Cholesterol Content and Methods for Cholesterol Determination in Meat and Poultry

Authors

  • Thu T. N. Dinh,

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author
  • Leslie D. Thompson,

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author
  • Michael L. Galyean,

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author
  • J Chance Brooks,

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author
  • Kristine Y. Patterson,

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author
  • L. Mallory Boylan

    1. Authors Dinh, Thompson, Galyean, and Brooks are with Dept. of Animal and Food Sciences, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Author Patterson is with Nutrient Data Laboratory, United States Dept. of Agriculture, Beltsville, MD 20705, U.S.A. Author Boylan is with Dept. of Nutrition, Hospitality, and Retailing, Texas Tech Univ., Lubbock, TX 79409, U.S.A. Direct inquiries to author Thompson (E-mail: leslie.thompson@ttu.edu).
    Search for more papers by this author

Abstract

Abstract:  Available data for cholesterol content of beef, pork, poultry, and processed meat products were reported. Although the cholesterol concentration in meat and poultry can be influenced by various factors, effects of animal species, muscle fiber type, and muscle fat content are focused on in this review. Oxidative red muscles tend to have greater total lipid and cholesterol contents, although differences in the same types of muscles or cuts have been reported. Moreover, contradictory results among various studies suggest that unless there are pronounced changes in muscle structure and composition, cholesterol content is unlikely to be affected. Second, multiple issues in cholesterol analysis, including sample preparation, detection, and quantification, were evaluated. Cholesterol content of meat and poultry has been determined mostly by colorimetry and chromatography, although the latter has become predominant because of technological advances and method performance. Direct saponification has been the preferred method for hydrolyzing samples because of cost- and time-effectiveness. The extraction solvent varies, but toluene seems to provide sufficient recovery in a single extraction, although the possible formation of an emulsion associated with using toluene requires experience in postsaponification manipulation. The most commonly used internal standard is 5α-cholestane, although its behavior is not identical to that of cholesterol. Cholesterol can be analyzed routinely by gas chromatography (GC)-flame ionization detector without derivatization; however, other methods, especially high-performance liquid chromatography (HPLC) coupled with different detectors, can also be used. For research purposes, HPLC-ultraviolet/Visible/photodiode array detector with nondestructiveness is preferred, especially when cholesterol must be separated from other coexisting compounds such as tocopherols. More advanced methods, such as GC/HPLC-isotope dilution/mass spectrometry, are primarily used for quality control purposes.

Introduction

Since the relationship between plasma cholesterol concentration and atherosclerosis was demonstrated in rabbits in 1913 (Vance and van den Bosch 2000), the interest in cholesterol content in foods has been driven by the awareness of the association between dietary cholesterol and human disease. As a result, cholesterol has become an important component in composition studies on meat and poultry products. The USDA Natl. Nutrient Database for Standard Reference (SR) reports cholesterol content of various beef, pork, and chicken products and updates the information periodically (USDA 2011). The SR provides public information on nutritional composition and such information can be used for the mandatory nutrition labeling (FDA 1993). Cholesterol data are usually accompanied by fat content, fatty acid composition, and type of meats (species or cuts; Rhee and others 1982a, 1982b; Kregel and others 1986; Mustafa 1988; Park and others 1991; Perkins and others 1992; Abu-Tarboush and Dawood 1993; Akinwunmi and others 1993; Lan and others 1993; Lewis and others 1993; Rule and others 1997, 2002; Bales and others 1998; Dorado and others 1999; Rowe and others 1999; Ayerza and others 2002; Baggio and others 2002; Girolami and others 2003; Alfaia and others 2006; Padre and others 2006; Polak and others 2008; Ponte and others 2008; Muchenje and others 2009). Cholesterol intake is also emphasized in the Dietary Guidelines for Americans, which recommends that less than 300 mg/d be consumed (USDA/HHS 2010).

Fundamental research on biological functions and chemical properties of cholesterol has been the driving force for the development and application of analytical technologies for the determination of cholesterol. Sophisticated techniques, such as isotopic tracing and mass fragmentation, were originally developed for qualitative purposes to study biosynthetic pathways or absorption and metabolism of cholesterol, although they have been adopted for quantitative purposes. Before the 1990s, much of the cholesterol data for meat and poultry products were generated using colorimetric measurement (Sweeney and Weihrauch 1976; Bohac and Rhee 1988; Hoelscher and others 1988; Beyer and Jensen 1989; Browning and others (1990); Sweeten and others 1990; Swize and others 1992; Akinwunmi and others 1993). Most chromatographic methods for cholesterol analysis in meat and poultry were initially developed for blood serum (Abell and others 1952; Gambert and others 1979; Cohen and others 1980; Iwata and others 1987; Pelletier and others 1987; Takatsu and Nishi 1987; Eckfeldt and others 1991; Nakajima and others 1995). The AOAC Intl. adopted the 1st validated cholesterol determination method for foods in 1976 (AOAC International 1996). Although the method was later modified by employing direct saponification (AOAC Official Method 994.10; Klatt and others 1995), it was validated with only one meat matrix. The use of these methods to analyze cholesterol in meat and poultry samples over the years on a routine basis and for research purposes has clarified and resolved some of the most important technical issues such as extraction solvent, saponification conditions, postsaponification manipulation, necessity of derivatization, use of internal standard, and effects of the sample matrix.

With U.S. per capita consumption of beef, pork, and poultry being approximately 29.5, 22.7, and 47.1 kg, respectively (AMI 2010), the cholesterol content of meat and poultry is of great public interest. The demand for accurate cholesterol determination has been met with recently developed technologies for chromatography, detection, and measurement, which has allowed for incredible specificity, sensitivity, accuracy, and precision. This review examines the cholesterol content of meat and poultry in light of recent data and discusses the controversial effects of fat content and type of meats on cholesterol content. Also discussed are several problematic issues that relate to cholesterol analytical procedures, especially with application to meat and poultry matrices.

Cholesterol Content of Meat and Poultry

Overview of cholesterol content in beef, pork, poultry, and processed meats

Major sources of cholesterol in the human diet are meat from domestic livestock, although seafood is also rich in cholesterol. Cholesterol content of raw and cooked meat and poultry products ranges from 40 to 90 mg/100 g (Chizzolini and others 1999; Mourot and Hermier 2001; Piironen and others 2002; Valsta and others 2005; Bragagnolo 2009; Honikel 2009). Recent data from our laboratory indicate upper concentrations of up to 150 mg/100 g for cooked chicken dark meat (USDA 2011, data under review). Bone marrow and organs, such as liver, kidney, or brain, contain a much greater content, up to several hundred milligrams per 100 g (Kunsman and others 1981; Mustafa 1988; Park and others 1991; Mourot and Hermier 2001; Williams 2007). Processed meat products contain from less than 50 mg/100 g to more than 150 mg/100 g, depending on their formulation (Piironen and others 2002; Valsta and others 2005; Bragagnolo 2009; Honikel 2009).

Cholesterol content of raw and cooked bovine meats ranges from 43 to 84 mg/100 g and 57 to 101 mg/100 g, respectively (Table 1; Chizzolini and others 1999; Dinh 2010). Multiple factors affect the cholesterol content of beef, such as gender, animal maturity, degree of marbling, subcutaneous fat thickness, animal breed, dietary energy level, different feeding treatments (restricted diet or ad libitum), and muscle location (type of cut) (Rhee and others 1982b; Eichhorn and others 1986a, 1986b; Wheeler and others 1987; Bohac and Rhee 1988; Agboola and others 1990; Browning and others 1990; Abu-Tarboush and Dawood 1993; Akinwunmi and others 1993; Lewis and others 1993; Morris and others 1995; Stromer and others 1966; Chizzolini and others 1999; Engle and others 2000; Andrae and others 2001; Bragagnolo and Rodriguez-Amaya 2003b; Cifuni and others 2004; Padre and others 2006; Bragagnolo 2009; Duckett and others 2009; Muchenje and others 2009). Cholesterol content of pork, 30 to 81 mg/100 g for raw pork and 56 to 113 mg/100 g for cooked pork (Slover and others 1987; Bales and others 1998; Buege 1998; Chizzolini and others 1999; Dorado and others 1999; Piironen and others 2002; Bragagnolo 2009; Sinclair and others 2010), is generally lower than that of beef, although some studies indicated no significant difference between the 2 types of meat (Bohac and Rhee 1988; Bragagnolo 2009). Cholesterol in pork is influenced by maturity, type of cut, fat thickness, animal diet, degree of marbling, and genetic variation (Kellogg and others 1977; Harris and others 1993; Fernandez and others 1995; Buege 1998; Hernandez and others 1998; Dorado and others 1999; Fernandez 1999; Bragagnolo and Rodriguez-Amaya 2002; Bragagnolo 2009; Cannata and others 2010). Cholesterol content of pork also varies depending on specific treatment conditions, such as the use of ractopamine, a repartitioning agent, which was suggested to decrease the cholesterol concentration of pork longissimus muscles because of its effects on fat deposition (Perkins and others 1992). However, Engeseth and others (1992) could not obtain similar results in pork longissimus muscles.

Table 1–.  Cholesterol concentration, fat content, and muscle oxidative pattern of various muscles and cuts from various meat and poultry species.
Source description (species, muscles, or cuts)Oxidative pattern1Fat content (%)Cholesterol content2 (mg/100 g)Ref.3
  1. *Cholesterol contents of PG and PO muscles did not differ (P > 0.05) and were pooled.

  2. 1Claasification (N/A = not applicable; PG = predominantly glycolytic; PO = predominantly oxidative; IO = intermediately oxidative) and description of muscle oxidative patterns: in the same reference as cholesterol content, and by the following authors: chicken: Barnard and others 1982, Smith and others 1993; pork: Laborde and others 1985, Flores and others 1996, Klont and others 1998, Karlsson and others 1999; beef: Klont and others 1998; Kirchofer and others 2002; rhea and ostrich: Velotto and Crasto 2004; deer: Tarrant and others 1972, Jurie and others 1995, Brandstetter and others 1997, Jurie and others 2007.

  3. 2Determined by chromatographic method (GC or HPLC).

  4. 3Reference: Alasnier and others 1996 (1); Alfaia and others 2006 (2); Ang and Hamm 1982 (3); Ayerza and others 2002 (4); Baggio and others 2002 (5); Cannata and others 2010 (6); Costa and others 2006 (7); Costa and others 2009 (8); Dinh 2010 (9); Dorado and others 1999 (10); Hamm and Ang 1984 (11); Hernandez and others 1998 (12); Horbariczuk and others 1998 (13); Komprda and others 2003 (14); Muchenje and others 2009 (15); Padre and others 2006 (16); Paleari and others 1998 (17); Piironen and others 2002 (18); Polak and others 2008 (19); Ponte and others 2004 (20); Ponte and others 2008 (21); Pratiwi and others 2006 (22); Rowe and others 1999 (23); Rule and others 2002 (24); Ryan and Gray 1984 (25); Sales 1998 (26); Sales and others 1999 (27); Salvatori and others 2004 (28); Salvatori and others 2008 (29); Sinclair and others 2010 (30); Solomon and others 1990 (31); Solomon and others 1991 (32).

