Application of visible and near infrared spectroscopy to ...



Near infrared reflectance spectroscopy predicts the content of polyunsaturated fatty acids and biohydrogenation products in the subcutaneous fat of beef cows fed flaxseed

Running title Estimation of fatty acid composition in cow subcutaneous fat by NIR spectroscopy

N. Prieto1, M.E.R. Dugan2, O. López-Campos2, T.A. McAllister3, J.L. Aalhus2, B. Uttaro2

1Instituto de Ganadería de Montaña (Consejo Superior de Investigaciones Científicas – Universidad de León). Finca Marzanas. E-24346 Grulleros, León, Spain.

2Lacombe Research Centre, Agriculture and Agri-Food Canada, 6000 C&E Trail, Lacombe, Alberta, T4L 1W1, Canada.

3Lethbridge Research Centre, Agriculture and Agri-Food Canada, 1st Avenue South 5403, P.O. Box 3000, Lethbridge, Alberta T1J 4B1.

*Corresponding author: Nuria Prieto. Instituto de Ganadería de Montaña (CSIC–ULE). Finca Marzanas. E-24346 Grulleros, León (Spain). Tel +34 987 317 064, Fax +34 987 317 161, E-mail: nuria.prieto@eae.csic.es

Abstract

This study examined the ability of near infrared reflectance spectroscopy (NIRS) to estimate the concentration of polyunsaturated fatty acids and their biohydrogenation products in the subcutaneous fat of beef cows fed flaxseed. Subcutaneous fat samples at the 12th rib of 62 cows were stored at -80 ºC, thawed, scanned over a NIR spectral range from 400 to 2498 nm at 31 ºC and 2 ºC, and subsequently analyzed for fatty acid composition. Best NIRS calibrations were with samples at 31 ºC, showing high predictability for most of the n-3 (R2: 0.81-0.86; RMSECV: 0.11-1.56 mg. g-1 fat) and linolenic acid biohydrogenation products such as conjugated linolenic acids, conjugated linoleic acids (CLA), non-CLA dienes and trans-monounsaturated fatty acids with R2 (RMSECV, mg. g-1 fat) of 0.85-0.85 (0.16-0.37), 0.84-0.90 (0.21-2.58), 0.90 (5.49) and 0.84-0.90 (4.24-8.83), respectively. NIRS could discriminate 100 % of subcutaneous fat samples from beef cows fed diets with and without flaxseed.

Keywords: near infrared reflectance spectroscopy, subcutaneous fat, fatty acid, flaxseed.

1. Introduction

Today’s health conscious consumers are interested in fat composition as scientific evidence suggests that diets high in saturated fat are associated with increased levels of blood total and low density lipoproteins, which are associated with increased risk of cardiovascular disease (Webb & O'Neill, 2008). Coronary heart disease is a major public health concern, as it accounts for more deaths than any other disease or group of diseases (British Heart Foundation, 2006). Thus, a lower saturated fatty acids (SFA) and a higher polyunsaturated fatty acids (PUFA) intake, especially of n-3 fatty acids (FA) to achieve an appropriate n-6/n-3 ratio (30 months of age) non-lactating, non-pregnant beef cows with body weight averaging 620 ± 62 kg were used. Cows were cared for according to Canadian Council on Animal Care guidelines (CCAC, 1993) and fed at the Lethbridge Research Centre. Cows were randomly assigned to one of four diets, with four pens of four cows per diet. Cows had ad libitum access to feed and water. Diets were designed to meet or exceed nutrient requirements for mature cows (Nassu et al., In Press; NRC, 2000) and consisted of 50:50 forage to concentrate (dry matter basis) and were fed as total mixed rations. Diets included hay control, barley silage control, hay plus flaxseed and barley silage plus flaxseed. Flaxseed was ground together with barley in a 7:3 ratio and flaxseed diets contained 15% flax substituted for dry rolled barley (dry matter basis). Diets were fed for 20 weeks. Duringthe study two animals were withdrawn due to lameness, one each from the silage and the silage plus flaxseed treatments.

2.2. Slaughter and sample collection

Animals were slaughtered at the Lacombe Research Centre. At 24 h post mortem, approximately 200 g of subcutaneous fat was removed from the 12th rib and stored at -80 ºC for subsequent fatty acid determinations and NIR spectral analysis.

2.3. Fatty acid analysis

From the subcutaneous fat collected, five grams were freeze dried and subsampled for fatty acid analysis according to Aldai, Dugan, Rolland, and Kramer (2009).

