We evaluated the glomerular filtration rate (GFR) during the second postpartum week in 22 healthy women who had completed an uncomplicated pregnancy. We used physiological techniques to measure GFR, renal plasma flow, and oncotic pressure and computed a value for the two-kidney ultrafiltration coefficient (Kf). We compared these findings with those in pregnant women previously studied on the first postpartum day as well as nongravid women of reproductive age. Healthy female transplant donors of reproductive age permitted the morphometric analysis of glomeruli and computation of the single-nephron Kf. The aforementioned physiological and morphometric measurements were utilized to estimate transcapillary hydraulic pressure (ΔP) from a mathematical model of glomerular ultrafiltration. We conclude that postpartum day 1 is associated with marked glomerular hyperfiltration (+41%). A theoretical analysis of GFR determinants suggests that depression of glomerular capillary oncotic pressure, the force opposing the formation of filtrate, is the predominant determinant of early elevation of postpartum GFR. A reversal of the gestational hypervolemia and hemodilution, still evident on postpartum day 1, eventuates by postpartum week 2. An elevation of oncotic pressure in the plasma that flows axially along the glomerular capillaries to supernormal levels ensues; however, GFR remains modestly elevated (+20%) above nongravid levels. An analysis of filtration dynamics at this time suggests that a significant increase in ΔP by up to 16%, an ∼50% increase in Kf, or a combination of smaller increments in both must be invoked to account for the persistent hyperfiltration.

  • pregnancy
  • glomerular hemodynamics
  • ultrafiltration coefficient
  • transcapillary hydraulic pressure gradient

pregnant women exhibit marked glomerular hyperfiltration. During the second half of pregnancy, the glomerular filtration rate (GFR) is elevated above normal, nongravid levels by 40-60% (3, 14, 17, 18, 26, 45, 52, 53). This hyperfiltration appears to result primarily from depression of the oncotic pressure in the plasma that flows axially along the glomerular capillaries (πGC). The latter pressure is, of course, the force that opposes the formation of glomerular filtration (20). The reduction of πGC in pregnancy is attributable to two phenomena. The first is a hypervolemia-induced hemodilution that lowers the protein concentration and oncotic pressure of plasma entering the glomerular microcirculation (30, 47, 49, 50). The second is an elevated rate of renal plasma flow (RPF) (3, 14, 18, 52). An ensuing hyperperfusion of glomeruli blunts the extent to which the axial protein concentration and oncotic pressure can increase along the glomerular capillaries during filtrate formation (11).

In contrast to the gravid state, there is scant knowledge about GFR and its determinants during the puerperium. We have reported that the GFR remains elevated above normal, nongravid levels by ∼40% on the first postpartum (PP) day (38). Although oncotic pressure remained significantly decreased association with persistent hemodilution, the elevation of RPF observed during gestation was no longer evident at this time. Two other studies have reported that GFR remains elevated during the first PP week but is restored to nongravid levels by 1 mo after delivery (25, 36). However, these latter studies have several limitations, including suboptimal techniques for the determination of GFR and failure to evaluate any of the determinants of GFR.

In addition to increasing our understanding of the physiology of the course of gestational hyperfiltration, precise knowledge of GFR and its determinants during the puerperium has implications for the evaluation of pregnancy-induced glomerular disease. Pertinent examples include preeclampsia, which may worsen, and preexisting lupus nephritis, which may flare immediately after delivery (4, 31, 41, 46). Determination of the extent to which such glomerular injuries depress the GFR requires precise knowledge of the normal range for GFR at corresponding point in time during the puerperium. To this end, we evaluated the GFR and its determinants during the second PP week in 22 healthy women completing an uneventful pregnancy. Our findings form the basis of this report.


Study Populations

The experimental group was composed of 22 healthy women who had completed an uncomplicated pregnancy. Each consented to undergo a study of glomerular function that had been approved by the Institutional Review Board at Stanford University. It was conducted during the second PP week. In each case, the gestation had reached full term, 37-41 wk. A subset of seven members of the experimental group also underwent serial plasma measurements of active renin and angiotensin on three occasions: before delivery during the third trimester, on the second or third PP day, and with the renal function study (PP week 2).

