At Ending Age-Related Diseases 2021, Michael Lustgarten discussed what he does to keep his biological age substantially below the average according to multiple biomarkers.
Aging and disease are biochemical processes that happen over many decades. With that in mind, if we track well-established biomarkers of organ and systemic health, can aging and disease risk be slowed? With that in mind, I measured my biological age. I got blood tested for the third time in 2021.
I’ve got those nine biomarkers listed right there. This is using using Morgan Levine’s phenotypic age calculator, which is a measure of biological age. Just to run through really quickly for people who haven’t seen this before, it includes albumin, creatinine, glucose, C reactive protein, the percentage of lymphocytes, mean red blood cell volume, red blood cell distribution width, alkaline phosphatase, white blood cells, and chronological age.
When entering in all of that data, my biological age comes up as 35.2. When considering my chronological age is 48 and a half, that puts me at 13.2 years younger than my chronological age.
An obvious question that should arise is how good is this biological age calculator. Here, we’re looking at the correlation for phenotypic age, again, as a metric of biological age versus chronological age. By the red line in the figure, you can see that it’s almost perfectly linear, which means as close as you can get, or getting as close to 1, a correlation of 1 or -1, but that’s not applicable in this situation.
Anyway, it’s pretty close to one, as you can see on the left, in two separate study cohorts in NHANES III, the correlation for this biological age calculator with chronological age was 0.94. In a separate study, NHANES IV, it was 0.96, so very strong correlations for this biological age calculator with chronological age.
Another question that often arises is how does this compare with the best epigenetic clocks, and the epigenetic clock that consistently is shown to have the strongest correlation with biological age is the Horvath clock or a couple of different Horvath clocks, which is what’s shown here. We’re looking at epigenetic age on the Y axis plotted against chronological age on the X axis. This is a composite of epigenetic age against chronological age for 16 different cell types. We’re looking at that on the right, so breast cells, cheek cells, cerebellum, so brain, colon, blood, skin, etc.
When looking at the composite measure for epigenetic age of those cell types against chronological age, we can see also a very strong correlation of 0.94. In other words, when considering Levine’s correlations with chronological age, and the best performing epigenetic clock from the Horvath group, we can see that they’re pretty close to similar for their ability to predict chronological age.
Aging data and being able to predict chronological age is interesting, but how does it associate with disease risk or risk of death for all causes: all-cause mortality risk? So we’ve shown here all-cause mortality data, and for every one-year increase in biological age, using Levine’s phenotypic age calculator, we can see a 9% increase in all-cause mortality risk. That association is true whether looking at young, middle-aged, or older adults, as shown there. Again, older biological ages based on Levine’s test significantly increased risk for all-cause mortality.
More specifically, it’s also associated with disease-specific mortality, including an increased risk for heart disease, cancer, chronic lower respiratory disease, interestingly, not cerebral vascular disease or stroke, an increase in mortality related to diabetes, influenza or pneumonia-related death.
As a side note, in a paper that Morgan Levine’s group published last year, an older biological age up to 13 years before the onset of the pandemic was associated with an increased mortality risk in COVID-19 patients. Last but not least, having an older biological age by Levine’s test is associated with an increased risk of death for kidney-related issues, including nephritis and nephrosis.
It’s important to note that Levine’s test isn’t the only blood-based biomarker panel that can easily assess biological age and all-cause mortality risk. I’m sure this crowd already knows about it, but just to run through it, the other one that I’m going to show some data for is aging.ai, in particular, aging.ai 3.0.
In contrast with the nine parameters on Levine’s test, this includes 19 different blood-based variables, and you can obtain them by getting the standard chemistry panel and complete blood count, CBC. In the US, that’s relatively cheap, it’s about a $35 test. Also note that the standard chemistry panel and CBC can also be used to to measure eight of the nine components on Levine’s test with the exception of high-sensitivity C reactive protein, which is usually a separately ordered test, which can cost up to about $45.
For $80 USD or less, depending on where you get it, you can do Levine’s test, and for $35 USD, you can do the blood test to calculate your own aging.ai age. Aging.ai isn’t as strong for its correlation with chronological age as Levine’s test, with a correlation coefficient of 0.8.
What about all-cause mortality risk? In a study that was published a few years ago by the aging.ai group, they found that biological ages that were younger than their corresponding chronological age were associated with reduced all-cause mortality risk.
That’s what’s shown here. First, they looked at a Canadian cohort. For people who had a five year older biological age relative to their chronological age, you can see they had about a two-fold increase in all-cause mortality risk.
On the other side of that, for people who had a greater than five-year reduction in their biological age relative to their chronological age, there was approaching a 50% decrease for all-cause mortality risk. These findings were also replicated in a US-based cohort. Again, NHANES, where we can see the same pattern.
With this in mind, what’s my data? I’ve put all my data up here, so anybody can double check my numbers, and based on my numbers for my test that I got done at the end of July, we can see that my predicted biological age with aging.ai is 29. Note that I’m close to 49, so this is about a 20-year difference for my biological age relative to the chronological age.
