PREDICTION OF PORK CARCASS COMPOSITION USING SUBCUTANEOUS ADIPOSE TISSUE MOISTURE OR LIPID CONCENTRATION 1 E. D. Aberle 2, T. D. Etherton a and C. E. Allen 3 Purdue University, West Lafayette, IN 47907 and University of Minnesota, St. Paul 55108 SUMMARY The usefulness of adipose tissue composition as a predictor of carcass composition was studied in pork carcasses selected at a commercial packing plant. Moisture and lipid concentrations in the middle subcutaneous adipose tissue dorsal to the last lumbar vertebra were used singly or in conjunction with measures of subcutaneous adipose tissue depth or muscle mass in regression equations to predict lipid content of the rear leg or ham, which closely parallels lipid content of the body. Percent lipid in the ham was negatively correlated with percent adipose tissue moisture and positively related to adipose tissue lipid. Simple linear correlations ranged from.80 to.83. However, the relation between adipose tissue composition and ham lipid was not linear over the range of samples studied. Transformation of percent lipid in the ham to logl 0 improved all correlations to values greater than.84. Multiple correlations from regression analyses between percent lipid in the ham vs adipose tissue moisture combined with subcutaneous fat depth or muscle size were greater than.907. The highest multiple correlation (.962) and lowest standard error of the estimate was obtained when percent adipose tissue moisture, subcutaneous fat depth at the 10th rib and weight of the biceps femoris muscle were used to estimate log percent lipid in the ham. (Key Words: Adipose Tissue, Body Composition, Swine, Growth, Composition.) ~Joumal Paper No. 6353, Purdue Agricultural Experiment Station and Journal Paper No. 9609, Minnesota Agricultural Experiment Station. Department of Animal Sciences, Purdue University. adepartment of Animal Science, University of Minnesota. I NTRODUCTION Measurements of gross body composition are often required in studies of growth and development and assessment of nutritional or disease states such as human obesity. Gross body composition may be defined as the fat, water and solids content of the body (Berlin, 1961). Accurate estimates of body composition can be obtained by chemical analyses of the animals's body or of its carcass, but this procedure is costly and time consuming and results in destruction of the carcass. Nondestructive methods are available for estimation of composition in either live animals or their carcasses and include measurements of subcutaneous adipose tissue thickness (Hazel and Kline, 1952; Doornenbal et al., 1962), body or carcass density (Kraybill et al., 1952;Whiteman et al., 1953), total body water (Soberman, 1950; Kraybill et al., 1951), and body potassium content (Spray and Widdowson, 1950; Reines et al., 1953). Each procedure has special requirements in equipment and experimental skills. Often, the cost of equipment is quite high, which may prohibit the use of a procedure. Estimates of composition obtained by nondestructive procedures may be quite variable in accuracy. There is a continuing need for simple and relatively inexpensive techniques to accurately estimate body composition. In animals of comparable age within a species, the proportions of water, protein and ash in the fat-free body are relatively constant (Moulton, 1923; Murray, 1922). However, fat content of the body is extremely variable. As animals become fatter, chemical composition of the adipose tissues changes. Adipose tissue from lean animals contains more water and protein and less fat than adipose tissue from fat animals (Lush, 1926; Callow, 1948). This is caused by dilution of intracellular components in each fat cell as the size of the lipid droplet in each cell 449 JOURNAL OF ANIMAL SCIENCE, Vol. 46, No. 3,1977
