This Article Abstract Full Text (PDF) All Versions of this Article: 26/6/775    most recent Final Manuscript Author Manuscript Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lindsay, J. R. Articles by Nieman, L. K. Search for Related Content PubMed PubMed Citation Articles by Lindsay, J. R. Articles by Nieman, L. K. Pubmed/NCBI databases Medline Plus Health Information Adrenal Gland Disorders High Risk Pregnancy

The Hypothalamic-Pituitary-Adrenal Axis in Pregnancy: Challenges in Disease Detection and Treatment

Reproductive Biology and Medicine Branch, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland 20892-1109

Correspondence: Address all correspondence and requests for reprints to: L. K. Nieman, M.D., Reproductive Biology and Medicine Branch, National Institute of Child Health and Human Development, National Institutes of Health, Building 10, Clinical Research Center, Room 1–3140, 10 Center Drive, Bethesda, Maryland 20892-1109. E-mail: niemanl{at}mail.nih.gov

	Abstract

Top Abstract I. Introduction II. HPA Axis Physiology... III. Cushing’s Syndrome... IV. Adrenal Insufficiency in... V. Summary References

 
Pregnancy dramatically affects the hypothalamic-pituitary-adrenal axis leading to increased circulating cortisol and ACTH levels during gestation, reaching values in the range seen in Cushing’s syndrome (CS). The cause(s) of increased ACTH may include placental synthesis and release of biologically active CRH and ACTH, pituitary desensitization to cortisol feedback, or enhanced pituitary responses to corticotropin-releasing factors. In this context, challenges in diagnosis and management of disorders of the hypothalamic-pituitary-adrenal axis in pregnancy are discussed.

CS in pregnancy is uncommon and is associated with fetal morbidity and mortality. The diagnosis may be missed because of overlapping clinical and biochemical features in pregnancy. The proportion of patients with primary adrenal causes of CS is increased in pregnancy. CRH stimulation testing and inferior petrosal sinus sampling can identify patients with Cushing’s disease. Surgery is a safe option for treatment in the second trimester; otherwise medical therapy may be used.

Women with known adrenal insufficiency that is appropriately treated can expect to have uneventful pregnancies. Whereas a fetal/placental source of cortisol may mitigate crisis during gestation, unrecognized adrenal insufficiency may lead to maternal or fetal demise either during gestation or in the puerperium. Appropriate treatment and management of labor are reviewed.

I. Introduction

II. HPA Axis Physiology in Normal Pregnancy

A. Circulating hormone levels and their origins

B. Regulation of the HPA axis

III. Cushing’s Syndrome (CS) in Pregnancy

A. Frequency

B. Maternal and fetal morbidity and mortality

C. Causes

D. Screening and diagnosis

E. Treatment of CS

IV. Adrenal Insufficiency in Pregnancy

A. Overview

B. Frequency

C. Causes

D. Maternal and fetal morbidity and mortality

E. Diagnosis

F. Treatment

V. Summary

	I. Introduction

Top Abstract I. Introduction II. HPA Axis Physiology... III. Cushing’s Syndrome... IV. Adrenal Insufficiency in... V. Summary References

 
NORMAL HUMAN GESTATION dramatically affects the maternal hypothalamic-pituitary-adrenal (HPA) axis. Increasing placental production of estrogen stimulates hepatic corticosteroid-binding globulin (CBG) production, thus stimulating cortisol production and increasing circulating levels of bound cortisol. However, both circulating and urinary free cortisol levels also increase steadily during gestation, reaching values that are in the range seen in Cushing’s syndrome (CS). Plasma ACTH levels parallel the rise in cortisol. The cause(s) of this increase in ACTH is not clear, but may include placental synthesis and release of biologically active CRH and ACTH, pituitary desensitization to cortisol feedback, or enhanced pituitary responses to corticotropin-releasing factors such as vasopressin and CRH. These possibilities will be discussed in the context of the decreased suppression of the HPA axis by exogenous glucocorticoids, blunted diurnal rhythm of cortisol and blunted response of ACTH to exogenous CRH, a normal response to stressors of venipuncture and labor, and the enhanced cortisol response to exogenous ACTH.

CS occurs rarely in pregnancy, with fewer than 150 cases in the world literature. When untreated, fetal mortality is nearly 20%; treatment reduces, but does not abolish, this adverse outcome. Maternal morbidity includes hypertension, hyperglycemia, and eclampsia.

The clinical diagnosis may be missed because of the overlapping features of weight gain, hypertension, fatigue, hyperglycemia, and emotional changes that occur in pregnancy. The biochemical diagnosis is difficult to establish because of the normal hypercortisolism of pregnancy. The proportion of patients with primary adrenal causes of CS is increased in pregnancy. This poses diagnostic problems because the increased ACTH levels of normal pregnancy are not suppressed by the hypercortisolism; thus, in contrast to nonpregnant patients, an undetectable ACTH level cannot be used as a criterion for this diagnosis. We and others have used ovine CRH and inferior petrosal sinus sampling (IPSS) to identify patients with Cushing’s disease (CD). Surgery is a safe option for treatment in the second trimester; otherwise, medical therapy may be used, which must be chosen carefully to avoid adverse maternal and fetal effects.

Women with known adrenal insufficiency (AI) that is appropriately treated can expect to have uneventful pregnancies of normal length without fetal compromise. However, if unrecognized, AI often leads to maternal or fetal demise either during gestation or in the puerperium. Emesis, fatigue, and altered food preferences of pregnancy contribute to a lack of clinical recognition of AI. Excessive emesis, hypoglycemia, and hyponatremia are important clues to its presence. Women are at increased risk for adrenal crisis postpartum, implying a potential contribution of a fetal/placental source of cortisol to prevention of crisis during gestation. Appropriate treatment, including increased sensitivity to mineralocorticoid replacement and management of labor, is reviewed.

	II. HPA Axis Physiology in Normal Pregnancy

Top Abstract I. Introduction II. HPA Axis Physiology... III. Cushing’s Syndrome... IV. Adrenal Insufficiency in... V. Summary References

 
A. Circulating hormone levels and their origins
1. Circulating and urinary glucocorticoids and CBG.
Pregnancy is associated with a state of increased HPA axis function (1, 2) as shown by elevations in urine free cortisol (UFC), plasma 17-hydroxysteroids [17-hydroxycorticosteroids (17-OHCS)], total and free plasma cortisol, and CBG values during pregnancy (3, 4, 5, 6, 7, 8, 9, 10). It is assumed that increased circulating estrogens from the placenta stimulate hepatic production of CBG, which remains elevated until at least the 12th postpartum day (10). Presumably, free cortisol concentrations drop transiently, as CBG increases, reducing negative feedback and increasing ACTH stimulation so that cortisol production increases to maintain a normal free cortisol level. However, as described below, free cortisol levels also are elevated, particularly in the second and third trimesters (5, 7, 8, 10, 11).

