On Cerebrotoxicity of Antihypertensive Therapy and Risk Factor Cosmetics

Franz H. Messerli; Chirag Bavishi; Adrian W. Messerli; George C.M. Siontis


Eur Heart J. 2021;42(7):758-760. 

The large database of the UK Biobank cohort allows the separate assessment of the impact of past blood pressure (BP) and concurrent BP on white matter hyperintensities (WMHs) as was done in the study of Wartolowska and Webb published in the issue of the European Heart Journal.[1] Past values of both systolic (SBP) and diastolic blood pressure (DBP) were positively associated with a higher WMH load later in life; however, their longitudinal analyses documented WMH load to be more strongly associated with past DBP than with past SBP, especially under the age of 50. WMH load was also strongly associated with concurrent BP, both systolic and diastolic, following adjustment for the other BP measure, age, sex, cardiovascular risk factors, source of BP measurements, and the assessment centre. History of diabetes, smoking and increased arterial stiffness were also associated with WMH. Although the effect of SBP diminished with adjustment, concurrent SBP remained a stronger predictor than concurrent DBP.

Graphical Abstract

For any given blood pressure (BP) level, treated hypertensive patients have significantly more white matter hyperintensities (WHMs) than untreated patients. Even well-controlled treated patients with a BP <120–130/70–80 mmHg do worse than untreated patients with uncontrolled hypertension (BP >140/90 mmHg).

None of these findings is particularly surprising, although we certainly have to appreciate that in this study, DBP has left its Cinderella status to which it had been relegated by some powerful editorials in the past. In 1982, a lead editorial of the New England Journal of Medicine stated that in men aged 30–40 with a DBP between 95 and 105 mmHg the suggested approach was not antihypertensive therapy but to improve their lifestyle![2] In addition, 'systolic pressure is all that matters' was the credo put forward in the Lancet a dozen years ago by Drs Williams, Lindholm, and Sever.[3] This provocative commentary was simply an encore of one of the author's 'Abandoning diastole' a decade earlier.[4] This was accompanied by the call to apply the '150' rule, although, in the uncomplicated patient and those older than 60, the treatment threshold was raised to 160 mmHg. Importantly, these recommendations have to be seen in their proper context. At that time the Health Survey for England, using the most conservative definitions of BP control (<160/95 mm Hg), reported that ~70% of people with known hypertension were uncontrolled.

In this day and age, after SPRINT,[5] the pendulum has been precipitously swinging the other way, to the extent that we fear it unexpectedly may come off the hook. The Lancet recently suggested that 'Updating (NICE guidelines) so that all patients with a systolic blood pressure (SBP) above the 120 mm Hg ideal and a cardiovascular risk score of 10% or higher would be offered treatment would have shown true grit'.[6] We begged to differ and dared to say so.[7]

Of particular interest in the study of Wartolowska and Webb are the effects of antihypertensive therapy.[1] Their data (only found as Supplementary material online) seemingly indicate that antihypertensive therapy per se, regardless of BP achieved, is a powerful risk factor for WMHs. Treated patients whose BP was very well controlled consistently had a higher WMH burden than patients with untreated hypertension. In fact, individuals under hypertensive treatment, irrespective of whether hypertension was well controlled or not, showed a considerably higher weighted average increase in WMHs compared with individuals without antihypertensive treatment. This was true regardless of whether BP values at baseline or BP values at follow-up were analysed. Even patients whose BP was very well controlled (<120/70 mmHg) had a higher WMH burden than untreated and uncontrolled patients with BP >140/90 mmHg (1.545 and 1.398, respectively). Hence patients who had their BP lowered with antihypertensive medication by >20 mmHg systolic or diastolic elicited a substantial increase in WMH burden. This indicates that antihypertensive therapy per se, independent of its effect on BP, was a powerful risk factor for WMHs. Apparently, antihypertensive therapy must be cerebrotoxic!

