Best Practices for Evaluation of Bone Marrow in Nonclinical Toxicity Studies

Regulatory Guidance

Few detailed recommendations are available from regulatory agencies on the assessment of the hematopoietic system. Nonclinical toxicity evaluation of test articles must usually include evaluation of a rodent and a nonrodent species. Therefore, pathologists must be familiar with the normal microarchitecture and cytological features of the bone marrow of multiple species, including but not limited to the mice, rats, dogs, and nonhuman primates. Concurrent controls should be evaluated together with the appropriate test article–treated groups. Successful evaluation of the hematopoietic system will depend on examination of good-quality samples by an experienced individual correlating the findings in the hematologic, histological, toxicokinetic, and in-life data as well as cytological and or flow cytometric data, if available.

Histological Evaluation

Histological evaluation alone will provide an assessment of overall hematopoietic cellularity, can identify subtle focal or multifocal lesions not easily identified by smear evaluation (i.e., necrosis or inflammation), and may allow for a general estimation of the proportion and maturation of granulocytic and erythroid cells, as well as megakaryocyte cellularity. Histological evaluation may provide data not obtained by cytologic evaluation, such as changes to the structural organization of the hematopoietic environment, endosteum, bone, interstitium, adipose tissue, and vasculature.

Cytological Evaluation

The need for cytological evaluation of bone marrow smears should be decided on a case-by-case basis. If the evaluation of the combined study data indicates that characterization of the hematopoietic system was, or will be, inadequate using hematology and histopathology, then cytologic evaluation of the bone marrow may be considered. Noteworthy situations that may warrant cytological examination of bone marrow smears include: a need to differentiate early hematopoietic precursors, the desire to investigate incompletely characterized increases or decreases in hematopoietic cellularity of the bone marrow relative to the peripheral blood cell numbers, and the requirement to determine whether histological changes in bone marrow cellularity reflect alterations in lymphoid versus erythroid cells. Cytological assessment may also be considered if there are uncharacterized atypical or abnormal blood cells in circulation, or abnormalities in erythrocyte indices (mean corpuscular volume [MCV], mean corpuscular hemoglobin concentration [MCHC], etc.) suggestive of abnormal erythropoiesis. Cytologic evaluation may not be necessary when decreased bone marrow cellularity is caused by decreases in food consumption or when peripheral blood lymphocyte counts alone are decreased. Cytological examination also may not be necessary when there are appropriate (responsive) increases in peripheral blood cell counts and/or bone marrow cellularity, or if there are granulocyte or platelet dysfunction disorders.

Flow Cytometric Evaluation

Flow cytometry can be used in addition to or instead of cytological examination to further characterize changes in the bone marrow. Indications for flow cytometric examination include criteria similar to those stated above for cytological assessment. However, if the changes present in the bone marrow need detailed morphological assessment, then cytological examination would be the preferred technique.

Differentiation of Early Hematopoietic Precursors

Histology alone is insufficient to differentiate between all types of immature hematopoietic precursors. Therefore, if differentiation or identification of early stages of cellular development is deemed necessary by the pathologists in consultation with the study director, then evaluation of bone marrow smears may be considered. Examples of situations in which detailed evaluation of early precursors may be considered include atypical morphology of leukocytes that is unrelated to inflammation, sepsis, antigenic stimulation, known pharmacologic effect of the drug, and/or drug-induced phospholipidosis (vacuolation), and cases of suspected drug-induced hematopoietic neoplasia that are not sufficiently classified by evaluation of blood smears, hematology data, and/or histopathology.

Additional Characterization of Effects on Hematopoietic Cell Lines

Significant decreases in single or multiple blood cell lines may warrant cytological examination of the bone marrow if routine evaluation of hematology and histology data was insufficient to characterize the effect. Examples include significant decreases in blood neutrophil counts in a nonclinical toxicity study, with no evidence of an underlying response to acute tissue inflammation or known antiproliferative effect of the test article; marked, persistent, and/or consistent decreases in peripheral eosinophil counts with no evidence of physiological stress or glucocorticoid-like effects; significant decreases in platelet counts with no evidence of an underlying use, sequestration, or destruction of platelets; or marked, persistent, and/or consistent decreases in monocyte counts. Significant increases in blood cells may also warrant cytological examination of the bone marrow if routine evaluation of hematology and histology data was insufficient to characterize the effect. For example, significant increase in the numbers of red blood cells (absolute erythrocytosis) that is not caused by dehydration or hemoconcentration, pharmacology (i.e., recombinant erythropoietin, hormones, cytokines), or known modulation of erythropoietin production by the test article; marked increases in neutrophils that are not caused by inflammation or necrosis, excitement/stress response (i.e., catecholamine/glucocorticoid-related), or administration of pro-inflammatory cytokines; and persistent or marked increases in platelet counts that are not caused by increased muscular activity, blood loss, iron deficiency, inflammation, and/or administration of growth factors or cytokines.

