Hypothermic Perfusion Mitigates Extended Cold Ischemia Time in Kidney Transplantation

Visual Abstract

Abstract

Key Points

• Compared with ≤ 12 hours of static cold storage, hypothermic machine perfusion up to 36 hours did not increase death-censored graft failure.

• Hypothermic machine perfusion reduced delayed graft function for kidneys with ≤36 hours of cold ischemia time.

• When cold ischemia time exceeded 36 hours, hypothermic machine perfusion no longer affected outcomes.

Background

Extended cold ischemia time (CIT) is associated with higher incidences of delayed graft function (DGF) and graft failure. Kidneys on hypothermic machine perfusion (HMP) perform better than comparable kidneys in static cold storage (SCS). This study aimed to determine whether hypothermic perfusion was associated with lowering the deleterious effects of extended CIT.

Methods

Data from the United Network of Organ Sharing database were examined to identify adult first-time, kidney-only recipients from 2005 to 2022. Multivariable models for death-censored graft failure (dcGF), uncensored graft failure, DGF, and patient mortality were constructed and adjusted for donor, recipient, and transplant factors. Kidneys with ≤12 hours SCS, an accepted standard for optimal transplantation, served as the reference group for each model.

Results

Among 120,438 allografts, 63% were in SCS and 37% were preserved with hypothermic perfusion. DcGF was higher in a dose-dependent fashion as CIT increased. Kidneys on HMP for ≤24 hours had less dcGF than reference (≤12 hours in SCS) kidneys. Perfusion for 24–36 hours did not differ from reference. After 36 hours, all kidneys had higher dcGF, regardless of the storage method. HMP lowered the incidence of DGF at every level of CIT.

Conclusions

HMP is an effective strategy for lowering the negative effect of prolonged CIT, providing transplant teams with greater flexibility to optimize donor and recipient logistics, without compromising long-term graft outcomes.

Introduction

Extended cold ischemia time (CIT) is associated with higher incidences of delayed graft function (DGF), primary nonfunction, and graft failure.^1–4 Most studies have reported that CIT beyond 15–24 hours has detrimental effects on outcomes.^3,4 Marginal kidneys, such as those with a kidney donor profile index (KDPI) >85% or from donors after circulatory death (DCD), are particularly sensitive to cold ischemia, and prolonged CIT often occurs due to delays in acceptance and placement.⁵

The circulation of a cold preservation solution during hypothermic perfusion preserves endothelial and parenchymal integrity, reduces vasospasm, and removes waste products. Reperfusion injury is reduced, which is associated with significantly less DGF and better graft survival.^6,7 Kidneys on hypothermic machine perfusion (HMP) perform better than similar kidneys preserved with static cold storage (SCS), including high KDPI and DCD kidneys.^8,9 HMP has contributed to lowering discard rate of marginal kidneys.¹⁰

In 2021, changes to the kidney allocation system aimed to reduce geographical disparities by prioritizing deceased donor kidney offers within a 250 nautical mile radius of the donor hospital. An unintended consequence of this policy was higher CIT for recipients outside of the primary radius.¹¹ During the corona virus disease of 2019 pandemic, travel restrictions, reallocation of health care resources, and limitations on transplanting infected patients (even when asymptomatic) all contributed to prolonged cold ischemia.¹²

The safe upper limit of CIT for HMP kidneys has not been established. In simultaneous liver-kidney transplants, hypothermic perfusion enables kidney implantation to be delayed until the recipient's hemodynamics stabilize, in which case CIT may exceed 48 hours. However, the kidneys used in these multiorgan transplants are often of higher quality compared with kidney-only allografts.¹³

The balance between the benefits of HMP and the disadvantage of adding CIT versus proceeding with SCS and less CIT remains unclear, and the safe upper limit of CIT for kidneys on HMP has not been established. Understanding the implications of prolonged CIT is essential for optimizing decisions during kidney offer evaluations.

We analyzed adult recipients of deceased donor kidney transplants between 2005 and 2022 who received standard immunosuppression, to compare graft and patient survival between kidneys preserved with HMP versus SCS, taking CIT into account.

