Bone marrow necrosis




In the medical community, little is known regarding bone marrow necrosis (BMN) as a clinicopathologic entity, although to the authors' knowledge it was described for the first time more than 50 years ago. To identify the rate of prevalence, the symptoms and signs, the underlying disease associations, and the usefulness of diagnostic procedures, an extensive literature search was made.


Only cases identified as extensive necrosis and diagnosed during life were selected. Two hundred forty cases met these criteria.


Bone pain (75%) and fever (68.5%) were the most important symptoms, whereas anemia (91%) and thrombocytopenia (78%), associated with a leukoerythroblastic picture (51%), were the most frequent hematologic abnormalities. Nearly 50% of patients showed elevated lactate dehydrogenase and alkaline phosphatase levels. In 90% of the patients an underlying malignancy was identified.


Bone marrow necrosis is caused by hypoxemia after failure of the microcirculation. Given the high rate of malignancy as an underlying disease association, an extensive search for neoplastic disease is justified whenever BMN is diagnosed. Pancytopenia and embolic processes are major complications that should be managed with supportive measures until effective treatment of the underlying disease has been administered. When necrosis resolves, repopulation of the bone marrow cavity with normal hematopoiesis is observed. Cancer 2000;88:1769–80. © 2000 American Cancer Society.

Bone marrow necrosis (BMN) was described for the first time as far as we know by Wade and Stevenson, in 1942, in a patient with sickle cell disease who died of cerebral infarction.1 Starting from a pathology study, Nies et al. tried, in 1965, to define the clinical features of BMN in patients with acute leukemia.2 It was not until the early 1970s that BMN was frequently diagnosed before death. This was due to the rare character of the disease, the underawareness by the medical community of its existence, and the sparse reporting in the hematologic literature.

As more case reports and small series of patients have been published, this clinical-pathologic presentation has become a separate entity, distinct from avascular bone necrosis and marrow aplasia. The goal of this article is to summarize the state of knowledge on BMN, its prevalence, the association with underlying diseases, and the usefulness of diagnostic procedures. A review therefore was made of 240 published cases, identified in an extensive literature search, by using a uniform definition of BMN.


Bone marrow necrosis is best defined as “necrosis of myeloid tissue and medullary stroma in large areas of the hematopoietic bone marrow”.3 On a trephine biopsy, BMN is characterized by a disruption of the normal bone marrow architecture with a considerable loss of fat spaces. This contrasts with the findings of aplastic anemia in which there is only a loss of myeloid tissue and no destruction of the reticular structure.4 Usually there is no destruction of the spicular architecture as is seen in aseptic necrosis.5


A Medline search for BMN was performed for the years 1980 to 1999. Furthermore, the references in these articles were screened to also collect articles on this subject published before 1980. To be included in our study, these case reports had to fulfill the following conditions: 1) BMN had to be diagnosed during life to exclude the possibility of necrosis as a postmortem phenomenon, and 2) BMN had to be extensive, which was defined by typical bone marrow aspiration cytology and/or by necrosis of > 50% of the diameter of a trephine biopsy specimen.


From the clinical point of view, BMN seems to be extremely rare as is evidenced by the few (240) reports in which BMN was diagnosed during life in the literature so far. This is in contrast to the findings in autopsy reports, in which it is encountered more commonly and in which rates vary widely.

In 1965, Nies was the first as far as we know to try to disclose the clinical features of the pathologic entity of BMN. He reviewed the marrow histology in 316 patients who had died of acute leukemia and found marrow necrosis in 34 cases (10.75%). At the time of initial interpretation, this was recognized in only nine biopsies.2

In 1976, Kiraly and Wheby published 13 cases of BMN in 664 subsequent bone marrow aspirations and/or biopsies (2%), taken during a 12-year period7

In 1979, Norgard reviewed 368 biopsies and diagnosed BMN retrospectively in 120 cases (32.5%). BMN was diagnosed at first pathology reading in four patients only. The degenerative and necrotic areas were graded on the basis of their size. Small necrotic lesions (< 25% of the dimension of the biopsy) and intermediate BMN (< 50% of the dimension of the biopsy) were found, respectively, in 47 (39%) and 29 (24%) cases. In 44 cases (12%), the grade of BMN was large (> 75% of the dimension of the biopsy specimen) or extensive (throughout the dimension of the specimen).8

