1 Introduction

The need for kidney transplantations worldwide continues to grow, and the increasing demand leads to the use of more marginal kidneys from extended criteria donors, including donors after circulatory death (DCD). The use of marginal kidneys increases the risk of delayed graft function (DGF) [1, 2]. The rate of DGF in the literature depends on the definition used, and the reported rates vary between 19% and 55% [3-12], with higher rates being reported in extended criteria and DCD donors. Studying the nature and management of DGF is essential as it is associated with higher risks for graft loss [12-15] and acute rejection [11, 14, 15].

DGF is a complex process, typically thought to originate from ischemia-reperfusion injury, causing acute kidney injury. There are multiple processes associated with the occurrence of ischemic-reperfusion injury: cold and warm ischemia time, activation of the renin-angiotensin-aldosterone system, deficiency of the ACE2 enzyme, and formation of inflammation and fibrosis [16-18]. These cellular processes and consequential injuries are also part of processes resulting in acute rejection.

The occurrence of acute rejection is mostly concentrated to the immediate period following transplantation. Since acute rejection is associated with worse outcomes [19], early recognition is important. Differentiating between other causes of DGF and acute rejection is clinically challenging, and biopsies serve as the diagnostic method. The indication for biopsies in the early post-transplant period is most often DGF, but other findings include loss of graft function, recurrence of primary disease, drug toxicity, or thrombotic microangiopathy [7, 20]. Kidney Disease Improving Global Outcomes (KDIGO) guidelines from 2009 suggest taking protocol biopsies in 7–10-day intervals during DGF since DGF is associated with higher risks of acute rejection [20]. This suggestion is, however, based on low evidence, and is based on studies done before the current era of immunosuppressive treatment [21-23].

The optimal timing of early protocol biopsies taken during DGF has not been specified, and the frequency of pathological findings in these early biopsies has not been characterized in detail. Additionally, by considering donor and recipient characteristics, we aim to target sequential biopsies only to risk groups. We study this hypothesis using early indication biopsies taken within 30 days post-transplant of kidneys with DGF at our institution.

2 Methods and Materials

3 Results

3.1 Characteristics of the Study Population

A total of 3577 kidneys were transplanted during the study period to adult recipients. Early graft function and non-function kidneys were excluded (n = 2556 and n = 84, respectively), as well as cases with missing data on graft function (n = 11). Living donors and DCD donors were not part of the study (n = 5 and n = 25, respectively). Pre-emptive transplantations were excluded from the groups mentioned above. Kidneys that did not have biopsies taken within the first 30 days post-transplant were excluded (n = 209). Inadequate biopsies (<3–5 glomeruli) were also excluded (n = 34). A flow diagram of the study population, along with inclusion and exclusion criteria, is shown in Figure 1. The final study population consisted of 1022 biopsies from the 678 recipients.

The median recipient age was 56 years and 69% of the recipients were male, while 58% of the donors were male and the median donor age was 61 years. DSAs were present in 11% of the cases while 26% had peak calculated panel reactive antibodies >30%. Acute rejection during the first 30 days was more common among younger recipients, in patients with DSAs, and retransplantations as well as among patients with tacrolimus as their first calcineurin inhibitor (Table 1). One hundred seventy-eight (26%) patients had acute rejection within the first 30 days, of which 3% were ABMR, 1% mixed, 14% TCMR, and 8% borderline (Table 2). In the study cohort, one biopsy had characteristics of coagulative tubular necrosis, and one biopsy showed frank infarction.

TABLE 1. Study population characteristics by occurrence of acute rejection during the first 30 days.

Variable	Overalla N = 678	No ARa N = 500 (74%)	ARa N = 178 (26%)	p valueb
Recipient age (years)	56 (48, 65)	57 (49, 65)	54 (44, 62)	0.002*
Recipient sex				0.001*
Female	211 (31%)	138 (28%)	73 (41%)
Male	467 (69%)	362 (72%)	105 (59%)
Cold ischemia time (hours)	20.7 (17.0, 23.8)	21.1 (17.4, 24.1)	20.0 (16.0, 23.5)	0.012*
Donor-specific antibodies present	76 (11%)	36 (7%)	40 (23%)	<0.001*
Missing	2	2	0
Peak calculated panel reactive antibodies >30	175 (26%)	109 (22%)	66 (37%)	<0.001*
Missing	3	3	0
Recipient kidney disease				0.909
Diabetes	168 (25%)	120 (24%)	48 (27%)
Glomerulonephritis	174 (26%)	128 (26%)	46 (26%)
Hypertension	32 (5%)	23 (5%)	9 (5%)
Polycystic	126 (18%)	96 (19%)	30 (17%)
Other	178 (26%)	133 (27%)	45 (25%)
Donor age (years)	61 (53, 67)	62 (54, 67)	60 (50, 67)	0.318
Donor sex				0.363
Female	287 (42%)	206 (41%)	81 (46%)
Male	391 (58%)	294 (59%)	97 (55%)
Previous kidney transplant	123 (18%)	77 (15%)	46 (26%)	0.003*
Induction				0.002*
Anti-thymocyte globulin	22 (3%)	11 (2%)	11 (6%)
Basiliximab	120 (18%)	79 (16%)	41 (23%)
Corticosteroids	536 (79%)	410 (82%)	126 (71%)
First calcineurin inhibitor				<0.001*
Cyclosporine A	325 (53%)	281 (62%)	44 (27%)
Tacrolimus	293 (47%)	176 (38%)	117 (73%)
Missing	60	43	17
First antiproliferative				0.468
Azathioprine	4 (1%)	2 (0.4%)	2 (1%)
Mycophenolate	613 (99%)	454 (99%)	159 (99%)
Missing	61	44	17
HLA AB mismatch	2 (1, 3)	2 (1, 3)	2 (1, 3)	0.5
HLA DR mismatch	1 (0, 1)	1 (0, 1)	1 (1, 1)	<0.001*

