What Twenty Years of “Borderline” LDL Actually Costs You
Imagine a 35-year-old whose LDL is 145. Their physician calls it “borderline high” and suggests a recheck in a year. Maybe a vague nod toward exercise and a Mediterranean diet. Nothing in the conversation feels alarming.
Now imagine that same individual, twenty years later, being carted into a cath lab. The plaque causing the flow-limiting occlusion didn’t appear overnight. It accumulated, every year, while their cholesterol was “borderline.” The tests at 35 didn’t show anything that would have made the conversation more urgent. The tests at 50 didn’t either. The problem doesn’t announce itself until it does, usually as a heart attack, or, for the lucky ones, as an incidental finding on a CT scan that catches the vulnerable plaque before it ruptures.
The numbers on this are sobering. In long-running cardiac epidemiology, sudden death is the first manifestation of coronary disease in a substantial share of patients, who had no warning symptoms before the event.3 And of the people who do have a first myocardial infarction, the case-fatality is far from trivial, both in those who reach the hospital and in the substantial fraction who don’t. Numbers like these are why the snapshot model fails so completely: by the time the snapshot finally tells you something is wrong, the answer has often already arrived.
The problem isn’t the LDL number. The problem is that the standard cardiology screening framework measures LDL as a snapshot, treats borderline as low-stakes, and acts as if the years between the snapshots aren’t accumulating damage. They are. And the cleanest evidence we have on this question, Mendelian randomization on 312,000 people plus a thirty-five-year cardiovascular cohort study, says that what’s happening between your snapshots may be the largest determinant of your cardiac future you’ll ever face.
Last week I walked through the four hidden assumptions baked into “metabolically healthy”. This week I want to focus on LDL specifically. The cumulative-exposure piece deserves its own chapter, and it’s a chapter I should have read myself, fifteen years ago. Starting around age thirty I was repeatedly told my LDL was “borderline high.” Each visit I could have made changes, or pushed for a low-dose statin, or even just asked a different question. That isn’t how the story played out. The math I’m about to walk through is the math nobody walked through with me.
What “borderline” actually means in current practice
The current screening framework treats LDL like a thermostat reading. You walk in, get a number, and the number gets bucketed:
- Under 100: optimal
- 100 to 129: near optimal
- 130 to 159: borderline
- 160 to 189: high
- 190+: very high
The action threshold for starting a statin in someone without prior cardiac disease usually requires LDL above 190, or LDL between 70 and 189 with a 10-year ASCVD risk score above some cutoff (typically 7.5% to 10%, depending on the guideline body). The ASCVD score does what it was designed to do — estimate ten-year event risk in middle-aged adults — and it does that reasonably well. The problem is what happens when it’s applied outside that window. Most thirty-five-year-olds with “borderline” LDL never hit those thresholds, because their 10-year risk score is dominated by age. The score doesn’t see fifteen more years of accumulating exposure. It sees the next ten years.
So the conversation tends to go: borderline, recheck in a year, watch your diet, see you in twelve months. The patient leaves without alarm, because in the framework’s terms, there’s no alarm to sound.
And the diet advice doesn’t close the gap
The dietary advice the borderline patient receives is, on average, undersized for the magnitude of the problem. Trial data on standard cardio-protective diets, including Mediterranean-style eating, generally show modest LDL reductions in well-controlled settings, with the magnitude shrinking further once you account for real-world adherence over years. More aggressive plant-forward approaches can push the number further, though long-term adherence in diet trials drops substantially over multi-year follow-up.
None of this means diet is irrelevant. Diet matters enormously, in both directions. Keto and high-saturated-fat patterns can raise LDL substantially in metabolically vulnerable people, which is the argument I made in Post 2 about the keto trial that got retracted, and individual response varies widely. The point here is narrower. The generic “eat Mediterranean and recheck in a year” advice given for borderline LDL doesn’t, on average, move the cumulative-exposure curve at the magnitude the math is asking for.
The cigarette analogy
Consider how we talk about a different kind of cumulative exposure. A pack of cigarettes a day for twenty years is twenty pack-years. No physician looks at twenty pack-years and says “your smoking is borderline.” The cumulative number is the diagnosis. Lung cancer risk, COPD risk, even smoking-attributable heart disease risk all scale with pack-years, not with whether the patient happened to be smoking on the morning of their appointment. Pack-years sit inside the medical vocabulary because the arithmetic is what predicts outcome.