In 1986, Pennaforte et al. looked only for extensive bone marrow necrosis and found it in four cases (0.3%) of 1149 biopsies.6

In 1988, Maisel et al. found some degree of BMN in 37% of 270 examined bone marrow biopsies. Small foci of BMN (< 20% of the dimension of the biopsy) were found in 26.5%, whereas moderate BMN (between 20 and 50% of the dimensionameter of the biopsy) was found in 7.5% of the biopsies reviewed. Only in 3% was severe BMN (> 50% of the dimension of the biopsy) mentioned.9

Macfarlane and Tauro and Pui et al. reported on the rate of BMN in children. MacFarlane and Tauro found necrosis in 4 (1%) out of 379 acute lymphoblastic leukemia (ALL) patients,10 whereas Pui et al. reported 7 cases (0.5%) of 1419 cancer patients.11

Dunn et al. published their experience with BMN in 1993. Forty cases (0.4%) were found out of 10,856 aspirates and/or trephines during 14 years. In 38 cases, a malignancy was found.12

Difference in BMN rates (0.3% vs. 37%) can be explained in part by the use of different grading scales. When looking only for extensive BMN, the gap narrows (0.3% vs. 12%). The population studied also can explain the difference in BMN rates. Nies, in 1965, reported on BMN in acute leukemia. The histology material was taken at autopsy, and the extent of necrosis was not mentioned. The absence of chemotherapeutics and antibiotics in the early 1960s may account for the high rate of BMN (10.75%).2 Although Kiraly and Wheby,7 Macfarlane and Tauro,10 and Pui et al.,11 and Dunn et al.12 did not mention the extent of BMN, all the cases they reported had symptoms and signs of an extensive necrosis. In contrast, the high percentages of BMN found by Norgard et al.8 and Maisel et al.9 cannot be correlated with clinical data. The rate of extensive BMN on bone marrow aspirate and/or biopsy in an unselected population as reported by Pennaforte et al.6(0.3%) and Dunn et al.12 (0.4%) appears more realistic. The data from Pui et al.11 (0.4% in childhood acute leukemia) and Macfarlane and Tauro10 (1% of BMN in childhood ALL) agree with the above-mentioned rates.


The underlying disease associations of the 240 cases are shown in Table 1. Malignancy (n = 218, 91%) is the most common underlying disease of BMN. Hematologic malignancies were found in 145 (60%) records with acute leukemia in 98 patients (41%) as the leading cause.2, 3, 6, 7, 10, 11, 13–54

Table 1. Underlying Disease Associations of BMN, Diagnosed Intravitally
DisorderNo. of patients
  • BMN: bone marrow necrosis, HIV: human immunodeficiency virus; ATRA: all-transretinoic acid; G-CSF: granulocyte-colony stimulating factor.

  • a

    Two solid tumors together with myelofibrosis and a lymphoma.

  • b

    In 6 of 17 cases with infection.

  • c

    In two of eight cases with drug exposure.

  • d

    In 8 of 12 cases with various diseases, no other underlying disorder was found.

Malignant disorders218
 Hematologic malignancies145
 Acute leukemia98
  Acute lymphoblastic leukemia(43)
  Acute myeloid leukemia(31)
  Acute leukemia(24)
 Chronic myeloproliferative disorders13
  Chronic myeloid leukemia(11)
  Essential thrombocythemia(1)
  Non-Hodgkin lymphomas(26)
  Chronic lymphocytic leukemia(2)
  Hairy cell leukemia(1)
  Multiple myeloma(1)
  Hodgkin disease(5)
 Solid tumors72a
  Primary tumor of unknown origin(12)
Nonmalignant disorders22
  Q fever(1)
  Mycobacterium tuberculosis(2)
 Sickle cell disease6
  Anorexia nervosa(1)
  Hemolytic uremic syndrome(1)
  Antiphospholipid syndrome(3)
  Diffuse intravascular coagulation(4)

Acute lymphoblastic leukemia is the most common underlying disorder of BMN both in the total population (n = 43, 18%)10, 11, 13–39 as well as in children in whom 18 of the 25 cases with BMN were associated with ALL.10, 11, 14, 15, 17–20, 33 Data on the time relation of BMN with ALL were available in 37 patients. Bone marrow necrosis developed before diagnosis in 26 out of 37 patients, after induction chemotherapy in 1 of 37 patients and at the time of recurrence in 10 of 37 patients.