As stated above, we have previously reported the findings in 12 healthy mothers undergoing a study of glomerular function after uncomplicated pregnancy on PP day 1 (38). They serve in the present study as a comparison group for the findings in our experimental group.

Two groups of healthy female volunteers of reproductive age (20-48 yr) served as nongravid controls. One group of 40 women underwent the same study of glomerular pressures and flows as our experimental group. These subjects served as functional nongravid controls. A separate group was composed of 12 living kidney transplant donors who underwent biopsy at the time of transplantation. Glomeruli in the biopsy core were subsequently subjected to a morphometric analysis of glomerular structure. These subjects served as structural nongravid controls. Formal studies of glomerular pressures and flows were not undertaken in the structural control group. However, each was meticulously investigated for possible renal abnormalities before kidney donation, and all were found to have a normal creatinine clearance.

All 86 subjects in our experimental, comparison, and control groups denied a history of renal disease, hypertension, or diabetes mellitus. During the pregnancy that preceded the PP studies and at the time of each evaluation, each subject was found to be normotensive and normoglycemic and to have normal levels of serum creatinine and urinary protein.

Physiological Evaluation

Women in the experimental and comparison groups and functional nongravid controls were admitted to the General Clinical Research Center at Stanford University Medical Center. The subset of the experimental group undergoing renin and angiotensin assays maintained an upright posture for 30 min before blood samples were drawn. All subjects had blood drawn for a baseline determination of plasma afferent oncotic pressure (πA) and Hct. Urine was voided spontaneously after diuresis had been established with an oral water load (10-15 ml/kg). A priming dose of inulin (50 mg/kg) and PAH (12 mg/kg) was then administered. Thereafter, inulin and PAH were given by continuous infusion to maintain plasma levels constant at ∼20 and 1.5 mg/dl, respectively.

Sixty minutes after the priming infusion, arterial blood pressure was determined. Four timed urine collections were then made, each of which was bracketed by a blood sample drawn from a peripheral vein. The GFR was expressed as the average value for the four timed inulin clearances. The rate of RPF was estimated by dividing the corresponding clearance of PAH by an assumed renal arteriovenous extraction ratio for PAH of 0.9, a value that we and others have shown to be typical of the healthy kidney in both the gravid and the nongravid state (6, 12, 58). Both GFR and RPF were adjusted for body surface area as determined from height and weight measured at the first obstetrical visit. Inulin and PAH concentrations were determined with colorimetric methods using a Technicon Auto Analyzer II (6). Plasma oncotic pressure was measured directly using a Wescor 4400 membrane osmometer (Wescor, Logan, UT).

Both plasma active renin and angiotensin levels were determined by radioimmunoassay (kit 40-6050, Nichols Institute Diagnostics, San Juan Capistrano, CA, and kit 001-RK-A22, Alpco Diagnostics, Windham, NH, respectively). Differences in plasma volume from the third trimester to the second or third PP day and to the renal function study (PP week 2) were calculated using the equation Math1

where Hct1 is the Hct predelivery, and Hct2 is either the Hct in the early PP period or during PP week 2.

Morphometric Evaluation

Light microscopy. Two cores of renal tissue were taken by open biopsy from the structural nongravid controls. The biopsy was taken immediately before the renal blood supply was clamped and the kidney was removed for transplantation. All glomeruli in a single, 1-μm-thick section stained with periodic acid Schiff reagent were analyzed at the light microscopic level. On average, 19 (range, 5-58) glomeruli were examined. A dedicated computer system (Southern Micro Instruments, Atlanta, GA), consisting of a video camera and monitor, microscope, and digitizing tablet, was used to perform the measurements. The outline of each glomerular tuft in the section was traced onto the digitizing tablet and the mean tuft cross-sectional area was determined using computerized planimetry. Glomeruli that had undergone global sclerosis were rare (<1%). The tuft area was measured only in patent glomeruli. Glomerular volume (VG) was calculated from the average tuft cross-sectional area (AG) as follows Math2

where β is a dimensionless shape coefficient (β = 1.38 for spheres), d is a size distribution coefficient (d = 1.1), which is used to adjust for variations in glomerular size (57), and f is a correction factor for the tissue shrinkage associated with paraffin embedding (fs = 1.64) (44).

