Citrulline has been advocated as a marker for acute cellular rejection (ACR) in intestinal transplantation; however, its significance as a forewarning in the long-term follow-up remains unknown. This study aimed to investigate the association between citrulline levels and the grading of ACR to establish a cutoff point that accurately predicts ACR beyond 3 months posttransplant in the pediatric patient population. During a 16-year period (1995–2011), a total of 13 499 citrulline samples were prospectively collected from 111 consecutive pediatric intestinal/multivisceral transplant recipients: 2155 were obtained concurrently with intestinal biopsies. There were 185 ACR episodes observed among 74/111 (67%) patients (median follow-up: 4.4 years). Citrulline levels were inversely proportional to the severity of ACR. Negative predictive values for any type of ACR (cutoff, 20 μmol/L) and moderate/severe ACR (cutoff, 10 μmol/L) were 95% and 99%, respectively. When patients were divided according to graft size, diagnostic accuracy using the same cutoff was identical. Similarly, subgroup analysis by the timing of citrulline measurement prior to biopsy varying from 1 to 7 days demonstrated comparable results. Citrulline is a potent indicator as a danger signal for ACR, being an exclusionary, noninvasive biomarker with excellent negative predictive values in the long term after pediatric intestinal/multivisceral transplant.
High incidence of acute cellular rejection (ACR) remains Achilles’ heel of intestinal and multivisceral transplantation and its early detection and treatment is paramount for long-term survival (1). To date, frequent endoscopy and biopsy of the intestinal allograft is the only way to identify ACR before it becomes severe and irreversible (2). Although this is the gold standard for the diagnosis of ACR, the procedure carries risks of bleeding and perforation, requiring experienced hands.
Recently, serum citrulline has been advocated as a candidate noninvasive biomarker for ACR in intestinal transplant with conflicting results (3–7). In different conditions such as short bowel syndrome, villous atrophy diseases and intestinal damage due to myeloablative therapy, citrulline levels has emerged as an innovative quantitative biomarker of functional enterocyte mass reduction (8–10). The main source of citrulline is intestinal de novo synthesis within the enterocytes (11). It is an end product of glutamine metabolism through a complex sequence of the glutamate-to-ornithine pathway in enterocyte mitochondria (12,13). Citrulline is released into to the circulation from enterocytes and then converted to arginine by the kidneys (11,14). The enzymes of this incomplete urea cycle (mitochondrial but not cytosolic) are upregulated when the body has to spare nitrogen: thus citrulline is described as a key player in nitrogen homeostasis and protein synthesis (15). We reported that citrulline is a useful marker for detecting moderate or severe ACR in a study including 37 pediatric and 20 adult intestinal recipients (7). However, the sensitivity and predictive value for detecting mild ACR were limited and insufficient citrulline data obtained immediately prior to biopsy with relatively small number of rejection episodes precluded the assessment of citrulline as a forewarning (7,16).
In the current study, we expanded our body of data among the pediatric patients and aimed to investigate the association between citrulline levels and the grading of ACR. Our goal is to establish a clinically practical cutoff point that accurately predicts ACR in the longterm (beyond 3 months) following pediatric intestinal/multivisceral transplant. This is important in three aspects. First, the blood concentration of citrulline is significantly low during the first days of life and increase with aging in healthy adults, but its level in the pediatric population is largely unknown (14,17,18). Second, because citrulline levels gradually increase posttransplant and they achieve a plateau only after 90 days, when the recovery of the intestinal allograft from ischemia-reperfusion injury takes place (5,19). Third, the patients always have a standard ileostomy during this early posttransplant period and they stay in close vicinity to the transplant center; noninvasive markers are not compelling since endoscopic surveillance can be performed easily and safely. If we succeed to define diagnostic criteria based on the citrulline values that determine who demands urgent endoscopy with histopathologic examination, it will become a big stride toward noninvasive surveillance and diagnosis of ACR in intestinal/multivisceral transplant.
