Extramedullary Relapse of Acute Leukemia
after Allogeneic Hematopoietic Stem Cell
Transplantation: Different Characteristics
between Acute Myelogenous Leukemia and
Acute Lymphoblastic Leukemia
, Fan Ye1
, Xinliang Mao2
, Jia Chen1
, Aining Sun1
, Huiying Qiu1
, Zhengming Jin1
, Miao Miao1
, Xiao Ma1
, Feng Chen1
, Shengli Xue1
, Depei Wu1,*
, Xiaowen Tang1,* 1Department of Hematology, First Afﬁliated Hospital of Soochow University, Jiangsu Institute of Hematology, Collaborative Innovation Center of Hematology, Key Laboratory of Thrombosis and Hemostasis, Ministry of Health, Suzhou, China
2Cyrus Tang Hematology Center, Jiangsu Institute of Hematology, First Afﬁliated Hospital of Soochow University, Soochow University, Suzhou, China
Article history: Received 8 January 2014 Accepted 25 March 2014 Key Words:
Acute lymphoblastic leukemia Acute myelogenous leukemia Allogeneic hematopoietic stem cell transplantation
a b s t r a c t
Extramedullary relapse (EMR) of acute leukemia (AL) after allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a contributor to post-transplantation mortality and remains poorly understood, especially the different characteristics of EMR in patients with acute myelogenous leukemia (AML) and those with acute lymphoblastic leukemia (ALL). To investigate the incidence, risk factors, and clinical outcomes of EMR for AML and ALL, we performed a retrospective analysis of 362 patients with AL who underwent allo-HSCT at the First afﬁliated Hospital of Soochow University between January 2001 and March 2012. Compared with patients with AML, those with ALL had a higher incidence of EMR (12.9% versus 4.6%;P¼.009). The most common site of EMR was the central nervous system, especially in the ALL group. Multivariate analyses identiﬁed the leading risk factors for EMR in the patients with AML as advanced disease status at HSCT, hyperleukocytosis at diagnosis, history of extramedullary leukemia before HSCT, and a total body irradiationebased conditioning regimen, and the top risk factors for EMR in the patients with ALL as hyperleukocytosis at diagnosis, adverse cytogenetics, and transfusion of peripheral blood stem cells. The prognosis for EMR of AL is poor, and treatment options are very limited; however, the estimated 3-year overall survival (OS) was signiﬁcantly lower in patients with AML compared with those with ALL (0 versus 18.5%;P¼.000). The characteristics of posteallo-HSCT EMR differed between the patients with AML and those with ALL, possibly suggesting different pathogenetic mechanisms for EMR of AML and EMR of ALL after allo-HSCT; further investigation is needed.
Ó2014 American Society for Blood and Marrow Transplantation.
Although allogeneic hematopoietic stem cell trans-plantation (allo-HSCT) is a potentially curative treatment for patients with acute leukemia (AL), relapse remains the most frequent cause of treatment failure and mortality. In partic-ular, a signiﬁcant rate of extramedullary relapse (EMR) after allo-HSCT has been reported [1-6], with a poor prognosis. Although risk factors for EMR have been described in patients with acute myelogenous leukemia (AML), few studies have compared the different characteristics of EMR in patients with AML and patients with acute lymphoblastic leukemia (ALL) in the post-HSCT setting.
In an effort to better understand post-HSCT EMR, we performed a retrospective analysis on 362 patients with AL who underwent allo-HSCT in the First afﬁliated Hospital of Soochow
University between January 2001 and March 2012. We studied the incidence, risk factors, treatments, outcomes, and mecha-nisms of post-HSCT EMR in patients with ALL and AML. PATIENTS AND METHODS
A total of 362 patients with AL who underwent allo-HSCT in our institution between January 2001 and March 2012 were enrolled in this retrospective study, including 208 patients with AML, 147 with ALL, and 7 with acute mixed lineage leukemia (AMLL). These patients included 204 males and 158 females, with a median age of 32 years (range, 3 to 63 years). The graft donor sources for allo-HSCT were unrelated in 27.1% of cases, sibling in 58.3%, and haploidentical in 14.6%; 79.6% of the transplants were HLA-matched. Stem cell sources included bone marrow (BM) in 45.6%, peripheral blood (PB) stem cells (PBSCs) in 36.5%, and BM plus PBSCs in 17.9%. In this study, 260 patients (71.8%) received a modiﬁed busulfan (Bu)/cyclophosphamide (Cy) conditioning regimen, 89 (24.6%) received a total body irradiation (TBI)/Cy regimen, and 13 (3.6%) received a non-myeloablative conditioning regimen. Patient characteristics are summarized inTable 1. Patients were treated according to clinical protocols approved by the Institutional Review Board and/or institutional care practice guidelines of Soochow University.
Of the 362 patients, 260 patients received a modiﬁed Bu/Cy regimen (Me-CCNU 250 mg/m2/day on day -10, Ara-C 4 g/m2/day on days9 to8, Financial disclosure:See Acknowledgments on page 1046.
*Correspondence and reprint requests: Xiaowen Tang and Depei Wu, Department of Hematology, First Afﬁliated Hospital of Soochow University, Jiangsu Institute of Hematology, 188 Shizi Rd, Suzhou 215006, PR China.
E-mail addresses: firstname.lastname@example.org (D. Wu), xwtang1020@ 163.com(X. Tang).
