Comparative Study of the Impact of Human Leukocyte Antigens on Renal Transplant Survival in Andalusia and the United States

Citation: Talaminos Barroso, A.; Reina Tosina, J.; Roa, L.M.; Calvillo Arbizu, J.; Pérez Valdivia, M.A.; Medina, R.; Rocha Castilla, J.L.; Castro-de-la-Nuez, P. Comparative Study of the Impact of Human Leukocyte Antigens on Renal Transplant Survival in Andalusia and the United States. Diagnostics 2023, 13, 608. https://doi.org/10.3390/ diagnostics13040608 Academic Editors: Marijn Speeckaert and Eric Deconinck Received: 8 November 2022 Revised: 23 January 2023 Accepted: 27 January 2023 Published: 7 February 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). diagnostics Article Comparative Study of the Impact of Human Leukocyte Antigens on Renal Transplant Survival in Andalusia and the United States Alejandro Talaminos Barroso 1,* , Javier Reina Tosina 1,2 , Laura M. Roa 1, Jorge Calvillo Arbizu 1,3 , Miguel Angel Pérez Valdivia 4, Rafael Medina 4, Jose Luis Rocha Castilla 4and Pablo Castro-de-la-Nuez 5 1Biomedical Engineering Group, Department of Signal Theory and Communications, University of Seville, 41004 Seville, Spain 2Department of Signal Theory and Communications, University of Seville, 41004 Seville, Spain 3Department of Telematics Engineering, University of Seville, 41004 Seville, Spain 4Urology and Nephrology Department, University Hospital Virgen del Rocío, 41012 Seville, Spain 5Autonomous Regional Transplant Coordination in Andalusia, 41012 Seville, Spain *Correspondence: [email protected] Abstract: Renal transplantation is the treatment of choice for patients suffering from chronic renal disease, one of the leading causes of death worldwide. Among the biological barriers that may increase the risk of acute renal graft rejection is the presence of human leukocyte antigen (HLA) incompatibilities between donor and recipient. This work presents a comparative study of the influence of HLA incompatibilities on renal transplantation survival in the Andalusian (South of Spain) and United States (US) population. The main objective is to analyse the extent to which results about the influence of different factors on renal graft survival can be generalised to different populations. The Kaplan–Meier estimator and the Cox model have been used to identify and quantify the impact on the survival probability of HLA incompatibilities, both in isolation and in conjunction with other factors associated with the donor and recipient. According to the results obtained, HLA incompatibilities considered in isolation have negligible impact on renal survival in the Andalusian population and a moderate impact in the US population. Grouping by HLA score presents some similarities for both populations, while the sum of all HLA scores (aHLA) only has an impact on the US population. Finally, the graft survival probability of the two populations differs when aHLA is considered in conjunction with blood type. The results suggest that the disparities in the renal graft survival probability between the two populations under study are due not only to biological and transplantation-associated factors, but also to social–health factors and ethnic heterogeneity between populations. Keywords: renal transplantation; survival analysis; blood type; human leukocyte antigens; socioeconomic factors 1. Introduction Transplantation is the most appropriate treatment for end-stage renal disease, one of the deadliest diseases in the world. The low availability of organs for transplantation has led to long waiting lists and transplant tourism in some countries. In Europe and the United States (US), legislative, regulatory and humanitarian services are provided to expedite the transplantation process and the appropriate choice between donor and recipient [ 1 ]. However, organ shortages, under-represented minority groups, difficulties in obtaining consent, lack of information, general ethical concerns, religion, problems of organ transportation and other socio-cultural and political factors have led to a lack of organ availability [ 2 ] and are counted among the main non-biological barriers that lengthen the waiting time before transplantation. Diagnostics 2023,13, 608. https://doi.org/10.3390/diagnostics13040608 https://www.mdpi.com/journal/diagnostics Diagnostics 2023,13, 608 2 of 14 The process of admission, surgery, and post-transplant treatment differs between the US and most European countries. In 2019, 60% of transplant recipients in the US had public coverage under the federal Medicare program, while 32% were covered by private insurance [ 3 ]. Medicare coverage ends 3 years after transplantation, except for patients