Bone marrow necrosis also is frequently associated with acute myeloid leukemia (AML) (n = 31, 13%).3, 6, 7, 16, 21–24, 37, 40–51 All French–American–British groups (M0–M7) have been described in association with bone marrow necrosis with no particular predilection. In the AML setting, BMN is again most commonly seen at diagnosis (16 out of 24 evaluable cases).

Bone marrow necrosis is less frequently associated with chronic myeloproliferative disorders (n = 13, 5%). Chronic myeloid leukemia (CML) has been reported in 11 cases7, 13, 21, 23, 46, 55–59 and essential thrombocythemia60 and myelofibrosis7 each in 1 case.

Bone marrow necrosis was described in the context of lymphoma in 35 patients (15%).4, 7, 14, 21, 23, 27, 35, 39, 46, 52, 61–74 The association was observed with aggressive lymphoma, with indolent lymphoid malignancies like chronic lymphocytic leukemia4, 65 or hairy cell leukemia,72 and with moderately aggressive disorders like follicle center lymphoma69 and multiple myeloma.66 Furthermore, BMN has been reported in association with Hodgkin disease in five cases.39, 46, 63, 70

The association of BMN with solid tumors is less striking (n = 72, 30%).7, 11–13, 21, 23, 27, 39, 53, 67, 75–88 Often, the primary origin of the tumor is not found (12 of 43), even after an extensive search. Bone marrow necrosis has been described in patients with carcinoma of the stomach (14 cases),7, 21, 23, 39, 74, 84–86, 88 lung (3 cases),7, 39, 80 ovary (3 cases),7, 76 breast,83 prostate,78 and esophagus.13 Case reports describe the association of BMN with Kaposi sarcoma,81 carcinoid tumor,39 neuroblastoma,11 and medullablastoma.82 Only 22 (9%) cases could be related only to nonmalignant causes. Sickle cell disease, infections, and certain drugs are the most important entities.

The first case as far as we know of BMN described in literature (1942) was associated with a sickle cell crisis.1 Since this report, some cases of BMN have been attributed to this hematologic disorder.75, 89–92 In 1967, Charache and Page75 looked for the frequency of BMN in patients with sickle cell disease who sought medical attention because of exquisite tenderness in bones. They stated that one of six patients with a sickle cell crisis develop some degree of BMN.75 The paucity of descriptions reflects not a low rate but a low frequency of examination of marrow in painful areas during crises. Not all patients with localized pain seek medical attention, and some reports only mention localized BMN. Usually there is full recovery after the end of the hemolytic crisis.

The causal relation of infection is more controversial. In 22 patients, active infection is believed to have caused BMN. These comprise sepsis due to Escherichia coli,13Streptococcus,20, 93Staphylococcus,14, 17Citrobacter freundii,92Salmonella,94 and other bacteria.13, 23, 45, 62, 95 Case reports of mucormycosis96 and Q fever97 have been described in association with BMN in the absence of any underlying malignancy. Furthermore, Mycobacterium tuberculosis61, 98 has been described as a responsible agent. Association between parvovirus infections and BMN was the subject of publication in four cases.32, 90, 99 However, 3 of these patients had sickle cell disease, and in 1 patient ALL was diagnosed 2 months later. In one publication, BMN was mentioned in an human immunodeficiency virus positive patient, in whom an underlying high grade lymphoma was diagnosed postmortem.68 The occurrence of BMN in a CML patient with a pneumococcal pneumonia has been reported.58 In conclusion, of 22 patients with BMN and documented infection, 16 patients were found to have an underlying hematologic disorder (CML, ALL, lymphoma, sickle cell disease, solid tumor) near the time of BMN occurrence, whereas in 1 case a CML was diagnosed 1 year after BMN.98 In the other cases, BMN could appear early after onset of infection93 as well as later when antibiotic treatment was postponed.94

The association of BMN with drugs is rare. A case report of BMN after treatment with sulphasalazine was published in the 1980s.100 Sulphathiazol, another member of the sulphamide family, also has been suspected of inducing BMN.13 Of more recent date are the records of BMN associated with interferon-α,19, 59, 101 granulocyte-colony stimulating factor,51 all-transretinoic acid with or without hydroxyureum,47–49 and fludarabine,71 all used in patients with underlying hematologic malignancies.