Electron microscopy. For transmission electron microscopy, tissue was fixed in 2.5% glutaraldehyde and 2.0% paraformaldehyde in 0.1 M cacodylate buffer and then embedded in Epon. Toluidine blue-stained sections were surveyed to locate blocks with patent glomeruli present entirely within the block. An ultrastructural analysis was performed on two glomerular profiles in each patient. Ultrathin sections (∼90 nm) of the selected glomeruli were stained with lead citrate and uranyl acetate. A complete montage of each glomerulus (×2,850 magnification) was prepared, and line-intercept counting was used to calculate the fractional surface density of the peripheral capillary wall by standard stereological methods (57). Six to eight images of peripheral capillary loops in each of the glomerular profiles were then photographed at ×11,280 to evaluate the frequency of epithelial filtration slits and the thickness of the peripheral glomerular basement membrane. Filtration slit frequency was determined by counting the total number of epithelial filtration slits and dividing that number by the total length of the peripheral capillary wall at the epithelial interface (23). The harmonic mean basement membrane thickness (δbm) was calculated for each individual using the method of orthogonal intercepts (34) Math3 where δbm′ is the apparent harmonic mean basement membrane thickness.

Theoretical Analysis

The relationship between GFR and its determinants was assessed using the following equation Math4

where ΔP is the transcapillary hydraulic pressure difference, πGC is the mean glomerular intracapillary oncotic pressure, and Kf is the glomerular ultrafiltration coefficient, i.e., the product of hydraulic permeability (k) and filtration surface area (S). To compute πGC, we first calculated oncotic pressure of plasma entering the efferent arteriole from the glomerular tuft (πE) Math5

where πA is afferent oncotic pressure, and FF is the filtration fraction. Equation 5 assumes that oncotic pressure rises linearly during axial plasma flow along the glomerular capillary loops, an assumption that we have shown previously to be accurate within 0.5 mmHg (13). We then calculated πGC as the mean of πA and πE (13). We next estimated the Kf for two kidneys as follows Math6 where 1.4 × 106 is the most accurate assessment available of the mean number of glomeruli in women with healthy kidneys between the ages of 20 and 50 yr (48), and SNKf is computed from the above-described morphometric analysis of glomeruli in our structural nongravid controls.

The total filtration surface area in a single glomerular tuft (S) was calculated from Math7

where Sv and VG are, respectively, the filtration surface density and glomerular tuft volume. In making this calculation, we corrected for the effect of immersion fixation to decrease glomerular dimensions relative to in situ perfused glomeruli (44). The intrinsic hydraulic permeability (k) of the glomerular capillary wall was then estimated from the filtration slit frequency (FSF) and the basement membrane thickness by using a hydrodynamic model of viscous flow that has been described in detail elsewhere (19, 23). Finally, SNKf was calculated from the product of k and S (23).

We then assumed that, as in the normal rat, Kf does not change during pregnancy (7, 8), and extrapolated the computed value for Kf from our structural nongravid controls to our comparison and experimental puerperal groups. To take into account the possibility that pregnancy and the puerperium in humans might be associated with glomerular enlargement, we also performed a sensitivity analysis using larger values for VG in calculating S from Eq. 7. To the best of our knowledge, there are no reported determinations of VG in normal pregnancy. We accordingly assumed that VG in normal pregnancy might enlarge by 50 or 100% and recalculated S to provide a range of values and a likely upper bound for our computation of Kf in the two puerperal groups in the present study.