aged 65 or older, or for disability [ 4 ], making it difficult the access to expensive immunosuppressive drugs, which are key to long-term graft survival. In late 2020, US Congress passed the Comprehensive Immunosuppressive Drug Coverage for Renal Transplant Patients Act (Immuno Bill) [ 5 ] to provide lifetime coverage of immunosuppressive drugs for renal transplant recipients. In Europe, most countries offer almost universal publicly funded renal transplant coverage, with the exceptions of Germany and Liechtenstein, where part of the costs are covered by private insurance [6]. Once patients have been added to the waiting list, access to transplantation is determined by the renal allocation scheme. Countries such as the United Kingdom, France and Spain have developed their own national allocation algorithms, while other countries such as Austria, Belgium, Croatia, Germany, Hungary, Luxembourg, the Netherlands and Slovenia are involved in international organ exchange initiatives such as Eurotransplant and Scandiatransplant [ 7 ]. In Spain, the search for compatible deceased donors is carried out using predictive tools that analyse the compatibility between donor and recipient [ 8 ], reducing the availability of organs for some of the patients such as patients with blood group B and highly sensitized patients. Even so, the percentage of renal grafts discarded in Spain is close to 25% in the case of brain death (DBD) and controlled asystole donation, reaching 44% in the case of uncontrolled asystole donation, with significant variability between centres [ 8 ]. In line with other countries, some regions of Spain, such as Andalusia, have also initiated a national program for the distribution of donor grafts in DBD that prioritizes patients with a higher degree of immunization. The pooling of organs and the creation of centralised waiting lists optimise the human leukocyte antigen (HLA) match between donor and recipient, which is generally considered the gold standard for donor–recipient matching [ 9 ]. Additionally, another important barrier in transplantation is blood group compatibility between donor and recipient [ 10 ]. The incidence of blood groups and HLA types varies across different countries [ 11 ], reducing the availability of organs for recipients. There are two HLA molecules [ 12 ]: class I and class II antigens. The HLA-A and HLA-B antigens, belonging to class I, and the HLA-DR antigen, belonging to class II, are generally considered for the study of renal survival. In particular, a higher degree of HLA mismatch is associated with a higher frequency of failure [ 9 ]. The values of these HLA incompatibilities (HLA-A, HLA-B and HLA-DR) are quantified as class 0, 1 and 2, where class 0 represents a high degree of compatibility between donor and recipient, while class 2 indicates a low degree. In Europe, weights are assigned to A, B, and DR incompatibilities, while in the US, DR incompatibility is primarily considered [9]. Studies of this nature have been characterised by the independent analysis of HLA incompatibility and blood group, and to the best of the authors’ knowledge, there is no study that combines them or that analyses the effects of the sum of HLA incompatibilities (aHLA) considering two very different population databases. In this sense, the aim of this work is to analyse the effect of HLA on the probability of renal survival in the American and Andalusian populations, while considering the different HLA-A, HLA-B and HLA-DR incompatibilities in isolation and, subsequently, jointly in different time horizons and together with other factors such as blood group. The work attempts to discuss some of the possible causes of the differences in behaviour between the two populations and the difficulties involved in the analysis of two such heterogeneous populations. In a previous paper [ 13 ], the authors presented a preliminary study limited to the Andalusian population, and now the study is extended to include two distinct populations. The paper is structured as follows: the methodology followed in this study is briefly described in Section 2. Section 3starts with a description of the available data and subsequently presents the results obtained from the Kaplan–Meier estimator and the Cox model, comparing the results with those obtained by other authors. Section 4discusses Diagnostics 2023,13, 608 3 of 14 the outcomes. Finally, the conclusions highlight the differences in the probability of renal survival between the populations studied and the need for exploratory algorithms that combine multiple factors. 