Methods

We used United Network for Organ Sharing (UNOS) files to identify first-time, adult transplant recipients between January 1, 2005, and December 31, 2022. The research protocol was approved by the Institutional Review Board of the University of Chicago with a waiver of informed consent. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the “Declaration of Istanbul on Organ Trafficking and transplant Tourism.”

Analysis was restricted to patients who received a single deceased-donor kidney without any other organs and were discharged on conventional maintenance immunosuppression with tacrolimus and mycophenolate, to lower variability in underlying immunologic risk and minimize the effect of post-transplant factors. Patients were excluded if the preservation strategy was hybridized between HMP and SCS, defined as discordance between the variable indicating whether a kidney was perfused and the variable indicating if the kidney was received as SCS or HMP, and also if data were missing or impossible for CIT, graft survival, or patient survival. Patients were followed until the end of the study period or death. Baseline characteristics were compared according to preservation strategy, summarized as means, medians, and percentages. T tests, Wilcoxon rank-sum tests, and chi-squared tests were used for comparisons as appropriate. Cohen d was calculated for standardized mean differences.

CIT was categorized as follow: ≤12 hours, 12.01–24 hours, 24.01–36 hours, and>36 hours. Cox proportional hazards models were constructed for the primary outcome of death-censored graft failure (dcGF), as well as patient mortality and uncensored graft failure. A logistic model was constructed for DGF, defined as dialysis during the first postoperative week, regardless of whether there was primary nonfunction, defined as graft loss for reasons other than death within the first 90 days.

Each outcome was first compared with kidneys received with SCS ≤12 hours, a reference group representing a commonly accepted timeframe for optimal allograft function. Each model was also analyzed incorporating two-way stratifications between CIT and perfusion status, to calculate custom hazard ratios for the use of perfusion at each time category. A zero-truncated negative binomial model was chosen for adjusted length of stay (LOS) due to over dispersed data with no negative or zero values.

Models were adjusted for donor factors including: age, sex, race, diabetes, donor eGFR, cause of death, left/right kidney, KDPI, and DCD versus donation after brain death (DBD). Recipient factors included age, sex, race, diabetes, waiting time, dialysis years (including preemptive transplantation), calculated panel reactive antibody, body mass index, and private insurance. Models were also adjusted for incompatibilities at human leukocyte antigen tissue type (HLA, HLA-DR), blood groups (A, B, and O), and cytomegalovirus status and also for the use of lymphocyte depleting induction therapy and maintenance steroids. Data were missing for <5% of observations, except for 11% missing data for calculated panel reactive antibody, with six variables available in the dataset for imputation. Missing data were handled using multiple imputation using the Markov chain Monte Carlo method, with generation of ten imputation datasets per model.¹⁴ We examined interactions between strata of storage strategy and CIT with potentially important donor variables (age, KDPI: <20% versus 20% to 85%, and DCD versus DBD) for the primary outcome.

We performed several sensitivity analyses to evaluate potential confounders. We performed our analyses on a dataset eliminating any restrictions regarding immunosuppressive regimens. We added variables for Organ Procurement Organization (OPO) and for time, both as adjustment variables and stratification variables. We also analyzed matched subsets developed with a propensity score model and a subset with matching by donor pair. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc, Cary, NC).

Results

The flow chart for study inclusion is shown in Figure 1. There were 120,438 recipients available for analysis. Of these, 75,479 (63%) allografts were preserved with SCS and 44,959 (37%) were maintained on HMP. The median follow up was 50.2 (interquartile range [IQR], 22.6–93.5) months. Over the study period, 37,632 (31%) grafts failed and 32,106 (27%) of patients died, but 78,451 patients (65%) were alive with functioning allografts at study end. DGF occurred in 33,055 (27%) of patients, of whom 1652 (1%) met our criteria for primary nonfunction. The median CIT was 16.7 (IQR, 11.4–22.2) hours. The use of hypothermic perfusion showed steady linear growth, increasing from 10% of transplants in 2005 to 48% in 2020, with marked increases in use for allografts with KDPI >85%, and those from expanded criteria donors and DCD (Figure 2). The overall use of HMP was 24% between 2005 and 2011, 39% between 2012 and 2016, and 46% from 2017 onward.

Figure 1.

Flow diagram for inclusion into the study. CIT, cold ischemia time; HMP, hypothermic machine perfusion; MMF, mycophenolate; SCS, static cold storage.

Figure 2.

