- Email: [email protected]
Analytical bone calcium and bone ash from mature laying hens correlates to bone mineral content calculated from quantitative computed tomography scans
Analytical bone calcium and bone ash from mature laying hens correlates to bone mineral content calculated from quantitative computed tomography scans C. I. Robison∗,1 and D. M. Karcher† Department of Animal Science, Michigan State University, East Lansing, MI 48824, USA; and † Department of Animal Sciences, Purdue University, West Lafayette, IN 47907, USA
ABSTRACT Laying hen skeletal health continues to be an industry priority. Bone ash and bone Ca quantification in laying hen long bones provides valuable information on skeletal health. Unfortunately, these measurements can only be accomplished by sacrificing hens, thus making longitudinal measurements on the same hen impossible. Quantitative computed tomography (QCT), used with a calcium hydroxyapatite phantom, has been used to determine bone density of wings and legs as well as live hens throughout the production cycle by scanning with a calcium hydroxyapatite phantom. QCT has also been used to scan live hens throughout the production cycle. The purpose of this study was to determine how QCT calculated bone mineral content (QCT BMC) corresponds to analytical bone Ca and bone ash. Wing and leg quarters from 72-wk-old W-36 hens were QCT scanned along with a QCT Phantom. After scanning, humeri, femurs, and tibias were cleaned, divided into epiphysis (E) and diaphysis (D), fat extracted, ashed, and digested under nitric acid, and Ca was determined by atomic absorption spectroscopy was used to determine E, D, and
whole bone Ca. Four bones/type were used for E and D, while 6 bones/type were used for whole bone measurements. A second set of bones were prepared to determine correlation of BMC to bone ash. QCT scans were analyzed with Mimics software (Materialise NV, Leuven, Belgium) to calculate bone volume and density in Hounsfield units. Utilizing the QCT phantom and bone volume, BMC was calculated for E, D, and whole bone. Data were analyzed with regression analysis and Pearson correlation coefficients were determined. Analytical Ca was correlated to QCT BMC for E (R = 0.84, P < 0.01), D (R = 0.99, P < 0.01), and whole bone (R = 0.97, P < 0.01). Whole bone ash was highly correlated to QCT BMC for femur (N = 47, R = 0.97, P < 0.001), tibia (N = 50, R = 0.94, P < 0.001), and keel (N 50, R = 0.94, P < 0.001). Whole bone ash and QCT BMC values of femur and tibia were not different (P = 0.39 and 0.22 respectively). Based on this information, QCT could provide relative quantitative assessment of total bone mineral in live birds proving useful in long-term studies.
Key words: laying hen, bone mineral, computed tomography, calcium 2019 Poultry Science 0:1–6 http://dx.doi.org/10.3382/ps/pez165
INTRODUCTION
tracking of the same bird. Alternatively, bone mineral content calculated from densitometry can provide similar information as bone ash. Dual-energy x-ray absorptiometry (DEXA), quantitative computed tomography (QCT), and peripheral QCT have been employed to determine bone mineral density and estimate bone mineral content in vivo and ex vivo (Korver et al., 2004; Saunders-Blades et al., 2009; Regmi et al., 2015). Several issues arise with these approaches: (1) long scan time requiring bird sedation, (2) limitation on numbers of animals that can be used for data collection as a result of lengthy scan time, and (3) reproducibility will decrease when an area less than the whole bone is selected (Kim et al., 2012). Another approach that could be used is clinical QCT. Compared to micro-CT or peripheral QCT image quality is lower, but an increase in the number of animals
Calcium demands on laying hens during production have been well documented and any imbalance may result in birds with poor structural bone quality. In 50wk-old layers fed varying levels of Ca, bone breaking force was found to be highly correlated to bone ash weight and bone volume (Zhang and Coon, 1997). The monitoring of bone ash and bone volume could provide valuable insight into the skeletal health of the laying hen throughout production. Unfortunately, bone ash determination requires bird sacrifice, potentially altering cage density and does not allow for longitudinal C 2019 Poultry Science Association Inc. Received November 9, 2018. Accepted March 12, 2019. 1 Corresponding author: [email protected]
1
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
∗
2
ROBISON AND KARCHER
Table 1. Whole bone ash correlated to bone mineral calculated by quantitative computed tomography (QCT BMC) scans when masked at specific Hounsfield units (HU). Paired t-tests examine the difference between bone ash and scanned mineral.
N
Mask (HU)
Femur Tibia Humerus Keel
47 50 23 50
480 600 120 120
1 2
Ash (g)1 2.41 2.93 1.46 2.31
± ± ± ±
0.09 0.09 0.02 0.06
Paired t-test P-value
Pearson correlation Prob |r|
Correlation P-value
± ± ± ±
0.387 0.217 < 0.001 < 0.001
0.974 0.939 0.455 0.944
< 0.001 < 0.001 0.03 < 0.001
2.46 2.84 1.12 2.57
0.15 0.15 0.05 0.1
Data represent the mean ± SEM ash in grams for each type of bone. Data represent the mean ± SEM calculated bone mineral for each type of bone.
scanned, reduction in scan time, and the ability to scan the entire bird may compensate for this shortcoming. Clinical QCT scan quality can accurately create 3D models of bones that can be used for morphometric shape analysis as well as determining cortical thickness, cross-sectional area, volumetric BMD, bone volume, and morphological measurements such as length, width, angle, and circumference (Chargo et al., 2018; Regmi et al., 2015, 2017). Recently researchers were able to scan 16 laying hens per clinical QCT scan without anesthesia providing researchers with whole body data (Chargo et al., 2018). In the current study, the objective was two-fold. First, to determine how analytical bone Ca correlated to bone ash from the epiphysis, diaphysis, and whole bone for the tibia, humerus, and femur. Second, to determine if digital calculation of mineral content from humeri, tibia, femur, and keel bones from mature laying hens correlated to total mineral content measured as fat-free ash.
MATERIALS AND METHODS All procedures were approved by the MSU Animal Care and Use Committee. Bones were collected from ninety 85-wk-old Hy-Line W36 hens. The left wing, left leg quarter, and keel bone were tagged using a numbered plastic tag (Ketchum Manufacturing, Ontario, Canada) and tagging gun. The parts were frozen at −20◦ C until analysis. A power analysis was conducted to determine 11 samples that were needed with expected correlation of 0.85 and 0.95.
