The purpose of this study was to reveal the clinical characteristics of nonleukemic granulocytic sarcoma (GS) and an association between the therapeutic regimens and the nonleukemic period.
The purpose of this study was to reveal the clinical characteristics of nonleukemic granulocytic sarcoma (GS) and an association between the therapeutic regimens and the nonleukemic period.
Clinical records of 2 patients reported here and 72 patients gathered using a literature search on Medline from other institutions were analyzed. The patients consisted of 57 patients who preceded acute nonlymphoblastic leukemia (ANLL) and 17 patients who did not develop ANLL. These patients were divided into 3 groups by therapeutic regimens; Group I included 12 patients who received only biopsy or surgical resection of the tumor, Group II was 20 patients who received local irradiation for the tumor, and Group III consisted of 42 patients who received systemic chemotherapy. The nonleukemic periods between these groups were compared. In Group III, the period in the patients who were treated with chemotherapy given to ANLL was compared with that in the patients who received chemotherapy used for malignant lymphoproliferative disorders (MLPDs).
Thirty-five patients (47%) initially were misdiagnosed, and the disease was most often malignant lymphoma. Preferential sites of GS were the small intestine, mediastinum, epidural site, uterus, and ovary, which often are difficult for the detection and diagnosis in addition to the skin and lymph nodes known commonly. The nonleukemic period after the diagnosis of GS was significantly longer in Group III than in the other groups (median, 12 months in Group III vs. 3 and 6 months in Groups I and II, respectively). The aggressive chemotherapy given to ANLL led to a longer nonleukemic period than the chemotherapy used for MLPDs.
To reduce the risk of subsequent ANLL in patients with nonleukemic GS, it is important that accurate histologic diagnosis is established initially for GS and that all isolated cases of GS, even those that appear to be cured by resection or irradiation of the tumor, are treated with intensive chemotherapy similar to that used to treat ANLL during the nonleukemic period as soon as possible. Cancer 2002;94:1739–46. © 2002 American Cancer Society.
Granulocytic sarcoma (GS) has been defined as a localized tumor mass composed of immature cells of granulocytic series or the cells of each maturation step in extramedullary sites. Granulocytic sarcoma develops in 2–8 % of patients with acute nonlymphoblastic leukemia (ANLL).1, 2 It usually occurs concomitantly with or after the onset of ANLL.3, 4 On rare occasions, it evolves before the onset of ANLL, and among these cases it often is misdiagnosed as malignant lymphoma.3, 5 We report two patients with GS of left testicular and intestinal involvements, respectively, that preceded leukemic involvement of the peripheral blood and bone marrow. To our knowledge, there has been no previous review comparing treatments for patients with GS in the nonleukemic stage preceding ANLL. In addition to these 2 patients, we reviewed the cases of 55 patients with GS preceding ANLL4–34 and 17 patients without the transformation during the follow-up period3, 35–45 that were well documented in the literature. Our review focused an analysis of the clinical characteristics and the relation between the therapeutic regimens of GS and the nonleukemic period.
Two patients (1%) who developed GS before the onset of ANLL were identified among 210 patients with ANLL who were treated at the Department of Hematology in Tokai University Hospital between January 1980 and December 1996. In addition, 55 patients who developed GS at least 1 month before the blood or bone marrow evidence of ANLL4–34 and 17 patients without leukemic transformation during the follow-up period3, 35–45 were gathered through a literature search on MEDICINE. Granulocytic sarcoma was diagnosed in these patients either by the positive results of naphtol ASD chloroacetate esterase (NCAE), by immunoperoxidase stainings with monoclonal antibodies for myeloperoxidase, by lysozyme or CD34 antigen,34 by electron microscopic findings of the tumor cells, or retrospectively by the transformation to ANLL. We excluded the patients who developed GS concomitantly with or after the onset of ANLL and the patients with chronic myeloproliferative disorder (CMPD). The records of 74 patients were used for clinical analysis. Because clinical data in 72 patients were gathered from the literature, we do not have enough primary data to perform a close analysis in the blood and bone marrow of these patients. We acknowledge the resultant limitation in our ability to draw conclusions in this study.
