Diagnosis of intestinal transplant rejection depends on clinical assessment, endoscopy and most importantly, histology of intestinal biopsies. Plasma citrulline levels (P-Cit) reflect functional enterocyte mass in nontransplant patients and have been evaluated in two small series after transplant. This study was designed to determine the sensitivity and specificity of P-Cit as diagnostic tool for allograft injury, especially to distinguish between viral enteritis and rejection. We prospectively collected 403 P-Cit samples within 24 h of intestinal biopsy in 49 patients. P-Cit levels were correlated with the mucosal damage and histopathological diagnoses. P-Cit levels in bowels with significant mucosal damage (i.e. moderate or severe rejection, viral enteritis, PTLD, ischemia reperfusion injury, allergic enteritis) were significantly lower than in intestines with no or mild injury (i.e. indeterminate or mild rejection, nonspecific enteritis): 22.9 ± 15.4 versus 38 ± 23.2 nmol/mL (p < 0.0001). Sensitivity and specificity of the test were 80% and 58.1% for rejection, and 56.5% and 66% for viral enteritis, thereby unable to distinguish between both entities. In conclusion, P-Cit reflects the extent of mucosal injury regardless of the etiology, but does not seem to be a predictive marker for rejection or viral enteritis, as its values may decline only when diffuse mucosal damage has occurred.
In 1981 Windmueller and Spaeth (1) confirmed previous observations postulating that the intestine is the main source of plasma citrulline (P-Cit). P-Cit is synthesized mainly from glutamine in the intestine, released from the enterocyte into the circulation and converted to arginine by the kidneys to complete the cycle (2). Furthermore, it was demonstrated that dietary amino acids are the major energy sources for the intestinal mucosa, playing also an essential role in maintaining its integrity (3).
In 2003, Crenn et al. described that post-absorptive P-Cit concentration is a marker of absorptive enterocyte mass in patients with intestinal failure. Humans P-Cit levels correlated with the remnant small bowel length, with 20 nmol/mL being the cut-off value for P-Cit distinguishing transient from permanent intestinal failure (4).
The diagnosis of rejection in intestinal transplant recipients depends on the combination of clinical assessment, endoscopy and most importantly, on the histology of intestinal biopsies. Frequent endoscopies and biopsies are potentially hazardous due to procedure-related complications. There are no serum assays and noninvasive tests available to assess mucosal allograft dysfunction (i.e. rejection, viral enteritis) analogue to plasma creatinine for kidney and serum transaminases for liver transplantation.
We (5) and others (6) proposed the use of P-Cit as such a marker. Pappas et al. demonstrated a correlation between P-Cit levels and the degree of intestinal rejection (6). We described that pathologies other than rejection, such as viral enteritis, can also affect P-Cit levels. We therefore proposed that P-Cit was not only a marker for rejection, but possibly a general indicator of intestinal mucosal dysfunction after transplantation (5). We also defined the normal range of P-Cit levels during the first year after intestinal transplantation (7).
The aim of the present study was to analyze the sensitivity and specificity of P-Cit as a diagnostic tool for histological allograft dysfunction, specifically to distinguish between viral enteritis and rejection, the two most common complications after isolated intestinal transplantation (IITx). This distinction is vitally important as both rejection and viral enteritis can result in terminal allograft dysfunction but require a completely different treatment approach.
Material and Methods
Between October 2001 and March 2005, we prospectively measured P-Cit in 49 intestinal transplant recipients within 12 h before or after endoscopic biopsies (i.e. before initiation of potentially required therapy). Biopsies were performed under our protocol (twice weekly for 6 weeks, once weekly for another 6 weeks, followed by every other week until 6 months, then monthly until 1 year posttransplant) beginning 6 days postintestinal transplant or when clinically indicated. Endoscopies were performed using a GIF XP-160 (5.9 mm) or GIF 160 (8.5 mm) (Olympus America Inc, Melville, NY, USA) endoscope for pediatric and adult patients, respectively. At least eight biopsies were taken from the distal 60 cm of the ileal graft.
