Kyoo-Hyung Lee

Jiyeon Kim, M.D., Miyoung Kim, M.D., Ph.D., Daehyun Chu, M.D., Young-Uk Cho, M.D., Ph.D., Sang-Hyun Hwang, M.D., Ph.D., Seongsoo Jang, M.D., Ph.D., Eul-Ju Seo, M.D., Ph.D., Eun-Ji Choi, M.D., Ph.D., Han-Seung Park, M.D., Ph.D., Jung-Hee Lee, M.D., Ph.D., Je-Hwan Lee, M.D., Ph.D., Dok Hyun Yoon, M.D., Ph.D., Heounjeong Go, M.D., Ph.D., Chan Sik Park, M.D., Ph.D., Kyoo-Hyung Lee, M.D., Ph.D., Chan-Jeoung Park, M.D., Ph.D., and Seungwoo Hwang, Ph.D.. Ann Lab Med -0001;0:. https://doi.org/10.3343/alm.2025.0395

Central nervous system (CNS) relapse in acute myeloid leukemia (AML) is uncommon but associated with poor outcomes. This retrospective study analyzed 960 AML patients who underwent hematopoietic cell transplantation (HCT) at Asan Medical Center between 2000 and 2020 to assess the efficacy of prophylactic intrathecal chemotherapy. Per protocol, patients received up to four intrathecal methotrexate (MTX). CNS relapse occurred in 2.8%, with a median onset of 6.3 months post-HCT. The 5-year CNS relapse rate was lower in the prophylactic MTX group compared to the non-MTX group (2.7% vs. 7.4%, p = 0.132), though not statistically significant. Increased number of intrathecal treatments did not reduce CNS relapse. Risk factors for CNS relapse included high white blood cell count, extramedullary disease, FAB M3/M4, and poor-risk cytogenetics. Prior CNS involvement was significant only in univariate analysis. These findings suggest intrathecal prophylaxis may benefit high-risk patients but is not necessary for all AML patients undergoing HCT.

The role of allogeneic hematopoietic cell transplantation (alloHCT) as first-line therapy in elderly patients with acquired aplastic anemia (AA) remains uncertain. While immunosuppressive therapy (IST) is the standard initial treatment for patients aged ≥40 years, recent evidence suggests that upfront alloHCT may offer improved outcomes. We conducted a retrospective multicenter study of 173 Korean patients aged ≥40 years who underwent alloHCT between 1999 and 2017. Patients received either upfront alloHCT without prior IST (n=66) or delayed alloHCT following IST failure or relapse (n=107). The upfront group had a higher rate of matched sibling donors (78.8% vs. 50.5%, p<0.001) and shorter time to transplant (median, 3.4 vs. 8.3 months; p=0.046). Rates of engraftment, hematopoietic recovery, and acute and chronic graft-versus-host disease were comparable. Five-year overall survival was higher in the upfront group (73.5% vs. 56.4%, p=0.059). In multivariate analysis, upfront alloHCT (hazard ratio [HR], 2.476; p=0.004) and ECOG performance status ≤1 (HR, 4.736; p<0.001) were independently associated with improved survival. These findings suggest that upfront alloHCT may be a viable first-line option for selected elderly AA patients and support re-evaluation of treatment algorithms that currently prioritize IST.

Clinical effect of donor-derived natural killer cell infusion (DNKI) after HLA-haploidentical hematopoietic cell transplantation (HCT) was evaluated in high-risk myeloid malignancy in phase 2, randomized trial. Seventy-six evaluable patients (aged 21–70 years) were randomized to receive DNKI (N = 40) or not (N = 36) after haploidentical HCT. For the HCT conditioning, busulfan, fludarabine, and anti-thymocyte globulin were administered. DNKI was given twice 13 and 20 days after HCT. Four patients in the DNKI group failed to receive DNKI. In the remaining 36 patients, median DNKI doses were 1.0 × 10⁸/kg and 1.4 × 10⁸/kg on days 13 and 20, respectively. Intention-to-treat analysis showed a lower disease progression for the DNKI group (30-month cumulative incidence, 35% vs 61%, P = 0.040; subdistribution hazard ratio, 0.50). Furthermore, at 3 months after HCT, the DNKI patients showed a 1.8- and 2.6-fold higher median absolute blood count of NK and T cells, respectively. scRNA-sequencing analysis in seven study patients showed that there was a marked increase in memory-like NK cells in DNKI patients which, in turn, expanded the CD8⁺ effector-memory T cells. In high-risk myeloid malignancy, DNKI after haploidentical HCT reduced disease progression. This enhanced graft-vs-leukemia effect may be related to the DNKI-induced, post-HCT expansion of NK and T cells. Clinical trial number: NCT02477787.

