Patients and methods

Patients

We retrospectively studied 35 consecutive patients who underwent UD-SCT for childhood ALL in CR3. The transplants were performed between July 1990 and March 2002 in the Bristol Royal Hospital for Children, Bristol, UK. Patients characteristics and transplantation details are shown in Table 1. All patients were less than 18 years of age at diagnosis. One patient was transplanted aged over 18 years (initially diagnosed at age 17.4 years).

Table 1 - Patient characteristics and clinical features.

Full table

HLA matching

The methodology for the HLA typing of patients and UDs changed during the study period. For patients transplanted prior to 1993, serological testing was performed using two-stage National Institute of Health (NIH) complement-dependent microlymphocytotoxicity for HLA-A and HLA-B loci and restriction fragment length polymorphism (RFLP) for HLA-DR and -DQ (n=6). After 1993, HLA typing was by polymerase chain reaction using sequence-specific primers (PCR-SSP) for HLA-A, -B, -C, -DRB1 and -DQB1.

In all, 25 patients were HLA matched at HLA-A, -B, -DRB1 and -DQB1. All were typed for HLA-C, except seven patients. Two of these were subsequently found to be mismatched at the HLA-C locus. HLA mismatches are summarized in Table 2.

Table 2 - Type of HLA mismatches.

Full table

Preparative regimen

Of the 35 patients, 32 received cyclophosphamide (Cy) and total body irradiation (TBI). Cyclophosphamide was given intravenously (i.v.) once daily for 2 days on days -6 and -5 at a dose of 60 mg/kg/day. TBI was given as eight fractions in 4 days on days -3 to 0, with a total dose 14.4 Gy. Additional radiotherapy was given to the site of extra medullary relapse, 6 Gy in four fractions to the whole brain (n=7), or 6 Gy in three fractions to the testes (n=1) in the week preceding TBI. Three patients were ineligible to receive TBI and therefore received non-TBI-containing preparative regimens. Of these three patients, two received busulfan and cyclophosphamide (BuCy) conditioning. Bu was given at 4 mg/kg/day twice daily for 4 days in divided doses, followed by Cy at 50 mg/kg/day i.v. for 4 days. The third patient received BuCy/VP-16, Cy 60 mg/kg/day once daily i.v. for 2 days, Bu 4 mg/kg/day in four divided dose twice daily for 4 days, and one day of VP-16 40 mg/kg i.v. Two of these patients had received TBI in an autologous transplant in CR2, and one had received cranial radiotherapy twice (at diagnosis and at first relapse).

Preparation of the bone marrow graft

One patient received PBSC collected using a Cobe Spectra apheresis device after G-CSF mobilization. The PBSC collection included in the study was CD34 selected using the CliniMacs procedure (Miltenyi Biotech) as described previously.¹⁷ T-cell depletion (TCD) was performed in 30 of 35 patients. In 29 patients, this was performed using anti-CD-52 monoclonal antibodies (CAMPATH-1G, four patients, or CAMPATH-1M, 25 patients) as described previously.¹⁸ TCD using CD34 selection was used in one patient. Samples of marrow or PBCS were taken prior to and at the end of purging/CD34 selection to determine the total nucleated cell count (TNC), mononuclear cell count (MNC), CD34+ and CD3+ cell counts. The TNC was measured using a Sysmex SE9000 hematology analyzer and the MNC, CD34 and CD3 cell counts were determined using an EPICS MCL Flow Cytometer (Beckman Coulter) using the ISHAGE gating strategy.

Graft-versus-host disease (GVHD) prophylaxis

In addition to ex vivo TCD, in vivo CAMPATH-1M 5–20 mg/day was administered to all patients for 5 days (-9 to -5). Cyclosporin A (CsA) was given to all patients, except one who received methotrexate (MTX) alone. CsA was given alone in 26 patients, and was given with MTX in eight patients. CsA was administered starting from day -1 at a dose of 1.5 mg/kg twice daily, and adjusted according to drug level thereafter. CsA taper was begun between 3 and 6 months post-SCT in those patients who did not develop GVHD. MTX was given at a dose of 15 mg/m² on day +1 and a dose of 10 mg/m² on days +3, +6 and +10. Six patients received only three doses of MTX and one patient received two doses.

