Pilot study of ⁸⁹Zr-bevacizumab positron emission tomography in patients with advanced non-small cell lung cancer

Bevacizumab was labeled with zirconium-89 using N-succinyl-desferrioxamine (N-suc-Df) as described previously [18],[19]. All procedures were performed under aseptic conditions in a shielded laminar flow cabinet. In short, desferrioxamine (Desferal (Df), Novartis, Basel, Switzerland) was converted to N-succinyl-Df. Next, the hydroxamate groups were blocked with iron, and the succinic acid group converted to its tetrafluorophenol ester. The resulting Fe-N-suc-Df-TFP ester was reacted with bevacizumab under basic conditions (pH 9.5 to 9.7) for 30 min at room temperature. Subsequently, iron was removed using an excess of ethylenediaminetetraacetic acid (EDTA) for 30 min at 35°C, and the resulting N-suc-Df-bevacizumab was purified over a e and radiolabeled with ⁸⁹Zr for 60 min in HEPES buffer at room temperature. Finally, ⁸⁹Zr-N-suc-Df-bevacizumab was purified over a PD-10 column using 5 mg/mL gentisic acid in 0.9% NaCl (pH 4.9 to 5.4). The mean labeling efficiency was 71.1% ± 5.3%. The product was formulated and filter sterilized to 5 mg bevacizumab and 37 MBq Zr-89 per patient injection. These procedures resulted in a sterile final product with endotoxin levels <2.5 EU/mL. The radiochemical purity was >97% according to iTLC and HPLC, and the immunoreactivity as determined by an ELISA assay was >75%, which is optimal for this assay.

All patients underwent a routine FDG PET/CT scan within 4 weeks prior to starting therapy. One week prior to therapy, the patients were injected intravenously with 36.4 ± 0.9 MBq ⁸⁹Zr-bevacizumab, and 10-min static PET/CT scans of the thoracic NSCLC lesion were performed on days 4 and 7 post-injection. These parameters were based on previous ⁸⁹Zr-trastuzumab analysis, showing appropriate doses and timing for visualization and quantification of uptake in HER2-positive tumors to be 37 MBq and days 4 to 5 after injection, respectively [20].

The scans were performed on a Gemini TF-64 PET/CT scanner (Philips Medical Systems, Cleveland, USA). PET data were normalized, and all appropriate corrections were applied for dead time, decay, randoms, and scatter. Reconstruction of PET data was performed using the BLOB-OS-TF reconstruction algorithm with CT-based attenuation correction, resulting in a final voxel size of 4 × 4 × 4 mm³, matrix of 144 × 144 × 45, and a spatial resolution of 5 to 7 mm full-width at half maximum [21]. Additional smoothing was performed, as this was shown to optimize quantitative accuracy and harmonized image quality [22].

FDG PET/CT scans were used to identify the exact location of the tumors. The PET data were analyzed using in-house-developed analysis tools within the IDL environment (Interactive Data Language Virtual Machine 6.2, RSI Inc., Boulder, CO, USA). Volumes of interest (VOI) were drawn manually on the CT images around the contours of the primary tumor (PT) and, if present, lymph node metastases (LNM) and non-lymph node metastases (NLNM). Additionally, VOI were drawn within the descending aorta (AD) and non-tumor tissue muscle (M), healthy lung (HL), and fatty tissue (FT). Standardized uptake value (SUV) parameters were calculated by normalizing VOI activity concentrations to injected dose and patient weight: SUV_peak (spheric VOI of 1.2-cm diameter positioned around the voxel with the highest uptake) and SUV_mean (mean activity in VOI/cc) from the AD VOI to calculate the tissue-to-blood ratio (TBR). SUV_peak and TBR for the abovementioned VOI were then compared for the scans at days 4 and 7.

Statistical analysis was performed using SPSS software (SPSS for Windows 15.0, SPSS, Inc.). Correlations were explored using Spearman’s correlation coefficient (r _s ). A two-tailed probability value of P < 0.05 was considered significant. Values in the text and tables are presented as mean ± standard deviation (SD) unless stated otherwise.

Prior to this study, it was unclear whether NSCLC tumors could be imaged by ⁸⁹Zr-bevacizumab. The results of this pilot study show that all tumor lesions (PT, LNM, and NLNM) had a higher ⁸⁹Zr-bevacizumab uptake as compared to non-tumor background tissues, allowing for visualization and analysis of tumor ⁸⁹Zr-bevacizumab uptake.

This increased uptake may be caused by accumulation of VEGF in the tumor, due to high paracrine expression and subsequent binding to extracellular matrix glycoproteins such as heparan sulfate proteoglycans (HSPG) and neuropilins (NRP). These glycoproteins act as non-signaling co-receptors that facilitate binding of VEGF to VEGFR molecules [6]. Another mechanism that may contribute is the internalization of ⁸⁹Zr-bevacizumab into cells within the tumor. After internalization, the ⁸⁹Zr label may become trapped in the lysosomes and show up on the PET scan [23].