Total and free plasma cortisol concentrations rise in parallel across gestation (11, 12), with plasma cortisol reported as 2- to 3-fold elevated compared with nonpregnant controls (5, 13). The increases in plasma cortisol are noted as early as the 11th week of gestation (12). In one series there was an almost 5-fold increment between the first trimester and delivery (Fig. 1) (3). As shown by Mukherjee and Swyer (14), there is a wide range of normal variation in the third trimester plasma cortisol from 16.3–55 µg/dl (450–1518 nmol/liter). The circadian rhythm of cortisol is preserved, although it may be partly blunted (3, 4, 5, 10, 14, 15).

View larger version (25K): [in this window] [in a new window]   FIG. 1. Serial increases in serum cortisol () and ACTH (•) during pregnancy in normal controls throughout pregnancy. This graph was modified from the series of five normal pregnant women from the series by Carr et al. (3 ) in 1981. The bars adjacent to the right axis are summary data derived from a recent series (173 ) to denote the range of serum cortisol observed in CS in pregnancy (, median and range; n = 52); ACTH values for CD (, median and range; n = 18) and adrenal CS (, median and range; n = 17). [Reprinted and modified with permission from B. R. Carr et al.: Am J Obstet Gynecol 139:416–422, 1981 (3 ). © Elsevier.]

Urine cortisol and its metabolites also increase in parallel with cortisol throughout gestation. In 1953, Gemzell (1) demonstrated that 17-OHCS levels were elevated 4-fold in pregnancy; this increase is mainly due to increased cortisol (9). Mean 24-h UFC is elevated at least 180% during gestation compared with nonpregnant levels (4). The aforementioned elevations in cortisol and its metabolites are consistent with a hypothesis that the maternal adrenals and the fetal-placental unit, in addition to estrogen-stimulated CBG elevations, all contribute to hypercortisolism in pregnancy (10, 18).

One explanation for the elevation in free cortisol is that pregnancy may represent a state refractory to cortisol action (7). Allolio et al. (17) demonstrated significant correlations between serum progesterone and salivary cortisol during late pregnancy. They suggested that elevated free plasma cortisol levels may result from antiglucocorticoid effects of elevated progesterone concentrations in pregnancy (17). Other theories include an altered set point to the negative feedback mechanism controlling ACTH secretion (4, 5). An alternate hypothesis is that placental ACTH represents an autonomous continuous source that is superimposed upon normal pituitary ACTH production (4).

The fetus is protected in early gestation from the effects of maternal hypercortisolism by placental 11-ß hydroxysteroid dehydrogenase 2 (11ß-HSD 2), which converts active glucocorticoids, cortisol, and corticosterone to their inactive 11-keto metabolites (19, 20). The enzyme is located in the syncytial trophoblastic cells. The capacity of placental 11ß-HSD 2 is sufficient to ensure that fetal cortisol levels are much lower than maternal levels (19). Whereas fetal cortisol concentrations are affected by 11ß-HSD 2 enzyme activity, approximately three fourths of fetal cortisol originates from fetal adrenal gland production in term infants (21, 22, 23). Dexamethasone, in contrast, is a poor substrate for 11ß-HSD 2 and can cross the placenta readily (20). In nonpregnant subjects conversion of cortisol to cortisone predominates; however, in late gestation there is a reversal of this reaction in the uterus, which favors production of the active hormone (24). These effects may favor late fetal development, including lung maturation (24). Altered 11ß-HSD 2 activity has been implicated in fetal programming, and this role has been a focus of research in the pathogenesis of adult disease, including the metabolic syndrome (20, 25, 26). Impaired activity of the enzyme and possible excessive fetal glucocorticoid exposure are observed in intrauterine growth retardation and preeclampsia, which are commonly associated with preterm infants (22).

2. Plasma ACTH.
ACTH is a 39-amino acid peptide normally derived, in the pituitary corticotropes, from successive cleavage of a larger precursor peptide, proopiomelanocortin (POMC). This reaction gives rise to a series of related peptides including ß-endorphin and -MSH (27). Parallel rises in plasma ACTH, ß-endorphin, and ß-lipotropin are observed through pregnancy, consistent with their origin from POMC (28). A placental source of ACTH was postulated for many years before the demonstration of ACTH and immunoreactive ß-endorphin and lipotropin within the placenta in the 1970s (29, 30, 31). Demura et al. (32) later showed the presence of equimolar concentrations of ACTH and ß-endorphin in trophoblastic tissues consistent with their origin from a common precursor. Short mRNA related to the gene encoding POMC was subsequently detected in human placenta (33), and trophoblastic cells synthesize POMC-derived peptides in vitro (34, 35). Whether POMC itself has a specific action in pregnancy is unknown (3).

In one series, plasma POMC was undetectable in nonpregnant women but became detectable by the third month and then steadily increased toward midgestation (36). Plasma POMC correlated with plasma CRH but showed no diurnal variation, was not suppressed by glucocorticoid administration, and did not correlate with plasma ACTH or cortisol (36).

Plasma ACTH levels rise through pregnancy, reaching maximal levels during labor and delivery (Fig. 1); in one study, levels increased almost 3-fold from the end of the first to the third trimester (23–59 pg/ml measured by RIA; 5–13 pmol/liter) (3). Compared with healthy nongravid women, basal plasma ACTH levels in pregnancy have been varyingly reported as low (3, 14) or high (37) using RIA. Diurnal patterns of plasma ACTH and ß-endorphin concentrations parallel each other and are preserved throughout pregnancy, and circulating cortisol and ACTH levels are strongly correlated (14, 15, 17). The elevated ACTH levels observed in late pregnancy suggest that a source of ACTH exists that is not subject to normal feedback control (3). Placentally derived ACTH may be a significant contributor to hypercortisolism in pregnancy. In vitro stimulation of ACTH production from superfused human placenta was first described in 1986 (38), and release of bioactive ACTH has been demonstrated in early and late gestation (39). Petraglia et al. (40) demonstrated that CRH modulates release of placental ACTH.

3. Plasma CRH and CRH-binding protein (CRH-BP).
CRH was isolated from human placenta in 1988 by Sasaki et al. (41) and is identical to hypothalamic CRH in structure, immunoreactivity, bioactivity, and transcriptional sites (42). It has since been demonstrated in extracts of placenta and in fetal plasma and amniotic fluid (43, 44, 45, 46). Placental CRH mRNA was identified between wk 7 and 40 of gestation; it increased more than 20-fold in the 5 wk preceding parturition in parallel with rising plasma CRH concentrations (47, 48).

