We recently showed that empiric weight-based ATG dosing in heavier patients (pts) prior to ex vivo CD34-selected allogeneic hematopoietic cell transplantation (allo-HCT) is associated with high ...non-relapse mortality (NRM) and poor survival. A more precise method is to assess the relationship between ATG pharmacokinetic (PK) exposure and outcomes. Therefore, we estimated ATG exposure using a population PK model and hypothesized that peri-HCT ATG overexposure would similarly result in unfavorable outcomes secondary to poorer immune reconstitution (IR).
We studied 415 consecutive adult pts who underwent first myeloablative ex vivo CD34-selected allo-HCT with ATG (Thymoglobulin) between 2006 and 2017. We estimated pre-/post-HCT and total ATG area under the curve (AUC, in mg*d/L) for all pts using an established PK model (Admiraal et al., Lancet Hematol 2017). We evaluated the association between ATG exposure and outcomes using a nonparametric estimate of median relapse-free survival (RFS) by ATG exposure to define an optimal ATG AUC cut point (highest RFS) that was used to estimate RFS and NRM. We also evaluated how CD4+ T cell IR related to NRM in a subgroup of 232 pts with sufficient data. IR was defined according to prior studies as absolute CD4+ ≥ 50 at ≥ 2 time points before day +100 and 1 subsequent time point within 60 days.
Among all pts, median age was 55 (range, 19-73), 228 (55%) were male, 204 (49%) had AML, 105 (25%) MDS, 50 (12%) ALL, and 56 (14%) other hematologic diseases. HCT-CI was 0 in 85 (20%), 1-2 in 146 (35%), and ≥ 3 in 184 (45%) of pts. Allografts were HLA matched related in 150 (36%), matched unrelated in 202 (49), and mismatched in 63 (15%). CMV serostatus was positive in 251 (60%). Conditioning regimens were chemotherapy based in 279 (67%) and TBI based in 136 (33%). All but 2 pts, who were inevaluable given early deaths, engrafted. All pts received 1-3 doses of ATG 2.5 mg/kg between days -4 and -1: 361 (87%) received 2 doses, 47 (11%) 3 doses, and 7 (2%) 1 dose. Among all pts, median estimated ATG AUC values were: pre-HCT 19 (range, 9-36), post-HCT 48 (range, 17-101), and total 66 (29-125). Median follow-up among survivors was 3.9 years (range, 0.7-12). A post-HCT ATG AUC of >60 was associated with poorer median predicted RFS (Figure 1). Pts with an estimated post-HCT ATG AUC of >60 and <60, respectively, had 5-year RFS of 31% and 55%, p=0.008 and higher 5-year NRM of 51% and 22%, p<0.001 (Figure 2). Pts who did not have or had IR, respectively, by day +100 post-HCT had higher 5-year NRM, 34% versus 8%, p<0.001 (Figure 3).
Population PK modeling confirms that post-HCT ATG overexposure in certain pts undergoing ex vivo CD34-selected allo-HCT leads to inferior RFS and markedly higher NRM possibly due to poor IR. Identifying the optimal ATG exposure and individualizing ATG dosing may improve early CD4+ IR and ultimately improve survival in these pts.
Studies evaluating the impact of 11q23 / KMT2A / MLL translocations on post-transplant outcomes as compared to other intermediate- and adverse-risk cytogenetic characteristics are lacking.
We ...considered 8709 AML patients from the CIBMTR database for inclusion. We evaluated relapse, non-relapse mortality (NRM), disease-free survival (DFS), and overall survival (OS) for patients who were transplanted in CR1 stratified by presence of 11q23 rearrangement, intermediate- or adverse-risk disease. Assignment of cytogenetic risk was performed by hand, based on the 2010 AML-MRC Classification System. A multivariate analysis was performed using a Cox proportional hazards model.