Computed Tomography Scans The parts were thawed for 24 h in a chiller at 6◦ C and then taken to the Michigan State University College of Veterinary Medicine for clinical QCT scans. Wings, leg quarters, and keels were carefully arranged on plexiglass in rows such that the parts were lined up end to end perpendicular to the gantry. All parts were scanned with feathers and muscle tissue intact. A solid calcium hydroxyapatite phantom (Image Analysis, Inc; Columbia, KY) consisting of rows representing 0, 75, and 150 mg/cm3 Ca was included in each scan. Each QCT scan used the following settings: 120 kV,
320 mAmp, 0.625 mm slices, and spine position using a GE Revolution Evo scanner (General Electric Healthcare, Princeton, NJ). Each row of bones from the QCT scan was reconned with a 7-cm field of view using Imageworks software (General Electric Healthcare, Princeton, NJ) with a bone algorithm resulting in a DICOM (Digital Imaging and Communications in Medicine) file for each row scanned. The DICOM files were imported into Mimics software (Materialise, Plymouth, MI) for analysis. An appropriate threshold Hounsfield units (HU) was selected to create a mask for the whole bone (Table 1). Then the region growing tool was used to separate each bone into its own mask. The multiple slice edit tool was used to remove the patella from the femur, the fibula from the tibia, and the ribs and other extra bone from the keel bone. Once bones were edited, the average HU and bone volume were recorded. Average HU was determined for each of the 3 concentrations of calcium hydroxyapatite at 10 locations along the length of the phantom for each scan. The HU were averaged for each concentration and a scatter plot was created with mg Ca hydroxyl apatite/cm3 versus HU. The equation of the regression line was used to convert density in HU to mg Ca hydroxyapatite/cm3 . Bone mineral content was calculated from the CT scans by multiplying the density of the bone in mg Ca hydroxyapatite/cm3 by the bone volume.
Analytical Chemistry Bones were prepared for analysis by placing the frozen parts into plastic autoclave tubs and autoclaving at 120◦ C for 19 min using a gravity cycle (733HCMC; Getinge, Wayne, NJ). Upon removal from the autoclave skin, muscle, and tendon were removed from each bone. For the keel bone, ribs and the proximal end was trimmed along the spina externa. Each long bone was measured in mm using 6” digital calipers (68304; Pittsburgh, PA). The calipers were used to mark the distance 25% from the distal and proximal ends of each bone, which was designated as the epiphysis (Epi). The remaining middle 50% of the bone was designated as diaphysis (Dia). Bones were cut, fat extracted, and dried as described in Van Wyhe et al. (2012). Crucibles were ashed in a Thermolyne furnace (30400; Barnstead
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Bone
QCT BMC (g)2
3
DIGITAL BONE MINERAL ANALYSIS
Table 2. Amount of bone mineral calculated by quantitative computed tomography scans (QCT BMC), analytical Ca, and ash quantity from 85-wk-old Hy-Line W36 laying hens. N
QCT BMC (g)2
Ash (g)
Analytical Ca (g)
% Ca in Ash
P-value1
12 4 4 4
0.904 0.949 1.3 0.466
± ± ± ±
0.11 0.03 0.10 0.05
0.872 0.928 1.12 0.569
± ± ± ±
0.07 0.01 0.08 0.02
0.315 0.324 0.412 0.208
± ± ± ±
0.03 0.01 0.03 0.01
36.1 35.0 36.8 36.5
± ± ± ±
0.3 0.3 0.4 0.2
0.42 0.30 0.01 0.058
Epiphysis Femur Tibia Humerus
12 4 4 4
1.19 1.51 1.42 0.64
± ± ± ±
0.14 0.14 0.17 0.08
1.34 1.40 1.70 0.927
± ± ± ±
0.11 0.13 0.11 0.05
0.485 0.498 0.620 0.338
± ± ± ±
0.04 0.04 0.04 0.02
36.2 35.8 36.3 36.5
± ± ± ±
0.3 0.6 0.3 0.3
0.048 0.25 0.02 0.017
Whole bone Femur Tibia Humerus
16 5 6 5
2.01 2.08 2.76 1.06
± ± ± ±
0.20 0.05 0.23 0.08
2.22 2.20 2.88 1.46
± ± ± ±
0.16 0.11 0.14 0.06
0.81 0.782 1.06 0.533
± ± ± ±
0.06 0.04 0.05 0.09
36.3 35.6 36.8 36.4
± ± ± ±
0.3 0.6 0.3 0.1
0.003 0.12 0.36 0.001
1 2
P-value indicates significance of the paired t-test between the QCT BMC and ash. Data is presented as mean ± SEM.
International) at 600◦ C for 10 h and weighed to determine ash percentage. All bones were QCT scanned and ashed; however, only a subset of bones were analyzed for Ca concentration. Samples analyzed for epiphyseal or diaphyseal Ca were randomly selected such that the same bone was not used for both analyses to allow for greater variation in the sample population. In order to determine Ca, ash was ground using a mortar and pestle and then 0.4-g ash was weighed into Teflon-lined digestion vials for microwave digestion as described by Shaw et al. (2002). Samples were allowed to digest overnight in 10 mL of 70% nitric acid (OmniTrace; MilliporeSigma, Temecula, CA). Samples were digested using a microwave accelerated reaction system (MARS-5; CEM Corp., Matthews, NC) as described by Lavin et al. (2013). Calcium concentrations in bone were determined by atomic absorption spectroscopy (AA-7000, Shimadzu, Kyoto, Japan). A peach leaves standard (National Institute of Standards and Technology, Gaitherburg, MD) was digested and analyzed with the samples (atomic absorption standard: ACAIKN-100, Ricca Chemical Co., Arlington, TX).
