Automatic-interactive measurement of resorption cavities in transiliac bone biopsies and correlation with deoxypyridinoline

Bone Vol. 17, No. 2

August 1995:153-156

ELSEVIER

Automatic-Interactive Measurement of Resorption Cavities in Transiliac Bone Biopsies and Correlation With Deoxypyridinoline J. P. R O U X , M. E. A R L O T , E. G I N E Y T S , P. J. M E U N I E R , and P. D. D E L M A S INSERM Research Unit 403, Lyon, France

contrast, histological assessment of bone resorption suffers from the lack of such a dynamic marker. For many years the resorption cavities have been studied in terms of their relative and absolute surface extent. ~'7`l°'tl'~4,ls Such eroded surfaces can be also classified as osteoclast positive or osteoclast negative. ~3 The number of osteoclast and/or of tartrate resistant acid phosphatase positive cells, expressed per unit of cancellous surfaces, is another index of bone resorption. 5 Erosion depth was first describe by Eriksen et al. 4 by counting eroded lamellae beneath the trabecular surface in resorption cavities with reference to the type of cells present in the cavity. A computerized method for estimation of erosion depth and eroded volume using a cubic spline reconstruction has also been reported. 8 Another approach is to measure erosion depth only on lacunae covered by osteoid, 2 in order to ensure that measurement is restricted to cavities in which resorption has been completed. Many of these techniques are time consuming and difficult to standardize, except for the cubic spline method. 8 Resorption can be estimated by the measurement of urinary deoxypyridinoline, a nonreducible pyridinium crosslink present in the mature form of the collagen that has only been found in significant amounts in the type I collagen of bone. It is excreted in urine in peptide bound and free forms and their total amounts can be measured by HPLC after acid hydrolyses. 6 Because of the need for a rapid and reliable method to assess the various aspects of bone resorption, we have developed an automatic and interactive method using an automatic color analyzer. Measurement of erosion depth was compared to the reference technique4 and the various parameters of bone resorption were correlated with urinary deoxypyridinoline, the most sensitive marker of bone resorption.

Measuring bone resorption accurately by histomorphometry of bone biopsies is a challenge. Several techniques have been proposed including the measurement of eroded surfaces and resorption depth, but they have not been compared between themselves nor with biochemical assessment of bone resorption. In addition, there is a need for a rapid method that could be used more routinely. We describe here an automatic interactive method using a color analyzer (Visiolab ®, BIOCOM ®, France) with a specific software for the evaluation of erosion depth, eroded volume, eroded surface, osteoclast number, and surface. Thirty transiliac undecalcified bone biopsies stained with Goidner's trichrome were used in this study, taken from subjects suffering from osteoporosis or primary hyperparathyroidism. At the time of the biopsy a 2 h fasting morning urine sample was collected for measurement by HPLC of total deoxypyridinoline, the most sensitive marker of bone resorption. There was a highly significant correlation between maximum erosion depth measured directly and the one calculated according to the count of eroded lamellae (E. F. Eriksen, et al. Metab Bone Dis Relat Res 5:243-252; 1984) (r -- 0.76; p = 0.0001). A significant correlation was found between urinary deoxypyridinoline and eroded volume/bone volume in cancellous and endocortical bone measured with the automatic interactive technique (r = 0.48; p = 0.007). [n contrast, other histological indexes of bone resorption did not correlate with urinary deoxypyridinoline. The volume of resorption cavities appears to be the most valid index of bone resorption rate as it was correlated with the urinary excretion of total deoxypyridinoline. This histological parameter of bone resorption can be measured with a convenient automated method using an image analyzer. (Bone 17:153-156; 1995)

Material and Methods

Bone Histomorphometry

Key Words: Bone; Resorption; Histomorphometry; Deoxypyridinoline; Crosslinks.

