Targeted Dual-Modality Imaging in Renal Cell Carcinoma: An Ex Vivo Kidney Perfusion Study

In oncological surgery, radical tumor resection is crucial for treatment outcome and patient survival (1–6). During surgery, differentiation of tumor tissue from nontumorous tissue with the naked eye may be challenging. Intraoperative imaging techniques that can distinguish tumor from normal tissue will help the surgeon to achieve complete tumor resection.

One of these techniques, radio-guided surgery, has already been implemented in clinical practice, for example, to detect sentinel lymph nodes (7–9). The high tissue penetration depth of gamma radiation allows accurate localization of tumors, almost regardless of tissue depth. However, exact tumor delineation with a gamma probe is difficult, and for precise real-time tumor delineation during surgery, an optical signal may be more beneficial. For this purpose, a fluorescent probe can be used, but because the penetration depth of light in biologic tissue is limited, fluorescence imaging is mainly useful for the detection of superficially located tumors (10, 11). Therefore, dual-modality image-guided surgery, combining the advantages of radioguided and fluorescence-guided surgery, may be a synergistic combination (12–15). A dual-labeled tumor-targeting imaging probe could allow intraoperative tumor detection, tumor delineation, and assessment of resection margins and remnant disease. Results from preclinical studies are promising (16–19), and tumor-targeted dual-modality imaging is awaiting its first use in patients.

Monoclonal antibodies against tumor-associated antigens tagged both with a radiolabel and a fluorophore may be used as probes during image-guided surgery (20). Girentuximab, a chimeric monoclonal IgG1 antibody, is an excellent vehicle to target clear cell renal cell carcinoma (ccRCC). It specifically recognizes carbonic anhydrase IX (CAIX), a cell surface antigen that is expressed abundantly in more than 95% of ccRCC and is absent in normal kidney tissue (21, 22). Previous research has shown that after administration of radiolabeled girentuximab, both primary tumors and metastases can be detected by PET/CT or SPECT/CT in patients (23–25). Animal studies using ¹¹¹In-girentuximab-IRDye800CW have illustrated the potential of dual-modality imaging (17, 18). Dual-labeled girentuximab accumulates highly and specifically in CAIX-expressing SK-RC-52 subcutaneous tumors (18). Next, a proof-of-principle study was performed to show the feasibility of image-guided surgery using an intraperitoneal tumor model (17). After administration of dual-labeled girentuximab, submillimeter CAIX-expressing tumor nodules could be detected preoperatively with SPECT/CT. Subsequently, during surgery, superficially located tumor nodules could be visualized and resected based on the fluorescent signal.

However, animal studies only partly reflect the clinical situation. To test the translation of dual-modality imaging to the clinic, models using clinical imaging systems and intact tumors obtained from patients are required. In the present study, tumorous kidneys resected from patients with ccRCC were connected ex vivo to a perfusion system via the renal artery as has been described previously (26). Subsequently, the tumorous kidneys were perfused with dual-labeled girentuximab and investigated using the clinical imaging system to be used during image-guided surgery. In preparation of the first targeted dual-modality image-guided surgery study in patients, this study aimed to show the feasibility of tumor-targeted dual-modality imaging in renal cell carcinoma using dual-labeled girentuximab in an ex vivo kidney perfusion model.

The aim of this study was to assess the feasibility of dual-modality imaging in patients with ccRCC in an ex vivo kidney perfusion study. Patients suspected of ccRCC and scheduled for a radical nephrectomy that signed informed consent were included. Exclusion criteria were a known subtype other than ccRCC or the administration of a radioisotope within ten physical half-lives prior to surgery. The first human tumorous kidney was perfused with DOTA-girentuximab-IRDye800CW not labeled with Indium-111, to estimate the radiation safety risks of the procedure. Next, four human tumorous kidneys were perfused ex vivo with dual-labeled girentuximab. To demonstrate specific binding of girentuximab, two additional tumorous kidneys were perfused with a mixture of dual-labeled ¹³¹I-girentuximab-IRDye800CW and a ¹²⁵I-labeled control antibody-IRDye800CW. After perfusion, radionuclide and fluorescence imaging were performed on a 1-cm lamella of the specimen using clinical and preclinical imaging systems. Subsequently, samples of tumor and normal tissue were taken for further analysis: quantification of antibody accumulation, fluorescence imaging, autoradiography, CAIX staining, and hematoxylin and eosin (H&E) staining. The primary outcome measure was the ability to detect a fluorescent and radioactive signal in tumor tissue. Secondary outcome measures were the antibody accumulation in tumor and normal kidney tissue expressed as percentage of injected dose per gram (%ID/g) and as the tumor-to-kidney tissue ratio. This study was approved by the regional ethical review board (CMO, region Arnhem-Nijmegen).

