Long-Term Outcomes of Patients with Acute Myelogenous Leukemia Treated with Myeloablative Fractionated Total Body Irradiation TBI-Based Conditioning with a Tacrolimus- and Sirolimus-Based Graft-versus-Host Disease Prophylaxis Regimen: 6-Year Follow-Up from a Single Center

Abstract

Cyclophosphamide (Cy)/etoposide combined with fractionated total body irradiation (FTBI) or i.v. busulfan (Bu) has been the main conditioning regimens for allogeneic hematopoietic cell transplantation (alloHCT) for young patients with acute myelogenous leukemia (AML) eligible for a myeloablative conditioning (MAC) regimen. Recent data has suggested that i.v. Bu could be the preferred myeloablative regimen in patients with myeloid malignancies. However, Bu-based regimens are associated with higher rates of sinusoidal obstruction syndrome. Here we report long-term survival outcomes of patients with AML receiving FTBI combined with Cy or etoposide before undergoing alloHCT at City of Hope (COH). We obtained a retrospective review of a prospectively maintained institutional registry of clinical outcomes in 167 patients (median age, 41 years; range, 18 to 57 years) with AML in first or second complete remission who underwent alloHCT at COH between 2005 and 2015. Eligible patients received a MAC regimen with FTBI (1320 cGy) and Cy (120 mg/kg) for unrelated donor transplantation or etoposide (60 mg/kg) for related donor transplantation. Graft-versus-host disease (GVHD) prophylaxis was provided with tacrolimus and sirolimus. In this retrospective study, 6-year overall survival was 60% and nonrelapse mortality was 15%. The GRFS rate was 45% at 1 year and 39% at 2 years. We also describe late metabolic effects and report the cumulative incidence of secondary malignancies (9.5%). Overall, in this young adult patient population, our results compare favorably to chemotherapy-based (i.v. Bu) conditioning regimens without significant long-term toxicity arising from TBI-based regimens.

INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (HCT) is the preferred consolidation strategy for treating patients with intermediate-risk or high-risk acute myelogenous leukemia (AML) [1,2]. Myeloablative conditioning (MAC) is the most commonly used regimen, associated with better leukemia-free survival (LFS) in younger patients and no major comorbidities. However, MAC regimens are associated with higher rates of nonrelapse mortality (NRM), contributed by regimen-related toxicities and graft-versus-host disease (GVHD) [3]. Most of the current myeloablative regimens are built around either fractionated doses of total body irradiation (TBI) or i.v. busulfan (Bu), and the choice of regimen is based on the transplantation center’s or physician’s experience or on patient-specific factors [4].

Based on the results of an initial study reported by Thomas et al [5], TBI and cyclophosphamide (TBI/Cy) has become the most widely used preparative regimen in patients with acute leukemias. Non-TBI-containing conditioning regimens, namely Bu/Cy regimens, have also been developed and refined for the treatment of AML [6–9]. Because higher doses of Bu are associated with sinusoidal obstruction syndrome (SOS) and lower doses are associated with relapse and graft failure [10–12], further refinements in drug dose have been made based on the results of pharmacokinetics studies. These refinements have resulted in more precise drug delivery, allowing individualized dosing in patients with reduced regimen-related toxicities and lower risk of SOS [3,13,14].

Earlier comparisons of TBI/Cy and Bu/Cy regimens favored the TBI-containing regimens, demonstrating superior overall survival (OS) and lower rates of relapse and transplantation-related mortality (TRM) [15–19]. More recently, 2 large registry studies in the United States reported that i.v. Bu may be superior to [20–22] or at least noninferior to [22] TBI-containing regimens. However, along with their nonrandomized study design, both trials had several other limitations, including heterogeneity in patient populations and GVHD prophylaxis and, most importantly, the inconsistency of TBI dose, which ranged from a nonmyeloablative dose of 550 cGy to as high as 1440 cGy. Another retrospective review by the European Society for Blood and Marrow Transplantation (EBMT) comparing i.v. Bu with TBI in patients with AML showed no differences in 3-year OS, LFS, or NRM between the 2 cohorts but reported improved LFS in patients with adverse cytogenetics who received TBI/Cy [23]. Despite these valuable registry studies, long term follow-up data for this patient population does not exist, and whether i.v. Bu-based conditioning is clearly superior to TBI-based conditioning remains unclear.