The patients were arbitrarily divided into 3 groups by the therapeutic regimens; Group I included 12 patients who received no treatment by systemic chemotherapy or irradiation after surgical resection or biopsy of the tumor, Group II included 20 patients who received local irradiation for the tumor, and Group III included 42 patients who received any regimen of chemotherapy consisting of cytotoxic drugs in addition to the surgical resection and/or local irradiation. Three groups were compared to reveal the association between the therapeutic regimens and the nonleukemic period.
Among 42 patients in Group III, 39 patients whose chemotherapeutic regimens were reported in detail were arbitrarily divided into 2 subgroups by a combination of the drugs. Twenty patients were treated with cyclophosphamide, vincristine, and prednisone (COP) or chemotherapy including anthracycline in addition to COP (CHOP46), L-asparaginase, methotrexate, bischloroethyl nitrosourea, and melphalan, i.e., agents often used for the treatment of non-Hodgkin lymphoma (NHL) or multiple myeloma (chemo. I), and the other patients received more intensive combination chemotherapy including cytosine arabinoside (Ara-C) or behenoyl Ara-C and anthracycline, i.e., agents given for ANLL47 or bone marrow transplantation (chemo. II).
The nonleukemic periods after the diagnosis of GS in the three groups were calculated according to the Kaplan and Meier method.48 The log-rank test49 was used for testing the differences among the curves. The chi-square test was used to estimate the distribution of the nonleukemic period between chemo. I and II in Group III, and P value less than 0.05 was considered to be statistically significant.
A 64-year-old man noticed swelling of his left testis 5 months before admission (October 1995). The patient underwent high orchidectomy of the left testis (March 1996). The resected testis was hardened uniformly and weighed 46 g. Microscopically, the tumor was composed of a diffuse infiltration of large cells with conspicuous nucleoli and many mitoses. Neutrophils and eosinophils infiltrated sporadically between the tumor cells. Staining using NCAE was strongly positive in the cytoplasm of the tumor cells. Immunoperoxidase stainings with monoclonal antibodies for myeloperoxidase and lysozyme were positive in the tumor section. The specimen first was interpreted as NHL and subsequently as GS. Hematologic study showed a leukocyte count of 11,800/μ L with a normal differential count (1.3% eosinophils). Repeated peripheral blood films and the first bone marrow aspiration test on right iliac bone were all negative for cytologic evidence of acute leukemia. Neither systemic chemotherapy with cytotoxic agents nor local irradiation was given after the resection. Hematologic study on June 17, 1996 revealed blast cells circulating in 11% of 6600/μL in leukocyte count. The second bone marrow aspiration test revealed an increase (65.8%) of myeloid blast cells, and the karyotypic analysis revealed 49, XY,+8, t(17;18)(p11;q11), +20, and +22. The patient received a diagnosis of ANLL classified as M2 according to the French–American–British (FAB) classification.50 Induction chemotherapy given for the initial remission consisted of daunorubicin, enocitabine, 6-mercaptopurine, and prednisolone. The patient died of cerebral hemorrhage during the first complete remission.
A 39-year-old man was admitted in June 1984 for evaluation of abdominal pain. Hematologic study showed a leukocyte count of 6900/μL with a normal differential count. Abdominal echogram revealed a tumor of 3 cm in dimension localized in ileum. On July 29, 1984, the patient underwent a right hemicolectomy and partial resection of ileum due to tumor infiltration from the ileum to the ascending colon. Immature cells resembling those observed in Case 1 infiltrated in the tumor. The cells were stained positively with NCAE and myeloperoxidase antibody. Eosinophils were observed sporadically between the infiltrated cells. The tumor was diagnosed as GS. Bone marrow aspiration test showed neither maturation arrest nor chromosomal abnormality. The patient received neither irradiation nor chemotherapy after the resection. On December 1, 1984, hematologic study showed a leukocyte count of 2900/μ L with 17% blast cells. The patient died of multiple organ failure due to leukemic cell infiltration after irradiation therapy unresponsive to the relapsed tumor.