In a prospective, explorative study, the donor-source difference of haploidentical family (HF), matched sibling (MS), and unrelated donors (UD) was evaluated for the outcome of haematopoietic cell transplantations (HCT) in 101 patients with acute myeloid leukaemia (AML) in complete remission (CR). To eliminate compounding effects, a uniform conditioning regimen containing antithymocyte globulin (ATG) was used. After transplantation, there was a significantly higher cumulative incidence of acute graft-versus-host disease (GVHD) in HF-HCT patients (49%, 7%, and 16% for HF-, MS- and UD-HCT respectively; p < 0.001). A quarter of acute GVHD cases observed in HF-HCT patients occurred within three days of engraftment and were characterized by diffuse skin rash, fever, weight gain, and hypoalbuminaemia. This peri-engraftment acute GVHD was not observed in MS-HCT or UD-HCT patients. Additionally, a significantly higher proportion of HF-HCT patients achieved complete donor chimaerism in the peripheral mononuclear cells at one month (88%, 46%, and 69% for HF-, MS- and UD-HCT respectively; p = 0.001). There was no significant difference in engraftment, chronic GVHD, leukaemia recurrence, non-relapse mortality, and patient survival. In patients with AML in CR who received HCT using ATG-containing conditioning, stronger donor-patient alloreactivity was observed in HF-HCT, in terms of increased acute GVHD and higher likelihood of complete donor chimaerism.

Background: Scoring systems developed for predicting survival after allogeneic hematopoietic cell transplantation (HCT) show suboptimal prediction power, and various factors affect posttransplantation outcomes. Objective: A prediction model using a machine learning–based algorithm can be an alternative for concurrently applying multiple variables and can reduce potential biases. In this regard, the aim of this study is to establish and validate a machine learning–based predictive model for survival after allogeneic HCT in patients with hematologic malignancies. Methods: Data from 1470 patients with hematologic malignancies who underwent allogeneic HCT between December 1993 and June 2020 at Asan Medical Center, Seoul, South Korea, were retrospectively analyzed. Using the gradient boosting machine algorithm, we evaluated a model predicting the 5-year posttransplantation survival through 10-fold cross-validation. Results: The prediction model showed good performance with a mean area under the receiver operating characteristic curve of 0.788 (SD 0.03). Furthermore, we developed a risk score predicting probabilities of posttransplantation survival in 294 randomly selected patients, and an agreement between the estimated predicted and observed risks of overall death, nonrelapse mortality, and relapse incidence was observed according to the risk score. Additionally, the calculated score demonstrated the possibility of predicting survival according to the different transplantation-related factors, with the visualization of the importance of each variable. Conclusions: We developed a machine learning–based model for predicting long-term survival after allogeneic HCT in patients with hematologic malignancies. Our model provides a method for making decisions regarding patient and donor candidates or selecting transplantation-related resources, such as conditioning regimens.

BACKGROUND Scoring systems developed for predicting survival after allogeneic hematopoietic cell transplantation (HCT) show suboptimal prediction power, and various factors affect posttransplantation outcomes. OBJECTIVE A prediction model using a machine learning–based algorithm can be an alternative for concurrently applying multiple variables and can reduce potential biases. In this regard, the aim of this study is to establish and validate a machine learning–based predictive model for survival after allogeneic HCT in patients with hematologic malignancies. METHODS Data from 1470 patients with hematologic malignancies who underwent allogeneic HCT between December 1993 and June 2020 at Asan Medical Center, Seoul, South Korea, were retrospectively analyzed. Using the gradient boosting machine algorithm, we evaluated a model predicting the 5-year posttransplantation survival through 10-fold cross-validation. RESULTS The prediction model showed good performance with a mean area under the receiver operating characteristic curve of 0.788 (SD 0.03). Furthermore, we developed a risk score predicting probabilities of posttransplantation survival in 294 randomly selected patients, and an agreement between the estimated predicted and observed risks of overall death, nonrelapse mortality, and relapse incidence was observed according to the risk score. Additionally, the calculated score demonstrated the possibility of predicting survival according to the different transplantation-related factors, with the visualization of the importance of each variable. CONCLUSIONS We developed a machine learning–based model for predicting long-term survival after allogeneic HCT in patients with hematologic malignancies. Our model provides a method for making decisions regarding patient and donor candidates or selecting transplantation-related resources, such as conditioning regimens.