Use of hypothermic perfusion over time, by donor type. All HMP, all transplants utilizing HMP; DBD-ECD, donation after brain death/expanded criteria donor; DBD-SCD, donation after brain death/standard criteria donor; DCD, donation after circulatory death; KDPI >85%, kidney donor profile index score >85%.

Baseline characteristics by preservation strategy are presented in Tables 1 and 2. Compared with recipients of kidneys preserved with SCS, recipients of HMP kidneys were more likely to be male and diabetic, and less likely to have private insurance. Compared with kidneys preserved with SCS, HMP kidneys had higher KDPI, lower eGFR, higher donor age, higher prevalence of DCD, with more donation due to anoxia and less due to trauma. Kidneys on HMP had longer CIT (Figure 3). Only 670 (0.6%) allografts had >36 hours CIT on SCS, compared with 2803 (2%) with >36 hours CIT on HMP.

Table 1.

Baseline donor and recipient characteristics, by organ preservation strategy

Characteristic	ALL N=120,438	SCS n=75,479	HMP n=44,959	Standardized Difference
Recipient age, mean (SD)	53 (13)	52 (13)	55 (13)	0.17
Female recipient, %	40	41	38	−0.08
Recipient race, %
African American	35	34	37	0.06
Asian	8	8	7	−0.06
Hispanic	19	20	18	−0.07
Others	2	2	2	0.02
White	36	36	36	0.02
Recipient diabetes, %	37	36	40	0.10
Dialysis mo; median (IQR)	45.1 (20.3–72.5)	44.9 (20.0–73.9)	45.3 (20.8–70.6)	−0.07
Recipient BMI, mean (SD)	28.5 (5.4)	28.3 (5.4)	28.8 (5.4)	0.08
Recipient diagnosis, %
Cystic kidney disease	8	8	8	−0.002
Diabetic nephropathy	31	30	34	0.10
Hypertensive nephropathy	28	28	28	−0.005
Primary GN	15	15	14	−0.03
Secondary GN	5	5	5	−0.03
Other/missing	13	14	11	−0.08
cPRA category, %
0%	63	61	68	0.14
1%–84%	27	28	25	−0.05
85%–100%	10	11	7	−0.15
Recipient private insurance, %	23	25	21	−0.07
CIT, hours mean (SD)	17.5 (8.6)	15.3 (7.3)	21.1 (9.3)	0.72
Donor left kidney, %	47	47	48	0.01
Donor age mean (SD)	39 (15)	37 (15)	42 (13)	0.33
Female donor, %	39	39	38	−0.02
Donor race, %
African American	14	14	14	0.12
Asian	2	3	2	−0.01
Hispanic	16	17	13	−0.12
Others	1	1	1	−0.06
White	67	65	70	−0.02
Diabetic donor, %	8	6	10	0.12
Donor eGFR, ml/min per 1.73 m2, mean (SD)	89 (38)	92 (37)	84 (37)	−0.23
Donor BMI, kg/m2, mean (SD)	27.9 (6.9)	27.4 (6.8)	28.8 (7.1)	0.19
Donor circulatory death, %	18	9	33	0.60
Donor cause of death, %
Anoxia	36	33	41	0.12
Cancer	0.4	0.5	0.3	−0.02
Stroke	29	29	29	0.02
Trauma	32	36	27	−0.16
Other	3	2	3	0.05

P values: Chi-squared tests for categorical variables; t tests for normally distributed, and Wilcoxon rank-sum tests for non-normally distributed continuous variables. Cohen d was used for standardized differences.

BMI, body mass index; cPRA, calculated panel reactive antibody; HMP, hypothermic machine perfusion; IQR, interquartile range; SCS, static cold storage;

Table 2.