The median age at diagnosis was 32 years in male (range, 16–84 years) and 34 years in female (range, 16–74 years) patients. Most patients (68%) received diagnosis between the ages of 20 and 44 (Fig. 1 and Table 1).
|Characteristic||No. of patients|
|Total no. of patients||74|
|Median age (yrs) (range)||33 (16–84)|
|Misdiagnosed (%)||35 (47)|
|Eosinophilia (> 0.3 × 109/L) (yes/no)||7/22|
|Eosinophilic infiltrate in tumor (yes/no)||22/9|
|Myeloid marker in tumor|
|Lysozyme (IPS) (yes/no)||8/1|
|Azurophilic granules (EM) (yes/no)||11/2|
|Type of ANLL/MDS transforming from GS|
|FAB type M0/M1/M2/M3/M4/M5/M6/M7||1/2/9/3/0/2/0/0|
|MDS (RAEB in T)||2|
Thirty-five (47%) of 74 patients initially were misdiagnosed. Among them, 31 patients were mistaken for malignant lymphoproliferative disorders (MLPDs) (NHL, lymphosarcoma, histiocytic lymphoma, reticulum cell sarcoma, lymphocytic sarcoma, thymoma, and myeloma), and 4 were mistaken for eosinophilic sarcoma and carcinoma. The misdiagnoses were not corrected until acute leukemia was suspected by bone marrow and/or blood examinations.
Myeloid origin of tumor cells was determined by NCAE staining in 36 (80%) of 45 patients examined, immunohistochemical staining for lysozyme in 8 (90%) of 9 patients, and electron microscopic study that revealed azurophilic granules consistent with promyelocytes in 11 (84%) of 13 patients. The tumor cells in four of seven patients whose cells did not stain with NCAE demonstrated lysozyme activity, and the other demonstrated azurophilic granules or immunohistochemically CD34 antigen.34
Eosinophilia of more than 300/μ L was observed in 7 (24%) of 29 patients assessed for the blood leukocyte differentials. Among 31 tumors described about pathologic histology, 22 (73%) had infiltration of eosinophilic series.
The FAB subtypes of ANLL were described in 17 of 57 patients with transformation to ANLL. Among them, 8 reviewed patients30–33 and a patient of Case 1 had M2 subtype, 39, 21, 22 had M3, 219 had M5, 226, 33 had M1, and 134 had M0. M4, M6, and M7 subtypes were not reported. Three patients23, 31 had meningeal leukemia with no bone marrow involvement. Two patients28, 29 developed refractory anemia with excess of blasts in transformation.51
As diagnosed initially, 102 tumors occurred in 74 patients. Fifteen (15%) involved the lymph nodes, 14 (14%) involved the skin, 13 (13%) the head and spinal cord, 11 (11%) the small intestine, 10 (10%) the mediastinum, 9 (9%) the bone, and 9 (9%) the ovary and uterus. Most of the skin lesions occurred on areas of the trunk and face. Cervical, supraclavicular, and axillary lymph nodes often were involved. The testicular lesion reported in Case 1 was the only one of nonleukemic GS reviewed. Fifty-one patients had single-organ involvement, and 19, 3, and 1 patient had multiple organ involvements in 2, 3, and 4 organs, respectively (Table 2).
|Tumor location||No. of patients|
|Head, spinal cord||13|
Cytogenetic analyses on the bone marrow cells of 10 patients are shown in Table 3. Chromosomal abnormalities were various among the reports. An inversion of chromosone 16 that may be associated with eosinophilia and eosinophil infiltration in GS tumor was not reported. The karyotype on the bone marrow cells of one patient29 who developed ANLL of FAB type M2 involved an 8;21 chromosome translocation that was reported previously on the tumor cells of GS.52 The analysis on the tumor cells of GS was not included in this review.
