Bone and Mmeral, ll(l990)
335-345
335
Efsevier BAM 00323
‘Rheumatology and RehabilitationResearch Unit, 36 Clarendon Road, Leeds LS2 9NZ, Enginnd and 2GeneratClinicalResearch Centre, UniversityHosplralIV59?,926 W. Michigan Street, Indianapolis,IN 46223, USA
(Received 12 January 1990) (Accepted 27 July 1990)
Summary The architecture of cancellous bone as well as quantity and quality is considered important in maintaining mechanical integrity. To determine whether abnormalities of architecture occur in femoral neck fracture we measured trabecular width and number in iliac crest bone biopsies of 68 women with femoral fracture and compared them with data from a postmortem series of age-matched women without known bone disease. Cortical thickness was measured in 27 of the fracture patients and 17 controls. After exclusion of ten biopsies with increased osteoid surface no significant difference was seen in mean trabecular width or number between fracture patients and subjects without fracture. Both thinning of trabeculae and Lossof trabeculae contributed to low bone volume in femoral fracture patients. Direct measurements of trabecular number correlated with calculated mean trabccu!:!srplate density (P < O.UOl),and the percentage of trabeculae at any one of a range of trabecular widths in the fracture biopsies was similar to that in non-fracture subjects. No difference was seen in the architecture of cancellous bone with age, fracture trauma or between subcapital and intertructtiarrtericfracture in fracture subjects. Cortical thickness, however, was related to age in both fracture and control subjects (P < 0.05). There was no difference in cortical thickness between age-matched fracture and control subjects. Patients with intertrochanteric fracture had lower cortical thickness (P < 0.02) and were older (P < 0.01) than patients with subcapital fracture.
Key words: Femoral fracture; Iliac crest biopsy; Trabecular width; Trabecular number; Cortical thickness; Osteoporosis
Femoral neck fracture is an important cause of morbidity and mortality in the elderly [1,2]. Its pathogenesis is disputed, but is probably multifactorial, depending on Correspondence to: Dr L.,n . Iiordon, Rheumatology and Rehabilitation Research Unit, 36 Clarendon Road, Leeds LS2 9NZ, England. 0169-60#/90/%03.50@I990 Elsevier Science Publishers B.V, (Biomedical Division)
336 the fall, the age-related impairment of protective responses to falling [3], and the quality and quantity of the bone in the femoral neck when the fall occurs. In many studies [4-81 patients with femoral neck fracture as a group have lower bone mass compared with age-matched subjects who have not fractured. However there is substantial overlap in measurements between fracture and non-fracture subjects, and some fracture subjects have more bone than expected for their age [7,9] suggesting that factors other than bone mass also play an important role in fracture. Less attention has been paid to the architecture of cancellous bone in femoral fracture patients using histomorphometric techniques although architecture is probably as important as bone mass in maintaining the integrity of the skeleton [lo]. One study of the femoral heads from patients with femoral fracture showed thinner trabeculae than in patients with osteoarthritis [ 111.Calculation of mean trabecular plate thickness (MTPT) and mean trabecular plate density (MTPD) can be performed using measurements of total perimeter and bone area in thin sections from iliac crest biopsies [12]. One study using this method showed a reduction in MTPD in vertebral fracture patients and a reduction in MTPT in femoral fracture patients [ 121.However, this was not confirmed by another study [ 131which showed reduced MTPT in intertrochanteric fractures only whilst cervical fractures showed higher bone volume and greater trabecular surface density compared with intertrochanteric fractures. However in these three studies [ll-131 controls were not well matched for age and in one study [ 131were not matched for sex. Whilst histomorphometric studies have shown a reduction in cortical thickness in transiliac biopsies in femoral fracture patients compared with younger controls [13,14] it is not clear whether cortical thickness is reduced compared with age- and sex-matched subjects. The aims of this study were to examine the architecture of cancellous and cortical bone in femoral neck fracture patients and to compare this with data from a postmortem series of age- and sex-matched subjects without fracture, to determine whether trabecular loss and/or trabecular thinning were responsible for low bone volume in femoral neck fracture and to examine whether abnormalities in architecture explained why some patients with normal bone volume for their age sustain femoral neck fracture. We also wished to determine whether differences in architecture occurred between subjects with subcapital and intertrochanteric fractures, and between subjects with traumatic and non-traumatic fractures.
Materials and Methods Patients Sixty-eight women aged 52-99 years (mean 78) admitted to Leeds General Infirmary with femoral neck fracture from March 1984 to February 1986 were studied. A11female patients admitted under one of four orthopaedic teams sharing acute orthopaedic admissions were considered for study. During this period a total of 329
337 women were admitted with femoral neck fracture and 229 of these patients received an operation. Our sample therefore represents 21% of the fracture population and 30% of those operated on for this condition. Patients considered unfit for surgery because of severe dementia or physical illness were not biopsied for ethical reasons. A total of 72 women were biopsied, but four samples were inadequate for analysis and these four patients were excluded from the study. Patients were classified radiologically into two groups, those with subcapital and transcervical fractures, and those with basal cervical and intertrochanteric fractures. Fractures resulting from a fall from standing height or less were further classified as ‘non-traumatic’ and fractures resulting from falls greater than standing height, e.g., fall down steps, minor road accidents, as ‘traumatic’. Patients with subtrochanteric fractures, and those known to have malignant disease were excluded. Informed consent was obtained from all subjects. Published data on mean trabecular width and number/field from a postmortem series of age-matched women who died suddenly and who had no medical condition or drug therapy predisposing to bone disease were used for comparison [15]. In addition cortical thickness was measured in transiliac biopsies from 17 w0mc.n aged 55-81 (mean 66 years) comprising I3 subjects from the postmortem series ;-nd four women undergoing minor surgery who had no medical condition or drug therapy predisposing to bone disease. Methods Transilial bone biopsies were obtained from a site 1 inch below and 1 inch behind the anterior superior iliac spine at the time of operation using an 8 mm Bordier trephine. The bone biopsies were fixed in 10% phosphate-buffered formalin for 48 h, dehydrated through graded alcohois, cleared in xylene, embedded in methylacrylate and left to polymerise until hard enough for sectioning. Between 30 and 48 longitudinal sections 10 pm thick were cut from each biopsy using an L.K.B. 2258 PMV heavy duty microtome. Sections were stained with Goldner to demonstrate osteoid surfaces. Bone v&me (BViTV) The corticomedullary junction was defined as the endocortical surface. Measurements on cancellous bone were made on the entire cancellous area. In a small minority of biopsies a transitional zone intermediate between cortical and cancellous bone was present and in accordance with our laboratory practice was excluded from cancellous measurements. The amount of cancellous bone per unit area was determined on 64 fields, four fields on each of 16 sections. Each field was examined using a 25-point Zeiss integrating eyepiece and a magnification of x50 and a point on the grid coinciding with a trabecula was scored as a hit. The number of hits in the four fields of each section was taken as a measure of the percentages of bone in that section. The nizan value of the 16 sections was expressed as a percentage volume of bony tissue, assuming the biopsy to be homogeneous.