Beef, brisketN/A145218
Beef, chuckN/A6.85518
Beef, longissimus musclePG1.1 to 3.250 to 537, 8, 18, 24
Beef, longissimus musclePG1.74616
Beef, longissimus musclePG3.44616
Beef, longissimus thoracisPG1.1 to 1.536 to 412, 15
Beef, pectineusPG4.55017
Beef, semitendinosusPG0.9352
Beef, biceps femorisPO1.4 to 1.850 to 557, 8
Beef, supraspinatusPO2.1 to 2.453 to 587, 8
Bison, longissimus musclePG1.1 to 2.144 to 5424
Pork, bellyN/A10.3 to 31.970 to 12010
Pork, bellyN/A35.15418
Pork, spare ribsN/A5.1 to 17.646 to 10210
Pork, biceps femorisIO3.15212
Pork, cutsPG or PO1.1 to 2.343 to 6030
Pork, longissimus musclePG1.6 to 2.739 to 4912, 29
Pork, loin, chop, longissimus musclePG1.1 to 7.131 to 6210, 18
Pork, loin, longissimus musclePG2.0 to 3.6102 to 1096
Pork, tenderloin, psoas majorPG3.3 to 7.445 to 9110
Pork, legPO2.2 to 12.943 to 7610
Pork, triceps brachiiPO3.25112
Chicken, hand-deboned meatN/A11.6 to 29.875 to 983
Chicken, mechanically deboned meatN/A13.2 to 25.294 to 1293
Chicken, breast meatPG1.0 to 2.747 to 5611, 14, 18, 20, 21
Chicken, breast meatPG0.85924
Chicken, white meatPG6.5 to 9.150 to 514
Chicken, dark meatPO13.3 to 19.054 to 574
Chicken, leg and thighPO11.28418
Chicken, thigh meatPO7.28314
Turkey, breast meatPG1.55314
Turkey, separable breast meatPG0.5275
Turkey, separable leg meatPO1.1355
Turkey, separable wing meatPO0.9465
Turkey, thigh meatPO3.83717
Turkey, thigh meatPO2.46214
Lamb, longissimus musclePG3.8 to 6.960 to 7023, 28, 31
Lamb, longissimus musclePG4.9 to 5.270 to 7732
Lamb, longissimus musclePG10.85823
Lamb, semimembranosusPG2.4 to 2.955 to 7528
Lamb, semimembranosusPG4.0 to 4.472 to 7832
Lamb, triceps brachiiPO4.7 to 5.471 to 7632
Goat, castrated, longissimus thoracisPGN/A55 to 6022
Goat, castrated, biceps femorisPON/A65 to 8322
Goat, castrated, infraspinatusPON/A71 to 8822
Elk, longissimus musclePG0.85024
Ostrich, gastrocnemiusPG1.3 to 1.566 to 6813
Ostrich, gastrocnemiusPGN/A5926
Ostrich, iliofibularisPO1.4 to 1.563 to 6613
Ostrich, iliofibularisPON/A61.526
Ostrich, thighPO1.63417
Rabbit, gastrocnemius lateraleIO1.7671
Rabbit, longissimus lumborumPG1.2451
Rabbit, psoas majorPG1.2501
Rabbit, semimembranosus propriosusPO4.4801
Rabbit, soleusPO4.8781
Red deer, hinds, semitendinosusPG1.77619
Red deer, hinds, triceps brachiiPO1.38719
Red deer, stags, semitendinosusPG1.17319
Red deer, stags, triceps brachiiPO1.27419
Rhea, meat (glycolytic gastrocnemius; oxidative iliofibularis)*Pooled PG and PO1.2 to 1.355 to 5927
Fat, beef and pork, 50 of 50, separableN/A70.0 to 71.0759
Fat, beef, subcutaneousN/A85.9959
Fat, beef, tallowN/AN/A14025
Fat, lamb, subcutaneousN/A62.5 to 76.371 to 8031
Fat, lamb, subcutaneousN/A57.9 to 79.872 to 9032
Fat, pork, separableN/A44.6 to 77.138 to 7630
Skin, turkeyN/A12815

It is more difficult to compare cholesterol content of poultry to that of beef and pork because poultry products sometimes contain skin, which is high in cholesterol (approximately 80 to more than 100 mg/100 g; Bragagnolo 2009). In general, raw poultry meat has approximately 27 to 90 mg cholesterol/100 g and cooked poultry meat contains around 59 to 154 mg/100 g (Chizzolini and others 1999; Bragagnolo 2009). Poultry has a comparable cholesterol content to that of beef and pork (Horbariczuk and others 1998; Paleari and others 1998; Chizzolini and others 1999; Piironen and others 2002; Rule and others 2002; Hur and others 2007; Bragagnolo 2009); however, a statistically meaningful comparison could not be made. A significant factor affecting cholesterol content of poultry is type of retail cut because of the difference between dark and white chicken meat and the presence of skin in many retail cuts. Poultry skin has the greatest cholesterol concentration compared with poultry meat or poultry fat. Cholesterol content of visible fat and breast meat is similar to or lower than that of dark meat (Ang and Hamm 1982; Smith and others 1993; Wong and others 1993; Baggio and others 2002; Komprda and others 2003). Moreover, the difference in cholesterol content between white and dark poultry meat is more pronounced than that between white (predominantly glycolytic) and red (predominantly oxidative) muscles in beef and pork (Browning and others 1990; Smith and others 1993; Buege 1998; Sinclair and others 2010). Nonetheless, the difference between dark and white meat was not always consistent among studies (Ayerza and others 2002). Van de Bovenkamp and Katan (1981) also suggested that the high cholesterol content of chicken skin that had been reported up to that date was erroneous, and they reported an analytical value for chicken skin of 71 mg/100 g of raw wet tissue. In contrast, data collected in our laboratory (using gas chromatography with flame ionization detector, GC-FID, for cholesterol determination) consistently showed greater concentrations of cholesterol in raw and cooked chicken skin (more than 100 mg/100 g). It is noteworthy, however, that the cholesterol content of the skin from enhanced chicken was about 10 to 15 mg/100 g lower than that of the skin from nonenhanced chicken (USDA 2011, data under review). Mechanically deboned poultry meat has a greater cholesterol content than hand-deboned meat (Ang and Hamm 1982; King and others 1988; Wong and others 1993; Baggio and others 2002). Such an increase in cholesterol content is probably caused by the presence of bone marrow, which is similar to mechanically deboned beef (Kunsman and others 1981). Dietary treatments, such as feed rations, also have been studied extensively in an effort to improve lipid quality and decrease cholesterol content in poultry meats (Konjufca and others 1997; Ayerza and others 2002; Ponte and others 2004, 2008; Hur and others 2007; Simsek and others 2009).

Cholesterol contents of processed meat products greatly vary from 23 to 144 mg/100 g (Bragagnolo 2009), primarily as a result of variation in ingredients, formulation, type of meats or muscles used, cooking or heating processes applied, storage, and oxidation of cholesterol. Cooked or processed meat products usually have a greater cholesterol content than raw meat because of moisture loss with cholesterol being retained in the tissues (Rhee and others 1982a, 1982b; Kregel and others 1986; Browning and others 1990; Bragagnolo and Rodriguez-Amaya 2003b; Baggio and Bragagnolo 2006), despite the fact that some of cholesterol is also lost during cooking. The migration of cholesterol from fat tissues to muscle tissues was used as an explanation for greater cholesterol content of cooked meats (Swize and others 1992). When reviewing cholesterol data of cooked and raw chicken dark meat (skin-on and skinless) for the USDA Natl. Nutrient Database for SR, we also noted some slight decrease in the cholesterol:solid ratio in the separable meat portion, which might be attributed to migration of fat from skin during cooking. However, trimming has not been documented to affect cholesterol concentrations in cooked meats (Rhee and others 1982a; Swize and others 1992; Akinwunmi and others 1993; Chizzolini and others 1999; Bragagnolo 2009). Cholesterol in processed meat products can be artificially modified by changing the proportions of meat and nonmeat ingredients. Extensive summaries of cholesterol contents of muscle foods by Chizzolini and others (1999), Piironen and others (2002), and Bragagnolo (2009) suggested that differences in the cholesterol content of meat and poultry among various studies could be caused by analytical methodology.

Effects of species on cholesterol content

Cholesterol is present in muscle as well as adipose tissues because it is an essential component of cell membranes and can be stored as cholesterol esters in lipid droplets (Dessi and Batetta 2003). The differences in cholesterol content among species are generally explained by variations in absorption and biosynthesis of cholesterol, lipoprotein metabolism, diet, muscle fiber type distribution, genetic variation, subcutaneous and intramuscular fat, body weight (Klein and Rudel 1983; Dietschy 1984; Dietschy and others 1993; Spady and others 1993; Rule and others 1997; Chizzolini and others 1999; Dorado and others 1999; Mourot and Hermier 2001; Piironen and others 2002; Komprda and others 2003; Salvatori and others 2004; Padre and others 2006; Bragagnolo 2009), as well as cell size. With a significant portion of cholesterol in meat residing in the cell membrane, a difference in cell size and number of cells per unit of muscle volume or weight can lead to a divergence in total membrane surface area and ultimately the content of membrane components, including cholesterol. Le Lay and others (2001) reported a possible link between size of adipocytes and their metabolic activities, which provided evidence of size of adipocytes influencing lipid raft formation and adipocyte metabolism by altering the cholesterol content of membranes.

Among animal species, the rate of cholesterol synthesis in the body is a function of body weight, which inversely and markedly influences the synthesis and turnover of cholesterol in the cellular membrane. Animal size also tremendously affects the removal rate of cholesterol in low-density lipoprotein (LDL) from plasma, and ultimately the LDL cholesterol turnover, to an even greater extent than it does the cholesterol synthesis (Dietschy and others 1993). The inverse effect of body weight on LDL cholesterol uptake was detected largely in the liver (receptor-mediated process). The uptake pattern and extent seem to vary in adrenal gland and ovary (receptor-mediated) regardless of animal size; however, it is similar in other tissues such as the intestines (receptor-independent), adipose tissues, and muscles. Muscle tissues synthesize most of their cholesterol demand (Dietschy and others 1993). An interesting review by Spady and others (1993) indicated that more than 50% of de novo synthesized cholesterol in animals and humans comes from extra-hepatic organs. In fact, if dietary cholesterol consumption ranges from 100 to 300 mg/1000 kcal intake, more than 90% of biosynthetic cholesterol is from extra-hepatic tissues because cholesterol synthesis in the liver is suppressed. Nonetheless, de novo synthesis of cholesterol varies among species, depending on the response of the liver to dietary cholesterol. Moreover, ruminant species are typically fed low cholesterol diets; therefore, the de novo synthesis in their intestines and extra-hepatic tissues greatly contributes to total body cholesterol (Vernon and Flint 1988). Interestingly, Pratiwi and others (2006) observed an increase in total cholesterol concentration in muscles taken from younger and lighter castrated Boer goats compared with older and heavier counterparts, which might indicate an effect of weight × age interaction on the cholesterol content of muscle. Although acknowledging that cholesterol content of muscles varies among different species, Chizzolini and others (1999) claimed that the magnitude of the differences was relatively low. Beef, veal, pork, and mutton muscles were determined to have cholesterol concentrations of 60 to 70 mg/100 g, whereas cholesterol contents of their intermuscular fat ranged from 75 to 99 mg/100 g (Chizzolini, and others 1999). Data from the 1990s and earlier indicated that bovine meats generally have greater cholesterol than chicken and pork (Mourot and Hermier 2001). Adding more evidence to cholesterol variation among species, Paleari and others (1998) reported a markedly greater cholesterol content of bovine meats compared with poultry meats (ostrich and turkey). On the other hand, Rule and others (2002) found that the cholesterol concentration of chicken breast meat was greater than that of longissimus and semitendinosus muscles from bison, beef cattle, and elk. Considerable evidence suggests that within- and among-species differences in cholesterol content greatly depend on type or location of muscles (Table 1; Chizzolini and others 1999; Bragagnolo 2009).

Muscle oxidative patterns and cholesterol content

Difference in muscle fiber types (such as among muscles from beef chuck and round; Kirchofer and others 2002) has been reported to cause divergence in cholesterol content of meat from different species and meat from different anatomical locations of the same species (Table 1). Effects of muscle fiber type have been found across species such as beef cattle, swine, chicken (Chizzolini and others 1999; Bragagnolo 2009), and others (rabbit, Alasnier and others 1996; ostrich, Sales 1998; goat, Pratiwi and others 2006; deer, Polak and others 2008). The effects of fiber type seem to be caused by size of muscle fibers and are associated with the amount of lipids accumulating in the muscles, which have shown that oxidative muscles with red muscle fiber, smaller fiber diameter, and greater fat content tend to have more total cholesterol. In a study focused on fiber types of rabbit muscles (glycolytic: longissimus lumborum and psoas major; oxidative: soleus and semimembranosus propriosus; intermediate: gastrocnemius laterale), Alasnier and others (1996) concluded that oxidative muscles contained more total lipids, triglycerides, and cholesterol than the glycolytic ones. Smith and others (1993) revealed that broiler pectoralis with 100% white fiber had 50% lower cholesterol content than duckling pectoralis with 16% white fiber. The broiler pectoralis also had significantly lower fat and collagen contents (1.6% and 1.27 mg/g, respectively) compared with the duckling pectoralis (2.3% and 1.75 mg/g, respectively). The authors also reported that broiler white muscle fibers were significantly bigger than duckling white muscle fibers. When examining the lipid composition of predominantly glycolytic longissimus lumborum and predominantly oxidative semispinalis capitis of Large White pigs, Fernandez and others (1995) found the same pattern with muscle consisting predominantly of red fibers having greater total lipids and cholesterol. Similarly, Hernandez and others (1998) found differences in the cholesterol content of porcine skeletal muscles with different oxidative patterns (classified by Laborde and others 1985 and Flores and others 1996). The intermediately oxidative and the predominantly oxidative muscles (biceps femoris and triceps brachii, respectively) had greater cholesterol concentrations than the glycolytic muscle (longissimus dorsi). Among 10 beef muscles that were investigated for effects of leanness on muscle characteristics, Browning and others (1990) reported that the cholesterol content of raw supraspinatus muscles (76.29 mg/100 g; intermediate, Kirchofer and others 2002) was greater than that of other muscles except for infraspinatus muscles (75.18 mg/100 g; oxidative, Kirchofer and others 2002). Cholesterol content on dry matter basis also varied among muscles (Browning and others 1990). In agreement, Rule and others (2002) reported the highest cholesterol concentration in supraspinatus compared with longissimus dorsi (glycolytic, Kirchofer and others 2002) and semitendinosus (glycolytic, Kirchofer and others 2002) from both feedlot bison and steers. Meat from castrated Boer goats also differed in total cholesterol concentration, depending on muscle type. Again, cholesterol content of predominantly oxidative biceps femoris and infraspinatus was greater than that of the predominantly glycolytic longissimus thoracis (Pratiwi and others 2006).

