Effect of pretransplant dialysis modalities on pancreas-kidney transplant outcomes: a systematic review and meta-analysis

Background: The impact of different pretransplant dialysis modalities on post-transplant outcomes for pancreas-kidney transplantation is currently unclear. This study aims to assess the association between pretransplant dialysis modalities [hemodialysis (HD) and peritoneal dialysis] and outcomes following pancreas-kidney transplantation. Methods: The authors searched PubMed, EMBASE, and the Cochrane Library for relevant studies published from inception until 1 December 2023. The authors included studies that examined the relationship between pretransplant dialysis modalities and clinical outcomes for pancreas-kidney transplantation. The primary outcomes considered were patient, pancreas and kidney graft survival, and intra-abdominal infection. Results: A total of 13 studies involving 1503 pancreas-kidney transplant recipients were included. Pretransplant HD was associated with improved pancreas graft survival (hazard ratio = 0.71, 95% confidence interval: 0.51–0.99, I²=12%) and a decreased risk of intra-abdominal infection [odds ratio (OR)=0.69, 95% CI: 0.51–0.93, I²=5%). However, no significant association was found between the dialysis modalities and patient or kidney graft survival. Furthermore, pretransplant HD was linked to a reduced risk of anastomotic leak (OR=0.32, 95% CI: 0.161–0.68, I²=0%) and graft thrombosis (OR=0.56, 95% CI: 0.33–0.96, I²=20%). Conclusion: Pretransplant HD is the preferred dialysis modality while awaiting pancreas-kidney transplantation, although well-designed prospective studies are needed to confirm these findings.


Introduction
Simultaneous pancreas-kidney (SPK) or pancreas after kidney (PAK) transplantation is the preferred treatment for patients with insulin-dependent diabetes mellitus and end-stage renal disease, because it restores long-term glycemic control and can reduce secondary diabetic complications [1][2][3] .Compared to insulin replacement, pancreas transplantation currently remains the most effective and reliable way to achieve sustained normal blood glucose, prevent hypoglycemia and ketoacidosis, and normalize hemoglobin A1c levels [4] .However, the number of pancreas transplants (including SPK and PAK) has declined globally in recent years [5] .This may be due to the fact that the early transplant failure rate of pancreas transplantation is the highest among all solid organ transplants [6] .This is related to the surgical complexity of pancreas transplantation and susceptibility to ischemia-reperfusion injury [6] .Therefore, it is important to identify the risk factors for poor prognosis, which may guide the improvement of prognosis.
Nowadays, the vast majority of pancreas transplants are combined kidney transplants (including SPK and PAK), the proportion of pancreas transplantation alone is very small [6] .In addition, due to a shortage of donors, most patients must undergo renal replacement therapy with dialysis before receiving a transplant.Hemodialysis (HD) is the most commonly used

HIGHLIGHTS
• The effect of pretransplant dialysis modalities on posttransplant outcomes in patients receiving pancreas-kidney transplantation is uncertain.• In this systematic review and meta-analysis of 13 observational studies involving 1503 patients, patients undergoing hemodialysis had a significantly decreased risk of pancreas graft failure, intra-abdominal infection, anastomotic leak, and graft thrombosis than patients undergoing peritoneal dialysis.• The findings suggest that hemodialysis is the preferred dialysis mode of choice during the waiting period for pancreas-kidney transplantation.
form of dialysis worldwide, with ~89% of patients with end-stage kidney disease receiving HD [7] .However, peritoneal dialysis (PD) has advantages such as preserving residual kidney function, better quality of life, and lower cost compared to HD [8][9][10] .The choice of PD or HD is influenced by factors like patient autonomy, comorbidities, vascular or peritoneal diseases, and dialysis center factors.PD appears to be a more effective transitional treatment for kidney transplant recipients prior to kidney transplantation.A recent systematic review revealed that kidney transplant recipients who underwent PD prior to transplantation had decreased risk of overall kidney graft failure and kidney delayed graft function [11] .While this study also encompassed some studies of SPK recipients, its primary focus was on kidney graft outcomes.Further investigation is needed to assess the outcomes of pancreas graft and specific complications associated with pancreas transplantation.Additionally, the unique nature of SPK or PAK in comparison to kidney transplantation alone (KTA) may also contribute to differences in the impact of dialysis modalities [1] .
The influence of pretransplant dialysis modalities (HD or PD) on prognosis in pancreas-kidney transplant recipients has been concerned and investigated.However, these studies have not reached a consistent conclusion.A previous study reported that compared with PD, pretransplantation HD is associated with decreased risk of pancreas graft failure, anastomotic leak, and thrombosis [12] .However, other studies have found that these outcomes are similar in patients undergoing pretransplant HD and those on PD [13,14] .These conflicting findings may be attributed to differences in selection criteria, sample size, or study design.
Therefore, the aim of this study was to conduct a systematic review and meta-analysis to determine the effects of pretransplant dialysis modalities on patients receiving SPK or PAK.