2. Methods and Materials 2.1. Methodology In order to carry out this study, the general methodology for survival analysis was followed: 1. Bibliographic search of survival analysis studies that consider blood group and HLA incompatibilities as factors that have an impact on renal transplant survival. 2. Pre-processing of the available databases by performing cleaning and quality improvement process, including standardisation of date-type variables, identification of null values, elimination of unimportant variables, renaming of some variables in order to better describe their content, standardisation of values using the scaling technique and conversion of categorical variables. 3. Analysis of renal graft survival using the Kaplan–Meier estimator considering the different HLA incompatibilities, both in isolation and together as well as in combination with the blood group. 4. Obtaining the coefficients of the Cox model for each of the identified factors with greater significance identified in the previous step. 5. Graphical representation of the survival probabilities obtained and discussion of the results. 2.2. Data Characteristics The renal transplant data that have been analysed in this work come from the Autonomous Transplant Database of Andalusia (SICATA), while the US data have been provided by the United Network for Organ Sharing (UNOS). For both databases, single renal transplants from cadaveric donors between 2006 and 2019 (both inclusive) were selected. The Regional Transplant Coordination Agency in Andalusia started database compilation for renal transplantation in 2006. No data are available for previous years. On the other hand, there has been a continued improvement in long-term success of renal transplantation in the last 10–15 years. Therefore, the situation before 2006 is not significant in light of today’s procedures. Likewise, both databases have been designed to cover the same time interval in order to ensure the greatest possible homogenization between them. Transplants from later dates have been discarded to avoid the possible influence of the COVID-19 pandemic on graft survival probability. The number of transplants for SICATA is 3478, while for UNOS, there are 112,877 transplants. The number of transplant centres for SICATA is 6, while for UNOS, it is 270. The centres with the highest number of patients in SICATA are Virgen del Rocío Hospital (Seville) with 974 transplants, Puerta del Mar Hospital (Cadiz) with 765 transplants and Virgen de las Nieves Hospital (Granada) with 676 transplants. In UNOS, the names of the centres are encrypted. The centre with the highest number of operations reaches 2208 transplants. On the other hand, the average number of days on the waiting list for SICATA is 2.6 years, while for UNOS, it is 3.6 years [14]. The data refer to single renal transplants from donors older than 3 years and recipients older than 18 years, considering only donors and recipients with the same blood type. Table 1presents the main characteristics for both databases, showing the homogeneity between percentages except in the proportion between donors and recipients of groups A and O. These values are the average of the two databases, not allowing for statistical analysis of them. Diagnostics 2023,13, 608 4 of 14 Table 1. Donor and recipient characteristics. Characteristic UNOS (112,877 Cases) SICATA (3478 Cases) Age of donor 38.93 (±15.42) 53.32 (±15.02) Gender of donor (% men) 68,903 (61.04%) 2143 (61.62%) BMI of donor (kg/m2)27.77 (±6.82) 27.4 (±15.02) Age of recipient 52.81 (±13.14) 52.74 (±12.89) Gender of recipient (% men) 68,289 (60.5%) 2205 (63.4%) BMI of recipient (kg/m2)28.11 (±5.4) 27.18 (±7.87) Blood group A–A 20,461 (18.13%) 1624 (46.69%) Blood group B–B 18,570 (16.45%) 335 (9.63%) Blood group O–O 71,386 (63.24%) 1386 (39.85%) Blood group AB–AB 2460 (2.18%) 133 (3.82%) HLA-A: class 0 14,966 (13.26%) 568 (16.33%) HLA-A: class 1 42,965 (38.06%) 1760 (50.6%) HLA-A: class 2 54,946 (48.68%) 1150 (33.06%) HLA-B: class 0 10,216 (9.05%) 247 (7.1%) HLA-B: class 1 27,492 (26.36%) 1140 (40.54%) HLA-B: class 2 75,169 (66.59%) 1821 (52.36%) HLA-DR: class 0 19,658 (17.42%) 647 (18.6%) HLA-DR: class 1 51,656 (45.76%) 1965 (56.5%) HLA-DR: class 2 41,563 (36.82%) 866 (24.9%) In SICATA, there is an equal distribution between these two groups, both for donors and recipients, whereas in UNOS, there is a predominance of group 0. This characteristic does not correspond to the distribution of blood groups in the Spanish and American populations, which have a similar distribution. Another notable dissimilarity between the two populations is in the age of donor, where there is a difference of approximately 14 years. HLA typing from the SICATA database is based on DNA-based methods, specifically