In childhood, bone marrow fills all cancellanous marrow spaces including long bones (Fig. 2b). The transition to adult begins at adolescence when there is accelerated skeletal growth and the involution occurs with recession in long bones except for proximal portions of humerus and femur. The adult marrow tends to be more central in the skull, vertebrae, pelvis, ribs, clavicle, and sternum. Although variations occur, the major functional sites are the pelvis and vertebrae that make up approximately 60 % of the total body marrow. The ribs, sternum, skull, scapulae, and proximal portions of the femur and humerus also contain functioning marrow (Atkinson 1962). In children, the humerus, femur, and other long bones are active, but marrow retraction from the peripheral (appendicular) toward central (axial) skeleton, and from diaphyseal to metaphyseal in individual long bones gradually occurs, so that by age 20 years the mature adult distribution pattern is present (Fig. 2b, c).

The geographic distribution of the bone marrow is particularly relevant to understanding radiation effects (Fig. 2b, c). Although variations occur, the major functional sites are the pelvis and vertebrae. In the past, this was determined based upon semiquantitative histopathologic studies, but recently, FLT-PET imaging has been utilized to determine marrow distribution (Hayman et al. 2011). In addition to the pelvis and vertebrae, the ribs, sternum, skull, scapulae, and proximal portions of the femur and humerus contain virtually all of the functioning marrow (Hayman et al. 2011). In children, the humerus, femur and other long bones are active, but marrow retraction from the peripheral (appendicular) toward central (axial) skeleton, and from diaphyseal to metaphyseal in individual long bones occurs. By age 20 years, the mature adult distribution pattern is present (Custer and Ahlfedt 1932).

In the adult body, there are two kinds of hemopoietic tissues: red bone marrow and lymphatic tissue (Fig. 3b, c). The former produces erythrocytes, leukocytes, and platelets, and the latter is the source of lymphocytes. The red bone marrow is composed chiefly of sinusoids and three cell types: reticular cells, adipose cells, and hemopoietic cells (blood-forming cells). Along with reticular fibers (not shown in Fig. 3), the reticular cells, which have ovoid nuclei and several processes, form a framework of reticular tissue in which the hemopoietic cells are contained. Adipose cells are the largest cells in the bone marrow, occupying more and more of the bone marrow cavity with age. Hemopoietic cells differentiate into erythrocytes and leukocytes. They are in different developmental stages, and some of them are seen undergoing mitotic division (in mitosis). In a marrow section, stages of hemopoietic differentiation are difficult to see, but they can be recognized in the bone marrow smear (Zhang 1999).

The formation of the hematopoietic cells of bone marrow are in different stages of development from both the erythrocytic series and the granulocytic series. The erythrocytic series contains the proerythroblast, basophilic erythroblast, polychromatophilic erythroblast, orthochromatophilic erythroblast, and reticulocyte; the granulocytic series contains the myelocytes and metamyelocytes. These cells are drawn from a human bone marrow smear stained by the Wright method. The characteristic features of these cell types are briefly described (Zhang 1999) (Fig. 3c).

Recently, both vascular and osteoblastic niches have been described in the murine marrow microenvironment. Whether such distinct niches are operative in human marrow is uncertain as direct studies of microenvironmental function are more difficult in humans. Hematopoietic stem cells are found in close proximity to osteoblasts which are affected by parathyroid/parathyroid receptor interactions. When murine stroma is treated with parathyroid hormone, it secretes increased stem cell factor (SCF, c-kit ligand), stroma derived factor (SDF)-1 alpha/CXCL12, and IL-6. Long-term culture initiating cells and hematopoietic stem cells are also increased (Calvi et al. 2003). C-terminal PTH also increases vascular endothelial growth factor expression in osteoblastic cells (Esbrit et al. 2000), which could also impact hematopoiesis. These effects can be blocked with gamma-secretase inhibitors, suggesting that Notch and its ligands, Delta and Jagged, play a role in PTH effects (Weber et al. 2006). An interesting laboratory study that supports stromal contribution was the observation of past radiation of the lower extremity that iron uptake increased at 3 months, however, the bone marrow was ablated histologically. A radioautograph of the tibia revealed the iron was concentrated in the bone cortex in 3 months. Over the next 3 months the bone marrow regenerated as a sleeve juxta-opposed to the cortex before following the entire bone marrow cortex (Rubin et al. 1997).

