INTRODUCTION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

THE CURRENT ENERGY INTAKE recommendation during pregnancy is an extra 300 kcal/day, or a total cumulative increase of 80,000 kcal (11). The recommendation is based on the assumption that a woman deposits 3.3 kg of fat during gestation, equivalent to ~32,000 kcal or almost one-half of the total cumulative cost for pregnancy. This fat gain was estimated indirectly as the weight not accounted for by tissues directly involved with reproduction or by extracellular water. Numerous studies have attempted to quantify fat deposition during pregnancy to validate this indirect estimate (7, 12, 16, 27). However, measurement of changes in body composition during pregnancy is confounded by a number of factors. One problem is obtaining an appropriate baseline measurement. Because body composition can change as early as in the first trimester (7, 11, 14, 15, 25) and a woman may never again achieve her prepregnancy body composition, "baseline" measurements obtained postpartum or early in pregnancy may not represent the prepregnancy composition. Another problem encountered in quantifying gestational fat gain is that common methods of estimating body composition are based on assumptions that are invalid during pregnancy. Most methods of estimating body fat are based on the two-component model, which assumes that the densities of fat mass (FM) and fat-free mass (FFM) are constant and known (22). During pregnancy, the accumulation of body water results in a decrease in the FFM density, producing a significant error if the usual equations are applied. In 1988, van Raaij et al. (26) published modified equations for estimating FM in pregnancy, which were based on the average changes in density and composition of the FFM taking place in a group of 42 women throughout pregnancy. These equations, and other methods of accounting for the altered FFM composition, have been used recently in studies of body composition during pregnancy (1, 2, 4, 5, 7, 13, 17, 27). The accuracy of these methods in quantifying FM in individual women is still questionable, however, because of the large variability in the amount of body water accumulated among women. It is clear that more valid methods of quantifying FM in individual women during pregnancy are needed to assess the composition of gestational weight gain and to verify the current estimate on which energy-intake recommendations are based.

One way to avoid the limitations of previous studies is to obtain a baseline measurement before conception and to use a four-component model of body composition to estimate FM during pregnancy. Measurement of body water, bone density, and FM separately avoids the problematic assumption of a constant and known composition of FFM. Although the four-component model has been used to estimate body composition in nonpregnant individuals, this model had never before been applied longitudinally to pregnant subjects. Thus the purpose of this study was to develop a four-component model to estimate fat deposition during pregnancy and to apply that model to a group of 10 well-nourished pregnant women followed from before conception to 6 wk postpartum. The inclusion of measurements of body composition before conception, used as each subject's own baseline, was critical to this study.

	SUBJECTS AND METHODS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

Subjects

Sixteen healthy, nonsmoking subjects who were planning pregnancies were recruited from the San Francisco Bay Area and participated in the study day before conception. Of these 16 subjects, 10 became pregnant within 3 mo of their pre-conception measurement and completed the longitudinal study. One subject could not perform the densitometry procedure and is not included in this discussion of body composition. The physical characteristics of the remaining nine subjects and their gestational outcomes are shown in Table 1. All women were of average body weight-for-height, as defined by a prepregnancy body mass index between 19.6 and 26.0 kg/m², and all were primi- or multiparous. None had extreme dietary behaviors (i.e., fasting, bingeing, purging, or pica), which might have influenced food intake, gestational weight gain, and change in body composition. None of the subjects developed preeclampsia, hypertension, or gestational diabetes, and only two complained of peripheral edema (subjects 3 and 4). All subjects carried their pregnancies to term (39-42 wk) and delivered vaginally, except subject 7, who had a Cesarean section because of prolonged labor. The birth weights of the four male and five female infants averaged 3.6 kg, with a range from 2.7 to 4.4 kg.

Experimental Design

Body density. Body density was measured by densitometry. After changing into a bathing suit, voiding and removing all jewelry, each subject was weighed in air to the nearest 0.01 kg on a beam balance scale. Residual lung volume (RLV) was measured in duplicate by oxygen dilution, with subjects in a sitting position, using the method of Wilmore et al. (31). Subjects were then weighed underwater on an overhead spring balance until three successive readings agreed within 50 g. Body density (D_b) was determined by the equation D_b = M_air/{[(M_air  M_water)/D_water]  RLV}, where M is body mass in air or water and D is density. Water density was read off a chart based on water temperature.

