Personalized Biomarkers to Predict Surgery Outcomes

Where does this passion stem from? You know, I see all of these like webinars and news articles and, and, you know, you’re, you’re quoted in there. Where does this, this passion come from?

Chris Ames (00:51.118)

Well, I’ll tell you this story. It’s been really, I’ve been a spine surgeon now at UCSF for about 20 years. And over time, my practice has increasingly involved spinal deformity, especially in the aging population. And really, if you look across the world at the demographic and economic data, we see that we have an aging population in the United States, in Asia, in areas of Latin America as well, and in Europe. And

Scoliosis, adult onset scoliosis, is really also a disease of aging because we don’t see degenerative scoliosis in the younger population. What we have though, however, we have a high prevalence of the disease and we have an aging population, but we don’t have a good mechanism to risk stratify those patients and to inform our treatment planning and trying to optimize.

those patients. So I began a really deep dive in large-scale registry data probably about 12, 13 years ago when I joined the largest North American registry for adult scoliosis. And we combined forces with the European registry, and I led that data effort. And the idea was if we took a really deep dive into the data that we were collecting,

we could potentially identify specific, disease-specific frailty scores, for example, and we can talk about frailty a bit later, that may be predictive of complication and may lead us in a direction that may allow us to do some patient optimization. From that, we realized that if we only looked at frailty apart from other aspects of the patient, including the procedure itself,

we were missing a large part of the variation and our predictive models were not very accurate. So then we started doing sort of whole data, AI-based machine learning and predictive modeling. And even when we did that, we were still missing about 50%. We thought initially, well, probably what we’re really missing is the number of patients. And we had some…

Hannah Went (03:04.179)

Wow.

Chris Ames (03:12.85)

Criticism of that work saying, well, you guys really need 5,000, 10,000 patients. But it’s interesting when you look at other large-scale registries like hip and knee in certain countries that have hundreds of thousands of patients, they weren’t getting a lot more accurate either. And so we added more of our traditional data points, things that we collect, like general assays of their disability, their ASA grade, their comorbidity score, their bone density. And we were not…

and we were collecting more and more patients, and you know what, we weren’t getting more accurate. So that told us really this is a new world, a new phase of our research, and if we’re gonna make these procedures safer, more accessible for patients across the world, and we’re really gonna have to look at different data fields. And that’s what started us really into what I call now the data hunt phase, or the…

Hannah Went (03:45.447)

Hmm.

Hannah Went (04:03.623)

Sure.

Hannah Went (04:09.36)

Mm-hmm.

Chris Ames (04:09.558)

the data point phase, the variable phase, biomarker phase, however you want to characterize it, to try to identify new and novel biomarkers that may help us to better predict the outcome of surgery. And I think to the point of maybe this podcast, even more importantly, to identify biomarkers that potentially we could track as we’re optimizing patients for surgery. So something that we could look at and say, you know what?

this variable which may predict complication is something we can follow through their prehabilitation process and their optimization process. And you know what? We’re having an impact. This patient is modifiable. This patient is getting better. So that’s probably a long-winded way of describing the journey of how we got to where we are today and why we’re doing some work with some of these aging biomarkers.

Hannah Went (05:04.163)

Yeah, no, that’s, that’s great. And again, I can see that, that passion coming through and shining through you and, uh, you’re, you’re really creating these, you know, I would almost call it like these personalized biomarkers to predict surgery outcomes. Do you think that would be an accurate way of saying that?

Chris Ames (05:19.55)

Yes, but not necessarily in isolation. So we’re looking always for new and novel biomarkers that are different from what we’re currently collecting and probably a lot of them will be biological and aging related. Some of them may be something as simple as the patient’s inflammatory state. So, you know, to just, I know we’re gonna talk about the aging biomarkers, but I wanna just put it

Hannah Went (05:43.731)

Mm-hmm.

Chris Ames (05:49.158)

in sort of the general context of things that we were missing when we initially went down this road. One of the things we were missing were physical markers of sarcopenia, which is a general state of muscular deterioration as patients age. And initially we were doing no objective physical assessment of their physical strength, of their walking speed, of their gait.

Hannah Went (05:54.971)

Yeah, please.

Hannah Went (06:11.609)

Mm-hmm.