450 ABERLE, ETHERTON AND ALLEN increases during fattening. This study was conducted to determine if composition changes in the adipose tissue alone or in conjunction with other carcass measures could be used for prediction of composition in swine carcasses from animals of comparable size but differing widely in fatness. MATERIALS AND METHODS Twenty-five pork carcasses were selected at a packing plant for this study. Carcass weighing between 68 and 82 kg with an average backfat thickness between 1.5 and 5.0 cm were studied. The right side of each carcass was wrapped and transported to the University of Minnesota Meat Science Laboratory. Carcass measurements (average backfat thickness, length, longissimus muscle area) were obtained and carcasses were cut into primal cuts as described by Christian et al. (1967). A single measurement of fat thickness over the longissimus muscle between the 10th and llth ribs was obtained as described by Fahey et al. (1975). A 5 to 10 g sample of the middle layer of subcutaneous adipose tissue was excised opposite the last lumbar vertebra and frozen for subsequent analysis. The middle layer was chosen because it undergoes much greater increase in thickness compared to the outer layer as pigs approach market weights (Anderson and Kauffman, 1973). Also, it is recognized that increases in backfat thickness over the lumbar region occur later than fat deposition over the shoulder. Thus, this site is probably one of the later maturing subcutaneous adipose tissue sites in the pig and might be expected to show a wider range in composition than other sites along the vertebral column among the carcasses selected for this study. The chemical composition of the untrimmed ham was used as an indicator of carcass composition. McMeekan (1940) and Doornenbal (1972) have demonstrated that composition of the untrimmed ham is highly correlated with composition of the whole carcass. After the biceps femoris muscle was dissected and weighed, bones were removed and the muscle, adipose tissue and skin of the ham were ground twice through a.95 cm plate and then thoroughly mixed. Approximately 1 kg of this coarsely ground material was then ground through a.32 cm plate three times and subsamples were analyzed for moisture, protein and ether extract (A.O.A.C., 1970). Nitrogen was converted to protein by multiplying by 6.25. Moisture in samples of middle subcutaneous adipose tissue was determined by two procedures. Slices 1 mm thick, weighing approximately 250 mg were extracted with 5 ml of anhydrous methanol in 50 ml Erlenmeyer flasks. Water was determined by the saponification procedure of Glass (1970) as modified by Hood et al. (1971) and Hood (1972). Moisture was also determined by drying adipose tissue slices in a vacuum oven at 85 C, 50 mm Hg for 16 hours. Fat content of adipose tissue was determined by Soxhlet extraction of oven-dried samples with anhydrous diethyl ether. Data were subjected to correlation and linear regression analyses as outlined by Snedecor and Cochran (1967). RESULTS Means, ranges and standard deviations for several carcass parameters and composition of the untrimmed ham and middle layer of lumbar subcutaneous adipose tissue are presented in table 1. These data indicate that the carcasses selected were nonuniform with respect to carcass measurements and gross composition. For example, the depth of subcutaneous adipose tissue over the longissimus muscle at the 10th rib ranged from 1.3 to 5.9 cm and the percent extractable lipid in the ham portion of the carcass ranged from 13.4 to 41.0. Nonuniformity of carcass composition among the sample population was desired for this study. It should be noted that adipose tissue moisture values were lower when determined by the saponification procedure as compared to values obtained by oven drying. The reason for this is not known. The simple correlation and regression coefficients between adipose tissue moisture or lipid and percent moisture, lipid or protein in the untrimmed ham are given in table 2. Absolute values for correlations between adipose tissue composition and percent moisture or extractable lipid in the ham were between.80 to.85; however, correlations with percent protein in the ham were lower than these values. Relationships between adipose tissue composition and untrimmed ham composition were evaluated for linearity over the range of samples studied. The regressions between adipose tissue moisture or extractable lipid and percent lipid in the ham deviated slightly from linearity so values for
