Potential of Noninvasive Serial Assessment of Acute Renal Allograft Rejection by ¹⁸F-FDG PET to Monitor Treatment Efficiency

DISCUSSION

In renal allografts, episodes of rejection are characterized by recruitment of activated leukocytes into the transplant (17,18,29). This is an integral part of the basic concept of the Banff classification, a commonly used score of renal rejection (7). Activated leukocytes strongly accumulate ¹⁸F-FDG, which can be quantified by PET (30).

On the basis of this concept, we recently established ¹⁸F-FDG PET as a noninvasive method for the detection and monitoring of renal allograft rejection using uninephrectomized rats (16). At present, monitoring of early treatment response is rather difficult and delayed, because up to 54% of patients with allograft rejection do not present with clinical signs (12,13). Biopsy as an invasive procedure usually applied in cases of persistent graft failure carries the risk of graft injury, is not feasible in patients taking anticoagulant medication, and might present false-negative results, such as when rejection is focal or patchy. Thus, the switch to a more effective therapeutic regimen could be delayed, aggravating graft failure (12,14). In this study, we hypothesized that ¹⁸F-FDG PET is a useful noninvasive tool for assessing treatment response early, in order to translate this approach to humans in the future. Using binephrectomized rats, we applied ¹⁸F-FDG PET in a renal transplant model closely mimicking the clinical situation. In this model, impairment of renal function leads to typical consequences of renal failure, such as disturbance of electrolyte and water homeostasis or azotemia (a milieu known to affect leukocyte function) (31)—features that are absent in the uninephrectomized model.

First, renal ¹⁸F-FDG uptake in all investigated renal injury and transplant models was quantitatively evaluated on PET images. The %ID in grafts developing rejection was significantly elevated starting on postoperative day 4, when compared with isografts or allografts under treatment with CSA. ¹⁸F-FDG uptake did not increase in models developing ATN or acute CSA toxicity, which are common differential diagnoses of acute allograft rejection. Consistently, increased ¹⁸F-FDG uptake in grafts undergoing rejection was already found in the lungs (32), liver (33), skin (34), and kidneys using a different, uninephrectomized, transplant model (16). Further, autoradiography showed the most intense ¹⁸F-FDG signaling in allograft rejection, whereas histology revealed increased inflammatory cell infiltrates most notably in the renal cortex of allograft rejection. In all control models used, ¹⁸F-FDG signal intensity was on the level of native kidneys whereas inflammation was absent, concentrated to small and distinct areas (ATN: spots in the outer medulla), or classified as mild (<10% of parenchyma). These results were supported by real time PCR analysis of proinflammatory cell markers within the kidneys. In control and sTX rats, we could not detect relevant amounts of mRNA of proinflammatory cells, whereas aTX rats on postoperative day 0 (the rats treated with CSA since the transplantation procedure) presented with very low levels of T-cell, B-cell, and monocyte/macrophage marker mRNA on postoperative day 4. However, the ti-score was still 0. Interestingly, on postoperative day 1 in aTX rats without an immunosuppressive regimen, the mRNA of proinflammatory cells had not yet increased (a ti-score of 0), whereas we found high expression of marker mRNA from cytotoxic T cells and monocytes/macrophages and lower but still significant levels of B cells, regulatory T cells, and dendritic cells on postoperative day 4 (a ti-score of 2–3). Because granzyme B and ficolin B gene expression were highly upregulated on postoperative day 4, one can assume that cytotoxic and regulatory T cells and monocytes/macrophages were highly activated (28), as is commonly found during allograft rejection (29). As expected, our findings suggest that ¹⁸F-FDG detects changes in renal glucose metabolism consistent with activated inflammatory infiltrates occurring in renal allografts. Notably, we found a significant correlation between the mRNA expression of granzyme B and ficolin B (activation marker genes of cytotoxic T cells and monocytes/macrophages) and ¹⁸F-FDG in all groups measured (Fig. 5).

Regarding renal functional data, creatinine and BUN had increased in all intervention groups by postoperative day 1. Recovery of renal function started immediately, leading to a continuous decrease of creatinine and BUN until postoperative day 4, with the highest levels in aTX rats developing allograft rejection (and therefore consecutively receiving further damage). Hence, ¹⁸F-FDG PET correctly diagnosed allograft rejection in rats with decreasing creatinine and BUN values on postoperative day 4, pointing to the capability of PET to detect allograft rejection early.

Morath et al. performed protocol biopsies on patients with delayed graft function 1 wk after transplantation (9). Allograft rejection was confirmed in 18% of patients with delayed graft function, compared with only 4% of patients with early graft function. Of allograft rejection episodes in patients with delayed graft function, 50% had remained clinically undiagnosed and were revealed only by the protocol biopsies, whereas allograft rejection would have remained undiagnosed on clinical grounds in all patients with delayed graft function who required dialysis. Similar results were reported by Shapiro et al. (35). Every fifth patient with normal or improving renal function had borderline histopathologic findings, and 1 of 4 had frank acute tubulitis (Banff 1A to 2A). Similar observations were published by Rush, who performed protocol biopsies 3 mo after transplantation (8). Thus, it was postulated that more than 50% of the allograft rejection episodes are subclinical at some time after renal transplantation without acute impairment of renal function. Therefore, we believe that ¹⁸F-FDG PET is a valuable diagnostic tool to avoid unnecessary biopsies, mainly in patients with inconspicuous laboratory results and clinical evolution.