338 Trabecular width and trabecularnumber/field These were measured directly by ahe methods described in detail by Aaron et al. [ 151. Trabecular number/field was measured on an area of section outlined by the rectangular field marker encompassing the focusing reticule in the viewing tube of Zeiss Photomicroscope II. This area had been previously selected [IS] such that the length of the field was equivalent to the shortest intercortical distance found among biopsy specimens from the postmortem series and measured 5.4 X 3.6 mm at a specimen magnification of x20. The number of individual trabeculae, defined as domains of bone extending between junctions, in this area was counted in duplicate in 10 sections from each sample. Biopsies in which one cortex was absent or damaged were included providing the area of intact trabecular bone continuous with the other cortex was large enough for the assessment of trabecular number per field. The width of trabeculae was determined using a Vickers Patholette microscope fitted with a projection illumination base and an overhead viewing screen. Attached to the screen was a clear perspex disc bearing the pattern of the Zeiss no. 1 integrating grid. At a magnification of x50, trabecular thickness was measured at the points where the grid intersected the trabecular bars, excluding junctions where several trabeculae fused and the junction of trabeculae with cortex. Non-overlapping fields were selected from throughout the biopsy specimen but areas of damage or artefact, particularly at the margins of the biopsy, were avoided. To check interobserver error and to assess the validity of comparison of the values derived from the femoral neck fracture biopsies with normal values derived from the postmortem study [15], 10 samples from the latter study were analysed. The precision was found to be well within the 5% considered to be acceptable variation for the technique. As well as a mean value for trabecular width for each biopsy the percentage of trabeculae in each biopsy at each of a range of trabecular widths was calculated. When data from all biopsies were combined this enabled a distribution curve of trabecular widths to be plotted and compared with that obtained from the previous study. In addition the mean trabecular plate density (MTPD) was calculated. It was derived from the ratio of per cent bone volume (TBV%) and trabecular width according to the equation: mean trabecular plate density (/mm) = TBV%/trabecular width (lim) x 10 [12]. Osteoidsurface (OS/.BS) This was assessed using the methods described by Aaron et al. [15] and the normal range for the percentage of trabecular surface covered in osteoid was defined as ~24% [7,16]. Osteoidthickness(0. th) This was assessed using the methods described by Aaron et al. [ 151and the normal range for O.th defined as 60% [ 181.
339 Cortical thickness (Ct. th) Only biopsies in which both cortices were intact were measured. Cortical thickness was measured on each cortex at four equidistant sites at a magnification of x50 using a calibrated eye-piece graticule, and a mean value for two cortices obtained. Cortical thickness in fracture and control subjects was measured by the same observer. Statistics Differences between groups were assessed using Student’s t-test. The correlation of two variables was assessed by linear regression analysis.
Trabecular architecture 01 the 68 biopsi.es, 10 showed increased osteoid surface and as in the postmortem series, were excluded from the analysis. Seven of these 10 biopsies fulfilled the criS ~60%). Values for teria for osteomalacia (OS/SS >24%, O.th >13ym and the excluded biopsies are given in Table 1. In the remaining 58 patients the mean trabecular width was 126.9 + 26.7 pm (SD) compared with 132.5 + 25.9 pm in 20 Table 1 Bone volume, mean trabecular width and trabecular number/field in women with femoral neck fracture and increased osteoid surface (n = 10) (results expressed as means fi SD) Bone volume W)
Mean trabecular width fpm)
Mean trabecular number/field
18.7 z!z5.7
170.6 + 58.2
19.6 AZ7.0
femoral
o--o
fracture
controls
Fig. 1. Distribution curves of trabecular width in women with femoral fracture and in normal elderly women.
340 .
73
.
femoral
0
normal
lraclure elderly
a .l
8
.
z 0
.
40
*
*
t
e224
10
90
110
130
170
150 MEAN
190
TRAEECULAR
WIDTH
um
Fig. 2. The relationship between mean trabecular number/field and mean trabecularwidth in women with femoral fracture(r = 0.006, NS) and in normal elderlywomen (r = -0.50, P < 0.05). normal women aged 56-94 (mean 75) years [15]. This difference was not signifi-
cant. The mean trabecular number/field for the 58 femoral fracture patients was 25.6 f 10.6 compared with 24.2 + 6.3 for age-matched normal women. This difference was not significant. In comparison with normal eldery women [ 151, the distribution curve of trabecular width in femoral fracture patients (Fig. 1) showed fewer trabeculae greater than 180pm in width and more in the region of 80-120pm. There was no relationship between mean trabecular number/field and mjean trabecular width (Fig. 2) in the fracture patients (r = 0.006, NS) although in the nor-
t: 5
.8@ 10
e. 15
BONE
“O~i!lME
%
25
30
35
Fig. 3. Therelationship between mean trabecularnumber/fieldand bone volume in women with femoral fracture(r = 0.42, P < 0.01).
341
5
160.