Finally, based on our determinations of GFR and πGC and our estimate of Kf, we rearranged Eq. 4 to compute ΔP, the only remaining determinant of GFR. To estimate ΔP from the three aforementioned quantities, we utilized a modification of the mathematical model of glomerular ultrafiltration of Deen et al. (20), which is described in detail elsewhere (1).

Statistical Analysis

To assess the significance of differences among the experimental, comparison, and control groups, we conducted three-way comparisons using ANOVA or the Kruskal-Wallis test. Either the Student-Newman-Keuls test or the Dunn procedure was used to make the respective post hoc comparisons. A repeated-measures design was utilized to assess for differences in the serum active renin and angiotensin levels before delivery, the early PP period, and PP week 2. All results are expressed as the means ± SD with the exception of the computed value for ΔP, which was expressed as the median (and 95% confidence interval). Proportions were compared using the χ2 test statistic or Fisher exact test where appropriate. SAS (version 8.0) and NCSS/PASS statistical software were utilized.


The observed range of GFR values is illustrated in Fig. 1. There was little overlap between the nongravid control and PP day 1 comparison groups. On average, GFR in puerperal subjects on PP day 1 exceeded corresponding nongravid control values by 42%, 149 ± 33 vs. 105 ± 15 ml·min-1·1.73 m2, respectively (P = 0.01; Table 1). GFR in the experimental group in PP week 2 remained elevated above the control nongravid value, averaging 125 ± 29 ml·min-1·1.73 m2 (P = 0.01). As shown in Fig. 1A, the distribution in PP week 2 GFR was shifted toward the lower end of the PP day 1 range, resulting in a significant lowering of GFR below PP day 1 levels (P = 0.01; Table 1).

Fig. 1.

Box plots comparing the glomerular filtration rate (GFR; A) and the glomerular oncotic pressure (πGC; B) in nongravid controls (left), postpartum (PP) patients studied on day 1 (middle), or during PP week 2 (wk2; right). *P < 0.05 vs. nongravid controls. ^P < 0.05 vs. PP day 1.

View this table:
Table 1.

Clinical characteristics

Several clinical features that bear on glomerular filtration are summarized in Table 1. Both ethnic distribution and calculated body surface area were similar in the three groups. Although age was similar in the two gravid groups, it was significantly higher than that of our functional nongravid controls and lower than that of our structural controls. Whereas determination of the creatinine clearance in healthy populations has suggested that the GFR begins to decline after age 20-30 yr (39, 56), studies that utilized inulin clearance to determine GFR reveal that the GFR does not exhibit a measurable age-related decline until after age 50 (5, 16, 22, 28, 59). Thus the disparity in age between the PP patients and nongravid controls is unlikely to be of biological significance. Although it tended to be lower PP, mean arterial pressure did not differ significantly among the three groups (Table 1). A parallel reduction in Hct and serum albumin concentrations below nongravid control levels is consistent with the persistence of gestational hypervolemia on PP day 1 (10, 32, 55). A parallel increase in Hct and serum albumin concentrations above the corresponding early PP levels in the experimental group by PP week 2 restored these values toward the nongravid control range, a change consistent with a 19% reduction in plasma volume between early PP and PP week 2 (Fig. 2).

Fig. 2.

Percent change in plasma volume (ΔP) from predelivery to the early PP period (left) and from the early PP period to PP week 2.

Glomerular Filtration Dynamics

Glomerular pressures and flows are summarized in Table 2. The hypervolemia on PP day 1 was associated with a trend toward higher RPF, 624 ± 108 vs. 560 ± 130 ml·min-1·1.73 m2 in nongravid controls (P = not significant; Table 2). The subsequent contraction of plasma volume was accompanied by a significant reduction in RPF below PP day 1 levels to 514 ± 109 ml·min-1·1.73 m2 in PP week 2 (P = 0.05 vs. PP day 1; Table 2). The disproportionate increase in GFR relative to RPF on PP day 1 and selective elevation of GFR on PP week 2 resulted in filtration fractions at both time points in the puerperium (0.24 ± 0.05 and 0.25 ± 0.06%, respectively) that were significantly elevated above the control nongravid value (0.19 ± 0.04%; Table 2).