Characteristics of 3779 patients, selected based on disease and donor type, HSCT in CR1, and availability of data, are shown in Figure 1. Of these, 426 patients had an 11q23-rearrangement, 2384 had intermediate-risk disease, and 969 had adverse-risk disease. Of the patients with 11q23 translocations, 112 (26%) had t(9;11), 62 (15%) had t(11;19), 41 (10%) had t(6;11), 28 (7%) had t(10;11), and 47 (11%) had other partners. No partner was provided in 136 cases (32%).
Figures 2 and 3 show the DFS, OS, NRM, and relapse data stratified by cytogenetic category. Based on the multivariate analysis, DFS was lowest in adverse-risk patients (HR for death 1.47, 1.26 for MLL, p < 0.001 and 0.002, respectively). OS was shorter for the MLL- and adverse-risk groups, with HR of 1.32 and 1.45, respectively, and p < 0.001 for both. The HR for NRM was significantly worse for patients with adverse-risk disease (HR 1.17, P=0.05), but not for patients with MLL disease. The HR for relapse in MLL patients was 1.27, p = 0.01 and 1.71, p < 0.001 in adverse risk.
Stratification of outcomes by translocation partner did not yield any statistically significant differences between groups, however a descriptive analysis of these data will be presented.
Measurable residual disease (MRD) status did not appear to play an independent role in NRM or in OS. However, the multivariate HR for relapse in patients with MRD-positivity was 1.23, p < 0.006. For DFS, the HR for death was 1.13 for MRD-positive patients (p = 0.04). Further data and figures will be presented.
Patients with MLL-rearranged AML had post-HSCT DFS and OS that were more similar to patients with adverse-risk disease than to those with intermediate-risk characteristics. MLL patients had a higher probability of relapse than intermediate-, but not adverse-risk, patients. NRM was similar among all groups. Both prognostic classification and MRD status independently contributed to the risk of relapse and DFS, but MRD was not an independent factor in predicting NRM or OS. Our study suggests that patients with MLL rearrangements do almost as poorly as those with adverse-risk disease and that novel approaches for this high-risk population represent an urgent unmet clinical need.
Acute graft-versus-host disease (aGVHD) remains a substantial cause of morbidity and mortality after unmodified and ex-vivo CD34+ selected/T-cell depleted (TCD) allogeneic hematopoietic stem cell ...transplantation (alloHSCT). We identified a cluster of patients who developed severe cutaneous aGVHD and expressed MHC class I HLA-A*0101.
To investigate if HLA-A*0101 expression correlates with an increased risk of severe cutaneous aGVHD after alloHSCT.
We evaluated 831 patients who received unmodified or TCD allograft at a single institution between 03/2010 and 02/2017. Patients who received cord blood transplants or <8/8 HLA-matched grafts were excluded. For patients with cutaneous aGVHD, we assessed donor-recipient HLA-typing, time of onset, grade at onset, and highest overall grade of cutaneous aGVHD.
HLA-A*0101 was expressed in 206 (25%) patients (98 TCD, 108 unmodified). These patients had similar demographics to those lacking HLA-A*0101 (Table). At day 180, patients expressing HLA-A*0101 had higher incidence of grade III-IV cutaneous aGVHD compared with patients lacking HLA-A*0101 expression in both the TCD (8% vs. 3%, p=0.027) and unmodified (11% vs. 4%, p=0.01) cohorts (Fig 1). In a multivariate cause-specific Cox model, the presence of HLA-A*0101 in the TCD alloHSCT cohort correlated with increased risk of grade III-IV cutaneous aGVHD HR=2.79 (95% CI: 1.07-7.28), p=0.036 after adjusting for related status of donor. In the unmodified alloHSCT cohort, presence of HLA-A*0101 allele correlated with increased risk of grade III-IV cutaneous aGVHD HR=2.68 (95% CI: 1.24-5.79), p = 0.012) after adjusting for myeloablative v non-myeloablative conditioning and donor relationship status. There were no statistically significant differences in OS or TRM between HLA-A*0101 expressing vs non-expressing patients in the cohorts.
In this group, donor/recipient expression of HLA-A*0101 correlated with an increased incidence and severity of cutaneous aGVHD after TCD and unmodified alloHSCT. If confirmed in a larger cohort, these findings have potential practical implications in the development of prophylaxis or early therapeutic strategies targeting the skin in this high-risk population.