Search strategy and selection of studies
Two researchers independently searched various studies and identified relevant full text from the following databases: PubMed, EMBASE, and Cochrane Library (from inception until 1 December 2023).Detailed retrieval is presented in Supplementary Materials (Table S1, Supplemental Digital Content 4 [SDC], http://links.lww.com/JS9/C453).
Detailed inclusion and exclusion criteria are present in Table S2, SDC (Supplemental Digital Content 4, http://links.lww.com/JS9/C453).In short, all studies investigating the relationship between dialysis modalities (HD and PD) and outcomes in pancreas-kidney transplant recipients (including SPK and PAK) were considered for inclusion, without limiting language.Clinical trials, cohort studies, and case-control studies are considered.In addition, in order to include as much relevant research as possible, conference abstracts, and letters with detailed data were also considered.Case reports, case series, cross-sectional studies, reviews, or studies without control groups were excluded.For studies with overlapping cohorts, consider the study with the most detailed data.

Data extraction, methodological assessment, and outcome measures
Two researchers independently extracted data from the final included studies, and the differences were resolved by the third researchers.The extracted data included study characteristics, recipient characteristics, donor and transplant characteristics, and outcome data.The quality of the included studies was independently assessed by two researchers using the Newcastle-Ottawa scale (NOS) [18] .NOS score ≥ 8 points is considered as high quality.
The primary outcomes of the study were patient survival, pancreas graft survival, kidney graft survival, and postoperative intra-abdominal infection.Secondary outcomes included relaparotomy, pancreas graftectomy, graft thrombosis, bleeding complications, graft pancreatitis, anastomotic leak, rejection, kidney delayed graft function, cytomegalovirus infection, wound infection, and the length of hospital stay.

Statistical analysis
Pooled hazard ratios (HRs) and 95% CIs were used to assess the effect of pretransplant dialysis modalities on the risk of mortality or pancreas/kidney graft loss.Use the adjusted HR if it is reported.For studies that did not report HR and 95% CI, we used the method described by Liu et al. [19] to obtain survival data from the reported Kaplan-Meier curve and calculate it.Pooled intraabdominal infections and other secondary outcomes, expressed as odds ratios (OR) and 95% CI, were calculated by including the number of total and events.Mean difference and 95% CI were used to pool length of hospital stay.Heterogeneity across studies was assessed by I² statistics, with significant heterogeneity defined as I² > 50%.If the heterogeneity is significant, the random effects model is used.Otherwise, the common effect model is used.
Preplanned subgroup analyses were performed according to year of publication, study location, or sample size.Univariate meta-regression analysis was performed to explore the association between study characteristics, recipient characteristics, donor and transplant characteristics, and pooled primary outcomes.Sensitivity analysis is performed by iteratively omitting single studies from pooled analyses to assess their role in the pooled results of primary outcomes.Publication bias was assessed by Egger' test and Begg' test, with P < 0.10 [20,21] .All outcomes were analyzed using R software (version 4.2.1).P-values of less than 0.05 were considered statistically significant.