Plasma CRH levels rise exponentially by 1000-fold as gestation progresses (49), beginning around 8 wk gestation (50, 51). At the 35th week there is a sharp increase to a peak of 4000 pg/ml at 40 wk gestation (5, 52) (Fig. 2), with normalization to nonpregnant values within 24 h of delivery (53, 54, 55, 56). CRH levels are significantly lower (20-fold) in umbilical cord plasma than in the maternal circulation and are close to the nonpregnant reference range (45). These data suggest that the placenta is the source of elevated circulating CRH during gestation (44, 45, 57, 58).

View larger version (16K): [in this window] [in a new window]   FIG. 2. Mean plasma CRH concentrations in seven women throughout pregnancy. Sequential samples were obtained at 1- to 2-wk intervals beginning at 12 wk gestation. [Reprinted with permission from R. S. Goland et al.: Am J Obstet Gynecol 159:884–890, 1988 (52 ). © Elsevier.]

Systemic maternal effects of elevated CRH in pregnancy are thought to be limited due to binding of free bioactive CRH to CRH-BP, a 322-amino acid glycoprotein (63). Whereas CRH-BP has been demonstrated primarily in the brain in mammals, in the human it is also present in the liver and placenta (64). Human CRH-BP binds to human but not ovine CRH (65). Circulating CRH-BP levels in early and midgestation are similar to nonpregnant levels, suggesting that, in contrast to CBG, CRH-BP is not stimulated by elevated estrogen levels in pregnancy (66). Between wk 34 and 35 of gestation, CRH-BP concentrations fall by around 60%, leading to elevations in free CRH (66) (Fig. 3). When given in vitro with CRH, at typical gestational concentrations, CRH-BP reduces the amount of ACTH released by the placenta but not the corticotrope, thereby potentially maintaining the maternal stress response during the third trimester (67).

View larger version (20K): [in this window] [in a new window]   FIG. 3. Comparison of the molar concentrations of CRH (]) and CRH-BP () in maternal plasma during the final 180 d of gestation in pregnancies ending in spontaneous term labor (37–42 wk gestation). Each point represents the mean (± SEM) of samples grouped by 10-d intervals calculated retrospectively from the day of delivery (mean of 59 samples at each time point). CRH and CRH-BP concentrations are significantly different (P < 0.002) at all points except at the intersection of the two curves, 20 d before delivery. [Reprinted with permission from M. McLean et al.: Nat Med 1:460–463, 1995 (71 ). © Nature Publishing Group (http://www.nature.com/).]

There is no correlation between plasma CRH and ACTH or total or free cortisol, suggesting either that placental CRH is not the sole regulator of the maternal pituitary-adrenal axis, or that regulation occurs in a paracrine fashion within the placenta (17, 45). These findings may be consistent with the concept that HPA axis function remains intact in normal pregnancy despite observations consistent with desensitization of maternal pituitary corticotrophs. The primary stimulus for the increase in activity of the HPA axis in the third trimester appears to be placental CRH.

Although CRH is a significant regulator of maternal and fetal HPA axes in pregnancy, it also plays a more general role in female reproduction (Table 1) (53). There is evidence that CRH facilitates decidualization, implantation, and ovarian function (53). Locally produced embryonic and endometrial CRH impedes rejection during implantation by inducing apoptosis in activated leukocytes carrying Fas ligand, thereby protecting the fetus from the maternal immune system (54, 70). Maternal CRH acts as a biological clock that determines the length of gestation (55, 71), and premature or accelerated activation of the placental CRH system may be associated with earlier onset of labor and delivery (72). Placental CRH may also be a marker of antepartum risk for preterm delivery (72). CRH is generally higher in women with spontaneous labor compared with those requiring induction, consistent with a central role in the onset of parturition (73, 74).

Plasma progesterone concentrations increase progressively throughout pregnancy to between 100 and 300 ng/ml in parallel with increases in plasma estradiol levels (87, 90). Acting as a mineralocorticoid receptor antagonist, progesterone reduces sodium reabsorption; it also contributes to reduced systemic vascular resistance, causing smooth muscle relaxation (91, 92). Conversely, increased estradiol and estriol levels in pregnancy are associated with elevated renin concentrations and up-regulation of the RAS (87, 93).

Against this backdrop of normal physiological changes occurring in pregnancy, elevations in mineralocorticoid levels appear necessary to maintain normal sodium balance and volume homeostasis. Although the RAS is markedly stimulated during pregnancy, both renin and aldosterone respond physiologically, albeit at an altered set point. Blockade of the mineralocorticoid receptor in animal models demonstrates that aldosterone and the RAS are of critical importance to fetal growth and development (94).

a. RAS.
The RAS comprises a cascade of events that proceeds from renin-mediated cleavage of the decapeptide angiotensinogen to angiotensin I, which is rate limiting. Angiotensin I can then be cleaved by angiotensin-converting enzyme to the octapeptide angiotensin II, which promotes aldosterone synthesis and secretion. Whereas renin is produced predominantly in the kidneys, the RAS is up-regulated during pregnancy, and the fetal-placenta unit is an important additional site of RAS activity (95, 96).

Plasma renin activity (PRA) increases early in the first trimester of normal pregnancy, reaching values almost 3- to 7-fold greater than the normal range by the third trimester (87, 96, 97) (Fig. 4). Approximately 50% of this increase is attributable to increased plasma renin substrate, and the changes observed in pregnancy are independent of sodium or potassium (87). A positive correlation exists between plasma renin substrate and plasma estriol and estradiol, supporting the view that increases are mediated by elevated estrogens during pregnancy (87, 93). Increased concentrations of renin are demonstrated within uterus, placenta, and amniotic fluid (95, 98, 99, 100, 101). The ovary produces renin and prorenin (102, 103). However, other factors, including changes in salt intake, blood pressure, effects of progesterone, increased renin substrate concentration, and the fetoplacental unit, influence the plasma renin concentration (87, 95, 104, 105). The response of plasma renin to posture or saline loading in pregnancy is similar in direction and magnitude compared with nonpregnant subjects, consistent with intact physiological regulation (106). However, urinary excretion of sodium before and after saline infusion is lower in pregnancy, in keeping with an increased sodium requirement for homeostasis (106).

View larger version (20K): [in this window] [in a new window]   FIG. 4. Sequential changes in PRA (•) and in PRA normalized to the postpartum (PP) substrate values () (mean ± SE) throughout pregnancy (*, P < 0.05; ***, P < 0.001). [Reprinted with permission from M. Wilson et al.: Am J Med 68:97–104, 1980 (87 ). © Elsevier.]

b. Aldosterone regulation.
In normal pregnancy, plasma and urinary aldosterone increase, in association with enlargement of the zona fasiculata (18, 86, 112). Plasma aldosterone concentrations are elevated 5- to 7-fold during the first trimester (18) and continue to increase until the 38th week of gestation when 10 to 20-fold elevations are reached (18, 86, 113). In contrast to desoxycorticosterone (DOC) and cortisol, aldosterone is not bound substantially to plasma proteins (113). There exists a disproportionate rise in plasma aldosterone concentrations compared with the magnitude of renin secretion, suggesting a possible increase of some other unknown pregnancy-associated factor that contributes to plasma aldosterone concentrations in pregnancy (97, 114).

