Characterization of the Pancreatic Tumor Microenvironment using Novel Quantitative Multiplex DSP

Hello and a very warm welcome to the final webinar in our multiplexing Master class program titled Characterization of the Pancreatic tumor microenvironment using novel quantitative Multiplex DPS. My name is Charlie Carter, and I'll be moderating today's presentation and I'm delighted to be joined by Dr Dana Adel Mustafa, assistant professor and group leader of the Tumor Immunopathology Laboratory at Erasmus University Medical Center.

Please feel free to submit your questions for the Q&A session at the left of your screen at anytime during the webinar. And remember to join us for the final session of our multiplexing master class, the Virtual Roundtable, which will take place on Tuesday, October 13th and features all our incredible speakers including Dana. Without further delay, I would like to hand over to Dana.

The Pancreas

Thank you, Charlie. Hello, everyone. I'm Dana Adel Mustafa and I'm going to present to you some work about the tumor microenvironment of pancreatic cancer. But before I start, I would like to thank Leica Biosystems for this invitation and for all the efforts to arrange that.

Pancreatic Cancer

A small reminder of the pancreas. It's an organ that belongs to the digestive system. We use it to digest our food. And it's 2 in 1 organs, so it serves the endocrine function by regulating blood sugar and the exocrine function by secreting in times to digest foods, especially livid. Therefore it consists of so many different types of cells, but the main 2 compartments are the islet cells and the acinar cells. Depending on the tumor or depending on the origin of the tumor, there are two types of cancers originating from the bacteria. Islet cells which are responsible to succeed unfeeling, and these are the same types of cells that give us diabetes if they are not functioning well. If they develop cancer, they develop a tumor called pancreatic endocrine tumor. The acinar cells that they secrete the digested material and they secrete the digestive enzymes in ducts. And in ducts they go to a main channel, the main ductal canal in the pancreas. They give pancreatic ductal adenocarcinoma.

Pancreatic Ductal Adenocarcinoma (PDAC)

In my presentation, I'm going to talk about pancreatic cancer. What I precisely mean is pancreatic ductal adenocarcinoma (PDAC). Why study it? Because it's mainly performed 90% of the solid neoplasm of surgeries. So most of the patients who develop pancreatic cancer, they develop pancreatic ductal adenocarcinoma. It's a silent disease, so when patients come to the clinic, it's already a bit late. But when they come to the clinic, they unfortunately come with a lot of pain because the cancer invades the nerves. The cancer invades the lymphatic system veins, and it's very well known for metastasis. Almost all patients passed away from pancreatic ductal adenocarcinoma. For that the five year survival rate is 5% and it's expected to be the leading cause of cancer related deaths in 2030. Here is a publication from 2014 in which I would like you to appreciate that pancreatic cancer incidence is rising and it's one of the most strongest cancers. Another publication from 2017 showed that in the year 2017 pancreatic cancer death is going to be more than breast cancer death in Europe. Currently in the Netherlands, pancreatic cancer death is more than this cancer unfortunately. Because we don't want anybody to die because of cancer. But the pancreatic cancer patients who are passing away are exceeding the numbers of breast cancers. That's what I'm trying to say. Now, from a histology point of view, what is remarkable about pancreatic ductal adenocarcinoma? First of all, it is an invasive malignant of epithelial or ductal differentiation. It's a highly heterogeneous disease like a lot of other types of cancer. And it's characterized by remarkable desmoplasia of poor blood supplies. This remarkable desmoplasia are not tumor cells, but sometimes it exceeds 50% of the tumor volume.

The Histopathological Hallmark of PDAC: Desmoplasia

So what do we mean about desmoplasia? This is a picture of or a figure taken from a review published very recently in Nature Reviews and which in green, you see tumor cells. These tumor cells are surrounded with desmoplasia and the desmoplasia is basically fibroblasts or cancer-associated fibroblasts. Currently we know at least four types of these fibroblasts, but if you divide them into two, they are the inflammatory and the myofibroblast cells. They secrete a lot of accessory analytics. And that, together with the immune infiltration of many different types of cells that basically cytokines, interleukins; all of this is desmoplasia. What I miss from this picture is actually the blood supply, which is not a lot in pancreatic cancer, but it is part of the desmoplasia or the microenvironment.