Chris Ames (06:15.782)

actually radiographic, what we call now radiomics markers, of their muscular integrity, the fatty infiltration of their muscle. We also realized that, as you know, Hannah, working in aging, there are patients that the aging scientists call inflamators, and some patients become more and more inflamed as they age and develop cardiovascular disease, and yet initially we weren’t collecting the C-reactive protein or the erythrocytes sedimentation rate.

which are markers of general levels of inflammation in patients. And they’ve also been linked to complication separately from any age-related marker, just the CRP level. We have a paper this year showing CRP is linked to risk of major complication after spine surgery. So the idea is not to say, OK, one biomarker. This is going to be directly linked to complication, but to say, look, when we initially

we were missing really a full characterization of the patient’s pace of aging, of the patient’s inflammatory state, of the patient’s physical strength and sarcopenic state. We weren’t really doing it in a proper way to data-fy that patient for prediction. So that’s what really changed the focus of a lot of our research and changed me from just kind of a registry researcher.

Hannah Went (07:19.315)

Mm-hmm.

Chris Ames (07:41.798)

into a biomarker hunter for the adult deformity population. And I might just wanna say one other word for the listeners who may not know a lot about adult deformity, but it’s a very good model for complex procedures in older patients. So if you look at it like a disease model to study, these are significant traumatic injuries for older patients to have these procedures.

Hannah Went (07:51.216)

Mm-hmm.

Chris Ames (08:08.618)

So the blood loss, I think, in our recent paper, was about one and a half to two liters. These cases take anywhere from three to six to eight hours sometimes. And these patients, on average, they’re ages 65, 75 years old. So it really is a model of studying the aging population under significant traumatic and physiologic stress. And this may be, things that might be close to it, for example,

cardiothoracic surgery, major surgery for cancer, maybe liver transplant or organ transplant surgery, all of which are growing in prevalence in the aging population. So it’s a nice model, and I don’t want the listener to think this is something super specific in the weeds where we’re just looking at this for adult deformity. This is something really that, if you look at it, I want the listeners to think about it more as a model of an older patient under significant surgical physiologic stress.

Hannah Went (08:57.424)

Right.

Chris Ames (09:08.803)

and what it’s telling us about how to predict that patient’s risk.

Hannah Went (09:13.123)

Yeah, no, no, that definitely makes sense. And, um, that’s why I love the work that you’re doing, right? It can be applicable hopefully across, you know, all, all different types of surgeries, even though you again are very specified and, um, kind of looking at it from the adult deformity standpoint. So I think this is great and, um, can be applied. I mean, you know, everyone knows someone who’s going through some type of surgery. For example, my, my mother actually just got, and I don’t know the formal name. Um,

Uh, but just got back surgery actually, um, a couple of days ago, um, on, on Sunday. So, you know, she’s going through that process. She’s 51. Um, but it does make me worry, you know, about her quality of life, her healing process and all of her biomarkers and how that’s just affecting her, you know, both short term and long term as well. So I think even this data and all of the work you’re doing can be applied to, again, more of those, those general surgeries too. So very big fan of your work.

Of course, that is the main focus for our chat today is that epigenetic age in the spinal deformity paper. So what else, Dr. Ames, why exactly did you want to conduct this particular study? I know you’ve done a lot of work beforehand, like you said with going through the data hunt phase, you’re looking into frailty, the AI predictive modeling. What was different about this study or if you wanna go deeper into why exactly you performed the study, although you touched on that some.

Chris Ames (10:37.45)

Well, for the listeners that aren’t clinicians and maybe aren’t epigenetic scientists, and they just think of themselves maybe as a provider or a third party sitting in front of a patient, or just their daily lives, we all see patients that are 40 years old or 50 or 60 years old, and some of them look like they’re 30, they look half their age, and some of them look double their age.

Hannah Went (10:41.82)

Mm-hmm.

Chris Ames (11:06.542)

probably, as we say in research, some low-hanging fruit that may be present there in terms of something, a predictive biomarker. So we started out by, and at the time, as you know, being in that business, there were increasingly direct-to-consumer test kits that were being offered to measure telomere length. And so the true story, you know…

The reality was I saw these kits advertised. There was a company at the time called Tilo years and I did my own, you know telomere length and I thought oh, it’s really interesting Quite different from my chronological age and I thought well, this is something that is potentially so a really valuable untapped variable for surgical prediction, so just true story and We realized at the time that we were

quite lucky because at UCSF we had Elizabeth Blackburn, who’s an Elba Prize winner, who identified telomeres and characterized, you know, her life’s work has been characterizing telomeres in a relationship to all sorts of biological variability. And so we just took a long shot and we went to her lab and we said, you know, could we

Hannah Went (12:09.881)

Oh yeah.

Chris Ames (12:31.054)

could we look at telomere length, which is at the time, as someone outside the field, I thought, well, this is maybe a good biomarker. So we went to her lab and we thought we were gonna be shut down, but it’s really true. At UCSF, everyone’s really very collaborative. And she said, sure, we’ll give you a good rate, give you one of my top technicians, and see what you can do with preoperative whole blood, quantitative PCR for telomeres.

