Correspondence: Wiebke Arlt MD DSc FRCP FMedSci, Centre for Endocrinology, Diabetes and Metabolism, School of Clinical & Experimental Medicine, University of Birmingham, Birmingham B15 2TT, UK. Tel.: +44-121-415-8716; Fax: +44-121-415-8712; E-mail email@example.com
A new diagnosis of primary adrenal insufficiency (PAI) during pregnancy is extremely rare and difficult to recognize as signs and symptoms such as nausea, fatigue and hypotension may resemble features of normal pregnancy. However, if the diagnosis is overlooked and steroid replacement delayed, subsequent adrenal crisis triggered by hyperemesis gravidarum, fever or delivery can cause severe maternal and foetal morbidity and even mortality. In case of clinical suspicion of PAI, we recommend to measure paired samples of cortisol and ACTH and, if clinically feasible, a short synacthen test. We propose trimester-specific pass cut-offs for the short synacthen test that take into account the rise of total and also free cortisol during pregnancy. Empirical hydrocortisone treatment should never be delayed if the clinical suspicion is high. All pregnant women with PAI should be monitored by a team of endocrine and obstetric specialists. The third trimester is physiologically associated with a rise not only in total but also free cortisol and thus requires regular adjustment of the glucocorticoid dose. Mineralocorticoid requirements may change during pregnancy due to the anti-mineralocorticoid properties of progesterone. As plasma renin physiologically increases in pregnancy, monitoring is limited to clinical assessment including blood pressure and serum electrolytes. It is crucial that a pregnant woman with PAI and her partner are well educated regarding the adjustment of glucocorticoid dose in intercurrent illness and that both are trained in hydrocortisone emergency injection techniques. The obstetric staff should be provided with clear and written guidance for hydrocortisone cover during labour and delivery. With the appropriate replacement therapy, PAI patients can expect to have an uneventful pregnancy and deliver a healthy infant.
Addison's disease or primary adrenal insufficiency, PAI, predominantly affects women in the third and fourth decade of life. This rare condition has a prevalence of 93–140 per million and an incidence of 4·7–6·2 per million; the main cause is an autoimmune destruction of the steroid hormone-producing cells in the adrenal cortex. PAI frequently occurs as part of an autoimmune polyendocrine syndrome, APS, most frequently APS2 associated with thyroid disease, premature ovarian failure and also type 1 diabetes.
PAI can present with an insidious set of chronic symptoms such as fatigue, anorexia and weight loss, or acutely with a life-threatening adrenal crisis. If the diagnosis is overlooked, and initiation of glucocorticoid and mineralocorticoid replacement is delayed, adrenal insufficiency (AI) can be lethal. Life-long hormonal substitution is essential for survival. However, Addison patients suffer from impaired subjective health and have been shown to have an increased mortality although the underlying causes are not entirely clear yet. During reproductive age, adrenal insufficiency may have consequences on fertility and reproduction. In a survey reporting on 269 Norwegian women with PAI, fertility was significantly reduced; the standardized incidence ratio (SIR) for childbirth was 0·97 in the women before being diagnosed with PAI, but dropped to 0·69 after the diagnosis had been established. This remained significantly reduced at 0·72 when excluding all women with premature ovarian failure. This could potentially be explained by the lack of adrenal androgen production in PAI women, as androgens are known to play a crucial role in normal folliculogenesis and consequently in fertility. A population-based cohort study conducted in Sweden identified 1188 women with PAI who delivered live infants, in comparison with 11 879 age-matched controls, between 1973 and 2006. The majority of PAI patients had normal pregnancy outcomes, however, presented with an increased risk of preterm delivery (OR 2·61, 1·69–4·05) and Caesarean section (OR 2·35, 1·68–3·27). Although the latter was thought to be due to physician decision, possibly aiming at the reduction in peripartal stress, the cause for increased rates of preterm delivery was unclear.
We believe that careful management of PAI throughout pregnancy can prevent any increase in maternal and foetal morbidity and mortality. In the following we aim to summarize current knowledge and give guiding recommendations for the management of pregnant patients with PAI.
