Urological obstructive complications (UOC) affect approximately 6% to 15% of patients after kidney transplantation (KTX) and occur in the immediate postoperative phase or within the first months.1-3 Although potentially reversible, they are associated with significant morbidity and may considerably affect the onset/evolution of adequate graft function.4 Most postrenal complications can be ruled out reliably by routine ultrasonography (US).5,6 However, particularly in the early posttransplant period, the expected high accuracy of US when excluding UOC may be reduced due to prolonged periods of delayed graft function (DGF) or decreased urinary output leading to false-negative sonographic imaging. Without hard evidence of high-grade hydronephrosis, the benefits of the immediate application of radiological imaging with higher sensitivity (ie, contrast-enhanced tomography) must be weighed against their potentially detrimental nephrotoxic effects. The exact prevalence of these—by current standards—“occult (=without relevant hydronephrosis)” UOC is unknown as the factors mentioned impede early diagnosis. The availability of further diagnostic tools would support the use of imaging methods with higher sensitivity to early diagnose “occult” UOC.
In this respect, renal allograft biopsy findings may provide essential differential diagnostic information as part of the routine evaluation of allograft dysfunction. However, a distinct histological phenotype indicating UOC is ill defined. In the early posttransplant period, biopsy findings are dominated by acute tubular injury (ATI) as a result of ischemia/reperfusion or by acute rejection features. A promising biopsy feature related to UOC may be the finding of tubular ectasia (TE), which has been shown to be an indicator of increased intratubular pressure in animal models of unilateral ureteral obstruction.7 In fact, in our center, the renal pathologist suspects occult UOC when TE occurs together with one or more of the following distinct tubular histomorphological features: acute tubular cell injury (tubular cell flattening and/or vacualization, loss of brush boarder) and tubular protein casts. In native kidneys, TE has been associated with ATI and worse renal function, while in smaller scale studies of renal allografts TE has been described as a nonspecific feature of DGF.8,9
However to our knowledge, there is no published large scale investigation regarding the value of TE or any other histomorphological features as an indicator of occult UOC or its association with renal outcomes.
This retrospective analysis was designed to evaluate the relationship between the finding of TE including other histological features and clinical endpoints in the early posttransplant phase as well as their accuracy for diagnosis of occult UOC.
MATERIALS AND METHODS
We included all recipients with early indication biopsy (n = 976 out of 1537 consecutive allografts; biopsy performed within first 6 mo for allograft dysfunction or proteinuria; all biopsies having adequate material for interpretation of tubular morphology) and KTX from January 1999 to December 2007 at the Medical University of Vienna (see Figure 1). We have chosen a common definition of early allograft biopsy10 in order to assess the time varying peaks of both the occurrence of DGF and early urological obstruction (which can also occur 100 days after Tx).11The patients’ medical records were used as source data concerning baseline variables, graft histology, transplantation-specific characteristics, and endpoint analysis. Our clinical routine at the time period consisted of antirefluxive ureteral implantation via extravesical submucosal tunneling, ureteral anastomosis with absorbable sutures, and no routine ureteral stent placement (UST). Renal ultrasonography has been performed on day 1 and every second day until achievement of adequate renal function and upon de novo transplant dysfunction. Relevant hydronephrosis (>grade I) was considered a contraindication for percutaneous biopsy and was routinely excluded with renal sonography.
All study procedures have been approved by the local ethics committee, assuring adherence to the Declarations of Helsinki and Istanbul.
Histological Features of UOC
All biopsy results were screened for the presence of TE suspicious of UOC as stated by the renal pathologist in the original histopathological report. The entire biopsies undergo formalin-fixed paraffin-embedded processing and all specimens are immediately fixed upon retrieval. Moreover, all biopsies undergo a rapid processing procedure at the pathology department, which is also available on weekends with resulting fixation times between 6 and 18 hours. For interpretation of tubular morphology, we used hematoxylin and eosin and periodic acid-Schiff staining according to standard methodology. TE considered by the nephropathologist as indicative of UOC was observed in 63 patients. Two patients with the necessity of intermittent self-catheterization before KTX and 3 patients (lost to follow-up) were excluded. Biopsies that were initially classified as suspicious for UOC were subsequently reevaluated by a single pathologist blinded to the clinical results (H.R.) in order to increase accuracy of tubular histomorphology suggestive of UOC. The reevaluation included severity of TE (I to III), overall degree of ATI (I to III), presence of tubular protein casts (>0%), and tubular vacuolization (>5%). Figure 2 depicts distinct histological features indicative of UOC.
