So I think we're ready to start. My name is André Lacroix. It's my pleasure to, from Montreal, to co-chair the session with Margaret Castro from Ribeiro Preto, Brazil. We have a very exciting symposium on diagnosis and pathogenesis of adrenal tumors. So it's my privilege to introduce our first speaker. I don't know why Margaret thought that I would be better to introduce Charlotte Ducy-Lepoutre from PSIS at Péterrière, who is going to present to us on H1MR spectroscopy as an innovative test for in vivo metabolic evaluation of pheochromocytoma and paraganglioma. Charlotte. Thank you very much for this introduction. So my talk today, first, I'm very pleased to be there and to begin this session, very nice session. So I will speak today about special sequence imaging to evaluate metabolic profile of pheochromocytoma and paraganglioma named PPGL. I have nothing to disclose. And this is the QR code for the virtual session. So as you know, about 40% of patients with a PPGL carry a germline mutation of one of the 20 PPGL susceptibility genes identified so far. And this is mostly tumor suppressor genes. So there is a germline mutation on one gene and you need to have the second event for the tumor development. Most frequently involved are SDH genes. These genes are encoding the first subunit of the succinate dehydrogenase, SDH A, B, C, and D. This is an enzyme in the mitochondria that is involved in both electron transport chain and the TCA cycle. In the TCA cycle, it is responsible for the oxidation of succinate into fumerate. And if there is a mutation of one of the four genes, the Wollinger enzyme is deficient and then succinate accumulates in the cell leading to the tumor development. It acts as an oncometabolite. This succinate accumulation can be detected in the tumor tissue by mass spectrometry and this detection of succinate is a very specific hallmark of SDH mutated tumor. So our objective was to determine if it is possible to detect it in vivo. We have in the lab a tumor model of SDH deficient tumor in mouse. This is a nalograft model. We used for that SDH B deficient cells coming from medulloadrenal of mouse and we holographed it to an MRI nude mice. So in this tumor model, we have a complete deficient of SDH activity and a huge accumulation of succinate in the tumor. What is now the technique? This is a special sequence. Spectroscopy is a special sequence of MRI. MRI is a procedure based on proton imaging and proton are mostly contained in water molecules. So this is an imaging of water molecules that permit to obtain this cross-sectional imaging. On the contrary, in spectroscopy, you will suppress this water signal and you'll be able to detect metabolites in a specific volume in a voxel. You won't obtain this image, but a spectral representation of the metabolite present in the tissue in vivo with a specific position of each metabolite. And succinate is positioned at 2.4 ppm. Moreover, this is a quantitative method as the area under the peak is proportional to the metabolic concentration measured in the tumor. So in the lab, we have a small animal dedicated device, a 4.7 Tesla MRI, and we try to develop and optimize this sequence in this specific device with our mouse model. So as you can see, we could detect succinate accumulation at 2.4 only in the tumor deficient model and not in the Y-type tumor model. We verified that succinate accumulation detected in vivo was completely correlated to the succinate measured by mass spectrometry in vitro. Then we transferred the sequence named Success for Succinate Estimation by Spectroscopy in tumor patients, thanks to a collaboration with the radiology department of our hospital. And our first patient was a patient with an SDHB mutation, but it was a variant of unknown significance. So we had to prove the pathogenicity of this mutation and to prove it at that moment, we only have technique in vitro. So we did it with these available techniques, so like genetic analysis to search from a loss of heterozygous in the tumor, the loss of protein expression by immunochemistry, and if you have a frozen tumor tissue, you can measure SDH activity and measure succinate by mass spectrometry. So everything was in favor of pathogenicity of this mutation, but needed tissue to prove it. And in fact, with our sequence, we already knew that the mutation was pathogenic because we observed a nice succinate peak, an accumulation of succinate in the cervical paraganglioma and in the abdominal paraganglioma of the patient. In contrast, in a sporadic flow chromocytoma, we did not observe any succinate accumulation, a positive immunochemistry, and no accumulation of succinate by mass spectrometry. So this was very encouraging result, and then we moved to validate this technique to a prospective translational research protocol on 50 tumors, including 24 head and neck paraganglioma, nine abdominal paraganglioma, and 17 pheochromocytoma. In these 15 tumors, we observed 20 succinate peak, mostly in head and neck paragangliomas, and all these patient with a succinate accumulation had an SDH deficiency. 17 had a germline mutation of one of the SDH genes, two had a somatic mutation on SDHD gene, and for one patient, we did not find the mutation at the moment, but he has clearly a deficient of SDH in the tumor by measuring activity and immunochemistry. In 30 paraganglioma, we did not observe any succinate peak. In this group, we have all the sporadic tumors, two patient with a VHL germline mutation, and we failed to detect a succinate peak in three SDH patient, and I will try to explain why after. So altogether, 50 tumors, all the four gene subunit involved 20 germline SDH mutation, two somatic SDHD mutation, and it leads to a very high specificity of 100% and quite good sensitivity of the method of 87%. Some examples of spectra obtained in paraganglioma, head and neck paraganglioma with SDHD mutation, an abdominal paraganglioma with an SDHB mutation, and also a paraganglioma head and neck with SDHD mutation. So unfortunately, we didn't make it for three patient with a germline mutation on SDH. Let's try to see why. The first one was a 35 years old patient with an SDHB mutation, and in fact, in her paraganglioma, we can see big vessels like carotid artery and jugular vein, and in that context, we cannot apply the sequence because it's too much fluid, and so it's too difficult to make the sequence on it. So we decided to try to explore a little metastatic lymph node just behind, and in fact, the lymph node was too small, and then that's why I think we did not observe quite good succinate peak. The second one had cervical paraganglioma, but completely necrotic or hemorrhagic one, and once again, we didn't observe anything, any metabolite. So this is really a failure of the technique when you observe that with nothing, no lipid, no choline, nothing. So this is really a failure of the sequence when it is completely necrotic. And the third one was a young patient with a small pheochromocytoma detected in a systematic screening in familial disease, and in fact, there is no necrosis, but the pheochromocytoma was small, and we had respiratory motion artifact on that. I think that's why we don't observe any succinate peak. Sorry. So these are our three false negatives, but it leads to a quite good sensitivity of 87%. What are the perspectives now? What are the questions? First one, are we able to apply this sequence beyond paraganglioma and pheochromocytoma for other tumors? Second, are we able to have a metabolic profile beyond SDH mutation in sporadic paraganglioma, for example, or in other mutation types? And third, is this succinate exploration is a good biomarker, a good quantitative biomarker to follow response to treatment? So as you know, SDH mutation predisposed to