PREDICTING BODY COMPOSITION 451 TABLE 1. CARCASS AND ADIPOSE TISSUE CHARACTERISTICS OF PORK CARCASSES SAMPLED Variable Meana Range S.D. Carcass weight, kg 71.9 60.8-89.8 6.2 Carcass length, cm 76.6 71.6-81.5 2.4 Average backfat thickness, cm 3.3 1.5-5.3 1.0 Fat thickness, 10th rib, cm 3.2 1.3-5.9 1.3 Longissiraus muscle area, cm 2 27.7 19.4 -- 34.6 4.5 Lean cuts, % 54.8 45.9 -- 63.9 5.1 Ham and loin, % 38.4 32.2-44.6 3.7 Untrimmed ham Moisture, % 56.5 45.0-67.3 5.9 Extractable lipid, % 26.8 13.4-41.0 7.5 Protein, % 15.3 12.3-18.3 1.8 Lumbar subcutaneous adipose tissue Moisture, saponification, % Moisture, oven drying, % Extractable lipid, % 5.62 4.00 -- 9.44 1.40 9.76 5.27 -- 18.53 2.87 87.75 76.69 -- 92.63 3.64 avalues are the mean of 25 observations. ham lipid were converted to log10 before correlation and regression analyses. Correlations were improved by this procedure (table 2). In figure la, percent lipid in the ham is plotted against percent adipose tissue moisture determined by oven-drying along with the regression line for the two variables, illustrating the deviation from linearity at either low or 40 t ~ 9 9 1.6 1.5 9 '25-9 -.\ -'J 9 o O_ 9.\ - 1.4 O O J ~ o 1.3 q 1.2 J t m I I I 6 9 12 15 18 % ADIPOSE TISSUE MOISTURE (a) ~.~ I I i I 9 12 15 18 % ADIPOSE TISSUE MOISTURE (b) Figure 1. Relation between percent lipid in the ham and percent adipose tissue moisture by oven drying. (a) linear plot of data (b) semilogarithmic plot of data.
452 ABERLE, ETHERTON AND ALLEN 0 Z m "8........ 9,.,-. o~ m "~z 0 ~' _~'~ om i',....," " ', ~,'1 ~176 q v g ~ m blo Z~ [.-, 0 c~,~.~ ~,~,~ o.~, o.~ o.~ ~B ~B BB B~ 0 0 0 0 0 0 A\ 2 q v,q,.-1 [... ~'~ m~ A\ ~ -~ u ~ "~, 9 ~, o o 0 ~ 2
PREDICTING BODY COMPOSITION 453 high values of adipose tissue moisture9 Figure lb is a semilogarithmic plot of the same data, showing that data points at either end of the curve fall closer to the regression line. Calculation of log-log equations did not improve the correlation coefficients over those for the semilogarithmic equations. Multiple regression equations were calculated using subcutaneous adipose tissue composition together with several carcass measurements to predict composition of the untrimmed ham. Only the equations in which percent lipid was the dependent variable are presented (table 3). All of these equations accounted for greater :~ than 82% of the variation in percent lipid in the untrimmed ham. Moisture values determined by ~_ oven drying gave slightly higher partial correlations than when values obtained by the saponification method were used in similar equations9 ~, The fourth equation, which included adipose tissue moisture, fat thickness (10th rib) and longissirnus muscle area, accounted for nearly ~ 89% of the variation in percent lipid of the Z ham. These measurements are easily obtained E and are nondestructive to the carcass. This ~ equation was improved upon by substituting weight of the bicepsfemoris muscle for longissi- ~ z rnus muscle area, equation 5. However, the ~ :~ disadvantage of this equation is that the biceps ~ fernoris muscle must be dissected from the O u ham. This is more time consuming and detracts,~ from the value of the ham. Equations predict- ~ ing log percent lipid in the ham accounted for a m Z slightly higher proportion of the variation Z (higher R z) than those predicting percent lipid. Percent moisture in the ham could be predicted ~z with accuracy comparable to that obtained for percent lipid by using the same independent z variables. For example, Y = 44.24 +.96 (%.