Keywordsantihypertensive drugs, pregnancy, preeclampsia, systematic reviews
Editors’ comment: The authors in updating this chapter continue to emphasize the need for better study designs regarding the ideal antihypertensive drugs for pregnant women. A plethora of national guidelines offer contrasting views regarding the hypertensive level at which drug therapy should be given during pregnancy and the authors in the initial sections of the chapter succinctly chronicle the reasons for this confusion. Additionally, systematic reviews of trials are discussed that further explain dilemmas in interpretation and provide guidelines for better-designed trials.
In the initial edition of this text, discussion of antihypertensive therapy in pregnancy was mainly historical; the drugs noted included veratrum alkaloids, opium and its derivatives, a host of sedatives, and even spinal analgesia – the older literature was often unclear whether such treatment was prescribed for hypertension per se or for an eclamptic convulsion. More emphasis, however, was devoted to diuretics – parenthetically vehemently opposed by Chesley – and to the recurring theme that hypertension in preeclampsia might, paradoxically, be treated by volume expansion. Space prohibits republishing these historical vignettes, which bear rereading. Of interest though, use of veratrum viride was incorporated into the treatment of eclampsia even before physicians were aware that a rise of blood pressure accompanied the “puerperal convulsion.” Also, for many years, use of veratrum was called the “Brooklyn treatment,” noted here because Chesley spent most of his career on the faculty of the State University of New York in Brooklyn.
There are several reasons for the paucity of information on antihypertensive treatment in the original Chesley monograph. Effective and tolerable drugs to lower blood pressure only emerged during the last four decades. Even then, many considered hypertension as “protective,” and that the increased pressure was needed to perfuse vital organs such as the kidney and brain in the setting of arteriosclerosis. This view persisted longer in the pregnancy literature, where it was feared that lowering blood pressure would decrease placental perfusion, and thus fetal nutrition and growth. This topic, still the subject of some controversy, will be addressed subsequently. More important was the absence of multicenter randomized trials of a scale large enough to assess the safety and efficacy of specific antihypertensive medications during pregnancy, the risks of congenital anomalies, and infant outcomes. This, too, will be revisited.
Goals of Antihypertensive Drug Therapy
Prior to the availability of safe and effective antihypertensive drugs, hypertension could only be defined by evaluation of the normal distribution of blood pressures within the population or by associations with morbid sequelae at different levels of pressure; the former remains the basis for definitions of hypertension in children, while the latter approach is taken in nonpregnant adults. Now, by contrast, hypertension could better be defined as a level of arterial pressure whose pharmacologic control would improve outcome for the population at risk.
In nonpregnant adults, blood pressure control can decrease the remote incidence of stroke, coronary heart disease, congestive heart failure, and cardiovascular mortality. Each of these outcomes – as well as others, for example, nephrosclerosis – is due only in part to hypertension per se, adding to or interacting with other risks such as smoking, dyslipidemia, diabetes mellitus, race, age, sex, and familial predisposition. In addition, the ability to detect improved outcome due to blood pressure control also depends on the severity of hypertension and the presence of target organ damage, the duration of follow-up observation, the completeness of blood pressure control, perhaps the pathophysiology leading to hypertension, and the drugs used to treat it. For example, treatment trials have been most easily designed to show decreases in stroke incidence in patients with severe hypertension; however, much larger, longer duration, or more restrictively designed trials were required to show benefits for coronary disease or in patients with only mild or moderate hypertension.
Hypertension in pregnant women is different. Here, the major goals of treatment are to safeguard the mother from acute dangers or irreversible insults during or immediately after the pregnancy while delivering a healthy infant. In this respect, one balances short-term maternal outcome against possible long-term consequences of intrauterine drug exposure or hemodynamic insult on fetal and childhood growth and development. While comorbidities such as diabetes mellitus, target organ dysfunction at baseline, and (uncommon) secondary causes of hypertension may certainly interact with maternal hypertension and alter strategies for its control, these concerns will not apply to most hypertensive pregnant women. The majority of women with blood pressures high enough to warrant treatment during pregnancy will either have chronic essential hypertension or preeclampsia leading to threatening levels of pressure which occur when there are compelling reasons to extend the length of the pregnancy.
It should be noted that the balance of risks and benefits for antihypertensive therapy will differ for women with chronic hypertension present from early in gestation, whose fetuses may have greater drug exposure during early stages of development, compared with women who develop hypertension, viz., either gestational hypertension or preeclampsia, closer to term. Well-designed and adequately powered trials, of which there are shamefully few and too many poorly designed – despite our pleas in previous editions – are still needed to account for this clinical heterogeneity.
Maternal risks which may justify pharmacotherapy include that of superimposed preeclampsia, which, with its morbid outcomes, appears to account for most complications ascribed to chronic hypertension. Additional risks are those of placental abruption, accelerated hypertension leading to hospitalization or to target organ damage, and cerebrovascular catastrophes. Risks to the fetus include death, growth restriction, and early delivery, the latter occurring in many cases due to concerns regarding maternal safety. As will be discussed later, there are no convincing data to currently suggest that antihypertensive treatment of women with mild to moderate chronic hypertension prevents any of these outcomes, except perhaps for worsened hypertension or the need for additional antihypertensive therapy.