Our findings are in accordance with the limited number of publications of previous preclinical and clinical data showing that high-VEGF-A-expressing tumors are associated with high tumor-to-background ⁸⁹Zr-bevacizumab uptake. Nagengast et al. [19],[24] showed that high-VEGF-producing SKOV-3 ovarian tumor xenografts had a higher uptake of ⁸⁹Zr-bevacizumab than of ⁸⁹Zr-labeled IgG, which served as a control. In a follow-up study, the same researchers found that ⁸⁹Zr-bevacizumab uptake decreased in (NVP-AUY922 sensitive) A2780 ovarian tumor xenografts after 2 weeks of antitumor therapy (using NVP-AUY922), while in the therapy-resistant CP70 xenografts, ⁸⁹Zr-bevacizumab uptake did not change. van der Bilt et al. [25] showed a decrease in ⁸⁹Zr-bevacizumab uptake in A2780 ovarian tumor xenografts after 2 weeks of everolimus therapy, while ⁸⁹Zr-bevacizumab uptake remained unchanged in other organs, matching ex vivo measures of VEGF-A levels [19],[24],[25]. Recently, a first clinical study using ⁸⁹Zr-bevacizumab was reported. In breast cancer patients, tumors with elevated VEGF-A levels showed higher ⁸⁹Zr-bevacizumab uptake than the background of healthy breast tissue [26].

We found that tumor-to-background ratios were high at days 4 and 7; however, tumor-to-blood ratios were higher on day 7, due to a relative decrease of blood activity concentrations as compared to tumor activity concentrations. However, image quality at day 7 was hampered due to physical decay of the tracer. The optimal time for ⁸⁹Zr-bevacizumab seems to be between 4 and 7 days post-injection.

Lymph node metastases and especially NLNM showed higher uptake than pulmonary lesions. This may be caused by the fact that pulmonary lesions suffered more from breathing movement-induced partial volume effects than LNM and NLNM, which were relatively fixed to bone and peritoneum. Although all tumoral lesions were visible in this study, the quantification of uptake in small-volume lesions needs to be interpreted with caution, as these tumors suffer even more from this partial-volume-induced underestimation of tracer uptake.

Although our results did not show a significant correlation between tracer uptake and PFS, a positive trend was observed. As tumor ⁸⁹Zr-bevacizumab uptake may represent the level of tumor VEGF, this technique might offer a predictive biomarker for bevacizumab treatment efficacy. At present, there is an absence of clinically useful predictive biomarkers. For example, several blood biomarkers, such as VEGF-A and NRP-1 using newly developed sensitive assays, have been proposed to predict bevacizumab treatment efficacy, but their value is still the subject of debate [6],[27].

As this was a pilot study, only a limited number of patients were included. However, the results obtained were consistent for all patients, indicating that larger clinical trials are warranted.

Another limitation was the absence of arterial blood sampling for blood and plasma activity and metabolite analysis, which could have provided a more accurate quantification of tracer uptake. However, because patients already underwent several PET/CT scans, additional blood sampling for research purposes was considered too high a burden. Furthermore, previously published data show a good correlation between image-derived activity from ⁸⁹Zr-labeled antibodies and blood activity [15],[17]. Additionally, ⁸⁹Zr-bevacizumab was found to be highly stable in plasma, as only a 6% decrease in protein-bound radioactivity was seen after 168 h (stored in serum at 37°C) [19]. Although NSCLC was initially diagnosed in all patients, no extra tumor biopsy was taken prior to scanning for VEGF-A staining, again because this was considered too burdensome.

The results of this study show tumor specific uptake of ⁸⁹Zr-bevacizumab. Future studies should consider not to include small lesions that could suffer from partial volume effects. Furthermore, the optimal timing should be investigated, as this can be expected to be between day 4 and day 7 post-injection. To better understand the physiological processes causing tracer accumulation, concurrent pathology data (e.g., angioproliferative markers) should be assessed by taking biopsies prior to scanning. Quantification of uptake may be improved by using blood sampling for plasma radioactivity assessment.

The effect of perfusion on ⁸⁹Zr-bevacizumab should be analyzed. In this study, an assessment of tumor blood perfusion, e.g., with [¹⁵O]H₂O PET scan, was not performed. Using [¹⁵O]H₂O PET scans, van der Veldt et al. [28] showed that the intratumoral distribution of docetaxel followed the tumor perfusion patterns. Therefore, tumor perfusion data may have provided additional understanding of the distribution of tumoral ⁸⁹Zr-bevacizumab uptake.

⁸⁹Zr-bevacizumab PET may be used as an imaging agent, as tracer uptake showed a trend towards a positive correlation with PFS and OS. It should be noted that patients were treated with carboplatin-paclitaxel-bevacizumab (CPB) followed by bevacizumab maintenance therapy. PFS was the result of both CPB therapy and bevacizumab. In future studies, ⁸⁹Zr-bevacizumab scans should ideally be performed at two time points, i.e., prior to CPB therapy and also prior to bevacizumab maintenance therapy. This is because the post-CPB-altered tumor VEGF status is unknown; however, this new post-CPB VEGF status may be a better predictor for sensitivity to subsequent bevacizumab maintenance therapy.