Multi omics in ICU_Dr Pradeep Rangappa

Added: 2026-06-02

[00:00:03]Hello friends.

[00:00:04]I'll be giving this very contemporary topic, multiomics in critical illness from molecules to meaning.

[00:00:12]Uh so, let me warn my listeners so this can be a very uh dry and quite a heavy-duty topic on our mind.

[00:00:22]So, so I was asked to deliberate this in Kolkata on 31st May 2026 in their annual ISCCM meet.

[00:00:30]Uh so, fairly challenging topic. It's a fairly contemporary, fairly futuristic, and a very theoretical, uh and very hypothetical also to some extent. Trying to make semblance as to how our health care will be uh navigating the future, and how we would be looking at it in next maybe 5 years. So, but uh it is imperative that we understand this whole dimension of multiomics because this is going to be the future of health care.

[00:01:03]And what is the current work that's happening in critical illness?

[00:01:07]So, I would say this is one of the component that would be a part of AI in critical illness. So, if you look into artificial intelligence as one big bubble, so multiomics is a part of that bubble. So, it is a it is embedded within that bubble. So, I think that is the best sort of a uh concept I can give you.

[00:01:25]Uh so, just to get and this can definitely be a question in DRNB or DM.

[00:01:30]So, please pay attention. It's not an easy topic to read.

[00:01:33]Uh so, what is omics?

[00:01:36]So, omics simply means a large data set.

[00:01:39]So, you take a large data set and look into the sort of a pattern or pattern recognition within that data, and uh make semblance with regards to your ability to predict certain things. So, that's the sort of sort of a crude understanding of what omics is. And multi is where you incorporate and integrate large data sets from different dimensions which would influence certain prognosis or therapeutic armamentarium in the health care. And in this particular case, I'll be talking about what are the omics that have been looked into in critical illness. So, this is what I would be looking at.

[00:02:18]So, why we need omics? So, the main reason why we need omics is we understand that any disease in critical illness is usually heterogen- heterogeneous. Like if you look into ARDS, that's a perfect example.

[00:02:32]Uh the lungs are very heterogeneous. And even the patient, the way they behave in ARDS, there is a lot of heterogeneity.

[00:02:39]And this is a problem in sepsis. And even if you take sepsis, it's a very heterogeneous problem because sepsis in one patient behaves very different to a sepsis in another patient. They don't behave similarly. And despite the best evidence-based sort of a medicine application in patients of sepsis, mortality still sits at around 30% in septic shock. It can go up to 40%.

[00:03:02]And why we omics is something that may have a bearing because there are more than 50 randomized controlled trials in last two to three decades that have failed to have a impactful difference in the outcome to the patients in sepsis or ARDS.

[00:03:18]And there are 50 trials in both put together which have failed over three decades of research.

[00:03:25]So, there we have looked into five large data sets which we call it as a five layers in omics and which on the face of it you wouldn't be able to make out unless we look at it in a systematic way. Which means these are layers which are existent within the host, within the disease, which are obviously not evident unless we look into the omic sort of a optics and make sense of it. So, it's a unique biology which is inevitable invisible to the general clinician these sort of a data which are present within us within the patient and within the disease sort of a factors.

[00:04:08]So, when you look into the multiomics landscape in critically ill we can divide it into five broad divisions. So, what are those five broad divisions?

[00:04:18]First one is genomic which means the genetic makeup of an individual or genetic makeup of the disease factors have an enormous bearing on how we respond to the disease or how the disease behaves.

[00:04:32]Then we have transcriptomics. So, what is transcriptomics? Transcriptomics simply means how the host variables how the host reacts to a disease reacts to a drug. So, it is the host factors how they behave when they get affected by a disease, they get affected by cancer, they get affected by an infection, they're exposed to the how host behaves.

[00:04:57]If you look into the sepsis definition it is life-threatening organ dysfunction due to dysregulated host response. That dysregulation that happens in the host and the factors that influence that regulation is what we call as transcriptomics.

