Clinical trials to improve the long-term outcome in renal transplantation are troubled by the very success of the procedure. Most grafts last 10 years or more, so the most meaningful outcome measurement—graft survival—cannot be used if results are to be provided in a reasonable time frame, with a practical number of patients and at acceptable cost (1).
The most widely used surrogate marker for graft survival is the incidence or severity of episodes of acute rejection. This is superficially appealing, although applicable only to immunosuppressive drugs. We know that acute rejection is an adverse prognostic factor for long-term survival. However, as Hunsicker and Bennett showed several years ago, acute rejection in the first year of engraftment is most closely associated with graft loss in the first year, but it correlates less well with the rate of graft loss after the first year (1). This analysis of registry data convincingly argued that acute rejection was not an adequate surrogate marker for long-term graft survival. Later episodes of acute rejection might be a better marker, were it not for their rarity, which again means that an adequately powered trial would need unattainable numbers of patients.
Good early graft function is associated with better survival in statistical analyses of large cohorts, but again the association is not close enough to make this a successful surrogate marker. As Kasiske et al. reported, when established grafts fail, a detectable impairment of function usually occurs late (2). In practice, the graph of inverse creatinine against time is rarely a straight line for the individual patient, and analysis of registry data confirms that graft function at 1 year is not a reliable predictor of long-term survival (1).
Consequently, for some years there has been interest in the use of renal allograft biopsy as a tool to provide information on the efficacy of treatment. Systems for evaluation, such as the “chronic allograft damage index,” were devised (3) and have since been refined. There are some problems with the reproducibility of subjective biopsy interpretation (4), but we recently demonstrated that the introduction of morphometric measurements can produce a dramatic decrease in the number of cases required to produce an adequately powered study, especially if two biopsies are available to allow the assessment of changes over time (5).
Such measurements should not be made on biopsies taken to investigate graft dysfunction, because the results would be distorted by any changes caused by whatever process is causing acute dysfunction. Protocol biopsies, taken during periods of stable graft function, are vital and may constitute the best strategy to design trials aimed at modifying the natural history of chronic allograft nephropathy with a reasonable number of patients. Chronic allograft nephropathy is the main cause of graft failure and can be detected on biopsy long before graft dysfunction becomes evident (6). However, many clinicians have been reluctant to perform a biopsy on a stable transplant for fear of causing complications, arguing that the risk is not ethically justifiable. Consequently, in clinical trials in which protocol biopsies have been planned at the outset, the compliance rate has sometimes been too low to be scientifically useful (P. Halloran, M.D., personal communication, 2001).
Those centers actively undertaking a protocol biopsy program express the view that the risks are small and, even for the individual patient, are outweighed by the potential benefits of detecting unexpected changes, such as donor-related vascular disease, “subclinical” acute rejection, or complications of therapy (7). However, the risks of renal transplant protocol biopsy have not been quantified. It is not appropriate to extrapolate from the risk of biopsies taken for graft dysfunction, in which many of the kidneys being assessed will be severely abnormal, some with preexisting acute vascular damage. The anecdotal experience of individual centers indicates that the risk of serious complications from protocol biopsy is low. Accurate quantification of a small risk requires large numbers of patients. To achieve such large numbers, we undertook an audit of complications of protocol biopsies in four major European transplant centers, which all have a long history of using the procedure.
MATERIALS AND METHODS
The questions to be answered in this audit were defined at the outset and are listed in Table 1. The time period during which the audit was to be conducted was left to the individual centers, to be selected to represent the most recent period when protocol biopsy practice had been relatively stable (Table 2).
The number of biopsy events studied in each center is indicated in Table 2. In Leicester, for logistic reasons, the study was split into two parts. Audit data on major complications (death, loss of graft, hemorrhage requiring operative intervention, and hemorrhage requiring transfusion) were available for 1,159 biopsies during a continuous 10-year period. Data on less serious complications were more labor intensive to collect, so this was limited to 518 biopsies during a 4-year period. Thus, 2,127 biopsy events were studied for less frequent and major complications, and 1,486 biopsies were studied for more frequent and less serious complications.
