BONE MARROW NECROSIS DISCOVERED IN A PATIENT WITH SUSPECTED THROMBOTIC THROMBOCYTOPENIC PURPURA

Abstract

A 48 year-old white man was hospitalized for evaluation of back pain. At time he reported that he had been in excellent health until three months before his hospitalization when he noticed difficulty walking when he got out of his car following a long trip. He said that “it just felt as though my legs wouldn’t move very well”. He did not see a doctor since this problem resolved within several days. Three weeks before his hospitalization, he thought he “had a viral illness” with cough and fatigue that persisted, together with subsequent abdominal discomfort. His primary care physician noted that his platelet and white blood cell counts were low which he attributed to a viral infection. An abdominal ultrasound reported minimal splenomegaly. He was treated with antibiotics and prednisone; all of his symptoms resolved and his platelet count increased. One week before his hospitalization he developed low back pain which made walking difficult. He also had fever and sweats. When these symptoms persisted he was admitted to the hospital. His physical examination was normal. His spleen was not palpable; he had no lymphadenopathy; his neurologic examination, including his gait, was normal. He had no back tenderness. His platelet count was 23,000/µL; white blood cell count, 3700/µL with a normal differential; hemoglobin, 13.5 gm/dL; creatinine, 1.7 mg/dL; LDH, 1737 U/L (normal, <190 U/L). Coagulation tests were normal; fibrinogen was 858 mg/dL. Examination of the peripheral blood smear demonstrated schistocytes and normal white cell morphology. Thrombotic thrombocytopenic purpura (TTP) was suspected because of the thrombocytopenia, red cell fragmentation, high serum LDH and creatinine, the history of fever, the possibility that the difficulty walking may have been a neurologic manifestation of TTP, and – most important – no apparent alternative etiology. Treatment with plasma exchange (PEX) and corticosteroids was begun.

Beginning PEX for clinically suspected TTP was appropriate for this patient.[1, 2] He had the clinical diagnostic features of TTP with no evidence for an alternative etiology. However the presenting complaint of back pain is rarely, if ever, a symptom of TTP; it has not occurred as a presenting symptom in our experience.[3] Back pain in association with microangiopathic hemolytic anemia and thrombocytopenia is suggestive of a systemic malignancy.[4] Also leukopenia is not a typical feature of TTP. When leukopenia occurs, it may suggest the presence of systemic lupus erythematosus (SLE), since the frequency of SLE in increased in patients with acquired, autoimmune TTP.[5]

In addition to the atypical clinical features, this patient also had symptoms related to his illness for several months. The transient difficulty walking three months ago could have been a symptom of TTP, but this was very unlikely. Among all of our patients, approximately one-third of patients had preceding symptoms attributed to TTP that had occurred one to four weeks preceding presentation. One patient had a transient ischemic attack and normal blood counts with left side weakness one month before presenting with TTP.[6] But the interval of three months suggests another disorder. Therefore, although beginning PEX was appropriate management, a high level of suspicion for an alternative disorder required continuing careful investigation for other etiologies.

Blood for a plasma sample to measure ADAMTS13 activity had been drawn before beginning PEX but the blood was clotted and the test was cancelled. Therefore ADAMTS13 activity was not measured before beginning PEX.

Although ADAMTS13 activity is typically measured in plasma, measurements in plasma and serum are equivalent.[7] Therefore ADAMTS13 activity could have been measured in the serum from this clotted tube. ADAMTS13 is very stable, therefore serum samples stored from previous days could also provide valid ADAMTS13 measurements. Also, in most patients with TTP, a severe deficiency of ADAMTS13 activity persists for several days after PEX is begun;[8] therefore an accurate measurement of ADAMTS13 activity may have been obtained in this patient from a plasma sample obtained after PEX had begun. However, in this patient, ADAMTS13 activity was not measured early in his course. This was not an important initial issue because the decision to begin or to stop PEX treatment should depend on clinical features, not only on the level of ADAMTS13 activity as has been previously suggested.[9]

A severe deficiency of ADAMTS13 activity with demonstration of an ADAMTS13 inhibitor supports the diagnosis of acquired autoimmune TTP, but exceptions occur. Patients can have acquired autoimmune TTP with normal ADAMTS activity and can have alternative disorders, such as a systemic malignancy or infection, with a severe deficiency of ADAMTS13 activity.[6, 7, 10] Therefore the decision to begin PEX should be based on the presence of clinical features of TTP and the absence of an alternative etiology. However in this patient, normal ADAMTS13 activity together with his atypical clinical features would have increased suspicion for an alternative disorder.