paraganglioma, but also to other tumors like GIST, renal cell carcinoma, and pituitary adenoma. For the last one, the literature is a bit discussed about the development of pituitary adenoma in SDH mutated patients. But for the two other, it's quite clear that SDH mutation predisposed to this tumor, and an English team already demonstrated that we can observe succinate accumulation, for example here, in a metastasis, in a liver metastasis of a GIST. So this is feasible in other tumor, even if we can imagine that there is a respiratory motion artifact in this location. Recently, we had the opportunity to explore a 40 years old patient with neck paraganglioma and a macroprolactinoma with an SDHC mutation. So as you can see, we observed the succinate accumulation in the neck paraganglioma, and we were very pleased that we observed also this succinate accumulation in the pituitary macroprolactinoma. So I think this is quite a nice result because it really linked the macroprolactinoma to SDH mutation, which was not so clear up to now. What about other PPGL with no SDH mutation? So this is an example of a right sporadic cervical PGL, and with a nice post-treatment of the spectrum, using software like LC model, for example, we can obtain a better discrimination of all metabolites present in this voxel, and probably there is a specific metabolic profile in this sporadic paraganglioma. So you can see that there is no succinate accumulation, but there is also other differences, and I just show you again the spectra of, the spectrum of the patient just before. There is also an accumulation of glutamine and probably aspartate, and so this is an ongoing study, but I'm sure we can define some profile for perhaps each mutation. Finally, is this a good biomarker for a treatment follower? So we tried to validate this hypothesis using sunitinib treatment. Why sunitinib treatment? Because it was the molecule used in the first randomized trial named FirstMap at that moment. So we used the same molecule, and trying to answer to this question, is succinate a good biomarker for response to treatment? It means that decrease in succinate level should be a result of anti-tumor therapy. So we went back to our mouse model, and we treated 40 mice with SDHB deficient tumors during three weeks, either with sunitinib or the vehicle, and we performed the sequence before treatment and after each week of treatment. We observed a good anti-tumor result with sunitinib with the stabilization of the tumor at the beginning during the first 10 days, and after quite an escape of tumor growth after 10 days. And we can correlate this tumor growth with our succinate measurement in vivo as at the beginning. It is correlated with a decrease in succinate level, but after two weeks of treatment, the succinate level stabilized, and this is the reflect of escapement indeed. So thanks to support of the French Society on Endocrinology, I am currently trying to do it in human, specifically in head and neck paraganglioma in patient who will underwent external beam radiation therapy. It is beginning, I have no result for the moment, but we will apply the sequence before and three and 10 months, 12 months after radiotherapy, because it is quite difficult to evaluate response to radiotherapy with MRI or with nuclear medicine. And so I think perhaps it could be interesting to add another marker. So in conclusion, in vivo metabolic evaluation is feasible in vivo. It can help to validate variant of anosignificant, in particular when patient do not underwent surgery. It helps to detect mutation, somatic mutations. It can link other tumor development to the primary SDH mutation. And finally, it probably will be a good biomarker for treatment evaluation. So I want to thank all the member of my lab headed by Judith Favier in Paris, the radiologist really involved in the sequence development, Bertrand Tavision in the lab, and Francesca Branzoli in Pitié-Salpêtrière Research Center, the French Endocrinology Society for his support and all of you for your attention. Thank you. Thank you very much for this fabulous presentation. Often one of the problem is to make the diagnosis of the very small lesions initially in patients who are genetically at risk. So what is the minimal size do you think of the lesion to be able to detect the accumulation of the metabolite? And is it also true that this is maintained in metastatic disease? Yes, for the first part of the equation, I think that it's the tumor size and also the box cell size. If the tumor is, well, just a nice spare, we can have a good box cell size. And I think the minimum is one cubic centimeter for the box cell size. So the tumor must be bigger. And the second part, I didn't try for the moment for the metastasis, but it has been done for just in liver metastasis. So it is feasible in metastasis. One thing that we could be careful is that the sequence for bone is not a good thing. So I think we won't be able to explore bone metastasis only if there is a big tissue metastasis, you know, a big metastasis. John, you want to press? Thank you, that was absolutely beautiful talk. Thank you. I've got two small questions, if that's all right. My first question is, is do you find a difference in the peak height of the succinate depending on the subunit type? No, we don't. We didn't. And in fact, in in vitro study, there is no such difference in succinate levels. So we are less sensitive than in vitro. So I think it's a result we cannot improve. And my second question is, if you just look at a liver, for example, in mutation positive carriers and in people who are not, do you find any differences at all just if you have a germline mutation? We did it in our first patient because it was a really interesting question. Is it specific of a tumor with a double mutation or a complete loss or not? And we did it on, we performed it of the liver of the first patient and there were no succinate peak. So this is really, you really need an enzyme deficiency to see it. Okay, thank you. Tobias. Hi, thank you very much. That was a really great talk. So do you need to, I have just a simple practical question. So do you need to run an extra sequence for getting these and how much time does it add to the scan? Good question. So at the beginning, we used a very long sequence. It was like 45 minutes of scanning and this is really specific. You cannot do it after this is a specific sequence You have to suppress water signal. It's not like radiomic where you can do it after So it was very long at the beginning and now we improved the sequence and it's like 20 minutes more but all together if it is just difficult to to make the shine the shimming I don't know if you know it's at the beginning of the MRI scanning You must be sure that everything is homogeneous and this can be quite long So all together is more than half an hour for each patient the patient I have to think patient and you have to be very In good terms with your radiologist And and this this time you add is that it is that only for the tumor specific area Or do you do the whole scan with it? No, this is in a vector in a voxel volume This is metabolized in a in a volume specific so you cannot screen the whole body. Okay, good. Thank you Thank you. I don't suppose you have been able to look at this in humans, but in your animal model Are there physiological conditions that enhance and facilitate the detection hypoxia or Glucose are there Physiological changes that will facilitate the detection of the succinate accumulation No, in fact, no, but this is dependent of temperature so if The temperature change the succinate peak will move a little bit So we have to be careful with that, but there is no enhancement or decrease in succinate peak with some condition In