2.4. Spectra collection

Subcutaneous fat for NIRS analysis was thawed overnight at +2 ºC. Duplicate intact circular fat cores were obtained with the help of a custom-constructed stainless steel device (Figure 1a) to enable consolidation of fat and produce fat discs of an appropriate diameter (38 mm) and thickness (7 mm) to fit the ring cups of the NIRS machine (Figure 1b). Each cold fat disc was placed in a ring cup, all visible air bubbles removed by squeezing, and the cup backed with thin black foam (Figure 1c). NIR spectra were collected when the subcutaneous fat samples were at 2 ºC, hereafter referred to as “cold samples”. Subsequently, the cold samples were placed in open plastic bags and heated in a water bath at 35º C. A DuaLogR model 600-1050 (Barnant Company Barrington, USA) thermocouple was inserted into the center of each fat sample for temperature monitoring during warming. As soon as the core sample reached the target endpoint temperature (31º C), samples were immediately removed from the water bath and NIR spectra were collected from these “warm samples”. The aim of using two temperatures was to know at which point in the slaughter chain NIR could be used on-line. The temperature of the warm samples approximates the temperature of subcutaneous fat immediately after skinning, and the temperature of the cold sample mirrored that which would be obtained after carcasses were stored in a cooler for 24 h. Subcutaneous fat sample was scanned 32 times over the range (400-2498 nm) using a NIRSystems Versatile Agri Analyzer (SY-3665-II Model 6500, FOSS, Sweden), and spectra averaged by the equipment software. Two fat samples per animal were scanned using two different cells, and each sample was scanned twice (resulting in four average spectra per cow). This approach increased the area of the subcutaneous fat scanned and reduced the sampling error (Downey & Hildrum, 2004). The four reflectance spectra of each sample were visually examined for consistency and then averaged, with the mean spectrum being used to predict the fatty acid content of each subcutaneous fat sample. The spectrometer interpolated the data to produce measurements in 2 nm steps, resulting in a diffuse reflectance spectrum of 1050 data points. Absorbance data were stored as log (1/R), where R is the reflectance. Instrument control and initial spectral manipulation were performed with WinISI II software (v1.04a; Infrasoft International, Port Matilda, MD).

2.5. Data analysis

Calibration and validation of the NIRS data were performed using The Unscrambler® program (version 8.5.0, Camo, Trondheim, Norway). The detection of anomalous spectra was accomplished using the Mahalanobis distance (H-statistic) to the centre of the population, which indicates how different a sample spectrum is from the average spectrum of the set (Williams & Norris, 2001). A sample with an H statistic of ≥ 3.0 standardized units from the mean spectrum was defined as a global H outlier and was eliminated from the population. In addition, some samples were removed from the initial data set as concentration outliers (T-statistic), which measures how closely the reference value matches the predicted value. Hence, the samples whose predicted values exceed 2.5 times the standard error of estimation were considered as T statistic outliers and excluded from the population. Spectral data were subjected to multiplicative scatter correction (MSC; Dhanoa, Lister, Sanderson, & Barnes, 1994) to reduce multicolinearity and the effects of baseline shift and curvature on spectra arising from scattering effects due to physical effects. First or second order derivatives (Shenk, Westerhaus, & Workman, 1992) were applied to the spectra to increase the resolution of spectral peaks, and heighten signals related to the chemical composition of subcutaneous fat samples (Davies & Grant, 1987). Partial least square regression type I (PLSR1) was used for predicting FA concentration using NIR spectra as independent variables. Internal full cross-validation was performed to avoid over-fitting the PLSR equations. Thus, the optimal number of factors in each equation was determined as the number of factors after which the standard error of cross-validation no longer decreased.

The predictive ability of the PLS calibration models was evaluated in terms of coefficient of determination (R2), root mean square error of cross-validation (RMSECV) (Westerhaus, Workman, Reeves III, & Mark, 2004) and ratio performance deviation (RPD) (Williams, 2001 & 2008). RMSECV and RPD are regarded as measures of precision and accuracy of prediction and are defined by:

where n is the number of samples in the calibration set, the yi represents the real (measured) responses, the [pic] represents the estimated responses obtained via cross-validation and SD is the standard deviation of the reference values of the calibration set. Williams (2001 & 2008) suggested that the RPD statistic should be equal or larger than 2, since lower RPD values could be attributed either to a narrow range of the reference values (giving a small SD) or to a large error in the estimation (RMSECV) compared to SD (Tøgersen, Arnesen, Nielsen, & Hildrum, 2003).

In order to discriminate among subcutaneous fat samples from beef cows fed different diets (hay/barley silage with or without flaxseed supplementation) by NIR spectra, discriminant analysis was performed using the dummy regression technique on the absorbance data with The Unscrambler® software (version 8.5.0, Camo, Trondheim, Norway) (Cozzolino, De Mattos, & Martins, 2002; Cozzolino, & Murray, 2004). The subcutaneous fat samples were identified with dummy variables (hay/barley silage = 1, hay/barley silage with flax = 2) and PLSR was used to generate a mathematical model that was cross-validated (leave one-out) to select the most relevant PLS components. According to this equation, a sample was classified as subcutaneous fat belonging to a specific category (hay/barley silage or hay/barley silage with flax) if the predicted value was within ±0.5 of the dummy value.

3. Results and discussion

3.1. Chemical data

Ranges, means, standard deviations (SD) and coefficients of variation (CV) of PUFAs and their biohydrogenation intermediates from subcutaneous fat are summarized in Table 1. In general terms, the concentrations of FA in the subcutaneous fat were within the normal range of variation reported by other authors in the subcutaneous adipose tissue of beef (Noci, Monahan, French, & Moloney, 2005; Dugan, Rolland, Aalhus, Aldai, & Kramer, 2008). The results revealed wide variability, which is important when searching for calibration equations to be used for predictions. The causes of such variability resulted from the different feeding regimes used in the study. Hence, the CV were higher than 50% for most of the FA and even higher than 100% for C20:3n-3, total conjugated linolenic acids (CLNA), c9,t11,t15-18:3 and c9,t11,c15-18:3.

The n-6:n-3 FA ratio is often used to evaluate the nutritional quality of fat. In this study, the n-6:n-3 ratio was 2.6 (Table 1), a value considered suitable according to the recommendation of the World Health Organization ( ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download