It is interesting that not all studies have come to such conclusive results with respect to the effects of muscle fiber type on cholesterol content. Triceps brachii (oxidative) from red deer (Cervus elaphus) had a greater total cholesterol concentration than semitendinosus (glycolytic, classified by Tarrant and others 1972; Jurie and others 1995; Brandstetter and others 1997; Jurie and others 2007) for hinds, but not for stags and calves. Overall, there was no clear muscle-type effect, but rather a significant interaction between muscle type and treatment groups (stags, hinds, and calves; Polak and others 2008). Browning and others (1990) found that although cholesterol content varied among muscles from different anatomical locations, it did not differ between longissimus and triceps brachii (intermediate, Kirchofer and others 2002). This result conflicts with the findings of Eichhorn and others (1986a, 1986b) and other previously discussed studies. Furthermore, some cholesterol values reported by Browning and others (1990) did not agree with those of Eichhorn and others (1986a), although both studies determined cholesterol concentration spectrophotometrically. The disagreement was found across species regardless of treatments of diet, gender, or breed. Sales and others (1999) found no difference in cholesterol content across 5 different muscles of 2 rhea species. It should be noted that the intramuscular fat contents of the 5 muscles under investigation also did not differ. Despite differing subspecies and feeding regimens, the cholesterol contents of gastrocnemius and iliofibularis from ostrich (glycolytic and oxidative, respectively, classified by Velotto and Crasto 2004) did not differ (Horbariczuk and others 1998). Later, Girolami and others (2003) came to the same conclusion that these 2 muscles, although differing in muscle fiber types, had similar cholesterol contents.

Within traditional meat animal species, such as cattle and swine, interesting results have also been reported. Although the proportion of oxidative muscle fibers was greater, the supraspinatus and the biceps femoris of Barrosa calves had the same cholesterol content as longissimus dorsi (Costa and others 2009). We summarized some interesting data (without meta-analysis) indicating that muscles with predominantly oxidative fibers do not always have greater cholesterol content than muscles with predominantly glycolytic fibers (Table 1). Cannata and others (2010) reported a cholesterol concentration of 102 to 109 mg/100 g for pork longissimus muscle (2.0% to 3.6% fat), and some pork tenderloins (psoas major, predominantly glycolytic) had up to 90 mg cholesterol/100 g (Dorado and others 1999), whereas most predominantly oxidative muscles or cuts have been found to have cholesterol concentrations of 50 to 70 mg/100 g (Table 1). Ayerza and others (2002) did not find any significant difference in cholesterol content of chicken dark and white meat (54 to 57 mg/100 g and 50 to 51 mg/100 g, respectively). Although Solomon and others (1991) did not design their study to determine differences in lipid composition among muscles from lamb, they did report cholesterol concentrations of 70 to 78 mg/100 g for predominantly glycolytic longissimus and semimembranosus muscles (4.0% to 5.2% fat) and 71 to 76 mg/100 g for predominantly oxidative triceps brachii muscle (4.7% to 5.4% fat). Hence, it was not surprising that Bohac and Rhee (1988) did not find any change in cholesterol content across 3 different anatomical locations (longissimus dorsi, semimembranosus, and semitendinosus) of steer and pig carcasses, regardless of dietary treatments. Cholesterol content of beef and pork muscles did not differ in the same study. The same 3 muscles from Podolian young bulls were determined to have similar cholesterol content by Cifuni and others (2004), although muscles from bulls and steers (Bohac and Rhee 1988) may not have comparable lipid content and composition. Even in the previously discussed study (Browning and others 1990), despite significant differences in cholesterol concentrations in various muscles, the longissimus dorsi, semimembranosus, and semitendinosus did not differ. It is important to recognize that the similarity emerging in the findings from Bohac and Rhee (1988), Browning and others (1990), and Cifuni and others (2004) results from the fact that the 3 muscles in all 3 studies belong to 1 species (bovine meat) and are all predominantly glycolytic (Kirchofer and others 2002), although the fiber-type distribution might vary among the muscles and the animals used in each study.

The contradictory findings among studies on cholesterol content of different muscles raise the questions of whether biological characteristics of a muscle have a significant impact on its lipid composition, especially cholesterol content, and whether the relationship among muscle characteristics, lipid content, and cholesterol concentration is predictable. To explain the difference in cholesterol content possibly caused by 3 different muscle oxidative patterns (oxidative, glycolytic, and intermediate), Alasnier and others (1996) hypothesized that cholesterol content had to be proportional to the amount of phospholipids to maintain proper membrane fluidity. The evidence of phospholipids:cholesterol ratio was also provided to support the hypothesis. Similarly, Leseigneur-Meynier and Gandemer (1991) reported that total phospholipid content was related to metabolic type. In fact, Yeagle and others (1988) and Sotomayor and others (2000) presented evidence of cholesterol:phospholipids mole ratio and cholesterol content of cell membrane influencing the activity of Na+/K+-ATPase, which was influenced by metabolic types of muscles (Briand and others 1981; Klont and others 1998; Lefaucheur 2010). In addition, in studies reporting contradictory data on the effects of muscle fiber type and lipid composition on cholesterol content (for example, no effect), the values of lipid content, phospholipid ratio, and fatty acid composition (degree of saturation, percentage of polyunsaturated fatty acids, P/S ratio, and so on) in the investigated muscles were similar entirely (Bohac and Rhee 1988; Browning and others 1990; Horbariczuk and others 1998; Sales and others 1999) or similar partially (Girolami and others 2003; Cifuni and others 2004; Polak and others 2008; Costa and others 2009); however, none of the studies differed in all values.

Rule and others (2002) found that muscles from bison and steers had greater total fatty acid concentrations than chicken breast, but the supraspinatus muscle from feedlot bison, steers, and chicken breast yielded the same cholesterol content. Results from Leseigneur-Meynier and Gandemer (1991) also suggested that intramuscular lipid content as well as triglyceride content was not strictly related to the metabolic type of muscle because the longissimus dorsi, a glycolytic muscle, had an intramuscular fat content close to that of masseter, an oxidative muscle. Hence, in addition to muscle fiber type, the composition of muscle lipids, not the lipid content itself, is more than likely another factor affecting cholesterol content of meat and poultry, and this factor may only be meaningful within a species.

Muscle fat and cholesterol content

Lipid and cholesterol contents of meat and poultry have been of great research interest for decades. This topic has been repeatedly studied and reviewed. The following discussion is conducted for lipid and cholesterol contents only in raw intact muscles because it is less meaningful for ground or processed muscle products, of which the lipid content can be artificially modified. Data on cooked muscles, if used, were only for interpretation purposes associated with cholesterol content of raw muscles.

There has been a definite decrease in fat content of meat and poultry over the years because of health consciousness; however, the trend has not been conclusive for cholesterol content (Chizzolini and others 1999; Jimenez-Colmenero and others 2001), although the fat content, fatty acid composition, and cholesterol content have usually been investigated together. There has been rarely an attempt to manipulate cholesterol content of meat because it is such an important component of the animal cell membrane. Data on the effect of lipid content, as with the effect of muscle fiber type, have been collected under different treatments and conditions (gender, breed, diet, species, cooking, and so on) and provided evidence of not only positive correlation but also insignificant relationship between fat and cholesterol contents (Table 1). Dinh and others (2008) reported a positive correlation between intramuscular fat and cholesterol content. The correlation was found in longissimus muscles from Angus, Brahman, and Romosinuano purebred cattle, among which the Angus longissimus muscles had almost a 50% greater fat content than longissimus muscles from the other 2 breeds. Rule and others (1997) suggested that the difference in cholesterol among bulls grouped by high and moderate growth rates might be caused by the amount of carcass fat. Alfaia and others (2006) reported that the beef longissimus had a greater cholesterol content than the semitendinosus. The authors attributed the contradiction between their results and those from other studies to the greater fat content of the longissimus compared with the semitendinosus. Dorado and others (1999) concluded that a greater fat content of Spanish commercial pork cuts led to a greater cholesterol content, although it is worth noting that some cuts had much more fat than others. Moreover, there is considerable consensus among studies supporting the finding that adipose tissues (70% to 90% extractable lipid) contains a much greater amount of cholesterol than muscle tissues (Rhee and others 1982a, 1982b; Eichhorn and others 1986a, 1986b; Wheeler and others 1987; Solomon and others 1990, 1991; Swize and others 1992; Abu-Tarboush and Dawood 1993; Lan and others 1993), and most of the studies agreed that the markedly greater lipid content of adipose tissues contributes to such a high accumulation of cholesterol. Rhee and others (1982a) found that cholesterol content of fat fractions was 50% (30 mg/100 g) to 100% (60 mg/100 g) greater than that of muscle fractions in beef (on a wet matter basis). A significantly positive correlation between fat and cholesterol contents on a wet matter basis was also reported. Averaged over all yield grades, the combined fat fractions (intermuscular and subcutaneous) constituted about 40% weight and 50% total cholesterol of the whole steak, which shows the significant contribution of fat within the muscle to the cholesterol pool of muscles when there is a marked change in lipid composition.

Browning and others (1990) revealed that although there was a significant difference in cholesterol concentration between 2 types of carcasses in their study (lean and typical; 3.98% and 5.89% fat, respectively; P < 0.05); however, the difference was only 1.84 mg/100 g, a minute amount from a practical standpoint. Among beef longissimus muscles with different marbling degrees, only the steaks with a Practically Devoid marbling score had less cholesterol than steaks with the other scores (Moderately Abundant, Slightly Abundant, Moderate, Modest, Small, Slight, and Traces; Rhee and others 1982b). An interesting point is that on a wet matter basis, the raw Moderately Abundant steaks had approximately 8 g/100 g lower moisture content, 9 g/100 g greater lipid content, and 13 mg/100 g greater cholesterol content than the raw Practically Devoid steaks; however, the cooked (60 and 75 °C) Moderately Abundant steaks had 9 to 10 g/100 g lower moisture content, 15 g/100 g greater lipid content, and an insignificantly different cholesterol content compared with the cooked Practically Devoid steaks. The cholesterol contents of the cooked steaks at the 2 extreme marbling degrees were surprisingly close (89.55 mg/100 g for Moderately Abundant and 85.57 mg/100 g for Practically Devoid). Based on the changes in moisture and lipid contents, it was obvious that highly marbled steaks lost less moisture and more lipid during cooking (gravimetric basis) than steaks with little marbling. Therefore, the similar concentrations of cholesterol in the cooked steaks at 2 extremely different marbling degrees could partially be caused by the loss of cholesterol in form of cholesterol esters in the lipid fraction, in addition to moisture loss. This observation was also supported by the cholesterol content on a dry matter basis. In raw steaks, the Moderately Abundant and Practically Devoid marbling categories had the same cholesterol content on a dry matter basis; however, the cooked Moderately Abundant steaks had significantly less cholesterol (on a dry matter basis) than the cooked Practically Devoid steaks. The change, therefore, seemed to confirm the impact of lipid loss from inter- and intramuscular fat on total cholesterol of the cooked steaks, in addition to moisture loss, during cooking because cholesterol in cell membrane is tightly bound by other lipid and protein components. Such a change was not found in other marbling categories above the Practically Devoid, which suggests that unless there is a dramatic difference in muscle fat, the fat content may not greatly affect cholesterol concentration in muscles.