Chimeric girentuximab (IgG1) was a kind gift from Wilex AG. The isotype-matched control human Immunoglobulin (IgG1) was purchased from Biotrend Chemikalien. The fluorophore IRDye800CW-NHS ester was purchased from LI-COR biosciences. IRDye800CW is a near-infrared dye that can be stably coupled to monoclonal antibodies (27). It emits 789-nm photons when excited at 774 nm. Near-infrared fluorescent dyes (700–1,000 nm) have a higher tissue penetration depth compared with dyes in the visible light range (28). The chelator DOTA-NHS ester (1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid mono-N-hydroxysuccinimide ester) was purchased from Macrocyclis.

DOTA-girentuximab-IRDye800CW was produced under metal-free conditions as described previously (29). In short, girentuximab (5 mg/mL) was incubated in 0.1 mol/L NaHCO₃, pH 8.5, at room temperature (RT) for 1 hour with a threefold molar excess of the IRDye800CW-NHS ester. Then, the mixture was incubated in NaHCO₃, pH 9.5, for 1 hour with a 10- to 20-fold molar excess of the DOTA-NHS ester. After conjugation, the reaction mixture was transferred into a Slide-A-Lyzer cassette (molecular weight cut-off: 10,000 or 20,000 Da; Thermo Scientific) and extensively dialyzed against 0.25 mol/L ammoniumacetate, pH 5.5, for 3 days with buffer changes to remove the unconjugated IRDye800CW and DOTA. The average number of IRDye800CW molecules that was conjugated to the antibody (substitution ratio or SR) was determined spectrophotometrically with the Ultrospec 2000 UV/Visible spectrophotometer (Pharmacia Biotech) and ranged from 1.3 to 1.5. The immunoreactive fraction of dual-labeled girentuximab was determined as described by Lindmo (30), within 2 weeks of each experiment and always exceeded 70%. DOTA-girentuximab-IRDye800CW was stored in the dark at 4°C until use.

For the control experiments, girentuximab-IRDye800CW and control IgG-IRDye800CW were prepared as described above. The SR of girentuximab-IRDye800CW and the irrelevant control IgG-IRDye800CW were 2.0 and 2.1, respectively.

On the day of the experiment, 1.2 mg DOTA-girentuximab-IRDye800CW was labeled with 3.5 to 7.0 MBq ¹¹¹InCl₃ (Mallinckrodt Pharmaceuticals) in two volumes of 0.5 mol/L 2-(N-morpholino)ethanesulfonic acid (MES). After 40 minutes of incubation at 45°C, 50 mmol/L ethylenediaminetetraacetic acid (EDTA) was added to a final concentration of 5 mmol/L to chelate unincorporated Indium-111. Preparations were purified on a PD10 disposable gelfiltration column eluted with PBS. Radiochemical purity was determined by Instant Thin Layer Chromatography (ITLC) on silicagel, using 0.1 mol/L sodiumcitrate buffer, pH 6.0, as mobile phase. Radiochemical purity of the Indium-111–labeled preparations exceeded 95%. After purification, the DOTA-girentuximab-IRDye800CW amount was adjusted to a total protein amount of 1.2 mg. Standards of the ID were prepared in triplicate to be able to quantify antibody accumulation corrected for radioactive decay. The injected activity dose was determined in a dose calibrator and ranged from 3.4 to 5.1 MBq.

In the two dual-isotope experiments, 100 μg of girentuximab-IRDye800CW was radiolabeled with 10 to 15 MBq Iodine-131 (PerkinElmer), and 100 μg of control IgG-IRDye800CW was labeled with 5 to 7 MBq Iodine-125 (PerkinElmer; ref. 31). Briefly, 100 μg of antibody conjugate was added to 10 μL of 0.5 mol/L sodium phosphate, pH 7.4, in a vial coated with 100 μg iodogen. After adding the radioactivity, the volume was adjusted to a total volume of 100 μL with 50 mmol/L sodium phosphate, pH 7.4. After 10 minutes of incubation at RT, the reaction mixture was transferred to a clean Eppendorf vial together with 100 μL of saturated tyrosine. After purification on a PD10 column radiochemical purity exceeded 95% and the amount of both antibodies was adjusted to 1.2 mg of protein with cold antibody. Standards of the ID were prepared in triplicate. Both Iodine-131-girentuximab-IRDye800CW (1.2 mg, ID 4.6, and 4.1 MBq) and Iodine-125-control IgG-IRDye800CW (1.2 mg, ID 1.0, and 1.2 MBq) were added to the perfusion reservoir.