General Principles in the Choice of Antihypertensive Agents
Blood pressure, in its simplest conceptualization, is determined as the product of cardiac output and systemic vascular resistance. The latter is sensitive to the structure of small arterial and arteriolar resistance vessels, activity of local vasodilator and vasoconstrictor systems, humoral influences such as the renin–angiotensin system (see Chapter 15 ), and the activity of the autonomic nervous system. The former is sensitive to changes in volume status and autonomic tone; as other influences on intrinsic myocardial contractility are usually minor in healthy women. These physiologic targets, sometimes obscured in the chronic state by vascular autoregulation or our limited ability to measure relevant volumes or pressures with precision, provide the rationale for each of the available pharmacologic strategies for control of hypertension. Further, due to the homeostatic nature of blood pressure control, even when pathologically elevated, this simple physiologic construct suggests likely mechanisms of apparent resistance to antihypertensive drugs, especially when used as single agents. In nonpregnant hypertensive patients, choice of a specific antihypertensive drug is usually rationalized by the severity of hypertension and immediate risk of end-organ damage, the desired time–action characteristics of the drug, specific comorbidities, spectrum of possible adverse drug effects, cost, and known secondary causes of the hypertension. Therapy may also be based on outcomes of well-conducted, large clinical trials, on the systematic review of many smaller randomized trials, or on broad, population-based assumptions regarding the likely physiologic mechanisms leading to hypertension in a given patient. A brief review of the above considerations follows below, but it is tempered by the knowledge that, in mild to moderate hypertension, most available agents appear effective in a similar proportion of patients when used as monotherapy, and most hypertension can be adequately controlled by a combination of two rationally paired drugs.
In nonpregnant adults with hypertensive emergencies defined by systolic BP≥180 mm Hg or diastolic BP≥120 mm Hg with evolving target organ damage, blood pressure is usually controlled acutely, albeit only to a target of ~160 mm Hg systolic so as not to compromise organ (especially cerebral) perfusion, by use of rapidly and short-acting, easily titrated parenteral agents. These include, most commonly, sodium nitroprusside (a nitric oxide donor), enalaprilat (an ACE inhibitor), hydralazine (a direct vasodilator), labetalol (a combined α- and β-adrenergic antagonist), nicardipine (a dihydropyridine calcium entry blocker), and fenoldopam (a specific dopamine-receptor antagonist). Short-acting (i.e., immediate-release) oral or sublingual nifedipine, used in the past for acute blood pressure control, is avoided due to unpredictable hypotensive responses, excessive autonomic activation, and precipitation of acute myocardial ischemia.
In the chronic setting, results of large randomized controlled trials provide compelling evidence that angiotensin-converting enzyme (ACE) inhibitors or angiotensin II (type 1; AT-1) receptor blockers (ARBs) should be the cornerstone of antihypertensive therapy in nonpregnant patients with established diabetic nephropathy because they significantly slow the progression of renal failure. Extrapolation from these findings and the results of other studies support the use of these drugs in diabetic patients with less evident nephropathy or in patients with proteinuric nondiabetic renal insufficiency. Similarly strong evidence supports the use of β-blockers (lacking intrinsic sympathomimetic activity), ACE inhibitors, ARBs, and aldosterone antagonists to prevent death in patients who have suffered a myocardial infarction and to improve both survival and functional status in patients with congestive heart failure, whether or not they are hypertensive. Finally, diuretics appear especially efficacious in avoiding a variety of morbid complications, especially stroke, in patients with isolated systolic hypertension, particularly in the elderly. Indeed, based on the results of very large trials such as ALLHAT – Antihypertensive and Lipid Lowering Heart Attack Trial – diuretics are considered first-line agents in most nonpregnant patients with essential hypertension who lack a “compelling indication” for a different drug. Another evolving literature recognizes that the large population of patients with more severe hypertension – defined as>20 mm Hg systolic or>10 mm Hg diastolic above treatment targets in the Joint National Committee (JNC) 7 guidelines – will require two or more agents for blood pressure control, advocating initial therapy with rationally chosen drug combinations rather than focusing on the drugs most likely to achieve control when used as monotherapy. This change in thinking has further focused research on the comparative effectiveness of such drug combinations.
In the absence of compelling data from controlled trials, drug choice may be influenced by reasoned extrapolation of the impact of known pharmacology of specific antihypertensive agents on medical comorbidities in individual patients. For example, β-blockers, even those with relative β selectivity, are routinely avoided in asthmatic patients due to their ability to provoke bronchospasm. They are also avoided in some patients due to their potential to acutely exacerbate systolic heart failure, their ability to mask autonomic symptoms of hypoglycemia in diabetics, and their capacity to worsen atrial–ventricular conduction defects. Conversely, β-blockers are rational agents in hypertensive patients also suffering from angina, supraventricular tachyarrhythmias, benign essential tremor, or migraine, as they are often useful in these latter conditions even when hypertension is absent. Thiazide diuretics are reasonably avoided in some patients due to their capacity to exacerbate hyperuricemia and gout, to impair glucose tolerance, or to worsen hypercalcemia in patients with primary hyperparathyroidism. Potassium-sparing diuretics, ACE inhibitors, or ARBs are all reasonably avoided in patients with a severe potassium excretory defect, most commonly due to hyperkalemic distal renal tubular acidosis and associated with diabetes mellitus, sickle-cell nephropathy, or obstructive uropathy.
Chronic hypertension in young men is more often associated with increased cardiac output and is often responsive to monotherapy with β-blockers. Likewise, and in spite of a striking lack of conformity in the clinical assessment of salt sensitivity, African-Americans and elderly women are more often apt to manifest low-renin hypertension, relative defects in the renal excretion of a salt load, and/or enhanced blood pressure increments with volume expansion. It is not surprising, therefore, that these groups demonstrate a greater tendency to effective blood pressure control with diuretic or calcium entry blocker monotherapy, with lesser rates of response to β-blockers or ACE inhibitors. In spite of these population differences, careful titration of even haphazardly selected antihypertensive drugs usually leads to acceptable blood pressure control in individual patients, casting some doubt on the importance of such “physiologically based” drug choice strategies in the absence of outcome data from well-designed prospective treatment trials.
All of these insights should lead to some humility as we choose drugs for use in hypertensive pregnant women, in whom both pathophysiologic and clinical trial data are even more limited.
