Meta-analysis of the effects of the dietary application of exogenous alpha-amylase preparations on performance, nutrient digestibility, and rumen fermentation of lactating dairy cows

Abstract

Several studies have evaluated the effects of the dietary application of exogenous alpha-amylase preparations (AMA) as a strategy to increase total tract starch digestibility (TTSD) and milk yield (MY) in dairy cows, but the results have been inconsistent. Thus, the objective of this study was to evaluate the effects of the dietary application of AMA on the performance, digestibility, and rumen fermentation of lactating dairy cows using a meta-analytic method. A total of 18 peer-reviewed manuscripts (N = 32 treatment comparisons) from 2003 to 2019 were systematically identified following the PRISMA method. The weighted raw mean differences between dietary AMA and control treatments were compared with a robust variance estimation. Likewise, diet characteristics like crude protein (CP) content, NDF content, starch content, days in milk (DIM), experimental design (Latin square and continuous), and AMA dose (0 to 732 Kilo Novo units [KNU]/kg TMR) were used as covariates in a meta-regression, subgrouping, and dose–response analysis. Compared to the control, dietary AMA increased (P < 0.05) DM digestibility (69.32% vs. 68.30%), TTSD (94.62% vs. 94.10%), milk protein concentration and yield (3.11% vs. 3.08%; 1.14 vs. 1.10 kg/d) and tended to increase (P = 0.09) fat-corrected milk (35.96 vs. 35.10 kg/d), but no effects were observed on DM intake (22.99 vs. 22.90 kg/d) and feed efficiency (1.50 vs. 1.48). Dietary AMA tended (P = 0.10) to reduce rumen pH (6.27 vs. 6.30). Both the enzyme dose and DIM strongly influenced (P < 0.05) the effects of AMA on digestibility and performance. The dose–response analysis revealed that feeding 600 KNU/kg to high-producing early lactation (< 70 DIM) dairy cows increased FCM and milk protein. Accounting for the type of experimental design was associated with a lower between-studies-variance among comparisons. Overall, this meta-analysis supports the hypothesis that dietary AMA supplementation is associated with a better lactational performance in dairy cows. However, these effects are only suitable for high-producing early lactation dairy cows.

• Feeding exogenous alpha-amylases to dairy cows is associated with an increase in starch digestibility, milk protein content, and fat-corrected milk.

• Results from this meta-analysis suggest that lactation stage, dietary starch content, and enzyme dose are the main factors associated with the response to dietary supplementation of alpha-amylases.

• Doses of 600 KNU/kg of exogenous alpha-amylase are associated with a greater starch digestibility, milk yield, and protein content in early lactation dairy cows.

Introduction

Dietary starch plays a pivotal role in dairy production systems as total-tract starch digestibility influences milk yield (MY) and feed efficiency by lactating dairy cows (Oba and Allen, 2003; Ferraretto et al., 2013; Fredin et al., 2014). Corn grain is one of the most important sources of starch for dairy production systems worldwide (Fredin et al., 2014). However, several factors influence starch digestibility of dairy cows fed corn grain-based diets, such as conservation and processing methods and endosperm type (Rémond et al., 2004; Lopes et al., 2009; Ferraretto et al., 2013). Considering the growing prices of corn grain in recent years, increasing the digestibility of dietary starch is of utmost importance to the dairy industry (Rémond et al., 2004; Ferraretto et al., 2013).

Biological treatments like exogenous alpha-amylase preparations (AMA) have been developed as a strategy to increase starch digestibility in dairy cow diets (Harrison and Tricarico, 2007; EFSA, 2012). Alpha-amylases are a group of enzymes used to cleave alpha-1,4 glycosidic linkages holding the linear (amylose; alpha-1,4) and branched (amylopectin; alpha-1,4 and alpha-1,6) sections of starch (Kunamneni et al., 2005; Karataş et al., 2013). Furthermore, the crosslinking of starch with prolamin proteins hinders starch digestibility by ruminant animals (Kotarski et al., 1992). Thus, dietary AMA can be blended or mixed with exogenous proteases whose catalytic role improves the accessibility to starch by hydrolyzing these prolamin proteins (Tricarico et al., 2005; Karataş et al., 2013; Amaro et al., 2021). Dietary AMA is an alternative to complement physical processing methods of corn grain (Miller et al., 2008; Boehlke et al., 2015) and the development of commercial enzymatic preparations propelled the investigation of the dietary effects of AMA on the performance of dairy cows (Harrison and Tricarico, 2007; Miller et al., 2008). Furthermore, dietary AMA could increase the intake of digestible starch and reduce fecal starch losses in high-producing dairy cows (Fredin et al., 2014, 2015; Andreazzi et al., 2018).

Several studies have evaluated the effectiveness of AMA on performance, digestibility, and rumen fermentation in lactating dairy cows, but the results have been inconsistent (Harrison and Tricarico, 2007; Nozière et al., 2014; Zilio et al., 2019). Although AMAs are commercially available, their effectiveness in increasing milk yield remains inconclusive (EFSA, 2012). Therefore, we hypothesized that a quantitative analysis would provide insights into the overall effects of AMA supplementation in dairy cow diets. Thus, the objective of this study was to estimate the effects of dietary supplementation of AMA on the performance, nutrient digestibility, and rumen fermentation of dairy cows following a meta-analytic approach.

Materials and Methods

Literature search and data extraction

A systematic database was constructed by conducting a comprehensive literature search on Google scholar, The Web of Science, ScienceDirect, Scopus, PubMed, and the Directory of Open Access Journals databases. The systematic search considered the following keywords: 1) alpha-amylases, 2) dairy cows, 3) milk yield, 4) starch, 5) digestibility, 6) intake, and 7) fermentation. The comprehensive database ranged from 2003 to 2019.