1083-8791/$esee front matterÓ2014 American Society for Blood and Marrow Transplantation. http://dx.doi.org/10.1016/j.bbmt.2014.03.030
American Society for Blood
Bu 0.8 to 1 mg/kg/6 hours on days7 to5, and Cy 1.8 g/m2/day on days4 to3) as a preparative regimen, and 89 patients received a TBI/Cy regimen consisting of fractionated TBI 12 Gy on days8 to6, Ara-C 2 g/m2/day on day5, and Cy 1.8 g/m2/day on days4 to3. Thirteen patients received nonmyeloablative Bu/ﬂudarabine (Flu) conditioning regimen (Bu 0.8 mg/kg/ 6 hour on days6 to5, Flu 30 mg/m2/day on days7 to2, and Ara-C 0.5 g/m2/day on days8 to4).
Graft-Versus-Host Disease Prophylaxis
Acute graft-versus-host disease (aGVHD) and chronic GVHD (cGVHD) were classiﬁed according to the criteria proposed by Przepiorka et al.and Sullivan. GVHD prophylaxis included cyclosporine A with short-term methotrexate for related identical transplantation and cyclosporine A, methotrexate, mycophenolate mofetil, and human antithymocyte globulin for unrelated and haploidentical transplantation. The haploidentical grafts were noneT cell depleted.
Central Nervous System Relapse Prophylaxis after Transplantation Prophylactic intrathecal chemotherapy (with methotrexate or cytosine arabinoside and dexamethasone) each month for 6 months was
administered to all patients with ALL and high-risk AML in theﬁrst year after allo-HSCT.
Deﬁnitions and Statistical Analysis
EMR included isolated EMR and EMR with concurrent bone marrow relapse (BMR); isolated EMR was deﬁned as EMR without concurrent BMR. EMR was diagnosed in most cases by magnetic resonance imaging, computed tomography, or positron emission tomography. Histological conﬁrmation was performed whenever possible. Central nervous system (CNS) relapse was diagnosed when leukemic cells were identiﬁed in the cerebrospinalﬂuid. Isolated CNS relapse was deﬁned as CNS relapse without any other sites of leukemia relapse. Clinical remission (CR) was deﬁned as <5% blasts in the BM with normal complete blood count values. High-risk cytogenetics was deﬁned as complex karyotype, 5q-, monosomy 7/7q-, and/or FLT-3epositive for AML and MLL gene rearrangement and/or Philadelphia chromosome in ALL. Hyperleukocytosis was deﬁned as a peripheral WBC count>50109/L at diagnosis of AML, B lymphoblastic leukemia as>30109/L, and T lymphoblastic leukemia as>100109/L.
The cumulative incidence estimation for leukemia relapse was deter-mined by treating death as a competing risk and was compared using the method of Gray. Overall survival (OS) was calculated from the date of relapse post-transplantation to death or last follow-up for censored cases, and was estimated using the Kaplan-Meier method and compared with a log-rank test. Continuous variables were compared with the Mann-Whitney test, and categorical variables were compared with the chi-square test. A comprehensive multivariate analysis of risk factors for relapse was esti-mated using the Cox proportional hazards regression model. All statistical analyses were performed with SPSS version 16.0 (SPSS Inc, Chicago, IL) and R version 2.15.1 (R Project for Statistical Computing, Vienna, Austria).
We retrospectively analyzed a consecutive series of 362 patients with AL who underwent allo-HSCT in our single institution. Of these 362 patients, 163 of 208 with AML, 110 of 147 with ALL, and 5 of 7 with AMLL were inﬁrst CR status (CR1) at the time of allo-HSCT. After transplantation, 40.1% patients experienced aGVHD, including 34.0% with mild (grade I-II) and 6.1% with severe (grade III-IV) aGVHD, and 28.2% patients developed cGVHD. After a follow-up ranging from 1 month to 139 months (median, 17 months), 90 pa-tients (24.9%) developed either EMR or BMR; 64 of these relapsed patients had isolated BMR only, 11 had isolated EMR, and 15 had EMR with concurrent BMR.
Incidence and Characteristics of EMR in AL
The estimated 10-year cumulative incidence of overall relapse posteallo-HSCT was 27%. That of EMR was 7.9%, with 3.1% of patients experiencing isolated EMR. Compared with patients with AML, those with ALL had a higher estimated 10-year cumulative incidence of EMR (12.9% versus 4.6%;
P¼.009) (Figure 1). Table 1
Clinical Characteristics of 362 Patients with AL Post-HSCT
Characteristic AML (n¼208) ALL (n¼147) AMLL (n¼7) Age, yr, median (range) 35 (3-63) 25 (4-57) 26 (20-54) Sex, n
Male 116 83 5
Female 92 64 2
Hyperleukocytosis at diagnosis, n 66 52 1 EM leukemia before HSCT, n 14 13 0 High cytogenetic risk, n 20 65 3 Disease status at HSCT, n CR1 163 110 5 CR2 23 36 0 NR 22 1 2 Conditioning regimen, n Bu/Cy 185 72 3 TBI/Cy 11 74 4 NST 12 1 0
Stem cell source, n
BM 100 62 3 PB 72 58 2 BMþPB 36 27 2 Donor type, n Sibling 142 66 3 URD 43 53 2 Haploidentical 23 28 2 aGVHD grade, n I-II 67 51 5 III-IV 13 9 0 cGVHD, n 61 38 3
NR indicates no remission; NST, nonmyeloablative conditioning regimen; URD, unrelated donor.
Figure 1.(A) Cumulative incidence of EMR after allo-HSCT for AL. EMR with or without BMR (dotted line), 7.9% (n¼26); isolated EMR (solid line), 3.1% (n¼11); EMR with concurrent BMR (dashed line), 4.7% (n¼15). (B) Cumulative incidence of EMR after HSCT for AML versus ALL (4.6% versus 12.9%;P¼.009).