Material and Methods
During a 16-year period (1995–2011), we prospectively collected serum citrulline levels in all consecutive pediatric (under 18 years of age) transplant recipients including an intestinal allograft. Citrulline samples were obtained according to the following protocol: biweekly during the first month posttransplant, then weekly for 3 months, and monthly thereafter (6,7). In addition to routine measurements, citrulline levels were checked whenever intestinal biopsies were performed for surveillance or under clinical suspicion of ACR and 1–2 times weekly during ACR episodes. Endoscopic surveillance with biopsies were performed 2–3 times weekly during the first month postoperatively, followed by once weekly for 2–3 months, then monthly until 4–6 months, and once every 3–6 months subsequently. During ACR episodes, endoscopies were done as frequently as every other day (1). Citrulline levels were reviewed retrospectively; therefore, they did not have any influence on the timing of biopsies. Diagnosis of ACR was defined as the presence of clinical symptoms with histopathologic evidence confirmed by the biopsy as well as the initiation of antirejection treatment. Standard grading scale was applied for histopathologic evaluation of ACR: no ACR, indeterminate for rejection, mild, moderate and severe (20). The maximum grade of rejection observed by biopsy was matched to the serum citrulline level that was retrieved at the closest possible date of endoscopic examination (range, within 14 days). If there were multiple citrulline concentrations available, the value immediately prior to the date of biopsy was selected. Since previous publications suggested that the citrulline level is a marker of functional intestinal mass, we used donor body surface area (matched to the recipient) as an approximation for the intestine allograft size (8,19). The estimated glomerular filtration rate was calculated using the Schwartz equation (21). To investigate the effect of individual variability, we assumed that the ‘baseline’ citrulline level for each patient would be represented as the average of the values that corresponded to no ACR by histopathologic examination. The fall in citrulline levels during ACR episodes was expressed as a % reduction from the ‘baseline’. Diagnosis of other posttransplant conditions, i.e. viral enteritis, posttransplant lymphoproliferative disease (PTLD), graft-versus-host disease (GVHD) and chronic rejection, was determined on the basis of preceding reports (5,22–25). This study conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was conducted under the approval of the University of Miami Institutional Review Board (#20080229).
Values were expressed as mean ± standard deviation or median (range) when appropriate. Student's t-test was used to compare the means of two groups. Spearman's rank correlation coefficient provided a distribution free test of independence between two variables. Receiver-operating characteristic (ROC) analyses were performed to assess the discriminative power of citrulline for detecting any type of ACR and moderate/severe ACR. Statistical significance was defined as a p value < 0.05.
A total of 111 pediatric patients were identified. Indications for intestinal transplant were gastroschisis (N = 29), necrotizing enterocolitis (N = 20), Hirshsprung's disease (N = 12), megacystis microcolon (N = 9), midgut volvulus (N = 9), intestinal atresia (N = 9), intestinal pseudo-obstruction (N = 8), microvillous inclusion disease (N = 6), tumor (N = 3), trauma (N = 2) and others (N = 4). Background information of the patients is displayed in Table 1. The smallest infants weighed 3.7 kg (N = 2). None of the patients had chronic kidney disease stages 4 or 5 at the time of transplant. Vast majority of the study subjects (81%) were recipients of a liver allograft from the same donor. Portal venous drainage of the allograft was employed in most of the cases (91%). Daclizumab was the mainstay of induction immunotherapy up until 2008 and it was switched to antithymocyte globulin thereafter. All patients received tacrolimus and steroids for maintenance immunosuppression. During a median follow-up period of 4.4 years (range, 4 months–16 years), there were 185 acute rejection episodes observed among 74/111 (67%) patients. The median number of rejection episodes per patient was 1 (range, 0–15). Eighty-one/111 (73%) patients had at least one episode of bacteremia. PTLD developed in 19/111 (17%) patients and most of the lesions were found in the gastrointestinal tract (N = 14). GVHD was observed in 18/111 (16%) patients: the skin was involved in all cases, accompanied by gastrointestinal manifestations in 2 and pulmonary in 1.
Table 1. Patient demographics
*Data presented as mean ± standard deviation for continuous variables and median (range) for discrete variables.
CMV = cytomegalovirus.
2.6 ± 3.5*
60 (54): 51 (46)
Body weight, kg
11.7 ± 9.8*
Body surface area, m2
0.48 ± 0.28*
Donor/recipient body surface area ratio
1.06 ± 0.33*
Pretransplant creatinine, mg/mL
0.36 ± 0.17*
Intensive care unit
Human leukocyte antigen match
CMV donor positive/recipient negative
Type of transplant
Liver–intestine +/− pancreas
Cold ischemia time, minutes
432 ± 84*
Venous drainage of the allograft
These patients provided a total of 13 499 citrulline levels. Of these, 2155 samples met the study inclusion criteria that were collected at a median time of 16 months (range, 3–179 months) posttransplant. Citrulline levels were measured before biopsy within 24 h in 68.5% (1477/2155) of subjects and within 24–72 h in 7.0% (150/2155). Only 1.5% (33/2155) was collected beyond 1-week interval from the time of biopsy. The median number of citrulline samples contributed per patient was 11 (1–167). Strong correlation was found between the level of citrulline and the severity of rejection. It was inversely proportional and as the citrulline levels decreased the rejection grade worsened (Table 2). Figure 1 illustrates the distribution of all 2155 citrulline levels divided into five categories in an increment of 5 μmol/L and their association with the histopathologic grades of acute cellular rejection. When rejection of any degree was present, 94.7% (482/509) of citrulline samples showed levels less than 20 μmol/L; however, substantial overlap was observed with episodes of no acute cellular rejection or indeterminate for rejection, 73.4% (1208/1646) of them falling below the 20 μmol/L line. ROC analysis revealed that this cutoff point of 20 μmol/L provided a negative predictive value of 94.5% with a negative likelihood ratio of 0.19 for detecting rejection of any degree (sensitivity, 95.1%; specificity, 26.3%: Figure 2A). If we set the cutoff point at 10 μmol/L, the citrulline levels equal to or higher than 10 μmol/L excluded moderate or severe rejection with 98.5% probability (negative likelihood ratio, 0.19; sensitivity, 87.7%; specificity, 63.0%: Figure 2B).
