Chronic rejection (CR) of pancreas allografts needs to be accurately defined and diagnosed. We propose a grading scheme designed for percutaneous needle biopsies (C0–CIII). Grading is based on the semi-quantitative determination of fibrosis and corresponding proportional loss of exocrine parenchyma. Pancreas biopsies (n = 141) from 46 patients were studied. Twenty-six patients lost graft function after a mean time of 25.5 months, whereas 20 patients retained good graft function during a mean follow-up of 67.7 months. Sequential biopsies showed gradual progression of CR over time (p = 0.0001), and good correlation was found between the CR grade and the time elapsed from transplantation (p = 0.0001). The CR grade was predictive of the remaining time of graft function (54.3 months for C0, 24.6 months for CI, 9.7 months for CII and 1.6 months for CIII p = 0.00001). Preceding episodes of acute rejection (AR) were more frequent and more severe, and often occurred late in patients with graft loss due to CR (p = 0.04). Reproducibility among pathologists from different institutions was excellent for grades C0 (Kappa 0.9) and CIII (0.87), substantial for Grade CII (0.61) and moderate for Grade CI (0.59). The proposed grading scheme provides a reliable and reproducible tool for the assessment of CR in percutaneous needle biopsies, with definite prognostic significance.
Biopsy-proven chronic rejection (CR) is the largest single cause of late pancreas allograft loss (1–3). The clinical presentation of CR is nonspecific, with loss of glycemic control being the main feature. Hyperglycemia may develop progressively or may be unmasked by infection or other physiologic stresses (1). However, the clinical usefulness of this feature is limited since with the development of hyperglycemia due to chronic rejection, the beta cell function is in general already irretrievably lost (1). The diagnostic specificity of hyperglycemia is also problematic, because this abnormality may be caused by other processes in addition to CR. These include development of insulin resistance and islet cell toxicity due to tacrolimus or cyclosporine use (4). An increase in pancreatic enzymes in serum is also of limited value due to its lack of specificity for the accurate diagnosis of the cause of graft dysfunction (5). In addition, with advanced CR the pancreatic enzymes are often within normal limits (1). The available measures of monitoring insulin secretory reserve, such as postprandial glucose tolerance, first-phase insulin release following glucose or arginine simulation, and the slope of glucose disappearance following glucose tolerance testing, although proven to be useful in specific circumstances, are difficult to apply to the routine clinical assessment of pancreas transplant recipients. A decline in C-peptide levels might reflect the progressive loss of insulin secretory reserves, but could also result from potentially reversible drug toxicity (e.g. calcineurin inhibitors) (6–9).
Thus, there is no clinical marker for the monitoring of progressive loss of pancreas functional reserve, comparable to the serial measurements of serum creatinine or glomerular filtration rate in kidney transplantation. Pancreas graft sclerosis can be demonstrated by imaging studies, indicating progressive decrease in graft size on ultrasound and MRI or ultrasound evidence of decreased parenchymal perfusion (10). As with other transplanted organs, percutaneous pancreas needle biopsies represent the gold standard for the diagnosis of graft dysfunction and in order to guide appropriate therapeutic interventions and determine prognosis (11–20). We have previously shown that the progressive nature of CR in the pancreas allografts can be assessed through the evaluation of serial biopsies (21).
The histological features of chronic pancreas rejection have been well described in animal models (22–24). As in other vascularized grafts, CR is characterized by progressive fibrosis and chronic arterial vascular lesions. Untreated acute allograft rejection in canine and rodent models is followed by interstitial fibrosis, that begins in the septa (perivascular areas). The fibrosis is associated with mononuclear inflammation and progressive disappearance of the acinar cells. Eventually, most of the graft is replaced by dense fibrous tissue in which rare residual ducts and islets are embedded. The islets of Langerhans are affected only when the fibrosis is very advanced (25,26). Clinical/pathological studies have shown that in addition to the presence of progressive fibrosis and proportional acinar loss, vascular changes characterized by narrowing of arterial lumina and concentric fibroproliferative endarteritis are an integral part of the pattern of pancreas chronic rejection (3,27,28). These vascular changes are similar to those observed in kidney and heart transplants. An additional important histological feature associated with pancreas graft sclerosis (chronic rejection) is the presence of recent and organized thrombosis in arteries and veins (3).