Transiliac bone biopsies were obtained in 30 adults (11 males, 19 females) referred for osteoporosis (N = 26) or primary hyperparathyroidism (N = 4). They were obtained with a 7.5 mm internal diameter trephine, not decalcified, then embedded in methylmethacrylate and sectioned with a Polycut ® microtome (Reichert-Leica) equipped with a carbide tungsten knife. Three Goldner's trichrome stained sections, 12 7 Ixm thick, were obtained at 200 ~,m intervals. Goldner's trichrome was used because bone and osteoid show a good contrast, the lamellar structure is visible in polarized light, and osteoclasts are easy to identify due to their red color and dark nuclei. Resorption parameters were measured by two different methods:

Introduction Measurement of bone formation rate can be performed with sensitivity and accuracy on iliac crest biopsy because the double tetracycline labeling of the mineralizing front allows a dynamic assessment of the anabolic activity of bone forming cells. In Address for correspondence and reprint requests: J. P. Roux, INSERM Unit 403, Facult6 A. Carrel, rue G. Paradin, 69372 Lyon Cedex 08,

France. © 1995 by Elsevier Science Inc.

153

8756-3282/95/$9.50 SSDI 8756-3282(95)00174-C

154

J.-P. Roux et al. Resorption cavities and deoxypyridinoline

(i) An automatic color analyzer (Visiolab®, BIOCOM ®, France) with a specific software for bone histomorphometry written in collaboration with the f'u'm was used. A 3 CCD color camera captured the image of all identifiable resorption lacunae at every stage of the resorptive phase of remodeling under polarized light (x230 or x460) to obtain a digital image that was recorded in the computer. Then, the image was displayed on the high definition video monitor. The method was similar to that of Garrahan et al. s with differences between the software used. The computer automatically drew the trabecular surface (automatic segmentation according to a recorded threshold for Goldner stained bone). The operator rebuilt the resorption cavity by drawing a line with the help of the nonresorbed lamellae of the lacuna. The analyzer then measured automatically the eroded length (length of interface between bone and lacuna in p,m), the mean erosion depth (measured automatically with eight equidistant segments), the maximum erosion depth (the longest segment), and the eroded volume (Figure 1). Osteoclast number and surfaces are also measured interactively according to their morphological identification. After automatic measurement of cancellous bone volume and cancellous bone surface, the results were expressed as: eroded volume/bone volume (%); eroded volume/ tissue volume (%); eroded surface/bone surface (%), osteoclast number/tissue volume (/mm2), osteoclast number/bone surface (/nun), osteoclast surface/bone surface (%). (ii) The number of eroded lamellae beneath the trabecular surface was counted in every resorption cavity according to Eriksen4 without cells classification. On each biopsy, the mean lamellar width was measured in completed walls using the orthogonal intercept method on at least 100 sites per biopsy. A magnification of ×460 or x920 under polarized light was used. Lamellar erosion depth (txm) was the number of eroded lamellae multiplied by mean lamellar thickness. The time required to select the resorption sites and to measure all resorption parameters for one biopsy ranged from 10 to 20 min with the direct method and from 20 to 30 min with the indirect method, including measurement of lamellar thickness and count.

Calcified bone I

I

Eroded volume measured automatically Eroded length measured automatically (p.m)

__~

Erosion depth (8 equidistant segments measured automatically)

Bone Vol. 17, No. 2 August 1995:153-156 All measurements were performed both in the cancellous and the endocortical envelopes. All width measurements (mean lamellar width, maximum and mean erosion depth) were multiplied by ~r/4 to correct for obliquity of the plane section and expressed in three dimensions.

Biochemistry At the time of the biopsy, a 2 h fasting morning urine sample was collected. Deoxypyridinoline was measured. The values were corrected for creatinine excretion according to a previously published method. 17 Briefly, the crosslinks were extracted from the hydrolyzed urine sample by cellulose chromatography, separated by reversed-phase HPLC, and identified by spectrofluorometry. The area of the fluorescent peak was quantified by comparison with calibrated deoxypyridinoline external standards purified from human cortical bone. All histomorphometric and biochemical measurements were performed in a blind fashion, and the code was broken after completion of the study.

Statistical Methods After analysis of distribution, a logarithmic transformation was performed for deoxypyridinoline. The following tests were used: paired t-test and linear regressions. Because of the number of statistical tests performed, the Bonferonni procedure was used to set the threshold of significance at a level of p = 0.02.