Hannah Went (12:45.971)

Mm-hmm.

Chris Ames (13:00.182)

So that was our initial study. It was a pilot study, it’s very labor intensive, and we did a pilot study of 43 patients, and we found at the time that the patients that had the shortest quartile of telomere length had a much higher risk of major complication at 90 days after spine surgery. And it was a very small, you know, pilot study, but it got a lot of attention, because I think the field…

of, and I’ll just again go beyond deformity, the field of musculoskeletal disease or musculoskeletal surgery knew that we needed new variables that were going to increase the accuracy of our predictive models. So that paper actually won the Hibbs Award at the Scoliosis Research Society, which is the top basic science award. It just shows like, and it was a small study, so it just showed that like, hey, the society is like, you know, we realize this is a pilot study.

but this is really good work. This is something where the field, you know, we need to go in this area for biomarker work. So that prompted us to say, you know, what other biomarkers of aging are out there? And it’s funny, at the time I saw TrueAge, you know, the company, your company, and True Diagnostics, and I said, oh, this is kind of a different approach utilizing epigenetics, and I did my own.

through diagnostic, through age test. And it was different from what we got from telomere aging. It was actually much more comprehensive and looked at different, looked at immune aging and other aspects of aging. And we realized that, you know, doing a deeper dive in telomeres, that there’s a lot of controversy about telomeres and it’s very like lab specific, it’s technique specific. There’s been data that shows that they are.

impactful and some that they aren’t as impactful in terms of prediction. So we wanted to move more into epigenetics and that’s where we found the collaboration with True Diagnostic. And we contacted True Diagnostic and we said we’d love to do epigenetic aging and see whether this is equally or better predictive variable. And that began our journey into the biological clocks that True Diagnostic offers.

Chris Ames (15:26.606)

And it’s a centralized lab, very standardized methodology. So we felt this was a good way to go rather than trying to do it in an independent lab, try to do it in more of a high throughput lab that offers direct to consumer and direct to research partnership. So that’s how we ended up at the point where we were going to do this study.

Hannah Went (15:48.691)

Yeah, I love hearing the background about how these studies come to fruition. Like you actually went and took the test yourself. You went through the process, you, you know, received your results back. So that’s been great. And I think if there’s one thing that I’ve learned from working with you and your team on this, uh, with true diagnostic UCSF is, it’s a very, very easy to work with and very collaborative. So you all make it easy on us, which I think is great. Um,

So we’re looking at these markers before and after, you know, these surgeries. Dr. Ames, what surgeries are we talking about here? Can you go into that maybe a little bit more? And probably something our listeners would be pretty unfamiliar with, so feel free to break that down as well.

Chris Ames (16:30.442)

Yeah, sure. So if we think about the study population that we included in the Dunedin PACE study that we just put online on a preprint server, that population was adult deformity patients. It all had… These aren’t the surgeries that, you know, typically your uncle or your friend had that have a one or two level fusion. These are patients that have, for example, the chin on chest that can’t look straight. The patients are way off.

leaning forward or way off to the side. They’re severely incapacitated, usually by their condition. We’ve done other comparative studies looking at their level of incapacitation. It’s the same as patients that have, for example, severe emphysema or chronic obstructive pulmonary disease, patients that have unilateral limb paralysis. In the cervical deformity population, what we call chin on chest, they’re on the level of patients that have blindness or severe visual impairment. So these are…

Hannah Went (17:29.903)

Oh, yeah.

Chris Ames (17:30.23)

very impaired patients and as I mentioned previously, the magnitude of the surgery is massive. So it’s a model of significant physiologic stress. But the interesting thing, Hannah, is that despite the magnitude of surgery and why we’re so passionate about it, is that it actually results in very significant clinical improvement for patients. So it increases their quality of life substantially.

compared to non-operative care. Once they’re moderately disabled and have failed non-operative care, surgery offers them the last hope to resume a functional life. And it has great success rates for improving quality of life. It’s been published many, many times by us and by the European group and by the Asian groups as well. However, the downside is it comes at a risk of high medical and surgical complication. So the idea is…

Can we identify populations of patients that may benefit from preoperative optimization? And the interesting thing about our predictive models was the one thing it told us in the pre-aging biomarker era was that more than half of the variability and outcome and complication were potentially modifiable variables. But many of these variables were subjectively assessed.

the patient’s not walking very well, or the anesthesiologist says, this patient looks pretty sick, you know, something in the ASA grade, it’s very, very subjective, or the patient reports something like, I feel imbalanced, or my social situation’s not good. So, a lot of it was very subjective reports. And so, the idea in the biomarker hunt is to identify objective variables and things that can be tracked.

