Issues in modern bone histomorphometry (2024)

Abstract

This review reports on proceedings of a bone histomorphometry session conducted at the Fortieth International IBMS Sun Valley Skeletal Tissue Biology Workshop held on August 1, 2010. The session was prompted by recent technical problems encountered in conducting histomorphometry on bone biopsies from humans and animals treated with anti-remodeling agents such as bisphosphonates and RANKL antibodies. These agents reduce remodeling substantially, and thus cause problems in calculating bone remodeling dynamics using in vivo fluorochrome labeling. The tissue specimens often contain few or no fluorochrome labels, and thus create statistical and other problems in analyzing variables such as mineral apposition rates, mineralizing surface and bone formation rates. The conference attendees discussed these problems and their resolutions, and the proceedings reported here summarize their discussions and recommendations.

Keywords: Histomorphometry, Bone remodeling, Over-suppression

Introduction

The understanding of bone physiology accelerated in the 1950s with the development of plastic-embedding technology for microscopic examination of undemineralized sections of bone tissue. Prior to then, bone histology required removal of its principal component, the mineral. This new technology was refined in several labs, in particular, the Radiobiology Lab at the University of Utah in studies of the skeletal effects of plutonium and radium, alpha radiation-emitting, bone-seeking metals. A benchmark occurred in 1960 with a publication by Frost [1] entitled, “Measurement of Bone Formation in a 57 Year-old Man by Means of Tetracyclines.” This opened the door for study of the complete composite of human bone, including both mineral and organic matrix. In 1983, a text was produced [2] which summarized the methodology and stereological theories behind the histomorphometric calculations. Another benchmark occurred in 1987 when the nomenclature, symbols and units for bone histomorphometry were standardized [3]. As a consequence, the use of in-life dual fluorochrome labels in humans and animals together with the application of bone histomorphometric techniques has become a standard method for studying the tissue level activities of bone modeling and remodeling. Starting in the late 1960s, Professor Webster Jee began the Sun Valley Workshops on musculoskeletal biology that continue to the present. In August 2010, the International Bone and Mineral Society, along with the National Institutes of Health, sponsored the 40th Annual IBMS/Sun Valley Workshop on musculoskeletal biology. At that meeting a half-day session was devoted to discussing issues in modern histomorphometry, 50 years after its introduction. The authors participated in the presentations and subsequent discussion.

Change in landscape

The success of potent anti-remodeling agents in reducing fracture risk in osteoporotic patients in recent years has presented new real world problems for bone histomorphometry, as it is applied to explain their tissue level mechanisms. Bone biopsies taken from patients, or specimens taken from animals treated with these agents, may have few or no fluorochrome labels, creating difficulty in the interpretation of the dynamic histomorphometric data and evaluation of patients’ or animals’ bone remodeling status. This experience has generated a number of important questions that have been debated in publications and scientific meetings over the past decade. Other issues involve use of proper referents for histomorphometric measurements, clarification of the utility of certain bone histomorphometric measurements and endpoints, sample size requirements for valid measurements, quantification of rare events, and the variables that should be reported to describe best the status of bone remodeling. The August 2010 Symposium sought to summarize the following questions and provide recommendations on how they might be answered:

1. How should one express and interpret mineralizing surface (MS/BS) and bone formation rate (BFR/BS) in the absence or paucity of double and/or single label in the standard sampling area?

2. How should mineral apposition rate (MAR) be expressed in the absence or paucity of double label in the standard sampling area? Should MAR be imputed (i.e., assigned a value)? If so, when?

3. Is an “extended label search” worth doing in the absence of label in the standard sampling area? (The definition of standard sampling area is 2–4 sections, ~100 μ apart, from the middle one third of a human transiliac biopsy specimen, or embedded animal specimen, sufficient to provide the minimum area for analysis as described later.) What should be the purpose of such a search? If label is found during the search, how should it be expressed?

4. Should MAR be measured and reported if found only on an extended label search? If so, in what manner?

5. What is the minimum and practical maximum number of double label width measurements required to achieve a reliable estimate of MAR? How should one handle situations with insufficient double label to achieve a reliable estimate of MAR?

6. What is the minimum number of measured sections, area, and perimeter to provide an adequate sample?

7. What is activation frequency? How should one interpret activation frequency? What assumptions are required? Is it a valid expression of remodeling rate?