RFLP-PCR (restriction fragment length polymorphism polymerase chain reaction) and matching at the antigen level. The serological tests for HLA typing and anti-HLA antibodies are performed before transplantation. Finally, the Regional Transplant Coordination Agency notes that there is excellent monitoring of vital status and type of renal replacement therapy, but they compile no data about post-transplant complications. With regard to the UNOS database, in 2011, the OPTN mandated DNA-based HLA typing methods for deceased donors [ 15 ]. Typical methods used for HLA typing of deceased donors in the US include reverse sequence-specific oligonucleotide (rSSO) probe hybridization, sequence-specific primer (SSP)-directed PCR amplification, and SSP-directed real-time PCR (RT-PCR) amplification [ 16 ]. Due to its simplicity and speed, the RT-PCR method is used as the general approach for HLA typing of deceased donors in most US histocompatibility laboratories [16]. 3. Results Results of the Data Analysis Figure 1shows the survival probabilities for UNOS and SICATA for all data (hereafter referred to as the base curves). The probability curves diverge almost from the first day, with the differences increasing progressively over the entire period. Diagnostics 2023,13, 608 5 of 14 Diagnostics 2023, 13, x FOR PEER REVIEW 5 of 14 Figure 1. Probability of graft survival for UNOS and SICATA. Figure 2 presents the evolution of the two-year renal survival probability at different time intervals for the UNOS database. The progressive improvement in the probability of graft survival is evident since 2006, reaching the highest differences between the period 2006–2010 and 2014–2018. It is expected that this trend will continue in the following years due to the Immuno Bill, mentioned in Section 1. These differences in behaviour between years are not observed for SICATA for the same time intervals. Figure 2. Probability of graft survival in the first two years for UNOS at different time intervals. Figure 1. Probability of graft survival for UNOS and SICATA. Figure 2presents the evolution of the two-year renal survival probability at different time intervals for the UNOS database. The progressive improvement in the probability of graft survival is evident since 2006, reaching the highest differences between the period 2006–2010 and 2014–2018. It is expected that this trend will continue in the following years due to the Immuno Bill, mentioned in Section 1. These differences in behaviour between years are not observed for SICATA for the same time intervals. Diagnostics 2023, 13, x FOR PEER REVIEW 5 of 14 Figure 1. Probability of graft survival for UNOS and SICATA. Figure 2 presents the evolution of the two-year renal survival probability at different time intervals for the UNOS database. The progressive improvement in the probability of graft survival is evident since 2006, reaching the highest differences between the period 2006–2010 and 2014–2018. It is expected that this trend will continue in the following years due to the Immuno Bill, mentioned in Section 1. These differences in behaviour between years are not observed for SICATA for the same time intervals. Figure 2. Probability of graft survival in the first two years for UNOS at different time intervals. Figure 2. Probability of graft survival in the first two years for UNOS at different time intervals. Diagnostics 2023,13, 608 6 of 14 With respect to the levels of HLA incompatibilities (HLA-A, HLA-B and HLA-DR), the differences in the probability of survival for UNOS and SICATA are not very relevant considering the different values in isolation. Figure 3shows the five-year survival probability values for UNOS (a) and SICATA (b), and the evolution of the 10-year survival probability for this incompatibility in UNOS is presented in Figure 4. Diagnostics 2023, 13, x FOR PEER REVIEW 6 of 14 With respect to the levels of HLA incompatibilities (HLA-A, HLA-B and HLA-DR), the differences in the probability of survival for UNOS and SICATA are not very relevant considering the different values in isolation. Figure 3 shows the five-year survival probability values for UNOS (a) and SICATA (b), and the evolution of the 10-year survival probability for this incompatibility in UNOS is presented in Figure 4. Figure 3. Five-year survival probability considering different values of HLA incompatibilities, for (a) UNOS and (b) SICATA. Figure 4. Ten-year survival probability for UNOS considering different HLA-DR mismatch values. Figure 3. Five-year survival probability considering different values of HLA incompatibilities, for (a) UNOS and (b) SICATA. Diagnostics 2023, 13, x FOR PEER REVIEW 