In the absence of proper regulation on the use of animals for research and animal welfare during experiments in Vietnam, the protocols were carried out according to the best practices usually accepted by the Ethical Committee of University of Liège (Belgium) when conducting similar experiments.

The RDP originated from a glutinous rice alcohol manufacturer in the Phu Loc village, Hai Duong province in the Red River Delta region in Vietnam. The RDP was obtained from rice distillation under traditional alcohol production process. Briefly, rice was heated over steam and then spread out on a flat surface to allow cooling. Yeast was then sprinkled over rice and both were well mixed. Preparation was placed in lidded buckets for incubation for about three days. The buckets were then filled with water and held for about one week. Finally, alcohol was distilled through a covered pot and steam ducted away to a water-cooled condenser. The residues were considered as RDP for pig feeding. The RDP samples were collected immediately after alcohol distillation for chemical analysis prior to diet formulation for the trial.

The study was carried out from January to March 2017 at a private farm specialized in pig production, in Hai Duong province, located 40 km from the Vietnam National University of Agriculture (VNUA), Hanoi, Vietnam. A total of 48 healthy weaned castrated male crossbred pigs (♂Duroc×♀[Landrace× Yorkshire]), originating from 12 sows to 4 male piglets per sow, 3 to 5 litter order, initial body weight (IBW) 7.54±0.97 kg (mean±standard deviation), and age about 4 wks, were used in this experiment. The piglets were individually ear tag numbered and randomly allocated into three diet-groups of four blocks according to similar IBW and sow origin by treatment. There were four replicate pens (block) per treatment and 4 pigs per concrete pen (3.6×1.1 m) with a height of 0.6 m, each pen equipped with stainless steel feeding trough and 2 automatic water-drinking nipples. In each pen, a (1.2×0.6 m) plywood piece and a 175W red infrared heat lamp bulb were used in order to maintain effective temperature for pig development. The room temperature was kept at approximately 30°C the first week, and afterwards was decreased by 1°C each week of the experiment, suitable for animal development. Piglets were fed one of three diets (a control diet without RDP, RDP0; and two experimental diets with 15%, RDP15; and 30%, RDP30; RDP on a DM basis) during a total of 35-d experiment for 2 stages (early phase, from d 0 to d 14; and late phase from d 15 to d 35). Chromium oxide (Cr₂O₃) added at 5 g/kg dry matter (DM) of diet was used as digesta flow marker to measure digestibility. On d 14 and d 35, two pigs per pen were randomly selected and transported 4 hours after feeding from the farm to Faculty of Veterinary Medicine, VNUA (about one hour quite travel in suitable vehicle). These pigs rested for one hour and were then killed for collection of digesta and intestinal tissue samples.

Raw feed ingredients were purchased all at once from a local feed company, except RDP which was delivered daily from the manufacturer. Feed was ground into flour through a 2 mm screen before mixing. The ingredients were formulated in order to obtain three iso-nitrogenous experimental diets. The RDP0 mainly consisted of corn, soybean, rice, fishmeal, and their by-products. Other diets (RDP15 and RDP30) were formulated separately with the same ingredients as the control diet, until RDP was extemporaneously included at 15% or 30% in the diets, on a DM basis (Table 1).

At each feeding, RDP0 diet was weighed and moistened with drinking water (pH 6.73) at a 1:1 ratio in order to prevent losses and to facilitate intake, and the two other ones were mixed with RDP. Prior to diet allocation, feed pH was measured, as possible covariate factor of other parameters such as palatability. Samples of RDP and complete diets at feeding were collected one time per week and stored at −20°C until the end of the experiment where they were thawed and homogenized for chemical analysis (Table 1). Daily feed allowance was divided into four equal amounts and was offered at 06:00, 11:00, 16:00, and 20:00 h during the entire experimental period. Diet amounts were increased gradually parallel to piglet development, at about 4% of live weight. Refusals were collected just before each feeding, weighed, stored at −20°C until the end of the experimental period where they were thawed and homogenized for DM analysis. Pigs always had free access to water by nipple drinkers.

