|
|
|
A Clinical Specialist in Hypertension Critiques JNC 7
Jay I. Meltzer
There are four major flaws: 1) the use of the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT)2 as an uncontested reference for the supremacy of diuretics; 2) failure to characterize accurately the downside risks of diuretic therapy; 3) the assertion in the treatment algorithm that diuretics are so supreme that even when ineffective, one should raise the dose and continue their use while adding drugs of other classes; and 4) failure to recognize any clinical role for pathophysiology. These are addressed below.
After publication and dissemination of ALLHAT, the medical community witnessed the birth of a new phenomenon: organized advocacy. Traditionally scientists published their work disinterestedly, allowing peers to judge it for themselves, often inserting attacks on their own hypotheses. Today, trialist investigators at the National Heart, Lung, and Blood Institute (NHLBI) complain that “huge trials, years of work, millions spent, results presented, published, and nothing happens.”3 But this is not the case when trials are convincing. For example, when doctors first saw the trial of penicillin in subacute bacterial endocarditis, they quickly changed practice although there were only seven patients in the study.4 Huge trials may not necessarily convince practicing physicians, because they know that the larger the trial, the smaller the effect that the investigators are trying to prove.
The ALLHAT investigators aggressively marketed their version of the study. The 600 participating research physicians became a sales force, instructed in techniques used by pharmaceutical companies. This novel behavior has been dubbed “academic detailing.”3 and 5
The NHLBI spends taxpayer money to “educate” doctors to conform with its conclusions. Does the worldwide medical community agree? Not if you read peer review journals, new studies, and European Society of Hypertension Guidelines.6, 7, 8, 9, 10, 11, 12 and 13 But before the world had time to study and critique ALLHAT, its flawed conclusions were speedily integrated into a speeded-up JNC 7 “EXPRESS.”14 That integration, facilitated by a 50% overlap in the coauthors of JNC 7 and ALLHAT, has been described as a potential conflict of interest.10
Much of JNC 7 is sound, and the sections on the heart and the kidney are scientifically excellent. The problems lie in the clinical logic of the drug treatment recommendations, the drug treatment algorithm, the return to stepped care, and the supremacy given to thiazide diuretics by exaggerating their benefits and minimizing their harms.
In JNC 7, under the subtitle “Rationale for Recommendation of Thiazide-Type Diuretics as the Preferred Initial Agent,” the only support was ALLHAT. Unable to show any difference in the strong primary endpoints, two major arguments were selected from weak secondary endpoints: 1) the increase in heart failure (CHF) risk in both the lisinopril and amlodipine arms, and 2) the increase in stroke in the lisinopril arm over chlorthalidone. In fact, the only endpoint favoring chlorthalidone over amlodipine was the increased CHF risk, which has been criticized for several reasons6, 9, 11 and 13: 1) the early appearance of a sharp divergence in its incidence; 2) the lack of prospective monitoring; and 3) the lack of increase in mortality in subjects with the CHF diagnosis, which should carry a mortality rate of about 25% per year.15 Nevertheless, the claim that chlorthalidone is superior to amlodipine reappears uncontested in JNC 7. Critics have claimed that the stroke difference is an unusable argument favoring chlorthalidone over lisinopril because of BP differences and unconventional step-up treatment sequences highly unfavorable to African American subjects.6, 9, 11 and 13 Nonetheless, this ALLHAT argument also appears unopposed in JNC 7.
The authors then comment on the Second Australian National Blood Pressure Study (ANBP2), “Cardiovascular outcomes were less in the ACEI group, compared with diuretics.” Left out was that, unlike ALLHAT, in which 90% of the subjects were previously treated and then switched without washout, ANBP2 patients began the study untreated, making this a true “first-step” study. Stroke and heart failure were both lower in the lisinopril arm, the exact opposite of ALLHAT. ANBP2 did not confirm ALLHAT; it contradicted it, but that did not moderate the JNC 7 claim of unequivocal diuretic supremacy.
The ALLHAT authors’ argument requires them to prove that diuretics must also be proven safer. However, both ALLHAT and JNC 7 have a biased view of the harms of diuretic therapy. Therefore they cannot come to a fair or logical conclusion as to the total health benefit of these agents.
Seven objective harms of diuretics were acknowledged by the ALLHAT authors in their article on rationale and design16: hypokalemia, increased blood sugar (including new-onset type 2 diabetes mellitus), lipids, and uric acid, hypomagnesemia, insulin sensitivity, and ventricular ectopy (including sudden cardiac death). Only the first three were monitored. ALLHAT found that chlorthalidone caused highly significant decreases in potassium, increases in blood sugar and new-onset diabetes, and higher serum cholesterol when compared with the effects of amlodipine and lisinopril. ALLHAT did not evaluate postural hypotension causing trauma, erectile dysfunction, or distressing polyuria and dysuria, common harms that practicing clinicians must confront.