P-Cit was measured using ion-exchange chromatography (Beckman Model 6300 Amino Acid Analyzer, Palo Alto, CA, USA), as previously described (7).
Biopsies were blindly classified as: (1) normal (N); (2) acute cellular rejection, subclassified as mild (MI), moderate (MO) and severe (S); (3) viral enteropathy (VE); (4) ischemia reperfusion injury (IRI); (5) nonspecific enteritis (NSE); (6) allergic enteritis (AE); (7) PTLD, based on established criteria (5). Biopsies have a low sensitivity and specificity for the diagnosis of VE (8–10). To increase the diagnostic yield for VE, we defined the diagnosis of VE in two ways: (1) histology only, (2) positive result of viral cultures, antigen detection or polymerase chain reaction (PCR) for virus in stool or serum. The term indeterminate (IND) for rejection was used for samples with slightly increased apoptotic body count that did not yet reach established criteria for mild rejection, or for apoptotic body count consistent with mild rejection but on a significant mixed cellular inflammatory background. Patients in this category were observed, not treated for rejection, and rebiopsied in 3–7 days depending on clinical symptoms.
All patients received a quadruple drug regimen consisting of tacrolimus, steroids, basiliximab and sirolimus, as previously described (7). Five adult patients who were transplanted despite a high panel reactive antibody titer and one patient who received a compatible mismatch after December 2003 received thymoglobuline, IVIg and mycophenolate mofetil in addition to tacrolimus and steroids. All patients received similar antimicrobial prophylaxis (ampicillin/sulbactam or piperacillin/tazobactam, fluconazole, ganciclovir) posttransplant. By protocol, enteral feeds were started between days 5 and 7 postintestinal transplant in all patients. In the early posttransplant period, pediatric patients received Vivonex RTF (5 g/dL of free amino acid) and the adults, Peptinex (5 g/dL of whey hydrolysate). Neither formula contains additional glutamine, no patient received glutamine supplementation.
Recipient body surface area (BSA) was estimated using Mosteller's formula (BSA m2= sqrt ((weight*height)/3600)). Patients were classified by BSA ≤1 or >1, as was previously described (7).
Values for descriptive data were expressed as mean ± SD. Student's t-test was used to compared P-Cit levels between no or mild mucosal injury (biopsies with none, IND, NSE or MI rejection) versus significant mucosal injury (biopsies with MO, S, VE, IRI, AE and PTLD). P-Cit in significant mucosal injury was analyzed in two groups according to BSA (≤1 or >1). Student's t-test was also used to compare P-Cit levels by type of viral disease. Pearson's and Spearman's rho correlation analyses were performed between P-Cit and each of the previously described diagnoses.
In addition, we analyzed P-Cit over time posttransplant comparing two groups: N, IND or NSE versus viral enteritis or rejection.
Sensitivity and specificity of the test was calculated overall, based on BSA and by time posttransplant. In addition, the ability of this biological marker to discriminate patients with normal pathology versus patients with disease was assessed by plotting receiver-operating characteristic (ROC) curves (i.e. sensitivity vs. 1-specificity) and reporting the estimated area under the curve.
Statistical analysis was performed with Software Package for Social Sciences (SPSS) for Windows (Release 12.0, Chicago, IL, USA). This study was reviewed and approved by the IRB of the Mount Sinai School of Medicine).
During the study period a total of 403 biopsies/P-cit pairs were collected from 49 intestinal transplant recipients. Three hundred seventy-eight biopsies were classified as no or mild loss of enterocyte mass (mucosal injury): N: 181 (45%); IND: 63 (16%); MI: 41 (10%); NSE: 93 (23%). Twenty-five biopsies were diagnosed as significant loss of enterocyte mass (mucosal injury): MO: 4 (1%); S: 8 (2%); VE: 5 (1%), based on histology only; IRI: 2 (0.5%); AE: 4 (1%); PTLD: 2 (0.5%).