LDL works exactly the same way. The screening framework just hasn’t absorbed the arithmetic yet.
What cumulative LDL exposure actually measures
Here’s what the framework doesn’t measure. Elevated LDL damages the artery each year it’s elevated. Not in proportion to today’s level, but in proportion to the integral of today’s level over time. Twenty years of LDL 145 isn’t twenty data points of “borderline.” It’s a continuous accumulation of ApoB-containing particles (ApoB is the protein marker that rides on the surface of these particles) crossing the inner artery wall, lodging in the intima, and initiating the slow inflammatory cascade that builds plaque. The unit that captures this is mg-years, your LDL number multiplied by the years you’ve carried it. Some lipidologists use “LDL-year” or “mmol-year.” Same idea, different units.
A 35-year-old at LDL 145 since age 20 has accumulated roughly 2,175 mg-years of exposure already. Their lipid panel today says “borderline,” but their artery has been logging the damage since college.
What Mendelian randomization actually proves
In 2012, Brian Ference and his colleagues published a Mendelian randomization study in the Journal of the American College of Cardiology that has, for my money, the cleanest evidence we have on the question of when matters as much as how much1. With my background in bioinformatics and epidemiology, I assign serious weight to study designs like this one. A well-conducted Mendelian randomization isn’t typical observational epidemiology; it’s a way to draw something close to a causal arrow from genetics to outcome.
Here’s the intuition. A regular observational study compares people who have low LDL because of something (diet, statin use, exercise, body weight, life circumstances) to people with higher LDL. Untangling the LDL effect from everything else is hard. A Mendelian randomization study compares people who inherited genetic variants that lower their LDL from birth, by chance, at conception, before they made a single lifestyle decision, to everyone else. The genetic version randomizes the exposure the way a clinical trial randomizes a drug, except the random assignment happens at conception and lasts a lifetime. Compare outcomes between carriers and non-carriers, and you get something close to a randomized experiment of “what if you had lower LDL from birth?” without ever running it.
Ference’s team combined nine such polymorphisms (small, naturally occurring genetic variants) across six different genes. The pooled meta-analysis included 312,321 participants.
The finding: each 1 mmol/L (about 38.7 mg/dL) lower LDL through these genetic variants was associated with a 54.5% reduction in coronary heart disease (CHD) risk (95% CI 48.8% to 59.5%, p = 8.43 × 10⁻¹⁹). All nine variants produced consistent reductions, with no evidence of heterogeneity.
Now compare that to a statin started later in life. Per Ference’s comparison against the long-running meta-analyses of statin trials, midlife treatment that achieves the same magnitude of LDL drop reduces CHD risk by roughly a third the size of the genetic effect. Both numbers are real. The drug works. The genetics work. They achieve the same delta in the LDL number.
But the genetic version is associated with roughly a 3-fold greater reduction in coronary risk per unit LDL than midlife statin therapy delivers (per Ference’s comparison to statin-trial meta-analyses).
The difference isn’t the molecule. The difference is the duration. The genetic carrier had lower LDL from conception. The midlife statin patient had lower LDL from age 50 onward. Same delta in the number, different starting point on the timeline, and the timeline is where most of the damage gets done.
(Honest caveat: a small portion of the genetic effect may also reflect developmental influences, not just longer exposure. Lifetime-low LDL from conception affects vascular development as well as decades-of-additional-exposure. The full causal story is probably “more years of lower LDL plus better-developed vasculature,” not pure exposure-time. The size of the gap is too big to be explained by developmental effects alone.)
The math the framework doesn’t show your thirty-five-year-old patient: a tripled benefit per unit LDL when the lowering starts early.
The CARDIA cohort: real-time confirmation in 4,958 humans
If Mendelian randomization gives us the gene’s-eye view, the CARDIA study gives us the longitudinal-cohort view. CARDIA, the Coronary Artery Risk Development in Young Adults study, enrolled 4,958 asymptomatic adults aged 18 to 30 in 1985 and 1986, and followed them for decades.
In 2020, Michael Domanski and colleagues published a CARDIA-based analysis that quantified, in real human longitudinal data, what cumulative LDL exposure does to cardiovascular event rates2.