Results For resorption parameters in the 30 biopsies, mean values, standard deviation, and ranges are listed in Table 1. There was a highly significant correlation between maximum erosion depth measured directly and the one calculated according to the count of eroded lamellae (r = 0.76; p = 0.0001) (Figure 2). However, the direct measurement gave significantly lower values than the count of lamellae (respectively, 18.9 --- 5.1 ~m and 25.0 --- 6.2 p.m; p = 0.0001). The lamellar thickness was 2.5 -+ 0.1 txm (range 2.0-2.7). 1275 cavities were found in the 30 biopsies. 1235 (97%) could be measured using the automatic method and only 827 (65%) using the count of lamellae. The mean value of deoxypyridinoline was 13.18 --- 10.87 nmol/mmol creat (range: 3.0--47.4). Because values were not normally distributed, they were log transformed before correlation with histomorphometric parameters of bone resorption. Coefficient of correlation between histomorphometric resorption parameters and deoxypyridinoline are listed in Table 1. As expected, the correlation was somewhat weaker after log transformation of deoxypyridinoline. A significant correlation was found between deoxypyridinoline and eroded volume expressed per bone volume (r = 0.48; p = 0.007) (Figure 3) and per tissue volume (r = 0.43; p = 0.018). Borderline correlations were found between deoxypyridinoline and mean erosion depth, maximum erosion depth, and eroded volume/tissue volume in cancellous + endocortical bone (Table 1). Correlation between deoxypyridinoline and osteoclast measurements were weaker with r values ranging from 0.22 to 0.31 after log transformation of deoxypyridinoline.

Max. depth (p.m) : the longest of the 8 segments. Mean depth (#m) : mean of the 8 segments,

Figure 1. Schematicrepresentationof the measured parameters of bone resorption cavity. The two arrows show the limit of the reconstruction line (--) interactivelydrawn by the operator.

Discussion Two findings emerged from this study: The maximum erosion depth measured directly correlated with erosion depth calculated according to the count of lamellae; and eroded volume/bone or

Bone Vol. 17, No. 2 August 1995:153-156

J.-P. Roux et al. Resorption cavities and deoxypyridinoline

155

Table 1. Cancellous + endocortical bone. Values of resorption parameters and correlation with deoxypyridinoline in 30 patients. Direct measurement with an automatic analyzer and indirect measurement by count of lamellae. Correlation coefficients and p values are shown before (r, p) and after (r', p') log transformation of deoxypyridinoline

Mean ± 1 SD Direct measurement Mean erosion depth (~,m) Maximum erosion depth (Ixm) Eroded volume/tissue volume (%) Eroded volume/bone volume (%) Eroded surface/bone surface (%) Osteoclast number/tissue volume (/mm2) Osteoclast number/bone surface (/mm) Osteoclast surface/bone surface (%) Indirect measurement Lamellar erosion depth (p,m)

12.2 18.9 0.22 1.95 5.01 0.13 0.05 0.17

± 3.0 - 5.1 ± 0.17 ± 1.86 ± 2.49 -± 0.10 ± 0.04 ± 0.13

25.0 ± 6.2

tissue volume but not other histomorphometric indices of bone resorption correlated significantly with deoxypyridinoline. For cancellous bone, our values of lamellar thickness are similar to those previously reported, ranging from 1.9 to 3.2 p,m. 2"4"9 The maximum erosion depth was correlated with the lamellar erosion depth derived from lamellae counting but with slightly lower value for the first method. The interactive method of the reconstruction line used to measure the maximum and mean erosion depth could underestimate the original surface as reported by Sod et al. 16 Conversely, when counting lamellae, a potential source of error is the assumption that the eroded lamellae are parallel to the surface, since counting lamellae lying obliquely to the surface would overestimate erosion depth. 2 Another source of error is the variability of lamellar thickness inside an osteon. Indeed, lamellar thickness at the middle of an osteon is higher than at its end because lamellae are joining together at each of its ends; thus, four eroded lamellae at the center of an osteon represent a significantly deeper erosion depth than four eroded lamellae at its end. With the direct method, the main problem for reconstruction of the line is represented by resorption cavities occurring at the end of trabeculae without curve lamellae; despite this limitation, we have been able to measure a

r = 0.76 p = 0.0001

35

Range

r

8.4-21.5 12.1-35.3 0.02-0.72 0.13-9.63 0.58-10.85 0.(K1-0.35 0.00-0.14 0.00-0.46

0.36 0.41 0.51 0.63 0.48 0.38 0.38 0.28

0.050 0.025 0.004 0.0002 0.007 0.038 0.036 0.130

0.37 0.41 0.43 0.48 0.34 0.31 0.31 0.22

0.047 0.025 0.018 0.007 0.068 0.092 0.097 0.236

11.1-40.2

0.26

0.171

0.25

0.182

p'