Twenty-six of the 362 patients experienced EMR after allo-HSCT, including 11 with isolated EMR and 15 with EMR and concurrent BMR. Of these latter 15 patients, 6 who initially had EMR later developed BMR, 4 with BMR later developed EMR, and the remaining 5 developed EMR and BMR almost simultaneously. In our center, the median time to BMR post-HSCT was 5 months (range, 1 to 60 months), and the median time to EMR was 4 months (range, 1 to 38 months; P¼.879). Involved extramedullary (EM) sites included the CNS, testis, skin, soft tissue, bone, lymph nodes, nasopharynx, and peritoneum. Multifocal involvement at EMR was observed in 5 of 26 patients (19.2%). CNS relapse (n¼18) was the most common type of EMR, with a cumu-lative incidence of 5.4%, whereas the cumucumu-lative incidence of isolated CNS relapse was 2.5%.
Among the 26 patients who developed EMR, 5 had a history of EM leukemia before undergoing allo-HSCT, all of whom relapsed at the same site as in the previous leukemia. Ten patients had mild aGVHD (grade I-II), 3 had severe aGVHD (grade III-IV), and 3 developed cGVHD before the onset of EMR. Characteristics of the 11 patients who had isolated EMR and the 15 patients who had EMR with con-current BMR are presented inTables 2 and 3.
Risk Factors for Overall Relapse and EMR in AL
We analyzed the variables of interest by Cox proportional hazard modeling to identify risk factors for AL relapse, including age, sex, disease type, donor type, stem cell source, hyperleukocytosis at diagnosis, cytogenetic risk, disease status at HSCT, HLA mismatch, conditioning regimen, EM leukemia before HSCT, aGVHD, and cGVHD. The results of our univariate and multivariate analyses are summarized in Tables 4 and 5.
In terms of overall relapse, our multivariate analyses showed that patients with high-risk cytogenetics, advanced disease status, TBI-based conditioning regimen, without cGVHD, and male sex were more likely to relapse (Table 4). The multivariate analyses also revealed a higher rate of EMR in patients with high-risk cytogenetics, advanced disease status, and male sex (relative risk [RR] ¼3.860, P¼.002; RR¼6.663,P¼.003; and RR¼2.844,P¼.038, respectively). A history of EM leukemia before undergoing HSCT and hyperleukocytosis at diagnosis also correlated with an increased risk of EMR (RR¼3.011;P¼.038 and RR¼3.382;
P ¼.004, respectively). Meanwhile, patients who received PBSC grafts were more likely to develop EMR (RR¼5.495;
P¼.002) (Table 5).
To identify differences in pathogenesis between BMR and EMR, we performed further analyses with separate consid-eration of isolated BMR. Patients with advanced disease status had a higher frequency of BMR, and also were more likely to develop EMR. Multivariate analyses also revealed that the patients without cGVHD were more likely to experience BMR (RR¼5.907;P¼.000); however, there was no signiﬁcant difference in the incidence of EMR between patients with cGVHD and those without cGVHD. In addition, the incidence of BMR was higher in patients who received a TBI-based conditioning regimen or nonmyeloablative conditioning regimen compared with patients who received a Bu-containing regimen (RR ¼ 2.100; P ¼ .012 versus RR¼3.159;P¼.031) (Table 5).
Differences in Risk Factors for EMR in AML and ALL To identify differences in the pathogenesis of EMR be-tween patients with AML and those with ALL, we performed further multivariate analyses with these 2 groups of patients separately. We found that in the patients with AML, advanced disease status, a history of EM leukemia before HSCT, use of a TBI-based conditioning regimen, and hyper-leukocytosis at diagnosis were associated with a higher fre-quency of EMR (RR¼16.264,P¼.001; RR¼10.416,P¼.011; RR¼10.455,P¼.035; and RR¼4.699,P¼.041, respectively), whereas in the patients with ALL, the top risk factors for EMR were receipt of PBSC grafts, hyperleukocytosis at diagnosis, and high-risk cytogenetics (RR¼6.015,P¼.007; RR¼2.996,
P¼.032; and RR¼2.889,P¼.042, respectively) (Table 6). Postrelapse Treatment and Clinical Outcomes in AL
Four patients refused any treatment and died of relapse. The remaining 22 patients who experienced EMR with or without BMR received various treatment modalities. Nine patients developed isolated CNS relapse and were treated with intrathecal chemotherapy and/or cranial irradiation. One patient who experienced EMR in both the CNS and peritoneum received local chemotherapy, and another pa-tient who developed EMR in lymph nodes was treated with chemotherapy and donor lymphocyte infusion (DLI); how-ever, both patients died of progressive disease. Eight patients received chemotherapy, followed by local irradiation and DLI; 7 of these patients achieved CR, and 2 were still sur-viving as of the date of this report. The other 5 patients died of GVHD, leukemia recurrence, or CNS relapse. Two patients received chemotherapy and local irradiation, and both died
Characteristics of Patients with AL Who Developed Isolated EMR after HSCT Patient Sex Age,
yr Disease Disease Status at HSCT EM Leukemia before HSCT
Donor Type Stem Cell Source
Relapse Site GVHD before Relapse
Results Direct Cause of Death
1 F 31 ALL CR1 No URD PB Yes CNS cGVHD Alive
2 M 22 ALL CR1 No URD PB Yes CNS aGVHD grade I Dead Infection
3 M 36 ALL CR1 No URD PB Yes CNS No Alive
4 F 22 ALL CR2 No Haploidentical PB No CNS aGVHD grade I Dead Relapse
5 M 48 ALL CR2 No Sibling PB Yes CNS aGVHD grade III
Alive 6 M 10 ALL CR3 CNS Haploidentical BMþPB No CNS aGVHD grade IV Alive
7 M 44 ALL CR4 No Sibling PB Yes CNS No Dead Infection
8 F 35 ALL CR1 No Sibling BMþPB Yes CNS, peritoneum aGVHD grade I Dead Relapse
9 M 18 ALL CR2 Lymph
URD PB Yes Lymph nodes No Dead Relapse
10 M 44 AML CR1 No Sibling BM Yes CNS No Dead Relapse
from leukemia recurrence. The remaining patient was treated with supportive care and died of progressive disease. In general, the estimated 3-year OS postrelapse was bet-ter in patients with isolated EMR than in patients with BMR only and those with EMR with concurrent BMR, but the difference was not statistically signiﬁcant (18.2% versus 12.0% versus 8.0%;P¼.206) (Figure 2). Although the median OS postrelapse was 9 months versus 4 months versus 4 months, these differences were not signiﬁcance either (P ¼.099). However, the estimated 3-year OS in these patients with EMR was signiﬁcantly lower in those with AML compared with those with ALL (0 versus 18.5%;P¼.000).