Our group has pioneered TBI with high-dose etoposide (VP-16) for matched related donor HCT [24,25]. Using this regimen, we previously reported a 3-year disease-free survival of 61% and relapse rate of 12%, demonstrating the efficacy of this preparative regimen for patients with AML undergoing allogeneic HCT in first complete remission (CR). Subsequently, after publication of initial reports from Dana-Farber Cancer Institute comparing tacrolimus and sirolimus with tacrolimus, sirolimus, and methotrexate as GVHD prophylaxis [26], our group reported reduced rates of acute GVHD and NRM with the use of tacrolimus and sirolimus for GVHD prophylaxis [27–29]. To date, there are no reports on long-term survival outcomes or late effects in patients with AML in remission who underwent allogeneic HCT using TBI-based conditioning and tacrolimus/sirolimus GVHD prophylaxis. In the present study, we sought to determine the survival outcomes and late effects of this regimen and to examine factors associated with these outcomes to develop further clinical modifications aimed at improving TBI-based HCT for patients with AML.

METHODS

Patients

We reviewed the medical records of 167 consecutive patients with AML who received a fractionated TBI-based conditioning regimen followed by allogeneic HCT while in remission between 2005 and 2015 at our institution. Recipients of matched sibling donor (MSD; n = 86) and matched unrelated donor (MUD; n = 81) transplants were included in the analysis. In accordance with institutional guidelines, an MUD was defined as any unrelated donor who was HLA-matched at high resolution (10/10 match at HLA-A, -B, -C, -DR-1, and -DQ). MUD grafts were obtained from the National Marrow Donor Program. The COH Institutional Review Board approved this study.

Transplantation Procedures

FTBI was given in 11 fractions of 120 cGy in a hyperfractionated fashion (total of 1320 cGy) using a linear accelerator. In combination with TBI, for MUD graft recipients, Cy was given in 2 doses of 60 mg/kg on days −4 and −3 pre-alloHCT, for a total of 120 mg/kg. For MSD graft recipients, a single dose of etoposide 60 mg/kg was given on day −3. GVHD prophylaxis consisted of tacrolimus (starting dose of 0.02 mg/kg by continuous infusion) started 3 days before HCT and sirolimus at a 12-mg loading dose, followed by 4 mg orally daily. Subsequent dosing was done based on therapeutic monitoring to maintain levels within the recommended therapeutic range. Immunosuppression taper after day +100 post-HCT was at the treating physician’s discretion. Keratinocyte growth factor (Palifermin) was administered for mucosal protection for all recipients of FTBI [30]. SOS prophylaxis was ursodiol and low-dose unfractionated heparin drip starting on the first day of conditioning and ending after neutrophil engraftment [31].

Data Collection

Long-term health-related outcomes were captured using the established prospective follow-up study at COH. As part of the study, if the date of last medical visit at COH was not recent (more than 6 months), or if there were any gaps in the patient’s history within the window of interest, then a standard protocol was used to obtain relevant details regarding patient health from a non-COH physician. If the non-COH physician was not available or unable to provide recent information, the patient was contacted to obtain this information. Health outcomes were coded using established definitions [32–35], with appropriate source documentation (eg, echocardiography report, pathology reports) available for confirmation of outcomes.

Statistical Analysis

Descriptive statistics were used to summarize baseline patient demographic, treatment, and disease characteristics. Kaplan-Meier curves and the log-rank test were used to evaluate OS, LFS, and GVHD-free, relapse-free survival (GRFS). OS was defined as the time from transplantation to death or censored at the last follow-up in patients known to be alive. LFS was calculated from the time of transplantation to the first observation of disease recurrence or death, whichever occurred first. LFS was censored at the last follow-up if the patient remained alive and leukemia-free. GRFS was defined as the time from transplantation to the earliest occurrence of grade III-IV acute GVHD, moderate or severe chronic GVHD, disease relapse, or death. GRFS was censored if patients were alive and free of events of interest. Cumulative incidence curves and the Gray test were used to examine the differences in relapse rates and NRM. Relapse/progression was defined as the time from transplantation to relapse or progression, with NRM as a competing risk. NRM was defined as the time from transplantation to death from any cause other than relapse/progression, with relapse or progression as a competing risk. Acute GVHD was defined as the time to onset of grade II-IV acute GVHD, with relapse and death as competing events. The cumulative incidence of de novo chronic health conditions (eg, subsequent neoplasm, cardiac disease [heart failure, myocardial infarction], stroke, cataract, diabetes mellitus, hypothyroidism, avascular necrosis) was calculated, taking into consideration competing risks for right-censored data.