Baseline transplant characteristics, by organ preservation strategy

Characteristic	ALL N=120,438	SCS n=75,479	HMP n=44,959	Standardized Difference
Preemptive transplant, %	10	10	9	−0.007
Period, %
2005–2010	25	31	15	−0.38
2011–2016	33	33	33	0.01
2017–2022	43	37	52	0.32
Lymphocyte depletion, %	71	69	74	0.04
Maintenance steroids, %	76	77	74	−0.05
CMV risk status, %
High (D+/R−)	18	18	18	0.009
Medium (D+/R+)	68	68	68	−0.005
Low (D−/R−)	12	12	13	0.004
Unknown	2	2	1	−0.02
HLA mismatches, %
0	6	8	3	−0.19
1	1	2	1	−0.06
2	4	4	4	−0.03
3	13	13	12	−0.01
4	27	26	28	0.04
5	33	32	35	0.06
6	16	15	17	0.04
H8LA-DR mismatches, %
0  1  2	17 47 36	19 46 35	13 49 38	−0.15 −0.06 0.05
KDPI, % of patients
<20%  20–85%  >85%	26 69 6	30 66 5	19 73 8	−0.25 0.17 0.13
ABO identical/compatible, %	99	99	99	−0.004

P values for categorical variables were calculated from chi-squared tests. Cohen d was used to calculate standardized differences.

ABO, blood groups; CMV, cytomegalovirus; D+, donor positive; D−, donor negative; HLA-DR, HLA-DR isotype; HMP, hypothermic machine perfusion; KDPI, kidney donor profile index; R+, recipient positive; R−, recipient negative; SCS, static cold storage.

Figure 3.

Histogram showing longer cold ischemia times for kidneys sent on HMP (white), versus kidneys sent in SCS (gray), with superimposed normal density curves.

Primary Outcome: dcGF

The results for dcGF are presented in Figure 4A and Table 3. Compared with the reference group (CIT ≤12 hours of SCS), dcGF was higher for SCS kidneys with CIT of 12–24 hours (hazard ratio [HR], 1.11; 95% CI, 1.06 to 1.15; P < 0.001), 24–36 hours (HR, 1.16; 95% CI, 1.09 to 1.23, P < 0.001), and >36 hours (1.21; 95% CI, 1.03 to 1.44; P = 0.02). Among HMP kidneys, dcGF was less than the reference group at ≤12 hours (HR, 0.87; 95% CI, 0.81 to 0.95; P = 0.001) and 12–24 hours (HR 0.95, 95% CI=0.91 to 1.00, P = 0.05) and did not differ from the reference category at 24–36 hours (HR, 1.01; 95% CI, 0.95 to 1.07; P = 0.76). HMP kidneys with >36 hours of CIT had inferior outcomes compared with the reference group (HR, 1.27; 95% CI, 1.16 to 1.39; P < 0.001). Full univariate and multivariable models are presented in Supplemental Table 1.

Figure 4.

Impact of organ preservation strategy on death censored graft failure and delayed graft function at different levels of cold ischemia time. Forest plots showing: (A) HR obtained from an adjusted Cox proportional hazards model for dcGF with 95% CI (B) OR obtained from an adjusted logistic model for DGF with 95% CI. All groups were compared with the reference group of kidneys with CIT ≤12 hours of SCS. CI, confidence interval; dcGF, death-censored graft failure; DGF, delayed graft function; HR, hazard ratios; OR, odds ratios.

Table 3.

Adjusted hazard ratios/odds ratios for graft and patient outcomes, compared with kidneys on ice for 12 hours

CIT, h		dcGF		DGF		Graft Failure (Uncensored)		Patient Mortality
Group	n	HR (95% CI)	P	OR (95% CI)	P	HR (95% CI)	P	OR (95% CI)	P
Static cold storage
0–12	28,374	REF		REF		REF		REF
12–24	38,247	1.11 (1.06 to 1.15)	<0.001	1.17 (1.13 to 1.21)	<0.001	1.07 (1.04 to 1.10)	<0.001	1.06 (1.03 to 1.09)	<0.001
24–36	8188	1.16 (1.09 to 1.23)	<0.001	1.89 (1.79 to 1.99)	<0.001	1.12 (1.07 to 1.17)	<0.001	1.09 (1.04 to 1.15)	<0.001
>36	670	1.21 (1.03 to 1.44)	0.03	2.39 (2.06 to 2.77)	<0.001	1.03 (0.91 to 1.16)	0.69	0.97 (0.85 to 1.10)	0.60
Hypothermic machine perfusion
0–12	5663	0.87 (0.81 to 0.96)	<0.001	0.50 (0.47 to 0.53)	<0.001	1.00 (0.95 to 1.06)	0.87	1.07 (1.01 to 1.13)	0.03
12–24	26,027	0.95 (0.91 to 1.00)	0.05	0.61 (0.59 to 0.64)	<0.001	1.04 (1.00 to 1.07)	0.02	1.09 (1.05 to 1.13)	<0.001
24–36	10,466	1.01 (0.95 to 1.07)	0.76	0.82 (0.78 to 0.86)	<0.001	1.08 (1.03 to 1.12)	<0.001	1.11 (1.06 to 1.16)	<0.001
>36	2803	1.27 (1.16 to 1.39)	<0.001	1.03 (0.95 to 1.11)	0.49	1.23 (1.15 to 1.31)	<0.001	1.17 (1.09 to 1.26)	<0.001