Center Technique and Basic Statistics
Predefined contraindications varied considerably between institutions, with systemic anticoagulation being the only unanimous exclusion criterion. The strictest contraindications were in place in Barcelona: anti-coagulation, bowel interposition, creatinine greater than 300 μmol/L, proteinuria greater than 1 g per day, and unstable renal function. This resulted in failure to biopsy as a result of contraindications in 10.3% of cases. The lowest dropout rate resulting from contraindications was in Hannover (1.3%).
Informed consent was a requirement in every center. However, the rate of patient refusals varied considerably between centers (9.8% in Barcelona and 1.6% in Hannover). In Leicester, patient refusal was not always documented. There were numerous biopsies not performed for unexplained reasons, so the refusal rate was probably higher. Unexplained failure to perform a biopsy was probably caused by resource problems, principally a lack of vacant beds and nurses for the necessary period of observation postbiopsy. However, it was also clear that some undocumented failures to perform a biopsy occurred when one trainee member of the transplant team, unknown to the senior staff, approached patients stating that the procedure was unnecessary, of no benefit to the patient, and likely to cause complications. By extrapolation, it seems likely that variations in the “consent” rate between other centers may have been caused by differences in the manner and content of information provision during the consent process.
The size of needle used varied between 14G and 18G, without detectable variation in the incidence of complications, as has been reported (8). Biopsies were taken using a spring-loaded gun under ultrasound guidance in every center except Antwerp, where a Tru-Cut needle was used with computed tomography imaging. Every center took an average of two cores per procedure, and the typical duration of the procedure was 20 min. Postbiopsy observation was typically 24 hr, except in Hannover, where the procedure was performed as a day case with a 4-hr observation.
Postbiopsy investigations varied considerably, from observation only in Leicester to postbiopsy computed tomography scan with hemoglobin estimation prebiopsy and 24-hr postbiopsy in Antwerp. This variation may account for some of the center-specific variation in the rates of less serious complications, as described next.
Major complications were recorded for 2,127 biopsies (Table 3).
There were no patient deaths attributable to biopsy. One transplant was lost as a consequence of postbiopsy hemorrhage. This was the most serious single complication in the series and therefore deserves detailed description.
The patient underwent a protocol transplant biopsy in the afternoon. Even according to the strictest set of contraindications used by any center in this study, there were no contraindications to the biopsy. Hematuria was noted soon after the biopsy, but initially its severity did not require anything more than observation. That night, during the 24-hr observation period, the hemorrhage recurred with a decrease in blood hemoglobin and hypotension. At this point, it was discovered that the hospital management had withdrawn the on-call interventional radiology service without informing the transplant team. The surgeon responsible that evening made the decision that the safest option was to excise the graft. During the subsequent investigation, it was agreed that the graft could almost certainly have been preserved with better management. An on-call interventional radiology service has now been restored.
Two patients experienced nonfatal peritonitis after the biopsy. One was the direct consequence of the biopsy needle puncturing the caecum. The other proved to be the result of a peptic ulcer of the stomach, which perforated in the hours after the procedure and arguably should not be regarded as a direct complication.
Three patients required intervention to control hemorrhage. In one case, the hemorrhage resulted from the biopsy needle disrupting the patient’s inferior epigastric artery rather than the hemorrhage coming from the graft. This patient and one other patient required operative intervention. The third patient’s hemorrhage was controlled by interventional radiology. In addition to these patients, three patients required transfusion without further operative intervention.
Despite the large number of biopsy events studied, the low rate of serious complications described does not permit any statistical evaluation of possible associations with different practice in different centers. In every case, the presence of a problem would have been recognized within the shortest period of observation used (4 hr), although, as described, the seriousness of the problem may not have become evident if the patient had not been subjected to 24-hr observation. No serious complications became evident after 24 hr.