Statistics Pearson’s correlation analysis was conducted to understand the association between analytical bone ash and QCT bone mineral content (QCT BMC). Pearson’s correlations coefficients (R) between measurements are presented and were calculated using the CORRELATION procedure in SAS 9.3 (SAS/STAT User’s manual, 2014). Pearson correlations were also calculated for each bone by part combination. Linear regression analysis was completed using the REGRESSION procedure in SAS 9.3 to examine the relationship between analytical bone mineral and QCT BMC. Paired t-tests were used to determine amount of difference between the measurements of bone mineral calculated analytically versus digitally using the TTEST procedure in SAS 9.3.
Table 3. Regression equations of calculated mineral of the epiphysis (E), diaphysis (D), and whole bone (W) to bone ash. Part
n
R2
Equation
Humerus
E D W
4 4 23
0.40 0.72 0.21
y = –0.23x +0.94 y = –0.68x + 2.01 y = –0.255x + 0.94
Femur
E D W
4 4 47
0.68 0.86 0.95
y = 0.27x + 0.89 y = –1.08x + 2.19 y = –1.23x + 1.53
Tibia
E D W
4 4 50
0.94 0.91 0.88
y = –1.10x + 1.48 y = –0.023x + 1.18 y = –1.78x + 1.57
Keel
W
50
0.87
y = –0.88x + 1.49
Bone
RESULTS Bone Ca to Bone Ash Average Ca content and QCT BMC are presented for each bone type by section with Pearson correlations and associated P-values (Table 2). Analytical bone Ca from the tibia and femur were highly correlated to QCT BMC regardless of the part (R = 0.73 to 0.96); however, the humerus only reported a moderate-to-high correlation (R = 0.52 to 0.83; Table 3). The aforementioned correlations only included 4 samples for each combination; therefore, in order to more thoroughly examine the relationship, all bone data were combined for an overall correlation by part. The Pearson correlation between diaphyseal Ca (N = 12) and QCT BMC was highly correlated with R = 0.989 (P < 0.01; Figure 1). The Pearson correlation between epiphyseal Ca (N = 12) and QCT BMC was R = 0.873 (P < 0.01; Figure 1B). Additionally, the Pearson correlation between whole bone Ca (N = 16) and QCT BMC was R = 0.974 (P < 0.001; Figure 1C).
Bone Ash to QCT BMC Femur QCT BMC was highly correlated to ash weight with a Pearson correlation coefficient of R = 0.97
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Diaphysis Femur Tibia Humerus
4
ROBISON AND KARCHER
Figure 2. The relationship between whole bone ash and bone mineral content calculated from quantitative computed tomography scans (QCT BMC) for the (A) femur (n = 47; P < 0.01), (B) tibia (n = 50; P < 0.01), and (C) keel (n = 50; P < 0.01).
(n = 47; P < 0.001; Figure 2A). QCT BMC of the tibia was also highly correlated to ash weight with a Pearson correlation coefficient of R = 0.939 (n = 50; P < 0.001; Figure 2B). Keel bone QCT BMC was correlated to ash weight with a Pearson correlation coefficient of R = 0.931 (n = 50; P < 0.001; Figure 2C). The QCT
BMC of the humerus was only moderately correlated to ash weight with a Pearson correlation coefficient of R = 0.455 (n = 23; P = 0.029; Table 1). Paired t-tests were used to determine the amount of difference between the QCT BMC and bone ash for each bone type (Table 1). The average difference for the
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Figure 1. The relationship of analytical bone Ca to bone mineral content calculated from quantitative computed tomography scans (QCT BMC) for the (A) diaphysis (P < 0.01), (B) epiphysis (P < 0.01), and (C) whole bone (P < 0.01). Plots A and B represent a compilation of femur (n = 4), tibia (n = 4), and humeri (n = 4). Plot C represents a compilation of femur (n = 5), tibia (n = 6), and humeri (n = 5).
5
DIGITAL BONE MINERAL ANALYSIS
femur was 0.053 ± 0.06 g (P = 0.387) and the average difference for the tibia was 0.09 ± 0.07 g (P = 0.217). Keel bones had an average difference of –0.15 ± .03 g (P < 0.001). Humerus paired t-test resulted in an average difference of 0.339 ± 0.05 g (n = 23, P < 0.001).
The composition of bone including mineral and collagen content influences the material properties (van der Meulen et al., 2001). Total mineral content is generally estimated using the ash content of the bone. Ash is the inorganic residue remaining after water and other materials are vaporized and organic substances are burned in the presence of oxygen, providing a direct measure of BMC. Bone breaking force was found to be highly correlated to bone ash weight (Zhang and Coon, 1997; Hester et al., 2004) and bone volume (R2 = 0.682) (Zhang and Coon, 1997). DEXA, QCT, and peripheral QCT have been employed to determine bone mineral density and estimate bone mineral content in vivo (Saunders-Blades et al., 2009; Regmi et al., 2015). Baird et al. (2008) compared in vivo BMC of tibias from 58-wk-old laying hens scanned using DXA to bone ash and found DXA underestimated BMC by 9.2% (Baird et al., 2008). In the current study, QCT BMC overestimated the BMC of the keel by 11% and underestimated the BMC of the humerus by 23%. Unlike the humerus and keel, the QCT BMC of femur and tibia were not different from the ash weight when the whole bone was used. In the current study, QCT BMC of the whole bone was highly correlated to ash weight in the femur, tibia, and keel while only moderately correlated in the humerus. There are some challenges with clinical QCT images. For example, the humerus was difficult to threshold effectively since the bone was equal to or less dense than the muscle tissue surrounding it, likely due to its hollow nature. The keel bone thresholds nicely at 120 HU; however, the calculated bone mineral was overestimated suggesting that either the HU threshold should be increased slightly or there was difficulty matching the shape of the digital image to the shape of the actual bone. On the other hand, the femur and tibia are easily imaged and there is minor difficulty in removing the patella and the fibula from the thresholded areas. Once proficiency is achieved with the software, digital data can be collected and calculated faster than bone ash can be conducted. Calcium content of the bone was measured to examine the correlation between Ca content and QCT BMC to determine if digital measurements could be used to estimate Ca content. The femur and tibia were highly correlated to Ca content while the humerus was moderately correlated. The Ca content accounted for approximately 37% of the ash weight regardless of bone. Both quality and quantity of bone are important to overall bone health. Neither whole bone ash nor analytic Ca
REFERENCES Baird, H. T., D. L. Eggett, and S. Fullmer. 2008. Varying ratios of omega-6: omega-3 fatty acids on the pre-and postmortem bone mineral density, bone ash, and bone breaking strength of laying chickens. Poult. Sci. 87:323–328. Chargo, N., C. I. Robison, S. L. Baker, M. J. Toscano, M. M. Makagon, and D. M. Karcher. 2018. Keel bone differences in laying hens housed in enriched colony cages. Poult. Sci. 98:1031–1036. ´ Cs´oka, G. Kov´ ¨ Petneh´ Donk´ o, T., A. Tischler, A. acs, M. Emri, O. azy, A. Szab´ o, V. Halas, J. Tossenberger, R. Garamv¨olgyi, and G. Bajzik. 2018. Estimation of bone mineral density and breaking strength of laying hens based on scans of computed tomography for body composition analysis. Br. Poult. Sci. 59:365–370.
