Fig 1.
During the bootstrapping process, virtual patients and donors were created based on haplotype frequencies. The initial waiting list was randomly populated matching the distributions provided by Eurotransplant. During the simulation, waiting patients were randomly removed from the waiting list (e.g. unfit for transplantation, died waiting) and new patients were randomly registered to the waiting list. Virtual donors were randomly registered and allocated to the current waiting list resulting in simulated transplantations.
Fig 2.
The basic ETKAS allocation consists of filtering steps and a cascade of sorting criteria. The dashed boxes refer to components involving HLA compatibility. These components were part of the altered allocation simulations. Other factors were unaltered.
Table 1.
Description of the allocation simulation models.
Fig 3.
(A) Graphical representation of the distribution of the 400 points for match grade in the PIRCHE-modified models ETKASPIR-B, -C, -D, -E, and -F. (B) Distribution of ETKAS points for HLA MMP on the bootstrapped waiting lists (red curve) shows a benefit for relatively few patients with rare genotypes. Unconverted, the PIRCHE-II risk profile (capped at PIRCHE-II weighted median of 200) assigns more points to average genotypes (blue curve), requiring fitting a polynomial (green curve) to match the regular ETKAS profile.
Fig 4.
Comparison of the bootstrapping results (blue dots) versus the data reported by ET in the years 2013–1017 (red dots). Waiting time was calculated from the start of dialysis. Significant differences (p < 0.01) are indicated with a "*".
Fig 5.
Simulation stability evaluation.
(A) HLA match grade of the ETKAS simulations, year 10 compared to the 1998 observations: JSD = 0.069, p<0.001, year 10 compared to 2015 ETKAS observations: JSD = 0.135, p<0.001. (B) Observed HLA match grade reported by ET annual reports, only ETKAS. (C) Course of the distribution of the countries of origin of virtual individuals. (D) Course of country balance depending on country and simulated allocation models. (E) Number of transplants per country remained stable between the simulated allocations. (F) distribution of recipient age in the individual PIRCHE groups of different simulation allocations.
Fig 6.
PIRCHE-II score distributions in simulations.
The log-transformed PIRCHE-II score distributions resulting from the different ETKAS simulation models. Density plots for the simulations were generated using a Gaussian smoothing kernel. Red lines in graphs A-C represent the PIRCHE-II scores in the ETKAS baseline simulations, the green line in graph A represents the previously reported PIRCHE-II scores as observed in the PROCARE cohort[20], blue lines in graphs B-C represent the PIRCHE-II scores in the PIRCHE-II modified ETKAS simulations for the ETKASPIR-E with priority by HLA match, assignment of points linearly descending, and inclusion of a PIRCHE-II based mismatch probability (MMP) (B), and ETKASPIR-F with priority allocation by PIRCHE-II score, assignment of points linearly descending, and inclusion of a PIRCHE-based MMP (C). Dashed lines indicate the previously reported strata by Lachmann et al. [19].
Table 2.
PIRCHE-II distribution per risk strata in percent.
Fig 7.
(A) After the simulation stabilized, the amount of cross-border donor exchange remained similar between the simulated ETKAS and ETKASPIR-E. Outliers in the burn-in period were suppressed for readability. (B) Average waiting times of transplanted patients was slightly lower in ETKASPIR-E, (C) whereas average waiting times of patients on the waiting list increased mildly. (D) number of match points given by match compatibility in the different simulations. Boxplots depict the median (horizontal line), mean (plus) and first to third quartile (box), the highest and lowest value within 1.5x IQR (whiskers), outliers (circles).
Fig 8.
(A) In ETKASPIR-F (green), the proportion of simulated transplants with a full HLA match reduces when compared to ETKAS (red). ETKASPIR-E (blue) compensates for that by maintaining the current ETKAS priority for HLA full matches. (B) PIRCHE-II scores are reduced in every HLA match group. Boxplots depict the median (horizontal line), mean (plus) and first to third quartile (box), the highest and lowest value within 1.5x IQR (whiskers), outliers (circles).