An experienced investigator screened the potential research studies by running a search strategy from inception to January 6, 2023, in seven electronic databases: MEDLINE, Embase, Web of Science, Scopus, Cochrane Central Register of Controlled Trials, EBSCOhost, and Latin American and Caribbean Health Sciences Literature Database (LILACS). The search strategy involved employing both keywords and free-text terms to explore relevant studies about external cooling devices to preserve a kidney graft during a transplant procedure. Both studies that tested a cooling device (CD) in vivo animal models or human patients were included. Terms such as “graft, kidney”, “cooling system”, “cooling apparatus” and “cooling device” were used during the search strategy. The results were complemented by screening reference lists, grey literature, and contacting experts in the field. The full search strategy is presented in Supplementary Material.

Studies with a randomized, interventional, and comparative design were included. For studies conducted with animal models, they were considered eligible if: (1) used swine as animal models, since they have numerous anatomical and functional similarities with humans, (2) maintained the animal subjects under standard conditions, with water and food provided ad libitum, (3) average weight for their age, and (4) were submitted to the transplant procedure under an appropriate anesthesia protocol. Common inclusion criteria included: (1) the use of an external cooling device during the kidney transplant in at least one of the study groups, and (2) at least one study group underwent the kidney transplant without the external cooling device, independently of the technique used (standard, robot-assisted, or laparoscopic). Studies were excluded if they had a single-arm design, compared two or more external cooling devices without including a control group, were an ex vivo animal model study, or had no outcomes of interest reported by the authors. We avoid including case-control studies, systematic reviews, meta-analyses, case reports, basic science research, conference abstracts, letters to the editor, and literature reviews. Studies were not excluded by terms of subjects' sex, year of publication, language, study setting, or time frame.

Studies without a control group were excluded to ensure the methodological integrity of the review and allow for valid comparisons of outcomes. The absence of a comparator group limits the ability to isolate the effect of the cooling intervention from confounding variables or natural variability in perioperative care. Including only comparative studies enhances internal validity and facilitates a more accurate assessment of the effectiveness and safety of external cooling devices during kidney transplantation.

Before every screening phase was conducted, a pre-screening pilot phase was carried out among four independent reviewers to adjust their attachment to the selection criteria. Between reviewers' agreement was estimated using the Fleiss' Kappa index, considering a value ≥0.70 as an acceptable cut-off value. The reviewers systematically screened all the studies, independently and blinded, by their abstracts and full text. Any disagreement was solved by consulting an additional reviewer who is an expert in the field (FRS). A detailed graphical representation of the global screening process is shown in (Figure 1).

A pre-designed extensive database was used for data extraction, based on the specifications described in the Cochrane Handbook for Systematic Reviews (16). For each reference included, descriptive and outcome-related data were extracted. In cases where data was only available in graphic resources (e.g., plots), a web-based digital tool (“WebPlotDigitizer”, https://apps.automeris.io/wpd/) was used to extract quantitative data.

Outcome-related data from primary and secondary outcomes were extracted as Mean ± Standard Deviation (SD) of the scores from the clinical tools used for their assessment. When data was reported as Mean ± Standard Estimate Mean (SEM), SD was obtained by multiplying the SEM by the square root of the sample size n of the group (17):SD=(SEM)(n)Incase a study omitted or unclearly reported an outcome, a letter was sent via email to the corresponding author to make contact and retrieve missing data.

Primary outcomes assessment included objective indicators of the kidney graft's functionality and integrity, such as histological preservation of the nephron, and urinary output after the transplant procedure.

These outcomes were evaluated through: 1.

Goujon's score: originally evaluated eight histopathological parameters (apical cytoplasm vacuolization, tubular necrosis, tubular dilatation, cell detachment, brush border integrity, intracellular edema, denuded basement membrane, mitochondria integrity) using an ordinal scale from 1 to 5, and consist in the following criteria: 1, no abnormality; 2, mild lesions affecting 10% or less of kidney samples; 3, lesions affecting 25% of kidney samples; 4, lesions affecting 50% of kidney samples; and 5, lesions affecting 75% or more of kidney samples (18). This score, and its modified version, which uses the same scale, but only assesses four histopathological parameters (percentage of glomerular flocculus retraction in Bowman's space, brush border loss, lumina of tubules with cellular debris, and tubular dilatation) are widely used for the histological integrity assessment of renal tissue, specifically after and intervention for its preservation is made.

Two reviewers independently assessed the risk of bias from the selected studies through one of two possible tools. For studies conducted in animal models, the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE tool) was used (19, 20). This tool is composed of ten signaling questions, which assess methodological bias in six domains (selection, performance, detection, attrition, reporting, and other type of bias).