Data arrangement and extraction were performed according to the PRISMA procedure (Moher et al., 2009) and summarized in Figure 1. After the initial screening, a total of 1,566 peer-reviewed manuscripts were added to the database and then reduced to 1,243 manuscripts after the removal of non-peer-reviewed and duplicated records. Only studies fulfilling these inclusion criteria were kept in the database: 1) peer-reviewed manuscripts published in English language, 2) studies conducted with lactating dairy cows, 3) studies reporting intake, the starch content of the diet, milk yield and, total-tract nutrient digestibility, 4) studies comparing the effects of AMA against control, 5) studies reporting the dose of AMA and the composition of the enzymatic preparation, and 6) studies reporting the standard error of the mean (SEM), standard deviation (SD), and the number of experimental units used per treatment.

Figure 1.

PRISMA workflow diagram from the meta-analysis of dietary supplementation of exogenous alpha-amylases on performance and rumen fermentation of dairy cows.

After the application of the inclusion criteria, a total of 18 studies were included in the final database generated with the total number of comparisons (AMA vs. Control) within studies, means, and SEM extracted from both control and AMA treatments. Similarly, dry matter intake (DMI), MY, milk fat, milk protein, total-tract digestibility of dry matter (DMD), total-tract starch digestibility (TTSD), total-tract crude protein digestibility (CPD), total-tract neutral detergent digestibility (NDFD), rumen pH, total volatile fatty acids (TVFA), ammonia (NH₃-N), molar proportions of acetate, propionate, and butyrate were recorded per treatment. Additionally, dietary crude protein (CP), neutral detergent fiber (NDF), and starch content, days in milk (DIM), experimental design (Latin square =1, Continuous design = 2), and enzyme doses were recorded and used as covariates. The AMA doses in the diet were expressed in Kilo Novo alpha-amylase units (KNU) per kg of dry matter. One KNU is the quantity of enzyme that hydrolyzes 5.26 g starch per hour at 37 °C, pH 5.6 (Olsen, 1995). When the units were not reported in KNU, the enzymatic activity (dextrose units or dextrose equivalents) was used as a reference to back-calculate the dose following Silano et al. (2018). The 18 studies used in the analysis of AMA ranged from 0 to 732 KNU/ kg DM with a total number of 32 treatment comparisons (experimental unit). The list of treatments with the corresponding doses and experimental units is reported in Table 1. Likewise, the list of variables measured in each manuscript is reported in Table 2 to clarify the inconsistencies in the number of comparisons analyzed. The raw data used for the analyses can be found at Supplementary Table 1.

Table 1.

Summary of the enzyme characteristics and doses of the exogenous alpha-amylases fed to lactating dairy cows

Reference	Treatment	Type of enzyme1	Form	Dose KNU/ kg of TMR2
Klingerman et al., 2009	Control	–	–	0
	Low amylase	EC 3.2.1.1.	Liquid	150
	High amylase	EC 3.2.1.1.	Liquid	350
	Sigma amylase	EC 3.2.1.1.	Liquid	300
Nozière et al., 2014	Control	–	–	0
	Amylase (High starch)	EC 3.2.1.1.	Dry	300
	Control	–	–	0
	Amylase (Low starch)	EC 3.2.1.1.	Dry	300
Weiss et al., 2011a	Control	–	–	0
	Amylase (low starch)	EC 3.2.1.1.	Liquid	300
	Control	–	–	0
	Amylase (high starch)	EC 3.2.1.1.	Liquid	300
Weiss et al., 2011b	Control	–	–	0
	Amylase (low starch)	EC 3.2.1.1.	Liquid	300
Gencoglu et al., 2010	Control	–	–	0
	Reduced starch + amylase	EC 3.2.1.1.	Liquid	300
Ferraretto et al., 2011	Control	–	–	0
	Reduced starch + amylase	EC 3.2.1.1.	Dry	600
Vargas-Rodriguez et al., 2014	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	600
Andreazzi et al., 2018	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	300
Silva et al., 2018	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	330
Silva et al., 2018	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	330
Takiya et al., 2017	Control	–	–	0
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Dry	150
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.5	Dry	300
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.5	Dry	450
McCarthy et al., 2013	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	732
DeFrain et al., 2005	Control	–	–	0
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Dry	662
Bachmann et al., 2018a	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	600
Bachmann et al., 2018b	Control	–	–	0
	Amylase	EC 3.2.1.1.	Dry	600
Harrison and Tricarico, 2007	Control	–	–	0
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.5	Dry	600
Gençoğlu et al., 2019; 2019	Control	–	–	0
(Two papers but same study)	Amylase	EC 3.2.1.1.	Dry	600
Tricarico et al., 2005	Control	–	–	0
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Dry	240
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Dry	480
	Amylase	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Dry	720
Miller et al., 2008	Control	–	–	0
	Amylase (Low)	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Liquid	200
	Amylase (High)	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Liquid	400
	Control	–	–	0
	Amylase (Low)	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Liquid	200
	Amylase (High)	EC 3.2.1.1., EC 3.2.1.8, EC 3.4.21.53	Liquid	400
Hristov et al., 2008	Control	–	–	0
	Amylase	EC 3.2.1.1.	Liquid	581
	Control	–	–	0
Zilio et al., 2019	Amylase	EC 3.2.1.1.	Dry	203

Enzyme activity according to BRENDA (https://www.brenda-enzymes.org/)

Enzyme doses were calculated following the method described by Olsen (1995) and Silano et al. (2018) considering the dextrose equivalent as reference unit.

Table 2.