• Abbreviations: AR, acute rejection; HLA, human leukocyte antigens.
• ^a n (%); median (25%, 75%).
• ^b Wilcoxon rank sum test for continuous; Pearson's Chi-squared test for categorical variables, Fisher's exact test.
• * p value statistically significant, <0.05.

TABLE 2. Biopsy diagnosis and histological lesion scores in first protocol biopsy, excluding acute tubular injury.

Variable	Overall N = 678 (100%)1
Biopsy diagnoses
No rejection	500 (74%)
T-cell mediated rejection	94 (14%)
Antibody-mediated rejection	23 (3%)
Borderline changes	57 (8%)
Mixed rejection	4 (1%)
Histological lesion scores
Interstitial inflammation 1;2;3	219 (33%); 73 (11%), 36 (5%)2
Tubulitis 1;2;3	80 (12%); 24 (4%); 5 (1%)3
Intimal arteritis 1;2;3	53 (8%); 30 (5%); 1 (0%)4
Glomerulitis 1;2;3	34 (5%); 17 (3%); 14 (2%)5
Peritubular capillaritis 1;2;3	34 (8%); 15 (4%); 2 (0%)6
C4d staining 1;2;3	32 (5%); 20 (3%); 11 (2%)7

• ¹n (%)missing n; ²18,³9,⁴15,⁵20,⁶269,⁷41.

Two patients had missing data on the presence of DSA at the time of transplantation, and three patients had missing data on peak calculated panel reactive antibodies. Three biopsies had missing information on the number of glomeruli. Missing data on histological lesion scores are found in the footnote of Table 2. Patients with missing data were excluded from relevant analyses.

3.2 Timeline of Biopsy Diagnosis

Figure 2 presents the number of sequential biopsies taken. Of the studied patients, 44% had two biopsies taken within the first month. The mean time between biopsies was 10 ± 3.7 days between the first and second biopsy and 9 ± 2.8 days between the second and third. Figure 3 presents the distribution of the biopsy diagnoses based on the post-transplant day. Rejection was most often diagnosed at 7–10 days post-transplant. Figure S1 presents the distribution of all rejection diagnoses based on post-transplant day. Figure 4a,b present the distribution of biopsy and rejection diagnoses based on the presence of DSAs. ABMR was more frequent when DSAs were present (20% vs. 1.3%, p < 0.001).

Sensitivity analysis was performed due to missing data on the lesion score PTC. The incidence of ABMR increased after 2012, when PTC scoring was systematically introduced at our center (Table S1 and Figure S2). Independent of the availability of PTC, rejection was most often diagnosed in early biopsies (Figure S3a,b).

In cases where more than one biopsy was taken post-transplantation, the findings in each sequential biopsy are visualized using a Sankey diagram (Figure 5). The diagram shows that in kidneys with no rejection in the first biopsy, subsequent biopsies revealed rejection in 10% of cases. Among 183 kidneys, the second biopsies found 3 cases of ABMR, 10 of TCMR, 5 borderline, and 165 without acute rejection. Additionally, borderline findings in the initial biopsy progressed to rejection in approximately one-third of the cases, despite treatment with steroids. Of the 42 third biopsies taken, two of the rejections were in kidneys with no previous rejection findings. Table S2 presents characteristics of cases with rejection in the second biopsy. Figure S4a,b presents similar results as Figure 5 regarding the evolution in protocol biopsies, based on the availability of PTC.

To further investigate the patients who were diagnosed with acute rejection after having no rejection in the first biopsy, we evaluated known risk factors for acute rejection. As there were so few cases of rejection, we studied the risks using univariable logistic regression models. We found no statistically significant risk factors for rejection in these cases (Table S3).