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
DISCUSSION
measurements describe the distribution of bone mineral or the quality of the bone (Korver et al., 2004). In addition to volumetric bone density, QCT can provide anatomical shape, which could be coupled with Finite Element analysis to analyze bone quality and the material properties of bone. Recent work using clinical CT with laying hens has demonstrated high correlations between calculated bone mineral density and maximum breaking strength (R = 0.78 to 0.80) in the tibiotarsal bone (Donk´ o et al., 2018). Clinical QCT allows researchers the ability to scan the entire bird quickly, potentially allowing researchers to take advantage of other digital calculations and analysis not previously available to them. Live birds can be scanned using a restraint device as described in Chargo et al. (2018) with up to 24 live birds on 1 scan. This set-up provides both a high-throughput and economical solution for collecting a considerable amount of skeletal data from live hens without the need for anesthesia. In a human study examining risk factors associated with femur fractures, researchers reported that volumetric BMD and cortical thickness represented the key differences between fractured and non-fractured bones (Raudaschl and Fritscher, 2017). Additionally, in a reanalysis of published data on osteoporosis, measurements of bone mass provided an indication of bone strength and fracture risk in humans with diminished bone mass associated with increased fracture risk regardless of measurement site (Ross et al., 1990). Images from clinical QCT can be used to create accurate 3D representations of the femur, tibia, and keel. Statistical shape models have been used in human medicine for therapy planning and fracture prediction using 3D distribution of BMD and shape (Whitmarsh et al., 2011). A recent publication demonstrated how 3D representations of the keel bone can be used to document changes in keel bone fractures and deviations in live birds during a longitudinal study using 3D part comparison (Chargo et al., 2018). The data support that bone ash can be successfully estimated from digital images acquired by QCT for laying hens. This method would provide researchers with a non-invasive tool to monitor bone mass and BMD throughout production. This approach can provide a quick, reliable method to monitor longitudinal bone health without sacrificing birds.
6
ROBISON AND KARCHER Ross, P. D., J. W. Davis, J. M. Vogel, and R. D. Wasnich. 1990. A critical review of bone mass and the risk of fractures in osteoporosis. Calcif. Tissue Int. 46:149–161. Saunders-Blades, J. L., J. L. MacIsaac, D. R. Korver, and D. M. Anderson. 2009. The effect of calcium source and particle size on the production performance and bone quality of laying hens. Poult. Sci. 88:338–353. Shaw, D. T., D. W. Rozeboom, G. M. Hill, and J. E. Link. 2002. Impact of vitamin and mineral supplement withdrawal and wheat middling inclusion on finishing pig growth performance, fecal mineral concentration, carcass characteristics, and the nutrient content and oxidative stability of pork1. J. Anim. Sci. 80:2920– 2930. van der Meulen, M. C., K. J. Jepsen, and B. Miki´c. 2001. Understanding bone strength: size isn’t everything. Bone 29:101–104. Van Wyhe, R. C., T. J. Applegate, M. S. Lilburn, and D. M. Karcher. 2012. A comparison of long bone development in historical and contemporary ducks. Poult. Sci. 91:2858–2865. Whitmarsh, T., K. D. Fritscher, L. Humbert, L. M. Del Rio Barquero, T. Roth, C. Kammerlander, M. Blauth, R. Schubert, and A. F. Frangi. 2011. A statistical model of shape and bone mineral density distribution of the proximal femur for fracture risk assessment. Med. Imag. Comput. Comput. Assist. Interv. 14:393–400. Zhang, B., and C. N. Coon. 1997. The relationship of various tibia bone measurements in hens. Poult. Sci. 76:1698–1701.
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Hester, P. Y., M. A. Schreiweis, J. I. Orban, H. Mazzuco, M. N. Kopka, M. C. Ledur, and D. E. Moody. 2004. Assessing bone mineral density in vivo: dual energy X-ray absorptiometry. Poult. Sci. 83:215–221. Kim, W. K., S. A. Bloomfield, T. Sugiyama, and S. C. Ricke. 2012. Concepts and methods for understanding bone metabolism in laying hens. Worlds Poult. Sci. J. 68:71–82. Korver, D. R., J. L. Saunders-Blades, and K. L. Nadeau. 2004. Assessing bone mineral density in vivo: quantitative computed tomography. Poult. Sci. 83:222–229. Lavin, T. E., B. D. Nielsen, J. N. Zingsheim, C. I. O’Connor-Robison, J. E. Link, G. M. Hill, and J. Shelton. 2013. Effects of phytase supplementation in mature horses fed alfalfa hay and pelleted concentrate diets. J. Anim. Sci. 91:1719–1727. Raudaschl, P., and K. Fritscher. 2017. Statistical shape and appearance models for bone quality assessment. Pages 409–443 in Statistical Shape and Deformation Analysis. G. Zheng, S. Li, and G. Sz´ekely, eds. Academic Press. Regmi, P., A. G. Cox, C. I. Robison, and D. M. Karcher. 2017. Correlation analysis of cortical geometry of tibia and humerus of white leghorns using clinical quantitative computed tomography and microcomputed tomography scans. Poult. Sci. 96:2950–2955. Regmi, P., T. S. Deland, J. P. Steibel, C. I. Robison, R. C. Haut, M. W. Orth, and D. M. Karcher. 2015. Effect of rearing environment on bone growth of pullets. Poult. Sci. 94:502–511.