8. Can one measure “erosion depth” reliably? Is it valuable to do so?

9. Is the “obliquity correction” required for width measurements?

10. Should one examine cortical bone in human transilial biopsies? If so, what variables should be measured?

11. What is the confidence in extrapolations from a transiliac biopsy (~1/14,000 of the skeleton) to the entire skeleton?

12. Can one define the term “bone anabolic agent” solely in histomorphometric terms? If not, what additional data are needed?

13. What is the definition of “over-suppression” of remodeling? Does “over-suppression of remodeling” exist in a practical sense? Do biochemical markers of bone formation and resorption assist in defining “over-suppression?”

Extended label search

When no double label, or insufficient double label from which to make MAR measurements (see Section Extended label search) is found in the standard analysis regions, we recommend that a limited extended label search be undertaken to confirm the presence or absence of double label in the specimen, or identify additional double labels to solidify the measurement of MAR. The most important reason to conduct any form of extended label search is that when label is found outside the standard analysis regions, it can be confirmed that the subject complied with the labeling schedule. When an extended label search is undertaken, this should be noted on the patient’s report.

The first place to examine is the trabecular bone outside the standard analysis regions defined as the area in sections outlined as in Fig. 1 to exclude the transitional zone and the bone fragments on the marrow edges in the two unstained sections that were analyzed (see Section Mineral apposition rate). If double label is found anywhere in this location, it can be assured that the patient took the fluorochrome label as directed and that, in a histomorphometric sense, some trabecular remodeling exists, despite the absence of double label in the standard region for histomorphometric analysis. This label can be used to solidify the measurement of MAR. The search is concluded.

If double label is not found as indicated earlier, the second place to examine is the cortices of the two unstained sections that contain the standard analysis regions. If double label is found in the cortices, it can be assured that the patient took both fluorochrome labels and has cortical remodeling, but not that the subject has trabecular remodeling. Cortical double labels cannot be used to solidify the measurement of MAR. It is just as certain that neither this limited search, nor any rational histomorphometric search, can ever confirm the absence of trabecular remodeling in the subject. In fact, experience has shown that most subjects, who lack double label in any part of the trabecular region from the first two levels of examination, have double label somewhere in the cortical region of those levels.

We believe that it is not warranted to continue to search for double labels in trabecular regions past this point. The discussion pertaining to the normal lower limit of remodeling provides a partial answer: total absence of double label anywhere in an iliac biopsy specimen did not occur in the standard analysis region of biopsies from a sizable cohort of normal premenopausal women in whom remodeling rates were quite low [8]. Thus, total absence of label in a specimen in an extended label search appears to mean that the subject’s remodeling rate is abnormally low, provided one is reasonably confident that the subject complied with the labeling schedule. Again, it is the authors’ contention that no search of the transilial biopsy specimen, including sectioning the specimen until exhaustion and examining repeated trabecular regions in detail, can assure the absence of trabecular remodeling. Our opinion is that it is certain that the finding of no double label in the standard analysis region indicates the existence of a low remodeling condition. The practical difference between this state and that indicated by sectioning a specimen until exhaustion and finding no double label, is too small to be worth the additional work. This finding of low remodeling is confirmed for a group when similar data emerge from multiple individuals within a treatment group.

However, if either single or double label is found outside the standard analysis region during an extended label search, how does one express the result for purposes of analyzing data from a cohort? In this case it is justified to use MAR calculated from trabeculae outside the standard analysis region in calculation of those variables that use the value as long as this is noted. Further, as noted earlier, standard histomorphometric report forms should include a location that, in the event of a zero MS/BS, records that an extended label search was performed and its outcome.

Although compliance with the labeling instructions can sometimes be confirmed by collecting and analyzing urines for tetracycline at the time of labeling, there are practical problems with this, including sample collection, shipping and analytical difficulties. We recommend that a patient or research subject be contacted by telephone on the days when fluorochrome label is to be consumed, to review specific details (i.e., timing of tablets; not to be taken within 2 h of a meal; and not to be consumed with calcium supplements, dairy products, or antacids in the surrounding 2 h) and obtain verification that the medication was properly taken. We also recommend that the investigators should: a) limit the time the specimen stays in aqueous fixative to 24 h, b) limit the exposure of the specimen and cut sections to ambient light, and c) use appropriate UV light sources for fluorescence microscopy. For animals, we recommend that fluorochrome label be given by subcutaneous, rather than intraperitoneal (IP) injection. IP injections have the potential for injection into the gut or bladder which will result in lack of incorporation.