fact, if there is a necrotic Part of the tumor it will go down if the tumor is big you can apply a big voxel and this is the more Enhancement you can have is the size of the voxel and the duration of the sequence More you accumulate more you have signal Hi, thank you for that Mike Patel NIH. I was wondering about the heterogeneity of the tumors You might have touched a little bit upon that on the conversation just now But were there different areas of the tumor that had different peak values in the same patient or the same mouse model or anything? In fact, this is not Sensitive enough to make to see tumor heterogeneity So we have this voxel and we try to position it on the most nice part of the tumors the most homogeneous without necrosis and with no big vessel, but We cannot be able to show it a radionicity of the tumor with this Sequence at the moment. Thank you So There are no questions coming from the remote audience So I want to thank you very much for initiating this session very excitingly Our second speakers is Julie Reggie She is a research Assistant professor at the University of Michigan and the title of his presentation will be genomic and transcriptomic analysis of archival adrenal cushing tissue So good afternoon everybody I want to thank the endocrine society for giving me this chance to present our work on such a big big platform So the title of my talk has changed a little bit from what went earlier So the work that pertain pertaining to the previous title was published. So I thought I could present something Present some newer data So nothing to disclose So the presentation will really be divided into three simple parts First I'll be talking about Giving a little background about what Cushing syndrome is and a little background about cortisol producing adenomas or CPAs after that, I'll be moving on to the methodology that we use in the lab to isolate RNA and DNA from FFP CPAs and Lastly, I'll be sort of touching on the utilization of this FFP material that we isolate from From the cortisol producing adenomas for mutation analysis as well as RNA-seq studies Well Cushing syndrome as everybody knows it represents the signs and symptoms that result from hypercortisolism So long-term glucocorticoid excess has been sort of known to correlate with Comorbidities such as hypertension, obesity, diabetes, recurrent infections, skin manifestations Well ACTH independent cortisol secretion can occur from either one or both the adrenals but dysregulated cortisol production is often seen in benign adrenal tumors and CPAs or cortisol producing adenomas are the major culprits in this scenario Cushing syndrome is sort of a prototype for metabolic syndrome and is more prevalent in women So as far as the subtype of Cushing syndrome is concerned first it's the overt type wherein which is hallmarked by the physical manifestations such as purple stry on the abdomen, fat pads on the clavicle and the back and It is very rare. It's present in two per million people per year Mild autonomous cortisol excess, MACE or subclinical hypercortisolism is more common It's present in about 0.2 to 2% of the population and it is it is sort of the evidence is biochemical cortisol excess, but without the presence of any specific clinical signs Now I'll be referring to this particular syndrome MACE from now on. It's just easier and quicker And as all everybody knows cortisol sort of spans Like a spectrum of severity so really putting rigid guidelines as to whether you know Just for diagnosis of Cushing as well as you know subtyping it. It's really very hard The other thing is recent data has sort of linked MACE with With The comorbidities that are associated with cortisol excess so we cannot take it very lightly Now in the last you know seven to eight years in the last decade there have been several groups that have tried to figure out the mutations using snap-frozen CPAs and and most of these mutations fall in the category or fall in the pathway of the cyclic AMPPK system or and Examples are PRKCA you have PRKR1A and GNAS and there's also beta-catenin from the Wnt pathway So with with the snap-frozen tissues About 50% of the CPAs have been found to come with a somatic hit and there are still half of the CPAs that have undefined causes as yet as you can see in this in this chart PRKCA and beta-catenin are really the Major drivers for for CPAs in this case now About roughly maybe two to three percent of the population is present with adrenal masses and of these benign adrenal benign adrenal adenomas sort of form a majority of these masses and The recent study by called Udenac by the NSAT has shown that you know It's 40% of these benign adrenal adenomas show some dysregulated cortisol secretion Which was much higher than what was earlier known So if we take these calculations into perspective, it really seems like about 1 million Americans have An adenoma which has a yet-to-be sort of it has a has an unknown new mutation It's sort of a yet-to-be-determined mutation. So if you keep these statistics into perspective It really seems like studying MACE Studying over pushing studying CPAs is really very important since we still have half the CPAs with with an unknown cause Now before I touch on the genomic analysis that or the methodology that we use in the lab I want to quickly touch base on how cortisol is produced in the human adrenal so you have cholesterol sort of metabolized through a series of steps and getting converted into cortisol in the zone adrenal zona fasciculata I want to point that these three important enzymes So we have three beta HST type 2 CYP17A1 and CYP11B1 So these the triad of these enzymes The adrenal localization or the adrenal cell localizations where all these three enzymes are present in the absence of aldosterone synthase Can be likely the source of cortisol So this particular enzymatic signature would be able to differentiate say cortisol producing tissue from aldosterone aldosterone producing tissue and even DHEA producing tissue So keeping this particular signature in mind we sort of developed an amino histochemistry guided sequencing approach Wherein we could use FFPE or formalin fixed paraffin ended embedded CPAs to define the causes What we basically do here is we go to we prepare a series of sections and we perform dual amino histochemistry for CYP17A1 and 3 beta HST to sort of define The cortisol producing regions of the tumor then we go in we capture those tumors and we get the DNA and the RNA So this this technique is really very crucial where you do, you know where you really cut cut those cut those cortisol producing regions and this technique has been mastered in a lab by our research technician Amy Blinder and Emma Zagayas who's our grad student and we've gone through scores of FFPE tissues just you know getting us nucleic acid material for analysis So coming back to the CPAs Once we get the CYP17A1 and 3 beta HST IHC done DNA RNA isolated the DNA goes towards targeted amplicon sequencing of this captured CPA material and Whatever sort of and this particular panel has the full sequence of known CPA related genes So whatever passes through that panel and the CPAs that are devoid of any mutation go towards the whole exome sequencing analysis RNA goes towards targeted and RNA-seq analysis so this has been this the mutation studies are being done in collaboration with our friend and the Department of Pathology at the University of Michigan Aaron Utica so the Department of Pathology has sort of long transitioned into doing their Molecular analysis using into sort of transitioned using using FFPE blocks And so the advantage being we can use really really very small amounts of DNA to get our mutation hits Now if I go into the depth of how we Do this technique like I said, we have