Indeed, Lewis and others (1993) suggested that an increased muscle fat content was not always an indicator of an increase in cholesterol content. Rowe and others (1999) showed that the longissimus dorsi of lambs fattened in drylot had a greater fat content (P < 0.01) but a lower cholesterol concentration (P < 0.05) than that of lambs fattened on pasture. Although having 7.1% and 35.1% fat, pork chop and pork belly were found to have cholesterol concentrations of 47 and 54 mg/100 g, respectively (Piironen and others 2002), compared with pork biceps femoris (3.1% fat and 52 mg cholesterol/100 g) and triceps brachii (3.2% fat and 51 mg cholesterol/100 g; Hernandez and others 1998). Akinwunmi and others (1993) found that steaks with Slight marbling had a greater cholesterol content than steaks with Modest marbling (101.3 against 96.2 mg/100 g). Rusman and others (2003) reported the same fat content but different cholesterol concentrations among 5 cattle breeds and 2 different muscles (longissimus thoracis and biceps femoris), and as did Salvatori and others (2008) for total lipid and cholesterol of longissimus dorsi muscles from gilts and barrows of the Italian autochthonous genotype Casertana and its crossbreds. Padre and others (2006) reported no difference in cholesterol contents of longissimus muscles from bulls and steers with significantly different fat contents, ranging from 1.7% to 3.4%. Wheeler and others (1987) provided data showing no correlation between proximate lipid content of beef longissimus muscles and their cholesterol concentrations. Greater cholesterol concentrations were only found in subcutaneous fat with a substantially different fat content compared with muscles tissues. Similar results can also be found in poultry meat (Table 1). Moreover, total lipid content of subcutaneous fat does not always correlate well with its cholesterol concentration. In a study on the effects of rapeseed and soybean on lipid composition of muscle and adipose tissues of rams, Solomon and others (1991) reported cholesterol concentrations of 85 to 95 mg/100 g in triceps brachii subcutaneous fat containing 58% to 64% total lipids. On the other hand, some subcutaneous fat samples from longissimus and semimembranosus muscles with 74% to 80% total lipids only had cholesterol concentrations of 72 to 73 mg/100 g. There was no statistical comparison of cholesterol contents among muscle and fat tissues.

In a study on the effects of fat trim level and cooking on cholesterol content, Swize and others (1992) showed that fat trim level (0 and 0.6 cm) generally did not affect cholesterol content of beef, pork, and lamb cuts. Where the significant differences in cholesterol content were found, the fat contents of the 2 compared trim levels were either similar (raw lamb semitendinosus, 7.4% against 10.6% fat levels, 75.4 and 86.6 mg cholesterol/100 g, respectively) or not of such an extreme difference (raw pork longissimus, 7.5% against 12.8% fat levels, 70.2 and 81.1 mg cholesterol/100 g, respectively), compared with fat contents of other cuts where there was no cholesterol variation between 2 trim levels (raw beef longissimus, 7.1% against 13.6% fat levels, 67.8 and 63.2 mg cholesterol/100 g, respectively; raw lamb longissimus, 8.4% against 16.6% fat levels, 72.4 and 75.6 mg cholesterol/100 g, respectively). Akinwunmi and others (1993) also reported that trim levels did not affect cholesterol content of steaks.

There is considerable evidence of significant effects of muscle fiber type and muscle lipid content on cholesterol concentration. Muscle fiber types of farm animals change during growth and aging depending on maturity, breed, gender, feeding system, feed supplementation, and genetic variations (Johnston and others 1981; Rehfeldt and others 2000; Picard and others 2002, 2006; Oksbjerg and others 2004; Schreurs and others 2008). Nonetheless, Rule and others (1997) underlined the fact that breed, nutrition, and gender might not affect cholesterol concentration in bovine skeletal muscle unless there are marked changes in structure of muscle cells associated with a marked redistribution of membrane fatty acids, which was supported by a finding of a substantially high proportion of cell cholesterol in cell membrane of an intact cell (Lange and Ramos 1983). Hoelscher and others (1988) found that about 60% to 80% of cholesterol in muscle tissues was located in the membrane fraction (20% to 40% from storage cholesterol in lipid of muscle tissues), whereas the membrane contribution was 8% to 12% in adipose tissues (about 90% from storage cholesterol in lipid of adipose tissues). Total cholesterol of adipose tissues was almost 100% greater than that of muscle tissues. In muscle tissues of different quality grades, although the contribution of cholesterol in muscle fat fraction became more important as quality grade increased, cholesterol concentrations in steaks with different quality grades (U.S. Prime, U.S. Choice, and U.S. Select) were not found to differ. The contributions of cholesterol in the muscle fat fraction were 36.0%, 20.5%, and 18.2% for Prime, Choice, and Select, respectively, and were not dramatically different between lean muscle and adipose tissues.

Sweeten and others (1990) further investigated the distribution of cholesterol in adipose tissues of both muscle (intramuscular adipose tissues) and subcutaneous fat (subcutaneous adipose tissues). The intramuscular adipose tissues had similar cholesterol contents to that of the subcutaneous adipose tissues; however, only 54% of the cholesterol in the intramuscular tissues was from cytoplasm (cholesterol in storage lipid), compared with a 90% proportion from cytoplasm in subcutaneous adipose tissues. Results from Hoelscher and others (1988) and Sweeten and others (1990) indicate that although marbling of steaks might contribute significantly to total cholesterol of the whole steaks (with the same cholesterol concentration as subcutaneous fat), the differences in fat content among quality grades might not be sufficient to cause a significant shift in cholesterol concentration; this is, probably an explanation for the results obtained by Akinwunmi and others (1993), Rowe and others (1999), Padre and others (2006), Rusman and others (2003), and Salvatori and others (2008). The contradictory results that were discussed previously, together with those from Hoelscher and others (1988) and Sweeten and others (1990), confirm that unless muscle type, muscle lipid, and their interactions with each other and with other contributory factors cause a dramatic change in muscle structure and composition, their effects on the cholesterol content of muscle might be limited.

Finally, a significant amount of valuable cholesterol information used in this discussion was determined by colorimetric methods (Bohac and Rhee 1988; Hoelscher and others 1988; Browning and others 1990; Sweeten and others 1990; Swize and others 1992; Akinwunmi and others 1993) compared with chromatographic determination. Some inconsistency may arise from the fact that colorimetric methods could overestimate the cholesterol content of meat and other food products (Sweeney and Weihrauch 1976; Beyer and Jensen 1989; Fenton 1992; Tonks 2006). Great care needs to be taken when evaluating data across different studies, and some validation of data might be needed for accurate interpretations. As a result, the data of cholesterol contents presented in Table 1 were collected from studies using chromatographic methods (GC or high-performance liquid chromatography [HPLC]) so that the methodological variation was minimized.

Cholesterol Analytical Methods in Meat and Poultry Products

Analytical methodology has been claimed to be a source of differences found in number of studies on cholesterol content of meat and poultry products (Bragagnolo 2009). Cholesterol determination procedures in foods and other biological materials usually involve lipid extraction, separation of cholesterol from interfering components or liberation of cholesterol into the free form, and measurement of isolated cholesterol (Sweeney and Weihrauch 1976; Fenton 1992; Hwang and others 2003; Dinh and others 2008). Cholesterol can be measured by means of gravimetry, titration, colorimetry, refractometry, fluorometry, and chromatography (Sweeney and Weihrauch 1976).

Extraction solvent and direct saponification

Before direct saponification was proposed and tested (Adams and others 1986; Al-Hasani and others 1990, 1993; Van Elswyk and others 1991; Klatt and others 1995), lipid extraction had been the 1st step of sample preparation (Sweeney and Weihrauch 1976; Fenton 1992). Of the most common extraction methods were the 2 proposed by Folch and others (1957) and Bligh and Dyer (1959), which were later modified in many other studies (Hoelscher and others 1988; Sweeten and others 1990; Rodriguez-Palmero and others 1994; Volin 2001; see summary of various extraction methods in Sweeney and Weihrauch 1976; see comparison between Folch and various extraction methods in Hubbard and others 1977 and in Fenton 1992). A mixture of polar and nonpolar solvents has been suggested to give better cholesterol extraction from food materials because cholesterol in these samples is usually bound by many other biological compounds such as lipoproteins, proteins, and phospholipids (Sweeney and Weihrauch 1976; Yeagle 2005; Dowhan and others 2008), and a multiple extraction approach was thought to be more suitable to remove membrane cholesterol (Sweeney and Weihrauch 1976). With saponification of extracted lipids or direct saponification of samples, however, results of other studies indicated that single or multiple extraction by hexane (Kovacs and others 1979; Al-Hasani and others 1990, 1993; Indyk 1990; Patton and others 1990; Fenton and Sim 1991; Fenton 1992) or by diethyl ether (Lognay and others 1989; Hwang and others 2003) or single extraction by toluene (Oles and others 1990; Dinh and others 2008) yielded sufficient recovery of cholesterol in foods, even in complex matrices such as meat and egg yolk. Hexane was usually the favored solvent because of its lower polarity compared with toluene (Gu and others 2004), which limits the formation of an emulsion; however, we have experienced frequent formation of a viscous aqueous phase when using methanolic KOH and hexane in our laboratory, which greatly interfered with the extraction and separation. A single extraction by toluene also has been used in our laboratory, and the recovery for cholesterol after direct saponification ranges from 94% to 102%. The benefits of toluene are that only a single extraction is needed and that cholesterol and internal standard stock solutions (1 mg/mL) in toluene can be stored for up to 2 y, compared with those in hexane, heptane, or diethyl ether (approximately 2 mo). A single extraction is time-efficient because solvent evaporation and reconstitution are not needed, especially when one is attempting to analyze cholesterol without derivatization.

Despite the benefits, the use of toluene is subject to a possible formation of emulsion, especially in the first 2 washing steps that involve addition of diluted KOH and water (Dinh 2010). Emulsion, if not eliminated, could lead to either decreased recovery or overestimated cholesterol concentration because the losses of toluene and cholesterol into the emulsion might not be proportional. Therefore, if toluene is the solvent of choice, postsaponification manipulation is very important to ensure distinct separation of toluene from the aqueous phase. Emulsion can be eliminated by a small amount of ethanol as described by Dinh (2010), Dinh and others (2008), and Richardson and others (1994). Walton and others (1989) proposed an increase in saponification time; however, this solution was not found to be effective in our laboratory (Dinh 2010, unpublished data). Adams and others (1986) washed the wall of separatory funnel during solvent cleanup, and such a manipulation also has been used effectively in our laboratory, especially when dealing with high-fat samples (more than 30%). Moreover, with high-fat samples, because of a large amount of potassium soap created after saponification, addition of ethanol (5 to 10 mL per 10 mL of toluene) immediately after saponification was found to counteract emulsion formation (Dinh 2010, unpublished data). The same treatment is used in the AOAC Official Method 994.10 (Klatt and others 1995; AOAC International 1996).

The saponification, not the extraction, probably plays a key role in liberating cholesterol from other components, especially the ones binding cholesterol in animal cell membrane or egg yolk (Al-Hasani and others 1990, 1993; Van Elswyk and others 1991; Dinh and others 2008). Saponification is essential to separate cholesterol and other unsaponifiable materials from fatty acids, thereby eliminating triglyceride interference (Fenton 1992; Bragagnolo 2009). When cholesterol is the only analyte of interest, direct saponification is chosen to simplify sample preparation (Perkins and others 1992; Mauricea and others 1994; Klatt and others 1995; Wu and others 1997; Fletouris and others 1998; King and others 1998; Piironen and others 2002; Bragagnolo and Rodriguez-Amaya 2003a; Pratiwi and others 2006; Dinh and others 2008). Direct saponification was investigated under various conditions, and results of most studies showed that it has superior recovery and accuracy than conventional lipid extraction and saponification (Adams and others 1986; Beyer and others 1989; Al-Hasani and others 1990, 1993; Fenton and Sim 1991; Van Elswyk and others 1991; Klatt and others 1995; Mazalli and others 2006; Prates and others 2006; Dinh and others 2008). Van Elswyk and others (1991) reported that direct saponification was more accurate in recovering total cholesterol content of SR materials (egg yolk and whole egg) obtained from the Natl. Inst. of Standards and Technology (NIST). Adams and others (1986) reported greater recovery (99.8%), excellent precision (coefficient of variation [CV] of 1.74%), and slightly greater cholesterol concentrations in meat matrices compared with the traditional procedure. Bragagnolo and Rodriguez-Amaya (2003a) showed that direct saponification led to significantly greater cholesterol content of eggs. Viturro and others (2010) reported that direct saponification (for an enzymatic assay) could also be used in milk samples with acceptable precision (CV of 4.8% and 9.1% for intra- and interassay variations, respectively) and recovery (98.1% to 106.3%). Moreover, in dairy products, direct saponification similarly delivered increased recovery compared with a conventional dairy fat extraction technique (Richardson and others 1994). In the same study, direct saponification improved precision for low-fat dairy products and cheese (CV of 1.49% against 3.16% of conventional fat extraction). Mazalli and others (2006) also presented an accurate and precise method employing direct saponification with a very low detection limit (0.002 to 0.079 μg/g). Nonetheless, there were also inconsistent results showing no significant difference between direct saponification and saponification after Folch extraction of lipids (in egg, Beyer and others 1989; in milk, Tsui 1989; in egg, Fenton 1992; in pork fat and muscle, Maraschiello and others 1996). Few comparison studies conducted with meat matrices have been reported.