[00:05:13]Then proteomics as the name suggests it's all the different different there are millions of proteins in our body protein signals so, how they respond to different disease patterns is what is proteomics.

[00:05:27]Metabolomics every intensivist is very adept to understanding. It's all the metabolic profile of an individual. When I say metabolic profile we are just looking at sodium, potassium, chloride, calcium, magnesium. There are other zillions. We look at the leucine levels, we look at isoleucine levels, we look at tryptophan levels. So, it's all basically making it very granular and looking into all the different elements is metabolomics. Then you have metagenomics. So these are the five important components in the landscape of multiomics.

[00:05:57]So let us start with genomics. Very simple. Genetic architecture or genetic makeup of a individual, we all understand has a bearing on acquiring disease, response to the disease, recovery from the disease, and wellness also.

[00:06:15]So one thing we have looked at is into SNP, which is single nucleotide polymorphism. So the single nucleotide polymorphism, see if figuratively if you see, if in a general population you have this, there is some changes that happen in one of the genetic alleles, which we call it as single nucleotide morphing polymorphism, which happens in certain individuals.

[00:06:38]That may make that individual vulnerable to the disease. They become susceptible to the disease because of the change in one of the genetic allele, which we call it a single It's a single nucleotide in the gene that undergoes certain polymorphism, which can make an individual susceptible to the disease.

[00:06:57]And which influences the individual and their response to the drugs that are given to the antibodies given, to some of the drugs, how they respond, and clinical outcome. So this polymorphism, this change in the genetic architecture, how it has bearing on the disease susceptibility, drug response, and clinical outcome is what the omics looks into. Then this is the terminology or nomenclature which all my listeners should possibly bear in mind because going into the future a future we'll be talking about GWAS. GWAS is genome-wide association studies. So what they do is they look They have three or four stages. There in the stage one they take a small cohort and they study the genes.

[00:07:43]And they do with a liberal P value, then they apply that into a larger group of patients with a stringent P value, and then they extrapolate into a much larger population with more stringency in the P value. So, this is a some sort of a adoption to try and look into the cohorts where certain genetic uh sort of a polymorphisms prevail and try to categorize them as to how they respond to the diseases. So, this is sort of genome This is a part of genome. So, which is GWAS, which is genome-wide association studies. So, the question is has this been applied in critically ill patients? So, we look into it.

[00:08:24]So, this GWAS, which is genome-wide associated studies, has been applied in patients with ARDS. So, what they have found is they have found this susceptibility locus. So, the single nucleotide polymorphism has been identified to this particular gene, 12p13.2, and they have identified two elements, which is BORCS5.

[00:08:47]BORCS5 is block one related complex subunit five within that nucleotide, and they have identified dual specific phosphatase 6. So, these are the two sort of a elements that they have looked at in SNP and looked into the patient's vulnerability by doing the GWAS, which is gene-wide associated sort of a survey to look into the vulnerability to ARDS. And they have found that patients who have this susceptibility locus of BORCS5 uh have shown increased lysosomal trafficking within the cells. There is increased lysosomal trafficking, which predisposes someone to ARDS. Then, there is lot of intracellular signaling mechanisms that tend to get activated because of this sort of a change that happens within the genetic loci which puts patients at a risk for ARDS because of this lysosomal sort of a trafficking and sort of abnormal cell signaling that tends to happen. And dual specific phosphatase 16 regulates so which is MAP kinase which is mitogen activated protein kinase. So this has a this is a sort of a chain that happens in the nucleotide which regulates MAP kinase because because of this dual specific phosphatase there is which which which which is an apparently that is present the patient become vulnerable for the cytokine storm and inflammatory cascade. So this is this is what has come out of GWAS analysis which is gene wide associated survey and and this leads to endothelial dysfunction also.

[00:10:26]So this is how genomics are helping us in trying to understand the pathobiology of identifying patients who are at a risk of developing ARDS by identifying this loci in the patients who may become vulnerable to ARDS.

[00:10:40]And what is the potential of these two?