View this table:
Table 2.

Filtration dynamics

Differences in πA among the three groups parallel those summarized in Table 1 for serum albumin concentration. The PP day 1 hypoalbuminemia was accompanied by a ≈7 mmHg mean reduction in measured πA, which averaged only 17.6 ± 1.3 vs. 24.3 ± 2.2 mmHg in nongravid controls (P = 0.005). Both serum albumin and πA (24.7 ± 1.7 mmHg) were largely restored by PP week 2 to normal nongravid values (Tables 1 and 2). Despite the elevated filtration fraction, the computed value for πGC was also significantly lower in PP day 1 subjects than in nongravid controls, 20.4 ± 1.7 vs. 27.1 ± 2.3 mmHg, respectively (P = 0.005; Fig. 1B, Table 2). On the other hand, the persistent elevation of the filtration fraction along with near normalization of the serum albumin in puerperal subjects in PP week 2 resulted in a significant elevation of πGC above PP day 1 levels to 28.9 ± 2.3 mmHg (P = 0.05, Fig. 1B, Table 2). Thus an increase in the pressure opposing the formation of filtrate presumably contributed to the significant reduction observed in GFR between PP day 1 and week 2 but was insufficient to prevent persistence of hyperfiltration of a moderate magnitude during PP week 2 (Table 1, Fig. 1A).

Theoretical Analysis of Kf and ΔP

Given the elevation of πGC above nongravid levels during PP week 2, the persistent elevation of GFR above nongravid levels by 20% at this time could only result from a corresponding elevation of either Kf and/or ΔP (see Eq. 4). In an attempt to evaluate a possible contribution by ΔP, we first used the morphometric analysis of glomeruli in our structural nongravid controls to estimate a likely lower limit for gestational single-nephron Kf. Mean VG was 2.24 ± 1.09 μm3 × 106, and filtration surface density was 0.12 ± 0.03 μm2/μm3. The product of these two quantities yielded a value for S of 4.09 ± 1.22 μm2 × 105/glomerulus (Eq. 7). Glomerular basement membrane thickness averaged 373 ± 37 nm, and the frequency of filtration slits was 1,265 ± 190 slits/mm of glomerular basement membrane length. Applying these quantities to our hydrodynamic model of viscous flow, we estimate k to be 4.1 ± 0.6 m·s-1·Pa-1 × 10-9. The product of S × k yields a nongravid control value for single-nephron Kf of 6.13 nl/(min·mmHg), and Eq. 6 yields a corresponding value for two-kidney Kf of 8.58 ml/(min·mmHg).

The range of values computed for ΔP in nongravid controls (Fig. 3, left) and the PP day 1 and PP week 2 puerperal groups (Fig. 3, center and right, respectively) are illustrated in Fig. 3. In each group, the measured values for GFR, RPF, and πA have been used to compute ΔP along with the estimated value for Kf in the structural nongravid control group. The box plots illustrate that the computed range for ΔP in the two puerperal groups is 1) far broader than in controls and 2) exhibits a non-Gaussian distribution. The controls are computed to have a median value (and 95% confidence interval) of 41.6 (39.8-42.9) mmHg. The PP day 1 comparison group has a similar median value of 42.1 and a 95% confidence interval of 34.2-42.6 mmHg (Table 5; P = not significant). In contrast, the PP week 2 experimental group is computed to have a significant increase in ΔP to 43.5 (42.6-49.9) mmHg (Table 5; P < 0.05 vs. both controls and the PP day 1 comparison group). Of note, although the excess in median ΔP above controls in the experimental group is a modest 1.9 mmHg, the corresponding excess in the upper end of the range is a much larger 7 mmHg (Fig. 3). Thus ΔP in the experimental group could have been elevated above the upper limit in nongravid controls by up to 16%.