a series of FFPE sections The first couple of sections go towards CYP17 and 3 beta HST immunohistochemistry this sort of gives us an idea where what could be the possible regions of cortisol excess in the tumor and Once that is sorted the next five to eight Unstained slides are used for capturing those regions and wherever possible. We also capture the adjacent adjacent adrenal Just for as a control so the other advantage about Using the IHC Guided or immunohistochemistry guided sequencing technology is that we can really distinguish the heterogeneous Regions of the tumor so I have example 2 here on the slide where you can see T1 has a more uniform Homogeneous 3 beta HST expression and we have T2 which has a slightly Lesser amount of 3 beta HST so we can actually go in and capture to those two regions Separately and we can also capture the adjacent adrenal Once we have the DNA like I mentioned earlier it goes towards sequencing but we first pass it through Sanger sequencing where it just where we can just look at the Somatic hotspots for PRKSE and the beta-catenin which are the major mutations Once we sort those samples the remaining samples go towards the ion torrent NGS or the targeted Next generation sequencing panel as I mentioned earlier We can detect mutations with just 5 nanograms of DNA or we can even use the somewhat degraded FFP samples to go through this panel the other advantages it's about a thousand to two thousand fold more sensitive than Sanger sequencing and We can choose what genes we want on the panel so in this case we have CPA related genes Such as PRKSE, we have beta-catenin, GNAS, PRKR-1A and PRKR-KSCB Now using this particular methodology we were able to detect somatic hits in about 72% of All the CPAs which is sort of an improvement about the snap-frozen tissue approach Wherein half the CPAs have been identified with a hit Oops As you can see in the in the plot beta-catenin alteration sort of form the majority the major bulk of the mutations and this is really because Mase CPAs form the major tissue type in our cohort So as you can see in the figure we have beta-catenin being the major Mutation type in all the Mase CPAs and in the over CPAs as we know PRKSE one is the major gene that is hit The mutation negative samples or the samples that are devoid of any known mutations are sort of comparable in both these two Cushing syndrome subtypes In the second study, I'll be talking a little bit about our targeted RNA-seq technology that we used for our FFP CPAs So, you know PRIOM genome-wide studies have shown that ACC's CPAs and APAs or ACC's meaning adrenal carcinomas APAs are aldosterone producing adenomas. They have distinct transcriptomic profiles But and this was done till recently using snap-frozen tissue, which also Came as a disadvantage because not all clinical samples have snap-frozen tissue So we sort of transitioned like I said into using FFP samples, which has really increased our access to CPAs not even only in the University of Michigan, but also with our collaborators Which also gives us the advantage of using archival CPA samples, which is another added advantage so we Aaron Udega with him Developed a targeted RNA-seq panel and this consisted of about 200 genes and These genes comprised of adrenal and adrenal tumor related transcript groups pertaining to steroidogenesis development differentiation and tumorigenesis, we also had Quite a few housekeeping genes in the panel so as to so that we could use them for normalization So using the RNA-seq panel we were able to see that You know the ACC's, the APAs and the CPAs could really be Distinguished into three separate groups. So this was this was this was very encouraging We next wanted to see if we could use the same 200 gene panel to sort of differentiate the CPA genotypes so here and I have Beta-catenin CPAs, PRK-CACPAs and Genas CPAs and again We could see two different two separate clusters for the beta-catenin CPAs and the PRK-CPAs The Genas CPAs fell mostly in line with the PRK-CPAs Which was not surprising since both of them are components of the cyclic AMPPKA pathway But we did have one Genas mutation that sort of went and clustered with the beta-catenin Now if we after we charted out the heat maps We looked at the steroidogenic enzymes first and we saw that the expression of the steroidogenic enzymes was higher in the in the mutations of the cyclic AMPPKA pathway So when we went into deeper analysis We saw that CYP11A1, 3-beta-HSD type 2, CYP17 and CYP21 Were really the the genes that were significantly different between the beta-catenin pathway mutations and the PKA pathway mutations and this sort of really told us that The Genas and the PRK-CPAs are are more steroidogenic And these findings were very similar to what Christina Ronchi got in her manuscript and Hironobu Susano They also did it with their studies. So this was again very encouraging When we went to the differentially expressed genes We saw that the direct Wnt pathway targets such as LEF1 We have LGR5 and APCCD1 which were higher in the beta-catenin and mutated CPAs Again going with the idea that Wnt activity is elevated at the beta-catenin and CPAs Conversely we saw that the sonic hedgehog pathway components such as GLEE and PATCH1 Had was suppressed in the beta-catenin and CPAs So the suppression of the sonic hedgehog pathway might suggest that the constitutive activation of the Wnt pathways perhaps sort of suppressing the sonic hedgehog activity and this has been shown to cause tumorigenesis in other tissues like glioblastomas So this may could be a reason why beta-catenin CPAs just because there's more tumorigenesis They can produce more cortisol We also looked at the several GPCRs and we found that Sistel R2 vasopressin receptor GPR38, LGR5 that I spoke about earlier were again high in the beta-catenin CPAs Whereas MC2R or the ACTH receptor was higher at the PRK CPAs Now I'll be coming to the last part of the last study wherein I'll be talking a little bit about how we select our samples our FFP CPA samples for whole exome analysis now we have been successful in in finding mutations in FFPE aldosterone producing adenomas using whole exome sequencing technology We have found hits in CACNA1H and CLCN2, but as everybody knows, whole exome sequencing technology can be expensive and more expensive for degraded samples, such as the FFPs, where material is also very less. So choosing the right kind of samples was really important. And again, for APAs, because now we only have about 5% mutation negative APAs, choosing samples for whole exome sequencing is perhaps a little easier, as opposed to CPAs, where we have about 28% mutation negative samples, as I just showed earlier. So selecting the right FFPE CPAs for whole exome sequencing is really crucial. So we decided to choose these samples on the basis of their transcriptomic signatures. So I have brought up this PCA plot again, that I had showed previously, where I showed different clusters of the CPA genotypes. So I have the beta-catenin CPAs clustered separately from the PRK CPAs. And along with these CPAs with known mutations, we also ran a cohort of FFPE CPA samples that were devoid of any known mutations. And of these, we chose the five CPAs that are marked in yellow circles. So as you can see, they sort of fall in the PC1 component of the PCA analysis, meaning they sort of align more with the beta-catenin CPAs. We got a whole exome sequence hit here, and the hit was a low-variant frequency Y300S mutation in ZNRF3. So this excited us a lot, because we know that ZNRF3 enhances wind receptor degradation, so it can dampen wind activation. But if this is an inactivating mutation, the wind activity is going to be enhanced. Well, Guillaume Assier and the group have shown the