Purpose:

Fluorescence molecular imaging, using cancer-targeted near infrared (NIR) fluorescent probes, offers the promise of accurate tumor delineation during surgeries and the detection of cancer specific molecular expression in vivo. However, nonspecific probe accumulation in normal tissue results in poor tumor fluorescence contrast, precluding widespread clinical adoption of novel imaging agents. Here we present the first clinical evidence that fluorescence lifetime (FLT) imaging can provide tumor specificity at the cellular level in patients systemically injected with panitumumab-IRDye800CW, an EGFR-targeted NIR fluorescent probe.

Experimental Design:

We performed wide-field and microscopic FLT imaging of resection specimens from patients injected with panitumumab-IRDye800CW under an FDA directed clinical trial.

Results:

We show that the FLT within EGFR-overexpressing cancer cells is significantly longer than the FLT of normal tissue, providing high sensitivity (>98%) and specificity (>98%) for tumor versus normal tissue classification, despite the presence of significant nonspecific probe accumulation. We further show microscopic evidence that the mean tissue FLT is spatially correlated (r > 0.85) with tumor-specific EGFR expression in tissue and is consistent across multiple patients. These tumor cell-specific FLT changes can be detected through thick biological tissue, allowing highly specific tumor detection and noninvasive monitoring of tumor EFGR expression in vivo.

Conclusions:

Our data indicate that FLT imaging is a promising approach for enhancing tumor contrast using an antibody-targeted NIR probe with a proven safety profile in humans, suggesting a strong potential for clinical applications in image guided surgery, cancer diagnostics, and staging.

This article is featured in Highlights of This Issue, p. 2199

Translational Relevance

Near-infrared (NIR) fluorescence imaging is currently being evaluated for enhancing tumor contrast using cancer-targeted probes to enable accurate surgical resection of tumors while minimizing loss of healthy tissue. Our work uses an NIR fluorescent probe, panitumumab-IRDye800CW, which is a conjugate of panitumumab, an FDA approved EGFR-targeted mAb, and IRDye800CW, an NIR fluorescent agent which has been validated for safety in multiple human trials. We have performed several clinical trials to evaluate the safety of panitumumab-IRDye800CW probe in humans. The optical imaging technique used in our work employs safe and nonionizing NIR light. Several NIR fluorescence devices have previously been FDA approved for image guided surgery, and could serve as predicates for future FDA clearance of fluorescence lifetime (FLT) imaging devices. The translational relevance of this work therefore lies in the fact that FLT imaging in conjunction with panitumumab-IRDye800CW can be potentially extended to intraoperative image-guided surgery upon successful validation in clinical trials.

Fluorescence molecular imaging, using tumor-targeted fluorescent dyes, is being widely explored for tumor detection during surgeries and for cancer staging (1–6). Besides allowing enhanced accuracy for tumor margin assessment and identifying distant metastases (e.g., lymph nodes) in various surgical settings, the use of probes excitable in the near infrared (NIR) can allow noninvasive imaging of cancer in superficial lymph nodes and in intact organs such as the breast (7–9). A wide range of fluorescent agents have been developed for tumor targeting, including receptor-targeted NIR dyes (3, 10–14), small molecule agents (15), and protease activatable imaging agents (16), with some of these agents currently in late stage clinical trials (1). The use of therapeutic antibodies for receptors overexpressed in cancers is a powerful approach for tumor targeting, given that antibodies are more likely to be retained in cancer cells and are less complex and less expensive to manufacture than activatable probes (1). One such agent that has shown great promise is pantitumumab-IRDye800CW, a conjugate of the FDA-approved therapeutic antibody for the EGFR, panitumumab, with IRdye800CW, an NIR dye that has been tested in multiple human trials (2, 3, 10, 17–20). EGFR is an ideal target for fluorescence imaging because it is overexpressed in several cancers, including head and neck (21), lung (22), gliomas (23), and metastatic colorectal cancer (mCRC; ref. 24). Several recent clinical trials have shown that panitumumab-IRDye800CW is safe for human use and can enhance tumor contrast during fluorescence guided-surgical resections compared with visual inspection and palpation, and can differentiate benign from metastatic lymph nodes in patients with head and neck squamous cell carcinoma (HNSCC; refs. 3, 20).

Traditional imaging systems detect total fluorescence intensity, which cannot readily distinguish fluorescence arising from tumor-bound probe from nonspecific fluorescence. Therefore, an essential requirement for the success of conventional intensity-based fluorescence imaging techniques is the selective labeling of tumor cells in vivo with minimal nonspecific uptake. However, despite over 30 years of effort in developing new imaging agents and many promising clinical trials, cancer cell-specific labeling has not yet been demonstrated using exogenous agents in humans. Nonspecific probe accumulation in normal or benign tissue remains a major problem that significantly lowers relative tumor brightness compared with background and results in poor signal to noise ratio, low specificity (false positives), and low sensitivity (false negatives). Specialized techniques have been explored to overcome the poor targeting specificity of molecular imaging agents, including administration of a concurrent loading dose to improve probe uptake (25), paired-agent imaging with targeted and untargeted probes (26), or referencing schemes (17). However, these techniques have not been demonstrated to be sufficiently robust for widespread clinical adoption and have an arduous path to clinical translation (27, 28) due to the need for administering multiple agents.

Another approach to enhance tumor contrast in the presence of nonspecific fluorescence arising from poor targeting specificity is to exploit changes in photophysical properties of the imaging probe. Fluorescence lifetime (FLT), measured in absolute units (typically nanoseconds) is one such parameter (29, 30) that is robust to measurement conditions and can potentially alleviate many of the above issues related to nonspecific probe uptake. We recently showed using macroscopic time domain (TD) imaging of preclinical breast tumor models that the FLTs of tumors labeled with fluorescently tagged EGFR-antibody are significantly longer than the FLTs of healthy tissue, providing a dramatic specificity and sensitivity improvement over standard fluorescence intensity-based detection for tumor versus normal classification in situ and in vivo (31). To our knowledge, the efficacy of FLT imaging for enhanced tumor contrast using exogenous agents has not yet been evaluated in human tissues. Here we present microscopic and macroscopic NIR FLT imaging of human oral cancer specimens, resected from patients systemically injected with panitumumab-IRdye800CW. We show that FLT imaging provides unprecedented tumor specificity at a single cell level and can achieve nearly 98% sensitivity and 98% specificity for tumor to normal classification in thick tissue specimens. Tissue regions with long FLTs are shown to spatially colocalize with areas of high EGFR expression, and the average tissue FLT is shown to strongly correlate with level of EGFR expression and is consistent across multiple patients. Given the demonstrated safety of panitumumab-IRDye800CW in recent clinical trials (32), our work indicates the high potential for clinical translation of this probe for both in vivo and intraoperative FLT imaging applications for a wide range of EGFR-overexpressing cancers.

Panitumumab-IRDye800CW conjugation

Conjugation of panitumumab-IRDye800CW was performed under cyclic guanosine 3′,5′-monophosphate (cGMP) conditions, as described previously (3). Briefly, panitumumab (Vectibix; Amgen) was concentrated, and pH-adjusted by buffer exchange to a 10-mg/mL solution in 50 mmol/L potassium phosphate, pH 8.5. IRDye800CW (IRDye800CW-N-hydroxysuccinimide ester, LI-COR Biosciences) was conjugated to panitumumab for 2 hours at 20°C in the dark, at a molar ratio of 2.3:1. After filtration with desalting columns (Pierce Biotechnology) to remove unconjugated dye and buffer exchange to PBS, pH 7, the final protein concentration was adjusted to 2 mg/ml. The product was sterilized by filtration and placed into single-use vials and stored at 4°C until used.

Cell lines and tissue culture

EGFR-overexpressing cell line MDA-MB-231 and EGFR-negative cell line MCF7 were purchased from ATCC and cultured in high glucose DMEM supplemented with 10% FBS and 1% penicillin–streptomycin (Life Technologies). The oral cancer cell line FaDu was purchased from ATCC. FaDu cells were maintained in RPMI culture media supplemented with 10% FBS and 1% penicillin–streptomycin. Cells were harvested at 80% confluency for tumor induction. Cell lines were used for up to 30 passages after thawing from frozen stocks and were tested free of Mycoplasma at the time of receipt.

In vitro experiments

FaDu cells were plated at 0.2 × 106 cells per well in a 12-well plate containing poly-D-lysine–coated glass coverslips and were allowed to adhere to the coverslips for 24 hours. Cells were then incubated with panitumumab-IRDye800CW (100 µg), IgG-IRDye800CW (100 µg), or PBS (pH 7.4) for 2 hours at 37°C. After probe incubation, cells were fixed in 4% paraformaldehyde (PFA) and mounted with ProLong Gold Antifade medium (Thermo Fisher Scientific) for confocal fluorescence lifetime imaging microscopy (FLIM).