Fig. 3.

Box plots comparing the calculated ΔP in nongravid controls (left), patients studied on PP day 1 (middle), and patients studied during the PP wk2 (right) where: *P < 0.05 vs. nongravid controls. ^P < 0.05 vs. PP day 1.

Table 3 also provides hypothetical values for ΔP in the event that the puerperium is associated with an increase in VG by either 50 or 100%, and hence in Kf, above control values. Each of the higher assumed Kf values lowers ΔP significantly below control values on PP day 1. However, given the diminished oncotic pressure on PP day 1, the modeled median ΔP of 35.3 mmHg (with a 50% increase in Kf) is sufficient to maintain a higher net filtration pressure than calculated for the control population. By PP week 2, the higher assumed Kf input values also lower ΔP. Neither the 50 nor the 100% increase in Kf lowers ΔP values significantly below control values, however. Moreover, for each of the three levels of Kf assumed for purposes of this sensitivity analysis, the computed ΔP in PP week 2 exceeds the corresponding value on PP day 1 (Table 3). Thus either an isolated increase in ΔP by ∼5-15% above control values or an isolated corresponding elevation in Kf by ≥50% can be invoked to explain the persistent glomerular hyperfiltration relative to nongravid control values in PP week 2.

View this table:
Table 3.

Modeled ΔP values

The aforementioned computations suggest that some elevation of ΔP could have contributed to persistent hyperfiltration during PP week 2. The combination of a decline in RPF along with an elevation of ΔP and the filtration fraction is consistent with stimulation of the renin-angiotensin system by plasma volume contraction during the PP period (40). However, serial measurements of plasma concentrations of active renin and angiotensin in a subset of the experimental group may not support this as a possible explanation. Plasma active renin before delivery was 51 ± 17 μU/ml. This level declined to 29 ± 9 μU/ml in the early PP and increased only slightly to 36 ± 11 μU/ml during PP week 2. Both PP values were significantly lower than the predelivery values (P < 0.05). The trend upward between the two PP values failed to meet statistical significance (Table 4). Corresponding PP angiotensin levels also declined significantly from predelivery levels of 41 ± 16 to 17 ± 4 pg/ml and 24 ± 14 pg/ml for PP day 1 and PP week 2, respectively (P < 0.05 vs. controls; Table 4).

View this table:
Table 4.

Serial active renin and angiotensin plasma levels


Studies of human pregnancy that have employed the most reliable filtration marker, inulin, have demonstrated that GFR increases progressively during the first half of pregnancy. It then reaches a peak level that exceeds nongravid levels by 40-60%, after which it remains constant until the end of pregnancy (3, 14, 17, 18, 26, 45, 52, 53). As judged by the comparison group in the present study, this level of hyperfiltration persists into the first PP day (38). As judged by our experimental group, hyperfiltration of lesser magnitude (±20%) then persists into the second PP week (Fig. 1A, Table 2).

The GFR can be equated with the product of Kf and the net pressure for ultrafiltration, i.e., ΔP - πGC (Eq. 4). Determinations of RPF, filtration fraction, and πA during peak gestational hyperfiltration in late pregnancy and on PP day 1 reveal that πGC is depressed by between 3.3 and 6.7 mmHg (38, 45, 52). In the event that neither ΔP nor Kf is altered by gestation, an equivalent increase in the net pressure for ultrafiltration could account exclusively for the observed level of hyperfiltration. However, Milne and co-workers (45) have recently interpreted an alteration in the transglomerular sieving of uncharged dextran macromolecules of broad size distribution to indicate that, in addition to depression of πGC by 3.3 mmHg, an ∼15% increase in Kf is required to account for the 38% increase in GFR observed by them in late pregnancy.