EMR after allo-HSCT is relatively rare and not well stud-ied. Although most previous studies reported on patients with AML, there are too little data comparing characteristics of EMR between patients with AML and patients with ALL. To our knowledge, this is theﬁrst report to examine the dif-ferences in EMR occurring after allo-HSCT in patients with AML and those with ALL. We found that signiﬁcant differ-ences between the 2 groups of patients.
The incidence of EMR after allo-HSCT varies among transplantation centers, ranging from 6% to 20% in single-institution reports [5,6,10-12]; however, in a retrospective European Group for Blood and Marrow Transplantation (EBMT) survey, the incidence of post-HSCT EMR was only 0.65% in patients with AML. The lower incidence of EMR found in that multicenter study may be related to under-reporting in retrospective registry data, as well as to the greater number of patients reported in recent series owing to longer follow-up, longer survival, and generally improved outcomes of HSCT. In general, the incidence of EMR is higher in patients with ALL compared with those with AML [4-6]. In our center, the estimated 10-year cumulative inci-dence of overall post-HSCT relapse in patients with AL was 27%; that of EMR was 7.9%, with 3.1% of patients experiencing isolated EMR. The incidence of EMR was 12.9% in patients with ALL and 4.6% in those with AML (P¼.009).
Historical risk factors for EMR after HSCT include Phila-delphia chromosomeepositive AL, AML subtype M4/M5, age
<18 years at diagnosis, EM leukemia before HSCT, adverse cytogenetics, relapse/refractory disease at transplantation, Bu/Cy preconditioning, and CD56 and T cell marker expres-sion[4,6,13,14]. In the present study, identiﬁed risk factors included EM leukemia before HSCT, adverse cytogenetics, and relapse/refractory disease at transplantation, consistent with previous studies[4,6,13]; however, AML subtype M4/ M5 and Bu/Cy preconditioning were not risk factors in our study cohort. In addition, we identiﬁed male sex, hyper-leukocytosis at diagnosis, TBI-based conditioning regimen, and transfusion of PBSCs as risk factors for EMR. These ﬁndings have not been reported previously in the post-HSCT setting; probable reasons for this include the following:
1. The incidence of EMR was high in our male patients owing to the relatively high rate of EMR with a testis leukemia subtype in our cohort (5 of 26).
2. Hyperleukocytosis has been identiﬁed as a risk factor for CNS involvement, but not in the post-transplantation setting ; we also found it corre-lated with an increased risk of EMR after allo-HSCT. Patients with hyperleukocytosis had a high tumor burden, which facilitates the inﬁltration of tumor cells to EM sites. Table 3 Char act eristics of P atients with AL Who Dev eloped EMR with Concurr ent BMR after HSCT Patient Sex Age, yr Disease Disease Status at HSCT EM Leukemia before HSCT Donor Type Stem Cell Source HLA Match Relapse Site Order of Relapse Month s b etween Relap ses First Relapse Remission GVHD before Relapse Results Direct Cause of Deat h 1 F 38 ALL CR1 No URD PB Yes BM, CNS * 3 1 Yes No Dead GVHD 2 M 18 ALL CR1 No Sibling BM þ PB Yes BM, CNS * 8 Yes aGVHD grade I Dead Relapse 3 M 21 ALL CR1 No URD PB Yes BM, CNS * 3 Yes aGVHD grade I Dead Relapse 4 M 25 ALL CR1 No URD PB Yes BM, CNS, nasopharynx * 2 Yes aGVHD grade II Dead Relapse 5 M 42 ALL CR1 No Sibling BM Yes BM, CNS, testis y 1 2 Yes aGVHD grade I Alive 6 M 17 ALL CR1 No Sibling BM Yes BM, testis y 3 Yes No Dead Relapse 7 M 8 ALL CR1 No Sibling BM Yes BM, testis, soft tissue z 0 N o Alive 8 M 18 ALL CR1 No URD PB Yes BM, bone * 8 N o N o Dead Relapse 9 M 42 AML CR1 No Sibling PB Yes BM, CNS * 3 Yes aGVHD grade I Dead Relapse 10 M 1 8 AML CR3 CNS URD PB No BM, CNS y 4 Yes No Dead Relapse 11 F 3 4 AML NR No URD PB Yes BM, CNS y 1 Yes aGVHD grade I Dead Relapse 12 M 4 1 AML CR1 No Sibling BM Yes BM, skin z 0 N o Dead Relapse 13 M 3 5 AML NR No Haploidentical BM þ PB No BM, skin z 0 aGVHD grade I Dead Relapse 14 M 2 5 AML NR Skin, testis Haploidentical BM þ PB No BM, testis z 0 N o Dead GVHD 15 M 1 1 AML NR No Haploidentical BM þ PB No BM, testis, soft tissue, sacrum z 0 aGVHD grade III and cGVHD Dead Relapse * EMR followed by BMR. yBMR followed by EMR. zBMR and EMR occurred at the same time.
