Longitudinal Imaging Using PET/CT with Collagen-I PET-Tracer and MRI for Assessment of Fibrotic and Inflammatory Lesions in a Rat Lung Injury Model

Non-invasive imaging biomarkers (IBs) are warranted to enable improved diagnostics and follow-up monitoring of interstitial lung disease (ILD) including drug-induced ILD (DIILD). Of special interest are IB, which can characterize and differentiate acute inflammation from fibrosis. The aim of the present study was to evaluate a PET-tracer specific for Collagen-I, combined with multi-echo MRI, in a rat model of DIILD. Rats were challenged intratracheally with bleomycin, and subsequently followed by MRI and PET/CT for four weeks. PET imaging demonstrated a significantly increased uptake of the collagen tracer in the lungs of challenged rats compared to controls. This was confirmed by MRI characterization of the lesions as edema or fibrotic tissue. The uptake of tracer did not show complete spatial overlap with the lesions identified by MRI. Instead, the tracer signal appeared at the borderline between lesion and healthy tissue. Histological tissue staining, fibrosis scoring, lysyl oxidase activity measurements, and gene expression markers all confirmed establishing fibrosis over time. In conclusion, the novel PET tracer for Collagen-I combined with multi-echo MRI, were successfully able to monitor fibrotic changes in bleomycin-induced lung injury. The translational approach of using non-invasive imaging techniques show potential also from a clinical perspective.

The procedure received from P. Caravan was performed but due to retention of activity on the C 18 -cartridge and the subsequent evaporation from ethanol was considered a too large drop in activity to perform the imaging study. In communications with Caravan the reason for using SepPak cartridges was to purify from residual Ge ions.1 As the Cu 64 is delivered with a specification of >99% purity as well as the ITLC analysis of the reaction proved the reaction to be higher yielding than 95%, it was deemed pure enough for a preclinical pilot study and the decision to inject the reaction solution as is after pH adjustment to 5.0 was made. The modified procedure was therefore changed according to discussion for adjusting the protocol without the use of SepPak.

B) Storage of CBP-NODAGA
The peptide was delivered to LBIC in December 2018 as a white powder. It was stored at -20°C. The vials were marked with 11/20/18, 0.61 mg, 8 doses and a vial number (#1-9) on delivery.

C) Radiolabelling with Cu 64
On the first radiolabelling of the week a vial containing CBP-NODAGA was thawed. To the vial was added 610 µL of 3.0 M NaOAc (pH 4.5) and the vial was shaken until all powder was dissolved. 300 µL of the solution was added directly into the vial containing Cu 64 . The remaining CBP-NODAGA solution was refrozen to be used on the second labelling the same week. The reaction vial was swirled gently to rinse down any activity remaining on the walls. pH was analyzed and if below 4.0 it was adjusted with additional 3.0 M NaOAc (pH 4.5) until pH approximately 4.1 was obtained. After correct pH was achieved the vial was sealed with parafilm and heated for 20 min to 60°C. Labelling efficiency was monitored by applying 1 µL to an ITLC strip and subsequent elution with 0.2 M citric acid. Radiolabelling yield was over 95% and solutions were used for injection and imaging without further purification. Note that these results prove that less than 5% of Cu 64 is free in solution, this does not necessarily mean that all peptide conjugates in the solution is labelled. After cooling for a few minutes the pH was adjusted to 4.5 using 3.0 M NaOH when needed, typically 40 µL. The vial, containing 4 doses of peptide and 600-800 MBq radioactivity in approximately 340µL buffer was then sealed and delivered as is to the imaging lab. For the first 2 production days, the delivery day was the day after production and the procedure used a SepPak to further purify the product, but as the study resumed in October 2019 it was delivered immediately after analysis as described above.

D) Analytical Results
The ITLC strips were analyzed according to the following procedure: 1 µL was applied to an ITLC strip ad it was eluted in 0.2M Citric Acid. The point of origin contains the radioconjugate and anything eluted with the solvent front on the strip is free Cu 64 according to figure 1. Figure   2 shows the elution profile for the labelled profile.

Stability
To ensure stability of the radiolabelled peptide a stability study was performed on the first radiolabelled product from 2019-04-02. The product solution was injected onto a 1260 Agilent HPLC, equipped with a µBondapakC-18 column. Eluent was MeCN/H2O with 1% TFA gradient system.
As can be seen from the radiosignal chromatograms on the following page the readout from the radioactivity did not change in respect to retention time during the 30 hours it was monitored making it reasonable to believe that the tracer solution was stable before in vivo administration. This result is accordance with the stability of the previous labelling stability with Ga 68 performed with ITLC.