Tumor Infiltrating Lymphocytes Ecosystem

Now one of the things that is very important is the tumor infiltration. It is accepted for years now that the cancer by itself is an ecosystem so it depends on itself to evolve and it attracts the cyber cells that they are important for its survival. So the immune infiltration or tumor infiltration infiltrating lymphocytes, they can predict response to therapy. They are good and bad. And the balance between the good and bad players is extremely important. We also need to understand the amount of infiltration and the location of infiltration of these lymphocytes.

TILs Predict Overall Survival & Progression-Free Survival

Our group published some data about cells and pancreatic cancer. In the upper picture, you see the infiltration or immunohistochemistry of CD8. These are a type of T cells and if you look at 500 or more tissue microarray, you can tell the distribution or the infiltration of t cells in the pancreatic cancer varies quite a lot. It varies from being completely absent to being high. What is important to know is the high CD8 to T gradually ratio is associated with favorable outcome.

PDAC Microenvironment

And since coming here, we are focusing not only in the minor microenvironment but in the immune compartment of the microenvironment. We have several questions that we'd like to answer. For example, what is the important cell type that infiltrates to the pancreatic cancer? How cells, or immune cells to be specific, interact with each other? What is the transcriptome of each cell? And is there any openings for Cw6 targets? And one of our questions is how these cells are organized in space? The reason for that is because it's again about the balance. So the balance of this infiltration and interaction will determine if the tumor progresses or it will be suppressed. At least it's one of the factors that will determine that.

Rationale

So what do you have together? All of what I told you as an introduction together. I'll give you the rationale of what I'm presenting today. Unfortunately, most of the patients who go under surgery, they die within couple of years. When we looked back at our archives, we found out that there are few patients who survived for longer than five years and this is extremely important because we would like to know why, why few patients survived for more for longer than five years. And our hypothesis was that is the tumor microenvironment and the immune cell infiltration may be the key for that longer survival. The aim of what I'm presenting today is to review the immune microenvironment and the pathway that may determine the long term survivorship in PDAC patients.

Study Design

To do that, we went, as I said, back to our archives and we selected patients that survived more than five years. And those who survived 6 months or less after surgery, but we excluded those who backed away because of surgical complications. Out of those patients, there are 20 patients, 20 samples of patients. We know that evaluation, we know when they develop metastases or when they progress and where they develop metastases. So we know the whole information, that clinical information, about those patients. What we did with the 20 samples is we measured them by the immunoprofiler of encountered NanoString technology. I'll talk about it and then we selected part of them to do the geometric DSP and I also explain it in detail.

NanoString Technology

For those of you who don't know about the NanoString technology, it's a targeted gene expression. In this picture is the RNA that you isolate from your samples. The NanoString technology comes with two groups. Each one of them is 50 bases, so we only need 100 bases of RNA to be able to measure the expression of the genes. So it is very suited for better of an embedded material which is known to have degraded RNA. One of the probes has a barcode and the barcode is 6 fluorescent dyes of four different colors. So this already tells you that we are allowed to do a Multiplex a day. Because if you combine these dyes in different sequence then you can measure different genes at the same measurement. What we are able to measure is 770 genes.

PanCancer Immune Profile Panel

NanoString arranges the panels in commercial available panels, but you can also customize your panels based on your question or genes of your interest. What I'm showing today is the presentation of our results of ten cancer immune profile panels. The reason why we chose that one because it has 770 genes representing the immune system and that allows us to identify 24 different types of immune cells and the number of immune pathways.

Workflow

What we do is we select our sample and then we study each sample with the pathologist. We enrich for tumor tissue and by that I mean tumor plus desmoplasia. But what we avoid is the normal adjacent tissue. And then we use commercial available kits to isolate the RNA, we check the quality of the RNA in order to make sure that we will have good quality data at the end. We hybridize them and we follow the protocol of NanoString.

Results: Deferentially Expressed Genes

After that, well, I'm presenting some data here, so I want you to know that I use the nSolver software available from NanoString technology and I like to use the advanced analysis module of it. Here is the full kernel plot of the differential expression in the Y axis that's in that look thin B values and in the X axis is there log2 fold change. The area of strength is the genes that they are overexpressed in the long survival and the left part is the genes that they are overexpressed with a good full exchange in the short survival. We have some significant genes. If you take the most significant 14 genes and when we perform principal component analysis we find as good separation between the short and long.