RESULTS

Patient and Transplantation Characteristics

A total of 167 patients were identified in the database who met the inclusion criteria for this retrospective analysis. Patients who received non-TBI-based myeloablative conditioning or GVHD prophylaxis other than tacrolimus/sirolimus were excluded. We primarily looked at outcomes of patients in 2 eras—2005 to 2009 (early cohort; n = 44) and 2010 to 2015 (late cohort; n = 123)—to account for changes in supportive care practices over a 10-year period. No differences were noted in survival, relapse, NRM, and progression-free survival between the 2 eras, and thus outcomes from the 2 groups were combined to report as a single cohort.

Patient and transplantation characteristics are summarized in Table 1. In brief, the median age at alloHCT was 41 years (range, 18 to 57 years), and the cohort was 50.9% female. Based on the European LeukemiaNet (ELN) recommendations [36], the majority of patients were at intermediate risk (n = 138), whereas 8 patients were at good risk and 13 were at high risk. AML patients underwent HCT in CR1 (n = 129; 77.2%) or CR2 (n = 38; 22.8%) from an MRD (n = 86; 51.5%) or MUD (n = 81; 48.5%) using peripheral blood stem cells (n = 148; 88.6%) or bone marrow (n = 19; 11.4%) as the graft source. The Karnofsky Performance Status score was 90 to 100 in 111 patients (66.5%). An HCT Comorbidity Index of 0 to 2 was seen in 126 patients (75.4%). According to our standard practice of using FTBI/VP-16 for MSD alloHCT and FTBI/Cy for MUD alloHCT, the majority of MUD HCT conditioning was done using TBI/Cy (n = 78; 46.7%) and TBI/VP-16 was given to 89 patients (53.3%), the majority of whom had an MSD graft [37]. Most patients received tacrolimus/sirolimus-based GVHD prophylaxis (n = 144; 86.2%).

Table 1.

Patient and Transplantation Characteristics

Characteristic	Value
Number of patients	167
Age at alloHCT, yr
Median (range)	41.0 (18.0–57.0)
IQR	31.0–47.0
Recipient sex, n (%)
Male	82 (49.1)
Female	85 (50.9)
Donor sex, n (%)
Male	101 (60.5)
Female	66 (39.5)
Female donor to male recipient, n (%)
Yes	32 (19.2)
No	135 (80.8)
Donor age at first alloHCT
Number	89 (18.0–56.0)
Median (range)	41 (18–57)
IQR	26.0–41.0
Year of HCT, n (%)
2005–2009	44 (26.4)
2010–2015	123 (73.6)
Disease status at transplantation, n (%)
CR1	129 (77.2)
≥CR2	38 (22.8)
ELN cytogenetic risk, n (%)
Not available	8 (4.8)
Favorable/intermediate	146 (87.4)
Adverse	13 (7.8)
Favorable	8 (4.8)
Intermediate	138 (82.6)
Karnofsky performance status, %, n (%)
90–100	111 (66.5)
70–80	33 (19.8)
Unknown	23 (13.8)
HCT-Comorbidity Index
Median (range)	0.0 (0.0–6.0)
IQR	0.0, 2.0
0, n (%)	81 (48.5)
1–2, n (%)	45 (26.9)
>2, n (%)	20 (12)
Unknown, n (%)	21 (12.6)
Graft source, n (%)
Bone marrow	19 (11.4)
Peripheral blood stem cells	148 (88.6)
Donor type, n (%)
MSD	86 (51.5)
MUD	81 (48.5)
ABO blood group compatibility, n (%)
ABO compatible	94 (56.3)
Minor mismatch (donor is O)	29 (17.4)
Major mismatch (recipient is O)	24 (14.4)
Bidirectional (neither is O)	20 (12)
Donor/recipient CMV serostatus, n (%)
D−/R−	13 (7.8)
D−/R+	39 (23.4)
D+/R−	9 (5.4)
D+/R+	106 (63.5)
Conditioning regimen, n (%)
CTX/FTBI	78 (46.7)
VP-16/FTBI	89 (53.3)
GVHD prophylaxis, n (%)
Tacrolimus/sirolimus	144 (86.2)
Tacrolimus/sirolimus/MTX	23 (13.8)