After five daily PEX treatments, the patient’s back pain resolved but his white blood cell count had decreased to 800/µL (61% lymphocytes, some appearing reactive; 39% neutrophils); occasional nucleated red cells were present. His hemoglobin had decreased to 10.1 gm/dL; platelet count was unchanged at 23,000/µL; LDH (194 U/L) and creatinine (0.9 mg/dL) were normal.

This patient’s initial response to treatment with PEX and corticosteroids had conflicting interpretations. The symptomatic response to PEX appeared support the clinical diagnosis of TTP. However, the additional treatment with corticosteroids makes the interpretation of a symptomatic response less certain, since another disorder mimicking TTP could respond to corticosteroids. The recovery of serum LDH and creatinine are encouraging but not specific; PEX simply removes LDH and the decreased creatinine may merely reflect rehydration. Anemia is not quickly corrected by PEX and may even worsen after PEX is begun. In this patient, the greatest concern was the increasingly severe leukopenia. This appeared to contradict all signs of improvement and to clearly indicate the presence of a disorder different from TTP.

A bone marrow biopsy was done. It demonstrated no viable cells. There was extensive necrosis throughout the marrow space that included necrosis of vessel walls; one vessel had an intravascular thrombus. Although the marrow was completely necrotic, the cellularity of the “ghost cells” appeared to be approximately 80%. The trabecular bone was viable.

A bone marrow biopsy is appropriate in a patient with suspected TTP who has atypical clinical features. A normal or hypercellular marrow would be consistent with TTP. The observation of bone marrow necrosis was totally unexpected. Although unexpected and dramatically atypical for TTP, the necrotic marrow revealed no evidence for an alternative etiology. The significance of the hypercellularity of the necrotic “ghost cells” was not apparent at this time. In the absence of another etiology, the descriptions of necrosis of the vessel walls and an intravascular thrombus appeared to provide some support to sustain the diagnosis of TTP. The rationale was that since TTP can cause ischemic injury to other organs,[11] maybe TTP could also cause bone marrow ischemia and necrosis. Also the description of necrosis of vessel walls and an intravascular thrombus was suggestive of the pathologic characteristics of thrombotic microangiopathy.[2]

Treatment for TTP was continued; rituximab was added to the PEX and corticosteroids. Over the next five weeks the patient continued to be asymptomatic. Multiple investigations revealed no alternative diagnosis. CT scans of the chest, abdomen, and pelvis were normal except for mild splenomegaly. Echocardiogram was normal. Cerebrospinal fluid was normal. Serologic tests for multiple infectious diseases, systemic lupus erythematosus, and scleroderma were normal. Tests for paroxysmal nocturnal hemoglobinuria and cryoglobulinemia were negative. A second marrow biopsy, performed one week after the initial bone marrow biopsy, demonstrated areas of normal cellularity among areas of aplasia. A third marrow biopsy, performed two weeks after the second bone marrow biopsy, again demonstrated necrosis. It was hypocellular with predominantly stromal elements and few lymphocytes and plasma cells. No etiology of the necrosis could be determined; there was no evidence of malignancy in any of the three bone marrow biopsies. Throughout these five weeks, the patient’s anemia (hemoglobin, 8.3–10.9 gm/dL), thrombocytopenia (platelet counts, 6000–41,000/µL) and leukopenia (white blood cell counts, 1100–3400/µL) persisted. Five weeks after his presentation, ADAMTS13 activity was 94% when his platelet count was 18,000/µL.