Hazard ratios for death-censored graft failure, uncensored graft failure and patient mortality obtained from adjusted Cox proportional hazards regression. Odds ratios for delayed graft function obtained by adjusted logistic regression.

Adjusted for: recipient age, sex, race, diabetes, body mass index, calculated panel-reactive antibody, primary kidney failure diagnosis, years of dialysis exposure, private insurance; donor age, sex, race, diabetes, eGFR, donation after circulatory death, cause of death; blood group compatibility, human leukocyte antigen and DR-locus mismatches, preemptive transplantation, lymphocyte-depleting induction, maintenance corticosteroids, and cytomegalovirus risk pairing.

CI, confidence interval; CIT, Cold ischemia time; dcGF, death-censored graft failure; DGF, delayed graft function; HR, hazard ratio; OR, odds ratio.

When HMP kidneys were compared with SCS kidneys in the same CIT timeframe, the use of HMP lowered dcGF compared with a kidney on SCS from the same CIT category, by 13%, 14%, and 13% at ≤12, 12–24, and 24–36 hours, respectively (P < 0.001 for each), but perfusion did not affect dcGF once CIT >36 hours (Table 4). Interaction variables with donor age, KDPI category, and DCD/DBD were NS.

Table 4.

Adjusted hazard ratios/odds ratios for outcomes of transplants treated with hypothermic perfusion, across each level of cold ischemia time

Cold Ischemia	0–12 h		12–24 h		24–36 h		>36 h
	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P	HR (95% CI)	P
dcGF	0.87 (0.81 to 0.95)	<0.001	0.86 (0.82 to 0.94)	<0.001	0.87 (0.81 to 0.94)	<0.001	1.09 (0.99 to 1.21)	0.07
Patient mortality	1.07 (1.01 to 1.13)	0.03	1.03 (1.00 to 1.07)	0.08	1.01 (0.96 to 1.07)	0.7	1.07 (0.99 to 1.16)	0.08
Uncensored graft failure	1.00 (0.95 to 1.06)	0.87	0.97 (0.94 to 1.00)	0.07	0.96 (0.93 to 1.02)	0.2	1.10 (1.02 to 1.18)	0.009

Cold ischemia	0–12 h		12–24 h		24–36 h		>36 h
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
DGF	0.67 (0.64 to 0.73)	<0.001	0.52 (0.50 to 0.55)	<0.001	0.43 (0.40 to 0.46)	<0.001	0.43 (0.36 to 0.52)	<0.001

Hazard ratios for death-censored graft failure, uncensored graft failure and patient mortality obtained from adjusted Cox proportional hazards regression. Odds ratios for delayed graft function obtained by adjusted logistic regression.

Adjusted for: recipient age, sex, race, diabetes, body mass index, calculated panel-reactive antibody, primary kidney failure diagnosis, years of dialysis exposure, private insurance; donor age, sex, race, diabetes, eGFR, donation after circulatory death, cause of death; blood group compatibility, human leukocyte antigen and DR-locus mismatches, preemptive transplantation, lymphocyte-depleting induction, maintenance corticosteroids, and cytomegalovirus risk pairing.

CI, confidence interval; dcGF, death-censored graft failure; DGF, delayed graft function; HR, hazard ratio; OR, odds ratio

DGF

DGF, shown in Figure 4B, was higher with longer CIT during SCS. However, compared with a kidney with <12 hours of SCS, HMP kidneys had less DGF at 0–12 hours (odds ratio [OR], 0.50; 95% CI, 0.47 to 0.53), 12–24 hours (OR, 0.61; 95% CI, 0.59 to 0.64), and 24–36 hours (OR, 0.82; 95% CI, 0.78 to 0.86), P < 0.001 for all (Table 3 and Supplemental Table 2). At >36 hours, DGF in HMP kidneys did not differ from reference kidneys with ≤12 hours of SCS (OR, 1.03; 95% CI, 0.95 to 1.11; P = 0.49).