Less Severe Complications
All other complications in the audit showed considerable variation in levels of reporting between different hospitals (Table 4). It is clear that much of this variation is caused by variable interpretation of when such complications should be reported and variations in the methods by which they were sought. This is most clear in the case of arteriovenous fistulae, in which one center followed the biopsy with a Doppler ultrasonography scan, specifically looking for this complication. The result was a much higher apparent incidence of arteriovenous fistulae (10.7%), but none of them required any treatment. With this exception, the incidence of all other complications remained relatively low.
Individual Patient Benefit
This aspect of the cost-benefit analysis was not formally evaluated. In Barcelona, the benefit to the individual patient was unequivocally zero, because the protocol biopsies were assessed independently of the process of patient care, and data were not shared with the clinical team. This arrangement has since been revised. In the other three centers, results from the protocol biopsies were made available as part of clinical management. Consequently, cases of subclinical acute rejection were detected as we have reported elsewhere (9). There was no consensus on how this should be managed or whether (as Rush et al. have suggested) treatment was required (7), but all centers believed that subclinical acute rejection deserved at least review, and in many cases some modification of the immunosuppressive regime was instituted. Furthermore, when a protocol biopsy revealed severe fibrosis, there was a tendency to attempt to reduce or even eliminate calcineurin inhibitors from the immunosuppressive regime. The finding of vascular lesions often led to an adjustment of the antihypertensive treatment or lipid-lowering drugs.
Such responses, intended to benefit individual patients, were not formally planned in this retrospective audit, and any resultant benefits were not assessed.
Some clinicians believe that renal transplant protocol biopsy poses a significant risk to graft survival, a risk that is greater than the potential benefit for the patient. This risk-to-benefit assessment is also directly relevant to the ethical review of any proposed clinical trials that include the study of protocol biopsies. Unfortunately, neither the risk nor the benefit have been clearly quantified.
The magnitude of any potential benefit for the individual patient remains a matter for debate. If it is confirmed that treatment of “subclinical acute rejection” prolongs long-term graft survival, then one benefit will be clear (7). Unfortunately, the original study proposing this benefit was relatively small, and it suffers from the very problem that makes protocol biopsies desirable in clinical trials; data on the “real” outcome measure, graft survival, will take several more years to become available. Similarly, the benefit of avoiding calcineurin inhibitors selectively in patients whose grafts demonstrate severe interstitial fibrosis or significant intimal fibrosis remains hypothetical without proof from long-term studies.
The problem that we have tried to resolve is the other side of the equation, an accurate estimation of the risk involved. The nature of the problem demands a multicenter study to provide an answer, and we have acknowledged that this introduces considerable variation in biopsy practice, follow-up, and investigation. Nevertheless, we believe we can confidently state that the incidence of clinically significant problems is low; the 2,127 biopsy events produced only eight major complications, with only one graft loss and no patient fatalities. Indeed, it is so low that even with a large multicenter study such as this we have insufficient serious complications to permit any statistical analysis of the circumstances under which they occurred. Nevertheless, we hope that the data presented here will assist clinicians in deciding whether it is justifiable to request a protocol biopsy and will also assist patients in deciding whether to give consent. It is our opinion that the risk of the procedure is sufficiently low to make it legitimate to request a protocol biopsy as part of a clinical trial even if there is an assumption of zero direct benefit to the patient concerned, assuming the patient has given consent on the basis of appropriate information, including a risk assessment.
It is perhaps relevant to compare our results with those of Wilczek (10), who reported three graft nephrectomies for biopsy-induced hemorrhage in a series of 1,129 transplant biopsies but noted that all three excised grafts were demonstrating severe acute rejection. This ties in with a local audit of native renal biopsies in Leicester, where postbiopsy hemorrhage requiring nephrectomy was invariably associated with severe parenchymal disease (P. Furness, unpublished data, 1990–2002). It is perhaps not surprising to find that the risk of a biopsy of “normal kidney” is lower than a biopsy of “inflamed kidney,” but previous audits have considered only overall complication rates. This will overestimate the risk of biopsy of a stable graft and therefore may explain the reluctance of some to undertake protocol biopsies.