presence of mutated ZNRF3 in cortisol-producing ACC. So this was another encouraging sign. And last but not the least, ZNRF3 activation has been shown in mouse adrenals to enhance wind signalings, and they've shown that it causes hyperplasia. So that was another encouraging sign. So now that we were so interested in this particular CPA, CPA number 79, we sent it for further histopathological analysis, which showed that this was actually an ACC which had a heterogeneous phenotype. So it had regions that resembled ACCs, and other regions that resembled adenomas. So, and we put it on a PCA plot again with CPAs, APAs, and ACC, and we can see CPA 79 sort of inching towards the ACC. So it's really becoming an outlier for the CPAs. We looked at the ACC marker, IGF2. Again, CPA 79 had an outlier expression for IGF2, and the expression also sort of told us it's probably going more towards the ACC phenotype. But the good thing, the encouraging signs here were we could use FFP archival material for whole exome sequencing. And the CPA that we chose for whole exome sequencing on the basis of its transcriptomic profile was indeed a pathway of the wind component. So these were the encouraging signs. And we thought that perhaps merging targeted NGS and whole exome could really help us in sort of for hunting down new mutations in CPAs. So more sequencing is now on the way. As summary, I just mentioned earlier, we carried out comprehensive immunohistochemistry-guided gene-targeted sequencing technology, and we got somatic mutation hits in about 72% of the CPAs. We saw that overt CPAs presented a different, a distinct mutation profile as opposed to the main CPAs, where PRK-CPAs dominated the overt CPAs, and beta-catenins dominated the main CPAs. We were also able to show that targeted RNA-seq could really differentiate between CPAs of different genotypes. Lastly, our data also shows that we can utilize FFP samples for whole exome sequencing, and we can really merge the targeted RNA-seq and whole exome sequencing technologies to sort of investigate and use it as a tool to investigate new cortisol driver mutations in CPAs. This research is really a team effort here, and I want to thank all our collaborators and the Rainey Lab at the University of Michigan. I would especially like to say thank you to Adina Turku Aaron Udiger and Tom Giordano, and all our international collaborators from across the world. So Irina Bancos at Mayo, Tracy Wong at Wisconsin, Mohamed Habra at the NIH, Matt Luth at Vanderbilt. We also have Martin Fasnacht and Anna Rister in Germany, Konstantin Stradakis in Greece, and of course Kazu Namba, who's been an integral part of this project from Japan. Thank you. Thank you very much for this very nice presentation. So you are essentially looking at somatic mutations and finding, and Christina also had some additional mutations, but did you look at copy number variants? Peter Kamenicki showed that in the patient with the cortisol, GIP-dependent cortisol adenoma, there's gene rearrangements. So chromosome 19 and 20 rearrangement leading to GIP overexpression. So did you look at possibilities of gene rearrangements or copy number variation? We really need to look really into our data. We have the data, but we haven't looked into that as yet. Hi, it's Maria Christina Zanaro. I have a question. So when you do FFP immunohistochemistry guided NGS or sequencing, but anyway, do you have very often those double nodules in cortisol producing adenoma? And do you think actually the better mutation detection, rate of mutation detection depends on the NGS being used instead of Sanger or the possibility of detecting this other nodules, the secondary nodules or whatever? I think it's a combination of both because with the immunohistochemistry, we can really dive in and really get into the regions that are positive for CYP17, 3-beta-HSD. And we also have done immunohistochemistry for CYP11A1B1, which sort of seems to track with the other two. And yes, definitely NGS, especially because we are using FFP samples. So a lot of these samples are degraded and the NGS technology being so effective, I think it's really a combination of both. Okay, can I ask another question? Yes, please. Yes, so the second question is, with regards to the degradation of FFP DNA, I mean, what is your success rates when doing whole exome sequencing on those samples? Is this working very well, like when you do panels, NGS panels? So we haven't done a whole lot of whole exome sequencing. We probably just done sequencing six or seven samples as yet. But we have had hits. As far as NGS goes, yes, we've had some failures as well. So say out of 100 CPAs, we have 5% that have been degraded and we couldn't use it. We went and sort of isolated more DNA and yeah, it just didn't work. So we do have a failure rate as well, but it's a really small one. But it's working. Well, that's encouraging. Thank you. I find it very interesting that the so-called mild dysregulated cortisol-serving adenomas have more of the beta-catenin mutations than the PKA, which is logical. But I would like to know in the tumors which are called so-called non-secreting, so if the levels are at 45 or 40 or 35, et cetera, do you see similar mutations in these samples? So actually that's the next project on the list. So we already have DNA isolated. It's probably being done for NGS right now as we talk. Yes, Larry. So yeah, Larry Kirshner from Ohio State. So I may have missed it, but how many genes were you actually sequencing in your targeted panel from the DNA analysis? So as far as CPA-related genes are concerned, we have about six, and we have another six or seven for the APA-related genes. So at the moment, we have about 13 or 14 genes that we can run simultaneously on the sequencing panel. So my follow-up question is, I mean, I think you're, I don't know if you've thought about this, you know, you're only gonna find the things that you're looking for. Yeah. So particularly those, you know, you had a number of samples where they didn't group with the PKA group or the, you know, have you thought about looking for secondary mutations in ones that already have one mutation, but they may have other things that would be quite interesting? Not as yet, but yeah, this is something really interesting, and yeah, it would be something. But then again, whole exome sequencing can again be such an expensive technology. So we haven't reached that part as yet, but yeah. Thank you. Since there is place for questions. So I just wonder in your CPA, which was an ACC, so actually did you sequence different regions? Do you have a genetic heterogeneity? So unfortunately for that one, we just had one tumor and it was just one big tumor and it had parts that seemed just different. So we did not, we were not able to get separate. And within the tumor, did you sequence this different? No, we just got the whole tumor as it is. And perhaps, you know, using that strategy would have really given us, you know, maybe two hits or we could have known what hit lies in the adenoma. So, but yeah, we need to keep our eyes more open it seems, you know, just getting under the microscope and just looking at the immunohistochemistry. Okay, thank you. Thank you. No other questions. Thank you very much. That was very interesting. Thank you. Thank you. Our next speaker is Christina Honk. She is from University of Birmingham and she's talking about diagnosis and pathogenesis of adrenal tumors. New pathogenic insights from genetic screening of adrenocortical tumors. Thank you very much for the introduction and good afternoon, everyone. I would like to thank the organizing committee for giving me the opportunity to talk about some of our data on the pathogenesis of adrenocortical tumors. Have nothing to disclose. And