Relapse of Leukaemia

Extramedullary relapse after allogeneic bone marrow transplantation for haematological malignancy


We describe the risk factors for and the natural history and response to treatment of extramedullary (EM) relapse in 183 patients who underwent allogeneic bone marrow transplantation (alloBMT) for a variety of haematological malignancies at our institution over a 7½ year period. Fifty-one patients relapsed; 15 had EM relapse either alone or in association with marrow involvement. A retrospective analysis found that the presence of chronic GVHD and a longer interval between transplant and relapse were independently associated with an increased risk of EM compared to marrow-only relapse. EM relapse was also associated with a longer post-relapse survival. Patients with EM relapse appeared to respond to cytotoxic therapy but not to DLI. EM relapse after alloBMT may be more common than previously thought and have a better prognosis than marrow-only relapse. While patients developing chronic GVHD after alloBMT have a lower overall relapse risk than those who do not, they may be more prone to delayed relapse at EM sites. Bone Marrow Transplantation (2000) 26, 1011–1015.


Allogeneic bone marrow transplantation (alloBMT) is potentially curative therapy for haematological malignancies including acute leukaemia, myelodysplastic syndrome (MDS), and chronic myeloid leukaemia (CML).123 However, 30–50% of patients transplanted for acute myeloid leukaemia (AML) beyond first remission eventually relapse – usually within 12 months.4 The post-transplant relapse rate for advanced acute lymphoblastic leukaemia (ALL) is higher, and accelerated phase CML carries a relapse risk of up to 60%.23

Disease relapse is usually in the bone marrow, but some patients develop extramedullary (EM) relapse either alone, or in association with marrow relapse. EM relapse after alloBMT has been thought to be rare and the prognosis relatively poor. A survey conducted by the European Group for Blood and Marrow Transplantation (EBMT) found a prevalence of EM relapse post allograft for AML, MDS and CML of only 0.45%.5 Two reports, however, suggest a higher prevalence in patients with relapsed AML post alloBMT; Mortimer et al4 noted an EM component in 20% of 95 patients and Simpson et al6 described EM relapse in 10 of 22 patients.

Little is known about predisposing factors, natural history and response to treatment of EM as compared to marrow relapse. We addressed these issues in a retrospective analysis of a cohort of patients with relapsed haematological malignancy following alloBMT. In this analysis, hepatosplenomegaly and lymphadenopathy were not considered manifestations of extramedullary disease. Leptomeningeal disease was not considered extramedullary unless positive cerebrospinal fluid cytology was accompanied by a structural lesion on radiological imaging.

Patients and methods

Between November 1990 and March 1998, 183 patients at the Royal Melbourne Hospital underwent alloBMT for acute leukaemia, MDS, myeloproliferative disorders or multiple myeloma. Relapse subsequently occurred in 51 patients. Of these patients, those with unrelated or less than ideal donors received cyclophosphamide/total body irradiation (TBI) (23 patients) as the preparative regimen. Other regimens used were busulphan/cyclophosphamide (21 patients), busulphan/cyclophosphamide/etoposide (six patients) and melphalan/TBI (one patient). All patients received cyclosporine/methotrexate as graft-versus-host disease (GVHD) prophylaxis. No patients received T cell-depleted grafts. Analysis was based on data collected to 30 November 1999.

A retrospective review of medical records of patients who relapsed noted the following: age at transplant, type and sites of disease at diagnosis, disease status at transplant, cytogenetic abnormalities, preparative regimen, donor source, presence of acute or chronic GVHD, interval between transplant and relapse, relapse site, length of survival post relapse and response to treatment. Sites of disease at diagnosis were noted for all 183 patients. Patients who never attained complete remission (CR) post alloBMT were excluded. Median follow up was 12.7 months (range 2–108) post transplant and total follow up was 1262 patient-months. The male:female ratio was 27:24 and the median age at transplant was 30.5 years (16–55). Other pretransplant characteristics for relapsed patients are presented in Table 1.

Table 1  Pretreatment characteristics of 51 patients with relapse following alloBMT

The following factors were analysed with regard to the relapse site: disease type, cytogenetics, disease status at transplant (first CR or chronic phase vs relapsed or progressive disease), conditioning regimen, donor source, grade of acute GVHD, stage of chronic GVHD, and time to relapse. Chronic GVHD was staged as absent, limited (including isolated oral GVHD) or extensive. Patients who died before 120 days post transplant were not analysed with respect to chronic GVHD.