The most suitable direct saponification conditions should leave an ethanolic KOH concentration of 0.33 to 0.5 M in the final saponification solution (Abell and others 1952; Fenton 1992) with saponification temperatures ranging from 55 to 75 °C (Fenton 1992) or up to ethanol/water mixture boiling point (Klatt and others 1995; Dinh and others 2008) and within 15 to 60 min. Small sample size (1 g) allowed for quick saponification (15 min) with a complex matrix such as meats (Dinh and others 2008; Dinh 2010). Fletouris and others (1998) found that increased saponification time could cause loss of cholesterol when using saturated methanolic KOH. The 2.0 M KOH also gradually decreased cholesterol recovery when saponification time increased from 15 to 90 min; however, this phenomenon was not observed in our laboratory with the use of 50% KOH. Our samples were saponified for at least 15 min; however, because of scheduled workload, some samples were occasionally boiled for an extended period (exceeding 90 min) with no apparent effect on cholesterol recovery. Many studies have used concentrated KOH for saponification (50%: Adams and others 1986, Indyk 1990, Van Elswyk and others 1991, Dinh and others 2008; 33% to 40%: Kaneda and others 1980; Fenton and Sim 1991). From our experience, the strength of KOH solution is important to the release of cholesterol. In our laboratory, a 1.0 N KOH solution was accidentally used to saponify a meat homogenate purchased from NIST (SRM1546, used for quality control purposes, NIST 2010) and the cholesterol recovery decreased by 50% compared with the regular use of a 50% (wt/wt) KOH solution, which resulted in an approximate ethanolic KOH concentration of 2 M. Although less concentrated KOH solutions can be used, the effectiveness of such solutions might depend on the nature of the samples. In the same incident of using 1.0 N KOH in our laboratory, the results of most chicken and beef samples seemed not to be affected (confirmed by reruns with 50% KOH), although there were more difficulties during extraction and cleanup.

Overview of cholesterol measurement

Much of earlier data on cholesterol content of foods before and during the early 1980s and even recently were produced by either spectrophotometric, gravimetric, or enzymatic methods, many of which were originally developed to measure cholesterol in blood serum (Man and Peters 1933; Sperry and Brand 1943; Coelho and Alves 1946; Nath and others 1946; Kenny 1952; Searcy and Bergquist 1960; Allain and others 1974; Sweeney and Weihrauch 1976; Rhee and others 1982a, 1982b; Hoelscher and others 1988; Auerbach and others 1990; Browning and others 1990; Fenton 1992; Abu-Tarboush and Dawood 1993; Morris and others 1995; Sales and others 1999; Caldironi and Manes 2006). Cholesterol in foods has been determined gravimetrically by precipitation using digitonin or tomatin; however, use of digitonin has been more common (Sweeney and Weihrauch 1976; Bragagnolo 2009). Digitonin is washed and the precipitated cholesterol is weighed and calculated using several common factors to account for the proportion of cholesterol in the precipitate (Weston 1914; Man and Peters 1933; Sweeney and Weihrauch 1976; Tonks 2006). The gravimetric measure of cholesterol digitonide, however, lacks specificity because of the effects of phospholipids on washing the excess digitonin (Sweeney and Weihrauch 1976). Moreover, the method is very time-consuming, expensive, and challenging because of strict experimental conditions (Man and Peters 1933; Sweeney and Weihrauch 1976; Tonks 2006). Gravimetric determination of cholesterol can follow a series of sample preparation steps from lipid extraction to purification; however, direct precipitation of cholesterol from the extract has been performed (Man and Peters 1933; Sweeney and Weihrauch 1976).

Enzymatic determination of cholesterol has been used for food samples, although it is used to a much greater extent in clinical laboratories (Auerbach and others 1990; Bragagnolo 2009). Enzymatic methods employ cholesterol esterase to cleave any ester bond and cholesterol oxidase to oxidize cholesterol to peroxide. The peroxide then reacts with peroxidase and 4-aminophenazone (or other reagents) to produce a pigment that can be measured spectrophotometrically (Allain and others 1974; Auerbach and others 1990; Usui and others 2002; Mizoguchi and others 2004; Rambaldi and others 2009). The enzymatic method has been used commonly for dairy products (Viturro and others 2010) and occasionally for other foods, including meats (Ulberth and Reich 1992; Girolami and others 2003; Cifuni and others 2004; Caldironi and Manes 2006). The enzymatic method was found to yield comparable results to the GC technique when used to determine cholesterol content of milk (not statistically tested, Viturro and others 2010) and processed foods containing primarily animal fat or fat of animal origin (P > 0.05, Ulberth and Reich 1992). Nonetheless, this method lacks specificity because other sterols with a 3β-OH group including phytosterols can also be oxidized and form similar pigments (Ulberth and Reich 1992; Bragagnolo 2009). Therefore, in processed foods containing primarily lipids of vegetable origin, the enzymatic method overestimated cholesterol content compared with GC (Ulberth and Reich 1992). Jiang and others (1991) evaluated 4 methods of cholesterol determination in egg yolk: colorimetric, enzymatic, GC, and HPLC. Results with the enzymatic method were found to be similar to those with GC and HPLC. All 3 methods, except colorimetry, achieved excellent precision.

In color-based methods, the application of the Liebermann–Burchard reaction (Liebermann 1885; Burchard 1890) or other color development reactions (Zlatkis and others 1953; Searcy and Bergquist 1960) is usually the key step after different extraction and de-esterification. Coloring reagents can be acetic acid and concentrated sulfuric acid reacting in different solvents such as chloroform or ether (Liebermann–Burchard reaction), paratoluenesulfonic acid or similar reagents with glacial acetic and concentrated sulfuric acid, or iron salt (iron chloride or iron sulfate) in sulfuric acid and glacial acetic (Zlatkis and others 1953; Searcy and Bergquist 1960; Tonks 2006). Color-producing reagents were added after extraction, saponification, and purification, or sometimes were added directly to the extract (Sweeney and Weihrauch 1976; Tonks 2006). Colorimetric methods usually suffer various interferences from unsaturated fatty acids, proteins, other steroids, or even vitamin A. Bilirubin, a component of red blood cells, can give a significant false positive estimation of cholesterol when present in samples (Sweeney and Weihrauch 1976; Tonks 2006). Cholesterol esters, if not hydrolyzed, can react differently with color reagents (Sweeney and Weihrauch 1976; Beyer and Jensen 1989; Tonks 2006). The color reactions depend on a double bond system, other functional groups, and the presence of a nonpolar side chain (Xiong and others 2007). Although Liebermann–Burchard reaction is used extensively in many clinical laboratories (because of minute interference in blood samples) and other analytical laboratories, its major reaction pathway in acid has only been clarified recently (Xiong and others 2007). Colorimetric methods were commonly used to measure cholesterol in meats (Weyant and others 1976; Rhee and others 1982a, 1982b; Bohac and others 1988; Bohac and Rhee 1988; Browning and others 1990); however, the lack of specificity and color stability, the issue of temperature dependency, and turbidity of final color-developed solution have made colorimetric methods subject to significant concern regarding accuracy (Searcy and Bergquist 1960; Sweeney and Weihrauch 1976; Beyer and Jensen 1989; Fenton 1992; Tonks 2006). Without saponification, Bohac and others (1988) showed that the colorimetric method overestimated cholesterol content of meats. Even with saponification, the presence of more unsaturated fatty acids (more double bonds) yielded greater cholesterol concentration without antioxidant protection. The cholesterol content of saponified samples with an antioxidant was similar to those analyzed by the GC technique. Beyer and Jensen (1989) confirmed that their colorimetric method overestimated the cholesterol content of egg yolk. Approximately 17.5% chromogens quantified by the colorimetric assay was not of cholesterol origin. Therefore, the purification of the cholesterol fraction before colorimetric determination was recommended. Jiang and others (1991) reported the same disadvantage of the colorimetric method with egg yolk. Both studies employed chromatography (HPLC, Beyer and Jensen 1989; GC and HPLC, Jiang and others 1991) as the reference method. Gravimetric, colorimetric, and enzymatic methods need strict control of analytical conditions for precise and accurate results (Sweeney and Weihrauch 1976; Tonks 2006; Bragagnolo 2009).

In recent years, among all quantitative techniques for cholesterol in foods, especially muscle food products, chromatography, primarily GC and HPLC with various detection methods, has been studied and used extensively (Adams and others 1986; Fenton 1992; Ulberth and Reich 1992; Thompson and Merola 1993; Rodriguez-Palmero and others 1994; King and others 1998; Lesellier 2001; Shimada and others 2001; Toivo and others 2001; Volin 2001; Abidi and others 2002; Bragagnolo and Rodriguez-Amaya 2003a; Hwang and others 2003; Pratiwi and others 2006; Dong and others 2007; Dinh and others 2008; Daneshfar and others 2009; Farwanah and others 2009; Griffiths and Wang 2009; Isidorov and Szczepaniak 2009). Chromatography is preferred for cholesterol analysis in foods because of its specificity to separate cholesterol from other unsaponifiable compounds based on their differences in physical and chemical properties and their interaction with stationary and mobile phases (Heftmann 1976; Sweeney and Weihrauch 1976; Fenton 1992; Kuksis 2004). Processed meats with nonmeat ingredients contain not only cholesterol but also plant sterols, tocopherol, tocotrienol, saturated hydrocarbons, squalene, aliphatic alcohols, terpene alcohols, triterpene alcohols, and steryl esters, all of which are unsaponifiable (Fenton 1992). Recently, HPLC has been used more than GC as it is thought to decrease cholesterol oxidation because of a lower operational temperature, and it allows for a nondestructive separation; however, the GC is still preferred for its higher sensitivity over HPLC (Bragagnolo 2009).

GC columns for cholesterol separation

Because of the interference of other unsaponifiable materials and cost and availability of highly specific detectors, the effectiveness of chromatographic separation and quantification of cholesterol greatly depend on the heart of the chromatographic system, the column. GC is still the most commonly used technique for the determination of sterols including cholesterol, especially when cholesterol is the sole compound of interest (Fenton 1992; Abidi 2001). Gas chromatographic columns have evolved dramatically from packed columns to capillary columns with wide variation in polarity. Packed columns have been used since the development of GC in the 1950s (Christie 1989). They can be made of glass or stainless steel and be packed with support material precoated with the liquid phase (Christie 1989). A 1.5- to 2-m well-packed column with support materials coated with 10% to 15% liquid phase can have 3000 to 5000 theoretical plates. Packed columns can be occasionally repacked with new material.

One of the 1st stationary phases used for GC separation of cholesterol and other sterols was methylpolysiloxane SE-30 supported on silanized Gas-Chrom P (Heftmann 1976). It is a nonselective stationary phase used for cholesterol determination (Sheppard and others 1977) because of its ability to separate a number of steroid hormones. A layer of 1% SE-30 was coated on 100 of 120 mesh Gas-Chrom Q (commercial form of Gas-Chrom P), which was then packed in a 152.4 × 4 mm i.d. glass column. The column, operated at 255 °C, gave the best separation among 7 packings evaluated (Sheppard and others 1977). This stationary phase has been used to separate both free cholesterol and its butyryl derivative (Sheppard and others 1977; Fenton 1992).