A further problem remains in justifying the use of protocol biopsies as a surrogate marker in clinical trials. There is as yet no agreed way in which to extract prognostic data from these samples. Subjective histologic grading systems (3,11) have proven validity and have been widely used, but there are problems with reproducibility between different centers (4). Measurements of interstitial or vascular fibrosis have been tested (5,12–14), but it has recently been suggested that more “acute” changes such as lymphocytic infiltration might have greater prognostic impact (15). Other possibilities, such as assessing gene expression or levels of cell senescence (16) have barely been explored. The most likely outcome will be that no single measurement provides the best possible measure of prognosis; instead, data from several measurements should be integrated with clinical, immunologic, and functional information, perhaps using logistic regression or neural network systems. To optimize this approach will require coordinated analysis of data from large numbers of protocol biopsies. In the meantime, we believe the evidence is sufficient to justify the introduction of renal transplant protocol biopsies into any new clinical trial that addresses long-term renal allograft survival.
1. Hunsicker LG, Bennett LE. Design of trials of methods to reduce late renal allograft loss: the price of success. Kidney Int 1995; 48 (Suppl 52): S120.
2. Kasiske BL, Massy ZA, Guijarro C, et al. Chronic renal allograft rejection and clinical trial design. Kidney Int Suppl 1995; 52: S116.
3. Isoniemi H, Taskinen E, Hayry P. Histological chronic allograft damage index accurately predicts chronic renal allograft rejection. Transplantation 1994; 58: 1195.
4. Furness PN, Taub N. International variation in the interpretation of renal transplant biopsies: report of the CERTPAP Project. Kidney Int 2001; 60 ( 5): 1998.
5. Moreso F, Lopez M, Vallejos A, et al. Serial protocol biopsies to quantify the progression of chronic transplant nephropathy in stable renal allografts. Am J Transplant 2001; 1 ( 1): 82.
6. Seron D, Moreso F, Ramon JM, et al. Protocol renal allograft biopsies and the design of clinical trials aimed to prevent or treat chronic allograft nephropathy. Transplantation 2000; 69 ( 9): 1849.
7. Rush D, Nickerson P, Gough J, et al. Beneficial effects of treatment of early subclinical rejection: a randomized study. J Am Soc Nephrol 1998; 9 ( 11): 2129.
8. Nicholson ML, Wheatley TJ, Doughman TM, et al. A prospective randomized trial of three different sizes of core-cutting needle for renal transplant biopsy. Kidney Int 2000; 58 ( 1): 390.
9. Jain S, Curwood V, White SA, et al. Sub-clinical acute rejection detected using protocol biopsies in patients with delayed graft function. Transpl Int 2000; 13 (Suppl 1): S52.
10. Wilczek HE. Percutaneous needle biopsy of the renal allograft. A clinical safety evaluation of 1129 biopsies. Transplantation 1990; 50 ( 5): 790.
11. Racusen LC, Solez K, Colvin RB, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int 1999; 55 ( 2): 713.
12. Seron D, Moreso F, Gratin C, et al. Automated classification of renal interstitium and tubules by local texture analysis and a neural network. Anal Quant Cytol Histol 1996; 18 ( 5): 410.
13. Woestenburg AT, Mensch PP, Bosmans JL, et al. Morphometric analysis of renal vessels. Advantage of the multisector area approach to cross-sections. Anal Quant Cytol Histol 2000; 22 ( 4): 285.
14. Bosmans JL, Woestenburg A, Ysebaert DK, et al. Fibrous intimal thickening at implantation as a risk factor for the outcome of cadaveric renal allografts. Transplantation 2000; 69 ( 11): 2388.
15. Furness PN. Chronic allograft nephropathy: what can the biopsy tell us? Transplant Proc 2001; 33 ( 7–8): 3357.
16. Halloran PF, Melk A, Barth C. Rethinking chronic allograft nephropathy: the concept of accelerated senescence. J Am Soc Nephrol 1999; 10 ( 1): 167.