Adequate sampling of a human transilial biopsy (Fig. 1)

The minimum acceptable total section area (Tt.Ar) in the standard analysis regions is 30 mm². The minimum acceptable total section perimeter (Tt.Pm) is 60 mm. The authors recommend obtaining two sets of sections as a standard. The first set is taken from an area a bit less than half way through the biopsy cylinder. The second is taken from an area ~300 μm further through the biopsy core. This leaves enough of the embedded block intact for additional sectioning if needed or for application of other analytical techniques (e.g., backscattered electron imaging to assess mineralization density).

Each set consists of three sections, one for Goldner’s stain (Fig. 1), one for Toluidine blue (or McNeal’s tetrachrome) stain and one left unstained for fluorescent microscopy. If the two levels do not provide standard analysis regions with at least 30 mm² of total section area (Tt.Ar) and 60 mm of total section perimeter (Tt.Pm), another set is taken ~300 μm further through the specimen, and more measurements are added to the dataset for that specimen. It is rare that an intact, well-fixed human iliac biopsy core taken with a 7.5 mm trephine contains less than 30 mm² in the standard analysis regions from two sets of sections.

Activation frequency and erosion depth: meaning and measurement

Bone remodeling or turnover is carried out by teams of juxtaposed osteoclasts (at the front) and osteoblasts (bringing up the rear), comprising temporary anatomical structures known as basic multicellular units (BMUs). In cortical bone, the BMUs drill tunnels or “cutting cones” through the compact tissue while in spongy, cancellous bone, they usually gouge across the trabecular surface forming a serpentine trench [11]. Bone turnover begins by conversion of a quiescent skeletal surface to a remodeling site, a process referred to as activation. The activation frequency (Ac.f) represents the probability (#/year) that a new remodeling cycle will be initiated at any point on the cancellous perimeter and is calculated by dividing the surface-based bone formation rate (BFR/BS) by the average amount of bone formed by a team of osteoblasts per activation event, i.e., the wall thickness (W.Th). Thus, Ac.f=((BFR/BS)/W.Th). The measurement is more than just the rate of BMU origination, as it corresponds to the product of the frequency of BMU origination and completed bone structural unit (BSU) thickness, rate of progression and lifespan. Ac.f is the best available two-dimensional histologic index of the intensity of bone remodeling or turnover and as such, also represents the number of BMUs currently active.

Ac.f is a highly derived variable whose determination presents issues in calculation (Table 2), assumptions, and interpretation. As discussed earlier, fluorescent labels are occasionally missing. In addition, even without an increase in bone formation rate, Ac.f will increase if W.Th decreases, as it does with age [12], and as it perhaps does with some anti-remodeling therapies. If W.Th decreases without a change in BFR/BS, Ac.f could increase even though the “rate of bone turnover” assessed by BFR/BS is apparently unchanged. Furthermore, interpretation of the Ac.f becomes complicated when an anabolic therapy increases wall thickness or thickens a trabecular profile by adding bone to a previously completed BSU, as occurs with the intermittent administration of parathyroid hormone. Because the calculation of Ac.f is based on measurements of MS/BS, it does not measure directly the activation of osteoclasts, the vigor of their work, or their lifespan.

Some prefer using just MS/BS alone as a measure of the bone remodeling rate. However, this approach has drawbacks: it is influenced greatly by the surface density and volume of existing trabeculae, and does not take into account the length of time required to complete the formation phase of remodeling sites. It is, however, a valid expression of remodeling rates, and is correlated with remodeling as expressed by measurement of circulating biochemical markers of bone formation and resorption [13]. As mentioned earlier, MS/BS is also robust even in the face of absence of double and single label, as zero is a reasonable value for MS/BS in such specimens.