Determining Whether Changes in Bone Marrow Cellularity Result from Changes in the Erythroid versus Lymphoid Population

Although it is possible to differentiate some granulocytic cells from erythroid cells on histological samples, differentiation of the erythroid series from the lymphoid cells is more challenging. This differentiation is of increased relevance in rodents since these species have greater numbers of lymphoid cells than do dogs or primates. Therefore, if differentiation of erythroid versus lymphoid precursors is deemed necessary, then bone marrow cytology should be considered.

Changes in Erythrocyte Indices or Red Cell Morphology

Changes in red cell indices such as MCV and MCHC may suggest abnormal erythropoiesis or neoplasia and may be accompanied by large or megalocytic erythroid cells in the blood. If detection/confirmation and detailed characterization of abnormal erythropoiesis (e.g., as suggested by megalocytes or megaloblastic changes) or other red cell developmental morphologic abnormalities (such as siderocytes or basophilic stippling) is deemed necessary, then bone marrow cytological assessment may be appropriate, since compared to the peripheral blood, large numbers of these abnormal erythrocytes may be found in the bone marrow.

Atypical Changes in Leukocyte Morphology in the Peripheral Blood That Are Unrelated to Inflammation (i.e., “toxic” changes), Antigenic Stimulation (i.e., lymphoid reactivity), or Known Effects of the Drug (i.e., vacuolation caused by phospholipidosis)

Atypical leukocytes observed in peripheral blood may be indicative of abnormal leukocyte maturation, neoplasia, and/or immune modulation. Atypical cells may be characterized by immaturity, altered size, abnormal cytoplasmic vacuolation, altered nuclei/cytoplasmic ratios, asynchrony of the nucleus and cytoplasm, altered chromatin, alterations in stain affinity, the presence of atypical granules, and/or other cytological alterations. In these situations, bone marrow cytological evaluation may help characterize the extent of the test article effect in hematopoietic precursors.

Changes in Bone Marrow Cellularity Occur Concurrently with Decrements in Food Intake

Decreased food intake may have a major impact on the hematopoietic system, particularly in rats, and these effects should not be confused with direct test article–related findings. Body weight and food consumption data should be evaluated in conjunction with hematological parameters and bone marrow histopathology. Several published studies have shown that the effect of decreased food consumption on the hematopietic system is proportional to the degree of food restriction and the length of the study. For example, 25% to 75% food restriction studies in rats (Wistar or Sprague Dawley) demonstrated marked changes in hematopoietic cell production. In studies lasting from two weeks to two years, important decreases in leukocytes (neutrophils, lymphocytes, and monocytes), decreases in reticulocytes and platelets, and reduction of erythroid precursors in bone marrow were observed, along with decreased body weight (Ogawa et al. 1985; Seki et al. 1997). In rats subject to significant food restriction (75%), necrosis and bone marrow degeneration were observed after only two weeks (Levin et al. 1993). In mice, decreased protein intake was associated with bone marrow cellular depletion and a halt of the precursor cell cycle (Borelli et al. 2009). Effects of food restriction on the hematopoietic system have been documented less often in other species, but similar effects of lower magnitude have been observed in dogs (Hill et al. 2005; Lawler et al. 2007).

Changes in Peripheral Blood Lymphocyte Numbers

Bone marrow lymphocyte density does not correlate well with peripheral lymphocyte counts (Yoffey and Courtice 1970), and the number of lymphocytes in peripheral blood may be influenced by rates of production, recirculation and use, or destruction of lymphocytes. Recirculating lymphocytes spend more time in specific organs than in peripheral blood, hence, increases or decreases in peripheral lymphocyte numbers do not necessarily reflect altered lymphopoiesis. Therefore, bone marrow cytological evaluation will likely not add value when assessing most changes in peripheral lymphocyte counts.