Results for one blood test, they’re interesting, but data for many more tests can tell the full story. With that in mind, I’ve been blood testing up to six times a year for a few years now. I have data for 12 blood tests since 2018, where I can calculate Levine’s biological age. That’s what’s shown here.
When looking at the different dates, we can see my chronological age, CA, my biological age or PhenoAge, and then the difference for my chronological age minus the biological age. As we can see, when taking that average, I’m about consistently 12 years younger over the last three-year period.
What about aging.ai? For aging.ai, I have many more blood tests, mostly because I just started measuring C reactive protein just a few years ago; I wasn’t measuring C reactive protein early on, more than 10 years ago. Looking at 27 blood tests in 2009, as shown here on the left, there were too many blood tests to put up all that data. I just took the average for every year going past 2016 to 2021.
Then from 2009 to 2013, I had three tests; I was testing about once a year. Nonetheless, we can see on the right my aging.ai average age, and then the average difference for my chronological minus my biological age. Notice that from 2009 to 2013, I had my lowest difference for CA minus BA around seven years.
Then in 2015, I started getting more serious about optimizing my biomarkers and optimizing my health in general. I started dietary tracking, weighing all my food, entering all that data into an online tool that then gave me macro, micronutrient data that are logged in an Excel file. Then I started to look at correlations between my diet and my blood biomarkers with the goal of optimizing the blood biomarkers.
My diet is completely driven by what I see in my blood. Since 2013, when after which I started dietary tracking, we can see my average difference between my chronological and biological age. I was consistently 15 years younger, and in three blood tests so far in 2021, I’m almost 19 years younger, so I’ve been able to get close to what biochemical youthfulness looks like.
The obvious question should be what’s contributing to this relatively low biological age. In terms of supplements, I was diagnosed with hypothyroidism in my 20s, so I have to take that. So I’ve been taking L-thyroxine for more than 20 years. Then, for the last blood test that I took in April, until my previous blood test in July, I was supplementing with methyl-B12 to help reduce homocysteine. If anybody’s interested in the why behind that, I’ve got a video on my YouTube channel, check it out.
What about fitness? I should mention that I take the average value in between blood tests, because that would be expected to correlate to the subsequent blood test. In between blood test number two and number three of this year, that 84-day period, I weigh myself every day, too.
In the morning, my average body weight was 156.7 pounds, which results in a BMI in the quote-unquote normal range, 24 and a half, so nothing too spectacular there. My resting heart rate, RHR, was around 47 beats per minute and heart rate variability, HRV, was about 53 milliseconds. These are cardiovascular-based, health-related metrics. The next question should be is, are these cardiovascular fitness metrics contributing to my relatively young biological age?
I should mention this, I wear a fitness tracker, and I’m not sponsored to promote them, but I wear Whoop. The reason that’s important is because this is Whoop data. Instead of comparing my data against the general population, where I’d likely see much larger changes, I’m comparing myself against people who are already fitness minded, so it’s a more relevant comparison.
Here, we’re looking at average resting heart rate and beats per minute on the Y axis plotted against age, so 20 up to about 50 years old. What we can see is that resting heart rate increases for both women and men, women in red, men in blue. So resting heart rate increases from the 20 to 50 years old age range. What about my resting heart rate, which was 47? I’d be off the chart if lower is indicative of youth, this would be indicative of a relatively youthful resting heart rate.
What about a low resting heart rate being associated with health or risk of disease, all-cause mortality risk? That’s what’s shown here, this is a meta-analysis of 46 studies that included more than 1.2 million subjects, and we’re looking at the relative risk for all-cause mortality on the Y axis plotted against resting heart rate on the X.
What we can see is that starting at 45, which was the lowest all-cause mortality risk, as you go above 45 beats per minute, there’s an increase in all-cause mortality risk. I’m at 47 right now. Based on the aging data, and based on the all-cause mortality data, that would suggest that this may be contributing to my relatively youthful biological age.
There’s a button there, so why is there a button there? This is Fitbit data, and in agreement with the Whoop data, we can see that average resting heart rate does indeed increase from 20 up until about 50 years of age, then notice that resting heart rate then also decreases in data for the women in green, men in blue. Resting heart rate decreases during aging. Is my relatively low resting heart rate indicative of youth and health, or is it low because of aging?
That’s where another metric for cardiovascular health can come in, which is known as heart rate variability. Heart rate variability is defined as the variation in time in between successive heartbeats.
What that verbiage means is, when your heart rate is 60 beats per minute, the assumption is it beats exactly one beat per second. But that’s not exactly true. It may be .9 seconds per beat, one second per beat, 1.1. The average would be one second per beat, but there is a variability in terms of the time that it takes for each beat, and that variability is the heart rate variability. In this case, for these three examples here, a heart rate variability, HRV, would be 67 milliseconds.
In order to get more context into the resting heart rate story in terms of being good or bad in my situation, and potentially contributing to a relatively young biological age, it’s important to see how heart rate variability changes during aging. Again, this is Whoop data, so comparing myself against other fitness-minded people.
We can clearly see looking at heart rate variability on the Y axis plotted against age and in this case, it’s up to 65 years as the maximum for the age range 20 to 65. We can clearly see by the black line that heart rate variability decreases during aging.