6 of 14 With respect to the levels of HLA incompatibilities (HLA-A, HLA-B and HLA-DR), the differences in the probability of survival for UNOS and SICATA are not very relevant considering the different values in isolation. Figure 3 shows the five-year survival probability values for UNOS (a) and SICATA (b), and the evolution of the 10-year survival probability for this incompatibility in UNOS is presented in Figure 4. Figure 3. Five-year survival probability considering different values of HLA incompatibilities, for (a) UNOS and (b) SICATA. Figure 4. Ten-year survival probability for UNOS considering different HLA-DR mismatch values. Figure 4. Ten-year survival probability for UNOS considering different HLA-DR mismatch values. Survival analysis for aHLA (HLA-A + HLA-B + HLA-DR), with values between 0 and 6, which offers differences in behaviour as pointed out by some authors [ 17 , 18 ]. Diagnostics 2023,13, 608 7 of 14 Figure 5shows the distribution of cases by percentage considering the different aHLA values in UNOS and SICATA. The most notable differences between UNOS and SICATA are seen at all levels, where UNOS reaches a higher percentage in incompatibilities 0 and 6 compared to SICATA, while in SICATA, the highest percentages compared to UNOS are in the intermediate values. Diagnostics 2023, 13, x FOR PEER REVIEW 7 of 14 Survival analysis for aHLA (HLA-A + HLA-B + HLA-DR), with values between 0 and 6, which offers differences in behaviour as pointed out by some authors [17,18]. Figure 5 shows the distribution of cases by percentage considering the different aHLA values in UNOS and SICATA. The most notable differences between UNOS and SICATA are seen at all levels, where UNOS reaches a higher percentage in incompatibilities 0 and 6 compared to SICATA, while in SICATA, the highest percentages compared to UNOS are in the intermediate values. Figure 5. Percentage of cases with respect to the total considering the different aHLA values in UNOS and SICATA. The 10-year survival probability for UNOS (a) and SICATA (b) considering aHLA is presented in Figure 6. Given the scarcity of data for SICATA, three groupings have been chosen instead of six, while in UNOS, one grouping is maintained for each aHLA value. Figure 6. Ten-year survival probability considering different aHLA values, where (a) UNOS (p < 0.005) and (b) SICATA (p > 0.1). Figure 5. Percentage of cases with respect to the total considering the different aHLA values in UNOS and SICATA. The 10-year survival probability for UNOS (a) and SICATA (b) considering aHLA is presented in Figure 6. Given the scarcity of data for SICATA, three groupings have been chosen instead of six, while in UNOS, one grouping is maintained for each aHLA value. Figure 6. Ten-year survival probability considering different aHLA values, where ( a ) UNOS (p< 0.005) and (b) SICATA (p> 0.1). The renal survival probability at one, three and five years considering the different aHLA combinations for UNOS and SICATA is shown in Tables 2and 3, respectively. The downward trend in the probability of survival in UNOS is notable for the different levels of aHLA in all years, with the differences at the extreme levels (one and six) becoming more Diagnostics 2023,13, 608 8 of 14 pronounced as the years progress. Individual analysis of the different HLA combinations reveal that transplants with zero HLA-DR mismatches had significantly better survival than those with one or two, with independence of the values of HLA-A and HLA-B mismatches. Table 2. Censored survival probability to renal graft failure at one, three and five years for UNOS from aHLA. aHLA Cases Probability of Survival 1 Year 3 Years 5 Years 1 1789 0.95 0.88 0.81 2 4927 0.94 0.87 0.78 3 14,627 0.94 0.85 0.75 4 16,236 0.93 0.85 0.74 5 36,085 0.93 0.84 0.73 6 17,644 0.92 0.83 0.72 Table 3. Censored survival probability to renal graft failure at one, three and five years for SICATA from aHLA. aHLA Cases Probability of Survival 1 Year 3 Years 5 Years 1 102 0.99 0.98 0.98 2 419 0.94 0.93 0.91 3 862 0.96 0.95 0.92 4 617 0.97 0.91 0.89 5 715 0.95 0.92 0.89 6 186 0.94 0.92 0.9 Combining multiple factors to make population groupings can reveal significant differences that are not obtained by individual factor analysis. In SICATA, blood type has virtually no relevance for survival probability (Figure 7a), nor does aHLA (Figure 6b). However, the combination of both factors reveals changes in behaviour (Figure 7b), with a difference of 7% between the highest (blood group O) and the lowest (blood group B) survival population group, while blood group A transplants have a virtually overlapping survival probability compared to the base curve (all cases). These results have been presented by the authors of the present work in a previous paper [ 13 ]. No such differences were observed in UNOS. Table 4presents the hazard ratios of the most studied factors in the literature extracted from the Cox model considering three-year censored renal graft survival. In SICATA, the most significant variables are the age of the donor and recipient, as reported by other authors for the same data [ 19 ], while in UNOS, in addition to the two previous variables, the sex of the donor and recipient, the BMI of the recipient and HLA-DR incompatibility are also relevant. This multivariate Cox regression analysis shows that HLA-DR incompatibility has the most significant impact on renal survival with respect to HLA-A and HLA-B, as demonstrated by the higher hazard ratio value (1.12 vs. 1.02). Diagnostics 2023,13, 608 9 of 14 Diagnostics 2023, 13, x FOR PEER REVIEW 9 of 14 Figure 7. Five-year survival probability considering by blood group for SICATA, where (a) captures all cases (p > 0.5), and (b) shows only cases where aHLA is three or lower (p < 0.05). Table 4 presents the hazard ratios of the most studied factors in the literature extracted from the Cox model considering three-year censored renal graft survival. In SICATA, the most significant variables are the age of the donor and recipient, as reported by other authors for the same data [19], while in UNOS, in addition to the two previous variables, the sex of the donor and recipient, the BMI of the recipient and HLA-DR incompatibility are also relevant. This multivariate Cox regression analysis shows that HLA-DR incompatibility has the most significant impact on renal survival with respect to HLA-A and HLA-B, as demonstrated by the higher hazard ratio value (1.12 vs. 1.02). Table 4. Cox multivariate proportional hazards analysis of censored graft survival to failure for SICATA and UNOS at three years for all data. Factor SICATA UNOS Hazard Ratio p Hazard Ratio p Age of donor 1.04 (1.02–1.05) <0.005 1.01 (1.01–1.01) <0.005 Gender of donor 0.96 (0.69–1.34) 0.81 1.11 (1.07–1.14) <0.005 BMI of donor 1.01 (0.97–1.04) 0.69 1 (1–1) 0.75 Age of recipient 0.98 (0.96–0.99) <0.005 1.01 (1.01–1.01) <0.005 Gender of recipient 1.47 (1.06–2.02) 0.02 0.88 (0.85–0.91) <0.005 BMI del receptor 1 (0.99–1.02) 0.76 1.01 (1.01–1.01) <0.005 HLA-A 1.1 (0.87–1.39) 0.44 1.02 (1–1.05) 0.09 HLA-B 1.26 (0.93–1.56) 0.1 1.02 (0.99–1.05) 0.2 HLA-DR 1.2 (0.93–1.54) 0.16 1.12 (1.1–1.15) <0.005 Finally, Figure 8 shows the hazard ratio obtained from a univariate Cox model for each of the aHLA levels, taking 0 as the reference value. For SICATA, the value 1 has been omitted due to the scarcity of data for this level. The linear relationship of hazard ratios in UNOS as aHLA increases was obtained by other authors [20], although with older data (up to 2013). This linear trend is also observed for SICATA, with the hazard ratio being higher than in UNOS at all levels, although statistically insignificant (p > 0.05). The error bars are the 95% confidence intervals for each value of the hazard ratio, omitted for SICATA given the magnitude of these. Figure 7. Five-year survival probability considering by blood group for SICATA, where ( a ) captures all cases (p> 0.5), and (b) shows only cases where aHLA is three or lower (p< 0.05). Table 4. Cox multivariate proportional hazards analysis of censored graft survival to failure for SICATA and UNOS at three years for all data. Factor SICATA UNOS Hazard Ratio pHazard Ratio p Age of donor 1.04 (1.02–1.05) <0.005 1.01 (1.01–1.01) <0.005 Gender of donor 0.96 (0.69–1.34) 0.81 1.11 (1.07–1.14) <0.005 BMI of donor 1.01 (0.97–1.04) 0.69 1 (1–1) 0.75 Age of recipient 0.98 (0.96–0.99) <0.005 1.01 (1.01–1.01) <0.005 Gender of recipient 1.47 (1.06–2.02) 0.02 0.88 (0.85–0.91) <0.005 BMI del receptor 1 (0.99–1.02) 0.76 1.01 (1.01–1.01) <0.005 HLA-A 1.1 (0.87–1.39) 0.44 1.02 (1–1.05) 0.09 HLA-B 1.26 (0.93–1.56) 0.1 1.02 (0.99–1.05) 0.2 HLA-DR 1.2 (0.93–1.54) 0.16 1.12 (1.1–1.15) <0.005 Finally, Figure 8shows the hazard ratio obtained from a univariate Cox model for each of the aHLA levels, taking 0 as the reference value. For SICATA, the value 1 has been omitted due to the scarcity of data for this level. The linear relationship of hazard ratios in UNOS as aHLA increases was obtained by other authors [ 20 ], although with older data (up to 2013). This linear trend is also observed for SICATA, with the hazard ratio being higher than in UNOS at all levels, although statistically insignificant (p> 0.05). The error bars are the 95% confidence intervals for each value of the hazard ratio, omitted for SICATA given the magnitude of these.