Background

When we prescribed a low-protein diet to the first pregnant patient with severe proteinuria and diabetic nephropathy (a case which gave us the opportunity to start a “joint venture” between Nephrology and Obstetrics), we did not foresee that fifteen years later our Unit would have followed-up a few hundred pregnancies, about 5 % of which involved subjects on a protein-restricted diet [1–5]. We also did not foresee that several large studies would have challenged the “meat eaters” in favour of Mediterranean or vegetable-based diets, thus leading to reconsider the role of protein intake in the overall population, as well as in CKD [6–14]. We were mainly worried about the patient’s increasing levels of proteinuria, and we did not know what else we could do besides keeping blood pressure under control, ordering bed rest (still a widely used procedure) and checking the baby’s growth curve [1].

On the basis of the available data on hyper-filtration in CKD and on the effect of low-protein diets in reducing the “work load” on the remnant nephrons, we chose to start her on the diet that we considered the “best” one available in our hands, i.e. a low-protein, vegan, supplemented diet [15–18].

After our patient delivered a healthy male baby, adequate for gestational age, at the 30th gestational week, we started prescribing a low-protein, vegan-vegetarian diet, with a simplified qualitative schema, to other pregnant patients with severe kidney function impairment or relevant proteinuria [1]. Our first results, involving 12 pregnancies, were promising enough to double the number of patients in a few years [2, 4]. The subsequent analysis on 22 live-born singleton deliveries showed the almost paradoxical finding of better growth in children delivered by on-diet mothers as compared to children of CKD mothers on an unrestricted diet [4].

At the time of our first experiences, 1–1.2 g of proteins /Kg day was considered the “normal” protein intake, and the intake in pregnant women was often higher, thus making our diets conflicting with the common beliefs in pregnancy. However, interest in vegan-vegetarian diets grew over the following years, and they are now considered safe in all phases of life, including pregnancy and lactation, provided that vitamins and microelements were controlled and integrated when needed [19–31].

Meanwhile, we gradually integrated the recommendation that patients should avoid both excessive weight gain; this was carried out by shifting from a purely qualitative diet prescription to the present qualitative-quantitative one [2, 4] (Appendix).

The main drawback of our previous studies was the difficulty of recruiting a homogeneous control group [2, 4]. Thus, the novelty of the present analysis, which is aimed at reviewing the results gathered over 15 years, is that the results of on-diet pregnancies are compared to a composite larger control group of pregnancies with similar clinical characteristics.

Methods

Definitions and control policies

CKD was defined and staged according to K-DOQI guidelines, whenever possible according to pre-conceptional data. Throughout pregnancy, GFR and proteinuria were assessed by 24-h urine collections, as specified more in detail elsewhere [5].

A newborn was defined as Small for Gestational Age (SGA) when birth weight was below the 5th or below the 10th centile, according to the birth weight references that were used [32–34]. Due to the specific interest in this point, we employed both the older Italian Parazzini charts and the newer INeS (Italian Neonatal Study) charts, and analysed the two cut-points at the 5th and 10th percentile [33, 34]. Preterm delivery, early preterm delivery and extremely preterm delivery were defined as before 37, 34 and 28 completed weeks of gestational age, respectively [32].

Hypertension was defined as per the current guidelines; the antihypertensive treatment was mainly based upon a combination of alphamethyl-dopa and nifedipine, adding doxazosine, small doses of diuretics or clonidine only when absolutely needed. Treatment was adjusted at every clinical visit with a target of 120–130/60–70 mmHg [5].

The study was performed in two Italian settings: Torino and Cagliari. These are the two Centers with the greatest experience of management of CKD in pregnancy in Italy, that keep a conjoint database (TOCOS: Torino Cagliari Observational Study [5]). For the sake of this study, the cases were recruited in Torino, the controls were selected in both settings, as further specified. In both settings of care, the frequency of nephrological and obstetric visits, of blood and urine tests and of biometric and Doppler studies of uterine and umbilical arteries are tailored to the individual patient (visits: 1 week–1 month, biometry every 2–3 weeks in case of SGA babies or at risk for foetal growth restriction; Doppler assessment two-three times weekly in case of Doppler anomalies), in keeping with the Italian best practices in pregnant CKD patients [35, 36].

The low-protein diet

The low protein diet consisted in an adaptation of the low-protein vegan diet employed in our centre, itself a simplification of the original scheme by Barsotti and Giovannetti [17, 18].