PCA Based on the 14 Most Differentially Expressed Genes

Short tiered is represented if each dot is a sample. Well in total we have 10 samples and they are in orange and long survivals are in gray and there is a good separation, not complete, but very good separation between the two groups.

How to Utilize the Gene Expression Results

How do we utilize the data more is by identifying sources of cells. And how do we take gene expressions to units of cells is by first checking the markers of their immune cells. Here I give an example of B cells. B cells are identified by CD20, CD19, CR2, and CD22. First we do a correlation between the expression of those genes in our samples and this is a perfect correlation. Then we take the averages of those genes to get value in the two groups.

Infiltration of Immune Cells

The left part shows that CD45, which is the expression of all lymphocytes-the immune cells infiltration in both groups- and it's very comparable. However, if we check the expression of B cell compartment from those lymphocytes then they are high significant, high in long term survival.

B Cells in PDAC

These were the only cells that they are significant between the two groups. Of course there is the questions about the B cells and pancreatic cancers and when we look at the literature, we'll find opposite results. So it's very well acceptable that B cells are immunosuppressive. In 2016, there were three publications in the same month talking about iso promote pancreatic tumorigenesis and they are not good for the survival of the patients. However, last year one of publications suggested the opposite and in our data, we see that B cells are infiltrated more in patients who survive longer. But we have a question. We have the question of the location, where do they infiltrate exactly? As I said, pancreatic cancer is very heterogeneous. And what are the other types of cells that they interact with? And for that, I'd like to remind you of what we did. We isolated RNA from the whole tissue of pancreatic cancer of all 20 samples. For us, one tissue was one RNA measurement. Usually I say we make a soup but fruits are nicer. You have fruits and from these fruits you make a cocktail and from the cocktails you try to identify the compartments of the cell types that they infiltrated in the tissue samples. However, what we want is a spatial location. The arrangement of these cells inside the tissue sample. Indeed I have two pictures of the same pie. If you look at the fruits, they are exactly the same fruits in these two pies. They are arranged differently and this is what we are interested and we are interested in the arrangement inside the juice.

GeoMx Digital Spatial Profiler (DSP)

To do that we use the GeoMX Digital Spatial Profiler. It will enable us to investigate our questions because it combines more political markers with hidden markers. Instead of studying one sample as one cocktail or one soup, now we come up with morphological markers. We study the tumor compartment or tumor cells (ductal cells) by yellow. We see them by pan cytokeratin fluorescent. We studied the stronger side by alpha-smooth muscle actin represented here in green and the immune compartments by CD45. And of course that is standing to represent cells. Now we are interested where exactly these cells are infiltrated inside the tumor. We study each compartment separated from the other neighbor. That’s one tissue and is not one sample. One tissue has many compartments which is what you, as a researcher, are interested in. This allows us to connect morphology to expression profiles. Why is that? How is that enabled? It’s all developed by NanoString. They take either the antibodies, so proteins, or the RNA, and they conjugate the antibodies to DNA oligo with a UV linkage. When we say that there's slide by tissue markers, as I showed you with four different compartments and we come up with hidden or invisible proteins connected to oligos connected to the DNA oligo. These are autoclavable, so we select our areas and we see these DNA oligo when we select our areas and we highlight or we give them a beam of UV light, then the DNA oligo is cleaved. We take that oligo and we hybridize it with the drops, so we make it quantitative and with very easy numbers that we deal with for data analysis.

Human IO Protein Panel Content

What I'm going to show you today, we measured in unit oncology panel and new oncology panel consists of invisible protein markers for profiling drug targets, immune activation markers and others. We take one section of four or five micrometers of embedded material. We come up with morphological markers maximum four, and we hybridize them at the same time with hidden antibodies now at 45 that I'm presenting in this slide. Then we study each one of these compartments separate from the others. We needed an example of one of our samples in which we studied each compartment different than the other. As a random example, here's region #2. If you see that little circle inside the second slide. ROI is region of interest. In my eyes, I think this area is poor with pan-cytokeratin or ductal cells. And if we check the expression of pan-cytokeratin is very little in this area. However it has green which is alpha smooth muscle actin and if we check the expression of alpha smooth muscle actin (ROI_02) then it is considerable and of course because it is desmoplasia or connective tissue and it has some parts of CD40. If we take region number six, it is very enriched for CD45, which we see here in region number six. CD45 expression is almost 150 and it is almost very poor for tumor cells or pan-cytokeratin and this is reflected by the graphs at the bottom.