ABSTRACT Laying hen skeletal health continues to be an industry priority. Bone ash and bone Ca quantification in laying hen long bones provides valuable information on skeletal health. Unfortunately, these measurements can only be accomplished by sacrificing hens, thus making longitudinal measurements on the same hen impossible. Quantitative computed tomography (QCT), used with a calcium hydroxyapatite phantom, has been used to determine bone density of wings and legs as well as live hens throughout the production cycle by scanning with a calcium hydroxyapatite phantom. QCT has also been used to scan live hens throughout the production cycle. The purpose of this study was to determine how QCT calculated bone mineral content (QCT BMC) corresponds to analytical bone Ca and bone ash. Wing and leg quarters from 72-wk-old W-36 hens were QCT scanned along with a QCT Phantom. After scanning, humeri, femurs, and tibias were cleaned, divided into epiphysis (E) and diaphysis (D), fat extracted, ashed, and digested under nitric acid, and Ca was determined by atomic absorption spectroscopy was used to determine E, D, and
whole bone Ca. Four bones/type were used for E and D, while 6 bones/type were used for whole bone measurements. A second set of bones were prepared to determine correlation of BMC to bone ash. QCT scans were analyzed with Mimics software (Materialise NV, Leuven, Belgium) to calculate bone volume and density in Hounsfield units. Utilizing the QCT phantom and bone volume, BMC was calculated for E, D, and whole bone. Data were analyzed with regression analysis and Pearson correlation coefficients were determined. Analytical Ca was correlated to QCT BMC for E (R = 0.84, P < 0.01), D (R = 0.99, P < 0.01), and whole bone (R = 0.97, P < 0.01). Whole bone ash was highly correlated to QCT BMC for femur (N = 47, R = 0.97, P < 0.001), tibia (N = 50, R = 0.94, P < 0.001), and keel (N 50, R = 0.94, P < 0.001). Whole bone ash and QCT BMC values of femur and tibia were not different (P = 0.39 and 0.22 respectively). Based on this information, QCT could provide relative quantitative assessment of total bone mineral in live birds proving useful in long-term studies.
Key words: laying hen, bone mineral, computed tomography, calcium 2019 Poultry Science 0:1–6 http://dx.doi.org/10.3382/ps/pez165
INTRODUCTION
tracking of the same bird. Alternatively, bone mineral content calculated from densitometry can provide similar information as bone ash. Dual-energy x-ray absorptiometry (DEXA), quantitative computed tomography (QCT), and peripheral QCT have been employed to determine bone mineral density and estimate bone mineral content in vivo and ex vivo (Korver et al., 2004; Saunders-Blades et al., 2009; Regmi et al., 2015). Several issues arise with these approaches: (1) long scan time requiring bird sedation, (2) limitation on numbers of animals that can be used for data collection as a result of lengthy scan time, and (3) reproducibility will decrease when an area less than the whole bone is selected (Kim et al., 2012). Another approach that could be used is clinical QCT. Compared to micro-CT or peripheral QCT image quality is lower, but an increase in the number of animals
Calcium demands on laying hens during production have been well documented and any imbalance may result in birds with poor structural bone quality. In 50wk-old layers fed varying levels of Ca, bone breaking force was found to be highly correlated to bone ash weight and bone volume (Zhang and Coon, 1997). The monitoring of bone ash and bone volume could provide valuable insight into the skeletal health of the laying hen throughout production. Unfortunately, bone ash determination requires bird sacrifice, potentially altering cage density and does not allow for longitudinal C 2019 Poultry Science Association Inc. Received November 9, 2018. Accepted March 12, 2019. 1 Corresponding author: [email protected]
1
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
∗
2
ROBISON AND KARCHER
Table 1. Whole bone ash correlated to bone mineral calculated by quantitative computed tomography (QCT BMC) scans when masked at specific Hounsfield units (HU). Paired t-tests examine the difference between bone ash and scanned mineral.
N
Mask (HU)
Femur Tibia Humerus Keel
47 50 23 50
480 600 120 120
1 2
Ash (g)1 2.41 2.93 1.46 2.31
± ± ± ±
0.09 0.09 0.02 0.06
Paired t-test P-value
Pearson correlation Prob |r|
Correlation P-value
± ± ± ±
0.387 0.217 < 0.001 < 0.001
0.974 0.939 0.455 0.944
< 0.001 < 0.001 0.03 < 0.001
2.46 2.84 1.12 2.57
0.15 0.15 0.05 0.1
Data represent the mean ± SEM ash in grams for each type of bone. Data represent the mean ± SEM calculated bone mineral for each type of bone.
scanned, reduction in scan time, and the ability to scan the entire bird may compensate for this shortcoming. Clinical QCT scan quality can accurately create 3D models of bones that can be used for morphometric shape analysis as well as determining cortical thickness, cross-sectional area, volumetric BMD, bone volume, and morphological measurements such as length, width, angle, and circumference (Chargo et al., 2018; Regmi et al., 2015, 2017). Recently researchers were able to scan 16 laying hens per clinical QCT scan without anesthesia providing researchers with whole body data (Chargo et al., 2018). In the current study, the objective was two-fold. First, to determine how analytical bone Ca correlated to bone ash from the epiphysis, diaphysis, and whole bone for the tibia, humerus, and femur. Second, to determine if digital calculation of mineral content from humeri, tibia, femur, and keel bones from mature laying hens correlated to total mineral content measured as fat-free ash.
MATERIALS AND METHODS All procedures were approved by the MSU Animal Care and Use Committee. Bones were collected from ninety 85-wk-old Hy-Line W36 hens. The left wing, left leg quarter, and keel bone were tagged using a numbered plastic tag (Ketchum Manufacturing, Ontario, Canada) and tagging gun. The parts were frozen at −20◦ C until analysis. A power analysis was conducted to determine 11 samples that were needed with expected correlation of 0.85 and 0.95.