The prevalence of age-related impairments in older patients (pts) undergoing hematopoietic cell transplantation (HCT), and its relationship with transplant outcomes is largely unknown.
In a pilot ...quality improvement project, assessment of functional, social, emotional, cognition, sensory, nutritional and polypharmacy domains was performed using an electronic rapid fitness assessment (eRFA) developed at MSKCC (Shahrokni JNCCN 2017). Older adults (age ≥ 60 years) undergoing HCT for hematologic malignancies from 2015 through 2018 in 2 transplant clinics were included. The prevalence of geriatric impairments, transplant-associated toxicities and survival were analyzed under an IRB approved waiver of authorization.
Patient characteristics are shown in Table 1. The median day from eRFA to HCT was 45 days (range 14-162). Median time to complete eRFA was 10 minutes. Table 2 has a summary of geriatric impairments. Majority of pts (n=33) had at least 1 impairment. The most prevalent were social, emotional, polypharmacy and functional. Median number of geriatric impairments per patient was 4 (range 0-9). Median toxicities per patients was 4 (range 0-12). The most common toxicities were: 25 metabolic, 22 infections, 21 oral/gastrointestinal, and 16 cardiopulmonary. The only pt admitted to the ICU was an allo-HCT pt with 5 impairments. Amongst the 8 pts (4 AHCT, 4 allo-HCT) readmitted to the hospital within the first 100 days of HCT, the median number of impairments was 4 (range 0-9). Seven patients died, 5 from disease, and 2 from GvHD. With a median follow up of 9 months (range 1-38) among survivors, the 18-months progression-free and overall survival were 0.71 (0.55-0.91) and 0.7 (0.51-0.95) respectively (Figure 1).
Performing pre-HCT eRFA among older transplant patients is feasible. With the median completion time of 10 minutes, the eRFA can identify geriatric impairments of this vulnerable population. Future larger studies should assess whether these impairments have association with the outcomes, and whether interventions to improve geriatric impairments also improve transplant outcomes.
•MAC with CD34+ cell selection results in similar DFS and OS as RIC with unmanipulated grafts in AML and MDS.•MAC and CD34+ cell selection showed lower risk of GVHD and higher CRFS than RIC with ...unmanipulated grafts.•Patients with MAC and CD34+ cell selection had a lower risk of relapse compared with patients with RIC with unmanipulated grafts.•For patients with low HCT-CI, MAC with CD34+ cell selection resulted in higher OS compared with unmanipulated RIC.
Reduced-intensity conditioning (RIC) and T cell depletion (TCD) through CD34+ cell selection without the use of post-transplantation immunosuppression are 2 strategies used to reduce nonrelapse mortality (NRM) in older patients after allogeneic hematopoietic cell transplantation (allo-HCT). To compare the efficacy of the RIC and TCD approaches, we evaluated the outcomes of patients age >50 years with acute myelogenous leukemia (AML) and myelodysplastic syndrome (MDS) who underwent allo-HCT from an HLA-matched donor with one of these strategies. Baseline characteristics were comparable in the patients receiving TCD (n = 204) and those receiving RIC (n = 151), except for a higher proportion of unrelated donors (68% versus 40%; P < .001) and a higher comorbidity burden (Hematopoietic Cell Transplantation Comorbidity Index HCT-CI ≥3: 51% versus 38%; P < .001) in the TCD cohort. Analysis of outcomes at 3 years showed a higher chronic graft-versus-host disease (GVHD)/relapse-free survival (CRFS) (51% versus 7%; P < .001), lower incidences of grade II-IV acute GVHD (18% versus 46% at day +180) and chronic GVHD (6% versus 55% at 3 years; P < .001), and a lower incidence of relapse (19% versus 33% at 3 years; P = .001) in the TCD group compared with the RIC group. Relapse-free survival (RFS), overall survival (OS), and NRM were similar in the 2 groups. Combining transplantation approach (RIC versus TCD) and comorbidity burden (HCT-CI 0-2 versus ≥3), patients with an HCT-CI score of 0-2 seemed to benefit from the TCD approach. In conclusion, in this retrospective study, the use of a CD34+ cell-selected graft and a myeloablative conditioning regimen was associated with higher CRFS and similar RFS and OS compared with unmodified allo-RIC in patients age >50 years with AML and MDS.