Responsive or Regenerative Changes in the Bone Marrow or Blood

In these situations, evaluation of hematologic and histopathologic data should be sufficient for assessment of the hematopoietic system. For example, increased cellularity of the bone marrow is often observed as an appropriate response to increases in cell turnover. Decreased numbers of red blood cells in the blood may be accompanied by histologically recognizable increases in the proportion or number of erythroid cells in the bone marrow (i.e., erythroid hyperplasia). Increases in the number or proportion of myeloid cells (i.e., granulocytic hyperplasia) is often associated with inflammation. Increases in the number (i.e., megakaryocytic hyperplasia) or changes in the morphology of megakaryocytes can develop in association with peripheral consumption of platelets and inflammation.

Platelet or Granulocyte Dysfunction

Platelet or granulocyte dysfunction are rarely associated with morphologic changes, so cytological assessment of the bone marrow would be of little value (Andreasen and Roth 2000; Boudreaux 2000). Platelet dysfunction is typically characterized by increased bleeding time with concurrent normal platelet counts, and normal coagulation times. Platelet aggregation testing and/or immunological evaluation of surface adhesion molecules are possible methods to further characterize platelet functional deficiencies. Acquired drug-related granulocyte functional defects may be caused by decreased adherence, chemotaxis, phagocytosis, or degranulation of leukocytes. Typically animals have persistent and progressive peripheral neutrophilia with hypersegmented neutrophils, decreased neutrophilic tissue infiltration, and recurrent pyogenic infections. The bone marrow granulocytic compartment is usually increased, and bone marrow histopathological assessment together with the hematology evaluation should be adequate to evaluate the bone marrow. Evaluation of neutrophil surface markers and/or granule content may also be useful.

Bone Marrow Collection and Processing

Histological Evaluation of the Bone Marrow

It is essential that the pathologist has a good working knowledge of bone marrow microanatomy in the test species and performs a systematic examination of all bone marrow components to achieve a thorough evaluation. It is also important that bone marrow from animals in the treated groups is compared to concurrent controls. Initially, the entire section should be scanned at low magnification (i.e., 40×) for a general evaluation of section and staining adequacy. At this point, bone marrow sections are evaluated for adequate size, composition (e.g., composed primarily of cortical bone with associated marrow), and evidence of crush or other artifactual change. At this magnification, an overall estimation of bone marrow cellularity (comparing the approximate proportion of hematopoietic to adipose tissue), including the distribution and numbers of megakaryocytes, should be performed. Bone abnormalities and focal marrow lesions (e.g., metastatic infiltrates, granulomas) may also be apparent at this magnification.

Following the initial low-magnification review, hematopoietic and stromal components, hemosiderin, and lesions observed on low magnification should be evaluated at higher magnifications (100×–400×). Trabecular bone thickness should be examined along with the presence and numbers of osteoblasts, osteoclasts, and Howship’s lacunae, particularly if the bone marrow section is the only bone tissue to be evaluated in a study.

Overall hematopoietic cellularity should be evaluated, and a general estimate of the proportions of granulocytic to erythroid cells (or myeloid to erythroid-lymphoid cells in rodents) in control and test article–treated animals should be assessed. This step should be followed by an examination of the individual hematopoietic tissue cell lineages. In paraffin–embedded, H&E–stained sections, some stages of the erythroid and granulocytic cells, adipocytes, macrophages, and megakaryocytes can be readily identified. However, hematopoietic stem cells, very immature granulocytic and erythroid cells, mast cells, lymphocytes, monocytes, and stromal cells may not be easily identified because of low cell numbers or indistinctive morphology. Generally, detailed evaluation of individual cells is better performed by light microscopic examination of bone marrow cytology preparations.

The granulocytic and erythroid series should be evaluated for cell morphology, relative proportions of immature and mature precursors, synchrony of maturation, and evidence of dysplasia or neoplasia. Abnormalities in the relative proportions of the specific granulocytic lineages (neutrophilic, eosinophilic, basophilic) should be evaluated. Megakaryocytes should be examined for overall cellularity, location, presence of clusters, and morphology, including size and nuclear lobularity.