Glucocorticoid synthesis in normal pregnancy
Normal pregnancy is accompanied by a progressive increase in circulating CRH and ACTH and in the third trimester also in free cortisol levels. This has been nicely documented in a longitudinal study recording diurnal salivary cortisol profiles in 12 healthy women across and after pregnancy (Fig. 1). Salivary cortisol is considered a valid marker of free, bioavailable cortisol, with modern tandem mass spectrometry assays available.[9, 10] A recent longitudinal study in 20 healthy pregnant women demonstrated a gradual increase in total plasma cortisol, corticosteroid-binding globulin (CBG) and 24-h urinary free cortisol, with circulating levels peaking during the third trimester to levels threefold higher than those in nonpregnant controls. The increase in total plasma cortisol concentration is primarily due to the oestrogen-stimulated increase in CBG concentrations, whereas the mechanisms underlying the rise in free cortisol during the later stages of pregnancy are less well understood. This increase may reflect alterations in the set point of the hypothalamic–pituitary–adrenal axis during pregnancy or an antiglucocorticoid effect exerted by the increasing progesterone levels, which have been shown to correlate with increases in salivary cortisol.
Mineralocorticoids in normal pregnancy
Pregnancy is characterized by activation of the renin-angiotensin-aldosterone system (RAAS), although only very few studies looking at this question are available. Figure 2 summarizes the current knowledge on maternal adaptations in RAAS and cardiovascular system during normal pregnancy. Aldosterone secretion is increased and leads to progressive plasma volume expansion. Progesterone acts as an antagonist at the level of the mineralocorticoid receptor, but despite elevated progesterone levels in pregnancy aldosterone can still act as a potent mineralocorticoid. This is partly explained by the efficient conversion of progesterone to 17-hydroxyprogesterone that has a weaker affinity to the mineralocorticoid receptor than progesterone. In addition, progesterone-induced inhibition of the enzyme 11β-hydroxysteroid dehydrogenase type 2 in the kidney results in decreased renal inactivation of cortisol to cortisone, and subsequently an increased activation of the mineralocorticoid receptor by cortisol. During normal gestation, renin release is increased in kidney and extra-renal tissues and oestrogen stimulates angiotensinogen secretion and during pregnancy a gradual, 3–7-fold increase in plasma renin activity has been reported. Although circulating levels of angiotensin II are raised, its vasopressor effect is decreased due to several compensatory mechanisms, in particular release of vasodilator prostaglandins from the uteroplacental unit, which leads to a decrease in systemic vascular resistance and arterial blood pressure.
Diagnosis of addison's disease during pregnancy
Most PAI women giving birth have had their diagnosis established before conception. UK data from the International Addison's Survey report that 14% of the female survey participants (n = 295) have been pregnant or delivered children since their diagnosis. The primary manifestation of newly diagnosed adrenal insufficiency during pregnancy is very rare, but can result in dramatic clinical circumstances if overlooked. Possible events that might trigger an acute adrenal crisis if not or not appropriately replaced might be hyperemesis (mostly in the first trimester), intercurrent illness (throughout pregnancy) and eventually the stress of labour and delivery.
Presenting symptoms are often nonspecific and include fatigue, lethargy, anorexia, nausea, gastric pain and dizziness. These have some similarity to pregnancy-induced conditions such as hyperemesis gravidarum. However, a personal or familial history of autoimmune disease should raise clinical suspicion of PAI in the context of typical symptoms. Hyperpigmentation is a more specific sign of PAI, resulting from increased ACTH stimulation of skin melanocortin 1 receptors. It sometimes is mistaken for chloasma of pregnancy, however, the latter is a melanosis that is restricted to sun-exposed areas only, whereas ACTH-driven hyperpigmentation typically affects areas of increased mechanical friction, such as scars, nipples, knuckles, toes and in severe cases also the oral mucosa in contact with teeth. Chronic mineralocorticoid deficiency may present as postural hypotension (or even syncope) in the presence of hyponatraemia and often metabolic acidosis, whereas hyperkalaemia is less frequently observed in newly diagnosed PAI.
Acute AI, i.e. adrenal crisis, is a life-threatening condition, presenting with severe hypotension or hypovolaemic shock, acute abdominal pain, vomiting and often fever. The onset of autoimmune hyperthyroidism (or thyroid replacement for newly diagnosed hypothyroidism) can precipitate adrenal crisis due to enhanced cortisol clearance.
If acute adrenal insufficiency is suspected in a pregnant patient not previously diagnosed with AI, it is important to obtain samples for measurements of paired plasma cortisol and ACTH levels at baseline, preferably followed by a short synacthen test (SST) with the injection of 250 μg ACTH1–24. In the setting of suspected adrenal crisis, this should be done while gaining intravenous access immediately before initiating treatment with i.v. glucocorticoids, which should not be delayed by awaiting results of investigations if the clinical suspicion is high. Plasma ACTH levels >22 pmol/l are consistent with PAI and this is also valid in pregnant women. The interpretation of cortisol levels is more challenging: In nonpregnant individuals, early morning plasma cortisol levels <165 nmol/l (depending on the local assay) is highly suggestive of AI in the setting of a typical clinical presentation. As total cortisol levels increase during pregnancy, it might be wise to assume higher cut-offs when investigating a pregnant woman suspected to have AI. On the basis of the data presented by Jung et al., which represents the largest cohort of pregnant women with cortisol levels measured by state-of-the-art methodology, we propose that random morning cortisol levels <300, <450 and <600 should raise the suspicion of AI when measured during the first, second and third trimester of pregnancy respectively.