The DM of RDP, diets, and digesta isolated from ileum were dried by oven drying at 70°C for 15 h, 90°C for 5 h and 102°C for 5 h consecutively, and they were milled separately through a 1 mm screen prior to analysis. The DM of feed ingredient was determined according to Method 934.01 from AOAC [27]. Feed ingredients and diets were analyzed for CP (Method 954.01, AOAC, 1990), ether extract (EE; Method 920.39, AOAC, 1990) with petroleum ether solvent, ash (Method 942.05, AOAC, 1990), crude fiber (CF; Method 962.09, AOAC, 1990), neutral detergent fiber and acid detergent fiber (ADF) (Method 973.18, AOAC, 1990) with fiber filter bags of Ankom technology F57, phosphorus (P; Method 965.17, AOAC, 1990) using a UV-vis spectrophotometer. Calcium (Ca) was determined by titration with a standardized solution of EDTA as previously described [28], and starch was determined from estimation of reducing sugars by dinitrosalicylic acid method under spectrophotometer UV-1800 (Kyoto, Japan) as previously described [29]. Chromium was analyzed using UV absorption spectrophotometry (UV-1800, Japan) according to [30]. The GE of samples were measured using a bomb calorimeter E2K (Germany). The pH value of RDP was determined using an electrode of a portable pH meter (HANNA, Singapore). Organic acids (OA) were determined by Vinger method (Ngoan, 2002). Ethanol in RDP was determined using high performance liquid chromatography (Agilen 1200 series; Agilent Technologies, Santa Clara, CA, USA) with Aminex HPX-87H column, RI detector, phase mobile including H₂SO₄ 10 mM; mobile phase flow rate was 0.5 mL/min, the column temperature was maintained at 60°C.

Data for growth performance and blood profiles were analyzed using the PROC MIXED procedure of SAS software (Version 9.4, Institute Inc., Cary, NC, USA). The statistical model included the diets (n = 3) as fixed effect, and the blocks (n = 4) as random effects. Pen was used as experimental unit for the performance data, and individual pig as experimental unit for visceral organ weights, gut pH and blood profiles. For repeated measures performed on the same experimental unit a similar model was used but including the effect of a compound symmetry structure of covariance. Orthogonal polynomials were performed to determine linear and quadratic effects of increasing level of the RDP in diets [31]. Data for pigs fed diets containing RDP were compared with data for pigs fed control diet using orthogonal contrasts. The multiple comparisons of least square means were performed according to PDIFF option. Significance was defined as p< 0.05 and 0.05<p<0.10 was considered as a trend. Due to unintentional feces and digesta pooling by diet, no statistical analysis could be performed on digestibility data.

Data for microbiota composition in the colon, the statis tical differences in bacterial diversity, bacterial richness and bacterial evenness between pig fed control diet and diets containing RDP were performed using Kruskal-Wallis with Benjamini, Krieger, and Yekutieli test (PRISM 7.0, Graph-Pad software, La Jolla, CA, USA). Beta diversity (Bray-Curtis dissimilarity) was compared using Kruskal-Wallis with Bonferroni correction (p≤0.003). To compare statistical differences in bacterial community abundance between diets, a non-parametric Kruskal-Wallis H tests was combined to a Storey False Discovery Rate followed by Tukey-Kramer post-hoc test (using STAMP 2.1.3 software) as previously described [32]. The significant level used for statistical tests was 0.05.

With inclusion levels of RDP, a linear trend (p = 0.07) for increasing ADG was observed in the two experimental periods, and FCR tended to decrease in the second one (p = 0.08) (Table 2). Final body weight (FBW) at d 35 thus increased linearly with dietary RDP inclusion. Besides, using orthogonal contrasts, increase in FBW was observed (p<0.05) when pigs fed RDP diets were compared with control pigs. As a rule, no significant differences in visceral organ weights were observed between groups, but heart weight at d 14 and empty stomach weight at d 35 were negatively and positively impacted, respectively, by RDP level (p = 0.04 and p = 0.03) (Table 3). With level of RDP incorporation, weights of visceral organs—except gut-relative to live weight decreased on d 14 so far that their sum decreased by 4.4 and 11.6%, in RDP15 and RDP30, respectively, when compared to the control group.