1 z 2
160.
a a -I 3 140. :: < I z l-20. a ii 100. go-
,
5
15
BONE
W?“ME
9b
25
30
35
Fig. 4. The relationship between mean trabecular width and bone volume in women with femoral fracture (r = 0.61, P c 0.001).
ma1 elderly women there was an inverse relationship between mean trabecular number/field and mean trabecular width (r = -0.50, P < 0.05). However there was a small group of eight femoral fracture patients with higher numbers of trabeculae than expected (mean ir 2 SD for normal women, 12.3-36.1). All had a mean trabecular width between 90 and 110pm and bone volume in all but one was .n.Lovethe mean of 16% for age-matched normal women from the postmortem series [IS]. There was a significant correlation between mean trabecular number/field and mean trabecular plate density (r = 0.76, P < 0.001) in the fracture patients. In the fracture patients, bone volume correlated with trabecular number/field (Fig. 3) (r = 0.42, P < 0.01) and trabecular width (r = 0.61, P < 0.001) (Fig. 4). No significant relationship of mean trabecular width, number/field or bone volume to age was seen in the fracture patients (r = -0.05, r = -0.01 and r = -0.14, respectively. There was no significant difference in mean trabecular width, number/field and btine volume between subcapital and intertrochanteric fractures (Table 2). However patients with intertrochanteric fracture were significantly older than those with subcapital fracture (P < 0.001). Table 2
Comparison of age, mean trabecular width, number/field and bone volume in subcapital and intertrochanteric fracture (results are expressed as means z!zSD)
Age (years) Bone volume (%) Trabecular width @m) Trabecular number/field a Difference between groups, P < 0.001.
Subcapital (n = 34)
Intertrochanteric (n = 24)
73.6 i: 15.8 k 128.1 f 26.0 f
81.8 + 15.4 + 124.4 + 25.0 +
9.3 5.8 27.4 9.3
5.9” 6.4 24.5 12.2
342 Table 3 Age, mean trabecular width, number/field and bone volume in women with femoral neck fracture in relation to fracture trauma (results are expressed as means + SD)
Age(yea4 Bone volume (%) Trabecularwidth @m) Trabecularnumber/field
Traumatic (n = 12)
Non-traumatic (n = 34)
74.2 f 16.7 f 131.8 f 23.8 +
79.1 + 15.4 + 125.8 f 25.9 f
4.8 4.8 23.2 8.3
9.3 7.0 29.7 11.7
In 46 patients the cause of fracture was sufficiently well documented to classify as traumatic (fall down steps (n = 9), knocked down by car (n = 2), fall from car roof (n = 1)) or non-traumatic (n = 34). Two traumatic fractures were alcohol-related. No significant difference was seen in age, mean trabecular width, number/field or bone volume between these two groups of patients (Table 3). Cortical thickness There was a significant relationship between age and cortical thickness in the 27 femoral neck fracture patients in whom cortical thickness was measured (r = -0.41, P < 0.05) and in the 17 controls (r = -0.50, P c 0.05) (Fig. 5). There was no significant relationship between bone volume and cortical thickness in either femoral fracture patients (r = -0.11, NS) or controls (r = 0.42, NS). As cortical thickness was age related, only femoral fracture patients and controls in the age range 60-82
-
femoral
0-0
d
fracture
normal
1200 0 0
e
e
e
50
60
70
60
90 AGE
years
Fig. 5. The relationship between corticalthickness and age in women with femoral fracture(r = -0.41, P < 0.05) and normal women (r = -0.50, P c 0.05).
343 Tabie 4
Comparison between cortical thickness in age matched femoral neck fracture patients and controls (results are expressed as means + SD)
Age(yea@ Bone volume (%) Cortical thickness @m)
Femoral fracture (n = 20)
Controls (n = 12)
73.1 f 5.8 15.7 -c 4.9 626.8 -c 286.2
69.3 + 7.6 15.2 + 4.2 716.9 + 284.4
Table 5 Comparison between age and cortical thickness in femoral fracture patients with subcapital and intertrochanteric fracture (results expressed as means f SD)
Age Cortical thickness km)
Subcapital (n = 16)
Intertrochanteric (n = 11)
72.9 f 6.8 677.8 & 335.6
80.2 + 4.6= 430.8 + 152.6b
Differences between groups: ‘P < 0.01; bP c 0.02.
were compared. There was no significant difference between age, bone volume or cortical thickness between the two groups (Table 4). In the subgroup of 27 femoral neck fracture patients in whom cortical thickness was mezsured, patients with intertrochanteric fracture had significantly lower cortical thickness (P C 0.02) and, as in the complete group (Table 2), patients with intertrochanteric fracture were significantly older than those with subcapital fracture (P < 0.01) (Table 5).
iscussio Age-related bone loss in normal women results mainly from a reduction in trabecular number [15,19,20], with trabecular thinning being less important [15,20]. It has been suggested that trabecular thinning in iliac crest biopsies [12] and femoral heads [II] from femoral neck fracture patients occurs relative to controls. We found no significant difference between mean trabeculsr width or number/field in the iliac crest between women with femoral fracture and age- and sex-matched controls. Both loss of trabeculae and trabecular thinning were apparent in femoral fracture patients with low bone volume. Variability in trabecular widths was large in both fracture and control subjects and as a result the power of the study to detect the observed difference of 5.6pm (~4%) between fractures and control groups was low. The power of the study to detect a 25 pm (=20%) difference was estimated as 0.95 at the 0.05 significance level. The use of postmortem biopsy samples as controls is not ideal. It has been suggested that sudden death series may contain more subjects with undiagnosed bone
344 c although every effort was made to exclude disease [20] than healthy volunteer.), such subjects from our series [15]. However the invasive nature of bone biopsy makes samples from healthy volunteers in their eighth and ninth decades difficult to obtain, both for practical and ethical reasons. We are able to say that there are no significant differences between bone architecture in age-matched women of similar bone volume with and without fracture. The correlation of trabecular number/field with MTPD suggests that isolation and complete removal of trabeculae rather than multiple perforation of trabecular plates into many small units [ 121was the major mechanism of bone loss in femoral fracture patients. The indirect method of calculating trabecular thickness (MTPT) [12] has the disadvantage of being unable to give information on the distribution of trabecular thickness within a biopsy. For example a fall in trabecular thickness in some regions may be offset by an increase in other regions. However we found the distribution curves of trabecular widths similar in the femoral fracture and control populations, with the percentage of very thin trabeculae (40pm) similar in fractures and controls. There was no significant difference between cortical thickness in age-matched femoral fracture patients and controls. However patients with intertrochanteric fracture had significantly lower cortical thickness than patients with subcapital fracture. We have shown that cortical thickness declines with age in both fracture and control populations, and, as seen in other studies [21,22]that patients with intertrochanteric fracture are older than those with subcapital fracture. Age-related cortical bone loss would therefore appear to determine whether a fracture is intertrochanteric or subcapital rather than any difference in cancellous bone volume or architecture. We conclude that the architecture of cancellous bone and cortical thickness at the iliac crest are similar in femoral fracture patients and age- and sex-matched subjects without fracture. Whilst cancellous architecture does not vary with age or fracture trauma in the femcbra1fracture population and is similar in subcapital and intertrochanteric fracture, cortical thickness declines with age and is lower in the more elderly patients with intertrochanteric fracture. Both loss of trabeculae and trabecular thinning contribute to low bone volume in femoral fracture patients.