Other stationary phases that have been used for cholesterol analysis by GC are OV-17 (phenyl methyl siloxane), SE-52 (methyl phenyl siloxane), Apiezon L (hydrocarbon grease), SP-2401 (trifluoropropyl methyl siloxane), and OV-101 (methyl siloxane) (Heftmann 1976; Fenton 1992). Similarly, Kunsman and others (1981) and Ryan and Gray (1984) used a nonpolar stationary phase (SP-2100, 100% dimethyl polysiloxane) to coat Supelcoport packed into a glass column for cholesterol determination. Cholesterol is typically classified as a nonpolar compound; therefore, a nonpolar stationary phase has been employed. However, cholesterol has one hydroxyl group (Figure 1) that gives it slight polarity, which allows for separation by both polar (for example, Silar 5-CP, cyanoalkyl phenyl siloxane, or OV-17) and nonpolar (for example, SE-30 or SP-2100) stationary phases (Fenton 1992). Indeed, Adams and others (1986) employed a highly polar stationary phase (SP-2401, 50% trifluoropropyl-methylpolysiloxane) for cholesterol determination by GC. A 183-cm column (2-mm i.d.) was packed with 5% SP-2401 on 100 of 120 mesh Supelcoport, which allowed for an operational temperature up to 214 °C. The method delivered great reproducibility with a CV of 1.74%.

Figure 1–.

Structures of cholesterol and its trimethylsilyl ether derivative (created using ACD/ChemSketch Freeware 12, Advanced Chemistry Development, Inc., Toronto, Ontario, Canada).

The separation of cholesterol, other sterols, and their derivatives on several GC packed columns has been reviewed by Heftmann (1976), Xu and others (1988), and Fenton (1992). Xu and others (1988) also presented additional information about the chromatographic behaviors of trimethylsilyl (TMS) derivatives of sterols by comparing GC separation and detection with other techniques such as HPLC and thin-layer chromatography (TLC). On the other hand, Itoh and others (1982) used glass capillary columns coated with OV-1 and OV-17 to collect information about the relative retention times of 168 acetate derivatives of sterols and triterpene alcohols. The OV-17-coated glass capillary column gave high resolution of a mixture of steroids (Heftmann 1976). Solomon and others (1990) successfully used the same OV-17 stationary phase coated on 100 of 120 mesh Gas Chrom Q packed in a glass column to measure cholesterol in meat.

Although the packed columns have provided useful analytical information and the later developed glass capillary columns have made the GC technique more efficient and robust for routine analysis (Christie 1989), fused silica wall-coated open-tubular (WCOT) columns have now mostly replaced them. Fused silica is an excellent medium for WCOT columns because it is an amorphous silicate material free of metal oxides and is therefore very inert (Christie 1989). The column has its surface chemically bonded with liquid stationary phases, and the molecules of polymeric liquid phases are chemically cross-linked to improve thermostability of the column (Christie 1989; Fenton 1992). The improved thermostability allows for increased operational temperatures with much less bleeding (Fenton 1992), which minimizes detector contamination, increases column life span, gives better linearity and reproducibility, and improves sensitivity. Film thickness of the stationary phase, column length, and inner diameter are important factors affecting resolution. Longer columns with smaller inner diameters and thicker stationary films usually give better resolution for more complex separations (Fenton 1992). Fletouris and others (1998) used a fused silica capillary column coated with SPB-1 (100% dimethylsiloxane) with 1.0-μm film thickness (thick film) and argued that a thick film covers the active silanol groups on the surface of the fused silica and prevents free cholesterol (hydroxyl group) from absorbing to the active region on the column surface, which decreases peak distortion and tailing. Similar phenomena were discussed by Fenton (1992) and Lesellier (2001). For the purpose of cholesterol determination, (5%-phenyl)-methylpolysiloxane (HP-5, HP-5 ms, ZB-5, SPB-5, DB-5, RTX-5, HP-Ultra2) with very low polarity has probably been the most commonly used stationary phase regardless of whether cholesterol was derivatized (Ulberth and Reich 1992; Rodriguez-Palmero and others 1994; King and others 1998; Toivo and others 2001; Baggio and others 2002; Bowden and others 2009; Isidorov and Szczepaniak 2009; Schummer and others 2009). The film thickness of choice was 0.25 μm except when there was no derivatization (1.05 μm, King and others 1998). Fewer studies have favored a midpolar phase of (50% phenyl)-methylpolysiloxane (DB-17, DB-1701, NB-17, CP-Sil 24CB, Eckfeldt and others 1991; Richardson and others 1994; Toivo and others 1998; Hwang and others 2003; Dinh and others 2008) or nonpolar phase of 100% dimethylpolysiloxane (HP-1, SE-30, CP-Sil 5CB, SPB-1, DB-1, Nawar and others 1991; Ballesteros and others 1996; Fletouris and others 1998; Paleari and others 1998; Muchenje and others 2009; Cannata and others 2010). As noted previously, the thin film thickness was chosen unless cholesterol was directly injected into the GC (Ballesteros and others 1996; Hwang and others 2003), except for Dinh and others (2008, free cholesterol, 0.15 μm). Thinner film columns have higher upper temperature limits because of low bleed, and they are more suitable for separating large molecules with high molecular weights and decreasing retention time; however, they make the cholesterol peak subject to possible tailing because of the interaction between hydroxyl group of cholesterol and silanol groups on the column surface (Figure 2). The GC columns, although very efficient at cholesterol separation, usually suffer from possible overlapping of cholesterol with other sterols (for example, cholestanol), and especially α-tocopherol, which co-elute with cholesterol on many GC systems (Fenton 1992). Peak overlapping can be resolved more efficiently by an HPLC system, especially when employing reverse-phase HPLC and a variety of special detection techniques such as photodiode array, fluorescence, and mass spectrometry.

Figure 2–.

Cholesterol peak-tailing (circled) by interaction with silanol groups compared with the 5α-cholestane peak, separated on a 50% phenyl-methylpolysiloxane, midpolar column.

HPLC for cholesterol separation

There are many published HPLC methods for cholesterol determination; however, most were not developed for routine analysis, but rather for separating cholesterol from other unsaponifiable compounds for specific research purposes (Beyer and Jensen 1989; Nakajima and others 1995; Nogueira and Bragagnolo 2002; Prates and others 2006; Hojo and others 2007; Daneshfar and others 2009). Because of the slight polarity caused by the hydroxyl group, either normal-phase (NP) or reversed-phase (RP) HPLC can be used for analysis of cholesterol. NP HPLC has been used to separate triglycerides, diglycerides, and cholesterol (Fenton 1992). As mentioned previously, the GC system is ineffective for separating cholesterol from tocopherols; however, Katsanidis and Addis (1999) successfully separated and quantified cholesterol, vitamin E, and its homologs by an NP-HPLC technique with a 25-cm Zorbax RX-Sil column (particle size of 5 μm). The compounds were detected with a ultraviolet (UV) detector at 295 nm for vitamin E and 202 nm for cholesterol. The column was made of ultra-clean porous silica microparticles. The mobile phase was 99% hexane and 1% isopropanol. Most NP-HPLC methods use an NP (polar stationary phase) column made of highly pure, porous silica microparticles (μPorasil, Kermasha and others 1994; Zorbax RX-Sil, Katsanidis and Addis 1999; InertSil ODS-2, Sion and others 2001; Zorbax RX-Sil, Ponte and others 2004, 2008; Spherisorb S5W, Costa and others 2006). Generally, the particle size is 5 μm, except for the column used by Kermasha and others (1994, 10 μm). Smaller particle size leads to significantly increased column pressure, but it also increases the resolution (theoretical plate number). More polar stationary phases, such as cyanopropylsilica and alcohol-bonded silica, have been used (Abidi 2001). The mobile phase for NP-HPLC has primarily been an isocratic phase of 1% to 3% isopropanol in hexane (Kermasha and others 1994; Katsanidis and Addis 1999; Abidi 2001; Costa and others 2006; Ponte and others 2004, 2008). Sion and others (2001), for the purpose of using light-scattering detection and improving peak shape, used a gradient mobile phase made of 98:2 methanol/water (vol/vol), 30:6:10 chloroform/methanol/water (vol/vol/vol), and 50:50 methanol/chloroform (vol/vol). The constant flow of 1 mL/min or lower has been used in most studies.

Although NP-HPLC was the initial technique used successfully to separate cholesterol in meat and poultry such as beef (Costa and others 2006), bovine tissues (Katsanidis and Addis 1999), and chicken meat (Ponte and others 2004, 2008), the RP-HPLC has been preferred because it offers a wider range of column selectivity and separable compounds based on polarity. The RP-HPLC method uses a nonpolar column or a column with low polarity and a polar mobile phase. Simple addition of some polar solvents such as methanol, ethanol, or water can dramatically change mobile phase polarity and alter the elution (Abidi 2001). Most RP-HPLC methods for cholesterol determination use acetonitrile as the polar organic solvent with the addition of either ethanol (Daneshfar and others 2009), methanol (Simsek and others 2009), or isopropanol (Komprda and others 2003). The RP-HPLC methods have typically employed silica (Abidi 2001) as 5-μm particle support material covalently bonded with either octadecyl (octadecylsilica, C18, C18H37-Nogueira and Bragagnolo 2002; Komprda and others 2003; Salvatori and others 2004, 2008; Pratiwi and others 2006; Simsek and others 2009; 10 μm in Beyer and Jensen 1989) or octyl (octylsilica, C8, C8H17-, Nakajima and others 1995; Daneshfar and others 2009). The RP-HPLC C18 column was found to be superior to the C8 column because the retention of cholesterol and other sterols is much stronger in the C18 column (Holen 1985; Fenton 1992). Therefore, a mobile phase that is less polar should be employed for cholesterol separation on a C18 column (Holen 1985). The HPLC technique, although superior to GC in terms of resolution of multiple unsaponifiable compounds, can also be improved by programmed temperature elution (Xu and others 1988; Fenton 1992).

Derivatization of cholesterol

In the GC technique, cholesterol is usually derivatized to improve volatility and thermostability, to optimize peak shape, to decrease retention time, and to increase sensitivity (Heftmann 1976; Fenton 1992; Kuksis 2004). The most commonly used derivative has been TMS ether (Anders and Mannering 1962; Adams and others 1986; Solomon and others 1990; Fenton 1992; Wong and others 1993; Richardson and others 1994; Rodriguez-Palmero and others 1994; AOAC Official Method 976.26 and 994.10, AOAC International 1996; King and others 1998; Abidi 2001; Shimada and others 2001; Toivo and others 2001; Hur and others 2007; Griffiths and Wang 2009; Isidorov and Szczepaniak 2009), although others such as butyryl ester (Sheppard and others 1977; Brumley and others 1985) or cholesteryl acetate (Anders and Mannering 1962; Avigan and others 1963; Itoh and others 1982; Reina and others 1997; Thiam and others 2000; Abidi 2001) have also been employed. The TMS ether derivative of cholesterol has high thermostability and low polarity. The low polarity, by preventing TMS derivatives from interacting with the polar sites (silanol groups) on GC columns because of conversion of hydroxyl group to TMS ether group (Figure 1), allows for an optimal peak shape without tailing (Fenton 1992; Fletouris and others 1998; Lesellier 2001; Shimada and others 2001). After reviewing published methods for the analysis of sterols in vegetable oils, Abidi (2001) reported that TMS and acetate derivatives are commonly used for volatilizing hydroxyl-containing analytes, enhancing resolution, and stabilizing unsaturated sterols. Compared with acetate derivatives, TMS derivatives are more suitable for GC analysis with mass spectroscopic (MS) quantification because they allow for more informative ion fragments upon ionization (Abidi 2001; Griffiths and Wang 2009). Cholesterol can be derivatized to the TMS ether by many different reagents (see summaries by Fenton 1992 and Abidi 2001). King and others (1998) compared the TMS derivatization process by hexamethyldisilane (HMDS)/trimethylchlorosilane (TMCS) (Sigma, St. Louis, Mo., U.S.A.) and Sylon BTZ (N-trimethylsilylimidazole [TSIM]/N,O-bis(trimethylsilyl)acetamide [BSA]/TMCS 3:3:2) (Supelco, Bellefonte, Pa., U.S.A.) and reported that the results were comparable (not statistically tested); however, Sylon BTZ reagent was faster and consumed less solvent. In contrast, Nawar and others (1991), when optimizing the silylating conditions for cholesterol and its oxides with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA), BSTFA/1% TMCS, and Sylon BTZ, found that BSTFA/1% TMCS gave the maximum conversion of cholesterol within 1 h at 80 °C and provided superior silylation for 7α- and 7β-diols compared with BSTFA only. Moreover, Bowden and others (2009) found that N-methyl-N-trimethylsilyl-trifluoroacetamide (MSTFA) was the most efficient derivatization reagent compared with BSTFA/1% TMCS and BSA. The TMS derivatives of cholesterol and other sterols are decomposed by the FID and form silicon dioxides on the detector, which eventually adversely affects the linearity of the method (Fenton 1992; Osman and Chin 2006). Therefore, although HMDS and TMCS are commonly used (Adams and others 1986; Richardson and others 1994; King and others 1998) and are recommended in the AOAC Official Methods 994.10 (AOAC International 1996), BSTFA is preferred because it produces hydrofluoric acid under FID decomposition and reacts with silicon dioxide to form volatile silicon tetrafluoride, which prevents silicon deposits on the GC detector (Solomon and others 1990; Nawar and others 1991; Fenton 1992; Toivo and others 2001; Hur and others 2007; Isidorov and Szczepaniak 2009). The 1% TMCS acts as an acid catalyst for the silylation process (Nawar and others 1991). The solvents commonly used for silylation are N,N-dimethylformamide (DMF) and pyridine because these solvents provide great solubility for cholesterol (Adams and others 1986; Solomon and others 1990; Wong and others 1993; Richardson and others 1994; Rodriguez-Palmero and others 1994; AOAC International 1996; King and others 1998; Toivo and others 2001; Hur and others 2007; Isidorov and Szczepaniak 2009). Pyridine, a solvent and a base, can also be an HCl-receptor to facilitate silylation effectively. Bowden and others (2009) also experimented with several enhancement techniques and found that the uses of solvent addition (DMF) and microwave heating improved the derivatization process. The TMS reagents and derivatives are easily hydrolyzed; therefore, a moisture-free environment to handle derivatized samples and a moisture trap for the GC system are usually required (Fenton 1992).