List of variables used to conduct the meta-analysis of dietary supplementation of exogenous alpha-amylase preparations in dairy cows

Reference		DMI	DMD	TTSD	NDFD	CPD	MY	FCM	MFat	MProtein	Lactose	FCM/DMI	1 Rumen fermentation
Klingerman et al., 2009	1	*	*	*	*	*	*	*	**	**	*	*
Nozière et al., 2014	2	*	*	*	*		*	*	*	*	*		*
Weiss et al., 2011	3	*	*	*	*	*	*		**	**	*	*
Gencoglu et al., 2010	4	*	*	*	*	*	*	*	**	**	*	*
Ferraretto et al., 2011	5	*					*	*	**	**	*
Vargas-Rodriguez et al., 2014	6	*					*		**	**	*
Andreazzi et al., 2018	7	*	*	*	*		*		**	**	*	*	* , +
Silva et al., 2018	8	*	*	*	*	*	*	*	**	**	*	*	* , +
Takiya et al., 2017	9	*	*	*	*		*	*	**	**	*	*	*
McCarthy et al., 2013	10	*	*	*	*	*	*	*	**	**	*	*
DeFrain et al., 2005	11	*					*		**	**	*		* , +
Bachmann et al., 2018	12						*		*	*	*
Harrison and Tricarico, 2007	13						*	*	**	**
Gençoğlu et al., 2019	14	*	*	*	*	*	*	*	**	**	*	*
Tricarico et al., 2005	15						*	*	**	**			*
Miller et al., 2008	16	*					*		**	**	*		*
Hristov et al., 2008	17	*	*	*	*								*
Zilio et al., 2019	18	*	*	*	*	*	*	*	**	**	*	*	*

Total Volatile fatty acids, acetate, propionate, butyrate, and ammonia nitrogen.

Variable included for the analysis.

Both % and yield reported.

Rumen pH was not reported.

Statistical analysis

Effect size

The effects of dietary application on performance and rumen fermentation were calculated using weighted raw mean differences (RMD) between AMA vs. Control treatments (effect size). The size effect was adjusted (weighting) by the inverse of the variance in a hierarchical effects model that used a robust variance estimation following Tipton (2015) and Arriola et al. (2021). The hierarchical effect model accounted for the variance estimation between-clusters and between-studies-within-cluster variance components.

Heterogeneity and variance

The heterogeneity was calculated using I² that corresponds to the proportion of the variance effects of the treatment divided by the total variance observed (Lean et al., 2018) proposed by Higgins (2008). However, to allow the readers to evaluate the variance components and the heterogeneity unambiguously, both the variance between clusters and between-studies-within-cluster were calculated (Borenstein et al., 2017; Lean et al., 2018; Arriola et al., 2021). Both τ² statistics and Ω² statistics were calculated to account for the between-clusters and between-studies-within-cluster variance components following a previously reported method (Hedges et al., 2010; Fisher et al., 2017). Briefly, each variance comparison between AMA vs. control treatments was calculated as the square of the pooled SD. The pooled SD for AMA vs. control comparison was calculated from the SEM reported considering $\mathrm{S}\mathrm{D}=\mathrm{S}\mathrm{E}\mathrm{M}\times \sqrt{n}$, where n represents the number of experimental units. Then, the overall effect size was calculated considering the weighted RMD (SEM and n), τ², and, Ω² statistics.

Publication bias and outliers

Publication bias was calculated and visualized using the asymmetry test by Egger’s regression method between RMD and SE using funnel plots (Egger et al., 1997; Oliveira et al., 2017). Likewise, Cook’s distances were used to remove outliers and influential points following the method described by Arriola et al. (2021). Briefly, the comparisons between AMA vs. Control had standardized residuals less than 2.3, and no influential points were detected in the dataset.

Meta-regression, subgrouping, and dose–response analysis

The meta-regression analysis was performed to evaluate the effects of the covariates (dietary components, DIM, experimental design, and AMA dose; Table 3) associated with the effect size on the response variables following the method described by Pech-Cervantes et al. (2021). The meta-regression analysis was performed following a robust variance estimation with a hierarchical effect model accounting for Ω² and τ² statistics (Hedges et al., 2010; Fisher et al., 2015; Arriola et al., 2021). When any of the covariates were significant (P < 0.05), the subset analysis was performed to examine the effects of the corresponding covariate. Additionally, the dose–response trends between the different doses of AMA and their corresponding weighted RMD were evaluated following the method proposed by Greenland (1992, 1995) and implemented by Farhat et al. (2022). Briefly, the calculated weighted effect size was used to estimate the efficacy of AMA on animal performance and fermentation. Then, different regression models (linear, quadratic, and cubic) were tested to select the best fitting with the lowest Akaike information criteria. Similar to Lean et al. (2018), the weighted RMD were fitted with the corresponding doses of AMA to let the readers contrast the differences in effect size among comparisons.

Table 3.

Chemical composition and descriptive statistics of the experimental diets and treatments for lactating dairy cows fed with exogenous alpha-amylase preparations

Item	N1	Mean	Std	Min	Max	Median
DM(% of as fed)	26	56.2	6.2	44.8	71.6	54.5
CP(% of DM)	32	16.3	1.1	13.9	18.8	16.3
NDF(% of DM)	32	33.8	3.8	28.1	43.4	34.1
ADF(% of DM)	17	19.2	2.3	15.6	23.5	18.6
Starch(% of DM)	32	24.6	4.6	15.5	32.2	25.5
AMA2 dose, KNU/ kg DM	32	256.4	256.1	0	732	240
DIM3 (d)	21	93.6	54.1	1	181	80
BW4 (kg)	32	643.7	66.2	539	733	663

Total number of treatments means (control and AMA treatments).

Exogenous alpha-amylase (kilo novo unit per kg of dry matter).

Days in milk.

Body weight.

