Co-administered antibody improves penetration of antibody-dye conjugate into human cancers with implications for antibody-drug conjugates

Abstract

Poor tissue penetration remains a major challenge for antibody-based therapeutics of solid tumors, but proper dosing can improve the tissue penetration and thus therapeutic efficacy of these biologics. Due to dose-limiting toxicity of the small molecule payload, antibody-drug conjugates (ADCs) are administered at a much lower dose than their parent antibodies, which further reduces tissue penetration. We conducted an early-phase clinical trial (NCT02415881) and previously reported the safety of an antibody-dye conjugate (panitumumab-IRDye800CW) as primary outcome. Here, we report a retrospective exploratory analysis of the trial to evaluate whether co-administration of an unconjugated antibody could improve the intratumoral distribution of the antibody-dye conjugate in patients. By measuring the multiscale distribution of the antibody-dye conjugate, this study demonstrates improved microscopic antibody distribution without increasing uptake (toxicity) in healthy tissue when co-administered with the parent antibody, supporting further clinical investigation of the co-administration dosing strategy to improve the tumor penetration of ADCs.

Conflict of interest statement

E.L.R. acts as consultant for LI-COR Biosciences Inc. and has equipment loans from this company. All other authors declare no competing interests.

Figures

Fig. 1. Overview of the study design.

Study patients received infusion of antibody–dye conjugate with or without a loading dose of unlabeled antibody. One to five days post infusion, patients underwent surgical tumor resection. Fresh tissue samples obtained from the primary tumor were homogenized to quantify antibody–dye concentration in the tissue. Whole-tumor specimen were then formalin-fixed and cut into 5 mm thick tissue sections for macroscopic imaging. Subsequently, 5 µm histological slides were prepared from each tissue paraffin block made from the 5 mm tissue sections for microscopic imaging. Antibody uptake and distribution were measured from tissue homogenates, macroscopic, and microscopic imaging. Some elements of this figure were created with BioRender.com.

Fig. 2. Co-administration of unlabeled antibody reduced muscle uptake while maintained tumor uptake.

a Muscle antibody uptake (%ID kg−1: percent injected dose per kilogram) decreased with increasing total monoclonal antibody (mAb) dose (owing to loading dose (LD)). b, c. Tumor and skin uptake (%ID kg−1) as normalized by muscle uptake (%ID kg−1) were significantly higher in the loading dose group than that in the non-loading dose group, suggesting that tumor and skin were not fully saturated within the total antibody dose range of 0.3 mg kg−1 to 2.6 mg kg−1. d When tumor uptake was normalized by skin uptake, there was no significant differences between groups. The number of independent patient samples available in b: n = 9 (LD), n = 11 (non-LD); cn = 8 (LD), n = 9 (non-LD); dn = 8 (LD), n = 10 (non-LD). p value = 0.010 b, 0.046 c. Mann–Whitney U test (two-tailed) were used for comparison in b–d. Graphs plotted mean with standard deviation. (*p < 0.05. **p < 0.01, ns: p > 0.05).

Fig. 3. Antibody–dye distribution demonstrated both macroscopic and microscopic heterogeneity within the solid tumor.

a Antibody–dye uptake (NIR: near-infrared) was higher in the tumor periphery or edge (especially the epithelium) than that in the tumor interior (Scale bar: 2 mm). b, c Antibody penetrated deeper into tumor nests located close to the edge of the bulk tumor than tumor nests located further away from the tumor edge, which was not attributable to EGFR (epidermal growth factor receptor) expression (Scale bar: 50 μm). d EGFR expression level is similar across the entire tumor (Scale bar: 2 mm). e The fluorescence intensity negatively correlated with increasing distance from the edge of the bulk tumor. f The fluorescence intensity plot depicted varying degree of antibody penetration into tumor nests: 1. fully saturated tumor nests at the periphery; 2. partially saturated tumor nests; 3. poorly penetrated tumor nests at the center of the tumor. (The shaded area in e and f means standard deviation). n = 3 tumor nests from the same patient. g Vessel staining (ERG) corresponding to region 1, 2, 3 in f (Scale bar: 50 μm).

Fig. 4. A loading dose of unlabeled antibody significantly reduced heterogeneity of antibody distribution in tumors, which can be captured by microscopic imaging but not macroscopic imaging.

a Schematic workflow of reconstructing the whole-tumor fluorescence distribution and measuring antibody delivery by macroscopic imaging. b–e The tumor uptake as measured by the MFI, and the tumor antibody distribution quantified by IQR,entropy, and uniformity from macroscopic imaging showed no significant difference between the two dosing groups. b–e Tissue specimens were available in n = 12 patients in the LD group and n = 12 in the non-LD group. f Schematic workflow of measuring antibody delivery in tissue histological sections and tumor microenvironmental factors using microscopic imaging. g–j The tumor uptake measured by the MFI from microscopic imaging also showed no difference between the two groups, but the antibody distribution measured by IQR, entropy, and uniformity from microscopic imaging showed significantly higher heterogeneity in the non-LD group (*p < 0.05, ns: p > 0.05). p value = 0.030 h, 0.036 i, 0.025 j. k–n Vessel area fraction, EGFR area fraction, αSMA area fraction, and tumor size were not statistically different between the LD and the non-LD group. (Scale bar in a and f: 2 cm; graphs plotted mean with standard deviation). g–n Tissue specimens were available in n = 12 patients in LD group and n = 10 in the non-LD group. Mann–Whitney U test (two-tailed) was used in b–e, g–n.

Fig. 5. Antibody–dye has more homogenous spatial distribution across tumor when co-administered with the unlabeled antibody.

a Each tumor was gridded into small square regions and antibody penetration within each square region was quantified as the ratio of fluorescent positive area to EGFR-positive area. b The ratio of fluorescence to EGFR over the entire tumor area was not significantly different between the LD and non-LD group (Scale bar: 5 mm). c–e Comparison of three metrics, including interquartile range (IQR), Pearson’s correlation coefficient between fluorescence and EGFR expression and CLCM correlation, all reflected less heterogeneity in antibody penetration across tumors in patients with loading dose (*p < 0.05; ns: p > 0.05). f, g Co-administration of parent antibody could overcome binding site barrier and improve antibody–dye distribution within the tumors. f As fluorescent antibody extravasates from blood vessels (red tube), it immediately binds EGFR in the tissue (represented by cells in the cube). g By co-administering non-fluorescent antibody, the same amount of fluorescent antibody reaches the tissue, but these antibodies compete for binding sites locally in the tissue. This results in improved distribution of fluorescent antibody at the microscopic scale (Scale bar: 250 μm). Mann–Whitney U test (two-tailed) was used in b–e. p value = 0.025 c, 0.025 d, 0.036 e. The number of independent patient samples available for b–e: n = 12 patients in the LD group and n = 10 in the non-LD group. Graphs plotted mean with standard deviation.

Fig. 6. The impact of loading dose on total antibody uptake.

a Tumor uptake (%ID kg−1) showed slightly decreasing trend with increasing total antibody dose. b Tumor uptake (%ID kg−1) normalized by vessel area fraction showed a constant trend with increasing total antibody dose, suggested the variations in vessel area between the two groups may have caused the downward trend in tumor uptake before normalization. c–h Examples from a LD patient and non-LD patient demonstrated partial saturation in the tumor periphery with loading dose. (Scale bar in c and f: 2 mm; scale bar in d, e, g, h: 100 μm). Linear regression was performed in a, b with the best-fit line (solid line), the 95% confidence bands (dotted line), goodness of fit (R2), and p value.