first of all, a brief introduction about adrenocortical tumors. As you know, they're mostly represented by benign adrenocortical adenomas. These are frequent in the general population and much more rarely by malignant adrenocortical carcinoma or ACC that are quite aggressive and generally associated with a poor prognosis. But talking about the adenomas, as you know, the large majority are endocrine inactive or non-functioning and are usually incidentally detected. But they can also, so the adenomas can also produce cortisol in excess. And in this case, the majority of them are associated with a mild cortisol autonomous secretion. And in a minority can be associated with overt Cushing syndrome, as you can see here. Of course, adenoma can also produce aldosterone in excess, but this will not be part of my talk today. We will concentrate on, oh, sorry. We concentrate on the non-aldosterone producing adrenocortical adenomas from now on called NAPACA. And if we look at what is currently known about the pathogenesis of adrenocortical tumors, as has been already introduced by Julie, about 50% of the endocrine inactive or non-functioning harbor alteration that induce an activation of the Wnt beta-catenin pathway. And similar alteration can also be observed in about one third of the ACC. So keep this in mind. And if we look at the cortisol producing adenomas, the majority of them harbor alterations in the CMP-PKA pathway. And with a different proportion, also depending if you're talking about overt Cushing or mild autonomous cortisol secretion. So here we come to two major questions in this field. One is, what are the molecular mechanisms involved in the early tumor development in the adrenal gland? And the other one is, what are the molecular pathogenetic mechanisms involved in autonomous cortisol secretion in the remaining proportion of CPA? Of course, if we look at the ACC, things are much more complicated with alterations also being present, such as copy number alterations, increased gene expression, especially in genes involved in the cell cycle, increased hypermethylation of the GPG islands, and also altered microRNA patterns. Some of these alterations have also been recognized to be potential prognostic markers or drug targets, but this will not be part of my talk today. So first of all, because I have to talk about genetic screening for the pathogenesis of adrenocortical tumors, I would like to present you some data about whole-exome sequencing, so the role of whole-exome sequencing in the pathogenesis of CPA associated with Hobart-Cushing syndrome. As Julie also has already shown, in 2014, our group, in collaboration with many centers in Europe and in the US, firstly detected mutations in the gene coding for the PRKAC, for the protein catalytic subunit of protein A, PRKAC gene. This was in one single hotspot somatic mutation that was detected in patients with Hobart-Cushing syndrome. Three other studies published in the same year demonstrated exactly the same mutation in the same population, and this was also confirmed by a large NSAT study that evaluated the presence of this mutation and the correlation with the clinical phenotype that corresponded with a more Hobart-Cushing syndrome. And our group also performed a functional validation of these mutations, as you can see here in these two panels. These experiments were done by FRET Microscopy in collaboration with the group of Davide Calibiro when he was still working in Wurzburg, and we could demonstrate, as you can see here in this panel, that the presence of PRKAC mutation were able to interfere with the binding between the catalytic and the regulatory subunit of the PKA, and therefore inducing a constitutive activation of the CMP-PKA pathway that therefore induced autonomous cortisol secretion by the adrenocortical cells. And I also wanted to highlight that this group, so our group, so-called PKA, that was a collaboration between the endocrinology and the pharmacology in Wurzburg also continued to work on this to further characterize the PRKAC mutations. And over the years, we were able to demonstrate that things are much more complicated than was thought at the beginning, with many more mutations being detected, all somatic mutations in the gene coded for the catalytic subunit, and that also these different mutations act through different mechanisms. So it's not only the interference between the catalytic and the regulatory subunit, but there are many other potential mechanisms, in particular an alteration of the substrate specificity. But now going back to the whole exome sequencing, we also performed this large multicentric study on behalf of the NSAT, where we investigated 99 NAPARCAS, including 74 CPA without the hotspot mutation in the PRKACA, and for the first time, also 25 endocrine inactive adenomas. And we were able to confirm presence of most frequent CTNNV1 mutations. We also described three new mutations in the PRKACA gene, and the rest of these are mutations in the GNAS gene. We also observed some mutation involved in the calcium metabolism of regulation, but these have not been validated further. I would like to focus your attention on this over 30% of cases where no mutations in driver genes have been detected. And this was also confirmed by an overview of the literature, similarly also to what Julie presented. Here we included all the studies available on whole exome sequencing, and you can see more than 30% of both CS-CPA and MAC-CPA did have no mutation in driver genes. Here you can also see the distribution of the most frequent somatic alteration, PRKACA gene, the GNAS, and the beta-catenin gene. So we wanted to investigate better those cases without mutations in known driver genes, so the so-called DAC better, and we wanted to try to answer to the questions about early tumor development and autonomous cortisol secretion. And to do this, we performed RNA-Seq on bulk tumor material, and gain using our NSAT cohort, and here we included 63 NAPACAs and also seven ACC for comparison. And this study was done by Guido Di Dalmazzi and Barbara Diri. And the first results of this study were about the correlation between the transcriptome profile and the known genetic background. And as you can see here in the principal component analysis, the cushion-associated CPA clustered apart from the other NAPACAs independently from the presence of mutations in the CMP-PKA pathway. While the MAC-CPA and the endocrine-inactive adenoma clustered together, again independently from the presence of beta-catenin mutations. The ACC clustered separately from all adenomas, but we also had a small group, only two, CPA with beta-catenin mutation that clustered somewhere between the adenomas and the ACC. And here are exactly the same results in the unsupervised cluster with heat map. So here is the mutation status and the presence of steroid secretion. And the second aim of the study was to detect another genetic alterations in adenomas without mutations in driver genes. And we could observe two further mutation in cushion-associated CPA, one in the PKACA gene and one in the JNS gene. And also one gene fusion between ACAP13 and PDE8A that you can see here, but it was not further validated. But we could conclude that there might be using different methods and different sequencing methodology, different results. And also we could conclude that there are likely still unrevealed alterations in both the CMPPKA pathway and in the Wnt-beta-catenin pathway that might be involved in the pathogenesis of adrenocortical tumors that we haven't detected yet using traditional methods. So at this point we decided it was time for something new and we decided to switch from bulk sequencing analysis to single-cell sequencing analysis. And I do not have