The diurnal rhythm of plasma aldosterone concentrations is preserved during pregnancy (115). Aldosterone responses to salt loading, posture, diuretics, volume depletion, and administration of mineralocorticoid suggest that the RAS is under tight physiological control (85, 116, 117). Furthermore, serum potassium levels remain constant in pregnancy despite increased plasma aldosterone, perhaps because of the mineralocorticoid antagonist effects of progesterone (96). Evidence in favor of a hypothesis that elevations in aldosterone levels are not excessive includes the observation of natriuresis after administration of an aldosterone inhibitor (118). Significantly, aldosterone levels are reduced in pregnancy-associated hypertension (96). Women with pregnancy-induced hypertension have a 2-fold greater increase in plasma aldosterone-plasma renin ratio compared with normal pregnant women (97), whereas in primary hyperaldosteronism, plasma aldosterone is increased in association with reduced renin (119).

c. Other mineralocorticoids.
Corticosterone, deoxycortisol, and cortisone parallel the 2- to 3-fold rise seen in cortisol during gestation (18). Plasma DOC, a potent mineralocorticoid, increases from 2-fold normal during the first trimester to peak levels of 60–100 ng/100 ml in the third trimester (120, 121, 122) and may contribute to sodium retention in pregnancy. Early studies showed increased responsiveness of urinary measures of DOC to ACTH stimulation during the first and second trimesters compared with nonpregnant controls; these observations suggest that DOC represents a substantial nonsuppressible source of mineralocorticoid that is relatively independent of the RAS (123). In the third trimester, whereas total DOC levels are unchanged after ACTH stimulation, free DOC levels are elevated, possibly due to displacement of free DOC from CBG-binding sites (121, 123, 124). Similarly, during the third trimester, DOC levels are not suppressed by salt intake or dexamethasone (121), lending credence to the hypothesis that DOC may promote sodium retention (107). The fetoplacental unit probably contributes to circulating DOC levels, as increased concentrations of DOC have been demonstrated in mixed cord blood (125). Nolten et al. (120) have speculated that placental progesterone might be converted to DOC by the fetal adrenals. Further support for this hypothesis is provided by the observation that DOC sulfate has been found in high concentrations in umbilical cord blood.

B. Regulation of the HPA axis
1. ACTH stimulation of cortisol secretion.
It was known as early as 1955 that, during late pregnancy, the adrenal glands have increased responsiveness to ACTH compared with nongravid women (126, 127, 128). Subsequent studies measuring urinary 17-oxogenic steroids or 11-hydroxycorticosteroids or plasma cortisol after im tetracosactin, corticotropin gel, or synthetic ACTH demonstrated 1- to 2-fold elevations in normal pregnancy compared with nonpregnant subjects (129). There was speculation that the apparent increased responsiveness might be due, in part, to delayed clearance of cortisol (126), because of a delayed peak response at 120 min. There also is a greater absolute rise in the unbound cortisol response, which increases as pregnancy advances (7).

A recent study examined aldosterone and cortisol responses to low-dose ACTH stimulation in normal pregnancy and preeclampsia (2, 3, 5, and 7 µg/h for 80 min), demonstrating a similar pattern of enhanced responsiveness of cortisol release in the third trimester of pregnancy compared with nonpregnant women (130). The mean maximum cortisol response in pregnancy was 34.9 µg/dl (963 nmol/liter) compared with 18.4 µg/dl (507 nmol/liter) in a group of nonpregnant controls, despite administration of lower doses (1, 2, 3, and 5 µg/h) to the control women to account for their lower relative plasma volume (130).

McKenna et al. (131) examined responses of six healthy women to 1 µg ACTH during the 24th to 34th weeks of gestation. The mean peak cortisol response was 44 µg/dl (1215 nmol/liter) (99% CI, 33.2–55.6 µg/dl; 917-1535 nmol/liter) and was attained at a mean of 27 min after cortrosyn.

2. Stimulation of ACTH secretion by CRH and vasopressin.
Exogenous human CRH, 1 µg/kg, failed to increase plasma cortisol or ACTH in seven pregnant women 1 wk before their expected delivery date (132). Although two women experienced transient flushing, no other maternal or fetal side effects were noted (132). In contrast, in the same women studied at 4–5 wk postpartum, there was a prompt ACTH response to administered CRH. Other investigators using a higher dose (2 µg/kg) during third trimester pregnancies demonstrated ACTH and cortisol increments that were similar to those of nonpregnant women (133). Whereas diminished CRH responsiveness may be due to effects of CRH-BP, in vitro studies of pituitary columns continuously perfused with CRH demonstrated initially brisk responses of ß-endorphin secretion, which gradually declined to baseline after a period of hours (134). These observations are consistent with a hypothesis proposed by Schulte et al. (132) that blunting of the CRH response may arise due to high endogenous cortisol concentrations with desensitization of the pituitary corticotrophs.

As noted earlier, plasma CRH levels are relatively nonvariant during the third trimester (17), suggesting that circadian and pulsatile secretion of ACTH from the corticotrope may be driven by another secretagogue (15). Arginine vasopressin has been postulated to fill this role, as it is secreted in a pulsatile fashion with a circadian increase in amplitude (15). Goland et al. (135) suggested that chronic placental CRH stimulation of the pituitary-adrenal axis during pregnancy leads to enhanced responsiveness to vasopressin and down-regulation of the response to exogenous CRH.

3. The stress response.
An individual’s ability to mount an appropriate stress response during the antenatal period is preserved in normal pregnancy (136). ACTH and cortisol levels are subsequently increased during the stress of labor (see below).

4. Suppression of the axis by glucocorticoids.
The HPA axis response to exogenous glucocorticoids during pregnancy is blunted. A range of reported dosing protocols and end points make interpretation of dexamethasone suppression tests more difficult in normal pregnancy. Early studies of human pregnancies showed suppression of urinary 17-OHCS of approximately 55% after 4–6 mg dexamethasone (129). Women in the third trimester treated with high-dose glucocorticoids (dexamethasone, 24 mg) before delivery exhibit suppressed ACTH levels within the first 24 h postpartum compared with untreated controls (137), but these effects are short lived (138). After iv administration of 4 mg dexamethasone to women in the second trimester with congenital adrenal hyperplasia, approximately 60% suppression of plasma cortisol was noted within 2 h that continued for up to 8 h. Up to 90% suppression was achieved after 12 mg dexamethasone given in a divided dose (139).