Bone marrow necrosis also can occur with anorexia nervosa.5 When nutrient delivery decreases slowly as in the common form of anorexia nervosa, gelatinous transformation predominates, visible on staining as pink amorphous, extracellular material consisting of acid mucopolysaccharides.

These replace the bone marrow fat that is mobilized for energy requirement. The acid mucopolysaccharides also may be responsible for the inhibition of hematopoiesis. Bone marrow necrosis in anorexia nervosa or with involuntary starvation occurs only when nutrient delivery suddenly is interrupted or when there is a superimposed insult. Due to the bone marrow hypocellularity, the extent of cell necrosis may be limited.5, 102

Association of BMN with a hemolytic uremic syndrome,103 the antiphospholipid syndrome,74, 104, 105 diffuse intravascular coagulation (DIC),106 and hyperparathyroidism107 has been reported. In two cases, no disease association was found, and they can be described as idiopathic BMN.8



Table 2 summarizes the frequency of signs and symptoms observed at the time of BMN in the patients reviewed. Because the information is not available on all patients, the denominator of examinable patients may vary.

Table 2. Symptoms and Laboratory Findings in 240 Patients with BMN, Diagnosed Intravitally
SymptomTotal (%)ALL (%)AML (%)CML (%)Lymphoma (%)Solid tumors (%)Sickle cell disease
  1. BMN: bone marrow necrosis; ALL: acute lymphoblastic leukemia; AML: acute myeloid leukemia; CML: chronic myeloid leukemia; LDH: lactic dehydrogenase; AP: alkaline phosphatase.

Bone pain163/209 (78)33/37 (89)22/28 (78.5)9/9 (100)16/31 (51.5)23/30 (76.5)6/6 (100)
Fever127/185 (68.5)23/33 (70)9/20 (45)6/9 (66.5)18/31 (58)18/30 (60)6/6 (100)
Anemia166/182 (91)30/33 (91)16/18 (89)9/10 (90)23/26 (88)28/30 (93)6/6 (100)
Thrombopenia116/148 (78)27/34 (79)21/23 (91)3/8 (37.5)21/26 (81)22/29 (76)4/5 (80)
 Normal46/149 (31)11/32 (34)3/19 (16)0/8 (0)7/25 (28)13/30 (43)4/6 (67)
 Low66/146 (45)18/32 (56)5/19 (26)2/8 (25)16/25 (64)8/30 (27)1/6 (17)
 Elevated37/146 (25)3/32 (9)11/19 (58)6/8 (75)2/25 (8)9/30 (30)1/6 (17)
Leukoerythroblastic  differential96/174 (55)9/33 (27)12/17 (70.5)3/8 (37.5)17/26 (65)23/29 (79)0/6 (0)
LDH elevation66/128 (51.5)18/32 (56)8/14 (57)0/8 (0)17/24 (71)13/23 (56.5)1/5 (20)
AP elevation51/125 (41)10/32 (31)4/13 (31)2/7 (28.5)15/24 (62.5)10/22 (45)0/5 (0)

Bone pain is an important diagnostic sign. It was first mentioned as far as we know by Kundel et al. in 1964 who noted this symptom in 6 ALL patients with BMN.108 Brown suggested that bone pain was seen more frequently when BMN was associated with hematologic malignancies or sickle cell disease.13 It seems less likely to occur when systemic infections or nonhematologic pathology were the underlying causes of BMN. In the first review as far as we know of BMN diagnosed in alive patients, Pennaforte et al. found bone pain in 32 of 54 cases.6 The pain was acute, intense, and usually located in the lower back. Bone pain was the most important reason for hospitalization in patients with BMN. In our database, bone pain was present in 78% (163 of 209) of the evaluable cases. It was a common symptom in patients with ALL (33 of 37), CML (9 of 9), AML (22 of 28), and solid tumors (23 of 30) but was less frequent in lymphoma patients (16 of 31). All patients with BMN and sickle cell disease in whom the information was available (6 of 6) suffered from bone pain. In patients with BMN secondary to infections or drug use, bone pain often was reported, but it was unclear whether this was only in patients with an underlying concomitant hematologic disorder. Bone pain is not a discriminating sign of BMN, although it is often more intense when BMN occurs together with the underlying disease. Bone pain is part of the presentation symptoms in childhood ALL in 23% of patients108 and in adult ALL in < 1%.109 The frequency of bone pain in both childhood and adult AML is very low. Bone pain is often mild in ALL but debilitating in BMN.