When regard to pH values of stomach, ileum, caecum and colon contents, no differences between groups were observed at both d 14 and d 35. However, the pH values decreased numerically in all segments at d 14 when increasing inclusion of RDP in diet (Table 4).

The AID and ATTD of CP, lipids, and gross energy numer ically decrease with inclusion of RDP, especially for energy and lipids (Figure 1).

On d 14, a linear decrease in amount of globulin was observed (p = 0.003) in pigs fed RDP diets when compared with pigs fed control diet. Quadratic effects were observed (p≤0.04) for WBC, lymphocyte, and albumin. On d 35, a trend for linear effect for RBC (p = 0.06) and a trend for quadratic effect for Hb (p = 0.09) were observed between diet groups. Other hematological parameters were not affect by pigs fed RDP diets compared with pigs fed control diet (Table 5).

The analysis of alpha diversity of colonic bacterial population showed that no statistical differences were found (p>0.05) between weaned pigs fed RDP diets and pigs fed control diet within the same period, as estimated by diversity indices including reciprocal Simpson Biodiversity, Chao1 richness, and Simpson Evenness (Figure 2). However, bacterial richness of colonic microbiota was greater in RDP30 on d 35 than on d 14 (p = 0.01). Beta diversity differed between diet groups as indicated by clear clustering on principal coordinates of microbial profiles on d 14 and d 35. Indeed, colonic microbiota communities differed between RDP0 and RDP30 on d 14 and d 35 (p = 0.002, and p<0.001), and between RDP15 and RDP30 on d 35 (p = 0.002).

Differences in colonic bacterial compositions at taxonomic levels (family and genus) are presented in Figure 3 and 4. When comparing relative abundance of family of diets in a contemporary, proportion of Lachnospiraceae in RDP30 was lower on d 14 (p = 0.002) and greater on d 35 (p<0.001). Proportion of Ruminococcaceae was greater in RDP0 (p<0.001) on d 14, whilst greater in RDP30 on d 35 (p≤0.03). Proportion of Bacteroidales_fa was lower in RDP15 on d 14 (p≤0.04), and greater in RDP30 (p<0.001) on d 35. And proportion of Prevotellaceae increased (p≤0.01) on d 14 and decreased (p≤0.05) on d 35 according to inclusion levels of RDP (Supplementary Table S1).

When comparing relative abundance of genus of diets in a contemporary, Lachnospiraceae_ge in RDP30 was respectively lower (p = 0.008) and greater (p<0.001) on d 14 and d 35. Ruminococcaceae spp. was greater in RDP0 on d 14, while greater in RDP30 on d 35 (p<0.001). Bacteroidales was lower in RDP15 on d 14 (p≤0.002), whilst greater in RDP30 on d 35 (p<0.001). Prevotellaceae increased with inclusion of RDP on both d 14 and d 35 (p<0.001). Prevotellaceae_NK3B31_group was greater in RDP15 on d 14 (p≤0.001) and in RDP30 on d 35 (p<0.001). Prevotella_1 was lower in DAR15 on d 14 (p≤0.02). Increases in Prevotella_2 and Prevotella_9 on d 14 (p≤0.04) and an inverse tendency on d 35 (p<0.001) were found with inclusion level of RDP (Supplementary Table S2).

Rice by-products classically are incorporated in diets for pig in Asia. Thus, one could argue some confounding effects between RDP levels and the other rice ingredients of the diets. Confounding bias cannot be completely avoided in such experiments. The main constraint aimed to preserve iso-nitrogenous characteristics of the diets. The RDP showed a high CP level, this constraint only could be fulfilled thanks to decreases in soybean by-products levels. With CF content close to 0.35%, broken rice could be considered to maintain similar levels of starch between the groups. Rice bran helped to compensate for CF provided by RDP. In this context, it should be kept in mind that the level of RDP incorporation in the diets was the main—and thus not the exclusive—factor of variation.