Forest and funnel plots

Forest plots were used to visualize the weighted effect size using the package “robumeta” and following the method previously described by Arriola et al. (2021). Likewise, funnel plots were used to visualize the bias following the methodology described in the package “metafor” and according to Oliveira et al. (2017).

All data analyses were performed in R and Rstudio following the methods described by Arriola et al. (2021) and Pech-Cervantes et al. (2021). The RMD, forest plot, and meta-regression analysis were performed using the robumeta (version 1.3.1093; https://cran.r-project.org/web/https://cran.r-project.org/web/packages/robumeta/robumeta.pdf) and metafor (version 1.3.1093; https://cran.r-project.org/web/packages/metafor) packages according to Fisher et al. (2015) and Viechtbauer (2010).

Results

Dietary composition, animal performance, and rumen fermentation

The descriptive statistics of the chemical composition of experimental diets, DIM, and BW are reported in Table 3. Although the chemical composition, DIM, and BW differed among comparisons, no influential points were detected (P > 0.05).

The effects of dietary AMA on digestibility and performance of lactating dairy cows are summarized in Table 4. Compared to the control, dietary AMA did not (P < 0.05) affect DMI (22.99 vs. 22.90 kg/d), NDFD (49.51% vs. 48.71%) or feed efficiency (1.50 vs. 1.48 kg of FCM/kg of DMI) in dairy cows. However, dietary AMA supplementation was associated with a higher (P < 0.05) DMD (69.32% vs. 68.30%) and TTSD (94.62% vs. 94.10%; Figure 2) and tended (P = 0.09) to increase FCM (35.96 vs. 35.10 kg/d; Figure 3) in dairy cows compared to the control. Among comparisons, dietary AMA supplementation was associated (P < 0.05) with a greater milk protein concentration (3.11% vs. 3.08%) and yield (1.12 vs. 1.08 kg/d) in dairy cows compared to the control. No effects were observed in CPD (68.16% vs. 67.31%), MY (35.36 vs. 34.9 kg/d), or milk fat concentration (3.64% vs. 3.62%). The variance analysis revealed a low to moderate Ω² for DMI and digestibility but a higher Ω² for FCM and MY among comparisons. Similarly, the funnel test revealed that most of the response variables had a moderate heterogeneity (I² < 50%), except for CPD, milk protein yield, and feed efficiency. Moreover, no significant publication bias was detected (P > 0.05) among comparisons (Supplementary Appendix 1; Supplementary Material 1). Furthermore, the forest plots of the effects of AMA on DMD and milk protein yield are reported in Supplementary Figures 1a and d.

Table 4.

Effect of dietary supplementation of exogenous alpha-amylases (AMA) on intake, digestibility, and performance by lactating dairy cows

Item	Control2			RMD3		Variance component4		Bias
	N1	Mean	STD	Effect size	P	Ω2	τ2	Funnel test5 (P-value)	I 2 (%)
DMI6 (kg/d)	26	22.9	4.5	0.09 (−0.44, 0.63)	0.70	0.82	0	0.80	52.01
DMD7 (%)	18	68.3	4.2	1.02 (0.16, 1.87)	0.03	2.71	0	0.72	35.42
NDFD8 (%)	18	48.7	11.1	0.80 (−0.53, 2.14)	0.19	7.74	0	0.10	16.19
CPD9 (%)	11	67.3	6.3	0.86 (−0.68, 2.41)	0.18	4.12	0	0.69	60.25
TTSD10 (%)	18	94.1	4.3	0.52 (0.2, 0.83)	<0.01	6.22	0	0.08	17.44
MY11 (kg/d)	31	34.9	7.3	0.46(−0.24,1.16)	0.18	18.5	0	0.63	22.35
3.5% FCM12 (kg/d)	19	35.1	6.4	0.86(−0.21, 1.92)	0.09	21.6	0	0.49	33.81
Milk fat (%)	31	3.62	0.42	0.02(−0.03,0.07)	0.35	0.09	0	0.14	16.28
Milk fat (kg/d)	27	1.25	0.21	0.01 (−0.03,0.05)	0.53	0.03	0	0.57	42.05
Milk protein (%)	31	3.08	0.18	0.02 (0.004,0.04)	0.03	0.15	0	0.73	0
Milk protein (kg/d)	27	1.08	0.17	0.04 (0.001,0.07)	0.04	0.03	0	0.22	82.52
Milk lactose (%)	25	4.81	0.17	−0.01(−0.02,0.006)	0.38	0.16	0	0.57	0
Efficiency (FCM/DMI)	14	1.48	0.24	0.02(−0.03,0.07)	0.40	0.007	0	0.19	84.79

Positive values in RMD indicate an increase by addition of AMA, whereas negative values in RMD indicate a decrease by addition of AMA respect to the control.

Total number of comparisons (AMA vs. Control).

No AMA in the diet.

Raw mean difference (weighted) between control vs. AMA treatment. The hierarchical model accounted for the comparison within the study.

Ω² = between-studies-within-cluster variance component; τ² = between-cluster variance component (Hedges et al., 2010; Fisher et al., 2017).

P-value for X² (Q) test for heterogeneity; I² = Proportion of total variation of size effect estimated due to heterogeneity.

Dry matter intake.

Total-tract dry matter digestibility.

Total-tract neutral detergent fiber digestibility.

Total-tract crude protein digestibility.

Total-tract starch digestibility.

Milk yield.

Fat corrected milk.

Figure 2.

Forest plot of the effect of dietary inclusion of exogenous alpha-amylases (AMA) on total-tract starch digestibility in lactating dairy cows. The squares represent the mean size effect for the comparison within study, the size of the square represents the relative weighting in each comparison, the connected lines with the squares represent the 95% confidence interval for the size effect, and the dotted line indicates the overall effect size. The diamond at the bottom represents the mean response across comparisons with the corresponding the 95% confidence interval.