Human marrow stroma contains pluripotent mesenchymal progenitor cells that give rise to many mesenchymal lineages, including chondroblasts, adipocytes, myoblasts, ligament, tendon, or osteoblasts. The differentiation of these cells toward a specific lineage is dependent on humoral and local factors activating specific transcription factors. The family of transforming growth factors, for example, may promote osteoblast differentiation and inhibit adipocyte conversion of rat marrow stromal cells in vivo (Ahdjoudj et al. 2004). The role that angiogenesis plays in marrow toxicity from chemotherapy and radiation is just beginning to be understood. Increased microvessel density in marrow of multiple myeloma, acute lymphoblastic leukemia, acute myeloid leukemia, myelodysplasia, and myeloproliferative disorders is well-described. Targeting these vessels with anti-angiogenesis agents will influence the nutrient and oxygen supply of tumors and may also reduce the growth stimuli provided by the endothelial cells. How angiogenesis protects or sensitizes to chemotherapy and radiation toxicity and effectiveness remains an area of active investigation (DeRaeve et al. 2004).

Although the alterations in Complete Blood Counts (CBCs) occur with most regionally localized cancers (brain tumors, head and neck cancers, lung and breast cancer, digestive tract, gland cancers, pelvic malignancies such as prostate and cervix), multiple factors including combined modality treatment, associated bleeding, and suppression of CBC elements occur (Fig. 6a). Figure 6a illustrates the classic figure of peripheral blood elements being depressed from time of total body irradiation (TBI) with variations depending on the dose and volume of bone marrow irradiated.

Severely granulocytopenic (or neutropenic) patients are generally considered to be those with an absolute granulocyte count (consisting of poly-morphonuclear cells and band forms) of 500/μL or less. These patients are at substantial risk for infection (David 2003). Counts between 500 and 1000/μL and declining due to antineoplastic therapy are considered by most oncologists to represent moderately severe neutropenia. Most patients clinically tolerate granulocyte counts between 1000 and 1500/μL.

Thrombocytopenia has been strictly defined as a platelet count below 140000/μL, but the manifestations of bruising, petechiae, and mucosal bleeding rarely occur with counts of 100000/μL or above, and in fact are infrequent unless counts fall below 20000/μL (Beutler 1993). Intracranial hemorrhage is particularly rare, and most episodes occur at counts less than 10000/μL.

Anemia prompting red blood cell transfusion is even more difficult to define because various clinical judgments are used. In practice, hematocrits as low as 23–25 % are tolerated by most patients. Those with cardiorespiratory difficulties may not tolerate hematocrits below 30 %.

As noted above, chronic marrow injury following cytotoxic therapy is poorly understood. The hematopoietic system appears to recover promptly after sub-ablative doses of chemotherapy and radiotherapy. However, heavily treated patients are known to have a reduced tolerance to additional therapy, showing lower nadirs of peripheral counts, particularly platelets. Decreased marrow functioning as reflected by marrow scanning after radiotherapy (Rubin et al. 1988; Rubin et al. 1973), or the CFU-GM pool after chemotherapy (Lohrmann and Schremi 1988) has been demonstrated up to 5 years following therapy. Hypoplastic and myelodysplastic syndromes have been observed at late intervals, but may represent a mutagenic or carcinogenic event.

The acute and chronic consequences of bone marrow injury by cytotoxic therapy can be considered in the context of these mechanisms. However, dissecting out the most relevant variables can be difficult due to limitations in the assessment of structure and function. Peripheral blood counts fail to demonstrate the true extent of marrow suppression or its likelihood of tolerating additional cytotoxic therapy because of the ability for the bone marrow to transiently compensate for injury. Progenitor cell cultures (Table 4) such as colony-forming units for granulocytes, macrophages (CFU-GM), burst-forming units for erythrocytes (BFU-E), colony-forming units for granulocytes, erythrocytes, macrophages, megakaryocytes (CFU-GEMM); histopathology (bone marrow aspirate and biopsy), radioisotopic scanning (¹¹¹In ⁵²Fe, ^99mTc-sulfur colloid reflecting the myeloid, erythroid, and reticuloendothelial compartments respectively); and stromal cell cultures such as colony-forming units for fibroblasts (CFU-F), are all used to evaluate various quantitative or functional aspects of the bone marrow but are all limited in scope. The long-term culture initiating cell assay and the NOD-SCID reconstituting cell assay can measure early multipotential progenitor cells (Bonnet 2003). Most of these assays are not helpful in real time as most except for peripheral blood counts take weeks to conduct and analyze. Levels of Flt3 ligand in plasma have been postulated to serve as a bio-indicator for radiation-induced aplasia (Bertho et al. 2001). There are some reports that nuclear magnetic resonance imaging may be useful to assess marrow damage from chemotherapy and radiation (Ollivier et al. 2006), and some of these changes may mimic metastases, so caution is advised in their interpretation in settings of malignancy since paratrabecular fibrosis and inflammatory cell infiltration after radiation can generate a T1 hypointense/T2 hyperintense signal on MRI (Kanberoglu et al. 2001). Studies of the use of positron emission tomography are also being evaluated as means to assess marrow function and health (Table 4) (Hayman et al. 2011).