In the first three days of the experiment, loose feces appeared on some experimental pens due to digestive disorders, this phenomenon resolved few days after feed intake amount was reduced. Therefore, diarrhea in piglets was not evaluated during the entire experimental period. All piglets in our experiment remained in good health, probably due to a good set-up with clean environment during the experimental period. Increasing inclusion level of RDP did not affect DM intake, and tended to improve ADG during d 0 to d 14 and d 15 to d 35, leading to higher final live weights in animals fed RDP. Interestingly, feed efficiency was improved only from d 15 to d 35. According to earlier studies, piglets fed diets with 10%, 15%, 20%, and 30% RDP had enhanced ADG and FCR compared with piglets fed control diet, but without effect in DM intake during a 42-d experiment [2,13], suggesting that RDP does not affect diet palatability in weaned pigs. Lower pH and higher lactic and acetic contents in RDP product could be beneficial to the gut health of piglets through a decrease in number of harmful bacteria such as Escherichia coli and total coliforms, and an increase in number of beneficial bacteria such as lactic acid bacteria [13]. This has been shown for pigs fed fermented diets [13,33]. In addition, short chain OA, mainly acetate, propionate, butyrate and lactate are produced in large intestine by fermentation of undigested protein and fiber fractions. These OA had deep effects on metabolism and gut health [34]. Acetate and propionate are energy substrates for lipogenesis and gluconeogenesis, respectively, and butyrate is used primarily by colonocytes as major energy source for their metabolic activities in pigs [35–38]. Hong et al [13] demonstrated that the concentrations of acetic acid and lactic acid in colonic digesta were greater for piglets fed RDP diet than those fed diet without RDP. Thus, our results suggest that the positive effect on growth performance of weaned piglets fed RDP diets may be associated with modulation of OA-production bacteria due to various diet components. In addition, no differences in almost all visceral organ weights among diet groups was observed. This is similar to previous report from [6] who demonstrated that there were no differences in heart, lung, liver, spleen, stomach, kidney, small intestine, and large intestine weights in growing pigs fed diets with different inclusion levels of RDP (0%, 7.5%, 15%, 22.5%, 30% DM). However, an increase in relative weight of empty stomach on d 35 was found with RDP diets, which could be due to soluble fiber content of RDP originated from glutinous rice [39] leading to increase in transit time of diets [40] from stomach to intestine. This could increase digestive fluids secretion and activity for breaking down feed that resulted in stomach enlargement [41]. Lower heart weight of pigs fed RDP diets on only d 14 is unclear and possibly due to fortuitous statistical significance but the more than 4% to 11% numerical decreases of the overall visceral—but stomach and intestines—proportions of live weight is noticeable. Such decreases suggest a more efficient viscera metabolism, may be consecutive to ready-to-use metabolites stemming from fiber fermentation, reducing thus metabolic burden for these organs. This hypothesis would merit to be further investigated.

Recent study [13] reported that a greater AID and a similar ATTD of CP was found for weaned pigs fed diet with 20% RDP than pigs fed control diet for a 42-d experiment. Moreover, piglets fed pectin-containing diet showed higher digesta viscosity with decrease ileal digestibility of protein [42]. Thus, and in the limit of variance indicators failure, the numerical decrease in AID and ATTD of CP in RDP diets could be related to high pectin content. The decrease in ATTD of CP with RDP inclusion could indicate a partial nitrogen flux shift from plasma (urea) to hindgut lumen, thanks to nitrogen conversion to microbial protein, the one thereafter excreted in feces. Additionally, a decrease in AID and ATTD of EE with inclusion of RDP, as observed in this study, has never been reported. According to [43] piglet diets supplemented with pectin could have negative effect on fat digestibility. A other study [44] revealed that weaning pigs fed diets containing OA had changed intestinal microbiota composition, with decreased amount of bile acid production, thus leading to negative impact on fat digestion. The OA issued from RDP diets possibly contributed to alter EE CATTD.