The SST is the most convenient investigation and a very valid measure of adrenal cortisol reserve.[27, 28] This also holds true in pregnancy. Although the BNF currently lists pregnancy as a contraindication for the use of synacthen, this may be historically explained by past use of repeated synacthen injections for induction of persistent hypercortisolaemia, e.g. for treatment of autoimmune conditions. However, there is no medical reason that would support abstaining from the use of a single dose of synacthen for diagnostic purposes in a pregnant woman. The interpretation of the SST during pregnancy has to take into account the fact that total cortisol levels rise throughout pregnancy with a concurrent rise of CBG. A previous study investigated five normal pregnant women and 18 women with secondary AI due to exogenous glucocorticoid treatment during the third trimester employing the low dose (1 μg) SST; results demonstrated that all normal women achieved a peak cortisol of >850 nmol/l and all women with secondary AI failed to mount a response exceeding the conventional cut-off of 550 nmol/l. We have carried out conventional dose (250 μg) SSTs in 37 normally cycling women and 13 women on oral contraceptives (OCP), and the results (Fig. 3) indicate that all women with physiological oestrogen levels pass the conventional SST cut-off of 550 nmol/l, whereas all patients on high-dose synthetic oestrogens pass a cut-off of 800 nmol/l. Jung et al. have demonstrated that OCP induce a 2·6-fold increase in CBG and they observed a 1·9-fold, 2·5-fold and 3·0-fold increase in CBG during the first, second and third trimesters of pregnancy, respectively; this would suggest that women on OCP are equivalent to second trimester pregnant women. Based on the above described data, which are the best available at present, we think that it is feasible to suggest SST pass cut-offs of 700, 800 and 900 nmol/l for the first, second and third trimester of pregnancy respectively. This may err slightly on the side of caution when evaluating cortisol levels 60 min rather than 30 min after ACTH, but given the limited amount of data available at present, we think this approach is most prudent. However, if unsure about the SST results and in the presence of high clinical suspicion, hydrocortisone treatment should always be initiated without delay to establish whether it improves the clinical picture. In principle, the insulin tolerance test has to be considered contraindicated in pregnancy because of potentially adverse effects on mother and child.
Treatment of PAI in pregnancy
The aim of PAI treatment in pregnancy is to achieve a physiological glucocorticoid replacement dose to ensure maternal and foetal health. Hydrocortisone (= cortisol) is efficiently inactivated in the placenta by amply expressed 11β-hydroxysteroid dehydrogenase type 2 and thus cannot cross the placenta to the foetus. The degree of inactivation of synthetic, long-acting glucocorticoids varies and dexamethasone can largely cross the placental barrier unaltered. Therefore, hydrocortisone is the treatment of choice for replacement therapy in pregnant AI patients. Daily doses used for glucocorticoid replacement in PAI generally vary between 20 and 25 mg hydrocortisone split into two or three doses, with 50–75% of the daily dose administered first thing in the morning to mimic the physiological cortisol secretion pattern. Because of the gradual increase in free cortisol observed during pregnancy, most PAI women need to have their daily dose of hydrocortisone increased by 20–40%, i.e. usually by 5–10 mg, during the third trimester.
As is the case for nonpregnant patients, glucocorticoid replacement surveillance is based on clinical grounds as no objective laboratory-based assessment has been proven reliable. The problem with clinical monitoring in pregnancy is that both signs of over-replacement such as weight gain or purplish stretch marks and signs of under-replacement such as fatigue and nausea are not only fairly nonspecific but also frequently observed in normal pregnancy. It is crucial to try and establish the appropriate hydrocortisone dose as over-replacement makes the pregnant PAI woman more prone to gestational diabetes and under-replacement bears the risk of triggering an adrenal crisis.