Acknowledgements
We would like to thank Dr Jean Aaron ‘forallowing us to use data and samples from her biopsy series [15] and for laboratory facilities, and the orthopaedic surgeons of Leeds General Infirmary Mr M.J. Abberton, Mr E.B. Longton, Mr M.A. Nelson and Mr F.F. Silk for allowing us to study their patients and Mr A. Wynne for obtaining the biopsies. The authors would also like to thank Mr P. Constable for statistical advice and Miss J.L. Chatten for secretarial assistance.
345 eferences 1 Knowelden J, Buhr AJ, Dunbar 0. Incidence of fracturesin persons over 35 years of age. Br J preventive Social Med 1964;18:130-141. 2 Wallace Wk. The increasing incidence of fractures of the proximal femur: an orthopaedic epidemic. Lancet 1983;i:l413. 3 Evans J, Grimle;r. Epidemiology of proximal femoral fractures. In: Recent Advances in Geriatric Medicine. B. Issacs (ed). Churchill Livingstone. Edinburgh. 1982201-214. 4 Horsman A, Nordin BEC, Simpson M, Speed R. Cortical and trabecular bone status in elderly women with femoral neck fracture. Clin Orthop Rel Res 1982;166:143-151. 5 Cooper C, Barker DJP, Morris J, Briggs RSJ. Osteoporosis, falls and age in fracture of the proximal femur. Br Med J 1987;295:13-15. 6 Riggs BL, Wahner HW, Seeman E, Offord KP, Dunn WL, Maze:s RB, Johnson KA, Melton LJ. Changes in bone mineral density of the proximal femur and spine with aging. J C1in Invest 1982;70:716-723. 7 Aaron JE, Stasiak L, Gallagher JC, Longton EB, Nicholson M, Anderson J, Nordin BEC. Frequency of osteomalacia and osteoporosis in fractures of the prk>ximalfemur. Lancet l974;i:223-229. 8 Evans RA, Ashwell JR, Dunstan CR. Lack of metabolic bb?nedisease in patients with fracture of the femoral neck. Australian NZ J Med 1981;11:158-161. 9 Faccini JM, Exton-Smith AN, Boyde A. Disorders of bone and fracture of the femoral neck. Lancet 1976;i:1089-1092. 10 Pugh JW, Rose RM, Radin EL Elastic and viscoelastic properties of trabecular bone: dependence on structure. J Biomechanics 1973;6:475-485. 11 Fazzalari NL, Darracott J, Vernon-Roberts B. Histomorphometric changes in the trabecular structure of a selected stress region in the femur in patients with osteoarthritis and fracture of the femoral neck. Bone 1985;6:125-133. 12 Parfitt AM, Matthews CHE, Villaneuva AR, Kleerekoper M, Frame B, Rao DS. Relationships between surface, voiume and thickness of iliac trabecular bone in aging and in osteoporosis. J Clin Invest 1983;72:1396-1409. 13 Uitewaal PJM, Lips P, Netelenbos, JC. An analysis of bone structure in patients with hip fracture. Bone Mineral 1987;3:63-73. 14 Johnston CC, Norton J, Khain MRA, Kernek C, Edouard C, Arlot M, Meunier PJ. Heterogeneity of fracture syndromes in post menopausal women. J Clin Endocrinol Metab 1985;61:551-556. 15 Aaron JE, Makins NB, Sagreiya K. The microanatomy of trabecular bone loss in normal aging men and women. Clin Orthop 1987;215:260-271. 16 Melvin KEW, Hopner GW, Bordier P, Neale G, Joplin GF. Calcium metabolism and bone pathology in adult coeliac disease. Q J Med 1970;153:83- 113. 17 Peacock M. Gsteomalacia and Rickets. In: Nordin BEC, ed. Metabolic bone and stone disease. Edinburgh, London, New York, Melbourne: Churchill Livingstone, 1984;71-74. 18 Aaron JE, Stasiak L, Gallagher JC, Langton EB, Nicholson M, Anderson J, Nordin BEC. Frequency of osteomalacia and osteoporosis in fracturesof the proximalfemur. Lancet 1974;i:229-233. 19 Compston JE, Me11ishRWE, Garrahan NJ. Age related changes in iliac crest microanatomic bone structure in man. Bone 1987;8:289-292. 20 Reeker RR, Kimme DB, Parfitt AM, Davies KM, Keshawarz N, Hinders S. Static and tetracyclineb;lseo bone histomofphometric data from 34 normal postmenopausal females. J Bone Mineral Res 1988;3:133-144. 21 Gallagher JC, Melton LJ, Riggs BL, Bergstrath E. Epidemiology of fractures of the proximal femur in Rochester, Minnesota. Clin Orthop Rel Res 1980;150:163-171. 22 Erocklehurst JC, Exton-Smith AN, Lempert Barber SM, Hunt LP, Palker MK. Fractureof the femur in old age: a two centre study of associated clinical factors and the cause of the fall. Age Agei% 1978;1:7-15.