While bone histomorphometric techniques work very well for the evaluation of bone formation status, they are less reliable when used to evaluate bone resorption. Though the concept of erosion depth is important and valid, there are multiple practical problems associated with its measurement. Erosion depth represents an index of osteoclast vigor that is otherwise difficult to determine. In theory, erosion cavities have an adverse effect on cancellous bone strength disproportionate to the decrease in bone mass that the cavities represent [14]. Furthermore, loss of bone strength due to perforation of trabecular profiles becomes more likely as erosion depth increases. Erosion depth has a wide frequency of distribution. Limited evidence indicates that it decreases with age more noticeably in women than in men [15]. Difficulty with the measurement itself occurs when erosion cavities have unusual shapes that defy efforts to reconstruct the original bone perimeter [16]. In addition, erosion cavities are often few and inconspicuous in specimens obtained from elderly patients with osteoporosis. Selection of only the cavities ideal for measurement may cause considerable measurement bias [17].

Eroded surface is a static measure that gives a valid snapshot of the morphologic picture of a specimen’s resorption surfaces, but no direct information concerning the dynamic status of resorption [13,18], because rate of osteoclast activity cannot be directly measured. The eroded surface is composed of the osteoclast surface (about 1% of the trabecular bone surface in humans) plus the reversal surface (about 9% of the trabecular bone surface in humans). Osteoclast surface has proven useful in identifying the occasional occurrence of large osteoclasts in persons treated with bisphosphonates [19,20]. Reversal surfaces increase with defective or delayed bone formation, as occurs with glucocorticoid excess, and thus may have little to do with the current amount of bone resorption. Moreover, most antiresorptive drugs decrease the ability of osteoclasts to erode bone, reduce the BFR, and actually increase the reversal surface, thereby, increasing eroded surface. Thus the interpretation of eroded surface in cases in which anti-remodeling drugs have been used can be misleading.

We recommend that osteoclast surface be included, because there is good inter-laboratory consensus on its identification and it can provide valuable information on cell number. We also recommend that despite limitations, eroded surface be included. While we acknowledge the concept of erosion depth, we recommend that it not be evaluated in transilial biopsies (Table 3), because the sample is usually small, the measurement can be quite variable, and no inter-laboratory consensus on the optimal method for obtaining erosion depth data exists.

Histomorphometric variables (Tables 1–3)

The variables, and their symbols, included in Tables 1–3 are taken from the 1987 reference that standardized the nomenclature, symbols and units for bone histomorphometry [3].

We recommend dividing histomorphometric variables into two classifications: measured variables (Table 1) and variables that should be reported (Table 2) in all publication of histomorphometric studies. However, the four cortical measurements can be omitted in any report that does not require assessment of cortical remodeling. The reported variables include a few measured variables used for quality assurance and a number of others that are calculated by previously published formulae that are recorded here. Since microscope variables refer to data that are collected from two-dimensional sections, they are always expressed in two dimensions. Since reported variables nearly always refer to calculations that are done to understand bone remodeling in a three-dimensional sense, they are expressed in three dimensions (3D). We feel that the variables listed in Table 2 are necessary and sufficient to show the complete picture of the histomorphometric findings in any human or animal experiment.

The thickness expressions in the calculated trabecular variables are converted to 3D by multiplying the width (microscope) measurements by π/4, the obliquity correction factor [3]. It is not necessary for cortical bone. Converting to 3D may aid in understanding the dynamic (tetracycline-based) histomorphometric expressions of remodeling, and correlation with non-histomorphometric measurements of dynamic bone remodeling and/or 3D measurements of micro-architecture using μ CT may be improved [21]. We recommend that authors indicate whether the obliquity correction factor was used in order to compare with data from other reports.

When widths (O.Wi, W.Wi, etc.) are measured at the microscope, it is important to report the number of width measurements of each type to obtain an average value (i.e., N.O.Wi). We consider the collection of some of the data in Table 1 optional and have so labeled them.

All variables in Table 2 should be included in reports and papers. As mentioned earlier, when imputed MAR is applied for any member of a group, the group data should be reported both with the imputed MAR data, and with MAR treated as missing data. The cortical thickness measurements have gained in importance recently as investigators attempt to understand the contribution of cortical bone to risk of fracture and the effect of therapeutic agents that reduce remodeling and cortical porosity.