Patients

Resection specimens were obtained from a previous clinical trial approved by the Stanford University (Stanford, CA) Institutional Review Board (IRB) and the FDA (NCT 02415881), which enrolled 64 patients in a phase I, open-label study evaluating the safety and pharmacokinetics of panitumumab-IRDye800 to detect head and neck cancer during surgery. Written informed consent was obtained from all the patients enrolled in the study. The study was performed in accordance with the Helsinki Declaration of 1975 and its amendments, FDA's International Council for Harmonization-Good Clinical Practice guidelines, and the laws and regulations of the United States. Consented patients were systemically injected with panitumumab-IRDye800CW between 43 to 48 hours prior to surgery. Further patient and tumor characteristics are summarized in Table 1. Specimens were transferred from Stanford to Massachusetts General Hospital (MGH, Boston, MA) using a material transfer agreement (MTA) associated with an MGH-IRB approved protocol (IRB protocol no. 2019P000732) for FLT imaging of resection specimens. Ex vivo tissue from 10 panitumumab-IRDye800CW injected patients (from Stanford) and 2 noninjected patients (from MGH) were formalin fixed, dissected, and paraffin embedded. We examined 26 representative specimens from the 12 patients selected to realize a wide range of intra- and interpatient EGFR levels. This sample size was more than the minimum necessary sample size (n = 6) to establish a statistically significant difference between tumor and normal tissue FLTs with 95% power and α of 0.001.

Table 1.

Summary of patient and tumor characteristics.

Patient characteristics
Gender 
 Male, n (%) 8 (67) 
 Female, n (%) 4 (33) 
Age, years (median; range) 68 (42–73) 
Cancer origin, n (%) 
 Oral cavity 9 (75%) 
 Nasal cavity 1 (8%) 
 Cutaneous 2 (17%) 
Tumor sites Tongue (4), buccal mucosa (3), maxilla (1), L maxillary sinus (1), R posterior neck (1), R RMT (1), melanoma (1) 
Weight, kg (median; range) 84.15 (41–96.6) 
Dose of panitumumab-IRDye80CW (mg) 50 
Time between infusion to surgery, hours (median; range) 46 (43–48) 
Patient characteristics
Gender 
 Male, n (%) 8 (67) 
 Female, n (%) 4 (33) 
Age, years (median; range) 68 (42–73) 
Cancer origin, n (%) 
 Oral cavity 9 (75%) 
 Nasal cavity 1 (8%) 
 Cutaneous 2 (17%) 
Tumor sites Tongue (4), buccal mucosa (3), maxilla (1), L maxillary sinus (1), R posterior neck (1), R RMT (1), melanoma (1) 
Weight, kg (median; range) 84.15 (41–96.6) 
Dose of panitumumab-IRDye80CW (mg) 50 
Time between infusion to surgery, hours (median; range) 46 (43–48) 

Animal models

All animal studies were approved by the Institutional Animal Care and Use Committee in accordance with the animal welfare guidelines at the MGH. Eleven (4- to 6-week-old) female nu/nu mice were purchased from Charles River Laboratories Inc and were housed at the animal facility in MGH. Animals were quarantined for 1 week and kept in a normal diet with 12-hour light and dark cycle. After 1 week, animals were anesthetized with 3% isoflurane and subcutaneously injected with 2 × 106 FaDu (n = 4, EGFR overexpressing), MDA-MB-231 (n = 5, EGFR overexpressing), or MCF7 (n = 2, EGFR-negative) cells in 1:1 PBS:Matrigel mixture. Mice with MCF7 cells were also implanted with a slow release estrogen pellet to expedite tumor growth. Tumors were measured once every 2 days until they reached 5- to 10-mm diameter in one dimension.

Wide-field TD imaging

A previously published custom-built TD imaging system was used for in vivo animal studies and ex vivo imaging of paraffin blocks of human OSCC specimens. The small-animal imaging system consisted of a Supercontinuum laser and tunable filter (EXR-20, SuperK Varia, NKT Photonics, 80 MHz repetition rate; 400–850 nm tuning range) providing 770- ± 30-nm excitation, a multimode fiber (Thorlabs) delivering light to the sample, and a gated intensified CCD (LaVision, Picostar, 500 V gain, 0.1 to 1 second integration time, 150 ps steps, 256 × 344 pixels after 4 × 4 hardware binning). A digital micromirror device was used to expand the output of the optical fiber and delivered to the surface of the animal. The average total power across the illumination area (approximately 6 × 8 cm) was 10 to 20 mW. Fluorescence was collected in reflectance mode using an 835/70-nm band-pass filter (AVR Optics). TD fluorescence imaging was performed with a gate width of 500 ps and 150 ps steps for a total duration of approximately 6 ns per laser duty cycle of 12.5 ns. In vivo animal imaging was performed 48 hours after i.v. injection of panitumumab-IRDye800CW (150 µL, 1 mg/mL). Animals were sacrificed after imaging and tumors were immediately frozen in optimal cutting temperature (OCT) compound for FLIM, histology and IHC staining.

Multispectral imaging

Paraffin blocks of ex vivo clinical specimens were imaged in an IVIS Spectrum CT imaging system (PerkinElmer) using a 710-nm excitation and 760- to 840-nm emission wavelengths. Camera integration time was automatically adjusted during image acquisition and the Living Image software was used to extract the fluorescence images normalized to integration time. True fluorescence emission spectra of panitumumab-IRDye800CW and tissue autofluorescence were used as basis functions to perform a linear deconvolution of the multispectral images and the amplitudes of panitumumab-IRDye800CW and tissue autofluorescence were extracted.

FLIM

A STELLARIS 8 FALCON (Leica) FLIM system was used for NIR FLIM of 10-μm–thin tissue sections (murine tumors and clinical specimens). Imaging was performed using 730-nm excitation with 750-nm notch filter and detected with a HyD R detector operating within 770- to 850-nm range. A 10x, 0.4 NA objective was used for image collection and digital images with 512 × 512 pixels (2.275 µm/pixel), four line repetitions, and four line averages were obtained. TD data was collected using time-correlated single photon counting.

Histopathology and IHC

OSCC tumors with surrounding normal tissue were fixed in 10% formalin, embedded in paraffin, sectioned (10-μm thickness), and stained with hematoxylin and eosin (H&E) or processed for IHC. For IHC, 10-μm–thick paraffin-embedded tissue sections were dewaxed in xylene and rehydrated in decreasing concentration of alcohol. Antigen retrieval was performed with EDTA (pH 9.0) at subboiling temperature for 15 minutes. Tissue sections were incubated in 1:50 dilution of anti-EGFR antibody (catalog no. 4267, Cell Signaling Technology) overnight at 4°C. Secondary antibody was applied for 30 minutes at 37°C and slides were developed with DAKO horseradish peroxidase (HRP)-compatible DAB (catalog no. SF-4100, Vector Laboratories) and counterstained with Harris Hematoxylin. Images of H&E- and IHC-stained tissue sections were obtained using an inverted Keyence BZ-X810 microscope (Keyence). A Plan Apo 10x, 0.45 NA air objective (Nikon) and a monochrome charge-coupled device (colorized with LC filter) were used to capture images. Histology images were graded by two experienced pathologists from Stanford and MGH.

Widefield TD data analysis

TD fluorescence images were analyzed in MATLAB (MathWorks) using a custom software. TD data from individual pixels were plotted as time gate versus log(counts) (Fig. 1) and the FLT was obtained by fitting the decay portion of TD fluorescence profiles to a single exponential function, e−t/τ(r), where r denotes pixel location and τ(r) constitutes a lifetime map. Histology images were co-registered with fluorescence intensity and FLT maps. Histologically confirmed regions of interest (ROI) for tumor and normal tissue were then mapped onto the co-registered fluorescence intensity and FLT images. The intensities and FLTs from pixels enclosed by the ROIs were used to calculate probability distributions (histograms, Fig. 5HJ) for pixels as normal or tumor. ROC curves were obtained by varying the threshold for intensity and FLT and computing sensitivity and specificity. Sensitivity = (no. of pixels within the tumor with intensity or FLT above the threshold) ÷ (total no. of pixels within the tumor). Specificity (=1 − false positive rate) = (no. of pixels outside the tumor below the threshold intensity or FLT) ÷ (total number of pixels outside the tumor).

Figure 1.

