Granulocytic Sarcoma (GS) Is Associated with Event-Free Survival Superior to That of Acute Myeloid Leukemia (AML).

Introduction. Granulocytic sarcoma (GS) is an extramedullary tumor composed of immature myeloid cells. Some GS patients develop acute myeloid leukemia (AML), but the outcomes of these two diseases have never been compared. The purpose of this retrospective analysis was to compare the rates of response, survival and event-free survival (EFS) of patients treated for GS and patients treated for AML, after adjusting for presenting characteristics.

Patients and Methods. We identified 1720 AML and 24 GS patients who presented at The University of Texas M. D. Anderson Cancer Center from 1990 to 2004. All the AML patients and 17 of the GS patients received ara-C + idarubicin or fludarabine. After comparing outcomes in these patients, we accounted for possible differences in underlying prognosis by matching treated GS and AML patients, using cytogenetics, age +/− 3 years, Zubrod performance status (0–2 vs. >2), and time of treatment +/− 3 years as the criteria for matching. Among 9 GS patients age ≤60 years, cytogenetic abnormalities were found at the sarcoma site or in the bone marrow in 5 patients; 4 of these 5 patients had a +8 abnormality, so we matched GS patients ≤60 years with cytogenetics that were “normal” in bone marrow but not assessed in GS tissue with AML patients with a +8 abnormality, assuming the “normal” karyotype in the GS patients was possibly non-informative and thus falsely negative. Among 15 GS patients <60 years old, cytogenetic abnormalities were inv(16) in 2, +8 in 2, 11q del in 1, −7 in 1, and miscellaneous in 2; 7 patients had normal bone marrow cytogenetics but no assessment in GS tissue. On the basis of the frequency of known cytogenetic abnormalities, we matched each of the 7 non-informative GS patients with better-, intermediate-, or worse-risk AML patients. EFS duration was compared in pair-mates of patients, and the Bayesian posterior probability that GS is associated with longer EFS duration was computed.

Results. Among treated patients (GS 17, AML 1720), complete response rates were 71% and 57% in GS and AML, respectively (p=0.28). The respective 2-year EFS rates were 82% and 63% (p=0.049), and 2-year overall survival rates were 94% and 77% (p=0.16). Matches could be found for 14 GS patients, who were matched with 91 AML patients (including 3 matched with 2 different GS patients each). The median number of matches was 4 (range, 3–21). EFS duration was longer in 27 AML pairmates, shorter in 55, and unevaluable owing to censoring in 9. Assuming a beta distribution and a priori that 41 matches would have favored GS and 41 AML if EFS was the same in GS and AML, the posterior probability that EFS duration was longer in GS was 0.99. Eliminating the matches involving 1 GS patient and 21 AML patients led to a 0.92 posterior probability of longer EFS duration in GS.

Conclusions. Our results suggest that GS is associated with better prognosis than AML based on the superior EFS of GS patients. Differences in rates of CR and survival did not reach statistical significance, possibly because of the small number of patients with GS.