Unlike the Barsotti and Giovannetti diets, our basic schemas are simplified: the food is chosen according to a qualitative approach (allowed-forbidden), not weighed, with a protein intake of 0.6 g/Kg/day (ideal weight), and 1–3 free meals per week. To allow the patient to follow a vegan diet without the need to use legumes and cereals in each meal, we added a supplementation of alpha-keto analogues and aminoacids (Alpha-Kappa or Ketosteril according to the availability over time): 1 pill/10 Kg of ideal body weight [37, 38].

In an empirical attempt to balance the potential advantages of low-protein diets in CKD and the habit of increasing protein intake in pregnancy, we initially adjusted the diet from 0.6 to 0.6–0.8 g/Kg/day of proteins, based on pre-conception weight, usually by increasing the protein intake from the first (0.6 g/Kg/day) to the last trimester (0.8 g/Kg/day). We also increased the amino and keto-acids supplementation from 1 pill each 10 Kg to 1 pill each 8 Kg, and in patients with low body weight, even up to 1 pill each 5 Kg in late pregnancy.

At the time of the first case, no report on these issues had been found or made available by the company; no report on safety concerns was available at that time or was found at the subsequent updates.

Since patients often missed milk and yoghurt in their diets, we allowed small quantities (100–150 mL per day) in selected cases, and changed the definition of “vegan” into “vegan-vegetarian”. On the basis of the functional status, of the proteinuria levels and of the patients’ needs and preferences, in keeping with the policy applied to non-pregnant patients, we allowed 1–3 unrestricted meals per week (without protein restriction but limited in unsaturated fats and short-chain sugars).

On the account of the lack of indications on salt restriction in pregnancy, we did not restrict salt; since salt intake cannot be controlled by the analysis of the 24 h excretion in pregnancy, due to the lack of referral standards, we limited our interventions to diet counselling in the cases with severe oedema or uncontrolled hypertension.

In addition to the biochemical tests (targeted at CKD), we progressively added iron status, B12, and 25-OH vitamin D to the routine monthly tests; vitamins and iron supplements were employed on the basis of the biochemical results. Erythropoietin was used when needed, with a haemoglobin target of 10 g/dL on account of the physiological haemodilution of pregnancy.

The most recent version of the diet is reported in the Appendix.

Indications for the diet and selection of controls

The main indications for the low-protein vegan-vegetarian diets in pregnancy were progressively broadened from the initial subjects with CKD stages 4-5 and/or nephrotic syndrome to include pregnancy in patients already on a supplemented vegetarian diet; CKD stages 3b or 3 with a progression trend before or during pregnancy; proteinuria above 3 g/day at any time of pregnancy, or proteinuria above 1 g/day at referral or in the first trimester, previous nephrotic syndrome, increase or development of proteinuria without any sign of preeclampsia, or a combination of any of these elements.

The controls were selected according to the same criteria from the Torino and Cagliari cohort. While the nephrologists’ approach was very similar, in keeping with our well-established cooperation, the Torino and Cagliari Units differed with regard to the Obstetric policy towards caesarean sections (more frequently performed in Cagliari [5]), therefore this outcome was not considered in the present study.

Statistical analysis

Descriptive analysis was performed as appropriate (mean and standard deviation for parametric and median and range for non-parametric data). Paired T-test, Chi-square test, Fisher’s test, Mid-p test, and Wilcoxon’s test were used for comparisons between patients and controls and to evaluate the differences from referral to delivery in patients and controls. Significance was set at <0.05.

Statistical evaluation was performed using SPSS vers18.0 for Windows (SPSS Chicago Ill, USA).

Ethical issues

Systematic counselling about the diet was provided. Patients were informed that few data on the supplemented diet during pregnancy were available outside of our group, furthermore, the limits and goals of the low-protein diets were extensively discussed. The importance of timely reporting of side effects or doubts was underlined; a written schema, progressively updated, was supplied. The first version is available elsewhere [5]. The most recent update is available in the Appendix.

The study was approved by the Ethics committee of the OIRM Sant’Anna (n° pratica 335; n° protocollo 11551/c28.2 del 4/3/2011). All patients signed a dedicate informed consent.

Results

Baseline data

The main baseline data of the 36 patients who followed the diet for at least one month and of the 31 patients who delivered a live-born singleton baby (excluded: 3 twin deliveries, 1 pregnancy termination following the mother’s wishes, 1 spontaneous miscarriage) are reported in Table 1. Two patients in the on-diet group undertook two pregnancies.

Table 1 Baseline data: “On-diet”: 36 pregnancies in patients who followed a supplemented vegan diet in pregnancy (31 singleton deliveries)
Full size table

Table 2 reports the baseline data in the control group of 47 pregnancies homogeneously selected according in Torino and Cagliari; there were 42 singleton deliveries and 5 spontaneous miscarriages.

