Liesbeth C. de Wreede

Allogeneic hematopoietic cell transplantation (allo-HCT) hinges on a delicate trade-off between graft-versus-tumor control and graft-versus-host disease (GVHD), mediated by donor T-cell recognition of antigens presented by recipient HLA molecules. We hypothesized that, beyond allele-level matching, sequence divergence at peptide-binding grooves across donor and recipient HLA loci shapes these responses. To this end, we evaluated the effect of HLA evolutionary divergence (HED), a metric quantifying amino acid variability at HLA peptide-binding sites, on selected hematological malignancies in 4695 patients undergoing allo-HCT from a 9 of 10 mismatched unrelated donor, reported to the European Society for Blood and Marrow Transplantation database. We examined (1) locus-specific recipient HED (HED-R) and (2) “HED-mismatch” (HED-MM), capturing immunopeptidome divergence at the mismatched locus. Although dichotomous mismatch status explained differences in survival and acute GVHD risk (with overall greater detriment for class I loci), HED metrics uncovered substantial within-mismatch heterogeneity. In DRB1-mismatched subgroup, HED-MM at this locus independently predicted inferior relapse-free survival (RFS) with an attenuating time-dependent association, further modulated by crosslocus HED-R. In this subgroup, higher HED-R at HLA-A and HLA-C associated with increased risks of acute GVHD and nonrelapse mortality, respectively. Among HLA-B–mismatched pairs, higher DRB1 HED-R associated with worse overall survival and RFS and higher relapse risk. In the HLA-A–mismatched subgroup, higher HED-R at HLA-A increased chronic GVHD risk. Collectively, HED-derived metrics complement conventional mismatch classification by capturing qualitative differences in donor-recipient immunopeptidome interactions and reveal a complex, nonlinear interplay among alleles across mismatch subgroups that modulates the clinical impact of mismatching.

We present a new framework to estimate excess cause‐specific mortality by integrating relative survival into a multi‐state model, motivated by the question of how vaccination and infection impacted different components of excess mortality for the older Dutch population during the first two years of the COVID‐19 pandemic in the Netherlands. The proportion of the study population who died from COVID‐19, 0.98%, was nearly three times the excess proportion of all‐cause mortality, 0.34%, reflecting a mortality substitution effect: other causes of death occurred less than expected. COVID‐19 infection increased short‐term mortality across all causes of death, with residual risk beyond 28 days postinfection. Vaccination was associated with reduced COVID‐19 mortality risk in both the acute and long‐term infection phases, with consistent results across landmark times. Results suggest a healthy vaccinee bias, specifically for cardiovascular diseases, mental & neurological disorders and neoplasms. However, this bias seems far smaller than the impact of vaccination on reduction of COVID‐19 mortality.

The Cox proportional hazards (PH) model remains the dominant approach in survival analysis, but it has two important limitations. First, it relies on the PH assumption, asserting that covariate effects act multiplicatively and remain constant over time. This assumption is often violated in practice, which complicates interpretation and compromises predictive accuracy. Second, the model is non‐collapsible, that is, marginal and conditional effects differ even in the absence of confounding, which leads to a lack of internal consistency and to the fact that the hazard ratio does not allow a causal interpretation in the presence of unobserved heterogeneity. In this article we review two alternative regression methods, the additive hazards model and the accelerated failure time (AFT) model, that address these limitations. Both models are collapsible and internally consistent, providing effect measures that retain their interpretation across conditioning sets. The additive hazards model provides a highly flexible, nonparametric formulation that naturally accommodates time‐varying covariate effects, while recent maximum likelihood methods further improve its practical performance. The AFT model offers a more intuitive interpretation by quantifying how covariates accelerate or decelerate event times; both parametric and semi‐parametric versions are discussed. We review theoretical properties, estimation methods, extensions, possible applications, and software implementations of both models. We illustrate similarities and differences between the models through a conceptual hazard and survival comparison and through a comparative data example.

Practical estimation of patient risk is often complicated by predictor values being unobserved or missing. In this paper we study the calibration and appli

The Coronavirus Diseases (COVID-19) pandemic led to excess mortality in many countries, i.e., to more deaths than expected. Older individuals generally had higher absolute risks of excess death. The complex dynamics of infection, vaccination and excess mortality have not yet been captured in one comprehensive model. With nationwide and unselected data from Statistics Netherlands, we analyzed the impact of documented infection and vaccination on excess mortality during 2020 and 2021 in the Dutch population aged over 63 on 1 January 2020 ( n = 3,826,770) by incorporating relative survival into a multi-state model considering COVID-19 (re)infection, vaccination and death. Background mortality was based on the observed mortality in 2015–2019 per sex, age and month of the year. All analyses were performed for the total cohort as well as stratified per sex and age category. The absolute excess mortality in 2020–2021 was 0.34% (95% confidence interval 0.32–0.37), comprising 4.41% of the observed mortality. It was higher in men (except in the youngest age group) and older individuals. Excess mortality occurred mostly during the first four weeks after a positive COVID-19 test, but also thereafter. If infection occurred after vaccination, excess mortality was still observed, but considerably less than without prior vaccination. These patterns were observed in all groups. In conclusion, these analyses demonstrate the substantial impact of COVID-19 on excess mortality during and after acute SARS-CoV-2 infection in individuals aged over 63 years. Moreover, the results show a reduction of excess mortality after vaccination for all groups in this cohort.