4 Discussion

Our study showcased the practice of taking biopsies of DGF kidneys at 7–10 days post-transplant, which consequently led to the finding of most rejections at this point. This is consistent with previous studies [8, 25]. Through studying sequential biopsies, we also found that in cases where additional biopsies were performed after the first one taken during DGF, rejection seldom occurred in later biopsies within the first month. This suggests that further biopsies may be less important when rejection is absent in the initial biopsy. Moreover, in the majority of cases, no further biopsies were performed after the first one. We could not identify any specific risk factors for acute rejection in later biopsies if the first biopsy did not have rejection. We did, however, find that the pre-existing DSAs are associated with glomerulitis and that ABMR is more prevalent in these kidneys, which is in line with earlier studies [12, 26].

In the present study, most acute rejection episodes occurred in the first biopsy post-transplant. The significance of the first biopsy during DGF has been previously highlighted by a smaller study [8], as well as a recent review [27]. One study, on the other hand, argued that early protocol biopsies do not necessarily offer clinically significant information in low-immunological risk recipients with induction therapy, and suggested postponing the rejection diagnostics until post-operative Day 15 [7]. Another argument for minimizing the number of biopsies taken by protocol during DGF is the complications associated with biopsies. However, the rate of these is considered quite small, and routine biopsies are a safe procedure [28]. A study about the histology of early indication biopsies found that inflammation in the biopsy is associated with alloimmune risk factors but not with nonimmune risk factors. Furthermore, they did not find a correlation between the inflammation in the biopsies and DGF [12]. The identification of risk factors requires further research.

The immunosuppressive regimens in the current era are associated with lower rates of acute rejection [29], and for example, the rates of acute rejection in the United States have decreased steadily during the last decade [3]. Even though ABMR is less common than TCMR in the acute phase, ABMR has been found to be the most common cause of late graft failure [30]. One study explored the histologic differences in DGF kidneys with different induction therapies, and while thymoglobulin was found to be preventative of inflammation compared to basiliximab, no differences in survival were found [31]. Our study observed a higher percentage of acute rejection with both anti-thymocyte globulin and basiliximab compared to corticosteroids alone. It is important to note that in Finland, induction with anti-thymocyte globulin and basiliximab has primarily been used for high-risk patients, which probably explains the association with a higher risk of rejection. Induction practices vary globally and have evolved over time. Previous research has suggested that induction with anti-thymocyte globulin may reduce the risk of acute rejection in the early post-transplant period [7, 32], but is associated with a higher risk of infections [33].

The selection of calcineurin inhibitors during the peri-transplant period requires balancing the risk of graft failure and rejection against the nephrotoxicity associated with these drugs. One article found increased rates of acute rejection with lower calcineurin inhibitor exposure [34]. A meta-analysis evaluating different strategies for calcineurin inhibitor use concluded that reduced exposure was effective in improving renal function without increasing the risk of rejection [35]. The studies referenced by KDIGO in their guidelines on protocol biopsies had variable immunosuppressive protocols. Shapiro et al. studied rejection rates using a protocol consisting of tacrolimus and steroids. Additionally, 64% received mycophenolate as induction. The rate of acute rejection was 46% in the early graft function cohort and 57% in the DGF group [21]. Giral-Classe et al. studied the rate and longevity of DGF using 843 kidney transplants with Cyclosporin A as the calcineurin inhibitor. Mycophenolate was used in less than 10% of cases. The observed rejection rates were 29%–37.4% with higher rates seen with longer DGF duration [22]. Qureshi et al. studied 410 kidney transplantations for the rate of silent acute rejection. Their immunosuppressive protocol changed during the study period, but Cyclosporin A was used in both protocols; 75.6% of the patients received mycophenolate. The observed rejection rate was 50.8% in DGF kidneys [23]. In Finland, tacrolimus has been used in high-risk patients earlier as the first calcineurin inhibitor as opposed to cyclosporin A. Since tacrolimus became the calcineurin of choice for most patients, the observed rejection rates have paradoxically slightly risen in DGF kidneys from brain-dead donors. This trend may be attributed to a decrease in other causes of DGF, such as prolonged CIT, rather than an actual increase in rejection rates. The overall number of DGF cases has decreased over time as CIT has been reduced, leading to a higher relative proportion of rejection-related DGF cases. Over the last decade, our center has seen a dramatic decrease in the incidence of DGF in kidneys from brain-dead donors, partly as a result of successful work to reduce cold ischemia times [36]. New ways to meet the demand for kidney transplants have led to an increase in DCD and extended criteria donations worldwide. DCD is associated with a high incidence of DGF, highlighting the importance of studying DGF, despite the decrease in DGF rates observed with brain-dead donations. In the United States, the rate of DGF is increasing [3], which is probably at least partly related to the increasing rate of DCD and extended criteria donors. We chose not to include DCD donations in our study, since the DCD program started in Finland only a few years ago, and kidneys from brain-dead donors still compose most kidney transplantations at our institution.

