Regulatory, Integrative and Comparative Physiology

Estrogen and progesterone effects on transcapillary fluid dynamics

Nina S. Stachenfeld, David L. Keefe, Steven F. Palter


The purpose of this study was to determine estrogen (E2) and progesterone (P4) effects on atrial natriuretic peptide (ANP) control of plasma volume (PV) and transcapillary fluid dynamics. To this end, we suppressed reproductive function in 12 women (age 21–35 yr) using a gonadotropin releasing-hormone (GnRH) analog (leuprolide acetate) for 5 wk. During the 5th week, the women either received 4 days of E2administration (17β-estradiol, transdermal patch, 0.1 mg/day) or 4 days of E2 with P4 administration (vaginal gel, 90 mg P4 twice per day). At the end of the 4th and 5th week of GnRH analog and hormone administration, we determined PV (Evans blue dye) and changes in PV and forearm capillary filtration coefficient (CFC) during a 120-min infusion of ANP (5 ng · kg body wt−1 · min−1). Preinfusion PV was estimated from Evans blue dye measurement taken over the last 30 min of infusion based on changes in hematocrit. E2 treatment did not affect preinfusion PV relative to GnRH analog alone (45.3 ± 3.1 vs. 45.4 ± 3.1 ml/kg). During ANP infusion CFC was greater during E2 treatment compared with GnRH analog alone (6.5 ± 1.4 vs. 4.9 ± 1.4 μl · 100 g−1 · min−1 mmHg−1,P < 0.05). The %PV loss during ANP infusion was similar for E2 and GnRH analog-alone treatments (−0.8 ± 0.2 and −1.0 ± 0.2 ml/kg, respectively), indicating the change in CFC had little systemic effect on ANP-related changes in PV. Estimated baseline PV was reduced by E2-P4treatment. During ANP infusion CFC was ∼30% lower during E2-P4 (6.0 ± 0.5 vs. 4.3 ± 4.3 μl · 100 g−1 · min−1mmHg−1, P < 0.05), and the PV loss during ANP infusion was attenuated (−0.9 ± 0.2 and −0.2 ± 0.2 ml/kg for GnRH analog-alone and E2-P4treatments, respectively). Thus the E2-P4treatment lowered CFC and reduced PV loss during ANP infusion.

  • plasma volume
  • interstitial fluid
  • capillary filtration coefficient

estrogen and progesterone modulate body water distribution across fluid compartments at rest (17, 27, 28) and during perturbations of normal body fluid homeostasis such as dehydration (23), sodium loading (22, 32), and hyponatremia (5). Although progesterone increases throughout pregnancy, the last trimester of human pregnancy is characterized by a rapid rise in estrogen concomitant with increases in plasma volume and interstitial fluid (9). In contrast, the midluteal phase of the menstrual cycle is characterized by a progesterone peak but is accompanied by a fall in plasma volume due to fluid and protein movement into the interstitium (17, 23). Plasma albumin and plasma colloid osmotic pressure (COPp) shifts are believed to be the primary causes for the changes in plasma volume related to estrogen (17) as plasma proteins move from the circulation into the interstitial fluid, resulting in net fluid shifts into the interstitium (17, 27, 28).

While estrogen and progesterone appear to play important roles in regulating fluid movement between the vascular and interstitial fluid compartments (17, 27, 28), the mechanism for these hormonal effects has not been elucidated. The vasodilatory properties of estrogen may be a primary mechanism for the direct estrogen effects on capillary filtration (6, 12, 20), simply by increasing the surface area of the vessels available for water and protein movement. Alternatively, estrogen may act on blood-borne substances such as atrial natriuretic peptide (ANP) to augment fluid shifts out of the vascular compartment. ANP acts on vessels in such a way as to alter Starling forces that control fluid movement across capillaries, augmenting capillary filtration coefficient (8) and protein permeability (33) and inducing extravascularization of fluid and proteins (8, 19). The ANP receptor subtype (ANP-R1) primarily responsible for the vasodilating actions of ANP (1) is upregulated by estrogen and downregulated by progesterone (13). Moreover, the smooth muscle-relaxant effect of ANP on rat myometrium is completely abolished by pregnancy or progesterone administration (18) and is associated with selective downregulation of the cGMP-coupled ANP-R1 receptor (18). In humans, high plasma progesterone levels are associated with lower plasma concentrations of ANP during sodium loading (30) and dehydration (23); plasma ANP concentration is lowest in the luteal phase; and estrogen may modify progesterone-mediated inhibition of ANP (22, 23). Therefore, modulation of ANP or ANP-mediated fluid-regulatory effects may be a mechanism by which estrogen and progesterone alter body fluid dynamics.