Fetal Safety and Drug Use in Pregnant Women
In the past, both regulations and ethical norms led industry to scrupulously avoid testing drugs in pregnant women. While regulatory changes made by the United States Food and Drug Administration (FDA) in 1993 reversed this policy by encouraging testing of some drugs in pregnant women, such clinical information remains unavailable for most currently prescribed agents. Indeed, rigorous evaluation of pharmacokinetics, biotransformation, maternal efficacy, fetal exposure, and long-term fetal effects of drugs used during pregnancy is generally lacking. Available information, save for assessment of teratogenicity in laboratory animals, is limited and selective. Over 5 years ago, the FDA proposed major changes to the pregnancy and lactation sections of drug labels, which would abandon use of its outmoded classification scheme to categorize potential fetal risks – shown in Table 19.1 – instead including more data and narrative rather than summary (letter) categories. Despite much enthusiasm, final guidelines have not been implemented as of this time, leaving us in limbo, so we include the current risk categories in this chapter, stressing, however, the need to take all clinical information into account when making a prescribing decision in a particular patient.
|A||Adequate, well-controlled studies in pregnant women have not shown an increased risk of fetal abnormalities.|
|B||Animal studies have revealed no evidence of harm to the fetus; however, there are no adequate and well-controlled studies in pregnant women.|
|or Animal studies have shown an adverse effect, but adequate and well-controlled studies in pregnant women have failed to demonstrate a risk to the fetus.|
|C||Animal studies have shown an adverse effect and there are no adequate and well-controlled studies in pregnant women.|
|or No animal studies have been conducted and there are no adequate and well-controlled studies in pregnant women.|
|D||Studies, adequate well-controlled or observational, in pregnant women have demonstrated a risk to the fetus. However, the benefits of therapy may outweigh the potential risk.|
|X||Studies, adequate well-controlled or observational, in animals or pregnant women have demonstrated positive evidence of fetal abnormalities. The use of the product is contraindicated in women who are or may become pregnant.|
Because data from human and animal studies are so limited, most drugs are listed in FDA category C, with the caveat that they should be used only if potential benefit justifies the potential risk to the fetus. This category is so broad as to be useless; it includes those antihypertensive drugs with the greatest history of safe use in pregnant women but, for many years, also included the ACE inhibitors, which are contraindicated in later pregnancy (Class D). Even when drugs are placed in category B, their presumed safety may be a function of the insensitivity of animal tests to predict subtle clinical effects, such as fetal ability to withstand hypoxic stress, changes in functional physiologic development, and altered postnatal neurocognitive development. Limited evidence of clinical safety is often extended injudiciously, such that drugs that lack significant teratogenic potential in early pregnancy may exert devastating effects on fetal organ function nearer to term; conversely, drugs that are safe in the third trimester may have irreversible effects on fetal growth or development when used earlier. By contrast, one can easily err against drug use in pregnancy, for example, when discontinuation of category D drugs might jeopardize both mother and fetus. Examples of this latter scenario would include discontinuing immunosuppressives in a pregnant renal transplant recipient or of antiepileptic agents in a pregnant woman with a seizure disorder.
Choice of an Antihypertensive Drug for Use in Pregnancy
Following discussion of the rationale for choosing an antihypertensive drug, we turn to consideration of each class of agents available in common practice. Each of these drug classes, their apparent mechanisms of action, and, in terms of this chapter’s primary goals, their suitability for use during pregnancy are described separately . Not considered are appropriate targets for blood pressure control during pregnancy, as these have not yet been established. Whether antihypertensive therapy in pregnancy should be targeted to specific hemodynamic endpoints other than systolic and diastolic arterial pressure, such as maternal cardiac output or measures derived from maternal pulse wave analysis, is also not discussed.
Sympathetic Nervous System Inhibition
Introduction of agents to decrease peripheral activity of the sympathetic nervous system marks, perhaps, the beginning of the modern era of effective antihypertensive therapy. Strategies for sympathoinhibition have included ganglionic blockade (e.g., guanethidine), depletion of norepinephrine from sympathetic nerve terminals, such as reserpine which is probably the first antihypertensive to have proven benefit in a prospective clinical trial, α-adrenergic agonists to decrease sympathetic outflow from the central nervous system (e.g., α-methyldopa and clonidine), and specific antagonists of α- or β-adrenergic receptors – including α-antagonists such as prazosin, terazosin, and doxazosin and β-blockers such as propranolol, atenolol, and metoprolol; and combined α/β antagonists including labetalol and carvedilol. Figure 19.1 shows a physiologic scheme for sympathetic nervous system influences on arterial pressure, along with sites of action for drugs such as α 2 -adrenergic agonists and peripheral adrenergic antagonists, which remain widely used in pregnancy.
Centrally Acting α 2 -Adrenergic Agonists
Methyldopa is the prototypical agent of this class. It is a prodrug metabolized to α-methylnorepinephrine, which then replaces norepinephrine in the neurosecretory vesicles of adrenergic nerve terminals. Because its efficacy is equivalent to that of norepinephrine at peripheral α 1 receptors, vasoconstriction is unimpaired. Centrally, however, it is resistant to degradation by monoamine oxidase, resulting in enhanced effect at the α 2 sites that regulate sympathetic outflow. Decreased sympathetic tone reduces systemic vascular resistance, accompanied by only minor decrements in cardiac output, at least in young, otherwise healthy hypertensive patients. Blood pressure control is gradual, over 6 to 8 hours, due to the indirect mechanism of action. There do not appear to be significant decreases in renin and, while the hypotensive effect is greater in the upright than supine posture, orthostatic hypotension is usually minor. Clonidine, a selective α 2 agonist, acts similarly.
Adverse effects are mostly predictable consequences of central α 2 -agonism or decreased peripheral sympathetic tone. These drugs act at sites in the brainstem to decrease mental alertness, impair sleep, lead to a sense of fatigue or depression in some patients, and decrease salivation to cause xerostomia. Peripheral sympathoinhibition may impair cardiac conduction in susceptible patients. Methyldopa can induce hyperprolactinemia and Parkinsonian signs in some patients. In addition, it can cause some potentially serious dose-independent adverse effects. Approximately 5% of patients receiving methyldopa will have elevated liver enzymes, with some manifesting frank hepatitis and, rarely, hepatic necrosis. Likewise, many patients will develop a positive Coombs test with chronic use, a small fraction of these progressing to hemolytic anemia.