TE was considered present if observed >5% of cortical tissue with at least mild severity. Since neither tubular diameters nor the height of epithelial cells is physiologically uniform, the definition of TE cannot simply be expressed as tubular diameter in micrometer. To express the complex relation of true tubular dilatation and/or epithelial cell attenuation in TE in a systematic but yet simple way, we chose to estimate the tubular lumen in relation to the vertical size of epithelial cells. TE severity was graded mild if the tangential diameter of the tubular lumen measured >2 times but <4 times the thickness of the adjacent tubular epithelial cells. A dilatation of >6 times the epithelial cell thickness was graded severe. Moderate were the cases in between mild and severe. The categorization was based on the dominant cortical phenotype. Interobserver concordance was assessed between an expert in renal allograft pathology and a trainee (H.R. and J.K.)
The term protein casts comprises all kinds of stainable, homogenous, finely granular material in tubular lumen and was considered positive when observed. Tubular vacuolization was defined as the appearance of optical empty cytoplasmic vacuoles in more than a few occasional epithelial cells per tubular cross section. No distinction was made between the so-called isometric or the other types of vacuoles. With this definition, we excluded protein reabsorption droplets. We categorized ATI as present based on the following lesions: epithelial cell flattening, detachment from the basement membrane, loss of brush boarder in proximal tubules, vacuolization of the cytoplasm, and tubular cell necrosis. As a sign of a low grade ATI, we considered flattening, loss of brush boarder, and vacuolization. We concluded a more severe lesion if cells were detached or necrotic (grade II and III). To estimate the overall injury, we also considered the overall area of affected tissue. This means tubular necrosis observed in a few cross sections might still have been graded as moderate injury and less advanced damage like loss of brush boarder and/or epithelial cell flattening that, however, affected the entire biopsy and might have been considered as indication of severe injury.
The outcomes measured were if TE classified as suspicious of UOC by the pathologist were as follows: (1) associated with “occult”, that is without high-grade hydronephrosis in renal ultrasonography (>grade I) UOC; and (2) related to DGF (defined as dialysis dependency within the first wk after KTX) or worse estimated glomerular filtration rate (eGFR) at 1 year (according to the Mayo equation12). UOC were defined as ureteral stenosis, ureteral necrosis, urinary leaks, and obstructions due to perinephric hematoma or perinephric lymphoceles within 4 weeks after biopsy and necessity of therapeutic intervention.
To assess the relevance of TE as an indicator of UOC, we performed a separate case-control study and compared patients with TE to randomly matched biopsied patients without TE. Controls were identified from the baseline cohort (n = 976) and were matched for the year of transplantation, donor age, and cold ischemia time. For 3 of the 58 subjects with TE, no suitable control was available (none of those had a UOC), leading to 55 cases versus 55 controls. Additionally, treatment data, eg, UST, percutaneous nephrostomy (PCN), ureterocystoneostomy (UCN) were assessed.