A number of investigators have speculated that the change in protein and fluid permeability is due to the actions of estrogen rather than progesterone (27, 28). However, the individual roles of estrogen and progesterone have been difficult to define in young healthy women because these hormones usually increase concurrently in women of reproductive age and often have opposite effects on body water regulation (14, 22, 23). In the present investigation, we suppressed endogenous production of estrogen and progesterone using the gonadotropin releasing-hormone (GnRH) analog, leuprolide acetate, and then administered estrogen alone, as well as combined estrogen and progesterone, to test the hypothesis that estrogen modulates ANP effects on transcapillary fluid dynamics. We hypothesized that estrogen would enhance fluid movement out of the vasculature into interstitial space and that the addition of progesterone would reverse these effects.


We recruited 12 healthy nonsmoking women with no contraindications to GnRH analog or reproductive hormone administration to participate in these experiments. All subjects were interviewed about their medical history and provided written confirmation of a negative Papanicolaou smear and normal physical examination within one year of being admitted to the study. They gave written informed consent to participate in the study, which had prior approval by the Human Investigation Committee of Yale University School of Medicine.

Experimental Design

During the month (early follicular phase) before beginning GnRH analog administration, plasma volume was determined with Evans blue dye dilution (see Physiological Outcome Variables). After this initial test, to suppress reproductive function for the duration of the study, the subjects received the GnRH analog leuprolide acetate (0.5 mg/day, Lupron, TAP Pharmaceuticals, Deerfield, IL) each day for 5 wk. At the end of the 5th week, the women either received 4 days of estrogen administration (17β-estradiol, transdermal patch, 0.1 mg/day, n = 6, 28 ± 1 yr) or 4 days of estrogen with progesterone administration (vaginal gel, 90 mg P4twice per day, n = 6, 25 ± 1 yr). Each hormone treatment was administered for 4 days, and group assignment was random. Experimental protocols were performed at the end of the 4th (GnRH analog alone) and 5th weeks (GnRH with hormone) of administration. This design permitted within-subject comparisons concerning hormone effects on transcapillary fluid dynamics without concern for slow hormone washout between trials.

GnRH Analog (Leuprolide Acetate)

This analog possesses greater receptor binding and decreased degradation than endogenous GnRH and acts as a potent inhibitor of gonadotropin secretion. When leuprolide acetate is given continuously, it downregulates the hypothalamic-pituitary-ovarian axis, with internalization and uncoupling of the GnRH receptors at the hypothalamic level. After an initial stimulation, chronic administration suppresses steroidogenesis via GnRH secretion, leading to low or undetectable estrogen and progesterone concentrations within 14 days. The GnRH analog administration began 7 days after the subject's luteinizing hormone peak. This peak precedes ovulation, usually days 12–14 of a 28-day menstrual cycle, and was determined individually by the use of ovulation prediction kits (OvuQuick, Quidel, San Diego, CA). The subjects self-administered daily subcutaneous injections of the GnRH analog (0.5 mg/day) after training by qualified medical personnel. This method of GnRH analog administration was chosen because it is easily discontinued in the event of uncomfortable side effects, such as headaches or hot flashes, and the suppression of the hypothalamic-pituitary-ovarian axis is reversed on cessation of drug therapy.

Reproductive Hormone Add-Back

During estrogen treatment alone, the subjects received 17β-estradiol administered using transdermal patches delivering 0.1 mg/day (Vivelle; CIBA Pharmaceuticals, Summit, NJ) for 4 days (22). During combined estrogen-progesterone administration, the subjects wore both the estradiol patches and used a vaginal gel containing 90 mg of progesterone twice per day (2.250 g of gel at 8% concentration, Crinone; Wyeth-Ayerst Pharmaceuticals, Philadelphia, PA) for 4 days (4).

Preliminary Procedures

To reduce individual variability in hydration levels, the subjects arrived fasted, having drunk 5 ml/kg of water the night before. The subjects were instructed to refrain from alcohol and caffeine for 12 h before the experiment and to maintain their normal diet and level of activity for 3 days before the experimental protocols. On arrival at the laboratory, the subjects were given a low-fat breakfast (∼300 kcal) with 300 ml orange juice. Fluid intake was limited to avoid the need to void during the ANP infusion. Volunteers arrived at the laboratory at 7:00 AM, and after eating, changed into shorts and a jogging bra, gave a baseline urine sample, and lay supine. Because naive volunteers were used, they were brought to the laboratory before the experimental day and familiarized with all routines to ensure that they were as comfortable as possible on the experimental days.