Chris Ames (19:28.89)

I’m not sure the listeners know much about prehabilitation, but the instantaneous pace of aging markers, like Dunedin Pays, they are not just markers of someone’s instantaneous pace of aging, but they’re also a marker of someone’s current physiologic state. They increase in pregnancy, for example, and other physiologic insults to…

to patients and people. And so what we thought was, you know, DNA in PACE may be as an instantaneous pace of aging marker, may really be telling us something current about the current physiologic state of that patient that may be even more predictable than one of the, you know, the first generation clocks like Horvath or Hannum or something like that that was really trained to chronological age. It’s telling us something different. And in fact, in our study,

the do need and pace was not even correlated to their chronological age. So it’s really telling us something really interesting about their physiology, their current physiologic state. And as you know, potentially interventions may lead to improvement, such as weight loss and diet and other aspects of stress reduction, potentially. We don’t know. You probably know more about that area than I do.

Hannah Went (20:33.395)

Mm-hmm.

Hannah Went (20:46.919)

Mm-hmm.

Chris Ames (20:56.778)

optimization programs, it may be a biomarker we could follow for that process of optimization.

Hannah Went (21:03.567)

Yeah, yeah, definitely. And again, everyone who’s listening, if you’re not familiar with that, Dunedin Pace of Aging, that is going to be my first episode actually with Dr. Terry Moffitt. So go back and listen to that, but it’s gonna be, hey, how quickly are you aging biologically for every one chronological year, right? And like you said, it’s gonna be related and very elevated in things like pregnancy, physiological changes to the body, which make it a very good measurement here.

I thought it was interesting as well, Dr. Ames, they’re starting to create more and more of these epigenetic DNA methylation biomarkers where we even have a gate speed predictor, grip strength, VO2 max, FEV1. I know you mentioned that you measure CRP as another biomarker and how that’s related to major complication after spine surgery. And there’s even a DNA methylation CRP predictor. So I’d love to see your data applied to…

those algorithms and what those outcomes look like, but we’ll have to chat about that more offline. But I really do see this DNA methylation, epigenetic modifications and changes as the future for measuring these outcomes. Of course, I’ll be the first to tell you that I don’t think it’s going to replace everything, but I think it’s a really good start with all of the vast information that it can give us.

Maybe this was a question I should have asked at the beginning, but what are the current methods to stratify risk in surgery? Can you talk about that, I guess, before these epigenetic methylation biomarkers?

Chris Ames (22:36.046)

Sure. The history is not good. If you look back, it’s kind of funny. But again, the story is interesting. Really, the history of risk stratification started with something called the ASA grade, which was a scoring system developed by anesthesia, which was quite subjective. It actually has done pretty well, basically trying to, you know,

Hannah Went (22:39.895)

Yeah, well there you have it.

Chris Ames (23:04.406)

the anesthesiologist’s opinion about how sick the patient looks. It’s basically, you know, are they going to die on the table or they look pretty good or are they somewhere between? And it’s interesting, you look back at the original paper, I think it was in the 40s or something like that. It says in the original paper, this is not meant to be used for wrist stratification. It’s really funny. And what did they do for over 80 years? That was like the main stratification that they used. And

Hannah Went (23:10.31)

Yeah.

Hannah Went (23:27.706)

Oh no.

Chris Ames (23:29.73)

The problem was that it really wasn’t linked to anything that was quite objective. And so what came next was something called the Charleston comorbidity score, which included some aspect, at least of chronological age and comorbidity burdens across cardiac disease and things like this. And it was really designed more as kind of a mortality predictor. It’s used quite a bit in cancer. Again, it was not meant for surgical risk stratification. So that was next.

but it started to be used for that. And then what came next really was this whole idea of frailty, which has become just a complete dark abyss. The deeper you go, the more complicated it gets. Initially, frailty was developed as a concept mostly in geriatrics for heightened susceptibility to physiologic insults, increased vulnerability, if you will, and it was meant to be some sort of general aspect of the patient.

Hannah Went (24:02.14)

Mm-hmm.

Chris Ames (24:30.758)

All the specialists started to try to use frailty, and the most commonly used one is something called the Edmonton Frailty Score. There are 25 of them now, general and disease specific or more. Everyone has their own sort of model. But the idea was to try to measure something that is generally indicating kind of a general vulnerability, and it usually went across different organ systems, everything from balance.

social system, mental cognitive scores, physiologic status, et cetera. Again, it was quite subjective. Then we moved from general frailty scores, Edmonton, into what’s called disease-specific scores. So we developed one with the North American registry for adult deformity called the ASDFI, the Adult Spinal Deformity Frailty Index. And I was the senior author on that paper. And that was…

Hannah Went (25:06.379)

Mm-hmm.