Biomarkers are used in clinical medicine to classify individuals into a binary disease state. It is often of interest to contrast multiple biomarkers to identify the one with the best discriminatory ...ability. Other times it is of interest to describe how the performance of a biomarker varies across covariates. The receiver operating characteristic (ROC) curve and functionals of the curve, such as the area under the curve (AUC) and partial-AUC, are commonly used to characterize the diagnostic ability of one or more biomarkers. Various parametric and semiparametric methods have been proposed in this setting to contrast how the ROC curve or various functionals change across different biomarkers or covariates. The required parameterization of these methods impose varying restrictions on the global shape of the ROC curve or biomarker distribution. This dissertation explores nonparametric alternatives to ROC curve regression. These methods use an adaptive modeling approach, such as regression trees, to flexibly model the covariate relationship on the ROC curve or the biomarker distribution. This allows for an entirely distribution-free approach that imposes minimal restrictions on the covariate form or shape of the ROC curve. These methods can be used to visually display the predictive space for a functional of the ROC curve and to estimate contrasts across different biomarkers or covariates. These nonparametric approaches are compared to current semiparametric models of the ROC curve and AUC. Through simulations, we quantify the efficiency loss of the nonparametric methods when the semiparametric methods are correctly specified, and demonstrate the added accuracy of the nonparametric methods under various levels of model misspecification for the semiparametric models. We find that these methods are able to estimate the predictive space well under a range of different covariate relationships and ROC curve scenarios. Lastly, these approaches are illustrated in two examples. The first example explores the discriminatory ability of two biomarkers—total PSA and ratio PSA—for diagnosing prostate cancer. The relative discriminatory ability of the biomarkers are compared at different time points prior to diagnosis. This illustrates how disease-specific covariates can potentially be incorporated in these methods. The second example evaluates the prognostic ability of C-reactive protein as a biomarker for all-cause mortality. The alternative methods are used to visualize and contrast the prognostic ability of C-reactive protein across key covariates.
Abstract A preparative regimen of reduced intensity that can reliably engraft cord blood (CB) and can be used as an alternative to either high-dose myeloablative or nonmyeloablative conditioning is ...needed. We evaluated double-unit CB transplantation in 30 patients (median age, 56 years; range, 18 to 69) with acute leukemia or myelodysplasia using a regimen of cyclophosphamide 50 mg/kg, fludarabine 150 mg/m2 , thiotepa 10 mg/kg, and 400 cGy total body irradiation with cyclosporine-A/mycophenolate mofetil immunosuppression. Ninety-seven percent of patients engrafted at a median of 26 days (range, 13 to 43), and 93% of patients had recovered platelets by day 180. Grades II to IV acute graft-versus-host disease (GVHD) incidence was 67% at day 180, and chronic GVHD was 10% at 1 year. Transplant-related mortality was 20% at day 180, and relapse was 11% at 2 years. Overall, 2-year disease-free survival (DFS) was 60% at 2 years. A hierarchy in DFS was seen according to the Sorror comorbidity score: 11 patients (median age, 55 years) with a score of 1 had a 2-year DFS of 82% compared with 62% in 9 patients (median age, 51 years) with a score of 2 to 3 and 40% in 11 patients (median age, 58 years) with a score of 4 to 5 ( P = .13). This reduced-intensity regimen combined with double-unit CB transplantation reliably facilitates sustained donor engraftment without antithymocyte globulin. Although other approaches are needed in patients with high comorbidity scores, this regimen is highly effective in patients ≥50 years old who are otherwise reasonably fit. It also represents a promising alternative to high-dose conditioning in younger patients.
PDF