Experimental Protocol

All testing was conducted in an environmental chamber (ambient temperature = 27°C). During a 60-min control period, the subjects were instrumented for the measurement of heart rate and cardiac stroke volume (cardiac impedance, Minneapolis Impedance Cardiograph, model 304B, Minneapolis, MN) and noninvasive arterial blood pressure (Colin Medical Instruments, Komaki, Japan). Catheters were inserted into large forearm veins for intravenous infusion (right arm) and for blood sampling and venous pressure measurements (left arm). During this control period, the subjects were also instrumented for transcapillary fluid dynamics (venous and interstitial hydrostatic pressures, venous and interstitial colloid osmotic pressures) measurements (see Physiological Outcome Variables). After the control period, resting heart rate, stroke volume, blood pressure, and venous and interstitial pressures were recorded and a blood sample was drawn, after which we began a 120-min infusion (5 ng · kg−1 · min−1, Harvard infusion pump, Harvard Apparatus, Holliston, MA) of ANP in sterile saline. A previous study found plasma ANP concentration to reach steady state levels by 30 min (8), so cardiovascular and capillary filtration coefficient (CFC; see Physiological Outcome Variables) measurements were begun after 30 min of the infusion and continued during the next 60 min, followed by plasma volume determinations. During the ANP infusion, blood samples were collected; cardiac output, heart rate, and blood pressure were measured at 30, 60, 90, and 120 min; and urine was collected at the end. Plasma volume using Evans blue dye (see Plasma volume) was determined during the final 30 min of ANP infusion. From the blood samples, we determined plasma osmolality (POsm), COPp, and concentrations of serum sodium (S[Na+]) and plasma proteins, albumin, and ANP.

To reduce the potential decrease in central venous pressure (∼3 mmHg, Ref. 3) and mean arterial pressure (∼7 mmHg, Ref.31) during ANP infusions, we used medical antishock trousers (MAST, David Clarke, Worcester, MA) to apply lower body positive pressure (LBPP). LBPP (+20 to +25 mmHg) was applied by enclosing the subject's legs in the antishock pants, which can be partially filled by a hand-held pump and monitored by a manometer. To avoid invasive central venous pressure measurements during the ANP infusion experiments, we used changes in stroke volume as an index of changes in central venous pressure. In pilot studies we measured central venous pressure during ANP infusion with a catheter in the superior vena cava while simultaneously measuring cardiac impedance during 120 min of ANP infusion. We demonstrated a strong relationship between stroke volume and central venous pressure (r = 0.95), with a 9.6-ml change in stroke volume indicative of a 1-mmHg change in central venous pressure. This noninvasive measurement of stroke volume is a well-established method of measuring changes in stroke volume and cardiac output (11), although less reliable for baseline, absolute measures.

Physiological Outcome Variables

Urine and blood sampling.

All blood sampling was done via a 20-gauge catheter placed in a large forearm vein. Subjects were supine during placement of the catheter and for at least 30 min before sampling. Blood sampling was done from free-flowing blood, taking care to discard fluid from the dead space of the system before sampling and to leave the dead space filled with heparinized saline (20 U/ml). Urine was collected via voluntary voiding.

Plasma volume.

This technique involves injection of a volume of Evans blue dye into the circulation and taking blood samples for determination of dilution. The dye itself attaches rapidly to plasma albumin and therefore becomes evenly mixed. Absolute plasma volume was measured after complete mixing of Evans blue dye (10 min). Blood was also sampled at 20 and 30 min to ensure that complete mixing had occurred by the 10-min sample. Plasma volume was determined from the product of the concentration and volume of dye injected divided by the concentration in plasma, taking into account 1.5% lost from the circulation within the first 10 min.

Changes in plasma volume (ΔPV) were estimated from changes in hematocrit (Hct) concentration from the baseline sample according to the equation (7)%ΔPV=100[(1Hcta×102)/(1Hctb×102)]100 where subscripts a and b denote measurements at time a and pre-ANP infusion, respectively. To estimate baseline plasma volume, we used preinfusion Hct and Evans blue dye measurement of plasma volume during the last 30 min of the ANP infusion.