To protect pregnant PAI women from life-threatening adrenal crisis, the medical team should remain vigilant in ensuring that patients are well aware of the sick-day rules, i.e. the need of doubling the daily doses for hydrocortisone in case of illness, fever, vomiting or severe mental stress until recovery. While planning pregnancy, it should be ensured that all PAI and their partners are provided with an emergency hydrocortisone injection set and receive training for its use.[12, 30] All PAI patients should have a steroid emergency card or bracelet.[25, 30] In particular, early pregnancy can be challenging as signs and symptoms of an adrenal crisis can be confused with hyperemesis gravidarum.[20, 23, 24] On the other hand, hyperemesis can easily elicit an adrenal crisis and appropriate cover with intravenous fluids and stress doses of hydrocortisone is of paramount importance in such a situation. Management of acute adrenal crisis consists of immediate intravenous bolus administration of 100 mg hydrocortisone, followed by injection of 50–100 mg hydrocortisone every 6–8 h (or continuous infusion of 200–300 mg per 24 h) and fluid resuscitation with i.v. physiological saline (initially 1 l/h, then 200 ml per hour) with regular monitoring of blood pressure, heart rate and serum electrolytes.[1, 31]
Fludrocortisone replacement in PAI is initiated at doses of 0·1 mg (with a range varying from 0·05 to 0·25 mg) and usually is continued at the same dose throughout pregnancy. However, progesterone production rises continuously during pregnancy and progesterone has been shown to exert antimineralocorticoid activity in vitro and in vivo.[14, 32] This can usually be compensated for by the increase in hydrocortisone dose during the third trimester as hydrocortisone also exerts some activity at the mineralocorticoid receptor, with 40 mg hydrocortisone equivalent to 0·1 mg fludrocortisone. However, in some cases additional increases in fludrocortisone dose may be required. Importantly, plasma renin cannot be used for monitoring the adequacy of mineralocorticoid replacement as renin secretion physiologically increases during pregnancy.[15, 33] The pregnant PAI patient should be regularly checked for significant postural hypotension (blood pressure supine/erect) and changes in serum electrolytes; if required, also urinary sodium excretion may be checked.
Pregnant PAI women require specialist supervision and ideally should be managed by a team with joint obstetric and endocrine expertise. If the patient is carefully monitored and the medication adjusted as required, the large majority of PAI pregnancies and deliveries will be uncomplicated for both mother and child.
The endocrinologist should provide the pregnant women and the obstetric staff with written instructions regarding glucocorticoid cover during labour and delivery, which should be equivalent to those recommended for major surgical stress and best to be initiated after the onset of the active phase of labour. The patient should receive parenteral administration of hydrocortisone 50–100 mg every 6–8 h (or continuous infusion of 200–300 mg per 24 h) throughout labour and both vaginal delivery or Caesarean section. Postdelivery, the predelivery hydrocortisone dose should be doubled for 24–48 h and then tapered quickly down to prepregnancy dose, if the new mother is clinically well. Assessment of adrenal function in infants from mothers with adrenal insufficiency who received appropriate physiological replacement is not necessary and there is no contraindication for breastfeeding.
Points for guidance
The diagnosis of PAI needs to be considered in a pregnant woman with persistent nausea, hypotension or excessive fatigue, especially if she has an associated autoimmune condition or a family history of endocrine autoimmune disease.
If clinical suspicion of PAI is high, paired serum samples of cortisol and ACTH and, if feasible, a short synacthen test (SST) should be carried out prior to empirical initiation of hydrocortisone replacement. Plasma ACTH will be invariably high in PAI. We recommend that random morning cortisol levels of <300, <450 and <600 should be treated as suspicious of AI when measured during the first, second and third trimester, respectively. We recommend SST peak cortisol pass cut-offs of >700, >800 and >900 nmol/l for the first, second and third trimester of pregnancy.
In case of adrenal crisis, intravenous hydrocortisone and fluid administration are life saving and should not be delayed by diagnostic work-up. Good patient education regarding stress-cover dosage of hydrocortisone and training in the use of a hydrocortisone emergency kit is crucial to prevent adrenal crisis.
Monitoring of hydrocortisone replacement is based on clinical grounds. Due to the physiological increase in free cortisol during late pregnancy, PAI women should receive a 20–40% increase in hydrocortisone dose between second and third trimester. Fludrocortisone dose does not require routine adjustment, but adequacy of dose should be monitored by postural blood pressure and serum electrolytes. Renin is not a valid indicator of mineralocorticoid requirements as it physiologically increases during pregnancy.
Peripartum management includes the initiation of parenteral administration of hydrocortisone in doses recommended for major surgical stress (50–100 mg hydrocortisone every 6–8 h), which should be initiated at the onset of the active phase of labour or prior to planned Caesarean section. After delivery, hydrocortisone dose can be quickly tapered to prepregnancy levels if all is well.
The authors are grateful to Bruno Allolio (University of Würzburg, Germany), Katherine White (Addison's Disease Self-Help Group, UK) and Kristien Boelaert and Shiao Chan (both University of Birmingham, UK) for helpful discussions.