Continuous data are presented as mean with standard deviation. Categorical data are presented as absolute count and percentage. We used Chi-squared test and Fisher exact test to analyze associations between categorical data and Mann Whitney U test to compare median values between the groups. To estimate the association of TE with DGF or occult UOC, univariate and multivariate logistic regression models were applied. Potential confounders were considered, if there was evidence of a statistical trend (P ≤ 0.2) between the variable and the outcome. We evaluated the following potential risk factors: TE, sex, donor age >60 years, cold ischemia time >14 hours, living donor, combined organ transplantation, recipient of prior allograft, latest complement-dependent cytotoxic panel reactive antibodies >10%, initial immunosuppression consisting of Cyclosporine A or Tacrolimus, initial immunosuppression consisting of an inhibitor of the mammalian target of rapamycin, early T-cell-mediated rejection, early C4d-positive active antibody-mediated rejection (AMR), and underlying vesicoureteral reflux as cause for end-stage renal disease. We categorized features of tubular injury to maximize their diagnostic accuracy and the sign was chosen to indicate a positive association with the outcome. We assessed the interobserver concordance for mild versus high grade TE with Kappa statistics. Because of a low prevalence of high-grade TE and a high probability of a prevalence-related bias (prevalence index = 0.6; values of 0 are considered having no prevalence related bias) in the interpretation of Kappa statistics, we have used the modification of Cohen’s kappa according to Byrt et al as recommended.13,14 We used multivariable logistic regression with UOC as the binary outcome (yes versus no), and histopathological features of tubular injury in binary categories as covariables. We investigated all first level interactions, tested for it using the likelihood ratio test, and entered interaction terms into models as appropriate. We used goodness-of-fit tests, calculated the Receiver operating characteristic (ROC)–area under the curve (AUC), and displayed ROC of the final multivariate models, according to standard methodology.
IBM SPSS Statistics (Version 24) and STATA (Version 14.0) were used for the statistical calculations. A two-sided P < 0.05 was considered as statistically significant.
In the current retrospective analysis, 976 renal allograft recipients with early indication biopsy were included (Figure 1). Overall, the rate of early ureteral stenosis or urinary leak/necrosis was 9.4%. The majority of the organs were of deceased donor origin (88%) with a median donor age of 51 years. The median recipient age was 52 years with a high prevalence of prior allograft recipients (18%) and presensitized patients (20%). Baseline immunosuppression mostly consisted of a calcineurin inhibitor (cyclosporine 61%, tacrolimus 30%). Early T-cell-mediated rejection and C4d-positive active AMR were diagnosed in 37% and 14%, respectively. DGF was observed in 29% of the included subjects.
TE suspicious of UOC was observed in 58 patients (5.9%) with a median of 13 days after KTX, in 4 of them after the first 3 months. Patients with TE were more frequently on tacrolimus therapy (42% versus 29%; P = 0.037) and had a significantly lower human leucocyte antigen -mismatch (P = 0.02), compared with patients without TE. Further baseline characteristics are shown in Table 1.
Tubular Ectasia Is not a Risk Factor for DGF or Lower eGFR
The relationship between TE and DGF was evaluated using all subjects with early biopsy (n = 976). Patients with TE showed a trend toward a higher rate of DGF with an odds ratio (OR) of 1.61 (P = 0.087).
In a subsequent multivariate logistic regression analysis including confounding variables as shown in Table 2, TE was not significantly related to DGF, while retransplantation and higher cold ischemia time remained independently related to DGF.
Moreover, TE was not related to worse eGFR after 1 year: median 50 mL/min/1.73 m2 (interquartile range [IQR]: 35–75.5) versus 51 mL/min/1.73 m2 (IQR 32–69); P = 0.62.
UOC and Their Treatment
Next, we analyzed the association of TE with UOC. For this purpose, we performed a case-controlled study as described in the Methods section.
Of the 55 patients with TE, 25 patients (45.5%) were diagnosed with UOC versus 24% without TE. Figure 3 depicts the distinct complications and their treatments. The most frequent UOC in the TE group was ureteral stenosis (48% versus 7% in the control group, P = 0.002). The rate of other UOC was not different between groups.
The initial treatment consisted mostly of UST (60%) and/or PCN (56%). Some patients with ureteral stenosis and all patients with ureteral necrosis/urinary leak were subjected to UCN (60%). Group comparisons of treatment strategies are shown in Figure 3. We observed a higher rate of PCN (P = 0.007), UST (P = 0.05), and UCN (P = 0.05) in the TE group compared to the control group.