Throughout these weeks, the patient remained asymptomatic. Intermittent PEX treatments, together with corticosteroids and rituximab, continued. Even though the persistent thrombocytopenia and the normal ADAMTS13 activity strongly suggested that TTP could not be the cause of his illness, no alternative disorder was apparent. Therefore diagnostic studies continued.

Six weeks after his presentation, a fourth marrow biopsy was performed when his white blood cell count had decreased to 400/µL and his platelet count to 5000/µL. The peripheral blood smear continued to demonstrate schistocytes, no immature or abnormal white blood cells, and few nucleated red blood cells. This biopsy had an interstitial infiltrate of small mononuclear cells that was not present in the peripheral blood. Acute lymphoblastic leukemia (ALL) was diagnosed by flow cytometry and immunohistochemical staining. He was begun on appropriate chemotherapy and achieved a complete remission with a normal bone marrow biopsy, without necrosis, in one month. Six months later he relapsed (with no marrow necrosis); 10 months later he died.

We assume that treatment for the initial diagnosis of TTP with corticosteroids and rituximab resulted in a partial response of the ALL that prolonged the patient’s course and delayed the correct diagnosis.

DISCUSSION

This patient’s puzzling course focused on the etiology of bone marrow necrosis (BMN). Necrosis is an uncommon and often unexpected observation in a bone marrow biopsy. Although many different disorders as well as adverse reactions to many drugs have been associated with BMN,[12] it continues to be a diagnostic problem.[13] In many patients, the cause is clear because the diagnosis of a disorder known to cause BMN, such as a hematologic malignancy, metastatic carcinoma, or sickle cell hemoglobinopathy, had been previously established. In some patients, the observation of BMN reveals a new, unexpected diagnosis. In occasional patients, the etiology of BMN is not apparent and treatment for an incorrect diagnosis continues; this was our experience. Our patient was initially diagnosed with thrombotic thrombocytopenic purpura (TTP) but he had atypical clinical features, notably severe neutropenia. Therefore a bone marrow biopsy was performed that, surprisingly, demonstrated necrosis. Because we could identify no apparent alternative etiology for the microangiopathic hemolytic anemia and thrombocytopenia, our initial assumption was that the BMN was caused by TTP and therefore treatment for TTP, with corticosteroids and rituximab in addition to PEX, continued. Six weeks after this patient’s presentation, a fourth bone marrow biopsy demonstrated ALL.

Following the diagnosis of ALL, we performed a systematic review of all previously published reports of BMN. Our literature search identified 766 articles and abstracts; 186 of these articles and abstracts reported data on 384 individual patients in whom BMN was documented by marrow biopsy. Among these 384 patients, BMN was attributed to 69 different etiologies. Hematologic malignancies were the most common etiology, accounting for 206 (54%) patients. ALL was the most common hematologic malignancy, occurring in 66 (32%) of the 206 patients with a hematologic malignancy. Other malignant disorders accounted for 74 (19%) of the patients. The most common diagnoses among non-malignant disorders were sickle cell hemoglobinopathies (20 patients). Among the remaining 84 patients, BMN was attributed to many diverse causes, including 13 different systemic infections and adverse reactions to 13 different drugs. Seven had patients no identified etiology in spite of extensive evaluations. The on-line Supplemental Material describes the methods and results of the literature search, with a table of data listing all reported etiologies of BMN together with the citations for all 186 articles.