Urine sample
A urine sample was obtained after the first imaging session (2019-04-02) which was analyzed on the same HPLC setup as previously mentioned on the day after the imaging study. As can be seen from the chromatograms below there is either a new adduct or free ions of Cu 64 shown as a peak at retention time 4.2 min. Even the day after urine collection there is also a significant amount of Cu 64 -CBP-NODAGA, retention time at 9.5 min. The ratio between the two is 68:32.

I) Organ distribution of tracer during live imaging
A pilot study was performed before the main study. Below in Figure 1, three examples of rats are presented from tracer injections and PET signal detection. This was mainly done for evaluation of approximate organ distribution at 1-24h post-injection. Both bleomycin challenged rats and Saline control were injected with the PET-tracer and were imaged at different time points. The kidney and liver showed marked high signal up-take at all time points. Figure 1: Images from PET overlaid on CT images. Organs distribution of the CBP-Cu 64 tracer during live imaging of one saline-challenged rat and two bleomycin-challenged rats, here shown in the coronal, sagittal and transverse plane, with all internal organs present within the FOV.

II) Organ distribution of tracer ex vivo
Organs were collected 24h post-injection, after the last PET imaging, weight was noted and counts measured on an Automatic Gamma counter (WIZARD® 1480 RiaCalc WIZ, PerkinElmer). There is a marked high up-take in the liver and the kidneys. However, the difference between the Saline and Bleomycin-challenged rats clearly differ when comparing the lung up-take.

Image analysis from MRI scans
The semi manually drawn regions of interest (ROI) were defined according to a template in each MRI slice, in a healthy rat, and followed as a map for all following scans. Example of the template ROI in three various slice positions are presented in Figure 1.
Once each ROI was completed within one animal scan, histograms of the data contained in the ROI were exported for both TESHORT and TELONG. The MRI histograms were batch analyzed using custom software written for the purpose in IDL 8.5 (Exelis VIS, Harris Corporation, Boulder, Colorado, USA). The software identifies the first peak in each histogram, representing the low-signal voxels in the normal lung. The point on the slope at the right side of this peak with the sharpest decline is then found and a linear function tangential to the slope is calculated. The sum of the volume in the histogram with signal above where the linear function intersects with the x-axis was defined as the MRI "high-signal" volume for each histogram, as shown in Supplementary Figure below. Quantification of the signal within the lung-ROI was done according to histogram analysis and threshold-based assessment in order to define the so called high-signal region within the lungregion. The two different TE were assessed for high-signal and the volume of these regions were acquired. Once defining the regions for TELONG and subtracting the regions representing the vessels (as only high-signal in the control animals), the "fluid" signal was quantified, representing the inflammation and edema within the lung region. The "tissue" signal was quantified by a subtraction of the TELONG image (fluid and vessels) from the TESHORT image (fluid, tissue and vessels). Thereby the fluid signal as well as signal from vessels are removed and only tissue remains

Image analysis from PET-CT scans
The MRI data as well as the ROI was resampled into isotropic voxels before the PET and CT scans of the same animal, at the same time point, were imported into the software session. To match these disparate datasets, MRI and ROI were used as reference while an automatic registration was performed of PET and CT datasets to the MRI datasets. The settings of Rigid translation, and Fine quality, was employed in VivoQuantTM 3.5 which uses a mutual information metric to fit the datasets together using translations and rotations during a maximum of 200 iterations. Post registration, the activity distribution in the PET image within the lung ROI was measured, both as activity uptake per mm 3 and total uptake in the whole lung, and divided by the decay-corrected injected dose per animal to find the fractional uptake. Subsequently, by using the histogram based analysis of the drawn total lung region, the "highsignal" from the MR-images were expressed. The high-signal areas from TELONG and TESHORT were then segmented as lesion-areas and PET signal was assessed within these regions as well as outside (e.g. health appearing lung areas). Figure 2: Histogram generated from one healthy lung region (Saline) and one Bleomycinchallenged rat, indicating the MRI signal in relation to the amount of voxels. High-signal voxels consist of signal from the vessels (alone in healthy lungs) and also lesions (besides the vessels in Bleomycin rats).