In 1976, Kiraly and Wheby mentioned the presence of fever as part of the clinical picture of BMN.7 It was reported in 16 of 54 cases of the Pennaforte group.6 It occurred in 68.5% (127 of 185) of cases in this review. There was no clear difference between the diagnostic groups: ALL (23 of 33), AML (9 of 20), CML (6 of 9), lymphoma (18 of 31), solid tumors (18 of 30).

Considering the rate of fever as a presentation symptom in 61% of childhood ALL108 and 25%–30% of adult leukemia, fever has no discriminating value in the diagnosis of BMN.

A rare but life-threatening complication of BMN is the embolization of fat and necrotic bone marrow to pulmonary arteries.39, 58 Similar to sickle cell disease, globules of fat or necrotic bone marrow, after being embolized to the pulmonary system, can pass pulmonary capillaries or arteriovenous shunts and enter the left circulation damaging organs.110, 111

Laboratory Results

The laboratory findings are also summarized in Table 2. In 1964, Kundel et al. observed that patients with BMN developed pancytopenia some time after the onset of bone pain.108 Kiraly and Wheby found pancytopenia with a leukoerythroblastic differential count in their 13 cases.7 Pennaforte et al. reported anemia in 36 of 54 cases.6 In this review, we observed anemia (166 of 182, 91%) and thrombocytopenia (116 of 148 patients, 78%) frequently in BMN. The leukocyte count was elevated in 37 of 149 patients (25%), normal in 46 of 149 patients (31%), and decreased in 66 of 149 patients (45%). A leukoerythroblastic differential was found in 96 of 174 (55%) patients.

Kiraly and Wheby described an elevation of lactic dehydrogenase (LDH), alkaline phosphatase (AP), uric acid, and glutamic oxaloacetic transaminase in his 13 cases.7 Pennaforte et al. noted an elevation of LDH and AP in 22 of the 54 cases.6 We could observe LDH and AP elevation, respectively, in 66 of 128 (51.5%) and 51 of 125 (41%) cases. Compared with an 84% rate of elevated LDH in adult ALL, none of these laboratory findings is pathognomonic. No representative data are available regarding the degree of LDH elevation or LDH isoenzymes in BMN.


Bone marrow aspiration can be unsuccessful so that multiple aspirates from different sites (sternal, posterior iliac crest, etc.) are necessary to obtain enough material.8 Aspiration can be serosanguineous, watery dark red or even clear fluid.14 The aspirate can smear very unevenly.96

Microscopy shows a background of amorphous extracellular eosinophilic proteinaceous material that surrounds cells that desintegrate8 (Fig. 1). The cells lose their normal staining characteristics and have irregular or indistinct cell margins on cytology. The cytoplasm can shrink or vacuolate, and the nucleus shows pyknosis, karyorrhexis, and karyolysis5 (Fig. 2).

Figure 1.

Bone marrow smear on light microscopy (original magnification ×10) shows a background of amorphous extracellular eosinophilic material that surrounds cells that disintegrate.

Figure 2.

Higher magnification (original magnification ×100) shows cells that lose their normal staining characteristics. These cells have irregular or indistinct margins; the cytoplasm can shrink or vacuolate, and the nucleus shows pyknosis, karyorrhexis, and karyolysis.