In the operating room, a vessel cannula was inserted on the bench in the renal artery and connected to a flushing system with Ringer's Lactate solution to rinse blood (see Supplementary Figs. S1 and S2). In case of multiple renal arteries, the two largest arteries were connected, and any others were left open. Also, the renal vein was left open. Then, the kidney with the vessel cannula(s) in situ was connected to a recirculating perfusion system, which included a peristaltic pump (30 mL/min). The flow in the system (30 mL/min) is considerably lower than the human renal blood flow (500 mL/min). This flow rate was selected based on our experience that at this maximum flow rate, the risk of leakage of the arterial connection was minimal. After ensuring the leakage-free connection to the kidney, the afferent and efferent hoses of the pump (Supplementary Fig. S1) were placed in a reservoir with 350 mL Ringer's Lactate solution containing 0.1 % BSA. This reservoir was cooled with ice (0–4°C) and protected against light. Subsequently, the dose of Indium-111-DOTA-girentuximab-IRDye800CW (1.2 mg, 3.6–5.2 MBq Indium-111) was added to the 350 mL reservoir, resulting in an antibody concentration of 3.4 mg/L. This concentration was chosen to mimic the serum concentration in patients that receive radiolabeled girentuximab in clinical studies (22–24).

The tumorous kidney was perfused during 11 to 15 hours with dual-labeled antibody. After this period, the kidney was rinsed with 5 to 7 liters of nonrecirculating Ringer's Lactate solution (2.5–4 hours) to wash out unbound antibody. For safety reasons, the first experiment was performed without radioactivity, and therefore antibody uptake was not quantified in this first kidney.

A central coronal slice (5–10 mm thick) of the perfused kidney containing both tumor and normal tissue was obtained for further analysis. Fluorescence imaging was performed using a real-time clinical imaging system (Storz D-light P laparoscopic setup). Furthermore, fluorescence imaging was performed with two nonclinical imaging systems, the IVIS Lumina closed-cabinet fluorescence imager (Caliper Life Sciences; recording time, 5 to 10 minutes; binning factor medium; F/stop 2; excitation filter 745 nm; emission filter ICG; FOV C; autofluorescence correction, 675 nm; and background correction) and the Odyssey CLx flatbed fluorescence scanner (LICOR biosciences; recording time, 20 to 30 minutes; 800 nm channel; focus 1.0 mm). Subsequently, the kidney slice (kept on ice) was exposed for 1 hour to a phosphor imaging plate for autoradiography. This plate was developed using the Typhoon FLA 7000 Phosphor Imager and analyzed with Aida Image Analyzer v. 4.21.

After dual-modality imaging, 1 cm³ samples of tumorous and normal tissue were taken from the kidney slice and fixed in 4% formalin for further analysis. The amount of Indium-111-DOTA-girentuximab-IRDye800CW in tissue was determined quantitatively by measuring these samples in a gamma counter (2480 WIZARD²; Perkin Elmer) together with the standards of the ID. In the control experiment, a dual-isotope protocol was used [Iodine-125 35 keV (window 15–85 keV), Iodine-131 360 keV (window 260–430 keV), spillover correction was applied]. The tracer uptake was expressed as percentage of the ID per gram tissue (%ID/g). In addition, the tumor-to-normal antibody uptake ratio was calculated.

Then, formalin-fixed, paraffin-embedded 5-μm sections were cut and were analyzed autoradiographically (2 weeks exposure to a phosphor imaging plate) and by fluorescence imaging (Odyssey CLx flatbed fluorescence scanner; 800 nm channel; recording time, 1 to 5 minutes; focus 1.0 mm). Subsequently, sections were stained for CAIX with the murine anti-CAIX antibody M75 (cell line HB-11128 obtained from the ATCC). Also, a classic H&E staining was performed on all slices.