Categorical variables were compared with Fisher's exact test. Times to relapse and survival post relapse were compared using the Kaplan–Meier estimator and the log-rank tests. Multivariate analyses and estimates of hazard ratios used logistic and Cox regression. Continuous variables used as covariates in logistic or Cox regressions were first log-transformed and tested for normality with the Wilk–Shapiro test. Significance was defined as two-sided α = 0.05 and all confidence intervals (CI) were 95%. Calculations were performed with Stata 6 (StataCorp, College Station, TX, USA).



Fifteen of 51 patients who relapsed post-allograft had an EM component to their relapse (see Table 1). EM relapse occurred in all diseases (see Table 2) and was biopsy proven in all but two patients (Nos 13 and 14 in Table 3). Both of these patients had multiple cortical bone lesions and pathological fractures demonstrated on bone scans and X-rays. Four of six patients with AML and EM relapse had myelomonocytic leukaemia, three of whom had M4eo with the characteristic inversion of chromosome 16. This was the only cytogenetic subgroup with more than one case of EM relapse. Of all 183 patients allografted, five had EM disease prior to alloBMT; all relapsed (Table 1). Four patients developed EM relapse and one relapsed in the marrow only.

Table 2  Site of relapse following alloBMT according to disease type
Table 3  Characteristics and clinical course of patients with EM relapse post alloBMT

Three patients with EM relapse had concomitant marrow relapse. Of the nine patients with isolated EM relapse initially, three subsequently developed marrow relapse 2, 2 and 6 months later; the remaining six had not developed marrow involvement at follow-up of 1, 2, 5, 6, 13 and 39 months. Three additional patients had marrow relapse followed 7, 7 and 11 months later by EM relapse.

The characteristics and clinical courses of the 15 patients with EM relapse are detailed in Table 4. Sites of EM relapse included bone (lytic cortical lesions in patients with ALL and AML), testis, soft tissue, pancreas, breast, meninges, paravertebral and nasopharynx. Four patients had two or more EM sites of disease. Median interval to first post-transplant relapse was 19 months (range 3–51) and 3 months (range 1–78) in the EM and marrow-only relapse groups, respectively.

Table 4  Relationship between the stage of chronic GVHD and the site of subsequent relapse

Factors associated with the development of EM relapse

The relationship between the stage of chronic GVHD and the site of relapse post alloBMT is detailed in Table 4. Statistical analysis was performed according to the stage of chronic GVHD, ie absent vs limited vs extensive. Patients with EM relapse were more likely to have had a more advanced stage of chronic GVHD than those who relapsed in marrow only (P = 0.043, odds ratio 2.08, CI 1.02–4.26). A longer interval between transplant and relapse was also associated with an EM component (P = 0.002, odds ratio multiplies by 2.32, CI 1.36–3.95 per doubling of time to relapse, Figure 1). In the multivariate analysis, time to relapse (P = 0.010, odds ratio multiplies by 2.23, CI 1.21–4.10 for each doubling of time to relapse) and stage of chronic GVHD (P = 0.042, odds ratio 2.36, CI 1.03–5.42) were independent predictors of EM relapse.

Figure 1

The relationship of the site to the time of relapse post-transplant.

There was no association between donor source and site of subsequent relapse (Table 5). Only on univariate analysis was there an association between pretransplant EM disease and EM relapse post transplant (P = 0.022, odds ratio 12.73, CI 1.28–126.14). There was no association between pretransplant EM disease and either time to relapse or survival post relapse. There were no statistically significant associations between other factors and the site of relapse.

Table 5  Donor marrow source and site of subsequent relapse following alloBMT


Treatment of patients with EM relapse post allograft varied widely and depended on duration of remission, relapse site, overall medical condition and patient preference. Management strategies included intensive chemotherapy, palliative chemotherapy, interferon, infusion of donor lymphocytes (DLI), radiotherapy and supportive care (Table 3). One patient (patient 11) received a second allograft for marrow relapse after achieving local CR with radiotherapy, but died from fungal infection prior to engraftment.

Five patients received DLI, four in combination with chemotherapy; three achieved CR lasting 5, 13 and 25 months. One of these three subsequently received intermediate dose i.v. melphalan for multiply recurrent plasmacytomata and remains in CR 39 months post relapse.7 One patient with M4eo AML and EM relapse achieved CR with intensive chemotherapy only and remains disease-free 84 months post relapse.8

Survival post relapse

The median post-relapse survival of patients who had EM relapse was 11 months (range 1–84) compared with 2 months (range 1–66) in patients with marrow-only relapse.