335-345
335
Efsevier BAM 00323
‘Rheumatology and RehabilitationResearch Unit, 36 Clarendon Road, Leeds LS2 9NZ, Enginnd and 2GeneratClinicalResearch Centre, UniversityHosplralIV59?,926 W. Michigan Street, Indianapolis,IN 46223, USA
(Received 12 January 1990) (Accepted 27 July 1990)
Summary The architecture of cancellous bone as well as quantity and quality is considered important in maintaining mechanical integrity. To determine whether abnormalities of architecture occur in femoral neck fracture we measured trabecular width and number in iliac crest bone biopsies of 68 women with femoral fracture and compared them with data from a postmortem series of age-matched women without known bone disease. Cortical thickness was measured in 27 of the fracture patients and 17 controls. After exclusion of ten biopsies with increased osteoid surface no significant difference was seen in mean trabecular width or number between fracture patients and subjects without fracture. Both thinning of trabeculae and Lossof trabeculae contributed to low bone volume in femoral fracture patients. Direct measurements of trabecular number correlated with calculated mean trabccu!:!srplate density (P < O.UOl),and the percentage of trabeculae at any one of a range of trabecular widths in the fracture biopsies was similar to that in non-fracture subjects. No difference was seen in the architecture of cancellous bone with age, fracture trauma or between subcapital and intertructtiarrtericfracture in fracture subjects. Cortical thickness, however, was related to age in both fracture and control subjects (P < 0.05). There was no difference in cortical thickness between age-matched fracture and control subjects. Patients with intertrochanteric fracture had lower cortical thickness (P < 0.02) and were older (P < 0.01) than patients with subcapital fracture.
Key words: Femoral fracture; Iliac crest biopsy; Trabecular width; Trabecular number; Cortical thickness; Osteoporosis
Femoral neck fracture is an important cause of morbidity and mortality in the elderly [1,2]. Its pathogenesis is disputed, but is probably multifactorial, depending on Correspondence to: Dr L.,n . Iiordon, Rheumatology and Rehabilitation Research Unit, 36 Clarendon Road, Leeds LS2 9NZ, England. 0169-60#/90/%03.50@I990 Elsevier Science Publishers B.V, (Biomedical Division)
336 the fall, the age-related impairment of protective responses to falling [3], and the quality and quantity of the bone in the femoral neck when the fall occurs. In many studies [4-81 patients with femoral neck fracture as a group have lower bone mass compared with age-matched subjects who have not fractured. However there is substantial overlap in measurements between fracture and non-fracture subjects, and some fracture subjects have more bone than expected for their age [7,9] suggesting that factors other than bone mass also play an important role in fracture. Less attention has been paid to the architecture of cancellous bone in femoral fracture patients using histomorphometric techniques although architecture is probably as important as bone mass in maintaining the integrity of the skeleton [lo]. One study of the femoral heads from patients with femoral fracture showed thinner trabeculae than in patients with osteoarthritis [ 111.Calculation of mean trabecular plate thickness (MTPT) and mean trabecular plate density (MTPD) can be performed using measurements of total perimeter and bone area in thin sections from iliac crest biopsies [12]. One study using this method showed a reduction in MTPD in vertebral fracture patients and a reduction in MTPT in femoral fracture patients [ 121.However, this was not confirmed by another study [ 131which showed reduced MTPT in intertrochanteric fractures only whilst cervical fractures showed higher bone volume and greater trabecular surface density compared with intertrochanteric fractures. However in these three studies [ll-131 controls were not well matched for age and in one study [ 131were not matched for sex. Whilst histomorphometric studies have shown a reduction in cortical thickness in transiliac biopsies in femoral fracture patients compared with younger controls [13,14] it is not clear whether cortical thickness is reduced compared with age- and sex-matched subjects. The aims of this study were to examine the architecture of cancellous and cortical bone in femoral neck fracture patients and to compare this with data from a postmortem series of age- and sex-matched subjects without fracture, to determine whether trabecular loss and/or trabecular thinning were responsible for low bone volume in femoral neck fracture and to examine whether abnormalities in architecture explained why some patients with normal bone volume for their age sustain femoral neck fracture. We also wished to determine whether differences in architecture occurred between subjects with subcapital and intertrochanteric fractures, and between subjects with traumatic and non-traumatic fractures.
Materials and Methods Patients Sixty-eight women aged 52-99 years (mean 78) admitted to Leeds General Infirmary with femoral neck fracture from March 1984 to February 1986 were studied. A11female patients admitted under one of four orthopaedic teams sharing acute orthopaedic admissions were considered for study. During this period a total of 329
337 women were admitted with femoral neck fracture and 229 of these patients received an operation. Our sample therefore represents 21% of the fracture population and 30% of those operated on for this condition. Patients considered unfit for surgery because of severe dementia or physical illness were not biopsied for ethical reasons. A total of 72 women were biopsied, but four samples were inadequate for analysis and these four patients were excluded from the study. Patients were classified radiologically into two groups, those with subcapital and transcervical fractures, and those with basal cervical and intertrochanteric fractures. Fractures resulting from a fall from standing height or less were further classified as ‘non-traumatic’ and fractures resulting from falls greater than standing height, e.g., fall down steps, minor road accidents, as ‘traumatic’. Patients with subtrochanteric fractures, and those known to have malignant disease were excluded. Informed consent was obtained from all subjects. Published data on mean trabecular width and number/field from a postmortem series of age-matched women who died suddenly and who had no medical condition or drug therapy predisposing to bone disease were used for comparison [15]. In addition cortical thickness was measured in transiliac biopsies from 17 w0mc.n aged 55-81 (mean 66 years) comprising I3 subjects from the postmortem series ;-nd four women undergoing minor surgery who had no medical condition or drug therapy predisposing to bone disease. Methods Transilial bone biopsies were obtained from a site 1 inch below and 1 inch behind the anterior superior iliac spine at the time of operation using an 8 mm Bordier trephine. The bone biopsies were fixed in 10% phosphate-buffered formalin for 48 h, dehydrated through graded alcohois, cleared in xylene, embedded in methylacrylate and left to polymerise until hard enough for sectioning. Between 30 and 48 longitudinal sections 10 pm thick were cut from each biopsy using an L.K.B. 2258 PMV heavy duty microtome. Sections were stained with Goldner to demonstrate osteoid surfaces. Bone v&me (BViTV) The corticomedullary junction was defined as the endocortical surface. Measurements on cancellous bone were made on the entire cancellous area. In a small minority of biopsies a transitional zone intermediate between cortical and cancellous bone was present and in accordance with our laboratory practice was excluded from cancellous measurements. The amount of cancellous bone per unit area was determined on 64 fields, four fields on each of 16 sections. Each field was examined using a 25-point Zeiss integrating eyepiece and a magnification of x50 and a point on the grid coinciding with a trabecula was scored as a hit. The number of hits in the four fields of each section was taken as a measure of the percentages of bone in that section. The nizan value of the 16 sections was expressed as a percentage volume of bony tissue, assuming the biopsy to be homogeneous.