Statistical analyses were performed using IBM SPSS 20.0. For each separate experiment, the antibody accumulation was expressed as %ID/g (maximum, mean, and SD). All samples that consisted macroscopically at least partly of tumor, also border, central cystic, or necrotic tumor regions, were considered as tumor tissue samples. Only samples macroscopically consisting of normal tissue were considered as normal tissue samples. Differences in accumulation between tumor and normal tissue samples were tested for significance within the single experiment using t tests. Furthermore, in the two control experiments, the difference between tumor accumulation of girentuximab compared with tumor accumulation of control IgG was tested for significance using the paired t test. Differences were considered significant at P < 0.05, two-sided. Tumor-to-normal (T:N) ratios were calculated by dividing the mean %ID/g in tumor tissue by the mean %ID/g in normal tissue. Furthermore, a range was given for the T:N ratio: minimum and maximum antibody accumulation in tumor tissue in relation to the mean antibody accumulation in normal tissue.

Between June 2014 and July 2015, seven radical nephrectomy specimens from patients with ccRCC were perfused ex vivo with dual-labeled girentuximab in Ringer's Lactate solution (experiment 1 was performed without Indium-111). Histopathology confirmed that all seven kidneys contained a CAIX-expressing ccRCC. Basic characteristics of these perfusions are shown in Table 1. 

Real-time fluorescence imaging of the lamella with the clinical fluorescence camera system revealed images in which the tumor was clearly delineated (Fig. 1 and Supplementary Video S1). Subsequently, non–real-time fluorescence imaging was performed with the Odyssey fluorescence flatbed scanner (Figs. 2D and 3B) and the IVIS Lumina fluorescence imager (Fig. 2B). Fluorescence imaging showed preferential accumulation of dual-labeled girentuximab in tumor tissue, whereas fluorescence in normal tissue was hardly detectable. As a result, a clear contrast was obtained between tumor and normal tissue. These results were consistent throughout all experiments. Most importantly, in one of the perfused kidneys, a small satellite tumor lesion was detected by fluorescence imaging that had not been recognized by macroscopic inspection, but proved to be ccRCC at histopathologic evaluation (Fig. 3A and B).

After the fluorescence measurements, the accumulation of Indium-111-girentuximab-IRDye800CW in tumor tissue could be clearly visualized by autoradiography of the 1-cm lamella (Fig. 2C, 1h exposure to a Phosphor imaging screen). Visual assessment revealed excellent colocalization of the fluorescent and radioactive signal in regions macroscopically designated as tumor tissue (Figs. 2 and 3).

The accumulation of Indium-111-girentuximab-IRDye800CW in tumor and normal tissue kidney samples is shown in Table 2. Indium-111-girentuximab-IRDye800CW accumulated preferentially in tumor tissue (mean, 0.12 %ID/g; range, 0.01–0.33 %ID/g). Accumulation in normal kidney parenchyma was low (mean, 0.02 %ID/g; range, 0.00–0.04 %ID/g). In the first kidney perfusion, no radioactivity was used and therefore antibody accumulation could not be quantified. In experiment 2, antibody accumulation in tumor tissue was not significantly higher than in normal tissue, most likely because the tumor tissue was largely necrotic. In that experiment, high antibody accumulation (0.15 %ID/g) was found in the renal vein tumor thrombus. In the other five kidney specimens, accumulation of girentuximab in tumor tissue was significantly higher than in normal kidney tissue (P < 0.05). The mean T:N ratio ranged from 4 to 14.

Perfusion of two radical nephrectomy specimens (#6 and #7) with a mixture of girentuximab-IRDye800CW and a control antibody-IRDye800CW labeled with I-131 and I-125 showed specific accumulation of dual-labeled girentuximab in tumor tissue. Uptake of Iodine-131-girentuximab-IRDye800CW in tumor tissue (mean, 0.14 %ID/g; range, 0.01–0.31 %ID/g) was significantly higher than uptake of the irrelevant control Iodine-125-IgG-IRDye800CW in tumor tissue (mean, 0.02 %ID/g; range, 0.01–0.03 %ID/g, P < 0.001). The latter concentration was in the same range as the concentration of girentuximab in normal kidney parenchyma (mean, 0.02 %ID/g; range, 0.01–0.02 %ID/g; Table 2; Supplementary Figs. S6 and S7).

Finally, microscopic analysis of the tissue sections confirmed that the intratumoral distribution of the radioactive signal was congruent with that of the fluorescent signal. Most importantly, both signals colocalized with CAIX expression (Fig. 4). In normal kidney parenchyma, the dual-labeled girentuximab was barely detectable (Supplementary Fig. S4). H&E staining of the sections showed that cell morphology after perfusion was normal, indicating that the continued perfusion did not lead to cell damage (Supplementary Fig. S4).

