Improved Discrimination of Tumors with Low and Heterogeneous EGFR Expression in Fluorescence-Guided Surgery Through Paired-Agent Protocols

Purpose The goal of fluorescence-guided surgery (FGS) in oncology is to improve the surgical therapeutic index by enhancing contrast between cancerous and healthy tissues. However, optimal discrimination between these tissues is complicated by the nonspecific uptake and retention of molecular targeted agents and the variance of fluorescence signal. Paired-agent imaging (PAI) employs co-administration of an untargeted imaging agent with a molecular targeted agent, providing a normalization factor to minimize nonspecific and varied signals. The resulting measured binding potential is quantitative and equivalent to in vivo immunohistochemistry of the target protein. This study demonstrates that PAI improves the accuracy of tumor-to-healthy tissue discrimination compared to single-agent imaging for in vivo FGS. Procedures PAI using a fluorescent anti-epidermal growth factor receptor (EGFR) affibody molecule (ABY-029, eIND 122,681) with untargeted IRDye 700DX carboxylate was compared to ABY-029 alone in an oral squamous cell carcinoma xenograft mouse model at 3 h after dye administration (n = 30). Results PAI significantly enhanced tumor discrimination, as compared to ABY-029 alone in low EGFR-expressing tumors and highly heterogeneous populations including multiple cell lines with varying expression (diagnostic accuracy: 0.908 vs. 0.854 and 0.908 vs. 0.822; and ROC curve AUC: 0.963 vs. 0.909 and 0.957 vs. 0.909, respectively) indicating a potential for universal FGS image thresholds to determine surgical margins. In addition, PAI achieved significantly higher diagnostic ability than ABY-029 alone 0.25–5-h post injection and exhibited a stronger correlation to EGFR expression heterogeneity. Conclusion The quantitative receptor delineation of PAI promises to improve the surgical therapeutic index of cancer resection in a clinically relevant timeline. Supplementary Information The online version contains supplementary material available at 10.1007/s11307-021-01656-3.


S2. Determination of paired-agent dose based on autofluorescence signal
In order to determine the effect of autofluorescence on the observed fluorescence emission intensity, mice (n = 3, per cell line) bearing tongue tumors were sacrificed without injection of fluorescent agents. The tongues were bisected along the raphe and imaged on the Odyssey CLx scanner, as described for paired-agent injected mice in the Methods. The average fluorescence intensity of the tumor and surrounding normal tongue were calculated and compared to mice with a humanequivalent micro-dose (30 nanomoles) of each agent (n = 3). Using the method of Reagan-Shaw (2007) [1], the mouse equivalent dose was determined to be 48.8 µg/kg and an average mass of 22 g was used for a final dose of 1.07 µg per mouse. So, 10 x dose is 10.7 µg per mouse. At a microdose, the ABY-029 signal was > 15 times the autofluorescence signal in the 800-channel; however, the IRDye 700DX signal was < 5 times the 700-channel autofluorescence. Therefore, 10 times human-equivalent microdose (300 nanomoles) of IRDye 700DX was tested and at 3-hours postadministration it was found that the fluorescence signal was ~6-10 times that of the autofluorescence (Fig 2b & Fig S1b). Therefore, it was determined that ABY-029 and IRDye 700DX will be administered in a 1:10 molar ratio (30:300 nanomoles). This discrepancy in the dose of the PAI agents is accounted for in the normalization factor (Equation 2).
In addition to determining the dose of paired-agent administration, the autofluorescence values of each tissue affects the observed binding potential values. Typically, autofluorescence is subtracted off of every post-injection image -on a pixel-by-pixel basis if images are taken without moving the animal, or on average if the animal is moved between pre-and post-administration images.
Here, it was found that pre-injection the BPs were -0.77 ± 0.03, -0.88 ± 0.004, -0.82 ± 0.008, and -0.83 ± 0.008 for normal tongue, FaDu, Detroit 562, and A431, respectively. The BP values for tumor are lower than that for normal tongue because the autofluorescence in the 700 nm channel is stronger than that in the 800 nm channel. This indicates that the BP values we report here are actually underestimating the real BP when autofluorescence is considered, and the difference between normal and tumor tissue would actually be greater if the autofluorescence was considered.

S3. Comparison of tumor-to-background ratio and contrast-to-variance ratio to measure image contrast
The normalization factor (NF) is an important part of calculating the binding potential (BP) as is could be artificially made to have high TBR values by simply altering the NF. CVR is a more robust measure that allows accurate comparisons between different imaging methods [2]. Note that the NF here is less than one because the IRDye 700DX is 10 times more concentrated than the ABY-029 for administration and therefore is brighter. The effect on TBR while changing the NF would be opposite if the targeted agent was brighter than the non-targeted agent.
For this study, the NF was determined in the tip of the tongue such that the resultant BP was equal to 0.5 (Equation 1). A BP of 0.5 was used instead of 0 in order to minimize pixels with a negative BP value in the normal tongue. For use during surgery, it is hypothesized that the surgeon could select a region of known normal tissue in order to determine the NF ratio for that patient. This would optimize visualization by creating a "zero background" signal in the normal surrounding tissues.

Figure S4. (a) PAI CVRs outperform ABY-029 SAI CVRs in most time points with less variance,
suggesting that BP provides a stable diagnostic accuracy and tumor contrast starting at early administration-to-imaging time. Figure S5. The effect of the normalization factor (NF) on image contrast. As the normalization ratio increases, the overall signal decreases. This results in a constant contrast-to-variance ratio (CVR) that is stable when NF is varied. However, tumor-to-background ratio (TBR) increases as  the signal in the normal tissue (the denominator) decreases. These results indicate that TBR is an unstable measure for binding potential and cannot be used to compare to single agent fluorescence.

S4. Spatial alignment of fresh tissue sections with pathological sections
Assessment of the spatial alignment performed between the IHC and the fluorescence images was performed by creating image overlays with the mshowpair function in Matlab ( Figure S3) [3].
"Checkerboard" creates an image with alternating rectangular regions from fluorescence and IHC images. Figure S6. Image overlays created using imshowpair in Matlab allow visual assessment of spatial alignment between IHC and BP images. In the Checkerboard overlay, IHC is in RGB (brown pixels) while the BP map is a grey scale image.

S5. Determination of EGFR per cell using flow cytometry
The methodology for the quantification of the number of EGFR per cell using flow cytometry has been described previously in detail [4]. Briefly, cells were purchased from and cultured as per  Table S1 and the average value plotted in Figure S5.