338 Trabecular width and trabecularnumber/field These were measured directly by ahe methods described in detail by Aaron et al. [ 151. Trabecular number/field was measured on an area of section outlined by the rectangular field marker encompassing the focusing reticule in the viewing tube of Zeiss Photomicroscope II. This area had been previously selected [IS] such that the length of the field was equivalent to the shortest intercortical distance found among biopsy specimens from the postmortem series and measured 5.4 X 3.6 mm at a specimen magnification of x20. The number of individual trabeculae, defined as domains of bone extending between junctions, in this area was counted in duplicate in 10 sections from each sample. Biopsies in which one cortex was absent or damaged were included providing the area of intact trabecular bone continuous with the other cortex was large enough for the assessment of trabecular number per field. The width of trabeculae was determined using a Vickers Patholette microscope fitted with a projection illumination base and an overhead viewing screen. Attached to the screen was a clear perspex disc bearing the pattern of the Zeiss no. 1 integrating grid. At a magnification of x50, trabecular thickness was measured at the points where the grid intersected the trabecular bars, excluding junctions where several trabeculae fused and the junction of trabeculae with cortex. Non-overlapping fields were selected from throughout the biopsy specimen but areas of damage or artefact, particularly at the margins of the biopsy, were avoided. To check interobserver error and to assess the validity of comparison of the values derived from the femoral neck fracture biopsies with normal values derived from the postmortem study [15], 10 samples from the latter study were analysed. The precision was found to be well within the 5% considered to be acceptable variation for the technique. As well as a mean value for trabecular width for each biopsy the percentage of trabeculae in each biopsy at each of a range of trabecular widths was calculated. When data from all biopsies were combined this enabled a distribution curve of trabecular widths to be plotted and compared with that obtained from the previous study. In addition the mean trabecular plate density (MTPD) was calculated. It was derived from the ratio of per cent bone volume (TBV%) and trabecular width according to the equation: mean trabecular plate density (/mm) = TBV%/trabecular width (lim) x 10 [12]. Osteoidsurface (OS/.BS) This was assessed using the methods described by Aaron et al. [15] and the normal range for the percentage of trabecular surface covered in osteoid was defined as ~24% [7,16]. Osteoidthickness(0. th) This was assessed using the methods described by Aaron et al. [ 151and the normal range for O.th defined as 60% [ 181.
339 Cortical thickness (Ct. th) Only biopsies in which both cortices were intact were measured. Cortical thickness was measured on each cortex at four equidistant sites at a magnification of x50 using a calibrated eye-piece graticule, and a mean value for two cortices obtained. Cortical thickness in fracture and control subjects was measured by the same observer. Statistics Differences between groups were assessed using Student’s t-test. The correlation of two variables was assessed by linear regression analysis.
Trabecular architecture 01 the 68 biopsi.es, 10 showed increased osteoid surface and as in the postmortem series, were excluded from the analysis. Seven of these 10 biopsies fulfilled the criS ~60%). Values for teria for osteomalacia (OS/SS >24%, O.th >13ym and the excluded biopsies are given in Table 1. In the remaining 58 patients the mean trabecular width was 126.9 + 26.7 pm (SD) compared with 132.5 + 25.9 pm in 20 Table 1 Bone volume, mean trabecular width and trabecular number/field in women with femoral neck fracture and increased osteoid surface (n = 10) (results expressed as means fi SD) Bone volume W)
Mean trabecular width fpm)
Mean trabecular number/field
18.7 z!z5.7
170.6 + 58.2
19.6 AZ7.0
femoral
o--o
fracture
controls
Fig. 1. Distribution curves of trabecular width in women with femoral fracture and in normal elderly women.
340 .
73
.
femoral
0
normal
lraclure elderly
a .l
8
.
z 0
.
40
*
*
t
e224
10
90
110
130
170
150 MEAN
190
TRAEECULAR
WIDTH
um
Fig. 2. The relationship between mean trabecular number/field and mean trabecularwidth in women with femoral fracture(r = 0.006, NS) and in normal elderlywomen (r = -0.50, P < 0.05). normal women aged 56-94 (mean 75) years [15]. This difference was not signifi-
cant. The mean trabecular number/field for the 58 femoral fracture patients was 25.6 f 10.6 compared with 24.2 + 6.3 for age-matched normal women. This difference was not significant. In comparison with normal eldery women [ 151, the distribution curve of trabecular width in femoral fracture patients (Fig. 1) showed fewer trabeculae greater than 180pm in width and more in the region of 80-120pm. There was no relationship between mean trabecular number/field and mjean trabecular width (Fig. 2) in the fracture patients (r = 0.006, NS) although in the nor-
t: 5
.8@ 10
e. 15
BONE
“O~i!lME
%
25
30
35
Fig. 3. Therelationship between mean trabecularnumber/fieldand bone volume in women with femoral fracture(r = 0.42, P < 0.01).
341
5
160.