Methyldopa has been compared with placebo or no treatment in eight trials focused on pregnant women, or with alternative hypotensive agents in 20 trials. The drug is also unique in that careful, albeit underpowered, studies have also assessed remote development of children exposed to this drug in utero (see “Systematic Analysis” below). Methyldopa is one of our preferred agents for non-emergency blood pressure control during pregnancy since no “modern” antihypertensive has proven more efficacious, better tolerated, or possesses a superior history of clinical safety. Indeed, observations of increased sympathetic nerve activity to the skeletal muscle vasculature in preeclamptic women, reverting to normal along with blood pressure after delivery, lends a compelling physiologic rationale to control of preeclamptic hypertension with agents that decrease sympathetic outflow.
Treatment with methyldopa decreases the subsequent incidence of severe hypertension, but not preeclampsia, and it is well tolerated by the mother without any apparent adverse effects on uteroplacental or fetal hemodynamics or on fetal well being. One placebo-controlled trial that included over 200 women with diastolic BP≥90 mm Hg at entry reported fewer midpregnancy losses in women randomized to methyldopa, an observation that was not confirmed in a later trial of similar size. It is important to note that birth weight, neonatal complications, and development during the first year were similar in children exposed to methyldopa compared with those in the placebo group. While Cockburn et al. noted somewhat smaller head circumference at 7 years of age in the subset of male offspring exposed to methyldopa at 16 to 20 weeks gestation, these children exhibited intelligence and neurocognitive development similar to controls.
Studies of clonidine have been more limited. One third-trimester comparative trial compared clonidine with methyldopa and showed similar efficacy and tolerability, while a small, controlled follow-up study of 22 neonates reported an excess of sleep disturbance in clonidine-exposed infants. Clonidine should be avoided in early pregnancy due to suspected embryopathy; later effects on fetal growth varied with drug-induced changes in maternal hemodynamics. There appears to be little justification for its use in preference to methyldopa given their similar mechanisms of action and the proven safety of the latter agent.
Peripherally Acting Adrenergic-Receptor Antagonists
Cardiac β 1 receptors mediate the chronotropic and inotropic effects of sympathetic stimulation, while receptors in the kidney modulate renin synthesis in response to renal sympathetic input. Activation of β 2 receptors leads to relaxation of airway smooth muscle and to peripheral vasodilatation. Acutely, nonselective β-blockade decreases cardiac output; but with limited change in arterial pressure, due to increased systemic vascular resistance. Over time, vascular resistance falls to predrug levels, resulting in a persisting hypotensive response that parallels the decrease in cardiac output. Moderate decrements in renin, and thus in angiotensin II and aldosterone, may contribute to the chronic antihypertensive efficacy of β-blockers in some patients, likely accounting for their greater efficacy in patient groups not believed to have salt-sensitive hypertension. Some agents, like pindolol or oxprenolol, are partial β-receptor agonists, viz. they possess some limited degree of “intrinsic sympathomimetic activity.” These drugs lead to lesser decrements in cardiac output and β 2 stimulation may even result in significant decrements of vascular resistance. Individual drugs may also possess selective potency at β 1 >β 2 receptors (e.g., atenolol and metoprolol), the additional capacity to block vascular α 1 receptors (e.g., labetalol), and may differ in their lipid solubility, such that hydrophilic agents gain less access to the central nervous system.
β-receptor antagonists are currently considered second-line agents for hypertension in nonpregnant adults (after diuretics) except when specific comorbidities suggest their initial use. They are preferred agents in patients who have experienced recent myocardial infarction, but appear to provide less benefit than diuretics in elderly patients with isolated systolic hypertension, in the prevention of stroke, or when compared with angiotensin receptor blockers in patients with multiple cardiovascular risk factors. Unlike with other agents, long-term use of β-blockers is not associated with (beneficial) remodeling of small resistance arteries which have been hypertrophied due to hypertension, the clinical consequences of this observation being unknown.
Adverse effects are predictable results of β-receptor blockade. They include fatigue and lethargy, exercise intolerance due mostly to (nonselective) β effects in skeletal muscle vasculature, peripheral vasoconstriction secondary to decreased cardiac output, sleep disturbance with use of more lipid-soluble drugs, and bronchoconstriction. While negative inotropy could worsen acute decompensated congestive heart failure or lead to heart block in susceptible patients, β-blockers exert paradoxical benefit in all stages of chronic heart failure with decreased systolic function.
β-blocking drugs have been used extensively in pregnancy and subjected to several randomized trials in pregnancy versus no treatment or placebo and versus alternative agents, and these are discussed further in the section “Systematic Analysis.” Long ago, animal studies and anecdotal clinical observations led to concerns that these agents could cause fetal-growth restriction, impair uteroplacental blood flow, and exert detrimental cardiovascular and metabolic effects on the fetus. However, most prospective studies which focused on drug administration in the third trimester and included a mix of hypertensive disorders have demonstrated effective control of maternal hypertension without significant adverse effects to the fetus. Further reassurance is derived from a 1-year follow-up study, which showed normal development of infants exposed to atenolol in utero . By contrast, when atenolol therapy for chronic hypertension was started between 12 and 24 weeks gestation, it resulted in clinically significant growth restriction, along with decreased placental weight; this observation was supported by subsequent retrospective series and reviews comparing atenolol with alternative therapies. Similarly, several more intensively monitored studies have demonstrated fetal and neonatal bradycardia, adverse influences on uteroplacental and fetal circulations, or evidence of other fetal insults following nonselective β-blockade; these effects may be mitigated with the partial agonist pindolol. By contrast, a more recent and reassuring systematic review showed no clear evidence of fetal or neonatal bradycardia due to maternally administered β-blockers.
Labetalol, which is also an antagonist at vascular α 1 receptors, appears to be the β-blocking drug most widely preferred and prescribed in pregnancy. The drug is administered parenterally to treat severe hypertension, especially the rapid rise that frequently occurs with the appearance or superimposition of preeclampsia. In such cases, its efficacy and tolerability appear equivalent to parenteral hydralazine. When administered orally in chronic hypertension, it appeared safe and equivalent to methyldopa, though high doses might result in neonatal hypoglycemia. Of further concern, its use was associated with fetal-growth restriction or neonatal difficulties in one placebo-controlled study.