~ adipose tissue moisture, oven drying) +.31 (longissirnus muscle area, cm~)-1.83 (10th rib fat thickness, cm), R 2 =.891, S.E.E. = 2.06. :~ Percent protein could not be predicted with the "q t~ same accuracy as eider percent lipid or moisture. For example, Y = 20.02--.12 (% adipose ~ tissue moisture, oven drying) +.015 (longissirnus muscle area, cmz)--l.26 (10th rib fat thickness, cm), R ~ =.621, S.E.E. = 1.17. Measurements of subcutaneous fat thickness, either average backfat thickness or fat thickness at the 10th rib, were better predictors of percent lipid in the ham (table 4) than were measures of adipose tissue composition. When either average backfat thickness or fat thickness at the 10th rib was used in combination with X 0 N - ~.,9 ~;,,.o u 0 9 -~ eq.... O.... O....,-....,. 9.C~O0... ~ 0. 9.xO... 0 0 9.,",.. 9.,",.. 9 u~ " ~ 0 ~ " 0 " 0 0 Z~4 ~-ie4 Z " ~ " " i~ 00 -=.~ o ~a
454 ABERLE, ETHERTON AND ALLEN MM~MM longissirnus muscle area, prediction of percent ham lipid was nearly as accurate as when adipose tissue composition" was used in combination with carcass measurements. Discussion Z t~,, z z.o to,~ to t~ [- 2 ~.'2. O i. v-4 t~ r ao~o~ o~ r eq r t~ r ~o am i,~ d '6 "O ~a The data reported above support the hypothesis that compositional changes in the adipose tissue can be useful parameters in predicting composition of the untrimmed ham. Since composition of the untrimmed ham has been shown to closely parallel composition of the whole carcass (McMeekan, 1940; Doornenbal, 1972), adipose tissue water or extractable lipid concentration would be useful in predicting carcass or whole body composition. Fat or moisture content would be predicted with higher accuracy than would protein content. Values for adipose tissue moisture obtained by either saponification or oven-drying procedures were of essentially equal value in prediction equations. Even though the procedures gave different values for moisture concentration, the results between the two techniques were directly correlated. The saponification technique would be preferred in applications where rapid results are needed or where sample size is limited, as in samples obtained by biopsy on anesthetized animals. The composition of adipose tissues varies among depots and sites within a depot (Lee and Kauffman, 1974). In preliminary experiments with pigs that were quite uniform in size and body composition, the outer layer of subcutaneous adipose tissue had higher moisture concentration (by the saponification technique) than the middle layer of all adipose sites sampled. Adipose tissue from the shoulder region had higher moisture concentrations than that from the lumbar region. Representative values were: shoulder, outer layer, 8.99 +.43%, middle layer, 6.86 +.21%; lumbar, outer layer, 6.92 +-.26% middle layer, 6.06 +.23% (values are mean + SEM, n = 6). Since percent adipose tissue lipid is inversely related to percent moisture, differences among depots and sites within a depot in percent adipose tissue lipid would be the reverse of differences observed for percent moisture. Thus, it is important that the depot site which is sampled be carefully standardized. Lush (1926) found a curvilinear relationship between percent lipid in the total body of cattle and percent extractable lipid in either
PREDICTING BODY COMPOSITION 455 perirenal or mesenteric adipose tissue. The range in adipose tissue lipid in his study was from less than 32% to greater than 96%. The relationship between percent body lipid and percent adipose tissue lipid was most closely approximated by the curve of a rectangular hyperbola. At very high values for percent adipose tissue lipid, the curve was parallel to the Y-axis and at very low values, it was parallel to the X-axis. The relationship between adipose tissue moisture or lipid and lipid concentration in the ham in the present study was also curvilinear. This agrees with the data of Lush (1926), but the range in adipose tissue composition was not sufficiently wide to justify fitting of curves beyond the log transformations employed. Hypertrophy of adipocytes due to enlargement of the intracellular lipid droplet is not the only process contributing to increased size of adipose