Long-term Survivor Sample

This is an example of a long-term survival sample, and this is the area #8 which is filled with immune cells. When you look at the expression of these immune cells, they were very high for CD20 which is a marker for B cells.

Short-term Survivor Sample

Another example is a short-term survivor. We chose the exact same area filled with CD45 and that area (#10) has a lot of expression of CD20. However, those two areas were in a completely different change. While they are expressed counts of above 5000, in the long term survival, they expre00ssed counts of 6 or 700 in the short term survival. If you put data from six long-term samples and six short term samples, you notice that the infiltration of the b cells to the long term survivor was much higher than in the short term survivor. However, that infiltration was not very heterogeneous and that is the beauty of using a technique like DSP.

Differentially Expressed Proteins in α-SMA ROIs after KH Normalization

The infiltration of the B cells were specifically in areas they are located with alpha smooth muscle actin or connective tissue and they work together with other types of cells like CD4, CD3 which are T cells, CD11c, CD8 another type of cells, so they were together with dendritic cells, with monocytes and with other types of T cells.

Differentially Expresed Proteins in PanCyto

If we check that pan-cytokeratin, in both groups we see the exact same, but it's not significant. On the left side of the volcano plot, you will see markers like CD3 and HLA-DR, CD20, CD14: the exact same volcano plot that I showed you earlier. However, they are far away from being significant.

Immune Infiltration in Long-Term Survivor PDAC

These cells infiltrate infant straight to the specific sites of the long term survival. First of all, these infiltrate higher and higher numbers and they infiltrate to their extracellular matrix or the desmoplasia not to the tumor side. They infiltrate together with other types of cells like T cells, monocytes and CD 11 positive cells.

The Important Role for B Cells in Promoting Anti-Tumor Immunity

At the beginning of 2020, B cells became of interest because there were three Nature papers in the same day talking about the B cells and the tertiary lymphoid structures in various types of cancers.

Tertiary Lymphoid Structures

The tertiary lymphoid structures are very important structures inside the tumors in which T cells infiltrate together with CD3 or T cells around them, so the T cells are in the middle and the CD3 T cells are around them. When the tumor develops these kind of structures, it's connected to the better over ocean markers.

Conclusions

The immune infiltration is one of the keys that determine the survival in pancreatic cancer. Infiltration was higher in numbers and in patients who survived longer than five years. The T cells were infiltrated together with T cells into macrophages desmoplasia. They were infiltrated in the connective tissue or in the desmoplasia but not in the tumor compartment. I hope I convince you that the GeoMX DSP can be used to map gene expressions with spatial and morphological features used in FFPE material.

Multiplex for Biology

Our talk is about Multiplex analysis and quantitative Multiplex analysis, so I'd like to explain that the DSP is not the only tool available for us as biologists. We actually have various tools or techniques. Each one of these techniques has positives and negatives. You can use it to answer specific questions, but one of the most important techniques we have, in my opinion is the quantitative immunofluorescence. It allows 3, 4, 5 sides or antibodies to be checked altogether. That was developed to give you Vectra system, which allows to 8 and even higher number of antibodies to be checked altogether. Another tool is like single cell sequencing because it gives a complete different types of measurements. 10X genomics, which is connecting sequencing to imaging, but from physical material and of course the geometric DSP that connects morphology to expressions from paraffin embedded material. It also can work in fresh frozen materials and the future for this technology is to connect it to the sequencing. I know that that NanoString already is generating data of connecting the morphology to sequence data. The difference between the DSP and any other technology is the DSP or the GeoMx DSP taking the technique or the measurement away from the tissue. So we took the oligos outside of the tissue and that's why we made it a quantity. Now, as a biologist, that is always where to go and which techniques to use. In my opinion, we need to study the type of tissue that we have like part of an embedded fresh frozen, the amount of money, the amount of time that we want to spend in any research and we need to check the facilities, the severity of the disability, the resolution of all the available tissues and based on all of these factors together we can choose which Multiplex technology we should use for our analysis.

Acknowledgement

With that, I'd like to conclude and I would like to appreciate the effort of all people who participated with me in this study, including and especially the patients who enabled and supported this work, because without donation of their tissue, we will not be able to conduct this work and I'd like to appreciates the financial support of Support Casper Foundation that they support me and my lab quite a lot. Thank you.