Because of technical shortcomings, notably the large coefficient of variation in measured dextran sieving coefficients, the precision of the aforementioned analysis by Milne et al. should be interpreted with caution (45). Another possible source of imprecision in the aforementioned study is that oncotic pressure in plasma was not measured directly by membrane osmometry but was calculated from serum protein concentration using an equation based on the relationship between this quantity and oncotic pressure in the rat (45). We have shown that the foregoing relationship fails to predict oncotic pressure in human plasma accurately (13). In addition to the consideration of species differences between rats and humans, we note with interest that the relationship may be altered by pregnancy. For example, there is a disparity between gestational plasma albumin concentration and oncotic pressure in the experimental group of the present study. Whereas measured πA is almost identical to that in controls, 24.7 ± 1.7 vs. 24.3 ± 2.2 mmHg (Table 2), the corresponding serum albumin is significantly lower, 3.2 ± 0.3 vs. 3.9 ± 0.5 g/dl (Table 1). This suggests that pregnancy and the PP state alter the plasma protein composition, such that elevated levels of circulating proteins other than albumin contribute to the prevailing oncotic pressure (50). As demonstrated by us previously for the nephrotic syndrome, we submit that attempts to calculate oncotic pressure from protein concentration using equations that do not take alterations in protein composition into account are likely to be inaccurate (13).

Baylis (7, 8) has used the micropuncture technique to examine the determinants of peak hyperfiltration in the pregnant rat. Among the determinants of GFR, she observed a significant change only in RPF, which was markedly elevated. πA and ΔP failed to change significantly (7, 8). Because the rats in Baylis's studies were at filtration pressure equilibrium, unique values of Kf could not be calculated, and conclusions about the elevation of Kf could not be drawn. Evidence that Kf elevation may nevertheless be implicated in hyperfiltration is provided by a more recent micropuncture study, in which Baylis compared filtration dynamics in gravid vs. nongravid rats with a remnant kidney (21). As in healthy pregnant rats, higher plasma flow and similar levels of oncotic pressure accompanied remnant hyperfiltration. However, an effect of pregnancy to reduce the hypertension associated with the remnant kidney model resulted in a ΔP value that was 10 mmHg lower than in nongravid remnant kidney rats. Because this experimental model is associated with filtration pressure dysequilibrium, Baylis was able to calculate a unique value for Kf. She reported a greater than twofold increase in this quantity, which accounted for the hyperfiltration in the pregnant animals, suggesting that Kf elevation could well be implicated in gestational hyperfiltration.

From the available evidence, we contend that a role for alteration of any GFR determinants other than depressed πGC in peak gestational hyperfiltration in humans is ambiguous. However, an examination of Eq. 4 reveals that elevation of either ΔP or Kf, or a combined elevation of both, may be implicated in the hyperfiltration that is observed in the first half of pregnancy (14) and during PP week 2. The hemodilution-induced fall in πGC during the second trimester is far smaller than in the third trimester. From the findings of Roberts et al. (52), the estimated reduction of πA at 16 wk of gestation was by only 1.1 mmHg. We compute the corresponding depression of πGC to be only 1.5 mmHg, despite a large increase in RPF. This suggests that increases in ΔP and/or Kf must be invoked to explain the corresponding elevation of GFR by 48% (52).

More certain is that only elevation of ΔP and/or Kf can explain the persistent hyperfiltration by ∼20% in PP week 2 in the present study. A sharp drop in plasma volume resulted in a reversal of hemodilution (Table 1). As a result, πGC actually exceeded the corresponding control value significantly, 28.9 ± 2.3 vs. 27.1 ± 2.3 mmHg, respectively (Fig. 1B, Table 2). In attempting to distinguish a role for ΔP elevation, we have used a modification of the mathematical model of glomerular ultrafiltration by Deen et al. (1, 20). We determined values for GFR and πGC and estimated a likely range of values for Kf. We subsequently rearranged Eq. 4 to yield the resulting value for ΔP.