3. There are paradoxical results from previous studies regarding pretransplantation conditioning regimens. One retrospective study found a higher incidence of isolated EMR associated with Bu/Cy regimens
compared with TBI-containing regimens . This result may be related to the low plasma levels of Bu achieved in some patients; however, other more controversial results failed to show such a difference Table 4
Univariate and Multivariate Analyses of Risk Factors for Overall Relapse in Patients with AL (n¼90)
Factor Univariate RR (95% CI) PValue Multivariate RR (95% CI) PValue
Sex, male/female 1.826 (1.168-2.853) .008 1.696 (1.082-2.657) .021 Age,18 yr/>18 yr 1.565 (0.960-2.551) .072 Disease AML 1.000 ALL 1.756 (1.154-2.672) .009 AMLL 1.598 (0.386-6.613) .518 Donor type Sibling 1.000 URD 1.522 (0.942-2.459) .086 Haploidentical 2.361 (1.375-4.052) .002 HLA mismatch/match 1.732 (1.084-2.767) .022
Stem cell source
PB 1.302 (0.817-2.077) .267
BMþPB 1.376 (0.791-2.393) .258
Hyperleukocytosis at diagnosis, yes/no 1.223 (0.799-1.873) .354
Cytogenetic risk, high/intermediate-low 2.190 (1.428-3.357) .000 1.704 (1.054-2.755) .030 Disease status at HSCT
CR1 1.000 1.000
CR2 2.621 (1.605-4.282) .000 1.838 (1.087-3.110) .023
NR 5.140 (2.844-9.289) .000 7.123 (3.855-13.163) .000
EM leukemia before HSCT, yes/no 1.617 (0.812-3.220) .172 Conditioning regimen Bu/Cy 1.000 1.000 TBI/Cy 2.147 (1.384-3.329) .001 1.846 (1.112-3.064) .018 NST 1.686 (0.610-4.658) .314 2.262 (0.806-6.349) .121 aGVHD, no/yes 0.857 (0.564-1.303) .471 cGVHD, no/yes 4.547 (2.282-9.059) .000 5.127 (2.551-10.304) .000 Table 5
Univariate and Multivariate Analyses of Risk Factors for EMR (n¼26) and BMR (n¼64) in Patients with AL
Factor EMR BMR Univariate RR (95% CI) PValue Multivariate RR (95% CI) PValue Univariate RR (95% CI) PValue Multivariate RR (95% CI) PValue Sex, male/female 3.486 (1.314-9.250) .012 2.844 (1.060-7.631) .038 1.543 (0.926-2.572) .096 Age,18/>18 yr 2.224 (0.967-5.117) .060 1.434 (0.780-2.636) .247 Disease AML 1.000 1.000 ALL 2.911 (1.296-6.539) .010 1.509 (0.916-2.484) .106 AMLL 0.000 (0.000-) .977 2.005 (0.480-8.376) .340 Donor type Sibling 1.000 1.000 URD 2.464 (1.025-5.926) .044 1.249 (0.697-2.236) .455 Haploidentical 3.236 (1.170-8.950) .024 2.098 (1.105-3.983) .024 HLA mismatch/match 1.735 (0.727-4.140) .214 1.737 (0.996-3.029) .052 Stem cell source
BM 1.000 1.000 PB 4.425 (1.605-12.199) .004 5.495 (1.880-16.062) .002 0.855 (0.484-1.510) .589 BMþPB 3.271 (0.998-10.719) .050 2.583 (0.752-8.870) .132 1.120 (0.588-2.135) .730 Hyperleukocytosis at diagnosis, yes/no 2.394 (1.107-5.176) .027 3.382 (1.484-7.707) .004 0.961 (0.566-1.630) .881 Cytogenetic risk, high/
intermediate-low 2.999 (1.385-6.495) .005 3.860 (1.647-9.045) .002 2.013 (1.201-3.375) .008 Disease status at HSCT CR1 1.000 1.000 1.000 CR2 2.770 (1.126-6.813) .026 1.154 (0.431-3.087) .775 2.598 (1.447-4.664) .001 2.052 (1.098-3.835) .024 NR 4.451 (1.467-13.505) .008 6.663 (1.923-23.085) .003 5.488 (2.728-11.043) .000 7.387 (3.593-15.189) .000 EM leukemia before HSCT, yes/no 3.427 (1.291-9.093) .013 3.011 (1.064-8.521) .038 1.029 (0.374-2.833) .955 Conditioning regimen Bu 1.000 1.000 1.000 TBI 2.645 (1.210-5.780) .015 1.959 (1.151-3.334) .013 2.100 (1.179-3.739) .012 NST 0.000 (0.000) .979 2.247 (0.802-6.292) .123 3.159 (1.114-8.956) .031 aGVHD, no/yes 0.635 (0.294-1.371) .248 0.958 (0.580-1.584) .868 cGVHD, no/yes 3.694 (1.108-12.320) .033 5.085 (2.193-11.791) .000 5.907 (2.531-13.784) .000
[10,17]. Oshima et al.reported a higher incidence of CNS relapse in patients who received TBI-containing preconditioning, possibly because a signiﬁcantly higher proportion of patients with CNS leukemia received such a regimen compared with those without CNS leukemia (81.5% versus 57.9%;P<.001). The role of conditioning regimen in preventing EMR has not been evaluated precisely. In our study, use of a TBI-containing regimen was associated with a higher incidence of EMR compared with use of a Bu/Cy regimen in the patients with AML, but not in those with ALL. This difference may be related to the fact that all 11 patients with AML who received a TBI-containing regimen were high-risk patients; 5 had a history of EM leukemia before HSCT, 3 patients were not in remission
at transplantation, 2 patients were in CR2 or greater (CR2) at transplantation, and 1 patient had adverse cytogenetics.