Furthermore, the ability of all these assays to demonstrate irradiation effects is particularly marginal because of volume and dose considerations. Small segments or large compartments of previously active marrow may become aplastic, hypoplastic, or hyperplastic, and previously quiescent areas may become active (Rubin and Scarantino 1978). Finally, chronic marrow damage from cytotoxic therapy depends on stromal or stem cell defects resulting from the underlying malignancy (e.g., leukemia) or the specific cytotoxic agents used. Normal numbers of mature hematopoietic cells probably result from an increase in stem cell cycling with amplification, and the long-term capacity for normal marrow to function or to respond to stress is not immediately apparent (Testa et al. 1985). Nevertheless, our current understanding of bone marrow damage following cytotoxic therapy is based on the aforementioned evaluations.

Another consideration in ultimate marrow damage from radiation is that as stem cells replicate through life, their telomere length is shortened. It has been noted that elevated telomerase activity and minimal telomere loss occur in cord blood during long-term cultures, but adult CD34+ cells show SCID reconstitution capacity for only 3–4 weeks with telomere shortening and low levels of telomerase (Weisel KC et al. 2004). Therefore, age at which stem cells are exposed to radiation or chemotherapy may determine extent of effect on long-term proliferation potential (Goytisolo et al. 2000). There is also interest in understanding how the timing of radiation and chemotherapy treatments relates to host circadian rhythmicity and how this might affect the toxic/therapeutic ratio (Haus 2002). Circadian variation exits in the proliferative activity of acute-reacting normal tissues like the gut and bone marrow, and a potential therapeutic gain can be realized by the chronomodulated administration of S-phase chemotherapeutic agents at a time when normal tissues are in a different cell cycle phase and may be spared (Rich et al. 2002).

Historically, it was thought that high doses of chemotherapy or radiation were required to create space for donor hematopoietic cells in allogeneic transplant situations. Recently, it has been shown that with non-myeloablative regimens, complete donor engraftment may occur. This would suggest that in cases where sufficient immune suppression allows ongoing presence of the graft, the emergence of stem cells may be based on cell cycle or other proliferative advantage as compared to a true hierarchical system (Giralt 2003).

It is not yet certain how acute versus low-dose chronic exposures differentially affect human marrow. Chromosomal painting studies in radiation workers exposed to plutonium had high frequencies or structural aberrancies with high exposures which correlated with marrow dose but not external dose. These aberrations could remain stable for decades after intake suggesting that chronic irradiation of hematopoietic precursors in marrow induced cytogenetically altered cells that persisted in blood (Livingston et al. 2006). There is also some evidence that a single exposure to a low dose of radiation might induce long-term damage with more apoptotic cells in marrow compartments 6 months after exposure than in control mice and in mice irradiated with higher doses. Marrow from mice with exposure to low doses was also more sensitive to extra in vitro irradiation (Giovanetti et al. 2003).

In humans, late effects of radiation on the hematopoietic system have been incompletely studied. Wong et al. reported long-term effects of radiation exposure on hemoglobin levels in Japanese atomic bomb survivors over a 40-year period for 1958–1998 who were estimated to have a 1 Gy marrow dose. These individuals had lower hemoglobin levels than controls at age 40 and age 80 (Wong et al. 2005). In children who were less than 6 years of age at the time of the Chernobyl accident and were followed from 1986 to 2000, a significant increase in leukemia risk was seen but all had <10 mGy to bone and most lived in the Ukraine away from the epicenter suggesting that prolonged exposure to very low radiation doses may increase leukemia risk as much as or even more than acute exposure (David 2003).