Odagiri et al. (13) demonstrated a 40% vs. 87% suppression of plasma cortisol and similar effects on UFC after 1 mg dexamethasone in normal second- to third-trimester pregnancy compared with nongravid controls (Fig. 5). Whereas the majority of nonpregnant women showed a consistent suppression of plasma cortisol, there was a wide range of variation in responses in pregnant women. Advancing gestation was associated with increasing loss of suppressibility after 1 mg dexamethasone (13). This decrease in the suppressive action of dexamethasone has been attributed to CBG effects on cortisol, tissue refractoriness to glucocorticoids, or resetting of the maternal HPA feedback mechanism (13). Other theories posit that antiglucocorticoid effects of progesterone might contribute to tissue resistance (13, 140). Other confounding factors, such as extrapituitary sources of ACTH and CRH, probably also contribute. Although pregnancy may alter the absorption of dexamethasone, there are contradictory reports examining its bioavailability. In one series, bioavailability via the oral route was 72% of the im route (141). In another series, the bioavailability of an 8-mg oral dose was similar to 6-mg im dosing (142).

View larger version (23K): [in this window] [in a new window]   FIG. 5. Change in plasma cortisol before and after the administration of 1 mg of dexamethasone (Dex.) in pregnant women (•). Blood was drawn at 0800 h. A single dose of dexamethasone was administered orally at 2300 h, and blood was drawn at 0800 h on the following day. [Reprinted from E. Odagiri et al.: Endocrinol Jpn 35:685–690, 1988 (13 ). © Endocrine Society of Japan.]

In the immediate postpartum period, plasma CRH, ACTH, and cortisol levels fall rapidly toward the nonpregnant range, consistent with their biological half-lives (145). Both CRH and ACTH normalize within 2 h from delivery whereas normalization of plasma cortisol levels is more protracted (58). In one series, mean postpartum 24-h plasma cortisol levels were 5.4 µg/dl (149 nmol/liter) compared with the second (18.8 µg/dl; 518 nmol/liter) and third trimesters (20.3 µg/dl; 560 nmol/liter) (4). Diurnal patterns of ACTH are present in the postpartum period (14, 15, 17).

In the immediate postpartum period 82% of women in one series did not have normal cortisol suppression after 1 mg dexamethasone (146). This abnormality may persist for up to 2–3 wk in a significant proportion of women (147). Owens et al. (147) observed normal responses to dexamethasone by the fifth postpartum week.

	III. Cushing’s Syndrome (CS) in Pregnancy

Top Abstract I. Introduction II. HPA Axis Physiology... III. Cushing’s Syndrome... IV. Adrenal Insufficiency in... V. Summary References

 
A. Frequency
CS is rarely associated with pregnancy, probably because hypercortisolism prevents normal follicular development and ovulation. The first description of CS occurring in pregnancy was reported by Hunt and McConahey (148) in 1953. Since then, at least 136 pregnancies have been reported in 122 subjects as individual cases and small case series (149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171). Multiple pregnancies occurred in about 10% of the patients (151, 172). The mean gestational age at diagnosis is approximately 18 wk (173).

B. Maternal and fetal morbidity and mortality
CS is associated with significant maternal morbidity and mortality in approximately 70% of cases. The most common complications in pregnancy are hypertension and diabetes or impaired glucose tolerance (158, 164, 174). In smaller numbers of cases, pregnancies were associated with poor wound healing, osteoporosis, fracture, severe psychiatric complications, maternal cardiac failure, and death (154, 156, 175, 176) (Table 3). Maternal death is rare: one death was reported in the month after delivery as a result of cerebrovascular disease and disseminated intravascular coagulation caused by pheochromocytoma (177). Another woman died due to complications from adrenalectomy and C-section (160).

C. Causes
The causes of CS can be broadly divided into excessive ACTH secretion by a corticotrope or ectopic tumor or autonomous adrenal hypersecretion of cortisol that is independent of ACTH (Table 4). Adrenal adenomas underlie a disproportionately high proportion of CS cases, accounting for approximately 40–50% of cases in pregnancy, compared with about 15% in nonpregnant women (149, 159). Conversely, CD appears to be less common in pregnancy, with rates of 58–70% in the general population compared with 33% in 122 pregnant women (149, 159, 179). Ectopic ACTH secretion has been reported to cause CS in four patients, two of whom had a diagnosis of pheochromocytoma (174, 177). Pheochromocytoma also was associated with one case of apparent ACTH-independent hypercortisolism in pregnancy (180). There was at least one case of CS where remission was observed during pregnancy (181). The increased incidence of adrenal CS in pregnancy is not understood. It is possible that women with CD are less ovulatory than those with primary adrenal disease, perhaps because they are more hyperandrogenic (182). Most patients with ectopic ACTH secretion have severe hypercortisolism and amenorrhea, which probably accounts for the reduced prevalence of this condition in pregnancy (158, 174, 177).

2. Screening tests.
In nongravid women, screening tests for CS establish enhanced cortisol production or a deranged diurnal rhythm, or document blunted suppression of cortisol after dexamethasone suppression. The normal gestational changes in the HPA axis alter these parameters and complicate the screening process for CS (3, 13, 133) (Fig. 1). As reviewed above, these changes include estrogen-dependent increases in CBG, increases in plasma cortisol and ACTH, and a 2- to 3-fold increase in plasma free cortisol and UFC (3, 133).

The mean morning plasma cortisol level of 37 µg/dl in pregnant women with CS is similar to the range observed by Carr and Simpson (3) in normal pregnancy (Fig. 1). Thus, as in the nonpregnant individual, morning plasma cortisol concentrations generally do not establish the diagnosis of CS.

The nocturnal nadir of plasma cortisol is lost in CS but is preserved in pregnancy, albeit with a higher absolute value (3, 4, 5, 176, 183, 184). An elevated midnight or evening plasma cortisol has helped to confirm hypercortisolism in some pregnant women (153, 161, 164, 179). However, no studies have developed a diagnostic threshold for interpretation of the test in pregnant patients. Similarly, salivary cortisol levels reflect serum levels and are elevated in patients with CS (185). However, there is only one case report that documents the potential utility of this noninvasive measure in pregnancy (185).

In nonpregnant women, UFC increases above 4-fold normal are virtually diagnostic of CS. Whereas UFC excretion is normal in the first trimester, it increases up to three times the upper limit of normal during the second and third trimesters. There is a mean 8-fold increase of UFC in pregnant CS patients (range, 2- to 22-fold) (173). This overlap of UFC values in pregnant women with and without CS suggests that only UFC values in the second and third trimester greater than three times the upper limit of normal can be taken to indicate CS (186). However, most studies characterized relatively few pregnant women using measurement of UFC by RIA. The current "gold standard" techniques for UFC measurement are structural assays such as mass spectroscopy, which have lower normative ranges than do antibody-based assays. Thus, it would be very helpful to have additional information on normative data using these modern methodologies.