Table 2 Baseline data: “controls”: 47 pregnant patients on unrestricted diet in pregnancy (22 singleton deliveries in Cagliari, 20 in Torino)
Full size table

The two groups are homogeneous with regard to the main clinical parameters: age (singletons only: on diet: 33 vs controls 33.5 years); and referral week (7 vs 9 weeks). CKD stage was non significantly lower in on diet patients (CKD 3-5: 48.4 % vs 64.3 %, p: 0.26), conversely, prevalence of hypertension was non significantly higher (51.6 % vs 40.5 %, p: 0.48). Nephrotic range proteinuria (16.1 % vs 12.2 %, p 0.74) was also non significantly higher in on diet patients. The combination of hypertension and proteinuria was present in 14/36 (38.9 %) on-diet patients and in 14/47 (29.8 %) controls (p = 0.35). There were more diabetic nephropathies in on-diet patients (on diet: 31 % vs controls: 5.3 %; p: 0.007) while lupus nephropathies were non-significantly higher in controls (on diet: 10.3 % vs controls 23.7 %; p: 0.28 (Fisher)), presumably as a reflection of the referral pattern of the individual Nephrology Units.

Pregnancy outcomes: kidney function and proteinuria

All of the patients on the diet followed it throughout pregnancy; no diet- or supplement- related side effects were reported and abdominal discomfort, when present, was not considered related to the diet itself. According to dietary recall, compliance was good; however, especially in the second period, in which the diet was more detailed and no more merely qualitative, some patients complained that it was very intrusive in their daily life.

An increase in serum creatinine leading to a shift towards a higher CKD stage was observed in 19.4 % on-diet and 9.5 % controls (p: 0.2 (Fisher)).

Proteinuria increased significantly in both patients and controls (new onset or doubling of proteinuria: 54.8 % of on-diet subjects and 50 % of controls; p: 0.5 (Fisher)). However, serum albumin and total proteins only moderately, and non significantly decreased at the end of pregnancy (diet group: total proteins: 6.5 g/dL at start vs 5.7 g/dL at delivery, albumin 3.75 g/dL at start vs 2.9 g/dL at delivery; control group: total proteins: 6.5 vs 6.1 g/dL, albumin 3.4 vs 3.24 g/dL) (Tables 3 and 4).

Table 3 Maternal data at delivery: “on-diet”: 31 singleton deliveries and 3 twin deliveries
Full size table
Table 4 Maternal data at delivery: “controls”: 42 singleton deliveries
Full size table

At 3 months after delivery serum creatinine increased and GFR decreased in both groups, in keeping with the reversal of pregnancy-related hyperfiltration. The decrease in proteinuria is probably due both to the reversal of the hyperflitraton phase, but other less known pregnancy-related permeability changes mechanisms may also play a role (Tables 3 and 4), Figs. 1, 2 and 3.

Fig. 1
figure 1

Performance of serum creatinine in on diet patients and controls

Full size image
Fig. 2
figure 2

Performance of GFR in on diet patients and controls

Full size image
Fig. 3
figure 3

Performance of proteinuria in on diet patients and controls

Full size image

Pregnancy outcomes: prevalence of small for gestational age and preterm babies

Tables 5 and 6 report the main data regarding birth-weight and timing of delivery in on-diet patients and in controls. No significant differences were observed for the overall prevalence of preterm delivery (<37 completed gestational weeks), which was over 70 % in both groups (on-diet singletons 77.4 %; controls 71.4 %; p: 0.76), or in the prevalence of children with birth-weight at or below 2.5 Kg (21/31: 66.7 % vs 25/42: 59.5 %, p: 0.32).

Table 5 Main Maternal-foetal outcomes and intrauterine growth: “on-diet”: 31 singleton deliveries and 3 twin deliveries
Full size table
Table 6 Main Maternal-foetal outcomes, and intrauterine growth: “controls”: 42 singleton deliveries
Full size table

The Figs. 4 and 5, based upon the original Parazzini charts that were the most commonly used references in Italy throughout the study period, summarize the relationship between birth-weight and prematurity in the two settings. Early preterm delivery (on diet: 32.3 % vs controls: 35.7 %) and extremely low birth-weight (on diet: 9.7 % vs controls: 23.8 %) were more common in control groups, and the only two extremely preterm deliveries were observed in the control group (p: 0.505).