Figure 3.

Forest plot of the effect of dietary inclusion of exogenous alpha-amylase enzyme preparations on 3.5% fat-corrected milk of dairy cows. The squares represent the mean size effect for the comparison within study, the size of the square represents the relative weighting in each comparison, the connected lines with the squares represent the 95% confidence interval for the size effect, and the dotted line indicates the overall effect size. The diamond at the bottom represents the mean response across comparisons with the corresponding the 95% confidence interval.

The effects of dietary AMA on rumen pH and fermentation of lactating dairy cows are shown in Table 5. Dietary application of AMA tended to decrease (P = 0.1) rumen pH (6.27 vs. 6.30) compared to the control without any effects observed (P > 0.05) on TVFA (112.8 vs. 111.9 mmol/L). Similarly, dietary application of AMA did not affect (P > 0.05) molar proportions of acetate (62.18% vs. 62.50%), propionate (19.97% vs. 20%), or butyrate (13.54% vs .13.40%) compared to the control. Furthermore, dietary AMA supplementation did not influence ruminal concentration of NH₃-N (20.94 vs. 21.0 mg/dL) compared to the control. The variance analysis showed a higher Ω² for total VFA concentrations but a lower Ω² for acetate, propionate, and butyrate molar proportions among comparisons. Unlike animal performance data, rumen fermentation data had a high heterogeneity (I² < 50%). Despite the high heterogeneity, no significant potential for bias was detected for pH and rumen fermentation, but significant potential for bias was detected (P < 0.01) for NH₃-N.

Table 5.

Effect of dietary supplementation of exogenous alpha-amylases (AMA) on rumen pH, volatile fatty acids, and ammonia nitrogen concentrations by lactating dairy cows

Item	Control2			RMD3		Variance component4		Bias
	N1	Mean	STD	Effect size	P	Ω2	τ2	Funnel Test5 (P-value)	I 2 (%)
pH	12	6.3	0.3	−0.03 (−0.06,0.01)	0.10	0	0	0.1	0
Total VFA6 (mmol/L)	16	111.9	45.6	0.86 (−4.56, 6.3)	0.66	9.06	0	0.07	73.48
Acetate (%)	17	62.5	3.3	−0.32 (−1.23,0.59)	0.38	2.18	0	0.62	42.90
Propionate (%)	17	20.0	2.6	0.03 (−1.41,1.46)	0.95	0.64	0	0.91	60.52
Butyrate (%)	17	13.4	1.9	0.14 (−0.65,0.93)	0.55	0.27	0	0.73	43.48
A:P ratio7	13	3.10	0.5	−0.10(−0.42,0.23)	0.41	0.15	0	0.38	73.59
NH3−N (mg/dL)	16	21.0	6.5	0.06(−0.78,0.91)	0.84	0	0	<0.01	71.63

Positive values in RMD indicate an increase by addition of AMA, whereas negative values in RMD indicate a decrease by addition of AMA respect to the control.

Total number of comparisons.

No AMA in the diet.

Raw mean difference (weighted) between control vs. AMA treatment.

Ω² = between-studies-within-cluster variance component, τ² = between-cluster variance component (Hedges et al., 2010; Fisher et al., 2017).

P-value for X² (Q) test for heterogeneity; I² = Proportion of total variation of size effect estimated due to heterogeneity.

Total volatile fatty acids.

Acetate:Propionate ratio.

Meta-regression, subset analysis, and dose–response analysis

The meta-regression analysis of the dietary application of AMA on performance and rumen fermentation of dairy cows is summarized in Table 6. Although most of the covariates did not influence the intercepts (P > 0.05), fitting the covariates was associated with a lower Ω² in the meta-regression. Among all the covariates, DIM and the doses of AMA were strongly associated (P < 0.05) with the responses observed in MY, FCM, feed efficiency, and TVFA. However, starch and NDF tended to influence (P = 0.08) DMI and MY (P = 0.08) among comparisons (Table 6). Moreover, the type of experimental design did not influence the responses observed in animal performance but was associated (P = 0.04) with the responses observed in rumen pH among comparisons (Figure 4). Furthermore, the subset analysis revealed that dietary AMA reduced rumen pH among comparisons of Latin square designs, but no associations were observed among comparisons of continuous designs (Figure 4).

Table 6.

Meta-regression of the effect of dietary nutrient concentrations and dietary dose of exogenous alpha-amylases (AMA) on raw mean differences (RMD) for performance, digestibility, and rumen fermentation of lactating dairy cows