In vitro and in vivo studies of panitumumab-IRDye800CW FLT in oral cancer. A, Confocal fluorescence intensity and lifetime microscopy images of FaDu oral cancer cell line after incubation with panitumumab-IRDye800CW (100 µg), IgG-IRDye800CW (100 µg), or PBS at 37°C for 2 hours, showing probe uptake and subsequent FLT enhancement of panitumumab-IRDye800CW treated cells only. Scale bar: 100 μm. B, Representative fluorescence decay curves of panitumumab-IRDye800CW (gray solid), IgG-IRDye800CW (black dashed), and PBS (gray dashed) in cancer cells are shown. FLTs are obtained as single exponential fits to the decay portion of the signal indicated by the red double arrow. C, Widefield FLT maps of culture media collected after incubation of imaging probes with cancer cells and panitumumab-IRDye800CW in PBS, showing comparable FLTs of panitumumab-IRDye800CW in culture media and in PBS. D, Representative fluorescence decay curves of panitumumab-IRDye800CW (gray solid) and IgG-IRDye800CW (black dashed) in culture media, and the stock solution of panitumumab-IRDye800CW in PBS (gray dotted). In vivo imaging of FaDu xenograft animal tumor model (∼5-mm tumor diameter, dotted outline), showing photograph (E), wide-field fluorescence intensity (F), and FLT (G) maps 48 hours after i.v. administration of panitumumab-IRDye800CW. Fluorescence intensity of the tumor was comparable with the liver and the bladder, while the wide-field FLT map delineated the tumor (black dotted outline) clearly with long FLT values observed only within the tumor. Histograms of fluorescence intensity (H) and FLTs (I) from ROIs of tumor (red) and normal (green) tissue are shown. J, ROC curves of tumor versus normal tissue discrimination based on fluorescence intensity (black dotted) and FLT (black solid) resulted in a significantly higher accuracy (AUC) of 0.99 with FLT compared with an accuracy of 0.71 for intensity-based classification. AU, arbitrary units.

Figure 1.

In vitro and in vivo studies of panitumumab-IRDye800CW FLT in oral cancer. A, Confocal fluorescence intensity and lifetime microscopy images of FaDu oral cancer cell line after incubation with panitumumab-IRDye800CW (100 µg), IgG-IRDye800CW (100 µg), or PBS at 37°C for 2 hours, showing probe uptake and subsequent FLT enhancement of panitumumab-IRDye800CW treated cells only. Scale bar: 100 μm. B, Representative fluorescence decay curves of panitumumab-IRDye800CW (gray solid), IgG-IRDye800CW (black dashed), and PBS (gray dashed) in cancer cells are shown. FLTs are obtained as single exponential fits to the decay portion of the signal indicated by the red double arrow. C, Widefield FLT maps of culture media collected after incubation of imaging probes with cancer cells and panitumumab-IRDye800CW in PBS, showing comparable FLTs of panitumumab-IRDye800CW in culture media and in PBS. D, Representative fluorescence decay curves of panitumumab-IRDye800CW (gray solid) and IgG-IRDye800CW (black dashed) in culture media, and the stock solution of panitumumab-IRDye800CW in PBS (gray dotted). In vivo imaging of FaDu xenograft animal tumor model (∼5-mm tumor diameter, dotted outline), showing photograph (E), wide-field fluorescence intensity (F), and FLT (G) maps 48 hours after i.v. administration of panitumumab-IRDye800CW. Fluorescence intensity of the tumor was comparable with the liver and the bladder, while the wide-field FLT map delineated the tumor (black dotted outline) clearly with long FLT values observed only within the tumor. Histograms of fluorescence intensity (H) and FLTs (I) from ROIs of tumor (red) and normal (green) tissue are shown. J, ROC curves of tumor versus normal tissue discrimination based on fluorescence intensity (black dotted) and FLT (black solid) resulted in a significantly higher accuracy (AUC) of 0.99 with FLT compared with an accuracy of 0.71 for intensity-based classification. AU, arbitrary units.

Close modal

FLIM and IHC image analysis

The FALCON/FLIM software was used to collect and analyze the FLIM data. Lifetime values at each pixel location were calculated by using a single exponential fitting of the fluorescence decay curves. Large area stitched FLIM and IHC images from each tissue slices were first co-registered using a custom MATLAB code. Images were then divided into multiple ROIs with a 300 × 300 pixel size. ROIs with less than 10% pixels represented by tissue were excluded from further analysis. IHC image ROIs were analyzed by color deconvolution using the IHC Toolbox in ImageJ (NIH, Version 1.48u) to extract EGFR-positive pixels within each ROI. EGFR expression level in the ROIs was represented as percentage of EGFR-positive pixels. Corresponding FLIM image ROIs were analyzed by averaging FLT values above 0.3 ns. EGFR expression and average FLT values of each pair of IHC and FLIM ROIs were compared using a scatter plot and correlation coefficient.

Statistics

Statistical analysis was carried out using Mann–Whitney U test (two-tailed) to estimate P values for box plots. P values less than 0.05 were considered significant: *, P < 0.05, and **, P < 0.01. The experiments were not randomized, and the investigators were not blinded to allocation during experiments and outcome assessment. Results are presented as mean ± SD.

Data availability statement

The data generated in this study are available upon request from the corresponding author.

In Vitro and in vivo cancer cell specificity of panitumumab-IRDye800CW FLT

We first present in vitro measurements to establish the tumor cellular specificity of FLT in a high EGFR-expressing head & neck cancer cell line, using FLIM. Figure 1A shows microscopic fluorescence intensity and FLT maps of FaDu cells, incubated with panitumumab-IRDye800CW (left), IgG-IRDye800CW (center), or PBS (right). Cells treated with panitumumab-IRDye800CW showed higher probe uptake and longer FLT values (∼0.75 ns) compared with the cells incubated with IgG-IRDye800CW, which showed low uptake and a short FLT (∼0.61 ns). Cells incubated with PBS showed basal level autofluorescence and an average FLT of 0.55 ns under the same experimental conditions. Representative fluorescence decay profiles (Fig. 1B) showed the longest decay time for panitumumab-IRDye800CW, followed by IgG-IRDye800CW and PBS in cells. Panitumumab-IRDye800CW and IgG-IRDye800CW in culture media collected from in vitro experiments showed (Fig. 1C and D) short FLTs of 0.58 ns and 0.56 ns respectively, which were comparable with the FLT of the stock solution of panitumumab-IRDye800CW in PBS (0.58 ns), indicating that environmental effects due to culture conditions did not cause the observed FLT increase of panitumumab-IRDye800CW in cancer cells. These in vitro experiments indicate an EGFR-specific uptake and FLT enhancement of panitumumab-IRDye800CW in cancer cells.

We next verified the specificity and accuracy of FLT for in vivo wide-field imaging of mouse xenograft model of FaDu tumors (n = 4 mice), following systemic injection of panitumumab-IRDye800CW. Figure 1E to J, shows a comparison of the performance of wide-field FLT and intensity imaging for tumor-normal classification. While FLT-based tumor versus normal classification resulted in 99% accuracy (measured as the AUC), intensity-based classification resulted in a significantly lower accuracy of 71%, attributed to the relatively small size of the tumor (∼5 mm) that resulted in a lower tumor uptake and a decreased tumor to normal fluorescence contrast compared with large tumors (Supplementary Fig. S1).

Tumor cell and EGFR expression-specific FLT enhancement in specimens from patients systemically injected with panitumumab-IRDye800CW

To determine the tumor specificity of panitumumab-IRDye800CW FLT in human head and neck specimens, we performed a retrospective imaging study of tissue specimens from surgery of patients with OSCC who received a systemic injection of panitumumab-IRDye800CW, 48 hours prior to surgery. Figure 2 shows FLIM (left), EGFR IHC (center), and H&E-stained histology (right) images from a representative specimen, illustrating the longer FLT in OSCC tumors corresponding to the higher EGFR expression in the tumor region. Figure 2A shows a large field of view ROI. Long FLT (red) is observed in two EGFR-overexpressing tumor clusters (Fig. 2A, arrows). Low EGFR-expressing normal tissue surrounding the tumor showed shorter FLTs (green/blue) consistent with the FLTs of nonspecific panitumumab-IRDye800CW and tissue autofluorescence. Figure 2B and C show higher magnification ROIs (shown as dashed boxes) from Fig. 2A and B, respectively. The FLIM images show long FLTs spatially colocalized within high EGFR-expressing tumor cells and the tumor specificity of FLT enhancement can be observed in individual OSCC cell clusters down to single-cell resolution (Fig. 2C, arrows).

Figure 2.