ALLHAT stated that hyperuricemia was an accepted side effect of diuretic therapy16; yet no assessment of its incidence was in the ALLHAT protocol, nor was gout. The JNC 7 also ignored the fact that uric acid is a recognized risk factor for cardiovascular disease.17
The JNC 7 reports the ALLHAT data showing that new-onset diabetes mellitus after 4 years of therapy was 11.8% for chlorthalidone, 9.6% for amlodipine, and 8.1% for lisinopril, but claims that “this did not translate to fewer events for the ACEI or CCB groups.” Also noted was “the tendency of diuretics to worsen hyperglycemia, but this effect did not produce more CV events compared with other drug classes.” But the authors did not do a subanalysis of those with either new-onset diabetes or increased blood sugar to try to detect an effect that might have been lost in the thiazide group as a whole. The JNC 7 asks us to believe that because no measurable differences occurred in 4 years, increased hyperglycemia and new-onset diabetes will not cause events in the future. Their Reference 185, used to argue diuretic safety, demonstrated the opposite, that serum potassium fell 4.1 to 3.5 mEq/L and uric acid rose 5.8 to 7.4 mg/dL, glucose 104 to 114 mg/dL, and serum cholesterol 203 to 212 mg/dL.18
In a regional meeting of the American Society of Hypertension (ASH) in New York, a member of the JNC 7 Committee addressed the question of new-onset diabetes in ALLHAT as follows: if patients develop diabetes while on chlorthalidone, they are already taking the best drug for cardiovascular protection.19 and 20 But is chlorthalidone still the best drug after causing diabetes when compared with the effect observed in subjects taking either lisinopril or amlodipine who never got diabetes? Although JNC 7 is meant to be a manual for practitioners, it never addresses the question, “What do you do when a patient develops new-onset diabetes while on a diuretic?” The issue regarding the long-term cardiovascular risk in patients with diuretic-induced diabetes mellitus has not been addressed by ALLHAT or any other study, nor has the cost related to monitoring and treating this complication.
The authors admit that “hypokalemia could contribute to increased ventricular ectopy and possible sudden death.”21 They also use the data from the subgroup analysis of the Systolic Hypertension in the Elderly Program (SHEP)22 showing loss of cardiovascular and stroke protection in those diuretic-treated patients with potassium <3.5mEq/L. But the authors counter these two careful large, long-term studies with the statement, “Other studies have not demonstrated increased ventricular ectopy as a result of diuretic therapy.” The reference for that assertion is not “other studies” but rather one study (Ref. 127 in their work),23 which consisted of only 4 weeks of diuretic treatment in 44 subjects. Can the reader conclude from the JNC 7 analysis that diuretic-induced hypokalemia is safe? This is another example in which JNC 7 authors attempt to blunt the findings of excellent large trials using poorly designed small trials because the latter justify their bias regarding diuretic therapy.
The JNC 7 states, “Trials longer than 1 year duration using modest doses of diuretics generally have not shown (cholesterol) increases in diuretic treated patients,” quoting two references (JNC 7 Refs. 124 and 125). Reference 124 in their study is an opinion piece, rather than a definitive reference. Reference 125 in JNC 724 is a multicenter, randomized, double-blind, parallel-group clinical trial in 15 Veterans Affairs centers, and its authors include members of the JNC 7 Coordinating Committee. It is not, as claimed in the text, a “trial longer than 1 year,” because it lasted exactly 1 year. Although it showed that cholesterol did not increase in those subjects whose BP responded to hydrochlorthiazide, the “nonresponders” had a significant rise in cholesterol of 10.9 mg/dL, P < .005. The authors24 did not recommend continuing diuretic treatment in nonresponders. This distinction assumes importance because their treatment algorithm mandates beginning everyone on a diuretic and, if there is no response or an inadequate response, raising the dose and continuing diuretics while adding drugs from other classes! This means that some patients will be subjected to increased serum cholesterol levels, increasing cardiovascular risk without clear BP benefit. It is widely accepted that a 1% increase in total cholesterol is associated with a 2% increase in cardiovascular risk.25 This is another example in which JNC 7 authors ignore a purported tenet of medical practice when it does not suit them.
The 20,000-subject nested ALLHAT study failed to show a significant event benefit for statins.2 This was explained by the fact that practically everyone in ALLHAT received statins, including “controls.” The widespread statin use could have blunted the expected cholesterol rise described in references 17 and 23 in JNC 7 and narrowed the still significant cholesterol rise between the diuretic and the lisinopril and amlodipine arms, thus favoring diuretics in yet another way.
In summary, the JNC 7 drug treatment recommendations are based on an analysis that exaggerated the benefits and minimized, distorted, and even ignored the harms of diuretics, thereby compromising their conclusions concerning diuretic supremacy.
The basic concept underlying the JNC 7 treatment algorithm is the belief that the universe of patients with hypertension of unknown cause is homogeneous. This view underlies the epidemiologic concept that hypertension is a risk factor and not a group of diseases of multifactorial pathogenesis. Treating risk, not disease, means treating all patients with essential hypertension in the same way.