higher number of lacunae (97%) using the direct method instead of 65% using the count of lamellae, because visibility of eroded lamellae was sometimes poor and the counting impossible. When more than one osteon were resorbed with different orientations between them, the count of lamellae was difficult or impossible, a restriction that might introduce a bias in the estimate of erosion depth. Deoxypyridinoline is the most sensitive marker of bone resorption. 17.,8 A somewhat low but significant correlation (r -0.45; p = 0.004) has been previously reported between osteoclast surfaces and deoxypyridinoline but not urinary hydroxyproline, a Deoxypyridinoline was measured on fasting urine specimens, rather than on a 24 h urine collection, because we have previously shown that the fasting expression of deoxypyridinoline, but not their 24 h excretion, was significantly correlated with the subsequent rate of bone loss assessed by repeated measurements of the radial bone mineral content in 37 postmenopausal women. ]8 Cortical bone represents 80% of the skeleton mass with a lower remodeling than cancellous bone. Deoxypyridinoline is likely to reflect resorption of the total skeleton in10

9

•

Correlation with deoxypyridinoline p r'

r = 0.48

•

p=0.007

•

3o ~ 6 •e 20

•

•

5

•Q

ID

15

-i

lO

N

5

~

0 0

5

10

15

20

25

30

35

40

Lamellar Erosion Depth (~m) Figure 2. Correlation between the maximum erosion depth measured directly with an automatic analyzer and the erosion depth calculated according to the number of eroded lamellae (Ref. 4). The equation of the correlation is maximum erosion depth = 0.62 lamellar erosion depth + 3.36.

i*o

20

3b 4'0 5'0

Deoxypyridinoline (nmol/mmol creat) Figure 3. Correlation between eroded volume/bone volume measured in cancellous + endocortical bone in 30 patients and deoxypyridinoline measured in fasting urine specimens. Because of non-normal distribution, deoxypyridinoline values were log transformed. The equation of the correlation is eroded volume/bone volume = 2.91 log deoxypyridinoline -0.99. After exclusion of the highest deoxypyridinoline value, the correlation was still significant (r = 0.43; p = 0.02).

156

J.-P. Roux et al. Resorption cavities and deoxypyridinoline

cluding cancellous, endocortical, and cortical envelopes. Conversely, the histomorphometric assessment of bone resorption is performed on a very limited amount of iliac crest tissue; which may not reflect resorption through the skeleton and is restricted to the cancellous and endocortical envelopes. These differences between the two techniques probably explain why we could not achieve higher r values when correlating the histomorphometric and biochemical assessment of bone resorption. Among the various histomorphometric parameters, the eroded volume expressed as a percentage of bone volume or tissue volume provided the best correlations with urinary deoxypyridinoline (r = 0.48; p = 0.007 and r = 0.43; p = 0.018). Such a finding could be expected because the eroded volume is an index of the fraction of bone that is resorbed. The mean and maximum erosion depth were also correlated with urinary deoxypyridinoline (r = 0.37 and 0.41, respectively), while the estimate of resorption depth by lamellar counting was not. This discrepancy could be related to the previously mentioned limitation of lamellar counting, and taken together, our data indicate that erosion depth should be measured directly rather than indirectly by lamellar counting. A significant correlation (r = 0.45; p = 0.004) has been previously reported between osteoclast surfaces and deoxypyridinoline.3 In the present study, the correlation with osteoclast surfaces and number were slightly weaker, with r values ranging from 0.28 to 0.38 before log transformation of deoxypyridinoline. It should be noted, however, that tartrate resistant acid phosphatase, a sensitive marker of osteoclasts, was not used in these two studies. In conclusion, we have described a convenient computerized method to measure bone resorption that is both automatic and interactive. It allows a direct measurement of resorption depth that appears to be more efficient than lamellar counting. Measurement of erosion volume might be the best index of the rate of bone resorption as it is correlated with the urinary excretion of deoxypyridinoline. References 1. Bordier, P., Matrajt, H., Miravet, L., and Hioco, F. Mesure histologique de la masse et de la rtsorption des travtes osseuses. Pathol Biol 12:1238-1243; 1964. 2. Cohen-Solal, M. E., Shih, M. S., Lundy, M. W., and Parfin, A. E. A new method for measuring cancellous bone erosion depth: Application to the cellular mechanisms of bone loss in postmenopausal osteoporosis. J Bone Miner Res 6:1331-1337; 1991. 3. Delmas, P. D., Shlemmer, A., Gineyts, E., Riis, B., and Christiansen, C. Urinary excretion of pyridinoline crosslinks correlates with bone turnover measured on iliac crest biopsy in patients with vertebral osteoporosis. J Bone Miner Res 6:639-644; 1991.