IQR indicates interquartile range; CMV, cytomegalovirus; D, donor; R, recipient;

Transplantation Outcomes

The median duration of follow-up was 74.8 months (range, 31.9 to 161.9 months) among survivors and 78.6 months (range, 31.9 to 161.9 months) for all patients. The 6-year OS and LFS were 60% (95% confidence interval [CI], 52% to 67%) and 56% (95% CI, 48% to 63%), respectively (Figure 1A and B). The 6-year cumulative incidences of relapse and NRM were 30% (95% CI, 23% to 37%) and 15% (95% CI, 10% to 21%), respectively (Figure 1C and D). The cumulative incidences of grade II-IV acute GVHD and grade III-IV acute GVHD were 61% (95% CI, 53% to 68%) and 17% (95% CI, 12% to 23%), respectively (Figure 2A). The cumulative incidence of chronic GVHD was 71% (95% CI, 64% to 78%) (Figure 2B). Neutrophil engraftment was achieved in 54% (95% CI, 46% to 61%) of the patients by day +14 post-alloHCT and by 99% (95% CI, 95% to 100%) by day +28 post-alloHCT (Supplementary Figure 1). The 100-day cumulative incidence of SOS was <1% (seen in only 1 patient). Finally, GRFS was 45% (95% CI, 37% to 53%) at 12 months and 39% (95% CI, 31% to 46% ) at 24 months (Supplementary Figure 2).

Figure 1.

Transplantation outcomes at 1 year and 6 years post-HCT. (A) OS. (B) LFS. (C) Relapse. (D) NRM.

Figure 2.

Cumulative incidences of acute GVHD, grade II-IV and grade III-IV, at 100 days post-HCT (A) and chronic GVHD at 6 years post-HCT (B).

Factors Associated with Transplantation Outcomes

The 6-year OS was similar in patients who underwent alloHCT in CR1 (62%; 95% CI, 53% to 70%) and CR2 or higher (55%; 95% CI, 38% to 69%; P = .20). The 6-year cumulative incidence of relapse was 29% (95% CI, 21% to 37%) for patients undergoing alloHCT in CR1 and 32% (95% CI, 18% to 47%) for those in CR2 or higher (P = .63).

As depicted in Figure 3, OS and LFS were not affected by donor type. The 6-year OS was 63% (95% CI, 52% to 73%) in MSD graft recipients compared with 57% (95% CI, 46% to 67%) in MUD graft recipients (P = .52). The 6-year LFS was 56% (95% CI, 45% to 66%) for MSD graft recipients and 55% (95% CI, 43% to 65%) for MUD graft recipients (P = .84). The 6-year cumulative incidence of relapse was 32% (95% CI, 22% to 42%) for MSD graft recipients and 27% (95% CI, 18% to 37%) for MUD graft recipients (P = .47) (Supplementary Figure 3).

Figure 3.

Comparison of 6-year survival outcomes in patients receiving transplants from matched sibling and matched unrelated donors. (A) OS. (B) LFS.

The 6-year overall survival for the non-high-risk cytogenetic group (per ELN guidelines) was 64% (95% CI, 55% to 71%), compared with 37% (95% CI, 13 % to 62%) in patients with high-risk cytogenetics (7.7%) – OS data (P = 0.054). The 6-year LFS was 58% (95% CI, 49% to 66%) in the non-high-risk cytogenetic group, compared with 38% (95% CI, 14% to 63%) in the high-risk group (P = .08) (Figure 4A). Finally, the 6-year cumulative incidence of relapse was 30% (95% CI, 23% to 37%) in the non-high-risk cytogenetic risk group versus 46% (95% CI, 18% to 71%) in the adverse-risk cytogenetic risk group (P = .18) (Figure 4B).

Figure 4.

Comparison of transplantation outcomes in patients with favorable-risk versus high-risk cytogenetics at 6 years post-HCT. (A) LFS. (B) Relapse.