An alternative view takes its lead from lessons learned from hypertension when the cause is known, as in the monogenic forms26 and in patients with “secondary hypertension.” It is clear from these examples that when we know the cause, we know the best drug. As noted by Deary et al,26 in the clinical universe of essential hypertension we usually do not know the cause, but that does not mean that universe may not contain a basket of relatively discrete oligogenic syndromes. If so, most patients may respond to one “best drug” that differs reproducibly among patients.
The powerful arguments for this second mechanistic approach call for strategies optimizing antihypertensive monotherapy based on an understanding of the renin-angiotensin system.27 This method takes time. The JNC 7 urges rapid addition of drugs to get BP down “in a timely manner.” What’s the rush? In a disease that lasts a lifetime, why not take the time to find an individual’s “best drug”? Any additional expense caused by increased office visits and N-of-1 trials at the start can be more than made up for by a more effective drug regimen throughout a lifetime.
Lessons from treatment responses in hypertension in which the cause is known suggests that the height of the BP bears little relation to the need for combination therapy. In malignant hypertension (MH), in which the pressure may reach 250/150, treatment with a single oral drug, captopril, will promptly bring the pressure down.28 and 29 Knowing the cause, uncontrolled renin secretion by the kidney, leads to finding the “best drug,” a blocker of angiotensin generation. Starting with two drugs just because the BP is >160 mm Hg has not proved to be more effective and may frustrate any chance of learning what that best drug is, complicating forever the patient’s drug treatment.
Initial combination therapy is arguably rational but is certainly not evidence based. Its rationale is merely that most patients end up on two drugs anyway. But the therapeutic sequencing upon which that sad fact is based was not rational to begin with. A tautology hardly forms a powerful logical basis for making a sad fact a rule.
In the section on identifiable causes of hypertension, the diagnostic test recommended for the diagnosis of primary hyperaldosteronism is a 24-h urinary aldosterone excretion rate (UAER). This recommendation misleads doctors because UAER cannot distinguish between the rare primary and the common secondary forms. The correct testing for the diagnosis of primary hyperaldosteronism is measurement of both UAER and plasma renin (PRA). The PRA is critical because it is high in secondary hyperaldosteronism and is suppressed in primary hyperaldosteronism. But JNC 7 never mentions renin. Such an omission is scientifically inexplicable.
In the same section, the diagnosis of renovascular hypertension (RVH) is defined only in terms of the anatomic diagnosis of renal artery stenosis. The fact that RVH is caused by “ischemia” stimulating the juxtaglomerular apparatus to release renin is ignored—a threshold effect around 80% stenosis.30 Attempts to diagnose RVH purely on anatomic grounds ignores Dr. Harry Goldblatt’s research and a huge scientific literature. Many patients with anatomic renal artery stenosis do not have hypertension.31 Only proof of renin stimulation can ensure the correct mechanistic diagnosis. For that, one must measure PRA and, when necessary, renal venous renin, with appropriate consideration for the effects of drugs, renal dysfunction, comorbidity, and the distinctions between unilateral and bilateral disease on the sensitivity and specificity of the measurements.
In the section on hypertensive crisis, classic clinical taxonomy is replaced by a scheme based on triage. The disease malignant hypertension (MH) is replaced by “hypertensive emergency,” as though all BP above a certain level should be treated the same way. This negates relating treatment to any known pathophysiologic mechanisms. In MH, PRA is very high. In other patients with “hypertensive crisis” and a BP of 250/150, renin may be low due to volume expansion. Both patients are pathophysiologically very different and should not be treated the same way.29
What do all of the above criticisms have in common? They all question the misguided effort of JNC 7 to simplify the treatment of hypertension, to make it easy, fast, and cheap. “Pathogenesis of Hypertension,”32 a short review of the known physiology of hypertension, concludes, “Current treatment guidelines generally recommend a generic approach to treating hypertension, with little emphasis on selecting therapy on the basis of the underlying pathophysiology.” Yet these current guidelines claim the name “Evidence Based Medicine.” Although it may be difficult to track the multifactorial perturbations that culminate in elevated BP, there is no reason not to apply what scientific evidence we do have in the service of best practices.
One cannot ignore malignant hypertension, curable with monotherapy, and not have patients suffer consequences. One cannot ignore decades of research on Goldblatt hypertension and the role of renin release in renovascular hypertension and not mistreat renal artery stenosis. One cannot diagnose primary hyperaldosteronism without understanding the science of hormonal biofeedback and renin suppression. The JNC 7 ignores the science in hypertension for the sake of statistical simplicity.
Why write this critique now? An erroneous
conclusion, propagandized authoritatively by admired governmental advisory
bodies, can carry the same weight as a sound one. But it is misplaced respect
for those in power if we concentrate solely on their sloganeered conclusions
like “cheaper is better” and not look to how they reached them. Therefore I
have, for the record, probed misleading elements in the data and flaws in the
JNC 7 logic used to reach questionable conclusions. If we ignore
pathophysiology to make treating BP simple, we fail to honor our covenant with
patients to provide each with the best possible care.
American Journal of Hypertension
Volume 18, Issue 7 ,
July 2005, Pages 894-898