tissue depots during fattening (development of obesity). There is also an increase in the absolute amount of non-lipid cellular and extra-cellular material contained in the depot (Lush, 1926; Thomas, 1962). Adipocyte numbers increase during growth and fattening in swine (Anderson and Kauffman, 1973; Hood and Allen, 1977) and to approximately 5 weeks of age in rats (Greenwood and Hirsch, 1974). Allen et al. (1974) and Allen (1976) reported that as swine and cattle become progressively more obese, the frequency distribution of adipocyte diameters assumes a bimodal character due to the development of a population of adipocytes with diameters less than 40 to 60 micrometers. It was postulated (Allen, 1976) that this distinct population of small adipocytes appears in response to the fact that the potential for further increases in diameter of the larger adipocytes is minimal. Further, the population of small adipocytes may have originated from reinitiation of hyperplasia in adipose tissue or from previously metabolically inactive adipocyte precursor cells which are entering the lipid filling phase of development. Recently, evidence was presented indicating that adipocyte hyperphasia was responsible for the bimodal distribution of adipocyte diameters observed in the Zucker obese female rat at 14 weeks of age (Johnson et al., 1976). These observations on adipose tissue cellularity in very obese animals are in partial accord with the non-linear relationship between adipose tissue composition and ham or carcass composition. Thus, as adipocytes approach their maximum size, there would be less change in their composition and a non-linear relationship with ham or carcass fat would be predicted. With the appearance of a bimodal population of cell sizes containing small adipocytes, one might expect the percent moisture in the adipose tissue to increase, even though the animal is more obese. This apparently did not happen in the carcasses used in this study since the most obese carcasses sampled did not have higher percent adipose tissue moisture than carcasses which were slightly less obese. Since cellularity data were not determined in this study, it is not known whether adipocyte diameters were mono- or bi-modally distributed. However, a more recent study in our laboratory indicates that the adipose tissue moisture content increases when the adipocyte diameter distribution becomes bimodal in the advanced stages of fattening (unpublished data). In conclusion, this study indicates that adipose tissue moisture or lipid content can be a useful predictor of composition in the ham or hind limb of the pig. A similar relationship would be expected between adipose tissue composition and carcass or body composition. It should be recognized that the variation in composition among carcasses in the sample population was probably greater than in a more closely defined population of pork carcasses such as may be present in a herd or an experiment. Thus, the correlations between adipose tissue composition and ham composition reported herein may be higher than similar correlations for such a population. Further studies are necessary to determine if adipose tissue moisture determined from adipose tissue biopsies can be equally useful in predicting body composition. The use of moisture in adipose tissue biopsies either singly or in combination with physical measurements on the animal or human to predict body composition may be a relatively simple and accurate method. LITERATURE CITED Allen, C. E. 1976. CeUularity of adipose tissue in meat animals. Fed. Proc. 35:2302. Allen, C. E., E. H. Thompson and P. V. J. Hegarty. 1974. Physiological maturity of muscle and adb pose cells in meat animalg Proc. Recip. Meat Conf. 27:8. Anderson, D. B. and R. G. Kauffman. 1973. Cellular and enzymatic changes in porcine adipose tissue