Blood and urine constituents.

Blood samples were separated immediately into aliquots. The first was analyzed for hemoglobin (Hb), Hct, and total protein. A second aliquot was transferred to a heparinized tube, a third to a tube containing EDTA and aprotinin, and a fourth to a tube without anticoagulant. The tubes were spun in a refrigerated centrifuge, and the plasma taken off the heparinized sample was analyzed for osmolality (freezing point depression, Advanced Instruments 3DII), albumin (colorimetric assay, Sigma Diagnostic Products), plasma total protein concentration (Biuret method), COPp (colloid osmometer), and plasma estrogen and progesterone concentrations. The EDTA-aprotinin plasma samples were analyzed for ANP concentration. Plasma concentrations of ANP, estrogen, and progesterone were measured by radioimmunoassay. Intra- and interassay coefficients of variation for the midrange standard for ANP (38.2 pg/ml) were 14.4 and 12.2% (Diasorin, Stillwater, MN); for 17β-estradiol (55 pg/ml) were 5.4 and 6.5%; and for progesterone (1.4 ng/ml) were 3.3 and 2.8% (Diagnostic Products). The tube without anticoagulant was used for the analysis of serum sodium and potassium (flame photometry, Instrumentation Laboratory, model 943).

Colloid osmotic pressure.

Colloid osmotic pressure was measured by placing a 10-μl-sample volume on a small-volume membrane colloid osmometer with pore restriction of 30,000 Da (PM30, Amicon). Before the sample measurements, the colloid osmometer was zeroed and calibrated at four pressures (0, 10, 20, and 40 mmHg) and four albumin standards (0, 2, 4, and 6 g/l).

Interstitial fluid hydrostatic pressure.

Interstitial fluid (ISF) hydrostatic pressure was measured from subcutaneous tissue in the forearm using a fluid-filled wick catheter (16, 17). The wick catheter assembly was filled with sterile saline and connected to an Ohmeda P23XL pressure transducer through 15-cm-long PE-50 tubing. The catheter was placed through the subcutaneous tissue of the forearm under local anesthesia (∼1 ml, 1% lidocaine) using 16-gauge intravenous placement unit (Jelco Labs, Rariton, NJ). Reference zero pressure level was set at the catheter tip, and communication between the wick-in-needle and ISF was verified by the response of the wick to light touch on the skin.

Interstitial colloid osmotic pressure.

Interstitial colloid osmotic pressure (COPi) was determined from ISF samples collected from forearm subcutaneous tissue using the empty wick catheter technique (16). Empty wick catheters consisted of 30-cm PE-50 tubing, which at one end had a 10-mm double 4-0 Dacon suture pulled in and anchored with a 6-0 nylon monofilament tether. The wick was wetted with a small amount of sterile saline and placed in the subcutaneous tissue using 16-gauge intravenous placement unit (Jelco Labs) after anesthesia (∼1 ml, 1% lidocaine). The empty wick catheter was placed below the insertion site with a syringe attached and pulled back to create light negative pressure. In most cases, enough sample (∼20 μl) was collected in 1 h for analysis of colloid osmotic pressure. The ISF samples from wick catheters that were bloody on gross inspection at removal were not used.


CFC was measured in the left forearm using venous occlusion plethysmography by relating the rate of change in limb girth to measured venous pressure assuming that increase in limb girth after cessation of venous filling is attributable to capillary fluid extravasation (10, 21). We used a Whitney mercury-in-Silastic strain gauge to register limb volume changes. A gauge was placed around the circumference of the forearm, which was then rested at shoulder level. A pneumatic congestion cuff was placed around the proximal forearm. After a resting measurement period of 5 min, we began a series of three pressure steps. The increment used in each step was 10 mmHg, and the cuff pressure was maintained 7 min at a pressure above venous and below diastolic (20–40 mmHg), with 5-min rest periods between each cuff pressure. This timing permitted five separate measurements, so two of three pressure levels were repeated, with the order of pressure measurements chosen randomly. As the arterial blood flowed in and the venous drainage occluded, the change in circumference was recorded on a strip-chart recorder and by computer. Forearm venous pressure was measured from a superficial vein catheter in close proximity to the strain gauge using a transducer (Ohmeda P23XL, Ohmeda Medical Devices, Oxnard, CA) placed at the vertical level of the catheter tip and connected to the strip-chart recorder and computer. CFC was calculated as the slope of the forearm girth measured starting after peripheral vein pressure stabilizes with limb occlusion pressure (at least 3 min), allowing for 4 min of data contributable to fluid extravascularization (21). We assumed at that point the superficial and deep vessels have filled, and therefore any change in limb girth was due to filtration of intravascular fluid into the interstitial space (Fig.1). The strain gauges were calibrated before and after each experiment by placement on a cylinder of appropriate circumference, changing the gauge length by known amounts and recording resistance changes.