Pancreas transplant percutaneous needle biopsies have been used routinely for the evaluation of graft dysfunction for several years (5,10–20). Similar to other solid organ transplants (29–31), the correct diagnosis and grading of chronic rejection in pancreas transplants has potentially important prognostic implications. Based on preliminary studies, we had previously proposed a 5-point grading scheme (C0-IV) for pancreas chronic allograft rejection in needle biopsies (21). Although overall prognostically useful, over time we found that this initially proposed scheme had reproducibility problems in the intermediate range (grades CII and CIII of the previous scheme) (32). Also, these intermediate grades presented occasionally out of the expected sequence of progression in serial biopsies, and based on the graft outcomes they appeared to identify a similar or identical group of patients. We concluded that these initially proposed intermediate grades represented not true pathological progression but rather random sampling variation. Using the results of the current study, we propose a clearly defined and simplified (4-point) grading system. With this study, we specifically sought to determine the usefulness of the histological grading system to assess the severity of CR in relationship to: (a) duration of subsequent graft survival; (b) progression of sclerosis in consecutive biopsies; and (c) relationship to previous acute rejection (AR) episodes. (d) We also assessed the reproducibility of the histological grading between pathologists from different centers.
Materials and Methods
Forty-six randomly selected patients were included in the study. They were divided into two groups: group A with 26 patients [12 simultaneous pancreas kidney transplant (SPK), 9 pancreas after kidney transplant (PAK), 4 pancreas transplant alone (PTA) and 1 simultaneous pancreas-living donor kidney transplant (SPLK)] who lost their grafts after an average time of 25.5 months (±23); and group B with 20 patients (13 SPK, 6 PAK, and 1 PTA) with functioning grafts after an average follow-up of 67.7 months (±19.4). Graft loss was defined as absence of glycemic control and return to insulin dependence.
Group A included 95 biopsies (1–8 per patient, mean 3.6 ± 1.9), whereas group B patients had 46 biopsies (1–8 per patient, mean 2.3 ± 1.7). All biopsies were evaluated by two pathologists (JCP, CBD) blinded to all clinical information. Acute allograft rejection was graded from 0 to V according to the previously described scheme (11). Chronic allograft rejection was graded from 0 to III based on the evaluation of the proportion of fibrosis and associated acinar atrophy/loss. A description of the proposed chronic rejection grading scheme used in the current study is presented in Table 1 and depicted in Figures 1–4.
Table 1. : Chronic Rejection Grading Scheme
(a) Inflammatory infiltrates associated with ongoing acinar cell injury, venous and/or arterial endotheliitis and ductal inflammation indicate active acute allograft rejection. Acute and chronic rejection should be graded independently based on the key histological features specified on each grading scheme (e.g. A IV, C II).
(b) Transplant arteriopathy closely parallels the degree of fibrosis, i.e. is absent in grade 0, minimal in grade I, mild to moderate in grade II and severe in grade III. Arterial branches may be absent in biopsy material; therefore the grading can be most consistently done by assessing the fibrosis and acinar atrophy/loss. Changes in islets are highly variable and not useful for grading.
Normal pancreas. The fibrous component is limited to normal septa and its amount is proportional to the size of the enclosed structures (ducts and vessels). The acinar parenchyma shows no signs of atrophy.
Expansion of fibrous septa; the fibrosis occupies less than 30% of the core surface. Most of the acinar lobules are reserved in their central areas, however, they show focal erosion and irregularity of their contours.
The fibrosis occupies 30–60% of the core surface. All acinar lobules show some degree of fragmentation and atrophy nd patchy drop-out. The exocrine atrophy affects the majority of the lobules in their periphery (irregular contours) but also in their central areas. The latter change appears as thin fibrous strands criss-crossing between individual acini.
The fibrotic areas predominate and occupy more than 60% of the core surface; only isolated areas of residual acini and/or islets are identified.