Chris Ames (25:27.986)

linked to major complication rate, but it started to move away from being a general measure of vulnerability to sort of a simplistic complication scoring system or risk of complication scoring system, and it lacked really all objective data. And the problem with these risk scoring systems like those, like frailty and ASA and comorbidity,

Hannah Went (25:46.547)

Mm-hmm.

Chris Ames (25:55.582)

it just doesn’t really work unless it’s married in the data to the size of the procedure, for example. What operations is the patient having? So that’s why the disease-specific scores usually perform a bit better, because they’re a little bit more tailored to the disease, the procedure the patient’s having and what’s really wrong with them at that time, how are they presenting. But it still was missing the procedure itself. So then it moved from.

Hannah Went (26:07.434)

Mm.

Chris Ames (26:24.174)

frailty to predictive modeling. And predictive modeling involved both the patient, the frailty, comorbidity scores, what operation the patient was having, the pre-op disability levels. And that’s where we really are today, is in this predictive modeling. And it was those models that demonstrated two things. One was adding more patients, adding simply more patients to those models. And these were not

Hannah Went (26:26.94)

Mm-hmm.

Chris Ames (26:52.934)

simple variable models. These were 100, 200 variable models chosen from a thousand variable registries. So this is serious data. They were not getting more accurate with more patients. And also those predictive models are where we found that the greatest impact on complication was the patient substrate itself, the patient’s physiology, and things that were potentially

Chris Ames (27:22.862)

from predictive modeling to saying, wait a minute, those models aren’t gonna get better unless we put more important state-of-the-art variables into those models. And that’s where Dunede and Pace and maybe some other aspects of the aging biomarker field will find their way into these predictive models. But I do believe that they’re always gonna have to be part of a multivariable model to be ultimately accurate. So you’re gonna get a certain level of accuracy.

Hannah Went (27:48.161)

Mm-hmm.

Chris Ames (27:51.49)

just looking at do need and pace, for example. But if you really wanna get accurate, you’re gonna have to take do need and pace with a lot of other patient variables, sarcopenic variables, radiographic variables, and then you’re gonna have to marry that to the procedure, what operations that patient having to see how they’ll withstand the stress of the surgery.

Hannah Went (28:10.607)

Yeah. Thank you for walking me through that. I’m learning so much as well. I was like taking notes and kind of taking my myself through the timeline. It almost seems comparable with like biological age, right? Because like back, back in the day, they used to measure biological age by how many cigarettes you smoked. They kind of give you this like score, right? So biological age gets better and better and better and better. Um, and the visibility as you, you know, add more lens, but I, I can’t believe that a lot of those

general frailty scores, you know, number one, a lot of it’s gonna be subjective, but then you said there were like 25 different frailty scores. So even when you’re communicating with people, I imagine that would be hard. Are you using the same kind of system that I’m using, you know, if you had to communicate that for whatever reason, but I love that you’re moving into this predictive modeling, still a little bit subjective, needing some more variables as well. And I, of course, love the epigenetic data. You’re lying on top of that. So…

I guess my next question for you would be how does the study of the epigenetic biological age and this adult deformity population have the potential to shed light on new molecular basis of frailty and improve those current risk assessment tools that you went through? Is it just more data, more power?

Chris Ames (29:27.894)

Yeah, one thing it is, but it’s also sort of an objective measurement of something that I don’t feel like previously. We were really accurately quantifying. And I’ll tell you, you know, the real story was I really felt like going into it. It was going to be more of maybe like a second generation clock like a phenol age or something that had some chronological age correlation.

Hannah Went (29:37.693)

Mm-hmm.

Chris Ames (29:56.014)

but also some biomarkers, but was sort of grounded sort of in the patient’s chronological age. I really felt like that was gonna be more linked to their frailty status. And I was a bit surprised that it was the pace of aging that was the most correlated to their frailty score, their disease-specific ASDFI, and also to their general frailty score, the Edmonton.

I don’t know if you have any theories about that. You’re definitely much more of an expert, you know, on those clocks than I am, but I was surprised that the instantaneous pace of aging was more linked to what we as clinicians subjectively assign as frailty than a clock that’s grounded in their chronological age and trained to different, you know, mortality or longevity.

Hannah Went (30:33.031)

Mm-hmm.

Hannah Went (30:48.911)

Yeah.