The above discussion of β-blockers focuses on control of systolic and diastolic blood pressure, without any attempt to further dissect maternal hemodynamics. While of uncertain significance, we draw brief attention to a separate literature whose authors have used noninvasive – Doppler echocardiography or impedence cardiography – technology to longitudinally assess cardiac output and systemic vascular resistance in hypertensive pregnant women. Such studies have noted supra-normal increments in cardiac output in many women with hypertensive pregnancy and in those destined to develop preeclampsia. These investigators have proposed that use of β-blockers – principally atenolol – to lower cardiac output towards gestational age-adjusted normal values would be beneficial and also that blood pressure control which did not excessively lower cardiac output would avoid uteroplacental hypoperfusion. Despite suggestive small studies reporting decreased incidence of preeclampsia and blunted increases in sFlt-1 following atenolol therapy targeted to hemodynamic endpoints, these interesting hypotheses remain largely untested.
Peripherally acting α 1 -adrenergic antagonists are third-line antihypertensive drugs in nonpregnant adults which have their clearest indication during pregnancy in the management of hypertension due to suspected pheochromocytoma (see Chapter 18 ). Both prazosin and phenoxybenzamine have been used, along with β-blockers as adjunctive agents, in the management of these life-threatening disorders. Because there is but limited and primarily anecdotal additional experience with these agents in pregnancy, their more routine use cannot be advocated.
These are probably the most commonly used antihypertensive agents worldwide. This is because they are among the drugs most clearly associated with beneficial outcome in randomized, controlled trials relating to hypertension in nonpregnant adults. They lower blood pressure by promoting natriuresis and subtle decrements of intravascular volume. Evidence for this mechanism includes observations that the acute hypotensive effect of thiazide diuretics parallels net sodium loss (usually 100–300 mEq, equivalent to 1–2 L of extracellular fluid) over the first few days of therapy, that the hypotensive effect is absent in animals or patients with renal failure, that a high-salt diet blocks the fall in blood pressure, that salt restriction perpetuates the hypotensive response even following diuretic withdrawal, and that infusion of salt or colloid solutions reverses the hypotensive effect. This early phase of volume depletion decreases cardiac preload and thus cardiac output. Decrements in blood pressure reflect the failure of counterregulatory mechanisms, including activation of the renin–angiotensin and sympathetic nervous systems, to raise peripheral resistance enough to defend elevated arterial pressure.
Curiously, the chronic antihypertensive effect of diuretics is maintained despite partial restoration of plasma volume and normalization of cardiac output, due to persisting decrements in systemic vascular resistance. Numerous studies have failed to reveal neurohumoral mechanisms, mediators, or direct actions on the vasculature which lead to this prolonged secondary vasodilator response; at present it is best ascribed to the phenomenon of total body autoregulation. Those patients most apt to have maintained responses to diuretics are those who most effectively achieve this secondary vasodilated state without persisting decrements in cardiac output leading to vasoconstrictor responses (i.e., salt-sensitive, low-renin hypertensives). When a patient appears diuretic-refractory, the addition of a β-blocker or ACE inhibitor is frequently effective; indeed, use of these pharmacologically rational combinations is encouraged in current guidelines.
The adverse effects of diuretics are mainly due to their precipitation of fluid, electrolyte, and solute disturbances, complications that could be predicted from their known effects on renal salt handling. By blocking renal tubular sodium reabsorption, thiazides enhance sodium delivery to distal sites in the nephron. These are sites at which sodium may be exchanged for potassium or protons; modest volume contraction resulting from diuretic use may thus be accompanied by hypokalemia and metabolic alkalosis. Likewise, volume depletion may evoke both thirst and nonosmotic secretion of vasopressin, favoring hyponatremia in some patients. Similarly, decreased renal perfusion may lead to azotemia if severe, but this is usually preceded by increased proximal reclamation of salt, fluid, and solutes, which commonly causes hyperuricemia.
There have been many prospective trials of diuretics or dietary salt restriction in pregnancy, primarily focused on prevention of preeclampsia rather than on treatment of hypertension. A meta-analysis published in 1985 suggested that diuretics did indeed prevent preeclampsia, noting, however, no decrease in the incidence of proteinuric hypertension; thus the claim of efficacy could be ascribed to use of an improper definition of the clinical outcome being studied. In addition, only limited data support the value of salt restriction, which some consider deleterious. Thus, neither of the above interventions appears successful, though diuretics may prevent much of the physiologic volume expansion which normally accompanies pregnancy. While volume contraction might be expected to limit fetal growth, outcome data have not supported these concerns. However, diuretic-induced hyperuricemia, as described above, may complicate the already difficult clinical diagnosis of superimposed preeclampsia. Also, observations of volume depletion and primary systemic vasoconstriction in preeclampsia make diuretics physiologically irrational agents in this disorder.
By contrast, diuretics are commonly prescribed in essential hypertension prior to conception and, given their apparent safety, there is general agreement, articulated first by the National High Blood Pressure Education Program (NHBPEP) Working Group on High Blood Pressure in Pregnancy, and retained in the recommendations of the Task Force convened by the American College of Obstetricians and Gynecologists and that included specialists from several disciplines. The Task Force, further described in Chapter 2 (p. 25), was specifically charged with updating the NHBPEP report, noted that diuretics may be continued through gestation or used in combination with other agents, especially for women deemed likely to have salt-sensitive hypertension. In this sense, the greatest utility for diuretics may be as second or third agents which might act synergistically with other antihypertensives in drug-resistant patients, particularly those with co-morbid chronic kidney disease or heart failure. By contrast, we note that gestational vasodilatation itself often normalizes the modestly elevated blood pressures most often controlled by diuretics, leading many to suggest that these agents be discontinued in most hypertensive patients lacking another indication for their continued use. Indeed, one of several small trials of diuretics in chronically hypertensive gravidas found no greater need for addition of antihypertensive medications in patients withdrawn from thiazides than in those who continued to receive diuretics through pregnancy.