For the total group of 51 relapsed patients, longer survival post relapse was found on univariate analysis to be strongly associated with longer time to relapse (P < 0.0005); the hazard ratio decreasing by 0.68 (CI 0.57–0.82) per doubling of time to relapse. EM relapse was also associated with longer post-relapse survival (P = 0.004, hazard ratio 0.43, CI 0.24–0.77; Figure 2). However, only time to relapse was an independent predictor of longer survival post relapse (P = 0.009), the hazard ratio decreasing by 0.76 (CI 0.63–0.93) per doubling of time to relapse.

Figure 2

Probability of survival post relapse according to the site of relapse post transplant, P = 0.004 on univariate analysis only.


The frequency of EM relapse following alloBMT for haematological malignancy in our single institution cohort was 29%. This is relatively high compared to the largest reported series and also contrasts with the very low incidence of EM relapse of AML following standard-dose chemotherapy.4569

Bone was the most common site of EM relapse, although many different sites were affected. Previously reported cases included almost all conceivable locations.510111213 In the present study, two patients with EM disease prior to transplant relapsed in similar areas to their original disease. There were three ‘sanctuary site’ relapses – two in the testis and one meningeal; two of these were patients with ALL. Sixty percent of patients with EM relapse developed marrow involvement at some stage post alloBMT, a frequency similar to that observed in the EBMT survey.

Our results suggest that EM disease is relatively resistant to graft-versus-tumour effects. There are three lines of evidence to support this. Firstly, an EM component was more commonly seen in patients who relapsed despite chronic GVHD than in patients who relapsed without chronic GVHD. This occurred in the context of the lower overall relapse risk in patients with chronic GVHD14 and suggests that in some patients, the graft-versus-tumour effect may preferentially maintain marrow remission without preventing EM relapse. A site-dependent graft-versus-tumour effect has been proposed for EM relapse following DLI for relapsed leukaemia.1516 Although the mechanism has yet to be defined, this could explain the relatively high incidence of EM relapse following alloBMT compared to standard-dose chemotherapy.

Secondly, patients who developed EM relapse did so after a longer interval post transplant than those with marrow-only relapse, consistent with the hypothesis that graft-versus-tumour effects may prevent early marrow relapse but not late relapse in other tissues. Finally, our observation that none of the five patients with EM relapse responded durably to DLI is consistent with this hypothesis. Our numbers, however, are too small to draw definite conclusions regarding the efficacy of adoptive immunotherapy in this cohort. The literature is also not conclusive on this issue. Three patients have been reported with EM relapse while in sustained marrow remission following DLI for myeloid leukaemia relapsing post alloBMT, two of whom had significant GVHD.1517 Zomas et al18 reported a patient with plasmacytomata which persisted despite DLI inducing marrow remission. Against the concept of site-dependent graft-versus-tumour effects, however, are reports of EM relapse of acute leukaemia responding to DLI.1920

Treatment following EM relapse tended to be more interventional compared to patients with marrow relapse. This was probably due to longer post-transplant remission and thus better performance status and overall medical condition. Initial response to treatment was generally good, with seven of 13 patients obtaining a CR with local radiotherapy or systemic treatment. This suggests that patients with EM relapse may have altered tumour biology resulting in both more indolent behaviour and greater chemo/radiosensitivity. Consistent with this is the observation that patients with EM relapse survived longer following relapse than patients with marrow-only disease, although this may also reflect the more intensive treatment strategies in the former group. While superior post-relapse survival in the EM relapse group could also be due to lower risk pre-transplant disease, there was no statistical correlation between pre-transplant disease status and post-transplant EM relapse.

EM relapse following alloBMT is a different disease entity compared with marrow relapse. It occurs later, is associated with superior post-relapse survival and often responds to chemotherapy and/or radiotherapy. This is important to recognise as EM relapse may be more common than previously reported.


  1. 1

    Clift RA, Buckner CD, Thomas ED et al. The treatment of acute non-lymphoblastic leukemia by allogeneic marrow transplantation Bone Marrow Transplant 1987 2: 243–258

  2. 2

    Doney K, Fisher LD, Appelbaum FR et al. Treatment of adult acute lymphoblastic leukemia with allogeneic bone marrow transplantation. Multivariate analysis of factors affecting acute graft-versus-host disease, relapse and relapse-free survival Bone Marrow Transplant 1991 7: 453–459

  3. 3

    Goldman JM, Apperley JF, Jones L et al. Bone marrow transplantation for patients with chronic myeloid leukemia New Engl J Med 1986 314: 202–207