Tumor specific FLT enhancement in a clinical OSCC specimen from the lateral tongue, resected from a patient systemically injected with panitumumab-IRDye800CW 48 hours prior to surgery. A, Confocal FLIM (left), EGFR IHC (center), and H&E (right) images are shown for a representative tissue section in large FOV (∼3 cm). B, Expanded view of FLIM (left), EGFR IHC (center), and H&E (right) images within the dashed rectangular ROIs outlined in (A), showing enhanced FLT in tumor areas with high EGFR expression. C, Higher magnification (20x) images of FLIM (left), EGFR IHC (center), and H&E (right) of the dashed rectangular ROIs outlined in (B), showing individual clusters with long FLT colocalized with EGFR-expressing tumor cells. Nontumor cells showed short FLT values indicating a tumor specific FLT enhancement of panitumumab-IRDye800CW.

Figure 2.

Tumor specific FLT enhancement in a clinical OSCC specimen from the lateral tongue, resected from a patient systemically injected with panitumumab-IRDye800CW 48 hours prior to surgery. A, Confocal FLIM (left), EGFR IHC (center), and H&E (right) images are shown for a representative tissue section in large FOV (∼3 cm). B, Expanded view of FLIM (left), EGFR IHC (center), and H&E (right) images within the dashed rectangular ROIs outlined in (A), showing enhanced FLT in tumor areas with high EGFR expression. C, Higher magnification (20x) images of FLIM (left), EGFR IHC (center), and H&E (right) of the dashed rectangular ROIs outlined in (B), showing individual clusters with long FLT colocalized with EGFR-expressing tumor cells. Nontumor cells showed short FLT values indicating a tumor specific FLT enhancement of panitumumab-IRDye800CW.

Close modal

Figure 3 (and Supplementary Figs. S2–S6) shows illustrative examples of the superior tumor contrast of FLT over fluorescence intensity in tissue regions with strong nonspecific uptake of panitumumab-IRDye800CW. In Fig. 3A, the spatial distribution of fluorescence intensity and FLT is shown in a large ROI of the specimen that includes high EGFR-expressing tumor regions and low or non–EGFR-expressing normal tissue. Corresponding histology and IHC images clearly delineate the tumor boundary (dashed line) from normal salivary glands, muscle, and a layer of connective tissue. While fluorescence intensities in tumor, the salivary gland, and muscle are comparable, FLTs within the tumor cell clusters are consistently longer than the FLTs of normal tissue, and the areas of long FLTs are spatially colocalized with the areas of high EGFR expression at a microscopic level. IHC shows the highest EGFR expression in tumor followed by the salivary gland and lowest expression in muscle. This trend is closely matched by the FLIM images, which show that the FLTs in the tumor front are significantly longer (0.9–1 ns), compared with the FLTs in the salivary gland (0.7 ns) and muscle (0.5 ns). It is noted that the low uptake and short FLTs in the interior of the tumor in Fig. 3A is due to the presence of non—EGFR-expressing connective tissue, high density of lymphocytes, and a heterogeneous tumor penetration of panitumumab-IRDye800CW. A more homogeneous tumor penetration of panitumumab-IRDye800CW may be achieved using a concurrent loading dose of unlabeled panitumumab with panitumumab-IRDye800CW, as described recently (25). It is noteworthy that the FLTs of panitumumab-IRDye800CW labeled tumors are consistent across multiple patients irrespective of the anatomic location of the head and neck tumors (further quantification is discussed below).

Figure 3.

Microscopic tumor specificity of FLT in the presence of high nonspecific uptake of panitumumab-IRDye800CW. Representative confocal fluorescence intensity and FLIM images along with corresponding H&E and EGFR IHC images from clinical specimens are shown with low magnification (A) and high magnification (B, C). The low magnification (10x) images represented in (A) show the fluorescence intensity and FLT of panitumumab-IRDye800CW in tumor, muscle, salivary gland, and connective tissue that separates the tumor from the muscle and the salivary gland. The dashed lines in (A) indicate the histology defined tumor boundary. It should be noted that the long FLT (>0.9 ns) is only observed within the high EGFR-expressing tumor boundary. B, Representative high magnification (20x) images showing panitumumab-IRDye800CW uptake only in tumor cell clusters (arrow). C, Representative high magnification (20x) images of a region with high background fluorescence intensity. The FLT in the cancer cells (corresponding to high EGFR expression) showed a significantly longer FLT than the surrounding lymphocytes (dotted arrow) and muscle tissue (arrowhead). T, tumor; M, muscle; SG, salivary gland; CT, connective tissue. AU, arbitrary units.

Figure 3.

Microscopic tumor specificity of FLT in the presence of high nonspecific uptake of panitumumab-IRDye800CW. Representative confocal fluorescence intensity and FLIM images along with corresponding H&E and EGFR IHC images from clinical specimens are shown with low magnification (A) and high magnification (B, C). The low magnification (10x) images represented in (A) show the fluorescence intensity and FLT of panitumumab-IRDye800CW in tumor, muscle, salivary gland, and connective tissue that separates the tumor from the muscle and the salivary gland. The dashed lines in (A) indicate the histology defined tumor boundary. It should be noted that the long FLT (>0.9 ns) is only observed within the high EGFR-expressing tumor boundary. B, Representative high magnification (20x) images showing panitumumab-IRDye800CW uptake only in tumor cell clusters (arrow). C, Representative high magnification (20x) images of a region with high background fluorescence intensity. The FLT in the cancer cells (corresponding to high EGFR expression) showed a significantly longer FLT than the surrounding lymphocytes (dotted arrow) and muscle tissue (arrowhead). T, tumor; M, muscle; SG, salivary gland; CT, connective tissue. AU, arbitrary units.

Close modal

Figure 3B shows a representative ROI with a single cluster of few OSCC cells distant from the primary tumor mass, observed at a higher magnification (20x). In this ROI, both fluorescence intensity and FLT show high contrast from the tumor cells due to low local nonspecific uptake of panitumumab-IRDye800CW. Figure 3C shows another ROI with high nonspecific fluorescence with three additional OSCC cell clusters (arrows) surrounded by muscle and lymphocytes. These EGFR-overexpressing OSCC cells are distinct only on the FLT image and are hardly distinguishable from the surrounding normal tissue based on fluorescence intensity. All three cell clusters show average FLTs of 0.96 ns, which is significantly longer than the muscle FLT of 0.5 ns. These data indicate that while fluorescence intensity-based imaging identifies certain tumor cell clusters, many tumor cell clusters are indistinguishable from background due to nonspecific fluorescence. On the other hand, the FLT of panitumumab-IRDye800CW is consistently longer in tumor cells and is specific to EGFR expression within tumor cells, providing a robust separation of tumor and normal tissue at a microscopic level. We also confirmed that the longer tumor FLT in the oral cancer specimens does not originate from endogenous tissue autofluorescence (Supplementary Fig. S7). It is noted that because these data were collected with formalin-fixed specimen, the FLT values could be altered. However, both tumor and normal tissue FLTs should be affected by the fixation process in the same manner, so that the tumor versus normal tissue FLT contrast is maintained.

Correlation of average fluorescence lifetime with tumor-specific EGFR expression in tissue

The FLT images in Figs. 2 to 3 (and Supplementary Figs. S2–S6) indicate that regions with higher EGFR expression show a longer FLT. While we expect increased FLTs at individual foci of EGFR binding on the cell membrane or cytoplasm, resolving individual molecules within subcellular compartments is not feasible using confocal imaging. Thus, the FLT at a given pixel of a microscopic image will be the spatial average over intracellular locations that include a range of EGFR expression levels and should correlate with the average EGFR expression within the tissue region corresponding to the pixel. To study this correlation, we quantified the relationship between EGFR expression and the mean FLT in multiple tissue sections within each patient and in tissue sections from multiple patients (n = 10). Figure 4A shows IHC of three representative ROIs from muscle, salivary glands, and tumor with increasing EGFR expression. While panitumumab-IRDye800CW accumulated in EGFR-overexpressing tumor cells, there is a significant nonspecific uptake in low EGFR-expressing muscle and moderate EGFR-expressing salivary glands, making the tumor indistinguishable from normal tissue based on fluorescence intensity, as evident from the high overlap of the intensity histograms for the three ROIs (Fig. 4B). On the other hand, the FLTs of panitumumab-IRDye800CW are shortest in low EGFR-expressing muscle region and the longest in high EGFR-expressing tumor cells (Fig. 4C). The average FLTs in the representative muscle, salivary gland, and tumor ROIs are 0.4 ns, 0.5 ns, and 0.88 ns, respectively. Figure 4D shows a scatter plot of average fluorescence intensity versus the percentage of area positive for EGFR in IHC across all the ROIs studied. The scatter plot shows poor correlation (r = −0.12) indicating that panitumumab-IRDye800CW fluorescence intensity cannot reliably be used to quantify EGFR expression in the presence of strong nonspecific uptake. The distribution of fluorescence intensities in IHC confirm EGFR-negative and -positive pixels (Fig. 4E), as obtained from co-registered IHC and FLIM images, are also not statistically different. The FLT distributions for each ROI within a patient show a strong positive correlation (r = 0.87) with the percent area positive for EGFR expression (Fig. 4F). This indicates that ROIs with higher EGFR expression have longer FLTs. Additionally, EGFR positive pixels showed significantly longer average FLTs than EGFR-negative pixels (Fig. 4G).