4. Transplantation of PBSCs was identiﬁed as a risk factor for EMR. A possible explanation for this may be the higher rate ofCR2/refractory disease status at the time of transplantation in the patients who received PBSC grafts compared with those who received BM grafts (28.8% versus 15.1%;P¼.004). Another possible explanation is that transfusion of PBSCs may be related to the presence of a graft-versus-leukemia (GVL) effect that did not protect EM sites after allo-HSCT. The increased incidence of GVHD in patients with EMR implies that GVL surveillance preferentially maintains remission in the BM while allowing leukemic cells in peripheral tissues to evade immune surveillance. Furthermore, we identiﬁed transfusion of PBSCs as a independent risk factor for EMR in patients with ALL, but not in those with AML, suggesting that the GVL effect may be less effective at EM sites in ALL. This may be related to the induction of T cell anergy by ALL cells , inadequate expression of important cos-timulatory molecules or adhesion molecules[20,21], or ALL resistance to killing by natural killer cells. Previous research also has shown that the GVL effect may be less effective at EM sites in patients with AML [10,23,24]. The patient heterogeneity and small sample size in a single transplantation center preclude draw-ing conclusions regarddraw-ing the association of EMR and these risk factors; multicenter studies with larger numbers of patients are needed.
In the present study, the development of cGVHD was protective against relapse in the BM, but this protective effect did not extend to EM sites. Multivariate analyses showed that the patients without cGVHD were more likely to experience BMR. Thisﬁnding suggests that the pathogenesis of EMR differs from that of BMR and identiﬁes the EM sites as potential sanctuary sites for leukemia cells, possibly because Table 6
Characteristics of Patients with AML and Patients with ALL
Characteristics EMR in AML (n¼9) EMR in ALL (n¼17) PValue
Cumulative incidence, % 4.60 12.90 .009 EMR site, n* CNS 5 13 .382 Testis 2 3 1.000 Skin/soft tissue 3 1 .104 Bone 1 1 1.000 Lymph nodes 0 1 1.000 Nasopharynx 0 1 1.000 Peritoneum 0 1 1.000 EM leukemia before HSCT, n 2 3 1.000 Risk factorsy
Hyperleukocytosis at diagnosis (RR¼4.699;P¼.041) Hyperleukocytosis at diagnosis (RR¼2.996; P¼.032)
Advanced disease status (RR¼16.264;P¼.001) High-risk cytogenetics (RR¼2.889;P¼.042) EM leukemia before HSCT (RR¼10.416;P¼.011) PB stem cell source (RR¼6.015;P¼.007) TBI-based conditioning regimen (RR¼10.455;P¼.035)
Type of therapy, n
Local 2 8 .399
Systemic 4 7 1.000
None 3 2 .302
Estimated 3-yr OS, % 0 18.50 .000
*Five of 26 patients (19%) presented with extramedullary disease in multiple sites.
yThe variables were analyzed by Cox proportional hazard modeling to identify the risk factors for relapse, including age, sex, donor type, stem cell source,
hyperleukocytosis at diagnosis, cytogenetic risk, disease status at HSCT, HLA mismatch, conditioning regimen, EM leukemia before HSCT, aGVHD and cGVHD, ALL with or without the Philadelphia chromosome, and AML subtype M4/M5.
Figure 2.Estimated 3-year OS for patients with AL relapse post-HSCT. Isolated EMR (solid line), 18.2%; isolated BMR (dashed line), 12.0%; EMR with concur-rent BMR (dotted line), 8.0%;P¼.206.
of decreased expression of HLA minor histocompatibility antigens and adhesion molecules.
To date, there are no standardized therapeutic strategies for EMR. Generally, some combination of systemic and local therapy should be considered, given that local therapy alone often results in subsequent systemic relapse. Local therapy includes surgical excision, intrathecal injection, and/or ra-diation, and systemic therapy involves immunotherapy with chemotherapy, DLI, and second allo-HSCT. The optimal therapeutic strategy remains controversial, however. Recent studies have reported good treatment response with some immune-targeting drugs, including gemtuzumab ozogami-cin , a recombinant humanized monoclonal antibody that targets the CD33 antigen, which is expressed in>90% of leukemic cells. Sorafenib, a multikinase inhibitor, has sig-niﬁcant activity against FLT3-ITDþ blasts in vitro and demonstrated encouraging clinical results as a single agent and in combination with chemotherapyfor patients with AML and the FLT3-ITD mutation. Either agent may be an effective therapeutic choice for EMR after allo-HSCT[29,30]. Recently, hypomethylating agents, such as 5-azacitidine and decitabine, were successfully used in the salvage treatment of patients with AML who relapsed or experienced EMR after allo-HSCT[31-33]. It was hypothesized that hypomethylating agents are directly cytotoxic and also might increase the GVL effect by inducing leukemic cell differentiation and expres-sion of HLA-DR to enhance the effects of DLI given concomitantly[33,34].
Owing to a lack of effective treatments, the prognosis for EMR after allo-HSCT is poor. Generally, patients with isolated EMR have a better prognosis than those with BMR. In the UK Medical Research Council’s UKALL12/ECOG 2993 study , analysis of outcomes in 609 adults with recurring ALL revealed a better 5-year OS in patients with EMR compared with patients with BMR and those with BMR and CNS relapse (14% versus 6% versus 0;P¼.004). The Children’s Oncology Group CCG-1961 studyreported obvious differences in the 3-year OS of childhood ALL in patients with BMR (with or without EMR), those with isolated CNS EMR, and those with other isolated EMR (29.7% versus 52.2% versus 68.2%;
P < .0001, log-rank test). In the multicenter trial Acute Lymphoblastic Leukemia Relapse Berlin-Frankfurt-Münster 90 , multivariate Cox regression analysis identiﬁed the site of relapse as an independent predictor for event-free survival (RR for subsequent event at an isolated EM site versus combined and isolated BM sites of relapse of 1, 1.7, and 2.3, respectively;P<.001). In the present study, in the post-HSCT setting, we also found a trend toward better prognosis in patients with isolated EMR compared with patients with BMR or systemic relapse (18.2% versus 12.0% versus 8.0%;
P¼.206). In addition, our patients with ALL had a signiﬁ -cantly higher estimated 3-year OS compared with those with AML, possibly related to the higher proportion of isolated EMR in the patients with ALL. Thus, treatment should focus on preventing systemic relapse.
