Presenting features and imaging in childhood acute myeloid leukemia with central nervous system involvement

1 INTRODUCTION

Central nervous system (CNS) is a fairly common site for extramedullary involvement in childhood acute myeloid leukemia (AML) manifesting itself as leukemic cells in the cerebrospinal fluid (CSF), meningeal infiltrates, a solid tumor in the CNS (myeloid sarcoma), or as neurological symptoms. Symptoms related to CNS involvement include headache, cranial nerve palsies, signs of spinal cord compression, and visual disturbance. Extramedullary CNS involvement is more common in patients with myelo-monoblastic and monoblastic subtypes of AML (AML French-American-British M4 and M5 morphology) and Inv(16).1, 2 It has been considered an adverse risk factor associated with a higher risk of relapse.3-5 However, other studies on children with AML and CNS involvement have shown either no effect on overall prognosis1, 2 or even better survival in patients with myeloid sarcoma in the CNS.6 The new pediatric AML protocols have successfully reduced prophylactic CNS irradiation,7 especially in patients without CNS involvement, and Nordic Society of Pediatric Haematology and Oncology (NOPHO) protocols do not recommend routine CNS irradiation even for patients with CNS involvement. However, irradiation may be considered in the few cases in which the effect of chemotherapy on CNS disease is unsatisfactory. In this study, we aimed to characterize the presenting clinical features and results of neuroimaging in patients with initial CNS involvement at diagnosis of AML in three Nordic countries (Sweden, Finland, and Denmark).

2 METHODS

Patients with CNS involvement at diagnosis were identified from the NOPHO AML registry among all patients aged 0–17.9 years at diagnosis of AML between 2000 and 2012 in Sweden, Denmark, and Finland and treated with the protocols NOPHO AML1993 and NOPHO AML2004. Details of the treatment protocols have been described previously.8, 9

CNS involvement was defined as CSF leukocyte count ≥5/μl with the presence of blasts on cytospin, and/or clinical signs of CNS leukemia such as cranial nerve palsy, seizures, symptoms of increased cranial pressure, and/or the detection of a leukemic infiltration in the CNS by imaging, either computed tomography (CT) or magnetic resonance imaging (MRI). Neuroimaging was not performed routinely on either of the protocols. Patients with lower CSF blasts counts at diagnosis (<5/μl, CNS2 status) without clinical symptoms of CNS involvement did not receive extra intrathecal treatment and were not included in the study.

Data concerning presenting symptoms and signs, CSF examination at diagnosis, and neuroimaging results were obtained through review of patient files. Copies of all available magnetic resonance images of the neuraxis were collected for central review by an experienced neuroradiologist (MP).

3 RESULTS

Questionnaires on presenting symptoms and results of imaging of the neuraxis at diagnosis were obtained from 33 of the 34 patients (97%) with CNS involvement. Among the patients with AML diagnosed between 2000 and 2012 treated according to NOPHO AML protocols, the frequency of CNS involvement was 10% (34/356). The clinical characteristics of the patients with CNS involvement are shown in Table 1. Fifty-eight percent (19/33) of the patients had symptoms related to CNS involvement at diagnosis: visual symptoms including exophthalmos (n = 8), vomiting (n = 6), cranial nerve palsy (facial palsy n = 6, trigeminal nerve symptoms n = 1, abducens nerve palsy n = 1), headache (n = 4), convulsions (n = 3), paresis (n = 3), and irritability (n = 3). In six patients, of whom four had facial palsy, CNS leukemia was diagnosed solely based on symptoms. Fundoscopy was performed in seven patients, of whom four had visual symptoms, showing papilledema in one patient and retinal hemorrhages in another.

MRI was performed in 13 cases and images were available for central review in nine patients. CT was performed in 14 cases, of which 5 were also investigated by MRI. Imaging was often performed on the day of diagnosis (median time 0 days, range −5 to 27 days). Of the 22 children with imaging, 5 had CNS involvement by imaging, 2 had solid intracranial tumors (Table 2, patients #1 and #5), 1 had orbital tumor with intracranial contrast enhancement (Table 2, patient #10), and 2 had spinal tumors (Table 2, patients #6 and #19). Four patients had orbital tumors without intracranial tumors (Table 2, patients #3, #4, #6, and #16). In two additional cases, imaging revealed suspicion of leukemic involvement: one with discrete cerebral swelling suggesting abnormal CSF circulation (Table 2, patient #2) and one with suspected meningeal involvement (Table 2, patient #21).

Imaging demonstrated intracranial bleeding in three cases (Table 2, patients #11, #13, and #22) and nonspecific findings assessed as ischemic or toxic encephalopathy in one patient (Table 2, patient #7). Of note, seven children had mastoid bone opacification without intracranial abnormalities (three cases detected by MRI, three by CT, and one by both MRI and CT; Table 2), which was considered an incidental finding in all cases. All but two patients with mastoid bone opacification had facial palsy. In contrast to the other patients with facial palsy, these two children had clinical signs of sinonasal inflammatory disease (Table 2, patients #1 and #18).

4 DISCUSSION

Our study demonstrates that most patients with known CNS involvement, either due to blasts in the CSF or neurological symptoms, had abnormal findings by neuroimaging. However, even if an orbital mass was observed, as in several patients, intracranial leukemic tumors or meningeal enhancement were rarely present. Given the high number of abnormal findings, our study supports further studies of CNS imaging in children with AML and CNS involvement at diagnosis. The significance of CNS imaging in patients with CNS2 is unknown.

Contrary to a previous report,10 facial palsy was a relatively common presenting symptom, and opacification of the mastoid cells by imaging with or without clinical symptoms of mastoiditis was associated with facial palsy in a majority of the cases.

Several clinical case reports have been published on patients with AML having mastoiditis and facial palsy. In 1984, Todd and Bowman11 published a case report of a child with atypical mastoiditis and facial palsy as presenting symptoms of AML. Howell et al.12 and Leite da Silveira et al.13 reported one adult patient with AML having mastoiditis and facial palsy at AML relapse. Our study is the first larger series of patients showing radiological signs of mastoid cell involvement in combination with facial palsy. Opacification of the mastoid cells is a common finding in children but is not usually associated with facial palsy. A recent study on MRI performed on 515 children for other indications than mastoiditis or otitis media demonstrated a high prevalence (21%) of mastoid opacification. Mastoid cell opacification rates decreased with age ranging from 45% in children younger than 2 years to 6.5% in adolescents older than 12 years.14 In our cohort, all but one of the patients with facial palsy and mastoid cell opacification was aged 8 or older.

In this retrospective study, the imaging was not targeted to evaluate the facial nerve, and radiographic resolution may have been too low to detect facial nerve enhancement on MRI, as this was not observed in any of the patients with facial palsy. Furthermore, the radiologist may have paid particular attention to the mastoid cell system in patients with facial palsy and thus noted mastoid cell opacification in the patients with known facial palsy, whereas in other cases it may have not been considered relevant and overlooked. To better clarify the significance of mastoid cell involvement in patients with AML having facial palsy, this observation should be verified in a larger patient series, preferably with thin section contrast-enhanced magnetic resonance scans.