Tubular Ectasia Is a Risk Factor for UOC
In the case–control analysis, TE was significantly associated with a more than 2-fold increased risk for occult UOC (OR 2.69 [IQR: 1.19–6.09]; P = 0.018). In a next step, we assessed further risk factors expected to be associated with occult UOC (=without high-grade hydronephrosis in renal ultrasound) in the early posttransplant period (see Table 3). However, none of the included covariables were significantly related to UOC including active Banff single lesions (ptc, g, i, t, v; data not shown). A trend was observed for early C4d-positive active AMR (OR 2.78; P = 0.2). In a subsequent multivariate logistic regression model including TE and early C4d-positive active AMR, TE remained independently related to UOC (OR 3.71 [IQR: 1.37–9.99]; P = 0.01).
Histomorphology of UOC
As TE was not the only feature of tubular injury triggering suspicion of UOC, all the biopsies that were historically classified as suspicious for UOC were reevaluated by a single pathologist (H.R.) in order to increase diagnostic accuracy. TE was considered present if observed >5% of cortical tissue with at least mild severity in the majority of the affected area.
Interobserver concordance has been assessed by an expert and a trainee in renal allograft pathology. The rate of interobserver agreement for mild (n = 42) versus high-grade TE (n = 11) was 69% with a fair interobserver concordance (κ = 0.36).
The reevaluation of biopsies included severity and extent of TE (I to III), degree of ATI (I to III), presence of tubular protein casts (>0%) and tubular vacuolization (>5%). The extent of cortical TE ranged from 5% to 70% of the biopsy specimen. There was no significant association between the dichotomized variable of the extent of TE according to the median of 30% and UOC (P = 0.35). While mild ATI correlated with UOC, higher degrees of ATI were inversely related to UOC. Comparing severity of TE, we found that mild TE was related to UOC, which was not the case for grade II/III TE. More severe TE was found in a higher rate in subjects with DGF: 46% (5/11), higher donor age (donor age >60 y): 30% (3/10), and higher cold ischemia time (>14 h): 60% (6/10) versus mild TE DGF: 36% (16/44), donor age >60 years: 19% (8/43), and cold ischemia time >14 hours: 50% (22/44).
Tubular protein casts and mild tubular vacuolization (<30%) were both related to UOC, which showed a significant first level interaction in the model. Subsequently, an interaction term was defined as the presence of tubular protein casts without tubular vacuolization, which was independently related to UOC. The final multivariate model included mild TE, mild ATI, mild tubular vacuolization, and tubular protein casts without tubular vacuolization and is shown in Table 4.
The ROC analysis of the final multivariate model showed a highly significant AUC for UOC (0.77, P = 0.001) (see Figure 4). The multivariate model showed a specificity of 78%, sensitivity of 61%, a positive predictive value of 67%, and a negative predictive value of 73%.
This is the first large-scale study investigating histomorphological features of tubular injury in the renal allograft and their relation to occult UOC. The current study provides for the first time the evidence that distinct tubular histomorphology, in particular TE, is associated with occult UOC. Moreover, we have developed a histomorphology-based multivariate model for exclusion of occult UOC with a high specificity and negative predictive value.
Specific histological characteristics indicative of UOC are ill defined in KTX. Most promisingly, TE has been described as a feature of urinary obstruction in animal models of unilateral ureteral obstruction.7,15,16 From a pathophysiological view, the dilatation and injury of the tubuli can be explained by the accumulation of urine and consecutive increase of intratubular pressure. In the current study, the finding of TE “suspicious of UOC” was independently related to occult UOC in support of the biological plausibility of TE as a marker of UOC also in humans. Routine US has misclassified all these cases while the current data provides evidence that distinct biopsy features have pointed toward UOC as a cause for renal dysfunction. The association of TE with UOC was mostly driven by a higher incidence of ureteral stenosis when compared to the control group, which is in line with the described animal models on unilateral ureteral obstruction. The second most common complications were ureteral necrosis and/or urinary leak. Accordingly, the most common interventions were placement of a ureteral stent and/or PCN and/or performance of an UCN.