So that the underlying diagnosis responsible for BMN is not overlooked, it is necessary to search through all the smears for a single focus of viable marrow.43 If the cytologist is familiar with the appearance of necrotic cells, the smudging of the cells will not be attributed to artifacts.13


The histology of BMN is characterized by a combination of gelatinous transformation and necrosis of the myeloid tissue. Gelatinous transformation is typical with severe malnutrition or starvation and is associated with areas of focal hypoplasia and a background of amorphous eosinophilic staining material.108 In BMN, this background of gelatinous transformation surrounds cellular debris with indistinct cellular margins, pyknotic nuclei, and abnormal eosinophilic staining cytoplasm. For the clinician, it is important that the pathologist grades BMN as the percentage of the diameter of the biopsy that shows necrosis. Although there is no destruction of the spicular architecture, several authors mention bone necrosis. The hallmark of bone necrosis is the loss of osteocytes, osteoblasts, and osteoclasts. Empty osteocytic lacunae in the affected trabeculae are rarely seen until 1 week after the initiating event. The mineral density and the trabecular structure remain unaltered for weeks.113 That may be the reason why bone changes are not frequently observed at the moment of diagnosis of BMN, and it explains the lack of concomitant radiographic changes of the bones affected by BMN. When the microvascular failure can be stopped, the repair reaction starts with the removal of necrotic elements by phagocytosis. Proliferating capillaries and fibroblasts reconstitute the bone marrow stroma. Repopulation of the necrotic bone marrow with normal hematopoietic tissue is the rule.75 In follow-up trephine biopsies, moderate fibrosis or reticulin staining is observed as an inflammatory reaction of the bone marrow to injury. The necrotic bone undergoes remodeling after being repaired.113

It is of utmost importance that the disintegration of cells is not attributed to artifacts. Bone marrow necrosis is a clinicopathologic entity when the necrosis is extensive. This means that large areas of the bone marrow cavity are filled with necrotic debris. This correlates with necrosis of > 50% of the dimension of the trephine biopsy. In the absence of a suggestive clinical picture, small foci of necrosis on trephine biopsy are not diagnostic. This implies that the pathologist should use a grading system when describing BMN.

Bone Marrow Scanning

Technetium 99m sulfur colloid and indium chloride localize selectively to the reticuloendothelial elements of the marrow, which correspond with areas of hematopoiesis. They allow changes in hematopoiesis to be visualized when the area exceeds at least 2 cm. In the case of BMN, there is little or no isotope uptake throughout the bone marrow cavity space by the lack of normal bone marrow activity. The diagnostic usefulness of bone marrow scintigraphy is double: it confirms the absence of hematopoiesis but also shows the existence of residual bone marrow activity from which material can be obtained by means of guided aspiration or biopsy. In cases of recovery, reappearance of normal hematopoiesis can be observed.14, 27

Nuclear Magnetic Resonance

Magnetic resonance imaging (MRI) provides a noninvasive method to evaluate a large fraction of bone marrow. Changes in bone marrow signal intensities are reflections of changes in the proportions of fat and water contained in the cellular elements.

BMN is characterized by an increase of water content due to watery changes of the bone marrow and replacement of the fat by serous material. This changes the contrast between fat and cellular tissue.114 Although MRI pictures are not diagnostic, they can show the extent of necrosis when it has been proved with a trephine biopsy. It also can serve as a guide to biopsy sites in which viable hematopoietic bone marrow is suspected. Magnetic resonance imaging is also able to document conversion from abnormal to normal bone marrow. This decrease or disappearance of necrosis can take years as was described in a case report by Weissman et al.70


The mechanism of BMN is poorly understood. Failure of the microcirculation must be considered as the critical event. This can be due to inflammatory damage or mechanical obstruction as in the case of DIC sickle cell disease, or tumor cell plugs.

Bernard et al. were the first as far as we know to prove that bone marrow microvasculature was completely lost or partly destroyed in a patient with BMN and AML, but without detectable leukemic infiltration of the bone marrow at time of necrosis. They postulated that the vascular lesions were induced through immune complexes.3

In the late 1960s and the early 1970s, several investigators induced bone marrow necrosis in animal models by injection with Hunstein virus115 and endotoxin.116 This relates to clinical observations of endotoxemia in an ALL patient with E. coli sepsis, documented at time of BMN and bone marrow cultures positive for bacteria in the case of BMN associated with sepsis.13 Occasionally, bacteria have been visualized in foci of necrotic bone marrow.93