to go too much into the details of the methodology, just wanted to mention that with bulk analysis we investigate, for instance, the gene expression of all genes in the entire tumor at once, but with single-cell analysis we first distinguish different subtypes of cell population and then investigate the gene expression in the different cell population. And these results are usually depicted using principal component analysis. So the aim of our study was first to investigate the transcriptome signaling of different cell subpopulation in the human adult normal adrenal gland. And from this then gain insights about the adrenal tumor heterogeneity and then from this obtain a comprehensive understanding of the adrenal tumorigenesis and the autonomous cortisol secretion. And this was a study in collaboration with Martin Fastrak in Wurzburg and Sascha Sauer in Berlin. And here's very briefly the method that we used for single-cell analysis. First of all, we used snap-frozen material and not fresh tumor material and we isolated single nuclei and not entire single cells using method well-described in the literature. Then we used an in-drop technology to obtain, to synthesize bar-coded cDNA in suspended single nuclei. The sequencing was performed using Illumina technology and then the most important part of the study, the bioinformatic analysis that was performed using a customized pipeline for transcriptomic data in single nuclei. And with this we obtained specific markers for each cell subtype and then we validated using two different methods, one targeted with immunochemistry that was performed by Barbara Altieri in Wurzburg and one untargeted with special transcriptomics that was performed by Karim Seshaner in Berlin. So first of all, I would like to show you our results on the normal human adult adrenal glands. And for this we used three samples coming from surrounding tissues of endocrine and active adenomas and three normal adrenal glands coming from patients operated for renal cell carcinoma. So these are the first results. This is a principal component analysis from the single cell analysis in normal adrenal gland. And you can see that we could distinguish a main cluster that corresponded to the adrenal cortex and we could also distinguish the three different zones in the adrenal cortex. Here are some examples with expression plots. You can see high expression here in yellow and that's only some examples of markers for the three zones. For instance, CYLT2A1 for the zona reticularis, CYLT17A1 for the zona fasciculata and CYLT11B2 for the zona glomerulosa. But we also could distinguish several satellite clusters. Just as an example, this represents the adrenal medulla and this is the fibroblast and connective tissue cluster that represents a capsule. And we also had lymphoid and myeloid cell cluster. I do not have time to show you the validation and expression plots for each of these sub-cluster. I only wanted to show you these two, probably the most interesting ones. One is the adrenal medullary progenitor cluster and here you can see some selected markers for this sub-cluster. Here you can see the adrenal medulla. Here is expression with the myelin plots of the chromogranin A. And here you can see synaptotagmin one and the neuroregulin one. And for the stem progenitor and vascular endothelial cell cluster, we also could select some markers. Here are some examples. This is the NR2F2 and this is the expression plot. This is a member of the nuclear receptor family and this is a recognized marker for adrenocortical stem cells. And this is expression of ID1 that is a known marker for mesenchymal stem cells. And here is just one slide about the validation that we performed with immunochemistry. Here is a dual staining for chromogranin A and synaptotagmin one for the validation of the adrenal medullary progenitors. And you can see here a co-expression of these two proteins in so-called sub-capsular cell clusters. And so from this and from other results that I do not have time to show, we could produce some evidence of the presence of cell clusters within adrenal medullary progenitors in the adult human adrenal glands, similarly to what has been already shown in mouse and in human fetal and postnatal adrenal glands. And this is the dual immunochemistry for the stem progenitors and vascular endothelial cells. This is dual staining for NR2F2 and ID1. And you can see very small adrenocortical so-called stem cell niche, also in a sub-capsular region. And from this we can hypothesize that ID1 could be a new marker of adrenocortical stem cells. And this is the validation that we performed by spatial transcriptomics, also in the normal adrenal glands. And we used for this 10 per vision spatial assay. And we used the method so-called module scoring, where multiple markers taken from the single cell RNA sequencing has been pulled together and then mapped onto two different slides of normal adrenal glands. And you can see we confirmed the localization of our markers for the three zones of the adrenal cortex. But maybe more interestingly, we could also confirm the localization of lymphoid and myeloid cluster. Lymphoid cluster mapping to the outer layers of the cortex. Myeloid cluster clustering to the blood vein in almost exclusive manner. And here you can see the stem progenitors and vascular endothelial cluster mapping to the sub-capsular region in this so-called stem cell niche. So to conclude about the normal adrenal gland, we could show the single nuclei RNA sequencing could represent a paradigm changing approach to provide insights in the physiology of the adult normal adrenal glands and also provide a kind of single cell atlas of human adult normal adrenal glands as shown here in the cartoon. We could also show evidence for two novel sub-capsular population, the adrenal medullary and Nr2F2 ID1 positive adrenocortical progenitors. And importantly, this could represent a resource for the entire research community that could help to identify mechanisms to contributing to disease as it has been recently published for the neuroblastoma. And exactly looking at this third point, I would like to show you just a couple of slides about our results on single nuclei RNA sequencing in NAPACA, so in adrenocortical adenomas. And for this, we investigated five endocrine inactive and seven, yes, CPA, so CPA associated with Dover-Cushion syndrome. And this is the integrated analysis for all the 18 samples, including the six normal adrenal glands and the 12 adenomas. And we could confirm the presence of the three zones of the adrenal cortex in all the samples and also the presence of the satellite clusters. But interestingly, in the adenomas, we also detected seven additional called adenoma-specific cell clusters. And here you can see the percentage of cell subpopulation in the different samples. Here we have the six normal adrenal glands, and here we have the 12 adenomas that are subdivided for diagnosis, so CPA and endocrine inactives. And here they are also separated for the genetic background. So we had two cases with PAKACA mutation, one case with EGNAS, three with no evident mutations in driver genes, and this with a beta-catenin mutation. Again, I do not have time to go into the details for each of the adenoma cluster, but I want to show you the results for the so-called tumor cluster one that is here represented in orange. And it was highly expressed in two samples affected by mutation in the beta-catenin gene. And we performed a pathway analysis of all the genes overexpressed in this tumor cluster, and you can see they were enriched in particular in splicing, extracellular matrix, and interestingly, in the cell proliferation and tissue development. And so we looked more in details in the genes involved