[00:10:43]This can be used as a biomarker for risk stratification of patients with ARDS trying to look into the phenotyping of the ARDS because we have understood that now there is hyper inflammatory ARDS and there is a hypo inflammatory ARDS so it may help us in identifying the phenotype and the future is to try and target for a precision medicine treatment. So that is where identification of these alleles have found to be helpful. And they have even done GWAS meta-analysis which is gene wide associated studies. They have done meta-analysis and they have found the cohorts from the US and the cohorts from the Europe behave distinctly differently with regard to this ARDS or any other diseases because of the changes in these sort of a polymorphism that prevails.

[00:11:30]And they have found that Mendelian randomization of the data sets has shown CRP and interleukin 10 also the levels also have a bearing based on these sort of a polymorphism that prevail and functional integration of downstream molecules and a high and upstream anchorage of these downstream molecules and functional integration of this has a bearing with regards to ascertaining the severity of ARDS. So these are all the future as to how the science is progressing by trying to identify certain more polymorphism and looking into the type of inflammatory cascades that individual may undergo is what has been looked at.

[00:12:13]So if you look into transcriptomics, so we finished the genomics. So let's move into the transcriptomics.

[00:12:20]So as I said, transcriptomics is trying to look into the host response to the inflammation or the insult that happens.

[00:12:28]So they've identified nine genes which are predictor of ARDS mortality and you'll see with those nine genetic sort of alleles they've found, the predictor was around AUC was 0.83. But when they have integrated integrated host and microbe classifier, the AUC was 0.79. And when they looked at host transcriptomic classifier in sepsis, the AUC was 0.75. And if you see the ability to predict using genomics and transcriptomics was much better than Apache 3. So basically this slide is only showing you by incorporating the genomics, your ability to predict ARDS mortality was better than looking at only the host transcriptomics in sepsis and all this was superior to Apache as a predictor.

[00:13:21]So that's what this one will show.

[00:13:23]And what does transcriptomics show? So the this is a neutrophil degranulation.

[00:13:29]So, the neutrophil degranulation releasing granules alongs with the T cell signaling. So, absence of T cells.

[00:13:36]So, there is a T cells which are meant to pass on the signals. Absence of T cell signaling and neutrophil degranulation are shown to be associated with mortality in sepsis. So, when So, all these are the host mechanism. So, in the host, they look at whether there is signaling happening from T cells or T cells have remained dormant. If there is no signaling that happens, then it means the defense response is not there, mortality is higher. And host and the neutrophils are getting degranulated.

[00:14:07]That That also has a bearing on sepsis mortality. And they have looked at suppression of interferons in the sepsis. So, interferons can be protective. If there is a suppression of interferons, it is shown to lead to hyperinflammatory sort of a sepsis. So, these are all the transcriptomics which is study of the host. So, in the store host, they look into this and and they look into alteration of the mitochondria. Alteration in the mitochondria and it's associated with mortality are shown to identify endotypes. So, your whole transcriptomics is to look into the response of the host to this sepsis or to the ARDS by looking into these elements. So, T cell trafficking, interferon suppression, mitochondrial sort of a dysfunction that may be happening or neutrophil degranulation.

[00:14:59]All these are the modalities they're looking at to look into the type of host response that tends to happen in different disease conditions. So, that is our understanding of transcriptomics.

[00:15:09]So, we finished genomics, we finished transcriptomics. So, let's move into proteomics. So, what does proteomics do?

[00:15:16]Proteomics, they take a blood sample, they look into all the different protein elements. They have identified 36 eight to 36 candidate proteins which have added bearing.

[00:15:29]And they have labeled each of these proteins. So, proteins is not like tryptophan, lysine, isoleucine, and this is all Greek and Latin for us. The type of proteins they've identified is they have looked into this protein validation, and these are the names of these proteins. VCAM, this big LDH.

[00:15:48]And MBLs. So, these are the proteins.