The pancreas biopsies studied were performed between January 1992 and January 1998. The main indications for biopsy were as follows: increase in serum amylase and lipase (31%), decrease in urinary amylase (43%), hyperglycemia (21%) and protocol surveillance (5%). The immunosuppression schemes were as follows: all patients received 10–14 days of induction therapy with either ATGAM (Upjohn) or OKT3 (Orthoclone, OrthoBiotech). Maintenance therapy was initiated when the nasogastric tube was discontinued (PAK, PTA) or when a dropping serum creatinine clearly indicated renal transplant function (SPK). Triple maintenance therapy consisted of either cyclosporine (Sandimune or Neoral) or tacrolimus (Prograf), prednisone and azathioprine (Immuran) or mycophenolate mofetil (Cellcept). Target blood concentrations of cyclosporine or tacrolimus in the immediate postoperative period were 300–400 ng/mL and 12–20 ng/mL, respectively; by 1 year the target levels were tapered to 200 ng/mL and 8–10 ng/mL, respectively. Rejection episodes were treated with 500 mg of intravenous methylprednisone followed by a taper period over 2 weeks. ATGAM (Upjohn) or OKT3 (Orthoclone, OrthoBiotech) was administered at standard doses for 10–14 days according to clinical parameters. The first episode of either minimal (AR grade II) or mild (AR grade III) acute rejection was treated with corticosteroids. Recurrent episodes of rejection of any grade as well as moderate or severe rejection (AR grades IV and V) were treated with a combination of corticosteroids and ATGAM or OKT3.
For the statistical analysis, the Student t-test, Fisher exact test, chi-square statistic and anova were employed as appropriate. SPSS (Chicago, IL) was the statistical computer program used for this study.
The reproducibility of the system was analyzed by grading of digital images by seven pathologists. Specifically, two digital images (one of the whole core, X20) and one medium power view (×200) from 10 cases were transmitted to the participating pathologists, who were provided with the grading system criteria as seen in Table 1. The participating pathologists were blinded to all other data. Statistical analysis was performed using the kappa-statistic (33).
Relationship of histological grading of chronic allograft rejection and graft loss
All biopsies (100%) of the patients in group B (i.e. good graft function after a mean follow-up of 67.7 months) lacked any significant fibrosis or acinar loss. All of these biopsies were classified as corresponding to Grades C0 and CI exclusively. In contrast, in the biopsy pool from patients who lost their grafts (Group A), the biopsies showed almost equal numbers of low and high grades of CR [Grades C0, CI 54/95 (57%) and Grades CII, CIII 41/95 (43%), respectively]; there was significant difference between groups A and B (p = 0.0001). The higher grades were observed at later post-transplant time points with evidence of ongoing progression in serial biopsies (see below). Overall, for all patients the histological degree of CR correlated well with the time elapsed from graft implantation, occurring at an average of 4.87 ± 6.7 months for grade C0, 9.35 ± 7.2 months for CI, 12.9 ± 8.6 months for CII and 17 ± 9.9 months for CIII (p = 0.0001) (Figure 5).
A direct correlation was also noted between the histological grade and average remaining time of graft function: 54.3 ± 27.5 months for C0, 24.6 ± 27.11 months for CI, 9.7 ± 10.34 months for grade CII and 1.6 ± 1.5 months for CIII (p = 0.00001) (Figure 5).
Relationship of chronic allograft rejection/graft loss to previous episodes of acute allograft rejection
Both groups A and B showed similar numbers of biopsies with no acute rejection [AR grades 0 and I 29/95 (30.5%) in group A and 14/46 (30.5%) in group B] or minimal/mild acute rejection [AR grades II and III 50/95 (53%) in group A and 29/46 (63%) in group B]. In contrast, the patients who lost their grafts (Group A) had more preceding biopsies with moderate or severe acute rejection (AR grades IV and V), 16/95 (17%) compared to 3/46 (6.5%) in group B (p = 0.11, NS). Also, the patients in group A had more episodes of histologically confirmed acute rejection per patient (group A: 68 episodes, i.e. 2.6 per patient, and group B: 32 episodes, i.e. 1.6 per patient).