This study demonstrated the feasibility of tumor-targeted dual-modality imaging in ccRCC using dual-labeled girentuximab ex vivo. Tumors were clearly distinguishable from normal kidney parenchyma both by fluorescence and radionuclide imaging when tumor-bearing kidneys were perfused with dual-labeled girentuximab. The specific localization of the dual-labeled antibody in CAIX-expressing tumor tissue allowed excellent tumor visualization. Colocalization was observed between the fluorescent signal, the radioactive signal, and CAIX expression in tumor tissue. These results, in a system closely resembling the clinical application, bridge the gap between preclinical research and clinical application of dual-labeled girentuximab and have led to the initiation of the first targeted dual-modality image-guided surgery study in patients with ccRCC (clinicaltrials.gov NCT02497599).

As more complex renal tumors are increasingly treated by partial nephrectomy these days (32, 33), intraoperative dual-modality imaging can be useful to improve intraoperative tumor detection and improve radical tumor resection. Particularly in more complex tumors, the percentage of positive surgical margins can be as high as 18% (34). Although no consensus exists about the prognostic impact of positive surgical margins, the uro-oncological surgeon should aim for radical tumor resection (34, 35). Targeted dual-modality imaging using dual-labeled girentuximab may be a valuable imaging technique during surgery of ccRCC patients. The high tissue penetration depth of gamma radiation for tumor localization combined with fluorescence imaging for real-time precise tumor delineation forms a powerful combination to improve surgical outcome. The advantage of the dual-labeling strategy over coinjection of separate radio- and fluorescently labeled probes is that both imaging signals originate from the same tracer molecule (20, 36).

Preclinical studies have already demonstrated the feasibility of dual-modality imaging using various monoclonal antibodies in several animal tumor models, for example, cetuximab (anti-EGFR; ref. 29), panitumumab (anti-HER1; ref. 37), trastuzumab (anti-HER2; refs. 37, 38), TRC105 (anti-CD105; ref. 39), D2B (anti-PSMA; ref. 16), and girentuximab (anti-CAIX; refs. 17, 18, 29) in head and neck cancer, breast cancer, prostate cancer, and renal cell carcinoma, respectively. However, extrapolation of results of these studies in mouse-tumor models to the clinical setting is cumbersome, because tumor models and the experimental setup only partly reflect the clinical situation. To facilitate clinical translation of dual-modality imaging, in this study, an ex vivo model was used to evaluate the in vivo targeting properties of dual-labeled antibodies (26, 40). This model has several advantages compared with the mouse-tumor models. Because human tumor-containing tissue is used, the tumor size, antigenic make-up, vascularization, and growth pattern of the tumor reflect the clinical situation. Furthermore, as the sensitivity of clinical imaging systems differs from preclinical imaging systems, we used both types of imaging systems in this study (i.e., closed cabinet-type fluorescence imaging systems often used in preclinical research and a hand-held laparoscopic fluorescence camera system used for clinical applications). By selecting an antibody concentration of 3.4 mg/L (1.2 mg in 350 mL of Ringer's Lactate solution), we mimicked the initial plasma concentration in clinical trials (administration of 5–50 mg girentuximab per patient; refs. 22–25). Clearance of the nonbound antibody from the circulation was mimicked by washing with Ringer's lactate. However, this does not reflect the complex clearance of antibodies in the human body. Nevertheless, we found a maximum accumulation of dual-labeled girentuximab of 0.3 %ID/g in tumor regions, which is in the same range as the accumulation that was found in clinical studies with radiolabeled girentuximab (up to 0.5 %ID/g; refs. 22, 41). Therefore, our results may be a good predictor for the intensity of the dual-modality signal that can be expected during surgery. However, one must realize that real-time intraoperative imaging involves many challenges that were not encountered in this imaging model of the lamella, for example, the limited movability of a camera during laparoscopy, 3D optical imaging, gamma probing, and an increased noise level because of tracer accumulation in the liver and blood. These aspects have to be evaluated in a clinical trial.