Causes of Death

All-cause mortality was 42% (n = 71) in patients who were analyzed for this retrospective study (Table 2). The majority of deaths occurred within the first 2 years post-HCT, beyond which the relapse rate plateaued. Among the 71 total deaths, relapse mortality was 48.2% (n = 33), NRM was 42.6% (n = 29), and 8.8% of patients (n = 6) were lost to follow-up and the cause of death could not be ascertained. The 29 cases of NRM included 14 due to GVHD, 14 due to infectious complications, and 1 from SOS. Of the 33 patients who relapsed after alloHCT, 7 (21%) had extramedullary relapse, with the most common site he central nervous system (n = 5). Other sites of relapse were spine, buccal mucosa, and liver (n = 1 for each).

Table 2.

Causes of Death (N = 68)

Cause	Number (%)
Relapse mortality	33 (48.5)
NRM	30 (42.6)
GVHD-related death	14
Infections	14
SOS	1
Unknown	6 (8)

HCT Late Effects

Subsequent neoplasms were diagnosed in 12 patients, with a 6-year cumulative incidence of 9.5% (95% CI, 5.1% to 15.7%). The most common second malignancy was skin cancer—squamous cell carcinoma (n = 7), basal cell carcinoma (n = 2), and melanoma (n = 2)—followed by cervical intraepithelial neoplasm (n = 2). The 6-year cumulative incidences of other chronic health conditions were as follows: avascular necrosis (6.4%; 95% CI, 3.1% to 11.4%), cataract (11.1%; 95% CI, 6.5% to 17.2%), cardiac disease (9.1%; 95% CI, 5.1% to 14.5%), diabetes mellitus (19.7%; 95% CI, 13.7% to 26.6%), hypothyroidism (1.7%; 95% CI, .3% to 5.5%), and stroke (n = 4; 2.2%; 95% CI, .6% to 5.9%).

DISCUSSION

To our knowledge, this retrospective study is the largest reported to date in young adults with AML receiving TBI-based conditioning with tacrolimus/sirolimus-based GVHD prophylaxis before undergoing alloHCT. With a median follow-up of 68.8 months, the patients had a 6-year OS of 61.1%. Rates of NRM and relapse were 13.7% and 30%, respectively. Despite the high incidence of chronic GVHD (71.2%), the mortality attributed to GVHD was low, indicating the need for a better GVHD prophylactic strategy to reduce morbidity, impaired quality of life, and resource utilization in patients with chronic GVHD [38]. The higher incidence of chronic GVHD in our patients compared with earlier studies [39] may also be related to the inclusion of unrelated donors and the longer duration of follow-up in our cohort. Despite of the higher cumulative incidence of chronic GVHD, the GRFS was 45% at 1 year and 39% at 2 years, indicating the dynamic nature of GVHD. Results of a recently completed randomized 3-arm (post-transplantation Cytoxan versus CD34⁺-selected graft versus tacrolimus/methotrexate) study (BMT-CTN 1301) exploring the best strategy to impact GVHD and relapse are eagerly awaited in this regard.

HCT outcomes have been traditionally worse in patients who undergo HCT in CR2 [40]. With the approval of several new drugs to treat AML and the increased access of patients with AML to clinical trials [41,42], more patients with relapsed AML are able to achieve CR2. In our study, the 6-year OS was similar in patients who underwent alloHCT in CR1 (62.6%; 95% CI, 54% to 70.1%) or CR2 (53.1%; 95% CI, 37.5% to 66.5%; P = .210). This suggests that TBI may be able to overcome the adverse risks associated with relapsed AML, and that TBI-based conditioning might be the preferred regimen for these high-risk patients. The <1% incidence of SOS in our cohort suggests that TBI conditioning is not associated with an increased risk of this complication compared with i.v. Bu-based regimens [3,14].

The majority of patients in our review were at ELN intermediate risk, which might be due to clustering of poor-risk cytogenetics in older patients [43]. We had a limited number of patients with high-risk cytogenetics in our database, and the long-term outcome of these patients was low, with a 6-year OS of 33%; indicating the need for better post-HCT maintenance strategies and the importance of maximizing disease control by getting them into minimal residual disease-negative status before alloHCT by novel conditioning regimens [44,45]. A drawback of our study is the absence of MRD data in patients undergoing alloHCT, owing to the lack of standardized methods and insufficient appreciation of its clinical significance at the time of this study.