in the cell proliferation that were overexpressed specifically in this tumor cluster, and you can see that three of them were also involved in the Wnt-beta-catenin pathway regulation. And in particular, AFF3 and ISM1 have been already described to be overexpressed in ACC. And among all of this, we thought that AFF3 is probably the most interesting one because it has been already reported to be a mediator of the oncogenic effect of beta-catenin in ACC. But I remember you, this was overexpressed in our cohort of adenomas. There were two adenomas with beta-catenin mutation. So we asked ourselves if this could be one of the mechanisms involved in early tumor development and potential involved in the rare event of an adenoma carcinoma sequence. So with this, I would like to conclude that our single nuclei RNA sequencing allowed us a comparative analysis of a series of normal adrenal gland and adenomas and revealed seven APAC-specific clusters and therefore can represent an innovative and an alternative approach to increase the resolution of molecular pathogenesis of adrenocortical tumors and help to answer to still open questions. And of course, now our next step will be to investigate the same approach, also a series of ACC, and this is already a work in progress and we hope that these results could help us to even better understand the early tumorigenesis of adrenocortical tumors. And with this, we'd like to thank, of course, all the persons that have been involved in the study, particularly Barbara Altieri and Karim Sechener and all the NSAT centers that contributed to the whole axon sequencing and the Adrenal Tumor Group in Birmingham where I'm working now. And thank you all for your kind attention. Thank you. Thank you very much, Christina, for this very exciting new technical approach to single-cell analysis and micro-dissection of those tumors. I see that Bill already has a question. Beautiful presentation, very exciting data. I had a couple of questions if there's time, but I'll start with one. With ACC, you get quite a variation in signatures and I noticed with your single cell, you're seeing differences in heterogeneity. Did you have an immune signature that stood out in either of the mutations, the CPA versus the PRK versus GNOS? Yes, very good question. Maybe, can I go back to, no, impossible. Can you put the slides back on, please? There are slight differences, yes, in both the lymphoid and the myeloid cluster. We could see some differences between CPA and endocrine-inactive. We are now very excited to see the results for ACC. I can ask you the other question. Yes. With the aldosterone-producing nodules, we're very lucky to have CYP11B2 to guide us toward them. You've done so many samples for RNA-seq, et cetera. Are you finding any markers that would benefit us in the FFPE world to help us guide us in these nodules? Or should we just go with the biggest one that has CYP11B1 and go with it? Is there any marker that you trust? Not really, unfortunately. We have a cluster that we call the cholesterol. Pre-regulated genes that are typically overexpressed in all CPA, but it's a kind of a pool of markers. It's not just one that you can select. Maybe I missed the point, but I thought you were going to show us, because you're doing single cell within the same tumor, what's the heterogeneity from cell to cell in terms of the underlying gene? So, PKA, and what percentage of the cells is it mutated, et cetera? Yeah, that's a very good question. Unfortunately, from this specific data that I showed you, we cannot tell this. It would have been nice. But we are planning now another study with combined RNA-seq and whole exome sequencing that will answer to this question. Because these are pure RNA sequencing data, and we used relatively old technology, and we couldn't detect a mutation with this. Joanna Price, Sheffield. Beautiful data, Christina. You have data now on normal adrenal glands, and you've got it very well characterized. A clinical observation is that you get more incidentalomas on the left adrenal gland than the right adrenal gland. Have you found any differences between the left and the right? We looked at sex and age, but not at the lateralization yet. Good point. Interior limb versus exterior limb. So, Christina, I was really impressed by your data, especially the spatial transcriptome. It's really exciting. I just wondered, when you look into your RNA-seq data from single nuclei, you could read the mutations, maybe? No, no. As I said, unfortunately, no. Oh, okay, I missed that? Okay, I'm sorry. No, but it's in the plan for the future. Okay, okay, sorry. Hi, Lauren Fishbein from Colorado. Thank you for sharing that talk. It was really great. I was curious about the normal medulla and then the precursor cells, and just an interesting kind of observations about how they're not close to each other on the PCA at all. And I was wondering about some thoughts about that. And then also, other markers for normal adrenal didn't show, like tyrosine hydroxylase or PNMT. Did those confirm some of the genes that you did show us? Yes, so starting with the second one, yes. There are many more markers that I didn't have time to show. It's more complex than it is. And for your first question, well, this principal common analysis is not always, the distance is not proportional with what you expect. And this adrenal medullary, they are a bit of a combination between the cortex and the medulla. So they cluster near to the cortex, but it doesn't mean that they are far away. So do you think those precursor cells are coming from that junction? That's kind of what I was guessing based on the data you were showing us. Yeah, that's what I was saying. What about the population of subcapsular cells? Have you looked specifically at subcapsular cells? Is there supposed to play such an important role in? Yes, yes, we do have different, at the end of the analysis and from the validation, we think that in the subcapsular region, there are multiple different types of cells coming from different origins. So we have these stem cells, we have the adrenal medullary cells, we also have the lymphoid cluster, they all are in the subcapsular region. Can I go ahead? Yeah, okay, still time. So coming from the aldosterone field, I just have a question. Your patients, I mean, your subjects from the normal adrenals, were those young or old subjects? And do you think you identify in your clusters some APCCs, for instance? So aldosterone-producing cell clusters? Yes, yes, for the first question, yes, we checked the age, and they are all between 40, 50, 60. So unfortunately, we do not have very young. It would have been nice to see differences. We also have few males. So again, we also checked the differences in sex, but there are not enough males to make it statistically significant. And APCC, yes, we can see them. So you told us you did not look at right or left adrenal, but you look at sex. So can you tell us anything about the difference between males and females? Yes, we concentrated on that, but unfortunately, we have only two males among the normal adrenals versus four females, and this makes statistical power, yeah. Are there any other questions? Well, if not, I think that on behalf of all the audience, we would like to thank all of our three speakers for a fantastic symposium, and the participants for excellent questions and comments. And this is, I think, one of the last, I think there's a social event now that everybody's invited to, so hope to see you there. Thank you.

adrenal tumors

H1MR spectroscopy

succinate accumulation

SDH mutated tumors

succinate exploration

genomic analysis

transcriptomic analysis

whole-exome sequencing

somatic mutations

RNA sequencing

cell clusters

single-cell RNA sequencing