Finally, there is an old literature, albeit anecdotal, reviewed in the first edition of this text, which called attention to cases of severe hyponatremia, hypokalemia, volume depletion, or thrombocytopenia in pregnant women treated with diuretics. Given that most of these are predictable consequences of diuretic pharmacology in susceptible patients, caution is still advised.
Each of the agents available in this drug class inhibits influx of Ca 2+ via voltage-dependent, slow L-type calcium channels. Principal cardiovascular sites of action include smooth muscle cells of arterial resistance vessels, cardiac myocytes, and cells of the cardiac conduction system. These drugs appear to have little effect on the venous circulation, do not act via the endothelium and, despite the presence of target channels in the central nervous system, appear to exert their hemodynamic effects peripherally. Because contraction of vascular myocytes is a direct function of free cytosolic Ca 2+ , the latter depending, in part, on influx via voltage-gated channels, these drugs act as direct vasodilators, antagonizing vasoconstriction, regardless of the original neural or humoral stimulus. Of the prototypical agents, verapamil is most selectively a negative chronotrope and negative inotrope, with significant but lesser effects as a direct vasodilator. Dihydropyridines, of which nifedipine is the prototype, are by contrast vasoselective agents with lesser effects on the heart; diltiazem is intermediate in its tissue selectivity.
Following calcium antagonist administration, blood pressure falls acutely due to decreased peripheral resistance; the response is blunted by reflex activation of the sympathetic nervous and renin–angiotensin systems. While these drugs often trigger dependent edema, probably due to local microvascular effects, they are only rarely the cause of compensatory renal volume retention. This observation suggests that the sustained antihypertensive effect of these agents includes substantial contributions by drug-induced natriuresis, the latter mediated primarily by effects on intrarenal hemodynamics including prominent afferent arteriolar vasodilatation, and perhaps at tubular sites as well. Indeed, increments in sodium excretion have paralleled decrements in blood pressure following administration of isradipine and other dihydropyridines, an effect opposite that observed with other classes of direct vasodilators. This secondary natriuretic effect likely explains results from population studies which demonstrate that calcium antagonist monotherapy is most likely to prove efficacious in those groups that respond preferentially to diuretics rather than to β-blockers, viz., salt-sensitive hypertensive patients. This is also consistent with some observations of limited benefit to combined therapy with calcium antagonists and diuretics; by contrast, either diuretics or calcium-channel blockers are effectively combined with ACE inhibitors.
A controversial report noted an apparent excess incidence of myocardial infarction and death in hypertensive patients with concomitant coronary artery disease who received higher doses of short-acting dihydropyridine calcium antagonists. Here the possibility that a precipitous fall in blood pressure, coupled with excessive sympathetic nervous system response, could lead to myocardial ischemia in patients with underlying coronary disease appears reasonable. Similarly convincing data fail to suggest such a risk for long-acting or sustained-release preparations of dihydropyridine calcium antagonists. While individual trials might be interpreted by some to suggest such risks, others appear to support the safety of these agents, even in high-risk populations.
Calcium-channel blockers have been used to treat both chronic hypertension and women with preeclampsia to prolong gestation, as well as women presenting late in gestation with accelerating hypertension due to preeclampsia. We could identify no adequate studies of calcium-channel blockers early in pregnancy, save for two multicenter retrospective studies whose results argue against significant teratogenic effect. Most investigators have focused on use of nifedipine, with sporadic reports on nicardipine, isradipine, nimodipine, and verapamil. Some have advocated oral or sublingual (immediate-release) nifedipine as a preferred agent in severely hypertensive preeclamptics; however, there is no difference in nifedipine pharmacokinetics or time–effect curves by these two routes of administration. Indeed, the results of small comparative studies suggest that this drug controls blood pressure in such patients with efficacy similar to that of parenteral hydralazine, with a similar spectrum of maternal adverse effects (mostly ascribed to vasodilatation). A more recent report focused on use of intravenous nicardipine (3–9 mg/h) in this setting with good result.
When treating severe hypertension, the data appear to conflict regarding the influence of calcium-channel blockers on uteroplacental blood flow and fetal well-being, especially in comparison with other agents. In two studies, one of which involved maternal hemodynamic monitoring via a pulmonary artery catheter, the authors claimed less fetal distress with nifedipine than with hydralazine. However, in one of these very limited trials, this observation may have been due to a greater hypotensive effect of the hydralazine doses used; several other studies have failed to discern such differences, finding no significant fetal benefit of nifedipine or alternative agents. Of interest too is an animal study where chronically instrumented pregnant sheep demonstrated fetal hypoxia and acidosis following high-dose maternal nifedipine infusion, unexplained by changes in maternal or uteroplacental hemodynamics; we are unaware of any corroborative clinical data for this worrisome observation.
Calcium-channel antagonists also relax uterine smooth muscle and have been used as tocolytics in preterm labor, but there appear to be no data to suggest that their use as antihypertensives compromises the progression of labor or leads to ineffective hemostasis following delivery. There have been concerns regarding use of calcium antagonists for urgent blood pressure control in preeclampsia because of the need to simultaneously infuse magnesium sulfate to prevent eclamptic seizures. Magnesium itself can interfere with calcium-dependent contractile signaling in excitable tissue and in muscle; its combined use with calcium antagonists could conceivably lead to increased risk of neuromuscular blockade or circulatory collapse. Indeed, there are isolated reports of such complications, while others argue against such adverse outcomes with routine therapy, this more reassuring conclusion echoed in a retrospective review comparing 162 women who received both nifedipine and magnesium with 215 who received magnesium without calcium-channel blocking drugs. In the Magpie Trial, in which magnesium was used to prevent preeclampsia ( n =10,110), 30% of women received nifedipine after trial entry and no associated adverse events were reported. For the few women who did have hypotension there was no association with the combination of nifedipine and magnesium sulfate. Finally, in accord with observations in nonpregnant patients with proteinuric renal disease, dihydropyridine calcium channel antagonists may increase proteinuria. This would favor the clinical diagnosis of preeclampsia and possibly create misdiagnosis; indeed, this is supported by results in a recent Cochrane review (see below).