Hannah Went (30:54.683)

That was actually my next question for you. I have it in my agenda. I’m looking, why do you believe we saw a significant association with the Dunedin pace? If I were to speculate, the Dunedin pace is definitely not, like you’re saying, trained on the chronological age, or you would expect it to maybe be more associated with pheno age, like those phenotypic outcomes. But with that Dunedin pace, just like we’re talking about making these…

surgical prediction outcomes more accurate and better, the Dunedin PACE is going to be powered by a lot more data. So it’s going to be powered by things like different brain imaging, the gate speed, the grip strength, different eye function. They even do some gum health testing and some dental testing to actually power that Dunedin PACE. So although it is not using chronological age

they are so gathering a lot of biological systems, along with 19 different blood-based biomarkers as well. So I think there’s more power behind that Duning-Dun pace than any clock that has been released to date. We do see the Duning-Dun pace getting ever so slightly increased as we age chronologically.

Although the association is not strong, the association is still there. But that would be my kind of two cents with it. Again, 50-year longitudinal trial really measuring just a lot of different biomarkers. So that could be what’s actually being more predictive of it and more predictive of those outcomes. Because we know, as you hinted again at the beginning, as the Denean and PACE increases, your balance gets worse.

You know, your gait speed gets worse. You have worse cognitive function. You even have faces that appear to look older as well, right? So that would kind of be my hypothesis behind it, if you will.

Chris Ames (32:56.022)

Yeah, that makes a lot of sense and also the idea that, you know, the divergence in pace of aging may increase as patients get older, right? That younger patients, although the instantaneous pace of aging may vary with some temporary physiologic stress, you know, it seems like the 20, 30…

35, 40 year old population is generally much more homogeneous in terms of their risk. If we’re going to get back to risk of surgery, then it’s a very heterogeneous population once you reach 60, 70, and 80 years old. And there’s probably a lot more physiologic variability going on at the tail end of life. And I think that’s where some of the original…

markers that were maybe just trained on chronological age or not capturing that quite as well as some of these more advanced markers that are including. If you look at Dunedin Pace, wow, what an unbelievable labor-intensive investment and study to do that. I would imagine that it’ll get even better over time. As the cohort gets older, you’re going to get a lot more insights.

as the New Zealand cohort ages. You’ll see a lot more interesting things there.

Hannah Went (34:20.851)

Mm-hmm.

Hannah Went (34:24.615)

Correct. Yeah, because they’re 52 years old now, Dr. Moffitt and her partner, Dr. Caspi and their team, they’re actually in Dunedin, New Zealand, as we’re speaking, capturing all of those biomarkers and values that I mentioned. But I think, you know, as long as funding allows, they’ll try and go back every so often to actually capture that. But it will be interesting to see as the group, you know, becomes older as they age, as they start to undergo their own

morbidities, mortalities themselves, what that means for the pace of aging. And I think it will make it a better model, but we’ll have to see. The study also has about a 96 retention rate. So again, very, very involved. And I hope we see another birth cohort like that at some point. I think that data is definitely necessary.

Chris Ames (35:14.958)

Let me turn it back on you and ask you a question about it. It seems to me that what you train these methylation, these principal component methylation markers to, you guys are doing much more of that and coming up with new and innovative things like the inflammatory state. What would you ever think about training the markers to surgical risk itself?

Chris Ames (35:43.67)

general instantaneous pace of aging. This kind of reminds me of my own journey from, you know, ASA grade to, you know, something very specific. Do you think we’re gonna see that same journey go on in the aging biomarker methylation field? Are we gonna have a surgical risk, you know, clock or something like that where it’s really trained specifically to the risk of a physiologic injury?

Hannah Went (36:11.619)

I do, I love that question. But I really, really do. You can make a clock for anything, I would say, as long as you have the data behind it. So as more data, like we’ve been kind of echoing this entire talk, the better. So we can have a different clock for skeletal muscle.

that’s physiologically trained, not chronologically trained. So we would need certain outputs depending on what’s available for skeletal muscle in creating that clock. So I do think it’ll get there eventually. There’s clocks for all sorts of different things. Internally, the study we’re doing right now with True Diagnostic in Harvard, which is very, very close to being done, we actually powered.

that study with about six different multiomic facets, if you will. And we would be able to use DNA methylation or epigenetics as a predictor of any of those facets. So being able to predict any metabolome, any protein level that’s in the body as well. So if you’re thinking more long term, what this means, especially for precision-based medicine is this can…

And it has been predicted by experts to even overtake conventional blood testing or hormone panel testing in the next decade or two. We know how slowly this field moves, but I think we’re already seeing the algorithms being created. It’s just a matter of validating those algorithms in cohorts in different populations and making sure that they are replicable, I think is very, very important and make sure that they’re accurate as well.