The variables in Table 3 can be calculated and may be reported. It is our experience that these variables are problematic. In some cases, their measurement is difficult and has a tendency to be biased. For cellular endpoints, inconsistent fixation may be a problem. In other cases (Tb.Th, Tb.N, Tb.Sp), better methods for their measurement exist, such as MicroCT, in which their values are calculated from 3D reconstructions. Fibrosis surface is a good index, but is only relevant for a few conditions. The periods (F.P, R.P, and Rm.P) and BFR/TV are heavily derived variables that have been of limited use in understanding how drugs work to reduce fracture rate in osteoporosis.

There are numerous additional histomorphometric variables that could be named [3] and added to Tables 1 and ​and2.2. However, the main use for histomorphometry is to understand the status of bone remodeling and mineralization. The variables in Tables 1 and ​and22 are those that are fundamental to that understanding, and thus justifying their inclusion.

Examples of measured variables used in other types of histomorphometric examinations include: Number of osteocyte lacunae per unit B.Ar, number/fraction of osteocyte lacunae containing normal osteocytes, number/fraction of osteocyte lacunae containing no osteocytes; number/fraction of osteocyte lacunae containing apoptotic osteocytes, number/fraction of osteocytes that are apoptotic, number of apoptotic osteoblasts/B.Pm, and area of woven bone/B.Ar. The publication on nomenclature [3] suggests the possibility of numerous useful calculated variables derived from the above microscope variables and others. We recommend that when needed to suit a particular examination, histomorphometrists report additional variables that are clearly-explained and appropriate for the purposes of the specific investigation.

Application of bone histomorphometry to animal models

When properly used, animal models for bone research, most notably for pharmacology studies, are considered among the most predictive models for human outcomes in any field of pre-clinical pharmacology [22]. Adult rodents (mice and rats) and adult large animals (e.g., non-human primates [NHPs: cynomolgus and rhesus], canines, sheep, etc.) are differentiated by the fact that only cortical bone of adult large animals displays Haversian remodeling like adult human cortical bone. Cortical bone of adult rodents has no ambient Haversian bone remodeling. [23,24]. Thus, the adult rodent is best used only as a model of trabecular bone remodeling.

Pre-necropsy dual fluorochrome labeling (subcutaneous calcein injection) with a five-to-seven day interlabel time period for adult rats and a two-to-five day interlabel time period for mice are routinely applied to produce double labels similar to those seen in humans. The adult ovariectomized (OVX) rat [25] is commonly used in pre-clinical pharmacology studies. Mice exhibit strain-related variation in bone behavior. They are widely-used today to study variation in bone phenotype due to genetic alterations. Adult NHPs aged 10–12 years display estrogen-deficiency-induced bone loss and are used both to evaluate cortical bone remodeling and study bone responses in the typical human osteoporotic fracture sites. Adult canines have appropriate trabecular and cortical bone remodeling, but do not display reliable estrogen deficiency-related bone loss.

Most variables used in the analysis of trabecular bone in transilial bone biopsy specimens from humans (Tables 1 and ​and2)2) not only can be measured/analyzed, but also have similar meaning in adult rodents. The exception is W.Wi. Though it can be evaluated in adult rodent vertebral trabecular bone, W.Wi cannot be readily measured in long bone metaphyseal trabecular bone because of the paucity of completed bone structural units. Fortunately, an excellent understanding of the effects of various agents (estrogens, bisphosphonates, PTH, etc.) and physiologic conditions (e.g., skeletal unloading, vitamin D deficiency, etc.) can be obtained in rodent experiments without evaluating W.Wi, by relying on MS/BS, BFR/BS, and related variables. Eroded perimeter (E.Pm) is considered more difficult to identify in rodents than in humans, making it less reliable than in humans. Therefore, osteoclast surface (Oc.S/BS) is recommended as the best index of bone resorption in rodents. Reversal perimeter, fibrosis perimeter, and erosion depth are generally not evaluated in rodents. Mineralization lag time is much shorter in rodents than in humans, resulting in very thin osteoid seams. Therefore, some variables such as osteoid thickness and volume are rarely reported in rodent studies, unless an agent impairs bone mineralization and induces osteoid accumulation. We recommend that MS/BS, MAR, and BFR/BS be handled as in humans with one exception. When double fluorochrome label is absent, since the imputed value for MAR (0.3 μm/day) discussed earlier was derived from human data; its use in animal studies is not recommended. We recommend that imputed/assigned MAR values for rodents not be used until an imputed MAR value derived specifically from rodents has been established.