The incidence of late (>1 year) acute rejection was significantly more frequent in group A, where 10/26 patients (38.5%) had repeated episodes of late acute rejection. In contrast, only 4/20 (20%) of patients in groups B had an isolated episode of late acute rejection (p = 0.04).
Histological progression of chronic rejection in consecutive biopsies
The CR histological grade in consecutive biopsies: (a) remained the same in 46 (48.5%) subsequent biopsies [29 (42%) in group A, 17 (65.5%) in group B]; (b) worsened in 40 (42%) of subsequent biopsies [35 (51%) in group A and 5 (19.5%) in group B]; (c) showed a lower grade of CR (i.e. sampling related ‘improvement’) in 9 (9.5%) subsequent biopsies [5 (7%) in group A and 4 (15%) in group B]. Overall the progression of fibrosis from earlier to subsequent biopsies was statistically significant (p = 0.0001).
The majority of patients in group A showed a progressive histological deterioration of the CR grade, as exemplified in a typical case of graft loss due to chronic rejection (Figure 6). This case also highlights the common pattern of recurrent episodes of moderate-severe acute rejection, including episodes of late acute rejection, in patients who lost graft function due to CR.
Reproducibility of the proposed chronic allograft rejection histological grading scheme
The agreement level (kappa statistic) for the seven participating pathologists was as follows: for C0 0.9 (almost perfect), CI 0.59 (moderate), CII 0.61 (substantial) and CIII 0.87 (almost perfect).
With the marked improvements achieved in the short-term outcome of pancreas allografts, mostly due to the decrease in the occurrence of AR, late graft loss and CR have become increasingly important issues to address (1). Whereas in the first months post transplantation graft losses are more often related to surgical complications, idiopathic thrombosis, AR and peripancreatic infections (3,34), after the first 12 months, CR – with or without superimposed AR – accounts for the majority of graft losses (3). In contrast to AR that presents with sudden graft dysfunction and can be prevented or successfully treated in the majority of cases, CR is characterized by a slowly progressive decline in graft function and does not respond well to treatment. Similarly to kidney allografts, in the pancreas CR is manifested histologically as graft sclerosis and loss of functional parenchyma (3,21–24). In many cases, progressive fibrosis and acinar loss are observed over a prolonged period of time in serial biopsies. The purpose of this communication is to describe a simple and reproducible scheme for the histological assessment of CR in pancreas allograft needle biopsies.
In the kidney the proposed histological schemes for the assessment of chronic damage have shown a definite prognostic value (29–31). Our results in the current study show similar prognostic value for the grading of CR in pancreas allograft needle biopsies, i.e. low CR grades are associated with lengthy graft survival, and high CR grades point to a limited time of remaining graft function. Furthermore, despite its notoriously patchy nature, the progression of pancreas allograft fibrosis can be reliably assessed through serial biopsies. A clear difference in the histological deterioration was seen between the patients who eventually lost their grafts vs. the ones that retained graft function. The majority (51%) of the subsequent biopsies in the former group showed a higher CR grade than the preceding biopsy, whereas the majority (65.5%) of the subsequent biopsies in the latter group showed persistence of the same grade sequentially. In either group, only a small part of the patients showed a factitious improvement in the degree of fibrosis in subsequent biopsies, thus indicating that sampling variation does not represent a significant problem overall.
A further goal of our study was to address the relationship of AR to CR in pancreas allografts. In the kidney it has been demonstrated that the risk for CR increases with the occurrence of interstitial and vascular rejection episodes, particularly the latter (35–37). AR episodes occurring late rather than within the first 3 months are also associated with increased risk of CR in the kidney (38,39). In pancreas allografts, late AR (occurring after 12 months post transplantation) is not unusual and appears to occur more often in solitary pancreas transplants (40,41). In the current study we observed that repeated episodes of AR, higher grades of AR, and late (>1 year) AR were all associated with an increased risk of CR and graft loss. The fact that more frequent and more severe episodes of AR lead to CR indirectly validates also the previously published AR histological grading scheme (15).