|Reference no.||Karyotype||Before or after overt ANLL|
|43||46, XY, t(12;13)(p12;q12)||Before|
|Case 2||46, XY||Before|
|5||47, XY, +G||After|
|29||45, X, −Y, −11q, t(8;21)(−q;+q)||After|
|31||47, XY, −3, +der3, −13, +der13, −14, +der14, +22[14 cells]||After|
|46, XY, −3, +der3, −13, +der13, −14, +der14[3 cells]|
|48, XY, −3, +der3, −13, +der13, +der13, +22[2 cells]|
|33||44, XX, −10, −11||After|
|34||47, XX, −5, +19, +20||After|
|Case 1||49, XY, +8, t(17;18)(p11;q11), +20, +22||After|
Figure 2 shows the association between therapeutic regimens and the nonleukemic periods in 74 patients. The medians of interval periods between GS and ANLL were 3 and 6 months in Groups I and II, respectively, with significant difference (P < 0.05). All of 32 patients in Groups I and II progressed finally to ANLL, and 26 (81%) of these patients did so within 11 months. Conversely, the period in Group III was 12 months with significant elongation as compared with the other groups (I vs. III, P < 0.0001; II vs. III, P < 0.0005). Eight patients (19%) in Group III had a nonleukemic period of more than 24 months, in contrast with only 1 (5%) in Groups I and II. This result revealed that systemic chemotherapy elongates the nonleukemic period as compared with resection or irradiation.
As shown in Table 4, 21 (60%) of 35 misdiagnosed patients received some type of chemotherapy. Among the 19 patients whose chemotherapies were reported in details, 17 were treated with chemo. I and 2 were treated with chemo. II. Twenty-one (54%) of 39 patients with a GS diagnosis received some type of chemotherapy. Among the 20 patients whose chemotherapies were reported in detail, 17 were treated with chemo. II and only 3 were treated with chemo. I.
|Initial diagnosis||Chemo. I||Chemo. II||No details in therapy||Not given|
|GS (n = 39)||3||17||1||18|
|MLPDs (n = 31)||15||2||2||12|
|Others (n = 4)||2||0||0||2|
Whether the nonleukemic period differs from chemo. I to chemo. II is reviewed in Figure 3. The period in the patients who received chemo. II was significantly longer than that in the patients who received chemo. I (P < 0.05). Thirty-two (43%) of 74 patients were not treated with any chemotherapy.
Granulocytic sarcoma occurs during the course of ANLL or before its onset. Most patients with the nonleukemic stage of GS in our review were between the ages of 20 and 44 years. This result was different from the increasing incidence with increasing age reported in ANLL.53 Therefore, nonleukemic GS may belong to a different category from GS appearing during or after ANLL.
Most of the misdiagnoses were made on patients whose tumor preceded the appearance of overt leukemia symptoms. An accurate initial diagnosis was made in only 39 (53%) of 74 patients with nonleukemic GS reviewed here. Most of the remaining patients were misdiagnosed as NHL. Consequently, incorrect treatment was performed under the misdiagnosis.