Because of small size and difficulty differentiating from erythroid precursors, unequivocal identification of individual lymphoid cells in decalcified, H&E–stained, paraffin-embedded sections of bone marrow is not readily accomplished. Despite the difficulty in visualizing individual lymphoid cells, aggregates of lymphocytes are straightforward to detect and have been reported in marrow sections of healthy dogs (Weiss 1986), in nonhuman primates with retroviral infections (Lowenstine 2003), and in wild-caught nonhuman primates (Ito et al. 1992). Lymphoid cell aggregates or lymphoid follicles are not usually found in normal rodents (Frith et al. 2000; Geldof et al. 1983).

For further detailed discussion of histological bone marrow examination, the reader is directed to other reviews on bone marrow evaluation (Andrews 1998; Bain et al. 2001; Buckley 1995; Elmore 2006; Travlos 2006b; Valli et al. 2002; Wickramasinghe et al. 1992).

Reporting of Histopathology Findings

The occurrence of background or incidental findings may be recorded in the individual animal data if desired, but in general, such data are not included in the pathology narrative unless the pathologist is unsure as to its relationship to test article administration. Samples of inadequate quality for evaluation should be recorded as such in the individual animal data together with a brief indication as to why it was nondiagnostic (for example, autolysis, not collected, etc.). The pathology narrative should indicate whether evaluation of the bone marrow at any given dose(s) was not possible because of too many nondiagnostic samples.

Test article–related changes should be recorded and graded in the individual animal data. As recommended in previous publications (Elmore 2006; Haley et al. 2005), whenever possible, treatment-related changes should be recorded in the individual animal data in descriptive or semiquantitative terms (e.g., decreased granulocytic cellularity, decreased hematopoietic cellularity, increased CD20 immunoreactivity) instead of clinical, interpretive, or diagnostic terms (e.g., “erythroid hyperplasia,” “granulocytic hyperplasia,” “aplasia”). Because multiple hematopoietic cell lines coexist in the bone marrow, it may be difficult to differentiate a true decrease in the absolute numbers of a particular lineage from a change in the proportion relative to the overall cellularity. An example of such a situation is when there is an increase in the proportion of granulocytic cells and the pathologist must try to determine whether it results from an increase in the numbers of granulocytic cells, a decrease in the numbers of erythroid cells, or both. A descriptive approach to recording the data would be encouraged in this situation. Interpretive terms such as atrophy, hyperplasia, hypoplasia, and so on may be reserved for the pathologist’s interpretation of the bone marrow findings within the context of the study in the narrative section of the report. The pathology narrative should summarize, describe, and interpret treatment-related changes and should be in context of other observations, including hematological data, in-life findings, and, if available, cytological or flow cytometric bone marrow assessment, in concurrent controls and treated animals.

Bone Marrow Harvesting

Bone marrow is typically collected at necropsy, following anesthesia and euthanasia. Bone marrow cells should ideally be the first sample collected at necropsy to optimize preservation of cytological detail. Because the marrow will quickly begin to clot and bone marrow cells will start to undergo autolysis, cytomorphology will be optimal if bone marrow specimens are collected within five minutes after euthanasia (Smith et al. 1994; Valli et al. 2002). However, a laboratory may want to determine the time after death, which may vary by species and which will provide adequate quality of bone marrow smears to perform cytological evaluation. The femur and sternum are acceptable sites for bone marrow collection from most small animal (rodent) species. The bone should be removed and, if femur is used, it should be opened longitudinally to expose the marrow cavity. It is important to visualize the marrow to ensure collection of active marrow, which is red, rather than yellow, less-active (fatty) sites. In older animals, especially nonrodent species, active marrow may not be present in the mid-shaft or distal femur. In larger animals (dog, cat, primate, rabbit, swine), the rib or sternum is a preferred site for bone marrow cytology because of the more uniform activity of the marrow at these sites and the ease of collection. The rib or sternum should be removed from the carcass and opened along the long axis to expose the marrow cavity. Once open, the bone is then squeezed with forceps, pliers, or a similar instrument to extrude some marrow from the exposed surface. Slides should be made immediately. Other collection sites may be successfully used depending on the experience and skill of the necropsy team. Although proximal femur may be used in large animals, one needs to be aware of the potential artifacts created by the harvesting technique, such as heat from a bone saw, and the longer time required to access this site to obtain active marrow.

