Volatile fatty acids (VFA) produced by rumen fermentation are the most important energy source for ruminant animals [1]. Rumen VFAs play important role in the carbohydrate nutrition of ruminants and have very close relationships with the composition and dynamics of the rumen microbial community. Furthermore, the proportions of individual VFA reflect different metabolic pathways. Propionate is the main source of gluconeogenesis, while acetate and butyrate are substrates for long-chain fatty acid synthesis [2]. Ruminal branched-chain VFA (isovalerate and isobutyrate), valerate and ammonia are products of ruminal protein degradation [3]. To understand the metabolism of rumen microbes and provide relevant numerical support for animal nutrition studies, accurate quantifications of rumen VFA are needed.
The standard methods for quantification of VFA include chromatography (gas phase or liquid phase), electromigration (capillary isotachophoresis), and spectrophotometric methods [4]. The traditional distillation process with subsequent titration has poor selectivity and productivity and is no longer widely used [5]. There have been reports of colorimetric methods applied to the analysis of organic acids, but which can only measure single acids [6]. Therefore, the chromatographic approaches of identification are more common.
Rumen VFA were first measured by high-performance liquid chromatography (HPLC) in 1988 [7] and HPLC is a simple, and reliable method to quantify rumen VFA. Nowadays, HPLC has been widely used in medicine, biochemistry, environmental protection, agriculture, and other scientific fields based on its many advantages. HPLC can analyse more than 70% of organic compounds [8]. The average time to analyse a sample is 15–30 minutes, but some samples can be analysed in less than 5 minutes, whereas others take up to an hour [9]. HPLC also has the benefit of allowing the chromatographic column to be reused and causing no damage to the sample.
HPLC also has its shortcomings. It is expensive, requires various packing columns, has a small capacity, is not well suited to analyse biological macromolecules and inorganic ions, and the eluent of mobile phase consumes a lot of most toxic chemicals [11]. Although HPLC is commonly used in the analysis of biological samples, the majority of applications are for targeted component analysis rather than overall metabolite fingerprinting [12].
As early as 1999, Attaelmannan et al. used proton nuclear magnetic resonance spectroscopy (NMR) to determine the main VFAs in rumen fluid [13]. Concentrations of acetate, propionate and butyrate in the rumen measured by NMR were not significantly different from the values determined by gas chromatography. NMR has several advantages for metabolite profiling: it is non-destructive, does not require chromatographic separation and minimal sample preparation is sufficient [14]. In addition, as a quantitative analytical technique that NMR can provide structural information about unknown compounds. The intensity of NMR signals is proportional to the concentration of compounds. This feature maximizes the information content of NMR spectra, even though excessive overlap of NMR signals can compromise this process. Matrix constituents and properties (i.e., salt, pH, additives, ionic strength) have a large effect on NMR analysis, but their variation can be controlled i.e., using buffers [15]. The fundamental disadvantage of NMR is its limited sensitivity, which typically limits to the µM range. The presence of significant peak overlap in the NMR spectra of complicated mixtures makes the measurement of low concentration metabolites difficult. The accuracy of quantification of overlapping compounds can be greatly improved by commercial software (such as the Chenomx NMR Suite 7.1) [16]. One drawback that cannot be ignored is the relative high acquisition and maintenance costs of NMR instrumentation, the need for specialised staff and expensive commercial software for quantitative analysis.
Pre-processing and pre-treatment are critical and necessary procedures before downstream data analysis, such as statistical models, for each quantitative metabolite approach. Because the pre-processing/pre-treatment determine whether findings from downstream data analysis accurately represent biological fact. Variability in metabolomic samples is primarily due to two factors: Technical variability is introduced during the pre-analytical and data processing steps, while biological variability is introduced owing to variability in the samples [17]. Biological variability varies according to the biological samples being studied. Due to variations in heredity, body weight, and nutrition, biological variability in biofluids such as urine or serum can be significantly larger [18]. For metabolite quantification, the allowable technical variability should typically not exceed 20% [19]. If the variations in sample concentrations between comparison samples are higher than the analytical technical variance, sample normalisation is required in a metabolomics study. A proper pre-processing method, according to related studies [20], can improve the observed fit.
This study comprehensively investigates the pros and cons of HPLC and NMR analysis methods for work procedures and empirical results on the determination of ruminal VFA. Our work aimed primarily to verify the repeatability, precision, and limits of methods to provide insight into the best approach for animal nutrition research.