Dependent variable	N1	Intercept	P	NDF2												Variance13
					P	CP3	P	Starch	P	DIM4	P	Dose5	P	Design14	P	Ω2	τ2
DMI (kg/d)	26	14.45	0.12	−0.24	0.08	−0.35	0.27	−0.03	0.08	0.001	0.81	0.001	0.86	−0.003	0.99	0.99	0
DMD6 (%)	18	16.10	0.30	0.07	0.61	0.74	0.18	0.06	0.60	−0.009	0.32	0.002	0.44	0.54	0.46	0.68	0
NDFD7 (%)	18	9.44	0.75	0.08	0.83	0.04	0.96	0.09	0.76	0.006	0.67	0.006	0.35	1.31	0.49	0	0
CPD8 (%)	11	−52.9	0.67	0.46	0.67	2.48	0.64	0.01	0.97	−0.006	0.85	0.001	0.95	−1.21	0.75	0	0
TTSD9 (%)	18	7.20	0.22	−0.11	0.19	−0.23	0.20	0.03	0.65	−0.002	0.82	−0.001	0.53	0.61	0.11	6.42	0
MY10 (kg/d)	31	8.02	0.21	−0.19	0.08	−0.20	0.45	0.14	0.15	−0.01	0.04	−0.001	0.38	0.40	0.53	18	0
3.5% FCM11 (kg/d)	19	0.16	0.98	−0.20	0.17	0.64	0.12	−0.02	0.81	−0.02	0.01	−0.006	0.03	0.76	0.24	20.7	0
Milk fat (%)	31	−0.54	0.63	−0.001	0.89	0.07	0.22	−0.02	0.16	0.001	0.94	−0.001	0.49	−0.13	0.17	0.04	0
Milk fat (kg/d)	27	0.09	0.73	−0.006	0.20	0.002	0.78	0.008	0.20	−0.001	0.07	−0.001	0.26	−0.03	0.40	0.01	0
Milk protein (%)	31	0.29	0.45	0.001	0.99	−0.02	0.39	0.001	0.86	−0.001	0.10	0.001	0.32	−0.02	0.44	0.08	0
Milk protein (kg/d)	27	0.14	0.71	−0.008	0.29	0.01	0.62	0.002	0.73	−0.004	0.30	−0.001	0.32	0.004	0.90	0.04	0
Milk Lactose (%)	25	−0.27	0.35	0.002	0.59	0.01	0.41	0.001	0.94	0.001	0.91	−0.001	0.75	0.02	0.25	0.07	0
Efficiency (FMC/DMI)	14	−1.39	0.16	0.01	0.21	0.06	0.28	0.004	0.57	−0.001	0.05	−0.001	0.34	0.05	0.37	0.004	0
pH	12	−0.12	0.62	0.004	0.39	−0.01	0.56	−0.003	0.20	0.001	0.08	0.001	0.73	0.15	0.04	0	0
Total VFA12 (mmol/L)	16	−27.3	0.47	0.59	0.34	−0.02	0.98	0.84	0.50	−0.09	0.05	−0.003	0.58	0.67	0.93	0	0
Acetate	17	−13.1	0.52	0.25	0.52	0.58	0.59	−0.19	0.29	0.006	0.74	−0.001	0.84	−0.83	0.64	1.97	0
Propionate	17	15.1	0.67	−0.25	0.68	−0.68	0.68	0.24	0.35	−0.01	0.67	−0.001	0.67	0.20	0.93	0	0.83
Butyrate	17	8.01	0.66	−0.03	0.90	−0.51	0.55	−0.05	0.41	0.01	0.33	0.001	0.52	1.17	0.30	0.09	0
A:P ratio	13	6.39	0.48	0.11	0.28	0.23	0.60	−0.04	0.21	0.001	0.80	−0.001	0.54	−0.13	0.80	0	0.02
NH3−N (mg/dL)	16	2.35	0.85	−0.04	0.88	−0.43	0.56	0.14	0.77	0.01	0.45	0.003	0.62	−0.15	0.97	0	0

Comparisons were performed between AMA vs. Control.

Number of comparisons.

Neutral detergent in the diet.

Crude protein in the diet.

Days in milk.

Dietary dose of alpha-amylase (KNU/ kg DM).

Total-tract dry matter digestibility.

Total-tract neutral detergent fiber digestibility.

Total-tract crude protein digestibility.

Total-tract starch digestibility.

Milk yield.

Fat corrected milk.

Total volatile fatty acids.

P-value for X² (Q) test for heterogeneity; I² = Proportion of total variation of size effect estimated due to heterogeneity.

Ω² = between-studies-within-cluster variance component, τ² = between-cluster variance component (Hedges et al., 2010; Fisher et al., 2017).

Latin square (1) or continuous design (2).

Figure 4.

Effect of type of experimental design (subsets = Latin square and Continuous) on rumen pH of lactating dairy cows fed with exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control). Error bars represent the 95% confidence interval. The symbol (*) indicates significant difference (P < 0.05). RMD, weighted raw mean difference between AMA vs. control.

The dose–response analysis revealed that low levels of starch in the diet (<24%) did not influence the effects of AMA on DMI among comparisons. Similarly, high levels of starch in the diet (>28%) did not influence the effectiveness of AMA on DMI. Conversely, average levels of starch in the diet (26%) were associated with greater efficacy of AMA on DMI among treatment comparisons (Figure 5). Unlike DMI, the lactational stage represented as DIM, and the dose of AMA influenced the responses observed in FCM among comparisons (Figure 6). The dose–response analysis revealed that feeding AMA to early lactation dairy cows (< 70 DIM) was consistently associated with an increase in FCM. Although only high doses of AMA (>500 KNU) increased FCM in dairy cows with more than 70 DIM, inconsistent responses were observed among comparisons. Furthermore, treatment comparisons conducted with mid-lactation dairy cows with more than 150 DIM did not benefit from dietary AMA supplementation.

Figure 5.

Dose–response plot of the effect of the dietary starch on dry matter intake (N = 26) of lactating dairy cows fed with different doses of exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control), the larger the marker, the greater the effect size of the treatment. DMI, dry matter intake kg/d; KNU, Exogenous alpha-amylase (kilo novo unit per kg of dry matter); RMD, weighted raw mean difference between AMA vs. control; y = −0.0045x3 + 0.322x2 − 7.5245x + 57.697; R² = 0.3921.

Figure 6.

Dose–response plot of the effect of days in milk on 3.5% fat-corrected milk yield (N = 19) raw mean differences of dairy cows fed with different levels of exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control); the larger the marker, the greater the effect size of the comparison. KNU, exogenous alpha-amylase (kilo novo unit per kg of dry matter); RMD, weighted raw mean difference between AMA vs. control; y = −1E-06x3 + 0.0002x2 − 0.0167x + 0.9179; R² = 0.2807.