Table 2. Cirtulline levels by the grade of rejection
Number of samples (%)
*Data presented as mean ± standard deviation.
17.4 ± 9.5
15.0 ± 9.7
Acute cellular rejection
10.0 ± 6.2
6.5 ± 4.0
5.5 ± 3.1
When the patients were divided into two groups according to the donor body surface area (< 0.5 m2 vs. ≥ 0.5 m2), the mean citrulline level of the smaller patients was significantly lower compared with the larger patients (13.7 ± 9.0 μmol/L vs. 15.5 ± 10.1 μmol/L, respectively, p < 0.001). There was a strong, positive correlation between donor and recipient body surface area (ρ= 0.666, p < 0.001). However, at a cutoff value of 20 μmol/L, the negative predictive values for all rejections in the smaller and larger patient groups were identical (95.1% and 94.1%, respectively). To estimate the significance of citrulline levels as an ‘alarm’ of any type of rejection, subgroup analysis was undertaken by limiting samples that were retrieved at three distinct intervals; within 24 h, 24–72 h and within 1 week before intestinal biopsies. Negative predictive values for all rejections were 92.7%, 95.6% and 92.7%, respectively, comparable to the results of the entire cohort. During the study period, only two patients suffered chronic kidney disease stages 4 or 5 posttransplant (both of them underwent deceased donor kidney transplant); therefore the correlation between citrulline levels and kidney function was not evaluated.
Next, the drop in citrulline levels from the ‘baseline’ for each patient was recorded individually (number of samples available for ‘baseline’ measurement, 8 ± 7 per patient). A 25% reduction yielded a negative predictive value of 69.0% with a negative likelihood ratio of 0.39 for detecting all ACR (sensitivity, 71.6%; specificity, 73.7%). Increasing the threshold to 50% resulted in less accuracy, excluding all ACR with merely a 57.3% chance (negative likelihood ratio, 0.64; sensitivity, 40.8%; specificity, 92.8%).
The mean citrulline levels for other etiologies were the following: viral enteritis (N = 7), 14.7 ± 11.7 μmol/L; PTLD (N = 16), 15.0 ± 8.1 μmol/L; GVHD (N = 3), 17.2 ± 16.2 μmol/L; chronic rejection (N = 5), 6.1 ± 3.1 μmol/L. Only the patients with chronic rejection had significantly lower citrulline values compared to those with no ACR (17.4 ± 9.5 μmol/L, p < 0.001). Among all ACR episodes, the incidence of concurrent diagnosis with one of the above was 2.4% (12/509 samples).
This is the largest cohort study to date on prospectively accrued, pediatric intestinal transplant recipients demonstrating citrulline as a very useful marker of ACR and its inverse correlation with the severity of rejection in the long term (beyond 3 months posttransplant). Our previous works (6,7) were successfully validated by the data presented herein that were collected in a series of greater than 100 pediatric patients with close to 200 episodes of ACR, observed over an extensive period of time (up to 16 years). One of the limitations of our preliminary results was that we were not able to evaluate the timeliness of citrulline data as a forewarning of rejection due to lack of sufficient number of rejection episodes with paired citrulline samples drawn just before the biopsy (7,16). To circumvent this problem and assess the feasibility of citrulline to accurately ‘predict’ rejection in clinical practice, we carefully matched the date of citrulline measurement to the date of intestinal biopsy so that the corresponding histopathologic diagnoses followed citrulline values in a chronological order. Our subgroup analysis verified that the preceding citrulline levels of up to 1 week before biopsy were clinically meaningful as an exclusionary marker for both all types of ACR (cutoff point, 20 μmol/L) and moderate or severe ACR (cutoff point, 10 μmol/L) with their excellent negative predictive values ranging from 93% to 99%. Of note, the discrepancies between these two cutoff points and our earlier result (the value of 13 μmol/L for detecting moderate or severe ACR) (7) were that the latter picked up the highest specificity/sensitivity combination, while we selected the cutoff parameters with a negative likelihood ratio of less than 0.2 in this series. Because the primary endpoint of the present study was to find a useful test to “rule out” ACR in order to avoid unnecessary endoscopies with maximum safety, the determined citrulline cutoff levels had to generate at least a moderate, if not conclusive, change from pretest to posttest probability (26). We have a low threshold for erroneously “ruling in” patients with no or indeterminate for rejection for further evaluation (moderate/low positive predictive value) taking into account the consequences of “overlooked” ACR. Therefore, although the percentage decline from the individual ‘baseline’ was not as useful as the cutoff points mentioned above, it still should be an alternative index. We propose the following criteria for performing a biopsy in an otherwise stable pediatric patient beyond 3 months after intestinal/multivisceral transplant: citrulline levels < 15 μmol/L or ≥ 25% reduction from the “baseline”. The intestinal allograft is very unforgiving and the margin of error is diminutive; moreover, the positive predictive value of citrulline is not very high. Given these restrictions, we believe it is prudent to confirm the diagnosis with a biopsy whenever possible. If a biopsy is not readily available, in the interest of time, citrulline at a low/reduced level could be used to initiate antirejection treatment, just like how elevated liver enzymes are applied in liver transplant and so is creatinine in kidney transplant.