Fig. 1.

Change in limb girth over time as a function of venous pressure.Top: a representative individual graph as used in the calculation of capillary filtration coefficient (CFC).Bottom: sample chart recording at 40-mmHg cuff pressure. Changes in slopes of limb girth responses over a range of venous pressures are an index of changes in CFC. In this case, the linear equation describing this relationship is y = 0.00028x − 0.0049, r 2 = 0.97.


CFC calculation.

CFC was calculated using the equationCFC=2×100×1,000limb girth×11+ψΔlimb girthtime×Δvenous pressure where ψ is the ratio of post- to precapillary resistance, taken to be constant at 0.16 (10), and Δ is change. The factors 2, 100, and 1,000 · limb girth−1 correct circumference measurement to a volume measurement, units of microliters, and standardization to limb volume, and Δlimb girth/(time × Δvenous pressure) is the slope of change in limb girth at 20-, 30-, and 40-mmHg venous pressure. A representative slope is shown in Fig. 1. CFC is expressed as microliters per 100 g per minute per mmHg. The infusion of ANP may have affected the pre- to postcapillary resistance ratio because of a reduction in precapillary pressure. We expect that this effect would be constant among the four treatments.


The variables over time (transcapillary fluid dynamics) were analyzed by conditions (GnRH analog alone, estrogen, or combined estrogen-progesterone) in separate ANOVA models for repeated measures, with Bonferonni post hoc testing applied to determine differences among means.

Sample size calculation.

We calculated sample size estimates for our outcome variable of interest and based our calculations on our least precise estimates. The desired statistical test is two sided at the 5% significance level, with 80% power to detect a difference from chance alone. On the basis of our previous work, 80% power is sufficient at P < 0.05 to detect significant alterations in transcapillary fluid dynamics. For a two-sided test, Z (α) = 1.96, and for 80% power, Z (β) = 0.84 (2). On the basis of effect sizes and standard deviations for COPp [2.0 mmHg and 1.7 (28)] and COPi [1.0 mmHg and 4.2 (28)], a sample size of six was sufficient for 80% power to determine effects in our variables of interest (2). Data were analyzed using BMDP statistical software (BMDP Statistical Software, Los Angeles, CA) and expressed as means ± SE.


Three subjects reported frequent and one subject reported occasional hot flashes during GnRH analog treatment. The subjects reported no other adverse effects due to the GnRH analog or hormone administration. In one subject (estrogen-progesterone group), plasma estrogen concentration was 100 pg/ml in the 4th week of GnRH analog-alone administration, indicating noncompliance with the GnRH analog protocol so her data were excluded from all analyses. Thus the data presented here are from six subjects in the estrogen group and five subjects in the estrogen-progesterone group. The reduction from six to five subjects in the estrogen-progesterone group reduced our statistical power to detect differences in CFC and COPpwithin that group to 77% at P < 0.05. We were able to attain COPi samples for only three subjects in the estrogen group and four subjects in the estrogen-progesterone group so we did not perform statistical tests on this variable.

In both the estrogen and the estrogen-progesterone groups, plasma estrogen and progesterone concentrations were at menopausal levels during GnRH analog-alone administration, and the estrogen patch and the vaginal progesterone administration increased the plasma levels of 17β-estradiol and progesterone to close to midluteal phase levels (Table 1). There were no differences noted in any of the blood (Tables2) or cardiovascular (Table3) variables between the groups before hormone administration.

View this table:
Table 1.

Subject characteristics

View this table:
Table 2.

Blood variables during estrogen and estrogen-progesterone administration

View this table:
Table 3.

Cardiovascular variables

Estrogen-Only Group

Estrogen administration did not affect Hct or Hb before ANP infusion or plasma volume as measured from Evans blue dye during the last 30-min of the infusion (Table 1). Furthermore, there were no preinfusion changes in any of the other blood variables (Tables 1 and2), including forearm venous pressure, interstitial hydrostatic pressure, COPp, COPi, or plasma ANP concentration during estrogen administration. Hormone administration did not affect any of the measured cardiovascular variables before the ANP infusion (Table 3).