[00:15:51]They've identified eight proteins. VCAM, LDH, BML, FLG2, so it's all the Greek and Latin. Or and LBP, I hope it is lipoprotein binding protein. So, these are the proteins which has correlation in prognosticating ARDS patients. This is proteomics. Proteomics is not studying the crude proteins that we have learned. It is identifying the elemental aspect of different proteins or the subunit of the proteins which are put that they study these eight proteins and see whether they it has a bearing with the outcomes in ARDS patients. And these proteins, eight proteins that were identified out of 36, were known to differentiate between the COVID-19 sort of an ARDS and the other bacteria-mediated ARDS.

[00:16:37]And by studying these pathway signals, one has an ability to look into the cell adhesion molecules, coagulation sort of a pathway, and sphingolipid signaling. So, basically these proteins are the ones which are needed for good cell adhesion, coagulation cascade functioning, and sphingolipid signaling which are all sort of a protective in nature. So, and if you say this is only theory, they have done studies to see if these eight proteins has had a bearing in predictive analysis, and they found that the area under curve by using proteomics in determining the prognosis was 0.893 with a clinical model as you see area under curve was only 0.784 and when they incorporated all these eight proteins it was 0.803. So it had a better predictive ability by looking into protein acid proteomics helps you to better stratify the risk proportion in ARDS. So we finished genomics, transcriptomics, proteomics. So let's look into metabolomics. So metabolomics is they have looked into amino acid dysregulation and they found this metabolic disruption has a linkage with the sepsis outcome.

[00:17:54]So here they have looked at integrating metabolic metabolomics which is metabolic profile with proteomics and had a good ability to predict the outcome in sepsis and they looked into one metabolic parameter serum mannose.

[00:18:08]The increase in the serum mannose has shown to put the patients at a lower risk for ARDS and having a high serum mannose has shown to improve 28-day and 60-day survival by doing Mendelian randomization. So maybe this is a project one can take on and discuss with your labs if you can measure serum mannose because if you have a high serum mannose it is shown to reduce. All this has been made possible by looking into millions of proteins and seeing the correlation of which proteins with regards to this disease patterns.

[00:18:41]And lipid signals they have shown that reduction in unsaturated long-chain phosphatidylcholines and increase in the kynurenine has shown to predict mortality in septic shock. So this is metabolomics where you look into certain metabolic elements. It's not the normal calcium magnesium it is these other elements that we look which are shown to have some correlation with mortality in sepsis and survival in ARDS friends.

[00:19:11]And they have looked into energy shift when there's increased glycolysis or absence of fatty acid oxidation, it is shown to increase the hyperinflammatory phenotype in sepsis. Which means if there is hyperglycolysis and absence of fatty acid oxidation, these were associated with hyperinflammatory type of sepsis and ARDS.

[00:19:35]So, we finished genomics, we finished proteomics, we finished transcriptomics, we finished metabolomics. Now, the science is how we integrate all these four. The proteomics, genomics, metabolomics, transcriptomics. The whole science is integration of this is shown to increase your ability, further increase your ability to identify. So, the GWAS, which is genome-wide association study, and integration with all other things has been able to predict ARDS mortality better. And as you saw, the integration area under curve was 0.79, which was much higher than Apache 3.

[00:20:10]So, that's about the whole meta meta all the omics multiomics. Then let's look into endotypes. So, they've looked into practical application of this. So, they've looked These are the two categories. There is hyperinflammatory sepsis or ARDS and hypo-inflammatory.

[00:20:26]Then there are biomarkers which are specific to this. So, interleukin 6 is high in hyper hyperinflammatory type phenotype. And interleukin 8 is high and tumor necrosis factor alpha. These biomarkers remain high in hyperinflammatory. In hypo-inflammatory, they were found to have low immune markers. And with regards to metabolism, there is increased glycolysis in hyperinflammatory and impaired mitochondrial function in hyperinflammatory.

[00:20:56]In hypo-inflammatory, there is absence of fatty acid oxidation. Outcome in hyperinflammatory, they had higher mortality. In hypo-inflammatory, they had they were found to have differential response to the fluid responsiveness.