1 z 2
160.
a a -I 3 140. :: < I z l-20. a ii 100. go-
,
5
15
BONE
W?“ME
9b
25
30
35
Fig. 4. The relationship between mean trabecular width and bone volume in women with femoral fracture (r = 0.61, P c 0.001).
ma1 elderly women there was an inverse relationship between mean trabecular number/field and mean trabecular width (r = -0.50, P < 0.05). However there was a small group of eight femoral fracture patients with higher numbers of trabeculae than expected (mean ir 2 SD for normal women, 12.3-36.1). All had a mean trabecular width between 90 and 110pm and bone volume in all but one was .n.Lovethe mean of 16% for age-matched normal women from the postmortem series [IS]. There was a significant correlation between mean trabecular number/field and mean trabecular plate density (r = 0.76, P < 0.001) in the fracture patients. In the fracture patients, bone volume correlated with trabecular number/field (Fig. 3) (r = 0.42, P < 0.01) and trabecular width (r = 0.61, P < 0.001) (Fig. 4). No significant relationship of mean trabecular width, number/field or bone volume to age was seen in the fracture patients (r = -0.05, r = -0.01 and r = -0.14, respectively. There was no significant difference in mean trabecular width, number/field and btine volume between subcapital and intertrochanteric fractures (Table 2). However patients with intertrochanteric fracture were significantly older than those with subcapital fracture (P < 0.001). Table 2
Comparison of age, mean trabecular width, number/field and bone volume in subcapital and intertrochanteric fracture (results are expressed as means z!zSD)
Age (years) Bone volume (%) Trabecular width @m) Trabecular number/field a Difference between groups, P < 0.001.
Subcapital (n = 34)
Intertrochanteric (n = 24)
73.6 i: 15.8 k 128.1 f 26.0 f
81.8 + 15.4 + 124.4 + 25.0 +
9.3 5.8 27.4 9.3
5.9” 6.4 24.5 12.2
342 Table 3 Age, mean trabecular width, number/field and bone volume in women with femoral neck fracture in relation to fracture trauma (results are expressed as means + SD)
Age(yea4 Bone volume (%) Trabecularwidth @m) Trabecularnumber/field
Traumatic (n = 12)
Non-traumatic (n = 34)
74.2 f 16.7 f 131.8 f 23.8 +
79.1 + 15.4 + 125.8 f 25.9 f
4.8 4.8 23.2 8.3
9.3 7.0 29.7 11.7
In 46 patients the cause of fracture was sufficiently well documented to classify as traumatic (fall down steps (n = 9), knocked down by car (n = 2), fall from car roof (n = 1)) or non-traumatic (n = 34). Two traumatic fractures were alcohol-related. No significant difference was seen in age, mean trabecular width, number/field or bone volume between these two groups of patients (Table 3). Cortical thickness There was a significant relationship between age and cortical thickness in the 27 femoral neck fracture patients in whom cortical thickness was measured (r = -0.41, P < 0.05) and in the 17 controls (r = -0.50, P c 0.05) (Fig. 5). There was no significant relationship between bone volume and cortical thickness in either femoral fracture patients (r = -0.11, NS) or controls (r = 0.42, NS). As cortical thickness was age related, only femoral fracture patients and controls in the age range 60-82
-
femoral
0-0
d
fracture
normal
1200 0 0
e
e
e
50
60
70
60
90 AGE
years
Fig. 5. The relationship between corticalthickness and age in women with femoral fracture(r = -0.41, P < 0.05) and normal women (r = -0.50, P c 0.05).
343 Tabie 4
Comparison between cortical thickness in age matched femoral neck fracture patients and controls (results are expressed as means + SD)
Age(yea@ Bone volume (%) Cortical thickness @m)
Femoral fracture (n = 20)
Controls (n = 12)
73.1 f 5.8 15.7 -c 4.9 626.8 -c 286.2
69.3 + 7.6 15.2 + 4.2 716.9 + 284.4
Table 5 Comparison between age and cortical thickness in femoral fracture patients with subcapital and intertrochanteric fracture (results expressed as means f SD)
Age Cortical thickness km)
Subcapital (n = 16)
Intertrochanteric (n = 11)
72.9 f 6.8 677.8 & 335.6
80.2 + 4.6= 430.8 + 152.6b
Differences between groups: ‘P < 0.01; bP c 0.02.
were compared. There was no significant difference between age, bone volume or cortical thickness between the two groups (Table 4). In the subgroup of 27 femoral neck fracture patients in whom cortical thickness was mezsured, patients with intertrochanteric fracture had significantly lower cortical thickness (P C 0.02) and, as in the complete group (Table 2), patients with intertrochanteric fracture were significantly older than those with subcapital fracture (P < 0.01) (Table 5).
iscussio Age-related bone loss in normal women results mainly from a reduction in trabecular number [15,19,20], with trabecular thinning being less important [15,20]. It has been suggested that trabecular thinning in iliac crest biopsies [12] and femoral heads [II] from femoral neck fracture patients occurs relative to controls. We found no significant difference between mean trabeculsr width or number/field in the iliac crest between women with femoral fracture and age- and sex-matched controls. Both loss of trabeculae and trabecular thinning were apparent in femoral fracture patients with low bone volume. Variability in trabecular widths was large in both fracture and control subjects and as a result the power of the study to detect the observed difference of 5.6pm (~4%) between fractures and control groups was low. The power of the study to detect a 25 pm (=20%) difference was estimated as 0.95 at the 0.05 significance level. The use of postmortem biopsy samples as controls is not ideal. It has been suggested that sudden death series may contain more subjects with undiagnosed bone
344 c although every effort was made to exclude disease [20] than healthy volunteer.), such subjects from our series [15]. However the invasive nature of bone biopsy makes samples from healthy volunteers in their eighth and ninth decades difficult to obtain, both for practical and ethical reasons. We are able to say that there are no significant differences between bone architecture in age-matched women of similar bone volume with and without fracture. The correlation of trabecular number/field with MTPD suggests that isolation and complete removal of trabeculae rather than multiple perforation of trabecular plates into many small units [ 121was the major mechanism of bone loss in femoral fracture patients. The indirect method of calculating trabecular thickness (MTPT) [12] has the disadvantage of being unable to give information on the distribution of trabecular thickness within a biopsy. For example a fall in trabecular thickness in some regions may be offset by an increase in other regions. However we found the distribution curves of trabecular widths similar in the femoral fracture and control populations, with the percentage of very thin trabeculae (40pm) similar in fractures and controls. There was no significant difference between cortical thickness in age-matched femoral fracture patients and controls. However patients with intertrochanteric fracture had significantly lower cortical thickness than patients with subcapital fracture. We have shown that cortical thickness declines with age in both fracture and control populations, and, as seen in other studies [21,22]that patients with intertrochanteric fracture are older than those with subcapital fracture. Age-related cortical bone loss would therefore appear to determine whether a fracture is intertrochanteric or subcapital rather than any difference in cancellous bone volume or architecture. We conclude that the architecture of cancellous bone and cortical thickness at the iliac crest are similar in femoral fracture patients and age- and sex-matched subjects without fracture. Whilst cancellous architecture does not vary with age or fracture trauma in the femcbra1fracture population and is similar in subcapital and intertrochanteric fracture, cortical thickness declines with age and is lower in the more elderly patients with intertrochanteric fracture. Both loss of trabeculae and trabecular thinning contribute to low bone volume in femoral fracture patients.