456 ABERLE, ETHERTON AND ALLEN during growth and fattening. J. Lipid Res. 14:160. A.O.A.C. 1970. Official Methods of Analysis. (11th Ed.) Association of Official Analytical Chemists. Washington, DC. Berlin, N. I. 1961. Somatometric approach to body composition: Introductory remarks. In J. Brozek and A. Henschel (Ed.) Techniques for Measuring Body Composition. National Academy of Sciences-National Research Council, Washington, DC. Callow, E. H. 1948. Comparative studies of meat; the changes in the carcass during growth and fattening and their relation to the chemical composition of the fatty and muscular tissues. J. Agr. Sci. 38:174. Christian, J. A., R. H. Ingram, M. D. Judge, R. A. Merkel, C. E. Shelby, J. R. Stouffer and C. L. Strong. 1967. Guides for pork carcass evaluation. In L. E. Orme (Ed.) Recommended Guides for Carcass Evaluation and Contests. Amer. Meat Sci. Ass., Chicago, IL. Doornenbal, H. 1972. Growth, development and chemical composition of the pig. II. Fatty tissue and chemical fat. Growth 36:185. Doornenbal, H., G. H. Wellington and J. R. Stouffer. 1962. Comparison of methods used for carcass evaluation in swine. J. Anim. Sci. 21:464. Fahey, T. J., D. M. Schaefer, R. G. Kauffman, J. R. Romans, R. J. Epley, D. G. Topel and G. C. Smith. 1975. A comparison of pork carcass evaluation systems. J. Anim. Sci. 41:290 (Abstr.). Glass, R. L. 1970. Saponification reaction and the determination of water. Anal. Biochem. 27:219. Greenwood, M. R. C. and J. Flirsch. 1974. Postnatal development of adipocyte cellularity in the normal rat. J. Lipid Res. 15:474. Hazel, L. N. and E. A. Kline. 1952. Mechanical measurement of fatness and carcass value in live hogs. J. Anim. SCi. 11:313. Hood, R. L. 1972. Adipose tissue cellularity and lipogenic activity in porcine and bovine animals. Ph.D. Thesis. University of Minnesota, St. Paul. Hood, R. L. and C. E. Allen. 1977. Cellularity of porcine adipose tissue: Effects of growth and adiposity. J. Lipid Res. (In press). Hood, R. L., C. E. Allen, R. D. Goodrich and J. C. Meiske. 1971. A rapid method for the direct chemical determination of water in fermented feeds. J. Anim. Sci. 33:1310. Johnson, P. R., J. Stern, R. Gruen, S. Blanchett-Hirst and M. R. C. Greenwood. 1976. Development of adipose depot cellularity, plasma insulin, pancreatic insulin release and insulin resistance in the Zucker obese female rat. Fed. Proc. 35:657 (Abstt.). Kraybill, H. F., H. L. Bitter and O. G. I-hnkins. 1952. Body composition of cattle. II. Determination of fat and water content from measurement of body specific gravity. J. Appl. Physiol. 4:575. Kraybill, H. F., O. G. Hankins and H. L. Bitter. 1951. Body composition of cattle. I. Estimation of body fat from measurement in vivo of body water by use of antipyrine. J. AppI. Physiol. 3:681. Lee, Y. B. and R. G. Kauffman. 1974. Cellularity and lipogenic activities of porcine intramuscular adipose tissue. J. Anita. Sci. 38:538. Lush, J. L. 1926. Practical methods of estimating the proportions of fat and bone in cattle slaughtered in commercial packing plants. J. Agr. Res. 32: 727. McMeekan, C. P. 1940. Growth and development of the pig, with special reference to carcass quality characters, J. Agr. Sci. 30:276. Moulton, C. R. 1923. Age and chemical development in mammals. J. Biol. Chem. 57:79. Murray, J. A. 1922. The chemical composition of animal bodies. J. Agr. Sci. 12:103. Reines, R., R. L. Schuch, C. L. Cowan, Jr., F. B. Harrison, E. C. Anderson and F. N. Hayes. 1953. Determination of total body radioactivity using liquid scintillation detectors. Nature 172: 521. Snedecor, G. W. and W. G. Cochran. 1967. Statistical Methods. (6th Ed.) Iowa State University Press, Ames. Soberman, R. J. 1950. Use of antipyrine in measurements of total body water in animals. Proc. Soc. Exp. Biol. Med. 74:789. Spray, C. M. and E. M. Widdowson. 1950. The effect of growth and development on the composition of mammals. Brit. J. Nutr. 4:332. Thomas, Lorette T. 1962. The chemical composition of adipose tissue of man and mice. Quart. J. Exp. Physiol. 47:179. Whiteman, J. V., J. A. Whatley and J. C. Hillier. 1953. Further investigation of specific gravity as a measure of pork carcass value. J. Anim. Sci. 12:859.