DGF is a common complication after kidney transplantation operations and is related to both short-term functional recovery and long-term survival of the transplanted kidney. Yarlagadda et al. [17] systematically reviewed the definitions of DGF, concluding that the combination of Scr reduction and dialysis needs constituted a reasonable definition. However, most centers have recently reached a consensus that the definition of DGF is the need for dialysis within the first week after transplantation [10-12]. As our study adopted the most common definition, DGF cases were relatively few. Consistent with the situation in America, the number of candidates waiting for kidney transplantation is increasing annually in China. Given the strain on kidney resources, the inclusion of ECD has been recognized internationally, even though the utilization of ECD is associated with increased cost and DGF [10]. A systematic review by Tingle et al. confirmed that machine perfusion reduced the incidence of DGF [18]; Patel et al. [19] also demonstrated that HMP can improve the utilization of kidneys. Thus, we used LifePort for marginal kidney perfusion to address such risk as much as possible. Despite the benefits of reducing the resistance parameter after perfusion, perfusion prolonged CIT; we chose 2 hours as the perfusion time based on the research results of Patel et al. [18]. We discarded some bad kidneys according to the parameters of LifePort. Although the included marginal kidneys perfused by LifePort showed statistical correlation with DGF, we still used the kidneys that might have been discarded, accompanied by good 1-year graft survival.
Risk factors related to DGF can be divided into donor factors and recipient factors, and the multiple interactions of these factors ultimately affect the recovery of graft function. Numerous studies have assessed the causes of DGF, yet there is no consensus to date. In our study, prolonged CIT and WIT were most strongly associated with DGF, consistent with previous reports [20-27]. Additionally, a longer duration of recipient pretransplant dialysis was likely to lead to DGF. Other donor predictive factors included primary cause of death and terminal Scr. KDPI was introduced in America to indicate the quality of a kidney based on the data from OPTN. Zens et al. [14] also showed that a higher KDPI increased the rate of DGF for kidney recipients, and resulted in shorter graft survival. Nonetheless, other centers have suggested that a high KDPI is not a reason for rejecting a kidney because it does not result in a long-term mortality risk [15,16]. In addition, a previous publication showed that the KDPI could not accurately predict pediatric donor kidney survival [28]. In our study, KDPI did not correlate with DGF and 1-year graft survival. The KDPI system does not include WIT and CIT, which may be one of the reasons for the difference. Additionally, the weight of each factor is fixed in the system; when the sample changes and decreases, accuracy also decreases.
Although immunity induction is an important step before surgery to avoid acute rejection, the use of ATG induction remains controversial. ATG may induce cytomegalovirus infections and hematological complications [29,30]. Popat et al. [31] reported a lower DGF rate in the ATG-induced group among 45 patients in a single-center study. However, ATG induction did not reduce the risk of DGF in the research of de Sandes-Freitaseta [32]. In our study, the use of ATG depended on the patient’s economic condition and the surgeon’s preferences because it is costly, and this use was therefore not predictive. The results of our study showed that donor factors were the main influencing factors of DGF, likely because graft quality strongly affects renal function after transplantation. Terminal Scr is the most direct indicator of kidney quality. It is generally believed that the lower is the value, the lower is the incidence of DGF, as observed by Helfer [27] and in the present study.
Although long-term graft survival is expected, numerous complicated factors can cause graft loss. The relationship between DGF and deceased graft survival has been demonstrated recently [33-34]. In a 3-year DCD kidney registry analysis, Lim et al [34] reported that the recipients of DCD kidneys with DGF experienced a higher incidence of acute rejection and overall graft loss. Gill et al. [33] observed that the DGF-associated risk of graft failure was greatest in the first posttransplant year, and a meta-analysis by Yarlagadda et al. [35] verified the association between DGF, acute rejection, and graft survival. The present study performed patient follow-up in the training cohort for one year, with findings consistent with those reported by Gill et al.
Our nomogram is a simple and visual prediction model of posttransplant factors. Maier et al. investigated the relationship between DGF and posttransplant indicators of neutrophil gelatinase-associated lipocalin (NGAL), reporting that early assessment of serum and urinary NGAL could predict DGF [36]. In their retrospective cohort study, Cardinal et al. [37] used multivariate analysis to examine predictors of DGF but did not distinguish the importance of each. Irish et al. [3,38] combined numerous donor and recipient factors, and applied nomogram scoring systems for predicting DGF in DD kidney transplantation, which were verified by ROC curves. Previous scoring systems are valuable; on this basis, we incorporated a certain percentage of marginal kidneys, especially young kidneys. Overall, young kidney transplantation was effective and safe, indicating a promising expansion of the donor pool. Infant donors younger than 5 months were excluded from this study, because our center used the novel method of en bloc kidney transplantation introduced by Dai et al. [39]. On the other hand, we investigated new indicators, including KDPI, LifePort and HCV history. The evaluation and therapeutic effects of LifePort on kidneys are worthy of affirmation, which is consistent with previous findings [40].
The present study applied a 10-fold cross-validation LASSO method to divide the data into 10 equal parts, with nine parts for the model and one part for validation. This process was repeated ten times to produce an accurate AUC.
In addition to identifying patients at higher risk of DGF before surgery, our model may also be used as a clinical tool to reduce the risk of DGF. CIT should be controlled when possible; for example, shortening the harvest and patient preoperative preparation times would reduce CIT, which would help to decrease the incidence of DGF. More specifically, the model can be used as a strategy to select suitable donors and recipients by identifying reasonable matches between recipients’ conditions and CIT. Additionally, the model can guide immunosuppression induction for high-risk DGF donors identified by the nomogram .
This study had some limitations. First, because this was a single-center study, the sample size was small and data were variable since. Second, the follow-up period of graft function was not long compared to 10–20 years. Finally, we did not provide solutions for predictors such as WIT. Future studies are needed to explore methods for shortening WIT and to investigate the factors influencing graft survival, as prolongation of graft survival is our ultimate aim.