In all organs, the issue of potential lack of reproducibility of histological grading in allograft biopsies has been of concern (42). The CR grading scheme proposed here appears to be reproducible, as shown by examination of images by pathologists in different centers. Reproducibility was moderate to excellent for all four grades, among participating pathologists.
In addition to the possibility of CR and insulin resistance, with the appearance of hyperglycemia in pancreas transplant recipients, recurrence of the original disease (diabetes) should be considered in the differential diagnosis (43–45). However, neither in this study nor in previous studies have we had the opportunity to observe evidence of recurrent autoimmune mediated diabetes mellitus (25). Although, clinically, patients with recurrent disease as well as those with progression of CR will present with loss of glucose homeostasis, the histological picture in these two conditions is very different. Whereas in CR the presence of pronounced fibrosis is the typical finding and the residual islets contain both identifiable alpha and identifiable beta cells, in patients with recurrent disease, there is no significant fibrosis and the diagnostic findings consist of selective loss of beta cells with or without insulitis, depending on the stage of the process. In the absence of inflammation the parenchyma may appear normal, and the diagnosis cannot be made without the aid of insulin and glucagon immunostains to identify selective loss of beta cells (45). Tissue evaluation is thus the only way to make the diagnosis with certainty (i.e. recurrent disease vs. CR).
The current study does not address the question of the pathogenesis of the CR in the pancreas. It is expected that in the pancreas as in other allografts, the fibrosis represents the end effect of cumulative injury (or injuries) of diverse origins, immunological and nonimmunological. It is unclear if in the pancreas similar mechanisms operate for the propagation of tissue damage once a critical amount of parenchymal mass is lost, as is the case in the kidney (46). The main histological findings in CR (septal fibrosis, acinar loss) are very similar to those of chronic pancreatitis in native organs, and this same grading scheme could potentially be used for the grading of chronic pancreatitis. From a pathophysiological point of view, the progressive tissue loss observed in chronic pancreatitis appears to be also the result of multiple factors, including ductal obstruction by concretions, altered pattern of protein secretion (e.g. lithostathine), oxidative stress (47), growth factors over-expression, activation of inflammatory mediators, deregulated immune responses, and altered nerve growth with specific neuroimmune interactions (48–51). Also, microcirculatory disturbances have been heavily implicated (52,53). Thus, the progress of pancreatic pathology in the native organ is apparently complex, self-perpetuating and with significant similarities to the chronic allograft rejection/sclerosis. It is conceivable that in addition to the immune-mediated acinar cell and vascular injury, similar mechanisms to the ones operating in chronic pancreatitis are likely to play some role in the pathogenesis of graft sclerosis. Accordingly, in order to cover the spectrum of pathogenetic possibilities resulting in chronic pancreatic allograft damage, particularly in patients in whom a clear history of preceding episodes of acute rejection cannot be elicited, we propose the more general term ‘chronic pancreas allograft rejection/graft sclerosis’.
The role of vascular injury in pancreas allograft pathology is unequivocal. We have previously demonstrated that vascular thrombosis superimposed on acute (endotheliitis) or chronic (transplant arteriopathy) arterial damage is associated with parenchymal fibrosis/sclerosis (3). Although arterial changes are typical of chronic pancreatic rejection, in this scheme we rely principally on the fibrosis and acinar loss for the grading of CR, due to the relatively unreliable presence of arterial branches in pancreas allograft biopsies.
Based on this retrospective study of a relatively limited number of patients, we conclude that the proposed grading scheme provides a valuable tool for the diagnosis and quantitation of the progressive pancreas allograft fibrosis that finally leads to graft failure. The grade of chronic pancreas allograft rejection/graft sclerosis has important long-term prognostic implications independently of the probably multifactorial etiology of the process. The conclusions of this work should be ideally validated with a prospective study based on protocol biopsies.
We wish to thank Dr Olga B. Ioffe and Dr Kathleen Romaine for participating in the reproducibility part of the study.