Cholesterol can be analyzed without derivatization, however, with the cholesterol peak possibly being subject to distortion or tailing on the chromatogram because of its interaction with silanol groups on the column surface (Fenton 1992; Shimada and others 2001). The derivatization, on the other hand, can create artifacts that contaminate detectors, especially FID (Fenton 1992; Osman and Chin 2006), and interfere with detection and quantification. With improvement in thermostability of fused silica columns with bonded and cross-linked stationary phase, it seems unnecessary to prepare cholesterol derivatives in many cases (Ballesteros and others 1996; King and others 1998; Baggio and others 2002; Hwang and others 2003; Osman and Chin 2006; Dinh and others 2008). In these studies, free cholesterol peaks were well shaped with good reproducibility. Osman and Chin (2006) used a high-temperature column and reported that the cholesterol peak was sharp. Even with packed columns, the separation and quantification of free cholesterol were found to be satisfactory (Kovacs and others 1979; Kunsman and others 1981; Ryan and Gray 1984). Kovacs and others (1979) found no significant difference between recoveries of the cholesterol TMS-derivative and the free cholesterol. Hwang and others (2003) reported excellent recovery and precision for nonderivatized cholesterol. Van Delden and others (1981) and Dinh and others (2008) indicated that injection of free cholesterol yielded high precision and linearity. Kaneda and others (1980), after a collaborative study of several sample treatments for cholesterol quantification in foods, suggested that pretreatments, such as preparative chromatography or derivatization, were not required because pretreatment could lead to loss of analyte. The authors reported that results obtained by GC analyses of free cholesterol and cholesterol derivatives (TMS and acetate) did not differ. Free cholesterol has been analyzed by GC-MS system and especially an HPLC system coupling with a variety of detectors (UV detector in particular) unless there is a reason to enhance the level of detection to trace concentrations by derivatization (Dong and others 2007; Lin and others 2007).

Detection of cholesterol and cholesterol derivatives

Cholesterol derivatives can be detected and quantified by either FID or MS (Fenton 1992; Abidi 2001; Shimada and others 2001). The FID gives good sensitivity and a wide range of linearity, and it is widely used for cholesterol determination in foods including meat and poultry products (Kovacs and others 1979; Fenton 1992; Richardson and others 1994; Rodriguez-Palmero and others 1994; King and others 1998; Paleari and others 1998; Toivo and others 2001; Hur and others 2007; Isidorov and Szczepaniak 2009; Cannata and others 2010). In recent years, the MS detector has become increasingly important for both GC and HPLC to provide better quantitative information when cholesterol co-elutes with other unsaponifiable compounds (Fenton 1992; Abidi 2001; Shimada and others 2001; Bowden and others 2009). Isidorov and Szczepaniak (2009) used both FID and MS detectors to obtain information about the GC retention indices of 389 biologically and environmentally important organic compounds (including cholesterol) and indicated that this modern approach was most effective when analyzing a complex mixture. The success of GC-MS technique requires efficiency of derivatization procedures, especially on steroids containing hydroxyl groups because the conversion of hydroxyl groups to TMS ether groups makes the compounds more responsive to the ionization process (Bowden and others 2009). GC with mass spectrometer (GC-MS) has been used for many food samples including meat and poultry (processed foods, Ulberth and Reich 1992; turkey meat and products, King and others 1998; turkey meat, Baggio and others 2002), and its primary purpose was to confirm the analyte peak or to compare with the detection and quantification of cholesterol with HPLC-UV detection (Baggio and others 2002), GC-FID (King and others 1998), or enzymatic methods (Ulberth and Reich 1992). An MS detector is typically coupled with an HPLC system for cholesterol determination and neutral lipid profiling in foods (egg, Bragagnolo and Rodriguez-Amaya 2003a; vegetable oil, Carretero and others 2008; olive oil, Canabate-Diaz and others 2007; chicken breast tissues, Jazbec and others 2009) and other biological samples (sterol mixture, Riddle and Guiochon 2006; human fibroblasts, Farwanah and others 2009; neurosterols, Griffiths and Wang 2009). Nonetheless, because sterols are highly hydrophobic with very few functional groups (only one OH group for cholesterol), it is difficult to use atmospheric pressure ionization (chemical ionization, APCI; and electrospray, ESI) to create ion fragments for MS detection; therefore, electron impact ionization (EI) is the most commonly used ionization technique (King and others 1998; Abidi 2001; Shimada and others 2001; Bowden and others 2009; Isidorov and Szczepaniak 2009).

Other detection techniques, including UV, fluorescence detection (FD), evaporative light-scattering detection (ELSD), infrared detection, nuclear magnetic resonance, and electrochemical detection (ECD), also have been reported for cholesterol determination (Fenton 1992; Abidi 2001; Shimada and others 2001). These detection methods are mainly coupled with HPLC for the determination of other unsaponifiable compounds such as tocopherols, tocotrienols, and plant sterols in addition to cholesterol (Abidi 2001; Shimada and others 2001). Thus far, an HPLC system equipped with either a UV/UV-Visible (UV-Vis) detector or a photodiode array (PDA) detector has been the most common alternative to GC for the analysis of cholesterol and other sterols in foods including meat products with acceptable sensitivity (UV-Vis, bovine tissues, Katsanidis and Addis 1999; UV-Vis, egg noodles, Nogueira and Bragagnolo 2002; PDA, egg, Bragagnolo and Rodriguez-Amaya 2003a; UV, chicken, turkey, fish, Komprda and others 2003; PDA, broiler meat, Ponte and others 2004, 2008; UV, goat meat, Pratiwi and others 2006; UV, bovine meat, Costa and others 2006, 2009; UV, broiler serum, Simsek and others 2009; UV, milk, egg yolk, olive oil, Daneshfar and others 2009). The UV wavelengths for cholesterol absorption used in these studies ranged from 202 to 210 nm. Sterols generally exhibit UV absorbance between 190 and 210 nm (Shimada and others 2001); however, cholesterol was found to absorb at 203 to 214 nm, with maximum absorbance at 205 nm (Hurst and others 1983; Xu and others 1988; Fenton 1992) because of the unsaturated center and the hydroxyl group. Bragagnolo and Rodriguez-Amaya (2003a), with slight difference, scanned the entire region of 190 to 300 nm and decided to choose the wavelength of 210 nm.

The determination of cholesterol and other related sterols by HPLC-UV is not very convenient because of the issues of specificity and sensitivity. Most sterols, including cholesterol, do not have specific UV absorptive properties and strong absorption in the UV region (Fenton 1992; Shimada and others 2001). Their UV molar absorption coefficients vary depending on numbers and position of double bonds (Abidi 2001; Shimada and others 2001); therefore, only some sterols such as cholesterol can be determined directly with a UV detector; the others must be derivatized before injection (Shimada and others 2001).

The PDA is much more helpful than UV or UV-Vis because it can facilitate multichannel detection (multiple wavelengths) at the same time; therefore, the absorbance information is stored at multiple wavelengths and is scanned over the whole run time. The information can be used to detect compounds other than cholesterol such as tocopherols, tocotrienols, or β-carotene that do not have the same absorptive region as cholesterol. Prates and others (2006) and Ponte and others (2008) simultaneously collected absorbance information on both cholesterol (202 nm) and β-carotene (450 nm) in bovine and chicken meats, respectively, using an HPLC-PDA system. In addition, a fluorescent detector was used to quantify tocopherols and tocotrienols. By employing PDA and FD at the same time, all 4 compounds, which have very different chromatic properties, were measured simultaneously. In their studies, there was no derivatization of any compound under investigation. Other studies, however, have explored special derivatization reactions besides ester and ether derivatization targeting specific quantitative purposes such as creating sterol derivatives allowing for fluorescent (Iwata and others 1987; Fenton 1992; Abidi 2001; Lin and others 2007; Griffiths and Wang 2009) or UV-enhanced detections (Dong and others 2007). Fluorogenic reactions have been used for HPLC quantification by means of either pre- or postcolumn derivatization and allow for the detection at trace levels with great specificity (Ohkura and others 1994). Fluorogenic reactions are either fluorescence-generating or fluorescence-tagging; the former can be used in both pre- and postcolumn derivatization, whereas the latter can only be used in precolumn derivatization (Ohkura and others 1994). Iwata and others (1987) used 3,4-dihydro-6,7-dimethoxy-4-methyl-3-oxo-quinoxaline-2-carbonyl azide in benzene to convert cholesterol and cholestanol extracted from human serum by hexane into fluorescent carbamic esters. The levels of detection (LOD, based on signal:noise ratio) were 2 and 3 pg for cholesterol and cholestanol, respectively. Lin and others (2007) investigated the same principle with HPLC-fluorescent detection with a different fluorogenic reagent (naproxen acyl chloride) in cow milk, soy milk, saliva, and urine samples. The sensitivity was high, with an LOD of 25 nM. Using a simple precolumn derivatization with chromic and sulfuric acids in acetone (the Jones oxidation) to convert cholesterol into cholest-4-en-3,6-dione, Dong and others (2007) were able to enhance the UV absorbance of the cholesterol derivative to such a strong level (at a much more favorable wavelength of 250 nm) that it allowed for a very effective HPLC analysis of cholesterol in biological samples at trace levels. The enhancement gave the method a very high sensitivity (LOD of 0.2 pmol) and precision (CV of 0.5%). Daneshfar and others (2009) reported an exceptional sensitivity of an HPLC-UV method at 210 nm employing dispersive liquid–liquid microextraction with 0.01 μg/L LOD (0.1 pmol based on sample volume and molecular weight of cholesterol) and 3.1% CV.

In recent years, the ELSD has been widely used in lipid analysis in combination with HPLC (Kermasha and others 1994; Abidi 2001; Seppanen-Laakso and others 2001; Shimada and others 2001; Lin and others 2007). The ELSD measures the intensity of the light scattering by droplets of analyte when effluent evaporates after leaving the HPLC column. Although requiring the same type of calibration for quantification as that of the UV technique, ELSD is a generic detector (lack of specificity) and less sensitive than UV (Abidi 2001; Shimada and others 2001). Kermasha and others (1994) reported that a laser light-scattering detector gave 2.5 times higher LOD (less sensitive) of cholesterol and cholestane (common internal standard in cholesterol determination) than a UV detector. The laser light-scattering detection was also less precise for determining cholesterol and cholestane than was the UV detection (CV of 18.2% and 3.0% compared with 1.8% and 1.5%, respectively). However, UV detection was less sensitive for determining other cholesterol oxidation products compared with the laser light-scattering detection. Seppanen-Laakso and others (2001) validated an HPLC-ELSD method for routine analysis of major lipid classes in serum and reported a CV of 4.5% in both low and high cholesterol concentrations. The accuracy of HPLC-ELSD method for free cholesterol was also greatest compared with other lipid compounds in serum.