Computed Tomography Scans The parts were thawed for 24 h in a chiller at 6◦ C and then taken to the Michigan State University College of Veterinary Medicine for clinical QCT scans. Wings, leg quarters, and keels were carefully arranged on plexiglass in rows such that the parts were lined up end to end perpendicular to the gantry. All parts were scanned with feathers and muscle tissue intact. A solid calcium hydroxyapatite phantom (Image Analysis, Inc; Columbia, KY) consisting of rows representing 0, 75, and 150 mg/cm3 Ca was included in each scan. Each QCT scan used the following settings: 120 kV,
320 mAmp, 0.625 mm slices, and spine position using a GE Revolution Evo scanner (General Electric Healthcare, Princeton, NJ). Each row of bones from the QCT scan was reconned with a 7-cm field of view using Imageworks software (General Electric Healthcare, Princeton, NJ) with a bone algorithm resulting in a DICOM (Digital Imaging and Communications in Medicine) file for each row scanned. The DICOM files were imported into Mimics software (Materialise, Plymouth, MI) for analysis. An appropriate threshold Hounsfield units (HU) was selected to create a mask for the whole bone (Table 1). Then the region growing tool was used to separate each bone into its own mask. The multiple slice edit tool was used to remove the patella from the femur, the fibula from the tibia, and the ribs and other extra bone from the keel bone. Once bones were edited, the average HU and bone volume were recorded. Average HU was determined for each of the 3 concentrations of calcium hydroxyapatite at 10 locations along the length of the phantom for each scan. The HU were averaged for each concentration and a scatter plot was created with mg Ca hydroxyl apatite/cm3 versus HU. The equation of the regression line was used to convert density in HU to mg Ca hydroxyapatite/cm3 . Bone mineral content was calculated from the CT scans by multiplying the density of the bone in mg Ca hydroxyapatite/cm3 by the bone volume.
Analytical Chemistry Bones were prepared for analysis by placing the frozen parts into plastic autoclave tubs and autoclaving at 120◦ C for 19 min using a gravity cycle (733HCMC; Getinge, Wayne, NJ). Upon removal from the autoclave skin, muscle, and tendon were removed from each bone. For the keel bone, ribs and the proximal end was trimmed along the spina externa. Each long bone was measured in mm using 6” digital calipers (68304; Pittsburgh, PA). The calipers were used to mark the distance 25% from the distal and proximal ends of each bone, which was designated as the epiphysis (Epi). The remaining middle 50% of the bone was designated as diaphysis (Dia). Bones were cut, fat extracted, and dried as described in Van Wyhe et al. (2012). Crucibles were ashed in a Thermolyne furnace (30400; Barnstead
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Bone
QCT BMC (g)2
3
DIGITAL BONE MINERAL ANALYSIS
Table 2. Amount of bone mineral calculated by quantitative computed tomography scans (QCT BMC), analytical Ca, and ash quantity from 85-wk-old Hy-Line W36 laying hens. N
QCT BMC (g)2
Ash (g)
Analytical Ca (g)
% Ca in Ash
P-value1
12 4 4 4
0.904 0.949 1.3 0.466
± ± ± ±
0.11 0.03 0.10 0.05
0.872 0.928 1.12 0.569
± ± ± ±
0.07 0.01 0.08 0.02
0.315 0.324 0.412 0.208
± ± ± ±
0.03 0.01 0.03 0.01
36.1 35.0 36.8 36.5
± ± ± ±
0.3 0.3 0.4 0.2
0.42 0.30 0.01 0.058
Epiphysis Femur Tibia Humerus
12 4 4 4
1.19 1.51 1.42 0.64
± ± ± ±
0.14 0.14 0.17 0.08
1.34 1.40 1.70 0.927
± ± ± ±
0.11 0.13 0.11 0.05
0.485 0.498 0.620 0.338
± ± ± ±
0.04 0.04 0.04 0.02
36.2 35.8 36.3 36.5
± ± ± ±
0.3 0.6 0.3 0.3
0.048 0.25 0.02 0.017
Whole bone Femur Tibia Humerus
16 5 6 5
2.01 2.08 2.76 1.06
± ± ± ±
0.20 0.05 0.23 0.08
2.22 2.20 2.88 1.46
± ± ± ±
0.16 0.11 0.14 0.06
0.81 0.782 1.06 0.533
± ± ± ±
0.06 0.04 0.05 0.09
36.3 35.6 36.8 36.4
± ± ± ±
0.3 0.6 0.3 0.1
0.003 0.12 0.36 0.001
1 2
P-value indicates significance of the paired t-test between the QCT BMC and ash. Data is presented as mean ± SEM.
International) at 600◦ C for 10 h and weighed to determine ash percentage. All bones were QCT scanned and ashed; however, only a subset of bones were analyzed for Ca concentration. Samples analyzed for epiphyseal or diaphyseal Ca were randomly selected such that the same bone was not used for both analyses to allow for greater variation in the sample population. In order to determine Ca, ash was ground using a mortar and pestle and then 0.4-g ash was weighed into Teflon-lined digestion vials for microwave digestion as described by Shaw et al. (2002). Samples were allowed to digest overnight in 10 mL of 70% nitric acid (OmniTrace; MilliporeSigma, Temecula, CA). Samples were digested using a microwave accelerated reaction system (MARS-5; CEM Corp., Matthews, NC) as described by Lavin et al. (2013). Calcium concentrations in bone were determined by atomic absorption spectroscopy (AA-7000, Shimadzu, Kyoto, Japan). A peach leaves standard (National Institute of Standards and Technology, Gaitherburg, MD) was digested and analyzed with the samples (atomic absorption standard: ACAIKN-100, Ricca Chemical Co., Arlington, TX).
Statistics Pearson’s correlation analysis was conducted to understand the association between analytical bone ash and QCT bone mineral content (QCT BMC). Pearson’s correlations coefficients (R) between measurements are presented and were calculated using the CORRELATION procedure in SAS 9.3 (SAS/STAT User’s manual, 2014). Pearson correlations were also calculated for each bone by part combination. Linear regression analysis was completed using the REGRESSION procedure in SAS 9.3 to examine the relationship between analytical bone mineral and QCT BMC. Paired t-tests were used to determine amount of difference between the measurements of bone mineral calculated analytically versus digitally using the TTEST procedure in SAS 9.3.