A major strength of our analysis is that in order to increase diagnostic accuracy of features of tubular damage, all biopsies that were initially categorized as suspicious for UOC were reevaluated by one single pathologist. Several features of tubular morphology that might be indicative of UOC were systemically analyzed and graded. The final multivariate model showed that mild TE, mild ATI, mild tubular vacuolization, and tubular protein casts without tubular vacuolization were related to occult UOC. The c-characteristic of the final multivariate model was highly significant with a high specificity and negative predictive value for subsequent UOC. These findings provide a valuable clinical decision tool, which, however, needs to be confirmed by future prospective multicenter studies, and should also include interobserver correlation analysis and additional measures such as renal scintigraphy and/or Whitaker test. Interobserver concordance analysis of the current cohort showed a fair interobserver agreement, which was similar to established lesions in allograft pathology such as peritubular capillaritis.17,18 Moreover, our data underline also feasibility of the scoring approach amongst experts and trainees in renal allograft pathology. However, the assessment of the full spectrum of interobserver aspects of TE should be ideally analyzed in a prospective cohort. Although the conditions of tissue fixation were uniform in the current study, an important aspect in the study design for a multicenter trial to increase comparability between centers is standardization of time-to-fixation and fixation length; both these factors may considerably affect tubular morphology and its interpretation.
Of note, in animal models distinct histomorphology develops in a retrograde manner, starting in the collecting duct, connecting tubule, and distal tubule and then spreading to proximal parts.15 This pathophysiology may partly explain why the extent of the TE was not related to UOC in the current study, and potentially the extent of TE in the renal medullary area may also represent a subtle diagnostic criterion for occult UOC. We also aimed to analyze the renal medullary area but only a few biopsy samples contained renal medulla (data not shown).
The results of this large scale analysis argue strongly against a dominant role of TE as a nonspecific feature of DGF, as suggested in previous smaller scale studies.9 Although TE showed a trend to a higher rate toward DGF in our patients, this finding could not be confirmed in subsequent multivariate models. Furthermore, TE was not associated with reduced eGFR. This is in line with reports that described UOC, contrary to vascular complications, are not associated with long-term graft outcome.19
When interpreting our data on UOC, it is important to acknowledge that most studies define those as ureteral stenosis and ureteral necrosis. The rate of ureteral stenosis and leak/necrosis was 9.4% overall and 10% in the control group, which is similar to the previously published cohorts.1,4,20 In contrast, however, we used a broader definition of UOC based on their clinical relevance and also included perinephric hematoma and lymphocele with obstruction of the urinary system, which accounted for more than a third of all the observed UOC. In our view, the fact that these complications necessitated therapeutic intervention acknowledges their clinical impact. Equally important is that the peritransplant ureteral stent placing was not part of the clinical routine at the respective time period; therefore, our findings must be cautiously interpreted in the setting of routine ureteral stent placing.14
We are aware of the limitations of our study lying in its retrospective design. We corrected for this by including a large sample-size of a nonselected cohort of all early indication biopsies, which allowed for adjustment of various relevant confounders. Moreover, the strict design of the case-control study allowed the exclusion of DGF as a relevant confounder for diagnosis of occult UOC. Another important aspect is the lack of systematical biopsy results of patients with known obstruction and typical hydronephrosis. As high-grade hydronephrosis is considered a contraindication for percutaneous ultrasound-guided biopsy by most centers,21 these data are not available. However, we have tried to compensate this by analyzing 7 patients who had a known UOC and underwent intraoperative biopsy in course of revision surgery. Six had a urinary leak, and 1 patient had a ureteral necrosis. All 7 patients had typical obstructive histomorphology with a variant degree of TE, tubular casts, and tubular vacuolization underlining the biological plausibility of the applied approach.
Finally, it has to be noticed that baseline ultrasonography has not triggered suspicion of UOC in these cases. However, the subsequent diagnosis of UOC was not based on the same advanced imaging techniques as some had computed tomography or magnetic resonance imaging (data not shown). However, these are not part of a routine work-up and the appropriate imaging procedure has to be chosen according to the clinical presentation and potential contraindications for the particular technique.
Our findings demonstrate that clinically relevant but occult UOC may be identified by distinctive routinely obtained biopsy findings including mild TE together with additional signs of tubular injury. Notably other causes of tubular damage also need consideration as high-grade TE was related to other causes of tubular injury such as higher rate of DGF, higher donor-age, and higher cold-ischemia time indicating that more severe tubular injury may be rather related to ischemia/reperfusion injury than UOC. The individual features of tubular injury may be unspecific if seen separately, but a combined view can yield a high specificity and negative predictive value. Accordingly, these findings should trigger more detailed investigations on potential UOC, preferably with radiologic imaging techniques of highest sensitivity.