Because of the experience with our patient, we focused on the six reports of seven patients with BMN who had an initial diagnosis of TTP (Table). BMN was subsequently attributed to metastatic carcinoma in five patients[14–18] and to hemoglobin S-β thalassemia in two patients.[19] In each of the five patients in whom metastatic carcinoma was subsequently diagnosed, the authors concluded that the patients had both TTP and carcinoma. However, we believe that it is more consistent with current concepts of thrombotic microangiopathies that the metastatic carcinoma itself caused the microangiopathic hemolytic anemia and thrombocytopenia and thereby merely mimicked the clinical features suggesting the diagnosis of TTP.[2] Four of these five patients presented with back pain as a primary symptom, similar to our patient. Although abdominal pain is a common presenting symptom of patients with TTP, in our experience back pain has not been a presenting symptom.[3] Therefore, a presenting symptom of back pain should suggest the diagnosis of a systemic malignancy.[4, 20] In the other two patients with BMN who were initially diagnosed with TTP (Table), the subsequent diagnosis was hemoglobin S-β thalassemia; one of these patients also presented with back pain.[19]

Table I.

Patients with bone marrow necrosis who were initially diagnosed as having thrombotic thrombocytopenic purpura (TTP)

Case	Citation	Age Sex	Clinical features					Final diagnosis
			Symptoms	Hgb	Platelets	Cr	LDH
1	2002[14]	33M	Back pain,1 month	6.3	66,000	0.8	7455	Gastric carcinoma
2	2003[15]	49M	Back pain, memory loss, 2 weeks	4.6	16,000	NR	2659	Signet ring carcinoma, unknown primary
3	2004[16]	67M	Back pain, confusion, 1 month	9.7	23,000	0.8	1239	Colon carcinoma
4	2007[17]	51F	Back pain, 2 weeks	6.7	40,000	NL	3350	Gastric carcinoma
5	2007[18]	53M	Fever, weakness, 1 month	4.3	10,000	NL	1814	Non-Small Cell Lung Carcinoma
6	2012[19]	25 M	Abdominal pain, 2 weeks, then coma	8.4	44,000	1.2	2118	Sickle-β thalassemia
7		53M	Back pain, 3 days, then coma	9.0	37,000	0.9	2114	Sickle-β thalassemia
8*	2014*	48M	Back pain, fatigue, 3 weeks	13.3	23,000	1.7	1737	Acute lymphoblastic leukemia

ADAMTS13 activity was not measured in patients 1–5. ADAMTS13 activity was normal (reported to be >60%) in patients 6 and 7. Abbreviations: Hgb, hemoglobin concentration, gm/dL; Platelets, platelet count/µL; Cr, serum creatinine, mg/dL; LDH, serum lactic dehydrogenase, U/L (normal values not reported); NR, not reported; NL, normal.

^*Patient 8 is described in this article. His ADAMTS13 activity was also normal (94%); although the sample was drawn 5 weeks after plasma exchange had begun, he was still severely thrombocytopenic (platelet count, 18,000/µL).

In one additional patient, a 36 year-old man reported in 1984,[21] BMN was attributed to hemolytic-uremic syndrome (HUS), diagnosed by microangiopathic hemolytic anemia, thrombocytopenia, liver function abnormalities, and anuric acute kidney injury which became end-stage kidney disease. Biopsies of both liver and kidney demonstrated arteriolar thrombi. No etiology of the HUS or potential alternative etiology for the BMN was described.[21] Although HUS has some similarities with TTP, anuric acute kidney injury and end-stage kidney disease rarely occur in patients with TTP and therefore this case description suggests the possibility of complement or coagulation-mediated TMA.[2]

CONCLUSION

How could we have made the diagnosis of ALL in this patient sooner? The knowledge from our systematic review of etiologies of BMN plus the clinical features that were atypical for TTP could have supported even more attempts to sample multiple marrow locations and to attempt additional diagnostic technologies to detect malignant marrow cells. Another consideration could have been to withhold corticosteroids and rituximab. In patients with clinically suspected TTP who have atypical features that create doubt about the diagnosis, PEX alone may be sufficient initial treatment. Withholding immunosuppressive agents may prevent a partial response of an occult hematologic malignancy that could delay the correct diagnosis or prevent an exacerbation of an occult infectious disease.[22]

Our experience emphasizes the difficulty of determining the cause of BMN. The observation of BMN requires extensive evaluation to search for hematologic or other malignant disorders. All of the malignant disorders reported to be associated with BMN may mimic the clinical features of TTP.[20]