Understanding DGF requires knowledge of multiple aspects of this process. A study about the proteomic profile of DGF kidneys in DCD donations found that there is a difference between the proteomic profiles in short versus long DGF duration as well as comparing these two groups to early graft function kidneys [37]. They showed that the molecular and metabolic features make the kidneys less able to cope with the stress reaction caused by the ischemia-reperfusion injury, which prolongs the time for the kidneys to recover post-transplant. Different biomarkers have been studied as predictors for slow graft function and DGF. For example, donor neutrophil gelatinase-associated lipocalin has been studied as a predictor and in some studies shown some predictive power [38, 39]. Because of the complexity, there is no one solution for minimizing the risk of DGF or its associated complications. Even though the rates of acute rejection are currently lower owing to effective regimens of immunosuppression and the availability of early detection of rejection through follow-up laboratory tests and biopsies, we know that DGF is associated with poorer long-term graft outcomes [15, 40] and may also be associated with poorer patient outcomes [14]. Continued research regarding the topic and a striving to understand the mechanisms will therefore aid in the development of clinical guidelines.

There are many strengths in our study; however, there are also limitations to consider. First, the study is retrospective, and thus, the causality of the findings cannot be proved. Our study population is quite large, has an acceptable amount of missing data, and the findings are in line with earlier research. These factors improve the power of our study and strengthen the conclusions. Second, we chose to use the lesion scores from the time of the biopsies and not score them again using the current Banff classification guidelines. This could lead to discrepancies in the results, as the rejection criteria have been updated continuously throughout the study period, especially concerning ABMR [26, 41]. A Spanish study compared the changes in ABMR diagnoses from the Banff 2009 criteria to the Banff 2013 criteria on 68 kidney transplant recipients. They found a change in diagnostic category for 18% of the studied biopsies and concluded that the ABMR diagnosis in the 2013 update reduced the amount of ambiguous ABMR diagnoses [42]. Callemeyn et al. studied the prognostic value of changes to the ABMR criteria in the 2013 and 2017 Banff updates. They found that the 2013 revision improved predictive accuracy, while the 2017 update removed the ABMR diagnosis from kidneys with worse graft survival, that previously were diagnosed as suspicious for ABMR [43]. Furthermore, in the following Banff updates, Notable is, however, that the diagnostic criteria for acute TCMR have not changed significantly. The criteria for borderline findings have been updated multiple times, which has led to heterogeneous descriptions of borderline findings in the literature [44]. In the 2019 updated, the criteria for interstitial inflammation in borderline was raised from i0 to i1 [45]. Based on our data, no conclusions about the optimal cut-offs for borderline changes can be made. The immunosuppression in our cohort is relatively mild, and the findings cannot be directly generalized to other cohorts, in which, for example, lymphocyte-depleting induction is routinely used. In sensitivity analysis regarding the missing data concerning the lesion score PTC, we observed a slight increase in the incidence of ABMR, but not a change in the timeline of rejection diagnosis. Last, different pathologists have analyzed the biopsies over the years, and it is well known that interobserver agreement is far from optimal in kidney transplant pathology [46]. On the other hand, our analyses describe the clinical practice and usefulness of the biopsies at the time of biopsy, not retrospectively.

In conclusion, the first biopsy 7–10 days post-transplant in DGF kidneys is important in identifying early acute rejection. If the first biopsy shows no acute rejection, the risk of finding acute rejection in sequential biopsies is low. These data reflect the rate of early TCMR in deceased donor kidney transplants when 79% of patients did not receive an anti-T cell induction antibody.

Author Contributions

Amanda Ahlmark contributed to the data curation, formal analysis, funding acquisition, investigation, methodology, project administration, visualization, resources, writing -original draft, and writing review & editing. Ville Sallinen contributed to conceptualization, project administration, resources, supervision, validation, and writing – review & editing. Ilkka Helanterä contributed to conceptualization, funding acquisition, methodology, project administration, resources, supervision, validation, and writing – review & editing. Jouni Lauronen contributed to data curation and writing – review & editing. Kaisa Ahopelto contributed to writing – review & editing. Marko Lempinen contributed to conceptualization and writing & editing. Anne Räisänen-Sokolowski contributed to writing – review & editing.

Conflicts of Interest

I.H. holds research grants from Hansa Biopharma and MSD and has received consultancy fees from AstraZeneca, Hansa Biopharma, Takeda, and MSD. J.L. has received consultancy fees from Hansa Biopharma. K.A. has received consulting fees from Hansa Biopharma and Astellas. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.