Table 3. Regression equations of calculated mineral of the epiphysis (E), diaphysis (D), and whole bone (W) to bone ash. Part
n
R2
Equation
Humerus
E D W
4 4 23
0.40 0.72 0.21
y = –0.23x +0.94 y = –0.68x + 2.01 y = –0.255x + 0.94
Femur
E D W
4 4 47
0.68 0.86 0.95
y = 0.27x + 0.89 y = –1.08x + 2.19 y = –1.23x + 1.53
Tibia
E D W
4 4 50
0.94 0.91 0.88
y = –1.10x + 1.48 y = –0.023x + 1.18 y = –1.78x + 1.57
Keel
W
50
0.87
y = –0.88x + 1.49
Bone
RESULTS Bone Ca to Bone Ash Average Ca content and QCT BMC are presented for each bone type by section with Pearson correlations and associated P-values (Table 2). Analytical bone Ca from the tibia and femur were highly correlated to QCT BMC regardless of the part (R = 0.73 to 0.96); however, the humerus only reported a moderate-to-high correlation (R = 0.52 to 0.83; Table 3). The aforementioned correlations only included 4 samples for each combination; therefore, in order to more thoroughly examine the relationship, all bone data were combined for an overall correlation by part. The Pearson correlation between diaphyseal Ca (N = 12) and QCT BMC was highly correlated with R = 0.989 (P < 0.01; Figure 1). The Pearson correlation between epiphyseal Ca (N = 12) and QCT BMC was R = 0.873 (P < 0.01; Figure 1B). Additionally, the Pearson correlation between whole bone Ca (N = 16) and QCT BMC was R = 0.974 (P < 0.001; Figure 1C).
Bone Ash to QCT BMC Femur QCT BMC was highly correlated to ash weight with a Pearson correlation coefficient of R = 0.97
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Diaphysis Femur Tibia Humerus
4
ROBISON AND KARCHER
Figure 2. The relationship between whole bone ash and bone mineral content calculated from quantitative computed tomography scans (QCT BMC) for the (A) femur (n = 47; P < 0.01), (B) tibia (n = 50; P < 0.01), and (C) keel (n = 50; P < 0.01).
(n = 47; P < 0.001; Figure 2A). QCT BMC of the tibia was also highly correlated to ash weight with a Pearson correlation coefficient of R = 0.939 (n = 50; P < 0.001; Figure 2B). Keel bone QCT BMC was correlated to ash weight with a Pearson correlation coefficient of R = 0.931 (n = 50; P < 0.001; Figure 2C). The QCT
BMC of the humerus was only moderately correlated to ash weight with a Pearson correlation coefficient of R = 0.455 (n = 23; P = 0.029; Table 1). Paired t-tests were used to determine the amount of difference between the QCT BMC and bone ash for each bone type (Table 1). The average difference for the
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Figure 1. The relationship of analytical bone Ca to bone mineral content calculated from quantitative computed tomography scans (QCT BMC) for the (A) diaphysis (P < 0.01), (B) epiphysis (P < 0.01), and (C) whole bone (P < 0.01). Plots A and B represent a compilation of femur (n = 4), tibia (n = 4), and humeri (n = 4). Plot C represents a compilation of femur (n = 5), tibia (n = 6), and humeri (n = 5).
5
DIGITAL BONE MINERAL ANALYSIS
femur was 0.053 ± 0.06 g (P = 0.387) and the average difference for the tibia was 0.09 ± 0.07 g (P = 0.217). Keel bones had an average difference of –0.15 ± .03 g (P < 0.001). Humerus paired t-test resulted in an average difference of 0.339 ± 0.05 g (n = 23, P < 0.001).
The composition of bone including mineral and collagen content influences the material properties (van der Meulen et al., 2001). Total mineral content is generally estimated using the ash content of the bone. Ash is the inorganic residue remaining after water and other materials are vaporized and organic substances are burned in the presence of oxygen, providing a direct measure of BMC. Bone breaking force was found to be highly correlated to bone ash weight (Zhang and Coon, 1997; Hester et al., 2004) and bone volume (R2 = 0.682) (Zhang and Coon, 1997). DEXA, QCT, and peripheral QCT have been employed to determine bone mineral density and estimate bone mineral content in vivo (Saunders-Blades et al., 2009; Regmi et al., 2015). Baird et al. (2008) compared in vivo BMC of tibias from 58-wk-old laying hens scanned using DXA to bone ash and found DXA underestimated BMC by 9.2% (Baird et al., 2008). In the current study, QCT BMC overestimated the BMC of the keel by 11% and underestimated the BMC of the humerus by 23%. Unlike the humerus and keel, the QCT BMC of femur and tibia were not different from the ash weight when the whole bone was used. In the current study, QCT BMC of the whole bone was highly correlated to ash weight in the femur, tibia, and keel while only moderately correlated in the humerus. There are some challenges with clinical QCT images. For example, the humerus was difficult to threshold effectively since the bone was equal to or less dense than the muscle tissue surrounding it, likely due to its hollow nature. The keel bone thresholds nicely at 120 HU; however, the calculated bone mineral was overestimated suggesting that either the HU threshold should be increased slightly or there was difficulty matching the shape of the digital image to the shape of the actual bone. On the other hand, the femur and tibia are easily imaged and there is minor difficulty in removing the patella and the fibula from the thresholded areas. Once proficiency is achieved with the software, digital data can be collected and calculated faster than bone ash can be conducted. Calcium content of the bone was measured to examine the correlation between Ca content and QCT BMC to determine if digital measurements could be used to estimate Ca content. The femur and tibia were highly correlated to Ca content while the humerus was moderately correlated. The Ca content accounted for approximately 37% of the ash weight regardless of bone. Both quality and quantity of bone are important to overall bone health. Neither whole bone ash nor analytic Ca
REFERENCES Baird, H. T., D. L. Eggett, and S. Fullmer. 2008. Varying ratios of omega-6: omega-3 fatty acids on the pre-and postmortem bone mineral density, bone ash, and bone breaking strength of laying chickens. Poult. Sci. 87:323–328. Chargo, N., C. I. Robison, S. L. Baker, M. J. Toscano, M. M. Makagon, and D. M. Karcher. 2018. Keel bone differences in laying hens housed in enriched colony cages. Poult. Sci. 98:1031–1036. ´ Cs´oka, G. Kov´ ¨ Petneh´ Donk´ o, T., A. Tischler, A. acs, M. Emri, O. azy, A. Szab´ o, V. Halas, J. Tossenberger, R. Garamv¨olgyi, and G. Bajzik. 2018. Estimation of bone mineral density and breaking strength of laying hens based on scans of computed tomography for body composition analysis. Br. Poult. Sci. 59:365–370.