TBW. TBW was measured by deuterium dilution. After collection of a baseline urine sample, each subject drank 100 mg/kg body wt of deuterium (99.7 atom %excess). Spot urine samples were collected midmorning on days 1, 5, 10, and 14 postdose. Samples were stored frozen at 80°C, and isotope enrichments were measured in the laboratory of Dr. William Wong (USDA/Agricultural Research Service, Children's Nutrition Research Center, Houston, TX) by gas-isotope-ratio mass spectrometry. The deuterium space was calculated from the zero time intercept of the isotope-disappearance curve measured over the 2-wk interval, assuming single-pool kinetics (21). TBW was estimated as deuterium space/1.04 to account for deuterium exchange with acidic body proteins.

Bone mineral content (BMC). BMC was determined by using a dual-energy X-ray absorptiometer (LUNAR DPX, Madison, WI). This method is based on the differential attenuation of X-rays at two discrete energy levels (70 and 140 keV) as they pass through soft tissue and bone. The emerging X-ray beams at initial intensity (I₀) undergo attenuation with a resultant exponential decrease in intensity by absorption in tissues, such that where U is mass attenuation coefficient and M is mass of soft tissue (st) and bone mineral (bm). The X-ray intensities I are measured directly, and the attenuation coefficients are known by calibration; thus the two equations can be solved for the masses of soft tissue and bone mineral. The data were analyzed by using the system software, version 3.6.

Body Composition Calculations

FM at each time point was also estimated from four previously used methods: the standard two-component TBW and densitometry models, equations of van Raaij et al. (26) derived for pregnant subjects, and Siri's (23) three-component model (Table 3). The change in FM between T₀ and T₃₆, as obtained by our four-component model, was compared with that obtained by using each of these four methods.

Energy intake (EI). EI was estimated at each time point by using 3-day weight food records. Records were analyzed by using Nutritionist III software (version 7.2, N-Squared Computing, Salem, OR), and energy intake and macronutrient content were estimated from the 3-day average value.

Resting metabolic rate (RMR). RMR was measured between 0800 and 0830 under standard conditions after a 10-h fast, using a metabolic cart system with a ventilated canopy (Sensormedics, Yorba Linda, CA). Measurements were made every minute for a 30-min period while the subjects were awake but at complete rest. Energy expenditure (kcal/min) was calculated from measurements of oxygen consumption and carbon dioxide production by using the classic Weir equation (29).

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

Table 4 shows the mean body weights, body densities, TBW, BMC, FFM, FM, RMR, and EI values at each time point for the nine subjects. T₀ body weight averaged 64.7 kg, did not change in the first trimester, then increased 7.4 kg by T₂₆ and an additional 3.8 kg by T₃₆. Body density averaged 1.031 g/ml at T₀ and decreased ~1% to 1.022 by T₂₆ and to 1.024 g/ml by T₃₆. TBW increased 3.0 kg by T₂₆ and by an additional 2.6 kg by T₃₆, for an average total increase of 5.6 kg by T₃₆. The mean T₀ BMC was 2,525 g and dropped 2.5% to 2,463 g at T_post, which was not statistically significant. T₀ FFM averaged 46.3 kg and increased 6.5 kg to 52.8 kg by T₃₆. FM increased from 20.2 to 24.3 kg over the course of pregnancy for an average increase of 4.1 kg. At 4-6 wk postpartum, body weight and FM were the only measured parameters that were still significantly higher than the corresponding T₀ values; 3.3 kg of body weight and 1.8 kg of FM were retained at T_post.

The individual values for gestational weight gain, TBW accumulation, FFM and FM deposition, and the change in RMR and EI by T₃₆ are shown in Table 5. There was a large interindividual variation in all of these values. Weight gain by T₃₆ ranged from 4.5 to 20.2 kg, TBW gain from 1.1 to 10.7 kg, FFM gain from 1.8 to 11.7 kg, and FM ranged from a loss of 0.6 kg to a gain of 10.6 kg. The increase in RMR ranged from 109 to 810 kcal/day, and EI changed from a reduction of 205 to an increase of 414 kcal/day.

Table 6 shows the average FM gained by these nine subjects, as calculated by using the four-component model, compared with the four standard methods. The standard TBW method, assuming a hydration of FFM of 0.73, underestimated the change in FM during gestation by an average of 0.6 kg, compared with the four-component model; the correlation coefficient for individual subjects was only 0.83. The standard densitometry method overestimated FM gain by 1.7 kg, with a correlation coefficient of 0.91. The equations derived by van Raaij et al. (26), based on body density values and taking into account the change in hydration of FFM, underestimated the FM change by 0.5 kg, with a correlation coefficient of 0.90. Siri's three-component model (23) provided results closest to those of the four-component model, overestimating FM gain by 0.4 kg, with a correlation coefficient of 0.95. The standard densitometry method was the only method used to estimate body fat change that differed significantly from the four-component value.