Definitions of engraftment

Patients were evaluable for engraftment if they survived more than 28 days post transplant. Neutrophil engraftment was defined as ANC 0.5 10⁹/l for 3 consecutive days. Primary graft failure was defined as failure to reach ANC 0.5 10⁹/l at any time. And secondary graft failure was defined as sustained loss of neutrophils <0.1 10⁹/l after initial engraftment had occurred.

Statistical analysis

Kaplan–Meier curves were used to estimate event-free survival (EFS). For the two 'competing' end points, relapse and transplant-related mortality (TRM), 'cumulative incidence rates' were calculated.^{19, 20} The latter effectively 'adjusted' for one another, that is, relapse was assessed 'assuming there is no TRM' and vice versa. The two sets of incidences sum to give the cumulative losses estimated by the overall Kaplan–Meier curve (see Supplementary Table 3).

A series of univariate Cox's 'proportional hazards' analyses were used to look for variables associated with overall outcome. Each risk factor variable was split into (usually) two subgroups; where no prior division was indicated, the cut-point used was the median value. One subgroup was chosen as the baseline (reference) category and the hazard rate ratio (RR) calculated for the other subgroup(s) relative to the reference subgroup; RRs> or <1, therefore, indicated increased or decreased risk, respectively, compared with the reference.

The overall study size was considered too small for multivariate analysis.

A series of Cox's models were fitted that looked at 'competing' relapses and TRMs simultaneously, using Lunn and McNeil's²¹ 'augmented data' approach. Their 'Model B' was used, which stratified by type of event, that is, made the not unrealistic assumption that the underlying hazard curve was fundamentally different for relapses and TRMs. The RRs for each type of end point were calculated within the same analysis.

Results

Infused cell dose

The median total nucleated cells (TNC) dose was 4.6 10⁸/kg (range 2.45–16.31). The median mononuclear cells (MNC) dose was 1.18 10⁸/kg (range 0.62–14.54). The CD34 dose was available on 33 patients with a median CD34 dose of 2.99 10⁶/kg (range: 0.17–12.08). The number of infused T cells was available on 32 patients. The median T-cell dose infused was 0.96 10⁶ cells/kg, (range 0.0019–65.5). In 25 of the 30 T-cell-depleted grafts, T-cell dose ranged from 0.1 10⁶ to 20 10⁶ cells/kg. Three patients received T-cell doses higher than 20 10⁶ cells/kg and two received T-cell dose lower than 0.1 10⁶ cells/kg.

Engraftment

Median time to engraftment was 17 days (range 12–29). In all, 34 patients were evaluable for engraftment.

Six patients had primary or secondary graft failure. Three of the six patients had primary graft failure: two mismatched unrelated donor (MMUD), and one matched unrelated donor (MUD). All three died, two within less than 100 days of transplant. The third patient received autologous rescue and died later at 1.1 years post-SCT of multiorgan failure. Of the six patients with graft failure, three had secondary graft failure. All underwent a subsequent UD-SCT. Two are alive and disease free (DF). One is DF 3.2 years after a second UD-SCT. He initially had autologous BM rescue at day +40 and then had a second UD-SCT at 1.2 years after the first SCT. The other had BM reinfusion at day +69 and is alive and DF 9.2 years post-SCT. The third patient had autologous rescue and then underwent second UD-SCT, but died of relapse 0.6 years post first SCT.

Transplant-related mortality

All transplant-related deaths were within 1.1 years of SCT. The estimated cumulative incidence of TRM was 23% (Figure 1). Variables associated with an increased risk of TRM (Table 3) were a shorter interval from initial diagnosis to SCT, <6.33 years (P=0.032), a shorter duration of first complete remission (CR1), <2.5 years (P=0.019), and a shorter duration of CR2 <2.5 years (P=0.009), Table 3. None of the other variables were significantly related to TRM. Causes of TRM are shown in Table 4.

Figure 1.

Cumulative incidence of relapse and TRM. The estimates of relapse and TRM at 3 years were 42 and 23%, respectively.