Nevertheless, the authors attempt to reassure us by stating 'the lower WMH in treated patients with lower BP could imply an effect of treatment'. True, WMH did increase with progressive severity of treated hypertension, but patients with untreated hypertension still faired consistently much better and exhibited a similar, although distinctly steeper, pattern between BP and WMH. The authors further state 'Our results suggest that to ensure maximal prevention of WMH in late life, control of DBP may be required in early mid-life, even for DBP below 90 mmHg, whilst control of SBP may be more important in late life'. This is usually done by intensifying antihypertensive therapy, i.e. either by up-titrating current antihypertensive drugs or by adding on another class. Do the authors truly mean that we now should give more of these 'cerebrotoxic' antihypertensive agents to even better control BP, diastolic before and systolic after age 50? Should we not rather stop all antihypertensive therapy immediately regardless of BP, to prevent further brain damage?

In sharp contrast to the present data, a recent review and meta-analysis of 209 prospective studies indicated that the use of antihypertensive medications conferred a 21% reduction in dementia risk.[8] With regard to BP exposure in late life, high SBP, low DBP, excessive BP variability, and orthostatic hypotension were all associated with an increased dementia risk. The U-shaped dose–response curve indicated that the protective window of DBP was between 90 and 100 mmHg for the lowest risk of dementia.

How do we get out of these paradoxes? We respectfully suggest the following. (i) The fact that an increase in past DBP and in concurrent SBP is associated with an increased WMH burden does not allow the inference that lowering such BP will lower WMH burden. This simply may be wishful thinking and what we call risk factor cosmetics! Admittedly, the observed 'cerebrotoxicity' may be the result of bias by indication, as suggested, but since it is very consistent, a detrimental effect of BP lowering on cerebral function or direct toxicity of some specific drugs or drug classes cannot be excluded. As a corollary to the present study, Mahinrad et al. recently reported that cumulative exposure to higher BPs from young adulthood to mid-life, even at levels below the clinical definition of hypertension, was associated with worse gait and cognitive function in mid-life.[9] However, neither this observational study nor the present study allows any conclusion regarding possible benefits of antihypertensive therapy. (ii) The effects of long-term exposure to elevated BP and/or its treatment on target organ disease such as WMH burden may no longer be reversible even when normotension is achieved by antihypertensive therapy. In the REGARDS cohort, the risk of stroke remained almost 2.5-fold higher in hypertensive patients treated with three drugs despite SBP being <120 mmHg compared with untreated subjects with the same SBP level.[10] Despite the most efficacious antihypertensive therapy, there remained a substantial residual risk. (iii) By definition, all antihypertensive drugs lower BP. However, with regard to possible benefits of antihypertensive therapy or absence thereof, not all antihypertensive drugs are created equal. For any given fall in mmHg, the effect on target organs and to a lesser extent on morbidity and mortality has been shown to differ from one drug to another on numerous instances. After a follow-up of 12 months only, a recent prospective randomized trial of 176 participants with mild cognitive impairment demonstrated candesartan to be superior to lisinopril on executive function and memory outcomes, effects seemingly independent of and possibly additive to their BP-lowering effects.[11] (iv) Both BP and WMHs are merely surrogate endpoints. What clinically matters is cognitive impairment and dementia. Leapfrogging from one surrogate endpoint (BP) across the other (WMHs) to the true endpoint of dementia and back to BP for therapeutic recommendations is prone to confounding and fraught with incalculable errors.

So, do we actually believe what the study data show, namely that antihypertensive therapy increases WMH burden, i.e. is cerebrotoxic? The answer is no, simply because there is presently no rational pathophysiological correlate explaining such an assumption. Do we believe the authors' suggestion that antihypertensive therapy lowering DBP to below 70 mmHg in patients younger than 50 will lower WMH burden, i.e. is cerebroprotective? The answer again is no; although there may be a rational pathophysiological correlate for this assumption, lowering DBP to these levels by antihypertensive therapy has never shown to confer benefits. Nevertheless, large-scale analyses have shown that there is an extensive overlap in risk factors even among seemingly diverse diseases, underscoring the importance of quantifying the magnitude of confounding. In the era of big data, risk factor cosmetics remain one of the most common fallacies in medicine; when a risk factor is elevated, its 'correction' must be beneficial. As the present study from the UK Biobank cohort shows, 'It ain't necessarily so!'[12]