We evaluated the studies reporting the use of renal cooling devices, regardless of whether a control group was included, surgical approach, experimental model, and outcome. For each study, we identified and compiled the technical specifications and reported properties of the cooling devices, as described by the inventors or authors. This approach allowed us to compare the various techniques implemented to protect the kidney from warm ischemia. The following characteristics were assessed: device material, manufacturing process, tubing configuration, sterilization methods, type of flow control system, coolant type, temperature monitoring mechanisms, presence of a hilum access window, compatibility with different surgical approaches, and the experimental model in which the device was tested.

Analysis of the overall histological preservation was made by pooling the weighted mean treatment effect, based on the standardized mean difference (between-groups Hedge's g) between the Goujon scores (simple or modified) obtained by both intervention and control groups at follow-up. The mean difference was standardized using the pooled SDΔ and corrected for bias due to small sample size. For the interpretation of the effect's magnitude, the following cut-off values were considered (Cohen, 1988): ≥0.8 = large; 0.2–0.5 = medium; and ≤0.2 = small. Since significant heterogeneity between studies was expected due to numerous methodological aspects, a Random-Effects Model was used to pool all overall outcomes.

For the secondary outcomes, the analysis of the weighted mean treatment differences in the evaluation of other functional indicators of the graft (urinary output) and surgical management (surgical time, time of anastomosis, length of cold and warm ischemia periods, and graft temperature at reperfusion) was made through the pooling of the standardized mean difference between measures obtained by both intervention and control groups.

To appraise heterogeneity, Cochran's Q-statistic was employed, adopting a significance threshold of p < 0.10. Additionally, the I²-statistic quantified the proportion of variability attributed to heterogeneity, with values surpassing 50% denoting substantial heterogeneity. Forest plots were used to represent the calculated mean differences graphically. No sensitivity or subgroup analyses were made due to the restricted number of articles used for quantitative analysis. Data analysis was made using meta and dmetar packages in the R statistical version R.4.2.3 software (Posit PBC, Boston, United States).

A general description of the included studies is presented in (Table 1). All animal models used a porcine model. Three studies included a cooling jacket device with a cold between 0 and 4°C liquid infusion circuit to preserve temperature. One study used an isolation bag preservation (organ pocket), the solutions varied between studies. From the animal studies, 2 used an open technique and 2 used minimal-invasive techniques, including robotic and laparoscopic techniques. Only two animal studies used a direct measurement technique to describe the mean renal temperature after reperfusion. The surgical time in the animal models was similar between groups. Warm ischemia time was reported in all animal studies. Histological outcomes were reported in two of the animal studies using the Goujon score.

The quality assessment of animal studies is demonstrated (Table 2). Only one of the animal studies generated a random sequence with unknown allocation concealment; moreover, only one study used blinding for the intervention.

Across the 36 studies reviewed, we identify 21 with a relevant description. A considerable variability in the design and reporting of renal cooling devices. Common materials included plastic polymers (silicone, polyethylene, polylactic acid), with one using aluminum (38). Manufacturing processes were inconsistently detailed; some used 3D printing (31, 35) or homemade sealed bags (21, 31, 34). Silicone or latex tubing was prevalent, but tubing configurations and flow control systems varied, ranging from manual irrigation to peristaltic pumps. Ice-cold saline was the standard coolant. Hilum access windows were frequently included. Devices were tested with open, laparoscopic, or robotic approaches and tested in animal or human models. Renal temperature monitoring during surgery was assessed using different methods across studies (Table 3).

This study has several limitations. First, the evidence based on preclinical studies with animal models, which may not be replicated in clinical studies, particularly for the lack of long-term follow-up. While a high number of single-arm studies and case series don't report adverse outcomes using cooling devices, the absence of randomized controlled trials limits the strength of the conclusions that can be drawn. Second, we have excluded a high amount of studies due to the strict criteria of inclusion of studies with the intervention without a control group. This, means that we only obtained a four preclinical studies in animals, limiting the performance of the meta-analysis. Third, the screened records exhibited a high risk of outcome reporting bias, which may impact the reliability of our findings. To mitigate this during the study selection process, we employed a comprehensive and systematic approach based on predefined inclusion criteria. Specifically, we included all studies that reported any clinical or experimental evaluation of cooling devices, irrespective of whether the outcomes were positive, negative, or inconclusive. Additionally, none of the studies have explored the short-term and long-term security profile of the devices including the prolonged anastomosis time, increased risk of wound infection, elevated costs, or other associated complications. Finally, a significant limitation across the included studies is the insufficient reporting of key technical aspects of the cooling devices. Most publications lacked detailed descriptions of the device design, the biomaterials employed, sterilization methods. Also, the high variety of strategies used to monitor thermal transfer at the renal surface. This lack of standardized reporting impairs the ability to critically assess the safety and efficacy of the inventions.