Figure 4.

Quantification of intra- and interpatient mean fluorescence intensity (MFI) and FLT with varying EGFR expression. A, IHC images of ROIs within muscle, salivary gland, and tumor, shown in increasing order of EGFR expression (percentage of area positive for EGFR IHC). Scale bar: 100 μm. B, Corresponding confocal fluorescence intensity images and histograms showing comparable MFI for the three regions. C, Confocal FLIM images and FLT histograms of the same ROIs as in (A) show an increasing trend of FLTs with increasing EGFR expression. Black dotted lines in the histograms (B and C) represent the mean, and green shaded areas represent the SDs. D–G, Correlation analysis of fluorescence intensity and FLT against EGFR expression within specimens from a single representative patient. H–K, Correlation analysis of fluorescence intensity and FLT against EGFR expression in specimens from the entire study population (n = 10 patients). Scatter plots of average fluorescence intensity versus the percent area positive for EGFR in IHC across all ROIs for single (D) and multiple (H) patients. The trend line (gray) shows the association between fluorescence intensity and EGFR expression. Box plots indicating the distribution of fluorescence intensities in EGFR-negative and -positive pixels for single (E) and multiple (I) patients. Scatter plots of average FLT versus percentage of area positive for EGFR in IHC across all ROIs for single (F) and multiple (J) patients. Box plot of FLTs in EGFR-negative and -positive pixels of the same ROIs for single (G) and multiple (K) patients. The Pearson correlation coefficient (R) is shown in the inset in (D), (F), (H), and (J). Mann–Whitney U test (two-tailed) was used to estimate P values for the box plots. *, P < 0.01; **, P < 0.001. AU, arbitrary units.

Figure 4.

Quantification of intra- and interpatient mean fluorescence intensity (MFI) and FLT with varying EGFR expression. A, IHC images of ROIs within muscle, salivary gland, and tumor, shown in increasing order of EGFR expression (percentage of area positive for EGFR IHC). Scale bar: 100 μm. B, Corresponding confocal fluorescence intensity images and histograms showing comparable MFI for the three regions. C, Confocal FLIM images and FLT histograms of the same ROIs as in (A) show an increasing trend of FLTs with increasing EGFR expression. Black dotted lines in the histograms (B and C) represent the mean, and green shaded areas represent the SDs. D–G, Correlation analysis of fluorescence intensity and FLT against EGFR expression within specimens from a single representative patient. H–K, Correlation analysis of fluorescence intensity and FLT against EGFR expression in specimens from the entire study population (n = 10 patients). Scatter plots of average fluorescence intensity versus the percent area positive for EGFR in IHC across all ROIs for single (D) and multiple (H) patients. The trend line (gray) shows the association between fluorescence intensity and EGFR expression. Box plots indicating the distribution of fluorescence intensities in EGFR-negative and -positive pixels for single (E) and multiple (I) patients. Scatter plots of average FLT versus percentage of area positive for EGFR in IHC across all ROIs for single (F) and multiple (J) patients. Box plot of FLTs in EGFR-negative and -positive pixels of the same ROIs for single (G) and multiple (K) patients. The Pearson correlation coefficient (R) is shown in the inset in (D), (F), (H), and (J). Mann–Whitney U test (two-tailed) was used to estimate P values for the box plots. *, P < 0.01; **, P < 0.001. AU, arbitrary units.

Close modal

We next performed the same analysis using data from specimens across all the patients studied. The correlation of intensity with EGFR expression was not consistent across patients (Supplemental Fig. S8), with the interpatient analysis (Fig. 4H and I) showing a significantly lower correlation (r = 0.37) for intensity compared with FLT. Remarkably, FLT showed a high correlation (r = 0.85) with EGFR expression even at a patient level (Fig. 4J and K). The mean FLT within EGFR-expressing cancer cells was similar across multiple patients (0.97 ± 0.12 ns) to within the experimental uncertainty of FLT estimation from TD measurements (33). The results shown in Fig. 4 confirm that the FLT of panitumumab-IRDye800CW in human tissue can be used as a robust, absolute parameter to provide quantitative estimates of tissue EGFR expression in the presence of strong nonspecific uptake.

Widefield FLT imaging enhanced accuracy for tumor versus normal classification in macroscopic resection specimens

With the cellular specificity of FLT enhancement of panitumumab-IRDye800CW established, we next evaluated the ability of wide-field FLT imaging for tumor-normal classification in thick macroscopic tissue, which is relevant for in situ imaging intraoperatively and for ex vivo imaging of large resection specimens. Figure 5 shows wide-field TD reflectance imaging of the same tissue block that contained the slides used for the microscopic FLIM and histology analysis shown in Fig. 3A. The color photograph of the specimen in paraffin block (Fig. 5A) was co-registered with histology (Fig. 5B) and the tumor ROI was outlined (black dotted) by pathologists. Widefield fluorescence imaging shows a broad and heterogeneous distribution of fluorescence intensity (Fig. 5C) inside and outside the histologically defined tumor boundary (white dotted), indicating a high level of nonspecific fluorescence (white arrows in Fig. 5C) in uninvolved muscle and salivary glands that is nearly indistinguishable from the tumor fluorescence. We verified that spectral unmixing (34) using predetermined tumor and normal spectral basis functions could not clearly distinguish the tumor and normal regions (Fig. 5D and E). However, the FLTs within the tumor region are significantly longer than the FLTs of normal tissue (Fig. 5F), showing little overlap. Histograms of fluorescence intensity, spectral unmixing amplitudes, and FLTs within and outside tumor boundary are shown in Fig. 5H, I, and J, respectively. The distributions show highly overlapping fluorescence intensities and spectral amplitudes but distinct FLTs for tumor and normal ROIs with minimal overlap in the corresponding FLT distributions. An ROC analysis using the pixel intensity and FLT distributions shown in Fig. 5K, resulted in an AUC of 0.98 for FLT-based tumor/normal classification compared with an AUC of 0.32 for intensity-based classification (Fig. 5K). While the use of spectral unmixing improves the accuracy over intensity, the performance of FLT imaging is far superior to that of spectral classification (AUC = 0.58). The high accuracy of FLT imaging for tumor normal classification is confirmed by the excellent agreement of the macroscopic FLT imaging-based tumor-normal boundary with the corresponding boundary from the microscopic FLIM image of a thin section of the same slide (Fig. 5G, reproduced from Fig. 3A for ease of comparison) and the histology (Fig. 5B). While the FLT images thus far were obtained using thin tissue slices or in thick excised tissue, detection of FLT changes in vivo in biological tissue in the presence of strong optical absorbing and scattering is also feasible under a wide range of conditions (35). We have also extensively validated the accuracy of FLT imaging for detecting EGFR-overexpressing tumors in vivo using animal tumor models (ref. 31; Supplementary Fig. S1).

Figure 5.

Macroscopic wide-field TD imaging and classification of tumor versus normal tissue in clinical specimen. Representative wide-field fluorescence intensity and FLT images of oral cavity cancer specimen are shown from patients systemically injected with panitumumab-IRDye800CW 48 hours prior to surgery. The wide-field fluorescence imaging was performed in whole paraffin blocks before sectioning for histology. A, photograph, B, H&E image of a 10-µm section, and C, wide-field fluorescence intensity are shown. The tumor is indicated by dotted outlines in (A–C). Arrows in (B) indicate the locations of tumor and salivary gland. Strong fluorescence intensity was observed in nontumor areas as indicated by the arrows in (C). Multispectral imaging was performed to separate the contributions from panitumumab-IRDye800CW and tissue autofluorescence to the fluorescence intensity image presented in (C). Panitumumab-IRDye800CW amplitude (D) and tissue autofluorescence amplitude (E) are shown after spectral unmixing. F, Wide-field FLT image of the specimen showing the tumor boundary (dotted line) from the co-registered H&E image presented in (B). G, FLIM of an ROI (dashed rectangle in C–F) confirms the accuracy of tumor-normal boundary (dashed line) at the microscopic level (reproduced from Fig. 3A). Solid arrow shows muscle and dashed arrow indicates salivary glands. Distribution of fluorescence intensity (H), panitumumab-IRDye800CW amplitude measured from spectral unmixing (I), and FLTs (J) within histology defined tumor (red) and normal (green) tissue. K, ROC curves for tumor versus normal tissue classification using FLT (black solid), fluorescence intensity (black dashed), and spectral unmixing (gray solid) based on the H&E ground truth. The AUC are shown in the inset. T, tumor; S, salivary gland. AU, arbitrary units.

