictor of response to immunotherapy

DR LOVE: I did want to ask you about other checkpoints and other predictors that are being looked at.

I think there was a trial recently, where they added a second different type of checkpoint inhibitor and they had positive results.

PROF PETERS: I know what you mean. In probably it’s lung cancer, you used another important checkpoint inhibitor, which is the TGIT. TGIT is another checkpoint which has been shown to switch off not only T cells, but also NK cells in the tumor, which might be interesting because some tumors are depending on NK cells, like small cell lung cancer, for example. So it’s quite interesting, and in that trial they combined atezolizumab and 1 anti-TGIT, which is called tiragolumab, versus atezolizumab in patients naïve from treatment, but with positive more or equal to 1% PD-L1 on tumor cells. And they could show that TGIT and atezolizumab was way better than atezolizumab. But interestingly they were showing this benefit was only seen in very high PD-L1 again, so in a tumor with more than 50% PD-L1. And this is now in Phase III, called SKYSCRAPER-01, where they compare atezo and tiragolumab versus atezo as front-line treatment in high PD-L1.

So no new biomarker there, it’s still PD-L1, probably because I would guess the TGIT expression very often moves parallel to PD-L1. And we still are not very good in making good immunohistochemistry for TGIT. So I would guess the biomarker might be TGIT, but we are still not really able to measure it.

The other biomarkers to come, where there are many, but there are 2 which I like. First of all, we need to understand more about other cells which are present in the tumor. So what we call the Ts, the T-cell infiltration, right? It’s not only about T cells, it’s what kind of T-cell is there. What kind of other cells, like macrophages and natural-killer cells? By defining you can do it by multispectral immunofluorescence, right? You could go to do immunohistochemistry or immunofluorescence of many cell types to describe who are the actors here in the field of the tumor that you can activate. So that’s a very important topic.

The other thing is probably going beyond TMB. TMB is a little kind of rough counting of something you ignore. We are more and more in the field where by looking at the mutation you can predict which ones are really neoantigens, which ones will be presenting on your major histocompatibility complex through the T cells.

So I think we have to move from TMB to neoantigens. And this is interesting because once you know about neoantigens it’s about vaccines, it’s about CAR T cells, it’s about T-cell therapy. So when you know what other neoantigens you can develop many other strategies. So I think we should leave TMB at some point, maybe in some years, to look at what are the visible neoantigens of the immune system.

DR LOVE: You had that 1 slide looking at PD-L1 and TMB, and I think you used the term double positive.

I was reflecting, I remember when we first started with targeted therapy at one point they were talking about triple-negative: ALK, EGFR, ROS1. And I was flashing, we’ve been doing to San Antonio Breast Cancer Symposium for like 20 years, but last year we did 1 symposium on HER2, 1 on triple negative, 1 on ER positive.

It’s like 3 different diseases. Do you think that kind of we’re moving in that direction in lung cancer except maybe there’s going to be 30 subsets or something?

PROF PETERS: Yeah, I think so. What is quite difficult for oncologists, even for PD-L1, even if you have a negative PD-L1 status, still you will have this patient who received pembrolizumab and is alive 5 years from now. So the problem is a negative predictive value. All our biomarkers have a good positive predictive value. They predict the benefits, but we still don’t have any good biomarker predicting the absence of benefit. And probably it’s not going to be able to be single negative, but a double or a triple negative.

For example, in lung cancer now if you have a low TMB and a negative PD-L1, then you will not be fine with pembro only. You need something else than pembro only. But in all cancer types I think the status of absolutely nothing allowing for an immune response to happen will be something to define, it will be double negative, triple negative. So I think we have more promises in excluding preventing these patients to be exposed to immunotherapy only then the other way around.

And keep in mind we have been seeing it. Remember in most of the trials using only immunotherapy you have what we hate is this early crossing of the curves, what we call the early excessive death. In the first 3 to 6 months of treatment there are patients in the immunotherapy arm that immediately progress and die. So for these patients you have done a mistake. You have probably led to some harm because this patient would have probably been alive with chemo, and by giving IO they have progressed and died very fast. So they are probably the double negative, triple negative, the ones that you wouldn’t like to expose to immunotherapy, and not only for economic reasons, for efficacy reasons.

DR LOVE: I mean I’m assuming that they didn’t progress any faster, it’s just they didn’t have the benefit of chemo.

PROF PETERS: Absolutely. The standard arm would have been a better choice. These are the patients for whom the standard arm would have been a better choice.

DR LOVE: Right. So again, I’m just kind of thinking about, I guess there’s not going to be much alternative then, as you were saying, then just kind of knowing the data on each cancer.

Anything else you want to add to what we’ve said today? Any thoughts that popped into your mind? Any, I’m not going to say mistakes, but I know you see a lot of second opinions, things that people do in practice related to this that maybe they ought to rethink?

PROF PETERS: No. The only thing I’d like to stress is we should be careful not to make any final decisions on a biomarker until you exactly know what the biomarker means. We will be seeing in the future probably dozens of trials with TMB. But always question how was TMB measured. Does it correspond to the other set of data you’ve been seeing? What might be the limitations? So it’s very difficult because you ask general oncologists, community oncologists, to have this kind of critical point of view, telling it’s not a simple trial. It’s an unusual biomarker which needs to be analyzed with caution, right? And I think for PD-L1 we have been seeing it, right? For many tumor types we have been telling oh, PD-L1 is of no value. And now we see some value, but you need to look at it differently.

So for TMB it will be the same. When you see the data coming you have to be sure how was it looked at, with all the details of the methodology behind, and where do we go from there? Because sometimes by making short ways we give up too early, right, for a concept which biologically makes sense. So we need to continue this work, and it will take some years.

