Simultaneous pancreas and kidney transplantation for end-stage kidney disease patients with type 2 diabetes mellitus: a systematic review and meta-analysis

Purpose The indications for patients with type 2 diabetes mellitus (T2DM) combined with end-stage kidney disease (ESKD) undertaking simultaneous pancreas and kidney transplantation (SPK) remain an unresolved issue. This study aimed to systematically review the survival outcomes of SPK among T2DM-ESKD patients. Methods Online databases including PubMed, MEDLINE, EMBASE, and the CENTRAL Library, CNKI, Chinese Biomedical Literature Database, and Wan-Fang database were used to locate the studies of ESKD patients with T2DM undertaking SPK up to May 2021. A third reviewer was consulted if there were disagreements. Data were analyzed with STATA (15.0). Results Nine cohort studies were identified. The pooled 1-year, 3-year, and 5-year patient survival rates of patients with T2DM and ESKD after SPK were 98%, 95%, and 91% respectively. Comparing the treatment effect of SPK between type 1 diabetes mellitus (T1DM) and T2DM, the survival estimates were comparable. For T2DM patients, SPK had a survival advantage compared with KTA. Conclusions The synthesized clinical outcomes of T2DM patients with ESKD after SPK were relatively better than KTA, but a subset of T2DM-ESKD patients who would benefit the most from SPK was to be defined. PROSPERO registration number CRD42019118321. Date of registration: 14 Jan 2019 (retrospectively registered) Supplementary Information The online version contains supplementary material available at 10.1007/s00423-021-02249-y.


Introduction
It was estimated that there were more than 463 million people were living with diabetes mellitus (DM) worldwide, and more than 90% of them were diagnosed with type 2 diabetes (T2DM) [1,2]. In Europe, the number of DM is estimated to be 58 million [2]. Over the past years, China has witnessed a surging prevalence of diabetes, with the largest number of diabetic patients in the world [3] and ranked number one in the 2019 International Diabetes Federation Diabetes Atlas Report [2]. Furthermore, diabetes is the leading cause of end-stage kidney disease worldwide; in conjunction with hypertension, it resulted in at least 80% end-stage kidney disease (ESKD) [4]. In the USA, Japan, New Zealand, and Singapore, about 50% of ESKD are primarily due to DM [4].
Since the first pancreas transplantation was done at Minnesota University in 1966, with the improvement of surgical techniques and introduction of immunosuppressive agents of cyclosporine, the number of pancreas transplantation has increased steadily, especially for simultaneous pancreas transplantation [5,6]. SPK has been a medically effective and cost-effective method for T1DM, but there was no consensus on SPK for the T2DM population, especially in the aspect of selection criteria [6,7]. In the 2020 Kidney Disease: Improving Global Outcomes (KDIGO) guideline, patients with ESKD and T1DM were recommended for SPK, while there were no suggestions for those with T2DM [8]. Data on SPK outcomes in T2DM patients began appearing in the annual International Pancreas Transplant Registry (IPTR) reports since the mid-1990s [9,10]. The number has gradually increased with treatment outcomes equivalent to or better than other treatment alternatives on T2DM nephrological patients [11]. The cases of SPK were steadily increasing in Europe as well [12]. Considering the growing size of T2DM-ESKD recipients receiving SPK and the need to synthesize existing knowledge to inform clinical practice, we sought to review systematically and summarize available survival data in these patients. We planned to (i) synthesize the risk of death after SPK for T2DM-ESKD patients; (ii) assess the quality of available epidemiological data; (iii) summarize the hazard risk of mortality between SPK T2DM recipients and their counterparts; and (iv) estimate the relative risk of commonly reported complications between SPK T2DM recipients and their counterparts.

Methods
This meta-analysis was written in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses [13].

Eligibility criteria
This review included cohort studies estimating the survival outcome of SPK for T2DM patients combined with ESKD since no trials were available. All studies that reported SPK survival outcomes of T2DM-ESKD patients in English or Chinese were included. There were no restrictions on the type of setting. The year of publication was limited for Chinese studies. Those conducted before 2010 were excluded during the study selection process, considering the implementation of the Donation after Citizen's Death in 2010 [14]. Primary outcomes were patients' and grafts' survival rates. Secondary outcomes were hazard ratio between T1DM and T2DM, SPK, and KTA, and risk ratio of complications was recorded as well.