Study selection
A total of 1414 records were retrieved from the three databases, of which 97 were duplicates.After filtering the titles and abstracts of the remaining records, 1296 records were excluded.After eligibility determination of the full text of 21 studies, 8 studies were excluded.Finally, 13 studies were included in the systematic review and meta-analysis [12][13][14][22][23][24][25][26][27][28][29][30][31] . The PRIMA flowchart of the study selection is presented in Figure 1.

Study characteristics and risk of bias
Table 1 and Table S3, SDC (Supplemental Digital Content 4, http://links.lww.com/JS9/C453)summarize the characteristics of the 13 included studies.The study sample size ranged from 27 to 200 and involved a total of 1503 pancreas-kidney transplant recipients.All studies were designed retrospectively.Most of the studies were conducted in SPK, and only one was in a mixed SPK and PAK cohort [12] .The mean recipient age ranged from 24.96 to 43.66 years, and the male proportion ranged from 48.1 to 72%.Dialysis duration ranged from 13 to 29.7 months, and diabetes duration ranged from 23.6 to 32.61 years.The NOS score ranges from 6 to 9, with five studies (38.46%) rated as high quality (Table S4, SDC, Supplemental Digital Content 4, http://links.lww.com/JS9/C453).

Subgroup analysis, meta-regression, sensitivity analysis, and publication bias
The association between pretransplant HD and pancreas graft survived was observed when subgroup analyses were based on studies published before 2015, studies with a sample size ≤ 100, or studies conducted in European.While the association between HD and intra-abdominal infection was observed when subgroup analyses were limited to studies published before 2015, studies with a sample size > 100, or studies conducted in European.With respect to patient survival and kidney graft survival, no differences were observed between subgroups based on publication year, sample size, or study location (Table S5, SDC, Supplemental Digital Content 4, http://links.lww.com/JS9/C453).
Sensitivity analyses for primary outcomes, patient survival and kidney graft survival were robust.Results for intra-abdominal infections are relatively robust and not affected by most studies.The outcome of pancreas graft survival was not robust (Table S7, SDC, Supplemental Digital Content 4, http://links.lww.com/JS9/C453).
The results of publication bias are presented in Table S8, SDC (Supplemental Digital Content 4, http://links.lww.com/JS9/C453).There may be publication bias in the analysis of length of hospital stay (P = 0.0108 for Egger' test, P = 0.1742 for Begg' test).In addition, no publication bias was found for other outcomes (both Begg' and Egger' tests > 0.10).