1. Shoskes DA, Hanbury D, Cranston D, et al. Urological complications in 1,000 consecutive renal transplant recipients. J Urol. 1995; 153118–21
2. Koçak T, Nane I, Ander H, et al. Urological and surgical complications in 362 consecutive living related donor kidney transplantations. Urol Int. 2004; 723252–256
3. Nie ZL, Zhang KQ, Li QS, et al. Urological complications in 1,223 kidney transplantations. Urol Int. 2009; 833337–341
4. Zavos G, Pappas P, Karatzas T, et al. Urological complications: analysis and management of 1525 consecutive renal transplantations. Transplant Proc. 2008; 4051386–1390
5. Inci MF, Ozkan F, See TC, et al. Renal transplant complications: diagnostic and therapeutic role of radiology. Can Assoc Radiol J. 2014; 653242–252
6. Rodgers SK, Sereni CP, Horrow MM. Ultrasonographic evaluation of the renal transplant. Radiol Clin North Am. 2014; 5261307–1324
7. Cortell S, Gennari FJ, Davidman M, et al. A definition of proximal and distal tubular compliance. Practical and theoretical implications. J Clin Invest. 1973; 5292330–2339
8. Solez K, Morel-Maroger L, Sraer JD. The morphology of “acute tubular necrosis” in man: analysis of 57 renal biopsies and a comparison with the glycerol model. Medicine (Baltimore). 1979; 585362–376
9. Olsen S, Burdick JF, Keown PA, et al. Primary acute renal failure (“acute tubular necrosis”) in the transplanted kidney: morphology and pathogenesis. Medicine (Baltimore). 1989; 683173–187
10. Walsh RC, Brailey P, Girnita A, et al. Early and late acute antibody-mediated rejection differ immunologically and in response to proteasome inhibition. Transplantation. 2011; 91111218–1226
11. Englesbe MJ, Dubay DA, Gillespie BW, et al. Risk factors for urinary complications after renal transplantation. Am J Transplant. 2007; 761536–1541
12. Rule AD, Larson TS, Bergstralh EJ, et al. Using serum creatinine to estimate glomerular filtration rate: accuracy in good health and in chronic kidney disease. Ann Intern Med. 2004; 14112929–937
13. Hallgren KA. Computing inter-rater reliability for observational data: an overview and tutorial. Tutor Quant Methods Psychol. 2012; 8123–34
14. Byrt T, Bishop J, Carlin JB. Bias, prevalence and kappa. J Clin Epidemiol. 1993; 465423–429
15. Hiatt MJ, Ivanova L, Trnka P, et al. Urinary tract obstruction in the mouse: the kinetics of distal nephron injury. Lab Invest. 2013; 9391012–1023
16. Kida Y, Sato T. Tubular changes in obstructed kidney of adult mice evaluated using immunohistochemistry for segment-specific marker. Histol Histopathol. 2007; 223291–303
17. Gibson IW, Gwinner W, Bröcker V, et al. Peritubular capillaritis in renal allografts: prevalence, scoring system, reproducibility and clinicopathological correlates. Am J Transplant. 2008; 84819–825
18. Kozakowski N, Herkner H, Eskandary F, et al. An integrative approach for the assessment of peritubular capillaritis extent and score in low-grade microvascular inflammation-associations with transplant glomerulopathy and graft loss. Nephrol Dial Transplant. 2019; 341166–174
19. Tisserand B, Doré B, Touchard G, et al. Long-term outcome of renal transplantation: impact of surgical complications on graft survival. Prog Urol. 2013; 232113–120
20. Streeter EH, Little DM, Cranston DW, et al. The urological complications of renal transplantation: a series of 1535 patients. BJU Int. 2002; 907627–634
21. Hogan JJ, Mocanu M, Berns JS. The native kidney biopsy: update and evidence for best practice. Clin J Am Soc Nephrol. 2016; 112354–362