Chris Ames (38:00.254)

I totally agree with that and I also agree that it needs to be part of a more complete picture. It’s interesting you mentioned skeletal muscle. If you ask, like, where are we going now that we’ve done that demedian pace, we’ve realized that we’re still leaving a lot of data on the table that potentially can be linked to the epigenetic data because one aspect that we’re not looking at yet is the structured and unstructured radiographic data that exists in…

For example, spine CT, spine MRI, abdominal CT, or even on plane images. And to be able to link some of that, what we call radiomics data, back to the epigenetic data that we have, I think is going to be really interesting and will be a field that you’ll see grow in the future as well. And so what are we doing in that regard? So now we used to just take pre-op blood.

And that’s where our collaboration with True Diagnostic came from. Now we take multiple vials of preop blood that we use for metabolomics. We take saliva, preop, and whole blood, and then we postoperatively, or intraoperatively I should say, we then take tissue. So we have muscle, fat, disc, bone, synovium.

from all the spine patients that volunteer to be part of our tissue database. And then post-operatively, we collect their post-operative blood very early on in the surgical recovery process. And that will allow us to do potentially epigenetic studies of their post-operative, post-injured state, but also then to look longer term at their one-year post-operative, for example, their pace of aging.

Hannah Went (39:52.464)

Yeah.

Chris Ames (39:53.854)

and see whether the surgical intervention, which is massive, actually improves their physiologic state, improves their ability to be active, improves their mental state and their mood to the point that we see some impact on their instantaneous pace of aging, due date and pace at one year post-op.

Hannah Went (40:13.371)

That’s super exciting. You know, a lot of these studies again, are going to be very entry level, right? This entire field is super new, but the reason we need these studies is so we can do the testing a year later, five years, a decade later. So the follow-up studies always get me even more excited than the initial studies themselves, because again, it gives us more to the picture. And Dr. Ames, I think something to think about too,

and future research that I would love to chat about more is Raghav Sehgal, he’s from Yale. What he’s actually doing and he’s taking more of again an aging approach, but he created different ages for 11, I believe different organ systems in the body and then there’s kind of one age that’s driven by all of those 11 organ systems. So I think when we apply that

model to your work, it could be that you need to look at different markers and recreate methylation, epigenetic scores for maybe so many different subtypes and then create them as one predictive surgical outcome score, if you will, or something like that, right? Because there’s a lot of moving parts and pieces and we need to get everything in the puzzle for it to actually make sense or be a better predictor. So that’d be an interesting, I think, thought process to go through.

Chris Ames (41:35.018)

Yeah, we agree. To me, a lot of this current work really involves casting the widest possible data net as early as possible so that we preserve our ability to do these follow-up studies. We started our tissue collection just a few months ago. We’re already at 100 specimen. So we have basically like 100 patients have volunteered their tissue to be used for these studies. And we have

you know, six, seven different tissue types per patient. So we can start to do bone age, synovial age, figure out why some patients develop degenerative scoliosis as they age and others don’t. Their spines degenerate symmetrically versus others degenerate asymmetrically. And then link this to the impact of trauma. If you get back to what is the original definition of frailty, it’s the increased vulnerability to injury or insult, right? And so now,

by collecting both preoperative blood and immediate postoperative blood. We even collect fluid from the drains. So what is the tissue producing after the injury? We’re even collecting that to try to measure that exact variability. So it’s not just the general sort of physiologic variability and then postoperatively the end point is did they or did they not have a complication? No. We’ll actually be able to measure all these metabolites and all these different epigenetics.

create an epigenetic map of all the different tissue types and metabolism derangements that occur after these major insults to patients.

Hannah Went (43:08.475)

Wow, yeah. I’m excited about that one. What do you think? I know you said you’re still collecting samples, and it seems like you’ve collected a lot already. What do you think the timeline on that study is?

Chris Ames (43:18.89)

Well, it’s interesting. We are going to, so we started at UCSF and we are expanding the study now to five other U S sites and also to, um, I think it’s now six or seven sites in Europe. So, um, they will all be collecting all of these different tissue types and blood and saliva, because ideally we want to create a test that can also be based in saliva as well. Um, it’s easier for direct to consumer.

Hannah Went (43:48.051)

Mm-hmm.

Chris Ames (43:48.394)

do saliva tests. And that study will be ongoing for the next couple of years. But given the, these are the busiest sites in North America. So it’s Johns Hopkins, it’s a spine hospital in New York, Auck Hospital, Washington University. And we also wanted to generate, important for aging, we wanted to have some data spread over the younger population. So one of our sites is San Diego Children’s Hospital.