	DISCUSSION

TOP ABSTRACT INTRODUCTION SUBJECTS AND METHODS RESULTS DISCUSSION REFERENCES

An accurate assessment of body fat gain during pregnancy is essential in estimating the additional energy needed to support a full-term pregnancy. In addition, weight gain recommendations are based on the assumption that a certain amount of fat is deposited during pregnancy. Until recently, however, very little valid data have been available to accurately assess body fat deposition in pregnant women. The main reason for this is that most previous studies have used body composition methods that are not valid during pregnancy. The three most commonly used methods for measuring body composition are TBW, total body potassium, and densitometry, all of which rely on a two-component model of body composition. These models assume that FFM is composed of 73% water, 20% protein, and 7% bone mineral. During pregnancy, the composition of the FFM gained can be composed of up to 90% water; in addition, the amount of water in FFM gained is highly variable. Lederman et al. (13) recently showed that, compared with a multicomponent model of body composition, the TBW method underestimated FM gain in pregnant women by 2.3 kg and the densitometry and total body potassium methods overestimated the gain by 3.0 and 5.5 kg, respectively. In a study comparing two-, three-, and four-component models of estimating body FM during pregnancy, Hopkinson et al. (10) found that two-component models varied from underestimating FM by 9% to overestimating FM by 22%, compared with the four-component model. Three-component models provided much more accurate FM values, within 1% of the four-component model.

In the present study, we used a four-component model of body composition to estimate FM gained during pregnancy. This model was based on direct measurements of body water and body density at each time point, bone mineral measurements before and after pregnancy, and estimates of body protein. Using this model, we calculated that 4.1 kg of body fat was deposited by 36 wk of gestation. Our longitudinal data on body weight, TBW, and body density allowed us to compare body fat changes obtained by using standard models of body composition with the changes calculated by using our four-component model. It is not surprising that the standard TBW and densitometry methods under- and overestimated FM gain during pregnancy, respectively. Neither of these methods takes into account the disproportionate amount of water accumulated in the FFM during pregnancy. It is concerning, however, that the equations derived by van Raaij et al. (26) for pregnant women did not fare any better when compared with the results obtained from our four-component model. This may be because these equations were derived for groups and are not accurate when applied to individual women, due to the large interindividual variation in water accumulation among women. We also compared our four-component model with Siri's (23) three-component model to assess whether body density and TBW measurements were adequate to estimate body composition during pregnancy. The correlation, 0.95, was the highest of any method. This could be, in large part, because Siri's three-component model uses the same density and TBW data as our four-component model for each subject.

These comparisons imply that a three-component model is not only superior to the two-component model in estimating body fat in states of altered hydration but is adequate to measure body composition during pregnancy. For practical purposes, the three-component model has the advantage over the four-component model in that it does not require the measurement of bone density, a fairly time-consuming procedure that requires expensive equipment. In addition, the three-component model does not require an estimate of body protein.

There is some concern in using a four-component model that the propagation of measurement errors associated with body density, TBW, and bone mineral measurements offsets the greater validity associated with the measurement of more components. A worst-case scenario, calculated by assuming that the squared errors are independent and additive, shows this not to be the case (18). In fact, the additive measurement errors in both three- and four-component methods do not offset the improved accuracy, compared with the two-component methods.

The total gestational fat gain estimated by our four-component model in these nine women, 4.1 kg, is higher than most previous studies have reported. One reason for this is that, as mentioned, previous studies used methods that were unmodified for pregnant subjects (12, 16, 27) or they used modified methods that may not have been accurate on an individual basis (2, 4, 17, 26). Another explanation for the discrepancy is that many studies of body composition during pregnancy used an early pregnant measurement, i.e., 8-12 wk after conception, as their "baseline" (2, 3, 15). Although our women on average did not deposit weight or fat in the first trimester, others have reported significant changes in body composition as early as 12 wk of gestation (7, 14). An early-pregnancy baseline would underestimate the total gestational change if changes had already occurred by that time. Finally, some studies have used a cross-sectional design (8, 22). Because of the wide differences in the amount of fat gained among women, changes as a result of pregnancy can only be assessed with serial measurements. Gain in FM reported in other studies may also have varied from that in this study simply as a result of the typically small sample sizes.

It could be argued that our subjects' fat and weight gains were extraordinarily high, especially given that their activity levels were low and food was always readily available. However, their total weight gain, 11.2 kg by 36 wk of gestation, correlated well with that of other studies (2, 4, 7, 27) and with Hytten and Leitch's (11) estimated weight gain of 12.5 kg by 40 wk, indicating that this gain was not excessive. In addition, mean dietary intakes increased only 180 kcal/day by the third trimester and averaged a modest 30% of calories from fat. However, food records are notoriously inaccurate, even in compliant subjects (19), and it is possible that their energy and fat intakes were indeed higher than reported.