ANP infusions increased plasma ANP concentration during GnRH analog-alone administration and estrogen treatment, with no difference between the experimental days (Table 2). ANP infusion increased Hct more rapidly during estrogen administration compared with GnRH analog-alone administration through 90 min of ANP infusion (Table2 and Fig. 2, P < 0.05). Plasma protein concentration and COPp increased during ANP infusion with and without estrogen, but plasma total protein concentration was lower during much of the infusion with estrogen administration compared with GnRH analog alone (Table 2). During ANP infusion, CFC was greater during estrogen treatment (6.5 ± 1.4 μl · 100 g−1 · min−1 · mmHg−1) compared with GnRH analog alone (4.9 ± 1.4 μl · 100 g−1 · min−1 · mmHg−1, Fig. 3, P < 0.05).

Fig. 2.

Changes from baseline (preinfusion) in plasma volume (ml/kg body wt), plasma albumin concentration (plasma[alb]), and plasma albumin content over time during gonadotropin-releasing hormone (GnRH) analog alone, estrogen (GnRH analog with estradiol administration), and estrogen-progesterone (GnRH analog with estradiol and progesterone administration). § Significant difference from preinfusion. Data are means ± SE. Significance was accepted atP < 0.05.

Fig. 3.

Change in CFC during atrial natriuretic peptide infusion between GnRH analog administration alone and the respective hormone treatments. Change in CFC from GnRH alone due to estradiol administration (open bar; left) and due to estradiol and progesterone administration (filled bar; right) is shown. * Significant difference between GnRH analog alone and hormone administration. Data are means ± SE. Significance was accepted atP < 0.05.

Despite the use of antishock pants, ANP infusion reduced stroke volume and cardiac output (Table 3, P < 0.05) during both the GnRH analog and estrogen experiments; however, the magnitude of the changes was not influenced by estrogen administration. Urine sodium excretion over the course of ANP infusion was unaffected by estrogen administration (24.4 ± 4.0 vs. 31.8 ± 9.4 μeq/min), as was urine potassium excretion (9.3 ± 1.4 vs. 7.1 ± 1.3 μeq/min) for the GnRH analog-alone and estrogen treatments, respectively.

Estrogen-Progesterone Group

Estrogen-progesterone added to GnRH analog administration did not affect baseline Hct and Hb, but plasma volume calculated from postinfusion Evans blue dye measurement and baseline Hct indicated preinfusion plasma volume was lower after hormone treatment (Table 1). In addition, POsm, S[Na+], and plasma protein concentration were unchanged relative to GnRH analog-alone administration. The baseline plasma volume decrease in the combined treatment was associated with a significant fall in plasma albumin content relative to GnRH analog alone (Table 2). Cardiovascular function was unaffected by estrogen-progesterone treatment before ANP infusion (Table 3).

Plasma ANP concentration increased similarly during infusion in the GnRH analog-alone and hormone treatments (Table 2). Plasma volume fell during ANP infusion with GnRH analog-alone treatment (Table 2, Fig. 2,P < 0.05). Even though plasma volume and plasma albumin content remained lower during estrogen-progesterone treatment relative to GnRH analog alone throughout the infusion, they did not decrease further in response to ANP stimulation (Table 2, Fig. 2,P < 0.05). CFC was lower during ANP infusion with combined estrogen-progesterone (4.3 ± 0.6 μl · 100 g−1 · min−1 · mmHg−1) treatment compared with GnRH analog alone (6.0 ± 0.5 μl · 100 g−1 · min−1 · mmHg−1, Fig. 3, P < 0.05).

Infusion of ANP was associated with reductions in stroke volume and cardiac output during both the GnRH analog-alone and estrogen-progesterone experiments (Table 3, P < 0.05). These reductions remained throughout recovery as well, and their magnitudes were similar between GnRH analog-alone and estrogen-progesterone administration. Urine sodium excretion was 19.9 ± 3.2 vs. 19.9 ± 3.9 μeq/min for the GnRH analog-alone and estrogen-progesterone treatments, respectively.