In summary, despite apparently widespread use, these agents are under-studied other than in late pregnancy. There are numerous, albeit size-limited, studies suggesting that they are relatively safe and effective antihypertensive agents in chronically hypertensive or preeclamptic pregnant women and only more mechanistically defined diagnosis would allow us to determine whether they increase preeclampsia. In spite of these reassuring reports, the limited literature leads us to consider them preferred second-line agents.
Commonly prescribed direct vasodilators exert their antihypertensive effects by one of three pharmacologic mechanisms. Diazoxide, like the active metabolite of minoxidil, opens ATP-sensitive K + channels, hyperpolarizing and relaxing arteriolar smooth muscle, with little effect on capacitance vessels. Sodium nitroprusside is a direct nitric oxide (NO) donor; the spontaneously released NO nonselectively relaxes both arteriolar and venular vascular smooth muscle, principally due to activation of soluble guanylyl cyclase and cyclic guanosine 3′,5′-monophosphate (cGMP) accumulation. Organic nitrates act similarly, though with greater effect on capacitance vessels than arterioles. By contrast, hydralazine and related phthalazine vasodilators selectively relax arteriolar smooth muscle by an as yet uncertain mechanism. These agents all have their greatest utility in the rapid control of severe hypertension, or as third-line agents for multidrug control of refractory hypertension.
Hydralazine-induced vasodilatation leads to striking reflex activation of the sympathetic nervous system, increments in plasma renin, and compensatory fluid retention. The sympathetic activation, combined with hydralazine’s lack of effect on epicardial coronary arteries, can precipitate myocardial ischemia or infarction in hypertensive patients with coronary atherosclerosis. These same compensatory mechanisms rapidly attenuate the hypotensive effect, requiring combination therapy with sympatholytic agents and diuretics for long-term blood pressure control. Hydralazine is effective orally or intramuscularly; parenteral administration is used for rapid control of severe hypertension. Adverse effects are mainly those due to excessive vasodilatation or sympathetic activation, such as headache, nausea, flushing, or palpitations. Chronic use can lead to a pyridoxine-responsive polyneuropathy or to a variety of immunologic reactions including a drug-induced lupus syndrome.
Nitroprusside, administered only by continuous intravenous infusion, is easily titrated because it has a nearly immediate onset of action, whose duration of effect is only about 3 minutes. Cardiac output tends to fall during nitroprusside administration in patients with normal myocardial function due to decreased preload, while it increases in those with systolic heart failure due to afterload reduction. Reflex sympathetic activation is the rule. Nitroprusside metabolism releases cyanide, which can reach toxic levels with high infusion rates; cyanide is metabolized to thiocyanate, whose own toxicity usually occurs after 24 to 48 hours of nitroprusside infusion, unless its excretion is delayed due to renal insufficiency.
Use of diazoxide is limited to urgent parenteral control of severe hypertension. Minoxidil is the alternative agent for prolonged use due to intolerable adverse effects when diazoxide was used chronically. Since diazoxide leads to profound activation of the sympathetic and renin–angiotensin systems, its hypotensive effect is enhanced by use of sympatholytic agents and diuretics. An intravenous bolus of diazoxide lowers blood pressure within 30 seconds, with maximum effect in 5 minutes, allowing easy titration by repeated administration. Most side effects are due to excessive vasodilatation, as with hydralazine, though stimulation of ATP-sensitive K + channels in the pancreas can inhibit insulin secretion, leading to striking hyperglycemia.
Hydralazine is the direct vasodilator most often used in pregnant women, either as a second agent for hypertension uncontrolled following methyldopa (or a β-blocker) or, more commonly, as a parenteral agent for control of severe hypertension. Its use is justified not by pharmacologic selectivity but by long clinical experience with tolerable side effects. Several studies of hydralazine or related compounds in preeclamptic women monitored with pulmonary artery catheters have highlighted concerns regarding its safety, including the occurrence of precipitous decreases in cardiac output and blood pressure with oliguria; these complications are predictable with knowledge of its pharmacology in the setting of the primary vasoconstriction and relative volume contraction which is characteristic in these patients. Effects on uteroplacental blood flow are unclear, likely due to variation in the degree of reflex sympathetic activation, though fetal distress may result with precipitous control of maternal pressure. Neonatal thrombocytopenia has been reported following intrauterine hydralazine exposure. Many investigators have suggested that urgent blood pressure control might be better achieved with less fetal risk by use of other agents such as labetalol or nifedipine; as will be discussed in greater detail below, objective outcome data currently fail to support the choice of one agent over another in this setting.
Diazoxide, even when dosed carefully, can lead to excessive hypotension, arrested labor, and hyperglycemia and interpretation of some animal studies suggests that this may compromise uterine blood flow. It is possible, however, that slow infusion or small dose increments may avoid these complications. Indeed, in a well-conducted recent randomized trial comparing 15 mg “miniboluses” of diazoxide with hydralazine, there was quicker blood pressure control observed with diazoxide with no excess of maternal, fetal, or neonatal distress. Finally, there are concerns from animal studies that repeated administration could result in pancreatic islet cell degeneration, an important observation that does not seem to have been addressed by follow-up studies in humans. Currently, this drug seems to have fallen into relative disuse in both pregnant and nonpregnant populations.
Nitroprusside has only been used sporadically in pregnancy, usually in life-threatening refractory hypertension. Adverse effects include those due to excessive vasodilatation, apparently including cardioneurogenic, i.e., paradoxically bradycardic, syncope in volume-depleted preeclamptic women. The risk of fetal cyanide intoxication remains unknown. Given the long experience with hydralazine and alternative utility of calcium-channel blockers or parenteral labetalol, both nitroprusside and diazoxide must be considered agents of last resort.