The anatomical sites most frequently used for histomorphometric study in adult rats are the proximal tibial and distal femoral metaphyses. When the metaphyses are too osteopenic, something that is more likely to occur after ovariectomy and with advancing age, the lumbar vertebral body can and should be substituted. The region of interest for histomorphometric measurements intended to provide data that address adult human bone remodeling should always exclude the primary spongiosa adjacent to the growth plate, which is composed of newly-formed juvenile bone that compares poorly to adult human bone. The anatomical sites most frequently used in adult mice to provide data about adult human bone remodeling are the distal femoral metaphysis, proximal tibial metaphysis, and L1–L4 vertebral bodies. As for rats, when the metaphyses are too osteopenic (e.g., after ovariectomy and in aged mice), the lumbar vertebral body can be substituted. These sites are smaller, but generally more uniform, than regions of the human transilial biopsy specimen. The minimum acceptable Tt.Ar in rat trabecular bone is approximately 6–8 mm², while the minimum acceptable B.Pm is approximately 25 mm, both less than in the human. Histomorphometric measurements are usually performed in two sections froma skeletal site separated by at least 50 μm. The minimum acceptable Tt.Ar and B.Pm in mouse trabecular bone is approximately 60% of that in rats, due to the smaller size of murine bones. Cortical bone studies in rodents are done in cross-sections taken at either the mid-point of the femur or the region located 1–2 mm proximal to the tibiofibular junction. These areas are used for the evaluation of periosteal and endocortical surface endpoints. The internal cortex of rodents cannot be used for the evaluation of Haversian remodeling. When conducting rodent experiments involving bone, it is better to use large group sizes, rather than analyze more section area and perimeter froma small number of animals. Although we recommend minimum group sizes of 6–8 rodents for bone histomorphometric studies as a starting point, it is best for group sizes in individual experiments to be determined by a power analysis.

The anatomical sites most frequently evaluated in adult non-human primates (NHP) are the typical human osteoporotic fracture sites, the vertebral body and the proximal femur. Since these sites can never be sampled systematically in experiments involving living humans, most investigators feel that studying these sites leads to the most direct understanding of how agents work at osteoporotic fracture sites in humans. The NHP iliumis also used for a direct comparison to the usual sampling site in humans. The osteoporotic fracture sites in NHPs yield enough trabecular bone that the minimums for sampling the human transilial biopsy specimen apply. On the other hand, the NHP ilium generally yields a smaller trabecular bone sample than the human transilial biopsy specimen. Its minimum acceptable Tt.Ar is approximately 8 mm², while its minimum acceptable B.Pm is approximately 25 mm. Multiple sections at 250 μm intervals should be studied to assure that these minimums are met. Cortical bone studies of periosteal and endocortical surface endpoints in NHPs are done using cross-sections taken at the seventh rib, the mid-point of the femur, the mid-point of the femoral neck, and the center of the lumbar vertebral body. The interior of the cortex at these sites is used to evaluate Haversian remodeling. When doing NHP experiments that involve bone, the group size is often enlarged to accommodate mechanical testing of the osteoporotic fracture sites, providing a generous sample size for histomorphometry.

Extrapolating data from a transilial biopsy specimen to the whole skeleton

The trabecular bone of the ilium is a red marrow site with a remodeling rate of ~10–15%/year. When considering bone remodeling activity, it appears to be representative of similar red marrow skeletal sites, such as the vertebral body [26]. It is not representative of fatty marrow sites with their low remodeling rates. Its relationship to the proximal femur has not been studied in any detail.

Transilial bone biopsies are not the best way to evaluate bone mass. While BV/TV in large groups of individuals may actually give reasonable data and be useful in the experimental setting, the small sample for analysis is a problem. DXA (dual energy X-ray absorptiometry) and CT (computed tomography) not only sample much larger areas, but also work in the osteoporotic fracture sites and are non-invasive.

We do not believe the Haversian portion of the cortical bone of a transilial biopsy specimen can be sampled in a manner that allows calculation of reliable bone histomorphometric data, because of the problems with consistent orientation during sectioning of the specimen. The periosteal and endocortical surfaces of the two cortices are suitable for study, but tend to provide limited perimeter. The data from each of the two cortices should be collected and reported separately.