The effects of DIM and different doses of AMA on milk protein concentration are summarized in Figure 7. Overall, dietary AMA supplementation was associated with higher milk protein concentration among comparisons conducted with early lactation dairy cows (< 120 DIM). However, comparisons conducted with mid-lactation dairy cows (>120 DIM) did not benefit from dietary AMA. Dietary AMA was associated with higher feed efficiency in early lactation dairy cows with less than < 70 DIM, but inconsistent effects were observed among treatment comparisons conducted with mid-lactation (> 150 DIM) dairy cows (Figure 8). The dose–response analysis suggests that early lactation dairy cows (< 120 DIM) had higher TVFA concentrations compared to mid-lactation dairy cows (Figure 9) among treatment comparisons. Moreover, the dose of AMA did not influence the responses observed in TVFA among treatment comparisons.

Figure 7.

Dose–response plot of the effects of days in milk (DIM) on milk protein concentration (%, N = 31) of dairy cows fed with different levels of exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control); the larger the marker, the greater the effect size of the comparison. Dose, KNU, exogenous alpha-amylase (kilo novo unit per kg of dry matter); RMD, weighted raw mean difference between AMA vs. control; y = −6E-08x3 + 2E-05x2 − 0.0014x + 0.0635; R² = 0.3074.

Figure 8.

Dose–response plot of the effects of days in milk (DIM) on feed efficiency (FCM/DMI, N = 14) raw mean differences of dairy cows fed with different levels of exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control), the larger the marker, the greater the effect size of the comparison. Dose, KNU, exogenous alpha-amylase (kilo novo unit per kg of dry matter); RMD, weighted raw mean difference between AMA vs. control; y = 1E-07x3 − 3E-05x2 − 0.0002x + 0.1408; R² = 0.4127.

Figure 9.

Dose–response plot of the effects of days in milk (DMI) on total volatile fatty acid concentrations in the rumen of dairy cows fed with different doses of exogenous alpha-amylases (AMA). The effect size (weighted raw mean difference) was calculated and weighted in each comparison (AMA vs. Control), the larger the marker, the greater the effect size of the comparison. Dose, KNU, exogenous alpha-amylase (kilo novo unit per kg of dry matter); RMD, weighted raw mean difference between AMA vs. control; y = 5E-06x3 − 0.0013x2 + 0.0691x + 2.5667; R² = 0.2854.

Discussion

Achieving optimal starch digestion throughout the gastrointestinal tract is the core idea behind the inclusion of AMA in dairy diets. However, both physiological and enzymatic factors influenced the effects of AMA observed on DMI, FCM, and TVFA concentrations. Therefore, it is not surprising that DIM and the dose were associated with the effectiveness of AMA in the present meta-analysis. Similar to the present study, previous meta-analyses concluded that exogenous enzymes were associated with higher milk protein concentration and tended to increase milk yield among treatment comparisons (Arriola et al., 2017; Tirado-González et al., 2018). However, unlike previous meta-analyses, results from this study highlighted that the lactation stage expressed as DIM was strongly associated with the effectiveness of AMA among comparisons. Thus, results from this meta-analysis underscored that treatment comparisons feeding AMA to early lactation dairy cows (from 1 to 70 DIM) with normal levels of starch (Ferraretto et al., 2013; Moharrery et al., 2014; Hatew et al., 2015) in the diet (26%) increased DMI, FCM, and milk protein content compared to control cows. In contrast, comparisons conducted with mid-lactation cows (> 120 DIM) did not affect animal performance regardless of the dose of AMA. Previous research reported that early lactation cows (from 53 to 81 DIM) had greater milk yield and efficiency compared to mid-lactation cows (> 100 DIM), and nutritional interventions have a greater impact on high-producing cows (DeVries et al., 2011; Boerman et al., 2015). These results could help us to explain the differences in FCM, feed efficiency, and milk protein concentration observed between comparisons in the present meta-analysis.

Collectively, the results of this meta-analysis demonstrated that compared to the control, a greater milk protein concentration, total-tract digestibility, and a tendency to increase FCM were observed in dairy cows fed diet supplemented with AMA The synergistic effects of amylases and rumen microbes on the hydrolysis of starch probably explain the effects on digestibility and milk protein observed in this meta-analysis (Kunamneni et al., 2005; Graminha et al., 2008). Unlike high purity enzymes, enzymatic preparations like AMA are produced by solid-state or liquid-state fermentation to reduce the cost of production (Salim et al., 2017). Thus, the batch-to-batch variation in AMA could help us to explain the differences in enzymatic activity among studies and the inconsistencies observed among comparisons with similar doses. Previous research in monogastric animals demonstrated the effectiveness of AMA on performance and starch digestibility (Omogbenigun et al., 2004; Amerah et al., 2017). Likewise, increased in vitro dry matter, starch degradability, and fermentation of dent corn grain were observed with the supplementation of exogenous amylases or proteases (Amaro et al., 2021). These results support some of the effects observed among comparisons in the present meta-analysis.