As discussed earlier, suppression of both plasma and urinary free cortisol by dexamethasone is blunted in pregnancy (4, 29). Thus, the 1-mg dexamethasone suppression test has more limited utility in pregnancy than in the general population because of increased risk for false-positive results.

In summary, standard screening is likely to yield a higher proportion of false-positive diagnoses unless pregnancy-specific cutoff points are developed for UFC and the 1-mg dexamethasone suppression test. Midnight plasma or salivary cortisol may be better, but require further study.

3. Tests for the differential diagnosis.
Hypercortisolism, regardless of the cause, inhibits ACTH secretion by normal corticotropes. As a result, plasma ACTH levels are suppressed in nonpregnant patients with autonomous adrenal disorders and are inappropriately normal or increased in those with tumoral ACTH production. In such patients, a two-site immunoradiometric assay reliably discriminates low (<10 pg/ml; 2.2 pmol/liter) or suppressed (< 5 pg/ml; 1.1 pmol/liter) ACTH levels (187) to identify ACTH-independent primary adrenal causes of CS. In that setting no further biochemical testing is needed, and imaging of the adrenal glands will localize the abnormality to a unilateral adrenal adenoma or carcinoma or bilateral adrenal disorders.

However, pregnant patients with adrenal causes of CS do not consistently have suppressed plasma ACTH values, probably reflecting effects of placental CRH that is not suppressed by hypercortisolism (see above). As a result, the recommended diagnostic ACTH thresholds for adrenal CS in the general population are not valid in pregnancy and may lead to missed diagnoses (187).

In nonpregnant individuals, the 8-mg overnight dexamethasone suppression test distinguishes CD from ectopic ACTH secretion with a sensitivity ranging from 60–80% and a specificity of more than 80% when a cutoff point of plasma cortisol suppression above 80% is used (187, 188). However, some authors advocate abandoning the test altogether; although it can detect patients with CD with relatively high sensitivity, it does not accurately exclude those with ectopic ACTH secretion due to a wide range of suppression of plasma cortisol for each diagnosis (188). The efficacy of the 8 mg dexamethasone suppression test for the differential diagnosis of ectopic ACTH secretion in pregnancy is unknown due to the limited number of reported cases (161, 163, 171, 178, 179, 189, 190, 191). The test may help discriminate adrenal forms of CS from CD, which may be useful given the difficulties in interpretation of plasma ACTH and the increased prevalence of adrenal disorders in pregnancy. In a recent systematic review, no patient with a primary adrenal cause of CS showed suppression, whereas four of seven patients with CD did (173).

In nonpregnant individuals with CD, the tumor corticotropes retain ACTH (and hence cortisol) responsiveness to CRH stimulation, whereas adrenal tumors and the majority of ectopic ACTH-producing tumors do not respond (192). Ovine CRH (the analog available in the United States) is a Food and Drug Administration (FDA) category C drug, recommended for use in pregnancy only when absolutely clinically indicated. Animal studies showed no teratogenic or adverse behavioral effects after 100 µg human CRH during organogenesis (193). Plasma ACTH responses to human CRH, 1 µg/kg, were reduced in third-trimester normal pregnancies (132). Although the CRH stimulation test has not been systematically studied in CS in pregnancy, in reports in the literature (and from our personal experience from three patients tested), there was a substantial rise in plasma cortisol (44–130%), consistent with surgically confirmed CD (161, 164, 165), and no adverse effects were observed (161, 164, 165).

For those pregnant women with CRH and dexamethasone test responses consistent with CD, and pituitary lesions larger than at least 6 mm, usually no additional testing is necessary, just as in the nonpregnant population. For others, IPSS may be warranted. The test involves catheterization of the petrosal sinuses draining the pituitary gland and simultaneous sampling from these and a peripheral vein for ACTH measurement before and after administration of CRH. The central-to-peripheral ACTH gradient in patients with CD is not found in other causes of CS, providing a very high diagnostic accuracy in the differential diagnosis of ACTH-dependent CS in the nonpregnant population (194). CS in pregnancy may represent one spectrum of disease in which the test may have special value given the difficulties with differentiation of normal physiological changes of pregnancy. The perceived risk of ionizing radiation probably has limited its use in pregnancy, reflected by only one published case in the literature using IPSS (165). Two additional cases have since been undertaken at our institution, indicating that the test can be used safely and effectively in a center with clinical expertise (173). Specific precautions, including a direct jugular approach for catheter insertion and use of additional lead barrier protection, are necessary during pregnancy. We advocate that IPSS should only be considered during pregnancy after completion of careful noninvasive assessment and only in centers with special expertise using the technique. Also, because it is not known whether pregnant patients with adrenal disease have complete pituitary suppression, the usual criteria for interpretation may not exclude these patients.

In summary, although no diagnostic algorithm has been developed prospectively, we recommend a combination of UFC and assessment of midnight salivary cortisol for screening of CS in pregnancy. In patients with confirmed CS, a low ACTH should prompt imaging of the adrenals. However, in cases with borderline ACTH, a combination of the 8-mg dexamethasone suppression test and CRH stimulation testing is suggested to establish the presence of, and distinguish between, the ACTH-dependent forms. IPSS may be necessary in a portion of cases with discordant biochemical or imaging findings.

4. Imaging
a. Adrenal.
Early reports of patients with adrenal CS were characterized by either the absence of imaging or reliance on x-ray tomography or pyelography (154, 169, 195). In other patients imaging was deferred until the postpartum period (196). Despite inadequate tumor definition using these modalities, several women had successful localization and surgery (148, 170). In more recent reports, about 50% of women had detailed ultrasound imaging, which is safe and effective in most. However, ultrasound appears to be less sensitive at smaller tumor size so that several cases required additional modalities for tumor localization (166). Magnetic resonance imaging (MRI) and computed tomography (CT) have been used effectively, although the former is preferred during pregnancy due to the risk of ionizing radiation (153, 180, 198). Specific precautions for the use of MRI are detailed below.

b. Pituitary.
Pituitary MRI should be obtained in all nonpregnant patients with ACTH-dependent CS (187). A recent consensus statement concluded that pituitary MRI may provide a definitive diagnosis in the setting of responses to CRH and dexamethasone consistent with CD when a greater than 6-mm pituitary adenoma is identified (187). However, the use of MRI is not routine in pregnant women because of safety issues. Because of potential (but unproven) teratogenic effects of MRI in the first trimester during organogenesis, it is considered contraindicated at that time, but is considered safe after 32 wk gestation. Between 12 and 32 wk, the potential and largely unknown risks of MRI must be balanced with the potential benefit, recognizing that MRI will detect an incidental tumor (6 mm) in up to 10% of healthy individuals. Evidence of a size criterion for pituitary incidentaloma stems from nonpregnant series (199). However, as the normal pituitary increases in size up to 2-fold by the third trimester, there may be an increased number of incidentalomas identified in pregnancy using these criteria compared with the nonpregnant population. The use of the contrast agent gadopentetate dimeglumine (gadolinium) is contraindicated in pregnancy, because it is FDA category C. In one series of nonpregnant individuals, the sensitivity of MRI for detection of CD decreased from 52% with contrast to 38% without (200). Pituitary MRI alone correctly identified an adenoma during pregnancy in five of eight patients with CD, three of whom had macroadenomas (161, 163, 164, 165, 171). This was not sufficiently sensitive for detection of microadenomas (171). Of interest, pituitary macroadenomas, reported in about half of those with reporting of imaging or operative findings, are overrepresented compared with nonpregnant series (163, 165, 191, 201, 202).