Figure 5.

Macroscopic wide-field TD imaging and classification of tumor versus normal tissue in clinical specimen. Representative wide-field fluorescence intensity and FLT images of oral cavity cancer specimen are shown from patients systemically injected with panitumumab-IRDye800CW 48 hours prior to surgery. The wide-field fluorescence imaging was performed in whole paraffin blocks before sectioning for histology. A, photograph, B, H&E image of a 10-µm section, and C, wide-field fluorescence intensity are shown. The tumor is indicated by dotted outlines in (A–C). Arrows in (B) indicate the locations of tumor and salivary gland. Strong fluorescence intensity was observed in nontumor areas as indicated by the arrows in (C). Multispectral imaging was performed to separate the contributions from panitumumab-IRDye800CW and tissue autofluorescence to the fluorescence intensity image presented in (C). Panitumumab-IRDye800CW amplitude (D) and tissue autofluorescence amplitude (E) are shown after spectral unmixing. F, Wide-field FLT image of the specimen showing the tumor boundary (dotted line) from the co-registered H&E image presented in (B). G, FLIM of an ROI (dashed rectangle in C–F) confirms the accuracy of tumor-normal boundary (dashed line) at the microscopic level (reproduced from Fig. 3A). Solid arrow shows muscle and dashed arrow indicates salivary glands. Distribution of fluorescence intensity (H), panitumumab-IRDye800CW amplitude measured from spectral unmixing (I), and FLTs (J) within histology defined tumor (red) and normal (green) tissue. K, ROC curves for tumor versus normal tissue classification using FLT (black solid), fluorescence intensity (black dashed), and spectral unmixing (gray solid) based on the H&E ground truth. The AUC are shown in the inset. T, tumor; S, salivary gland. AU, arbitrary units.

Close modal

Many fluorescent contrast agents and imaging systems have been developed in the past few decades for cancer diagnosis, image guided surgery, and drug development. Despite advances in identifying new cancer-specific molecular targets and imaging probes, no agent has been widely adopted for clinic use, primarily due to poor pharmacokinetics and relatively low tumor uptake (6). Conventional fluorescence imaging systems detect total fluorescence intensity reemitted from the sample. Fluorescence intensity depends on a product of probe concentration and fluorescence lifetime, and therefore cannot easily distinguish tumor specific fluorescence from nonspecific dye accumulation of probe in healthy tissue. Further, fluorescence intensity is affected by tumor size and probe uptake, making it difficult to detect small lesions in the surgical bed with sufficient accuracy. Fluorescence intensity is also strongly affected by tissue attenuation and system-specific measurement parameters. As a result, fluorescence intensity measurements cannot be readily compared across multiple specimens, subjects, and imaging systems on an absolute scale, thereby hindering standardization and ease of adoption.

Our work presents the first clinical evidence that the FLT of a cancer-targeted NIR probe systemically injected in patients is longer in cancer cells compared with normal tissue. We showed that the increased tumor FLT of the probe leads to unprecedented accuracy for tumor delineation both at a microscopic level in thin tissue sections and in macroscopic thick tissue specimens, and also allows accurate quantification of receptor expression in tissue. To our knowledge, this is the first report to demonstrate cancer specificity at a cellular level in human tissue using exogenous cancer-targeted agents, although cancer cellular uptake has previously been demonstrated in vitro and in animal models (15). Besides an improved accuracy for tumor detection, an important feature of FLT is that it is not affected by system-specific measurement parameters, such as light illumination power and camera sensitivity. Therefore, FLT can serve as an absolute parameter that can be readily compared across multiple imaging systems and studies, facilitating better standardization in image guided surgery (4).

The cellular specificity of FLT to cancer has relevance beyond microscopic imaging of thin tissue sections, and can be exploited for imaging tumors in deep tissue. FLT measurements are unaltered by tissue light propagation under a wide range of conditions (33, 35) and can be estimated in the presence of thick tissue without the knowledge of tissue optical properties, which can often be challenging to estimate. Therefore, the cellular specificity of FLT to cancer guarantees that FLTs measured through thick biological tissue can be attributed solely to tumor-bound probe, provided the FLTs are above a predetermined threshold. In comparison, fluorescence intensity is strongly attenuated by tissue light propagation, thereby requiring a full knowledge of tissue optical properties and tissue thickness to accurately quantify probe uptake. The ability to measure FLTs through deep tissue can be useful when the detection of tumors embedded in thick macroscopic tissue is necessary, such as for the evaluation of margin depth in resection specimens (2, 17, 19) or when imaging deep seated tumors noninvasively in whole organs (8). We have shown that in vivo FLTs can be detected and localized in deep tissue using tomographic reconstruction algorithms (36). These methods exploit the independence of FLT to tissue scattering and absorption for in vivo FLTs that are longer than the intrinsic light diffusion timescales in tissue (∼0.2 ns; ref. 35). The latter condition is well satisfied for many NIR fluorophores including IRDye800CW. Diagnostic systems can be envisaged that quantify cancer-related biomarkers using whole body measurements, or to provide rapid on/off “optical switch” readouts based on predetermined FLT thresholds.

FLT imaging has been previously applied for preclinical studies at the microscopic and whole animal level (37–40). While visible FLIM has been evaluated for image guided surgery exploiting endogenous FLTs of tissue components (41, 42), endogenous FLT contrast between tumor and normal tissue is inherently poor (also see Supplementary Fig. S7), resulting in significantly lower sensitivity and specificity for tumor versus normal classification compared to FLT imaging with exogenous agents. Further, endogenous fluorescence imaging systems use visible light and are therefore more susceptible to ambient light interference. Visible fluorescence also precludes the ability to image subsurface tumors due to strong tissue attenuation, thereby limiting intraoperative applications to exposed surface tumors. NIR agents can exploit the greater depth sensitivity of NIR light for intraoperative or deep tissue imaging. In addition, exogenous targeted agents are essential for reporting on molecular expression markers. Nevertheless, endogenous FLIM in the visible spectrum can clearly delineate various tissue structures that could provide valuable morphologic information to complement the NIR tumor signal from exogenous agents.

Our data suggests that spectral unmixing techniques can alleviate poor tumor contrast due to nonspecific uptake to some extent and can be useful when FLT imaging systems are not readily available. The tumor versus normal spectral contrast is still not sufficiently high to provide a tumor detection accuracy comparable with that using FLT. It should be noted that spectral contrast is essentially an intensity based measure, and therefore suffers from the same limitations as intensity, viz., dependence on measurement parameters, tissue absorption, and scattering, and is thus harder to quantify in thick biological tissue (43).

The utility and safety of EGFR antibody-labeled NIR probes has been extensively studied for OSCC, pancreatic, and brain tumors. Although these studies show significant improvement in sensitivity compared with visual identification and palpation, intensity is not always reliable because nonspecific uptake in tissue is heterogeneous and can vary across multiple regions within a given specimen. This is clearly illustrated by our data where intensity performs well in some oral tissue regions with good tumor uptake, while high nonspecific uptake is present in other areas of oral tissue such as salivary glands, significantly diminishing tumor contrast. FLTs of tumors, on the other hand are consistently and uniformly longer than normal oral tissue and therefore provide a robust measure of tumor uptake.

We are actively investigating the mechanism for FLT increase in tumors, which likely involves two steps: probe internalization, followed by the influence of local intracellular environment on the residualized probe. It is well known that following internalization and endosomal uptake of receptor-bound panitumumab-IRDye800CW, IRDye 800CW is residualized into lysosomes (44). Therefore, we expect the tumor lysosomal environment, including local solvent polarity, pH, and viscosity, to play a role in the observed FLT enhancement. Because IRDye 800CW does not possess sulphate and amine groups that can be protonated at the physiologic pH range, the local pH variations between tumor and normal tissue are likely insufficient to cause the significant FLT enhancement observed in our preclinical and clinical studies. On the other hand, lysosomal viscosity is increased in tumor cells (∼40 cP) compared with normal cells (∼30 cP; ref. 45). As higher viscosity is known to enhance the quantum yield and FLT optical probes (30, 46), lysosomal viscosity could potentially be a key factor contributing to the lifetime increase of IRDye 800CW in tumor cells. Further studies to evaluate the role of viscosity and pH in the observed lifetime increase in tumor cells are underway and will be reported in future work.