They computed each participant’s area under the LDL-by-age curve, which is essentially mg-years, and tracked cardiovascular events after age 40. Median follow-up after age 40 was 16 years. There were 275 events: nonfatal coronary heart disease, stroke, transient ischemic attack, heart-failure hospitalization, cardiac revascularization, peripheral arterial disease intervention, or cardiovascular death.
After adjusting for sex, race, and traditional risk factors, the analysis produced two findings worth reading carefully, because together they change how you should think about a “borderline” reading at thirty-five.
Finding one. Cumulative LDL exposure was significantly associated with cardiovascular event risk: hazard ratio 1.053 per 100 mg/dL × years (p < 0.0001). Translated: every additional 100 mg-years of LDL accumulation raised the hazard of a cardiovascular event by 5.3%, in a linear, dose-response fashion. The relationship had no inflection point. More exposure, more risk, all the way down.
Finding two: the one most people miss. The same area-under-the-curve carried more risk when it accumulated earlier in life than when it accumulated later (HR 0.797 per mg/dL/year of slope, p = 0.045). Translated, what this means in plain language: an early-life accumulation of 2,000 mg-years done by the time you’re thirty-five is more dangerous than the same 2,000 mg-years accumulated between forty-five and sixty.
The total exposure was identical. The risk was different, and the years that mattered most were the early ones.
This is what Mendelian randomization had implied biologically; CARDIA confirmed it observationally. Two different methods, two different decades, both pointing at the same finding: time is not just a multiplier on LDL exposure. Time at the wrong age is a force multiplier.
Honest framing: this is one observational cohort. The methods are good but the design is observational. The findings need replication. The coherence with Ference’s genetic data, with Navar-Boggan 2015’s young-adult hyperlipidemia paper (cited in Post 6), and with what we know mechanistically about plaque biology means multiple lines of evidence are pointing at the same place, and the broader literature on lifetime cholesterol exposure has many more studies aimed at the same target.
What this means for a 35-year-old with LDL 145
Let me run the math three ways.
Case A. A 35-year-old whose LDL has hovered at 145 since age 20. Their cumulative exposure by age 35 is 15 × 145 = 2,175 mg-years. By age 45, assuming the number doesn’t change, they reach 25 × 145 = 3,625 mg-years.
Case B. Same patient, same starting LDL of 145 at age 20. But at age 30, they (or their physician) take it seriously. Lifestyle changes plus, if needed, intervention bring LDL to 100. By age 45, their cumulative exposure is 10 × 145 (ages 20 to 30) + 15 × 100 (ages 30 to 45) = 1,450 + 1,500 = 2,950 mg-years.
The difference between Case A and Case B by age 45 is 675 mg-years — about 19% lower lifetime LDL load.
Apply Domanski’s hazard ratio of 1.053 per 100 mg-years. The 675 mg-years of avoided exposure works out to roughly a 30% lower hazard of cardiovascular events for Case B compared to Case A, a meaningful drop, before you even count the additional benefit Domanski found from intervening earlier on the slope of accumulation.
Case C. Same 35-year-old. Acts at age 35 instead of 30. Cumulative by age 45 is 15 × 145 + 10 × 100 = 2,175 + 1,000 = 3,175 mg-years.
Five years later than Case B. The five-year delay costs about 225 mg-years — roughly a third of the savings Case B was on track to capture, given up to inaction. Five years of “we’ll watch it” is not a neutral choice. It’s a third of the available risk reduction, walked away from.
Caveat clearly: these numbers are illustrative, not prescriptive. Real lifetime LDL doesn’t stay flat at any number. It drifts up with age, varies with diet, gets perturbed by every event in life. Real-world risk depends on dozens of variables this calculation doesn’t capture. But the direction is right, the magnitude is plausible, and the lesson is this: the timeline of your LDL is not a backdrop to your cardiac risk. It’s the primary driver.
The math your physician probably hasn’t shown you is this: at thirty-five, a “borderline” reading isn’t a holding pattern. It’s an unfolding cost.
The 35-year window
Most cardiology screening tools, the ASCVD risk score, the Framingham risk equation, the QRISK calculator, were designed for people in their fifties. They estimate ten-year event risk based on traditional factors. They are reasonably good at telling a fifty-five-year-old whether to start a statin.
They are the wrong tool for a thirty-five-year-old.