Information sources, search strategy, and records management
Only quantitative studies were searched. PubMed, MED-LINE (1946 onwards), EMBASE (1947 onwards), the CEN-TRAL trials registry of the Cochrane Collaboration (1948 onwards), China National Knowledge Infrastructure (CNKI, 1994 onwards), Chinese Biomedical Literature Database (CMB, 1978 onwards), and Wan-Fang database (1998 onwards) were searched to May 2021. The specific search strategies were created by two team members in consultation with an expert in medical informatics. Search strategies were included in Supplemental digital contents Table 1  (SDC-Table S1). As relevant studies were identified, the reviewers checked for additional relevant articles. Records identified through the database were managed with NoteExpress, which is an information manager for researchers and designed to help organize research notes and bibliographic references and generate bibliographies automatically (http:// www. inote expre ss. com/ aegean/).

Study selection
The titles and abstracts of all of the references generated by search strategies were screened independently by two review members to identify eligible studies. Disagreements were resolved by discussion and consensus between the two reviewers. If disagreement persisted, the final decision was made by consensus with the involvement of the third member of the team. The full-text articles of included abstracts and uncertain abstracts were retrieved and reviewed by two members for inclusion separately. Reasons for study exclusion were recorded.

Data collection process
Two authors independently extracted and record data based on a standardized data extraction form (EXCEL form) designed by YXF and YC. The following items were extracted from the identified articles, name of the first author, publication year, title, study purpose, country, city/ region, data source, study design, definition of T2DM, operation technique, number of cases, study period, age, BMI, sex, duration of DM, pre-operation comorbidities, induction agents, immunosuppressive agents, follow-up period, definitions and rates of complications, definition of graft failure, survival rate, and HR. Adjusted data were preferentially selected if available [20]. When the eligible studies failed to provide specific survival data and HR with 95% CI, the data in figures were extracted using Engauge Digitizer (version 10.11 http:// marku mmitc hell. github. io/ engau ge-digit izer/), a free publicly available software, and the HR with 95%CI was calculated using methods suggested by Tierney et al. [21].

Risk of bias in individual studies
The Newcastle-Ottawa Scales (NOS) recommended by the Cochrane handbook were adopted for quality assessment [22,23]. Giving that enrolled studies were There was statistically significant between SPK1 and SPK2 @ There was no statistically significant between SPK2 and KTA2 @@ There was statistical significant between SPK2 and KTA2 *** P < 0.05 for chi-squared tests comparing differences between SPK,P + , SPK,P − , DD-KA, and LD-KA groups; Meta-A: meta-analysis of synthesizing survival outcome of T2MD after SPK.
Meta-B: meta-analysis of summarizing hazard ratio between T1DM and T2DM after SPK, Meta-C: meta-analysis of synthesizing hazard ratio between SPK and KTA among T2DM) First Table S2. Two reviewers independently appraised the study quality. Disagreements between the reviewers over the risk of bias were resolved by discussions with a third reviewer (YXF).

Data synthesis
Survival rates and HRs were combined with the random effect model. All statistical syntheses and analyses were performed using STATA (15.0). Statistical heterogeneity was assessed using the I 2 statistic (< 0%: very slight heterogeneity; 30% to 60%: may represent moderate heterogeneity; 50% to 90%: may represent substantial heterogeneity; 75% to 100%: considerable heterogeneity) [24]. If the heterogeneity was above 50%, sensitivity analysis would be conducted. Evidence surrounding definitions of DM, baseline characteristics of recipients, definitions of graft failure, and complications which might play a role in survival outcomes was synthesized qualitatively. Funnel plots were not used to visualize the publication bias in each group since the number of included studies in the meta-analysis was less than 10 [22,25].

Search results
The search yielded abstracts for 1677 publications. After excluding duplicate articles and screening the abstracts, 1394 were remained for further review. Then, 153 copies of the full published version of each study were obtained, after excluding records which did not refer to SPK among recipients with T2DM in the title or abstract. Sixty-nine full texts were excluded next due to lack of survival outcomes of patients or grafts, leaving 16 eligible publications. Next, among the 16 studies, 9 were involved in the difference data synthesis process [11,[15][16][17][18][19][26][27][28]. Considering the study quality and data completeness on primary outcomes, 6 were included for synthesizing survival outcomes of T2MD patients after SPK [11, 15-17, 19, 26-28], 6 studies were included for summarizing the hazard ratio between T1DM and T2DM after SPK [15][16][17][18][19]28], 3 studies were included for synthesizing hazard ratio between the SPK group and the KTA group [16,26,28], and each meta-analysis had no overlapping samples. (Fig. 1, Table 1). The details of excluded studies with overlapping samples were in the SDC- Table S3.

