CTS Newsletter 2:2021: Epitope versus traditional HLA matching

CTS analyses continue to underline the importance of HLA matching in deceased-donor kidney transplantation. In kidney allocation systems, recipients and donors are traditionally matched based on alleles from three different HLA loci, namely HLA-A, -B, and -DRB₁, at broad or split antigen level. During the rejection process, however, the recipient's immune system does not recognize the whole sequence of the foreign HLA allele molecule but rather small parts of it, called epitopes. Moreover, the same antigenic epitope can be present on different HLA alleles. Matching of HLA antigens at epitope instead of allele level has therefore the potential to further improve the outcome in kidney transplantation.

Several theoretical and experiment-based algorithms have been developed for epitope matching. The Predicted Indirectly ReCognizable HLA Epitopes (PIRCHE-II) algorithm is one of these theoretical approaches which allows calculation of an 'epitope load' score based on donor HLA epitopes that are not present on the recipient's own HLA alleles but can be presented by his/her HLA class II molecules to his/her immune system. In a recent CTS publication, we analyzed whether an epitope-based matching of donor-recipient pairs using the PIRCHE-II algorithm is superior to the currently applied traditional HLA allele matching (doi.org/10.3389/fimmu.2021.631246). More than 65,000 first deceased-donor kidney transplantations performed during 1990 – 2016 in Europe were analyzed. To simulate currently used allocation algorithms, typing information at split antigen level for the HLA loci A, B, and DRB₁ was used for calculation of both, PIRCHE II scores as well as HLA mismatches.

**Figure 1.** Correlation between HLA A+B+DRB₁ mismatches and PIRCHE-II scores in deceased-donor kidney transplantations performed during 2000 – 2016.

For this Newsletter, we concentrated on the more recent 2000 – 2016 period. As illustrated in Figure 1, there is a significant correlation between the number of HLA A+B+DRB₁ mismatches and PIRCHE-II scores (Spearman’s rank correlation coefficient 0.63; P < 0.001). Such a strong correlation is not surprising because both parameters measure the same characteristic, namely the grade of HLA incompatibility between the donor and the recipient.

The PIRCHE-II score has the capacity to be more precise than the traditional HLA antigen matching, solely due to the fact that it involves a higher range of values (0 – 211 in our study) than the possible seven mismatches (0 – 6) in the traditional HLA-A, -B, -DRB₁ matching. For a fair evaluation of superiority, we adjusted the PIRCHE-II scores to the 0 – 6 range of HLA mismatches and compared the influence of both parameters on outcome in three different Cox models. Figure 2 demonstrates that, if analyzed separately, both parameters are risk factors for overall graft failure, death-censored graft failure as well as patient death with significantly increased hazard ratios (HRs). When analyzed in one model simultaneously, the overall effect is split between the two confounders because they partly share the same information. Therefore, the magnitude of their influence decreases slightly; it remains, however, with the exception of their impact on patient mortality, statistically highly significant.

**Figure 2.** Impact of matching for PIRCHE-II score and HLA A+B+DRB₁ mismatches on 5-year (A) overall graft failure, (B) death-censored graft failure, and (C) patient death. Multivariable Cox regression analyses were performed in which the two parameters were evaluated either separately in the absence of the other parameter or simultaneously in one model. P-values and hazard ratios with 95% confidence intervals are shown.

For a more precise evaluation of superiority, we also studied the additional influence of PIRCHE-II scores and HLA mismatches in different categories of the other parameter. As illustrated in Figure 1, the variance in PIRCHE-II score increases with the higher number of HLA mismatches. Among recipients with 4 – 6 HLA mismatches, there are subgroups of patients who are confronted with a relatively low or a relatively high PIRCHE-II score, i.e. epitope load. In contrast, the variance in PIRCHE-II is extremely low in recipients of grafts with 0 HLA mismatches. Figure 3A demonstrates that the influence of PIRCHE-II score on death-censored graft failure is almost double as high in the presence of high (4 – 6) numbers of HLA mismatches as in the presence of low (0 – 3) mismatch numbers (HR = 1.113 and 1.059; P = 0.004 and 0.015). HLA mismatches also increase the risk of graft loss additionally, however, only if the PIRCHE-II score is above 25, whereas their impact does not reach statistical significance in the presence of a low PIRCHE-II score of 0 – 25 (HR = 1.070 and 1.015; P < 0.001 and = 0.57; Figure 3B). These findings altogether indicate that both parameters deliver complementary information regarding the outcome, especially in the higher score categories of the other parameter.