Because panitumumab-IRDye800CW has been extensively tested for safety in humans (3), the results presented here have important clinical relevance for intraoperative surgical guidance in EGFR-overexpressing cancers. Over 90% of head and neck cancers overexpress EGFR (47). Besides the multiple clinical trials of anti-EGFR antibody-labeled probes for OSCC, clinical trials of EGFR-antibody–based probes have been conducted in brain (48), colorectal (49), and pancreatic cancers (50). FLT imaging using panitumumab-IRDye800CW is therefore likely to strongly impact surgical guidance for these cancers as well. In addition to EGFR targeting, FLT contrast is also likely to benefit tumor imaging using other receptor-targeted probes. Early preclinical studies have shown tumor specific FLT changes of fluorescently labeled affibody for HER2 in mice (51). Also, our ongoing work in animals indicates that probes targeted to immune expression markers exhibit longer lifetimes in tumor cells compared with normal tissue. As novel probes to target cancer-specific molecular markers continue to be developed, we anticipate that FLT will be applicable to these newly developed agents as well. Given its powerful and unique benefits, it is conceivable that FLT imaging using targeted molecular imaging agents would play an important role in a wide range of clinical settings ranging from cancer diagnostics to surgical therapy.

A.T.N. Kumar reports a patent for 63/272,847 pending to The General Hospital Corporation. No disclosures were reported by the other authors.

R. Pal: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M.E. Hom: Resources, data curation, validation, visualization, writing–review and editing. N.S. van den Berg: Resources, data curation, validation, investigation, writing–review and editing. T.M. Lwin: Data curation, investigation, visualization, writing–review and editing. Y.-J. Lee: Resources, data curation, validation, visualization. A. Prilutskiy: Formal analysis, validation. W. Faquin: Resources, validation, investigation. E. Yang: Resources, validation. S.V. Saladi: Resources, validation. M.A. Varvares: Conceptualization, resources, validation, investigation. E.L. Rosenthal: Resources, data curation, supervision, funding acquisition, validation. A.T.N. Kumar: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We acknowledge the following funding sources: NIH grants R01-CA211084, R01-CA260857, and MGH-Executive Committee on Research Interim Support Fund. We acknowledge Philipp Isermann, Daniel Tom, and Linda Liao from Leica Microsystems, and the Harvard Medical School (Boston, MA) MicRoN for the loan of Stellaris 8 used in this work, and for the extensive assistance in imaging.

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.