Diagnostic distinction from a high-grade NHL, especially large cell lymphoma, is difficult, especially in patients without bone marrow involvement. In an analysis of 61 tumors of GS biopsied in 26 patients with ANLL and 24 patients with CMPD, Nieman et al.4 found that the 29 (48%) tumors assessed as the blastic category showed no morphologic evidence of granulocytic differentiation and exhibited features mimicking lymphoma. The differentiated tumors may be infiltrated by eosinophilic series that provide the chance to diagnose GS but the blastic tumors hardly contained the series. They reported that 42 (75%) of 56 tumors were positive for NCAE and 42 (89%) of 47 tumors were positive for antilysozyme immunoperoxidase stains. Staining using NCAE immunophenotyping of myeloid markers such as myeloperoxidase, lysozyme, and CD34 antigen, and ultrastructural study of tumor sections have a diagnostic value for identifying myeloid origin.27, 31, 34, 55, 56 The immunophenotyping is more important in cases in which NCAE staining is negative because of the immature nature of the tumor cells.30 If the serum lysozyme concentration is determined routinely at the diagnosis of the tumor, the increase may alert the physician to the need for the strict examination to confirm GS. Several publications of NHL with coexppression of myeloid and lymphoid markers55 and GS with mixed phenotypes56 may make the choice of chemotherapeutic regimens difficult. After all, the histologic diagnosis of GS can be difficult if it is not suspected.
Acute nonlymphoblastic leukemia with an 8;21 chromosome translocation is characteristic of predisposition to solid tumor formation, young adult onset, and occasionally eosinophilia57 The translocation also has been reported on tumor cells of GS that developed after the onset of ANLL52 and on the bone marrow cells of ANLL subsequent to GS in 1 of 10 patients whose karyotypes were analyzed in this review. The association between an inversion of chromosome 16 and acute myelomonocytic leukemia with abnormal eosinophils (FAB M4Eo) was reported previously.58 Although eosinophilia and eosinophil infiltration in the tumor also occurs frequently in patients with GS, the inversion was not found in this review.
Although GS can occur virtually anywhere in the body, the most common sites in nonleukemic GS were the bone, lymph node, and skin as well as those in the tumors appearing during or after ANLL.2, 3 This review revealed that the other preferential sites were small intestine, mediastinum, epidural site, uterus, and ovary, i.e., sites often difficult for the detection and diagnosis.
Neiman et al.4 reported that, among 15 patients who showed no discernible hematologic abnormality at the initial diagnosis and received no chemotherapy, 13 (87%) developed ANLL at a mean of 10.5 months after the diagnosis of GS. In this series, 28 (88%) of 32 patients with the same conditions transformed to ANLL within 11 months. On the contrary, 25 (58%) of 42 patients given chemotherapy remained in the nonleukemic stage for more than 11 months. Meis et al.5 reported that at least 4 (25%) of 16 patients who received chemotherapy did not develop ANLL during a follow-up period of 3.5–16 years after diagnosis of GS. Among the cases reviewed here, most misdiagnosed patients were treated with cytotoxic agents used for NHL, sarcoma, or multiple myeloma, and most patients who initially were diagnosed with GS received aggressive chemotherapy coincident with ANLL. The latter chemotherapy prolonged the nonleukemic period significantly. Allogeneic bone marrow transplantation or peripheral blood stem cell transplantation also may be considered during hematologic remission.41
To our knowledge, no controlled study has been reported regarding the treatment of nonleukemic GS. However, this review and the previous reports3, 5 suggest, first, that most tumors left untreated before leukemic symptoms will almost always progress to ANLL, and second, that longer disease free intervals occur in patients initially treated with the same regimen as given for ANLL and lengthen survival period. One report54 suggests that GS occurring before manifest ANLL responds better to cytotoxic treatment and has a better prognosis than leukemic GS. Although irradiation therapy for the tumor is effective for the high radiosensitivity, it did not improve the disease free interval and the prognosis, suggesting that GS is a partial manifestation of a systemic disease. This review stresses that all isolated cases of GS, even these that appear to be cured by resection or irradiation, should be treated with intensive chemotherapy similar to that used to treat ANLL. Hayashi et al.59 demonstrated AML1/MTG8 fusion mRNA by reverse transcription–polymerase chain reaction (RT-PCR) in the bone marrow cells of a patient with nonleukemic GS. The RT-PCR may give us a useful tool in deciding a start of chemotherapy for patients with nonleukemic GS.
Initial correct diagnosis of GS and early start of antileukemic therapy would promise a longer survival for these patients who are mostly young.