Because of the rapid loss of cytological integrity after death, collection of bone marrow for cytology from animals that have died on study (found dead) is without value. Bone marrow cytology smears obtained from moribund animals should be examined judiciously since the cytology may reflect moribund-related changes, which could be mistakenly concluded to be test article related rather than a consequence of physiological failure.

Bone marrow collection from live animals is rarely required and is a consideration only for large animals, such as dogs. A number of references may be consulted for information regarding collection techniques (Freeman 2000; Harvey 2001; Smith et al. 1994).

Slide Preparation Techniques and Staining

There are various techniques that, with practice, result in acceptable and representative bone marrow smears. The choice of technique may depend on the experience of the individuals who are collecting the samples. With practice, any of these techniques will produce good-quality slides. With any of the following techniques, excess diluent should be avoided, since doing so will cause background staining and will prolong the drying process. Prolonged drying from excess diluent or unduly thick samples will result in cell condensation, which compromises subsequent cell identification. Excessive cell lysis may be caused by too little diluent or when samples have started to clot. Acceptable performance of the collection method in the species of interest should be demonstrated by the individuals who will collect the samples prior to use in a study.

Paint Brush Technique

Clean, natural-bristle (sable) brushes are used to obtain the bone marrow sample and apply it to the slide. The brush is dipped in 5% bovine serum albumin (BSA), allogenic serum, or fetal calf serum mixed 2:1 with 7.5% EDTA to slightly wet the bristles and serve as a diluent. Small (e.g., 1 mL) aliquots of a diluent can be frozen in individual vials and thawed as needed for studies. The brush bristles are blotted to remove extra diluent and create a narrow tip. The tip then is used to obtain some marrow from the exposed cavity, which is then applied to a clean glass microscope slide in 2–4 wavy lines (Valli et al. 2002; Figure 1 ).

OPEN IN VIEWER

Figure 1.

Bone marrow smear preparation: paint brush technique.

To a limited degree, the size of the animal should dictate the size of the brush needed. There is no uniform size numbering system for brushes among vendors, so the size will need to be confirmed prior to conducting a study. It is necessary to use a very small brush to access the marrow cavity of a mouse, which limits the amount of bone marrow applied to a slide. A larger brush should be used for other species. To prevent cross-contamination between animals, separate brushes should be used for each animal, or brushes must be rinsed thoroughly with distilled water to lyse residual cells, and then meticulously dried prior to use on another animal during necropsy.

Push Slide Technique

This technique is similar to that used to make cytology or hematology slides. First, one drop of diluent (as described above) is placed at one extremity of a glass slide. Marrow is collected as described above or with a small forceps and gently mixed with the diluent on the slide. The suspension is then spread in an even film using a second glass slide (held at an angle) as a spreader. The spreader slide is slid forward the length of the slide to produce a smear (Provencher Bolliger 2004; Valli et al. 2002; Figures 2 and 3 ).

OPEN IN VIEWER

Figure 2.

Bone marrow smear preparation: push slide technique.

OPEN IN VIEWER

Figure 3.

Bone marrow smear preparation: squash or pull technique.

Cytological Evaluation

It is critical that the person cytologically evaluating the bone marrow smears is an experienced evaluator. Bone marrow smears should not be examined in a blinded fashion, as it is necessary to consider the variability and establish a baseline for control animals in a specific study before it is possible to critically evaluate smears from treated animals. Samples are usually collected on all animals in the study. All groups may be evaluated, or one may choose to evaluate animals in the control and high-dose groups and then proceed down to mid- and low-dose groups until a NOEL is determined. Bone marrow cytology slides must be shown to be of adequate quality (i.e., adequate area(s) with uniform staining of cells arranged in a monolayer) in order to confidently interpret that the specimens reflect the physiological state of the animal and not extensive artifactual alterations that might have been induced by sample collection or slide preparation techniques. Slides that are of inadequate quality should not be evaluated.