E. Treatment of CS
As cited previously, untreated CS is associated with significant maternal morbidity, including diabetes, hypertension, heart failure, and preeclampsia (191, 196, 203), and adverse fetal outcomes, including premature births, spontaneous abortions, stillbirth, perinatal death, and intrauterine growth retardation (149, 159). It is assumed that these outcomes could be prevented by reducing UFC excretion to the upper part of the range observed in normal pregnancy (186, 204, 205). However, treatment for pregnant patients with CS tends to have been implemented sporadically, generally late in the course of the pregnancy. As a result, the ability of treatment to prevent adverse outcomes is not well established. We recently reviewed 136 pregnancies in which treatment outcomes were available. When no active treatment was given, there were 59 live births (76%) compared with 50 live births (89%) in women in whom treatment was instituted at a mean gestational age of 20 ± 1 wk (173). Even in cases with apparent remission after successful treatment, the progression to eclampsia and premature delivery in a case treated at our institution illustrates that successful treatment may not prevent adverse outcomes (173).

Most patients underwent adrenalectomy for adrenal adenomas, although several had adrenal carcinoma (153, 157, 174). The live birth rate after unilateral or bilateral adrenalectomy is approximately 87%; although the patient group is heterogeneous, adrenalectomy appears beneficial (149, 157, 158, 173).

Forty women, including four who were treated at our institution, have been reported with CD. Approximately 20% underwent transsphenoidal surgery (164). The remainder received medical therapy and/or adrenalectomy, and one case of unrecognized pregnancy had external pituitary irradiation (148, 158, 206, 207, 208). A high proportion, either presenting late in pregnancy or before modern management, was left untreated (209, 210). In contrast to medical therapy, which is discussed below, surgery seems to be more uniformly successful (161, 163, 164, 179).

Primary medical therapy was given to 20 women, usually to prolong pregnancy or to prepare for delivery (171, 211). Metyrapone, which seems generally well tolerated, has been used most often (155, 204) and has had no adverse effects on maternal hepatic functioning or fetal development in the small number of cases reported to date. There is one report of fetal hypoadrenalism after metyrapone (151). However, although metyrapone is effective, there is the potential for exacerbation of hypertension and progression to preeclampsia, which may limit its use (155, 178). Ketoconazole has been used successfully without adverse event in three anecdotal reports of pregnancies (211, 212, 213), including in an individual who had discontinued contraception while using ketoconazole, 600-1000 mg, for CD (211). Despite known antiandrogenic effects through inhibition of aromatase activity, a normal male infant was delivered at 37 wk (211). In the rat, ketoconazole crosses the placenta and is teratogenic and abortifacient, so that the drug is FDA category C. Although ketoconazole has been advocated recently as a potential option in patients requiring medical therapy, we recommend its use only in individuals who are intolerant of metyrapone and are in need of emergency medical therapy. Cyproheptadine appeared safe in three women, but is not effective (214, 215, 216). Aminoglutethimide is avoided because it can induce fetal masculinization (217). Similarly, mitotane is contraindicated as it has teratogenic effects (202).

Thus, we recommend surgical treatment of CS in pregnancy, except perhaps late in the third trimester, with medical treatment being a second choice. There does not appear to be a rationale for supportive treatment alone. Perhaps the mixed experience with treatment of CS indicates that this disease is not recognized early enough during the course of pregnancy to impact outcome. Regardless of the chosen treatment strategy, the prognosis for the fetus remains guarded when hypercortisolism persists. An increased suspicion for diagnosis of this rare disease would likely facilitate early treatment and result in improved outcome for both mother and fetus.

	IV. Adrenal Insufficiency in Pregnancy

Top Abstract I. Introduction II. HPA Axis Physiology... III. Cushing’s Syndrome... IV. Adrenal Insufficiency in... V. Summary References

 
A. Overview
Although AI in pregnancy is uncommon, it is important to recognize it to optimize maternal and fetal outcomes. AI can present acutely or with a more insidious set of chronic symptoms. Primary AI (so-called Addison’s disease) refers to intrinsic adrenal pathology with atrophy of the adrenal cortex and insensitivity to ACTH and angiotensin II stimulation, resulting in impairment of aldosterone and/or cortisol secretion. Urinary and plasma cortisol and aldosterone levels are low or undetectable (218). Plasma aldosterone-to-renin ratios are reduced in association with elevated PRA (218, 219). Secondary or tertiary AI arises from impaired ACTH or CRH secretion due to hypothalamic or pituitary disease or, more commonly, as a result of exogenous corticosteroid administration. However, secondary AI is not associated with mineralocorticoid deficiency, as the zona glomerulosa remains responsive to the action of the RAS.

B. Frequency
The prevalence of primary AI in the predominantly Caucasian nonpregnant population is estimated to range between 39 and 117 per million (220, 221, 222). Although the majority of cases of primary AI affect women (92%) (223), the exact prevalence of AI occurring in association with pregnancy is unknown. By 1953, there were approximately 50 cases of AI in pregnancy reported, and since then a similar number have been published (148, 224, 225). In one of the largest series, during a 12-yr period between 1976 and 1987 in Tromso, Norway, five women with AI gave birth to six children. From this series of 15,700 deliveries, the estimated incidence of pregnancy in women with AI was 1:3000 births per 12-yr period (225). In 1968, Mason et al. (221) estimated one case of AI in pregnancy per 12,000 gestations.

C. Causes
The presentation and causes of AI have been reviewed extensively (226, 227). Autoimmune adrenalitis is the most common cause of primary AI in developed countries, whereas tuberculosis is a more common etiology worldwide. Whereas the glands are small in autoimmune primary adrenal disease, they are large in tuberculous or fungal infection, bilateral metastases, hemorrhage, or infarction. A recent Italian survey illustrated the current prevalence and etiology in a group of 322 patients with AI presenting between 1969 and 1999. Most patients were female, and 83% of them had an autoimmune cause for AI. The mean age at presentation was 30 yr and, although tuberculosis was relatively uncommon (12%), that condition was more prevalent in males who had a mean age of presentation of 53 yr (227). The association of AI with type 1 diabetes mellitus has been well described in the general population and in pregnancy (228, 229, 230, 231).