Fig. 4
figure 4

Relationship between birth-weight and prematurity in on diet patients and controls: females

Full size image
Fig. 5
figure 5

Relationship between birth-weight and prematurity in on diet patients and controls: males

Full size image

The birth-weighty centiles, assessed by the Parazzini chart, reference in most of the period of study, showed a lower prevalence of babies below the 10th centile or extremely preterm (below 28 weeks) in on diet patients versus in controls; the difference (one tailed Fisher exact test) reaches statistical significance (12.9 % vs 33.3 % p: 0.04). If centiles are calculated with INeS charts, the figures are similar (below 10th centile: 9.7 % on diet vs 26.2 % controls, but the difference doesn’t reach statistical significance (p: 0.068)).

Conversely, gestational age and birth weight did not differ in the two cohorts (Tables 5 and 6 and Figs. 1 and 2). One twin child diet of on-diet mother died (cerebral haemorrhage after heart surgery for cardiac malformation); none of the singletons died in the on-diet series, while two perinatal deaths occurred in the control group (p = 0.505).

Discussion

An often-cited quote by Feuerbach states: “a man is what he eats”; indeed there are good reasons to reflect on Feuerbach’s clever and polemic sentence in the era of epigenetics and of rediscovery of the importance of what we eat to prevent diseases and possibly to cure them.

Low protein diets are a well-known tool for contrasting absolute or relative hyperfiltration in the case of nephrotic syndrome or diabetes, and in the remnant nephrons in CKD patients [14–16, 39–41]. Pregnancy is another well-acknowledged condition of physiological hyperfiltration, which may exert a negative effect on kidney function or increase proteinuria in CKD patients [42–47].

Control of hyperfiltration and of proteinuria were the potential advantages we hoped to achieve by a low-protein diet in pregnancy, when this experience started, at a time when pregnancy in CKD was often discouraged and the common practice was to increase protein intake in pregnancy [48].

Almost unexpectedly, the finding of equivalent or better foetal growth in on-diet patients shifted our attention from the maternal kidneys to the maternal-foetal exchanges, suggesting a potential effect on the utero-placental axis [4]. While the low numbers, and the lack of a homogeneous control group limited the interest in our findings, this larger cohort with a well-matched larger control group may allow us to refine the previous results.

Similarly to our previous studies, in the present series there is a trend towards better preserved foetal growth, that reaches statistical significance for the combined outcome of extremely preterm delivery and small for gestational age baby (below the 10th centile) (Tables 5 and 6). Preterm delivery was over 70 % in cases and controls, witnessing the relevance of the renal impairment; such prevalence is in line with available studies on patients with advanced CKD [3, 5, 49–51].

In our analysis the differences between cases and controls regard the “harder” and partially overlapping outcomes, which include early preterm delivery, small for gestational age (SGA) and extreme preterm babies, “very small” babies (birth-weight is at or below1,500 g). The lower incidence of SGA has to be contextualised with the similar incidence of early preterm delivery (32.3 % vs 35.7 % in controls), since SGA is a reason for anticipating delivery [52]. This reinforces our previous findings, of a better foetal growth in children of on-diet CKD mothers (Figs. 1 and 2, Tables 5 and 6).

Our study has several limitations, which are partly shared by other studies on pregnancy: first of all, it is not randomised. However, randomization of the diet is hardly feasible outside of pregnancy and may be ethically unsound in pregnancy.

Secondly, we deal with a small number of patients, even if ours is the only study to date dedicated to this issue in CKD pregnancies.

Further research, involving a greater number of subjects is needed to highlight the differences suggested by our studies and to analyse placental vascularization and development, thus possibly offering insights into the pathogenesis of adverse pregnancy-related outcomes in CKD mothers. Theoretically, a positive effect could be due to a decrease in “vaso-toxic” elements or to an increase in “vaso-protective” ones; both are present in the study diet. A growing amount of data suggests that red meat consumption is associated with an increase in cardiovascular risk, while diets that are rich in vegetables, legumes and grains (especially those with a low glycaemic index) may be protective against endothelial dysfunction [53–63].

The specific advantage of vegetable proteins and of supplementation with ketoacids may have played an important role, as it has been suggested in experimental models, which show a protective endothelial effect of ketoacids in rats with kidney disease and a decrease in the risk of CKD in the offspring of rats with genetic kidney diseases that are fed a soya rich diet [64, 65].

In the absence of a randomised controlled trial that could present ethical limitations in pregnancy, we hope that our data may stimulate new research on this important issue.

Conclusion

Vegan-vegetarian diets with moderate protein restriction, supplemented with amino and keto-acids, are safe in pregnancy and may be followed without appreciable side effects. A favourable trend towards improving foetal outcomes was observed for growth and timing of delivery, and reached statistical significance for the combined outcome of small for gestational age babies and extremely preterm delivery, which are also the most robust predictors of future health.

While waiting for further studies to highlight the underlying mechanisms, we hope that this positive finding may raise awareness to the important issue of diet, CKD and pregnancy.