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
DISCUSSION
measurements describe the distribution of bone mineral or the quality of the bone (Korver et al., 2004). In addition to volumetric bone density, QCT can provide anatomical shape, which could be coupled with Finite Element analysis to analyze bone quality and the material properties of bone. Recent work using clinical CT with laying hens has demonstrated high correlations between calculated bone mineral density and maximum breaking strength (R = 0.78 to 0.80) in the tibiotarsal bone (Donk´ o et al., 2018). Clinical QCT allows researchers the ability to scan the entire bird quickly, potentially allowing researchers to take advantage of other digital calculations and analysis not previously available to them. Live birds can be scanned using a restraint device as described in Chargo et al. (2018) with up to 24 live birds on 1 scan. This set-up provides both a high-throughput and economical solution for collecting a considerable amount of skeletal data from live hens without the need for anesthesia. In a human study examining risk factors associated with femur fractures, researchers reported that volumetric BMD and cortical thickness represented the key differences between fractured and non-fractured bones (Raudaschl and Fritscher, 2017). Additionally, in a reanalysis of published data on osteoporosis, measurements of bone mass provided an indication of bone strength and fracture risk in humans with diminished bone mass associated with increased fracture risk regardless of measurement site (Ross et al., 1990). Images from clinical QCT can be used to create accurate 3D representations of the femur, tibia, and keel. Statistical shape models have been used in human medicine for therapy planning and fracture prediction using 3D distribution of BMD and shape (Whitmarsh et al., 2011). A recent publication demonstrated how 3D representations of the keel bone can be used to document changes in keel bone fractures and deviations in live birds during a longitudinal study using 3D part comparison (Chargo et al., 2018). The data support that bone ash can be successfully estimated from digital images acquired by QCT for laying hens. This method would provide researchers with a non-invasive tool to monitor bone mass and BMD throughout production. This approach can provide a quick, reliable method to monitor longitudinal bone health without sacrificing birds.
6
ROBISON AND KARCHER Ross, P. D., J. W. Davis, J. M. Vogel, and R. D. Wasnich. 1990. A critical review of bone mass and the risk of fractures in osteoporosis. Calcif. Tissue Int. 46:149–161. Saunders-Blades, J. L., J. L. MacIsaac, D. R. Korver, and D. M. Anderson. 2009. The effect of calcium source and particle size on the production performance and bone quality of laying hens. Poult. Sci. 88:338–353. Shaw, D. T., D. W. Rozeboom, G. M. Hill, and J. E. Link. 2002. Impact of vitamin and mineral supplement withdrawal and wheat middling inclusion on finishing pig growth performance, fecal mineral concentration, carcass characteristics, and the nutrient content and oxidative stability of pork1. J. Anim. Sci. 80:2920– 2930. van der Meulen, M. C., K. J. Jepsen, and B. Miki´c. 2001. Understanding bone strength: size isn’t everything. Bone 29:101–104. Van Wyhe, R. C., T. J. Applegate, M. S. Lilburn, and D. M. Karcher. 2012. A comparison of long bone development in historical and contemporary ducks. Poult. Sci. 91:2858–2865. Whitmarsh, T., K. D. Fritscher, L. Humbert, L. M. Del Rio Barquero, T. Roth, C. Kammerlander, M. Blauth, R. Schubert, and A. F. Frangi. 2011. A statistical model of shape and bone mineral density distribution of the proximal femur for fracture risk assessment. Med. Imag. Comput. Comput. Assist. Interv. 14:393–400. Zhang, B., and C. N. Coon. 1997. The relationship of various tibia bone measurements in hens. Poult. Sci. 76:1698–1701.
Downloaded from https://academic.oup.com/ps/advance-article-abstract/doi/10.3382/ps/pez165/5438336 by Serials Dept user on 14 April 2019
Hester, P. Y., M. A. Schreiweis, J. I. Orban, H. Mazzuco, M. N. Kopka, M. C. Ledur, and D. E. Moody. 2004. Assessing bone mineral density in vivo: dual energy X-ray absorptiometry. Poult. Sci. 83:215–221. Kim, W. K., S. A. Bloomfield, T. Sugiyama, and S. C. Ricke. 2012. Concepts and methods for understanding bone metabolism in laying hens. Worlds Poult. Sci. J. 68:71–82. Korver, D. R., J. L. Saunders-Blades, and K. L. Nadeau. 2004. Assessing bone mineral density in vivo: quantitative computed tomography. Poult. Sci. 83:222–229. Lavin, T. E., B. D. Nielsen, J. N. Zingsheim, C. I. O’Connor-Robison, J. E. Link, G. M. Hill, and J. Shelton. 2013. Effects of phytase supplementation in mature horses fed alfalfa hay and pelleted concentrate diets. J. Anim. Sci. 91:1719–1727. Raudaschl, P., and K. Fritscher. 2017. Statistical shape and appearance models for bone quality assessment. Pages 409–443 in Statistical Shape and Deformation Analysis. G. Zheng, S. Li, and G. Sz´ekely, eds. Academic Press. Regmi, P., A. G. Cox, C. I. Robison, and D. M. Karcher. 2017. Correlation analysis of cortical geometry of tibia and humerus of white leghorns using clinical quantitative computed tomography and microcomputed tomography scans. Poult. Sci. 96:2950–2955. Regmi, P., T. S. Deland, J. P. Steibel, C. I. Robison, R. C. Haut, M. W. Orth, and D. M. Karcher. 2015. Effect of rearing environment on bone growth of pullets. Poult. Sci. 94:502–511.