Study characteristics
Characteristics of the included studies were presented in Table 1 and Table 2 (additional information in SDC- Table S4). All included studies were retrospective cohort studies. In accordance with the eligible criteria, all reports included a cohort of T2DM ESKD patients undertaking SPK. Considering the geographical coverage of the study, included studies were from the USA, Argentina, Germany, Austria, South Korea, and China.

Definition of T2DM and definition of renal failure and pancreas failure
As shown in Table 2, a variety of DM definitions has been witnessed among the included studies. Studies adopted center-specific criteria in selecting DM candidates for SPK [16,17,[26][27][28], which classified DM around the C-peptide level, BMI of 30 kg/m 2 , age, and pancreatic antibodies [15-17, 19, 29-31]. UNOS's definition of T2DM was based on the SPK transplant recipient registration form and the diagnosis of end-stage pancreas disease (ESPD) [15]. A consensus on the definition of renal failure among enrolled studies was witnessed, which was kidney retransplantation, returning to dialysis, or patient death. However, the definitions of pancreas graft failure varied. Most studies defined pancreas graft failure as insulin resumption, patient death, or pancreas graft removal.

Methodological quality of included studies
Methodological quality scores ranged from 4 to 8 on a modified scale of 0 to 11 (Table 3). A majority of studies showed good quality in patient selection and outcome assessment [11, 15-19, 26, 28]. The main heterogeneity between studies might arise from the sample size disparity, poor comparability of cohorts on the basis of study design or analysis, and insufficient reporting of follow-ups. Specifically, most studies [16-19, 27, 28] had a sample size of T2DM undertaking SPK below 100 and the number in UNOS studies [11,15,26] was more than 500; only 3 studies [15,26,27] reported adjusted hazard ratio on T1DM vs. T2DM or SPK vs. KTA; the majority of studies neither reported median or mean follow-up period, nor described the details of the follow-up-losses (Table 1 and
Since there was substantial heterogeneity in the 3-year survival rate analysis, sensitivity analyses were conducted and the result indicated that Fu et al.' s study [27] was the reason of heterogeneity which might be due to the short median follow-up period (SDC- Figure S1).