**Figure 3.** Impact of adjusted PIRCHE-II scores and HLA mismatches on 5-year death-censored graft failure in patients with (A) low (0 – 3; n = 16,093) or high number (4 – 6; n = 20,836) of HLA mismatches and (B) low (0 – 25; n = 16,694) or high (> 25; n = 35,111) PIRCHE-II scores. Hazard ratios with 95% confidence intervals are shown. HLA mismatches and adjusted PIRCHE-II scores were analyzed simultaneously in a multivariable Cox regression model.

Figure 4 demonstrates that the slightly higher benefit of PIRCHE-II score over HLA mismatches is only present during the first post-transplant year, irrespective of whether these two parameters were analyzed separately in the absence of the other parameter or simultaneously in one model.

**Figure 4.** Impact of adjusted PIRCHE-II scores and HLA mismatches on death-censored kidney graft failure during the (A) first year and (B) years 2 – 5 post-transplant. Multivariable Cox regression analyses were performed in which the two parameters were evaluated either separately in the absence of the other parameter or simultaneously in one model. P-values and hazard ratios with 95% confidence intervals are shown.

In the subgroup of pre-sensitized patients with cytotoxic panel reactivity (PRA), the difference between PIRCHE-II scores and HLA mismatches is stronger (HR = 1.153 vs. 1.104; P < 0.001 for both) than in the larger group of patients without PRA (HR = 1.131 vs. 1.101; P < 0.001 for both), when the influence on outcome is analyzed separately. The same trend is also visible in the simultaneous consideration of both parameters (Figure 5).

**Figure 5.** The impact of adjusted PIRCHE-II scores and HLA mismatches on 5-year death-censored graft failure in subgroups of kidney transplant recipients (A) with and (B) without pre-transplant panel reactive cytotoxic antibodies (PRA). Multivariable Cox regression analyses were performed in which the two parameters were evaluated either separately in the absence of the other parameter or simultaneously in one model. P-values and hazard ratios with 95% confidence intervals are shown.

Every additional effort contributing to a better compatibility between donor and recipient has the potential to improve the outcome further. While our findings suggest that PIRCHE-II score cannot fully replace the traditional HLA matching, it could be a decent tool to decide whom to give the organ preferentially if there were several potential recipients with the same number of HLA mismatches. Such a combinatory approach could especially be useful in donor-recipient constellations with a high grade of HLA incompatibility; in the current analysis, recipients with 4 – 6 mismatches made up as much as 40% of the studied cohort. Pre-sensitized recipients with a generally higher alloreactivity also appear to profit from an additional PIRCHE-II matching. Whether HLA or epitope matching covering additional HLA loci typed at a higher resolution level can fully replace the traditional HLA matching must be addressed in further, albeit costly studies of large cohorts.

I would like to thank those of you who supplied us with pre- and post-transplant Covid-19 information by filling out the 'Covid-19 Information Form' (https://www.ctstransplant.org/downloads/centerarea/Covid-19.pdf), participating in the Covid-19 serum studies, and informing us on hospitalizations for SARS-CoV-2 in the regular CTS follow-up forms. More than 2,000 patients with post-transplant Covid-19 disease have already been registered in the CTS database and we will soon be able to perform the first outcome analysis using this information. Please continue informing us and check whether the list of Covid 19 cases of your center, which you will soon receive in a separate mail, is complete.

Newsletter 2:2021

May 1, 2021