Consideration of the benefits and risks of immunotherapy for cancer

DR LOVE: So one final comment, just curious what you think about it.

About 6 months ago we had a case from a general community oncologist of a patient with metastatic non-small cell, PD-L1 zero, but a history of multiple sclerosis and was treated. But okay, stable, and of course he’s agonizing. Now the woman goes through second-line chemo, very intelligent, goes for a second opinion, everybody’s like are we going to really try this PD-L1 in this woman, particularly because her PD-L1 level was negative. So finally the patient said let’s do it. I don’t have to tell you what happened. She had a great response, and she’s still in response. The MS stayed the same. No problem with the MS, et cetera. But just amazing. It’s important, as you said, to remember that you do see these patients who respond in spite of that.

PROF PETERS: Remember in the very early clinical trials because we are of course the pharma industry we are very cautious in how to use. The worst for a compound is to kill patients, right? So if you start with a broad inclusion criteria kind of panel, then the risk is that some of the patients do very badly, and then you stop the developments, and you hold the developments. And this shouldn’t happen.

So in the first trial patients with brain mets, patients with tuberculosis, patients with HIV, patients with hepatitis, patients with any kind of infectious disease, patients with any history of previous autoimmune diseases were all excluded. And that was the, I would say, evidence-based trials which defined a new landscape. But now we come from there and start to enlarge the benefit we have been observing to additional communities. And now it’s kind of a paradigm. Look at hepatitis. We now know that hepatocarcinoma, which comes from hepatitis kind of inflammation, do better under immunotherapy than the ones which are related to alcohol. So it’s interesting to see that what was an exclusion criteria is almost now an inclusion criteria for immunotherapy in hepatocarcinoma.

You can treat some stable autoimmune diseases with checkpoints, but you need to discuss with your patient. You can treat HIV patients with checkpoints, but you need to discuss with your patients. So all these kinds of things need a second look, but of course I understand the strategy to start, which is about bringing new strategies to the market. And we also benefit from that, the patient benefits from that.

But now we are revisiting all that. And this MS patient is a good example. At some point you sit with your patient and you say these are the risks and these are the potential benefits, what are we doing now? And we are creating kind of real-world evidence around that, right?

DR LOVE: Well the thing I was thinking about is everybody focuses on the issue of can you safely give it? Are they going to have their autoimmune problem get worse? But I was just kind of curious about what the efficacy is of immunotherapy in these patients who have already some type of alteration in their immune system. I could imagine they would respond better, putting aside the toxicity issue.

PROF PETERS: Yeah. There are some data showing that the few patients who are characterized by a too strong immune system, right, an immune system which tends to be activated for nothing, do a little better on immunotherapy. But of course you will imagine these are only retrospective matched series because nobody randomized these patients in any trial because no ethical committee would ever accept it, right?

But it looks like this very strong immune system might also be more reactive to immunotherapy. But it’s also in line with what we know about unexpected toxicities. When you treat a thousand patients with immunotherapy and maybe 50 of them will present with Grade 3 and 4 toxicities, unfortunately they have diarrhea, pneumonitis, but these patients, like in my case, do amazingly well. And maybe they were some kind of super-immune patients, basically, and they developed autoimmune responses, which also leads to antitumor responses, right? And this has been shown in multiple publications now. Maybe 10 in lung cancer with very strong data. So toxicity’s bad, but we tend to like it in lung cancer, at least.

DR LOVE: No. I mean when I saw that slide pop of the pneumonitis I was like probably this patient’s — I mean I don’t know if there’s that much data to support it, but you hear so many cases of people who get 1 cycle, they have this horrendous autoimmune problem, and then they stay in remission for a couple years.

PROF PETERS: Yeah. And the refined strategy now is probably to try to stay away, for the quality of life of your patients, to the very high dose of steroids, right? What is very difficult for patients is to start to have all the side effects of steroids. So now you can be a bit more refined. We have published quite a lot on that with the immunologists here around to measure the cytokines and to try to more personalize anti-immunity kind of treatment or immunosuppressive treatment.

Imagine you have a huge rise in IL-6, which you observe quite often in diarrhea, but also when you have this arthritis. So you can use tocilizumab, right, and potentially try to shorten or minimize the amount of steroids you might use. Sometimes TNF is the mechanism in diarrhea. Using early infliximab might help you giving less steroids. So all of these kinds of things are promoting some better way to address this toxicity without the side effects and also potentially leaving a little more room for T cells to survive, right?

So we need to, that’s probably one of the points, to refine how to address the toxicities. And that’s a very interesting field.

DR LOVE: We’re presenting so many cases nowadays it’s just hard to believe. But I just flashed on an incredible case that makes the point you made we had of a patient who got a checkpoint inhibitor and got diabetes-related autoimmune issue and colitis at the same time. So they tried to give the steroids for the colitis, and then the diabetes, of course, was getting worse. And so the investigator brought up the issue of maybe using the tocilizumab earlier to reduce there, the issue of the diabetes. And of course a lot of patients have diabetes, and as you say, have issues with corticosteroids.

PROF PETERS: I think that sometimes you only address the primary endpoint of your trial, but now that there have been seeing these toxicities, there’s probably a secondary endpoint to address. It’s how can you conserve and preserve first of all the immune response, second the quality of life of your patient, having the best benefits still present. And that might be about tocilizumab, and all of these fancy drugs, we have dozens of them, being anti-cytokine responses, right? So that’s probably for tomorrow.

DR LOVE: This concludes our program. Special thanks to Dr Peters, and thank you for listening. This is Dr Neil Love for Oncology Today.