Four weeks of GnRH analog administration reduced the 17β-estradiol and progesterone blood levels of healthy young women to those of postmenopausal women. With selective hormone administration, we were then able to isolate specific effects of estrogen, with and without progesterone, on body fluid regulation in healthy young women. Adding 4 days of estrogen administration to the GnRH analog treatment did not change resting plasma volume or plasma albumin content. During estrogen administration ANP infusion caused a greater CFC in the forearm and a more rapid plasma volume loss relative to GnRH alone, although the magnitude of plasma volume loss by the end of ANP infusion was unaffected. Thus estrogen did not modify ANP effects on whole body fluid dynamics. In contrast, progesterone administered with estrogen prevented plasma volume loss induced by ANP infusion, likely by decreasing transcapillary fluid movement as indicated by a decrease in forearm CFC. The changes in body fluids under both hormonal conditions occurred without changes in renal excretory variables or changes in body weight. Thus progesterone blunts ANP-induced plasma volume loss by altering transcapillary forces rather than through changes in sodium retention.

Our finding that estrogen treatment did not change baseline plasma volume is in contrast to other studies (22, 23). In an earlier study, 28 days of combined oral contraceptive administration led to a 3.2% plasma volume increase in young women (23), and 14 days of transdermal 17β-estradiol administration resulted in a 4.5% plasma volume increase in postmenopausal women (22). In these earlier studies the plasma volume measurements were carried out in subjects in the upright seated position, whereas in the present study the plasma volume measurements were carried out with subjects in the supine position, in which plasma volume changes would have been suppressed (15). For example, Nagashima et al. (15) demonstrated that exercise-induced plasma volume and protein content expansion were apparent only in the upright position. The posture effect is likely due to greater right atrial and lymph outflow pressures in the supine position, resulting in lower lymph return of water and proteins (26). Estrogen may magnify the posture effect because estrogen administration to animals reduces the lymph outflow pressure at which lymph flow decreases, thereby requiring less of a rise in right atrial pressure to cause a reduction in lymph flow and increase capillary filtration (29). Taken together, these studies indicate that plasma volume expansion in high-estrogen states is likely to be attenuated when measured in the supine position.

Our data indicate that estrogen augments the ANP effect on transcapillary fluid movement in some vascular (i.e., muscle and skin) beds, but the systemic effect on vascular volume is transient. CFC during ANP infusion was greater with estrogen treatment compared with GnRH analog alone and was associated with a greater plasma volume loss over the first 90 min of ANP infusion (Fig. 2). However, by 120 min of ANP infusion, the extent of plasma volume change was similar between treatments. The unchanged plasma volume suggests that while CFC was increased in the forearm, there was likely compensation in other vascular beds, preventing fluid from leaving the vasculature in other parts of the body. In other words, other Starling forces limit the changes in plasma volume so that the magnitude of the whole body plasma volume change is the same with and without estrogen despite the local CFC increase. Moreover, the greater CFC in the forearm does not necessarily indicate changes in systemic transcapillary fluid movement. For example, in an earlier study (17) COPi was 25% lower in the ankle compared with 12% lower in the thorax in the luteal vs. the follicular menstrual phase (17). Oral contraceptive administration caused adjustments in the Starling forces (e.g., COPp-COPi gradient) in the leg, but not in the thorax (28). Thus compensation in other vascular beds may minimize the effect of local increases in CFC on plasma volume and whole body fluid dynamics.

Preinfusion plasma volume, as estimated from Evans blue dye and baseline Hct, was lower after progesterone-estrogen administration relative to GnRH analog alone. The levels of estrogen and progesterone achieved by our hormone administration were similar to endogenous levels of the hormones in the midluteal phase and are consistent with earlier studies that have reported a fall in plasma volume of 7.3% (23) and 4.9% (24) in the midluteal vs. early follicular phase. We found no change in preinfusion Hct, COPp, plasma albumin content, or total protein concentration, indicating no significant protein loss from the plasma and suggesting that any plasma volume loss would have been associated with a loss of whole blood. Conversely, Oian et al. (17) reported a reduction in COPp during the midluteal phase, as did Tollan et al. (28) during 3–6 mo of oral contraceptive administration, indicating protein loss from the circulation in both cases. The primary difference between these earlier studies and the present study is the length of time the women were exposed to the hormones, suggesting estrogen- or progesterone-related protein loss may only occur during chronic exposure to estrogen and progesterone.

Renal increases in water and sodium retention can compensate for a sustained decrease in plasma volume, such as during our estrogen-progesterone treatment or during the menstrual cycle (17). However, under these circumstances, the greater water retention should lead to lower COPp. This scenario may have occurred in our study, and alterations in the COPimay have compensated over the period of hormone administration to compensate for changes in COPp (17, 28). Unfortunately our COPi samples were too inconsistent to evaluate this hypothesis.