1.
Azari
F
,
Kennedy
G
,
Bernstein
E
,
Hadjipanayis
C
,
Vahrmeijer
AL
,
Smith
BL
, et al
.
Intraoperative molecular imaging clinical trials: a review of 2020 conference proceedings
.
J Biomed Opt
2021
;
26
:
050901
.
2.
Van Keulen
S
,
Van Den Berg
NS
,
Nishio
N
,
Birkeland
A
,
Zhou
Q
,
Lu
G
, et al
.
Rapid, non-invasive fluorescence margin assessment: Optical specimen mapping in oral squamous cell carcinoma
.
Oral Oncol
2019
;
88
:
58
65
.
3.
Gao
RW
,
Teraphongphom
N
,
De Boer
E
,
Berg
NSVD
,
Divi
V
,
Kaplan
MJ
, et al
.
Safety of panitumumab-IRDye800CW and cetuximab-IRDye800CW for fluorescence-guided surgical navigation in head and neck cancers
.
Theranostics
2018
;
8
:
2488
.
4.
Dsouza
AV
,
Lin
H
,
Henderson
ER
,
Samkoe
KS
,
Pogue
BW.
Review of fluorescence guided surgery systems: identification of key performance capabilities beyond indocyanine green imaging
.
J Biomed Opt
2016
;
21
:
080901
.
5.
Lee
JYK
,
Cho
SS
,
Stummer
W
,
Tanyi
JL
,
Vahrmeijer
AL
,
Rosenthal
E
, et al
.
Review of clinical trials in intraoperative molecular imaging during cancer surgery
.
J Biomed Opt
2019
;
24
:
120901
.
6.
Weissleder
R
.
Molecular imaging in cancer
.
Science
2006
;
312
:
1168
71
.
7.
Sevick-Muraca
EM
,
Sharma
R
,
Rasmussen
JC
,
Marshall
MV
,
Wendt
JA
,
Pham
HQ
, et al
.
Imaging of lymph flow in breast cancer patients after microdose administration of a near-infrared flurophore: Feasibility study
.
Radiology
2008
;
246
:
734
41
.
8.
Corlu
A
,
Choe
R
,
Durduran
T
,
Rosen
MA
,
Schweiger
M
,
Arridge
SR
, et al
.
Three-dimensional in vivo fluorescence diffuse optical tomography of breast cancer in humans
.
Opt Express
2007
;
15
:
6696
716
.
9.
Poellinger
A
,
Burock
S
,
Grosenick
D
,
Hagen
A
,
Lüdemann
L
,
Diekmann
F
, et al
.
Breast cancer: early- and late-fluorescence near-infrared imaging with indocyanine green—a preliminary study
.
Radiology
2011
;
258
:
409
16
.
10.
Rosenthal
EL
,
Warram
JM
,
De Boer
E
,
Chung
TK
,
Korb
ML
,
Brandwein-Gensler
M
, et al
.
Safety and tumor specificity of cetuximab-IRDye800 for surgical navigation in head and neck cancer
.
Clinical Cancer Res
2015
;
21
:
3658
66
.
11.
Boogerd
LS
,
Hoogstins
CE
,
Schaap
DP
,
Kusters
M
,
Handgraaf
HJ
,
van der Valk
MJ
, et al
.
Safety and effectiveness of SGM-101, a fluorescent antibody targeting carcinoembryonic antigen, for intraoperative detection of colorectal cancer: a dose-escalation pilot study
.
Lancet Gastroenter Hepatol
2018
;
3
:
181
91
.
12.
Samkoe
KS
,
Sardar
HS
,
Bates
BD
,
Tselepidakis
NN
,
Gunn
JR
,
Hoffer‐Hawlik
KA
, et al
.
Preclinical imaging of epidermal growth factor receptor with ABY-029 in soft-tissue sarcoma for fluorescence-guided surgery and tumor detection
.
J Surg Oncol
2019
;
119
:
1077
86
.
13.
Hoogstins
CES
,
Tummers
QRJG
,
Gaarenstroom
KN
,
De Kroon
CD
,
Trimbos
JBMZ
,
Bosse
T
, et al
.
A novel tumor-specific agent for intraoperative near-infrared fluorescence imaging: a translational study in healthy volunteers and patients with ovarian cancer
.
Clin Cancer Res
2016
;
22
2929
38
.
14.
Tummers
QRJG
,
Hoogstins
CES
,
Gaarenstroom
KN
,
De Kroon
CD
,
Van Poelgeest
MIE
,
Vuyk
J
, et al
.
Intraoperative imaging of folate receptor alpha positive ovarian and breast cancer using the tumor specific agent EC17
.
Oncotarget
2016
;
7
:
32144
.
15.
Shen
D
,
Xu
B
,
Liang
K
,
Tang
R
,
Sudlow
GP
,
Egbulefu
C
, et al
.
Selective imaging of solid tumours via the calcium-dependent high-affinity binding of a cyclic octapeptide to phosphorylated Annexin A2
.
Nat Biomed Eng
2020
;
4
:
298
313
.
16.
Smith
BL
,
Gadd
MA
,
Lanahan
CR
,
Rai
U
,
Tang
R
,
Rice-Stitt
T
, et al
.
Real-time, intraoperative detection of residual breast cancer in lumpectomy cavity walls using a novel cathepsin-activated fluorescent imaging system
.
Breast Cancer Rese Treat
2018
;
171
:
413
20
.
17.
Van Keulen
S
,
Nishio
N
,
Birkeland
A
,
Fakurnejad
S
,
Martin
B
,
Forouzanfar
T
, et al
.
The Sentinel margin: intraoperative ex vivo specimen mapping using relative fluorescence intensity
.
Clin Cancer Res
2019
;
25
:
4656
62
.
18.
Warram
JM
,
de Boer
E
,
van Dam
GM
,
Moore
LS
,
Bevans
SL
,
Walsh
EM
, et al
.
Fluorescence imaging to localize head and neck squamous cell carcinoma for enhanced pathological assessment
.
J pathol Clin Res
2016
;
2
:
104
12
.
19.
Teraphongphom
N
,
Kong
CS
,
Warram
JM
,
Rosenthal
EL
.
Specimen mapping in head and neck cancer using fluorescence imaging
.
Laryngoscope Investig Otolaryngol
2017
;
2
:
447
52
.
20.
Rosenthal
EL
,
Kulbersh
BD
,
King
T
,
Chaudhuri
TR
,
Zinn
KR
.
Use of fluorescent labeled anti-epidermal growth factor receptor antibody to image head and neck squamous cell carcinoma xenografts
.
Mol Cancer Ther
2007
;
6
:
1230
8
.
21.
Pomerantz
RG
,
Grandis
JR
.
The role of epidermal growth factor receptor in head and neck squamous cell carcinoma
.
Curr Oncol Rep
2003
;
5
:
140
6
.
22.
Selvaggi
G
,
Novello
S
,
Torri
V
,
Leonardo
E
,
De Giuli
P
,
Borasio
P
, et al
.
Epidermal growth factor receptor overexpression correlates with a poor prognosis in completely resected non-small-cell lung cancer
.
Ann Oncol
2004
;
15
:
28
32
.
23.
Xu
H
,
Zong
H
,
Ma
C
,
Ming
X
,
Shang
M
,
Li
K
, et al
.
Epidermal growth factor receptor in glioblastoma
.
Oncol Lett
2017
;
14
:
512
6
.
24.
Lofgren
JA
,
Dhandapani
S
,
Pennucci
JJ
,
Abbott
CM
,
Mytych
DT
,
Kaliyaperumal
A
, et al
.
Comparing ELISA and surface plasmon resonance for assessing clinical immunogenicity of panitumumab
.
J Immunol
2007
;
178
:
7467
72
.
25.
Moore
LS
,
Rosenthal
EL
,
De Boer
E
,
Prince
AC
,
Patel
N
,
Richman
JM
, et al
.
Effects of an unlabeled loading dose on tumor-specific uptake of a fluorescently labeled antibody for optical surgical navigation
.
Mol Imaging Biol
2017
;
19
:
610
6
.
26.
Li
C
,
Xu
X
,
McMahon
N
,
Alhaj Ibrahim
O
,
Sattar
HA
,
Tichauer
KM
.
Paired-agent fluorescence molecular imaging of sentinel lymph nodes using indocyanine green as a control agent for antibody-based targeted agents
.
Contrast Media Mol Imaging
2019
;
2019
:
7561862
.
27.
Pogue
BW
,
Rosenthal
EL
.
Review of successful pathways for regulatory approvals in open-field fluorescence-guided surgery
.
J Biomed Opt
2021
;
26
:
030901
.
28.
Tummers
WS
,
Warram
JM
,
Tipirneni
KE
,
Fengler
J
,
Jacobs
P
,
Shankar
L
, et al
.
Regulatory aspects of optical methods and exogenous targets for cancer detection
.
Cancer Res
2017
;
77
:
2197
206
.
29.
Berezin
MY
,
Achilefu
S
.
Fluorescence lifetime measurements and biological imaging
.
Chem Rev
2010
;
110
:
2641
84
.
30.
Lakowicz
JR
.
Principles of fluorescence spectroscopy
. 2nd ed.
Switzerland
:
Springer
;
1999
.
31.
Pal
R
,
Kang
H
,
Choi
HS
,
Kumar
ATN
.
Fluorescence lifetime-based tumor contrast enhancement using an EGFR antibody–labeled near-infrared fluorophore
.
Clin Cancer Res
2019
;
25
:
6653
61
.
32.
Pei
J
,
Juniper
G
,
Van Den Berg
NS
,
Nisho
N
,
Broadt
T
,
Welch
AR
, et al
.
Safety and stability of antibody-dye conjugate in optical molecular imaging
.
Mol Imaging Biol
2021
;
23
:
109
16
.
33.
Raymond
SB
,
Boas
DA
,
Bacskai
BJ
,
Kumar
TN
.
Lifetime-based tomographic multiplexing
.
J Biomed Opt
2010
;
15
:
046011
.
34.
Xu
H
,
Rice
BW
.
In-vivo fluorescence imaging with a multivariate curve resolution spectral unmixing technique
.
J Biomed Opt
2009
;
14
:
064011
.
35.
Kumar
AT
,
Raymond
SB
,
Boverman
G
,
Boas
DA
,
Bacskai
BJ
.
Time resolved fluorescence tomography of turbid media based on lifetime contrast
.
Opt Express
2006
;
14
:
12255
70
.
36.
Rice
WL
,
Shcherbakova
DM
,
Verkhusha
VV
,
Kumar
ATN
.
In vivo tomographic imaging of deep seated cancer using fluorescence lifetime contrast
.
Cancer Res
2015
;
75
:
1236
43
.
37.
Bacskai
BJ
,
Skoch
J
,
Hickey
GA
,
Allen
R
,
Hyman
BT
.
Fluorescence resonance energy transfer determinations using multiphoton fluorescence lifetime imaging microscopy to characterize amyloid-beta plaques
.
J Biomed Opt
2003
;
8
:
368
75
.
38.
Bloch
S
,
Lesage
F
,
Mcintosh
L
,
Gandjbakhche
A
,
Liang
K
,
Achilefu
S
.
Whole-body fluorescence lifetime imaging of a tumor-targeted near-infrared molecular probe in mice
.
J Biomed Opt
2005
;
10
:
054003
.
39.
Nothdurft
RE
,
Patwardhan
SV
,
Akers
W
,
Ye
Y
,
Achilefu
S
,
Culver
JP
.
In vivo fluorescence lifetime tomography
.
J Biomed Opt
2009
;
14
:
7
.
40.
Goergen
CJ
,
Chen
HH
,
Bogdanov
A
,
Sosnovik
DE
,
Kumar
ATN
.
In vivo fluorescence lifetime detection of an activatable probe in infarcted myocardium
.
J Biomed Opt
2012
;
17
:
056001
.
41.
Farwell
DG
,
Meier
JD
,
Park
J
,
Sun
Y
,
Coffman
H
,
Poirier
B
, et al
.
Time-resolved fluorescence spectroscopy as a diagnostic technique of oral carcinoma: validation in the hamster buccal pouch model
.
Arch Otolaryngol Head Neck Surg
2010
;
136
:
126
133
.
42.
Weyers
BW
,
Marsden
M
,
Sun
T
,
Bec
J
,
Bewley
AF
,
Gandour‐Edwards
RF
, et al
.
Fluorescence lifetime imaging for intraoperative cancer delineation in transoral robotic surgery
.
Transl Biophotonics
2019
;
1
:
e201900017
.
43.
Hou
S
,
Bacskai
BJ
,
Kumar
ATN
.
Comparison of tomographic spectral and lifetime multiplexing
.
Opt Lett
2016
;
41
:
5337
40
.
44.
Cilliers
C
,
Liao
J
,
Atangcho
L
,
Thurber
GM
.
Residualization rates of near-infrared dyes for the rational design of molecular imaging agents
.
Mol Imaging Biol
2015
;
17
:
757
62
.
45.
Chowdhury
R
,
Saha
A
,
Mandal
AK
,
Jana
B
,
Ghosh
S
,
Bhattacharyya
K
.
Excited state proton transfer in the lysosome of live lung cells: normal and cancer cells
.
J Phys Chem B
2015
;
119
:
2149
-
56
.
46.
Miao
W
,
Yu
C
,
Hao
E
,
Jiao
L
.
Functionalized BODIPYs as fluorescent molecular rotors for viscosity detection
.
Front Chem
2019
;
7
:
825
.
47.
Kalyankrishna
S
,
Grandis
JR
.
Epidermal growth factor receptor biology in head and neck cancer
.
J Clin Oncol
2006
;
24
:
2666
72
.
48.
Zhou
Q
,
Van Den Berg
NS
,
Rosenthal
EL
,
Iv
M
,
Zhang
M
,
Vega Leonel
JCM
, et al
.
EGFR-targeted intraoperative fluorescence imaging detects high-grade glioma with panitumumab-IRDye800 in a phase 1 clinical trial
.
Theranostics
,
2021
;
11
:
7130
.
49.
Marston
JC
,
Kennedy
GD
,
Lapi
SE
,
Hartman
YE
,
Richardson
MT
,
Modi
HM
, et al
.
Panitumumab-IRDye800CW for fluorescence-guided surgical resection of colorectal cancer
.
J Surg Res
2019
;
239
:
44
51
.
50.
Lu
G
,
van den Berg
NS
,
Martin
BA
,
Nishio
N
,
Hart
ZP
,
van Keulen
S
, et al
.
Tumour-specific fluorescence-guided surgery for pancreatic cancer using panitumumab-IRDye800CW: a phase 1 single-centre, open-label, single-arm, dose-escalation study
.
Lancet Gastroenter Hepatol
2020
;
5
:
753
64
.
51.
Ardeshirpour
Y
,
Chernomordik
V
,
Hassan
M
,
Zielinski
R
,
Capala
J
,
Gandjbakhche
A
.
In vivo fluorescence lifetime imaging for monitoring the efficacy of the cancer treatment
.
Clin Cancer Res
2014
;
20
:
3531
9
.

Supplementary data