Patients with orbital myeloid sarcoma had a better survival than those with CNS MS, non-CNS MS, as well as a better survival than patients with CNS leukemia and patients with neither myeloid sarcoma nor CNS leukemia. We are missing the location of the myeloid sarcoma in approximately one third of cases, but despite this limitation, this study represents the largest analysis of pediatric patients with myeloid sarcoma. In a previous study of extramedullary leukemia in children with newly diagnosed AML (2
), patients with non-skin extramedullary leukemia (including the CSF) had a better survival than patients with extramedullary leukemia involving the skin. We reinforced this finding since in this study, 48 of the 57 patients with non-CNS MS had skin disease, representing the majority of the non-CNS MS patients. This should provide reassurance to clinicians caring for patients with myeloid sarcoma involving CNS or orbital sites.
Our results of improved survival in patients with orbital myeloid sarcoma are in contrast to a Turkish study which found that children with AML and orbital granulocytic sarcoma had a very poor outcome with a mean survival time of 8.7 months (9
). Another study though, of children with orbital granulocytic sarcoma had similar results to ours, with all 9 of these patients surviving their disease (10
). Thus, this set of results combined with the results of this study are reassuring.
One recent study of extramedullary infiltration (EMI) in children with AML included leukemic infiltration of organs other than the liver, spleen and lymph nodes, and also included CNS leukemia in their EMI definition (11
). They found that this occurred in 56(23%) of their 240 cases and the patients with extramedullary infiltration had a higher white cell count, higher proportion of M4/M5, and a higher incidence of inv16 and 11q23 abnormalities. The EMI was isolated as skin involvement in 13(23%) of their cases. Other studies also have shown that MS is often associated with t(8;21) and M2 subtype (12
). A small study also found a high incidence of abnormalities of chromosomes 8 and 21 in patients with granulocytic sarcomas (10
). Nine of their 15 patients with granulocytic sarcoma had these chromosomal findings and all 9 of these patients had periorbital involvement of the granulocytic sarcoma, thus orbital MS. We found that there was a higher incidence of t(8;21) in patients with orbital and CNS MS, and these 2 groups of patients had a higher incidence of M2 morphology.
The significantly higher incidence of t(8;21) in the orbital MS and CNS MS patients in our study compared to the non-CNS MS, the CNS3 and the CNS 1/2 without MS patients is not the explanation for the superior outcome of these patients, based on the Cox regression analyses done. This correlates with data on the presence of t(8;21) in two small studies: A Turkish study did not find that the presence of t(8;21) had a favorable effect on prognosis of children with orbital MS (9
). As well, a small Argentinean study examined children with t(8;21) and found the survival in these patients was 70% if they had no MS, and 58% if they had a MS (15
). The 8 patients with MS in this study had an orbital MS in 5 cases and a CNS MS in 3 cases (including 1 with orbital and CNS MS). Their orbital MS and CNS MS survival was significantly lower than in our study, perhaps due to the small number of patients in their study.
The incidence of MS (orbital, CNS and non-CNS) and CNS3 in our study was 243 of 1459 (16.7%) which is slightly lower than the Kobyashi study (11
) of only 240 cases. As well, in our larger study, we found that patients with all types of MS had relatively low median white counts (8.4–13.1×103
/μL), while patients with CNS3 had a significantly higher median white blood cell count (71.1×103
Kobayashi et al found that overall survival did not differ between patients with or without extramedullary infiltration (11
). They also found that EFS was significantly lower in children with CNS leukemia compared to those without EMI or with “myeloblastoma.” We found that patients with orbital MS, followed by CNS MS had a better overall survival and even free survival than the non-CNS MS patients and patients with CNS3 status, and patients with no EML (CNS1/2 patients with no MS).
In comparing this group of patients, to the entire cohort of patients treated on trials CCG 2891 and CCG 2961, the overall survival on these protocols was 51% and 52% respectively (6
). Thus, the overall survival of 92% for patients with orbital MS and 73% for patients with CNS MS is significantly better than the cohort treated on these protocols. In fact, in comparing orbital and CNS MS patients to patients with no EML (CNS1/2 and no MS), the overall survival was significantly higher (92% and 73% respectively vs 50%).
Patients on these protocols who had MS were supposed to be treated with radiation therapy at a dose of 2000 cGy but we know that of patients on protocols 2861 and 2891, only 36% of patients with MS are known to have received radiation therapy (2
). As well, on protocol 2961 radiation therapy was prescribed to be given to all patients with myeloid sarcoma, but data was not collected documenting the administration of radiation therapy and so we cannot comment on its role in the therapy of MS. In a recent small study of pediatric granulocytic sarcoma, none of the 15 patients received local radiation therapy and these patients had an improved DFS compared to patients with AML and no granulocytic sarcoma (10
). The recent COG de novo AML protocol AAML0531 did not recommend the use of radiation therapy to treat patients with MS unless the mass was causing a significant deficit, as clinicians are trying to avoid the long term complications of radiation therapy. It will be important to capture the data on radiation therapy for patients with MS to definitively determine if the difference seen in the CNS MS patients compared to the non-CNS MS patients is related to utilization of radiation therapy.
This study represents the largest cohort by far of pediatric patients with AML examined for the presence of myeloid sarcoma at diagnosis. Patients with MS involving orbital and CNS sites had a better survival than patients with non-CNS MS as well as those with leukemia in the CSF and those with neither myeloid sarcoma nor CNS leukemia. The reason for this improved survival is not due to the presence of t(8;21), and further investigation into this group of patients is warranted.