Hypertension, NSAIDs, and Lessons Learned

[Guest guest]

Hypertension, Nonsteroidal Antiinflammatory Drugs, and Lessons Learned

Journal of Rheumatology

June 2004

Full Text Editorial

JANET E. POPE, MD, MPH, FRCPC,

Associate Professor of Medicine,

Division of Rheumatology, Department of Medicine,

The University of Western Ontario,

London, Ontario, Canada

Address reprint requests to Dr. J.E. Pope, Rheumatology Centre, St.

ph's Health Centre, 268 Grosvenor Street, Box 5777, London, ON N6A

4V2. E-mail: janet.pope@...

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

Newer guidelines have suggested that in patients over 50 years of age,

systolic blood pressure (BP) is more important than diastolic BP as a

cardiovascular risk factor1. Also, the chance of developing hypertension

(HTN) over their remaining lifetime in normotensive people at age 55

years is as high as 90%. A reduction of 5 mm Hg in systolic BP decreases

mortality by 7% (the reduction is greater for cardiovascular mortality).

In a study of 463 people with rheumatoid arthritis (RA), BP was not

always measured in the charts of patients. Nineteen percent had no BP

recorded, 37% of those with readings had HTN, and the BP was

uncontrolled (worse than 140/90) in 77%2. Nonsteroidal antiinflammatory

drugs (NSAID) are most apt to increase systolic BP. It appears that

rheumatologists are not doing a good job with respect to adequate

recognition and treatment of HTN.

In this issue of The Journal, Wolfe, et al discuss the prevalence of

self-reported HTN and edema in the National Data Bank for Rheumatic

Diseases among traditional NSAID users and coxib users3. The following

key messages will be highlighted in this editorial: where these findings

fit in with other studies of BP destabilization with NSAID; the

potential biases of the Wolfe, et al study; the high prevalence of HTN

in our arthritis patients; and the role of this information leading to

good clinical practice.

For over 2 decades, the possibility of NSAID aggravating BP has been

recognized4,5. In fact, much of the work on captopril, an early

angiotensin converting enzyme (ACE) inhibitor, was done to determine why

it was attenuated with NSAID, as well as to determine its prostaglandin

effects6-8. NSAID alter prostaglandins (PG) and result in decreases in

PGI2 and changes in urinary measurements of PGE and PGF1a9. This change

can result in renovascular vasoconstriction and resultant decreased

renal blood flow and can cause or aggravate HTN in some patients who are

dependent on the renin system10,11.

Patients using NSAID at risk for HTN include those with baseline HTN or

borderline BP elevations, the elderly, and likely those with compromised

renal compensatory mechanisms4. Two metaanalyses were published

comparing various traditional NSAID and their effects on BP4,5. The

average incremental increases in mean arterial pressure were small, but

this was because most subjects did not have significant BP aggravation;

however, hidden in the mean values were those who had large and

significant BP elevations. Indomethacin was most likely to increase BP,

and sulindac (an inactive prodrug at the kidney) did not have any

elevations in BP. The take-home message was that patients are usually

lucky, but occasionally significant HTN will occur, particularly in high

risk individuals.

As the first 2 coxibs differentiated themselves in the marketplace,

research was undertaken to compare these agents head to head in the area

of HTN. Consistently, celecoxib showed less effects on BP than

rofecoxib12,13, but discussions regarding dose equivalence and timing of

BP readings followed. HTN as a side effect is uncommon in healthy

normotensive individuals, so differences between coxibs can also be

hidden or magnified, depending on the patient population studied. Thus,

trials comparing normotensive patients would rarely show BP differences,

whereas in elderly hypertensive subjects, the between-drug differences

would be more obvious14,15. From one large study, one could conclude

that clinically important increases in BP in hypertensive patients were

found in 15% of rofecoxib 25 mg od users compared to 7% of celecoxib 200

mg od users, with the majority of effects observed being those on

systolic BP13.

One interpretation of these results would be to follow BP carefully,

particularly early on, in patients with HTN using coxibs. Options to be

considered in those with worsening BP control could be stopping the

NSAID, switching to another drug, lowering the dose of NSAID, and/or

increasing hypertensive medications if the benefit of the NSAID merits

continued treatment.