Bone Voi. 17, No. 2 August 1995:153-156 4. Eriksen, E. F., Gundersen, H. J. G., Melsen, F., and Mosekilde, L. Reconstruction of the formative si~ in iliac trabecular bone in 20 normal individuals employing a kinetic model for matrix and mineral apposition. Metab Bone Dis Relat Res 5:243-252; 1984. 5. Evans, R. A., Dunstan, C. R., and Baylink, D. J. Histological identification of osteoclasts in undecalcified sections of human bone. Miner Electrolyte Metab 2:179-185; 1979. 6. Eyre, D. R., Koob, T. J., and Van Ness, K. P. Quantification of hydroxypyridinium crosslinks in collagen by high-performance liquid chromatography. Anal Biochem 137:380-388; 1984. 7. Frost, H. M., and Vinanueva, A. R. Human osteoclastic activity: Qualitative histological measurement. Henry Ford Hosp Med Bull 10:229-237; 1962. 8. Garrahan, N. J., Croucher, P. I., and Compston, J. E. A computerised technique for the quantitative assessment of resorption cavities in trabecular bone. Bone 11:241-245; 1990. 9. Mal'otti, G. A new theory of bone lamellation. Calcif Tissue Int 53(suppl 1):$47-$56; 1993. 10. Melsen, F., Melsen, B., Mosekilde, L., and Bergman, S. Histomorphometric analysis of normal bone from the iliac crest. Acta Pathol Microbiol Scand 86:70-81; 1978. 11. Meunier, P., Edouard, C., and Courpron, P. Morphometric analysis of trabecular resorption surfaces in normal iliac bone. Jaworski, Z. F. G., ed. Bone Histomorphometry. Ottawa, Canada: University of Ottawa Press; 1976; 156160. 12. Meunier, P. J. Histomorphometry of the skeleton. Peck, W. A., ed. Bone and Mineral Research. Annual 1. A yearly survey of developments in the field of bone and mineral metabolism. Amsterdam: Excerpta Medica; 1983; 191-222. 13. Parfitt, A. M., Drezner, M. K., Glorieux, F. H., Kanis, J. A., Malluche, H., Meunier, P. J., Ott, S. M., and Recker, R. R. Bone histomorphometry: Standardization of nomenclature, symbols, and units. J Bone Miner Res 2:595610; 1987. 14. Schenk, R. K., Merz, W. A., and Muller, J. A quantitative histological study on bone resorption in human cancellous bone. Acta Anat 74:A,A, A,8; 1969. 15. Schultz, A. and Delling, G. Age-related changes of bone resorption parameters in iliac crest trabecular bone. Jaworski, Z. F. G., Ed. Bone Histomorphometry. Ottawa, Canada: University of Ottawa Press; 1976; 161-162. 16. SOd, E. W., Juhlin, K. D., Li, J., and Lundy, M. W. Comparison of bone resorption depth measurement methods that are based on intact surface estimation. J Bone Miner Res 9(suppl. 1):$247; 1994. 17. Uebelhart, D., Gineyts, E., Chapuy, M. C., and Delmas, P. D. Urinary excretion of pyridium crosslinks: A new marker of bone resorption in metabolic bone disease. Bone Miner 8:87-96; 1990. 18. Uebelhart, D., Schlemmer, A., Johansen, J. S., Gineyts, E., Christiansen, C., and Delmas, P. D. Effect of menopause and hormone replacement therapy on the urinary excretion of pyridium cross-links. J Clin Endocrinol Metab 72: 367-373; 1991.

Date Received: January 9, 1994 Date Revised: April 18, 1994 Date Accepted: April 21, 1995