Hannah Went (43:58.321)

Yeah.

Chris Ames (44:17.102)

So we will actually get a younger population to measure all the baseline physiology and bone, epigenetic variability in bone, muscle, fat, and then we’ll have it up to patients. We have one patient in our tissue database that’s 89 years old. So we’ll really have the full spectrum painted. And if you can think of anything else that we’re not covering, let me know because we are really trying to cover everything. And I’ll just give you one example, which is so…

Hannah Went (44:38.035)

Yeah.

Hannah Went (44:42.876)

Yeah.

Chris Ames (44:44.782)

crazy to me, the huge elephant in the room. I don’t know if your listeners or you know much about the fact that fusion or trying to get two bones to grow together is a huge part of surgical care. So it’s involved in everything from neurosurgery, brain surgery where you’re trying to get the skull to grow back to spine surgery, obviously we’re fusing bones together to correct deformities to oral maxillofacial surgery, orthopedic surgery, obviously we’re trying to…

you know, heel hip fractures, et cetera, plastic surgery, and yet we have no datification of the age of bone. And so we actually, we have no ability to create precision bone cocktails to specific patients based on their specific physiology. We treat every patient, every patient’s bone nearly exactly the same way, and we don’t have a good idea.

about how that bone is aging. And there are patients that probably have defects in calcification, mineralization. Some have defects in resorption, density, obviously osteoporosis. And we probably are treating some patients that we shouldn’t be trying to treat with fusion because they have no physiologic ability to actually fuse their bones, and yet they’re still having these fusion procedures. So we really wanna make a hard move, a very deep move into the…

Hannah Went (45:53.324)

Yeah.

Chris Ames (46:10.654)

trying to characterize better the physiologic state of that patient’s bone healing. And so applying epigenetics to bone itself, I think will be really interesting and really improve the quality of care and safety of care in patients around the world.

Hannah Went (46:17.02)

Yeah.

Hannah Went (46:30.051)

Oh, that’s so interesting. As far as the biomarkers go, I think you have it all. In all of the work that I’ve done and data that I’ve seen, what I would say is missing from that is the younger population usually, right? But you already have that pediatric site, so I think that’s amazing. You’re really going to have the full age spectrum. Then we may be able to see different correlations with algorithms, chronological age, et cetera. And then, yes, I think we can do like a bone clock.

or bone, we can come up with a better name than that. But I think that opens so many different avenues and doors in terms of research. So I would be super interested in learning more about your vision there. But this has been great, Dr. Ames. I wish we could chat all day. We’re coming to the end of the show though. I’m gonna link all of the studies that we talked about in the show notes, the original ones with the telomere length, even the new preprint that just came out.

What about for anyone who wants to learn more or they want to learn more about your work or connect with you? Where can they find you?

Chris Ames (47:35.822)

Well, I’m on LinkedIn, so that has my email. And True Diagnostics certainly knows how to find me. And happy to have you direct anyone to us. Also, my UCSF email is on the UCSF website. So I’m pretty easy to find. Anyone that’s interested in collaboration, especially with our tissue data that’s growing, we’d be very happy to discuss potential collaborations.

Hannah Went (47:54.011)

Pretty easy to find, yeah.

Hannah Went (48:04.297)

Awesome. And I have one last question, Dr. Ames. This is my fun question at the end of every show. If you could be any animal in the world, what would you be and why?

Chris Ames (48:14.582)

Well, as a surgeon, Hannah, I would say a rhinoceros because you have to have a very strong skin to survive in surgery. And so we always say have the skin of a rhino and we have to tolerate a lot of going through our learning curves and a lot of harsh criticism and certainly in clinical research, a lot of the things you try don’t work and you have to keep going. So I would say my favorite animal would be a rhino.

Hannah Went (48:23.62)

Yeah.

Hannah Went (48:28.292)

Oh.

Hannah Went (48:43.747)

I like it. I like that answer. No one has had a rhino so far. So, um, no, this is great. Dr. Ames, I really, really appreciate the work you’re doing, the dedication to the research. Um, you know, it’s, it’s not easy, but I definitely see you making massive, massive improvements in this field. So, um, super appreciative of you being on here and to everyone who’s listening. Thank you so much for joining me and Dr. Ames at the everything epigenetics podcast. Remember you have control over your DNA. So tune in next time to learn.

Thanks so much, Dr. Ames.

Chris Ames (49:14.35)

Thank you, Hannah, and thanks to True Diagnostic for all the great work and services you guys are providing. It’s extremely helpful to our clinical work.

Hannah Went (49:23.043)

Yeah, absolutely. Appreciate the kind words.