ECD enhances the use of HPLC technique in determining cholesterol to a sensitivity level beyond that of UV detection. Nakajima and others (1995) used 2-[2-(isocyanate)ethyl]-3-methyl-1,4-naphthoquinone as precolumn derivatization reagent to convert cholesterol into an electrochemically active ester of carbamic acid. After separated in an HPLC column, the derivative was reduced in a platinum catalyst reduction column online with the HPLC column. The final derivative was quantified by an electrochemical detector in oxidation mode with a glassy carbon working electrode and an Ag/AgCl reference electrode (0.7 V applied potential). The LOD was 6.6 pg (17 fmol) at 10-μL injection volume for cholesterol, which is comparable to FD previously discussed. Interestingly, Hojo and others (2007) showed that cholesterol could be electrochemically determined without any precolumn derivatization. A preliminary experiment was conducted to determine that cholesterol could be oxidized on a glassy carbon electrode in a nonaqueous solvent (isopropanol/acetonitrile) with 50 mM LiClO4 and an applied potential of 1.9 V (Ag/AgCl reference electrode). The LOD was 0.36 μM (1.8 pmol). Although the sensitivity was not as great as that of the method proposed by Nakajima and others (1995), the one investigated by Hojo and others (2007) was simpler because it did not require derivatization. The latter was also claimed to be more robust than the GC-MS method; however, there was not a direct comparison in the study (Hojo and others 2007).

Internal standards for cholesterol quantification

For quantitative purposes, an internal standard is important for cholesterol determination to compensate for any loss occurring during sample preparation and quantification. It is suggested that an internal standard should be added at the earliest step possible (Fenton 1992). An internal standard is especially important in the GC technique to correct injection error (Dinh and others 2008) because injection volume is usually very small, typically around 1 μL. An internal standard is also added to an external standard series at the same concentration as that of samples (routine analysis). Moreover, the internal standard is used to determine the response factor and monitor the linearity of the response, especially the response of the flame in FID (Fenton 1992; Araujo and others 2006). The most common internal standard for GC determination of cholesterol is 5α-cholestane (Kovacs and others 1979; Ryan and Gray 1984; Adams and others 1986; Ulberth and Reich 1992; Richardson and others 1994; Rodriguez-Palmero and others 1994; AOAC Official Method 994.10, AOAC International 1996; Ballesteros and others 1996; King and others 1998; Paleari and others 1998; Rowe and others 1999; Hwang and others 2003; Padre and others 2006; Pratiwi and others 2006; Polak and others 2008; Muchenje and others 2009; Viturro and others 2010; Cannata and others 2010). Error compensation with an internal standard is based on the assumption that the internal standard is chemically similar to the analyte; however, 5α-cholestane is an alkane that does not resemble cholesterol. Cholesterol-resembling compounds, such as plant sterols, however, are present in various foods, especially processed foods, for which most cholesterol analytical methods are usually developed. Therefore, 5α-cholestane has become the typical choice because it has a structure close to that of cholesterol, and it is absent in foods. Araujo and others (2006) suggested that the most suitable response factor for using 5α-cholestane in cholesterol determination (GC-FID) is 0.68. At GC conditions and cholesterol and 5α-cholestane concentrations that result in such a value, the GC determination of cholesterol could be performed without standard series. In our laboratory, a response factor from 0.64 to 0.69 was established with cholesterol concentration ranging from 0.008 to 0.020 mg/mL and a 5α-cholestane concentration of 0.005 mg/mL. However, the response factor was slightly increased when cholesterol concentration increased, probably caused by a small tail of cholesterol peak (Figure 2). The difference in peak shape (width and tailing) highlighted the disadvantages of using 5α-cholestane as internal standard for cholesterol determination because it may not behave the same as cholesterol, especially during extraction and separation. Although suggested by Fenton (1992) as a common practice, the addition of 5α-cholestane in the beginning of sample preparation was found to be unnecessary (Dinh 2010). Sorenson and Sullivan (2007) came to the same conclusion in their collaborative study on GC determination of phytosterols.

Another available and excellent choice for an internal standard is an isotope of the analyte for isotope dilution/MS (ID/MS). ID/MS is a highly reliable and accurate method for quantification. It has become the definitive method of analysis, which has been used to certify concentrations of food-related analytes in SR materials (Ellerbe and others 1989; Welch and others 2001; NIST 2010). The ID/MS was initially evaluated for determining cholesterol in serum only (Gambert and others 1979; Cohen and others 1980; Takatsu and Nishi 1987). The isotopically labeled cholesterol such as cholesterol-(3,4–13C2), cholesterol-d7, or cholesterol-(25,26,2713C3) is added to the sample as an internal standard. The sample then undergoes a preparation processes (with or without derivatization), separation by GC or HPLC, and ionization in the MS system. The ion fragments are collected by a mass collector and are analyzed for ion intensity (Fenton 1992). An isotope is almost the absolute resemblance of an analyte because it has the same number of protons, which defines their identical chemical properties. Isotopically labeled cholesterol has identical chemical properties to cholesterol because the chemical composition is not altered, just the mass of one or more atoms in the labeled cholesterol molecule. Therefore, the isotopically labeled internal standard and the analyte behave similarly through all preparation and quantitative steps and co-elute, except that they have different masses. For example, the cholesterol-(25,26,2713C3) (Welch and others 2001) and cholesterol-d7 (Ellerbe and others 1989) after TMS derivatization and electron impact ionization are monitored at m/z 461 and m/z 465, respectively, whereas the TMS derivative of food cholesterol is monitored at m/z 458 (Welch and others 2001). The information about mass and ion intensity is used to quantify analytes. Gambert and others (1979) reported a 0.997 correlation between GC and ID/MF (ID mass fragmentation with chemical ionization) when using cholesterol-(3,4–13C2) as an internal standard. Cohen and others (1980) described GC-ID/MS (using cholesterol-d7) as a highly accurate and precise method for measuring cholesterol in serum with a CV of 0.36%. Takatsu and Nishi (1987), using cholesterol-(3,4–13C2), showed that ID/MS could be coupled with an HPLC system for the same accuracy as GC-ID/MS (1% difference between HPLC-ID/MS and GC-ID/MS); however, the HPLC-ID/MS had a slightly greater CV. Pelletier and others (1987) similarly proposed an optimized GC-ID/MS [using cholesterol-(3,4–13C2)] for serum cholesterol. Eckfeldt and others (1991) developed a GC-ID/MS method for serum cholesterol with a benchtop quadrupole MS and cholesterol-(25,26,2713C3) as an internal standard and found that their method was as precise and accurate as the national definitive method developed by NIST using a magnetic sector MS for serum cholesterol. The studied method was also less labor intensive so that a single analyst could analyze up to 30 samples in 8 h. The ID/MS technique has been shown to be superior to other quantification methods and been the only definitive method for cholesterol analysis in foods (Ellerbe and others 1989; Welch and others 2001; NIST 2010); however, this technique is more expensive, more time-consuming, and therefore impractical for routine analysis and is only used as reference method and for research purposes. Recent advances in MS and ID/MS techniques for the analyses of sterols and food-related compounds were reviewed by Shimada and others (2001), Careri and others (2002), and Griffiths and Wang (2009).

A method can use several internal standards at the same time (Richardson and others 1994). Other internal standards that have been used occasionally for cholesterol analysis are 5,7-dimethyltocol, cholesteryl n-butyrate, cholestanol, betulin, stigmasterol, and epicoprostanol (Fenton 1992; Richardson and others 1994; Toivo and others 1998; Hojo and others 2007). Moreover, to determine cholesterol, a method can employ both external and internal standard calibrations. Fletouris and others (1998) tested the efficiencies of external and internal calibration methods and found that they both had the same degree of linearity and reliability. For practical purposes, it is recommended that an external standard calibration be used because it is much more cost-effective and productive for routine analysis.

Conclusions

Cholesterol content of meat and poultry has been important in making nutritional decisions. Selection of protein sources always includes the consideration of fat and cholesterol, the inseparable components of meat and poultry, because meat and poultry are obviously the most important protein sources with abundance and affordability. Recent updates of the USDA Natl. Nutrient Database for SR on meat and poultry composition have provided an opportunity to review the cholesterol content of many retail cuts of beef, pork, and chicken in the U.S. Rigorous analyses and verification and coordinated effort by multiple laboratories brought some new interesting data such as cholesterol content of chicken dark meat and the possibility of fat and cholesterol migration during cooking. Most of the changes recently found in the cholesterol content of meat and poultry have resulted from improved methodology and instrumentation rather than an effort by the meat and poultry industries to alter cholesterol content. Nonetheless, cholesterol content could also be affected by industry efforts to produce high-yielding animals and leaner products because such efforts can lead to changes in muscle composition and muscle fiber size. Currently developed and available instrumentation and SR materials of animal origin not only expand the analytical options for cholesterol determination, including some definitive methods employing ID/MS, but also increase the certainty of the analytical results because of better quality control.

The selection of a specific method for cholesterol determination is a complicated task. Analytical purposes, nature and composition of samples, method performance parameters (accuracy, precision, LOD, and LOQ - level of quantification), productivity, skills and knowledge of analyst, and available capital and equipment all affect the decision. Research has compared cholesterol analytical methods (colorimetric, enzymatic, and chromatographic) in foods, including meat products. Many procedural steps (fat extraction, derivatization, saponification, sample injection, detection, and quantification) have been improved or eliminated over the years because of the development of new technologies and the demands on new matrices with more complications and challenges (Bohac and others 1988; Jiang and others 1991; Richardson and others 1994; Kermasha and others 1994; Maraschiello and others 1996; Murphy and others 1996; Bragagnolo and Rodriguez-Amaya 2003a; Hwang and others 2003; Osman and Chin 2006; Dinh and others 2008; Bragagnolo 2009; Schummer and others 2009). A variety of choices for cholesterol analysis in research exist from colorimetric to GC/HPLC-ID/MS; however, at the time, GC-FID and HPLC-FID/UV/PDA are probably the best choices for routine analysis of cholesterol in foods, including meat and poultry. The colorimetric method tends to overestimate cholesterol content, the enzymatic method requires much more labor and strict analytical conditions, and both methods are not specific. Other methods, such as TLC (preparative or coupled with FID), supercritical fluid chromatography (SFC), and capillary electrochromatography (CEC), have been used for the determination of cholesterol and other sterols (Fenton 1992; Abidi 2001; Lesellier 2001). The former sometimes has unacceptable precision and accuracy (Fenton 1992). CEC features high resolution, high efficiency, high speed, and minimal solvent consumption. It combines electrophoresis and HPLC and has better separation power than HPLC and SFC (Abidi 2001; Abidi and others 2002; Abidi 2004); however, CEC, although a promising technique, is still in its developmental phase. Current literature indicates the precision and sensitivity of cholesterol determination method in the descending order of GC, HPLC, SFC, and CEC (Abidi 2001, 2004). GC and HPLC do not vary much in terms of precision and accuracy; however, the GC-FID system has a superior LOD and LOQ. The HPLC system occasionally has better precision and resolution than the GC system. When there are additional compounds to analyze with cholesterol, the HPLC-PDA system is probably the optimal choice for routine analysis. Saponification is necessary to liberate cholesterol from the ester form and cell membrane, and direct saponification can work for any type of meat or food samples. Moreover, derivatization is not essential for cholesterol determination and it has been confirmed that free cholesterol can be injected directly into a chromatographic system for effective separation and quantification.

Acknowledgment

This review uses important data from the analysis of cholesterol in beef, pork, and chicken for the USDA Natl. Nutrient Database for Standard Reference (SR) at the Texas Tech Univ. The studies to update the SR were funded in part by USDA ARS Cooperative Agreement 58-1235-8-155 and NIH Agreement Y1-HV-8116-15. The authors acknowledge the review of data and the contribution by the USDA Nutrient Data Laboratory. The authors also acknowledge the contribution and support of the Natl. Cattleman's Beef Assn. through the Beef Check-Off program and the Natl. Pork Board.

Ancillary