Full figure and legend (14K)

Table 3 - Risk factors for EFS, relapse and TRM.

Full table

Table 4 - Causes of death.

Full table

Graft-versus-host disease

Grade II–IV acute GVHD (aGVHD) occurred in 10 of the 32 evaluable patients. Grade III–IV aGVHD occurred in three of the 32 evaluable patients. Of the 30 evaluable patients, limited and extensive chronic GVHD (cGVHD) occurred in four and two patients, respectively. Of the three patients who developed grade III–IV aGVHD, one had T-cell replete graft. One of the two patients who developed severe cGVHD had T-cell replete graft. The second patient had a T-cell-depleted graft and had also developed severe aGVHD. Of the five patients who received T-cell replete grafts, two had developed severe GVHD.

Relapse

The estimated cumulative incidence of relapse was 42% at 3 years post-SCT (Figure 1). All but three of 15 patients who relapsed died of relapse. These three patients were alive with disease at last follow-up 3.83, 2.56 and 1.30 years post transplant. The median time from SCT to relapse was 0.7 years (range 0.31–1.48).

A shorter duration of CR1 (<2.5 years) was associated with increased risk of relapse (P=0.023) (Table 3). The significance of the site of first and of second relapse was difficult to assess because of small numbers. A first isolated extramedullary relapse was associated with increase risk of further relapse compared to isolated bone marrow (P=0.019).

Nine of the 10 patients who had a first extramedullary relapse eventually failed SCT in CR3. Six of the failures were due to relapse. Eight of the 10 patients had a short duration to first relapse (CR1 <2.5 years). Seven of the first isolated extramedullary relapse sites were in the central nervous system (CNS). Five patients had prior CNS radiotherapy at initial diagnosis. Based on the available numbers, further analysis of compounding factors such as duration of CR1, particular site of extramedullary relapse and prior radiotherapy in this subgroup was not possible.

No other variables analyzed were significantly related to relapse risk.

Event-free survival

At the time of the analysis, 12 of the 35 patients were alive and in remission. The Kaplan–Meier estimate of EFS at 3 years was 35% (SE 8%) (Figure 2). Among the variables analyzed (Table 3), short duration of CR1 (<2.5 years) and short duration of CR2 (<2.5 years) were associated with adverse survival (P=0.001 and 0.003, respectively). The site of first relapse was significantly related to EFS, with extramedullary relapse significantly worse than bone marrow (P=0.012). None of the other variables analyzed were significantly related to EFS.

Figure 2.

Event-free survival (EFS). At the time of analysis, 12 of 35 patients were alive and in remission. The Kaplan–Meier estimate at 3 years was 35% (s.e. 8%).

Full figure and legend (9K)

Discussion

We report 35 consecutive patients who underwent UD-SCT for childhood ALL in CR3 in a single institution. The EFS was 35% (SE 8%). The major causes of failure were relapse, occurring in 15 patients and TRM in eight patients. These results compare favorably to a recently published series of 10 patients transplanted from unrelated donors in CR3. The 3-year EFS, relapse rate and TRM were 20, 20, 60%, respectively.¹⁵ In another series of 33 patients transplanted from matched siblings (MS), the EFS at 6 years was 48% (SE 9%). TRM was 30% (10 of 33), and six of 33 patients relapsed.¹³ These apparent differences among studies could be explained by the small numbers of patients, the different methods of UD-TCD , intensity of preceding therapy and donor type.

On the other hand, an outcome of UD-SCT comparable to MS stem cell transplantation (MS-SCT) has been reported.⁵ Weinsdorf published 36 patients transplanted from matched related donors in CR3 or beyond in a single institution, with an EFS of 25% (SE 15%).¹⁴ Their analysis of prognostic factors was limited by the inclusion of several phases of disease. The apparent inferior outcome may be related to the inclusion of patients in subsequent remission beyond CR3.