Many clinical studies are based on registry analyses, but exact approaches of data extraction and pre-processing are rarely included, while this is critical for reliability and reproducibility of results. We aimed to develop an open-source data extraction pipeline which generates a ready-to-analyze dataset focused on relevant determinants of outcomes after hematopoietic stem cell transplantation (HSCT). This pipeline was developed using EBMT registry data, including 54,457 allogeneic and 63,651 autologous HSCT procedures. The pipeline determines HLA matching from molecular data, assesses cytogenetic risk for acute myeloid leukemia and myelodysplastic syndrome, processes molecular markers, assigns the hematopoietic cell transplantation comorbidity index (HCT-CI) based on comorbidities, and maps disease states to simplified categories. We prospectively assessed the recently developed disease risk stratification system (DRSS), showing that the pipeline produces consistent results with previous studies. The hazard ratio correlation between our cohort and the original DRSS derivation cohort was 0.92 with a 2-year AUC of 0.616, indicating similar effects and predictive performance. We aim to establish a new standard by promoting transparent, standardized and uniform extraction of registry data, enhancing reproducibility in registry studies.

Allogeneic hematopoietic cell transplantation (allo-HCT) hinges on a delicate trade-off between graft-versus-tumor control and graft-versus-host disease (GvHD), mediated by donor T-cell recognition of antigens presented by recipient human leukocyte antigen (HLA) molecules. We hypothesized that, beyond allele-level matching, sequence divergence at peptide-binding grooves across donor and recipient HLA loci shapes these responses. To this end, we evaluated the effect of HLA evolutionary divergence (HED), a metric quantifying amino acid variability at HLA peptide-binding sites, on selected hematological malignancies in 4,695 patients undergoing allo-HCT from a 9/10 mismatched unrelated donor (MMUD), reported to the EBMT database. We examined (i) locus-specific recipient HED (HED-R) and (ii) “HED-mismatch” (HED-MM), capturing immunopeptidome divergence at the mismatched locus. While dichotomous mismatch status explained differences in survival and acute GvHD risk (with overall greater detriment for class I loci), HED metrics uncovered substantial within-mismatch heterogeneity. In DRB1 mismatched subgroup, HED-MM at this locus, independently predicted inferior relapse-free survival (RFS) with an attenuating time-dependent association, further modulated by cross-locus HED-R. In this subgroup, higher HED-R at HLA-A and HLA-C associated with increased risks of acute GvHD and non-relapse mortality, respectively. Among HLA-B–mismatched pairs, higher DRB1 HED-R associated with worse overall survival (OS) and RFS and higher relapse risk. In the HLA-A–mismatched subgroup, higher HED-R at HLA-A increased chronic GvHD risk. Collectively, HED-derived metrics complement conventional mismatch classification by capturing qualitative differences in donor–recipient immunopeptidome interactions and reveal a complex, non-linear interplay among alleles across mismatch subgroups that modulates the clinical impact of mismatching. Keypoints: In mismatched unrelated HCT, baseline risk varies across mismatch constellations, with class I mismatches more detrimental than class II. HED complements conventional HLA mismatch classification by capturing qualitative donor-recipient immunopeptidome interactions.

Discussion remains concerning the safety and tolerability of anti-thymocyte globulin (ATG)-based immunosuppressive therapy (IST) in older patients with aplastic anaemia (AA). Using data of 127 consecutive patients from the Dutch adult AA registry, we evaluated long-term treatment success of standard ATG-based IST as first-line treatment with a multi-state model. Only one death was potentially associated with ATG. Overall survival at 5 years was 79%. We defined Transplantation-, Treatment- and Disease-Free Survival (TT-DFS) as the ultimate success of this treatment. This means that a patient is alive, is currently transfusion-independent without having received an allogeneic stem cell transplantation, has not developed AML or MDS, and has stopped all medication for AA. The probability of TT-DFS was 42% at 5 years after start of IST. In patients younger than 40 years ( n = 36) and in patients aged 60 or above ( n = 53), this was 58% and 34%, respectively. Older age, more severe AA and absence of a PNH-clone of ≥ 1% all reduced the likelihood of reaching TT-DFS. These analyses on unselected nationwide data indicate that ATG-based IST is effective and safe also in older patients. They suggest that age, AA severity and presence of a PNH-clone should be taken into account when considering this treatment in older patients.

Donor lymphocyte infusion (DLI) after allogeneic stem cell transplantation (alloSCT) can boost Graft-versus-Leukaemia (GvL) reactivity but may induce Graft-versus-Host-Disease (GvHD). It is essential to understand which factors besides timing, donor type, and dose influence DLI alloreactivity. We previously identified viral infections, ≥ 5% patient cells in bone marrow chimerism, and lymphopenia at the time of DLI as relevant factors for GvHD after DLI following alemtuzumab-based T-cell depletion. Here, we investigated these factors and the alloreactivity after DLI following alloSCT with posttransplant cyclophosphamide in 83 patients with acute leukaemia/myelodysplastic syndrome receiving a prophylactic or preemptive DLI. 5% had viral infections close to DLI, 6% had ≥ 5% mixed chimerism, and 17% had lymphopenia. 2-year cumulative incidence of GvHD requiring systemic treatment was low: 7% (95%-confidence interval 1–14%). 22 of the 28 patients with ≥ 1% mixed chimerism at the time of DLI (79%) converted to full-donor chimerism. None of these responders relapsed, indicating achievement of GvL despite the low incidence of GvHD. Our data show that DLI alloreactivity is determined by the conditions at the time of DLI which are influenced by the transplantation strategy. Adjusting the DLI dose based on these conditions may improve the balance between GvHD and GvL.