[00:21:10]And treatment target in hyper inflammatory is you can use immunomodulation like steroids, immunoglobulins, so on and so forth. And for hypo inflammatory, immune reconstitution has been put in place.

[00:21:22]And transcriptomics, as I said, in hyper inflammatory, there is suppression of interferon, which I mentioned earlier.

[00:21:28]In hypo inflammatory, T cells become dormant. There is no signals coming from the T cells because T cells are the defense cells. They remain dormant and there is no signaling activation that happens, and that becomes altered in hyper inflammatory. So, this is the endotypes that were derived from multiomics.

[00:21:46]So, what are the challenges and the translational gap by in applying omics for our future as a speed because at this point of time, there's no point of care for omics, no rapid point of care, and we don't know the turnaround time. And these tests are not standardized. There are gaps in standardization. There is platform heterogeneity that tends to happen.

[00:22:14]So, and there is platform heterogeneity, and there is diverse analytics that tends to happen, which impede cross-center reproducibility. So, these are some So, standardization is an issue, so which has an bearing on these reproducibility. And none of these are validated on large scale, and these biomarkers uh are not fully validated, and they don't come from prospective RCTs. So, that is also something that we need to So, most of it is from a retrospective data, and the the operational sort of a inefficiency remains, data privacy remains, computational complexity remains with these new tests. And in whether how effectively can these be integrated into the clinical pathways are all the test of time at this point of time. So, there's no standardization, no validation of these biomarkers, and operationally, we do not know how feasible it is, data privacy, computational quantity, how how effectively we can integrate into our clinical pathways. All these remains to be seen. So, just coming to the last one or two slides, future directions. What is the future directions towards precision? So, there are no no standardized pipelines, and integration of the clinical pathways with the prospectively validated interventions is what needs to be put in place to make this successful, and real-time analytics have to happen with by integrating machine learning algorithms with the omics for rapid actions that need to be put in place, and network analysis, integration with multiomics, and to prioritize these certain conditions like septic cardiomyopathy and ARDS is the future, and collaborative cohort with large diverse longitudinal data to create harmonized protocols to make it more generalizable is the future for this multiomics, and using Bayesian models to integrate genomics into early warning signs for septic shock outcome. So, these are some of the future sort of a work that is happening.

[00:24:21]So, we need standardized pipeline, real-time analytics, network analysis, collaborative cohort with a harmonized protocol for generalizable Bayesian models. So, the key takeaway, friends, I know it's a very complex, very contemporary dry topic, but I've tried to simplify to the best of my ability.

[00:24:38]The key take takeaway is there is molecular heterogeneity. So, what we have understood is molecular heterogeneity is what is leading to clinical heterogeneity in the patients, and the disease we may say 10 patients have ARDS, but 10 patients have different biology and different outcomes. So, if you understand this, this is where omics will help us to calibrate our precision medicine. If you have understood that at the end of my talk, I would have done justice. And each omics, when I say each omics, it's genomics, proteomics, transcriptomics, metabolomics, each one has different blueprint altogether, unique biology, and all these have very different signals that are conveyed. But, integration of these four mix or multiomics is the way forward in being more precise in our approach.

[00:25:33]And as I said, integrating this omics has increased our area under curve and we have seen in some of the studies in ARDS and sepsis. And there have been 14 points gained by integrating omics.

[00:25:47]That's why multiomics is better than one single omics. And endotype-guided treatment is the next frontier. And and this is found to be very valuable in importantly two subgroups, trauma patients, sepsis patients, and ARDS. And this this remains as a proof of concept.

[00:26:05]But, what is a challenge is validation and standardization appears extremely critical. And we need more prospective sort of a data on randomized controlled trials and molecular enrichment for validating some of the biomarkers that emanate from these omics, friends. So, that's about it, friends. So, request you all to submit your valuable work to Journal of Critical Care. And you can visit my website to react to this lecture. Thank you. Thanks again, everyone.