Radioisotope Scans (RIS) provide a means of dynamically assessing bone anatomically as well as functionally; and with serial RIS the status of HBM can be reassessed over time longitudinally from initial suppression to eventual regeneration and compensatory expansion to ablation. Invivo/invitro allows labeling of different compartments i.e., myeloid (¹¹¹In) erythroid (Fe⁵²) and reticuloendothelial (^99MTc) (Fig. 6b). Bone marrow distribution is often determined utilizing 18F-flurodeoxythymidine (FDT), PET scans.

MRI is preferable to CT for visualized bone marrow and the reverse is true for imaging cortical and cancellous bone. MRI can assess replacement of fat for radiation or chemotoxic ablation of HBM which is normal in bone architecture. The MRI appearance of the acute response of bone marrow to radiation injury is very similar to the changes seen after chemotherapy. Within a few days, tissue edema leads to a drop in signal intensity of T1-weighted images. The edema is then replaced by fatty transformation, which is seen as an increase in T1 signal intensity. Unlike the changes after radiation injury, however, bone marrow changes after chemotherapy are temporary. After 3–4 weeks, with reseeding of the hematopoietic red marrow, the MRI appearance of bone marrow returns to its normal low signal intensity on T1-weighted images (Husband and Reznek 1998). Unlike other normal tissue postradiation which undergoes late fibrosis, bone marrow is the only tissue to be replaced by fat, adipose cells.

If the appropriate assay is used, residual effects of all but the lowest doses of radiation therapy can be detected; their clinical importance, however, varies according to the fascinating capacity of the bone marrow to sense and compensate for such injury (Fig. 7d, e and Table 6). An appreciation of the extensive communication network that apparently exists in this organ is necessary to understand the patterns of bone marrow function following radiation therapy:

The contrasting radiosensitivities of the bone marrow stroma and hematopoietic stem cells, and the elaboration of growth factors following irradiation, underlie the mechanisms for compensation of damage. Early investigations in mice by Fried et al. demonstrated that the stromal cells supporting hematopoietic regeneration were more radioresistant than the CFU-S; single whole-body doses of 500 cGy or less did not consistently affect the ME but destroyed more than 99 % of CFU-S (Fried et al. 1976). The stroma both proliferates and supports hematopoiesis, and these capacities are diverse in their radiosensitivity. Low single doses of less than 600 cGy can prevent the establishment of a stromal layer in long-term culture systems (Laver et al. 1986) and prevent stromal cell proliferation (Zuckerman et al. 1985). Conversely, greater than 10000 cGy to an established stroma are tolerated in terms of the ability to produce growth factors (Zuckerman et al. 1985; Gualtieri and McGraw 1985), or support hematopoietic progenitor cell proliferation and adherence (Zuckerman et al. 1985). The influence of dose fractionation and rate is also relevant because of the contrasting ability of the stromal and hematopoietic precursors to repair damage. In vitro studies by Laver provide D_o values of 130 and 115 cGy for CFU-F and CFU-GM respectively; however, the CFU-F survival curve exhibits a shoulder (n = 1.3), whereas the CFU-GM curve does not (n = 1.0) (Laver et al. 1986).* Similar D_O values for CFU-S and other progenitors have been demonstrated by Hendry (Fig. 7d) (Hendry 1985). Various survival curves have been demonstrated for the stroma, apparently dependent on the assay used (Fig. 7e) (Bierkens et al. 1989). Greenberger et al. have shown the relevance of low (5 cGy/min) and high (120 cGy/min) dose rates to several parameters of marrow stromal cell support function for engrafted hematopoietic stem cells (Greenberger et al. 1988). Conversely, the radiosensitivity of multipotential stem cells is independent of dose rate (Hendry). The understanding of these data must await further investigate because of the demonstrated differences between the ability for marrow stroma to regenerate and repair damage when assayed in vitro versus in vivo (Chertkov et al. 1983; Knopse and Husseini 1986).

A cascade of direct and indirect responses of both irradiated and unirradiated bone marrow regions underlies the patterns of marrow activity observed after therapy. On the most basic level, the individual radiosensitivities of the ME and hematopoietic stem cells determine the potential for any locally irradiated region to maintain function. On a higher level, the interactive capabilities of treated and untreated bone marrow volumes determine the potential for the organism to sustain radiation injury. Table 6 is an attempt to categorize what must be a continuum of events in this regard.