More intensive prior therapy followed by relapse confers higher risk of TRM and post-SCT relapse.¹⁵ Patients treated in more recent eras who develop relapse after more intensive initial therapy may have disease more resistant to treatment with both chemotherapy and SCT. Even though in our study the period in which SCT was performed (before and after 1996) did not influence outcome (data not shown), other studies have shown a higher relapse rate after transplant in patients transplanted in more recent years.¹⁵

Age at SCT did not influence the outcome in this study. However, there was a trend for younger patients (under 10 years of age) at SCT to have higher TRM. This trend did not reach statistical significance and has to be interpreted with caution. In fact, it is known that younger age at SCT is a favorable feature, and is accompanied by decreased risk of TRM.¹⁵ Since the analysis in our study was limited to univariate analysis, and since the transplants were performed in CR3, the trend toward higher TRM in younger age in our study may partly reflect other factors, like short duration of previous remissions and shorter interval from diagnosis to SCT.

There was a significantly increased risk of TRM in patients who had short CR1, those who had short CR2 and patients who had short duration from initial diagnosis to SCT. This may reflect a rapid accumulation of toxic effects of therapy preceding SCT.

Our results indicate that a short CR1 is significantly associated with increased risk of relapse after SCT in CR3. This is consistent with the concept that shorter remission is a marker of more aggressive disease. Patients transplanted in CR2 who have a short previous CR1 have a higher relapse rate and higher TRM.¹²

The impact of T-cell depletion on engraftment, GVHD, relapse and therefore UD-SCT outcome was previously studied by members of our group.^{22, 23} The incidence of GVHD was low in this study. This may be due to the GVHD prophylaxis strategy that combined both ex vivo and in vivo T-cell depletion using CAMPATH in the majority of the patients. However, TCD has been shown to be associated with an increased relapse rate due to loss of graft-versus-leukemia effect. This may explain the high relapse rate of 42%.

There was no significant difference between the outcomes of MMUD and MUD SCT in our study. Previously, our group had reported a comparable outcome of MMUD and MUD-SCT in CR2.¹⁸ This has also been reported by other investigators.^{15, 24}

In this study, patients with a first isolated extramedullary relapse had a significantly worse outcome when compared with those with isolated BM relapse. These data have to be interpreted with caution, in view of the small numbers in each relapse site. However, poor outcome due to high relapse rate after SCT following extramedullary relapse has been reported previously.^{2, 25} The dismal outcome of patients who had isolated extramedullary first relapse highlights the importance of identifying subgroups of patients with isolated first relapse in an extramedullary site, who have a very poor prognosis. Early first isolated extramedullary relapse (CR1 <2.5 years) occurred in eight of 10 patients in our study and should be considered an important risk factor. Higher death rate has been reported among patients who developed isolated CNS relapse after prior CNS directed radiotherapy at diagnosis than in patients who did not receive this therapy.²⁶ On the other hand, many transplant centers defer SCT in patients who have a first relapse isolated to an extramedullary site since the salvage rates with conventional chemotherapy and radiotherapy in many studies are promising.^{26, 27, 28, 29} The selection of patients with first relapse occurring at an extramedullary site, who need more intensive, novel therapy or SCT, needs further study. Post-SCT-extended intrathecal chemotherapy has been suggested as an effective strategy to improve outcome of CNS relapses.²⁵

In conclusion, we have shown that allogeneic unrelated SCT is an effective treatment of childhood ALL in CR3. The major limiting factors for success are TRM and a high relapse rate. The short duration of CR1 and of CR2 is a particularly adverse prognostic factor. SCT should be performed in CR2 when CR1 is short. Mismatched-UD should also be considered, since our results show a similar outcome of MUD-SCT and MMUD-SCT. Our findings also suggest that more studies of specific risk features of patients with first extramedullary relapse and identification of the best therapeutic options for certain subgroups with first extramedullary relapse are necessary.

References

References

Acknowledgements

We thank the consultants of the bone marrow transplantation unit in the Royal Bristol Hospital for children who planned and provided the management of the patients. Special thanks to Dr Peter Shaw of the children's hospital of Westmead, Sydney Australia for editing the paper. We are grateful to Dr Michael Pulsipher and Dr Finn Petersen of the University of Utah for reviewing the paper, and to Ms Darlene Sylverster for her assistance with formatting the manuscript.

Supplementary Information accompanies the paper on Bone Marrow Transplantation website (http://www.nature.com/bmt)