Acknowledgements
We would like to thank Dr Jean Aaron ‘forallowing us to use data and samples from her biopsy series [15] and for laboratory facilities, and the orthopaedic surgeons of Leeds General Infirmary Mr M.J. Abberton, Mr E.B. Longton, Mr M.A. Nelson and Mr F.F. Silk for allowing us to study their patients and Mr A. Wynne for obtaining the biopsies. The authors would also like to thank Mr P. Constable for statistical advice and Miss J.L. Chatten for secretarial assistance.
345 eferences 1 Knowelden J, Buhr AJ, Dunbar 0. Incidence of fracturesin persons over 35 years of age. Br J preventive Social Med 1964;18:130-141. 2 Wallace Wk. The increasing incidence of fractures of the proximal femur: an orthopaedic epidemic. Lancet 1983;i:l413. 3 Evans J, Grimle;r. Epidemiology of proximal femoral fractures. In: Recent Advances in Geriatric Medicine. B. Issacs (ed). Churchill Livingstone. Edinburgh. 1982201-214. 4 Horsman A, Nordin BEC, Simpson M, Speed R. Cortical and trabecular bone status in elderly women with femoral neck fracture. Clin Orthop Rel Res 1982;166:143-151. 5 Cooper C, Barker DJP, Morris J, Briggs RSJ. Osteoporosis, falls and age in fracture of the proximal femur. Br Med J 1987;295:13-15. 6 Riggs BL, Wahner HW, Seeman E, Offord KP, Dunn WL, Maze:s RB, Johnson KA, Melton LJ. Changes in bone mineral density of the proximal femur and spine with aging. J C1in Invest 1982;70:716-723. 7 Aaron JE, Stasiak L, Gallagher JC, Longton EB, Nicholson M, Anderson J, Nordin BEC. Frequency of osteomalacia and osteoporosis in fractures of the prk>ximalfemur. Lancet l974;i:223-229. 8 Evans RA, Ashwell JR, Dunstan CR. Lack of metabolic bb?nedisease in patients with fracture of the femoral neck. Australian NZ J Med 1981;11:158-161. 9 Faccini JM, Exton-Smith AN, Boyde A. Disorders of bone and fracture of the femoral neck. Lancet 1976;i:1089-1092. 10 Pugh JW, Rose RM, Radin EL Elastic and viscoelastic properties of trabecular bone: dependence on structure. J Biomechanics 1973;6:475-485. 11 Fazzalari NL, Darracott J, Vernon-Roberts B. Histomorphometric changes in the trabecular structure of a selected stress region in the femur in patients with osteoarthritis and fracture of the femoral neck. Bone 1985;6:125-133. 12 Parfitt AM, Matthews CHE, Villaneuva AR, Kleerekoper M, Frame B, Rao DS. Relationships between surface, voiume and thickness of iliac trabecular bone in aging and in osteoporosis. J Clin Invest 1983;72:1396-1409. 13 Uitewaal PJM, Lips P, Netelenbos, JC. An analysis of bone structure in patients with hip fracture. Bone Mineral 1987;3:63-73. 14 Johnston CC, Norton J, Khain MRA, Kernek C, Edouard C, Arlot M, Meunier PJ. Heterogeneity of fracture syndromes in post menopausal women. J Clin Endocrinol Metab 1985;61:551-556. 15 Aaron JE, Makins NB, Sagreiya K. The microanatomy of trabecular bone loss in normal aging men and women. Clin Orthop 1987;215:260-271. 16 Melvin KEW, Hopner GW, Bordier P, Neale G, Joplin GF. Calcium metabolism and bone pathology in adult coeliac disease. Q J Med 1970;153:83- 113. 17 Peacock M. Gsteomalacia and Rickets. In: Nordin BEC, ed. Metabolic bone and stone disease. Edinburgh, London, New York, Melbourne: Churchill Livingstone, 1984;71-74. 18 Aaron JE, Stasiak L, Gallagher JC, Langton EB, Nicholson M, Anderson J, Nordin BEC. Frequency of osteomalacia and osteoporosis in fracturesof the proximalfemur. Lancet 1974;i:229-233. 19 Compston JE, Me11ishRWE, Garrahan NJ. Age related changes in iliac crest microanatomic bone structure in man. Bone 1987;8:289-292. 20 Reeker RR, Kimme DB, Parfitt AM, Davies KM, Keshawarz N, Hinders S. Static and tetracyclineb;lseo bone histomofphometric data from 34 normal postmenopausal females. J Bone Mineral Res 1988;3:133-144. 21 Gallagher JC, Melton LJ, Riggs BL, Bergstrath E. Epidemiology of fractures of the proximal femur in Rochester, Minnesota. Clin Orthop Rel Res 1980;150:163-171. 22 Erocklehurst JC, Exton-Smith AN, Lempert Barber SM, Hunt LP, Palker MK. Fractureof the femur in old age: a two centre study of associated clinical factors and the cause of the fall. Age Agei% 1978;1:7-15.