A thirty-five-year-old’s ten-year risk is dominated by age. The score will almost always come back low, because the next ten years (35 to 45) genuinely don’t carry much absolute event risk. But the next thirty years, the years where Mendelian randomization and CARDIA both say the most damage compounds, are not on the score’s clock.
The right question for a thirty-five-year-old isn’t “what’s my ten-year risk?” It’s “what’s my thirty-year exposure trajectory, and what’s the leverage of intervening now versus in fifteen years?”
The Ference and Domanski math both say the answer to that second question is: a lot.
This is the leverage window. It’s also the window where most people with borderline numbers do nothing, because the screening tools tell them they don’t need to.
What I’m telling people in their thirties and forties
Three things, in plain practical terms:
Know your ApoB and Lp(a) in addition to your LDL-C. Once. By age 30 if you can. ApoB tells you the actual particle count, which tracks with risk better than the cholesterol mass; Lp(a) is a separate genetic risk factor that most cardiologists won’t order without prompting. I covered the ApoB-vs-LDL-C distinction in Post 6. Both tests are widely available; many physicians don’t order them by default. Ask.
Reframe the conversation from “is my LDL borderline?” to “what’s my mg-year trajectory?” The first question gets you a snapshot answer. The second one forces a conversation about time. If you walk into a primary care visit and ask “what’s my cumulative LDL exposure look like over the next twenty years if we don’t change anything?”, the question itself forces even a physician who’s never thought in mg-years to engage with you on those terms. You’re moving the conversation from “is the number high enough” to “is the trajectory acceptable.”
If your borderline number has been borderline for ten or more years already, that’s not “monitor and recheck.” That’s data. Bring the time-series to the appointment. Twelve readings of LDL 145 over fifteen years is not a borderline finding. It’s a confirmed cumulative-exposure pattern. Your physician’s instinct will be to look at the most recent point. As I wrote earlier in this series about the fight to get the right cardiac scan, the system rewards patients who come in with the full picture and ask for what they actually need. Make sure your physician sees the full series, not just the most recent point.
When I look at my own labs from age 30 to 43, the mg-year arithmetic explains the timeline almost perfectly. The conversations I had at thirty would have been the cheapest cardiovascular intervention I could ever have made. None of them happened, because nothing in the framework gave anyone, including me, a reason to think they were urgent.
What’s next
Last week I named four hidden assumptions baked into “metabolically healthy”: tissue-level inflammation, borderline blood pressure, undiagnosed autoimmune contributions, and genetic endothelial vulnerability. The cumulative-exposure piece I just walked through is the LDL-specific deepening of that argument. Next week I’m going to take borderline blood pressure and put it through the same kind of evidence-deep treatment, including the SPRINT trial and what it actually said about a number most people in primary care still call “low one-thirties over high seventies, not horrible.”
The story keeps moving. The math is harder than the framework wants to admit. And the clock is the variable nobody put on the chart.
More to come next week.
For more evidence-based analysis of the claims shaping your health decisions, visit calibratedsignal.com. Hard science, delivered honestly. No sponsors. No supplement deals. Just signal.
This post is educational, not medical advice. See Medical Disclaimer and Disclosures.
References
-
Ference BA, Yoo W, Alesh I, Mahajan N, Mirowska KK, Mewada A, Kahn J, Afonso L, Williams KA Sr, Flack JM. Effect of long-term exposure to lower low-density lipoprotein cholesterol beginning early in life on the risk of coronary heart disease: a Mendelian randomization analysis. J Am Coll Cardiol. 2012;60(25):2631-2639. PMID: 23083789
-
Domanski MJ, Tian X, Wu CO, et al. Time Course of LDL Cholesterol Exposure and Cardiovascular Disease Event Risk. J Am Coll Cardiol. 2020;76(13):1507-1516. PMID: 32972526
-
Myerburg RJ, Junttila MJ. Sudden cardiac death caused by coronary heart disease. Circulation. 2012;125(8):1043-1052. PMID: 22371442
Nick Hanson, MS, RN, CEN
Mayo Clinic Board Certified Emergency Department RN
MS Bioinformatics & Computational Biology | University of Minnesota / Mayo Clinic
APRN-FNP Candidate | Duke University
PhD Candidate Bioinformatics & Computational Biology | University of Minnesota / Mayo Clinic
Published Epigenetics and Oncology Researcher
Former Health & Wellness Industry CEO (15+ years)
Certified Fitness Trainer (ISSA)