Compared across CIT categories, HMP kidneys had less DGF at CIT <12 hours (OR, 0.67; 95% CI, 0.64 to 0.73), 12–24 hours (OR, 0.52; 95% CI, 0.50 to 0.55), 24–36 hours (0.43; 95% CI, 0.40 to 0.46), and >36 hours (OR, 0.36 to 0.52), P < 0.001 for all (Table 4).

Uncensored Graft Failure (Including Death with Functioning Allograft)

Compared with the reference group (≤12 hours of SCS), the hazard ratio of SCS kidneys for uncensored graft failure was higher at CIT 12–24 hours (HR, 1.07; 95% CI, 1.04 to 1.10; P < 0.001) and CIT 24–36 hours (HR, 1.12; 95% CI, 1.07 to 1.12; P < 0.001), but not when CIT >36 hours (HR, 1.03; 95% CI, 0.91 to 1.16; P = 0.69). HMP kidneys did not differ significantly from the reference group at CIT <12 hours or 12–24 hours but had higher uncensored graft failure at CIT 12–24 hours (HR, 1.08; 95% CI, 1.03 to 1.12) and CIT >36 hours (HR, 1.23; 95% CI, 1.15 to 1.31), P < 0.001 for both (Table 3 and Supplemental Table 3). Comparing HMP to SCS within the same CIT timeframe (Table 4), uncensored graft failure did not differ significantly at ≤12, 12–24, or 24–36 hours but had more uncensored graft failure at CIT >36 hours (HR, 1.10; 95% CI, 1.02 to 1.18; P = 0.009).

Patient Mortality

Compared with the reference group (≤12 hours of SCS), death rate was higher if the kidney arrived on SCS with CIT of 12–24 hours (HR, 1.06; 95% CI, 1.03 to 1.09; P < 0.001), 24–36 hours (HR, 1.09; 95% CI, 1.04 to 1.15; P < 0.001), but did not differ for CIT >36 hours (HR, 0.97; 95% CI, 0.85 to 1.10; P < 0.001). HMP kidneys had higher hazard ratios for death, compared with the reference group of patients with ≤12 hours of SCS (Table 3 and Supplemental Table 4).

When HMP was compared with SCS within the same CIT category, the hazard ratio for death was NS at 12–24, 24–36, or >36 hours, but death rate was higher (HR, 1.07; 95% CI, 1.01 to 1.13; P = 0.03) in recipients of HMP kidneys when CIT was ≤12 hours (Table 4).

Hospital LOS

The median LOS was 5 (IQR, 4–7) days for all recipients with CIT <36 hours, 6 (IQR, 4–8) days for HMP kidneys with CIT >36 hours, and 7 (IQR, 4–10) days for kidneys with >36 hours of SCS. Across equivalent CIT categories, recipients of HMP kidneys had slightly shorter LOS (10.7, 11.2, 12.2, and 13.5 hours at <12, 12–24, 24–36, and >36 hours, respectively, P < 0.001 for all).

Sensitivity Analyses

Expanding the data set to include all post-transplant immunosuppressive regimens (N=143,011) did not change results (Supplemental Tables 5–8). In models with inclusion of OPO (Supplemental Tables 9 and 10 and 17 and 18), period (Supplemental Tables 11 and 12), or year of transplant (Supplemental Tables 19 and 20) as an adjustment variable or a stratification variable, we observed higher patient mortality at>36 hours for recipients of HMP kidneys compared with recipients of SCS kidneys but did not observe higher mortality across the 0–12-hour stratum for HMP kidneys versus SCS kidneys. Parallel models performed on the subsets matched by propensity score (Supplemental Tables 13–16; N=70,008) or by donor-pair (Supplemental Tables 21 and 22; N=69,986) did not show higher mortality across any stratum of CIT but were otherwise consistent with the primary analysis. Significance for higher uncensored graft failure across the >36-hour stratum was not seen in the matched models.