Second, renal ¹⁸F-FDG uptake in aTX rats treated with CSA remained low and stable during treatments, suggesting that ¹⁸F-FDG PET can be used to monitor the efficacy of immunosuppressive therapy as a noninvasive method, in contrast to biopsy. However, there are other methods, such as assessment of renal serum parameters, for evaluation of transplant function. To enlarge the potential use of ¹⁸F-FDG PET for treatment follow-up, we included a group of rats that developed allograft rejection, which was treated by CSA since postoperative day 4. Notably, ¹⁸F-FDG signaling decreased only 24 h after commencement of therapy (postoperative day 5) to levels nearly those of controls and stayed at these levels until the end of follow-up. Analysis of functional renal parameters revealed that creatinine and BUN decreased 24 and 48 h later (postoperative days 6–7), respectively. Moreover, renal histology showed a degree of infiltration that was still relevant (ti-score of 1–2) in these grafts on postoperative day 7, potentially leading to the diagnosis of active, therapy-refractory allograft rejection. However, in our model these cells responded to treatment as confirmed by PCR analysis of gene expression profile, such as decreased expression of CD8a/b and ficolin B and increased expression of pro-apoptotic caspase 3. Therefore, detection of treatment response was 24–48 h earlier when ¹⁸F-FDG PET was applied. Consistently, Lopes de Souza et al. used ^99mTc-labeled mononuclear leukocytes for detection of renal allograft rejection in humans. In their study, scintigraphy detected allograft rejection on postoperative day 5 (but not on postoperative day 1), which was around 2–3 d before clinical manifestations (36). The fact that ¹⁸F-FDG PET correctly diagnosed treatment response in rats with increasing creatinine and BUN values points to the capability of PET to detect early treatment response.

These findings might yield consequences for the management of patients undergoing allograft rejection, because it might be possible to react to treatment response or nonresponse earlier. In the case of treatment response, one might reduce immunosuppressive drug doses and thereby limit their side-effects, because a successful response to therapy is classically defined as a relative serum creatinine concentration no more than 110% of the concentration on the day of diagnosed rejection (day 0) and returning to the day 0 value or less within 5 d of therapy (12). Even more important, in the case of treatment-unresponsive allograft rejection, escalation of the immunosuppressive regimen can start earlier. Earlier escalation might prevent graft damage by shortening allograft rejection episodes. At present, steroid-resistant rejection—an allograft rejection episode that has been treated with up to 1 g of methylprednisolone and in which the creatinine concentration fails to stabilize or lessen after 3 d of therapy—is diagnosed late (12).

Two potential problems in applying ¹⁸F-FDG PET to the clinical setting should be addressed. Urinary excretion of the tracer might induce artificially high renal ¹⁸F-FDG uptake. Thus, the renal pelvis was carefully excluded from the measurements. Various techniques such as hydration or administration of diuretics have been used to improve the diagnostic yield of PET by increasing diuresis for dilution of urinary ¹⁸F-FDG (37). In view of the results of preliminary dynamic ¹⁸F-FDG investigations, we chose late acquisitions 3 h after injection to reduce the tracer concentration in the urine during the PET scan. The impact of renal function on ¹⁸F-FDG uptake seems to be minimal in our model because ¹⁸F-FDG uptake did not correlate with creatinine and BUN (Fig. 6). Moreover, in a previous study we did not find a correlation between fluoride clearance (a method for noninvasive assessment of renal function (21)) and ¹⁸F-FDG uptake (16). Nevertheless, significant renal ¹⁸F-FDG accumulation was present in allograft rejection only, whereas renal insufficiency was observed in other groups, too (Table 2). However, we cannot exclude the possibility that a part of the PET signal might refer to excreted ¹⁸F-FDG. When renal biopsy needs to be performed despite suspected allograft rejection, such as in cases of patchy rejection patterns, ¹⁸F-FDG PET might help to determine the appropriate location.

The second problem is the inevitably impaired specificity in diagnosing allograft rejection: although ¹⁸F-FDG uptake certainly detects enhanced metabolic activity, ¹⁸F-FDG uptake is not disease-specific. Thus, graft infection or tumors might also cause ¹⁸F-FDG uptake on PET. However, if clinical symptoms pointing to malignancy (e.g., weight loss or night sweating) or to urinary tract infection (e.g., fever or dysuria) are present, target-orientated additional diagnostics such as ultrasound, blood tests, or urine tests should be applied.