Pancreatic graft thrombosis (PAT) remains one of the major surgical complications and causes of graft loss in pancreatic transplantation. The reported incidence ranges from 1% to 40% [1–3] as the entity of thrombosis, ranging from partial to complete, and its extension, diagnosis, and treatment are still not well defined. In addition, partial thromboses are often underestimated, even though they are potential precursors of complete thrombosis [4, 5]. In this case, their early detection could be essential to prevent graft failure. Ultrasound and/or computed tomography angiography (CTA) are usually used to detect PAT, either routinely or when clinical symptoms develop [5–8]. The usefulness of systematic PAT detection using CTA is still debated [7–9]. Hakeem et al [10] recently proposed the Cambridge Pancreas Allograft Thrombosis (CPAT) grading system (Figure 1), which provides prognostic and therapeutic recommendations. The authors reported their experience of PAT in 103 patients who received pancreas transplantation between 2014 and 2017. In this study, CTA was performed only for biochemical/clinical reasons but not routinely. PAT was retrospectively graded on the basis of CTA to identify the risk of graft loss and outline a management algorithm through a retrospective review of these cases.

The aim of the present study was to retrospectively assess CTA examinations performed routinely in simultaneous pancreas kidney (SPK) transplantation recipients to implement the CPAT grading system [10] and to study its association with the recipients’ outcomes.

In total, 344 consecutive first-time SPK transplantations were performed during the study period (from February 2005 to December 2019).

At the retrospective reading by the two readers, 215 of the 319 patients had PAT on CTA, with a thrombosis grade of 1, 2, or 3 in 106, 85, and 24 patients, respectively. The 104 remaining patients had no sign of PTA (grade 0). The thromboses were diagnosed as arterial in 86 patients, venous in 51 patients, or mixed in 78 patients (Table 1).

CTA signs of PAT were found in 215 patients, while 104 did not show such signs. These two groups of patients were compared. The patients who did not undergo CTA were not included in this comparative study (Figure 2).

There was no difference between the patients with or without PAT on CTA in terms of the donor’s age, BMI, cause of death, anoxia brain damage, cardiac arrest, and period spent in the ICU (Table 2). The only statistically significant difference was a larger proportion of donor men in the thrombosis- vs. the non-thrombosis group (73.5% vs. 62.1%; p = 0.039).

There was no difference between the two groups in terms of the recipient’s gender, age, BMI, duration of diabetes, dialysis status, and number of HLA mismatches (Table 2); preservation solution; cold ischemia time; anastomosis time; and operative procedures (Table 3).

As shown in Table 4, graft survival in the first 30 post-operative days was significantly lower in the thrombosis group (pancreas loss was 14.4% in the thrombosis group vs. 2.9% in the non-thrombosis group, p < 0.002). The association between graft loss due to PAT and grade of thrombosis proved highly significant (17/25 graft losses occurred in patients with grade 3 thrombosis vs. 7/25 with grade 2 thrombosis in the first 30 post-operative days, p = 0.0000 with the chi-square test). Whatever the cause of pancreatic graft loss, it was significantly correlated to the grade of thrombosis (Figure 3). Graft losses due to PAT occurred within the first 5 post-transplant days.

During the first 30 post-operative days, there was no difference between the two groups in the number of kidney graft losses, deaths, and ICU and hospitalization days (Table 4).

There was a statistically significant difference in the number of thrombotic complications and re-interventions between the groups, (11.2% in the thrombosis group vs. 1.9% in the non-thrombosis group, p < 0.001 and 14.4% vs. 1%, p < 0.001, respectively), and a significantly higher number of transplantectomies in the thrombosis group (13% in the thrombosis group vs. 0% in the non-thrombosis group, p < 0.001) (Table 5). There was no difference between the two groups regarding the number of other surgical complications (35.6% in the non-thrombosis group vs. 34.9% in the thrombosis group) or surgical re-exploration (34% in the non-thrombosis group vs. 34.4% in the thrombosis group). There was no significant difference between the incidence of surgical complications or surgical re-explorations or graft loss due to bleeding between the thrombosis group, which received anticoagulation, and the non-thrombosis group (Table 5).

Two patients in the non-thrombosis group suffered from peritonitis which prompted re-intervention and PAT detection.

As shown in Table 5, the risk of pancreatic graft loss during the follow-up was higher in the thrombosis group (27.9% vs. 18.3% in the non-thrombosis group, p < 0.052). This result was also confirmed by the Kaplan-Meier analysis (Figure 4).

The main cause of pancreatic graft loss was thrombosis (13% in the thrombosis group vs. 1% in the non-thrombosis group, p < 0.001); other causes included bleeding, peritonitis, acute and chronic rejection, and diabetes recurrence, with no significant difference between the two groups (Table 5).

Patient survival was not correlated to pancreatic graft thrombosis (Figure 5).

The multivariate analysis (Cox proportional hazard model) showed that the risk of pancreatic graft loss was significantly associated with the recipient’s age, the development of hyperglycemia, hemorrhage, abdominal pain, and thrombosis grade 2 or 3, while there was no increase in the risk of graft loss in recipients with PAT grade 0 or 1 (Table 6).

Only in the 12 patients who developed PAT intraoperatively or during the first post-operative hours (they were not included in the thrombosis group) there was a correlation between PAT occurrence and donors’ age and the recipients’ duration of diabetes. The Kruskal-Wallis analysis showed that the donor’s age was significantly higher in patients who developed PAT intraoperatively than in those who developed it after the first post-operative days but within 30 post-transplantation days (41 vs. 32 years, p = 0.02) or in the patients who did not develop PAT (41 vs. 29 years, p = 0.01). Similarly, the duration of the recipient’s diabetes was significantly higher in the patients who developed PAT intraoperatively than in those who developed PAT after the first post-operative hours but within 30 days (32 vs. 26 years, p = 0.01) or in the patients who did not develop PAT (32 vs. 24 years, p = 0.01).

