30 yr study of women's cardiovasc risk by baseline hs-CRP,LDL,Lp(a)

 A recent study found that a single initial measurement of LDL cholesterol, high-sensitivity C-reactive protein, and Lipoprotein (a) was predictive of future cardiovascular events 30 years later (see cad women inflam chol lip(a) 30 yr NEJM2024 in dropbox, or DOI: 10.1056/NEJMoa2405182.

 

Details:

--39,876 US female health professionals in the Women’s Health Study (WHS), a prospective cohort of healthy US women at least 45 years old who were randomized to low-dose aspirin or vitamin E, enrolled from 1992-1995

    -- these women were followed for 10 years with this intervention, but then continued to have followup prospectively through 2023, allowing for assessment of 30-year outcomes

    -- 27,939 women in this cohort volunteered to have baseline blood drawn and stored in liquid nitrogen, later being assessed for high-sensitivity CRP levels (hs-CRP), LDL cholesterol, and Lipoprotein (a) (LP(a))

-- baseline demographics/etc:

    -- mean age 55

    -- hypertension in 25%, diabetes 2.5%, parental history of MI before age 65yo 14%,  white 94%, BMI 26

    -- in their statistical adjustments, they comment that models were "adjusted for age, initial randomization treatment group, smoking (current, past, never), presence of diabetes, and Framingham blood pressure categories", though there was no elaboration of what the prevalence was for the several of these items at baseline or for those specifically in this study (70% of the original cohort)

 

-- primary endpoint: a composite of incident myocardial infarction, incident stroke, coronary revascularization, and death from cardiovascular disease during 30 years of follow-up.  All outcomes were confirmed.

-- median follow-up 27.4 years

Results:

-- the correlation between the results from each of these individual biomarkers was minimal (Spearman correlation coefficient between 0.01 to 0.17 for all interactions)

-- quintiles of the biomarkers:

    -- hs-CRP:

        -- quintile 1: <0.65 mg/L

        -- quintile 2: 0.65 to <1.47 mg/L

        -- quintile 3: 1.47 to <2.75 mg/L

        -- quintile 4: 2.75 to <5.18 mg/L

        -- quintile 5: at least 5.18 mg/L

    -- LDL:

        -- quintile 1: <96.1 mg/dL

        -- quintile 2: 96.1 to <113.5 mg/dL

        -- quintile 3: 113.5 to <129.7 mg/dL

        -- quintile 4: 129.7 to <150.7mg/dL

        -- quintile 5: at least 150.7 mg/dL

    -- Lp(a):

        -- quintile 1: <3.6 mg/dL

        -- quintile 2: 3.6 to <7.6 mg/dL

        -- quintile 3: 7.6 to <15.5 mg/dL

        -- quintile 4: 15.5 to <44.1mg/dL

        -- quintile 5: at least 44.1 mg/dL

 

-- first major confirmed cardiovascular event occurred in 3662 women

-- first composite event by a single biomarker in top vs bottom quintile, by covariate-adjusted hazard ratios:

    -- hs-CRP: 70% increase, HR 1.70 (1.52-1.90)

    -- LDL: 36% increase, HR 1.36 (1.23-1.52)

    -- Lp(a): 33% increase, HR 1.33 (1.21-1.47)

        -- similar findings for coronary heart disease and stroke individually

-- overall, the risks of hs-CRP and LDL did attenuate marginally over time, but not so for Lp(a)

-- risk for Lp(a) was mostly found in those in quintile 5 (ie at least 44.1 mg/dL)

-- the graphs were very similar for the 3 different combinations of 2 biomarkers, but there were increased numbers of events with 2 very high biomarkers

-- the combined effects of all 3 biomarkers for women who had 0, 1, 2, or 3 biomarkers in quintile 5 (the highest levels) vs those with no biomarkers at that level as the reference:

    -- 1 biomarker in quintile 5: 27% increased risk, HR 1.27 (1.19-1.37)

    -- 2 biomarkers in quintile 5: 66% increased risk, HR 1.66 (1.51-1.83)

    -- 3 biomarkers in quintile 5: 2.63-fold incidence, HR 2.63 (2.16-3.19)

-- similar results for coronary heart disease  and stroke combined:

    -- 1 biomarker in quintile 5: 68% increased risk, HR 1.68 (1.14-2.48)

    -- 2 biomarkers in quintile 5: 3.71-fold incidence, HR 3.71 (2.94-4.68)

 

--age-adjusted and competing risk-adjusted cumulative incidence curves for the first major cardiovascular event, coronary heart disease event, and stroke during 30 years of follow-up:

 

-- Sensitivity analysis:

    -- statin use: over the 30 years of the study, 16,053 women (57%) received at least one prescription for statins, but there were no high-level data in WHS on therapy adherence, dose, achieved LDL, or duration. An analysis done up to the time of the first reported use of statins (ie, before statins was an issue) found 2151 of the 3662 had first major cardiovascular events; there were similar results when comparing quintile 5 to quintile 1:

        -- hs-CRP: 65% increase, HR 1.65 (1.43-1.90)

        -- LDL: 62% increase, HR 1.62 (1.41-1.86)

        -- Lp(a): 42% increase, HR 1.42 (1.25-1.62)

    -- and for the results of combining the 3 biomarkers in quintile 5:

        -- 3.21-fold, HR 3.21 (2.41-4.27) for the primary endpoint

        -- 2.87-fold, HR 2.87 (1.71-4.84) for stroke

        -- 4.08-fold, HR 4.08 (2.88-5.77) for coronary heart disease

 

Commentary:

-- the 3 atherosclerosis risk factors they assessed seemed to be largely independent of each other but additive to the risk of actual cardiovascular outcomes:

    -- hs-CRP:

        -- the JUPITER study and others have found that even in people with lowish LDL levels (median of 108 mg/dL in JUPITER) but mildly elevated hsCRP levels (median of 4.3 mg/L), then given rosuvastatin 20mg or placebo for 1.9 years (though the study was supposed to last 5 years), led to LDL of 55 vs 109 mg/dL in controls and hsCRP of 1.8 vs 3.3 mg/L in controls and a huge benefit in the outcome combo of nonfatal MI, nonfatal stroke, unstable angina, arterial revascularization and all-cause mortality. limitations of this study, however,  include that these outcomes may have been from decreasing LDL as much or more than decreasing hs-CRP levels, and a systematic review of achieved LDL levels found that an LDL of 55 mg/dL as in the rosuvastatin group would be much more protective than an LDL of 109 mg/dL: https://gmodestmedblogs.blogspot.com/2018/08/very-low-ldl-levels-benefit-without-harm.html 

            -- the MESA study of 950 patients who met the JUPITER inclusion criteria (LDL<130, hsCRP>2, no known cardiovascular disease) did find that by assessing coronary artery calcium (CAC) scores, there was no added value of CRP determinations

        -- anti-inflammatory meds also can decrease hs-CRP levels: colchicine (eg see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7458326/ ). the FDA in 2023 did formally approve colchicine for decreasing cardiovascular events based on a couple of studies (https://gmodestmedblogs.blogspot.com/2024/03/colchicine-decreases-cardiovasc-events.html ). And there was a study using the monoclonal antibody canakinumab (which targets IL-1b) finding decreased cardiovascular events without changing LDL levels (https://www.nejm.org/doi/full/10.1056/NEJMoa1707914 )

        -- other meds decrease hs-CRP: ARBs (valsartan, irbesartan, olmesartan and telmisartan, but  not losartan, candesartan which had mixed effects), pioglitazone, aspirin, coxibs, platelet inhibitors (clopidogrel), statins/ezetimibe/fenofibrate, ACE inhibitors (ramipril, captopril, fosinopril but not enalapril and trandolapril): https://pubmed.ncbi.nlm.nih.gov/16939632/

        -- CRP is also lowered with a healthy diet and exercise

        -- and, by the way, higher CRP levels are associated with young-onset dementia...

(https://gmodestmedblogs.blogspot.com/2024/05/young-onset-dementia-risk-factors.html )

        – unfortunately, we do not have head-to-head comparisons of the different meds and non-pharmacologic interventions to assess their relative value in cardiovascular risk reduction

 

    -- LDL cholesterol:

        -- there are many concerns with LDL being the appropriate surrogate marker for lipid-associated increased cardiovascular risk (see https://gmodestmedblogs.blogspot.com/2024/06/using-surrogate-markers-for-disease-are.html , a blog that comments on the different LDL particles with different levels of atherogenicity). It has been clear for decades that apolipoprotein B, cholesterol/HDL ratio, and non-HDL are better markers for cardiovascular risk

 

    -- Lp(a):

    -- Lp(a) is a genetically determined lipid fraction and is stable for one’s life after about age 5

    -- however, the PCSK9 inhibitors do lower Lp(a) levels by 20-30%, as well having profound effects on LDL levels

        – For example, the well-done Fourier study of 25,096 patients  found that Lp(a) conferred a 22% increased risk of adverse cardiovascular events independent of LDL levels, that the PCSK9 inhibitor evolocumab was associated with a 27% decrease in Lp(a) levels, and that there was a 23% decrease in adverse cardiovascular events by the med in those with baseline Lp(a) levels  > median but only 7% in those with levels <median (https://www.ahajournals.org/doi/10.1161/CIRCULATIONAHA.118.037184 ). By the way, the updated Fourier study also found cardiovascular benefit in lowering the LDL to <20 mg/dL (see cad high risk LDL less than 20 fourier studyCirc2023 in dropbox, or DOI: 10.1161/CIRCULATIONAHA.122.063399)

        -- niacin, which is largely not used anymore but was the mainstay of lipid management in the past, also does lower Lp(a) 35-40%

        -- and newer Lp(a)-lowering agents are being developed

 

-- this study did confirm that each of these 3 risk factors measured once in 55yo healthy women were associated with adverse cardiovascular outcomes up to 30 years later, and the combination of high values in 2 or, even more so with 3 of them, was associated with much more cardiovascular risk. But there are a few issues:

    -- it is really great to have 30yr follow-up on patients, since we know that atherosclerotic disease begins early in life in the US, that it is progressive, that preventing disease is important (outcomes are better in primary prevention if the atherosclerosis is dealt with earlier on), and that a 10-year timeframe for a healthy 55yo is pretty likely useless in most cases (the results are very likely below the arbitrary cutpoints of 5%,10%, 20%)

        -- but the problem with a 30-year study is that things change a lot over the 30 years: diet, exercise, smoking, adiposity, cardiotoxic exposures (stress, pollution, microplastics), comorbidities that increase cardiovascular risk (diabetes, hypertension, metabolic syndrome, inflammatory diseases…), medications taken (some helpful such as statins or anti-inflammatories such as colchicine; some harmful such as some cancer drugs, some prior drugs such as rosiglitazone and likely some newer ones that we will find out about later). And all of this would affect the results of any long-term study, since the initial understanding of cardiovascular risk factors at the beginning of the study has changed so much in the past 30 years, so they did not have the full set of granular data that we would collect now

    -- it does make sense that a single value of Lp(a) at baseline will hold long-term

    -- but this is not clear for hs-CRP: a baseline value may not be reflective of the hs-CRP load over time:

        -- CRP can be elevated if one has a short-term inflammatory condition (eg infection), so it would make sense to recheck it a few months later to see if it normalized

        -- CRP would increase in those who subsequently develop long term inflammatory illnesses (rheumatologic, inflammatory bowel disease, chronic infections, etc) , so these individuals would have much more exposure to high CRP levels  by the 30-year mark than their initial value