In general, the greatest attenuation of antihypertensives from NSAID is

observed with ACE inhibitors, and less with beta blockers, calcium

channel blockers, and diuretics. One must be aware of the other renal

effects of combining some of these medications with NSAID, such as

worsening renal insufficiency, and in the case of ACE, angiotensin II,

potassium-sparing drugs, and beta blockers, a risk of hyperkalemia. In a

study of NSAID in the elderly, it was concluded at 2 weeks that BP

elevations were similar between rofecoxib, naproxen, and celecoxib

(funded by Merck)15. This study was small (n = 67) and the subjects were

elderly but normotensive, so worsening BP would be rare in these

patients. Thus trial design can play a role to minimize or magnify drug

differences. It is unknown what role the design of Wolfe's study, which

was funded by Pfizer, may have played in shaping conclusions about the

effects of BP, as even in rofecoxib trials sponsored by Merck, which

were based on normotensive subjects, rofecoxib seemed to have a slightly

increased mean BP with a dose-response relationship when compared to

other NSAID such as naproxen (where a majority of data had been obtained

in phase III and IV studies16,17).

The database in Wolfe's study consists of 9226 patients (77% with RA),

of whom 8538 were eligible for this study3. The prevalence of HTN was

high (about 45%) and it was lowest in the users of traditional NSAID,

possibly due to their younger age, but comparable with the prevalence of

HTN in a recent study of RA of 37%2. Fluid retention for most of the

patients in Wolfe, et al was clinically irrelevant (frequently reported

but not leading usually to discontinuation or changing of medications)

and thus will not be discussed.

Any study has limitations. The patients may or may not be generalizable

to a specific practice, but the large sample size ensures some degree of

" certainty " regarding the observations. A major limitation of the Wolfe,

et al study is self-reporting of HTN and its exacerbation. However, the

authors tried to confirm whether changes in BP medications were made.

Indeed, 91% of self-reported HTN patients were using antihypertensive

medications, and 75% reported changes in dose when they had an

exacerbation of their BP control. The pooling of traditional NSAID data

is not straightforward, as past studies have suggested that the effects

of BP are different with several traditional NSAID4,5. Also, this

methodology would not necessarily justify the separation of coxibs as a

class. Treatment biases are not easily determined in data sets. For

instance, a large Canadian data set suggested that severe asthma

exacerbation was less in asthmatics taking beta blockers19. Of course,

beta blockers can exacerbate asthma, but it is likely that only those

with the most mild disease would be prescribed beta blockers, so the

observation is confounded by mild disease. Similarly, the unknown biases

of prescribing patterns may have affected the results. Wolfe's Table 7

displays the crude HTN rates with specific NSAID. Indomethacin had only

20% increases in HTN and sulindac had 43%, which is in direct contrast

to previous metaanalyses of randomized controlled trials with these

agents, and reveals a quirk in the data. One wonders if mostly younger

normotensive patients were prescribed indomethacin and hypertensive

patients received sulindac, as sulindac is known to be an NSAID that

does not have BP effects. Thus the attribution of HTN worsening in that

group could be due to other factors. Thus one becomes less certain as to

the conclusions of the individual coxibs, although the age and morbidity

of the users of celecoxib and rofecoxib were similar. One explanation

could be that physicians were aware of results of trials showing

exacerbation of HTN with rofecoxib and thus were more likely to measure

BP in that group (detection bias). They tried to adjust for this by

subsetting the data into years before and after the greater awareness of

BP effects of rofecoxib. In the Wolfe, et al study the efficacy of the

NSAID treatment was not measured, so the tradeoffs of risks and benefits

of any drug cannot be determined. Results demonstrated more HTN

exacerbations in the rofecoxib group. We do not know the dose

comparability between rofecoxib and celecoxib.

So what should a clinician do with this knowledge? We must be aware that

HTN is common and underdiagnosed and undertreated in our patients, and

that in hypertensive patients, NSAID may aggravate their HTN control.

One can decide that some drugs are more or less apt to destabilize BP,

but good clinical practice would warrant a common sense approach: being

aware, monitoring BP, particularly early in chronic NSAID users, and

responding to significant elevations of BP.

http://www.jrheum.com/subscribers/04/06/1035.html

I'll tell you where to go!

Mayo Clinic in Rochester

http://www.mayoclinic.org/rochester

s Hopkins Medicine

http://www.hopkinsmedicine.org