The mean P-Cit level for the 403 samples obtained was 36.5 ± 23.0 nmol/mL. P-Cit levels in significant mucosal injury (i.e. moderate or severe rejection, viral enteritis, PTLD, ischemia reperfusion injury, allergic enteritis) were significantly lower than in no or mild mucosal injury (i.e. indeterminate or mild rejection, nonspecific enteritis): 22.9 ± 15.4 versus 38 ± 23.2 nmol/mL (p < 0.0001). These findings corresponded histologically also to the number of enterocytes undergoing apoptosis, expressed as number of apoptotic bodies per 10 contiguous crypt cross sections: quantitative increase of apoptosis in biopsies with significant mucosal injury when compared with no or mild mucosal injury (7.1 ± 8.1 vs. 2.8 ± 2.6 p = 0.0001).
Mean P-Cit levels corresponding to each histological diagnosis were as follows: no or mild mucosal injury: N: 36.5 ± 21.2 nmol/mL; IND: 37.1 ± 21.1 nmol/mL; MI 38.7 ± 29.2 nmol/mL; NSE: 40.1 ± 24.5 nmol/mL. Significant mucosal injury: MO: 20.7 ± 21.7 nmol/mL; S: 28.1 ± 16.9 nmol/mL; VE: 18.0 ± 24.0 nmol/mL (adenovirus: 1 and calicivirus: 4); IRI: 13.0 ± 7.0 nmol/mL; AE: 29.5 ± 11.7 nmol/mL; PTLD: 13.0 ± 8.4 nmol/mL. There was a significant difference between biopsies with viral enteritis versus normal histology or nonspecific enteritis (p = 0.015) (Figure 1A).
All patients with enteritic symptoms were also evaluated with viral cultures, antigen testing and PCR in serum and stool obtained at the time of the biopsy. If signs for VE were histologically absent (biopsies read as nonspecific enteritis, rejection, etc.), but one of the above viral markers became positive, diagnosis was changed to VE. Using these criteria, we observed an increased number of VE from 5 to 24 samples. The additional 19 samples of P-Cit that were shifted from previous histopathological diagnoses (N, IND, MI and NSE) (Figure 1B) were positive for adenovirus (6), calicivirus (12 samples) and rotavirus (1 sample). The P-Cit values changed after this observation and were used for the rest of the analyses as follows: N 36.7 ± 21.5 nmol/mL, IND: 37.7 ± 21.2 nmol/mL, MI 41.8 ± 30.8 nmol/mL, NSE 40.1 ± 24.7 nmol/mL and VE 23.3 ± 15.2 nmol/mL. These broader diagnostic criteria of viral enteritis including culture and PCR with negative histology resulted in an increased mean P-Cit level and the loss of the significance between viral enteritis versus normal or nonspecific enteritis. The histologically negative viral enteritis might represent an initial stage in the enterocyte injury without enough damage to be reflected in the P-Cit level (Figure 1B).
When we analyzed P-Cit within the group of patients with viral enteritis, we observed that carriers of calicivirus had significantly higher levels of P-Cit than patients with adenovirus enteritis (28.6 ± 15.4 nmol/mL vs. 11.0 ± 5.8 nmol/mL; p < 0.05). These findings reflect again the more pronounced cytopathic effect and enterocyte death during adenovirus infection when compared to the clinically milder calicivirus infection.
P-Cit is not a diagnostic tool to differentiate disease entities. The sensitivity and specificity of P-Cit in the diagnosis of VE in pediatric patients were 56.5% and 66%, respectively, the area under the ROC curve was 0.6. The sensitivity and specificity of P-Cit test for the diagnosis of rejection in adults were 80% and 58.1%, respectively, the area under the ROC curve was also 0.6.
Plasma citrulline has recently been found to be a reliable marker of functional enterocyte mass (2). It has been shown that patients with short bowel syndrome have a lower plasma citrulline level and that it not only correlates with the remnant small bowel length but also is able to discriminate between permanent and transient intestinal failure 2 years after resection (2). Furthermore, in patients with small bowel villous atrophy due to celiac disease and nonceliac disease the reduction of citrulline levels correlated in those patients with similar severity or extent of the histological lesions (4).