Similar inflammatory vascular damage can be held responsible for BMN induced after the injection of foreign protein in sensitized animals117, 118 and after high dose irradiation. This led to the disappearance of the sinusoidal microvasculature in animal experiments.119

Mechanical obstruction of the microcirculation may be the mechanism in sickle cell disease whereby aggregates of deformed sickle cells occlude the bone marrow capillaries during crisis.13, 120 In DIC, the obstruction is based on the formation of fibrin plugs.105 In metastatic carcinoma, aggregates of carcinomatous cells could be found inside the bone marrow vessels when BMN occurred.78 In patients with leukemia, the leukemic population may outgrow its blood supply and impinge upon the vascular structures.13

The common denominator in the pathophysiology of BMN may be the toxicity and the release of toxins, cytokines, or vasoactive substances from damaged cells.47, 48, 73 Tumor necrosis factor (TNF) could play a role in BMN. Production of TNF has been reported in two patients with metastatic carcinoma and BMN in the absence of infection. Tumor necrosis factor could be responsible for the prothrombotic effect involved in the vascular damage.67


The survival of BMN depends greatly on the underlying disorder. In 1965, Nies was the first as far as we know to comment on the outcome of nine patients with acute leukemia and BMN. The median survival was 7 months (range, 4–13 months). Three patients reached a complete remission though this remission did not last longer than 1 month in two patients.2

Kiraly and Wheby reported a median survival from the diagnosis of BMN of only 22 days (range, 5 days to 11 months) for his 14 patients. Four patients survived longer than 1 month. One patient with ovarian carcinoma recovered from cytopenia while receiving chemotherapy. Neither BMN nor tumor cells could be detected on repeated bone marrow examinations in this patient. In contrast, one patient with CML continued to show extensive necrosis on repeated BM examinations.7

Pui et al., in 1985, noted symptomatic improvement (bone pain and fever) after treatment in seven children with BMN. A complete response was obtained in all patients, and subsequent BM aspirates at previous sites of necrosis showed normal cellularity during a period of 1 to 10 months. With a follow-up ranging between 10 and 51 months, two patients relapsed, but they had a second remission.11

In 1986, Macfarlane and Tauro reported a complete remission in their four child patients with ALL and BMN. One patient remained in continued complete remission, and three patients died after recurrence between 1 and 6 years after the diagnosis of BMN.10

Cassileth and Brooks, in 1987, followed nine adult patients with acute leukemia and BMN. Survival data ranged between less than 1 and 8.5 months, respectively. No patient had a complete remission.22 Hamidou et al. also reported on the bad outcome of BMN in hematologic malignancies: six of seven patients died and one relapsed.46 Scudla et al. reported on seven patients with BMN and malignancy in the period 1987–1992. Survival data ranged between 4 and 52 weeks.23, 42, 87 Dunn et al., in 1993, observed a survival between 10 months and 8 years for patients with acute leukemia in which complete response (CR) could be induced. When there was no response to chemotherapy, the median survival was shortened to 6 weeks.12

In 148 of the 240 BMN patients reviewed in our study, individual survival data were available. These data are summarized in Table 3. One hundred three of 145 patients (69%) had died at the time of publication. Only in approximately 30 patients was this early death due mainly to the development of BMN rather than to the underlying condition.

Table 3. Survival Data in 148 of 240 Patients with BMN, Diagnosed Intravitally
DisorderSurvival data availableAliveDeathMedian survivalRange of survivalReferences
  1. BMN: bone marrow necrosis; ALL: acute lymphoblastic leukemia; AML: acute myeloid leukemia; MPD: myeloproliferative disorders.