In conclusion, our results reported here demonstrate that EMR after allo-HSCT poses a signiﬁcant challenge for trans-plantation physicians. Our patients with ALL had a higher cumulative incidence of EMR compared with those with AML. CNS relapse is the most common subtype of EMR. Pa-tients with high-risk cytogenetics, advanced disease status, history of EM leukemia before allo-HSCT, hyperleukocytosis at diagnosis, receipt of PBSCs blood stem cell, and male are at increased risk for EMR. These patients should be closely evaluated for early evidence of EMR. 18F-ﬂuorodeoxyglucose
positron emission tomography/computed tomography may be a useful tool for this purpose. Prophylactic intrathecal chemotherapyand low-dose azacitidine or decitabine administered after allo-HSCT[33,40]may reduce the rate of recurrence. The prognosis for EMR after allo-HSCT is poor, and efﬁcacious treatment strategies are lacking. In addition to an early diagnosis with new modalities, clinical studies using new agents that may offer systemic activity while preserving the GVL effect are warranted in an effort to improve clinical outcomes. Considering the limitations of this study and previous related studies, including single institution experience, small sample size of patients with EMR, and patient heterogeneity, future studies with larger numbers of patients are warranted to further deﬁne the incidence, risk factors, and appropriate therapeutic strategies for EMR after allo-HSCT.
Financial disclosure:This work was supported by National
Natural Science Foundation of China (Grant 81270645), the Natural Science Foundation of Jiangsu Province (Grant BK2012627), the Natural Science Foundation of the Higher Education Institutions of Jiangsu Province (Grant 11KJB320015), the Priority Academic Program Development of Jiangsu Higher Education Institutions, Jiangsu Provincial Health Ofﬁce (Grant H201125), the Jiangsu Province Key Medical Center (Grant ZX201102), Jiangsu Provincial Special Program of Medical Science (BL2012005), and the National Public Health Research Foundation (Grant 201202017).
Conﬂict of interest statement:There are no conﬂicts of
in-terest to report.
1. Bekassy AN, Hermans J, Gorin NC, Gratwohl A. Acute and Chronic Leukemia Working Parties of the European Group for Blood and Marrow Transplantation. Granulocytic sarcoma after allogeneic bone marrow transplantation: a retrospective European multicenter survey.
Bone Marrow Transplant. 1996;17:801-808.
2. Au WY, Kwong YL, Lie AK, et al. Extra-medullary relapse of leukemia following allogeneic bone marrow transplantation. Hematol Oncol. 1999;17:45-52.
3. Berthou C, Léglise MC, Herry A, et al. Extramedullary relapse after favorable molecular response to donor leukocyte infusions for recur-ring acute leukemia.Leukemia. 1998;12:1676-1681.
4. Lee KH, Lee JH, Choi SJ, et al. Bone marrow vs extramedullary relapse of acute leukemia after allogeneic hematopoietic cell transplantation: risk factors and clinical course. Bone Marrow Transplant. 2003;32: 835-842.
5. Shimoni A, Rand A, Shem-Tov N, et al. Isolated extra-medullary relapse of acute leukemia after allogeneic stem-cell transplantation (SCT): different kinetics and better prognosis than systemic relapse.Biol Blood
Marrow Transplant. 2009;15:59.
6. Harris AC, Mageneau J, Braun T, et al. Extramedullary relapse in acute leukemia following allogeneic hematopoietic stem cell transplantation: incidence, risk factors and outcomes. Biol Blood Marrow Transplant. 2010;16:S177-S178.
7. Przepiorka D, Weisdorf D, Martin P, et al. 1994 Consensus Conference on Acute GVHD Grading.Bone Marrow Transplant. 1995;15:825-828. 8. Sullivan KM. Acute and chronic graft-versus-host disease in man.Int J
Cell Cloning. 1986;4:42-93.
9. Gray RJ. A class of K-sample tests for comparing the cumulative inci-dence of a competing risk.Ann Stat. 1988;16:1141-1154.
10. Solh M, DeFor TE, Weisdorf DJ, Kaufman DS. Extramedullary relapse of acute myelogenous leukemia after allogeneic hematopoietic stem cell transplantation: better prognosis than systemic relapse.Biol Blood
Marrow Transplant. 2012;18:106-112.
11. Lee JH, Choi SJ, Lee JH, et al. Anti-leukemic effect of graft-versus-host disease on bone marrow and extramedullary relapses in acute
12. Shi J-M, Meng X-J, Luo Y, et al. Clinical characteristics and outcome of isolated extramedullary relapse in acute leukemia after allogeneic stem cell transplantation: a single-center analysis. Leuk Res. 2013;37: 372-377.
13. Clark WB, Strickland SA, Barrett AJ, Savani BN. Extramedullary relapses after allogeneic stem cell transplantation for acute myeloid leukemia and myelodysplastic syndrome.Haematologica. 2010;95:860-863. 14. Simpson DR, Nevill TJ, Shepherd JD, et al. High incidence of
extra-medullary relapse of AML after busulfan/cyclophosphamide condi-tioning and allogeneic stem cell transplantation. Bone Marrow
15. Lazarus HM, Richards SM, Chopra M, et al. Central nervous system involvement in adult acute lymphoblastic leukemia at diagnosis: re-sults from the international ALL trial MRC UKALL XII/ECOG E2993.