The bone marrow cytological preparation should be scanned at low magnification (40×–100×) to determine areas of appropriate cell density and adequate staining to perform a cytological evaluation. Detailed cytological evaluation of hematopoietic cells should be done under oil immersion magnification (500×–1,000×) to assess the cytomorphology of the hematopoietic cells present and to assess adequacy of differentiation in each lineage (Ryan 2001). All the major lineages should be evaluated. Megakaryocytes are not enumerated, but their morphology can be assessed. The cytological bone marrow assessment may be done by either a qualitative assessment, similar to the histological evaluation, or by a quantitative method. The decision of whether qualitative or quantitative assessment must be performed for a particular study should be made by the study clinical pathologist in consultation with the study director and other scientists who are knowledgeable of the activity, toxicologic effects, and target of the test article. These methods are outlined below. All of these methods can be used successfully to further characterize the changes noted histologically, but they give different levels of detail of this assessment.

A qualitative assessment of cell maturation and morphology is key and is the foundation of the quantitative assessment. For some studies, the qualitative assessment may be used as the only method of evaluation, but it should also accompany each type of quantitative evaluation. Qualitative assessment alone in lieu of a quantitative assessment can be useful to further characterize the histological change. In addition to abnormalities in morphology, the relative proportions and maturation of the different cell lineages can be assessed. Qualitative assessment can also be used to obtain valuable information when there is an adverse effect on overall cellularity, or when one or several cell lines is present in very low numbers, or when cytological changes preclude definitive cell classification.

For all quantitative approaches, it is recommended that 300 to 500 nucleated cells be evaluated (Ryan 2001). Enumeration of 500 cells per slide may be preferable when there are few animals in each treatment group. In rodent studies in which group sizes are typically larger (ten/sex/group), enumeration of 300 to < 500 cells may be sufficient. Bare nuclei or lysed cells should not be included in the numerical evaluation.

If the definitive quantitative approach is chosen, the nucleated cells should be categorized into granulocytic, erythroid, and lymphoid lineages. Identification of all maturation stages for neutrophilic and erythrocytic precursors is suggested to provide a basis for determining whether there is a shift to less-mature (left) or more-mature (right) stages. A myeloid (granulocytic)-to-erythroid (M:E) ratio can be calculated from the quantitative differential count. It is not usually necessary to identify developmental stages for eosinophilic, basophilic, lymphoid, or other cell types. A modified quantitative approach is to do a partial differential, which categorizes the cells into the major categories (lymphoid, erythroid, and granulocytic) but does not enumerate each specific maturation stage for the erythroid or granulocytic cell type.

In addition, a further differentiation may be used to classify cells into proliferative and maturation (nonproliferative) phases of development. For granulocytic cells, myeloblasts, promyelocytes, and myelocytes are categorized into the proliferative population, whereas the metamyelocytes, band cells, and segmented cells are considered part of the maturation phase. Erythroid cells also may be categorized into proliferative phase (rubriblasts, prorubricytes, and basophilic rubricytes) versus maturation phase (polychromatophilic rubricytes and metarubricytes). Because partial differential evaluations group several developmental stages together, this approach is not as complete as a full differential but is often adequate to characterize the major cytological findings in the bone marrow. The proportions of proliferating versus nonproliferating cells can be used to calculate an erythroid maturation index, myeloid maturation index, and overall maturation index. Maturation indices are sometimes used for semiquantifying left or right shifts in hematopoietic cell maturation.

Reporting of Cytological Data

A report of bone marrow cytology should include a statement of specimen adequacy, quantitative and qualitative data that were generated during the evaluation, and an interpretation by a qualified and experienced clinical pathologist that correlates the bone marrow cytology findings with all of the study data. As with histology, samples of inadequate quality for evaluation should be recorded as such in the individual animal data together with a brief indication as to why the evaluation was nondiagnostic (for example, excessive “smudge” cells, not collected). The clinical pathology narrative should indicate whether evaluation of the bone marrow at any given dose(s) was not possible because of too many nondiagnostic samples. Use of diagnostic interpretive terminology (hypocellularity, hyperplasia, etc.) should be avoided or used very judiciously. Instead, morphological descriptions of changes, such as increases or decreases relative to control, should be favored over the use of diagnostic interpretive terminology. Often the spectrum of changes in nonclinical toxicity studies may not fall into a well-described or established clinical category and the use of clinical interpretive terminology could be misleading. Statistics sometimes may be used to assist in identifying effects. The clinical pathology narrative should summarize, describe and interpret treatment-related changes in the context of other findings including hematology data, histopathology changes, and in-life findings.