The overall effects among comparisons showed that dietary AMA supplementation was associated with a higher TTSD, milk protein concentration, and FCM compared with the control. These effects could be explained by the pre-ingestive and post-ingestive hydrolysis of dietary starch and prolamins in dairy cow diets (Larson and Hoffman, 2008; Gallo et al., 2016; Amaro et al., 2021). Hydrophobic prolamins are starch-encapsulating proteins that reduce starch degradation in the rumen (Larson and Hoffman, 2008; Nellis et al., 2013). Possibly, the synergy of the effects of proteases and alpha-amylases increased the degradation of encapsulated starch by rumen microbes which increased the total VFA concentration in the rumen, but further research is warranted to evaluate this premise. The dose–response analysis showed that a dose of 600 KNU/kg of AMA was associated with higher DMI, FCM, milk protein concentration, and feed efficiency among comparisons conducted with early lactation cows (< 120 DIM). However, inconsistent effects were observed on DMI, milk protein concentration, and efficiency with lower doses of AMA (< 400 KNU/kg) among comparisons. Several hypotheses have been proposed to explain the lack of effects of AMA on the performance of ruminants like compensatory starch digestion in the hindgut, low enzymatic activity in AMA, competition with rumen microbes, and processing method (DiLorenzo et al., 2011; Gallo et al., 2016; Takiya et al., 2017). These hypotheses could help us to explain the inconsistencies observed across the literature in the present meta-analysis. Moreover, treatment comparisons with doses of 600 KNU/kg to early lactation (< 70 DIM) dairy cows were associated with higher milk yield and milk protein yield, implying that dietary application of AMA benefits high-producing dairy cows.

Fitting the covariates in a robust meta-regression analysis successfully reduced the between-studies variance (Ω²) and explained the variation observed among comparisons. Previous research highlighted the importance of the explanatory variables in hierarchical models like this meta-analysis (Goldstein et al., 2000). Likewise, Lean et al. (2018) recommended using a robust regression model to weight the effect size in dairy cow meta-analyses. Although some of the results of this meta-analysis agreed with these studies, inconsistent responses were observed in FCM and milk protein concentration with high and low doses of AMA. Thus, caution is advised extrapolating the differences in digestibility and performance between lactation stages. Based on the current meta-analysis, we hypothesize that feeding 600 KNU/kg to high-producing early lactation dairy cows could increase starch digestibility and milk yield when dietary starch is not higher than 26%.

Results of the present meta-analysis showed that starch levels in the diet below 20% and above 27% did not show any response associated with AMA among treatment comparisons. Recent research showed that providing high dietary starch content (29.2%) reduced the ruminating time, DMD, and did not improve TVFA concentrations in lactating dairy cows compared to lower levels of starch (22.5%) in the diet (Akhlaghi et al., 2022). Increasing the starch content in the diet reduces the forage to concentrate ratio which favors the proliferation of starch-degrading bacteria at the detriment of fiber-degrading bacteria thereby reducing fiber fermentation in the rumen (Zhang et al., 2022). These results could help us to explain the effects among comparisons observed in the present meta-analysis.

The subset analysis showed that the effects of AMA were associated with a lower rumen pH among comparisons with Latin square designs without significant associations in continuous designs (parallel studies). These differences could be related to the lower variation in the residual error observed in Latin square designs and a potential carry-over effect of periods. Latin square designs require special attention in meta-analyses due to the carry-over effects among periods, the potential for exclusion bias (between periods), and a much smaller variance (Lean et al., 2009). Thus, trimming of the SEM has been implemented in meta-analyses to avoid overweighting the effect size of comparisons with Latin square designs (Roman-Garcia et al., 2016; Arriola et al., 2021). Although some authors have challenged the idea of combining studies with different designs, we accounted for the effect of the type of experimental design in this meta-analysis to further understand the causes of heterogeneity among comparisons within studies (Sauvant et al., 2008, 2020; Lean et al., 2009). Unlike rumen pH, the type of experimental design was not associated with the effects of AMA on animal performance, rumen fermentation profiles, and TVFA concentrations among treatment comparisons.

Results of DMI, FCM, and milk protein observed among comparisons suggest that feeding 600 KNU/kg AMA to early lactation dairy cows (< 70 DIM) is the best strategy to increase animal performance. Although doses of 300 KNU/kg AMA were associated with better animal performance, their effects were inconsistent among comparisons. Thus, considering that amylolytic preparations like AMA are commercially available as feed additives (EFSA, 2012; Silano et al., 2018), the results of this meta-analysis provide a concrete strategy to maximize the effects of AMA in dairy operations. However, these recommendations can only be applied to high-producing dairy operations in which neither dietary nor lactational factors limit milk yield (Sato et al., 2005). Moreover, additional studies are warranted to estimate the cost-benefit of the dietary application of AMA in high-producing dairy cows. Future studies should focus on elucidating the mode of action of dietary AMA on the rumen microbiome and metabolome.

Conclusion

Dietary application of AMA was associated with an increase in digestibility and milk protein content in lactating dairy cows among treatment comparisons. Compared to the control, higher milk protein yield and FCM were observed when the dose of AMA was 600 KNU/kg in early lactation dairy cows. Moreover, inconsistent responses among comparisons were observed with 300 KNU/kg of AMA. The effectiveness of AMA was strongly associated with the lactation stage, starch content in the diet, and enzyme dose. Accounting for the type of experimental design reduced the variance (Ω²) observed among comparisons in the meta-regression analysis. Although this meta-analysis supports the hypothesis that dietary AMA improves lactational performance, these responses are limited to high-producing early lactation dairy cows. Future studies are required to elucidate the mode of action of AMA in the rumen and their influence on the rumen microbiome.

Glossary

Abbreviations

AMA

exogenous alpha-amylase preparations

ADF

acid detergent fiber

TTDS

total-tract starch digestibility

MY

milk yield

PRISMA

Preferred Reporting Items for Systematics Reviews and Meta-analyses

RMD

raw mean difference (AMA vs. Control)

CP

crude protein

CPD

crude protein digestibility

DM

dry matter

DIM

days in milk

DMI

dry matter intake

DMD

dry matter digestibility

FCM

fat-corrected milk

KNU

kilo novo alpha-amylase units

NDF

neutral detergent fiber

NDFD

neutral detergent fiber digestibility

TVFA

total volatile fatty acids

Conflict of Interest Statement

The authors declare no real or perceived conflicts of interest.