In order to identify a biochemical marker able to predict posttransplant intestinal dysfunction, we looked at citrulline as a noninvasive diagnostic tool to screen patients in whom endoscopy and biopsy would be indicated. An important difference between P-Cit in the nontransplant setting versus posttransplant is that patients with already established short gut or mucosal damage will usually not have a continuous process of ongoing mucosal damage and repair like transplanted intestines can undergo. This introduces a new variable of enterocytes metabolism and hence P-Cit levels.
Therefore, we established in our previous study normal range of P-Cit level after IITx and its evolution over time. We found that P-Cit is lower than expected during the first 3 months posttransplant when compared with nontransplant patients. After 3 months, P-Cit reached a plateau similar to healthy individuals (7). We also correlated P-Cit with different pathologies (i.e. rejection, viral infection) and demonstrated that P-Cit was lower in samples of patients not only with rejection but also with viral enteritis (5). The findings of Pappas et al. corroborated our data on rejection; however, viral enteritis was not examined in that study (8).
Several reports focused on the impact of viral infection in intestinal transplantation (9–12). It is well known that this entity mainly affects pediatric patients, but not exclusively, and can cause severe allograft dysfunction. It is essential to differentiate viral enteritis from rejection because a timely reduction of immunosuppression is sometimes the only way to cure the disease. Importantly, viral enteropathies, like adenovirus, can cause enterocyte apoptosis, the hallmark for the histological diagnosis of rejection (9–11). Intestinal transplant rejection is a focal process and it becomes diffuse only in severe exfoliative rejections. Viral enteritis on the other hand tends to be a more diffuse phenomenon at its inception, but the degree of disease varies between patients with enteritis without significant epithelial injury to patients with clear mucosal damage.
We examined if P-Cit levels could contribute to predict or differentiate viral enteritis from rejection. The data, however, demonstrated clearly that P-Cit was not a reliable diagnostic tool to distinguish different causes of mucosal injury, but that it correlated only with the extent of mucosal damage regardless of the underlying etiology. This is also illustrated by the fact that adenovirus enteritis with high cytopathic effects resulted in lower P-Cit than less aggressive calicivirus infections. Furthermore, P-Cit levels are initially very low in severe ischemia reperfusion injury, usually presenting with significant posttransplant bleeding and diffuse mucosal damage, but follow, after recovery, a similar curve as bowels without initial injury. In conclusion, P-Cit is a marker for intestinal mucosal injury and regeneration in the posttransplant setting until steady state is achieved (13).
Comparison of P-Cit between patients is difficult as baseline enterocyte mass varies over time. A more promising approach would be to serially (daily) measure P-Cit where patients can serve as their own controls. This would be feasible with the recently available P-Cit determination in blood dry spots (fingersticks) (14). Other potential markers like Perforin or Granzyme B have recently been proposed with a similar objective. Both apoptosis inducing molecules are released by activated cytotoxic T lymphocytes and NK cells, and are up-regulated in intestinal transplant recipients with acute cellular rejection. Therefore, it was thought that their concentrations may change before apoptosis occurs. Preliminary reports have shown an 84% specificity but only a 44% sensitivity for these markers to predict rejection (15). Calprotectin, a marker of T-cell-mediated inflammation, measured by ELISA in the stool of IITx patients has been found to be higher in patients with rejection than in cases of viral enteritis or normal biopsy with a sensitivity of 83% and a specificity of 77%. Elevation of baseline levels occurred 6–18 days prior to a biopsy proven rejection (16).
It is conceivable that a trend of slowly decreasing daily P-Cit levels in combination with an increasing Perforin or Granzyme B and a positive Calprotectin test in the stool may herald impending mucosal damage in an individual patient indicating the need for an endoscopy, thereby reducing the number of protocolized procedures.
Obvious limitations of our study were the small sample size. A larger cohort of patients, probably only in the setting of a multicenter trial, is necessary to conduct adequately powered trials in the evolving field of intestinal transplantation. The identification of one or a combination of biochemical markers with predictive value for intestinal dysfunction remains a priority.
To Dr. Martin Rumbo for his careful revision of the manuscript.