There is one important caveat to using the plasma volume and blood volume Evans blue dye measurements to calculate preinfusion plasma volume and blood volume. We estimated preinfusion plasma volume from the Evans blue dye measurement over the last 30 min of the infusion. Evans blue dye attaches to albumin, and as such is diluted in the blood. Thus ANP, which will alter the escape rate of albumin, may have had an effect on the plasma volume measurement itself. If albumin escape rate were reduced, which is consistent with lower CFC during estrogen-progesterone, the dilution of Evans blue dye attached to albumin would appear lower, thus falsely lowering the calculated plasma volume. On the other hand, the greater CFC during the estrogen-only administration did not seem to have this same effect of falsely increasing the plasma volume measurement.

Irrespective of baseline plasma volume, there was a clear attenuation of ANP-induced plasma volume loss with progesterone-estrogen administration relative to the GnRH analog alone. This CFC response may help to explain earlier findings demonstrating a lower limit to plasma volume loss during passive heating and exercise in the midluteal phase vs. early the follicular phase (25). In our study, progesterone limited plasma volume loss during ANP infusion most likely by limiting the ANP-related rise in CFC and capillary permeability. Thus our data suggest that, rather than a lower limit to plasma volume loss, transcapillary fluid exchange may be what is being regulated in conditions such as heat stress and exercise, both conditions under which ANP is increased (23). Studies with CFC measurements using exercise and heat stress to induce body water loss with controlled hormone levels remain to be done to test this hypothesis.

Our study did not elucidate the mechanism for the differences in transcapillary fluid movement between estrogen and progesterone; however, progesterone may impact ANP modulation of CFC and of plasma volume by altering ANP vasodilating effects. The subtype ANP-R1 receptors with cGMP as second messenger are believed to be primarily responsible for the vasodilating actions of ANP (1). The ANP-R1 receptor cells are decreased in the adrenal cortex of ovariectomized rats but increase with estrogen administration (13). Moreover, the smooth muscle relaxant effects on myometrium of rats were completely abolished by pregnancy and progesterone administration, which was associated with selective downregulation of the cGMP-coupled ANP-R1 receptor (18). Taken together, these data suggest the ANP-R1 cells are downregulated by progesterone and upregulated by estrogen. However, the ANP receptor subtypes mediating ANP-induced changes in endothelial permeability have not been definitively identified.

The GnRH analog, leuprolide acetate, is a viable, well-tolerated method to temporarily suppress reproductive hormones in healthy women. This protocol is a controlled method to study specific effects of estrogen and/or progesterone in healthy young women while eliminating interference from hormone fluctuations over the menstrual cycle. In our study plasma volume did not change in response to estrogen administration and appeared reduced during combined estrogen and progesterone administration. ANP led to plasma volume losses during GnRH analog alone and with estrogen administration. However, the addition of progesterone prevented an ANP-induced plasma volume loss, likely by changing transcapillary forces to favor fluid retention in the vasculature as indicated by the lower CFC. The lower plasma volume and CFC may be due to a reduction in ANP-mediated vasodilation.


The leuprolide hormone add-back protocol enables us to do safely in humans what has thus far only been done in animals: to temporarily “oophorectomize” young, healthy women to isolate specific effects of estrogen and progesterone on a physiological system. Not only will this protocol enable us to draw conclusions about these effects on the fluid-regulatory system, but this model can also be applied to study sex hormone effects on most physiological systems. Because this study had the distinct advantage of being performed in humans, it may lead to therapeutic advances in the treatment of symptoms related to estrogen administration, both in young and older women, such as water retention, bloating, and hypertension. Furthermore, the finding that progesterone with estrogen administration led to reduced transcapillary filtration rate suggests that this hormone combination may be studied as a possible pharmaceutical intervention for a variety of disease states that cause edema. The treatment of acute conditions such as postsurgery hyponatremic encephalopathy, which primarily affect young women, may also benefit from this protocol.


We gratefully acknowledge the technical support of C. Weseman, J. Evenski, and D. Blair. We are also grateful to G. W. Mack for help in designing the methods for these experiments and for help in the preparation of this manuscript.


  • This work was supported in part by a grant from the Ethel F. Donaghue Women's Health Investigator Program at Yale University School of Medicine.

  • Address for reprint requests and other correspondence: N. S. Stachenfeld, The John B. Pierce Laboratory, 290 Congress Ave., New Haven, CT 06519 (E-mail: nstach{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract