Prenatal Amino Acid Supplementation to Improve Fetal Growth: A Systematic Review and Meta-Analysis

Aberrant fetal growth remains a leading cause of perinatal morbidity and mortality and is associated with a risk of developing non-communicable diseases later in life. We performed a systematic review and meta-analysis combining human and animal studies to assess whether prenatal amino acid (AA) supplementation could be a promising approach to promote healthy fetal growth. PubMed, Embase, and Cochrane libraries were searched to identify studies orally supplementing the following AA groups during gestation: (1) arginine family, (2) branched chain (BCAA), and (3) methyl donors. The primary outcome was fetal/birth weight. Twenty-two human and 89 animal studies were included in the systematic review. The arginine family and, especially, arginine itself were studied the most. Our meta-analysis showed beneficial effects of arginine and (N-Carbamyl) glutamate (NCG) but not aspartic acid and citrulline on fetal/birth weight. However, no effects were reported when an isonitrogenous control diet was included. BCAA and methyl donor supplementation did not affect fetal/birth weight. Arginine family supplementation, in particular arginine and NCG, improves fetal growth in complicated pregnancies. BCAA and methyl donor supplementation do not seem to be as promising in targeting fetal growth. Well-controlled research in complicated pregnancies is needed before ruling out AA supplements or preferring arginine above other AAs.


Overview Performed Meta-Analyses
We performed meta-analyses on fetal/birth weight following supplementation with AA in the arginine family, BCAA, and methyl donors. Regarding maternal BP and development of SGA, the arginine family was the only AA group for which a meta-analysis could be performed. A meta-analysis on the development of other pregnancy complications was not possible. Data for gestational weight gain were not pooled because, (1) without individual participant data, we had to estimate the mean weight gain and SD for studies that did report gestational weight at two different time points during pregnancy per group; (2) studies reported gestational weight gain over different gestational time periods, which did not consistently match the supplementation periods; and (3) gestational periods are very different between species. Too few studies reported on glucose levels to pool these data. No data on insulin resistance (HOMA-IR) were found. Results on glucose and gestational weight gain are described in the Appendix.

Meta-Analyses
Supplementation of prenatal AA from the arginine family increases birth weight by 6% (1.06 (1.02; 1.11)) in complicated pregnancies ( Figure 1). No effect was observed in normal-growth pregnancies (1.01 (0.98; 1.05)) or the risk population (1.07 (0. 93; 1.22)). In animal studies only, no differences were observed in normal pregnancies and arginine increased birth weight by 8% (1.08 (1.03; 1.13)) in complicated pregnancies. There were no at-risk studies conducted in animals. In human studies only, no differences were observed in normal-growth pregnancies and an increase in at-risk pregnancies (ROM 1.08 (1.02; 1.13) or MD 219 g (65; 374)) and complicated pregnancies (ROM 1.07 (1.03; 1.11) or MD 162 g (69; 255)). Meta-analysis on prenatal supplementation of amino acids from the arginine family on fetal/birth weight (BW). While there was no effect of prenatal supplementation of amino acids from the arginine family in normal-growth pregnancies, it increased birth weight ratio in a risk population and in complicated pregnancies. The data is ordered within each amino acid from smallest to largest animal. Data represent pooled estimates expressed as ratio of means (ROM) with 95% confidence interval (CI) using a random effect model. Residual I 2 is shown. Some studies had multiple cohorts and are distinguishable in this figure by: * supplementation during full pregnancy in this upper line compared to partial in next line; † this upper line is female offspring compared to next line in male offspring; ‡ in this upper line supplementation period was shorter compared to the next line(s); § in this upper line the daily dose is lower compared the next line(s); || in this upper line primigravid animals were used compared to the next two lines of multigravida animals, in the last two lines the dose differed with the first one being the highest dose. FGR, fetal growth restriction; I 2 , heterogeneity; NCG, N-Carbamylglutamate; PE, preeclampsia; PIH, pregnancy induced hypertension.
Within complicated pregnancies, arginine and NCG appeared to be the most effective AAs in the arginine family ( Figure 2A). The largest increase was noted in sheep ( Figure 2B), in which supplementation consisted of either arginine or NCG. For humans the effect was also significant (increase of 9%). The effect was comparable between different (induced) pregnancy complications ( Figure 2C). AAs from the arginine family appeared to be more effective when supplemented during only one phase of pregnancy, but only two studies supplemented AAs during full pregnancy ( Figure  2D). The administration scheme (continuous vs. interval) was not of influence ( Figure 2E). We observed no effect of a preventive approach versus a therapeutic approach ( Figure 2F). Note that while we did not see clear differences between isonitrogenous vs. non-isonitrogenous control diets, most studies (including all human studies) failed to use isonitrogenous control diets ( Figure 2G). Meta-analysis on prenatal supplementation of amino acids from the arginine family on fetal/birth weight (BW): While there was no effect of prenatal supplementation of amino acids from the arginine family in normal-growth pregnancies, it increased the birth weight ratio in a risk population and in complicated pregnancies. The data is ordered within each amino acid from smallest to largest animal. Data represent pooled estimates expressed as a ratio of means (ROM) with a 95% confidence interval (CI) using a random effect model. Residual I 2 is shown. Some studies had multiple cohorts and are distinguishable in this figure by the following: * supplementation during full pregnancy in this upper line compared to partial in the next line; † this upper line is female offspring compared to the next line which is male offspring; ‡ in this upper line, the supplementation period was shorter compared to the next line(s); § in this upper line, the daily dose is lower compared to the next line(s); || in this upper line, primigravid animals were used compared to the next two lines of multigravida animals; and in the last two lines, the dose differed with the first one being the highest dose. FGR, fetal growth restriction; I 2 , heterogeneity; NCG, N-(Carbamyl) glutamate; PE, preeclampsia; PIH, pregnancy-induced hypertension.
Within complicated pregnancies, arginine and NCG appeared to be the most effective AAs in the arginine family ( Figure 2A). The largest increase was noted in sheep ( Figure 2B), in which supplementation consisted of either arginine or NCG. For humans, the effect was also significant (increase of 9%). The effect was comparable between different (induced) pregnancy complications ( Figure 2C). AAs from the arginine family appeared to be more effective when supplemented during only one phase of pregnancy, but only two studies supplemented AAs during full pregnancy ( Figure 2D). The administration scheme (continuous vs. interval) was not influential ( Figure 2E). We observed no effect of a preventive approach versus a therapeutic approach ( Figure 2F). Note that, while we did Nutrients 2020, 12, 2535 6 of 54 not see clear differences between isonitrogenous vs. non-isonitrogenous control diets, most studies (including all human studies) failed to use isonitrogenous control diets ( Figure 2G). Interpretation of the significance of each meta-regression remained unchanged after p-value correction for the 7 modifiers (p = 0.05/7 = 0.007). A dose-response relation for birth weight was absent with an effective daily dose already reached at the lowest tested dose (Figure 3).
Nutrients 2020, 12, x FOR PEER REVIEW 6 of 58 Interpretation of the significance of each meta-regression remained unchanged after p-value correction for the 7 modifiers (p = 0.05/7 = 0.007). A dose-response relation for birth weight was absent with an effective daily dose already reached at the lowest tested dose ( Figure 3).   Data represent pooled estimates expressed as ratio of means (ROM) with a 95% confidence interval (CI) using a random effect model. Residual I 2 is shown, and in grey is the residual I 2 after removal of the outlier Sharky et al. FGR, fetal growth restriction; I 2 , heterogeneity; PE, preeclampsia; PIH, pregnancy-induced hypertension.
Nutrients 2020, 12, x FOR PEER REVIEW 6 of 58 Interpretation of the significance of each meta-regression remained unchanged after p-value correction for the 7 modifiers (p = 0.05/7 = 0.007). A dose-response relation for birth weight was absent with an effective daily dose already reached at the lowest tested dose ( Figure 3).   There is no dose-response relation between prenatal supplementation of amino acids from the arginine family and birth weight ratio (p slope = 0. 81). An increase of 10% was already reached at the lowest dose. The sensitivity analysis identified the rat study by Sharkey et al. [51] as a sensitive case ( Figure A2). Removing this study resulted in an increase in body weight by 9% (5; 12) in complicated pregnancy and a reduction of I 2 from 93% to 77%. Visual inspection of the funnel plot suggested publication bias ( Figure A3). However, Eggers regression did not confirm this (p = 0.26 for all studies and p = 0.29 for studies in complicated pregnancies).

Meta-Analyses
While prenatal supplementation with AAs from the arginine family did not affect BP in normal-growth pregnancies or the risk population, it reduced BPs, with 25 mmHg (−34; −17) in complicated pregnancies ( Figure A4). However, this reduction was completely driven by animal studies. In human studies only, no significant BP reduction was observed in either normal-growth (−8 (−21; 5)), at-risk (−5 (−14; 5)), or complicated (−2 (−10; 6)) pregnancies. The BP difference was comparable for the type of BP (mean arterial pressure or systolic BP; data not shown; p = NS) [7,31]. Meta-regression showed high interspecies difference in the ten rat and three human study cohorts including pregnancy complications, thus we did not consider further meta-regression analysis rational ( Figure A5). In contrast to birth weight outcome, higher doses did result in larger BP differences ( Figure A6). Sensitivity analysis did not reveal specific influential cases ( Figure A7).

Study Characteristics
Prevention of pregnancy complications in human risk populations was mostly studied after arginine supplementation (n = 8) [78,79,83,[85][86][87]95,96] Table A7). All studies reported on the prevalence of SGA. Neri et al. [83] assessed different cut-offs for SGA and showed that, with the same treatment strategy, the risk of developing SGA was lower when a lower cut-off for birth weight was used. This means that especially the more severe FGR was prevented. Only the cut-off of <p10 was included in our meta-analysis. Some of the cohorts also reported lower risk of preterm birth (n = 3) [78,79,83] and PE (n = 2) [78,79] and no effect on GDM risk (n = 1) [78], but there were too few studies to pool data for individual pregnancy complications. The human studies supplementing arginine were performed in Poland (n = 4) [85][86][87]95], Mexico (n = 2) [83,96], and Italy (n = 2) [79,83].

Meta-Analyses
The odds ratio for developing SGA in a risk population between prenatal supplementation of arginine and placebo was 0.45 ((0.27; 0.75); p = 0.002) (Figure 4). The treatment strategies were similar in these studies (interval, partly, and non-isonitrogenous control diet). Therefore, further meta-regression analysis could not be performed. Based on the sparse data-points, mostly centered around the dose of 0.04 mg/kg, there does not appear to be a clear dose-response relationship ( Figure A8). . Meta-analysis on prenatal supplementation of arginine on development of small for gestational age (SGA) in human risk population. The odd ratio (OR) for developing SGA in a risk population was 0.45 following arginine supplementation during pregnancy compared to placebo (95% confidence interval (CI) 0.27; 0.75) using a random effect model. Residual I 2 for heterogeneity is shown.

Meta-Analyses
Prenatal BCAA supplementation did not improve fetal/birth weight in normal-growth (0.98 (0.95; 1.01)) or complicated pregnancy (1.05 (0.98; 1.13), p = 0.24, I 2 = 69%; Figure A9). We were unable to perform meta-regression because Brunner et al. [43] was the only study performed in pregnancy complications (PKU-induced FGR). Brunner et al. [43] tested different dosages and showed that the highest tested dose of leucine and isoleucine were more effective in pregnancy complications. The dose-response curve showed that higher doses of leucine resulted in exponentially higher birth weight in all pregnancies ( Figure A10). This effect was less clear for valine or for isoleucine. Sensitivity analysis showed that Viana et al. [103], the only mouse study, was an influential case ( Figure A11); removing this study had no significant effect on the pooled effect estimate (0.97 (0.95-0.99); p < 0.01), but did reduce I 2 to 30%.

Meta-Analyses
Prenatal BCAA supplementation did not improve fetal/birth weight in normal-growth (0.98 (0.95; 1.01)) or complicated pregnancy (1.05 (0.98; 1.13), p = 0.24, I 2 = 69%; Figure A9). We were unable to perform meta-regression because Brunner et al. [43] was the only study performed in pregnancy complications (phenylketonuria (PKU)-induced FGR). Brunner et al. [43] tested different dosages and showed that the highest tested dose of leucine and isoleucine were more effective in pregnancy complications. The dose-response curve showed that higher doses of leucine resulted in exponentially higher birth weight in all pregnancies ( Figure A10). This effect was less clear for valine or for isoleucine. Sensitivity analysis showed that Viana et al. [103], the only mouse study, was an influential case ( Figure A11); removing this study had no significant effect on the pooled effect estimate (0.97 (0.95-0.99); p < 0.01), but did reduce I 2 to 30%.

Meta-Analyses
Overall, methyl donor supplementation during normal-growth (0.97 (0.92; 1.02)), risk population (0.98 (0.83; 1.15)), or complicated pregnancy (0.98 (0.93; 1.04)) did not alter birth weight (p = 0.46; I 2 = 96%; Figure 5). The two Egyptian studies were the only human studies showing an improvement in birth weight. The dose-response curve showed that higher (excess) doses of methionine and cysteine resulted in a larger reduction of birth weight as was also visible in the forest plot for prenatal methionine in normal-growth pregnancies ( Figure A12). Meta-regression showed a lack of effect for all three methyl donors in complicated pregnancies ( Figure A13A). Methyl donor supplementation in the two overgrowth (risk) animal studies induced by excess energy and high fat diet failed to influence birth weight [113,132]. However, methyl donors appeared to increase birth weight especially in human pregnancies complicated by PE ( Figure A13B,C). Meta-regression did not identify a more effective treatment strategy ( Figure A13D-F). Interpretation of the significance of each meta-regression remained unchanged when the p-value was corrected for the 6 modifiers (p = 0.05/7 = 0.008). There was no clear publication bias visible in the funnel plot ( Figure A14), which was supported by Eggers regression (p = 0.67). Sensitivity analysis showed that Mori et al. [98] was an influential case ( Figure A15). Removing this study had no significant effect on the pooled effect estimate (0.99 (0.95; 1.02), p = 0.19, I 2 = 91%) in normal-growth pregnancies. We speculate that the difference in effect in this study is caused by the high dose of methyl donor.

Meta-Analyses
Overall, methyl donor supplementation during normal-growth (0.97 (0.92; 1.02)), risk population (0.98 (0.83; 1.15)) or complicated pregnancy (0.98 (0.93; 1.04)) did not alter birth weight (p = 0.46; I 2 = 96%; Figure 5). The two Egyptian studies were the only human studies showing an improvement in birth weight. The dose-response curve showed that higher (excess) doses of methionine and cysteine resulted in a larger reduction of birth weight as was also visible in the forest plot for prenatal methionine in normal-growth pregnancies ( Figure A12). Meta-regression showed a lack of effect for all three methyl donors in complicated pregnancies ( Figure A13A). Methyl donor supplementation in the two overgrowth (risk) animal studies induced by excess energy and high fat diet failed to influence birth weight [113,132]. However, methyl donors appeared to increase birth weight especially in human pregnancies complicated by PE ( Figure A13B,C). Meta-regression did not identify a more effective treatment strategy ( Figure A13D-F). Interpretation of the significance of each meta-regression remained unchanged when the p-value is corrected for the 6 modifiers (p = 0.05/7 = 0.008). There was no clear publication bias visible in funnel plot ( Figure A14), which was supported by Eggers regression (p = 0.67). Sensitivity analysis showed that Mori et al. [98] was an influential case ( Figure A15). Removing this study had no significant effect on the pooled effect estimate (0.99 (0.95; 1.02), p = 0.19, I 2 = 91%) in normal-growth pregnancies. We speculate that the difference in effect in this study is caused by the high dose of methyl donor. Prenatal supplementation of methyl donors did not affect birth weight in normal growth, risk populations or complicated pregnancies. Data is ordered within each amino acid (AA) from smallest to largest animal. Data represent pooled estimates expressed as ratio of means (ROM) with 95% confidence interval (CI) using a random effect model. Residual I 2 is shown. Only two studies included (risk of) overgrowth as their study population (bold). Some studies had multiple cohorts split up by sex indicated by *, in which the upper line represents male offspring compared to the next line in female offspring. FGR, fetal growth restriction; I 2 , heterogeneity; NAC, N-acetyl Cysteine; PE, preeclampsia; DM, diabetes mellitus. Prenatal supplementation of methyl donors did not affect birth weight in normal-growth, risk populations or complicated pregnancies. Data are ordered within each amino acid (AA) from smallest to largest animal. Data represent pooled estimates expressed as a ratio of means (ROM) with a 95% confidence interval (CI) using a random effect model. Residual I 2 is shown. Only two studies included (risk of) overgrowth as their study population (bold). Some studies had multiple cohorts split up by sex indicated by *, in which the upper line represents male offspring compared to the next line which represents female offspring. FGR, fetal growth restriction; I 2 , heterogeneity; NAC, N-acetyl Cysteine; PE, preeclampsia; DM, diabetes mellitus.

Study Quality and Risk of Bias Assessment
The items to determine the risk of bias in animal studies were poorly reported and mostly unclear ( Figure A16 and Table A10). The reporting of key indicators of study quality was poor. Especially blinding at any level of the experiment (3%) and power calculations (0%) were hardly reported. The item on experimental unit was important to detect potential statistical errors in the data analysis. In 51% of the studies, it was unclear whether respectively the mothers, or the individual offspring were used as a statistical unit. For risks of bias, a high risk of bias was most often observed for attrition bias (53%), followed by selection bias based on group similarity at baseline (38%). Nearly all studies had an unclear risk of bias for items concerning blinding and randomization (98-100%), because blinding and randomization were either not mentioned at all, or because the methodology used was not described. In human studies, attrition bias constituted the highest risk of bias as well. In addition, the methods used to achieve randomization and blinding were frequently unclear, as was the risk of potential conflict of interest ( Figure A17). Only one study had a low risk of bias on all parameters, and the worst score included 3 high risk, 3 unclear risk, and 1 low risk item ( Figure A18).

Discussion
This systematic review and meta-analysis are unique in providing an elaborate overview of prenatal AA supplementation on fetal growth and related pregnancy complications in both humans and animals. Almost all studies focused on the effect of supplementation to target fetal undergrowth. Although 12 of the 14 searched AAs were included, arginine was by far the most studied for all outcome parameters.

Fetal Undergrowth
None of the three AA supplementation groups affected fetal growth in normal-growth pregnancies. Specifically, the arginine family improved fetal growth by 6% in complicated pregnancies. BCAA and methyl donors did not indicate an effect on fetal undergrowth; however, these data were sparse with, for example, only one BCAA study performed in growth-restricted pregnancies and no human studies at all. Within the competent arginine family, arginine and NCG were identified as the most potent, but due to co-linearity in sheep studies, and potential confounding by total nitrogen intake, we cannot conclude this with certainty. The beneficial effect of prenatal arginine supplementation on fetal growth was also reflected by the reduced risk of SGA development in the at-risk human population.
Our observed reduction of BP in hypertensive disorders during pregnancy could prolong pregnancy, thereby improving fetal growth. While there was a strong dose-response curve observed when data across species were combined, no effect or even a potential worsening of BP was observed after supplementation with arginine in women with PE and/or FGR. Arginine might therefore be indicated at low doses to prevent FGR but not as maternal indication to directly treat hypertensive disorders of pregnancy.
The effects evaluated in the present studies might be related to the ability of the placenta to secure adequate essential AA supply towards the fetus, assuming that maternal protein (and nitrogen) intake is of adequate quantity and quality. This would plead for a combined intervention with multiple AAs. Beneficial effects of arginine family supplementation might be mediated through the NO pathway [22]. However, at this stage, we cannot rule out that the effects partially result from arginine stimulating placental nutrient transport or (fetal) protein synthesis through the mTOR pathway [24,139]. This would align with the mTOR-mediated alleviation of FGR observed after leucine supplementation [140].

Fetal Overgrowth
We hypothesized that methyl donors could potentially normalize overgrowth. Unfortunately, only two studies used methyl donor supplementation (and one used arginine) in overgrowth (risk) pregnancies, leaving the answer to the research question inconclusive. Methionine at (very) high doses reduced fetal growth in normal-growth pregnancies. This is potentially due to reduced maternal food intake and a reduction in ovarian steroidogenic pathway activity that could be rescued by administration of exogenous estrone and progesterone [98,116,119]. However, even in rats administered estrone and progesterone, fetal weight was still reduced compared to pair-fed controls, so additional mechanisms may be involved [116,119]. Several human studies have also reported side effects of methionine at extremely high levels [141].
Very little research was performed in diabetic pregnancies regarding the possible effect of AA supplementation on glucose and insulin levels. However, oral administration of choline prior to and during pregnancy in mouse models of maternal obesity has been reported to reduce fetal overgrowth [26,142], which is a common complication in diabetic pregnancy.

Strengths and Limitations
The major strength of this meta-analysis involves integration of data across species. This relatively novel but increasingly used methodology has been shown to be of great value to improve translation from animal studies to humans in several fields since (1) they provide insight on the safety of interventions because of the larger range of dosages, (2) they aid in determining factors influencing the effect size, (3) they reveal biases thus leading to less misinterpretation, and (4) they clarify differences in design between animal and human studies [143,144]. For instance, we previously showed that a large RCT might not have not observed benefits of a treatment due to underdosage [31]. In this integrated meta-analysis, we additionally combined different groups of AAs that act through different pathways, but included only oral supplementation, and different dosages, all to get one step closer to the bedside. This was a valuable approach for the arginine family, but the relative scarcity of studies performed in complicated pregnancy settings compared to normal-growth ones for the BCAAs and methyl donors limited our ability to draw conclusions about which AAs would be most efficient.
Higher heterogeneity in this integrated type of meta-analysis compared to clinical meta-analyses is inevitable due to the inclusion of different experimental designs. Of note, the aim of a meta-analysis of animal studies alone or combined with human studies is not to pinpoint the effect estimated to directly drive clinical practice. Rather, their goal is to investigate factors influencing treatment efficacy, by determining sources of heterogeneity. As such, high heterogeneity provides the chance to explore its source, and the results generate new hypotheses on how to improve efficacy of the intervention or design of future (human) studies. However, the relatively high heterogeneity in our meta-analysis could not always be fully explained by the performed meta-regression. Socio-economic status taken as a surrogate for baseline nutritional status could influence in particular the results of human supplementation, but included studies were performed either in countries with a similar socio-economic status/ethnics division or that did not have a different impact on effect size. Furthermore, animal models represent a part of a complex syndrome and could influence the results, with our main concern regarding studies supplementing with arginine in compromised animal models by a manipulated NO-pathway However, we could not identify an effect on birth weight or blood pressure when excluding studies using L-NAME-induced animal models.
Our risk of bias tool revealed that most human and animal studies failed to report on quality items or risk of bias items. The unclear risk of bias must be taken into account when interpreting the results (of the individual studies and of our meta-analysis). As we did not exclude any studies based on their risk of bias score, this may have contributed to the high heterogeneity (although it was unclear to what extent, as they were not reported). One study [51] could be considered an influential case in our meta-analysis, since removal of this study would result in a significant drop in I 2 value in both the overall meta-analysis and meta-regressions on the effect of arginine family supplementation and birth weight. However, we could not find any reason for the apparent atypical result found in this study (and have therefore not excluded this study from analysis).
Furthermore, fetal/birth weight is an interesting direct pregnancy outcome, but it does not necessarily correlate with other important obstetric, neonatal, and developmental programming outcomes related to improved long-term health. Hence, BCAA or methyl donors could have no effect on birth weight while still having beneficial or adverse developmental programming effects [145,146]. This was beyond the scope of our meta-analysis.

Perspectives
Overall, this systematic review gives a broad overview of the reported effects of oral prenatal AA supplementation on fetal growth and related pregnancy outcomes. We conclude that none of the AA groups had any adverse effects on fetal growth at low doses. Supplementation with AAs from the arginine family improved birth weight in complicated pregnancies, and reduced risk of SGA development in a human risk population. However, the potency on maternal BP was less clear and the arginine family might not be indicated as maternal treatment for hypertensive disorders of pregnancies. Based on this systematic review and meta-analysis, we formed recommendations for future research, which are summarized in Table 1. We plead for better and well-controlled study designs by using the most suitable study population and animal models, isonitrogenous control diets, and similar baseline nutritional state. In addition, the risk of bias could be reduced by a preplanned protocol describing the intended outcomes, and blinding and randomization methods. Supplementation of BCAA and methyl donors requires more research in animal studies to subsequently determine their potential on fetal growth, blood glucose, and HOMA-IR in models of pregnancies complicated by GDM or fetal overgrowth. The optimal combination of several AAs complemented with potential co-factors should be determined in future research. However, the beneficial effects that this review presents encourages a human RCT on supplementation of arginine family members, with an isonitrogenous control diet, to treat and prevent fetal growth restriction. The screening of hits was conducted by two independent investigators, first based on their title and abstract and, subsequently, eligible articles were screened for final inclusion based on their full-text. A third investigator was consulted when consensus was not reached.

. Exclusion Criteria
Studies were excluded in cases of combined intervention (e.g., supplementation with two or more AAs in the treatment arm), other administration routes than oral, intervention not during pregnancy, pre-conceptional administration, supplementation other than the 14 specified AAs, no control treatment group present, non-mammals, no outcome of interest as previously mentioned, irretrievable full-text or meeting abstract irretrievable, or if the research articles on the studies did not contain unique primary data.
Data was extracted on study characteristics, including species, strain, animal model, pregnancy complication, and maternal weight. Pregnancies complicated by placental insufficiency were labelled as one or a combination of the following: fetal growth restriction (FGR), preeclampsia (PE), or pregnancy-induced hypertension (PIH). Regarding supplementation strategy, we extracted data on the dose in grams per kg body weight per day, the duration of supplementation, the timing during pregnancy (partly or full), the administration scheme (continuous versus interval), the intervention type (prevention or treatment), and whether an isonitrogenous control diet was provided. Maternal weight was used to calculate the dose in grams per kg body weight per day, and maternal weight was estimated when not provided. For birth/fetal weight, the number of offspring and sex was also extracted. For maternal BP, the method and type of measurement were extracted. We also extracted whether BP measurements were performed under stressful condition; in humans, whether it concerned a 24 h or office BP measurement; and in animals, whether BP was measured under restrained or unrestrained conditions.
When data was only presented graphically, we used a graph digitizer to extract the data (http://arohatgi.info/WebPlotDigitizer/). We contacted corresponding authors once per email in case of missing data. SEM and pooled SEM were converted to SDs. The Hozo formula was used to estimate the mean and SD when the median was reported [147].

. Amendments to Protocol
The following amendments to the review protocol were made post hoc: dose-response curves, meta-regression in type of pregnancy complication, and isonitrogenous versus non-isonitrogenous control diet in arginine family. We also changed the categories early, mid, late, and full gestational to partly vs. full gestational, because most studies reported overlapping parts during pregnancy (e.g., early-mid) which resulted in multiple categories with only one study per category. Also, pregnancy "trimesters" and the stage of development were difficult to compare between species. We extracted data on basal protein intake, but we could not perform our initial planned meta-regression. Since the cut-off of when basal protein intake is too low differs per species, and the individual intake and maternal weight (gain) were not reported, we could not convert the extracted data into a unit of measurement that we could pool.
Appendix A.1.5. Adjustments Made to the Risk of Bias Tools To the SYRCLE tool, we added the reporting item whether the correct experimental unit was used. For the item of comparable baseline characteristics, we assessed (1) whether induction of the animal model occurred at the same gestational age, (2) whether the age or weight of pregnant animal was similar (<10% difference), and (3) whether parity (virgin or multiple pregnancies) was similar between groups. Other risks of bias within the Cochrane tool entailed a statement of no conflict of interest.
None of the studies reported HOMA-IR levels, and only seven animal studies reported blood glucose levels: two rat [45,49], one sheep [57], and three pig studies [67,74,77] (Table A9). All studies supplemented with arginine and two studies had an extra cohort with NCG supplementation [57,77]. None of the studies in normal-growth pregnancies (n = 3), complicated (n = 2), or at-risk pregnancies (n = 1) reported significant effects of arginine supplementation on maternal blood glucose levels.

Appendix A.2.2. Effect of Prenatal BCAA on Maternal Weight Gain and Glucose Levels
Maternal weight gain was reported in four rat studies; of these, Brunner [43], Matsueda [97], and Mori [98] tested in different treatment arms the effects of valine, leucine, and isoleucine and Ventrucci [101] studied only leucine supplementation (Table A8). Almost all studies included normal-growth pregnancies in which no effect was observed. The one study including pregnancies complicated by FGR found some effects in high-dose groups: a reduction in gestational weight gain following valine, an increase following leucine, and no significant effects of isoleucine [43]. Glucose levels could only be extracted from one study, using leucine supplementation in normal-growth pregnant rats [150]. In this study, leucine supplementation increased maternal blood glucose levels. As for the arginine family and maternal weight gain, we were unable to pool the data.
We included four animal studies reporting on maternal blood glucose level in response to methyl donor supplementation [107,113,132,152] (Table A9). Maternal blood glucose levels remained in the same range following choline supplementation in two cow studies including normal-growth, FGR, or overgrowth risk [132,152]. However, the two cysteine studies reported significant increases in maternal blood glucose levels in a streptozotocin-induced pre-existent diabetes mellitus (DM) type 1 mice model [107] and an overgrowth risk rat model using high fat diet [113].

Appendix B
Nutrients 2020, 12, x FOR PEER REVIEW 15 of 58 FGR found some effects in high dosed groups: a reduction in gestational weight gain following valine, an increase following leucine and no significant effects of isoleucine [43]. Glucose levels could only be extracted from one study, using leucine supplementation in normal-growth pregnant rats [150]. In this study, leucine supplementation increased maternal blood glucose levels. As for the arginine family and maternal weight gain, we were unable to pool the data.
We included four animal studies reporting on maternal blood glucose level in response to methyl donor supplementation [107,113,132,152] (Table A9). Maternal blood glucose levels remained in the same range following choline supplementation in two cow studies including normal growth, FGR or overgrowth risk [132,152]. However, the two cysteine studies reported significant increases in maternal blood glucose levels in a streptozotocin-induced pre-existent diabetes mellitus (DM) type 1 mice model [107] and an overgrowth risk rat model using high fat diet [113].
Appendix B Figure A1. Flow chart of the study selection process. Our search strategy retrieved 17329 unique hits of which we included 111 studies reporting on amino acid (AA) supplementation in our systematic review. Of these, 63 studies reported on arginine supplementation, 11 on branched-chain amino acids (BCAA) supplementation and 38 on methyl donor supplementation. We pooled data on the effect of arginine supplementation on birth weight (BW) in 57 studies, on maternal blood pressure (BP) in 15 studies, and on small for gestational (SGA) development in risk populations in 8 studies. We were able to pool data on the effect of BCAA on BW in 10 studies and of methyl donor supplementation in 36 studies. Adapted from PRISMA [29].    Sharkey et al., as an influential case, was highlighted by the colour red [51].
Nutrients 2020, 12, x FOR PEER REVIEW 17 of 58 Figure A4. Meta-analysis on prenatal supplementation of arginine on maternal blood pressure. Blood pressure was unaffected in normal growth pregnancies following arginine supplementation, but was reduced in the risk population and complicated pregnancies. The data is ordered within each amino acid (AA) from smallest to largest animal. Blood pressure difference (BP diff) data represent pooled estimates expressed as mean difference (MD) with 95% confidence interval (CI) using a random effect model. Residual is shown. FGR, fetal growth restriction; I 2 , heterogeneity; PE, preeclampsia; PIH, pregnancy-induced hypertension. Figure A5. Species meta-regression of amino acids (AA) in arginine family on maternal blood pressure (BP) in pregnancy complications. Meta-regression revealed large interspecies differences, although only two human study cohorts versus seven rat study cohorts reported on the effect of prenatal supplementation of AA of the arginine family in complicated pregnancies. Data represent pooled estimates expressed as mean difference (MD) with 95% confidence interval (CI) using a random effect model. I 2 , heterogeneity. Figure A4. Meta-analysis on prenatal supplementation of arginine on maternal blood pressure: Blood pressure was unaffected in normal-growth pregnancies following arginine supplementation, but was reduced in the risk population and complicated pregnancies. The data is ordered within each amino acid (AA) from smallest to largest animal. Blood pressure difference (BP diff) data represent pooled estimates expressed as mean difference (MD) with 95% confidence interval (CI) using a random effect model. Residual is shown. FGR, fetal growth restriction; I 2 , heterogeneity; PE, preeclampsia; PIH, pregnancy-induced hypertension.
Nutrients 2020, 12, x FOR PEER REVIEW 17 of 58 Figure A4. Meta-analysis on prenatal supplementation of arginine on maternal blood pressure. Blood pressure was unaffected in normal growth pregnancies following arginine supplementation, but was reduced in the risk population and complicated pregnancies. The data is ordered within each amino acid (AA) from smallest to largest animal. Blood pressure difference (BP diff) data represent pooled estimates expressed as mean difference (MD) with 95% confidence interval (CI) using a random effect model. Residual is shown. FGR, fetal growth restriction; I 2 , heterogeneity; PE, preeclampsia; PIH, pregnancy-induced hypertension. Figure A5. Species meta-regression of amino acids (AA) in arginine family on maternal blood pressure (BP) in pregnancy complications. Meta-regression revealed large interspecies differences, although only two human study cohorts versus seven rat study cohorts reported on the effect of prenatal supplementation of AA of the arginine family in complicated pregnancies. Data represent pooled estimates expressed as mean difference (MD) with 95% confidence interval (CI) using a random effect model. I 2 , heterogeneity.                                  Fourteen amino acids during pregnancy: arginine, citrulline, glutamate, glutamine, asparagine, aspartic acid, proline, ornithine, (iso-)leucine, valine, cysteine, methionine, and choline. The search was performed on 25 July 2018, with 14,169 records.  Fourteen amino acids during pregnancy: arginine, citrulline, glutamate, glutamine, asparagine, aspartate, proline, ornithine, (iso-)leucine, valine, cysteine, methionine, and choline. The search was performed on 25 July 2018, with 10,393 records.   OR "aminoglutaric acid" OR "aminopentanedioic acid" OR acidogen or acidoride OR acidothym OR acidulin OR aciglumin OR aciglut OR aclor OR antalka OR flanithin OR gastuloric OR glusate OR glutadox OR glutamidin OR "glutamin acid" OR "glutaminic acid" OR glutaminol OR glutan OR glutansin OR glutasin OR glutaton OR hydrionic OR hypochylin OR levoglutamate OR "levoglutamic acid" OR muriamic OR pepsdol or glutamine OR "aminoglutaramic acid" or acutil OR "adamin G" OR glumin OR glutamin OR levoglutamide OR levoglutamine OR nutrestore or citrulline OR "ureidopentanoic acid" OR ureidonorvaline OR "ureidovaleric acid" OR citrullin OR carbamylornithine OR asparagine OR asparagin OR "aminosuccinamic acid" OR "aspartic acid" OR aspartate OR Magnesiocard OR Mg-5-Longoral OR Mg 5 Longoral OR Mg5 Longoral OR panangin OR astra 2045 OR "aminosuccinic acid" OR "asparagic acid" OR asparaginate OR "asparaginic acid" OR aspartyl OR aspatofort OR "levoaspartic acid" OR proline OR prolin OR levoproline OR "pyrrolidinecarboxylic acid"OR pyrrolidine carboxylate OR ornithine OR ornithin OR "Diaminopentanoic Acid" OR "diaminovaleric acid"  Edema Hypertension Gestosis OR Edema Proteinuria Hypertension Gestosis OR EPH Gestosis or EPH Toxemia * OR EPH Complex OR "placenta insufficiency" OR "placental insufficiency" OR "placenta insufficiencies" OR "placental insufficiencies" OR "placenta deficiency" OR "placental deficiency" OR "placenta deficiencies" OR "placental deficiencies" OR "placenta failure" OR "placental failure" OR Macrosomia * OR high birth weight OR "overweight infant" OR "overweight infants" OR "overweight newborn" OR "overweight newborns" OR "overweight neonate" OR "overweight neonates" OR diabetes gravidarum:ti,ab,kw Fourteen amino acids during pregnancy: arginine, citrulline, glutamate, glutamine, asparagine, aspartate, proline, ornithine, (iso-)leucine, valine, cysteine, methionine, and choline. The search was performed on 25 July 2018, with 819 records.    [134] Human Early onset severe PE/HELLP PE 0.025 a X X Shahin (2009) [137] Human Previous preterm labor Risk 0.008 a X X Methionine Abdel-Wanhab (1999) [115] Rat Control Normal 0.043 X Brunner (1978) [43] Rat PKU-induction FGR 1.570 b X X Control Normal 1.570 b X X Chandrashekar (1977) [116] Rat Control Normal NA X Fujii (1971) [117] Rat Control Normal 0.029 X  [138] Human Alcohol Risk 0.035 X X X X Ross (2013) [135] Human Control Normal 0.012 a X X X Yan (2012) [136] Human Control Normal 0.007 a X X Ordered according to species per amino acid. a Dose in g/kg/day was calculated using the estimated mean maternal weight or based on b estimated food intake. "Normal" in the pregnancy complication column indicates the normal-growth group. AA, amino acid; BP, blood pressure; BW, birth weight; dev. compl; development of pregnancy complication in risk population; DM, diabetes mellitus; EFW, estimated fetal weight; FGR, fetal growth restriction; GWG, gestational weight gain; L-NAME, (ω)-nitro-L-arginine methyl ester; LPS, lipopolysaccharides; NA: not applicable or available; HELLP, hemolysis, elevated liver enzyme, and low platelet syndrome; HTN, hypertension; PE, preeclampsia; PKU; phenylketonuria; PIH, pregnancy-induced hypertension; RUPP, reduced uterine pressure perfusion; UA, uterine artery; UOL, unilateral oviduct ligation; WT, wild type.      Ordered according to species form small to large per amino acid. a The supplementation scheme was continuous (C) or an interval (I). b Intervention type was treatment (T) or prevention (P). c When there is nothing in brackets reported in the table for sex of offspring (F or M) it means that the weight was measured in a mixed population. d Isonitrogenous control diet was used. e Dose in g/kg/day was calculated using the estimated mean maternal weight or based on f estimated food intake. "Normal" in the pregnancy complication column indicates the normal-growth group. AA, amino acid; BW, birth weight; DM, diabetes mellitus; EFW, estimated fetal weight; FGR, fetal growth restriction; GD, gestational day; L-NAME, (ω)-nitro-L-arginine methyl ester; LPS, lipopolysaccharides; NA: not applicable or available; HELLP, hemolysis, elevated liver enzyme, and low platelet syndrome; HTN, hypertension; P0, postnatal day 0 or day of P0; PE, preeclampsia; PKU; phenylketonuria; PIH, pregnancy-induced hypertension; RUPP, reduced uterine pressure perfusion; UA, uterine artery; UOL, unilateral oviduct ligation; WT, wild type; wk, weeks.  Ordered according to species per amino acid. a The supplementation scheme was continuous (C) or an interval (I). b Intervention type was treatment (T) or prevention (P). There were no studies using an isonitrogenous control diet. c Dose in g/kg/day was calculated using the estimated mean maternal weight or based on d estimated food intake. "Normal" in the pregnancy complication column indicates the normal-growth group. AC, abdominal circumference; BW, birth weight; DBP, diastolic blood pressure; GA, gestational age; FGR, fetal growth restriction; HTN, hypertension; IP, intraperitoneal; L-NAME, (ω)-nitro-L-arginine methyl ester; LPS, lipopolysaccharides; MAP, mean arterial pressure; Mg, magnesium; mo, months; NA, not applicable or available; P0, birth day; PE. preeclampsia; PIH, pregnancy-induced hypertension; RUPP, reduced uterine pressure perfusion; SBP, systolic blood pressure; SD, Sprague-Dawley; SHHF, spontaneous hypertension and heart failure; UA, umbilical artery; wks, weeks.    Ordered according to species per amino acid. a The supplementation scheme was continuous (C) or an interval (I). b Intervention type was treatment (T) or prevention (P). c Isonitrogenous control diet was used. d Dose in g/kg/day was calculated using the estimated mean maternal weight or based on e estimated food intake. "Normal" in the pregnancy complication column indicates the normal-growth group. BW, birth weight; FGR: fetal growth restriction; GD: gestational day; LPS, lipopolysaccharides; HTN, hypertension; PE: preeclampsia; PIH, pregnancy-induced hypertension; PKU; phenylketonuria; mo, month; NA: not applicable or available; P0, birth day; SD, Sprague Dawley; wk, wks.   Quality assessment using the SYRCLE risk of bias tool: The first five columns represent the reporting of key study quality indicators; the last ten columns entail the risk of bias assessment. Y = yes, reported; N = no, not reported; H = High risk of bias; L = Low risk of bias; ? = unclear risk of bias.