This study addressed PAT that occurred following SPK transplantation in a large series of patients who underwent transplants at a single center. The study included only patients with type 1 diabetes and end-stage renal disease who received a SPK transplantation for the first time, in order to exclude additional risks of PAT.

To our knowledge, this is the first study that implemented the CPAT grading system in clinical practice after Simonis SA et al [11] assessed the applicability and the reproducibility of this system.

The large majority of the recipients (319/344, i.e., 93%) underwent systematic CTA to detect early signs of thrombosis. CTA was not performed in 25 recipients, 21 of whom had lost their pancreatic allograft intraoperatively or within the first hours after the transplantation, and 4 of whom had shown poor renal function recovery.

Although there is no consensus on when systematic CTA should be performed [8–10], we decided to perform it within the first 7 post-operative days or sooner when the patients presented signs of complications (i.e., hyperglycemia). CTA was chosen for its high specificity and sensitivity, and non-operator dependence [8, 12–14]. It was well tolerated without a significant decrease in renal function [9–11]. It was not performed only in a few patients to avoid further kidney injury. PAT was detected by CTA except in 12 patients who developed it in the operating room. Moreover, it was not diagnosed by protocol CTA in two patients, in whom PAT was detected in the operating theater during a re-operation for other causes.

In the present study, the incidence of PAT was high because all the recipients underwent CTA and all grades of thrombosis were considered, contrasting with the majority of studies where CT scans were not performed routinely in all recipients but merely in those showing graft dysfunction, or following the appearance of symptoms [8–10]. Moreover, in the study of Simonis SA et al [11], 80%–90% of the re-analyzed CT scans showed signs of thrombosis.

In the present study, the retrospective analysis of CTA showed 106 grade 1, 85 grade 2, and 24 grade 3 thromboses, which were all included in our analysis, while grade 1 thromboses were not considered in the majority of the previous studies [8–10].

The demographic data of the two groups (thrombosis and non-thrombosis) did not show significant differences, except for the higher proportion of male donors in the thrombosis group (73.5% in the thrombosis- vs. 62.1% in the non-thrombosis group) and a higher incidence of thrombosis in patients with Roux-en-Y enteric drainage, but the number of these patients is too small to be considered informative. Shahrestani S et al [15] also found that the risk of thrombosis increased by 25.6-fold in the case of male donors. Interestingly, the donor’s age and the duration of the recipient’s diabetes were significantly associated with the risk of developing PAT only in the 12 patients who developed it intraoperatively or during the first post-operative hours. These risk factors have been reported in many studies [6, 16–19], but in our study, we found a significant association between them and the occurrence of PAT only in these 12 patients.

The majority of the grade 1 thromboses were arterial (81.1%), while the thromboses graded 2 or 3 were either venous or mixed (77.1%).

In the present study, patients with grade 1 thrombosis had a favorable course (none of them lost their graft of PAT). Indeed, the survival analysis showed that the risk of graft loss was the same in recipients with grade 0 or grade 1 thrombosis; conversely, patients with PAT grades 2 or 3 were at a significantly higher risk of graft loss due to PAT (7/25 and 17/25, respectively) compared to patients with grades 0 or 1. Moreover, even though there was no significant difference between the two groups in the number of surgical complications, whatever the cause of pancreatic graft loss (including bleeding and pancreatitis), the risk was significantly associated with PAT grades 2 and 3.

Currently, no standard protocol exists that is able to consistently prevent thrombosis of the arterial or venous anastomosis sites or within the extension grafts following transplantation [9, 10, 20, 21]. In the present study, the patients did not receive prophylaxis with LMWH but were treated for a few days with IV heparin before undergoing the protocol CTA [20–23].

The retrospective review of the CTA scans and the use of the grading system allowed us to grade the thromboses and address the management of PAT. Indeed, only 7/85 (8.2%) of the patients who developed grade 2 thrombosis and 17/24 (70.8%) of those who developed grade 3 thrombosis lost their pancreatic graft. Although the indications for anticoagulation remain to be studied [7, 10, 17, 23–25], we suggest treating patients with grade 2 thrombosis with LMWH and VKA for 3–6 months but not introducing specific treatment (VKA) for grade 1. However, treatment only with LMWH in grade 1 could be recommendable. Despite the low success rate, surgical/endovascular management has to be considered in grade 3 thrombosis [26, 27] followed by VKA. Moreover, careful donor selection and prophylaxis with LMWH in preventing thrombosis could be useful. In the present study, 12 patients lost their pancreatic graft in the immediate post-operative period before performing CTA and starting any anticoagulation treatment. This group of patients is important, rendering necessary a better knowledge of donor and recipient characteristics (i.e., thrombophilia abnormalities) to identify the high-risk patients before transplantation (i.e., in the group of patients not on dialysis).

Our study has some limitations. Firstly, the study is retrospective and includes patients who underwent transplants over a long period of time by different surgeons with different experience and we have to consider that some cases of PAT might be associated with the surgical procedure. Moreover, some difficulties were experienced in the implementation of the CPAT grading system, particularly in the differentiation between grades 0 and 1, already experienced by Simonis SA et al [11].

In conclusion, the CPAT grading system was successfully implemented in a large series of SPK transplantations and showed its applicability in clinical practice. We suggest an early protocol CTA to detect PAT and a large prospective study introducing subgrouping in the CPAT system to better establish clear indications for PAT prophylaxis and treatment (28).