16. Hassan M, Ljungman P, Bolme P, et al. Busulfan bioavailability.Blood. 1994;84:2144-2150.
17. Ringdén O, Ruutu T, Remberger M, et al. A randomized trial comparing busulfan with total body irradiation as conditioning in allogeneic marrow transplant recipients with leukemia: a report from the Nordic Bone Marrow Transplantation Group.Blood. 1994;83:2723-2730. 18. Oshima K, Kanda Y, Yamashita T, et al. Central nervous system relapse
of leukemia after allogeneic hematopoietic stem cell transplantation.
Biol Blood Marrow Transplant. 2008;14:1100-1107.
19. Cardoso AA, Schultze JL, Boussiotis VA, et al. Pre-B acute lymphoblastic leukemia cells may induce T cell anergy to alloantigen.Blood. 1996;88: 41-48.
20. Galandrini R, Albi N, Zarcone D, et al. Adhesion moleculeemediated signals regulate major histocompatibility complexeunrestricted and CD3/T cell receptoretriggered cytotoxicity.Eur J Immunol. 1992;22: 2047-2053.
21. Han P, Story C, McDonald T, et al. Immune escape mechanisms of childhood ALL and a potential countering role for DC-like leukemia cells.Cytotherapy. 2002;4:165-175.
22. Ruggeri L, Capanni M, Urbani E, et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants.
23. Yoshihara S, Ikegame K, Kaida K, et al. Incidence of extramedullary relapse after haploidentical SCT for advanced AML/myelodysplastic syndrome.Bone Marrow Transplant. 2011;47:669-676.
24. Choi SJ, Lee JH, Kim S, et al. Treatment of relapsed acute myeloid leukemia after allogeneic bone marrow transplantation with chemotherapy followed by G-CSFeprimed donor leukocyte infusion: a high incidence of isolated extramedullary relapse.Leukemia. 2004;18: 1789-1797.
25. Kolb HJ. Graft-versus-leukemia effects of transplantation and donor lymphocytes.Blood. 2008;112:4371-4383.
26. Takeshita A. Efﬁcacy and resistance of gemtuzumab ozogamicin for acute myeloid leukemia.Int J Hematol. 2013;97:703-716.
27. Metzelder S, Wang Y, Wollmer E, et al. Compassionate use of sorafenib in FLT3-ITDepositive acute myeloid leukemia: sustained regression
before and after allogeneic stem cell transplantation.Blood. 2009;113: 6567-6571.
28.Ravandi F, Cortes JE, Jones D, et al. Phase I/II Study of combination therapy with sorafenib, idarubicin, and cytarabine in younger patients with acute myeloid leukemia.J Clin Oncol. 2010;28:1856-1862. 29.Ando T, Mitani N, Matsunaga K, et al. Gemtuzumab ozogamicin
therapy for isolated extramedullary AML relapse after allogeneic hematopoietic stem-cell transplantation. Tohoku J Exp Med. 2010; 220:121-126.
30.Safaian NN, Czibere A, Bruns I, et al. Sorafenib (NexavarÒ) induces molecular remission and regression of extramedullary disease in a patient with FLT3-ITDþacute myeloid leukemia.Leuk Res. 2009;33: 348-350.
31.Antar A, Otrock ZK, Kharfan-Dabaja M, et al. Azacitidine in the treat-ment of extramedullary relapse of AML after allogeneic hematopoietic cell transplantation.Bone Marrow Transplant. 2013;48:994-995. 32.Graef T, Kuendgen A, Fenk R, et al. Successful treatment of relapsed
AML after allogeneic stem cell transplantation with azacitidine.Leuk Res. 2007;31:257-259.
33.Singh SN, Cao Q, Gojo I, et al. Durable complete remission after single agent decitabine in AML relapsing in extramedullary sites after
allo-SCT.Bone Marrow Transplant. 2011;47:1008-1009.
34.Jabbour E, Giralt S, Kantarjian H, et al. Low-dose azacitidine after allogeneic stem cell transplantation for acute leukemia.Cancer. 2009; 115:1899-1905.
35.Fielding AK, Richards SM, Chopra R, et al. Outcome of 609 adults after relapse of acute lymphoblastic leukemia (ALL); an MRC UKALL12/ECOG 2993 study.Blood. 2007;109:944-950.
36.Freyer DR, Devidas M, La M, et al. Postrelapse survival in childhood acute lymphoblastic leukemia is independent of initial treatment in-tensity: a report from the Children’s Oncology Group.Blood. 2011;117: 3010-3015.
37.Tallen G, Ratei R, Mann G, et al. Long-term outcome in children with relapsed acute lymphoblastic leukemia after time-point and site-of-relapse stratiﬁcation and intensiﬁed short-course multidrug chemotherapy: results of Trial ALL-REZ BFM 90.J Clin Oncol. 2010;28: 2339-2347.
38.Stolzel F, Rollig C, Radke J, et al. 18F-FDG-PET/CT for detection of extramedullary acute myeloid leukemia. Haematologica. 2011;96: 1552-1556.
39.Thompson CB, Sanders JE, Flournoy N, et al. The risks of central nervous system relapse and leukoencephalopathy in patients receiving marrow transplants for acute leukemia.Blood. 1986;67:195-199.
40.de Lima M, Giralt S, Thall PF, et al. Maintenance therapy with low-dose azacitidine after allogeneic hematopoietic stem cell transplantation for recurrent acute myelogenous leukemia or myelodysplastic syndrome.