To the best of our knowledge, no determination of glomerular volume has been made during normal pregnancy. However, given the demonstration of increased kidney size during pregnancy (29), we have considered the possibility that glomeruli may hypertrophy with an attendant increase in filtration surface area and hence, Kf. We thus selected a hypothetical upper limit for glomerular volume that is twofold larger than the corresponding value in healthy age-matched women. Our selection was based on a glomerular volume of 5.23 ± 2.16 μm 106 observed by us in pregnant women with preeclampsia, a value 2.3-fold larger than the corresponding value in the structural control group in the present study (37). Our preeclamptic subjects also exhibited considerable endocapillary cell proliferation and hyperplasia, more markedly in glomerular endothelial than mesangial cells (37). We surmise that the expansion of the endocapillary cell compartment must have contributed significantly to the glomerular enlargement. Any glomerular hypertrophy that might be attributable to normal pregnancy alone is of substantially smaller magnitude and thus in all likelihood, by less than a factor of 2.0. Our sensitivity analysis reveals that an increase of Kf by 50% or more can account exclusively for the maintenance of the increased GFR by 20% in postpartum week 2 (Table 3). Alternatively, should a gestational increase in Kf not eventuate in humans, an isolated elevation of ΔP by up to 16% above the normal range must be invoked to explain the persistent elevation of GFR in PP week 2 (Table 3).

During the present study, we observed that plasma volume declines sharply between early PP and PP week 2. We accordingly proceeded to study the circulating renin-angiotensin system serially in a subset of our experimental group. We wished to test the hypothesis that stimulation of this system in response to plasma volume contraction might lead to efferent arteriolar vasoconstriction and thereby explain the decline in RPF along with a possible elevation of calculated ΔP. Paradoxically, however, active renin and angiotensin levels in plasma declined during the puerperium (Fig. 2 and Table 4). Nevertheless, it remains possible that enhanced angiotensin II action led to efferent arteriolar vasoconstriction in PP week 2. Decreased vascular sensitivity to the pressor effects of angiotensin during pregnancy has been well documented (9, 33, 42, 51). A possible mechanism includes alteration to receptor number and/or affinity (27). Animal data support the possibility that vascular responsiveness to angiotensin is restored by the PP day 5 (15, 35, 43). Thus one possibility is that the resistance to angiotensin II that is observed in pregnancy may have resolved by PP week 2, with the result that angiotensin action was enhanced despite low ambient levels in plasma. Another possibility is that PP week 2 was accompanied by a paradoxical stimulation of a paracrine, intrarenal renin-angiotensin system, notwithstanding an opposite trend in the systemic circulation (2). It is worthy of note that angiotensin I infusion has been shown in the rat to lower Kf modestly (<5%), putatively by stimulating mesangial cell contraction (54). It follows that any effect of enhanced angiotensin II action to elevate GFR in PP week 2 by increasing ΔP would have to be of sufficient magnitude to overcome any concomitant reduction of Kf.

To summarize, we conclude that PP day 1 is associated with marked glomerular hyperfiltration (+41%) comparable in magnitude to peak levels reported in the second half of pregnancy. A theoretical analysis of GFR determinants suggests that depression of πGC is predominantly or exclusively responsible for early elevation of PP GFR. A reversal of the gestational hypervolemia and hemodilution, still evident on PP day 1, eventuates by PP week 2. An elevation of πGC to supernormal levels ensues, yet GFR remains modestly elevated (+20%) above nongravid levels. An analysis of filtration dynamics at this time suggests that either a significant increase in ΔP by up to 16%, an ∼50% increase in Kf, or a combination of smaller increments in both ΔP and Kf must be invoked to account for the persistent hyperfiltration. PP depression of plasma renin and angiotensin levels points away from, but does not exclude, a glomerulopressor effect of this system as a possible mediator of the glomerular hyperfiltration that persists into the PP week 2.


This study was supported by National Institutes of Health Grant DK-52876 and General Clinical Research Center Grant M01-RR-00070. M. Hladunewich's postdoctoral fellowship was supported by the Satellite Dialysis Centers Postdoctoral Fellowship and Education Fund and by the Northern California Chapter of the National Kidney Foundation.


  • 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