Table 7 shows how the composition of weight gain changed with time in these nine women. In the second trimester, the subjects gained 65% of their total weight gain; more than one-half of this weight was FM. The subjects gained an additional 3.8 kg in the third trimester, none of which, according to our model, was FM. During the third trimester, the fetus may require such a large proportion of the available maternal energy that the women did not deposit any additional FM.

By 4-6 wk postpartum, an average of 1.8 kg of FM was retained over prepregnancy values. Similarly, van Raaij et al. (27) reported that an extra 1.5 kg of FM was retained at 9 wk postpartum, compared with the subjects' average FM values before conception. Forsum et al. (5), however, reported a higher FM retention of 3.2 kg by 6 mo postpartum in 22 women whose gestational weight gain, i.e., 11.7 kg, was similar to our subjects' average gain.

When the energy cost of the fat gained in our nine subjects is calculated, using an energy density of 9.3 kcal/g fat, an average of 38,130 kcal was deposited, close to Hytten's estimate of 32,000 kcal. However, the energy cost in individual subjects ranged from a loss of 5,600 kcal (subject 3) to a deposition of almost 100,000 kcal (subject 9). We carried out a multiple-regression analysis to explain this large interindividual variation in FM deposition. Interestingly, EI and the change in EI from T₀ to T₂₄, and from T₀ to T₃₆, were not correlated to FM gain. Prepregnancy factors, such as weight, FM, and FFM also did not predict the amount of fat deposited. However, the correlation between FM gain and prepregnancy RMR approached significance (r = 0.66, P < 0.06), indicating that women with higher RMRs in the prepregnancy state gained more fat with their pregnancies. Weight gain was strongly correlated to FM gain (r = 0.69, P = 0.04), indicating that women who gain a large amount of weight deposit more fat. This relationship and the regression equation are shown in Fig. 1. The only other measured factor that was correlated with FM gain was the change in RMR. We found that the more RMR increased during pregnancy, the less fat was deposited (r = 0.56); however, this failed to reach significance (P = 0.12). Table 8 summarizes the regression analysis of these factors on FM gain.

View larger version (8K): [in this window] [in a new window]   Fig. 1.   Relationship between weight gain and fat mass (FM) gain in 9 pregnant women. FM gain was estimated by 4-component model. Regression equation: FM gain = (0.543 × weight gain)  1.9.

The large variation in fat gained during pregnancy using this four-component model substantiates previous research in which less sophisticated methods of estimating body composition were used. The reasons behind this variability remain elusive, and fat gain is seemingly unpredictable from prepregnancy factors. The weak negative correlation between FM deposited and the change in RMR may indicate that individual women favor either an increase in RMR or fat deposition; however, the mechanism directing energy utilization during pregnancy is unknown. Further studies correlating hormonal changes to energy utilization among individual women may help to resolve these issues.

In regard to estimating the energy "requirement" of pregnancy, it is unclear what value for the energy cost of fat deposition should be used. We have shown that the value of 32,000 kcal, on which current recommendations are based, is close to the average amount of energy stored but that individual values range dramatically. It has always been assumed that a certain amount of fat deposition is essential for optimal gestational outcome; however, we found no correlation between maternal FM gained and infant birth weight in our sample of ten women. In addition, gestational fat gained was highly predictive of FM retained by 4-6 wk postpartum (r = 0.86, P = 0.003). Because excess fat gain during pregnancy may lead to maternal obesity and its resulting health problems, it may actually be advantageous to limit fat deposition in the gestational period. Further studies on the effects of a low-fat diet and/or aerobic exercise on body composition and gestational outcome are needed to assess the short- and long-term health benefits of limiting fat gain during pregnancy.

We have shown that standard two-component methods of assessing body composition, even those that attempt to correct for altered hydrational status, are not valid during pregnancy. A four-component model including measurements of bone mineral is ideal to estimate body composition in the pregnant state, but, at the very least, a three-component model using measurements of TBW and body density should be used. The variability among women in the amount of fat deposited during pregnancy makes it impossible to derive a single incremental energy requirement that is applicable to all pregnant women. This variability in FM gained was not explained by prepregnancy weight, body composition, or by energy intake during pregnancy, and women who gained more fat did not have bigger babies. The strong positive correlation between weight gained and fat gained, and the tendency to retain weight and fat in the postpartum period, indicate that it may be appropriate to limit fat gained during pregnancy. Future studies should focus on the effect of hormonal changes on energy utilization and on the effect of exercise and diet on body composition in pregnant women.