At least seven pregnancies in association with autoimmune polyglandular syndrome (APS) type 2 or Schmidt’s syndrome (primary autoimmune hypoadrenalism, type 1 diabetes mellitus, thyroid autoimmune disease), have been reported since the syndrome was originally described in 1926 (225, 232, 233, 234, 235, 236, 237). This condition is more common in women and is more common than the other forms of APS. APS 2 has a complex inheritance pattern with varying degrees of genetic susceptibility. A high index of clinical suspicion should be present for the diagnosis in offspring of individuals with APS (232, 233, 238). The prevalence of APS 2 is probably overrepresented in the literature of AI in pregnancy, because it is a unique multisystem endocrine disease. Three cases with APS presented as a new diagnosis of AI during pregnancy (233, 234, 236). An awareness of the association of type 1 diabetes or thyroid disease with AI is necessary to ensure adequate screening and recognition of APS before or during pregnancy. In addition to the morbidity associated with AI, untreated hypothyroidism is associated with higher incidence of infertility and miscarriage as well as gestational hypertension and low birth weight (235, 236). Macrosomia and eclampsia are common complications of uncontrolled gestational diabetes, and appropriate management poses a particular challenge beyond that of isolated hypoadrenalism in APS 2 (232).

The most common cause of secondary AI in the adult population is administration of exogenous corticosteroids for conditions such as asthma and inflammatory bowel and dermatological or rheumatic diseases (239, 240). The adverse effects of exogenous steroids and their contribution to fetal growth retardation, suppression of the fetal HPA axis, and effects on neurological functioning have previously been reviewed extensively in both animal models and in humans (241, 242, 243, 244, 245). The true prevalence of AI after long-term glucocorticoid replacement in either the nonpregnant or pregnant population is unknown. An assessment of the HPA axis is warranted for women receiving at least 5 mg prednisone or equivalent per day for more than 3 wk (246). In these cases glucocorticoid reserve should be tested formally before discontinuing a tapering regimen (see below), and stress dosing of glucocorticoids should be administered as clinical suspicion arises. These patients are at particular risk in times of stress and may be at increased risk during pregnancy.

Asthma complicates approximately 4% of pregnancies, and current guidelines support the use of inhaled or systemic corticosteroids for treatment in pregnancy (247). Although chronic oral or high-dose corticosteroid use for asthma in pregnancy is associated with gestational diabetes, preterm labor, and preeclampsia, there have been few reports of adrenal crisis in pregnancy (248). Similarly, whereas steroid dependency is common in up to 36% of patients with Crohn’s disease, there have been only isolated cases presenting in adrenal crisis (249). In contrast, recent series highlight the potential risks of maternal adrenal suppression in women treated with standard short-term doses of betamethasone for preterm delivery (131, 250).

Postpartum pituitary necrosis (Sheehan’s syndrome) is a well-recognized complication of pregnancy, which results after obstetric shock and usually presents with failure to lactate or to resume normal menses in the postpartum period (251). The diagnosis should be considered in postpartum women with hypoglycemia or coma or in stable cases at longer-term follow-up (252). Approximately 20% of cases of Sheehan’s syndrome arise due to antepartum hemorrhage (253). Although it is the most widely cited cause of hypopituitarism in association with pregnancy, this condition has become less common with improved obstetric care (253, 254).

Lymphocytic hypophysitis has considerable overlap in clinical presentation with Sheehan’s syndrome, and these two conditions are the primary differential diagnoses for postpartum hypopituitarism (255). Lymphocytic hypophysitis was first described in 1962 by Goudie and Pinkerton (197), and since then at least 130 cases have been described (256). Approximately 90% of cases present in the last trimester of pregnancy or in the early postpartum period. Lymphocytic hypophysitis is characterized by inflammatory lesions of the pituitary, which simulate a pituitary space-occupying lesion, and often is diagnosed at biopsy of what was considered to be a tumor. The presentation may occur with symptoms of hypoadrenalism or hypothyroidism or other autoimmune conditions, such as pernicious anemia, and may be responsive to glucocorticoids in a proportion of cases.

Other causes of primary hypopituitarism in the adult population are pituitary or other intracranial neoplasms and their associated treatments. In one large series of hypopituitarism in the United Kingdom, 77% had been treated with surgery and 35% with pituitary radiotherapy (257). Iatrogenic causes of secondary AI are important, given the potential for early identification. Careful follow-up after transsphenoidal surgery or pituitary irradiation is recommended. In macroadenomas, ACTH deficiency usually occurs late, in association with a progressive decline in GH, gonadotropin reserve, and TSH production. These all contribute to diminished reproductive function, ensuring that pregnancy is rare except in cases undergoing assisted reproduction. However, since the availability of ovulation induction with gonadotropins, women with established hypopituitarism can expect near normal fecundity, although their pregnancies are considered high risk (258).

D. Maternal and fetal morbidity and mortality
Early reports of AI in pregnancy highlighted the potential risks of mortality (259, 260). Cohen (261) reported a 35% mortality rate for AI in pregnancy in the 70 yr before 1930, which decreased to 18% between 1940 and 1947. In one of the largest early series, Brent (224) observed high rates of adrenal crisis and mortality (45%) in 39 cases of AI in pregnancy before 1946. In contrast, Hendon and Melick (262) subsequently found only one death in 14 cases in 1955. Indeed, more recent series demonstrate the potential for successful maternal outcome after the availability of cortisone in the 1950s. There have been no reported maternal deaths since the 1950s (219, 263). Several subsequent cases have illustrated the potential for safe outcomes for both mother and fetus in previously undiagnosed and untreated cases, probably reflecting less severe AI and improved obstetric care (225, 235, 264). Significantly, unrecognized cases may be protected by transplacental passage of cortisol from fetus to mother. Primate studies showed that up to 60% of fetal cortisol is normally transmitted to the mother, representing 6.6% of total maternal cortisol under normal conditions (236). Consequently, the need for treatment of AI may be recognized only in the immediate postpartum period (236). In cases with AI during pregnancy, careful attention to management of glucocorticoid replacement is required to enhance maternal outcomes and avoid adrenal crisis. Unfortunately, adrenal crisis can occur despite appropriate titration of glucocorticoid replacement, emphasizing the importance of close and careful clinical follow-up (225). Whereas maternal hypotension is a presentation of adrenocortical failure, side effects of treatment for AI include hypertension and exacerbation of preeclampsia (225, 265). It is also important to recognize that, whereas the early emphasis is on careful antenatal care, follow-up in the distant postpartum period is critical given a report of late maternal death at 8 months postpartum (225).

Intrauterine growth retardation and low birth weight are the most commonly reported adverse effects for the fetus from mothers with untreated AI. Osler and Pedersen (228) demonstrated the association of fetal growth retardation with AI in a series of 15 cases in 1962; their observations have since been confirmed in a series of additional case reports by ultrasound (228, 236,