Discussion

HMP effectively lowered DGF and dcGF when CIT was ≤36 hours, outperforming SCS. More importantly, our findings indicate that kidneys preserved with HMP for up to 36 hours had comparable outcomes to kidneys in SCS for ≤12 hours.

Variable definitions of “long” CIT complicate the transplant outcome literature. Few studies have specifically examined the effects of very prolonged CIT. Opelz et al. reported a 21% higher risk of graft failure when CIT exceeded 18 hours, although this finding was not statistically significant.¹⁵ Salahudeen et al. demonstrated a significantly worse graft survival for patients with a CIT higher than 30 hours.¹⁶

In a paired prospective study, Kox et al. demonstrated that HMP lowered DGF, but not death-censored graft survival, when CIT was<10 hours, and observed no benefits with longer CIT.¹⁷ By contrast, our data showed that HMP lowered dcGF, up to CIT of 36 hours. A possible explanation for the discordant findings was our very large study cohort, which allowed us to adjust for multiple donor and recipient characteristics and provided sufficient statistical power to analyze the less common graft failure outcomes.

A paired analysis from the Scientific Registry of Transplant Recipients by Kayler et al. concluded that increasing CIT up to 15 hours in expanded criteria donor kidneys does not adversely affect overall graft survival.⁹ These results differ from those of Cashion et al., who identified the KDPI as an independent risk factor for graft failure, separate from CIT; they included CIT >24 hours.² In our model, we observed a 9% higher dcGF for each 10% higher KDPI, independent of CIT, consistent with Cashion et al.

In our study, HMP was not associated with better outcomes for CIT >36 hours. Higher patient mortality and uncensored graft failure with the use of HMP after 36 hours were observed in some of our models, highlighting the need for caution when extending storage times beyond this threshold. There were relatively few transplants in this very high CIT category, suggesting that clinicians are intuitively hesitant to use kidneys after 36 hours.

In our cohort, kidneys preserved with HMP had overall significantly less advantageous baseline characteristics than those in SCS, which suggests underlying decision-making about the deployment of HMP.

Consistent with previous reports, we observed significantly lower DGF rates in HMP kidneys compared with those in SCS across all CIT categories. Gill et al. similarly reported a significant lowering in DGF for kidneys preserved with hypothermic perfusion up to 36 hours in both standard and expanded criteria donors, although they noted no such benefit in DCD donors. They did not analyze graft survival.¹⁸

DcGF is the preferred outcome measure for assessing graft quality, as it excludes deaths with a functioning graft—events often influenced by factors unrelated to graft performance. In our analysis, uncensored graft failure among patients with >36 hours of CIT was higher among recipients of HMP kidneys, compared with SCS kidneys, a finding that persisted in most of our sensitivity analyses. These observations will need further investigation.

As in previous studies,^6,19 we found no significant differences in patient survival within the same CIT categories when comparing HMP with SCS kidneys, aside from a slightly higher risk of death in the ≤12 hour CIT HMP group (HR, 1.07; 95% CI, 1.01–1.13), which did not persist in many of our sensitivity analyses.

One reason to avoid using HMP is the additional cost associated with organ acquisition. However, when the economic burden of DGF is considered—including longer hospital days, dialysis, and follow-up appointments—the initial cost of HMP may be less consequential. Previous studies on the cost-effectiveness of HMP have shown significant cost savings due to less DGF and less primary nonfunction.²⁰ Analyses that evaluate the cost-benefit effects of HMP across varying cold ischemia times are needed.

The strength of this study is the large dataset, with examination of many patients and adjustments for many factors. The limitations of this study include its retrospective design and the reliance on registry data. In our cohort, HMP kidneys had overall significantly less advantageous baseline characteristics that SCS kidneys, which suggests underlying clinical decision making about the use of HMP. There are no standardized criteria for using HMP, and utilization is variable across transplant centers. Schold et al. found that kidneys from DCD, Black donors, and donors with prolonged CIT, elevated terminal creatinine, and diabetes were more likely to have HMP.¹⁰ We have no data on the criteria used to select organ preservation strategies, introducing the possibility of indication bias. However, as most of the characteristics of the HMP group were inferior to those in SCS, the expected direction of this bias would have been expected to lower the effect sizes. The primary finding of lower dcGF was observed consistently in our propensity-matched and donor-matched cohorts, which was reassuring.