We investigated two potential confounders of citrulline values, the size of the intestinal allograft and pre/posttransplant kidney function. The donor body surface area used as an approximation for the functional intestinal mass was directly proportional to the citrulline level, which was in accordance with a previous publication (19). However, the diagnostic accuracy did not show significant improvement when the patients were divided by their donor body surface area above or below 0.5 m2. In the pediatric patient population, citrulline level adjustments for body size are probably not needed for most patients. With regard to kidney function, none of the patients had severe impairment pretransplant and only two went into end-stage kidney disease posttransplant, requiring kidney transplant at 4.4 years and 11 years after intestinal transplant. Currently, we have 45 long-term survivors (> 5 years after intestinal/multivisceral transplant) and still growing in the pediatric patient subset: we will continue to monitor their kidney function and citrulline values. Other prognostic variables can theoretically affect citrulline levels, but our study subjects were primarily homogeneous in terms of operative techniques and immunosuppression regimens, and life-threatening posttransplant complications were either common (73% of patients had documented bacteremia) or infrequent (< 20% with PTLD or GVHD); thus their impact on citrulline values were considered to be trivial from the clinical standpoint, where citrulline should be used as an ‘exclusionary’ marker of ACR. Indeed, the mean citrulline values for viral enteritis, PTLD and GVHD (calculated from a limited number of samples) were similar to that for no rejection, further supporting this argument in part. Nevertheless, we found significantly lower citrulline levels in patients with chronic rejection, which is underdiagnosed and reported due to insufficient clinical data (27). Longer follow-up and accumulation of evidence is required to delineate the role of citrulline in its diagnosis.
Recently, we reported promising novel measures using gene expression profiles or donor-specific antibody levels for detecting acute rejection in intestinal transplant (28,29). Other authors described intriguing data on T cell alloresponses, stool calprotectin and molecular markers (30–34). The ultimate goal is to increase the positive likelihood ratio and specificity by integrating citrulline with these novel diagnostic tools on the horizon to the level of definitive diagnosis. If an algorithm comprising noninvasive surveillance and screening as well as confirmation of acute rejection is established, it will obviate the need for the endoscopic biopsy, the current gold standard for the diagnosis of ACR. At the present time, final diagnosis lies solely in the hands of transplant pathologists and only a prospective multicenter study will allow us to reach a general consensus.
A drawback of this study is that the citrulline samples were not obtained together with precise clinical information, i.e., presence/absence of symptoms, and we were not able to see if citrulline could predict asymptomatic rejection, before the injury to the intestinal allograft has become irreversible. It has been also suggested that there is a relatively wide intraindividual variability in citrulline readings (19), but because our series did not include the citrulline samples that were obtained beyond the 2-week mark of endoscopic biopsy for the study purpose, we do not know if the “baseline” citrulline level we used as an approximation for a patient in steady state truly serves as a control for each patient in the diagnosis of ACR. Prior reports have demonstrated that liver function did not significantly influence citrulline concentration, that liver transplant failed to correct acitrullinemia in an inborn error of metabolism, and that net flux of citrulline across the liver was not significantly different from zero (35–37). Consequently, the potential bias introduced by our patient population (81% received liver allografts) is deemed negligible.
In conclusion, serum citrulline level is a potent surrogate indicator which significantly correlates with the severity of ACR of the intestine in 3 months or longer follow-up after pediatric intestinal/multivisceral transplants. It is an excellent exclusionary biomarker and a danger signal for any type of ACR (cutoff point, 20 μmol/L) and for moderate or severe ACR (cutoff point, 10 μmol/L) with their negative predictive values ranging from 93% to 99%.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.