        -- and, since CRP also varies with visceral adiposity, diet, exercise, stress, depression, etc, the measured CRP would track according to positive or negative changes in any of the many conditions that are associated with systemic inflammation

    -- same with LDL, which very likely changes over time:

        -- this is true for both nonpharmacologic reasons (changes in diet, exercise, visceral adiposity, etc) as well as medications (statins lower LDL levels, as well as inflammation independent of lowering the LDL levels) and the actual achieved LDL level should be tracked over the 30 years to have a true representation of what the vasculature is exposed to) 

 

Limitations:

-- the results from a study of healthy young professional women may not be generalizable to the general population of women, or extendable to men. however, it is certainly welcome to have more analyses of risk factors for women, since women tend to be underdiagnosed and undertreated for cardiovascular symptoms

-- it is not clear what the role of being assigned to take aspirin was, since it is anti-inflammatory and could affect outcomes given that one of the major risk factors was an increased CRP level that aspirin lowers

-- was there a selection bias in the women who volunteered to have blood drawn (70% of them)? Were they different than the others in substantial ways? Were they even more different from the average 55yo healthy women in the general population?

-- we do not have the important granular data over the 30-year period to have really solid results: so many of the potential risk factors likely have changed a lot, as per above, including diet, exercise, smoking, visceral adipiosity, development of inflammatory conditions (diabetes, rheumatologic, depression, stress,....

-- and we do not have important cutpoints as to what specific level of hs-CRP, LDL, or Lp(a) begins to be associated with increased cardiovascular risk (the broad quintiles are helpful, but it would be more useful clinically to know what specific values are associated with increased risk)

 

So, this study does bring up several issues:

-- they addressed one of the major clinical issues in cardiovascular disease prevention: it is important to have a long-term predictive models (ie 30 years instead of 5-10 years) of cardiovascular risk, since atherosclerosis is a long-term disease that begins in individuals early on, progresses to clinical disease manifesting itself typically several decades later, and we know that:

    -- essentially everyone aged 15-34 has lipid streaks in their coronaries, 2% of men aged 15-19 have advanced atherosclerotic lesions, and 20% of men and 8% of women have advanced lesions by age 30-34

    -- the first cardiac event has a significant mortality associated with it (on the order of 30 to 40%, much of it occurring prior to reaching the hospital: Law MR. Arch Intern Med. 2002; 162: 2405-2410), and should be aggressively avoided

    -- subsequent cardiac events in the survivors are at a much higher likelihood, despite statin therapy,  than those who have never had had a cardiac event

-- there are independent contributions of LDL cholesterol, hs-CRP, and Lipoprotein (a) in risk for atherosclerotic disease

    -- these 3 biomarkers are useful to measure early on in order to address problems non-pharmacologically and pharmacologically as needed at that time, prior to the advent of more advanced atherosclerotic disease

    -- though, as noted, we do not have much to do at this time for high Lp(a) levels, though high Lp(a) in association with high LDL and/or hs-CRP levels seems to confer an additional significant increase in cardiac risk; ie, these individuals are at higher risk because of the Lp(a) component and should therefore have even more aggressive nonpharm (diet, exercise, smoking cessation, stress reduction, weight reduction....) and med treatments for their more treatable  LDL and hs-CRP levels.  

-- And these issues should be addressed early in life... (and not waiting to check lipids until age 35 or even older in men and 45  in women, for example)

geoff

-----------------------------------

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@bidmc.harvard.edu

  

to get access to all of the blogs:  go to http://gmodestmedblogs.blogspot.com/ to see the blogs in reverse chronological order

  -- click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

  -- or you can just click on the magnifying glass on top right, then type in a name in the search box and get all the blogs with that name in them

if you would the article, please email me.

please feel free to circulate this to others. also, if you send me their emails (gmodest@bidmc.harvard.edu), i can add them to the list