My average heart rate variability from my second blood test until my third for that period was about 53 milliseconds. That would put me close to the top of the range for my chronological age category. At best, it would put me at the low end of the range for a 20-something year old. As a relatively quick review for my resting heart rate and heart rate variability, my resting heart rate is relatively low in conjunction with the heart rate variability that is higher than the median for my age at worst or at the low end of the range for a 23-year-old at best.
When considering that a low resting heart rate and a high heart rate variability are found in youth, and, in contrast, a low resting heart rate but also a low heart rate variability are found in older adults, to answer the question, is cardiovascular fitness contributing to a relatively young biological age, I posit that my data suggests that it’s indicative of youth. If my heart rate variability was, say, 30 milliseconds as indicated there, which would be below the median for my age range or chronological age, then one can make the argument that a relatively low resting heart rate plus a low heart rate variability would be indicative of poor cardiovascular fitness.
That doesn’t seem to be the case, at least for me right now. I should mention, since my last blood tests, my heart rate variability for the past month or so is around 58, so I’m even better than I was going into my last blood test.
What I haven’t mentioned yet is how diet’s contributing to my relatively low biological age. NAD is involved in the health and function of virtually every organ system in the body, and unfortunately for us, it declines during aging. One reason for that is because of the age-related increase in a protein that degrades NAD, which is known as CD38. There are CD38 inhibitors, fortunately, for example, apigenin, quercetin, luteolin and kuromanin as shown there, which can inhibit CD38, increase NAD, or slow the age-related NAD decline.
Note that these metabolites are found in food. By eating an abundance of foods that contain CD38 inhibitors, potentially we can maximize NAD, thereby keeping us youthful for longer. With that in mind, I’m going to take a look at some of my own data in terms of dietary CD38 inhibitors and whether or not they’re correlated with my biological age.
As I mentioned, I’ve been weighing all of my food since 2015. It wasn’t until more recently that I started quantifying food intake. In 2015, I was tracking macro, micronutrients, but then I had the thought, well, I’m reducing food to its components but the fact is, I should be tracking actual food amounts that I’m taking in. I’ve been doing that since July 2018.
Since then, I have more than 1100 days of food intake that correspond to 16 blood tests during that time. That’s what’s shown here on the left. Just to run through this approach real quick, and if anybody wants more info, I have a video on this too, on my YouTube channel, lots of these videos on my YouTube channel, so check it out.
Notice the date. On July 10, starting at the top, 2018, I got a blood test on that day. Since I was fasting, that day starts day one, for the period that corresponds to my next blood test, which was then on 8/21/2018. From July 10 to August 20, I have an average dietary intake, since I’m recording all of my food during that period that corresponds to blood test data.
Because I have enough dietary periods that correspond with blood test, I can then look at correlations between my diet with blood biomarkers. Just going across the list for each of the four dietary CD38 inhibitors, I calculate how much I was taking in on average every day for each of the four inhibitors and then sum them.
Note that it’s not all the same for my dietary CD38 intake. In terms of my average daily intake, I have a range of 160 to about 390 milligrams of these dietary CD38 inhibitors per day, which should allow for some decent correlations if they exist. Whereas apigenin and quercetin are kind of the more pop culture way to inhibit CD38 and potentially raise NAD, in contrast, about 70% of my dietary CD38 inhibitor intake comes from kuromanin, and the major reason for that is because I eat a lot of blackberries and kuromanin is enriched in blackberries.
With all this in mind, is the sum of these dietary CD38 inhibitors significantly correlated with biological age and/or its component biomarkers? First looking at Levine’s test, and I’ve got how many blood tests that correspond to dietary intake with the little n next to each variable. I have between 11 and 16 blood tests that correspond to the dietary CD38 data.
First, note that for the sum of the dietary CD38 inhibitor intake, that’s not significantly correlated with the overall biological age score, the PhenoAge. You can see the P value’s not even close to statistically significant, but interestingly, we can see that the higher my dietary CD38 inhibitor intake is, the lower my glucose levels are. That’s a significant correlation, but for whatever reason, that correlation isn’t enough to drive an overall reduction for biological age, at least using Levine’s test.
All right, so what about aging.ai? At least it’s going in the right direction, we can see an inverse correlation, meaning the higher my dietary CD38 inhibitor intake, that’s correlated with a younger aging.io age, but note that the P value is not statistically significant. There’s no significant correlation for my dietary CD38 inhibitor intake with aging.ai age.
Note that glucose, as glucose is also included in aging.ai, we see the same correlation. Then also note that the higher that my dietary CD38 inhibitor intake is, that’s correlated with lower levels of platelets and blood cells, which is probably a bad thing because levels of platelets and red blood cells both decline during aging, so that would be sending them in the wrong direction.
One biomarker going in the right direction, two going in the wrong direction. Again, the net effect of those three biomarkers is not enough to affect the overall aging.ai age in terms of its correlation with dietary CD38 inhibitor intake. When considering this data for the dietary CD38 inhibitors, I posit that’s not impacting my overall reduction for biological age. That’s all I’ve got for now, I’ve just got posted here everywhere that I’m online. With that, I’m ready for questions.