Defining and detecting “oversuppression” of remodeling

The bone remodeling rate ordinarily remains within in a certain range that maintains health of the skeleton. It is measurable by bone histomorphometric techniques and, to some degree, by biochemical markers for formation and resorption. High remodeling rates, as in postmenopausal women, patients with primary hyperthyroidism, and some cases of renal osteodystrophy, cause bone loss, decreased bone strength, and increased fracture risk. These elevated rates have been measured by both biochemical markers of formation and resorption and bone histomorphometric techniques. The well-known age-related rise in fracture risk is associated with increased remodeling as expressed by activation frequency, particularly around menopause. It is, thus, not surprising that anti-remodeling agents work by a tissue-level mechanism that decreases activation frequency, and reduces fracture risk.

Low bone remodeling rates represent the other side of that coin. It must also be remembered that defining and detecting “oversuppression” of remodeling are two very different tasks. In theory, oversuppression is defined as a condition (perhaps caused by a pharmacologic intervention) that results in levels of remodeling activity low enough to cause harm to the individual in the form of increased risk of fracture, much in the same way that high levels of remodeling activity are associated with increased risk of fracture. Though speculation about the existence of ultra-low remodeling rates that might be called oversuppression has emerged recently, no skeletal adverse consequences have been linked convincingly to such ultra low remodeling rates. In the absence of a clear relationship of low bone remodeling rate to adverse outcomes, we feel that it is difficult to define a specific value for rate of bone remodeling that can be considered “oversuppression.” That being said, we further believe that the detection of a low remodeling state is problematic, given the tools available for judging remodeling rates. As we have stated, while transilial bone biopsy-based histomorphometry can be used to detect a range of bone remodeling activity, when BFR/BS and Ac.f are reported as zero, the correct interpretation is that remodeling is low, but not necessarily, or not even likely, zero, anywhere in either the specimen or the skeleton. Furthermore, individuals who have BFR/BS and Ac.f reported as zero tend to display a range of bone biochemical marker values from normal to low. Importantly, the biochemical marker values rarely or never reach zero.

One method of analysis that can be used when there are no fluorochrome labels in a biopsy specimen is illustrated in Table 4. We have recommended that a finding of no double label in the standard sampling areas of an individual patient always be taken as a certain signal of a low remodeling state, but never be taken to mean that the patient completely lacks remodeling. We have also recommended that the main method for presenting the data from such patients in studies with multiple treatment groups, is that they be included for MS/BS (value of zero) and BFR/BS (value of zero) when no label of any kind is found, but excluded from other important calculations such as mineral apposition rate (MAR) and all others that depend upon MAR, due to the inability to calculate MAR. Excluding non-double label patients from calculations of central tendency, creates an inaccurate upward bias in the group BFR/BS and Ac.f values (less likely for MAR because it is so constant), in groups containing patients without double label.

There is a way to compensate partially for this bias. The number of patients in a group excluded from MAR and BFR/BS calculation due to lack of double label can serve as a useful index of group remodeling state. Based on published data [8,27], investigators can and should determine whether a treatment group has more subjects without double label than: a) any logical comparator group within the study, b) healthy premenopausal women [8], and c) untreated post-menopausal osteoporotic women [27].

In Table 4, we present group sizes and the minimum numbers of subjects who lack double label in the test group to establish significant differences (P<.05 and P<.01). We also add a comparison that uses typical group sizes for pre-clinical studies. When the difference is significant, a significantly low remodeling state exists in the test group. We use a typical group size of 30–32 for human osteoporosis studies. For example, it is of interest to know whether a treated group has lower bone formation rate than healthy pre-menopausal women. If three subjects have no double label, remodeling rate is likely to be significantly lower than in premenopausal women (P<.05). If five subjects lack double label, remodeling rate is even more likely to be significantly lower than in premenopausal women (P<.01). We use the group sizes presented [8,27] for the published data. This analysis was done with the Chi Square test, and can be performed similarly for any comparator group size.

Acknowledgments

Footnotes

^☆Presented at the Fortieth International IBMS Sun Valley Skeletal Tissue Biology Workshop; August 1, 2010.

References