ALL (adult + childhood)35/4317184 mos2 days to 6.5 yrs10, 11, 13, 15, 17, 18, 20, 22–25, 27, 30, 32–36, 38
Childhood ALL18/251352.5 yrs4 mos to 6 yrs10, 11, 14, 15, 17, 18, 20, 33
AML25/315202 mos8 days to 3 yrs3, 6, 7, 16, 22, 24, 40–45, 47, 48, 50, 52
AL22/24022Not available2 mos to 2 yrs2, 12, 22, 53
MPD7/13163 mos5 days to 25 mos42, 56–60
Lymphoma24/357171 moEarly to 8 mos4, 14, 23, 27, 35, 46, 52, 62–72, 74, 75
Solid tumors18/723155 wksEarly to 4 mos11, 13, 23, 27, 53, 67, 76–78, 84, 85, 87
Sickle cell disease6/66075, 89–92
Infections5/6233 wksEarly to 29 days93, 94, 96–98
Others5/8412.5 wks1–4 wks8, 104, 106, 107

The median survival data of the hematologic malignancies range between 1 and 4 months. Historic survival data of the different hematologic malignancies without BMN are certainly better. However, we lack information on other prognostic markers in most of the case reports to conclude that BMN alone is responsible for the bad outcome. If we select only patients with childhood ALL, CR was achieved in all patients and 13 of 18 were alive at time of publication. Survival ranged between 4 months and 6 years for the patients who died. This suggests that BMN in childhood and certainly in childhood ALL does not compromise CR and survival.

In eight childhood ALL patients of whom a rich bone marrow could be obtained despite of BMN at other sites, the median percentage of lymphoblasts was 80% (range 72%-95%). This is not different from the initial percentage of lymphoblasts seen at diagnosis of ALL without BMN.110 In 18 patients with solid tumors of whom survival data were available, survival time was short.

Most patients with solid tumor and BMN had widespread metastasis and involvement of the bone marrow. Thus, BMN in a patient with a solid tumor seems to be a sign of generalized disease and predicts a short survival.

In six patients with sickle cell disease, BMN resolved and hematopoietic recovery was proven during a follow-up period that ranged from 1 to 10 months. Five of 12 patients with BMN due to infection, drug therapy, or other underlying disorders died in the first month after a diagnosis of BMN was made.

Although available data do not allow to prove or deny that BMN is an independent poor prognostic factor, this study suggests that BMN is not always an ominous sign and that vigorous supportive care together with disease specific treatment must be continued to permit adequate time for spontaneous recovery of the normal hematopoietic tissue. Transfusions of red blood cells and platelets must be given as long as cytopenia is a problem. Adequate antibiotic treatment must be started immediately when there is suspicion of infection. In sickle cell disease, hydration, alkalinization, and oxygen are necessary supportive measures. When an underlying malignancy is found, chemotherapy must be started as soon as possible.


The diagnosis of BMN is made by the typical cytologic findings on bone marrow aspirate and/or the striking histologic picture on trephine biopsy. This cytohistologic diagnosis correlates best with the clinical findings when necrosis is extensive.

Bone pain (75%) and fever (68.5%) are the most important symptoms, whereas anemia (91%) and thrombopenia (78%) are the most frequent hematologic abnormalities. A leukoerythroblastic picture often is observed (51%) whereas the white blood cell count can change in any direction. These and other recurrent laboratory findings, however, are not specific because they can be caused by the underlying illness. In BMN, the nature of bone pain is, however, more intense and debilitating.

An underlying malignancy was noted in 90% of the patients. Most frequent were hematologic neoplasias (60%), whereas solid tumors were seen in 30% of the cases. Bone marrow necrosis can develop before the diagnosis of these malignancies, after chemotherapy or at the time of recurrence. Sickle cell disease was the most common nonmalignant condition underlying BMN (6 of 22 patients).

Furthermore, sepsis and certain drugs were mentioned in case reports as a possible causes of BMN, although there was often a concomitant malignancy. Tissue hypoxemia after failure of the microcirculation can be considered as the most important mechanism in developing necrosis of the bone marrow, irrespective of the underlying condition.

Given the high rate of malignancies, an extensive search for neoplastic disease is justified whenever BMN is diagnosed. Multiple bone marrow aspirates or biopsies at different places may be required for the evaluation of viable bone marrow. Scanning or MRI of the bone marrow is helpful in finding places of nonnecrotic tissue and in determining optimal puncture sites. Pancytopenia and embolic processes are the major complications of BMN, and should be managed with supportive measures until treatment for the underlying disorder is effective.

Survival generally depends on the prognosis of the underlying disease. When the necrosis resolves, repopulation of the bone marrow cavity with normal hematopoietic cells is seen, leaving small fibrotic scars.