Effect of RDP incorporation in weaned pig diet on blood profiles has not been published yet. This study found no effect of RDP on hematological profiles such as WBC, Hb, and lymphocyte percentage, and biochemical parameters such as serum concentrations of albumin, IgG, and IgM, except serum globulin concentration on d 14 and RBC count on d 35. Despite decreased RBC and globulin values with inclusion of RDP, all results were found to be within the normal ranges for weaned pigs, as reported previously [45–48]. The difference in values over time were considered as normal owing to the pig developmental stage [40]. Additionally, according to [44], IgA, IgG, and IgM concentrations were not influenced by supplementation of OA in weaning piglet diets, which in line with our results. The largest immunologically competent organ in the body is gut or its associated lymphoid system, and the development of immune system is associated with composition of indigenous micro flora [49]. With inclusion of RDP, the decrease in serum globulin concentration on d 14 and of RBC count on d 34 was not clear and could be due to stochastic or incidental effects without relationships with RDP. Further studies are required to drain more conclusion.

Intestinal microbiota, that affect nutrient metabolism and immune system development, play a significant role in intestine and host health of pigs [50]. Gut microbiota composition is affected by diet composition, in particular fiber components [51]. In this study, we compared the microbial population structure and microbiota composition from colonic digesta of the weaned pigs. To the best of our knowledge, this is the first report to evaluate the effect of RDP inclusion on the colonic microbiota of weaned pigs using 16S rRNA gene sequence analysis. On d 35, colonic microbial richness increased in piglets fed RDP30 in the later phase of feeding trial, suggesting that inclusion of 30% RDP may stimulate the growth of colonic bacteria in a long-term intervention. Moreover, beta diversity analysis indicated different clustering of microbial community structure between RDP diets and control diet. In addition, the microbial compositions significantly varied with inclusion of RDP at family and genus levels. The change in relative abundance of colonic microbiota composition could be related to the chemical components of RDP, especially fiber fraction. Most of the time, the composition of bacterial communities in colonic contents presented opposite evolution when comparing d 14 and d 35 values. This could due to an adaption phase of immature digestive tract of weaned piglets. An increase in relative abundance of Prevotellaceae_NK3B31_group was found in pigs fed diets containing high component of arabinoxylan on d 28 [52], which also was found in our results. Besides that, a considerable decrease in Prevotella_1, Prevotella_2, and Prevotella_9 proportions was observed in piglets fed RDP30 on d 35. The reason for this variation might be associated with increased microbiota compositions of Prevotellaceae_ge, Prevotellaceae_NK3B31_group. Indeed, arabinoxylan fiber, an important content of plan cell walls of rice, is considered as a prebiotic effect on intestinal health of pigs stimulating the growth of the latter bacteria genera [53,54]. Moreover, members of Lachnospiraceae family —whose proportion increased at d 35 in our experiment—was positively correlated with intestinal epithelial cell energy metabolism and butyrate production [55, 56]. Ruminococcacecae spp. also was related to butyrate production [55]. Finally, a higher abundance of Bacteroides spp., Lachnospiraceae spp., and Ruminococcacecae spp. was linked to obese animals [57–59]. All these arguments are in line with the better performances observed at the end of the experimental period in our experiment. Also, piglets fed RDP diet had a higher acetic acid concentration in colon [13], and butyrate-producing bacteria are able to use acetate [60]. For example, a positive correlation between concentration of acetic acid and butyric acid in colon of pigs fed diets with increasing inclusion of chicory forage was reported by a previous finding [60], and [61] reported that OA induced adipose phenotypes. Thus, RDP diets alter not only composition of colonic microbiota but also microbiota metabolism. In addition, [51] found that microbiota can improve host fat storage. The OA controlled metabolic regulation by signal via G-protein-coupled receptors (GPCRs), such as GPR41-relative to the adipose tissues, and effects of the microbiota on fat deposition depends on this OA receptor [62]. However, further studies are needed to clarify the effects of microbiota composition on colonic gene expression profiling.