Discussion
Previously, Chan et al. has attempted to address the question about the controversy of conducting SPK on T2DM patients in a review which vaguely concluded that the efficacy of SPK for T2DM remained controversial in 2016 [32]. Al-qaoud et al. concluded that the outcomes of strictly selected T2DM recipients mirrored those of T1DM in a literature review [33]. Hitherto, no high-quality evidence was available for T2DM patients with SPK, neither the precise survival risks of T2DM patients undertaking SPK compared with T1DM SPK patients or T2DM KTA patients. With several studies from different countries emerging between 2016 and 2020 [17,18,[26][27][28], this study conducted a systematic review and meta-analysis to identify, collect, and synthesize all evidence reporting the survival outcomes of T2DM-ESKD recipients undertaking SPK worldwide, and make comparisons with their T1DM and KTA counterparts. This systemic review included 9 studies comprising 811 T2DM-SPK recipients. The meta-5-year survival rates of patients and kidney grafts were above 90%, and the meta-5-year survival rate of pancreas graft was 81%.
The survival outcomes of T2DM were identical to those of T1DM. For comparison of survival outcomes between SPK and KTA, the patients' and grafts' survival rates in the SPK group were superior to those in the KTA group. Although studies from a different geographical area with different organ distribution systems, the I 2 s were very low showing good homogeneity. The survival estimates of pancreas graft should be interpreted with caution given the various definitions reported in each program. Even though UNOS approved a standard definition in 2015 and the new policy was implemented in 2018, these were not reflected in the included studies [34][35][36].
The synthesized survival comparisons between T1DM and T2DM verified that the overall survival outcomes of T2DM recipients were comparable with those of T1DM, despite that the baseline characteristics of T1DM and T2DM were notably different, with T2MD recipients of older age and higher BMI [21,29,31,32]. The   [28]. Margreiter et al. [16] concluded that T2DM-ESKD patients with low coronary risk profile and age ≤ 55 years may have a favorable outcome from SPK. Previously, several attempts had been made for this issue. Sener et al. [38] had proposed criteria for pancreas transplantation in T2DM in 2010, suggesting C-peptide level, BMI, and pre-operation cardiovascular disease be considered. Dean et al. [38] mentioned T2DM with low insulin requirements would probably benefit from pancreas transplantation. Previously, mostly mentioned factors were C-peptide level, BMI, onset of DM [21][22][23], and recipient's age [24]. While in 2018, UNOS amended the qualifying criteria and abandoned the maximal allowable BMI and C-peptide, and the policy was implemented in 2019 [33,39]. Additionally, the novel subgroups of DM proposed in 2018 with a refined classification based on glutamate decarboxylase antibodies, age at diagnosis, BMI, HbA1c, and homeostatic model assessment 2 estimates of β-cell function and insulin resistance could also serve as a reference in making criteria [40]. For the comparison of SPK and KTA among T2DM patients, kidneys and patients' survival outcomes after SPK were more favorable than those after KTA. The meta-HRs were dominated by Alhamad et al.'s study [26] with a weight of about 80%. Alhamad et al.'s study [26] was a retrospective design based on the national database, with multiple factors adjusted in the survival analysis. However, some covariables like the duration of diabetes, insulin dose before transplantation, waiting time, and other factors reflecting diabetes-related comorbidities of recipients and donor factors which might be significantly different were not reported and adjusted. Therefore, the significant hazard ratio of KTA compared with SPK should be interpreted with caution. Prospective randomized studies which could control for confounders were still lacking.
Surgical, infectious, and immunological complications after SPK have been tricky issues for a long  Hau [28], 2020, Germany period [41][42][43][44][45]. The present analysis indicated that the T2DM group had a higher risk of renal graft DGF. DGF of kidney graft was reported to be significantly associated with weight [15]. Most T2DM recipients were overweight or obese compared with T1DM patients, which could cause a higher DGF risk. The rejection rate, infection rate, and DGF rate of pancreas graft were not significantly inferior. The estimates were limited by the insufficient description of definitions of each complication and definition of pancreas graft failure, further hindering the comparison between groups. As pointed out by Dean et al. [6], currently, a lack of uniform definition regarding complications limited the broader application of collected data. The integrated results about complication risk ratio should be interpreted with caution.

Strengths and limitations
This study has several strengths. The major one is that it is the first attempt to integrate the survival rate of patients and grafts of T2DM after SPK with software (Engauge Digitize) and HR calculation spreadsheet (by Tierney et al. [21]) with rigorous methodology. In addition, the studies included in this meta-analysis were drawn from a variety of countries that increased its applicability across many populations and organ transplantation centers. Next, there was slight between-study heterogeneity, indicating that estimates of mortality varied significantly beyond chance. There were some limitations, and the major one was that the absence of clear definitions of complications and pancreas graft failure hindered the interpretation of meta-results. Additionally, a small number of studies were enrolled in the meta-analysis of complications risk, and the meta-estimate should be interpreted with caution. Another main pitfall was that some included studies failed to provide multivariable-adjusted data, which might increase the risk of Type 2 error [46]. Besides, though capturing survival data from figures in articles made a quantitative synthesize of time-to-event data possible and had been used widely [47][48][49], still it would have a slight degree of error. Therefore, transparency of original studies is advocated.

Conclusions
The synthesized survival estimates of T2DM-ESKD patients after SPK were above 90%. Specifically, survival outcomes of T2DM patients are comparable with that of T1DM, and for T2DM, SPK is superior to KTA. However, a uniform criterion of T2DM subsets that would benefit the most from SPK and clearly defined diagnosis standards of SPK-related complications are urgent to be made.

Availability of data and code
The datasets or code used or analyzed during the current study are available from the corresponding author on reasonable request. More stars (*) indicate higher quality of the study. S1, representativeness of exposed cohort; S2, selection of nonexposed cohort; S3, ascertainment of exposure; S4, study was published within 5 years (after 2016); C1, comparability of the cohort on basis of design or analysis; O1, assessment of outcome; O2, was followup long enough for outcomes to occur ; O3, adequacy of follow-up; NA, not applicable; For details, please refer to SDC- Table S2. Author