Modulators of the Renin–Angiotensin–Aldosterone Axis
Angiotensin-converting enzyme inhibitors block ACE, which is a kininase II. This enzyme is responsible for conversion of angiotensin I – cleaved from angiotensinogen by renin – to angiotensin II (AII). Inhibition of ACE, localized to the endothelium and most abundant in the pulmonary microvasculature, predictably leads to decrements in AII and also in levels of aldosterone, whose synthesis in the zona glomerulosa of the adrenal cortex is AII-dependent. These agents lower blood pressure primarily by blocking AII-induced vasoconstriction; their efficacy in patients with low circulating renin and AII serves to reveal contributions of a parallel “tissue” renin–angiotensin axis distributed in the arterial wall. In some hypertensive patients, accumulation of bradykinin, an endothelium-dependent vasodilator, and also an endogenous substrate for ACE, may contribute to the antihypertensive effect. Indeed, adverse effects, including the common ACE inhibitor-induced dry cough, appear to be due, at least in part, to bradykinin. Long-term blood pressure control may be favored by an apparent natriuretic effect of these drugs, partially due to decrements in aldosterone synthesis, but mainly to a resetting of the normal relationship between salt excretion and renal perfusion pressure. It is unclear what long-term benefits may be derived from nonhemodynamic effects of ACE inhibition, as AII is a potent growth factor for a variety of cardiovascular cells.
Antagonists acting at the AT-l subtype of AII receptors (ARBs) exhibit antihypertensive pharmacology virtually identical to that of ACE inhibitors, save for hypersensitivity reactions and those effects due to bradykinin accumulation. Indeed, the fidelity with which these agents recapitulate the beneficial effects of ACE inhibition casts doubt on significant contributions of either bradykinin accumulation or AT-2 receptor activation to the blood pressure-lowering effect of ACE inhibitors in most patients. An emerging literature with relevance to pregnancy has noted an alternative pathway, depending on the ability of ACE-2 to cleave angiotensin I to the Ang 1–7 fragment, which apparently exerts vasodilator activity via the mas receptor; its levels appear to be increased in normal pregnancy and decreased in preeclampsia. Effective drugs modulating this system are not yet available in clinical practice. Aliskiren, a direct renin inhibitor in use since 2007, has been the subject of much study focused on achieving more complete blockade of the renin–angiotensin system with recent disappointments due to unanticipated harm when used in combination with other agents in high-risk patient populations; given existing concerns regarding ACE inhibitors, we do not expect these drugs to be used (intentionally) during pregnancy.
As predicted, ACE inhibitors and ARBs are most efficacious in patients with renin-dependent hypertension, such as those with unilateral renal artery stenosis. Their benefit in diabetic nephropathy, their ability to decrease proteinuria in patients with glomerular diseases, and their apparent “renal protective” effect in other (principally proteinuric) diseases leading to progressive renal insufficiency derive from their effects in the renal microvasculature. Since AII acts selectively to constrict the efferent arteriole (its effect in the afferent arteriole being antagonized by locally synthesized NO), its absence, or antagonism, lowers intraglomerular pressure out of proportion to any effect on systemic arterial pressure. It is important to note that in those cases where preservation of glomerular filtration rate in the face of decreased renal perfusion depends on selective efferent arteriolar vasoconstriction by AII – examples include renal artery stenosis, decreased effective arterial volume, and congestive heart failure – ACE inhibition may lead not only to exaggerated hypotension, but also to acute renal failure. In addition to having “compelling indications” for the treatment of hypertension in patients with diabetes and proteinuric renal disease, ACE inhibitors and ARBs are similarly indicated following myocardial infarction and in all stages of congestive heart failure.
Finally, there has been a recent resurgence in the use of aldosterone antagonists such as spironolactone and eplerenone, with clinical trial data demonstrating their specific benefit in the postmyocardial infarction, congestive heart failure, and refractory hypertension populations. The aldosterone antagonists are generally avoided in pregnancy due to concerns regarding possible antiandrogenic effects in the fetus.
All circulating elements of the renin–angiotensin system are increased in normal human pregnancy. Preeclampsia, where AII levels are lower than in normal gestation, is characterized by the appearance of autoantibodies which activate the AT-1 receptor, which, itself, appears to be upregulated. Thus, both ARBs and ACE inhibitors might have seemed attractive antihypertensive agents for treating hypertensive pregnant women, indeed, a conclusion apparently supported by one small case series. Concerns regarding use of these drugs might have been anticipated, however, from excess fetal wastage in animal studies. Indeed, use of drugs from these classes in human pregnancy has been associated with frequent reports of a specific fetopathy – renal dysgenesis and calvarial hypoplasia; oligohydramnios – likely a result of fetal oliguria; fetal-growth restriction, and neonatal anuric renal failure leading to death. For these reasons, these drugs are contraindicated following midpregnancy.
Then, in 2006, a study which linked (Tennessee) Medicaid prescription records with maternal and infant medical and vital records identified a 2.7-fold increase in congenital malformations, due entirely to increases in cardiovascular and central nervous system malformations following ACE-inhibitor first-trimester exposures. These observations have led to an avoidance of their use in pregnancy as well as in women of childbearing potential. Subsequently, several much larger studies have failed to confirm concerns regarding first-trimester exposure, some suggesting that any increased risks of malformation may be related to maternal hypertension, rather than its treatment. Likewise, secondary analysis of 208 pregnancies which occurred during a randomized clinical trial of candesartan in type 1 diabetics suggested no excess risk due to first-trimester ARB exposure. So, while it seems prudent to discontinue their use when pregnancy is first confirmed, and many may argue against their use in women with prepregnancy essential hypertension, due to the high rates of unintended pregnancy and of delayed or limited antenatal care, these risks must be weighed against the significant potential benefits of their use in women of childbearing potential with compelling indications, such as those with underlying diabetes or proteinuric renal disease, and in patients with difficult to control hypertension, which might recur following drug withdrawal.