Discussion
This is the first systematic review and meta-analysis to summarize the effects of pretransplant dialysis modalities on pancreas-kidney transplant outcomes.The study found that patients who underwent HD prior to transplantation had better pancreas graft survival and decreased risk of intra-abdominal infection, anastomotic leak, and graft thrombosis compared with patients who underwent PD.However, the overall risk of bias in the included studies was moderate, and these results should be interpreted with caution.The elevated risk of complications in PD patients may contribute to the decreased survival rate of pancreas grafts.Graft thrombosis is the primary cause of early pancreas graft loss, accounting for ~60% of graft losses within three months after transplantation [32] .Our study also revealed that patients who underwent PD prior to transplantation had a higher risk of graft thrombosis compared to patients who underwent HD.Graft thrombosis often leads to subsequent graft pancreatectomy [33] .Despite the findings of this study, no effect of dialysis modalities on kidney graft outcomes, including kidney delayed graft function and kidney graft survival, was found in pancreas-kidney transplant patients.However, this conclusion is based on limited studies and sample sizes.It is important to carefully consider the impact of dialysis modalities on kidney graft outcomes and conduct prospective studies with larger sample sizes, especially considering previous evidence that pretransplantation PD generally provides better preservation of residual native kidney function compared to HD [34][35][36] .Although this study did not find a significant difference in patient mortality among pancreas-kidney transplant recipients with different pretransplant dialysis modalities.However, previous studies have found that HD may increase the risk of cardiovascular events after renal transplantation through intermittent and nonphysiological volume changes [37] .
In contrast to KTA, SPK, or PAK procedures require abdominal manipulation, easier access to a stronger immunosuppressive regimen, and history of diabetes mellitus [1] .As a result, intraabdominal infection is a common complication of these procedures [38] .The peritoneal catheter communicates with the skin and external environment, and long-term PD can lead to peritoneal fibrosis and decreased peritoneal defense function, potentially increasing the risk of intra-abdominal infection in PD patients [22,39] .Therefore, more frequent monitoring for infection in PD patients after surgery is recommend.Effective infection prevention programs need be considered, but multidrugresistant infections need to be vigilant.Furthermore, peritoneal thickening and fibrosis caused by long-term PD may lead to adhesions that affect anastomotic healing [39] .PD patients have an increased incidence of intraperitoneal infection and a secondary rupture of the anastomosis, potentially contributing to the increased incidence of anastomotic leak in PD patients.
Compared with kidney grafts, pancreas grafts are more susceptible to graft thrombosis [1,38] .In PD patients, the procoagulant system and blood concentration are thought to be more prone to thrombosis than in HD patients, potentially explaining the higher risk of graft thrombosis in PD patients [40,41] .However, potential confounding bias should be carefully considered.Patients with a hypercoagulable state or poor vascular condition before dialysis may be more likely to choose PD, making these patients themselves more susceptible to thrombosis.Our results suggest that active anticoagulant prophylaxis may be needed in patients with PD.However, this must be weighed against the increased risk of bleeding, although no difference in baseline bleeding risk was found between HD and PD patients in our study.
The current evidence is derived from retrospective studies and is of limited quality.However, it is currently challenging to conduct randomized clinical trials to compare the effects of HD and PD on the outcome of pancreas-kidney transplantation.In patients with diabetes and end-stage renal disease awaiting pancreas-kidney transplantation, we believe that HD is a reasonable replace treatment option.However, a comprehensive assessment of blood vessel and thrombosis status, as well as the patient's willingness, needs to be considered.Therefore, nephrologists, transplant surgeons, endocrinologists, and patients need to make an integrated decision before initiating dialysis.For patients who received PD before transplantation, more potent infection prevention protocols and anticoagulant therapy regiments need to be considered to reduce the risk of intra-abdominal infection and graft thrombosis.Well-designed prospective studies are needed in future to validate the findings of this study.
This systematic review and meta-analysis must acknowledge certain limitations.Firstly, the included studies were designed retrospectively, and their limitations may introduce potential bias.Some studies excluded patients who had undergone dialysis switching, making it impossible to determine outcomes in this populations.Secondly, it is important to note the clinical heterogeneity between studies.The majority of the studies focused on SPK, with only one study including a mixed cohort of SPK and PAK.The studies also differed significantly in technical features, including endocrine and exocrine drainage.Therefore, caution should be exercised when interpreting the conclusions of this study in relation to other populations.However, despite significant clinical heterogeneity, statistical heterogeneity was not significant in most analyses, leading us to speculate that it has little effect on the outcome of dialysis mode.Thirdly, the data for intra-abdominal infection outcome and all secondary outcomes were not adjusted, potentially overlooking the influence of potential confounding factors and exaggerating the effect of dialysis modalities on outcomes.
Lastly, the overall quality of the studies is moderate and uneven.Therefore, the results of this meta-analysis should be interpreted with caution.

Conclusion
Compared with PD, patients who received HD before transplantation had comparable patient survival and kidney graft survival.However, patients who received HD had better pancreas graft survival and a decreased risk of abdominal infection, anastomotic leak, and graft thrombosis.The results of this study suggest that the outcome of pancreas-kidney transplantation in HD is better than that in PD.Unlike patients waiting for a kidney transplant alone, current evidence recommends HD as the preferred dialysis modality for patients waiting for pancreas-kidney transplantation.However, due to the limited quality of the evidence, it is necessary to conduct future prospective large sample studies.

Figure 2 .
Figure 2. Forest plots on patient survival (A), pancreas survival (B), kidney survival (C), and intra-abdominal infection (D) in patients undergoing hemodialysis compared to peritoneal dialysis.

Table 1
Baseline characteristics of the included studies.