  4. 4

    Mortimer J, Blinder MA, Schulman S et al. Relapse of acute leukemia after marrow transplantation: natural history and results of subsequent therapy J Clin Oncol 1989 7: 50–57

  5. 5

    Bekassy AN, Hermans J, Gorin NC, Gratwohl A . Granulocytic sarcoma after allogeneic bone marrow transplantation: a retrospective European multicentre survey. Acute and chronic leukemia working parties of the European Group for Blood and Marrow Transplantation Bone Marrow Transplant 1996 17: 801–808

  6. 6

    Simpson DR, Nevill TJ, Shepherd JD et al. High incidence of extramedullary relapse of AML after busulfan/cyclophosphamide conditioning and allogeneic stem cell transplantation Bone Marrow Transplant 1998 22: 259–264

  7. 7

    Grigg AP . Multiply recurrent extramedullary plasmacytomas without marrow relapse in the context of extensive chronic GVHD in a patient with myeloma Leuk Lymphoma 1999 34: 635–636

  8. 8

    Thomas DM, Seymour JF, Griffiths JD et al. Prolonged disease-free survival with specific re-induction therapy in acute myeloid leukaemia relapsing following bone marrow transplantation Aust N Z J Med 1997 27: 712–713

  9. 9

    Byrd JC, Edenfield JW, Shields DJ, Dawson NA . Extramedullary myeloid cell tumors in acute nonlymphocytic leukemia: a clinical review J Clin Oncol 1995 13: 1800–1816

  10. 10

    Ueda S, Kanamori H, Sasaki S et al. Isolated extramedullary relapse in knee joint after allogeneic bone marrow transplantation for Ph ALL Bone Marrow Transplant 1998 21: 319–321

  11. 11

    Riseberg DA, Mulvey KP, Dahut WL et al. Late extramedullary relapse following bone marrow transplant for AML presenting as acute renal failure and leukemic ascites Bone Marrow Transplant 1994 14: 1009–1010

  12. 12

    Au WY, Chan CA, Lie AK et al. Isolated extramedullary relapse after allogeneic bone marrow transplantation for chronic myeloid leukemia Bone Marrow Transplant 1998 22: 99–102

  13. 13

    Millot F, Facon T, Kerckaert JP et al. Unusual recurrence of chronic myelogenous leukemia following bone marrow transplantation Bone Marrow Transplant 1991 7: 393–395

  14. 14

    Sullivan KM, Weiden PL, Storb R et al. Influence of acute and chronic graft-versus-host disease on relapse and survival after bone marrow transplantation from HLA-identical siblings as treatment of acute and chronic leukemia (published erratum Blood 1989; 74: 1180) Blood 1989 73: 1720–1728

  15. 15

    Porter D, Schenkein D, Rauch A et al. Unusual pattern of relapse after donor leukocyte infusions for relapsed CML: lack of graft-versus-leukaemia effect for isolated chloromas Blood 1997 90: (Suppl. 1) 384b

  16. 16

    Au WY, Lie AK, Liang R, Kwong YL . Isolated extramedullary relapse of acute lymphoblastic leukaemia after allogeneic bone marrow transplantation Bone Marrow Transplant 1999 24: 1137–1140

  17. 17

    Berthou C, Leglise MC, Herry A et al. Extramedullary relapse after favourable molecular response to donor leukocyte infusions for recurring acute leukemia Leukemia 1998 12: 1676–1681

  18. 18

    Zomas A, Stefanoudaki K, Fisfis M et al. Graft-versus-myeloma after donor leukocyte infusion: maintenance of marrow remission but extramedullary relapse with plasmacytomas Bone Marrow Transplant 1998 21: 1163–1165

  19. 19

    Lawson SE, Darbyshire PJ . Use of donor lymphocytes in extramedullary relapse of childhood acute lymphoblastic leukaemia following bone marrow transplantation Bone Marrow Transplant 1998 22: 829–830

  20. 20

    Kottaridis P, Ketley N, Peggs K et al. An unusual case of intrapulmonary granulocytic sarcoma presenting as interstitial pneumonitis following allogeneic bone marrow transplantation for acute myeloid leukaemia and responding to donor lymphocyte infusion Bone Marrow Transplant 1999 24: 807–809

Download references

Author information

Correspondence to A Grigg.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Chong, G., Byrnes, G., Szer, J. et al. Extramedullary relapse after allogeneic bone marrow transplantation for haematological malignancy. Bone Marrow Transplant 26, 1011–1015 (2000).

Download citation


  • allogeneic bone marrow transplantation
  • extramedullary relapse
  • graft-versus-host disease
  • donor lymphocyte infusion

Further reading