Another important limitation is that UNOS registry data do not capture the duration of HMP relative to total CIT. Not all kidneys receive HMP immediately, as some have to be transported from local hospitals to an OPO before being perfused. In addition, our findings cannot be generalized to kidneys that were initially placed on HMP but later transported with SCS, as these were excluded due to ambiguity regarding the proportions of time spent in each preservation method. Machines do travel across the country, but this is dependent on availability and OPO policies. Resistance index levels were not available. Additional confounding by unmeasured factors, such as warm ischemia time, medical events, and surgical complications, cannot be ruled out.

There is evidence that oxygenation may enhance organ preservation. The COMPARE study demonstrated that oxygenated HMP led to improved kidney transplant outcomes compared with nonoxygenated HMP.²¹ Oxygenation is not captured in our dataset but is not likely to have occurred in many of our subjects, since introduction of this technique in the United States did not occur until 2022. Future study of the effect of oxygenated HMP on graft survival and clinical outcomes is warranted.

Kidney transplantation has faced increasing challenges over time, including changes to the allocation system, with transportation of organs over longer distances. Advancing age and inclusion of patients with more comorbid conditions has increased the medical complexity of kidney recipients. Increasing rates of retransplantation require time-consuming crossmatching. This complexity contributes to longer CIT. HMP effectively lowered the adverse effects of prolonged CIT, allowing for greater flexibility in timing, enabling surgical teams to schedule complex surgeries during daytime hours, when optimal staffing and resources are available, potentially improving outcomes for high-risk patients.

In conclusion, HMP is a valuable strategy for lowering the negative effect of prolonged CIT, providing transplant teams with the flexibility needed to optimize donor and recipient logistics.

Supplementary Material

cjasn-21-108-s001.pdf ^{(512.1KB, pdf)}

cjasn-21-108-s002.pdf ^{(2.6MB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/CJN/C451.

Author Contributions

Conceptualization: Rita L. McGill, Angelica Perez-Gutierrez.

Formal analysis: Rita L. McGill.

Methodology: Rita L. McGill.

Visualization: Rita L. McGill.

Writing – original draft: Rita L. McGill, Angelica Perez-Gutierrez.

Writing – review & editing: Piotr Bachul, Rolf N. Barth, Jessica Cao, John Fung, Rita L. McGill, Angelica Perez-Gutierrez.

Funding

None.

Declarative Statements

This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. The need to obtain informed patient consent was waived. This study includes clinical experimentation and complies with the Declaration of Helsinki.

Data Availability Statements

Data belong to a third party, and authors are not authorized to share the data. Identity of Third Party: UNOS. Reason for Restriction: UNOS Star Files can be made available free of charge to any individual who fills out a Data Use Agreement. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the Declaration of Istanbul on Organ Trafficking and Transplant Tourism.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/C452.

Supplemental Table 1. Unadjusted and adjusted models for dcGF.

Supplemental Table 2. Unadjusted and adjusted models for DGF.

Supplemental Table 3. Unadjusted and adjusted models for uncensored graft failure.

Supplemental Table 4. Unadjusted and adjusted models for patient mortality.

Supplemental Table 5. Baseline donor/recipient characteristics, with no limitations on immunosuppressive regimen.

Supplemental Table 6. Baseline transplant characteristics, with no limitations on immunosuppressive regimen.

Supplemental Tables 7 and 8. Adjusted models for transplant outcomes, compared to 12 hours SCS, with no limits on immunosuppressive regimen.

Supplemental Tables 9 and 10. Adjusted models for transplant outcomes, with OPO as an adjustment variable.

Supplemental Tables 11 and 12. Adjusted models for transplant outcomes, with period as an adjustment variable.

Supplemental Tables 13 and 14. Covariate balance after propensity score matching.

Supplemental Tables 15 and 16. Models for transplant outcomes performed on a propensity-matched dataset.

Supplemental Tables 17 and 18. Adjusted models for transplant outcomes, with OPO as an stratification variable.

Supplemental Tables 19 and 20. Adjusted models for transplant outcomes, with period as an stratification variable.

Supplemental Tables 21 and 22. Adjusted models for transplant outcomes, performed on recipient pairs matched by donor.