The kidneys included in this study contained large renal cell carcinomas, including tumor thrombi. This differs from the small renal masses where we envision that intraoperative dual-modality imaging might be particularly useful. However, inclusion of specimens with smaller tumor masses was not possible as in these cases usually partial nephrectomy is indicated, and partial nephrectomy specimens lack the renal artery to perform the perfusion. However, the added value of dual-modality imaging for tumor detection and resection was elegantly illustrated after perfusion of specimen #5. In this kidney lamella, a small satellite ccRCC lesion was identified by fluorescence imaging (see Fig. 3B) that had not been recognized on primary macroscopic inspection. Histopathologic analysis confirmed that this was a ccRCC lesion. According to literature, satellite lesions are found in approximately 7% of patients, depending on tumor stage (42).

A challenge for the clinical application of targeted imaging techniques will be intratumoral heterogeneity. Large variation in antibody accumulation was seen within tumors. This might be attributable to differences in CAIX expression, tumor viability, vascularization, and/or perfusion (41). We found an excellent spatial overlap between CAIX expression and the fluorescent and radioactive signal of dual-labeled girentuximab. Furthermore, antibody accumulation was found to be high at the tumor borders, which are clinically the most relevant regions. Another concern was that the accumulation of girentuximab could be partly due to the enhanced permeability and retention (EPR) effect, because of the enhanced vascular permeability in tumor tissue (43). Therefore, we performed a dual-isotope (¹³¹I and ¹²⁵I) control experiment in two resected tumorous kidneys. Because accumulation of the control IgG in tumor tissue was much lower than accumulation of girentuximab (Table 2), we conclude that the girentuximab accumulation is mainly due to specific interaction of girentuximab with the CAIX antigen in the tumor tissue. Because the concentration of the dual-labeled antibody is similar to the plasma concentration and because the concentration in tumor tissue ex vivo and in vivo is in the same order of magnitude, the accumulation of the antibody in the tumor due to the EPR effect is expected to be similar ex vivo and in vivo. This also implies that targeted imaging using girentuximab can only be used in CAIX-expressing tumors. However, with a suitable tumor-targeting agent, dual-modality imaging can be used in any type of cancer surgery.

The first clinical trials using fluorescently labeled tumor-targeting probes [folate-FITC (11), cetuximab-IRDye800CW (44), and GE-137 (45)] have clearly demonstrated the potential of targeted intraoperative fluorescence imaging for improved intraoperative or endoscopic tumor visualization, which may lead to improved surgical outcome in clinical oncology. Furthermore, improved delineation of tumor margins may improve locoregional control, shorten duration of surgery, and reduce morbidity because a more limited resection may be performed (44). Fluorescence imaging is particularly useful for superficially located tumors, because the penetration depth of the fluorescent signal is limited (28). Deeper located tumor lesions, for example intraparenchymal tumors or metastatic lymph nodes covered with fat, may be missed by fluorescence imaging. For these applications, the addition of a radiolabel to the imaging probe can be valuable. When the tumor has been localized intraoperatively using the signal emitted by the radionuclide, resection can be performed using fluorescence-guided surgery (12). Finally, the surgical cavity can be examined for remnant disease with dual-modality imaging. Another advantage of dual-modality imaging is that a PET/CT or SPECT/CT can be obtained preoperatively. This gives the surgeon an indication of the location of the primary tumor and/or metastases. Hence, preoperative and intraoperative imaging can be performed with the same imaging probe. Thus, targeted dual-modality imaging could revolutionize surgical oncology.

No potential conflicts of interest were disclosed.

Conception and design: M.C.H. Hekman, O.C. Boerman, E. Oosterwijk, P.F.A. Mulders, M. Rijpkema

Development of methodology: M.C.H. Hekman, O.C. Boerman, M. de Weijert, E. Oosterwijk, P.F.A. Mulders, M. Rijpkema

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.C.H. Hekman, D.L. Bos, P.F.A. Mulders, M. Rijpkema

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.C.H. Hekman, E. Oosterwijk, H.F. Langenhuijsen, P.F.A. Mulders, M. Rijpkema

Writing, review, and/or revision of the manuscript: M.C.H. Hekman, O.C. Boerman, D.L. Bos, E. Oosterwijk, H.F. Langenhuijsen, P.F.A. Mulders, M. Rijpkema

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.C.H. Hekman, M. de Weijert, D.L. Bos, E. Oosterwijk

Study supervision: M.C.H. Hekman, O.C. Boerman, E. Oosterwijk, H.F. Langenhuijsen, P.F.A. Mulders, M. Rijpkema

The authors thank Anja van Wincoop for her assistance in logistics on the operation room to obtain the radical nephrectomy specimens. They also thank Arie Maat for the preparation of the kidney lamellae and Tim van Oostenbrugge for his support during the experiments.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.