Multiple trials have compared TBI/Cy and Bu/Cy regimens [16,17,46,47]. However, given the small sample sizes, varying Disease Risk Index values, and center-specific features, including variations in GVHD prophylaxis and administration of FTBI (radiation dose and fractionation schedules) or Bu (oral formulation and lack of pharmacokinetic-based dosing), results of these trials are difficult to interpret. Some of the early comparative trials using TBI/Cy (120 mg/kg) and oral Bu/Cy showed equivalence of regimens in setting of CML [48,49], but favored TBI-based conditioning in patients with AML based on a nonsignificant improvement in survival [15]. More recent registry data are more conflicting, with some studies favoring i.v. Bu [20,22] and another study reporting equivalent outcomes with the use of TBI/i.v. Bu [23].

The Acute Leukemia Working Party of the EBMT compared i.v. Bu/Cy to Cy/TBI in 1659 adult patients with AML who underwent alloHCT between January 2004 and December 2010 [23]. They reported a lower mean incidence of relapse in patients who received conditioning with Cy/TBI compared those who received i.v. Bu/Cy (21 ± 1% versus 26 ± 2%; P = .012). This significant difference was noted even though the group assigned to the Cy/TBI arm had a higher proportion of patients with poor-risk cytogenetics. LFS was similar in the Cy/TBI (64 ± 2%) and Bu/Cy (61 ± 2%) groups. Subgroup analysis favored Cy/TBI in patients with poor-risk cytogenetics (LFS, 60 ± 5% versus 43 ± 9%) owing to the higher relapse rate in the i.v. Bu group. Those data and our present findings reinforce the continued benefit of TBI for all subgroups of AML patients eligible for myeloablative HCT.

A major concern in patients after alloHCT is long-term toxicities resulting in late morbidity and mortality [50–53]. In our study, it was reassuring that the risk of second malignant neoplasms was low at 9.5%, which is comparable to previous published reports [54]. The majority of secondary malignant neoplasms were skin cancers, including 2 cases of melanoma that were cured by wide excision only. Most of the patients with secondary skin cancers required a single excision for complete removal; however, 2 patients required multiple excisions (13 procedures in one patient and 8 procedures in the other) for disease control. One of these patients was on prolonged treatment with voriconazole for previous Aspergillus pneumonia, and the second patient had sclerodermatous chronic GVHD, both of which are known risk factors for secondary skin cancers [55,56]. The 2 patients with melanoma had early-stage disease cured by wide excision alone. Close follow-up with dermatology for annual skin examination and skin protective measures are needed, especially in patients with active chronic GVHD [57], are on voriconazole [55] or are not on sirolimus-based therapy for chronic GVHD [58]. Given late cardiac and metabolic events in allogeneic HCT recipients, multidisciplinary follow-up is required to manage these complications.

In conclusion, our retrospective study investigating long-term outcomes of alloHCT in young patients with AML with favorable characteristics (CR, good performance status, and intermediate-risk cytogenetics) TBI-based conditioning with tacrolimus/sirolimus GVHD prophylaxis demonstrated excellent 6-year OS of 61% with no increase in long-term complications and low risk of SOS. Based on our results, in centers with expertise in radiation delivery and management of complications arising thereof, TBI-based conditioning regimens should remain the preferred choice for myeloablative alloHCT for this patient population. Continued refinements in radiation techniques to deliver targeted forms of TBI, such as the use of volumetric modulated arc therapy (IMRT) or helical tomographic IMRT with dose escalation, may further improve outcomes through targeted delivery of higher doses of radiation and improved control of underlying disease [44,59,60]. In addition, these techniques entail reduced radiation exposure to critical organs while improving dose coverage of target organs [61], thereby reducing late effects in long-term survivors. Better GVHD prophylactic strategies with the use of post-transplantation Cy [62], and graft engineering with in vivo or ex vivo T cell depletion [63,64] may help reduce the chronic GVHD burden in patients receiving TBI-based conditioning. Prospective studies in which eligible patients have an equal chance of randomization to either arm (TBI/Cy or i.v. Bu-based) [14,65] will be best to determine the superior myeloablative regimen for patients with AML.